1 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3 PR rtl-optimization/82185
4 * expmed.c (emit_store_flag_int): Only test tem if it has been
7 2017-09-12 Richard Biener <rguenther@suse.de>
10 * match.pd ((FTYPE) N CMP CST): Fix typo.
12 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
14 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16 (mips_near_type_p): Add 'short_call' attribute as a synonym
18 * doc/extend.texi (short_call): Document new function attribute.
20 2017-09-12 Jakub Jelinek <jakub@redhat.com>
23 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
24 assertion check that in the condition.
25 (get_atomic_generic_size): Likewise. Before testing if parameter
26 has pointer type, if it has array type, call for C++
27 default_conversion to perform array-to-pointer conversion.
29 2017-09-12 Richard Biener <rguenther@suse.de>
31 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
32 for operations we cannot scalarize.
34 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
36 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
37 vectors heap vectors. Clean up comments.
38 Make visited_bbs a reference.
39 (profitable_jump_thread_path): Make GC
40 vectors heap vectors. Clean up comments.
42 (convert_and_register_jump_thread_path): Make GC vectors heap
44 (check_subpath_and_update_thread_path): Same. Clean up comments.
45 Make visited_bbs a reference.
46 (handle_phi): Abstract common code to to
47 register_jump_thread_path_if_profitable.
48 Rename VAR_BB to DEF_BB.
50 Make GC vectors heap vectors.
51 Make visited_bbs a reference.
52 (handle_assignment): Same.
53 (register_jump_thread_path_if_profitable): New.
54 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
56 Make GC vectors heap vectors. Clean up comments.
57 Make visited_bbs a reference.
58 (find_jump_threads_backwards): Make visited_bbs live in the stack.
59 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
62 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
65 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
66 words of E_DImode object are reachable by xtensa_uimm8x4 access.
68 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
70 Revert r251800 and r251799.
72 2017-09-11 Martin Jambor <mjambor@suse.cz>
75 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
77 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
78 use it to find predecessor edges.
79 (naive_outof_ssa): Collect vector of predecessors.
81 2017-09-08 Jason Merrill <jason@redhat.com>
83 PR c++/70029 - ICE with ref-qualifier and -flto
84 * langhooks.h (struct lang_hooks_for_types): Add
86 * attribs.c (build_type_attribute_qual_variant): Use it.
87 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
89 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
90 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
92 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
95 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
96 (*mulsi3_sp64): New instruction.
97 (mulsi3): New expander.
99 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
101 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
103 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
105 * sancov.c: Include memmodel.h.
107 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
110 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
113 2017-09-07 Carl Love <cel@us.ibm.com>
115 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
116 the sldi instruction.
118 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
120 * sancov.c: Include tm_p.h.
122 2017-09-07 Jakub Jelinek <jakub@redhat.com>
125 * output.h (switch_to_other_text_partition): New declaration.
126 * varasm.c (switch_to_other_text_partition): New function.
127 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
128 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
129 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
130 to the other text partition before emitting LCL label and switch back
131 after emitting the word after it.
133 2017-09-07 Richard Biener <rguenther@suse.de>
135 * passes.def (pass_split_crit_edges): Remove instance before PRE.
136 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
137 critical edges here, after loop init.
138 (pass_data_pre): Remove PROP_no_crit_edges flags.
139 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
140 for valueization of call args to avoid leaking VN_TOP.
141 (visit_use): Assert we do not visit default defs.
142 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
143 Use error_mark_node to more easily detect leaking VN_TOP.
144 All default-defs are varying, not VN_TOP. Mark them visited.
145 (run_scc_vn): Make code match comment.
147 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
149 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
150 OPTION_MASK_FLOAT128_KEYWORD.
151 (POWERPC_MASKS): Likewise.
152 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
153 support for the -mfloat128-type option, and make -mfloat128
154 default on PowerPC Linux systems. Define or undefine
155 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
156 Define __float128 to be __ieee128 if IEEE 128-bit support is
157 enabled, or undefine it.
158 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
159 Delete defining __FLOAT128_TYPE__.
160 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
161 -mfloat128-type option and make -mfloat128 default on PowerPC
163 (TARGET_FLOAT128_TYPE): Likewise.
164 (-mfloat128-type): Likewise.
165 * config/rs6000/rs6000.c (rs6000_option_override_internal):
166 Delete the -mfloat128-type option and make -mfloat128 default on
167 PowerPC Linux systems. Always use __ieee128 to be the keyword for
168 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
169 128-bit floating point is enabled. Change tests from using
170 -mfloat128-type to -mfloat128.
171 (rs6000_mangle_type): Use the correct mangling for the __float128
172 type even if normal long double is restricted to 64-bits.
173 (floatn_mode): Enable the _Float128 type by default on VSX Linux
175 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
176 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
177 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
179 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
180 documentation for -mfloat128.
182 2017-09-06 Olivier Hainque <hainque@adacore.com>
184 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
186 2017-09-06 Wish Wu <wishwu007@gmail.com>
187 Jakub Jelinek <jakub@redhat.com>
189 * asan.c (initialize_sanitizer_builtins): Add
190 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
191 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
192 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
193 BT_FN_VOID_UINT64_PTR variables.
194 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
195 (BT_FN_VOID_UINT16_UINT16): Likewise.
196 (BT_FN_VOID_UINT32_UINT32): Likewise.
197 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
198 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
199 (BT_FN_VOID_UINT64_PTR): Likewise.
200 * common.opt (flag_sanitize_coverage): New variable.
201 (fsanitize-coverage=trace-pc): Remove.
202 (fsanitize-coverage=): Add.
203 * flag-types.h (enum sanitize_coverage_code): New enum.
204 * fold-const.c (fold_range_test): Disable non-short-circuit
205 optimization if flag_sanitize_coverage.
206 (fold_truth_andor): Likewise.
207 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
208 * opts.c (COVERAGE_SANITIZER_OPT): Define.
209 (coverage_sanitizer_opts): New array.
210 (get_closest_sanitizer_option): Add OPTS argument, handle also
211 OPT_fsanitize_coverage_.
212 (parse_sanitizer_options): Adjusted to also handle
213 OPT_fsanitize_coverage_.
214 (common_handle_option): Add OPT_fsanitize_coverage_.
215 * sancov.c (instrument_comparison, instrument_switch): New function.
216 (sancov_pass): Add trace-cmp support.
217 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
218 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
219 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
220 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
221 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
222 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
223 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
224 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
225 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
226 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
228 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
230 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
231 error. Only quit immediately if parsing is complete.
232 (BEGIN): Initialize fatal_err and parse_done.
233 (begin fpu, end fpu): Check number of arguments.
234 (begin arch, end arch): Likewise.
235 (begin cpu, end cpu): Likewise.
236 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
237 (optalias): Likewise.
239 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
241 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
242 * config/arm/arm-isa.h: Delete. Move definitions to ...
243 * arm-cpus.in: ... here. Use new feature and fgroup values.
244 * config/arm/arm.c (arm_option_override): Use lower case for feature
246 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
247 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
248 * config/arm/parsecpu.awk (END): Add new command 'isa'.
250 (print_isa_bits_for): New function.
251 (gen_isa): New function.
252 (gen_comm_data): Use print_isa_bits_for.
253 (define feature): New keyword.
254 (define fgroup): New keyword.
255 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
256 (arm-isa.h): Add rule to generate file.
257 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
258 case for feature bit names.
260 2017-09-06 Richard Biener <rguenther@suse.de>
262 * tree-ssa-pre.c (NECESSARY): Remove.
263 (create_expression_by_pieces): Do not touch pass-local flags.
264 (insert_into_preds_of_block): Likewise.
265 (do_pre_regular_insertion): Likewise.
266 (eliminate_insert): Likewise.
267 (eliminate_dom_walker::before_dom_children): Likewise.
268 (fini_eliminate): Do not look at inserted_exprs.
269 (mark_operand_necessary): Remove.
270 (remove_dead_inserted_code): Replace with simple work-list
271 algorithm based on inserted_exprs and SSA uses.
272 (pass_pre::execute): Re-order fini_eliminate and
273 remove_dead_inserted_code.
275 2017-09-06 Olivier Hainque <hainque@adacore.com>
277 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
278 for VxWorks 7. Adjust surrounding comments.
280 2017-09-06 Richard Biener <rguenther@suse.de>
282 * gimple-ssa-strength-reduction.c
283 (find_candidates_dom_walker::before_dom_children): Also allow
286 2017-09-06 Richard Biener <rguenther@suse.de>
288 PR tree-optimization/82108
289 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
290 for gap in the non-permutation SLP case.
292 2017-09-06 Martin Jambor <mjambor@suse.cz>
294 PR tree-optimization/82078
295 * tree-sra.c (sort_and_splice_var_accesses): Move call to
296 add_access_to_work_queue...
297 (build_accesses_from_assign): ...here.
298 (propagate_all_subaccesses): Make sure racc is the group
299 representative, if there is one.
301 2017-09-06 Jakub Jelinek <jakub@redhat.com>
304 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
307 2017-09-06 Richard Biener <rguenther@suse.de>
309 * gimple-ssa-strength-reduction.c
310 (find_candidates_dom_walker::before_doom_children): Use a
311 type and not a mode check.
313 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
316 * config/arm/predicates.md (arm_general_adddi_operand): Create new
318 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
320 2017-09-05 Jeff Law <law@redhat.com>
322 PR tree-optimization/64910
323 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
324 cases where we have 3 or more operands.
326 2017-09-05 Jakub Jelinek <jakub@redhat.com>
329 * omp-low.c (lower_omp_for): Recompute tree invariant if
330 gimple_omp_for_initial/final is ADDR_EXPR.
333 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
334 into gimple val before gimplification fo the COND_EXPR.
336 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
338 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
339 REGION_COPY argument.
340 (thread_through_all_blocks): Remove unused argument to
341 duplicate_thread_path.
343 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
344 Alan Hayward <alan.hayward@arm.com>
345 David Sherwood <david.sherwood@arm.com>
347 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
348 Take a scalar_mode rather than a machine_mode.
349 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
350 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
351 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
352 (aarch64_gen_adjusted_ldpstp): Likewise.
353 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
355 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
356 Alan Hayward <alan.hayward@arm.com>
357 David Sherwood <david.sherwood@arm.com>
359 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
360 Take a scalar_int_mode instead of a machine_mode.
361 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
362 (aarch64_output_scalar_simd_mov_immediate): Likewise.
363 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
364 (aarch64_simd_attr_length_rglist): Delete.
365 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
366 a scalar_int_mode instead of a machine_mode.
367 (aarch64_add_offset): Likewise.
368 (aarch64_internal_mov_immediate): Likewise
369 (aarch64_add_constant_internal): Likewise.
370 (aarch64_add_constant): Likewise.
371 (aarch64_movw_imm): Likewise.
372 (aarch64_rtx_arith_op_extract_p): Likewise.
373 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
374 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
375 Remove assert that the mode isn't a vector.
376 (aarch64_output_scalar_simd_mov_immediate): Likewise.
377 (aarch64_expand_mov_immediate): Update calls after above changes.
378 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
379 (aarch64_and_bitmask_imm): Check for scalar integer modes.
380 (aarch64_move_imm): Likewise.
381 (aarch64_can_const_movi_rtx_p): Likewise.
382 (aarch64_strip_extend): Likewise.
383 (aarch64_extr_rtx_p): Likewise.
384 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
385 a CONST_INT when the mode parameter is VOIDmode.
386 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
388 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
390 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
391 * stor-layout.c (bitwise_mode_for_mode): Likewise.
392 (bitwise_type_for_mode): Update accordingly.
394 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
396 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
397 * stor-layout.c (mode_for_size_tree): Likewise.
398 (mode_for_array): Update accordingly.
399 (layout_decl): Likewise.
400 (compute_record_mode): Likewise. Only set the mode once.
402 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
404 * target.def (get_mask_mode): Change return type to opt_mode.
406 * doc/tm.texi: Regenerate.
407 * targhooks.h (default_get_mask_mode): Return an opt_mode.
408 * targhooks.c (default_get_mask_mode): Likewise.
409 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
410 * optabs-query.c (can_vec_mask_load_store_p): Update use of
411 targetm.get_mask_mode.
412 * tree.c (build_truth_vector_type): Likewise.
414 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
416 * machmode.h (mode_for_vector): Return an opt_mode.
417 * stor-layout.c (mode_for_vector): Likewise.
418 (mode_for_int_vector): Update accordingly.
419 (layout_type): Likewise.
420 * config/i386/i386.c (emit_memmov): Likewise.
421 (ix86_expand_set_or_movmem): Likewise.
422 (ix86_expand_vector_init): Likewise.
423 (ix86_get_mask_mode): Likewise.
424 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
426 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
427 * expmed.c (extract_bit_field_1): Likewise.
428 * expr.c (expand_expr_real_2): Likewise.
429 * optabs-query.c (can_vec_perm_p): Likewise.
430 (can_vec_mask_load_store_p): Likewise.
431 * optabs.c (expand_vec_perm): Likewise.
432 * targhooks.c (default_get_mask_mode): Likewise.
433 * tree-vect-stmts.c (vectorizable_store): Likewise.
434 (vectorizable_load): Likewise.
435 (get_vectype_for_scalar_type_and_size): Likewise.
437 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
439 * machmode.h (mode_for_int_vector): New function.
440 * stor-layout.c (mode_for_int_vector): Likewise.
441 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
442 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
443 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
444 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
445 (s390_expand_vcond): Likewise.
447 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
449 * machmode.h (opt_machine_mode): New type.
450 (opt_mode<T>): Allow construction from anything that can be
452 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
453 (mode_for_size): Return an opt_machine_mode.
454 * stor-layout.c (mode_for_size): Likewise.
455 (mode_for_size_tree): Update call accordingly.
456 (bitwise_mode_for_mode): Likewise.
457 (make_fract_type): Likewise.
458 (make_accum_type): Likewise.
459 * caller-save.c (replace_reg_with_saved_mem): Update call
461 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
462 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
463 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
464 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
465 * expmed.c (extract_bit_field_1): Likewise.
466 * reload.c (get_secondary_mem): Likewise.
467 * varasm.c (assemble_integer): Likewise.
468 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
471 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
473 * machmode.h (decimal_float_mode_for_size): New function.
474 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
475 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
476 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
477 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
478 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
479 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
481 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
483 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
484 (get_builtin_sync_mode): Likewise.
485 (expand_ifn_atomic_compare_exchange): Likewise.
486 (expand_builtin_atomic_clear): Likewise.
487 (expand_builtin_atomic_test_and_set): Likewise.
488 (fold_builtin_atomic_always_lock_free): Likewise.
489 * calls.c (compute_argument_addresses): Likewise.
490 (emit_library_call_value_1): Likewise.
491 (store_one_arg): Likewise.
492 * combine.c (combine_instructions): Likewise.
493 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
494 * config/arm/arm.c (arm_function_value): Likewise.
495 (aapcs_allocate_return_reg): Likewise.
496 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
497 * config/i386/i386.c (construct_container): Likewise.
498 (ix86_gimplify_va_arg): Likewise.
499 (ix86_expand_sse_cmp): Likewise.
500 (emit_memmov): Likewise.
501 (emit_memset): Likewise.
502 (expand_small_movmem_or_setmem): Likewise.
503 (ix86_expand_pextr): Likewise.
504 (ix86_expand_pinsr): Likewise.
505 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
506 * config/microblaze/microblaze.c (microblaze_block_move_straight):
508 * config/mips/mips.c (mips_function_value_1) Likewise.
509 (mips_block_move_straight): Likewise.
510 (mips_expand_ins_as_unaligned_store): Likewise.
511 * config/powerpcspe/powerpcspe.c
512 (rs6000_darwin64_record_arg_advance_flush): Likewise.
513 (rs6000_darwin64_record_arg_flush): Likewise.
514 * config/rs6000/rs6000.c
515 (rs6000_darwin64_record_arg_advance_flush): Likewise.
516 (rs6000_darwin64_record_arg_flush): Likewise.
517 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
518 (sparc_function_value_1): Likewise.
519 * config/spu/spu.c (adjust_operand): Likewise.
520 (spu_emit_branch_or_set): Likewise.
521 (arith_immediate_p): Likewise.
522 * emit-rtl.c (gen_lowpart_common): Likewise.
523 * expr.c (expand_expr_real_1): Likewise.
524 * function.c (assign_parm_setup_block): Likewise.
525 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
526 * reload1.c (alter_reg): Likewise.
527 * stor-layout.c (mode_for_vector): Likewise.
528 (layout_type): Likewise.
530 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
532 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
533 (spu_convert_move): Likewise.
534 * lower-subreg.c (resolve_simple_move): Likewise.
536 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
539 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
540 define_insn to a define_expand.
541 (altivec_vsum2sws_direct): New define_insn.
542 (altivec_vsumsws): Convert from a define_insn to a define_expand.
544 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
546 * config/arm/arm.c (arm_option_params_internal): Improve setting of
549 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
554 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
555 realign stack if stack alignment needed is less than incoming
558 2017-09-05 Marek Polacek <polacek@redhat.com>
561 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
564 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
566 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
568 2017-09-05 Richard Biener <rguenther@suse.de>
570 PR tree-optimization/82084
571 * fold-const.c (can_native_encode_string_p): Handle wide characters.
573 2017-09-05 Richard Biener <rguenther@suse.de>
575 PR tree-optimization/82102
576 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
578 2017-09-05 Martin Liska <mliska@suse.cz>
580 PR tree-optimization/82032
581 * tree-cfg.c (generate_range_test): New function.
582 * tree-cfg.h (generate_range_test): Declared here.
583 * tree-cfgcleanup.c (convert_single_case_switch): New function.
584 (cleanup_control_expr_graph): Use it.
585 * tree-switch-conversion.c (try_switch_expansion): Remove
587 (emit_case_nodes): Use generate_range_test.
589 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
592 * config/i386/i386.md (*<btsc><mode>_mask): Add
593 TARGET_USE_BT to insn constraint.
594 (*btr<mode>_mask): Ditto.
596 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
598 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
599 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
601 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
604 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
605 TARGET_NEON and TARGET_IWMMXT.
606 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
607 TARGET_NEON and TARGET_IWMMXT.
608 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
610 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
612 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
613 (ix86_rewrite_tls_address): Ditto.
614 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
615 (ix86_rewrite_tls_address_1): Ditto.
616 (ix86_rewrite_tls_address): Ditto.
617 * config/i386/predicates.md (tls_address_pattern): New predicate.
618 * config/i386/i386.md (TLS address splitter): New splitter.
620 2017-09-04 Richard Biener <rguenther@suse.de>
622 PR tree-optimization/82084
623 * fold-const.h (can_native_encode_string_p): Declare.
624 * fold-const.c (can_native_encode_string_p): Factor out from ...
625 (native_encode_string): ... here.
626 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
627 vectorizing stores from constants we later cannot handle.
629 2017-09-04 Marek Polacek <polacek@redhat.com>
632 * doc/invoke.texi: Update -Wtautological-compare documentation.
634 2017-09-04 Jeff Law <law@redhat.com>
636 PR tree-optimization/64910
637 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
638 swap the first and last operand if the last is a constant.
640 2017-09-04 Marek Polacek <polacek@redhat.com>
643 * convert.c (do_narrow): When sanitizing signed integer overflows,
644 bail out for signed types.
645 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
647 2017-09-04 Richard Biener <rguenther@suse.de>
649 PR tree-optimization/82060
650 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
651 Move devirtualization after stmt folding and before EH/AB/noreturn
652 cleanup to get the stmt refs canonicalized. Use a bool instead
653 of gimple_modified_p since that doesn't work for NOPs. Schedule
654 NOPs generated by folding for removal.
656 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
657 Alan Hayward <alan.hayward@arm.com>
658 David Sherwood <david.sherwood@arm.com>
660 * coretypes.h (pad_direction): New enum.
661 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
662 (FUNCTION_ARG_PADDING): Likewise.
663 * target.def (function_arg_padding): New hook.
664 * targhooks.h (default_function_arg_padding): Declare.
665 * targhooks.c (default_function_arg_padding): New function.
666 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
667 (TARGET_FUNCTION_ARG_PADDING): ...this.
668 * doc/tm.texi: Regenerate.
669 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
670 instead of direction.
671 (compute_argument_addresses): Likewise.
672 (load_register_parameters): Likewise.
673 (emit_library_call_value_1): Likewise.
674 (store_one_arg): Use targetm.calls.function_arg_padding instead
675 of FUNCTION_ARG_PADDING.
676 (must_pass_in_stack_var_size_or_pad): Likewise.
677 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
678 (emit_group_store): Likewise.
679 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
680 instead of FUNCTION_ARG_PADDING.
681 (emit_push_insn): Likewise, and propagate enum change throughout
683 * function.h (direction): Delete.
684 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
686 * function.c (assign_parm_find_stack_rtl): Likewise.
687 (assign_parm_setup_block_p): Likewise.
688 (assign_parm_setup_block): Likewise.
689 (gimplify_parameters): Likewise.
690 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
691 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
693 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
694 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
695 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
696 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
697 (aarch64_function_arg_padding): ...this new function.
698 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
699 (TARGET_FUNCTION_ARG_PADDING): Redefine.
700 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
701 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
702 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
703 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
704 (arm_pad_arg_upward): Replace with...
705 (arm_function_arg_padding): ...this new function.
706 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
708 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
709 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
710 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
711 (ia64_hpux_function_arg_padding): Replace with...
712 (ia64_function_arg_padding): ...this new function. Use pad_direction
713 instead of direction. Check for TARGET_HPUX.
714 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
715 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
716 (iq2000_function_arg_padding): New function.
717 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
718 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
719 (mips_function_arg_padding): ...this new function.
720 (mips_pad_reg_upward): Update accordingly.
721 (TARGET_FUNCTION_ARG_PADDING): Redefine.
722 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
723 targetm.calls.function_arg_padding.
724 (FUNCTION_ARG_PADDING): Delete.
725 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
726 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
727 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
728 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
729 (nios2_block_reg_padding): Return pad_direction instead of direction.
730 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
731 instead of direction.
732 (nios2_function_arg_padding): Likewise. Make static.
733 (TARGET_FUNCTION_ARG_PADDING): Redefine.
734 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
735 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
736 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
737 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
738 (pa_function_arg_padding): Make static. Return pad_direction instead
740 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
741 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
742 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
743 instead of direction. Use targetm.calls.function_arg_padding.
744 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
745 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
746 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
747 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
748 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
750 (function_arg_padding): Rename to...
751 (rs6000_function_arg_padding): ...this. Make static. Return
752 pad_direction instead of direction.
753 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
754 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
755 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
756 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
757 instead of direction. Use targetm.calls.function_arg_padding.
758 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
759 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
760 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
761 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
762 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
763 (function_arg_padding): Rename to...
764 (rs6000_function_arg_padding): ...this. Make static. Return
765 pad_direction instead of direction.
766 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
767 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
768 * config/s390/s390.c (s390_function_arg_padding): New function.
769 (TARGET_FUNCTION_ARG_PADDING): Redefine.
770 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
771 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
772 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
773 (function_arg_padding): Rename to...
774 (sparc_function_arg_padding): ...this. Make static. Return
775 pad_direction instead of direction.
776 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
777 * config/spu/spu.c (spu_function_arg_padding): New function.
778 (TARGET_FUNCTION_ARG_PADDING): Redefine.
779 * system.h (FUNCTION_ARG_PADDING): Poison.
781 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
782 Alan Hayward <alan.hayward@arm.com>
783 David Sherwood <david.sherwood@arm.com>
785 * target.def (modes_tieable_p): New hook.
786 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
787 (TARGET_MODES_TIEABLE_P): ...this.
788 * doc/tm.texi.in: Regenerate.
789 * hooks.h (hook_bool_mode_mode_true): Declare.
790 * hooks.c (hook_bool_mode_mode_true): New function.
791 * combine.c (subst): Use targetm.modes_tieable_p instead of
793 * dse.c (find_shift_sequence): Likewise.
794 * expmed.c (extract_low_bits): Likewise.
795 * lower-subreg.c: Include target.h.
796 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
798 * rtlanal.c (rtx_cost): Likewise.
799 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
800 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
801 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
802 (TARGET_MODES_TIEABLE_P): Redefine.
803 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
804 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
805 (TARGET_MODES_TIEABLE_P): Redefine.
806 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
807 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
808 (arc_modes_tieable_p): New function.
809 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
810 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
811 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
812 (arm_modes_tieable_p): Make static.
813 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
814 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
815 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
816 (TARGET_MODES_TIEABLE_P): Redefine.
817 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
818 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
819 (TARGET_MODES_TIEABLE_P): Redefine.
820 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
821 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
822 (cr16_modes_tieable_p): New function.
823 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
824 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
825 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
826 (TRULY_NOOP_TRUNCATION): Update comment.
827 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
828 (TRULY_NOOP_TRUNCATION): Update comment.
829 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
830 (frv_modes_tieable_p): New function.
831 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
832 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
833 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
834 (TARGET_MODES_TIEABLE_P): Redefine.
835 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
836 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
837 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
838 (TARGET_MODES_TIEABLE_P): Redefine.
839 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
840 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
841 (ia64_modes_tieable_p): New function.
842 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
843 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
844 (iq2000_modes_tieable_p): New function.
845 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
846 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
847 (lm32_modes_tieable_p): New function.
848 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
849 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
850 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
851 (TARGET_MODES_TIEABLE_P): Redefine.
852 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
853 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
854 (m32r_modes_tieable_p): New function.
855 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
856 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
857 (m68k_modes_tieable_p): New function.
858 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
859 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
860 (mcore_modes_tieable_p): New function.
861 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
862 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
864 (TARGET_MODES_TIEABLE_P): Redefine.
865 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
866 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
867 * config/mips/mips.c (mips_modes_tieable_p): Make static.
868 (TARGET_MODES_TIEABLE_P): Redefine.
869 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
870 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
871 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
872 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
873 (mn10300_modes_tieable_p): ...this and make static.
874 (TARGET_MODES_TIEABLE_P): Redefine.
875 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
876 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
877 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
878 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
879 (msp430_modes_tieable_p): Make static.
880 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
881 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
882 (TARGET_MODES_TIEABLE_P): Redefine.
883 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
884 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
885 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
886 (TARGET_MODES_TIEABLE_P): Redefine.
887 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
888 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
889 * config/pa/pa.c (pa_modes_tieable_p): Make static.
890 (TARGET_MODES_TIEABLE_P): Redefine.
891 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
892 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
893 (pdp11_modes_tieable_p): New function.
894 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
895 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
896 (rs6000_modes_tieable_p): New function.
897 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
898 * config/powerpcspe/powerpcspe.md: Update comment.
899 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
900 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
901 (TARGET_MODES_TIEABLE_P): Redefine.
902 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
903 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
904 (rl78_modes_tieable_p): New function.
905 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
906 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
907 (rs6000_modes_tieable_p): New function.
908 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
909 * config/rs6000/rs6000.md: Update comment.
910 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
911 * config/rx/rx.c (rx_modes_tieable_p): New function.
912 (TARGET_MODES_TIEABLE_P): Redefine.
913 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
914 * config/s390/s390.c (s390_modes_tieable_p): New function.
915 (TARGET_MODES_TIEABLE_P): Redefine.
916 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
917 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
918 (sh_modes_tieable_p): New function.
919 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
920 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
921 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
922 (sparc_modes_tieable_p): Make static.
923 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
924 * config/spu/spu.c (spu_modes_tieable_p): New function.
925 (TARGET_MODES_TIEABLE_P): Redefine.
926 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
927 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
928 (TARGET_MODES_TIEABLE_P): Redefine.
929 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
930 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
931 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
932 * config/v850/v850.c (v850_modes_tieable_p): New function.
933 (TARGET_MODES_TIEABLE_P): Redefine.
934 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
935 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
936 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
937 (visium_modes_tieable_p): New function.
938 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
939 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
940 (xtensa_modes_tieable_p): New function.
941 * system.h (MODES_TIEABLE_P): Poison.
943 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
944 Alan Hayward <alan.hayward@arm.com>
945 David Sherwood <david.sherwood@arm.com>
947 * target.def (hard_regno_mode_ok): New hook.
948 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
949 (TARGET_HARD_REGNO_MODE_OK): ...this.
950 * doc/tm.texi.in: Regenerate.
951 * hooks.h (hook_bool_uint_mode_true): Declare.
952 * hooks.c (hook_bool_uint_mode_true): New function.
953 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
955 * genpreds.c (write_insn_preds_c): Add an include of target.h.
956 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
957 instead of HARD_REGNO_MODE_OK.
958 * caller-save.c: Include target.h.
959 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
961 * combine.c (can_combine_p): Likewise.
962 (combinable_i3pat): Likewise.
963 (can_change_dest_mode): Likewise.
964 * expr.c (init_expr_target): Likewise.
965 (convert_move): Likewise.
966 (convert_modes): Likewise.
967 * ira.c (setup_prohibited_class_mode_regs): Likewise.
968 (setup_prohibited_mode_move_regs): Likewise.
969 * ira.h (target_ira): Likewise.
970 * lra-assigns.c (find_hard_regno_for_1): Likewise.
971 * lra-constraints.c (process_alt_operands): Likewise.
972 (split_reg): Likewise.
973 * recog.c (peep2_find_free_register): Likewise.
974 * ree.c (combine_reaching_defs): Likewise.
975 * regcprop.c (maybe_mode_change): Likewise.
976 * reginfo.c (init_reg_sets_1): Likewise.
977 (choose_hard_reg_mode): Likewise.
978 (simplifiable_subregs): Likewise.
979 * regrename.c (check_new_reg_p): Likewise.
980 * reload.c (find_valid_class): Likewise.
981 (find_valid_class_1): Likewise.
982 (reload_inner_reg_of_subreg): Likewise.
983 (push_reload): Likewise.
984 (combine_reloads): Likewise.
985 (find_dummy_reload): Likewise.
986 (find_reloads): Likewise.
987 * reload1.c (find_reg): Likewise.
988 (set_reload_reg): Likewise.
989 (allocate_reload_reg): Likewise.
990 (choose_reload_regs): Likewise.
991 (reload_adjust_reg_for_temp): Likewise.
992 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
993 (simplify_subreg_regno): Likewise.
994 * sel-sched.c (init_regs_for_mode): Likewise.
995 * varasm.c (make_decl_rtl): Likewise.
996 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
997 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
999 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1000 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1002 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1003 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1004 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1005 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1006 (arc_mode_class): Delete.
1007 (HARD_REGNO_MODE_OK): Delete.
1008 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1009 (arc_hard_regno_mode_ok): Rename old array to...
1010 (arc_hard_regno_mode_ok_modes): ...this.
1011 (arc_conditional_register_usage): Update accordingly.
1012 (arc_mode_class): Make static.
1013 (arc_hard_regno_mode_ok): New function.
1014 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1015 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1016 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1017 (arm_hard_regno_mode_ok): Make static.
1018 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1020 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1021 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1022 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1024 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1025 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1026 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1027 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1028 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1029 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1030 * config/bfin/predicates.md (valid_reg_operand): Use
1031 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1032 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1033 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1034 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1035 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1036 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1037 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1038 (cr16_hard_regno_mode_ok): Make static and return a bool.
1039 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1040 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1041 (cris_hard_regno_mode_ok): New function.
1042 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1043 (epiphany_mode_class): Delete.
1044 (HARD_REGNO_MODE_OK): Delete.
1045 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1046 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1047 (hard_regno_mode_ok): Rename to...
1048 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1049 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1050 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1052 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1053 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1054 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1055 (frv_hard_regno_mode_ok): Make static and return a bool.
1056 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1058 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1059 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1060 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1061 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1063 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1064 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1065 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1066 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1068 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1069 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1070 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1071 (ia64_hard_regno_mode_ok): New function.
1072 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1073 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1074 (iq2000_hard_regno_mode_ok): New function.
1075 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1076 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1077 (lm32_hard_regno_mode_ok): New function.
1078 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1079 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1080 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1081 instead of HARD_REGNO_MODE_OK.
1082 (m32c_hard_regno_ok): Rename to...
1083 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1084 (m32c_cannot_change_mode_class): Update accordingly.
1085 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1086 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1087 (m32r_mode_class): Delete.
1088 (HARD_REGNO_MODE_OK): Delete.
1089 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1090 (m32r_hard_regno_mode_ok): Rename to...
1091 (m32r_hard_regno_modes): ...this.
1092 (m32r_mode_class): Make static.
1093 (m32r_hard_regno_mode_ok): New function.
1094 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1095 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1096 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1097 (m68k_hard_regno_mode_ok): Make static.
1098 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1099 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1100 (mcore_hard_regno_mode_ok): New function.
1101 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1102 (HARD_REGNO_MODE_OK): Delete.
1103 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1105 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1106 (microblaze_hard_regno_mode_ok): New function.
1107 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1108 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1109 (mips_hard_regno_mode_ok): Delete.
1110 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1111 (mips_hard_regno_mode_ok_p): ...this and make static.
1112 (mips_hard_regno_mode_ok_p): Rename to...
1113 (mips_hard_regno_mode_ok_uncached): ...this.
1114 (mips_hard_regno_mode_ok): New function.
1115 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1116 of HARD_REGNO_MODE_OK.
1117 (mips_option_override): Update after above name changes.
1118 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1119 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1120 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1121 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1122 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1123 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1124 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1125 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1126 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1127 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1128 (msp430_hard_regno_mode_ok): Make static and return a bool.
1129 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1130 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1131 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1133 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1134 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1135 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1136 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1137 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1138 (PA_HARD_REGNO_MODE_OK): ...this
1139 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1140 (PA_HARD_REGNO_MODE_OK): ...this.
1141 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1142 (pa_hard_regno_mode_ok): New function.
1143 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1144 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1145 (pdp11_hard_regno_mode_ok): New function.
1146 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1147 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1149 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1151 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1152 (rs6000_hard_regno_mode_ok): Rename to...
1153 (rs6000_hard_regno_mode_ok_uncached): ...this.
1154 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1155 (rs6000_hard_regno_mode_ok): New function.
1156 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1157 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1158 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1159 (riscv_hard_regno_mode_ok): ...this and make static.
1160 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1161 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1162 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1163 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1164 (rl78_hard_regno_mode_ok): Make static and return bool.
1165 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1166 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1168 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1169 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1170 (rs6000_hard_regno_mode_ok): Rename to...
1171 (rs6000_hard_regno_mode_ok_uncached): ...this.
1172 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1173 (rs6000_hard_regno_mode_ok): New function.
1174 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1175 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1176 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1177 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1178 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1179 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1180 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1181 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1182 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1183 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1184 (sh_hard_regno_mode_ok): Make static.
1185 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1186 instead of HARD_REGNO_MODE_OK.
1187 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1188 (sparc_mode_class): Delete.
1189 (HARD_REGNO_MODE_OK): Delete.
1190 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1191 (hard_regno_mode_classes): Make static.
1192 (sparc_mode_class): Likewise.
1193 (sparc_hard_regno_mode_ok): New function.
1194 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1195 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1196 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1198 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1199 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1200 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1201 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1202 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1203 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1204 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1205 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1206 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1207 (visium_hard_regno_mode_ok): New function.
1208 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1209 instead of HARD_REGNO_MODE_OK.
1210 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1211 (HARD_REGNO_MODE_OK): Delete.
1212 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1213 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1214 (xtensa_option_override): Update accordingly.
1215 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1216 (xtensa_hard_regno_mode_ok): New function.
1217 * system.h (HARD_REGNO_MODE_OK): Poison.
1219 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1220 Alan Hayward <alan.hayward@arm.com>
1221 David Sherwood <david.sherwood@arm.com>
1223 * target.def (hard_regno_call_part_clobbered): New hook.
1224 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1225 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1226 * doc/tm.texi: Regenerate.
1227 * hooks.h (hook_bool_uint_mode_false): Declare.
1228 * hooks.c (hook_bool_uint_mode_false): New function.
1229 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1230 * cselib.c (cselib_process_insn): Use
1231 targetm.hard_regno_call_part_clobbered instead of
1232 HARD_REGNO_CALL_PART_CLOBBERED.
1233 * ira-conflicts.c (ira_build_conflicts): Likewise.
1234 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1235 * lra-constraints.c (need_for_call_save_p): Likewise.
1236 * lra-lives.c: Include target.h.
1237 (check_pseudos_live_through_calls): Use
1238 targetm.hard_regno_call_part_clobbered instead of
1239 HARD_REGNO_CALL_PART_CLOBBERED.
1240 * regcprop.c: Include target.h.
1241 (copyprop_hardreg_forward_1): Use
1242 targetm.hard_regno_call_part_clobbered instead of
1243 HARD_REGNO_CALL_PART_CLOBBERED.
1244 * reginfo.c (choose_hard_reg_mode): Likewise.
1245 * regrename.c (check_new_reg_p): Likewise.
1246 * reload.c (find_equiv_reg): Likewise.
1247 * reload1.c (emit_reload_insns): Likewise.
1248 * sched-deps.c (deps_analyze_insn): Likewise.
1249 * sel-sched.c (init_regs_for_mode): Likewise.
1250 (mark_unavailable_hard_regs): Likewise.
1251 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1252 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1253 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1255 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1256 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1257 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1259 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1261 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1262 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1263 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1265 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1266 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1267 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1269 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1270 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1272 * config/powerpcspe/powerpcspe.c
1273 (rs6000_hard_regno_call_part_clobbered): New function.
1274 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1275 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1276 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1278 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1279 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1280 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1282 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1283 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1284 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1286 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1287 Alan Hayward <alan.hayward@arm.com>
1288 David Sherwood <david.sherwood@arm.com>
1290 * rtl.h (subreg_memory_offset): Declare.
1291 * emit-rtl.c (subreg_memory_offset): New function.
1292 * expmed.c (store_bit_field_1): Use it.
1293 * expr.c (undefined_operand_subword_p): Likewise.
1294 * simplify-rtx.c (simplify_subreg): Likewise.
1296 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1298 PR rtl-optimization/57448
1301 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1302 using atomic_load pattern.
1303 (expand_atomic_store): Likewise.
1305 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1308 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1309 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1310 handling. Use replace_call_with_value with NULL instead of
1311 gsi_replace, unlink_stmt_vdef and release_defs.
1313 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1316 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1318 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1321 * rtl.h (emit_library_call_value_1): Declare.
1322 (emit_library_call): Replace declaration with a series of overloads.
1323 Remove the parameter count argument.
1324 (emit_library_call_value): Likewise.
1325 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1326 with an "rtx_mode_t *".
1327 (emit_library_call_value): Delete.
1328 (emit_library_call): Likewise.
1329 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1330 (asan_emit_allocas_unpoison): Likewise.
1331 * builtins.c (expand_builtin_powi): Likewise.
1332 (expand_asan_emit_allocas_unpoison): Likewise.
1333 * cfgexpand.c (expand_main_function): Likewise.
1334 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1335 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1336 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1337 * config/arm/arm.c (arm_trampoline_init): Likewise.
1338 (arm_call_tls_get_addr): Likewise.
1339 (arm_expand_divmod_libfunc): Likewise.
1340 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1341 (smulsi3_highpart): Likewise.
1342 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1343 (c6x_expand_compare): Likewise.
1344 (c6x_expand_movmem): Likewise.
1345 * config/frv/frv.c (frv_trampoline_init): Likewise.
1346 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1347 (ix86_expand_divmod_libfunc): Likewise.
1348 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1349 (ia64_expand_compare): Likewise.
1350 (ia64_profile_hook): Likewise.
1351 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1352 (nonlocal_goto): Likewise.
1353 (restore_stack_nonlocal): Likewise.
1354 * config/m32r/m32r.c (block_move_call): Likewise.
1355 (m32r_trampoline_init): Likewise.
1356 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1357 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1358 (m68k_call_m68k_read_tp): Likewise.
1359 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1360 (microblaze_expand_divide): Likewise.
1361 * config/mips/mips.h (mips_args): Likewise.
1362 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1363 (MIPS_ICACHE_SYNC): Likewise.
1364 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1365 (nios2_trampoline_init): Likewise.
1366 * config/pa/pa.c (hppa_tls_call): Likewise.
1367 (pa_trampoline_init): Likewise.
1368 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1369 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1370 (expand_strn_compare): Likewise.
1371 (rs6000_generate_compare): Likewise.
1372 (rs6000_expand_float128_convert): Likewise.
1373 (output_profile_hook): Likewise.
1374 (rs6000_trampoline_init): Likewise.
1375 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1376 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1377 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1378 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1379 (rs6000_generate_compare): Likewise.
1380 (rs6000_expand_float128_convert): Likewise.
1381 (output_profile_hook): Likewise.
1382 (rs6000_trampoline_init): Likewise.
1383 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1384 * config/sh/sh.c (sh_trampoline_init): Likewise.
1385 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1386 (sparc_emit_float_lib_cmp): Likewise.
1387 (sparc32_initialize_trampoline): Likewise.
1388 (sparc64_initialize_trampoline): Likewise.
1389 (sparc_profile_hook): Likewise.
1390 * config/spu/spu.c (ea_load_store): Likewise.
1391 * config/spu/spu.md (floatunssidf2): Likewise.
1392 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1393 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1394 * config/visium/visium.c (expand_block_move_4): Likewise.
1395 (expand_block_move_2): Likewise.
1396 (expand_block_move_1): Likewise.
1397 (expand_block_set_4): Likewise.
1398 (expand_block_set_2): Likewise.
1399 (expand_block_set_1): Likewise.
1400 (visium_trampoline_init): Likewise.
1401 (visium_profile_hook): Likewise.
1402 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1403 (xtensa_setup_frame_addresses): Likewise.
1404 (xtensa_trampoline_init): Likewise.
1405 * except.c (sjlj_emit_function_enter): Likewise.
1406 (sjlj_emit_function_exit): Likewise.
1407 * explow.c (allocate_dynamic_stack_space): Likewise.
1408 (probe_stack_range): Likewise.
1409 * expr.c (convert_mode_scalar): Likewise.
1410 * optabs.c (expand_binop): Likewise.
1411 (expand_twoval_binop_libfunc): Likewise.
1412 (expand_unop): Likewise.
1413 (prepare_cmp_insn): Likewise.
1414 (prepare_float_lib_cmp): Likewise.
1415 (expand_float): Likewise.
1416 (expand_fix): Likewise.
1417 (expand_fixed_convert): Likewise.
1418 (maybe_emit_sync_lock_test_and_set): Likewise.
1419 (expand_atomic_compare_and_swap): Likewise.
1420 (expand_mem_thread_fence): Likewise.
1421 (expand_atomic_fetch_op): Likewise.
1423 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1425 * doc/generic.texi (OpenACC): Adjust URL.
1426 * doc/invoke.texi (C Dialect Options): Ditto.
1428 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1430 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1431 predicate for operand 1. Add (m,<S>) constraint.
1432 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1433 Prevent memory operand 1 with register operand 2.
1435 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1437 PR rtl-optimization/82024
1438 * combine.c (try_combine): If the combination result is a PARALLEL,
1439 and we only need to retain the SET in there that would be placed
1440 at I2, check that we can place that at I3 instead, before doing so.
1442 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1445 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1447 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1448 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1451 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1452 Jeff Law <law@redhat.com>
1454 * varasm.c (bss_initializer_p): Do not put constants into .bss
1455 (categorize_decl_for_section): Handle bss_initializer_p returning
1456 false when DECL_INITIAL is NULL.
1458 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1461 * config/s390/s390.c (s390_can_inline_p): New function.
1463 2017-09-01 Jeff Law <law@redhat.com>
1465 PR tree-optimization/82052
1466 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1467 Always initialize the returned slot after a hash table miss
1468 when INSERT is true.
1470 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1472 * config/s390/s390.md (mem_signal_fence): Remove.
1473 * doc/md.texi (mem_signal_fence): Remove.
1474 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1476 * target-insns.def (mem_signal_fence): Remove.
1478 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1481 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1484 * asan.c (create_odr_indicator): Strip name encoding from assembler
1485 name before appending it after __odr_asan_.
1487 2017-09-01 Martin Liska <mliska@suse.cz>
1489 PR tree-optimization/82059
1490 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1491 frequency only when an edge is redirected.
1493 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1495 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1496 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1497 (arc_conditional_register_usage): Remove ARC600 lp_count
1499 (arc_file_start): Emit Tag_ARC_CPU_variation.
1500 (arc_can_use_doloop_p): New conditions to use ZOLs.
1501 (hwloop_fail): New function.
1502 (hwloop_optimize): Likewise.
1503 (hwloop_pattern_reg): Likewise.
1504 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1505 (arc_reorg_loops): New function, calls reorg_loops.
1506 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1507 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1509 (arc_loop_hazard): Remove function, functionality moved into
1511 (arc_hazard): Remove arc_loop_hazard call.
1512 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1513 into hwloop_optimize.
1514 (arc_label_align): Remove ZOL handling.
1515 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1516 * config/arc/arc.md (doloop_begin): Remove pattern.
1517 (doloop_begin_i): Likewise.
1518 (doloop_end_i): Likewise.
1519 (doloop_fallback): Likewise.
1520 (doloop_fallback_m): Likewise.
1521 (doloop_end): Reimplement expand.
1522 (arc_lp): New pattern for LP instruction.
1523 (loop_end): New pattern.
1524 (loop_fail): Likewise.
1525 (decrement_and_branch_until_zero): Likewise.
1526 * config/arc/arc.opt (mlpc-width): New option.
1527 * doc/invoke.texi (mlpc-width): Document option.
1529 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1531 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1532 (arc_ccfsm_advance): Fix checking for delay slots.
1533 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1535 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1537 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1539 (movhi_insn): Update store instruction constraint which are saving
1540 6-bit short immediates.
1541 (movsi_insn): Consider also short scaled load operations.
1542 (zero_extendhisi2_i): Use Usd constraint instead of T.
1543 (extendhisi2_i): Add q constraint.
1544 (arc_clzsi2): Add type and length attributes.
1545 (arc_ctzsi2): Likewise.
1546 * config/arc/constraints.md (Usc): Update constraint, the
1547 assembler can parse two relocations for a single instruction.
1549 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1551 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1552 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1554 2017-08-31 Olivier Hainque <hainque@adacore.com>
1556 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1557 match as powerpc-wrs-vxworks*.
1559 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1561 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1562 register constraint for by-element operand.
1563 (aarch64_mls_elt_merge<mode>): Likewise.
1565 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1567 * config/arc/arc.c (arc_can_follow_jump): Check for short
1570 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1572 * config.gcc: Use g.opt for arc.
1573 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1574 functionality moved to ...
1575 (legitimate_scaled_address_p): New function, ...here.
1576 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1577 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1578 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1580 (arc_override_options): Handle G option.
1581 (arc_output_pic_addr_const): Correct function definition.
1582 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1583 (arc_decl_anon_ns_mem_p): Delete.
1584 (arc_in_small_data_p): Overhaul this function to take into
1585 consideration the value given via G option.
1586 (arc_rewrite_small_data_1): Renamed and corrected old
1587 arc_rewrite_small_data function.
1588 (arc_rewrite_small_data): New function.
1589 (small_data_pattern): Don't use pic_offset_table_rtx.
1590 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1591 * config/arc/simdext.md (movmisalignv2hi): Use
1592 prepare_move_operands function.
1594 (movmisalign*): Likewise.
1595 * doc/invoke.texi (ARC options): Document -G option.
1597 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1599 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1601 * config/arc/arc.c (arc_print_operand): Output scalled address for
1602 sdata whenever is possible.
1603 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1604 load/stores are available.
1605 (compact_sda_memory_operand): Check for the alignment required by
1606 code density instructions.
1607 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1609 * config/arc/constraints.md (Usd): Update constraint.
1610 (Us0): New constraint.
1611 (Usc): Update constraint.
1613 2017-08-31 Richard Biener <rguenther@suse.de>
1616 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1619 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1621 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1622 Resize type_signature.
1624 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1625 Alan Hayward <alan.hayward@arm.com>
1626 David Sherwood <david.sherwood@arm.com>
1628 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1629 subregs whose inner modes can be stored in GPRs.
1630 (aarch64_classify_index): Likewise.
1632 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1633 Alan Hayward <alan.hayward@arm.com>
1634 David Sherwood <david.sherwood@arm.com>
1636 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1637 (V_INT_EQUIV): ...this.
1638 (v_cmp_result): Rename to...
1639 (v_int_equiv): ...this.
1640 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1641 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1642 (copysign<mode>3): Likewise.
1643 (aarch64_simd_bsl<mode>_internal): Likewise.
1644 (aarch64_simd_bsl<mode>): Likewise.
1645 (vec_cmp<mode><mode>): Likewise.
1646 (vcond<mode><mode>): Likewise.
1647 (vcond<v_cmp_mixed><mode>): Likewise.
1648 (vcondu<mode><v_cmp_mixed>): Likewise.
1649 (aarch64_cm<optab><mode>): Likewise.
1650 (aarch64_cmtst<mode>): Likewise.
1651 (aarch64_fac<optab><mode>): Likewise.
1652 (vec_perm_const<mode>): Likewise.
1653 (vcond_mask_<mode><v_cmp_result>): Rename to...
1654 (vcond_mask_<mode><v_int_equiv>): ...this.
1655 (vec_cmp<mode><v_cmp_result>): Rename to...
1656 (vec_cmp<mode><v_int_equiv>): ...this.
1658 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1659 Alan Hayward <alan.hayward@arm.com>
1660 David Sherwood <david.sherwood@arm.com>
1662 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1664 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1665 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1666 (UNSPEC_LD4_DREG): New unspecs.
1667 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1668 (aarch64_ld2<mode>_dreg_be): Replace with...
1669 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1671 (aarch64_ld3<mode>_dreg_le)
1672 (aarch64_ld3<mode>_dreg_be): Replace with...
1673 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1675 (aarch64_ld4<mode>_dreg_le)
1676 (aarch64_ld4<mode>_dreg_be): Replace with...
1677 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1680 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1682 PR tree-optimization/81987
1683 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1684 insert an initializer in a location not dominated by the stride
1687 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1689 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1690 on the entire header of the finally block in the fallthru case.
1692 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1694 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1696 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1698 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1699 rs6000_emit_move_from_cr and call renamed function.
1700 (rs6000_emit_prologue): Call renamed functions.
1701 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1702 movesi_from_cr, remove volatile CRs.
1704 2017-08-30 Jon Beniston <jon@beniston.com>
1705 Richard Biener <rguenther@suse.de>
1707 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1708 of VECTOR_MODE_P check.
1709 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1710 element vector types.
1712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1714 * df.h (df_read_modify_subreg_p): Remove in favor of...
1715 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1716 const_rtx instead of an rtx.
1717 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1718 * df-problems.c (df_word_lr_mark_ref): Likewise.
1719 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1720 (mark_pseudo_reg_dead): Likewise.
1721 (mark_ref_dead): Likewise.
1722 * reginfo.c (init_subregs_of_mode): Likewise.
1723 * sched-deps.c (sched_analyze_1): Likewise.
1724 * df-scan.c (df_def_record_1): Likewise.
1725 (df_uses_record): Likewise.
1726 (df_read_modify_subreg_p): Remove in favor of...
1727 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1728 const_rtx instead of an rtx.
1730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1731 Alan Hayward <alan.hayward@arm.com>
1732 David Sherwood <david.sherwood@arm.com>
1734 * rtl.h (partial_subreg_p): New function.
1735 * caller-save.c (save_call_clobbered_regs): Use it.
1736 * calls.c (expand_call): Likewise.
1737 * combine.c (combinable_i3pat): Likewise.
1738 (simplify_set): Likewise.
1739 (make_extraction): Likewise.
1740 (make_compound_operation_int): Likewise.
1741 (gen_lowpart_or_truncate): Likewise.
1742 (force_to_mode): Likewise.
1743 (make_field_assignment): Likewise.
1744 (reg_truncated_to_mode): Likewise.
1745 (record_truncated_value): Likewise.
1746 (move_deaths): Likewise.
1747 * cse.c (record_jump_cond): Likewise.
1748 (cse_insn): Likewise.
1749 * cselib.c (cselib_lookup_1): Likewise.
1750 * expmed.c (extract_bit_field_using_extv): Likewise.
1751 * function.c (assign_parm_setup_reg): Likewise.
1752 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1753 * ira-build.c (create_insn_allocnos): Likewise.
1754 * lra-coalesce.c (merge_pseudos): Likewise.
1755 * lra-constraints.c (match_reload): Likewise.
1756 (simplify_operand_subreg): Likewise.
1757 (curr_insn_transform): Likewise.
1758 * lra-lives.c (process_bb_lives): Likewise.
1759 * lra.c (new_insn_reg): Likewise.
1760 (lra_substitute_pseudo): Likewise.
1761 * regcprop.c (mode_change_ok): Likewise.
1762 (maybe_mode_change): Likewise.
1763 (copyprop_hardreg_forward_1): Likewise.
1764 * reload.c (push_reload): Likewise.
1765 (find_reloads): Likewise.
1766 (find_reloads_subreg_address): Likewise.
1767 * reload1.c (alter_reg): Likewise.
1768 (eliminate_regs_1): Likewise.
1769 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1771 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1773 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1774 back to if statements, including unpack.
1776 2017-08-30 Martin Liska <mliska@suse.cz>
1779 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1781 (func_checker::compare_asm_inputs_outputs): ... this function.
1782 (func_checker::compare_gimple_asm): Use the function to compare
1783 also ASM constrains.
1784 * ipa-icf-gimple.h: Rename the function.
1786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1787 Alan Hayward <alan.hayward@arm.com>
1788 David Sherwood <david.sherwood@arm.com>
1790 * coretypes.h (complex_mode): New type.
1791 * gdbhooks.py (build_pretty_printer): Handle it.
1792 * machmode.h (complex_mode): New class.
1793 (complex_mode::includes_p): New function.
1794 (is_complex_int_mode): Likewise.
1795 (is_complex_float_mode): Likewise.
1796 * genmodes.c (get_mode_class): Handle complex mode classes.
1797 * function.c (expand_function_end): Use is_complex_int_mode.
1799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1800 Alan Hayward <alan.hayward@arm.com>
1801 David Sherwood <david.sherwood@arm.com>
1803 * coretypes.h (scalar_mode_pod): New typedef.
1804 * gdbhooks.py (build_pretty_printer): Handle it.
1805 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1806 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1807 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1808 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1812 Alan Hayward <alan.hayward@arm.com>
1813 David Sherwood <david.sherwood@arm.com>
1815 * machmode.h (mode_for_vector): Take a scalar_mode instead
1817 * stor-layout.c (mode_for_vector): Likewise.
1818 * explow.c (promote_mode): Use as_a <scalar_mode>.
1819 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
1821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1822 Alan Hayward <alan.hayward@arm.com>
1823 David Sherwood <david.sherwood@arm.com>
1825 * target.def (preferred_simd_mode): Take a scalar_mode
1826 instead of a machine_mode.
1827 * targhooks.h (default_preferred_simd_mode): Likewise.
1828 * targhooks.c (default_preferred_simd_mode): Likewise.
1829 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1830 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1831 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1832 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1833 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1834 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1835 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1836 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1838 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1839 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1840 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1841 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1842 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1843 * doc/tm.texi: Regenerate.
1844 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1848 Alan Hayward <alan.hayward@arm.com>
1849 David Sherwood <david.sherwood@arm.com>
1851 * target.def (scalar_mode_supported_p): Take a scalar_mode
1852 instead of a machine_mode.
1853 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1854 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1855 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1856 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1857 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1858 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1859 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1860 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1861 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1862 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1863 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1864 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1865 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1866 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1868 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1869 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1870 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1871 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1872 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1874 * doc/tm.texi: Regenerate.
1876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1877 Alan Hayward <alan.hayward@arm.com>
1878 David Sherwood <david.sherwood@arm.com>
1880 * coretypes.h (opt_scalar_mode): New typedef.
1881 * gdbhooks.py (build_pretty_printers): Handle it.
1882 * machmode.h (mode_iterator::get_2xwider): Add overload for
1884 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1886 * expr.c (convert_mode_scalar): Likewise.
1887 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1888 * optabs.c (expand_float): Likewise.
1889 (expand_fix): Likewise.
1890 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1893 Alan Hayward <alan.hayward@arm.com>
1894 David Sherwood <david.sherwood@arm.com>
1896 * optabs.c (expand_float): Explicitly check for scalars before
1897 using a branching expansion.
1898 (expand_fix): Likewise.
1900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1901 Alan Hayward <alan.hayward@arm.com>
1902 David Sherwood <david.sherwood@arm.com>
1904 * expr.c (convert_mode): Split scalar handling out into...
1905 (convert_mode_scalar): ...this new function. Treat the modes
1908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1909 Alan Hayward <alan.hayward@arm.com>
1910 David Sherwood <david.sherwood@arm.com>
1912 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1914 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1917 Alan Hayward <alan.hayward@arm.com>
1918 David Sherwood <david.sherwood@arm.com>
1920 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1921 rather than a machine_mode.
1922 (fixed_from_string): Likewise.
1923 (fixed_convert): Likewise.
1924 (fixed_convert_from_int): Likewise.
1925 (fixed_convert_from_real): Likewise.
1926 (real_convert_from_fixed): Likewise.
1927 * fixed-value.c (fixed_from_double_int): Likewise.
1928 (fixed_from_string): Likewise.
1929 (fixed_convert): Likewise.
1930 (fixed_convert_from_int): Likewise.
1931 (fixed_convert_from_real): Likewise.
1932 (real_convert_from_fixed): Likewise.
1933 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1936 Alan Hayward <alan.hayward@arm.com>
1937 David Sherwood <david.sherwood@arm.com>
1939 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1940 of separate mode class checks. Do not allow vector modes here.
1941 (immed_wide_int_const): Use as_a <scalar_mode>.
1942 * explow.c (trunc_int_for_mode): Likewise.
1943 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1944 (wi::shwi): Likewise.
1945 (wi::min_value): Likewise.
1946 (wi::max_value): Likewise.
1947 * dwarf2out.c (loc_descriptor): Likewise.
1948 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1952 Alan Hayward <alan.hayward@arm.com>
1953 David Sherwood <david.sherwood@arm.com>
1955 * tree.h (SCALAR_TYPE_MODE): New macro.
1956 * expr.c (expand_expr_addr_expr_1): Use it.
1957 (expand_expr_real_2): Likewise.
1958 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1959 (fold_convert_const_fixed_from_int): Likewise.
1960 (fold_convert_const_fixed_from_real): Likewise.
1961 (native_encode_fixed): Likewise
1962 (native_encode_complex): Likewise
1963 (native_encode_vector): Likewise.
1964 (native_interpret_fixed): Likewise.
1965 (native_interpret_real): Likewise.
1966 (native_interpret_complex): Likewise.
1967 (native_interpret_vector): Likewise.
1968 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1969 (simd_clone_adjust_argument_types): Likewise.
1970 (simd_clone_init_simd_arrays): Likewise.
1971 (simd_clone_adjust): Likewise.
1972 * stor-layout.c (layout_type): Likewise.
1973 * tree.c (build_minus_one_cst): Likewise.
1974 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1975 * tree-inline.c (estimate_move_cost): Likewise.
1976 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1977 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1978 (vectorizable_reduction): Likewise.
1979 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1980 (vect_recog_mixed_size_cond_pattern): Likewise.
1981 (check_bool_pattern): Likewise.
1982 (adjust_bool_pattern): Likewise.
1983 (search_type_for_mask_1): Likewise.
1984 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1985 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1986 (vectorizable_load): Likewise.
1987 (vectorizable_store): Likewise.
1988 * ubsan.c (ubsan_encode_value): Likewise.
1989 * varasm.c (output_constant): Likewise.
1991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1992 Alan Hayward <alan.hayward@arm.com>
1993 David Sherwood <david.sherwood@arm.com>
1995 * coretypes.h (scalar_mode): New class.
1996 * machmode.h (scalar_mode): Likewise.
1997 (scalar_mode::includes_p): New function.
1998 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1999 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2000 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2001 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2002 * expmed.c (store_bit_field_1): Likewise.
2003 (extract_bit_field_1): Likewise.
2004 * expr.c (write_complex_part): Likewise.
2005 (read_complex_part): Likewise.
2006 (emit_move_complex_push): Likewise.
2007 (expand_expr_real_2): Likewise.
2008 * function.c (assign_parm_setup_reg): Likewise.
2009 (assign_parms_unsplit_complex): Likewise.
2010 * optabs.c (expand_binop): Likewise.
2011 * rtlanal.c (subreg_get_info): Likewise.
2012 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2013 * varasm.c (output_constant_pool_2): Likewise.
2015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2016 Alan Hayward <alan.hayward@arm.com>
2017 David Sherwood <david.sherwood@arm.com>
2019 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2021 (expmed_mult_highpart_optab): Likewise.
2022 (expmed_mult_highpart): Likewise.
2024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2025 Alan Hayward <alan.hayward@arm.com>
2026 David Sherwood <david.sherwood@arm.com>
2028 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2029 instead of a machine_mode.
2030 (builtin_memset_read_str): Likewise.
2031 * builtins.c (c_readstr): Likewise.
2032 (builtin_memcpy_read_str): Likewise.
2033 (builtin_strncpy_read_str): Likewise.
2034 (builtin_memset_read_str): Likewise.
2035 (builtin_memset_gen_str): Likewise.
2036 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2037 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2038 instead of a machine_mode.
2039 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2041 (make_extraction): Likewise.
2042 (try_widen_shift_mode): Take and return scalar_int_modes instead
2044 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2045 a scalar_int_mode instead of a machine_mode.
2046 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2047 (avr_addr_space_pointer_mode): Likewise.
2048 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2049 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2050 (msp430_unwind_word_mode): Likewise.
2051 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2052 (spu_addr_space_pointer_mode): Likewise.
2053 (spu_addr_space_address_mode): Likewise.
2054 (spu_libgcc_cmp_return_mode): Likewise.
2055 (spu_libgcc_shift_count_mode): Likewise.
2056 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2057 (rl78_addr_space_pointer_mode): Likewise.
2058 (fl78_unwind_word_mode): Likewise.
2059 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2061 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2062 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2063 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2064 (mips_valid_pointer_mode): Likewise.
2065 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2066 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2067 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2069 (ft32_addr_space_address_mode): Likewise.
2070 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2071 scalar_int_mode instead of a machine_mode.
2072 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2074 (m32c_addr_space_address_mode): Likewise.
2075 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2076 (rs6000_eh_return_filter_mode): Likewise.
2077 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2078 (rs6000_eh_return_filter_mode): Likewise.
2079 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2080 (s390_libgcc_shift_count_mode): Likewise.
2081 (s390_unwind_word_mode): Likewise.
2082 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2084 * target.def (mode_rep_extended): Likewise.
2085 (valid_pointer_mode): Likewise.
2086 (addr_space.valid_pointer_mode): Likewise.
2087 (eh_return_filter_mode): Return a scalar_int_mode rather than
2089 (libgcc_cmp_return_mode): Likewise.
2090 (libgcc_shift_count_mode): Likewise.
2091 (unwind_word_mode): Likewise.
2092 (addr_space.pointer_mode): Likewise.
2093 (addr_space.address_mode): Likewise.
2094 * doc/tm.texi: Regenerate.
2095 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2097 (do_jump): Use scalar_int_mode for local variables.
2098 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2099 rather than a machine_mode.
2100 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2101 (scompare_loc_descriptor_wide): Likewise.
2102 (scompare_loc_descriptor_narrow): Likewise.
2103 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2105 * except.c (sjlj_emit_dispatch_table): Likewise.
2106 (expand_builtin_eh_copy_values): Likewise.
2107 * explow.c (convert_memory_address_addr_space_1): Likewise.
2108 Take a scalar_int_mode rather than a machine_mode.
2109 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2110 than a machine_mode.
2111 (memory_address_addr_space): Use scalar_int_mode for local variables.
2112 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2113 rather than a machine_mode.
2114 * expmed.c (mask_rtx): Likewise.
2115 (init_expmed_one_conv): Likewise.
2116 (expand_mult_highpart_adjust): Likewise.
2117 (extract_high_half): Likewise.
2118 (expmed_mult_highpart_optab): Likewise.
2119 (expmed_mult_highpart): Likewise.
2120 (expand_smod_pow2): Likewise.
2121 (expand_sdiv_pow2): Likewise.
2122 (emit_store_flag_int): Likewise.
2123 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2125 (extract_low_bits): Likewise.
2126 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2128 * expr.c (pieces_addr::adjust): Likewise.
2129 (can_store_by_pieces): Likewise.
2130 (store_by_pieces): Likewise.
2131 (clear_by_pieces_1): Likewise.
2132 (expand_expr_addr_expr_1): Likewise.
2133 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2134 (expand_expr_real_1): Likewise.
2135 (try_casesi): Likewise.
2136 * final.c (shorten_branches): Likewise.
2137 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2138 type of "mode" to machine_mode.
2139 * internal-fn.c (expand_arith_overflow_result_store): Take a
2140 scalar_int_mode rather than a machine_mode.
2141 (expand_mul_overflow): Use scalar_int_mode for local variables.
2142 * loop-doloop.c (doloop_modify): Likewise.
2143 (doloop_optimize): Likewise.
2144 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2145 than a machine_mode.
2146 (expand_doubleword_shift_condmove): Likewise.
2147 (expand_doubleword_shift): Likewise.
2148 (expand_doubleword_clz): Likewise.
2149 (expand_doubleword_popcount): Likewise.
2150 (expand_doubleword_parity): Likewise.
2151 (expand_absneg_bit): Use scalar_int_mode for local variables.
2152 (prepare_float_lib_cmp): Likewise.
2153 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2154 rather than a machine_mode.
2155 (convert_memory_address_addr_space): Likewise.
2156 (get_mode_bounds): Likewise.
2157 (get_address_mode): Return a scalar_int_mode rather than a
2159 * rtlanal.c (get_address_mode): Likewise.
2160 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2161 than a machine_mode.
2162 * targhooks.c (default_mode_rep_extended): Likewise.
2163 (default_valid_pointer_mode): Likewise.
2164 (default_addr_space_valid_pointer_mode): Likewise.
2165 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2166 than a machine_mode.
2167 (default_libgcc_cmp_return_mode): Likewise.
2168 (default_libgcc_shift_count_mode): Likewise.
2169 (default_unwind_word_mode): Likewise.
2170 (default_addr_space_pointer_mode): Likewise.
2171 (default_addr_space_address_mode): Likewise.
2172 * targhooks.h (default_eh_return_filter_mode): Likewise.
2173 (default_libgcc_cmp_return_mode): Likewise.
2174 (default_libgcc_shift_count_mode): Likewise.
2175 (default_unwind_word_mode): Likewise.
2176 (default_addr_space_pointer_mode): Likewise.
2177 (default_addr_space_address_mode): Likewise.
2178 (default_mode_rep_extended): Take a scalar_int_mode rather than
2180 (default_valid_pointer_mode): Likewise.
2181 (default_addr_space_valid_pointer_mode): Likewise.
2182 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2184 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2185 rather than a machine_mode.
2186 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2187 for local variables.
2188 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2189 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2190 than a machine_mode.
2192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2193 Alan Hayward <alan.hayward@arm.com>
2194 David Sherwood <david.sherwood@arm.com>
2196 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2197 the mode argument to scalar_int_mode.
2198 (do_jump_by_parts_zero_rtx): Likewise.
2199 (do_jump_by_parts_equality_rtx): Likewise.
2200 (do_jump_by_parts_greater): Take a mode argument.
2201 (do_jump_by_parts_equality): Likewise.
2202 (do_jump_1): Update calls accordingly.
2204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2205 Alan Hayward <alan.hayward@arm.com>
2206 David Sherwood <david.sherwood@arm.com>
2208 * is-a.h (safe_dyn_cast): New function.
2209 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2210 (jump_table_for_label): Likewise.
2211 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2212 instead of an rtx_insn *.
2213 (shorten_branches): Use dyn_cast instead of LABEL_P and
2214 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2215 rtx_jump_table_data::get_data_mode.
2216 (final_scan_insn): Likewise.
2218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2219 Alan Hayward <alan.hayward@arm.com>
2220 David Sherwood <david.sherwood@arm.com>
2222 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2223 trying to combine a full-register integer set with a subreg
2226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2227 Alan Hayward <alan.hayward@arm.com>
2228 David Sherwood <david.sherwood@arm.com>
2230 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2231 that is always either address_mode or pointer_mode.
2233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2234 Alan Hayward <alan.hayward@arm.com>
2235 David Sherwood <david.sherwood@arm.com>
2237 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2238 when the two are known to be equal.
2240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2241 Alan Hayward <alan.hayward@arm.com>
2242 David Sherwood <david.sherwood@arm.com>
2244 * simplify-rtx.c (simplify_const_unary_operation): Use
2245 is_a <scalar_int_mode> instead of checking for a nonzero
2246 precision. Forcibly convert op_mode to a scalar_int_mode
2247 in that case. More clearly differentiate the operand and
2248 result modes and use the former when deciding what the value
2249 of a count-bits operation should be. Use is_int_mode instead
2250 of checking for a MODE_INT. Remove redundant check for whether
2251 this mode has a zero precision.
2253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2254 Alan Hayward <alan.hayward@arm.com>
2255 David Sherwood <david.sherwood@arm.com>
2257 * optabs.c (widen_leading): Change the type of the mode argument
2258 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2259 (widen_bswap): Likewise.
2260 (expand_parity): Likewise.
2261 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2262 (expand_ffs): Likewise.
2263 (epand_unop): Check for scalar integer modes before calling the
2266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2267 Alan Hayward <alan.hayward@arm.com>
2268 David Sherwood <david.sherwood@arm.com>
2270 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2272 (expand_expr_real_1): Update call accordingly.
2274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2275 Alan Hayward <alan.hayward@arm.com>
2276 David Sherwood <david.sherwood@arm.com>
2278 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2279 value_mode arguments. Use scalar_int_mode internally.
2280 (store_bit_field_1): Rename the new integer mode from imode
2281 to op0_mode and use it instead of GET_MODE (op0). Update calls
2282 to store_split_bit_field, store_bit_field_using_insv and
2283 store_fixed_bit_field.
2284 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2285 Use scalar_int_mode internally. Use a bit count rather than a mode
2286 when calculating the largest bit size for get_best_mode.
2287 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2288 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2289 Remove assertion that OP0 has a scalar integer mode.
2290 (store_split_bit_field): Add op0_mode and value_mode arguments.
2291 Update calls to extract_fixed_bit_field.
2292 (extract_bit_field_using_extv): Add an op0_mode argument.
2293 Use scalar_int_mode internally.
2294 (extract_bit_field_1): Rename the new integer mode from imode to
2295 op0_mode and use it instead of GET_MODE (op0). Update calls to
2296 extract_split_bit_field, extract_bit_field_using_extv and
2297 extract_fixed_bit_field.
2298 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2299 to extract_split_bit_field and extract_fixed_bit_field_1.
2300 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2301 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2303 (extract_split_bit_field): Add an op0_mode argument. Update call
2304 to extract_fixed_bit_field.
2306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2307 Alan Hayward <alan.hayward@arm.com>
2308 David Sherwood <david.sherwood@arm.com>
2310 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2311 * explow.c (hard_function_value): Likewise.
2312 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2313 convert_to_mode call outside the loop.
2314 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2315 for the mode iterator. Require the mode specified by max_pieces
2317 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2319 (copy_blkmode_to_reg): Likewise.
2320 (set_storage_via_setmem): Likewise.
2321 * optabs.c (prepare_cmp_insn): Likewise.
2322 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2323 * stor-layout.c (finish_bitfield_representative): Likewise.
2325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2326 Alan Hayward <alan.hayward@arm.com>
2327 David Sherwood <david.sherwood@arm.com>
2329 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2330 * expr.c (convert_move): Use them.
2331 (convert_modes): Likewise.
2332 (store_expr_with_bounds): Likewise.
2334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2335 Alan Hayward <alan.hayward@arm.com>
2336 David Sherwood <david.sherwood@arm.com>
2338 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2339 parameter for the mode of "x". Remove the "known_x", "known_mode"
2340 and "known_ret" arguments. Change the type of the mode argument
2342 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2343 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2344 (reg_num_sign_bit_copies_for_combine): Likewise.
2345 * rtlanal.c (nonzero_bits1): Likewise.
2346 (num_sign_bit_copies1): Likewise.
2347 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2348 (reg_num_sign_bit_copies_general): Likewise.
2349 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2350 (reg_nonzero_bits_general): Likewise.
2352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2353 Alan Hayward <alan.hayward@arm.com>
2354 David Sherwood <david.sherwood@arm.com>
2356 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2357 than in subroutines. Return 1 for non-integer modes.
2358 (cached_num_sign_bit_copies): Change the type of the mode parameter
2360 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2361 classes. Handle CONST_INT_P first and then check whether X also
2362 has a scalar integer mode. Check the same thing for inner registers
2363 of a SUBREG and for values that are being extended or truncated.
2365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2366 Alan Hayward <alan.hayward@arm.com>
2367 David Sherwood <david.sherwood@arm.com>
2369 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2370 in subroutines. Return the mode mask for non-integer modes.
2371 (cached_nonzero_bits): Change the type of the mode parameter
2373 (nonzero_bits1): Likewise. Remove early exit for other mode
2374 classes. Handle CONST_INT_P first and then check whether X
2375 also has a scalar integer mode.
2377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2378 Alan Hayward <alan.hayward@arm.com>
2379 David Sherwood <david.sherwood@arm.com>
2381 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2382 Return a scalar_int_mode and assert that the size is greater than
2384 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2385 (op_by_pieces_d::op_by_pieces_d): Likewise.
2386 (op_by_pieces_d::run): Likewise.
2387 (can_store_by_pieces): Likewise.
2389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2390 Alan Hayward <alan.hayward@arm.com>
2391 David Sherwood <david.sherwood@arm.com>
2393 * combine.c (extract_left_shift): Add a mode argument and update
2395 (make_compound_operation_int): Change the type of the mode parameter
2396 to scalar_int_mode and update the call to extract_left_shift.
2398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2399 Alan Hayward <alan.hayward@arm.com>
2400 David Sherwood <david.sherwood@arm.com>
2402 * combine.c (simplify_and_const_int): Change the type of the mode
2403 parameter to scalar_int_mode.
2404 (simplify_and_const_int_1): Likewise. Update recursive call.
2406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2407 Alan Hayward <alan.hayward@arm.com>
2408 David Sherwood <david.sherwood@arm.com>
2410 * combine.c (simplify_compare_const): Check that the mode is a
2411 scalar_int_mode (rather than VOIDmode) before testing its
2413 (simplify_comparison): Move COMPARISON_P handling out of the
2414 loop and restrict the latter part of the loop to scalar_int_modes.
2415 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2416 and when considering SUBREG_REGs. Use is_int_mode instead of
2417 checking GET_MODE_CLASS against MODE_INT.
2419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2420 Alan Hayward <alan.hayward@arm.com>
2421 David Sherwood <david.sherwood@arm.com>
2423 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2424 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2425 shift_unit_mode and for modes involved in scalar shifts.
2427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2428 Alan Hayward <alan.hayward@arm.com>
2429 David Sherwood <david.sherwood@arm.com>
2431 * combine.c (force_int_to_mode): New function, split out from...
2432 (force_to_mode): ...here. Keep xmode up-to-date and use it
2433 instead of GET_MODE (x).
2435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2436 Alan Hayward <alan.hayward@arm.com>
2437 David Sherwood <david.sherwood@arm.com>
2439 * optabs-query.h (extraction_insn::struct_mode): Change type to
2440 opt_scalar_int_mode and update comment.
2441 (extraction_insn::field_mode): Change type to scalar_int_mode.
2442 (extraction_insn::pos_mode): Likewise.
2443 * combine.c (make_extraction): Update accordingly.
2444 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2445 (get_optab_extraction_insn): Likewise.
2446 * recog.c (simplify_while_replacing): Likewise.
2447 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2448 parameter to opt_scalar_int_mode.
2450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2451 Alan Hayward <alan.hayward@arm.com>
2452 David Sherwood <david.sherwood@arm.com>
2454 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2455 to a scalar_int_mode instead of a machine_mode.
2456 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2457 (get_best_mode): Return a boolean and use a pointer argument to store
2458 the selected mode. Replace the limit mode parameter with a bit limit.
2459 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2460 for the values returned by bit_field_mode_iterator::next_mode.
2461 (store_bit_field): Update call to get_best_mode.
2462 (store_fixed_bit_field): Likewise.
2463 (extract_fixed_bit_field): Likewise.
2464 * expr.c (optimize_bitfield_assignment_op): Likewise.
2465 * fold-const.c (optimize_bit_field_compare): Likewise.
2466 (fold_truth_andor_1): Likewise.
2467 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2468 Update for new type of m_mode.
2469 (get_best_mode): As above.
2471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2472 Alan Hayward <alan.hayward@arm.com>
2473 David Sherwood <david.sherwood@arm.com>
2475 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2476 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2477 (store_bit_field): Check is_a <scalar_int_mode> before calling
2478 strict_volatile_bitfield_p.
2479 (extract_bit_field): Likewise.
2481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2482 Alan Hayward <alan.hayward@arm.com>
2483 David Sherwood <david.sherwood@arm.com>
2485 * target.def (cstore_mode): Return a scalar_int_mode.
2486 * doc/tm.texi: Regenerate.
2487 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2488 * targhooks.h (default_cstore_mode): Likewise.
2489 * targhooks.c (default_cstore_mode): Likewise, using a forced
2491 * expmed.c (emit_cstore): Expect the target of the cstore to be
2494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2495 Alan Hayward <alan.hayward@arm.com>
2496 David Sherwood <david.sherwood@arm.com>
2498 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2500 (niter_desc): Likewise mode.
2501 (iv_analyze): Add a mode parameter.
2503 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2504 and change its type to scalar_int_mode.
2505 * loop-iv.c: Update commentary at head of file.
2506 (iv_constant): Pass the mode paraeter before the rtx it describes
2507 and change its type to scalar_int_mode. Remove VOIDmode handling.
2508 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2509 (iv_extend): Likewise.
2510 (shorten_into_mode): Likewise.
2511 (iv_add): Use scalar_int_mode.
2512 (iv_mult): Likewise.
2513 (iv_shift): Likewise.
2514 (canonicalize_iv_subregs): Likewise.
2515 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2516 it describes and change its mode to scalar_int_mode. Also change
2517 the type of the returned inner_mode to scalar_int_mode.
2518 (get_biv_step): Likewise, turning outer_mode from a pointer
2519 into a direct parameter. Update call to get_biv_step_1.
2520 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2521 iv_constant and get_biv_step.
2522 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2523 and change its type to scalar_int_mode. Don't initialise iv->mode
2524 to VOIDmode and remove later checks for its still being VOIDmode.
2525 Update calls to iv_analyze_op and iv_analyze_expr. Check
2526 is_a <scalar_int_mode> when changing the mode under consideration.
2527 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2528 Update call to iv_analyze_expr.
2529 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2530 inner register is not also a scalar_int_mode. Update call to
2532 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2533 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2534 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2535 separate mode class checks. Update calls to iv_analyze. Remove
2536 fix-up of VOIDmodes after iv_analyze_biv.
2537 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2538 don't have a scalar_int_mode. Update call to biv_p.
2540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2541 Alan Hayward <alan.hayward@arm.com>
2542 David Sherwood <david.sherwood@arm.com>
2544 * cfgexpand.c (convert_debug_memory_address): Use
2545 as_a <scalar_int_mode>.
2546 * combine.c (expand_compound_operation): Likewise.
2547 (make_extraction): Likewise.
2548 (change_zero_ext): Likewise.
2549 (simplify_comparison): Likewise.
2550 * cse.c (cse_insn): Likewise.
2551 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2552 (mem_loc_descriptor): Likewise.
2553 (loc_descriptor): Likewise.
2554 * expmed.c (init_expmed_one_mode): Likewise.
2555 (synth_mult): Likewise.
2556 (emit_store_flag_1): Likewise.
2557 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2558 of a comparison with size.
2559 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2560 (reduce_to_bit_field_precision): Likewise.
2561 * function.c (expand_function_end): Likewise.
2562 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2563 * loop-doloop.c (doloop_modify): Likewise.
2564 * optabs.c (expand_binop): Likewise.
2565 (expand_unop): Likewise.
2566 (expand_copysign_absneg): Likewise.
2567 (prepare_cmp_insn): Likewise.
2568 (maybe_legitimize_operand): Likewise.
2569 * recog.c (const_scalar_int_operand): Likewise.
2570 * rtlanal.c (get_address_mode): Likewise.
2571 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2572 (simplify_cond_clz_ctz): Likewise.
2573 * tree-nested.c (get_nl_goto_field): Likewise.
2574 * tree.c (build_vector_type_for_mode): Likewise.
2575 * var-tracking.c (use_narrower_mode): Likewise.
2577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2578 Alan Hayward <alan.hayward@arm.com>
2579 David Sherwood <david.sherwood@arm.com>
2581 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2582 * builtins.c (expand_builtin_signbit): Use it.
2583 * cfgexpand.c (expand_debug_expr): Likewise.
2584 * dojump.c (do_jump): Likewise.
2585 (do_compare_and_jump): Likewise.
2586 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2587 * expmed.c (make_tree): Likewise.
2588 * expr.c (expand_expr_real_2): Likewise.
2589 (expand_expr_real_1): Likewise.
2590 (try_casesi): Likewise.
2591 * fold-const-call.c (fold_const_call_ss): Likewise.
2592 * fold-const.c (unextend): Likewise.
2593 (extract_muldiv_1): Likewise.
2594 (fold_single_bit_test): Likewise.
2595 (native_encode_int): Likewise.
2596 (native_encode_string): Likewise.
2597 (native_interpret_int): Likewise.
2598 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2599 * internal-fn.c (expand_addsub_overflow): Likewise.
2600 (expand_neg_overflow): Likewise.
2601 (expand_mul_overflow): Likewise.
2602 (expand_arith_overflow): Likewise.
2603 * match.pd: Likewise.
2604 * stor-layout.c (layout_type): Likewise.
2605 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2606 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2607 * tree-ssanames.c (get_range_info): Likewise.
2608 * tree-switch-conversion.c (array_value_type) Likewise.
2609 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2610 (vect_recog_divmod_pattern): Likewise.
2611 (vect_recog_mixed_size_cond_pattern): Likewise.
2612 * tree-vrp.c (extract_range_basic): Likewise.
2613 (simplify_float_conversion_using_ranges): Likewise.
2614 * tree.c (int_fits_type_p): Likewise.
2615 * ubsan.c (instrument_bool_enum_load): Likewise.
2616 * varasm.c (mergeable_string_section): Likewise.
2617 (narrowing_initializer_constant_valid_p): Likewise.
2618 (output_constant): Likewise.
2620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2621 Alan Hayward <alan.hayward@arm.com>
2622 David Sherwood <david.sherwood@arm.com>
2624 * machmode.h (NARROWEST_INT_MODE): New macro.
2625 * expr.c (alignment_for_piecewise_move): Use it instead of
2626 GET_CLASS_NARROWEST_MODE (MODE_INT).
2627 (push_block): Likewise.
2628 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2630 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2633 Alan Hayward <alan.hayward@arm.com>
2634 David Sherwood <david.sherwood@arm.com>
2636 * postreload.c (move2add_valid_value_p): Change the type of the
2637 mode parameter to scalar_int_mode.
2638 (move2add_use_add2_insn): Add a mode parameter and use it instead
2640 (move2add_use_add3_insn): Likewise.
2641 (reload_cse_move2add): Update accordingly.
2643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2644 Alan Hayward <alan.hayward@arm.com>
2645 David Sherwood <david.sherwood@arm.com>
2647 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2649 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2650 * optabs.c (expand_unop): Likewise.
2652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2653 Alan Hayward <alan.hayward@arm.com>
2654 David Sherwood <david.sherwood@arm.com>
2656 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2657 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2658 (popcount_loc_descriptor): Likewise.
2659 (bswap_loc_descriptor): Likewise.
2660 (rotate_loc_descriptor): Likewise.
2661 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2662 calling the functions above.
2664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2665 Alan Hayward <alan.hayward@arm.com>
2666 David Sherwood <david.sherwood@arm.com>
2668 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2670 (try_combine): Likewise.
2671 (simplify_if_then_else): Likewise.
2672 * cse.c (cse_insn): Likewise.
2673 * dwarf2out.c (mem_loc_descriptor): Likewise.
2674 * emit-rtl.c (gen_lowpart_common): Likewise.
2675 * simplify-rtx.c (simplify_truncation): Likewise.
2676 (simplify_binary_operation_1): Likewise.
2677 (simplify_const_relational_operation): Likewise.
2678 (simplify_ternary_operation): Likewise.
2679 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2682 Alan Hayward <alan.hayward@arm.com>
2683 David Sherwood <david.sherwood@arm.com>
2685 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2686 * reload.c (push_reload): Likewise.
2687 (find_reloads): Likewise.
2689 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2690 Alan Hayward <alan.hayward@arm.com>
2691 David Sherwood <david.sherwood@arm.com>
2693 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2694 (make_compound_operation_int): Likewise.
2695 (change_zero_ext): Likewise.
2696 * expr.c (convert_move): Likewise.
2697 (convert_modes): Likewise.
2698 * fwprop.c (forward_propagate_subreg): Likewise.
2699 * loop-iv.c (get_biv_step_1): Likewise.
2700 * optabs.c (widen_operand): Likewise.
2701 * postreload.c (move2add_valid_value_p): Likewise.
2702 * recog.c (simplify_while_replacing): Likewise.
2703 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2704 (simplify_binary_operation_1): Likewise. Remove redundant
2705 mode equality check.
2707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2708 Alan Hayward <alan.hayward@arm.com>
2709 David Sherwood <david.sherwood@arm.com>
2711 * combine.c (combine_simplify_rtx): Add checks for
2712 is_a <scalar_int_mode>.
2713 (simplify_if_then_else): Likewise.
2714 (make_field_assignment): Likewise.
2715 (simplify_comparison): Likewise.
2716 * ifcvt.c (noce_try_bitop): Likewise.
2717 * loop-invariant.c (canonicalize_address_mult): Likewise.
2718 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2721 Alan Hayward <alan.hayward@arm.com>
2722 David Sherwood <david.sherwood@arm.com>
2724 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2725 is_a <scalar_int_mode> instead of != BLKmode.
2727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2728 Alan Hayward <alan.hayward@arm.com>
2729 David Sherwood <david.sherwood@arm.com>
2731 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2732 instead of != VOIDmode.
2733 * combine.c (if_then_else_cond): Likewise.
2734 (change_zero_ext): Likewise.
2735 * dwarf2out.c (mem_loc_descriptor): Likewise.
2736 (loc_descriptor): Likewise.
2737 * rtlanal.c (canonicalize_condition): Likewise.
2738 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2741 Alan Hayward <alan.hayward@arm.com>
2742 David Sherwood <david.sherwood@arm.com>
2744 * simplify-rtx.c (simplify_binary_operation_1): Use
2745 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2748 Alan Hayward <alan.hayward@arm.com>
2749 David Sherwood <david.sherwood@arm.com>
2751 * wide-int.h (int_traits<unsigned char>) New class.
2752 (int_traits<unsigned short>) Likewise.
2753 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2754 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2756 * combine.c (set_nonzero_bits_and_sign_copies): Use
2757 is_a <scalar_int_mode>.
2758 (find_split_point): Likewise.
2759 (combine_simplify_rtx): Likewise.
2760 (simplify_logical): Likewise.
2761 (expand_compound_operation): Likewise.
2762 (expand_field_assignment): Likewise.
2763 (make_compound_operation): Likewise.
2764 (extended_count): Likewise.
2765 (change_zero_ext): Likewise.
2766 (simplify_comparison): Likewise.
2767 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2768 (ucompare_loc_descriptor): Likewise.
2769 (minmax_loc_descriptor): Likewise.
2770 (mem_loc_descriptor): Likewise.
2771 (loc_descriptor): Likewise.
2772 * expmed.c (init_expmed_one_mode): Likewise.
2773 * lra-constraints.c (lra_constraint_offset): Likewise.
2774 * optabs.c (prepare_libcall_arg): Likewise.
2775 * postreload.c (move2add_note_store): Likewise.
2776 * reload.c (operands_match_p): Likewise.
2777 * rtl.h (load_extend_op): Likewise.
2778 * rtlhooks.c (gen_lowpart_general): Likewise.
2779 * simplify-rtx.c (simplify_truncation): Likewise.
2780 (simplify_unary_operation_1): Likewise.
2781 (simplify_binary_operation_1): Likewise.
2782 (simplify_const_binary_operation): Likewise.
2783 (simplify_const_relational_operation): Likewise.
2784 (simplify_subreg): Likewise.
2785 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2786 * var-tracking.c (adjust_mems): Likewise.
2787 (prepare_call_arguments): Likewise.
2789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2790 Alan Hayward <alan.hayward@arm.com>
2791 David Sherwood <david.sherwood@arm.com>
2793 * machmode.h (is_int_mode): New fuction.
2794 * combine.c (find_split_point): Use it.
2795 (combine_simplify_rtx): Likewise.
2796 (simplify_if_then_else): Likewise.
2797 (simplify_set): Likewise.
2798 (simplify_shift_const_1): Likewise.
2799 (simplify_comparison): Likewise.
2800 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2801 * cse.c (notreg_cost): Likewise.
2802 (cse_insn): Likewise.
2803 * cselib.c (cselib_lookup_1): Likewise.
2804 * dojump.c (do_jump_1): Likewise.
2805 (do_compare_rtx_and_jump): Likewise.
2806 * dse.c (get_call_args): Likewise.
2807 * dwarf2out.c (rtl_for_decl_init): Likewise.
2808 (native_encode_initializer): Likewise.
2809 * expmed.c (emit_store_flag_1): Likewise.
2810 (emit_store_flag): Likewise.
2811 * expr.c (convert_modes): Likewise.
2812 (store_field): Likewise.
2813 (expand_expr_real_1): Likewise.
2814 * fold-const.c (fold_read_from_constant_string): Likewise.
2815 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2816 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2817 * optabs.c (expand_binop): Likewise.
2818 (expand_unop): Likewise.
2819 (expand_abs_nojump): Likewise.
2820 (expand_one_cmpl_abs_nojump): Likewise.
2821 * simplify-rtx.c (mode_signbit_p): Likewise.
2822 (val_signbit_p): Likewise.
2823 (val_signbit_known_set_p): Likewise.
2824 (val_signbit_known_clear_p): Likewise.
2825 (simplify_relational_operation_1): Likewise.
2826 * tree.c (vector_type_mode): Likewise.
2828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2829 Alan Hayward <alan.hayward@arm.com>
2830 David Sherwood <david.sherwood@arm.com>
2832 * machmode.h (smallest_mode_for_size): Fix formatting.
2833 (smallest_int_mode_for_size): New function.
2834 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2835 instead of smallest_mode_for_size.
2836 * combine.c (make_extraction): Likewise.
2837 * config/arc/arc.c (arc_expand_movmem): Likewise.
2838 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2839 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2840 * config/s390/s390.c (s390_expand_insv): Likewise.
2841 * config/sparc/sparc.c (assign_int_registers): Likewise.
2842 * config/spu/spu.c (spu_function_value): Likewise.
2843 (spu_function_arg): Likewise.
2844 * coverage.c (get_gcov_type): Likewise.
2845 (get_gcov_unsigned_t): Likewise.
2846 * dse.c (find_shift_sequence): Likewise.
2847 * expmed.c (store_bit_field_1): Likewise.
2848 * expr.c (convert_move): Likewise.
2849 (store_field): Likewise.
2850 * internal-fn.c (expand_arith_overflow): Likewise.
2851 * optabs-query.c (get_best_extraction_insn): Likewise.
2852 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2853 * stor-layout.c (layout_type): Likewise.
2854 (initialize_sizetypes): Likewise.
2855 * targhooks.c (default_get_mask_mode): Likewise.
2856 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2859 Alan Hayward <alan.hayward@arm.com>
2860 David Sherwood <david.sherwood@arm.com>
2862 * machmode.h (opt_mode::else_blk): New function.
2863 (int_mode_for_mode): Declare.
2864 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2865 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2867 * cfgexpand.c (expand_debug_expr): Likewise.
2868 * combine.c (gen_lowpart_or_truncate): Likewise.
2869 (gen_lowpart_for_combine): Likewise.
2870 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2871 * config/avr/avr.c (avr_to_int_mode): Likewise.
2872 (avr_out_plus_1): Likewise.
2873 (avr_out_plus): Likewise.
2874 (avr_out_round): Likewise.
2875 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2876 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2877 (s390_expand_vcond): Likewise.
2878 * config/spu/spu.c (spu_split_immediate): Likewise.
2879 (spu_expand_mov): Likewise.
2880 * dse.c (get_stored_val): Likewise.
2881 * expmed.c (store_bit_field_1): Likewise.
2882 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2884 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2885 (extract_low_bits): Likewise.
2886 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2887 handling rather than repeating the check.
2888 (emit_group_store): Likewise.
2889 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2890 * optabs.c (expand_absneg_bit): Likewise.
2891 (expand_copysign_absneg): Likewise.
2892 (expand_copysign_bit): Likewise.
2893 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2894 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2895 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2896 * var-tracking.c (prepare_call_arguments): Likewise.
2897 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2898 int_mode_for_mode instead of mode_for_size.
2899 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2902 Alan Hayward <alan.hayward@arm.com>
2903 David Sherwood <david.sherwood@arm.com>
2905 * machmode.h (int_mode_for_size): New function.
2906 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2907 instead of mode_for_size.
2908 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2910 * combine.c (expand_field_assignment): Use int_mode_for_size
2911 instead of mode_for_size.
2912 (make_extraction): Likewise.
2913 (simplify_shift_const_1): Likewise.
2914 (simplify_comparison): Likewise.
2915 * dojump.c (do_jump): Likewise.
2916 * dwarf2out.c (mem_loc_descriptor): Likewise.
2917 * emit-rtl.c (init_derived_machine_modes): Likewise.
2918 * expmed.c (flip_storage_order): Likewise.
2919 (convert_extracted_bit_field): Likewise.
2920 * expr.c (copy_blkmode_from_reg): Likewise.
2921 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2922 * internal-fn.c (expand_mul_overflow): Likewise.
2923 * lower-subreg.c (simple_move): Likewise.
2924 * optabs-libfuncs.c (init_optabs): Likewise.
2925 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2926 * tree.c (vector_type_mode): Likewise.
2927 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2928 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2929 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2930 * tree-vect-stmts.c (vectorizable_load): Likewise.
2931 (vectorizable_store): Likewise.
2933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2934 Alan Hayward <alan.hayward@arm.com>
2935 David Sherwood <david.sherwood@arm.com>
2937 * coretypes.h (pod_mode): New type.
2938 (scalar_int_mode_pod): New typedef.
2939 * machmode.h (pod_mode): New class.
2940 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2941 * genmodes.c (emit_mode_int_n): Update accordingly.
2942 * lower-subreg.h (target_lower_subreg): Change type to
2943 scalar_int_mode_pod.
2944 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2945 scalar_int_mode_pod.
2947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2948 Alan Hayward <alan.hayward@arm.com>
2949 David Sherwood <david.sherwood@arm.com>
2951 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2952 machine_mode to scalar_int_mode.
2953 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2954 (rs6000_option_override_internal): Remove cast to int.
2955 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2956 machine_mode to scalar_int_mode.
2957 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2958 (rs6000_option_override_internal): Remove cast to int.
2959 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2960 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2962 * config/s390/s390.c (s390_expand_builtin): Likewise.
2963 * coretypes.h (scalar_int_mode): New type.
2964 (opt_scalar_int_mode): New typedef.
2965 * machmode.h (scalar_int_mode): New class.
2966 (scalar_int_mode::includes_p): New function.
2967 (byte_mode): Change type to scalar_int_mode.
2968 (word_mode): Likewise.
2969 (ptr_mode): Likewise.
2970 * emit-rtl.c (byte_mode): Likewise.
2971 (word_mode): Likewise.
2972 (ptr_mode): Likewise.
2973 (init_derived_machine_modes): Update accordingly.
2974 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2975 and MODE_PARTIAL_INT.
2976 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2977 opt_scalar_int_mode.
2979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2980 Alan Hayward <alan.hayward@arm.com>
2981 David Sherwood <david.sherwood@arm.com>
2983 * target.def (libgcc_floating_mode_supported_p): Take a
2985 * doc/tm.texi: Regenerate.
2986 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2988 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2989 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
2992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2993 Alan Hayward <alan.hayward@arm.com>
2994 David Sherwood <david.sherwood@arm.com>
2996 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2997 * doc/tm.texi: Regenerate.
2998 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2999 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3000 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3001 * targhooks.h (default_floatn_mode): Likewise.
3002 * targhooks.c (default_floatn_mode): Likewise.
3003 * tree.c (build_common_tree_nodes): Update accordingly.
3005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3006 Alan Hayward <alan.hayward@arm.com>
3007 David Sherwood <david.sherwood@arm.com>
3009 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3010 (mode_iterator::iterate_p): Likewise.
3011 (mode_iterator::get_wider): Likewise.
3012 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3015 Alan Hayward <alan.hayward@arm.com>
3016 David Sherwood <david.sherwood@arm.com>
3018 * coretypes.h (opt_scalar_float_mode): New typedef.
3019 * machmode.h (float_mode_for_size): New function.
3020 * emit-rtl.c (double_mode): Delete.
3021 (init_emit_once): Use float_mode_for_size.
3022 * stor-layout.c (layout_type): Likewise.
3023 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3026 Alan Hayward <alan.hayward@arm.com>
3027 David Sherwood <david.sherwood@arm.com>
3029 * output.h (assemble_real): Take a scalar_float_mode.
3030 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3031 * config/arm/arm.md (consttable_4): Likewise.
3032 (consttable_8): Likewise.
3033 (consttable_16): Likewise.
3034 * config/mips/mips.md (consttable_float): Likewise.
3035 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3036 * varasm.c (assemble_real): Take a scalar_float_mode.
3037 (output_constant_pool_2): Update accordingly.
3038 (output_constant): Likewise.
3040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3041 Alan Hayward <alan.hayward@arm.com>
3042 David Sherwood <david.sherwood@arm.com>
3044 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3045 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3046 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3047 (native_encode_real): Likewise.
3048 (native_interpret_real): Likewise.
3049 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3050 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3053 Alan Hayward <alan.hayward@arm.com>
3054 David Sherwood <david.sherwood@arm.com>
3056 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3057 <scalar_float_mode>. Simplify.
3058 (gen_extend_conv_libfunc): Likewise.
3060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3061 Alan Hayward <alan.hayward@arm.com>
3062 David Sherwood <david.sherwood@arm.com>
3064 * coretypes.h (scalar_float_mode): New type.
3065 * machmode.h (mode_traits::from_int): Use machine_mode if
3066 USE_ENUM_MODES is defined.
3067 (is_a): New function.
3069 (dyn_cast): Likewise.
3070 (scalar_float_mode): New class.
3071 (scalar_float_mode::includes_p): New function.
3072 (is_float_mode): Likewise.
3073 * gdbhooks.py (MachineModePrinter): New class.
3074 (build_pretty_printer): Use it for scalar_float_mode.
3075 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3076 (format_helper::format_helper): Turn into a template.
3077 * genmodes.c (get_mode_class): New function.
3078 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3079 or machine_mode if none.
3080 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3081 as_a <scalar_float_mode>.
3082 * dwarf2out.c (mem_loc_descriptor): Likewise.
3083 (insert_float): Likewise.
3084 (add_const_value_attribute): Likewise.
3085 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3086 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3087 (expand_unop): Update accordingly.
3088 (expand_abs_nojump): Likewise.
3089 (expand_copysign_absneg): Take a scalar_float_mode.
3090 (expand_copysign_bit): Likewise.
3091 (expand_copysign): Update accordingly.
3093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3094 Alan Hayward <alan.hayward@arm.com>
3095 David Sherwood <david.sherwood@arm.com>
3097 * coretypes.h (opt_mode): New class.
3098 * machmode.h (opt_mode): Likewise.
3099 (opt_mode::else_void): New function.
3100 (opt_mode::require): Likewise.
3101 (opt_mode::exists): Likewise.
3102 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3103 (GET_MODE_2XWIDER_MODE): Likewise.
3104 (mode_iterator::get_wider): Update accordingly.
3105 (mode_iterator::get_2xwider): Likewise.
3106 (mode_iterator::get_known_wider): Likewise, turning into a template.
3107 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3108 forcing a wider mode to exist.
3109 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3110 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3111 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3112 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3113 * lower-subreg.c (init_lower_subreg): Likewise.
3114 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3115 on the final iteration.
3116 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3117 a wider mode exists before asking for a move pattern.
3118 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3119 forcing a wider mode to exist.
3120 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3121 returning false if no such mode exists.
3122 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3123 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3124 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3125 Avoid checking for a MODE_INT if we already know the mode is not a
3127 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3128 forcing a wider mode to exist.
3129 (expmed_mult_highpart_optab): Likewise.
3130 (expmed_mult_highpart): Likewise.
3131 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3133 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3134 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3135 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3136 * internal-fn.c (expand_mul_overflow): Update use of
3137 GET_MODE_2XWIDER_MODE.
3138 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3139 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3140 GET_MODE_WIDER_MODE.
3141 (convert_plusminus_to_widen): Likewise.
3142 * tree-switch-conversion.c (array_value_type): Likewise.
3143 * var-tracking.c (emit_note_insn_var_location): Likewise.
3144 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3145 Return false inside rather than outside the loop if no wider mode
3147 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3148 and GET_MODE_2XWIDER_MODE
3149 (can_compare_p): Use else_void.
3150 * gdbhooks.py (OptMachineModePrinter): New class.
3151 (build_pretty_printer): Use it for opt_mode.
3153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3154 Alan Hayward <alan.hayward@arm.com>
3155 David Sherwood <david.sherwood@arm.com>
3157 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3158 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3161 Alan Hayward <alan.hayward@arm.com>
3162 David Sherwood <david.sherwood@arm.com>
3164 * machmode.h (mode_traits): New structure.
3165 (get_narrowest_mode): New function.
3166 (mode_iterator::start): Likewise.
3167 (mode_iterator::iterate_p): Likewise.
3168 (mode_iterator::get_wider): Likewise.
3169 (mode_iterator::get_known_wider): Likewise.
3170 (mode_iterator::get_2xwider): Likewise.
3171 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3172 (FOR_EACH_MODE): Likewise.
3173 (FOR_EACH_MODE_FROM): Likewise.
3174 (FOR_EACH_MODE_UNTIL): Likewise.
3175 (FOR_EACH_WIDER_MODE): Likewise.
3176 (FOR_EACH_2XWIDER_MODE): Likewise.
3177 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3178 * combine.c (simplify_comparison): Likewise
3179 * config/i386/i386.c (type_natural_mode): Likewise.
3180 * cse.c (cse_insn): Likewise.
3181 * dse.c (find_shift_sequence): Likewise.
3182 * emit-rtl.c (init_derived_machine_modes): Likewise.
3183 (init_emit_once): Likewise.
3184 * explow.c (hard_function_value): Likewise.
3185 * expmed.c (extract_fixed_bit_field_1): Likewise.
3186 (extract_bit_field_1): Likewise.
3187 (expand_divmod): Likewise.
3188 (emit_store_flag_1): Likewise.
3189 * expr.c (init_expr_target): Likewise.
3190 (convert_move): Likewise.
3191 (alignment_for_piecewise_move): Likewise.
3192 (widest_int_mode_for_size): Likewise.
3193 (emit_block_move_via_movmem): Likewise.
3194 (copy_blkmode_to_reg): Likewise.
3195 (set_storage_via_setmem): Likewise.
3196 (compress_float_constant): Likewise.
3197 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3198 * optabs-query.c (get_best_extraction_insn): Likewise.
3199 * optabs.c (expand_binop): Likewise.
3200 (expand_twoval_unop): Likewise.
3201 (expand_twoval_binop): Likewise.
3202 (widen_leading): Likewise.
3203 (widen_bswap): Likewise.
3204 (expand_parity): Likewise.
3205 (expand_unop): Likewise.
3206 (prepare_cmp_insn): Likewise.
3207 (prepare_float_lib_cmp): Likewise.
3208 (expand_float): Likewise.
3209 (expand_fix): Likewise.
3210 (expand_sfix_optab): Likewise.
3211 * postreload.c (move2add_use_add2_insn): Likewise.
3212 * reg-stack.c (reg_to_stack): Likewise.
3213 * reginfo.c (choose_hard_reg_mode): Likewise.
3214 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3215 * stor-layout.c (mode_for_size): Likewise.
3216 (smallest_mode_for_size): Likewise.
3217 (mode_for_vector): Likewise.
3218 (finish_bitfield_representative): Likewise.
3219 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3220 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3221 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3222 * var-tracking.c (prepare_call_arguments): Likewise.
3224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3225 Alan Hayward <alan.hayward@arm.com>
3226 David Sherwood <david.sherwood@arm.com>
3228 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3229 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3230 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3231 * machmode.h (mode_size): Move earlier in file.
3232 (mode_precision): Likewise.
3233 (mode_inner): Likewise.
3234 (mode_nunits): Likewise.
3235 (mode_unit_size): Likewise.
3236 (unit_unit_precision): Likewise.
3237 (mode_wider): Likewise.
3238 (mode_2xwider): Likewise.
3239 (machine_mode): New class.
3240 (mode_to_bytes): New function.
3241 (mode_to_bits): Likewise.
3242 (mode_to_precision): Likewise.
3243 (mode_to_inner): Likewise.
3244 (mode_to_unit_size): Likewise.
3245 (mode_to_unit_precision): Likewise.
3246 (mode_to_nunits): Likewise.
3247 (GET_MODE_SIZE): Use mode_to_bytes.
3248 (GET_MODE_BITSIZE): Use mode_to_bits.
3249 (GET_MODE_PRECISION): Use mode_to_precision.
3250 (GET_MODE_INNER): Use mode_to_inner.
3251 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3252 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3253 (GET_MODE_NUNITS): Use mode_to_nunits.
3254 * system.h (ALWAYS_INLINE): New macro.
3255 * config/powerpcspe/powerpcspe-c.c
3256 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3257 int for arg1_mode and arg2_mode.
3259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3260 Alan Hayward <alan.hayward@arm.com>
3261 David Sherwood <david.sherwood@arm.com>
3263 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3264 Prefix mode names with E_ in case statements.
3265 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3266 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3267 (aarch64_split_simd_move): Likewise.
3268 (aarch64_gen_storewb_pair): Likewise.
3269 (aarch64_gen_loadwb_pair): Likewise.
3270 (aarch64_gen_store_pair): Likewise.
3271 (aarch64_gen_load_pair): Likewise.
3272 (aarch64_get_condition_code_1): Likewise.
3273 (aarch64_constant_pool_reload_icode): Likewise.
3274 (get_rsqrte_type): Likewise.
3275 (get_rsqrts_type): Likewise.
3276 (get_recpe_type): Likewise.
3277 (get_recps_type): Likewise.
3278 (aarch64_gimplify_va_arg_expr): Likewise.
3279 (aarch64_simd_container_mode): Likewise.
3280 (aarch64_emit_load_exclusive): Likewise.
3281 (aarch64_emit_store_exclusive): Likewise.
3282 (aarch64_expand_compare_and_swap): Likewise.
3283 (aarch64_gen_atomic_cas): Likewise.
3284 (aarch64_emit_bic): Likewise.
3285 (aarch64_emit_atomic_swap): Likewise.
3286 (aarch64_emit_atomic_load_op): Likewise.
3287 (aarch64_evpc_trn): Likewise.
3288 (aarch64_evpc_uzp): Likewise.
3289 (aarch64_evpc_zip): Likewise.
3290 (aarch64_evpc_ext): Likewise.
3291 (aarch64_evpc_rev): Likewise.
3292 (aarch64_evpc_dup): Likewise.
3293 (aarch64_gen_ccmp_first): Likewise.
3294 (aarch64_gen_ccmp_next): Likewise.
3295 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3296 (alpha_emit_xfloating_libcall): Likewise.
3297 (emit_insxl): Likewise.
3298 (alpha_arg_type): Likewise.
3299 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3300 (arc_preferred_simd_mode): Likewise.
3301 (arc_secondary_reload): Likewise.
3302 (get_arc_condition_code): Likewise.
3303 (arc_print_operand): Likewise.
3304 (arc_legitimate_constant_p): Likewise.
3305 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3306 * config/arc/arc.md (casesi_load): Likewise.
3307 (casesi_compact_jump): Likewise.
3308 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3309 (cc_use_register): Likewise.
3310 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3311 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3312 (arm_init_iwmmxt_builtins): Likewise.
3313 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3314 (neon_expand_vector_init): Likewise.
3315 (arm_attr_length_move_neon): Likewise.
3316 (maybe_get_arm_condition_code): Likewise.
3317 (arm_emit_vector_const): Likewise.
3318 (arm_preferred_simd_mode): Likewise.
3319 (arm_output_iwmmxt_tinsr): Likewise.
3320 (thumb1_output_casesi): Likewise.
3321 (thumb2_output_casesi): Likewise.
3322 (arm_emit_load_exclusive): Likewise.
3323 (arm_emit_store_exclusive): Likewise.
3324 (arm_expand_compare_and_swap): Likewise.
3325 (arm_evpc_neon_vuzp): Likewise.
3326 (arm_evpc_neon_vzip): Likewise.
3327 (arm_evpc_neon_vrev): Likewise.
3328 (arm_evpc_neon_vtrn): Likewise.
3329 (arm_evpc_neon_vext): Likewise.
3330 (arm_validize_comparison): Likewise.
3331 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3332 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3333 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3334 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3335 (c6x_preferred_simd_mode): Likewise.
3336 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3337 (epiphany_rtx_costs): Likewise.
3338 * config/epiphany/predicates.md (proper_comparison_operator):
3340 * config/frv/frv.c (condexec_memory_operand): Likewise.
3341 (frv_emit_move): Likewise.
3342 (output_move_single): Likewise.
3343 (output_condmove_single): Likewise.
3344 (frv_hard_regno_mode_ok): Likewise.
3345 (frv_matching_accg_mode): Likewise.
3346 * config/h8300/h8300.c (split_adds_subs): Likewise.
3347 (h8300_rtx_costs): Likewise.
3348 (h8300_print_operand): Likewise.
3349 (compute_mov_length): Likewise.
3350 (output_logical_op): Likewise.
3351 (compute_logical_op_length): Likewise.
3352 (compute_logical_op_cc): Likewise.
3353 (h8300_shift_needs_scratch_p): Likewise.
3354 (output_a_shift): Likewise.
3355 (compute_a_shift_length): Likewise.
3356 (compute_a_shift_cc): Likewise.
3357 (expand_a_rotate): Likewise.
3358 (output_a_rotate): Likewise.
3359 * config/i386/i386.c (classify_argument): Likewise.
3360 (function_arg_advance_32): Likewise.
3361 (function_arg_32): Likewise.
3362 (function_arg_64): Likewise.
3363 (function_value_64): Likewise.
3364 (ix86_gimplify_va_arg): Likewise.
3365 (ix86_legitimate_constant_p): Likewise.
3366 (put_condition_code): Likewise.
3367 (split_double_mode): Likewise.
3368 (ix86_avx256_split_vector_move_misalign): Likewise.
3369 (ix86_expand_vector_logical_operator): Likewise.
3370 (ix86_split_idivmod): Likewise.
3371 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3372 (ix86_build_const_vector): Likewise.
3373 (ix86_build_signbit_mask): Likewise.
3374 (ix86_match_ccmode): Likewise.
3375 (ix86_cc_modes_compatible): Likewise.
3376 (ix86_expand_branch): Likewise.
3377 (ix86_expand_sse_cmp): Likewise.
3378 (ix86_expand_sse_movcc): Likewise.
3379 (ix86_expand_int_sse_cmp): Likewise.
3380 (ix86_expand_vec_perm_vpermi2): Likewise.
3381 (ix86_expand_vec_perm): Likewise.
3382 (ix86_expand_sse_unpack): Likewise.
3383 (ix86_expand_int_addcc): Likewise.
3384 (ix86_split_to_parts): Likewise.
3385 (ix86_vectorize_builtin_gather): Likewise.
3386 (ix86_vectorize_builtin_scatter): Likewise.
3387 (avx_vpermilp_parallel): Likewise.
3388 (inline_memory_move_cost): Likewise.
3389 (ix86_tieable_integer_mode_p): Likewise.
3390 (x86_maybe_negate_const_int): Likewise.
3391 (ix86_expand_vector_init_duplicate): Likewise.
3392 (ix86_expand_vector_init_one_nonzero): Likewise.
3393 (ix86_expand_vector_init_one_var): Likewise.
3394 (ix86_expand_vector_init_concat): Likewise.
3395 (ix86_expand_vector_init_interleave): Likewise.
3396 (ix86_expand_vector_init_general): Likewise.
3397 (ix86_expand_vector_set): Likewise.
3398 (ix86_expand_vector_extract): Likewise.
3399 (emit_reduc_half): Likewise.
3400 (ix86_emit_i387_round): Likewise.
3401 (ix86_mangle_type): Likewise.
3402 (ix86_expand_round_sse4): Likewise.
3403 (expand_vec_perm_blend): Likewise.
3404 (canonicalize_vector_int_perm): Likewise.
3405 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3406 (expand_vec_perm_1): Likewise.
3407 (expand_vec_perm_interleave3): Likewise.
3408 (expand_vec_perm_even_odd_pack): Likewise.
3409 (expand_vec_perm_even_odd_1): Likewise.
3410 (expand_vec_perm_broadcast_1): Likewise.
3411 (ix86_vectorize_vec_perm_const_ok): Likewise.
3412 (ix86_expand_vecop_qihi): Likewise.
3413 (ix86_expand_mul_widen_hilo): Likewise.
3414 (ix86_expand_sse2_abs): Likewise.
3415 (ix86_expand_pextr): Likewise.
3416 (ix86_expand_pinsr): Likewise.
3417 (ix86_preferred_simd_mode): Likewise.
3418 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3419 * config/i386/sse.md (*andnot<mode>3): Likewise.
3420 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3421 (*<code><mode>3): Likewise.
3422 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3423 (ia64_expand_atomic_op): Likewise.
3424 (ia64_arg_type): Likewise.
3425 (ia64_mode_to_int): Likewise.
3426 (ia64_scalar_mode_supported_p): Likewise.
3427 (ia64_vector_mode_supported_p): Likewise.
3428 (expand_vec_perm_broadcast): Likewise.
3429 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3430 (iq2000_function_arg_advance): Likewise.
3431 (iq2000_function_arg): Likewise.
3432 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3433 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3434 (m68k_libcall_value): Likewise.
3435 (m68k_function_value): Likewise.
3436 (sched_attr_op_type): Likewise.
3437 * config/mcore/mcore.c (mcore_output_move): Likewise.
3438 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3440 (microblaze_function_arg): Likewise.
3441 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3442 (mips_print_operand): Likewise.
3443 (mips_mode_ok_for_mov_fmt_p): Likewise.
3444 (mips_vector_mode_supported_p): Likewise.
3445 (mips_preferred_simd_mode): Likewise.
3446 (mips_expand_vpc_loongson_even_odd): Likewise.
3447 (mips_expand_vec_unpack): Likewise.
3448 (mips_expand_vi_broadcast): Likewise.
3449 (mips_expand_vector_init): Likewise.
3450 (mips_expand_vec_reduc): Likewise.
3451 (mips_expand_msa_cmp): Likewise.
3452 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3453 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3454 (cc_flags_for_mode): Likewise.
3455 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3456 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3457 (nds32_output_casesi_pc_relative): Likewise.
3458 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3459 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3460 (nvptx_gen_unpack): Likewise.
3461 (nvptx_gen_pack): Likewise.
3462 (nvptx_gen_shuffle): Likewise.
3463 (nvptx_gen_wcast): Likewise.
3464 (nvptx_preferred_simd_mode): Likewise.
3465 * config/pa/pa.c (pa_secondary_reload): Likewise.
3466 * config/pa/predicates.md (base14_operand): Likewise.
3467 * config/powerpcspe/powerpcspe-c.c
3468 (altivec_resolve_overloaded_builtin): Likewise.
3469 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3471 (rs6000_preferred_simd_mode): Likewise.
3472 (output_vec_const_move): Likewise.
3473 (rs6000_expand_vector_extract): Likewise.
3474 (rs6000_split_vec_extract_var): Likewise.
3475 (reg_offset_addressing_ok_p): Likewise.
3476 (rs6000_legitimate_offset_address_p): Likewise.
3477 (rs6000_legitimize_address): Likewise.
3478 (rs6000_emit_set_const): Likewise.
3479 (rs6000_const_vec): Likewise.
3480 (rs6000_emit_move): Likewise.
3481 (spe_build_register_parallel): Likewise.
3482 (rs6000_darwin64_record_arg_recurse): Likewise.
3483 (swap_selector_for_mode): Likewise.
3484 (spe_init_builtins): Likewise.
3485 (paired_init_builtins): Likewise.
3486 (altivec_init_builtins): Likewise.
3487 (do_load_for_compare): Likewise.
3488 (rs6000_generate_compare): Likewise.
3489 (rs6000_expand_float128_convert): Likewise.
3490 (emit_load_locked): Likewise.
3491 (emit_store_conditional): Likewise.
3492 (rs6000_output_function_epilogue): Likewise.
3493 (rs6000_handle_altivec_attribute): Likewise.
3494 (rs6000_function_value): Likewise.
3495 (emit_fusion_gpr_load): Likewise.
3496 (emit_fusion_p9_load): Likewise.
3497 (emit_fusion_p9_store): Likewise.
3498 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3499 (fusion_gpr_mem_load): Likewise.
3500 (fusion_addis_mem_combo_load): Likewise.
3501 (fusion_addis_mem_combo_store): Likewise.
3502 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3503 (fusion_gpr_mem_load): Likewise.
3504 (fusion_addis_mem_combo_load): Likewise.
3505 (fusion_addis_mem_combo_store): Likewise.
3506 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3508 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3509 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3510 (rs6000_preferred_simd_mode): Likewise.
3511 (output_vec_const_move): Likewise.
3512 (rs6000_expand_vector_extract): Likewise.
3513 (rs6000_split_vec_extract_var): Likewise.
3514 (reg_offset_addressing_ok_p): Likewise.
3515 (rs6000_legitimate_offset_address_p): Likewise.
3516 (rs6000_legitimize_address): Likewise.
3517 (rs6000_emit_set_const): Likewise.
3518 (rs6000_const_vec): Likewise.
3519 (rs6000_emit_move): Likewise.
3520 (rs6000_darwin64_record_arg_recurse): Likewise.
3521 (swap_selector_for_mode): Likewise.
3522 (paired_init_builtins): Likewise.
3523 (altivec_init_builtins): Likewise.
3524 (rs6000_expand_float128_convert): Likewise.
3525 (emit_load_locked): Likewise.
3526 (emit_store_conditional): Likewise.
3527 (rs6000_output_function_epilogue): Likewise.
3528 (rs6000_handle_altivec_attribute): Likewise.
3529 (rs6000_function_value): Likewise.
3530 (emit_fusion_gpr_load): Likewise.
3531 (emit_fusion_p9_load): Likewise.
3532 (emit_fusion_p9_store): Likewise.
3533 * config/rx/rx.c (rx_gen_move_template): Likewise.
3534 (flags_from_mode): Likewise.
3535 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3536 (s390_slb_comparison): Likewise.
3537 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3538 (s390_vector_mode_supported_p): Likewise.
3539 (s390_cc_modes_compatible): Likewise.
3540 (s390_match_ccmode_set): Likewise.
3541 (s390_canonicalize_comparison): Likewise.
3542 (s390_emit_compare_and_swap): Likewise.
3543 (s390_branch_condition_mask): Likewise.
3544 (s390_rtx_costs): Likewise.
3545 (s390_secondary_reload): Likewise.
3546 (__SECONDARY_RELOAD_CASE): Likewise.
3547 (s390_expand_cs): Likewise.
3548 (s390_preferred_simd_mode): Likewise.
3549 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3550 * config/sh/sh.c (sh_print_operand): Likewise.
3551 (dump_table): Likewise.
3552 (sh_secondary_reload): Likewise.
3553 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3554 * config/sh/sh.md (casesi_worker_1): Likewise.
3555 (casesi_worker_2): Likewise.
3556 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3557 (fcc_comparison_operator): Likewise.
3558 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3559 (emit_soft_tfmode_cvt): Likewise.
3560 (sparc_preferred_simd_mode): Likewise.
3561 (output_cbranch): Likewise.
3562 (sparc_print_operand): Likewise.
3563 (sparc_expand_vec_perm_bmask): Likewise.
3564 (vector_init_bshuffle): Likewise.
3565 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3566 (spu_vector_mode_supported_p): Likewise.
3567 (spu_expand_insv): Likewise.
3568 (spu_emit_branch_or_set): Likewise.
3569 (spu_handle_vector_attribute): Likewise.
3570 (spu_builtin_splats): Likewise.
3571 (spu_builtin_extract): Likewise.
3572 (spu_builtin_promote): Likewise.
3573 (spu_expand_sign_extend): Likewise.
3574 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3575 (tilegx_simd_int): Likewise.
3576 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3577 (tilepro_simd_int): Likewise.
3578 * config/v850/v850.c (const_double_split): Likewise.
3579 (v850_print_operand): Likewise.
3580 (ep_memory_offset): Likewise.
3581 * config/vax/vax.c (vax_rtx_costs): Likewise.
3582 (vax_output_int_move): Likewise.
3583 (vax_output_int_add): Likewise.
3584 (vax_output_int_subtract): Likewise.
3585 * config/visium/predicates.md (visium_branch_operator): Likewise.
3586 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3587 (visium_print_operand_address): Likewise.
3588 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3589 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3590 (xtensa_expand_conditional_branch): Likewise.
3591 (xtensa_copy_incoming_a7): Likewise.
3592 (xtensa_output_literal): Likewise.
3593 * dfp.c (decimal_real_maxval): Likewise.
3594 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3597 Alan Hayward <alan.hayward@arm.com>
3598 David Sherwood <david.sherwood@arm.com>
3600 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3601 (mode_nunits_inline): Likewise.
3602 (mode_inner_inline): Likewise.
3603 (mode_unit_size_inline): Likewise.
3604 (mode_unit_precision_inline): Likewise.
3605 (emit_insn_modes_h): Likewise. Also emit a #define of the
3607 (emit_mode_wider): Add an E_ prefix to mode names.
3608 (emit_mode_complex): Likewise.
3609 (emit_mode_inner): Likewise.
3610 (emit_mode_adjustments): Likewise.
3611 (emit_mode_int_n): Likewise.
3612 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3613 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3614 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3615 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3616 (CRC32_BUILTIN, ENTRY): Likewise.
3617 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3618 (aarch64_pop_regs): Likewise.
3619 (aarch64_process_components): Likewise.
3620 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3621 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3622 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3623 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3624 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3625 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3627 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3628 (pcmpistr): Likewise.
3629 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3630 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3631 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3633 * config/rl78/rl78.c (mduc_regs): Likewise.
3634 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3635 (htm_expand_builtin): Likewise.
3636 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3637 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3638 * config/xtensa/xtensa.c (print_operand): Likewise.
3639 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3640 (NUM_MODE_VECTOR_INT): Likewise.
3641 * genoutput.c (null_operand): Likewise.
3642 (output_operand_data): Likewise.
3643 * genrecog.c (print_parameter_value): Likewise.
3644 * lra.c (debug_operand_data): Likewise.
3646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3648 * dwarf2out.c (scompare_loc_descriptor_wide)
3649 (scompare_loc_descriptor_narrow): New functions, split out from...
3650 (scompare_loc_descriptor): ...here.
3651 * expmed.c (emit_store_flag_int): New function, split out from...
3652 (emit_store_flag): ...here.
3654 2017-08-30 Richard Biener <rguenther@suse.de>
3656 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3657 (dwarf2out_early_finish): Move setting of AT_pubnames from
3658 early debug output to early finish.
3660 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3662 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3663 and -mdata-region to the assembler.
3665 2017-08-30 Richard Biener <rguenther@suse.de>
3667 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3669 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3670 (dwarf2out_early_global_decl): Only generate a DIE for the
3671 abstract origin if it doesn't already exist or is a declaration DIE.
3672 (resolve_addr): Do not add the linkage name twice when
3673 generating a stub DIE for the DW_TAG_GNU_call_site target.
3675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3677 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3678 Use machine_mode rather than int for arg1_mode.
3680 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3683 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3684 that the second argument of the built-in functions to unpack
3685 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3686 switch statement instead a lot of if statements.
3687 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3688 Allow 64-bit values to be in Altivec registers as well as
3689 traditional floating point registers.
3690 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3692 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3694 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3695 MAX_REGS_PER_ADDRESS == 1.
3697 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3699 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3700 * config/i386/i386.c (ix86_option_override_internal): Simplify
3701 setting of opts->x_flag_entry.
3703 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3704 Jakub Jelinek <jakub@redhat.com>
3705 Richard Biener <rguenther@suse.de>
3707 PR tree-optimization/81503
3708 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3709 folded constant fits in the target type; reorder tests for clarity.
3711 2017-08-29 Martin Liska <mliska@suse.cz>
3713 * passes.def: Include pass_lower_switch.
3714 * stmt.c (dump_case_nodes): Remove and move to
3715 tree-switch-conversion.
3716 (case_values_threshold): Likewise.
3717 (expand_switch_as_decision_tree_p): Likewise.
3718 (emit_case_decision_tree): Likewise.
3719 (expand_case): Likewise.
3720 (balance_case_nodes): Likewise.
3721 (node_has_low_bound): Likewise.
3722 (node_has_high_bound): Likewise.
3723 (node_is_bounded): Likewise.
3724 (emit_case_nodes): Likewise.
3725 (struct simple_case_node): New struct.
3726 (add_case_node): Remove.
3727 (emit_case_dispatch_table): Use vector instead of case_list.
3728 (reset_out_edges_aux): Remove.
3729 (compute_cases_per_edge): Likewise.
3730 (expand_case): Build list of simple_case_node.
3731 (expand_sjlj_dispatch_table): Use it.
3732 * tree-switch-conversion.c (struct case_node): Moved from
3733 stmt.c and adjusted.
3734 (emit_case_nodes): Likewise.
3735 (node_has_low_bound): Likewise.
3736 (node_has_high_bound): Likewise.
3737 (node_is_bounded): Likewise.
3738 (case_values_threshold): Likewise.
3739 (reset_out_edges_aux): Likewise.
3740 (compute_cases_per_edge): Likewise.
3741 (add_case_node): Likewise.
3742 (dump_case_nodes): Likewise.
3743 (balance_case_nodes): Likewise.
3744 (expand_switch_as_decision_tree_p): Likewise.
3745 (emit_jump): Likewise.
3746 (emit_case_decision_tree): Likewise.
3747 (try_switch_expansion): Likewise.
3748 (do_jump_if_equal): Likewise.
3749 (emit_cmp_and_jump_insns): Likewise.
3750 (fix_phi_operands_for_edge): New function.
3751 (record_phi_operand_mapping): Likewise.
3752 (class pass_lower_switch): New pass.
3753 (pass_lower_switch::execute): New function.
3754 (make_pass_lower_switch): Likewise.
3755 (conditional_probability):
3756 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3757 * tree-pass.h: Add make_pass_lower_switch.
3759 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3762 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3765 2017-08-29 Richard Biener <rguenther@suse.de>
3767 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3768 we do not add a DW_AT_inline attribute twice.
3769 (gen_subprogram_die): Remove code setting DW_AT_inline on
3770 DECL_ABSTRACT_P nodes.
3772 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3774 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3775 calls to internal functions.
3776 (gimplify_modify_expr): Likewise.
3777 * tree-call-cdce.c (use_internal_fn): Likewise.
3778 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3779 (convert_to_divmod): Set the nothrow flag.
3780 * tree-if-conv.c (predicate_mem_writes): Likewise.
3781 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3782 (vectorizable_call): Likewise.
3783 (vectorizable_store): Likewise.
3784 (vectorizable_load): Likewise.
3785 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3786 (vect_recog_mask_conversion_pattern): Likewise.
3788 2017-08-29 Martin Liska <mliska@suse.cz>
3791 * gcc.c (driver_handle_option): Add new argument.
3792 * opts-common.c (handle_option): Pass
3793 target_option_override_hook.
3794 * opts-global.c (lang_handle_option): Add new option.
3795 (set_default_handlers): Add new argument.
3796 (decode_options): Likewise.
3797 * opts.c (target_handle_option): Likewise.
3798 (common_handle_option): Call target_option_override_hook.
3799 * opts.h (struct cl_option_handler_func): Add hook for
3800 target option override.
3801 (struct cl_option_handlers): Likewise.
3802 (set_default_handlers): Add new argument.
3803 (decode_options): Likewise.
3804 (common_handle_option): Likewise.
3805 (target_handle_option): Likewise.
3806 * toplev.c (toplev::main): Pass targetm.target_option.override
3809 2017-08-29 Richard Biener <rguenther@suse.de>
3810 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3812 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3813 life to the active subtree.
3815 2017-08-28 Jeff Law <law@redhat.com>
3817 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3818 derive_equivalences.
3819 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3821 (edge_info::derive_equivalences): New private member function
3823 * tree-ssa-dom.c (class edge_info): Changed from a struct
3824 to a class. Add ctor/dtor, methods and data members.
3825 (edge_info::edge_info): Renamed from allocate_edge_info.
3826 Initialize additional members.
3827 (edge_info::~edge_info): New.
3828 (free_dom_edge_info): Delete the edge info.
3829 (record_edge_info): Use new class & associated member functions.
3830 Tighten forms for testing for edge equivalences.
3831 (record_temporary_equivalences): Iterate over the simple
3832 equivalences rather than assuming there's only one per edge.
3833 (cprop_into_successor_phis): Iterate over the simple
3834 equivalences rather than assuming there's only one per edge.
3835 (optimize_stmt): Use operand_equal_p rather than pointer
3836 equality for mini-DSE code.
3838 2017-08-28 Nathan Sidwell <nathan@acm.org>
3840 * gcc.c (execute): Fold SIGPIPE handling into switch
3841 statement. Adjust internal error message.
3843 2017-08-28 Richard Biener <rguenther@suse.de>
3846 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3847 Do nothing for removed DIEs.
3849 2017-08-28 Richard Biener <rguenther@suse.de>
3851 PR tree-optimization/81977
3852 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3855 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3858 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3859 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3860 using targetm.gen_mem_thread_fence.
3862 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3865 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3866 predicate to register_operand. Reorder operands.
3867 (*btr<mode>): Ditto.
3868 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3869 (*btr<mode>_mask): Ditto.
3871 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3873 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3874 * config/rs6000/xmmintrin.h: New file.
3875 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3877 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3880 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3881 parameter and_insn and return it.
3882 (recombine_lvx_pattern): Insert a copy to ensure availability of
3883 the base register of the copied masking operation at the point of
3884 the instruction replacement.
3885 (recombine_stvx_pattern): Likewise.
3887 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3889 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3890 undocumented switches.
3891 (-mpower9-dform-vector): Likewise.
3892 (-mpower9-dform): Likewise.
3893 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3894 comments to delete references to -mpower9-dform* switches.
3895 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3896 Delete reference to -mpower9-dform* switches, test for
3897 -mpower9-vector instead.
3898 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3899 (OTHER_P9_VECTOR_MASKS): Likewise.
3900 (POWERPC_MASKS): Likewise.
3901 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3902 tests against -mpower9-dform* to -mpower9-vector. Delete code
3903 that checked for -mpower9-dform* consistancy with other options.
3904 Add test for -mpower9-misc to enable other power9 switches.
3905 (rs6000_init_hard_regno_mode_ok): Likewise.
3906 (rs6000_option_override_internal): Likewise.
3907 (rs6000_emit_prologue): Likewise.
3908 (rs6000_emit_epilogue): Likewise.
3909 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3910 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3911 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3913 (emit_fusion_p9_store): Likewise.
3914 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3915 resetting these macros if the assembler does not support ISA 3.0
3917 (TARGET_P9_DFORM_VECTOR): Likewise.
3918 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3919 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3921 2017-08-25 Alan Modra <amodra@gmail.com>
3923 PR rtl-optimization/81747
3924 * cse.c (cse_extended_basic_block): Don't attempt to record
3925 equivalences for degenerate conditional branches that branch
3926 to their fall-through.
3928 2017-08-24 Martin Sebor <msebor@redhat.com>
3931 * gimple-fold.c (size_must_be_zero_p): New function.
3932 (gimple_fold_builtin_memory_op): Call it.
3934 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3936 * config/rs6000/mm_malloc.h: New file.
3938 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3940 PR tree-optimization/81913
3941 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3942 analysis when either IVs in condition can wrap.
3944 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3946 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3947 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3949 2017-08-24 Richard Biener <rguenther@suse.de>
3952 * targhooks.c (default_target_can_inline_p): Properly
3953 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3954 is present and always compare.
3955 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3956 infer -mfpmath=sse from TARGET_SSE_P.
3957 (ix86_can_inline_p): Properly use target_option_default_node when
3958 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3960 2017-08-24 Richard Biener <rguenther@suse.de>
3963 * dwarf2out.c (output_die): Handle flag_generate_offload like
3965 (output_comp_unit): Likewise.
3966 (gen_array_type_die): Likewise.
3967 (dwarf2out_early_finish): Likewise.
3968 (note_variable_value_in_expr): Likewise.
3969 (dwarf2out_finish): Likewise. Adjust assert.
3970 * cgraphunit.c (symbol_table::compile): Move setting of
3971 flag_generate_offload earlier ...
3972 (symbol_table::finalize_compilation_unit): ... here, before
3973 early debug finalization.
3975 2017-08-24 Richard Biener <rguenther@suse.de>
3977 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3978 and ipa-fnsummary.h.
3979 (ix86_can_inline_p): When ix86_fpmath flags do not match
3980 check whether the callee uses FP math at all.
3982 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3985 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3986 instead of TYPE_PRECISION.
3988 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3989 Alan Hayward <alan.hayward@arm.com>
3990 David Sherwood <david.sherwood@arm.com>
3992 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3994 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3996 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3998 (ix86_frame::stack_realign_allocate): New field.
3999 (struct machine_frame_state): Modify comments.
4000 (machine_frame_state::sp_realigned_fp_end): New field.
4001 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4003 (sp_valid_at): Add assertion to assure no attempt to access invalid
4004 offset of a realigned stack.
4005 (fp_valid_at): Likewise.
4006 (choose_baseaddr): Modify comments.
4007 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4008 ix86_expand_prologue.
4009 (ix86_expand_prologue): Modify stack realignment and allocation.
4010 (ix86_expand_epilogue): Modify comments.
4011 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4012 avx2_runtime, avx512f, and avx512f_runtime.
4014 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4016 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4017 (mstackrealign): Do not init to -1.
4018 * config/i386/i386.c (ix86_option_override_internal):
4019 Check opts_set, not opts when setting default value of
4020 opts->x_ix86_force_align_arg_pointer.
4022 2017-08-23 Richard Biener <rguenther@suse.de>
4024 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4025 lang_hooks.decl_printable_name.
4026 * print-rtl-function.c (print_rtx_function): Likewise.
4027 * tree-pretty-print.c (dump_function_header): Likewise.
4029 2017-08-23 Richard Biener <rguenther@suse.de>
4032 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4033 -g0 at compile-time.
4035 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4038 * doc/sourcebuild.texi (Other hardware attributes):
4041 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4044 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4045 Fix single-use check.
4047 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4049 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4051 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4053 * doc/install.texi: Modify to add more details on running selected
4056 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4058 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4059 is combined with -mabi=ms.
4060 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4063 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4065 PR tree-optimization/81488
4066 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4067 and cached_basis fields.
4068 (MAX_SPREAD): New constant.
4069 (alloc_cand_and_find_basis): Initialize new fields.
4070 (clear_visited): New function.
4071 (create_phi_basis_1): Rename from create_phi_basis, set visited
4072 and cached_basis fields.
4073 (create_phi_basis): New wrapper function.
4074 (phi_add_costs_1): Rename from phi_add_costs, add spread
4075 parameter, set visited field, short-circuit when limits reached.
4076 (phi_add_costs): New wrapper function.
4077 (record_phi_increments_1): Rename from record_phi_increments, set
4079 (record_phi_increments): New wrapper function.
4080 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4081 (phi_incr_cost): New wrapper function.
4082 (all_phi_incrs_profitable_1): Rename from
4083 all_phi_incrs_profitable, set visited field.
4084 (all_phi_incrs_profitable): New wrapper function.
4086 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4087 Alan Hayward <alan.hayward@arm.com>
4088 David Sherwood <david.sherwood@arm.com>
4090 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4091 that takes the outer and inner modes.
4092 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4093 comparison as the canonical test for a paradoxical subreg.
4094 * combine.c (simplify_set): Use paradoxical_subreg_p.
4095 (make_extraction): Likewise.
4096 (force_to_mode): Likewise.
4097 (rtx_equal_for_field_assignment_p): Likewise.
4098 (gen_lowpart_for_combine): Likewise.
4099 (simplify_comparison): Likewise.
4100 * cse.c (equiv_constant): Likewise.
4101 * expmed.c (store_bit_field_1): Likewise.
4102 * final.c (alter_subreg): Likewise.
4103 * fwprop.c (propagate_rtx): Likewise.
4104 (forward_propagate_subreg): Likewise.
4105 * ira-conflicts.c (ira_build_conflicts): Likewise.
4106 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4107 * lra-constraints.c (curr_insn_transform): Likewise.
4108 (split_reg): Likewise.
4109 * lra-eliminations.c (move_plus_up): Likewise.
4110 (lra_eliminate_regs_1): Likewise.
4111 * recog.c (general_operand): Likewise.
4112 * ree.c (combine_reaching_defs): Likewise.
4113 * reload.c (push_reload): Likewise.
4114 (find_reloads): Likewise.
4115 * reload1.c (elimination_effects): Likewise.
4116 (compute_reload_subreg_offset): Likewise.
4117 (choose_reload_regs): Likewise.
4118 * rtlanal.c (subreg_lsb_1): Likewise.
4119 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4120 (simplify_subreg): Likewise.
4121 * var-tracking.c (track_loc_p): Likewise.
4122 * emit-rtl.c (byte_lowpart_offset): Likewise.
4123 (paradoxical_subreg_p): Delete out-of-line definition.
4125 2017-08-22 Jeff Law <law@redhat.com>
4127 PR tree-optimization/81741
4128 PR tree-optimization/71947
4129 * tree-ssa-dom.c: Include tree-inline.h.
4130 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4131 equivalences if one is more expensive to compute than the other.
4132 * tree-ssa-scopedtables.h (class const_or_copies): Make
4133 record_const_or_copy_raw method private.
4134 (class avail_exprs_stack): New method simplify_binary_operation.
4135 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4136 avail_exprs_stack::simplify_binary_operation as needed.
4137 (avail_exprs_stack::simplify_binary_operation): New function.
4139 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4141 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4142 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4143 (DOT_SYMBOLS): Likewise.
4144 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4145 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4146 (RS6000_ABI_NAME): Likewise.
4147 (TARGET_CMODEL): Likewise.
4148 (TOC_SECTION_ASM_OP): Likewise.
4149 (SET_CMODEL): New macro.
4150 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4152 2017-08-22 Richard Biener <rguenther@suse.de>
4154 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4155 to allow for free-lang-data replacements similar to verify_type_variant.
4157 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4158 Alan Hayward <alan.hayward@arm.com>
4159 David Sherwood <david.sherwood@arm.com>
4161 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4162 VOIDmode for the LABEL_REF.
4164 2017-08-22 Richard Biener <rguenther@suse.de>
4166 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4168 2017-08-22 Richard Biener <rguenther@suse.de>
4170 * common.opt (feliminate-dwarf2-dups): Ignore.
4171 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4172 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4173 same_die_p_wrap, compute_section_prefix,
4174 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4175 (comdat_symbol_id, comdat_symbol_number): Likewise.
4176 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4178 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4179 (output_die): Mark unreachable path unreachable.
4180 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4181 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4182 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4183 (dwarf2out_early_finish): Likewise.
4185 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4187 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4189 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4192 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4193 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4194 (avr_attribute_table) <io, io_low, address>: Initialize
4195 .decl_required with true.
4197 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4199 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4200 undocumented debugging options.
4201 (-mvsx-scalar-double): Likewise.
4202 (-mallow-df-permute): Likewise.
4203 (-mvectorize-builtins): Likewise.
4204 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4205 (rs6000_builtin_vectorized_function): Likewise.
4206 (rs6000_builtin_md_vectorized_function): Likewise.
4207 (rs6000_opt_vars): Likewise.
4209 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4212 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4213 (*btrq_imm): Rename from *btrq.
4214 (*btcq_imm): Rename from *btcq.
4215 (btsc): New code attribute.
4216 (*<btsc><mode>): New insn pattern.
4217 (*btr<mode>): Ditto.
4218 (*<btsc><mode>_mask): New insn_and_split pattern.
4219 (*btr<mode>_mask): Ditto.
4221 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4222 Alan Hayward <alan.hayward@arm.com>
4223 David Sherwood <david.sherwood@arm.com>
4225 * function.c (pad_below): Simplify padding calculation.
4227 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4228 Alan Hayward <alan.hayward@arm.com>
4229 David Sherwood <david.sherwood@arm.com>
4231 * target.def (function_prologue): Remove frame size argument.
4232 (function_epilogue): Likewise.
4233 * doc/tm.texi: Regenerate.
4234 * final.c (final_start_function): Update call to function_prologue.
4235 (final_end_function): Update call to function_epilogue.
4236 (default_function_pro_epilogue): Remove frame size argument.
4237 * output.h (default_function_pro_epilogue): Likewise.
4238 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4239 (arm_output_function_prologue): Likewise.
4240 * config/frv/frv.c (frv_function_prologue): Likewise.
4241 (frv_function_epilogue): Likewise.
4242 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4243 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4244 (ia64_output_function_epilogue): Likewise.
4245 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4246 (m32r_output_function_epilogue): Likewise.
4247 * config/microblaze/microblaze.c (microblaze_function_prologue)
4248 (microblaze_function_epilogue): Likewise.
4249 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4250 (mips_output_function_epilogue): Likewise.
4251 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4252 (mmix_target_asm_function_epilogue): Likewise.
4253 * config/msp430/msp430.c (msp430_start_function): Likewise.
4254 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4255 (nds32_asm_function_epilogue): Likewise.
4256 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4257 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4258 (pa_output_function_epilogue): Likewise.
4259 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4260 (rs6000_output_function_epilogue): Likewise.
4261 * config/rl78/rl78.c (rl78_start_function): Likewise.
4262 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4263 (rs6000_output_function_epilogue): Likewise.
4264 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4265 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4266 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4267 (sparc_asm_function_epilogue): Likewise.
4269 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4271 * tree.h (type_has_mode_precision_p): New function.
4272 * convert.c (convert_to_integer_1): Use it.
4273 * expr.c (expand_expr_real_2): Likewise.
4274 (expand_expr_real_1): Likewise.
4275 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4276 * match.pd: Likewise.
4277 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4278 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4279 * tree-tailcall.c (process_assignment): Likewise.
4280 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4281 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4282 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4283 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4284 (vectorizable_assignment): Likewise.
4285 (vectorizable_shift): Likewise.
4286 (vectorizable_operation): Likewise.
4287 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4289 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4291 * match.pd: Add pow (C, x) simplification.
4293 2017-08-21 Richard Biener <rguenther@suse.de>
4295 PR tree-optimization/81900
4296 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4297 for blocks with abnormal predecessors.
4298 (compute_antic): Do not set visited flag prematurely.
4300 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4303 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4305 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4307 * stor-layout.h (vector_type_mode): Move to...
4308 * tree.h (vector_type_mode): ...here.
4309 * stor-layout.c (vector_type_mode): Move to...
4310 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4312 2017-08-21 Richard Biener <rguenther@suse.de>
4314 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4315 register_external_die hooks.
4316 (debug_false_tree_charstarstar_uhwistar): Declare.
4317 (debug_nothing_tree_charstar_uhwi): Likewise.
4318 * debug.c (do_nothing_debug_hooks): Adjust.
4319 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4320 (debug_nothing_tree_charstar_uhwi): Likewise.
4321 * dbxout.c (dbx_debug_hooks): Adjust.
4322 (xcoff_debug_hooks): Likewise.
4323 * sdbout.c (sdb_debug_hooks): Likewise.
4324 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4325 * dwarf2out.c (macinfo_label_base): New global.
4326 (dwarf2out_register_external_die): New function for the
4327 register_external_die hook.
4328 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4329 (dwarf2_debug_hooks): Use them.
4330 (dwarf2_lineno_debug_hooks): Adjust.
4331 (struct die_struct): Add with_offset flag.
4332 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4333 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4334 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4335 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4336 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4337 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4338 defining section names for the early LTO debug variants.
4339 (reset_indirect_string): New helper.
4340 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4341 (print_dw_val): Add support for offsetted symbol references.
4342 (get_ultimate_context): Split out from is_cxx.
4343 (is_cxx): Use get_ultimate_context.
4344 (is_fortran): Add decl overload.
4345 (compute_comp_unit_symbol): Split out worker from
4346 compute_section_prefix.
4347 (compute_section_prefix): Call compute_comp_unit_symbol and
4348 set comdat_type_p here.
4349 (output_die): Skip DIE symbol output for the LTO added one.
4350 Handle DIE symbol references with offset.
4351 (output_comp_unit): Guard section name mangling properly.
4352 For LTO debug sections emit a symbol at the section beginning
4353 which we use to refer to its DIEs.
4354 (add_abstract_origin_attribute): For DIEs registered via
4355 dwarf2out_register_external_die directly refer to the early
4356 DIE rather than indirectly through the shadow one we created.
4357 Remove obsolete call to dwarf2out_abstract_function for
4358 non-function/block origins.
4359 (gen_array_type_die): When generating early LTO debug do
4360 not emit DW_AT_string_length.
4361 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4362 late when in LTO. As suggested place a gcc_unreachable for
4363 the DECL_ABSTRACT_P case.
4364 (gen_subprogram_die): Avoid another specification DIE
4365 for early built declarations/definitions for the late LTO case.
4366 (gen_variable_die): Add type references for late duplicated VLA dies
4368 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4369 we have the abstract instance already.
4370 (process_scope_var): Adjust decl DIE contexts in LTO which
4371 first puts them in limbo.
4372 (gen_decl_die): Do not generate type DIEs late apart from
4373 types for VLAs or for decls we do not yet have a DIE. Do not
4374 call dwarf2out_abstract_function late.
4375 (dwarf2out_early_global_decl): Make sure to create DIEs
4376 for abstract instances of a decl first.
4377 (dwarf2out_late_global_decl): Adjust comment.
4378 (output_macinfo_op): With multiple macro sections use
4379 macinfo_label_base to distinguish labels.
4380 (output_macinfo): Likewise. Update macinfo_label_base.
4381 Pass in the line info label.
4382 (note_variable_value_in_expr): When generating LTO resolve
4383 all variable values here by generating DIEs as needed.
4384 (init_sections_and_labels): Add early LTO debug flag parameter
4385 and generate different sections and names if set. Add generation
4386 counter for the labels so we can have multiple of them.
4387 (reset_dies): Helper to allow DIEs to be output multiple times.
4388 (dwarf2out_finish): When outputting DIEs to the fat part of an
4389 LTO object first reset DIEs.
4390 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4391 (modified_type_die): Check for decl_ultimate_origin being self
4393 (gen_type_die_with_usage): Likewise.
4394 (gen_typedef_die): Allow decl_ultimate_origin being self.
4395 (set_decl_abstract_flags): Remove.
4396 (set_block_abstract_flags): Likewise.
4397 (dwarf2out_abstract_function): Treat the early generated DIEs
4398 as the abstract copy and only add DW_AT_inline and
4399 DW_AT_artificial here and call set_decl_origin_self.
4400 If the DIE has an abstract origin don't do anything.
4401 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4402 if we have none yet (Go fails to build one, PR78628).
4403 (variably_modified_type_p): Prevent endless recursion for Ada
4404 cyclic pointer types.
4405 * lto-streamer-in.c: Include debug.h.
4406 (dref_queue): New global.
4407 (lto_read_tree_1): Stream in DIE references.
4408 (lto_input_tree): Register DIE references.
4409 (input_function): Stream DECL_DEBUG_ARGS.
4410 * lto-streamer-out.c: Include debug.h.
4411 (lto_write_tree_1): Output DIE references.
4412 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4413 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4414 (output_function): Stream DECL_DEBUG_ARGS.
4415 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4416 Stream DECL_ABSTRACT_ORIGIN.
4417 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4418 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4419 DECL_CONTEXT for file-scope decls.
4420 * lto-streamer.h (struct dref_entry): Declare.
4421 (dref_queue): Likewise.
4422 * cfgexpand.c (pass_expand::execute): Do not call the
4423 outlining_inline_function hook here.
4424 * lto-wrapper.c (debug_obj): New global.
4425 (tool_cleanup): Unlink it if required.
4426 (debug_objcopy): New function.
4427 (run_gcc): Handle early debug sections in the IL files by
4428 extracting them to separate files, partially linkin them and
4429 feeding the result back as result to the linker.
4430 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4431 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4432 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4433 sections into a separate segment.
4434 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4436 (darwin_asm_dwarf_section): Likewise.
4437 (darwin_asm_output_dwarf_offset): Likewise.
4438 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4440 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4441 Alan Hayward <alan.hayward@arm.com>
4442 David Sherwood <david.sherwood@arm.com>
4444 * read-md.h (md_reader::record_potential_iterator_use): Replace
4445 pointer argument with an rtx and an index.
4446 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4447 (apply_mode_iterator): Likewise.
4448 (apply_code_iterator): Likewise.
4449 (apply_int_iterator): Likewise.
4450 (apply_subst_iterator): Likewise.
4451 (record_iterator_use): Likewise.
4452 (record_attribute_use): Likewise.
4453 (md_reader::record_potential_iterator_use): Likewise. Update calls
4454 to record_iterator_use and apply_iterator.
4455 (iterator_use): Replace ptr with x and index.
4456 (attribute_use): Likewise.
4457 (apply_attribute_uses): Update calls to apply_iterator.
4458 (apply_iterators): Likewise. Update initialization of iterator_use.
4459 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4460 and record_potential_iterator_use.
4461 (rtx_reader::read_rtx_operand): Likewise.
4463 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4464 Alan Hayward <alan.hayward@arm.com>
4465 David Sherwood <david.sherwood@arm.com>
4467 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4470 2017-08-21 Richard Biener <rguenther@suse.de>
4473 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4474 at struct end conservatively when comparing common bases.
4476 2017-08-21 Richard Biener <rguenther@suse.de>
4478 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4479 (mem_ref_in_stmt): Remove.
4480 (determine_max_movement): Use ref index to get at the reference.
4481 (invariantness_dom_walker::before_dom_children): Deal with
4482 lim data already initialized.
4483 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4485 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4487 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4488 (TARGET_ISA_ROUND): Ditto.
4489 (TARGET_ROUND): Ditto.
4490 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4491 * config/i386/i386.md: Ditto.
4492 * config/i386/sse.md: Ditto.
4493 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4494 with OPTION_MASK_ISA_SSE4_1.
4496 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4499 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4500 __builtin_ia32_lzcnt_u16.
4502 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4505 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4506 (rs6000_set_current_function): Rewrite function to use it.
4508 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4511 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4512 and TYPE_WARN_IF_NOT_ALIGN.
4513 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4514 (handle_warn_if_not_align): New.
4515 (place_union_field): Call handle_warn_if_not_align.
4516 (place_field): Call handle_warn_if_not_align.
4517 Copy TYPE_WARN_IF_NOT_ALIGN.
4518 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4519 (layout_type): Likewise.
4520 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4522 (tree_decl_common): Add warn_if_not_align.
4523 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4524 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4525 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4526 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4527 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4528 * doc/extend.texi: Document warn_if_not_aligned attribute.
4529 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4531 2017-08-17 Martin Liska <mliska@suse.cz>
4534 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4535 (get_data_dependence): Use it as pointer type.
4536 (distribute_loop): Likewise.
4538 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4540 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4541 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4542 (p8_vmrgow_<mode>_direct): New define_insn.
4543 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4544 handle endianness for vmrgew and vmrgow permute patterns.
4546 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4548 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4549 * config/rs6000/rs6000-cpus.def: Remove comment.
4550 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4551 (POWERPC_MASKS): Likewise.
4552 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4553 use of TARGET_VSX_TIMODE.
4554 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4555 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4556 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4557 (rs6000_option_override_internal): Remove dead code.
4558 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4559 (rs6000_legitimize_reload_address): Likewise.
4560 (rs6000_legitimate_address_p): Likewise.
4561 (rs6000_opt_masks): Delete "vsx-timode".
4562 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4563 from function comment.
4564 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4565 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4566 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4568 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4569 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4570 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4573 (define_peephole2): Likewise.
4575 2017-08-17 Martin Sebor <msebor@redhat.com>
4578 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4579 past the end of an array.
4580 (test_pp_format): Add test cases.
4582 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4584 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4585 missing ECF_NOTHROW flags.
4587 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4590 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4591 operands residing in integer registers.
4592 (*vsx_le_perm_load_<mode>): Likewise.
4593 (*vsx_le_perm_store_<mode>): Likewise.
4594 (define_peephole2): Add peepholes to optimize the above.
4596 2017-08-17 Marek Polacek <polacek@redhat.com>
4599 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4600 to mimic what shorten_compare did. Change the return type to bool.
4601 (fold_cond_expr_with_comparison): Update call to
4602 operand_equal_for_comparison_p.
4603 (fold_ternary_loc): Likewise.
4605 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4607 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4609 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4611 2017-08-17 Richard Biener <rguenther@suse.de>
4613 * tree-ssa-structalias.c (solve_graph): When propagating
4614 to successors update the graphs succ edges and avoid duplicate work.
4616 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4619 * config/i386/i386.c (ix86_option_override_internal): Save target
4620 specific options after ix86_stack_protector_guard_reg was changed.
4622 2017-08-17 Richard Biener <rguenther@suse.de>
4624 PR tree-optimization/81827
4625 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4626 (new_var_info): Initialize it conservatively.
4627 (get_call_vi): Mark register vars.
4628 (new_scalar_tmp_constraint_exp): Likewise.
4629 (handle_rhs_call): Likewise.
4630 (handle_const_call): Likewise.
4631 (create_function_info_for): Likewise.
4632 (solve_constraints): Sort varinfos to separate register from
4633 non-register vars to pack points-to solution bitmaps during
4636 2017-08-17 Marek Polacek <polacek@redhat.com>
4638 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4640 2017-08-17 Richard Biener <rguenther@suse.de>
4642 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4643 to true when overflow is undefined and we saturated the result.
4645 2017-08-17 Alan Modra <amodra@gmail.com>
4648 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4649 Don't use store multiple if only one reg needs saving.
4650 (interesting_frame_related_regno): New function.
4651 (rs6000_frame_related): Don't emit frame info for regs that
4653 (rs6000_emit_epilogue): Likewise.
4655 2017-08-16 Nathan Sidwell <nathan@acm.org>
4657 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4658 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4659 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4660 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4661 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4662 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4663 (verify_type): Adjust for TYPE_BINFO move.
4664 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4665 process TYPE_BINFO directly.
4666 (hash_tree): Likewise.
4667 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4669 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4672 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4674 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4676 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4679 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4680 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4681 Add ix86_binary_operator_ok to insn constraint.
4682 (*iordi_1_bts): Ditto.
4683 (*xordi_1_btc): Ditto.
4684 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4685 Update corresponding peephole2 pattern.
4689 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4691 PR tree-optimization/81832
4692 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4693 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4695 2017-08-16 Marek Polacek <polacek@redhat.com>
4698 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4699 regarding size_zero_node.
4701 2017-08-16 Martin Liska <mliska@suse.cz>
4704 * config.gcc: Respect previously set extra_objs in case
4707 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4709 PR tree-optimization/81835
4710 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4711 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4712 not depend on the phi.
4714 2017-08-16 Alan Modra <amodra@gmail.com>
4716 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4719 2017-08-16 Alan Modra <amodra@gmail.com>
4721 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4722 (save_reg_p): ..into this. Update all callers.
4723 (first_reg_to_save): Simplify.
4725 2017-08-16 Alan Modra <amodra@gmail.com>
4727 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4730 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4734 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4735 constant count if that count is less than 32.
4737 2017-08-15 Nathan Sidwell <nathan@acm.org>
4739 * gcc.c (execute): Emit friendlier message if inferior is killed
4740 by an external cause.
4742 2017-08-15 Richard Biener <rguenther@suse.de>
4744 PR tree-optimization/81790
4745 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4746 CONSTRUCTORs from simplifying and VN.
4748 2017-08-14 Martin Sebor <msebor@redhat.com>
4750 * builtin-attrs.def: Add comments.
4752 2017-08-14 Martin Sebor <msebor@redhat.com>
4755 * doc/extend.texi (attribute nonstring): Document new attribute.
4757 2017-08-14 Martin Sebor <msebor@redhat.com>
4760 * tree-diagnostic.c (default_tree_printer): Handle %G.
4761 * gimple-pretty-print.h (percent_G_format): Declare new function.
4762 * gimple-pretty-print.c (percent_G_format): Define.
4763 * tree-pretty-print.c (percent_K_format): Add argument.
4765 2017-08-14 Martin Sebor <msebor@redhat.com>
4767 PR translation/79998
4768 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4769 Remove a stray space.
4771 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4774 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4775 (*iordi_1_bts): Ditto.
4776 (*xordi_1_btc): Ditto.
4778 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4781 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4782 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4784 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4785 quoted strings, and make more translator-friendly.
4786 (darwin_rs6000_override_options): Likewise.
4787 (rs6000_option_override_internal): Likewise.
4788 (rs6000_return_in_memory): Fix overlong line.
4789 (init_cmulative_args): Use quoted strings, and make more
4790 translator-friendly.
4791 (rs6000_pass_by_reference): Fix overlong line.
4792 (def_builtin): Use quoted strings.
4793 (altivec_expand_predicate_builtin): Use quoted strings, and make
4794 more translator-friendly.
4795 (htm_expand_builtin): Use quoted strings.
4796 (cpu_expand_builtin): Use quoted strings, and make more
4797 translator-friendly.
4798 (altivec_expand_builtin): Likewise.
4799 (paired_expand_predicate_builtin): Likewise.
4800 (rs6000_invalid_builtin): Likewise.
4801 (builtin_function_type): Use quoted strings.
4802 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4803 more translator-friendly.
4804 (rs6000_trampoline_init): Likewise.
4805 (rs6000_handle_altivec_attribute): Likewise.
4806 (rs6000_inner_target_options): Use quoted strings.
4807 (rs6000_disable_incompatible_switches): Likewise.
4808 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4809 strings, and make more translator-friendly.
4810 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4812 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4814 PR tree-optimization/81799
4815 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4816 cond_expr to simple gimple operand.
4818 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4821 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4823 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4827 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4828 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4829 TARGET_GASISR_PROLOGUES.
4830 * config/avr/avr.c (avr_option_override): Same.
4831 (avr_pass_pre_proep::execute): Same.
4833 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4836 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4837 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4839 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4841 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4842 define_insn_and_split. Split to a memory load from 0 in
4843 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4844 using PTR mode iterator.
4845 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4846 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4847 (*add_tp_<mode>): Redefine as define_insn_and_split.
4848 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4849 address space. Merge with *add_tp_x32 using PTR mode iterator.
4850 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4851 Split to an add with a memory load from 0 in
4852 DEFAULT_TLS_SEG_REG address space.
4854 2017-08-12 Andrew Pinski <apinski@cavium.com>
4856 * config/aarch64/aarch64-option-extensions.def (rdma):
4857 Fix feature string to what Linux prints out in /proc/cpuinfo.
4859 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4862 * dwarf2out.c (modified_type_die): For C typedef types that have
4863 an ultimate origin, process the ultimate origin instead of the
4865 (gen_typedef_die): Assert that input DECLs have no ultimate
4867 (gen_type_die_with_usage): For typedef variants that have an
4868 ultimate origin, just call gen_decl_die on the original DECL.
4869 (process_scope_var): Avoid creating DIEs for local typedefs and
4870 concrete static variables.
4872 2017-08-12 Alan Modra <amodra@gmail.com>
4876 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4877 match gnu-user.h startfile.
4878 (ENDFILE_LINUX_SPEC): Similarly.
4880 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4883 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4885 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4887 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4888 * config/aarch64/aarch64.md (mov<mode>): Change.
4889 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4890 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4891 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4893 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4895 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4896 for storage order barriers.
4898 2017-08-11 Martin Liska <mliska@suse.cz>
4901 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4902 variables of void type.
4904 2017-08-11 Martin Liska <mliska@suse.cz>
4906 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4907 TARGET_SUPPORTS_ALIASES.
4908 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4909 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4910 (optimize_weakref): Likewise.
4911 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4912 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4913 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4915 2017-08-11 Martin Liska <mliska@suse.cz>
4918 * config/i386/i386.c (make_resolver_func): Do complete
4919 refactoring of the function.
4921 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4924 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4925 * config/i386/i386.c (ix86_stack_protect_guard): Use
4926 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4927 * doc/invoke.texi (x86 Options): Document
4928 -mstack-protector-guard-symbol= option.
4930 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4932 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4933 * config/i386/i386.c (ix86_split_stack_guard): New function.
4934 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4935 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4936 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4937 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4938 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4939 (split_stack_space_check): Call ix86_split_stack_guard.
4941 2017-08-10 Martin Sebor <msebor@redhat.com>
4943 * print-tree.c (print_node): Print location using the established
4945 Replace spaces with colons.
4946 (debug_raw, debug): Ditto.
4948 2017-08-10 Martin Sebor <msebor@redhat.com>
4951 * pretty-print.c (pp_format): Correct the handling of %s precision.
4953 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4956 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4958 (ix86_finalize_stack_frame_flags): This. Also clear
4959 frame_pointer_needed if -fno-omit-frame-pointer is used without
4961 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4962 with ix86_finalize_stack_frame_flags.
4963 (ix86_expand_epilogue): Likewise.
4964 (ix86_expand_split_stack_prologue): Likewise.
4965 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4967 2017-08-10 Martin Liska <mliska@suse.cz>
4970 * c-attribs.c (handle_target_attribute):
4971 Report warning for an empty string argument of target attribute.
4973 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4976 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4978 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4979 or DECL_NONLOCAL labels.
4980 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4981 or DECL_NONLOCAL labels here.
4983 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4985 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4986 to indicate when early gimple folding has been disabled.
4987 (rs6000_gimple_fold_builtin): Add debug content.
4988 (rs6000_invalid_builtin): Fix whitespace.
4989 (rs6000_expand_builtin): Fix whitespace.
4990 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4992 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4995 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4996 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4998 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5000 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5001 (qdf24xx): Likewise.
5002 * config/aarch64/aarch64.md: Include falkor.md.
5003 * config/aarch64/falkor.md: New.
5005 2017-08-09 Marek Polacek <polacek@redhat.com>
5008 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5009 * diagnostic.c (emit_diagnostic): Add a comment.
5010 (emit_diagnostic_valist): New function.
5012 2017-08-09 Marek Polacek <polacek@redhat.com>
5015 * input.c (make_location): New overload.
5016 * input.h (make_location): Declare.
5018 2017-08-08 Alan Modra <amodra@gmail.com>
5019 H.J. Lu <hongjiu.lu@intel.com>
5022 * gcc.c (NO_PIE_SPEC): Delete.
5023 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5025 (LINK_PIE_SPEC): ..to here.
5026 (LINK_COMMAND_SPEC): Support -no-pie.
5027 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5028 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5030 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5031 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5032 (ENDFILE_CRTEND_SPEC): Similarly.
5034 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5037 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5038 (mstack-protector-guard-offset=): Ditto.
5039 * config/i386/i386.c (ix86_option_override): Handle
5040 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5042 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5043 ix86_stack_protect_guard_offset variables.
5044 (TARGET_STACK_PROTECT_GUARD): Always define.
5045 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5046 and -mstack-protector-guard-offset= options.
5048 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5050 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5051 boundary case for the last candidate.
5053 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5055 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5056 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5058 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5061 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5062 * config/aarch64/aarch64-builtins.c
5063 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5064 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5065 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5067 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5068 Andrew Pinski <pinskia@gmail.com>
5071 * internal-fn.def (XORSIGN): New.
5072 * optabs.def (xorsign_optab): New.
5073 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5074 (convert_expand_mult_copysign): New.
5075 (pass_optimize_widening_mul::execute): Call
5076 convert_expand_mult_copysign.
5078 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5080 PR tree-optimization/81354
5081 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5082 Insert on edges rather than explicitly creating landing pads.
5083 (analyze_candidates_and_replace): Commit edge inserts.
5085 2017-08-08 Richard Biener <rguenther@suse.de>
5088 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5089 (expand_simple_operations): Also look through ADDR_EXPRs with
5090 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5092 2017-08-08 Richard Biener <rguenther@suse.de>
5094 PR tree-optimization/81723
5095 * tree-vect-slp.c (struct bst_traits): New hash traits.
5096 (bst_fail): New global.
5097 (vect_build_slp_tree_2): New worker, split out from ...
5098 (vect_build_slp_tree): ... this now wrapping it with using
5099 bst_fail set to cache SLP tree build fails. Properly handle
5101 (vect_analyze_slp_instance): Allocate and free bst_fail.
5103 2017-08-08 Martin Liska <mliska@suse.cz>
5106 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5107 LABEL_DECLs that can be from a different function.
5109 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5111 PR tree-optimization/81744
5112 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5113 loop's number of iterations.
5115 2017-08-08 Martin Liska <mliska@suse.cz>
5117 * asan.c: Include header files.
5118 * attribs.c (build_decl_attribute_variant): New function moved
5120 (build_type_attribute_qual_variant): Likewise.
5121 (cmp_attrib_identifiers): Likewise.
5122 (simple_cst_list_equal): Likewise.
5123 (omp_declare_simd_clauses_equal): Likewise.
5124 (attribute_value_equal): Likewise.
5125 (comp_type_attributes): Likewise.
5126 (build_type_attribute_variant): Likewise.
5127 (lookup_ident_attribute): Likewise.
5128 (remove_attribute): Likewise.
5129 (merge_attributes): Likewise.
5130 (merge_type_attributes): Likewise.
5131 (merge_decl_attributes): Likewise.
5132 (merge_dllimport_decl_attributes): Likewise.
5133 (handle_dll_attribute): Likewise.
5134 (attribute_list_equal): Likewise.
5135 (attribute_list_contained): Likewise.
5136 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5137 (lookup_attribute_by_prefix): Likewise.
5138 * bb-reorder.c: Include header files.
5139 * builtins.c: Likewise.
5140 * calls.c: Likewise.
5141 * cfgexpand.c: Likewise.
5142 * cgraph.c: Likewise.
5143 * cgraphunit.c: Likewise.
5144 * convert.c: Likewise.
5145 * dwarf2out.c: Likewise.
5146 * final.c: Likewise.
5147 * fold-const.c: Likewise.
5148 * function.c: Likewise.
5149 * gimple-expr.c: Likewise.
5150 * gimple-fold.c: Likewise.
5151 * gimple-pretty-print.c: Likewise.
5152 * gimple.c: Likewise.
5153 * gimplify.c: Likewise.
5154 * hsa-common.c: Likewise.
5155 * hsa-gen.c: Likewise.
5156 * internal-fn.c: Likewise.
5157 * ipa-chkp.c: Likewise.
5158 * ipa-cp.c: Likewise.
5159 * ipa-devirt.c: Likewise.
5160 * ipa-fnsummary.c: Likewise.
5161 * ipa-inline.c: Likewise.
5162 * ipa-visibility.c: Likewise.
5164 * lto-cgraph.c: Likewise.
5165 * omp-expand.c: Likewise.
5166 * omp-general.c: Likewise.
5167 * omp-low.c: Likewise.
5168 * omp-offload.c: Likewise.
5169 * omp-simd-clone.c: Likewise.
5170 * opts-global.c: Likewise.
5171 * passes.c: Likewise.
5172 * predict.c: Likewise.
5173 * sancov.c: Likewise.
5174 * sanopt.c: Likewise.
5175 * symtab.c: Likewise.
5176 * toplev.c: Likewise.
5177 * trans-mem.c: Likewise.
5178 * tree-chkp.c: Likewise.
5179 * tree-eh.c: Likewise.
5180 * tree-into-ssa.c: Likewise.
5181 * tree-object-size.c: Likewise.
5182 * tree-parloops.c: Likewise.
5183 * tree-profile.c: Likewise.
5184 * tree-ssa-ccp.c: Likewise.
5185 * tree-ssa-live.c: Likewise.
5186 * tree-ssa-loop.c: Likewise.
5187 * tree-ssa-sccvn.c: Likewise.
5188 * tree-ssa-structalias.c: Likewise.
5189 * tree-ssa.c: Likewise.
5190 * tree-streamer-in.c: Likewise.
5191 * tree-vectorizer.c: Likewise.
5192 * tree-vrp.c: Likewise.
5194 * ubsan.c: Likewise.
5195 * varasm.c: Likewise.
5196 * varpool.c: Likewise.
5197 * tree.c: Remove functions moved to attribs.[ch].
5199 * config/aarch64/aarch64.c: Add attrs.h header file.
5200 * config/alpha/alpha.c: Likewise.
5201 * config/arc/arc.c: Likewise.
5202 * config/arm/arm.c: Likewise.
5203 * config/avr/avr.c: Likewise.
5204 * config/bfin/bfin.c: Likewise.
5205 * config/c6x/c6x.c: Likewise.
5206 * config/cr16/cr16.c: Likewise.
5207 * config/cris/cris.c: Likewise.
5208 * config/darwin.c: Likewise.
5209 * config/epiphany/epiphany.c: Likewise.
5210 * config/fr30/fr30.c: Likewise.
5211 * config/frv/frv.c: Likewise.
5212 * config/ft32/ft32.c: Likewise.
5213 * config/h8300/h8300.c: Likewise.
5214 * config/i386/winnt.c: Likewise.
5215 * config/ia64/ia64.c: Likewise.
5216 * config/iq2000/iq2000.c: Likewise.
5217 * config/lm32/lm32.c: Likewise.
5218 * config/m32c/m32c.c: Likewise.
5219 * config/m32r/m32r.c: Likewise.
5220 * config/m68k/m68k.c: Likewise.
5221 * config/mcore/mcore.c: Likewise.
5222 * config/microblaze/microblaze.c: Likewise.
5223 * config/mips/mips.c: Likewise.
5224 * config/mmix/mmix.c: Likewise.
5225 * config/mn10300/mn10300.c: Likewise.
5226 * config/moxie/moxie.c: Likewise.
5227 * config/msp430/msp430.c: Likewise.
5228 * config/nds32/nds32-isr.c: Likewise.
5229 * config/nds32/nds32.c: Likewise.
5230 * config/nios2/nios2.c: Likewise.
5231 * config/nvptx/nvptx.c: Likewise.
5232 * config/pa/pa.c: Likewise.
5233 * config/pdp11/pdp11.c: Likewise.
5234 * config/powerpcspe/powerpcspe.c: Likewise.
5235 * config/riscv/riscv.c: Likewise.
5236 * config/rl78/rl78.c: Likewise.
5237 * config/rx/rx.c: Likewise.
5238 * config/s390/s390.c: Likewise.
5239 * config/sh/sh.c: Likewise.
5240 * config/sol2.c: Likewise.
5241 * config/sparc/sparc.c: Likewise.
5242 * config/spu/spu.c: Likewise.
5243 * config/stormy16/stormy16.c: Likewise.
5244 * config/tilegx/tilegx.c: Likewise.
5245 * config/tilepro/tilepro.c: Likewise.
5246 * config/v850/v850.c: Likewise.
5247 * config/vax/vax.c: Likewise.
5248 * config/visium/visium.c: Likewise.
5249 * config/xtensa/xtensa.c: Likewise.
5251 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5254 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5255 constraints since the -mupper-regs-* switches have been
5257 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5258 into a vector from a double word element that was extracted from
5259 another vector, and eliminate extra XXPERMDI instructions.
5260 (vsx_concat_<mode>_2): Likewise.
5261 (vsx_concat_<mode>_3): Likewise.
5262 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5263 concat to allow optimizing inserts from previous extracts.
5265 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5267 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5268 memory reference to a SSP offset in TLS address space.
5269 (ix86_print_operand) <case '@'>: Remove.
5270 (ix86_print_operand_punct_valid_p): Remove '@' code.
5271 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5273 (stack_tls_protect_set_<mode>): Remove.
5274 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5275 (stack_tls_protect_test_<mode>): Remove.
5276 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5278 2017-08-07 Olivier Hainque <hainque@adacore.com>
5281 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5283 2017-08-07 Douglas Rupp <rupp@adacore.com>
5285 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5286 variable was referenced as multidir in command.
5288 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5291 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5294 2017-08-07 Martin Liska <mliska@suse.cz>
5296 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5297 * config/rl78/rl78.c: Add include of attribs.h.
5298 * config/sh/sh.c: Likewise.
5299 * config/v850/v850.c: Likewise.
5301 2017-08-07 Tom de Vries <tom@codesourcery.com>
5304 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5306 2017-08-07 Martin Liska <mliska@suse.cz>
5308 * config/mips/mips.c: Include attribs.h.
5310 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5313 * doc/invoke.texi: Document change in behvaior for -Ofast for
5316 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5318 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5320 (aarch64_pop_regs): Likewise.
5321 (aarch64_gen_load_pair): Likewise.
5322 (aarch64_save_callee_saves): Likewise.
5323 (aarch64_restore_callee_saves): Likewise.
5325 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5327 * config/i386/i386.c: Revert the last change.
5329 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5332 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5334 (ix86_finalize_stack_frame_flags): This. Also clear
5335 frame_pointer_needed if -fno-omit-frame-pointer is used without
5337 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5338 with ix86_finalize_stack_frame_flags.
5339 (ix86_expand_epilogue): Likewise.
5340 (ix86_expand_split_stack_prologue): Likewise.
5342 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5345 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5346 to P_AES for Westmere.
5348 2017-08-07 Jonathan Yong <10walls@gmail.com>
5350 * config/i386/mingw.opt (fset-stack-executable): Removed.
5351 * config/i386/cygming.opt (fset-stack-executable): Moved
5353 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5355 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5357 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5359 2017-08-07 Marek Polacek <polacek@redhat.com>
5362 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5364 2017-08-07 Martin Liska <mliska@suse.cz>
5366 * attribs.h (canonicalize_attr_name): New function.
5367 (cmp_attribs): Move from c-format.c and adjusted.
5368 (is_attribute_p): Moved from tree.h.
5369 * tree-inline.c: Add new includes.
5370 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5371 (private_is_attribute_p): Remove.
5372 (private_lookup_attribute): Likewise.
5373 (private_lookup_attribute_by_prefix): Simplify.
5374 (remove_attribute): Use is_attribute_p.
5375 * tree.h: Remove removed declarations.
5377 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5380 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5381 instead of computing it in the function. Formatting fix.
5382 (expand_case): Don't rely on default_edge being the first edge,
5383 clear it if removing it, pass default_edge to
5384 emit_case_dispatch_table.
5385 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5388 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5390 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5391 insn in the function, emit NOP after the insn.
5393 2017-08-06 Tom de Vries <tom@codesourcery.com>
5395 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5398 2017-08-06 Tom de Vries <tom@codesourcery.com>
5400 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5403 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5405 PR tree-optimization/57371
5406 * match.pd: New pattern.
5408 2017-08-04 Marek Polacek <polacek@redhat.com>
5411 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5412 perform the computation in offset_int.
5414 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5416 PR tree-optimization/81136
5417 * tree-vectorizer.h: Include tree-hash-traits.h.
5418 (vec_base_alignments): New typedef.
5419 (vec_info): Add a base_alignments field.
5420 (vect_record_base_alignments): Declare.
5421 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5423 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5424 (create_data_ref): Add an is_conditional_in_stmt argument.
5425 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5426 the is_conditional_in_stmt field.
5427 (data_ref_loc): Add an is_conditional_in_stmt field.
5428 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5429 (find_data_references_in_stmt): Update call to create_data_ref.
5430 (graphite_find_data_references_in_stmt): Likewise.
5431 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5432 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5433 (vect_record_base_alignment): New function.
5434 (vect_record_base_alignments): Likewise.
5435 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5436 for nested statements even if we fail to compute a misalignment.
5437 Use pooled base alignments for unconditional references.
5438 (vect_find_same_alignment_drs): Compare base addresses instead
5440 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5441 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5443 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5445 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5446 Add an explicit name for the enum. Use auto_vec for slp_instances
5448 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5450 (_bb_vec_info): Add a constructor and destructor.
5451 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5452 (destroy_loop_vec_info): Delete.
5453 (vect_destroy_datarefs): Likewise.
5454 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5455 (vec_info::vec_info): New function.
5456 (vec_info::~vec_info): Likewise.
5457 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5458 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5459 destroy_loop_vec_info.
5460 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5461 (_loop_vec_info::_loop_vec_info): ...this.
5462 (destroy_loop_vec_info): Replace with...
5463 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5464 the stmt_vec_infos. Leave handling of vec_info information to its
5465 destructor. Remove explicit vector releases.
5466 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5467 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5468 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5469 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5470 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5471 (destroy_bb_vec_info): Replace with...
5472 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5473 information to its destructor.
5474 (vect_slp_analyze_bb_1): Use new and delete instead of
5475 new_bb_vec_info and destroy_bb_vec_info.
5476 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5479 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5481 * tree-data-ref.h (subscript): Add access_fn field.
5482 (data_dependence_relation): Add could_be_independent_p.
5483 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5484 (same_access_functions): Move to tree-data-ref.c.
5485 * tree-data-ref.c (ref_contains_union_access_p): New function.
5486 (access_fn_component_p): Likewise.
5487 (access_fn_components_comparable_p): Likewise.
5488 (dr_analyze_indices): Add a reference to access_fn_component_p.
5489 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5491 (constant_access_functions): Likewise.
5492 (add_other_self_distances): Likewise.
5493 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5494 (initialize_data_dependence_relation): Use XCNEW and remove
5495 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5496 of access functions that have the same type. Allow the
5497 subsequence to end with different bases in some circumstances.
5498 Record the chosen access functions in SUB_ACCESS_FN.
5499 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5500 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5501 (subscript_dependence_tester_1): Likewise dra and drb.
5502 (build_classic_dist_vector): Update calls accordingly.
5503 (subscript_dependence_tester): Likewise.
5504 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5505 DDR_COULD_BE_INDEPENDENT_P.
5506 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5507 comp_alias_ddrs instead of may_alias_ddrs.
5508 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5510 (vect_analyze_data_ref_dependence): Use it if
5511 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5512 distance vectors if that fails.
5513 (dependence_distance_ge_vf): New function.
5514 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5515 LOOP_VINFO_MAY_ALIAS_DDRS.
5517 2017-08-04 Richard Biener <rguenther@suse.de>
5520 * fold-const.c (fold_binary_loc): Properly restrict
5521 minus_var0 && minus_var1 case when associating undefined overflow
5524 2017-08-04 Tom de Vries <tom@codesourcery.com>
5526 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5528 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5530 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5531 Don't start diagnostic messages with a capital letter.
5532 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5534 (rs6000_invalid_builtin): Likewise.
5535 (rs6000_trampoline_init): Likewise.
5537 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5540 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5541 after setting changeable df flags.
5543 2017-08-03 Richard Biener <rguenther@suse.de>
5545 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5546 up if the use is in USE - X.
5548 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5550 * toplev.c (dumpfile.h): New include.
5551 (internal_error_reentered): New static function. Use it...
5552 (internal_error_function): ...here to handle reentered internal_error.
5554 2017-08-03 Richard Biener <rguenther@suse.de>
5557 * fold-const.c (split_tree): Add minus_var and minus_con
5558 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5559 here but always use minus_*.
5560 (associate_trees): Assert we never associate with MINUS_EXPR
5561 and NULL first operand. Do not recurse for PLUS_EXPR operands
5562 when associating as MINUS_EXPR either.
5563 (fold_binary_loc): Track minus_var and minus_con.
5565 2017-08-03 Tom de Vries <tom@codesourcery.com>
5568 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5569 ACCEL_COMPILER, apply finish_options on
5570 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5572 2017-08-03 Tom de Vries <tom@codesourcery.com>
5575 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5576 function_entry_patch_area_size > 0.
5578 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5581 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5582 instead of 10??LU, perform unit multiplication in wide_int,
5583 don't change alloc_object_size_limit if the limit is larger
5586 PR tree-optimization/81655
5587 PR tree-optimization/81588
5588 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5589 the case when ranges[i].low and high are 1 for unsigned type with
5593 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5594 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5596 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5598 * tree-vrp.h: Add include guard.
5600 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5603 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5604 (ud2): New insn pattern.
5605 * config/i386/i386.c (ix86_expand_epilogue):
5606 For naked functions, generate ud2 instead of trap insn.
5608 2017-08-02 Marek Polacek <polacek@redhat.com>
5611 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5613 2017-08-02 Tom de Vries <tom@codesourcery.com>
5614 Cesar Philippidis <cesar@codesourcery.com>
5616 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5617 Add missing edge probabilities.
5619 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5621 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5624 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5627 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5628 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5629 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5630 prologue_seq sequences - if any.
5632 2017-08-02 Richard Biener <rguenther@suse.de>
5634 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5635 via vectors if supported, integer extracts via punning if supported
5636 or otherwise vector extracts.
5638 2017-08-02 Richard Biener <rguenther@suse.de>
5640 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5642 (bitmap_insert_into_set): ... this.
5644 2017-08-02 Richard Biener <rguenther@suse.de>
5646 PR tree-optimization/81633
5648 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5650 PR tree-optimization/71752
5651 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5653 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5655 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5656 (machine_function::call_ms2sysv_pad_out): Remove field.
5657 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5658 (ix86_compute_frame_layout): Likewise.
5660 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5663 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5664 attribute with interrupt attribute.
5666 2017-08-01 Andrew Pinski <apinski@cavium.com>
5668 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5669 BIT_INSERT_EXPR's operand 1
5670 to see if the types precision matches.
5672 2017-08-01 Martin Liska <mliska@suse.cz>
5675 * builtins.c (expand_builtin_memcpy_args): Remove.
5676 (expand_builtin_memcpy): Call newly added function
5677 expand_builtin_memory_copy_args.
5678 (expand_builtin_memcpy_with_bounds): Likewise.
5679 (expand_builtin_mempcpy): Remove last argument.
5680 (expand_builtin_mempcpy_with_bounds): Likewise.
5681 (expand_builtin_memory_copy_args): New function created from
5682 expand_builtin_mempcpy_args with small modifications.
5683 (expand_builtin_mempcpy_args): Remove.
5684 (expand_builtin_stpcpy): Remove unused argument.
5685 (expand_builtin): Likewise.
5686 (expand_builtin_with_bounds): Likewise.
5688 2017-08-01 Martin Liska <mliska@suse.cz>
5691 Make mempcpy more optimal (PR middle-end/70140).
5693 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5696 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5697 __builtin_vec_cmpne verify both arguments are compatible vectors
5698 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5699 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5700 move computation of aligned to after checking the argument types.
5704 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5705 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5708 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5709 Doug Rupp <rupp@adacore.com>
5710 Olivier Hainque <hainque@adacore.com>
5712 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5713 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5715 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5716 for TARGET_OS_CPP_BUILTIN.
5717 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5718 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5720 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5721 passing required abi options to the assembler for EABI configurations.
5722 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5723 of .text.hot and .text.unlikely sections for kernel modules when
5724 using ARM style exceptions.
5725 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5726 options. Add EXTRA_CC1_SPEC.
5727 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5729 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5731 (ARM_TARGET2_DWARF_FORMAT): Define.
5732 * config/arm/t-vxworks: Adjust multilib control to removal of the
5733 Diab command line options.
5735 2017-08-01 Martin Liska <mliska@suse.cz>
5737 PR gcov-profile/81561
5738 * gcov.c (unblock): Make unblocking safe as we need to preserve
5739 index correspondence of blocks and block_lists.
5741 2017-08-01 Richard Biener <rguenther@suse.de>
5743 PR tree-optimization/81181
5744 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5745 (compute_antic): ... end of iteration here.
5747 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5749 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5750 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5751 (ftree-slp-vectorize): Likewise.
5752 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5753 can no longer be set independent of flag_tree_loop_vectorize.
5754 * omp-general.c (emp_max_vf): Likewise.
5755 * opts.c (enable_fdo_optimizations): Remove references to
5756 flag_tree_vectorize, these are now implicit.
5757 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5758 and leave it for the options machinery.
5760 2017-08-01 Martin Liska <mliska@suse.cz>
5763 * builtins.c (expand_builtin_memcpy_args): Remove.
5764 (expand_builtin_memcpy): Call newly added function
5765 expand_builtin_memory_copy_args.
5766 (expand_builtin_memcpy_with_bounds): Likewise.
5767 (expand_builtin_mempcpy): Remove last argument.
5768 (expand_builtin_mempcpy_with_bounds): Likewise.
5769 (expand_builtin_memory_copy_args): New function created from
5770 expand_builtin_mempcpy_args with small modifications.
5771 (expand_builtin_mempcpy_args): Remove.
5772 (expand_builtin_stpcpy): Remove unused argument.
5773 (expand_builtin): Likewise.
5774 (expand_builtin_with_bounds): Likewise.
5776 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5779 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5780 print "ds:" only for immediates in generic address space.
5782 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5785 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5786 (ix86_function_ok_for_sibcall): Return false for naked functions.
5788 2017-08-01 Richard Biener <rguenther@suse.de>
5790 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5791 (compute_antic): Seed worklist with exit block predecessors.
5792 * cfganal.c (dfs_find_deadend): For a cycle return the source
5793 of the edge closing it.
5795 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5797 * config/aarch64/aarch64.c
5798 (aarch64_can_const_movi_rtx_p): Move 0 check.
5800 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5802 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5803 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5805 * match.pd: Ditto in address comparison pattern.
5807 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5809 PR tree-optimization/81627
5810 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5811 closed ssa form for store-store chain.
5813 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5815 PR tree-optimization/81620
5816 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5817 for store-store chain.
5819 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5821 PR tree-optimization/81588
5822 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5823 ranges[i].in_p, invert comparison code ccode. For >/>=,
5824 swap rhs1 and rhs2 and comparison code unconditionally,
5825 for </<= don't do that. Don't swap rhs1/rhs2 again if
5826 ranges[i].in_p, instead invert comparison code ccode if
5827 opcode or oe->rank is BIT_IOR_EXPR.
5830 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5831 a direct optab to conversion optab.
5832 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5833 with GET_MODE_INNER as last argument instead of optab_handler.
5834 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5835 vector extraction if possible and optab is available.
5836 * expr.c (store_constructor): Use convert_optab_handler instead
5837 of optab_handler. Use vector initialization from smaller
5838 vectors if possible and optab is available.
5839 * tree-vect-stmts.c (vectorizable_load): Likewise.
5840 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5842 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5843 of vec_init from half-sized vectors with the same element mode.
5844 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5845 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5846 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5847 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5848 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5849 after mode in gen_vec_extract* calls.
5850 (vec_extract<mode>): Renamed to ...
5851 (vec_extract<mode><ssescalarmodelower>): ... this.
5852 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5853 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5854 element mode after mode in gen_vec_init* calls.
5855 (VEC_INIT_HALF_MODE): New mode iterator.
5856 (vec_init<mode>): Renamed to ...
5857 (vec_init<mode><ssescalarmodelower>): ... this.
5858 (vec_init<mode><ssehalfvecmodelower>): New expander.
5859 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5860 (vec_extractv2sfsf): ... this.
5861 (vec_initv2sf): Renamed to ...
5862 (vec_initv2sfsf): ... this.
5863 (vec_extractv2si): Renamed to ...
5864 (vec_extractv2sisi): ... this.
5865 (vec_initv2si): Renamed to ...
5866 (vec_initv2sisi): ... this.
5867 (vec_extractv4hi): Renamed to ...
5868 (vec_extractv4hihi): ... this.
5869 (vec_initv4hi): Renamed to ...
5870 (vec_initv4hihi): ... this.
5871 (vec_extractv8qi): Renamed to ...
5872 (vec_extractv8qiqi): ... this.
5873 (vec_initv8qi): Renamed to ...
5874 (vec_initv8qiqi): ... this.
5875 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5876 (vec_init<mode>): Renamed to ...
5877 (vec_init<mode><VEC_base_l>): ... this.
5878 (vec_extract<mode>): Renamed to ...
5879 (vec_extract<mode><VEC_base_l>): ... this.
5880 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5881 (vec_initv2sfsf): ... this.
5882 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5883 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5884 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5885 element mode after mode in gen_vec_init* calls.
5886 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5887 (vec_init<mode><Vel>): ... this.
5888 (vec_extract<mode>): Renamed to ...
5889 (vec_extract<mode><Vel>): ... this.
5890 * config/aarch64/iterators.md (Vel): New mode attribute.
5891 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5892 Add element mode after mode in gen_vec_extract* calls.
5893 * config/s390/vector.md (non_vec_l): New mode attribute.
5894 (vec_extract<mode>): Renamed to ...
5895 (vec_extract<mode><non_vec_l>): ... this.
5896 (vec_init<mode>): Renamed to ...
5897 (vec_init<mode><non_vec_l>): ... this.
5898 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5899 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5901 * config/arm/iterators.md (V_elem_l): New mode attribute.
5902 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5903 (vec_extract<mode><V_elem_l>): ... this.
5904 (vec_extractv2di): Renamed to ...
5905 (vec_extractv2didi): ... this.
5906 (vec_init<mode>): Renamed to ...
5907 (vec_init<mode><V_elem_l>): ... this.
5908 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5909 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5910 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5911 Add element mode after gen_vec_extract* calls.
5912 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5913 (vec_init<mode><unitmode>): ... this.
5914 (vec_extract<mode>): Renamed to ...
5915 (vec_extract<mode><unitmode>): ... this.
5916 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5917 (vec_init<mode><unitmode>): ... this.
5918 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5919 (vec_initv2sfsf): ... this.
5920 (vec_extractv2sf): Renamed to ...
5921 (vec_extractv2sfsf): ... this.
5922 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5923 Add element mode after gen_vec_extract* calls.
5924 * config/mips/mips.md (unitmode): New mode iterator.
5925 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5926 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5927 * config/spu/spu.md (inner_l): New mode attribute.
5928 (vec_init<mode>): Renamed to ...
5929 (vec_init<mode><inner_l>): ... this.
5930 (vec_extract<mode>): Renamed to ...
5931 (vec_extract<mode><inner_l>): ... this.
5932 * config/sparc/sparc.md (veltmode): New mode iterator.
5933 (vec_init<VMALL:mode>): Renamed to ...
5934 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5935 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5936 (vec_initv2sisi): ... this.
5937 (vec_initv2sf): Renamed to ...
5938 (vec_initv2sfsf): ... this.
5939 (vec_extractv2sf): Renamed to ...
5940 (vec_extractv2sfsf): ... this.
5941 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5942 (vec_init<mode>): Renamed to ...
5943 (vec_init<mode><VEC_base_l>): ... this.
5944 (vec_extract<mode>): Renamed to ...
5945 (vec_extract<mode><VEC_base_l>): ... this.
5946 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5947 (vec_initv2sfsf): ... this.
5948 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5949 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5950 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5951 gen_vec_init* calls.
5953 2017-08-01 Richard Biener <rguenther@suse.de>
5955 PR tree-optimization/81297
5956 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5957 TREE_OVERFLOW from INTEGER_CSTs.
5959 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5961 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5963 2017-07-31 Carl Love <cel@us.ibm.com>
5965 * config/rs6000/rs6000-c: Add support for built-in functions
5966 vector signed char vec_xl_be (signed long long, signed char *);
5967 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5968 vector signed int vec_xl_be (signed long long, signed int *);
5969 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5970 vector signed long long vec_xl_be (signed long long, signed long long *);
5971 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5972 vector signed short vec_xl_be (signed long long, signed short *);
5973 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5974 vector double vec_xl_be (signed long long, double *);
5975 vector float vec_xl_be (signed long long, float *);
5976 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5977 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5978 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5980 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5981 (altivec_expand_builtin): Add switch statement to call
5982 altivec_expand_xl_be for each builtin.
5983 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5984 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5985 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5986 __builtin_vsx_le_be_v16qi.
5987 * doc/extend.texi: Update the built-in documentation file for the
5988 new built-in functions.
5990 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5993 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5995 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5997 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5999 * config.gcc: Add z14.
6000 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6001 CPU model numbers for z13s and z14.
6002 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6004 * config/s390/s390-opts.h (enum processor_type): Rename
6005 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6006 * config/s390/s390.c (processor_table): Add field for CPU name to
6007 be passed to Binutils.
6008 (s390_asm_output_machine_for_arch): Use the new field in
6009 processor_table for Binutils.
6010 (s390_expand_builtin): Replace arch12 with z14.
6011 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6012 (s390_get_sched_attrmask): Likewise.
6013 (s390_get_unit_mask): Likewise.
6014 * config/s390/s390.opt: Add z14 to processor_type enum.
6016 2017-07-31 Martin Jambor <mjambor@suse.cz>
6019 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6020 regardless of optimization level.
6022 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6023 Martin Liska <mliska@suse.cz>
6025 * predict.def: Remove old comment and adjust probability.
6026 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6029 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6032 * config/i386/i386.c (ix86_function_naked): New function.
6033 (ix86_can_use_return_insn_p): Return false for naked functions.
6034 (ix86_expand_prologue): Skip prologue for naked functions.
6035 (ix86_expand_epilogue): Skip epilogue for naked functions
6036 and emit trap instruction.
6037 (ix86_warn_func_return): New function.
6038 (ix86_attribute_table): Add "naked" attribute specification.
6039 (TARGET_WARN_FUNC_RETURN): Define.
6040 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6042 2017-07-31 Martin Liska <mliska@suse.cz>
6044 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6045 (dump_gimple_bb_header): Always dump BB info.
6046 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6048 2017-07-31 Martin Liska <mliska@suse.cz>
6051 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6052 also with current_function_decl non-null equality.
6054 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6057 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6058 change type to the element type, instead add eltype variable and
6059 use it where we are interested in the element type.
6061 PR tree-optimization/81603
6062 * ipa-polymorphic-call.c
6063 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6064 offset arithmetic in offset_int, bail out if the resulting bit offset
6065 doesn't fit into shwi.
6067 2017-07-31 Martin Liska <mliska@suse.cz>
6069 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6070 (gimplify_save_expr): Fix comment.
6072 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6075 * config/i386/i386.c (ix86_function_arg): Update arguments for
6077 (ix86_compute_frame_layout): Set the initial stack offset to
6078 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6079 INCOMING_FRAME_SP_OFFSET.
6080 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6081 stack before exception handler returns.
6082 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6083 the 'ERROR_CODE' for exception handler.
6085 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6087 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6088 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6089 (ASM_OUTPUT_REG_POP): Ditto.
6090 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6091 instead of asm_fprintf to output pure string.
6093 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6095 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6096 to imported_module_or_decl hook.
6097 (debug_nothing_tree_tree_tree_bool): Remove.
6098 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6099 * debug.c (do_nothing_debug_hooks): Use
6100 debug_nothing_tree_tree_tree_bool_bool instead of
6101 debug_nothing_tree_tree_tree_bool.
6102 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6103 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6104 * sdbout.c (sdb_debug_hooks): Likewise.
6105 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6106 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6108 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6109 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6110 attribute, don't add anything.
6112 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6114 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6115 (fold_build2_stat_loc): Likewise.
6116 (fold_build3_stat_loc): Likewise.
6117 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6118 (fold_build1_loc): Remove macro.
6119 (fold_build2_loc): Likewise.
6120 (fold_build3_loc): Likewise.
6122 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6124 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6125 (gimple_build_debug_bind_source_stat): Likewise.
6126 * gimple.h (gimple_build_debug_bind): Remove macro.
6127 (gimple_build_debug_bind_source): Likewise.
6129 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6131 * bitmap.c (bitmap_alloc): Adjust.
6132 (bitmap_gc_alloc): Likewise.
6133 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6135 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6137 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6138 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6139 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6140 (bitmap_gc_alloc_stat): Likewise.
6141 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6143 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6145 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6146 * rtl.h (shallow_copy_rtx): Remove macro.
6148 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6150 * emit-rtl.c (gen_raw_REG): Adjust.
6151 * gengenrtl.c (gendef): Likewise.
6152 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6153 * rtl.h (rtx_alloc): Remove macro.
6155 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6157 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6158 (build_tree_list_stat): Likewise.
6159 * tree.h (build_tree_list): Remove macro.
6160 (build_tree_list_vec): Likewise.
6162 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6164 * tree.c (make_vector_stat): Remove _stat from name.
6165 (build_vector_stat): Likewise.
6166 * tree.h (make_vector_stat): Remove macro.
6167 (build_vector_stat): Likewise.
6169 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6171 * tree.h (build_var_debug_value): Remove prototype.
6173 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6175 * tree.c (tree_cons_stat): Remove _stat from name.
6176 * tree.h (tree_cons): Remove macro.
6178 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6180 * tree.c (build_vl_exp_stat): Remove _stat from name.
6181 * tree.h (build_vl_exp): Remove macro.
6183 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6185 * tree.c (build_decl_stat): Remove _stat from name.
6186 * tree.h (build_decl): Remove macro.
6188 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6190 * gimple.c (gimple_build_with_ops_stat): Adjust.
6191 (gimple_alloc_stat): Remove _stat from name.
6192 * gimple.h (gimple_alloc): Remove macro.
6194 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6196 * tree.c (make_tree_vec_stat): Remove _stat from name.
6197 (grow_tree_vec_stat): Likewise.
6198 * tree.h (make_tree_vec_stat): Adjust prototype.
6199 (grow_tree_vec_stat): Likewise.
6200 (make_tree_vec): Remove macro.
6201 (grow_tree_vec): Likewise.
6203 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6205 * fold-const.c (fold_build1_stat_loc): Adjust.
6206 (fold_build2_stat_loc): Likewise.
6207 (fold_build3_stat_loc): Likewise.
6208 * tree.c (build0_stat): Remove _stat from name.
6209 (build1_stat): Likewise.
6210 (build2_stat): Likewise.
6211 (build3_stat): Likewise.
6212 (build4_stat): Likewise.
6213 (build5_stat): Likewise.
6214 * tree.h (build1_loc): Remove macro, and rename _stat function
6216 (build2_loc): Likewise.
6217 (build3_loc): Likewise.
6218 (build4_loc): Likewise.
6219 (build5_loc): Likewise.
6221 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6223 * tree.c (make_int_cst_stat): Remove _stat from name.
6224 * tree.h (make_int_cst_stat): Adjust prototype.
6225 (make_int_cst): Remove macro.
6227 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6229 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6230 * tree.h (make_tree_binfo_stat): Adjust prototype.
6231 (make_tree_binfo): Remove.
6233 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6235 * tree.c (copy_node_stat): Rename to copy_node.
6236 (build_distinct_type_copy): Adjust.
6237 * tree.h (copy_node_stat): Adjust prototype.
6238 (copy_node): Remove macro.
6240 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6242 * tree.c (make_node_stat): rename to make_node.
6243 (build_tree_list_stat): Adjust.
6244 (build0_stat): Likewise.
6245 (build2_stat): Likewise.
6246 (build3_stat): Likewise.
6247 (build4_stat): Likewise.
6248 (build5_stat): Likewise.
6249 (build_decl_stat): Likewise.
6250 * tree.h (make_node_stat): Adjust prototype.
6251 (make_node): remove macro.
6253 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6255 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6256 (PPC_FEATURE2_SCV): Likewise.
6257 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6259 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6261 * config/aarch64/aarch64.c
6262 (aarch64_internal_mov_immediate): Add new special pattern.
6263 * config/aarch64/aarch64.md (*movdi_aarch64):
6264 Add reg/32bit const mov case.
6266 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6267 Richard Sandiford <richard.sandiford@linaro.org>
6269 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6270 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6271 Add integer and movi cases.
6272 (movi-split-hf-df-sf split, fp16): New.
6273 (enabled): Added TARGET_FP_F16INST.
6274 * config/aarch64/iterators.md (GPF_HF): New.
6275 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6277 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6279 * config/aarch64/aarch64.c
6280 (aarch64_simd_container_mode): Add prototype.
6281 (aarch64_expand_mov_immediate): Add HI support.
6282 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6283 (aarch64_can_const_movi_rtx_p): New.
6284 (aarch64_preferred_reload_class):
6285 Remove restrictions of using FP registers for certain SIMD operations.
6286 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6287 (aarch64_valid_floating_const): Add integer move validation.
6288 (aarch64_simd_imm_scalar_p): Remove.
6289 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6290 (aarch64_legitimate_constant_p): Expand list of supported cases.
6291 * config/aarch64/aarch64-protos.h
6292 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6293 (aarch64_reinterpret_float_as_int): New.
6294 (aarch64_simd_imm_scalar_p): Remove.
6295 * config/aarch64/constraints.md (Uvi): New.
6296 (Dd): Split into Ds and new Dd.
6297 * config/aarch64/aarch64.md (*movsi_aarch64):
6299 (*movdi_aarch64): Add SIMD mov case.
6301 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6303 * tree-predcom.c: (struct chain): Handle store-store chain in which
6304 stores for elimination only store loop invariant values.
6305 (execute_pred_commoning_chain): Ditto.
6306 (prepare_initializers_chain_store_elim): Ditto.
6307 (prepare_finalizers): Ditto.
6308 (is_inv_store_elimination_chain): New function.
6309 (initialize_root_vars_store_elim_1): New function.
6311 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6313 * tree-predcom.c: Revise general description of the pass.
6314 (enum chain_type): New enum type for store elimination.
6315 (struct chain): New field supporting store elimination.
6316 (struct component): Ditto.
6317 (dump_chain): Dump store-stores chain.
6318 (release_chain): Release resources.
6319 (split_data_refs_to_components): Compute and create component
6320 contains only stores for elimination.
6321 (get_chain_last_ref_at): New function.
6322 (make_invariant_chain): Initialization.
6323 (make_rooted_chain): Specify chain type in parameter and record it.
6324 (add_looparound_copies): Skip for store-stores chain.
6325 (determine_roots_comp): Compute type of chain and pass it to
6327 (initialize_root_vars_store_elim_2): New function.
6328 (finalize_eliminated_stores): New function.
6329 (remove_stmt): Handle store for elimination.
6330 (execute_pred_commoning_chain): Execute predictive commoning on
6332 (determine_unroll_factor): Skip unroll for store-stores chain.
6333 (prepare_initializers_chain_store_elim): New function.
6334 (prepare_initializers_chain): Hanlde store-store chain.
6335 (prepare_finalizers_chain, prepare_finalizers): New function.
6336 (tree_predictive_commoning_loop): Return integer value indicating
6337 if loop is unrolled or lcssa form is corrupted.
6338 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6340 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6342 * tree-predcom.c (initialize_root): Delete.
6343 (execute_pred_commoning_chain): Initialize root vars and replace
6344 reference of non-combined chain directly, rather than call above
6347 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6349 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6350 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6352 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6354 * tree-predcom.c (struct chain): New field init_seq.
6355 (release_chain): Release init_seq.
6356 (prepare_initializers_chain): Record intialization stmts in above
6358 (insert_init_seqs): New function.
6359 (tree_predictive_commoning_loop): Call insert_init_seqs.
6361 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6363 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6365 2017-07-28 Richard Biener <rguenther@suse.de>
6367 * match.pd: Remove superfluous :c.
6368 * genmatch.c (simplify::id): Add member.
6369 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6371 (current_id): New global.
6372 (dt_node::parent): Move from ...
6373 (dt_operand::parent): ... here. Add for_id member.
6374 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6375 (decision_tree::find_node): Relax order requirement when
6376 merging DT_TRUE nodes to ones inbetween the current simplify
6377 and the one we try to merge with. Add diagnostic whenever
6378 we need to enforce pattern order by not merging.
6379 (decision_tree::insert): Set current_id.
6380 (decision_tree::print_node): Dump parent node and for_id.
6381 (parser::last_id): Add member.
6382 (parser::push_simplify): Assign unique id.
6383 (parser::parser): Initialize last_id.
6385 2017-07-28 Martin Liska <mliska@suse.cz>
6388 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6389 gimple_build_debug_bind.
6391 2017-07-28 Richard Biener <rguenther@suse.de>
6393 PR tree-optimization/81502
6394 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6396 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6398 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6399 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6400 for BIT_FIELD_REF args.
6401 * fold-const.c (make_bit_field_ref): Likewise.
6402 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6404 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6407 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6408 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6409 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6410 Or it into SANITIZER_UNDEFINED.
6411 * ubsan.c: Include gimple-fold.h and varasm.h.
6412 (ubsan_expand_ptr_ifn): New function.
6413 (instrument_pointer_overflow): New function.
6414 (maybe_instrument_pointer_overflow): New function.
6415 (instrument_object_size): Formatting fix.
6416 (pass_ubsan::execute): Call instrument_pointer_overflow
6417 and maybe_instrument_pointer_overflow.
6418 * internal-fn.c (expand_UBSAN_PTR): New function.
6419 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6420 * sanitizer.def (__ubsan_handle_pointer_overflow,
6421 __ubsan_handle_pointer_overflow_abort): New builtins.
6422 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6423 * internal-fn.def (UBSAN_PTR): New internal function.
6424 * opts.c (sanitizer_opts): Add pointer-overflow.
6425 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6426 * fold-const.c (build_range_check): Compute pointer range check in
6427 integral type if pointer arithmetics would be needed. Formatting
6430 2017-07-28 Martin Liska <mliska@suse.cz>
6433 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6434 parameters that are of a variable-length.
6436 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6438 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6440 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6441 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6442 (CRT_CALL_STATIC_FUNCTION): Likewise.
6443 (ASM_DEFAULT_SPEC): New define.
6444 (ASM_SPEC32): Likewise.
6445 (ASM_SPEC64): Likewise.
6446 (ASM_SPEC_COMMON): Likewise.
6447 (ASM_SPEC): Likewise.
6448 (INVALID_64BIT): Likewise.
6449 (LINK_OS_DEFAULT_SPEC): Likewise.
6450 (LINK_OS_SPEC32): Likewise.
6451 (LINK_OS_SPEC64): Likewise.
6452 (POWERPC_LINUX): Likewise.
6453 (PTRDIFF_TYPE): Likewise.
6454 (RESTORE_FP_PREFIX): Likewise.
6455 (RESTORE_FP_SUFFIX): Likewise.
6456 (SAVE_FP_PREFIX): Likewise.
6457 (SAVE_FP_SUFFIX): Likewise.
6458 (SIZE_TYPE): Likewise.
6459 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6460 (TARGET_64BIT): Likewise.
6461 (TARGET_64BIT): Likewise.
6462 (TARGET_AIX): Likewise.
6463 (WCHAR_TYPE_SIZE): Likewise.
6464 (WCHAR_TYPE): Undefine.
6465 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6466 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6467 (CPP_OS_RTEMS_SPEC): Delete.
6468 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6469 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6471 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6473 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6475 PR tree-optimization/81578
6476 * tree-parloops.c (build_new_reduction): Bail out if
6477 reduction_code isn't one of the standard OpenMP reductions.
6478 Move the details printing after that decision.
6480 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6482 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6483 related to reload_in_progress.
6484 (splat_input_operand): Likewise.
6485 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6487 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6489 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6490 (TARGET_INSTANTIATE_DECLS): Likewise.
6491 (legitimate_indexed_address_p): Delete reload_in_progress code.
6492 (rs6000_debug_legitimate_address_p): Likewise.
6493 (rs6000_eliminate_indexed_memrefs): Likewise.
6494 (rs6000_emit_le_vsx_store): Likewise.
6495 (rs6000_emit_move_si_sf_subreg): Likewise.
6496 (rs6000_emit_move): Likewise.
6497 (register_to_reg_type): Likewise.
6498 (rs6000_pre_atomic_barrier): Likewise.
6499 (rs6000_machopic_legitimize_pic_address): Likewise.
6500 (rs6000_allocate_stack_temp): Likewise.
6501 (rs6000_address_for_fpconvert): Likewise.
6502 (rs6000_address_for_altivec): Likewise.
6503 (rs6000_secondary_memory_needed_rtx): Delete function.
6504 (rs6000_check_sdmode): Likewise.
6505 (rs6000_alloc_sdmode_stack_slot): Likewise.
6506 (rs6000_instantiate_decls): Likewise.
6507 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6508 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6509 Delete reload_in_progress.
6510 (*vec_reload_and_plus_<mptrsize>): Likewise.
6511 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6512 (vsx_div_v2di): Likewise.
6513 (vsx_udiv_v2di): Likewise.
6515 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6517 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6518 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6519 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6520 (rs6000_debug_reg_global): Delete print of LRA status.
6521 (rs6000_option_override_internal): Delete dead LRA related code.
6522 (rs6000_lra_p): Delete function.
6523 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6525 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6527 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6528 * config/riscv/rtems.h: New file.
6530 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6531 Sudakshina Das <sudi.das@arm.com>
6533 * config/aarch64/aarch64.md
6534 (define_split for and<mode>3nr_compare): Move
6535 non aarch64_logical_operand to a register.
6536 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6537 register immediate operand to a register.
6538 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6540 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6543 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6545 2017-07-27 Richard Biener <rguenther@suse.de>
6547 PR tree-optimization/81573
6548 PR tree-optimization/81494
6549 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6550 multi defuse cycle case.
6552 2017-07-27 Richard Biener <rguenther@suse.de>
6554 PR tree-optimization/81571
6555 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6558 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6560 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6561 earlier and only if MASK_FPU is set. Adjust formatting.
6563 2017-07-27 Martin Liska <mliska@suse.cz>
6565 * opt-functions.awk: Add validation of value of Init.
6566 * optc-gen.awk: Pass new argument.
6568 2017-07-27 Martin Liska <mliska@suse.cz>
6570 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6571 Fix wrong condition.
6573 2017-07-27 Martin Liska <mliska@suse.cz>
6575 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6576 BBs and edges seen by autoFDO.
6578 2017-07-27 Richard Biener <rguenther@suse.de>
6580 PR tree-optimization/81502
6581 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6582 with incompatible but same sized type.
6583 (execute_update_addresses_taken): Likewise.
6585 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6587 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6588 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6590 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6593 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6594 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6595 Change s_operand to memory_operand.
6597 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6599 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6600 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6601 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6602 Emit instructions rather than returning an expression. Handle TFmode
6603 and KFmode by casting to TImode.
6604 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6605 (rs6000_emit_le_vsx_store): Likewise.
6606 * config/rs6000/vsx.md (VSX_TI): New iterator.
6607 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6608 (*vsx_le_undo_permute_<mode>): Likewise.
6609 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6610 emit the split sequence.
6611 (*vsx_le_perm_store_<mode>): Likewise.
6613 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6615 PR tree-optimization/81555
6616 PR tree-optimization/81556
6617 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6618 if true, force CHANGED for the recursive invocation.
6619 (reassociate_bb): Remember original length of ops array, pass
6620 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6622 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6623 attributes for noipa attribute. For naked attribute use
6624 lookup_attribute first before lookup_attribute_spec.
6625 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6627 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6628 for functions with noipa attribute.
6629 (cgraph_externally_visible_p): Return true for functions with noipa
6631 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6632 for functions with noipa attribute.
6633 * doc/extend.texi: Document noipa function attribute.
6634 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6635 also for functions with noipa attribute.
6636 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6638 2017-07-26 Andrew Pinski <apinski@cavium.com>
6640 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6641 vec_unalign_load_cost and vec_unalign_store_cost.
6643 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6645 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6646 -mvsx-small-integer option.
6647 (ISA_3_0_MASKS_IEEE): Likewise.
6648 (OTHER_VSX_VECTOR_MASKS): Likewise.
6649 (POWERPC_MASKS): Likewise.
6650 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6651 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6652 code, only testing for DImode being allowed in non-VSX floating
6654 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6655 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6657 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6658 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6659 TARGET_P8_VECTOR test.
6660 (rs6000_secondary_reload_simple_move): Likewise.
6661 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6662 since TARGET_P9_VECTOR was already tested.
6663 (rs6000_opt_masks): Remove -mvsx-small-integer.
6664 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6665 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6667 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6668 test for TARGET_VEXTRACTUB was used, and that uses
6670 (p9 extract splitter): Likewise.
6671 (vsx_extract_<mode>_di_p9): Likewise.
6672 (vsx_extract_<mode>_store_p9): Likewise.
6673 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6674 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6675 the elimination of TARGET_VSX_SMALL_INTEGER.
6676 (vsx_extract_<mode>_p8): Likewise.
6677 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6678 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6679 (vsx_set_<mode>_p9): Likewise.
6680 (vsx_set_v4sf_p9): Likewise.
6681 (vsx_set_v4sf_p9_zero): Likewise.
6682 (vsx_insert_extract_v4sf_p9): Likewise.
6683 (vsx_insert_extract_v4sf_p9_2): Likewise.
6684 * config/rs6000/rs6000.md (sign extend splitter): Change
6685 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6686 (floatsi<mode>2_lfiwax_mem): Likewise.
6687 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6688 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6689 since a test for TARGET_P9_VECTOR was used.
6690 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6691 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6692 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6693 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6694 TARGET_P8_VECTOR test.
6695 (fix_trunc<mode>si2_stfiwx): Likewise.
6696 (fix_trunc<mode>si2_internal): Likewise.
6697 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6698 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6700 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6701 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6702 TARGET_P8_VECTOR test.
6703 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6704 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6705 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6707 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6708 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6709 since a test for TARGET_P9_VECTOR was used.
6710 (splitter for loading small constants): Likewise.
6712 2017-07-26 Andrew Pinski <apinski@cavium.com>
6714 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6717 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6720 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6721 (fp_valid_at): Likewise.
6723 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6725 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6726 (qdf24xx_addrcost_table): Likewise.
6727 (cortexa57_tunings): Update to use generic_branch_cost.
6728 (cortexa72_tunings): Likewise.
6729 (cortexa73_tunings): Likewise.
6730 (qdf24xx_tunings): Likewise.
6732 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6734 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6735 (thunderx2t99_branch_cost): Likewise.
6736 (cortexa35_tunings): Update to use generic_branch_cost.
6737 (cortexa53_tunings): Likewise.
6738 (cortexa57_tunings): Likewise.
6739 (cortexa72_tunings): Likewise.
6740 (cortexa73_tunings): Likewise.
6741 (thunderx2t99_tunings): Likewise.
6743 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6745 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6746 (sparc_option_override): Honour MASK_FSMULD.
6747 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6748 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6749 * config/sparc/sparc.opt (mfsmuld): New option.
6750 * doc/invoke.texi (mfsmuld): Document option.
6752 2017-07-26 Marek Polacek <polacek@redhat.com>
6755 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6757 2017-07-26 Richard Biener <rguenther@suse.de>
6759 * gimple-match-head.c (do_valueize): Return OP if valueize
6761 (get_def): New helper to get at the def stmt of a SSA name
6763 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6764 do_valueize to get at the def stmt.
6765 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6767 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6770 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6772 2017-07-26 Martin Liska <mliska@suse.cz>
6775 * function.c (expand_function_start): Make expansion of
6776 nonlocal_goto_save_area after parm_birth_insn.
6778 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6780 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6781 from all CPU target flags enable members.
6783 2017-07-26 Richard Biener <rguenther@suse.de>
6785 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6786 (decision_tree::gen): Make 'type' const.
6787 (write_predicate): Likewise.
6789 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6791 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6792 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6793 (rs6000_option_override_internal): Likewise.
6794 (rs6000_expand_vector_set): Likewise.
6795 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6796 (TARGET_UPPER_REGS_SF): Likewise.
6797 (TARGET_UPPER_REGS_DI): Likewise.
6798 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6799 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6800 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6801 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6802 (Splitters for DI constants in Altivec registers): Likewise.
6803 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6804 (vsx_set_v4sf_p9): Likewise.
6805 (vsx_set_v4sf_p9_zero): Likewise.
6806 (vsx_insert_extract_v4sf_p9): Likewise.
6807 (vsx_insert_extract_v4sf_p9_2): Likewise.
6809 2017-07-25 Carl Love <cel@us.ibm.com>
6811 * doc/extend.texi: Update the built-in documentation file for the
6812 existing built-in functions
6813 vector signed char vec_cnttz (vector signed char);
6814 vector unsigned char vec_cnttz (vector unsigned char);
6815 vector signed short vec_cnttz (vector signed short);
6816 vector unsigned short vec_cnttz (vector unsigned short);
6817 vector signed int vec_cnttz (vector signed int);
6818 vector unsigned int vec_cnttz (vector unsigned int);
6819 vector signed long long vec_cnttz (vector signed long long);
6820 vector unsigned long long vec_cnttz (vector unsigned long long);
6822 2017-07-25 Andrew Pinski <apinski@cavium.com>
6824 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6825 accesses where the use is for the first operand of a BIT_INSERT.
6827 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6830 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6831 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6833 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6835 * config/i386/gstabs.h: Delete.
6836 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6838 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6840 * config/i386/i386.c (ix86_decompose_address): Do not check for
6841 register RTX when looking at index_reg or base_reg.
6842 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6844 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6846 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6847 to update EH info here.
6849 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6851 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6853 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6855 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6857 2017-07-25 Torsten Duwe <duwe@suse.de>
6859 * common.opt: Introduce -fpatchable-function-entry
6860 command line option, and its variables function_entry_patch_area_size
6861 and function_entry_patch_area_start.
6862 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6863 including a two-value parser.
6864 * target.def (print_patchable_function_entry): New target hook.
6865 * targhooks.h (default_print_patchable_function_entry): New function.
6866 * targhooks.c (default_print_patchable_function_entry): Likewise.
6867 * toplev.c (process_options): Switch off IPA-RA if
6868 patchable function entries are being generated.
6869 * varasm.c (assemble_start_function): Look at the
6870 patchable-function-entry command line switch and current
6871 function attributes and maybe generate NOP instructions by
6872 calling the print_patchable_function_entry hook.
6873 * doc/extend.texi: Document patchable_function_entry attribute.
6874 * doc/invoke.texi: Document -fpatchable_function_entry
6875 command line option.
6876 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6878 * doc/tm.texi: Re-generate.
6880 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6883 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6884 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6886 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6888 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6890 2017-07-25 Richard Biener <rguenther@suse.de>
6892 PR tree-optimization/81455
6893 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6894 not walk in cycles when looking for guards.
6896 2017-07-25 Richard Biener <rguenther@suse.de>
6898 PR tree-optimization/81529
6899 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6900 when optimizing backedge uses.
6902 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6904 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6906 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6907 to dl_section_ref. On AIX, append an expression to subtract
6908 the size of the section length to dl_section_ref.
6910 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6912 * configure.ac: If any of the config.* scripts fail, exit 1.
6913 * configure: Regenerate.
6915 2017-07-25 Richard Biener <rguenther@suse.de>
6918 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6919 of immediate uses, be more verbose on errors.
6921 2017-07-25 Richard Biener <rguenther@suse.de>
6923 PR tree-optimization/81510
6924 * tree-vect-loop.c (vect_is_simple_reduction): When the
6925 reduction stmt is not inside the loop bail out.
6927 2017-07-25 Richard Biener <rguenther@suse.de>
6929 PR tree-optimization/81303
6930 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6931 profitability check against LOOP_VINFO_NITERSM1.
6933 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6935 * domwalk.c (cmp_bb_postorder): Simplify.
6936 (sort_bbs_postorder): New function. Use it...
6937 (dom_walker::walk): ...here to optimize common cases.
6939 2017-07-25 Martin Liska <mliska@suse.cz>
6942 * ipa-visibility.c (function_and_variable_visibility): Make the
6943 redirection just on target that supports aliasing.
6944 Fix GNU coding style.
6946 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6949 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6951 * config/arm/rtems.h: Likewise.
6952 * config/bfin/rtems.h: Likewise.
6953 * config/i386/rtemself.h: Likewise.
6954 * config/lm32/rtems.h: Likewise.
6955 * config/m32c/rtems.h: Likewise.
6956 * config/m68k/rtemself.h: Likewise.
6957 * config/microblaze/rtems.h: Likewise.
6958 * config/mips/rtems.h: Likewise.
6959 * config/moxie/rtems.h: Likewise.
6960 * config/nios2/rtems.h: Likewise.
6961 * config/powerpcspe/rtems.h: Likewise.
6962 * config/rs6000/rtems.h: Likewise.
6963 * config/rtems.h: Likewise.
6964 * config/sh/rtems.h: Likewise.
6965 * config/sh/rtemself.h: Likewise.
6966 * config/sparc/rtemself.h: Likewise.
6968 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6971 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6972 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6973 * tree-ssa-structalias.c (alias_get_name): Same.
6975 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6978 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6979 instructions if no du chain is found.
6981 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6983 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6985 2017-07-25 Richard Biener <rguenther@suse.de>
6988 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6991 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6993 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6995 (ISA_2_7_MASKS_SERVER): Likewise.
6996 (ISA_3_0_MASKS_IEEE): Likewise.
6997 (OTHER_P8_VECTOR_MASKS): Likewise.
6998 (OTHER_VSX_VECTOR_MASKS): Likewise.
6999 (POWERPC_MASKS): Likewise.
7000 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7001 duplicate list of options.
7002 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7003 explicit -mupper-regs options.
7004 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7005 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7006 alias for -mupper-regs-df.
7007 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7008 (rs6000_init_hard_regno_mode_ok): Likewise.
7009 (rs6000_option_override_internal): Likewise.
7010 (rs6000_opt_masks): Likewise.
7011 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7012 options in terms of whether -mvsx or -mpower8-vector was used.
7013 (TARGET_UPPER_REGS_DI): Likewise.
7014 (TARGET_UPPER_REGS_SF): Likewise.
7015 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7016 -mupper-regs-* options.
7018 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7020 * passes.c (emergency_dump_function): Print some empty lines and a
7021 header before the RTL dump.
7023 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7025 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7027 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7030 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7031 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7033 2017-07-24 Carl Love <cel@us.ibm.com>
7035 * config/rs6000/rs6000-c.c: Add support for built-in functions
7036 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7037 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7038 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7039 vec_extract_fp_from_shortl): Add defines for the two builtins.
7040 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7041 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7043 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7044 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7045 * doc/extend.texi: Update the built-in documentation file for the
7046 new built-in function.
7048 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7051 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7053 * doc/generic.texi: Likewise.
7054 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7055 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7057 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7059 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7060 (aarch64_mls_elt_merge<mode>): Likewise.
7062 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7064 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7065 having __cxa_atexit.
7067 2017-07-23 Michael Collison <michael.collison@arm.com>
7069 * config/arm/arm.c (arm_option_override): Deprecate
7070 use of -mstructure-size-boundary.
7071 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7072 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7074 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7077 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7078 Reduce cost estimate for direct moves.
7080 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7083 * config/i386/i386.c (ix86_option_override_internal): Disable
7084 BMI, BMI2 and TBM instructions for -m16.
7086 2017-07-21 Carl Love <cel@us.ibm.com>
7088 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7089 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7090 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7091 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7092 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7093 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7094 VMULOSW): New enum "unspec" values.
7095 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7096 altivec_vmulosw): New patterns.
7097 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7098 VMULOSW): Add definitions.
7100 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7102 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7103 (qdf24xx): Likewise.
7104 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7105 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7106 (AARCH64_FL_V8_1): Renumber.
7107 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7108 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7109 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7110 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7111 rdma to feature modifiers list.
7113 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7116 * ipa-visibility (function_and_variable_visibility): Convert
7117 recursive PLT call to direct call if appropriate.
7119 2017-07-21 Andrew Pinski <apinski@cavium.com>
7121 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7122 operand 1 to see if the types precision matches.
7123 * fold-const.c (operand_equal_p): Likewise.
7125 2017-07-21 Richard Biener <rguenther@suse.de>
7127 PR tree-optimization/81303
7128 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7129 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7130 (vect_peeling_hash_get_lowest_cost): Adjust.
7131 (vect_enhance_data_refs_alignment): Likewise. Use
7132 vect_get_peeling_costs_all_drs to compute the penalty for no
7133 peeling to match up costs.
7135 2017-07-21 Richard Biener <rguenther@suse.de>
7137 PR tree-optimization/81500
7138 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7139 we didn't identify a reduction path.
7141 2017-07-21 Tom de Vries <tom@codesourcery.com>
7142 Cesar Philippidis <cesar@codesourcery.com>
7144 PR gcov-profile/81442
7145 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7148 2017-07-21 Tom de Vries <tom@codesourcery.com>
7151 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7153 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7154 nvptx_override_options_after_change.
7156 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7158 * dwarf2out.c (output_file_names): Avoid double testing for
7161 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7163 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7165 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7167 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7169 (try_crossjump_to_edge): Do not punt on partitioned functions.
7171 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7173 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7174 Put all BBs reachable only via paths crossing cold region to cold
7176 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7178 2016-07-21 Richard Biener <rguenther@suse.de>
7180 PR tree-optimization/81303
7181 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7182 into account prologue and epilogue iterations when raising
7183 min_profitable_iters to sth at least covering one vector iteration.
7185 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7187 * config/arm/arm.c (arm_test_cpu_arch_dat):
7190 2017-07-20 Nathan Sidwell <nathan@acm.org>
7192 Remove TYPE_METHODS.
7193 * tree.h (TYPE_METHODS): Delete.
7194 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7195 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7196 (dbxout_type_methods): Scan TYPE_FIELDS.
7197 (dbxout_type): Don't check TYPE_METHODS here.
7198 * function.c (use_register_for_decl): Always ignore register for
7199 class types when not optimizing.
7200 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7201 * tree.c (free_lang_data_in_type): Stitch out member functions and
7202 templates from TYPE_FIELDS.
7203 (build_distinct_type_copy, verify_type_variant,
7204 verify_type): Member fns are on TYPE_FIELDS.
7205 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7206 * tree-pretty-print.c (dump_generic_node): Likewise.
7208 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7211 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7212 V2TImode and V4TImode.
7213 (ix86_expand_vector_extract): Likewise.
7214 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7215 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7216 (ssescalarmode): Handle V4TImode and V2TImode.
7217 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7218 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7219 (VEXTRACTI128_MODE): New mode iterator.
7220 (splitter for *vec_extractv?ti first element): New.
7221 (VEC_INIT_MODE): New mode iterator.
7222 (vec_init<mode>): Consolidate 3 expanders into one using
7223 VEC_INIT_MODE mode iterator.
7225 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7227 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7228 non_spilled_static_chain_regno_p.
7230 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7232 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7234 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7236 * bb-reorder.c (connect_traces): Allow copying of blocks within
7239 2017-07-20 Richard Biener <rguenther@suse.de>
7241 * gimple.h (gimple_phi_result): Add gphi * overload.
7242 (gimple_phi_result_ptr): Likewise.
7243 (gimple_phi_arg): Likewise. Adjust index assert to only
7244 allow actual argument accesses rather than all slots available
7246 (gimple_phi_arg_def): Add gphi * overload.
7247 * tree-phinodes.c (make_phi_node): Initialize only actual
7249 (resize_phi_node): Clear memory not covered by old node,
7250 do not initialize excess argument slots.
7251 (reserve_phi_args_for_new_edge): Initialize new argument slot
7254 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7256 PR tree-optimization/81388
7258 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7260 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7261 by removing computation of may_be_zero.
7263 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7264 Tom de Vries <tom@codesourcery.com>
7267 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7268 when gimple level profile disagrees with what RTL expander did.
7270 2017-07-20 Richard Biener <rguenther@suse.de>
7272 PR tree-optimization/61171
7273 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7274 (vect_analyze_stmt): Add slp instance parameter.
7275 (vectorizable_reduction): Likewise.
7276 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7277 (vect_is_simple_reduction): Deal with chains not detected
7278 as SLP reduction chain, specifically not properly associated
7279 chains containing a mix of plus/minus.
7280 (get_reduction_op): Remove.
7281 (get_initial_defs_for_reduction): Simplify, pass in whether
7282 this is a reduction chain, pass in the SLP node for the PHIs.
7283 (vect_create_epilog_for_reduction): Get the SLP instance as
7285 (vectorizable_reduction): Get the SLP instance as arg.
7286 During analysis remember the SLP node with the PHIs in the
7287 instance. Simplify getting at the vectorized reduction PHIs.
7288 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7289 through SLP instance.
7290 (vect_slp_analyze_operations): Likewise.
7291 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7292 (vect_transform_stmt): Likewise.
7294 2017-07-20 Tom de Vries <tom@codesourcery.com>
7296 PR tree-optimization/81489
7297 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7298 read of phi arg location to before loop that modifies phi.
7300 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7302 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7305 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7308 * except.c (execute): Fix ordering issue.
7310 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7312 PR rtl-optimization/81423
7313 * combine.c (make_compound_operation_int): Don't try to optimize
7314 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7316 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7318 PR rtl-optimization/81423
7319 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7320 with a constant that is -1 in the truncated to mode.
7322 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7324 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7325 (determine_unlikely_bbs): ... here.
7326 * predict.h (propagate_unlikely_bbs_forward): Declare.
7327 * cfgexpand.c (pass_expand::execute): Use it.
7328 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7330 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7331 propagate_unlikely_bbs_forward.
7333 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7336 * except.c (maybe_add_nop_after_section_switch): New function.
7339 2017-07-19 Tom de Vries <tom@codesourcery.com>
7341 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7343 2017-07-19 Tom de Vries <tom@codesourcery.com>
7345 * gimple.h (gimple_phi_arg): Make assert more strict.
7347 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7349 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7350 * config/rs6000/mmintrin.h: New file.
7351 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7353 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7355 PR tree-optimization/81346
7356 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7358 2017-07-19 Tom de Vries <tom@codesourcery.com>
7360 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7362 2017-07-19 Tom de Vries <tom@codesourcery.com>
7364 * config/nvptx/nvptx-modes.def: Add V2DImode.
7365 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7366 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7367 (nvptx_output_mov_insn): Handle lack of mov.b128.
7368 (nvptx_print_operand): Handle 'H' and 'L' codes.
7369 (nvptx_vector_mode_supported): Allow V2DImode.
7370 (nvptx_preferred_simd_mode): New function.
7371 (nvptx_data_alignment): New function.
7372 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7373 nvptx_preferred_simd_mode.
7374 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7376 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7378 2017-07-19 Tom de Vries <tom@codesourcery.com>
7380 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7381 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7382 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7383 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7384 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7385 (mov<VECIM>_insn): New define_insn.
7386 (define_expand "mov<VECIM>): New define_expand.
7388 2017-07-19 Tom de Vries <tom@codesourcery.com>
7390 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7392 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7394 PR tree-optimization/81346
7395 * fold-const.h (fold_div_compare, range_check_type): Declare.
7396 * fold-const.c (range_check_type): New function.
7397 (build_range_check): Use range_check_type.
7398 (fold_div_compare): No longer static, rewritten into
7399 a match.pd helper function.
7400 (fold_comparison): Don't call fold_div_compare here.
7401 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7404 2017-07-19 Nathan Sidwell <nathan@acm.org>
7406 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7407 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7408 * tree.c (find_decls_types_r, verify_type): Use
7409 TYPE_{MIN,MAX}_VALUE_RAW.
7410 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7411 (hash_tree): Likewise.
7412 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7414 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7417 2017-07-18 Tom de Vries <tom@codesourcery.com>
7420 * omp-expand.c (expand_omp_for_static_chunk): Handle
7421 equal-argument loop exit phi.
7423 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7426 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7427 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7428 operand 2 predicate.
7429 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7430 operand 2 predicate.
7431 (ror,rol -> rorx splitters): Use const_int_operand as
7432 operand 2 predicate.
7434 2017-06-18 Richard Biener <rguenther@suse.de>
7436 PR tree-optimization/81410
7437 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7438 the gap in the ! slp_perm SLP case after each group.
7440 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7443 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7446 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7449 * predict.c (set_even_probabilities): Cleanup; do not affect
7450 probabilities that are already known.
7451 (combine_predictions_for_bb): Call even when count is set.
7453 2017-07-18 Nathan Sidwell <nathan@acm.org>
7455 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7458 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7461 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7462 optimization for loop niter analysis.
7464 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7467 * config/avr/avr.c (avr_optimize_casesi): Don't use
7468 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7470 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7472 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7473 body_cost_vec from _vect_peel_extended_info.
7474 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7475 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7478 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7480 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7482 2017-07-18 Richard Biener <rguenther@suse.de>
7484 PR tree-optimization/80620
7485 PR tree-optimization/81403
7486 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7487 info when re-using a VN table entry.
7489 2017-07-18 Richard Biener <rguenther@suse.de>
7491 PR tree-optimization/81418
7492 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7493 vectype_in. Verify that with lane-reducing reduction operations
7494 we have a single def-use cycle.
7496 2017-07-17 Carl Love <cel@us.ibm.com>
7498 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7500 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7501 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7502 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7503 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7504 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7505 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7506 VMULOSW): New enum "unspec" values.
7507 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7508 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7509 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7510 altivec_vmulosw): New patterns.
7511 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7512 VMULOSW): Add definitions.
7513 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7515 * config/alpha/alpha.c: Include predict.h.
7517 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7519 * tree-vrp.c (compare_assert_loc): Fix comparison function
7520 to return predictable results.
7522 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7524 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7527 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7528 * doc/invoke.texi (mexpand-adddi): Update text.
7530 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7532 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7533 that also clobbers the CC register. The old expand code is moved
7535 (*arc_clzsi2): ... here.
7536 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7537 the CC register. The old expand code is moved to ...
7538 (arc_ctzsi2): ... here.
7540 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7542 * config/arc/arc.opt (mindexed-loads): Use initial value
7543 TARGET_INDEXED_LOADS_DEFAULT.
7544 (mauto-modify-reg): Use initial value
7545 TARGET_AUTO_MODIFY_REG_DEFAULT.
7546 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7547 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7548 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7549 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7551 2017-07-17 Martin Liska <mliska@suse.cz>
7554 * opts.c (finish_options): Do not allow -fgnu-tm
7555 w/ -fsanitize={kernel-,}address. Say sorry.
7557 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7560 * tree-loop-distribution.c (classify_partition): Only assert on
7561 numer of iterations.
7562 (merge_dep_scc_partitions): Delete prameter. Update function call.
7563 (distribute_loop): Remove code handling loop with unknown niters.
7564 (pass_loop_distribution::execute): Skip loop with unknown niters.
7566 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7569 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7570 function sort_partitions_by_post_order.
7572 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7574 PR tree-optimization/81374
7575 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7576 the max index of basic blocks, rather than number of basic blocks.
7578 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7580 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7582 (arc_legitimate_pic_operand_p): Likewise.
7583 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7585 (arc_needs_pcl_p): Likewise.
7586 (arc_legitimate_pc_offset_p): Likewise.
7587 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7588 function is also used in constrains.md.
7589 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7590 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7591 PLUS. Only return true/false in known cases, otherwise assert.
7592 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7593 is already called in arc_legitimate_constant_p.
7594 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7596 (LEGITIMATE_PIC_OPERAND_P): Use
7597 arc_raw_symbolic_reference_mentioned_p function.
7598 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7603 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7604 Andrew Burgess <andrew.burgess@embecosm.com>
7606 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7607 (arc_return_address_register): New function.
7608 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7609 (arc_handle_fndecl_attribute): Add naked attribute.
7610 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7611 (TARGET_WARN_FUNC_RETURN): Likewise.
7612 (arc_allocate_stack_slots_for_args): New function.
7613 (arc_warn_func_return): Likewise.
7614 (machine_function): Change type fn_type.
7615 (arc_compute_function_type): Consider new naked function type,
7616 change function return type.
7617 (arc_must_save_register): Adapt to handle new
7618 arc_compute_function_type's return type.
7619 (arc_expand_prologue): Likewise.
7620 (arc_expand_epilogue): Likewise.
7621 (arc_return_address_regs): Delete.
7622 (arc_return_address_register): New function.
7623 (arc_epilogue_uses): Use above function.
7624 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7625 (arc_function_type): Change encoding, add naked type.
7626 (ARC_INTERRUPT_P): Change to handle the new encoding.
7627 (ARC_FAST_INTERRUPT_P): Likewise.
7628 (ARC_NORMAL_P): Define.
7629 (ARC_NAKED_P): Likewise.
7630 (arc_compute_function_type): Delete prototype.
7631 * config/arc/arc.md (in_ret_delay_slot): Use
7632 arc_return_address_register function.
7633 (simple_return): Likewise.
7634 (p_return_i): Likewise.
7636 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7638 PR tree-optimization/81428
7639 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7640 can't be built for those types.
7642 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7644 Remove stuff dead since r239246.
7646 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7647 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7648 (avr_inform_devices): Remove dead stuff.
7650 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7652 * config/arm/arm_neon.h: Fix softp typo.
7654 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7656 PR tree-optimization/81365
7657 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7658 aggregate moves onto bb predecessor edges, make sure there are no
7659 loads that could alias the lhs in between the start of bb and the
7662 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7665 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7666 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7667 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7669 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7671 PR tree-optimization/81396
7672 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7673 (init_symbolic_number): Initialize it to 1.
7674 (perform_symbolic_merge): Add n_ops from both operands into the new
7676 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7677 without base_addr as useless if they need more than one operation.
7678 (bswap_replace): Handle !bswap case for NULL base_addr.
7680 2017-07-17 Tom de Vries <tom@codesourcery.com>
7683 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7686 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7688 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7689 conditional builtin define __FIX_LEON3FT_B2BST.
7691 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7693 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7694 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7697 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7699 PR rtl-optimization/81424
7700 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7701 to remove potential trapping from operands if -fnon-call-exceptions.
7703 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7705 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7706 profile_proability for scalling.
7707 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7709 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7711 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7713 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7715 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7716 fixpoint arithmetics.
7718 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7720 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7721 fixpoint arithmetics.
7723 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7725 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7726 fixpoint arithmetics.
7728 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7730 * profile-count.h (profile_probability::from_reg_br_prob_note,
7731 profile_probability::to_reg_br_prob_note): New functions.
7732 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7733 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7734 * predict.c (probability_reliable_p): Update.
7735 (edge_probability_reliable_p): Update.
7736 (br_prob_note_reliable_p): Update.
7737 (invert_br_probabilities): Update.
7738 (add_reg_br_prob_note): New function.
7739 (combine_predictions_for_insn): Update.
7740 * asan.c (asan_clear_shadow): Update.
7741 * cfgbuild.c (compute_outgoing_frequencies): Update.
7742 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7743 (update_br_prob_note): Update.
7744 (rtl_verify_edges): Update.
7745 (purge_dead_edges): Update.
7746 (fixup_reorder_chain): Update.
7747 * emit-rtl.c (try_split): Update.
7748 * ifcvt.c (cond_exec_process_insns): Update.
7749 (cond_exec_process_if_block): Update.
7750 (dead_or_predicable): Update.
7751 * internal-fn.c (expand_addsub_overflow): Update.
7752 (expand_neg_overflow): Update.
7753 (expand_mul_overflow): Update.
7754 * loop-doloop.c (doloop_modify): Update.
7755 * loop-unroll.c (compare_and_jump_seq): Update.
7756 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7757 * predict.h: Update.
7758 * reorg.c (mostly_true_jump): Update.
7760 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7761 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7762 * config/arc/arc.c: (emit_unlikely_jump): Update.
7763 * config/arm/arm.c: (emit_unlikely_jump): Update.
7764 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7765 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7766 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7767 (ix86_print_operand): Update.
7768 (ix86_split_fp_branch): Update.
7769 (predict_jump): Update.
7770 * config/ia64/ia64.c (ia64_print_operand): Update.
7771 * config/mmix/mmix.c (mmix_print_operand): Update.
7772 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7773 (rs6000_expand_split_stack_prologue): Update.
7774 * config/rs6000/rs6000.c: Update.
7775 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7776 (s390_expand_vec_movstr): Update.
7777 (s390_expand_cs_tdsi): Update.
7778 (s390_expand_split_stack_prologue): Update.
7779 * config/sh/sh.c (sh_print_operand): Update.
7780 (expand_cbranchsi4): Update.
7781 (expand_cbranchdi4): Update.
7782 * config/sparc/sparc.c (output_v9branch): Update.
7783 * config/spu/spu.c (get_branch_target): Update.
7784 (ea_load_store_inline): Update.
7785 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7786 * config/tilepro/tilepro.c: Update.
7788 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7790 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7791 (gimplify_save_expr): Likewise.
7793 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7795 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7797 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7799 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7801 * timevar.def (TV_IPA_FNSUMMARY): Define.
7803 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7805 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7806 to back store errata sensitive sequence from being generated.
7807 (sqrtdf2_fix): Likewise.
7809 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7811 * tree-ssa-threadupdate.c (compute_path_counts,
7812 update_joiner_offpath_counts): Use profile_probability.
7814 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7817 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7819 * config/arm/arm-c.c (arm_cpu_builtins): Define
7820 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7822 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7824 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7825 array entries to represent __ieee128 versions of the
7826 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7827 scalar_extract_sig, and scalar_insert_exp built-in functions.
7828 (altivec_resolve_overloaded_builtin): Add special case handling
7829 for the __builtin_scalar_insert_exp function, as represented by
7830 the P9V_BUILTIN_VEC_VSIEDP constant.
7831 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7832 exponent support for __ieee128 argument.
7833 (VSESQP): Add scalar extract signature support for __ieee128
7835 (VSTDCNQP): Add scalar test negative support for __ieee128
7837 (VSIEQP): Add scalar insert exponent support for __int128 argument
7838 with __ieee128 result.
7839 (VSIEQPF): Add scalar insert exponent support for __ieee128
7840 argument with __ieee128 result.
7841 (VSTDCQP): Add scalar test data class support for __ieee128
7843 (VSTDCNQP): Add overload support for scalar test negative with
7845 (VSTDCQP): Add overload support for scalar test data class
7847 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7849 (UNSPEC_VSX_SIEXPQP): New constant.
7850 (xsxexpqp): New insn for VSX scalar extract exponent quad
7852 (xsxsigqp): New insn for VSX scalar extract significand quad
7854 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7855 precision with floating point argument.
7856 (xststdcqp): New expand for VSX scalar test data class quad
7858 (xststdcnegqp): New expand for VSX scalar test negative quad
7860 (xststdcqp): New insn to match expansions for VSX scalar test data
7861 class quad precision and VSX scalar test negative quad precision.
7862 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7863 special case operand checking to enforce that second operand of
7864 VSX scalar test data class with quad precision argument is a 7-bit
7866 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7867 prototypes and descriptions of __ieee128 versions of
7868 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7869 scalar_test_data_class, and scalar_test_neg built-in functions.
7871 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7873 PR tree-optimization/81162
7874 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7875 replace a negate with an add.
7877 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7879 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7881 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7883 * config/arm/arm-c.c (arm_cpu_builtins): Define
7884 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7886 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7888 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7889 (armv8-r): Set ARM Cortex-R52 as default CPU.
7890 * config/arm/arm-tables.opt: Regenerate.
7891 * config/arm/arm-tune.md: Regenerate.
7892 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7894 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7895 extension for -mcpu=cortex-r52.
7897 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7899 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7900 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7901 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7902 (fp-armv8): Define it as FP_ARMv8 only.
7903 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7904 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7906 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7907 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7908 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7909 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7910 than TARGET_FPU_ARMV8.
7911 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7912 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7913 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7915 * config/arm/neon.md (neon_vrint): Likewise.
7916 (neon_vcvt): Likewise.
7917 (neon_<fmaxmin_op><mode>): Likewise.
7918 (<fmaxmin><mode>3): Likewise.
7919 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7920 * config/arm/predicates.md (arm_cond_move_operator): Check against
7921 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7923 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7925 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7928 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7930 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7931 loop in comment with memset.
7933 2017-07-14 Martin Liska <mliska@suse.cz>
7935 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7936 * dwarf2out.c (is_java): Remove the function.
7937 (output_pubname): Remove usage of the function.
7938 (lower_bound_default): Remove usage of DW_LANG_Java.
7939 (gen_compile_unit_die): Likewise.
7940 * gcc.c: Remove compiler defaults for .java and .zip files.
7941 * gimple-expr.c (remove_suffix): Change as there's no longer
7942 extension than 4-letter one.
7943 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7944 (gimplify_save_expr): Likewise.
7945 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7946 as it's possible even for other languages than Java.
7947 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7948 * lto-opts.c (lto_write_options): Remove reference to Java.
7949 * opts.c (strip_off_ending): Update file extension handling.
7950 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7951 * tree-eh.c (lower_resx): Likewise.
7952 * tree.c (free_lang_data_in_type): Remove dead code.
7953 (find_decls_types_r): Likewise.
7954 (build_common_builtin_nodes): Remove Java from a comment.
7955 (verify_type): Remove dead code.
7956 * varasm.c (assemble_external): Remove Java from a comment.
7958 2017-07-14 Martin Liska <mliska@suse.cz>
7960 * opts.c (finish_options): Add quotes.
7961 (common_handle_option): Likewise.
7963 2017-07-14 Martin Liska <mliska@suse.cz>
7965 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7966 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7968 * dwarf2out.c (lower_bound_default): Do not handle
7970 (gen_compile_unit_die): Likewise.
7971 * gcc.c: Remove default extension binding for GNU Pascal.
7972 * stmt.c: Remove Pascal language from a comment.
7973 * xcoffout.c: Likewise.
7975 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7978 * diagnostic-show-locus.c (fixit_cmp): New function.
7979 (layout::layout): Sort m_fixit_hints.
7980 (column_range::column_range): Assert that the values are valid.
7981 (struct char_span): New struct.
7982 (correction::overwrite): New method.
7983 (struct source_line): New struct.
7984 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7985 calls in terms of classes source_line and char_span, and
7986 correction::overwrite.
7987 (selftest::test_overlapped_fixit_printing_2): New function.
7988 (selftest::diagnostic_show_locus_c_tests): Call it.
7990 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7992 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7993 early if there is no lhs.
7995 2017-07-13 Martin Liska <mliska@suse.cz>
7997 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7998 (gen_reference_type_die): Likewise.
7999 * stor-layout.c: Remove Pascal-related comment.
8001 2017-07-13 Martin Liska <mliska@suse.cz>
8003 * opts.c (finish_options): Add quotes to error messages.
8004 (parse_sanitizer_options): Likewise.
8006 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8008 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8010 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8012 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8014 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8016 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8018 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8020 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8023 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8024 provides the hardware capability bits, define the macro
8025 __BUILTIN_CPU_SUPPORTS__.
8026 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8027 if GLIBC does not provide the hardware capability bits. Add a
8028 gcc_unreachable call if the built-in cpu function is neither
8029 __builtin_cpu_is nor __builtin_cpu_supports.
8030 (rs6000_get_function_versions_dispatcher): Change the warning
8031 that an old GLIBC is used which does not export the capability
8032 bits to be an error.
8033 * doc/extend.texi (target_clones attribute): Document the
8034 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8035 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8036 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8037 the macros defined by GCC if the newer GLIBC is available.
8039 2017-07-12 Jeff Law <law@redhat.com>
8041 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8042 remaining includes slightly.
8043 * config/riscv/riscv-builtins.c: Include profile-count.h.
8045 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8048 * config/avr/avr.c (avr_set_current_function): In diagnostic
8049 messages: Quote keywords and (parts of) identifiers.
8050 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8053 2017-07-12 Carl Love <cel@us.ibm.com>
8055 * config/rs6000/rs6000-c.c: Add support for built-in functions
8056 vector bool char vec_revb (vector bool char);
8057 vector bool short vec_revb (vector short char);
8058 vector bool int vec_revb (vector bool int);
8059 vector bool long long vec_revb (vector bool long long);
8060 * doc/extend.texi: Update the built-in documentation file for the
8061 new built-in functions.
8063 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8065 * config/s390/s390.md: Remove movcc splitter.
8067 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8069 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8070 load/store on condition.
8072 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8075 * config/avr/avr.c (avr_encode_section_info)
8076 [progmem && !TREE_READONLY]: Error if progmem object needs
8079 2017-07-11 Michael Collison <michael.collison@arm.com>
8081 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8084 2017-07-11 Carl Love <cel@us.ibm.com>
8086 * config/rs6000/rs6000-c.c: Add support for builtins
8087 vector unsigned int vec_parity_lsbb (vector signed int);
8088 vector unsigned int vec_parity_lsbb (vector unsigned int);
8089 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8090 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8091 vector unsigned long long vec_parity_lsbb (vector signed long long);
8092 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8093 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8094 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8095 * doc/extend.texi: Update the built-in documentation file for the
8096 new built-in functions.
8098 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8100 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8101 (layout::m_primary_loc): New field.
8102 (layout::layout): Initialize new field. Move location filtering
8103 logic from here to...
8104 (layout::maybe_add_location_range): ...this new method. Add
8105 support for filtering to just the lines already specified by other
8107 (layout::will_show_line_p): New method.
8108 (gcc_rich_location::add_location_if_nearby): New method.
8109 (selftest::test_add_location_if_nearby): New test function.
8110 (selftest::diagnostic_show_locus_c_tests): Call it.
8111 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8114 2017-07-11 Tom de Vries <tom@codesourcery.com>
8116 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8117 (bb_first_real_insn): New function.
8118 (nvptx_single): Add extra initialization of broadcasted condition
8121 2017-07-11 Nathan Sidwell <nathan@acm.org>
8123 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8125 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8127 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8128 Binutils doc as TEXI will mess them up.
8129 * doc/invoke.texi (AVR Options): Same here.
8131 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8133 * config/sparc/sparc.opt (mfix-ut700): New option.
8134 (mfix-gr712rc): Likewise.
8135 (sparc_fix_b2bst): New variable.
8136 * doc/invoke.texi (SPARC options): Document them.
8137 (ARM options): Fix warnings.
8138 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8139 instructions to prevent sequences that can trigger the store-store
8140 errata for certain LEON3FT processors.
8141 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8142 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8143 * config/sparc/sparc.md (fix_b2bst): New attribute.
8144 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8146 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8149 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8151 (*rsqrtsf2_sse): Ditto.
8153 (div<mode>3): Macroize insn from divdf3 and divsf3
8154 using MODEF mode iterator.
8156 2017-07-10 Martin Sebor <msebor@redhat.com>
8158 PR tree-optimization/80397
8159 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8160 instead of testing for equality to INTEGER_TYPE.
8162 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8164 * config.gcc: Remove uclibc from arc target spec.
8166 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8168 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8170 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8173 * lto-wrapper.c (remove_option): New function.
8174 (merge_and_complain): Merge PIC/PIE options more realistically.
8176 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8178 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8182 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8183 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8184 * config.in: Regenerate.
8185 * configure: Regenerate.
8186 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8187 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8188 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8189 (TARGET_GASISR_PROLOGUES): ...target mask.
8190 * common/config/avr/avr-common.c
8191 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8192 Set -mgas-isr-prologues.
8193 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8194 INSERT_PASS_BEFORE for it.
8195 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8196 * config/avr/avr.c (avr_option_override)
8197 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8198 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8199 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8200 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8201 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8203 (make_avr_pass_pre_proep): New function.
8204 (emit_push_sfr) <treg>: Add argument to function and use it
8206 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8207 and set machine->gasisr.yes.
8208 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8209 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8210 __gcc_isr.n_pushed to .L__stack_usage.
8211 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8212 (avr_asm_final_postscan_insn): ...this new static function.
8213 * config/avr/avr.h (machine_function)
8214 <is_no_gccisr, use_L__stack_usage>: New fields.
8215 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8216 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8217 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8218 (gasisr, *gasisr): New expander and insn.
8219 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8220 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8221 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8223 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8225 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8228 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8230 Move jump-tables out of .text again.
8233 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8234 (ASM_OUTPUT_ADDR_VEC): New function.
8235 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8236 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8237 INSN_ADDRESSes as asm comment.
8238 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8239 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8240 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8241 * config/avr/avr.md (*tablejump): Adjust comment.
8242 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8243 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8245 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8246 (avr_output_addr_vec): New proto.
8247 (avr_log_t) <insn_addresses>: New field.
8249 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8252 * config/i386/i386.c (ix86_function_arg_advance): Set
8253 outgoing_args_on_stack to true if there are outgoing arguments
8255 (ix86_function_arg): Likewise.
8256 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8257 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8258 * config/i386/i386.h (machine_function): Add
8259 outgoing_args_on_stack.
8261 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8263 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8265 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8267 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8269 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8270 (REAL_H): Remove $(MACHMODE_H).
8271 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8273 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8274 $(MACHMODE_H) and double-int.h.
8275 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8277 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8280 2017-07-07 Andrew Pinski <apinski@cavium.com>
8282 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8283 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8285 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8287 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8288 Add warning if GCC was not configured to link against a GLIBC that
8289 exports the hardware capability bits.
8290 (make_resolver_func): Make resolver function private and not a
8291 COMDAT function. Create the name with clone_function_name instead
8292 of make_unique_name.
8295 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8296 correct operand in doing the split.
8298 2017-07-07 Carl Love <cel@us.ibm.com>
8300 * config/rs6000/rs6000-c: Add support for built-in function
8301 vector unsigned short vec_pack_to_short_fp32 (vector float,
8303 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8304 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8305 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8306 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8307 (convert_4f32_8i16): Add define_expand.
8308 * doc/extend.texi: Update the built-in documentation file for the
8309 new built-in function.
8311 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8313 * config/sparc/m8.md: New file.
8314 * config/sparc/sparc.md: Include m8.md.
8316 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8318 * config/sparc/sparc.opt: New option -mvis4b.
8319 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8320 (sparc_option_override): Handle VIS4B.
8321 (enum sparc_builtins): Define
8322 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8323 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8324 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8325 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8326 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8327 (check_constant_argument): New function.
8328 (sparc_vis_init_builtins): Define builtins
8329 __builtin_vis_dictunpack{8,16,32},
8330 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8331 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8332 __builtin_vis_fpcmpde{8,16,32}shl and
8333 __builtin_vis_fpcmpur{8,16,32}shl.
8334 (sparc_expand_builtin): Check that the constant operands to
8335 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8336 constant and in range.
8337 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8339 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8340 (SPARC_IMM5_P): Likewise.
8341 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8342 (enabled): Handle vis4b.
8343 (UNSPEC_DICTUNPACK): New unspec.
8344 (UNSPEC_FPCMPSHL): Likewise.
8345 (UNSPEC_FPUCMPSHL): Likewise.
8346 (UNSPEC_FPCMPDESHL): Likewise.
8347 (UNSPEC_FPCMPURSHL): Likewise.
8348 (cpu_feature): New CPU feature `vis4b'.
8349 (dictunpack{8,16,32}): New insns.
8350 (FPCSMODE): New mode iterator.
8351 (fpcscond): New code iterator.
8352 (fpcsucond): Likewise.
8353 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8354 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8355 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8356 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8357 * config/sparc/constraints.md: Define constraints `q' for unsigned
8358 2-bit integer constants and `t' for unsigned 5-bit integer
8360 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8362 (imm5_operand_dictunpack16): Likewise.
8363 (imm5_operand_dictunpack32): Likewise.
8364 (imm2_operand): Likewise.
8365 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8366 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8367 ditunpack* and fpcmp*shl builtins.
8369 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8371 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8372 * config.in: Add HAVE_AS_SPARC6 define.
8373 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8375 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8377 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8378 (CPP_CPU_SPEC): Handle m8.
8379 (ASM_CPU_SPEC): Likewise.
8380 * config/sparc/sparc-opts.h (enum processor_type): Add
8382 * config/sparc/sparc.c (m8_costs): New struct.
8383 (sparc_option_override): Handle TARGET_CPU_m8.
8384 (sparc32_initialize_trampoline): Likewise.
8385 (sparc64_initialize_trampoline): Likewise.
8386 (sparc_issue_rate): Likewise.
8387 (sparc_register_move_cost): Likewise.
8388 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8389 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8390 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8391 (CPP_CPU_SPEC): Handle M8.
8392 (ASM_CPU_SPEC): Likewise.
8393 (AS_M8_FLAG): Define.
8394 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8395 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8396 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8398 * configure: Regenerate.
8399 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8402 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8404 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8406 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8407 ("*movdi_insn_sp32"): Do not set v3pipe.
8408 ("*movsi_insn"): Likewise.
8409 ("*movdi_insn_sp64"): Likewise.
8410 ("*movsf_insn"): Likewise.
8411 ("*movdf_insn_sp32"): Likewise.
8412 ("*movdf_insn_sp64"): Likewise.
8413 ("*zero_extendsidi2_insn_sp64"): Likewise.
8414 ("*sign_extendsidi2_insn"): Likewise.
8415 ("*mov<VM32:mode>_insn"): Likewise.
8416 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8417 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8418 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8419 ("<vlop:code><VL:mode>3"): Likewise.
8420 ("*not_<vlop:code><VL:mode>3"): Likewise.
8421 ("*nand<VL:mode>_vis"): Likewise.
8422 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8423 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8424 ("one_cmpl<VL:mode>2"): Likewise.
8425 ("faligndata<VM64:mode>_vis"): Likewise.
8426 ("alignaddrsi_vis"): Likewise.
8427 ("alignaddrdi_vis"): Likweise.
8428 ("alignaddrlsi_vis"): Likewise.
8429 ("alignaddrldi_vis"): Likewise.
8430 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8431 ("bmaskdi_vis"): Likewise.
8432 ("bmasksi_vis"): Likewise.
8433 ("bshuffle<VM64:mode>_vis"): Likewise.
8434 ("cmask8<P:mode>_vis"): Likewise.
8435 ("cmask16<P:mode>_vis"): Likewise.
8436 ("cmask32<P:mode>_vis"): Likewise.
8437 ("pdistn<P:mode>_vis"): Likewise.
8438 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8440 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8442 * config/sparc/sparc.md ("subtype"): New insn attribute.
8443 ("*wrgsr_sp64"): Set insn subtype.
8444 ("*rdgsr_sp64"): Likewise.
8445 ("alignaddrsi_vis"): Likewise.
8446 ("alignaddrdi_vis"): Likewise.
8447 ("alignaddrlsi_vis"): Likewise.
8448 ("alignaddrldi_vis"): Likewise.
8449 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8450 ("fexpand_vis"): Likewise.
8451 ("fpmerge_vis"): Likewise.
8452 ("faligndata<VM64:mode>_vis"): Likewise.
8453 ("bshuffle<VM64:mode>_vis"): Likewise.
8454 ("cmask8<P:mode>_vis"): Likewise.
8455 ("cmask16<P:mode>_vis"): Likewise.
8456 ("cmask32<P:mode>_vis"): Likewise.
8457 ("fchksm16_vis"): Likewise.
8458 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8459 ("fmean16_vis"): Likewise.
8460 ("fp<plusminus_insn>64_vis"): Likewise.
8461 ("<plusminus_insn>v8qi3"): Likewise.
8462 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8463 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8464 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8465 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8466 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8467 ("*movqi_insn"): Likewise.
8468 ("*movhi_insn"): Likewise.
8469 ("*movsi_insn"): Likewise.
8470 ("movsi_pic_gotdata_op"): Likewise.
8471 ("*movdi_insn_sp32"): Likewise.
8472 ("*movdi_insn_sp64"): Likewise.
8473 ("movdi_pic_gotdata_op"): Likewise.
8474 ("*movsf_insn"): Likewise.
8475 ("*movdf_insn_sp32"): Likewise.
8476 ("*movdf_insn_sp64"): Likewise.
8477 ("*zero_extendhisi2_insn"): Likewise.
8478 ("*zero_extendqihi2_insn"): Likewise.
8479 ("*zero_extendqisi2_insn"): Likewise.
8480 ("*zero_extendqidi2_insn"): Likewise.
8481 ("*zero_extendhidi2_insn"): Likewise.
8482 ("*zero_extendsidi2_insn_sp64"): Likewise.
8483 ("ldfsr"): Likewise.
8484 ("prefetch_64"): Likewise.
8485 ("prefetch_32"): Likewise.
8486 ("tie_ld32"): Likewise.
8487 ("tie_ld64"): Likewise.
8488 ("*tldo_ldub_sp32"): Likewise.
8489 ("*tldo_ldub1_sp32"): Likewise.
8490 ("*tldo_ldub2_sp32"): Likewise.
8491 ("*tldo_ldub_sp64"): Likewise.
8492 ("*tldo_ldub1_sp64"): Likewise.
8493 ("*tldo_ldub2_sp64"): Likewise.
8494 ("*tldo_ldub3_sp64"): Likewise.
8495 ("*tldo_lduh_sp32"): Likewise.
8496 ("*tldo_lduh1_sp32"): Likewise.
8497 ("*tldo_lduh_sp64"): Likewise.
8498 ("*tldo_lduh1_sp64"): Likewise.
8499 ("*tldo_lduh2_sp64"): Likewise.
8500 ("*tldo_lduw_sp32"): Likewise.
8501 ("*tldo_lduw_sp64"): Likewise.
8502 ("*tldo_lduw1_sp64"): Likewise.
8503 ("*tldo_ldx_sp64"): Likewise.
8504 ("*mov<VM32:mode>_insn"): Likewise.
8505 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8506 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8508 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8510 * config/sparc/sparc.md ("type"): New insn type viscmp.
8511 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8513 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8514 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8515 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8516 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8518 ("n7_vis_logical_11cycle"): Likewise.
8519 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8520 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8521 * config/sparc/niagara.md ("niag_vis"): Likewise.
8522 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8523 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8525 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8527 * config/sparc/sparc.md: New instruction type `bmask'.
8528 (bmaskdi_vis): Use the `bmask' type.
8529 (bmasksi_vis): Likewise.
8530 * config/sparc/ultra3.md (us3_array): Likewise.
8531 * config/sparc/niagara7.md (n7_array): Likewise.
8532 * config/sparc/niagara4.md (n4_array): Likewise.
8533 * config/sparc/niagara2.md (niag2_vis): Likewise.
8534 (niag3_vis): Likewise.
8535 * config/sparc/niagara.md (niag_vis): Likewise.
8537 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8539 * ipa-comdats.c: Remove optimize check from gate.
8540 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8541 for functions not optimized.
8542 (ipa_fn_summary_read): Skip optimize check.
8543 (ipa_fn_summary_write): Likewise.
8544 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8546 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8548 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8549 (check_callers): Give up on caller which is not optimized.
8550 (inline_small_functions): Likewise.
8551 (ipa_inline): Do not give up when not optimizing.
8552 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8553 away unoptimizes cdtors.
8554 (whole_program_function_and_variable_visibility): Do
8555 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8556 * ipa.c (process_references): Do not check optimize.
8557 (symbol_table::remove_unreachable_nodes): Update optimize check.
8558 (set_writeonly_bit): Update optimize check.
8559 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8560 (pass_ipa_single_use::gate): Remove.
8562 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8564 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8565 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8566 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8567 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8568 permute_load, permute_store, adjust_extract, adjust_splat,
8569 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8570 replace_swap_with_copy, dump_swap_insn_table,
8571 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8572 recombine_lvx_pattern, recombine_stvx_pattern,
8573 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8574 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8575 to file rs6000-p8swap.c.
8576 * config/rs6000/rs6000-p8swap.c: New file.
8577 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8578 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8579 and rs6000*-*-* targets.
8581 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8583 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8585 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8587 * lto-wrapper.c (merge_and_complain): Do not merge
8588 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8589 fsigned_zeros, ftrapping_math, fwrapv.
8590 (append_compiler_options): Do not track these options.
8591 (append_linker_options): Likewie
8593 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8595 * cgraphunit.c (cgraph_node::finalize_function): When
8596 !flag_toplevel_reorde set no_reorder flag.
8597 (varpool_node::finalize_decl): Likewise.
8598 (symbol_table::compile): Drop no toplevel reorder path.
8600 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8602 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8603 edges; zero probability is not better than uninitialized.
8605 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8607 * asan.h (asan_sanitize_allocas_p): Declare.
8608 * asan.c (asan_sanitize_allocas_p): New function.
8609 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8610 (handle_builtin_alloca): Likewise.
8611 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8612 if !asan_sanitize_allocas_p.
8613 * params.def (asan-instrument-allocas): Add new option.
8614 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8615 * opts.c (common_handle_option): Disable allocas sanitization for
8618 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8620 * asan.c: Include gimple-fold.h.
8621 (get_last_alloca_addr): New function.
8622 (handle_builtin_stackrestore): Likewise.
8623 (handle_builtin_alloca): Likewise.
8624 (asan_emit_allocas_unpoison): Likewise.
8625 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8626 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8627 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8628 (instrument_builtin_call): Pass gimple iterator to
8629 get_mem_refs_of_builtin_call.
8630 (last_alloca_addr): New global.
8631 * asan.h (asan_emit_allocas_unpoison): Declare.
8632 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8633 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8634 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8635 if function calls alloca.
8636 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8637 * gimple-fold.h (replace_call_with_value): Declare.
8638 * internal-fn.c: Include asan.h.
8639 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8640 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8642 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8644 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8645 (C_SELFTEST_FLAGS): New.
8646 (CPP_SELFTEST_FLAGS): New.
8647 (SELFTEST_DEPS): New, from deps of s-selftest.
8648 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8649 (CPP_SELFTEST_DEPS): New.
8650 (selftest): Add dependency on s-selftest-c++.
8651 (s-selftest): Rename to...
8652 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8653 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8654 than SELFTEST_FLAGS.
8655 (selftest-gdb): Rename to...
8656 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8658 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8659 (selftest-valgrind): Rename to...
8660 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8662 (selftest-valgrind): Reintroduce as an alias for
8663 selftest-c-valgrind.
8664 (s-selftest-c++): New.
8665 (selftest-c++-gdb): New.
8666 (selftest-c++-valgrind): New.
8668 2017-07-06 Olivier Hainque <hainque@adacore.com>
8670 * gcc.c (process_command): When deciding if undefined variables
8671 should be ignored when processing specs, accept "gcc -v" as well.
8673 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8675 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8676 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8678 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8680 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8681 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8682 * config/arm/arm-tables.opt: Regenerate.
8683 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8685 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8687 2017-07-06 Carl Love <cel@us.ibm.com>
8689 * ChangeLog: Clean up from mid air collision
8691 2017-07-06 Carl Love <cel@us.ibm.com>
8693 * config/rs6000/rs6000-c.c: Add support for built-in functions
8694 vector signed int vec_subc (vector signed int, vector signed int);
8695 vector signed __int128 vec_subc (vector signed __int128,
8696 vector signed __int128);
8697 vector unsigned __int128 vec_subc (vector unsigned __int128,
8698 vector unsigned __int128);
8699 vector signed int vec_sube (vector signed int, vector signed int,
8701 vector unsigned int vec_sube (vector unsigned int,
8702 vector unsigned int,
8703 vector unsigned int);
8704 vector signed __int128 vec_sube (vector signed __int128,
8705 vector signed __int128,
8706 vector signed__int128);
8707 vector unsigned __int128 vec_sube (vector unsigned __int128,
8708 vector unsigned __int128,
8709 vector unsigned __int128);
8710 vector signed int vec_subec (vector signed int, vector signed int,
8712 vector unsigned int vec_subec (vector unsigned int,
8713 vector unsigned int,
8714 vector unsigned int);
8715 vector signed __int128 vec_subec (vector signed __int128,
8716 vector signed __int128,
8717 vector signed__int128);
8718 vector unsigned __int128 vec_subec (vector unsigned __int128,
8719 vector unsigned __int128,
8720 vector unsigned __int128);
8721 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8722 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8723 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8724 BU_ALTIVEC_OVERLOAD_X definitions.
8725 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8726 * doc/extend.texi: Update the built-in documentation file for the new
8729 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8732 * diagnostic-show-locus.c (layout::layout): Use start and finish
8733 spelling location for the start and finish of each range.
8734 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8735 Add unused aspect param.
8736 * input.c (expand_location_1): Add "aspect" param, and use it
8737 to access the correct part of the location.
8738 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8740 (expand_location_to_spelling_point): Likewise.
8741 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8742 param, and pass it to expand_location_1.
8744 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8746 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8747 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8748 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8749 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8750 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8751 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8752 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8753 _mm_maskz_getmant_ss): New intrinsics.
8754 (__builtin_ia32_getexpss128_mask): Changed to ...
8755 __builtin_ia32_getexpss128_round ... this.
8756 (__builtin_ia32_getexpsd128_mask): Changed to ...
8757 __builtin_ia32_getexpsd128_round ... this.
8758 * config/i386/i386-builtin-types.def
8759 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8760 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8761 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8762 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8763 __builtin_ia32_getmantss_mask_round): New builtins.
8764 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8765 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8766 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8767 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8768 * config/i386/sse.md
8769 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8770 avx512f_sgetexp<mode><mask_scalar_name>
8771 <round_saeonly_scalar_name> ... this.
8772 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8773 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8774 vgetexp<ssescalarmodesuffix>
8775 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8776 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8777 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8778 avx512f_vgetmant<mode><mask_scalar_name>
8779 <round_saeonly_scalar_name> ... this.
8780 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8781 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8782 vgetmant<ssescalarmodesuffix>
8783 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8784 %0<mask_scalar_operand4>, %1, %2
8785 <round_saeonly_scalar_mask_op4>, %3} ... this.
8786 * config/i386/subst.md (mask_scalar_operand4,
8787 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8788 round_saeonly_scalar_nimm_predicate): New subst attributes.
8790 2017-07-06 Julia Koval <julia.koval@intel.com>
8792 * config/i386/i386.c (ix86_erase_embedded_rounding):
8793 Remove code for old rounding pattern.
8795 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8797 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8799 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8801 * doc/sourcebuild.texi (Test Directives, Variants of
8802 dg-require-support): Add documentation for dg-require-stack-check.
8804 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8806 * config/i386/subst.md (mask_scalar, round_scalar,
8807 round_saeonly_scalar): New meta-templates.
8808 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8809 round_scalar_mask_operand3, round_scalar_mask_op3,
8810 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8811 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8812 round_saeonly_scalar_constraint,
8813 round_saeonly_scalar_prefix): New subst attribute.
8814 * config/i386/sse.md
8815 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8816 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8817 <round_scalar_name> ... this.
8818 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8819 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8820 <round_scalar_name> ... this.
8821 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8822 <sse>_vm<code><mode>3<mask_scalar_name>
8823 <round_saeonly_scalar_name> ... this.
8824 (v<plusminus_mnemonic><ssescalarmodesuffix>
8825 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8826 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8827 v<plusminus_mnemonic><ssescalarmodesuffix>
8828 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8829 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8830 (v<multdiv_mnemonic><ssescalarmodesuffix>
8831 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8832 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8833 v<multdiv_mnemonic><ssescalarmodesuffix>
8834 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8835 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8836 (v<maxmin_float><ssescalarmodesuffix>
8837 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8838 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8839 v<maxmin_float><ssescalarmodesuffix>
8840 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8841 %0<mask_scalar_operand3>, %1, %<iptr>2
8842 <round_saeonly_scalar_mask_op3>} ... this.
8844 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8846 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8847 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8849 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8850 Alan Hayward <alan.hayward@arm.com>
8851 David Sherwood <david.sherwood@arm.com>
8853 * combine.c (simplify_if_then_else): Remove "enum" before
8855 * compare-elim.c (can_eliminate_compare): Likewise.
8856 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8858 (aarch64_lookup_simd_builtin_type): Likewise.
8859 (aarch64_simd_builtin_type): Likewise.
8860 (aarch64_init_simd_builtin_types): Likewise.
8861 (aarch64_simd_expand_args): Likewise.
8862 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8864 (aarch64_reverse_mask): Likewise.
8865 (aarch64_simd_emit_reg_reg_move): Likewise.
8866 (aarch64_gen_adjusted_ldpstp): Likewise.
8867 (aarch64_ccmp_mode_to_code): Likewise.
8868 (aarch64_operands_ok_for_ldpstp): Likewise.
8869 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8870 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8872 (aarch64_min_divisions_for_recip_mul): Likewise.
8873 (aarch64_reassociation_width): Likewise.
8874 (aarch64_get_condition_code_1): Likewise.
8875 (aarch64_simd_emit_reg_reg_move): Likewise.
8876 (aarch64_simd_attr_length_rglist): Likewise.
8877 (aarch64_reverse_mask): Likewise.
8878 (aarch64_operands_ok_for_ldpstp): Likewise.
8879 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8880 (aarch64_gen_adjusted_ldpstp): Likewise.
8881 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8883 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8884 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8885 (arm_lookup_simd_builtin_type): Likewise.
8886 (arm_simd_builtin_type): Likewise.
8887 (arm_init_simd_builtin_types): Likewise.
8888 (arm_expand_builtin_args): Likewise.
8889 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8890 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8891 (ft32_setup_incoming_varargs): Likewise.
8892 (ft32_function_arg): Likewise.
8893 (ft32_function_arg_advance): Likewise.
8894 (ft32_pass_by_reference): Likewise.
8895 (ft32_arg_partial_bytes): Likewise.
8896 (ft32_valid_pointer_mode): Likewise.
8897 (ft32_addr_space_pointer_mode): Likewise.
8898 (ft32_addr_space_legitimate_address_p): Likewise.
8899 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8901 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8902 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8903 (iamcu_alignment): Likewise.
8904 (canonicalize_vector_int_perm): Likewise.
8905 (ix86_noce_conversion_profitable_p): Likewise.
8906 (ix86_mpx_bound_mode): Likewise.
8907 (ix86_operands_ok_for_move_multiple): Likewise.
8908 * config/microblaze/microblaze-protos.h
8909 (microblaze_expand_conditional_branch_reg): Likewise.
8910 * config/microblaze/microblaze.c
8911 (microblaze_expand_conditional_branch_reg): Likewise.
8912 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8914 (rs6000_reassociation_width): Likewise.
8915 (rs6000_invalid_binary_op): Likewise.
8916 (fusion_p9_p): Likewise.
8917 (emit_fusion_p9_load): Likewise.
8918 (emit_fusion_p9_store): Likewise.
8919 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8921 (riscv_hard_regno_mode_ok_p): Likewise.
8922 (riscv_address_insns): Likewise.
8923 (riscv_split_symbol): Likewise.
8924 (riscv_legitimize_move): Likewise.
8925 (riscv_function_value): Likewise.
8926 (riscv_hard_regno_nregs): Likewise.
8927 (riscv_expand_builtin): Likewise.
8928 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8929 (riscv_build_integer): Likewise.
8930 (riscv_split_integer): Likewise.
8931 (riscv_legitimate_constant_p): Likewise.
8932 (riscv_cannot_force_const_mem): Likewise.
8933 (riscv_regno_mode_ok_for_base_p): Likewise.
8934 (riscv_valid_base_register_p): Likewise.
8935 (riscv_valid_offset_p): Likewise.
8936 (riscv_valid_lo_sum_p): Likewise.
8937 (riscv_classify_address): Likewise.
8938 (riscv_legitimate_address_p): Likewise.
8939 (riscv_address_insns): Likewise.
8940 (riscv_load_store_insns): Likewise.
8941 (riscv_force_binary): Likewise.
8942 (riscv_split_symbol): Likewise.
8943 (riscv_force_address): Likewise.
8944 (riscv_legitimize_address): Likewise.
8945 (riscv_move_integer): Likewise.
8946 (riscv_legitimize_const_move): Likewise.
8947 (riscv_legitimize_move): Likewise.
8948 (riscv_address_cost): Likewise.
8949 (riscv_subword): Likewise.
8950 (riscv_output_move): Likewise.
8951 (riscv_canonicalize_int_order_test): Likewise.
8952 (riscv_emit_int_order_test): Likewise.
8953 (riscv_function_arg_boundary): Likewise.
8954 (riscv_pass_mode_in_fpr_p): Likewise.
8955 (riscv_pass_fpr_single): Likewise.
8956 (riscv_pass_fpr_pair): Likewise.
8957 (riscv_get_arg_info): Likewise.
8958 (riscv_function_arg): Likewise.
8959 (riscv_function_arg_advance): Likewise.
8960 (riscv_arg_partial_bytes): Likewise.
8961 (riscv_function_value): Likewise.
8962 (riscv_pass_by_reference): Likewise.
8963 (riscv_setup_incoming_varargs): Likewise.
8964 (riscv_print_operand): Likewise.
8965 (riscv_elf_select_rtx_section): Likewise.
8966 (riscv_save_restore_reg): Likewise.
8967 (riscv_for_each_saved_reg): Likewise.
8968 (riscv_register_move_cost): Likewise.
8969 (riscv_hard_regno_mode_ok_p): Likewise.
8970 (riscv_hard_regno_nregs): Likewise.
8971 (riscv_class_max_nregs): Likewise.
8972 (riscv_memory_move_cost): Likewise.
8973 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8974 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8975 (rl78_addr_space_address_mode): Likewise.
8976 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8978 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8979 (rs6000_reassociation_width): Likewise.
8980 (rs6000_invalid_binary_op): Likewise.
8981 (fusion_p9_p): Likewise.
8982 (emit_fusion_p9_load): Likewise.
8983 (emit_fusion_p9_store): Likewise.
8984 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8985 (ok_for_simple_move_operands): Likewise.
8986 (ok_for_simple_move_strict_operands): Likewise.
8987 (ok_for_simple_arith_logic_operands): Likewise.
8988 (visium_legitimize_reload_address): Likewise.
8989 (visium_select_cc_mode): Likewise.
8990 (output_cbranch): Likewise.
8991 (visium_split_double_move): Likewise.
8992 (visium_expand_copysign): Likewise.
8993 (visium_expand_int_cstore): Likewise.
8994 (visium_expand_fp_cstore): Likewise.
8995 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8996 (visium_function_arg): Likewise.
8997 (visium_function_arg_advance): Likewise.
8998 (visium_libcall_value): Likewise.
8999 (visium_setup_incoming_varargs): Likewise.
9000 (visium_legitimate_constant_p): Likewise.
9001 (visium_legitimate_address_p): Likewise.
9002 (visium_legitimize_address): Likewise.
9003 (visium_secondary_reload): Likewise.
9004 (visium_register_move_cost): Likewise.
9005 (visium_memory_move_cost): Likewise.
9006 (prepare_move_operands): Likewise.
9007 (ok_for_simple_move_operands): Likewise.
9008 (ok_for_simple_move_strict_operands): Likewise.
9009 (ok_for_simple_arith_logic_operands): Likewise.
9010 (visium_function_value_1): Likewise.
9011 (rtx_ok_for_offset_p): Likewise.
9012 (visium_legitimize_reload_address): Likewise.
9013 (visium_split_double_move): Likewise.
9014 (visium_expand_copysign): Likewise.
9015 (visium_expand_int_cstore): Likewise.
9016 (visium_expand_fp_cstore): Likewise.
9017 (visium_split_cstore): Likewise.
9018 (visium_select_cc_mode): Likewise.
9019 (visium_split_cbranch): Likewise.
9020 (output_cbranch): Likewise.
9021 (visium_print_operand_address): Likewise.
9022 * expmed.c (flip_storage_order): Likewise.
9023 * expmed.h (emit_cstore): Likewise.
9024 (flip_storage_order): Likewise.
9025 * genrecog.c (validate_pattern): Likewise.
9026 * hsa-gen.c (gen_hsa_addr): Likewise.
9027 * internal-fn.c (expand_arith_overflow): Likewise.
9028 * ira-color.c (allocno_copy_cost_saving): Likewise.
9029 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9030 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9031 (process_invariant_for_inheritance): Likewise.
9032 * lra-eliminations.c (move_plus_up): Likewise.
9033 * omp-low.c (lower_oacc_reductions): Likewise.
9034 * simplify-rtx.c (simplify_subreg): Likewise.
9035 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9036 (TARGET_CHKP_BOUND_MODE): Likewise..
9037 * targhooks.c (default_chkp_bound_mode): Likewise.
9038 (default_setup_incoming_vararg_bounds): Likewise.
9039 * targhooks.h (default_chkp_bound_mode): Likewise.
9040 (default_setup_incoming_vararg_bounds): Likewise.
9041 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9042 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9043 (have_whole_vector_shift): Likewise.
9044 * tree-vect-stmts.c (vectorizable_load): Likewise.
9045 * doc/tm.texi: Regenerate.
9047 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9049 Graceful degrade if Binutils PR21472 is not available.
9052 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9053 .rodata in flash test fails.
9054 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9055 * confgure: Regenerate.
9056 * config.in: Regenerate.
9057 * config/avr/avr.c (avr_asm_named_section)
9058 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9059 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9060 (avr_asm_init_sections): Same.
9062 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9064 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9065 (fma<VH:mode>4_intrinsic): Likewise.
9066 (*fmsub<VCVTF:mode>4): Likewise.
9067 (*fmsub<VH:mode>4_intrinsic): Likewise.
9069 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9072 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9073 Don't depend on "optimize > 0".
9074 (out_movhi_r_mr, out_movqi_mr_r): Same.
9075 (out_movhi_mr_r, out_movqi_r_mr): Same.
9076 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9077 io_address_operand on "optimize > 0".
9079 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9081 * tree-loop-distribution.c: Add general explanantion on the pass.
9082 (generate_loops_for_partition): Mark distributed loop.
9083 (pg_add_dependence_edges): New parameter. Handle alias data
9084 dependence specially and record it in the parameter if asked.
9085 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9086 (init_partition_graph_vertices, add_partition_graph_edge): New.
9087 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9088 (free_partition_graph_vdata, build_partition_graph): New.
9089 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9090 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9091 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9092 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9093 (version_for_distribution_p, finalize_partitions): New.
9094 (distribute_loop): Handle alias data dependence specially. Factor
9095 out loop fusion code as functions and call these functions.
9097 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9099 * tree-loop-distribution.c (classify_partition): New parameter and
9100 better handle reduction statement.
9101 (rdg_build_partitions): Revise comment.
9102 (distribute_loop): Compute statements in all partitions and pass it
9103 to classify_partition.
9105 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9107 * tree-loop-distribution.c (enum partition_type): New.
9108 (struct partition): New field type.
9109 (partition_merge_into): Add parameter. Update partition type.
9110 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9111 (build_rdg_partition_for_vertex): Compute partition type.
9112 (rdg_build_partitions): Dump partition type.
9113 (distribute_loop): Update calls to partition_merge_into.
9115 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9117 * tree-loop-distribution.c (struct ddr_hasher): New.
9118 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9120 (classify_partition): Call get_data_dependence.
9121 (pg_add_dependence_edges): Ditto.
9122 (distribute_loop): Release data dependence hash table.
9124 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9126 * tree-loop-distribution.c (ref_base_address): Delete.
9127 (similar_memory_accesses): Rename ...
9128 (share_memory_accesses): ... to this. Check if partitions access
9129 the same memory reference.
9130 (distribute_loop): Call share_memory_accesses.
9132 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9134 * tree-loop-distribution.c (struct partition): New field recording
9136 (partition_alloc, partition_free): Init and release data refs.
9137 (partition_merge_into): Merge data refs.
9138 (build_rdg_partition_for_vertex): Collect data refs for partition.
9139 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9141 (distribute_loop): Remve data refs from vertice data of partition
9144 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9146 * tree-loop-distribution.c (params.h): Include header file.
9147 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9148 (datarefs_vec): New global var.
9149 (create_rdg_vertices): Use datarefs_vec directly.
9150 (free_rdg): Don't free data references.
9151 (build_rdg): Update use. Don't free data references.
9152 (distribute_loop): Compute global variable for data references.
9153 Bail out if there are too many data references.
9155 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9157 * tree-loop-distribution.c (loop_nest): New global var.
9158 (build_rdg): Use loop directly, rather than loop nest.
9159 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9161 (distribute_loop): Compute global variable loop nest. Update use.
9163 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9165 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9166 (partition_merge_into): New parameter. Dump reason for fusion.
9167 (distribute_loop): Update use of partition_merge_into.
9169 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9171 * tree-loop-distribution.c (bb_top_order_index): New.
9172 (bb_top_order_index_size, bb_top_order_cmp): New.
9173 (stmts_from_loop): Use topological order.
9174 (pass_loop_distribution::execute): Compute and release topological
9175 order for basic blocks.
9177 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9179 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9182 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9184 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9185 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9186 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9187 * internal-fn.def (LOOP_DIST_ALIAS): New.
9188 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9189 (fold_loop_internal_call): ... this.
9190 (vect_loop_dist_alias_call): New function.
9191 (set_uid_loop_bbs): Call fold_loop_internal_call.
9192 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9195 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9198 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9199 Require dead FLAGS_REG at the beginning of a peephole.
9201 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9204 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9205 arguments in the call to __builtin_ia32_sbb_u32.
9206 (_subborrow_u64): Swap _X and _Y arguments in the call to
9207 __builtin_ia32_sbb_u64.
9209 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9212 * tree-vrp.c (compare_assert_loc): Turn into a function template
9213 with stable template parameter. Only test if a->e is NULL,
9214 !a->e == !b->e has been verified already. Use e == NULL or
9215 e != NULL instead of e or ! e tests. If stable is true, don't use
9216 iterative_hash_expr, on the other side allow a or b or both NULL
9217 and sort the NULLs last.
9218 (process_assert_insertions): Sort using compare_assert_loc<false>
9219 instead of compare_assert_loc, later sort using
9220 compare_assert_loc<true> before calling process_assert_insertions_for
9221 in a loop. Use break instead of continue once seen NULL pointer.
9223 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9225 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9226 Cortex-R7 and Cortex-R8 processors.
9228 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9230 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9231 uninitialized while src is not.
9233 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9235 * common/config/arm/arm-common.c: Adjust include path for
9237 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9238 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9239 (arm-cpu-data.h): Likewise.
9240 (arm-cpu-cdata.h): Likewise.
9241 * config/arm/arm-cpu.h: Delete.
9242 * config/arm/arm-cpu-cdata.h: Delete.
9243 * config/arm/arm-cpu-data.h: Delete.
9245 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9247 * config/arm/arm-cpus.in (cortex-a55): New.
9248 (cortex-a75): Likewise.
9249 (cortex-a75.cortex-a55): Likewise.
9250 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9252 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9253 * config/arm/arm-cpu-cdata.h: Regenerate.
9254 * config/arm/arm-cpu-data.h: Regenerate.
9255 * config/arm/arm-cpu.h: Regenerate.
9256 * config/arm/arm-tables.opt: Regenerate.
9257 * config/arm/arm-tune.md: Regenerate.
9259 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9261 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9263 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9265 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9268 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9270 PR tree-optimization/81292
9271 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9272 full_string_p, also call adjust_related_strinfos if the adjustment
9273 is simple, otherwise invalidate related strinfos.
9275 2017-07-04 Martin Liska <mliska@suse.cz>
9278 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9279 newly created variable as DECL_IGNORED_P.
9281 2017-07-04 Martin Liska <mliska@suse.cz>
9284 * ipa-inline.c (inline_small_functions):
9285 Use xstrdup_for_dump.
9287 2017-07-04 Tom de Vries <tom@codesourcery.com>
9289 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9291 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9294 * config/darwin.c (darwin_function_switched_text_sections):
9297 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9299 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9301 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9303 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9305 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9307 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9308 min_profitable_iters, and th as inclusive lower bounds.
9309 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9310 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9311 for min_profitable_iters and min_profitable_estimate.
9312 (vect_transform_loop): Treat th as an inclusive lower bound.
9313 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9315 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9318 * config/darwin.c (darwin_function_switched_text_sections):
9319 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9320 in two pieces, and suppress the use of buf.
9322 2017-07-03 Nathan Sidwell <nathan@acm.org>
9324 * hash-table.h (hash_table_mod1): Fix indentation.
9326 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9329 * predict.c (force_edge_cold): Be more careful about propagation
9331 * profile-count.h (profile_probability::guessed,
9332 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9334 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9336 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9338 * doc/invoke.texi (rcpc architecture extension): Document it.
9340 2017-07-03 Richard Biener <rguenther@suse.de>
9342 PR tree-optimization/60510
9343 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9344 the scalar reduction PHI and use it.
9345 (vectorizable_reduction): Properly guard the single_defuse_cycle
9346 path for non-SLP reduction chains where we cannot use it.
9347 Rework reduc_def/index and vector type deduction. Rework
9348 vector operand gathering during reduction op code-gen.
9349 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9350 chains dissolve the chain and leave it to non-SLP reduction
9353 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9355 * tree-data-ref.h (dr_alignment): Declare.
9356 * tree-data-ref.c (dr_alignment): New function.
9357 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9358 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9360 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9362 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9364 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9365 and base_misalignment fields.
9366 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9367 * tree-data-ref.c: Include builtins.h.
9368 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9369 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9370 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9371 * tree-vect-data-refs.c: Include tree-cfg.h.
9372 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9373 fields instead of calculating an alignment here.
9374 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9375 innermost_loop_behavior fields.
9377 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9379 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9381 (DR_STEP_ALIGNMENT): New macro.
9382 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9383 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9384 (create_data_ref): Print it.
9385 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9386 to tell whether the step preserves vector (mis)alignment.
9387 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9388 Move the check for an integer step and generalise to all INTEGER_CST.
9389 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9390 Print the outer step alignment.
9392 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9394 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9395 with offset_alignment.
9396 (DR_ALIGNED_TO): Delete.
9397 (DR_OFFSET_ALIGNMENT): New macro.
9398 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9399 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9400 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9401 (create_data_ref): Likewise.
9402 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9403 (vect_analyze_data_refs): Likewise.
9404 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9405 creating dummy innermost behavior.
9407 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9409 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9410 with a "innermost_loop_behavior *" and refeence tree.
9411 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9412 (create_data_ref): Update call accordingly.
9413 * tree-predcom.c (find_looparound_phi): Likewise.
9415 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9417 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9418 fields with dr_wrt_vec_loop.
9419 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9420 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9421 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9422 (vect_dr_behavior): New function.
9423 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9424 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9425 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9426 track whether the step preserves the misalignment.
9427 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9428 Use vect_dr_behavior.
9429 (vect_setup_realignment): Update call accordingly.
9430 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9431 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9432 call to vect_create_addr_base_for_vector_ref.
9433 (vect_create_cond_for_align_checks): Likewise.
9434 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9435 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9436 (vect_recog_mask_conversion_pattern): Likewise.
9437 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9438 (new_stmt_vec_info): Remove redundant zeroing.
9440 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9442 * common/config/arm/arm-common.c (arm_be8_option): New function.
9443 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9444 (ISA_ARMv6): Add isa_bit_be8.
9445 * config/arm/arm.h (arm_be8_option): Add prototype.
9446 (BE8_SPEC_FUNCTION): New define.
9447 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9448 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9449 (mlittle-endian): Similarly.
9450 (mbe8, mbe32): New options.
9451 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9452 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9454 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9456 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9458 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9460 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9461 (cleanup_tree_cfg_bb): Use it.
9462 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9464 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9466 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9469 * loop-doloop.c (add_test): Update profile.
9471 2017-07-03 Martin Liska <mliska@suse.cz>
9474 * sanopt.c (rewrite_usage_of_param): New function.
9475 (sanitize_rewrite_addressable_params): Likewise.
9476 (pass_sanopt::execute): Call rewrite_usage_of_param.
9478 2017-07-03 Richard Biener <rguenther@suse.de>
9480 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9481 back to using VIEW_CONVERT_EXPR.
9483 2017-07-03 Martin Liska <mliska@suse.cz>
9486 * doc/extend.texi: Document when a resolver function is
9487 generated for target_clones.
9489 2017-07-03 Martin Liska <mliska@suse.cz>
9491 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9492 and shadow memory of auto variables which are subject of
9493 use-after-scope sanitization.
9494 (asan_expand_mark_ifn): Add do set only when is_poison.
9496 2016-07-03 Richard Biener <rguenther@suse.de>
9498 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9500 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9501 (vectorizable_reduction): Perform reduction PHI creation when
9502 visiting a reduction PHI and adjust and simplify code generation
9503 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9504 (vect_transform_loop): Visit reduction PHIs.
9505 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9506 defs into the SLP tree.
9507 (vect_build_slp_tree): Reduction defs terminate the recursion.
9508 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9510 (vect_get_vec_defs_for_stmt_copy): Export.
9511 (vect_get_vec_defs): Likewise.
9512 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9514 (vect_get_vec_defs_for_stmt_copy): Declare.
9515 (vect_get_vec_defs): Likewise.
9517 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9519 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9520 parameter with a "loop" parameter and use it instead of the
9521 loop containing DR_STMT. Don't check simple_iv when doing
9522 BB analysis. Describe the two analysis modes in the comment.
9524 2017-07-03 Tom de Vries <tom@codesourcery.com>
9526 PR tree-optimization/69468
9527 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9528 (find_same_succ_bb): Handle ignore_edge_flags.
9530 2017-07-03 Tom de Vries <tom@codesourcery.com>
9532 PR tree-optimization/81192
9533 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9535 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9537 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9539 2017-07-03 Tom de Vries <tom@codesourcery.com>
9541 PR tree-optimization/81192
9542 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9543 BB_SAME_SUCC (bb) == NULL.
9545 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9547 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9550 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9552 * dumpfile.c: Include profile-count.h
9553 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9555 (insert_cond_bb): Update profile.
9556 * tree-cfg.h (insert_cond_bb): Update prototype.
9557 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9558 * tree-dump.c: Do not include tree-cfg.
9560 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9562 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9564 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9566 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9569 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9571 * tree-complex.c (expand_complex_div_wide): update profile.
9573 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9574 Alan Hayward <alan.hayward@arm.com>
9575 David Sherwood <david.sherwood@arm.com>
9577 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9578 (CORETYPES_H): New define.
9579 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9580 (insn-modes-inline.h, s-modes-inline-h): New rules.
9581 (generated_files): Add insn-modes-inline.h.
9582 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9583 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9584 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9585 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9586 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9587 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9588 (build/gencodes.o, build/genconditions.o): Likewise.
9589 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9590 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9591 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9592 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9593 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9594 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9595 * coretypes.h: Include everything up to real.h for generators.
9596 Include insn-modes.h first. Include wide-int-print.h after
9597 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9598 * machmode.h: Don't include insn-modes.h here.
9599 * function-tests.c: Remove includes of signop.h, machmode.h,
9600 double-int.h and wide-int.h.
9602 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9604 * optc-save-gen.awk: Likewise.
9605 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9606 * godump.c: Remove include of wide-int-print.h.
9607 * pretty-print.h: Likewise.
9608 * wide-int-print.cc: Likewise.
9609 * wide-int.cc: Likewise.
9610 * hash-map-tests.c: Remove include of signop.h.
9611 * hash-set-tests.c: Likewise.
9612 * rtl-tests.c: Likewise.
9613 * mkconfig.sh: Remove include of machmode.h.
9614 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9616 (emit_insn_modes_inline_h): ...this new function. Emit the code
9617 into an insn-modes-inline.h header file, adding appropriate
9618 include guards and end comments.
9619 (emit_insn_modes_c_header): Remove include of machmode.h.
9620 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9622 (main): Handle -i flag and call emit_insn_modes_inline_h when
9625 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9627 * tree-ssa-strlen.c (strinfo): Rename the length field to
9628 nonzero_chars. Add a full_string_p field.
9629 (compare_nonzero_chars, zero_length_string_p): New functions.
9630 (get_addr_stridx): Add an offset_out parameter.
9631 Use compare_nonzero_chars.
9632 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9633 (new_strinfo): Update after above changes to strinfo.
9634 (set_endptr_and_length): Set full_string_p.
9635 (get_string_length): Update after above changes to strinfo.
9636 (unshare_strinfo): Update call to new_strinfo.
9637 (maybe_invalidate): Likewise.
9638 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9639 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9640 as a uhwi instead of an shwi. Update after above changes to
9641 strinfo and new_strinfo.
9642 (zero_length_string): Assert that chainsi contains full strings.
9643 Use zero_length_string_p. Update call to new_strinfo.
9644 (adjust_related_strinfos): Update after above changes to strinfo.
9645 Copy full_string_p from origsi.
9646 (adjust_last_stmt): Use zero_length_string_p.
9647 (handle_builtin_strlen): Update after above changes to strinfo and
9648 new_strinfo. Install the lhs as the string length if the previous
9649 entry didn't describe a full string.
9650 (handle_builtin_strchr): Update after above changes to strinfo
9652 (handle_builtin_strcpy): Likewise.
9653 (handle_builtin_strcat): Likewise.
9654 (handle_builtin_malloc): Likewise.
9655 (handle_pointer_plus): Likewise.
9656 (handle_builtin_memcpy): Likewise. Track nonzero characters
9657 that aren't necessarily followed by a nul terminator.
9658 (handle_char_store): Likewise.
9660 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9662 PR tree-optimization/80769
9663 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9664 for malloc and calloc. Document the new invariant that all related
9665 strinfos have delayed lengths or none do.
9666 (verify_related_strinfos): Move earlier in file.
9667 (set_endptr_and_length): New function, split out from...
9668 (get_string_length): ...here. Also set the lengths of related
9670 (zero_length_string): Assert that chainsi has known (rather than
9672 (adjust_related_strinfos): Likewise.
9674 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9676 PR tree-optimization/81136
9677 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9678 assert that two references with the same misalignment have the same
9679 compile-time misalignment if those compile-time misalignments
9682 2017-07-01 Andi Kleen <ak@linux.intel.com>
9684 * print-tree.c (print_node): Print all attributes.
9686 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9688 * cfg.c (scale_bbs_frequencies): New function.
9689 * cfg.h (scale_bbs_frequencies): Declare it.
9690 * cfgloopanal.c (single_likely_exit): Cleanup.
9691 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9693 (scale_loop_profile): Likewise.
9694 (loop_version): Likewise.
9695 (create_empty_loop_on_edge): Update.
9696 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9697 scale_loop_frequencies, scale_loop_profile, loopify,
9698 loop_version): Update prototypes.
9699 * modulo-sched.c (sms_schedule): Update.
9700 * predict.c (unlikely_executed_edge_p): Also check probability.
9701 (probably_never_executed_edge_p): Fix typo.
9702 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9703 * tree-parloops.c (gen_parallel_loop): Update.
9704 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9705 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9706 * tree-ssa-loop-split.c (split_loop): Update.
9707 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9708 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9709 (vect_loop_versioning): Update.
9710 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9712 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9714 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9716 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9718 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9719 to keep profile consistent.
9721 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9723 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9724 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9725 * profile-count.h (max_safe_multiplier): Make unsigned.
9726 (profile_count::guessed_zero): New.
9728 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9730 * bb-reorder.c (fix_up_crossing_landing_pad,
9731 fix_crossing_conditional_branches): Use make_single_succ_edge
9732 to keep profile consistent.
9734 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9736 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9739 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9742 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9743 the right scopes, make sure cond_jump isn't preserved between multiple
9744 iterations. Search for fallthru edge whenever there are 3+ edges and
9745 use find_fallthru_edge for it.
9747 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9749 Patch by Alexander Monakov <amonakov@ispras.ru>
9750 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9751 probabilities consistently.
9753 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9755 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9756 emit_cmp_and_jump_insns.
9758 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9761 * tree-inline.c (expand_call_inline): Combine profile statuses.
9763 2017-06-30 Andrew Pinski <apinski@cavium.com>
9765 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9766 fold_stmt returned true.
9768 2017-06-30 Nathan Sidwell <nathan@acm.org>
9770 * ggc.h (empty_string): Delete.
9771 * cfgexpand.c (expand_asm_stmt): Use plain "".
9772 * optabs.c (expand_asm_memory_barrier): Likewise.
9773 * stringpool.c (empty_string): Delete.
9774 (digit_vector, digit_string): Delete.
9775 (ggc_alloc_string): Use plain "", don't optimize single digit
9776 strings. Use ggc_alloc_atomic.
9778 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9780 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9781 comparison set and one other set, use the cost of the non-comparison
9784 2017-06-30 Nathan Sidwell <nathan@acm.org>
9786 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9789 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9791 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9792 loops. Remove now unneeded calls to gimple_switch_set_label() that
9793 just set removed labels to NULL_TREE.
9795 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9797 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9798 (set_range_info): ...here. Only call set_range_info_raw if domain
9800 (set_nonzero_bits): Call set_range_info_raw.
9801 * tree-ssanames.h (set_range_info_raw): New.
9803 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9806 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9807 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9808 of nonimmediate_operand and <store_mask_constraint> instead of m
9809 for the input operand. For V8FI iterator, always split if input
9810 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9811 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9812 <store_mask_predicate> instead of register_operand and
9813 <store_mask_constraint> instead of v for the input operand. Make
9814 sure both operands aren't MEMs for if not <mask_applied>.
9816 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9818 * lto-wrapper.c (copy_file) Close both file descriptors before
9821 2017-06-30 Martin Liska <mliska@suse.cz>
9824 * multiple_target.c (create_dispatcher_calls): Make ifunc
9825 also for function that don't have calls or are not referenced.
9827 2017-06-30 Richard Biener <rguenther@suse.de>
9829 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9830 analyze the first scalar stmt. Move vector type computation
9831 for the BB case here from ...
9832 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9833 live operation processing in the SLP case properly.
9835 2017-06-30 Richard Biener <rguenther@suse.de>
9837 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9839 2017-06-30 Martin Liska <mliska@suse.cz>
9842 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9843 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9845 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9847 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9850 2017-06-30 Richard Biener <rguenther@suse.de>
9852 PR tree-optimization/81249
9853 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9854 condition reduction result to original scalar type.
9856 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9858 * profile-count.h (enum profile_quality): Fix typos and whitespace
9861 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9863 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9864 type for branch probabilities.
9866 2017-06-29 Julian Brown <julian@codesourcery.com>
9867 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9869 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9870 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9871 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9872 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9874 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9876 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9877 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9878 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9879 CC usage from generic code to here.
9880 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9881 CC usage into the target macros.
9883 2017-06-29 Maya Rashish <coypu@sdf.org>
9885 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9888 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9890 * arm/arm-builtins.c: Include profile-count.h
9891 * except.c (sjlj_emit_function_enter): Use
9892 profile_probability::unlikely.
9894 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9896 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9897 and tocrel_offset be pointer args rather than implicitly using
9899 (legitimate_constant_pool_address_p, rs6000_emit_move,
9900 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9901 tocrel_offset and use in toc_relative_expr_p call.
9902 (print_operand, print_operand_address): Use static tocrel_base_oac
9903 and tocrel_offset_oac.
9904 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9907 2017-06-29 Maya Rashish <coypu@sdf.org>
9909 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9911 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9913 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9914 objects, take into account only the alignment of 'op0' and 'mode1' if
9917 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9919 * ccmp.c (ccmp_tree_comparison_p): New function.
9920 (ccmp_candidate_p): Update to use above function.
9921 (get_compare_parts): New function.
9922 (expand_ccmp_next): Update to use new functions.
9923 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9925 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9926 take mode as argument.
9927 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9928 * expr.c (expand_expr_real_1): Pass mode as argument.
9930 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9932 * combine.c (combine_instructions): Print insns to dump_file, together
9935 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9937 * asan.c (asan_emit_stack_protection): Update.
9938 (create_cond_insert_point): Update.
9939 * auto-profile.c (afdo_propagate_circuit): Update.
9940 * basic-block.h (struct edge_def): Turn probability to
9941 profile_probability.
9942 (EDGE_FREQUENCY): Update.
9943 * bb-reorder.c (find_traces_1_round): Update.
9944 (better_edge_p): Update.
9945 (sanitize_hot_paths): Update.
9946 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9947 (make_single_succ_edge): Update.
9948 (check_bb_profile): Update.
9949 (dump_edge_info): Update.
9950 (update_bb_profile_for_threading): Update.
9951 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9952 probabilitycount to 0.
9953 * cfgbuild.c (compute_outgoing_frequencies): Update.
9954 * cfgcleanup.c (try_forward_edges): Update.
9955 (outgoing_edges_match): Update.
9956 (try_crossjump_to_edge): Update.
9957 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9958 (expand_gimple_tailcall): Update.
9959 (construct_init_block): Use make_single_succ_edge.
9960 (construct_exit_block): Use make_single_succ_edge.
9961 * cfghooks.c (verify_flow_info): Update.
9962 (redirect_edge_succ_nodup): Update.
9963 (split_edge): Update.
9964 (account_profile_record): Update.
9965 * cfgloopanal.c (single_likely_exit): Update.
9966 * cfgloopmanip.c (scale_loop_profile): Update.
9967 (set_zero_probability): Remove.
9968 (duplicate_loop_to_header_edge): Update.
9969 * cfgloopmanip.h (loop_version): Update prototype.
9970 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9971 (force_nonfallthru_and_redirect): Update.
9972 (update_br_prob_note): Update.
9973 (rtl_verify_edges): Update.
9974 (purge_dead_edges): Update.
9975 (rtl_lv_add_condition_to_bb): Update.
9976 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9977 * cgraphunit.c (init_lowered_empty_function): Update.
9978 (cgraph_node::expand_thunk): Update.
9979 * cilk-common.c: Include profile-count.h
9980 * dojump.c (inv): Remove.
9981 (jumpifnot): Update.
9982 (jumpifnot_1): Update.
9983 (do_jump_1): Update.
9985 (do_jump_by_parts_greater_rtx): Update.
9986 (do_compare_rtx_and_jump): Update.
9987 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9988 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9989 * dwarf2cfi.c: Include profile-count.h
9990 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9991 (sjlj_emit_dispatch_table): Likewise.
9992 * explow.c: Include profile-count.h
9993 * expmed.c (emit_store_flag_force): Update.
9994 (do_cmp_and_jump): Update.
9995 * expr.c (compare_by_pieces_d::generate): Update.
9996 (compare_by_pieces_d::finish_mode): Update.
9997 (emit_block_move_via_loop): Update.
9998 (store_expr_with_bounds): Update.
9999 (store_constructor): Update.
10000 (expand_expr_real_2): Update.
10001 (expand_expr_real_1): Update.
10002 * expr.h (try_casesi, try_tablejump): Update prototypes.
10003 * gimple-pretty-print.c (dump_probability): Update.
10004 (dump_profile): New.
10005 (dump_gimple_label): Update.
10006 (dump_gimple_bb_header): Update.
10007 * graph.c (draw_cfg_node_succ_edges): Update.
10008 * hsa-gen.c (convert_switch_statements): Update.
10009 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10010 (find_if_case_1): Update.
10011 (find_if_case_2): Update.
10012 * internal-fn.c (expand_arith_overflow_result_store): Update.
10013 (expand_addsub_overflow): Update.
10014 (expand_neg_overflow): Update.
10015 (expand_mul_overflow): Update.
10016 (expand_vector_ubsan_overflow): Update.
10017 * ipa-cp.c (good_cloning_opportunity_p): Update.
10018 * ipa-split.c (split_function): Use make_single_succ_edge.
10019 * ipa-utils.c (ipa_merge_profiles): Update.
10020 * loop-doloop.c (add_test): Update.
10021 (doloop_modify): Update.
10022 * loop-unroll.c (compare_and_jump_seq): Update.
10023 (unroll_loop_runtime_iterations): Update.
10024 * lra-constraints.c (lra_inheritance): Update.
10025 * lto-streamer-in.c (input_cfg): Update.
10026 * lto-streamer-out.c (output_cfg): Update.
10027 * mcf.c (adjust_cfg_counts): Update.
10028 * modulo-sched.c (sms_schedule): Update.
10029 * omp-expand.c (expand_omp_for_init_counts): Update.
10030 (extract_omp_for_update_vars): Update.
10031 (expand_omp_ordered_sink): Update.
10032 (expand_omp_for_ordered_loops): Update.
10033 (expand_omp_for_generic): Update.
10034 (expand_omp_for_static_nochunk): Update.
10035 (expand_omp_for_static_chunk): Update.
10036 (expand_cilk_for): Update.
10037 (expand_omp_simd): Update.
10038 (expand_omp_taskloop_for_outer): Update.
10039 (expand_omp_taskloop_for_inner): Update.
10040 * omp-simd-clone.c (simd_clone_adjust): Update.
10041 * optabs.c (expand_doubleword_shift): Update.
10042 (expand_abs): Update.
10043 (emit_cmp_and_jump_insn_1): Update.
10044 (expand_compare_and_swap_loop): Update.
10045 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10046 * predict.c (predictable_edge_p): Update.
10047 (edge_probability_reliable_p): Update.
10048 (set_even_probabilities): Update.
10049 (combine_predictions_for_insn): Update.
10050 (combine_predictions_for_bb): Update.
10051 (propagate_freq): Update.
10052 (estimate_bb_frequencies): Update.
10053 (force_edge_cold): Update.
10054 * profile-count.c (profile_count::dump): Add missing space into dump.
10055 (profile_count::debug): Add newline.
10056 (profile_count::differs_from_p): Explicitly convert to unsigned.
10057 (profile_count::stream_in): Update.
10058 (profile_probability::dump): New member function.
10059 (profile_probability::debug): New member function.
10060 (profile_probability::differs_from_p): New member function.
10061 (profile_probability::differs_lot_from_p): New member function.
10062 (profile_probability::stream_in): New member function.
10063 (profile_probability::stream_out): New member function.
10064 * profile-count.h (profile_count_quality): Rename to ...
10065 (profile_quality): ... this one.
10066 (profile_probability): New.
10067 (profile_count): Update.
10068 * profile.c (compute_branch_probabilities): Update.
10069 * recog.c (peep2_attempt): Update.
10070 * sched-ebb.c (schedule_ebbs): Update.
10071 * sched-rgn.c (find_single_block_region): Update.
10072 (compute_dom_prob_ps): Update.
10073 (schedule_region): Update.
10074 * sel-sched-ir.c (compute_succs_info): Update.
10075 * stmt.c (struct case_node): Update.
10076 (do_jump_if_equal): Update.
10077 (get_outgoing_edge_probs): Update.
10078 (conditional_probability): Update.
10079 (emit_case_dispatch_table): Update.
10080 (expand_case): Update.
10081 (expand_sjlj_dispatch_table): Update.
10082 (emit_case_nodes): Update.
10083 * targhooks.c: Update.
10084 * tracer.c (better_p): Update.
10085 (find_best_successor): Update.
10086 * trans-mem.c (expand_transaction): Update.
10087 * tree-call-cdce.c: Update.
10088 * tree-cfg.c (gimple_split_edge): Upate.
10089 (move_sese_region_to_fn): Upate.
10090 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10091 * tree-eh.c (lower_resx): Upate.
10092 (cleanup_empty_eh_move_lp): Upate.
10093 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10094 * tree-inline.c (copy_edges_for_bb): Update.
10095 (copy_cfg_body): Update.
10096 * tree-parloops.c (gen_parallel_loop): Update.
10097 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10098 (gimple_gen_time_profiler): Update.
10099 * tree-ssa-dce.c (remove_dead_stmt): Update.
10100 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10101 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10102 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10103 (unloop_loops): Update.
10104 (try_peel_loop): Update.
10105 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10106 * tree-ssa-loop-split.c (connect_loops): Update.
10107 (split_loop): Update.
10108 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10109 (hoist_guard): Update.
10110 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10111 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10112 (value_replacement): Update.
10113 * tree-ssa-reassoc.c (branch_fixup): Update.
10114 * tree-ssa-tail-merge.c (replace_block_by): Update.
10115 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10116 (create_edge_and_update_destination_phis): Update.
10117 (compute_path_counts): Update.
10118 (recompute_probabilities): Update.
10119 (update_joiner_offpath_counts): Update.
10120 (freqs_to_counts_path): Update.
10121 (duplicate_thread_path): Update.
10122 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10123 (struct switch_conv_info): Update.
10124 (gen_inbound_check): Update.
10125 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10126 (vect_do_peeling): Update.
10127 (vect_loop_versioning): Update.
10128 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10129 (optimize_mask_stores): Update.
10130 * ubsan.c (ubsan_expand_null_ifn): Update.
10131 * value-prof.c (gimple_divmod_fixed_value): Update.
10132 (gimple_divmod_fixed_value_transform): Update.
10133 (gimple_mod_pow2): Update.
10134 (gimple_mod_pow2_value_transform): Update.
10135 (gimple_mod_subtract): Update.
10136 (gimple_mod_subtract_transform): Update.
10137 (gimple_ic): Update.
10138 (gimple_stringop_fixed_value): Update.
10139 (gimple_stringops_transform): Update.
10140 * value-prof.h: Update.
10142 2017-06-29 Carl Love <cel@us.ibm.com>
10144 * config/rs6000/rs6000-c.c: Add support for built-in functions
10145 vector signed int vec_signed (vector float);
10146 vector signed long long vec_signed (vector double);
10147 vector signed int vec_signed2 (vector double, vector double);
10148 vector signed int vec_signede (vector double);
10149 vector signed int vec_signedo (vector double);
10150 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10151 instruction generator.
10152 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10153 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10154 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10156 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10157 vunsignede_v2df): Add define_expands.
10158 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10159 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10160 VEC_UNSIGNEDO): Add definitions.
10161 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10162 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10163 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10164 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10165 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10166 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10167 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10168 vec_unsignede, vec_unsignedo): Add builtin defines.
10169 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10171 * doc/extend.texi: Update the built-in documentation file for the
10172 new built-in functions.
10174 2017-06-29 Richard Biener <rguenther@suse.de>
10176 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10177 reduction chains to LOOP_VINFO_REDUCTIONS.
10178 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10179 SLP reductions after processing reduction chains.
10181 2017-06-29 Nathan Sidwell <nathan@acm.org>
10183 * builtins.c (fold_builtin_FUNCTION): Use
10184 lang_hooks.decl_printable_name.
10186 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10188 PR middle-end/81194
10189 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10190 with only one label.
10191 * stmt.c (expand_case): Assert NCASES is greater than one.
10193 2017-06-29 Richard Biener <rguenther@suse.de>
10195 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10197 (group_case_labels): Likewise.
10198 (find_taken_edge): Push sanity checking on val to workers...
10199 (find_taken_edge_cond_expr): ... here
10200 (find_taken_edge_switch_expr): ... and here, handle cases
10201 with just a default label.
10202 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10203 (group_case_labels): Likewise.
10204 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10205 group_case_labels does anything cleanup the CFG again.
10207 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10209 PR tree-optimization/81196
10210 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10211 exit condition comparing two IVs.
10213 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10215 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10216 profile to the dummy entry at the end of the list of architectures.
10217 * config/arm/arm-cpu-cdata.h: Regenerated.
10219 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10220 Michael Collison <michael.collison@arm.com>
10223 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10225 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10226 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10227 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10228 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10229 of shift when the shift amount is masked with constant equal to
10230 the size of the mode.
10231 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10234 2017-06-29 Martin Liska <mliska@suse.cz>
10236 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10238 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10241 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10244 (edge_growth_cache_entry::edge_growth_cache_entry): New
10246 (reset_edge_growth_cache): Update to use constructor.
10248 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10250 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10251 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10252 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10254 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10256 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10257 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10259 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10261 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10262 (*-linux-uclibc*): Add t-uclibc tmake_file.
10263 * config/t-musl: New.
10264 * config/t-uclibc: New.
10266 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10268 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10270 (gen_comm_data): Emit architectural setting of arch_prof.
10271 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10273 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10274 (armv8-m.base, armv8-m.main): Likewise.
10275 * arm-protos.h (arm_build_target): Add profile field.
10276 (arch_option): Likewise.
10277 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10279 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10280 arm_active_target.profile.
10282 2017-06-28 Richard Biener <rguenther@suse.de>
10284 PR middle-end/81227
10285 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10286 TYPE_OVERFLOW_WRAPS.
10287 * match.pd (negate_expr_p): Likewise.
10288 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10289 fold_build2, not fold_binary.
10291 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10293 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10294 Convert memory address to Pmode.
10295 (aarch64_print_operand): Assert MEM operands are always Pmode.
10297 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10300 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10301 Remove redundant if.
10302 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10303 * config/arm/aarch-common-protos.h
10304 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10305 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10307 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10310 * multiple_target.c (create_dispatcher_calls): Set the default
10311 clone to be static, not public.
10313 2017-06-28 Richard Biener <rguenther@suse.de>
10315 * tree-vect-loop.c (vectorizable_reduction): Move special
10316 cond reduction IV var creation ...
10317 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10318 parameter. Use STMT_VINFO_VECTYPE.
10319 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10322 2017-06-28 Martin Liska <mliska@suse.cz>
10325 * ipa-visibility.c (non_local_p): Handle visibility.
10327 2017-06-28 Martin Liska <mliska@suse.cz>
10330 * common.opt: Add IntegerRange to various options.
10331 * opt-functions.awk (integer_range_info): New function.
10332 * optc-gen.awk: Add integer_range_info to cl_options struct.
10333 * opts-common.c (decode_cmdline_option): Handle
10334 CL_ERR_INT_RANGE_ARG.
10335 (cmdline_handle_error): Likewise.
10336 * opts.c (print_filtered_help): Show valid interval in
10337 when --help is provided.
10338 * opts.h (struct cl_option): Add range_min and range_max fields.
10339 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10341 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10343 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10344 (x * C EQ/NE y * C): New transformation.
10346 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10348 * genmultilib (combination_space): Accept '+' in option names.
10350 2017-06-28 Martin Liska <mliska@suse.cz>
10353 * asan.c (instrument_derefs): Bail out inner references
10354 that are hard register variables.
10356 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10359 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10360 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10362 2017-06-28 Richard Biener <rguenther@suse.de>
10364 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10365 (vect_get_slp_defs): Adjust.
10366 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10368 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10370 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10371 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10372 (vectorizable_reduction): Adjust.
10373 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10375 (vect_get_slp_defs): Likewise.
10376 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10377 (vectorizable_bswap): Adjust.
10378 (vectorizable_call): Likewise.
10379 (vectorizable_conversion): Likewise.
10380 (vectorizable_assignment): Likewise.
10381 (vectorizable_shift): Likewise.
10382 (vectorizable_operation): Likewise.
10383 (vectorizable_store): Likewise.
10384 (vectorizable_condition): Likewise.
10385 (vectorizable_comparison): Likewise.
10387 2017-06-28 Michael Collison <michael.collison@arm.com>
10390 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10392 (arm_gen_movmemqi): Removed unused variable 'i'.
10393 Convert 'for' loop into 'while' loop.
10394 (arm_expand_prologue): Remove last unnecessary set of insn.
10395 (thumb_pop): Remove unused variable 'pushed_words'.
10396 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10398 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10400 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10401 * config/s390/s390-protos.h: Add prototype of
10402 s390_rel_address_ok_p.
10403 * config/s390/s390.c (s390_got_symbol): New function.
10404 (s390_rel_address_ok_p): New function.
10405 (legitimize_pic_address): Use s390_rel_address_ok_p.
10406 (s390_load_got): Use s390_got_symbol.
10407 (s390_option_override): Issue error if
10408 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10409 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10411 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10413 2017-06-27 Andrew Pinski <apinski@cavium.com>
10415 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10416 (X * copysign (1.0, X)): New pattern.
10417 (X * copysign (1.0, -X)): New pattern.
10418 (copysign (-1.0, CST)): New pattern.
10420 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10422 * genmultilib (combination_space): Remove variable.
10423 Validate reuse rules against regular expression for any sequence
10424 of multilib options in any order.
10426 2017-06-27 Michael Collison <michael.collison@arm.com>
10428 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10429 call aarch64_split_simd_combine.
10430 * (aarch64_combine_internal<mode>): Delete pattern.
10431 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10432 Allow register and subreg operands.
10434 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10436 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10437 specific need, just fallback on defaults.
10438 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10440 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10441 Olivier Hainque <hainque@adacore.com>
10443 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10445 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10446 targets. Pick a default if no particular attempt applied.
10447 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10450 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10452 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10453 (x86_64-wrs-vxworks7): Likewise.
10455 2017-06-27 Marek Polacek <polacek@redhat.com>
10458 * ubsan.c (instrument_null): Check get_base_address's result for null.
10460 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10462 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10464 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10466 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10467 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10468 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10469 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10470 New function types.
10471 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10472 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10473 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10474 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10475 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10476 BUILT_IN_FEUPDATEENV): New builtins.
10477 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10478 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10479 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10480 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10482 (builtin_structptr_types): Adjust size.
10483 * tree.c (builtin_structptr_types): Add four entries.
10485 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10486 Olivier Hainque <hainque@adacore.com>
10488 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10489 (TLS_SYM): New local macro, forcing reference to __tls__ on
10490 link command lines for VxWorks 7 RTPs, triggering initialization
10492 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10493 OS features TLS support, true for RTPs on VxWorks 7.
10494 * config/vxworks.c (vxworks_override_options): Setup emutls
10497 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10499 * predict.c (test_prediction_value_range): Use -1U instead of -1
10500 to avoid narrowing conversion warning.
10501 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10502 to avoid narrowing conversion warning.
10503 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10505 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10507 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10509 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10510 64bit configurations.
10511 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10512 (SIZE_TYPE): Likewise.
10513 * config/vxworks.c (vxworks_emutls_var_fields): Use
10514 long_unsigned_type_node instead of unsigned_type_node as the offset
10515 field type, which is "pointer" mode in emutls.c.
10517 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10520 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10522 PR middle-end/81207
10523 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10524 gimple_vuse copying separately from gimple_vdef copying.
10526 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10528 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10531 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10532 Olivier Hainque <hainque@adacore.com>
10534 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10535 for all vxworks7 targets.
10536 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10537 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10538 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10539 variations for VX6/VX7 and 32/64bits later on in ...
10540 (VXWORKS_LIB_SPEC): Leverage new macros.
10541 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10542 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10544 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10546 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10547 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10549 2017-06-26 Carl Love <cel@us.ibm.com>
10551 * config/rs6000/rs6000-c.c: Add support for built-in functions
10552 vector bool char vec_reve (vector bool char);
10553 vector signed char vec_reve (vector signed char);
10554 vector unsigned char vec_reve (vector unsigned char);
10555 vector bool int vec_reve (vector bool int);
10556 vector signed int vec_reve (vector signed int);
10557 vector unsigned int vec_reve (vector unsigned int);
10558 vector bool long long vec_reve (vector bool long long);
10559 vector signed long long vec_reve (vector signed long long);
10560 vector unsigned long long vec_reve (vector unsigned long long);
10561 vector bool short vec_reve (vector bool short);
10562 vector signed short vec_reve (vector signed short);
10563 vector double vec_reve (vector double);
10564 vector float vec_reve (vector float);
10565 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10566 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10567 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10568 (altivec_vreve): New pattern.
10569 * config/rs6000/altivec.h (vec_reve): New define.
10570 * doc/extend.texi (vec_rev): Update the built-in documentation file
10571 for the new built-in functions.
10573 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10575 PR tree-optimization/71815
10576 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10578 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10580 (slsr_process_phi): Likewise.
10581 (replace_uncond_cands_and_profitable_phis): Don't replace a
10582 multiply candidate with a stride of 1 (copy or cast).
10583 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10585 (lowest_cost_path): Likewise.
10586 (total_savings): Likewise.
10588 2017-06-26 Richard Biener <rguenther@suse.de>
10591 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10592 Use def_builtin_pure for all gather builtins.
10594 2017-06-26 Richard Biener <rguenther@suse.de>
10596 PR tree-optimization/81203
10597 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10598 non-dominating BBs.
10600 2017-06-26 Marek Polacek <polacek@redhat.com>
10603 * doc/invoke.texi: Document -Wmultistatement-macros.
10605 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10607 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10608 arm_neon_ok_no_float_abi effective target.
10610 2017-06-26 Richard Biener <rguenther@suse.de>
10612 PR tree-optimization/80928
10613 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10614 (copy_bbs): Set BB_DUPLICATED flag early.
10615 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10617 (execute_on_shrinking_pred): Likewise.
10618 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10619 BB_DUPLICATED blocks.
10620 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10621 iterate over all PHIs considering removal of *gsi.
10623 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10625 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10628 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10630 * config/rs6000/rs6000-string.c: (expand_block_clear,
10631 do_load_for_compare, select_block_compare_mode,
10632 compute_current_alignment, expand_block_compare,
10633 expand_strncmp_align_check, expand_strn_compare,
10634 expand_block_move, rs6000_output_load_multiple)
10635 Move functions related to string/block move/compare
10636 to a separate file.
10637 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10638 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10639 for this function which is now used in two files.
10640 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10641 * config.gcc: Add rs6000-string.o to extra_objs for
10642 targets powerpc*-*-* and rs6000*-*-*.
10644 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10647 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10648 32-bit, since indexed is not valid for DImode.
10649 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10650 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10651 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10652 (define_peephole2 for Altivec d-form store): Likewise.
10655 * multiple_target.c (create_dispatcher_calls): Only create the
10656 dispatcher call if the function is the default clone of a
10657 versioned function.
10659 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10661 PR middle-end/80902
10662 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10663 a call, force the call to not be a tail call.
10665 2017-06-23 Jeff Law <law@redhat.com>
10667 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10670 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10672 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10673 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10674 handling for early expansion of vector shifts (sl,sr,sra,rl).
10675 (builtin_function_type): Add vector shift right instructions
10676 to the unsigned argument list.
10678 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10680 rtl-optimizatoin/79286
10681 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10682 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10683 trap. PIC register plus a const unspec without offset can never trap.
10685 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10687 * tree.h (builtin_structptr_type): New type.
10688 (builtin_structptr_types): Declare new array.
10689 * tree.c (builtin_structptr_types): New array.
10690 (free_lang_data, build_common_tree_nodes): Use it.
10692 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10695 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10698 2017-06-22 Matt Turner <mattst88@gmail.com>
10700 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10701 Lake models to skylake case. Assume skylake for unknown
10702 models with clflushopt.
10704 2017-06-22 Jeff Law <law@redhat.com>
10706 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10707 frame sizes that do not satisfy aarch64_uimm12_shift.
10709 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10711 * profile-count.h (apply_probability,
10712 apply_scale, probability_in): Fix checks for zero.
10714 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10716 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10717 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10719 2016-06-22 Richard Biener <rguenther@suse.de>
10721 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10722 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10723 REDUC_MAX_EXPR support.
10724 (vectorizable_reduction): Likewise.
10725 (vect_create_epilog_for_reduction): Likewise.
10727 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10729 * match.pd (A / (1 << B) -> A >> B): New.
10730 * generic-match-head.c: Include optabs-tree.h.
10731 * gimple-match-head.c: Likewise.
10732 * optabs-tree.h (target_supports_op_p): New.
10733 * optabs-tree.c (target_supports_op_p): New.
10735 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10737 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10738 $gcc_cv_ld --help output.
10739 (gcc_cv_ld_demangle): Likewise.
10740 (gcc_cv_ld_eh_frame_hdr): Likewise.
10741 (gcc_cv_ld_pie): Likewise.
10742 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10743 (gcc_cv_ld_buildid): Likewise.
10744 (gcc_cv_ld_sysroot): Likewise.
10745 (ld_bndplt_support): Likewise.
10746 (ld_pushpopstate_support): Likewise.
10747 * configure: Regenerate.
10748 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10750 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10753 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10754 operands indexes to avoid gap between operands and match_dups.
10756 2017-06-21 Andrew Pinski <apinski@cavium.com>
10758 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10759 Increment Arith_shift and Arith_shift_reg by 1.
10760 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10762 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10763 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10764 (aarch64_strip_extend): Add new argument and test for it.
10765 (aarch64_cheap_mult_shift_p): New function.
10766 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10767 add a cost if it is true.
10768 Update calls to aarch64_strip_extend.
10769 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10771 2017-06-21 Andrew Pinski <apinski@cavium.com>
10773 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10775 (thunderxt88): Likewise.
10776 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10777 (thunderx_prefetch_tune): New variable.
10778 (thunderx2t99_prefetch_tune): Update for the correct values.
10779 (thunderxt88_tunings): New variable.
10780 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10781 generic_prefetch_tune.
10782 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10784 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10786 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10787 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10788 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10789 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10790 (aarch64_atomic_cas<mode>, GPI): Likewise.
10792 2017-06-21 Martin Liska <mliska@suse.cz>
10794 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10795 statements on cold and hot labels.
10796 * predict.c (tree_estimate_probability_bb): Remove the
10797 prediction from this place.
10799 2017-06-21 Martin Liska <mliska@suse.cz>
10801 PR tree-optimization/79489
10802 * gimplify.c (maybe_add_early_return_predict_stmt): New
10804 (gimplify_return_expr): Call the function.
10805 * predict.c (tree_estimate_probability_bb): Remove handling
10807 * predict.def: Update comment about early return predictor.
10808 * gimple-predict.h (is_gimple_predict): New function.
10809 * predict.def: Change default value of early return to 66.
10810 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10812 * passes.def: Put pass_strip_predict_hints to the beginning of
10815 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10817 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10818 FUNCTION_DECL declarations.
10819 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10821 (dwaf2out_decl): Likewise.
10822 * godump.c (go_early_global_decl): Skip call to the real debug hook
10823 for FUNCTION_DECL declarations.
10824 * passes.c (rest_of_decl_compilation): Skip call to the
10825 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10826 -fdump-go-spec is passed.
10828 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10830 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10831 Reorder for compactness.
10832 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10833 (def_builtin_pure, def_builtin_pure2): New functions.
10834 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10836 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10838 * match.pd (nop_convert): New predicate.
10839 ((A +- CST1) +- CST2): Allow some NOP conversions.
10841 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10844 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10845 with ctors/dtors if GOVD_SHARED is set.
10847 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10849 * config/aarch64/aarch64.md (movti_aarch64):
10850 Emit mov rather than orr.
10851 (movtf_aarch64): Likewise.
10852 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10853 Emit mov rather than orr.
10855 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10857 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10858 Swap alternatives, make integer dup more expensive.
10860 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10862 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10863 Return true for non-tls symbols.
10865 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10867 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10868 (cortex-a75): Likewise.
10869 (cortex-a75.cortex-a55): Likewise.
10870 * config/aarch64/aarch64-tune.md: Regenerate.
10871 * doc/invoke.texi (-mtune): Document new values for -mtune.
10873 2017-06-21 Tom de Vries <tom@codesourcery.com>
10875 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10876 stack_size feature.
10877 (Effective-Target Keywords, Other attributes): Suggest using
10878 dg-add-options stack_size feature to get stack limit in stack_size
10879 effective target documentation.
10881 2017-06-21 Julian Brown <julian@codesourcery.com>
10882 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10884 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10885 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10886 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10888 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10889 attribute type list for neon_multiply.
10890 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10891 attribute type list for neon_multiply.
10892 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10893 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10894 attribute type list for neon_multiply.
10895 * config/arm/types.md (crypto_pmull): Add.
10896 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10897 attribute type list.
10899 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10901 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10904 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10906 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10907 to make sure not to dereference a NULL cost_classes_ptr pointer.
10909 2017-06-20 Carl Love <cel@us.ibm.com>
10911 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10912 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10913 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10914 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10915 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10916 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10917 VMULOSW): New enum "unspec" values.
10918 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10919 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10920 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10921 altivec_vmulosw): New patterns.
10922 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10923 VMULOSW): Add definitions.
10925 2017-06-20 Julia Koval <julia.koval@intel.com>
10927 * config/i386/i386.c: Fix rounding expand for new pattern.
10928 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10930 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10932 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10933 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10935 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10937 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10940 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10942 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10943 architecture, then by implementer ID.
10944 * config/aarch64/aarch64-tune.md: Regenerate.
10946 2017-06-20 Richard Biener <rguenther@suse.de>
10948 PR middle-end/81097
10949 * fold-const.c (split_tree): Fold to type before negating.
10951 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10953 * diagnostic-show-locus.c
10954 (selftest::test_fixit_deletion_affecting_newline): New function.
10955 (selftest::diagnostic_show_locus_c_tests): Call it.
10957 2017-06-20 Andreas Schwab <schwab@suse.de>
10960 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10963 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10965 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10966 __ARM_FEATURE_COPROC according to support.
10968 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10970 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10971 Rewritten to avoid overflow for > 32-bit pointers.
10974 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10975 by removing enum keyword.
10976 (ubsan_type_descriptor): Likewise. Formatting fix.
10979 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10980 splitter): Require TARGET_SSE2 in the condition.
10982 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10985 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10986 for doing vector set of SFmode on ISA 3.0.
10987 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10988 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10990 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10991 SFmode value into a V4SF variable that was extracted from another
10992 V4SF variable without converting the element to double precision
10993 and back to single precision vector format.
10994 (vsx_insert_extract_v4sf_p9_2): Likewise.
10996 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10998 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10999 in UWHI to avoid undefined overflow.
11002 * ubsan.h (enum ubsan_encode_value_phase): New.
11003 (ubsan_encode_value): Change second argument to
11004 enum ubsan_encode_value_phase with default value of
11005 UBSAN_ENCODE_VALUE_GENERIC.
11006 * ubsan.c (ubsan_encode_value): Change second argument to
11007 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11008 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11009 create_tmp_var_raw instead of create_tmp_var and use a
11011 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11012 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11013 ubsan_encode_value callers.
11016 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11017 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11018 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11020 2017-06-19 Richard Biener <rguenther@suse.de>
11022 PR middle-end/81118
11023 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11024 estimates if we changed anything.
11026 2017-06-19 Richard Biener <rguenther@suse.de>
11028 PR tree-optimization/80887
11029 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11030 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11031 simplified lookups, then reset mprts_hook.
11032 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11034 (try_to_simplify): Likewise.
11036 2017-06-19 Martin Liska <mliska@suse.cz>
11039 * gimplify.c (gimplify_switch_expr):
11040 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11042 2017-06-19 Martin Liska <mliska@suse.cz>
11044 * doc/install.texi: Document that PGO runs in 4 stages.
11046 2017-06-19 Martin Liska <mliska@suse.cz>
11049 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11050 to dispatcher function name.
11051 * multiple_target.c (replace_function_decl): New function.
11052 (create_dispatcher_calls): Redirect both edges and references.
11054 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11056 * profile-count.c (profile_count::dump): Dump quality.
11057 (profile_count::differs_from_p): Update for unsigned val.
11058 * profile-count.h (profile_count_quality): New enum.
11059 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11061 2017-06-19 Richard Biener <rguenther@suse.de>
11063 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11064 struct function as arg.
11065 (estimate_numbers_of_iterations): Export overload with loop arg.
11066 (free_numbers_of_iterations_estimates_loop): Use an overload of
11067 free_numbers_of_iterations_estimates instead.
11068 * tree-cfg.c (remove_bb): Adjust.
11069 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11070 * tree-parloops.c (gen_parallel_loop): Likewise.
11071 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11073 (tree_unroll_loops_completely): Likewise.
11074 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11075 Use an overload instead and export.
11076 (estimated_loop_iterations): Adjust.
11077 (max_loop_iterations): Likewise.
11078 (likely_max_loop_iterations): Likewise.
11079 (estimate_numbers_of_iterations): Take struct function as arg
11081 (loop_exits_before_overflow): Adjust.
11082 (free_numbers_of_iterations_estimates_loop): Use an overload.
11083 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11084 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11086 2017-06-19 Richard Biener <rguenther@suse.de>
11089 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11090 RANGE_EXPR conservatively.
11092 2017-06-16 Carl Love <cel@us.ibm.com>
11094 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11095 definitions for vec_float, vec_float2, vec_floato,
11096 vec_floate built-ins.
11097 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11098 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11100 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11101 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11102 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11103 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11104 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11105 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11106 vec_floato): Add builtin defines.
11107 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11108 Update the built-in documentation file for the new built-in
11111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11113 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11114 (mthumb): Mark as the negative of -marm.
11116 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11118 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11120 (ARM Options, -mtune): Document that this accepts the same
11121 extension options as -mcpu.
11122 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11124 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11126 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11127 permitted extensions.
11129 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11131 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11132 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11133 (armv8-m.main): Add option +nodsp.
11134 * config/arm/arm-cpu-cdata.h: Regenerated.
11136 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11138 * config/arm/t-fuchsia: New file.
11139 * config.gcc (arm*-*-fuchsia*): Use it.
11141 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11143 * config/arm/t-symbian: Rewrite for new option infrastructure.
11145 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11147 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11148 (MULTILIB_REQUIRED): Likewise.
11150 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11152 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11153 (MULTILIB_RESUE): Likewise.
11154 (MULTILIB_MATCHES): Likewise.
11155 (MULTLIB_REQUIRED): Likewise.
11157 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11159 * config/arm/t-rtems: Rewrite for new option framework.
11161 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11163 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11164 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11165 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11166 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11167 * config/arm/t-multilib: ... here.
11168 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11169 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11170 armv7-a and armv8*-a when A-profile libraries have not been built.
11171 * config/arm/t-rmprofile: Rewrite.
11173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11175 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11176 with a backslash. Remove the backslash after substituting unescaped
11178 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11182 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11183 the driver to pass through the default setting of -mfloat-abi.
11184 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11186 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11187 (all_feat_combs): New rule.
11188 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11190 * config/arm/t-aprofile: Rewrite.
11192 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11194 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11195 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11196 fpu is not specified by the user/command-line.
11197 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11198 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11199 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11200 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11201 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11202 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11204 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11206 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11207 * config/arm/t-arm-elf: Rewritten.
11209 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11211 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11212 have some floating-point instructions.
11213 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11214 (TARGET_MAYBE_HARD_FLOAT): New macro.
11215 * config/arm/arm-builtins.c (arm_init_builtins): Use
11216 TARGET_MAYBE_HARD_FLOAT.
11217 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11219 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11221 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11222 (configargs.h): Include it.
11223 (arm_print_hint_for_fpu_option): New function.
11224 (arm_parse_fpu_option): New function.
11225 (candidate_extension): New class.
11226 (arm_canon_for_multilib): New function.
11227 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11228 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11229 (ARCH_CANONICAL_SPECS): New macro.
11230 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11232 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11234 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11235 are set after handling multilib fragments. Set target_cpu_default2
11238 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11240 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11242 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11243 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11245 (check_arch): Likewise.
11246 * config/arm/arm.c (arm_configure_build_target): Handle
11247 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11248 options in the default.
11250 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11252 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11253 when an option is an alias of another.
11254 * config/arm/parsecpu.awk (optalias): New parser token.
11255 (gen_comm_data): Mark non-alias options as such. Emit entries
11256 for extension aliases.
11257 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11258 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11259 (armv6kz, armv6zk, armv6t2): Likewise.
11260 (armv7): Make vfpv3-d16 an alias.
11261 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11263 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11264 Sort in canonical order.
11265 (armv8-a): Sort in canonical order.
11266 (armv8.1-a, armv8.2-a): Likewise.
11267 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11269 (cortex-a9): Sort in canonical order.
11270 * config/arm/arm.c (selftests.h): Include it.
11271 (arm_test_cpu_arch_data): New function.
11272 (arm_run_self_tests): New function.
11273 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11274 (targetm): Move declaration to the end of the file.
11275 * arm-cpu-cdata.h: Regenerated.
11277 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11279 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11280 call to target_mode_check describing the type of option passed.
11281 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11282 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11283 arm_parse_cpu_option_name to match parameters against list of
11285 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11286 arm_arch_core_flags data structure.
11287 * config/arm/arm-cpu_cdata.h: Regenerated.
11289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11291 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11293 (arm_print_hint_for_cpu_option): Likewise.
11294 (arm_print_hint_for_arch_option): Likewise.
11295 (arm_parse_cpu_option_name): Likewise.
11296 (arm_parse_arch_option_name): Likewise.
11297 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11298 of entries in the all_fpus list.
11299 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11300 (arm_parse_cpu_option_name): Declare.
11301 (arm_parse_arch_option_name): Declare.
11302 (arm_parse_option_features): Declare.
11303 (arm_intialize_isa): Declare.
11304 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11306 (gen_comm_data): ... here. Make definitions non-static.
11307 * config/arm/arm-cpu-data.h: Regenerated.
11308 * config/arm/arm-cpu-cdata.h: Regenerated.
11310 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11312 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11313 (cpu_arch_extension): New structure.
11314 (cpu_arch_option, arch_option, cpu_option): New structures.
11315 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11316 architecture types.
11317 (gen_data): Generate new format data tables.
11318 * config/arm/arm.c (cpu_tune): New structure.
11319 (cpu_option, processors): Delete.
11320 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11321 (arm_print_hint_for_cpu_option): ... this and ...
11322 (arm_print_hint_for_arch_option): ... this.
11323 (arm_parse_arch_cpu_name): Delete. Replace with ...
11324 (arm_parse_cpu_option_name): ... this and ...
11325 (arm_parse_arch_option_name): ... this.
11326 (arm_unrecognized_feature): Change type of target parameter to
11328 (arm_parse_arch_cpu_features): Delete. Replace with ...
11329 (arm_parse_option_features): ... this.
11330 (arm_configure_build_target): Rework to use new configuration data
11332 (arm_print_tune_info): Rework for new configuration data tables.
11333 * config/arm/arm-cpu-data.h: Regenerated.
11334 * config/arm/arm-cpu.h: Regenerated.
11336 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11338 * Makefile.in (OBJS): Move sbitmap.o from here ...
11339 (OBJS-libcommon): ... to here.
11341 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11343 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11344 (ISA_ALL_CRYPTO): New macro.
11345 (ISA_ALL_SIMD): New macro
11346 (ISA_ALL_FP): New macro.
11347 * config/arm/arm.c (fpu_bitlist): Update initializer.
11348 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11350 (arm9e): Add fpu. Add option for nofp
11351 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11352 (arm926ej-s, arm1026ej-s): Likewise.
11353 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11354 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11355 neon-fp16, neon-vfpv4, nofp and nosimd.
11356 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11357 (cortex-a8): Add fpu. Add option for nofp.
11358 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11359 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11360 (cortex-r4f): Add fpu.
11361 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11362 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11364 (cortex-r8): Likewise.
11365 (cortex-m4): Add fpu. Add option for nofp.
11366 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11367 (cortex-a17.cortex-a7): Likewise.
11368 (cortex-a32): Add fpu. Add options for crypto and nofp.
11369 (cortex-a35, cortex-a53): Likewise.
11370 (cortex-a57): Add fpu. Add option for crypto.
11371 (cortex-a72, cortex-a73): Likewise.
11372 (exynos-m1): Likewise.
11373 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11374 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11375 (cortex-m33): Add fpu. Add option for nofp.
11376 * config/arm/arm-cpu-cdata.h: Regenerated
11377 * config/arm/arm-cpu-data.h: Regenerated.
11379 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11381 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11382 (armv5te, armv5tej): Likewise.
11383 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11384 (armv7): Add options fp and vfpv3-d16.
11385 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11386 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11388 (armv7ve): Likewise.
11389 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11390 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11391 (armv8-a): Add nocrypto option.
11392 (armv8.1-a, armv8.2-a): Likewise.
11393 (armv8-m.main): add options fp, fp.dp and nofp.
11395 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11397 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11399 (armv8-a+crc): Delete.
11400 (armv8.1-a): Add options simd, crypto and nofp.
11401 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11402 (armv8.2-a+fp16): Delete.
11403 (armv8-m.main): Add option dsp.
11404 (armv8-m.main+dsp): Delete.
11405 (cortex-a8): Add fpu. Add nofp option.
11406 (cortex-a9): Add fpu. Add nofp and nosimd options.
11407 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11408 link to main cpu and architecture data structures.
11409 (gen_comm_data): Only put isa attributes from the main architecture
11411 (option): New statement for architecture and CPU entries.
11412 * arm.c (struct cpu_option): New structure.
11413 (struct processors): Add entry for options.
11414 (arm_unrecognized_feature): New function.
11415 (arm_parse_arch_cpu_name): Ignore any characters after the first
11417 (arm_parse_arch_cpu_feature): New function.
11418 (arm_configure_build_target): Separate out any CPU and architecture
11419 features and parse separately. Don't error out if -mfpu=auto is
11420 used with only an architecture string.
11421 (arm_print_asm_arch_directives): New function.
11422 (arm_file_start): Call it.
11423 * config/arm/arm-cpu-cdata.h: Regenerated.
11424 * config/arm/arm-cpu-data.h: Likewise.
11425 * config/arm/arm-tables.opt: Likewise.
11427 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11429 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11430 assembler when it is not -mfpu=auto.
11432 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11434 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11435 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11436 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11437 (ASM_CPU_SPEC): Rewrite.
11438 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11439 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11440 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11441 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11442 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11443 copied string is NUL-terminated. Also strip any characters prefixed
11445 (arm_rewrite_selected_arch): New function.
11446 (arm_rewrite_march): New function.
11448 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11450 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11451 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11452 (march, mcpu, mtune): Convert to string-based options.
11453 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11454 (arm_parse_arch_cpu_name): New function.
11455 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11456 identify selected architecture or CPU.
11457 (arm_option_save): New function.
11458 (TARGET_OPTION_SAVE): Redefine.
11459 (arm_option_restore): Restore string options.
11460 (arm_option_print): Print string options.
11462 2017-06-16 Martin Sebor <msebor@redhat.com>
11464 PR tree-optimization/80933
11465 PR tree-optimization/80934
11466 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11467 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11468 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11469 (gimple_fold_builtin): Call them.
11471 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11473 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11474 as unlikely; update profile.
11476 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11478 * predict.c (force_edge_cold): Handle declaring edges impossible
11481 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11483 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11485 (try_unroll_loop_completely): Fix reporting.
11487 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11489 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11491 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11494 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11495 if given a non-constant argument for an intrinsic which requires a
11498 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11500 * profile.c (compare_freqs): New function.
11501 (branch_prob): Sort edge list.
11502 (find_spanning_tree): Assume that the list is priority sorted.
11504 2017-06-16 Richard Biener <rguenther@suse.de>
11506 PR tree-optimization/81090
11507 * passes.def (pass_record_bounds): Remove.
11508 * tree-pass.h (make_pass_record_bounds): Likewise.
11509 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11510 make_pass_record_bounds): Likewise.
11511 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11512 not free niter estimates at the beginning but at the end.
11513 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11515 2017-06-16 Richard Biener <rguenther@suse.de>
11517 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11518 initializer to workaround ICE in host GCC 4.8.
11520 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11522 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11524 (clone_inlined_nodes): Update.
11526 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11528 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11529 prefetch settings, and enable prefetching by default at -O3.
11531 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11533 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11534 Set flag_prefetch_loop_arrays according to tuning data.
11536 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11538 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11539 New tune structure.
11540 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11541 [Unrelated to main purpose of the patch] Place the pointer field last
11542 to enable type checking errors when tune structure are wrongly merged.
11543 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11544 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11545 (thunderx2t99_prefetch_tune): New tune constants.
11546 (tune_params *_tunings): Update all tunings (no functional change).
11547 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11548 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11549 from tunings structures.
11551 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11554 * ubsan.c (instrument_null): Add T argument, use it instead
11555 of computing it based on IS_LHS.
11556 (instrument_object_size): Likewise.
11557 (pass_ubsan::execute): Adjust instrument_null and
11558 instrument_object_size callers to pass gimple_get_lhs or
11559 gimple_assign_rhs1 result to it. Use instrument_null instead of
11560 calling get_base_address and instrument_mem_ref. Handle
11561 aggregate call arguments for object-size sanitization.
11563 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11565 PR tree-optimization/81089
11566 * tree-vrp.c (is_masked_range_test): Validate operands of
11569 2017-06-15 Martin Sebor <msebor@redhat.com>
11572 * dumpfile.c (dump_register): Avoid calling memset to initialize
11573 a class with a default ctor.
11574 * gcc.c (struct compiler): Remove const qualification.
11575 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11576 * hash-table.h: Ditto.
11577 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11579 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11580 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11582 * params.h (struct param_info): Make struct members non-const.
11583 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11584 with default initialization.
11585 * vec.h (vec_copy_construct, vec_default_construct): New helper
11587 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11588 with vec_copy_construct.
11589 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11590 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11591 * doc/invoke.texi (-Wclass-memaccess): Document.
11593 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11595 * emit-rtl.h (is_leaf): Update comment about local
11596 register allocator.
11598 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11601 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11602 for a variable to have a section before checking if the section has a
11604 Set section to.persistent if persistent attribute is set.
11605 Warn if .persistent attribute is used on an automatic variable.
11607 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11609 PR rtl-optimization/80474
11610 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11612 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11614 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11617 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11619 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11620 (casesi): Emit bounds checking as RTL.
11621 (casesi_internal_mips16_<mode>): Remove bounds checking.
11623 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11625 * config/xtensa/xtensa.c (xtensa_option_override): Append
11626 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11627 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11628 xtensa_doloop_hooks): Define unconditionally.
11629 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11631 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11632 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11633 for it in xtensa_option_override.
11634 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11635 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11637 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11639 * doc/cppopts.texi: Document '-' special value to -MF.
11641 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11643 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11644 (cortex_a53_fconst): Likewise.
11645 (cortex_a53_fpmul): Likewise.
11646 (cortex_a53_f_load_64): Likewise.
11647 (cortex_a53_f_load_many): Likewise.
11648 (cortex_a53_advsimd_alu): Likewise.
11649 (cortex_a53_advsimd_alu_q): Likewise.
11650 (cortex_a53_advsimd_mul): Likewise.
11651 (cortex_a53_advsimd_mul_q): Likewise.
11652 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11653 Add missing fmul, r2f_cvt and fconst cases.
11655 2017-06-14 Richard Biener <rguenther@suse.de>
11657 PR middle-end/81088
11658 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11660 (fold_binary_loc): When associating do not treat pre-existing
11661 TREE_OVERFLOW on literal constants as a reason to allow
11662 TREE_OVERFLOW on associated literal constants.
11664 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11666 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11667 (MASK_FEATURES): New macro.
11668 * config/sparc/sparc.c (sparc_option_override): Remove the special
11669 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11671 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11673 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11674 a division of 0 if non-call exceptions are enabled.
11676 2017-06-14 Andrew Pinski <apinski@cavium.com>
11677 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11680 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11681 Improve vector initialization code gen for only variable case.
11683 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11685 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11687 2017-06-14 Richard Biener <rguenther@suse.de>
11689 PR tree-optimization/81083
11690 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11693 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11695 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11696 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11697 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11698 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11699 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11700 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11702 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11704 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11705 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11707 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11709 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11711 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11713 * config/rs6000/t-rtems: Don't handle SPE.
11715 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11717 * config/rs6000/t-linux: Don't handle SPE.
11719 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11721 * config/rs6000/eabispe.h: Delete file.
11723 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11725 * config/rs6000/t-spe: Delete file.
11727 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11729 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11730 (rs6000_legitimate_offset_address_p): Return false for anything in
11731 V2SImode or V2SFmode.
11733 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11735 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11736 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11737 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11739 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11740 (rs6000_legitimate_offset_address_p): Ditto.
11741 (rs6000_emit_move): Ditto.
11742 (rs6000_init_builtins): Remove V4HI_type_node.
11744 2017-06-13 Martin Liska <mliska@suse.cz>
11747 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11748 (gate_asan): Likewise.
11749 * asan.h (asan_no_sanitize_address_p): Remove the function.
11750 (sanitize_flags_p): New function.
11751 * builtins.def: Fix coding style.
11752 * common.opt: Use renamed enum value.
11753 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11754 * doc/extend.texi: Document no_sanitize attribute.
11755 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11756 to SANITIZE_UNDEFINED_NONDEFAULT.
11757 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11758 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11759 Use sanitize_flags_p.
11760 * gimplify.c (gimplify_function_tree): Likewise.
11761 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11762 * opts.c (parse_no_sanitize_attribute): New function.
11763 (common_handle_option): Use renamed enum value.
11764 * opts.h (parse_no_sanitize_attribute): Declare.
11765 * tree.c (sanitize_flags_p): New function.
11766 * tree.h: Declared here.
11767 * tsan.c: Use sanitize_flags_p.
11768 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11769 (instrument_mem_ref): Likewise.
11770 (instrument_bool_enum_load): Likewise.
11771 (do_ubsan_in_current_function): Remove the function.
11772 (pass_ubsan::execute): Use sanitize_flags_p.
11773 * ubsan.h: Remove do_ubsan_in_current_function
11774 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11775 (dump_function_to_file): Use it here.
11777 2017-06-13 Martin Jambor <mjambor@suse.cz>
11779 PR tree-optimization/80803
11780 PR tree-optimization/81063
11781 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11782 (propagate_subaccesses_across_link): Enqueue subtree whenever
11783 necessary instead of relying on the caller.
11785 2017-06-13 Martin Jambor <mjambor@suse.cz>
11787 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11788 that have a first_link.
11789 (sort_and_splice_var_accesses): Do not check first_link before
11791 (subtree_mark_written_and_enqueue): Likewise.
11792 (propagate_all_subaccesses): Likewise and do not stop at first
11793 parent with a first_link.
11795 2017-06-13 Martin Jambor <mjambor@suse.cz>
11797 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11800 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11802 * match.pd: New pattern.
11804 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11806 * tree-vrp.c (is_masked_range_test): New function.
11807 (register_edge_assert_for): Determine ranges for
11810 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11812 PR tree-optimization/67328
11813 * fold-const.c (maskable_range_p): New function.
11814 (build_range_check): Generate bittests if possible.
11816 2017-06-13 Martin Liska <mliska@suse.cz>
11818 * gimple-pretty-print.c (dump_probability): Add new argument.
11819 (dump_edge_probability): Dump both probability and count.
11820 (dump_gimple_label): Likewise.
11821 (dump_gimple_bb_header): Likewise.
11823 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11826 * config/avr/avr-devices.c: Fix indentation.
11827 * config/avr/gen-avr-mmcu-specs.c: Dito.
11829 2017-06-13 Richard Biener <rguenther@suse.de>
11831 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11832 instead get vector type from stmt_info.
11833 (vectorizable_reduction): Adjust. Remove dead code.
11835 2017-06-13 Richard Biener <rguenther@suse.de>
11837 PR middle-end/81065
11838 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11839 case of C * (x * C2 + C3).
11840 (fold_addr_of_array_ref_difference): Properly fold index difference.
11842 2017-06-12 David S. Miller <davem@davemloft.net>
11845 * config/sparc/sparc.md (return expander): Emit frame blockage if
11846 function uses alloca.
11848 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11850 * combine.c (make_field_assignment): Check len rather than the mode
11851 precision when calling force_to_mode.
11853 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11855 Support multilibs and devices that see flash in RAM address range.
11858 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11859 (avr_mcu_t) <flash_pm_offset>: New field.
11860 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11861 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11862 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11863 (AVR_TINY_PM_OFFSET): Remove macro.
11864 * config/avr/avr.opt (-mshort-calls): New option.
11865 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11866 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11867 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11868 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11869 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11870 instead of avr_arch->have_jmp_call.
11871 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11872 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11873 avr_arch->flash_pm_offset to define.
11874 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11875 new field flash_pm_offset. Add entry for avrxmega3.
11876 (avr_texinfo): Add entry for avrxmega3.
11877 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11878 attiny212, attiny214,
11879 attiny412, attiny414, attiny416, attiny417,
11880 attiny814, attiny816, attiny817,
11881 attiny1614, attiny1616, attiny1617,
11882 attiny3214, attiny3216, attiny3217.
11883 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11884 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11885 (avr_print_operand_address) [AVR_TINY]: Same.
11886 (avr_asm_init_sections) <readonly_data_section>: Only patch
11887 callback if avr_arch->flash_pm_offset = 0.
11888 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11889 for rodata if avr_arch->flash_pm_offset != 0.
11890 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11891 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11892 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11893 (m_options): Append opt_rcall.
11894 (m_dirnames): Append dir_rcall.
11895 * config/avr/t-multilib: Regenerate.
11897 * configure.ac [target=avr]: Check whether avrxmega3 default
11898 linker description file works as needed.
11899 * configure: Regenerate.
11900 * doc/avr-mmcu.texi: Regenerate.
11901 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11902 <__AVR_ARCH__>: Document avrxmega3 and 103.
11903 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11904 <__AVR_SHORT_CALLS__>: Document it.
11905 <__AVR_PM_BASE_ADDRESS__>: Document it.
11906 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11907 (AVR Variable Attributes) <progmem>: Document this is
11908 not needed for avrxmega3.
11909 (AVR Named Address Spaces) <__flash>: Dito.
11911 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11913 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11915 2017-06-12 Doug Rupp <rupp@adacore.com>
11917 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11918 Append vxworks-stdint.h to the tm_file list.
11919 * config/vxworks-stdint.h: New file.
11921 2017-06-12 Martin Liska <mliska@suse.cz>
11923 PR tree-optimization/81041
11924 * tree-profile.c (gimple_gen_ic_func_profiler):
11925 Create an extra BB in profile-generate
11926 (gimple_gen_time_profiler): Likewise.
11928 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11930 PR tree-optimization/81003
11931 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11932 (update_range_test): Use it instead of force_gimple_operand_gsi.
11934 2017-06-12 Richard Biener <rguenther@suse.de>
11936 PR tree-optimization/81053
11937 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11938 with backedge value not defined in loop. Simplify def stmt
11941 2017-06-11 Tom de Vries <tom@codesourcery.com>
11944 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11946 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11947 nvptx_cannot_force_const_mem.
11949 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11951 * opts.c (finish_options): Move test for flag_split_stack after
11952 it has been initialized.
11954 2017-06-11 Jason Merrill <jason@redhat.com>
11956 * tree.h (id_equal): New.
11957 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11958 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11959 instead of strcmp of IDENTIFIER_POINTER.
11961 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11963 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11964 (mark_all_inlined_calls_cdtor): Fix formating.
11965 (inline_transform): Rescale profile before inlining.
11967 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11969 * cgraph.h (cgraph_edge::clone): Update prototype.
11970 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11971 (cgraph_node::create_clone): Update.
11972 (cgraph_node::create_version_clone): Update.
11973 * tree-inline.c (copy_bb): Update.
11974 (expand_call_inline): Update.
11976 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11978 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11979 factored out from ...
11980 (rs6000_emit_prologue): ... here.
11982 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11984 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11985 factored out from ...
11986 (rs6000_emit_prologue): ... here.
11988 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11990 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11992 (handle_missing_profiles): Fix computation of tp_first_run.
11993 (counts_to_freqs): Do not touch freqs when count is 0.
11995 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11997 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12000 2017-06-10 Tom de Vries <tom@codesourcery.com>
12002 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12003 attributes): Document signal effective target.
12005 2017-06-10 Tom de Vries <tom@codesourcery.com>
12007 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12008 Document effective target stack_size.
12010 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12012 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12013 to the edit_context if they can be auto-applied.
12015 2017-06-9 Ian Lance Taylor <iant@golang.org>
12017 * opts.c (finish_options): If -fsplit-stack, disable implicit
12018 -forder-blocks-and-partition.
12019 * doc/invoke.texi (Optimize Options): Document that when using
12020 -fsplit-stack -forder-blocks-and-partition is not implicitly
12023 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12025 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12026 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12027 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12028 * builtins.def (abort, trap, unreachable): Declare cold.
12029 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12030 * tree-core.h (ECF_COLD): New.
12031 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12032 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12034 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12036 * predict.c (unlikely_executed_stmt_p): Cleanup.
12038 2017-06-09 Richard Biener <rguenther@suse.de>
12040 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12041 model if the ref is always written to.
12043 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12045 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12047 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12049 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12052 2017-06-09 Tom de Vries <tom@codesourcery.com>
12055 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12056 "target cannot support label values" when encountering LABEL_REF.
12058 2017-06-09 Martin Liska <mliska@suse.cz>
12060 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12061 (gimple_gen_ic_func_profiler): Emit direct comparison
12062 of __gcov_indirect_call_callee with NULL.
12063 (gimple_gen_time_profiler): Change probability from
12064 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12066 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12068 * profile.c (edge_gcov_counts): Turn to pointer.
12069 (compute_branch_probabilities, compute_branch_probabilities): Update.
12070 (branch_prob): Do not clear edge_gcov_count.
12071 * profile.h (edge_gcov_counts): Turn to pointer.
12072 (edge_gcov_count): Update.
12074 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12076 * gimple.h (gimple_check_failed): Mark cold.
12078 2017-06-09 Richard Biener <rguenther@suse.de>
12080 PR tree-optimization/66623
12081 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12082 refactor check_reduction into two parts, properly computing
12083 whether we have to check reduction validity for outer loop
12086 2017-06-09 Richard Biener <rguenther@suse.de>
12088 PR tree-optimization/79483
12089 * graphite-scop-detection.c (order): New global.
12090 (get_order): Compute bb to order mapping that satisfies code
12091 generation constraints.
12092 (cmp_pbbs): New helper.
12093 (build_scops): Start domwalk at entry block, sort generated
12096 2017-06-09 Richard Biener <rguenther@suse.de>
12098 PR middle-end/81007
12099 * ipa-polymorphic-call.c
12100 (ipa_polymorphic_call_context::restrict_to_inner_class):
12101 Skip FIELD_DECLs with error_mark_node type.
12102 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12105 2017-06-09 Martin Liska <mliska@suse.cz>
12107 * predict.c (struct branch_predictor): New struct.
12108 (test_prediction_value_range): New test.
12109 (predict_c_tests): New function.
12110 * selftest-run-tests.c (selftest::run_tests): Run the function.
12111 * selftest.h: Declare new tests.
12113 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12116 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12117 gen_add3_insn did not fail.
12118 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12119 r0, construct that number in a temporary reg and add that reg to r0.
12120 If asked to put the result in r0 as well, fail.
12122 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12124 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12125 for early expansion of vec_eqv.
12127 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12129 PR middle-end/81005
12130 * ubsan.c (instrument_null): Avoid pointless code temporary.
12131 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12133 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12137 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12140 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12143 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12145 * predict.c (unlikely_executed_edge_p): Move ahead.
12146 (probably_never_executed_edge_p): Use it.
12148 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12150 PR middle-end/79988
12151 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12152 gimple_call_builtin_p call.
12154 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12156 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12157 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12158 rtl_check_failed_type2, rtl_check_failed_code1,
12159 rtl_check_failed_code2, rtl_check_failed_code_mode,
12160 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12161 rtvec_check_failed_bounds, rtl_check_failed_flag,
12162 _fatal_insn_not_found, _fatal_insn): Likewise.
12163 * tree.h (tree_contains_struct_check_failed,
12164 tree_check_failed, tree_not_check_failed,
12165 tree_class_check_failed, tree_range_check_failed,
12166 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12167 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12168 tree_operand_check_failed, omp_clause_check_failed,
12169 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12172 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12174 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12175 flag_branch_probabilities.
12176 * ipa-inline.c (edge_badness): Likewise.
12177 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12178 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12179 * predict.c (maybe_hot_frequency_p): Likewise.
12180 (probably_never_executed): Likewise.
12181 * sched-ebb.c (schedule_ebbs): Likewise.
12182 * sched-rgn.c (find_single_block_region): Likewise.
12183 * tracer.c (tail_duplicate): Likewise.
12185 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12187 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12188 longer requires x_flag_profile_use.
12190 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12192 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12193 instead of flag_reorder_blocks_and_partition.
12194 * dbxout.c (dbxout_function_end): Likewise.
12195 * dwarf2out.c (gen_subprogram_die): Likewise.
12196 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12197 * hw-doloop.c (reorg_loops): Likewise.
12198 * varasm.c (assemble_start_function,
12199 assemble_end_function): Likewise.
12200 (decide_function_section): Do not check for
12201 flag_reorder_blocks_and_partition.
12203 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12205 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12207 (chkp_get_hard_register_fake_addr_expr): Ditto.
12208 (chkp_build_addr_expr): Add check for hard reg case.
12209 (chkp_parse_array_and_component_ref): Ditto.
12210 (chkp_find_bounds_1): Ditto.
12211 (chkp_process_stmt): Don't generate bounds store for
12214 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12216 * predict.c (maybe_hot_bb_p): Do not check profile status.
12217 (maybe_hot_edge_p): Likewise.
12218 (probably_never_executed): Check for zero counts even if profile
12220 (unlikely_executed_edge_p): New function.
12221 (unlikely_executed_stmt_p): New function.
12222 (unlikely_executed_bb_p): New function.
12223 (set_even_probabilities): Use unlikely predicates.
12224 (combine_predictions_for_bb): Likewise.
12225 (predict_paths_for_bb): Likewise.
12226 (predict_paths_leading_to_edge): Likewise.
12227 (determine_unlikely_bbs): New function.
12228 (estimate_bb_frequencies): Use it.
12229 (compute_function_frequency): Use zero counts even if profile is
12231 * profile-count.h: Fix typo.
12233 2017-08-08 Julia Koval <julia.koval@intel.com>
12235 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12236 _mm512_mask_cvtsepi16_storeu_epi8,
12237 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12238 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12239 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12240 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12241 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12242 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12243 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12244 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12245 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12246 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12247 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12248 __builtin_ia32_pmovuswb256mem_mask,
12249 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12250 __builtin_ia32_pmovwb512mem_mask): New builtins.
12252 2017-08-08 Julia Koval <julia.koval@intel.com>
12254 PR target/73350,80862
12255 * config/i386/subst.md (round): Fix round pattern.
12256 * config/i386/i386.c (ix86_erase_embedded_rounding):
12257 Fix erasing rounding for the fixed pattern.
12259 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12261 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12263 2017-06-08 Martin Liska <mliska@suse.cz>
12265 PR gcov-profile/80911
12266 * gcov.c (block_info::block_info): New constructor.
12268 2017-06-07 Carl Love <cel@us.ibm.com>
12270 * config/rs6000/rs6000-c: The return type of the following
12271 built-in functions was implemented as int not long long. Fix sign
12272 of return value for the unsigned version of vec_mulo and vec_mule.
12273 vector unsigned long long vec_bperm (vector unsigned long long,
12274 vector unsigned char)
12275 vector signed long long vec_mule (vector signed int,
12277 vector unsigned long long vec_mule (vector unsigned int,
12278 vector unsigned int)
12279 vector signed long long vec_mulo (vector signed int,
12281 vector unsigned long long vec_mulo (vector unsigned int,
12282 vector unsigned int)
12283 * doc/extend.texi: Fix the documentation for the built-in
12286 2017-06-07 Carl Love <cel@us.ibm.com>
12289 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12292 2017-06-07 Carl Love <cel@us.ibm.com>
12294 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12295 support, Generate doublehv for signed int/float for BE case only.
12297 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12299 * doc/invoke.texi (mcx16): Rewrite.
12301 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12303 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12304 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12305 *mov<mode>_softfloat, and an anonymous splitter): Use
12306 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12308 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12310 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12312 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12313 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12314 (rs6000_debug_reg_global): Adjust.
12315 (rs6000_init_hard_regno_mode_ok): Adjust.
12316 (rs6000_dbx_register_number): Adjust.
12317 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12318 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12319 Remove SPE_ACC and SPEFSCR.
12320 (REG_ALLOC_ORDER): Ditto.
12321 (FRAME_POINTER_REGNUM): Change to 111.
12322 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12323 (REG_CLASS_NAMES): Ditto.
12324 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12325 (REGISTER_NAMES): Ditto.
12326 (ADDITIONAL_REG_NAMES): Ditto.
12327 (rs6000_reg_names): Ditto.
12328 * config/rs6000/rs6000.md: Renumber some register number
12331 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12333 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12335 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12336 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12338 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12339 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12340 Delete the SPE high registers.
12341 (REG_ALLOC_ORDER): Ditto.
12342 (enum reg_class): Remove SPE_HIGH_REGS.
12343 (REG_CLASS_NAMES): Ditto.
12344 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12345 (REGISTER_NAMES): Ditto.
12346 (rs6000_reg_names): Ditto.
12347 * doc/tm.texi.in: Remove SPE as example.
12348 * doc/tm.texi: Regenerate.
12350 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12352 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12353 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12354 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12355 * config/rs6000/rs6000.md (type): Remove "brinc".
12357 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12359 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12360 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12361 * config/rs6000/linuxspe.h: Delete file.
12362 * config/rs6000/rs6000.md: Don't include spe.md.
12363 * config/rs6000/spe.h: Delete file.
12364 * config/rs6000/spe.md: Delete file.
12365 * config/rs6000/t-rs6000: Remove spe.md.
12367 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12369 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12370 (reg_or_none500mem_operand): Delete.
12371 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12372 instead of reg_or_none500mem_operand.
12374 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12376 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12377 handling of SPE flags.
12378 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12380 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12382 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12384 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12385 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12386 paired_divv2sf3): Similar.
12387 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12388 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12389 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12391 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12392 Rename the paired_* instruction patterns.
12393 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12395 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12396 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12397 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12398 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12399 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12400 PAIRED_VECTOR_MODE.
12401 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12402 (spe_func_has_64bit_regs_p): Delete.
12403 (spe_expand_predicate_builtin): Delete.
12404 (spe_expand_evsel_builtin): Delete.
12405 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12406 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12407 (invalid_e500_subreg): Delete.
12408 (rs6000_legitimize_address): Always force_reg op2 as well, for
12409 paired single memory accesses.
12410 (rs6000_member_type_forces_blk): Delete.
12411 (rs6000_spe_function_arg): Delete.
12412 (rs6000_expand_unop_builtin): Delete SPE handling.
12413 (rs6000_expand_binop_builtin): Ditto.
12414 (spe_expand_stv_builtin): Delete.
12415 (bdesc_2arg_spe): Delete.
12416 (spe_expand_builtin): Delete.
12417 (spe_expand_predicate_builtin): Delete.
12418 (spe_expand_evsel_builtin): Delete.
12419 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12420 (spe_init_builtins): Delete.
12421 (spe_func_has_64bit_regs_p): Delete.
12422 (savres_routine_name): Delete "info" parameter. Adjust callers.
12423 (rs6000_emit_stack_reset): Ditto.
12424 (rs6000_dwarf_register_span): Delete.
12425 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12426 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12427 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12429 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12431 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12432 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12433 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12434 mulv2sf3, divv2sf3): Delete expanders.
12436 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12438 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12440 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12442 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12443 * config/rs6000/rs6000.c: Ditto.
12445 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12447 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12448 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12449 comparison_operator.
12451 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12453 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12454 * config/rs6000/rs6000.opt: Ditto.
12455 * config/rs6000/t-rtems: Ditto.
12457 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12459 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12460 TARGET_E500_SINGLE by 0, simplify.
12461 * config/rs6000/rs6000.c: Ditto.
12462 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12463 (spe_build_register_parallel): Delete.
12464 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12465 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12466 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12467 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12468 (E500_CONVERT): Delete.
12469 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12471 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12473 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12474 * config/rs6000/dfp.md: Ditto.
12475 (negdd2, *negdd2_fpr): Merge.
12476 (absdd2, *absdd2_fpr): Merge.
12477 (negtd2, *negtd2_fpr): Merge.
12478 (abstd2, *abstd2_fpr): Merge.
12479 * config/rs6000/e500.h: Delete file.
12480 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12481 TARGET_FPRS by 1 and simplify.
12482 * config/rs6000/rs6000-c.c: Ditto.
12483 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12484 TARGET_DF_SPE by 0.
12485 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12487 * config/rs6000/rs6000.md: Ditto.
12488 (floatdidf2, *floatdidf2_fpr): Merge.
12489 (move_from_CR_gt_bit): Delete.
12490 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12491 (E500_CR_IOR_COMPARE): Delete.
12492 (All patterns that require !TARGET_FPRS): Delete.
12493 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12495 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12497 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12499 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12501 * graphds.c (add_edge): Intitialize edge's attached data.
12502 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12503 pointer parameter. Call pointed function on each edge during
12504 graph traversing. Skip traversing the edge when the function
12506 (graphds_dfs, graphds_scc): Ditto.
12507 (for_each_edge): New parameter. Pass the new parameter to callback
12509 * graphds.h (skip_edge_callback): New function pointer type.
12510 (graphds_dfs, graphds_scc): New function pointer parameter.
12511 (graphds_edge_callback, for_each_edge): New parameter.
12513 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12515 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12516 out code checking if runtime alias check is possible to below ...
12517 Call the new function.
12518 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12519 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12521 2017-06-07 Marek Polacek <polacek@redhat.com>
12524 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12525 TYPE_OVERFLOW_WRAPS check.
12527 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12529 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12530 if versioning is required.
12531 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12532 peeling with the check for versioning.
12534 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12536 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12537 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12538 Set true to new parameter if new ssa variable is defined.
12539 (vect_gen_vector_loop_niters): Refactor. Set range information
12540 for the new vector loop bound variable.
12541 (vect_do_peeling): Ditto.
12543 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12545 * tree-affine.c (ssa.h): Include header file.
12546 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12547 has wrapping overflow behavior.
12549 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12551 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12553 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12555 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12556 (tree_to_aff_combination): ... here.
12558 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12560 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12561 reg_pressure model function.
12562 (ivopts_global_cost_for_size): Delete.
12563 (determine_set_costs, iv_ca_recount_cost): Call new model function
12564 ivopts_estimate_reg_pressure.
12566 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12568 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12569 expensive than udiv. Remove floating point cases from mod.
12571 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12573 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12574 Increase idiv cost.
12576 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12578 * config/aarch64/aarch64.md
12579 (copysignsf3): Fix mask generation.
12581 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12583 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12585 (class dump_manager): Add register_dumps method.
12586 * dumpfile.c: Include langhooks.h.
12587 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12588 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12589 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12590 (dump_manager::dump_register): Start with 512 entries instead of 32.
12591 (dump_manager::register_dumps): New method.
12592 * toplev.c (general_init): Instead of invoking register_dumps
12593 langhook, invoke register_dumps method on the dump manager.
12594 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12597 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12599 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12600 Say that # requires an associated define_split to exist, and that
12601 the define_split must be suitable for use after register allocation.
12603 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12605 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12606 (compute_outgoing_frequencies): Also initialize zero counts.
12607 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12608 around loops; preserve more of profile when nothing changes.
12610 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12612 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12614 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12615 * config/arm/arm-cpu-cdata.h: Regenerate.
12616 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12617 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12618 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12619 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12620 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12622 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12624 * config/arm/t-rmprofile: Likewise.
12625 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12627 2017-06-06 David S. Miller <davem@davemloft.net>
12630 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12631 blockage if function uses alloca.
12633 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12635 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12636 New "uid" fields to hold pretty-print IDs of group and ref.
12637 Memory references are now identified as <group_id>:<ref_id>
12638 instead of using [random] addresses.
12639 (dump_mem_details): Simplify, no functional change.
12640 (dump_mem_ref): Simplify and make output more concise.
12641 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12642 (find_or_create_group): Initialize group uid.
12643 (record_ref): Initialize ref uid. Improve debug output.
12644 (prune_group_by_reuse, should_issue_prefetch_p,)
12645 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12646 (mark_nontemporal_store, determine_loop_nest_reuse):
12647 Improve debug output.
12649 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12651 * dbgcnt.def (prefetch): New debug counter.
12652 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12653 (schedule_prefetches): Stop issueing prefetches if debug counter
12656 2017-06-06 Tom de Vries <tom@codesourcery.com>
12658 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12659 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12661 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12663 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12664 Use aarch64_reg_or_zero predicate for operand 4.
12665 (aarch64_compare_and_swap<mode> define_insn_and_split):
12666 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12667 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12669 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12671 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12672 (arm_compute_save_core_reg_mask): This.
12673 (thumb1_compute_save_reg_mask): Rename into ...
12674 (thumb1_compute_save_core_reg_mask): This.
12675 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12676 (arm_compute_frame_layout): Likewise.
12678 2017-06-06 Richard Biener <rguenther@suse.de>
12680 PR tree-optimization/80974
12681 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12682 keep or clear leaders SSA info.
12684 2017-06-06 Tom de Vries <tom@codesourcery.com>
12686 * config/nvptx/nvptx.c (split_mode_p): New function.
12687 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12689 2017-06-06 Tom de Vries <tom@codesourcery.com>
12691 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12693 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12696 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12699 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12701 * shrink-wrap.c (handle_simple_exit): Update profile.
12702 (try_shrink_wrapping): Upate profile.
12704 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12706 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12707 (tree_guess_outgoing_edge_probabilities): New.
12708 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12709 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12711 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12713 * ipa-split.c (split_function): Initialize return bb profile.
12715 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12717 * profile.c (compute_branch_probabilities): Also initialize
12718 EXIT_BLOCK profile.
12720 2017-06-06 Richard Biener <rguenther@suse.de>
12722 PR tree-optimization/80928
12723 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12724 (vect_analyze_loop_operations): Properly guard analysis for
12726 (vect_transform_loop): Likewise.
12727 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12728 (vect_model_induction_cost): Do not cost for pure SLP.
12729 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12730 of induction in inner loop vectorization.
12731 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12732 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12733 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12735 (vect_analyze_slp_cost_1): Cost induction.
12736 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12737 (vect_get_slp_vect_defs): Likewise.
12738 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12739 (vect_transform_stmt): Handle SLP reductions.
12740 * tree-vectorizer.h (vectorizable_induction): Adjust.
12742 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12744 * config/rs6000/rs6000.c (make_resolver_func): Update
12745 init_lowered_empty_function call.
12747 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12749 * doc/invoke.texi: Document the -fprofile-abs-path option.
12750 * common.opt (fprofile-abs-path): New option.
12751 * gcov-io.h (gcov_write_filename): Declare.
12752 * gcov-io.c (gcov_write_filename): New function.
12753 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12754 * profile.c (output_location): Likewise.
12756 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12758 * shring-wrap.c: Revert accidental commit.
12760 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12762 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12764 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12766 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12768 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12769 profile in callgraph edge.
12770 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12771 (apply_scale): Likewise.
12772 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12773 Also scale profile when inlining function with zero profile.
12774 (initialize_cfun): Update exit block profile even when it is zero.
12775 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12776 when profile is read.
12778 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12780 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12781 (CLONE_*): New constants to define the processors we can generate
12782 code for with the target_clone attribute.
12783 (rs6000_clone_map): New array to identify which clone processors
12784 the current program is running on.
12785 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12786 target_clone attribute.
12787 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12788 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12789 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12790 (cpu_expand_builtin): Add support for target_clone attribute.
12791 (rs6000_valid_attribute_p): Allow "default" attribute.
12792 (get_decl_name): New debug function to simplify printing the
12793 current function name in debugging statements.
12794 (rs6000_clone_priority): New functions to support the target_clone
12795 attribute, and be able to generate code to switch between ISA 2.05
12796 through ISA 3.0 (power6 through power9).
12797 (rs6000_compare_version_priority): Likewise.
12798 (rs6000_get_function_versions_dispatcher): Likewise.
12799 (make_resolver_func): Likewise.
12800 (add_condition_to_bb): Likewise.
12801 (dispatch_function_versions): Likewise.
12802 (rs6000_generate_version_dispatcher_body): Likewise.
12803 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12804 (fusion_gpr_load_p): Fix a spacing issue.
12805 * doc/extend.texi (Common Function Attributes): Document that the
12806 PowerPC supports the target_clone attribute.
12808 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12810 * config/arm/arm.h: explain F symbol found in description of ARM
12811 register allocation in its legend.
12813 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12815 * config/mips/frame-header-opt.c: Include profile-count.h.
12816 * config/riscv/riscv.c: Include profile-count.h
12818 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12820 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12822 (sm_set_flag_if_changed): Add bbs field.
12823 (execute_sm_if_changed_flag_set): Pass BBS.
12824 (execute_sm): Update.
12826 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12828 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12831 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12833 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12834 (peephole2): New peephole2 to emit the above.
12835 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12837 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12839 * config/aarch64/aarch64.c (define_peephole2 above
12840 *sub_<shift>_<mode>): New peephole.
12842 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12844 * config/i386/i386.c (make_resolver_func): Update.
12845 * Makefile.in: Add profile-count.h and profile-count.o
12846 * auto-profile.c (afdo_indirect_call): Update to new API.
12847 (afdo_set_bb_count): Update.
12848 (afdo_propagate_edge): Update.
12849 (afdo_propagate_circuit): Update.
12850 (afdo_calculate_branch_prob): Update.
12851 (afdo_annotate_cfg): Update.
12852 * basic-block.h: Include profile-count.h
12853 (struct edge_def): Turn count to profile_count.
12854 (struct basic_block_def): Likewie.
12855 (REG_BR_PROB_BASE): Move to profile-count.h
12856 (RDIV): Move to profile-count.h
12857 * bb-reorder.c (max_entry_count): Turn to profile_count.
12858 (find_traces): Update.
12859 (rotate_loop):Update.
12860 (connect_traces):Update.
12861 (sanitize_hot_paths):Update.
12862 * bt-load.c (migrate_btr_defs): Update.
12863 * cfg.c (RDIV): Remove.
12864 (init_flow): Use alloc_block.
12865 (alloc_block): Uninitialize count.
12866 (unchecked_make_edge): Uninitialize count.
12867 (check_bb_profile): Update.
12868 (dump_edge_info): Update.
12869 (dump_bb_info): Update.
12870 (update_bb_profile_for_threading): Update.
12871 (scale_bbs_frequencies_int): Update.
12872 (scale_bbs_frequencies_gcov_type): Update.
12873 (scale_bbs_frequencies_profile_count): New.
12874 * cfg.h (update_bb_profile_for_threading): Update.
12875 (scale_bbs_frequencies_profile_count): Declare.
12876 * cfgbuild.c (compute_outgoing_frequencies): Update.
12877 (find_many_sub_basic_blocks): Update.
12878 * cfgcleanup.c (try_forward_edges): Update.
12879 (try_crossjump_to_edge): Update.
12880 * cfgexpand.c (expand_gimple_tailcall): Update.
12881 (construct_exit_block): Update.
12882 * cfghooks.c (verify_flow_info): Update.
12883 (dump_bb_for_graph): Update.
12884 (split_edge): Update.
12885 (make_forwarder_block): Update.
12886 (duplicate_block): Update.
12887 (account_profile_record): Update.
12888 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12889 (get_estimated_loop_iterations): Update.
12890 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12891 (single_likely_exit): Update.
12892 * cfgloopmanip.c (scale_loop_profile): Update.
12894 (set_zero_probability): Update.
12895 (lv_adjust_loop_entry_edge): Update.
12896 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12897 (purge_dead_edges): Update.
12898 (rtl_account_profile_record): Update.
12899 * cgraph.c (cgraph_node::create): Uninitialize count.
12900 (symbol_table::create_edge): Uninitialize count.
12901 (cgraph_update_edges_for_call_stmt_node): Update.
12902 (cgraph_edge::dump_edge_flags): Update.
12903 (cgraph_node::dump): Update.
12904 (cgraph_edge::maybe_hot_p): Update.
12905 * cgraph.h: Include profile-count.h
12906 (create_clone), create_edge, create_indirect_edge): Update.
12907 (cgraph_node): Turn count to profile_count.
12908 (cgraph_edge0: Likewise.
12909 (make_speculative, clone): Update.
12910 (create_edge): Update.
12911 (init_lowered_empty_function): Update.
12912 * cgraphclones.c (cgraph_edge::clone): Update.
12913 (duplicate_thunk_for_node): Update.
12914 (cgraph_node::create_clone): Update.
12915 * cgraphunit.c (cgraph_node::analyze): Update.
12916 (cgraph_node::expand_thunk): Update.
12917 * final.c (dump_basic_block_info): Update.
12918 * gimple-streamer-in.c (input_bb): Update.
12919 * gimple-streamer-out.c (output_bb): Update.
12920 * graphite.c (print_global_statistics): Update.
12921 (print_graphite_scop_statistics): Update.
12922 * hsa-brig.c: Include basic-block.h.
12923 * hsa-dump.c: Include basic-block.h.
12924 * hsa-gen.c (T sum_slice): Update.
12925 (convert_switch_statements):Update.
12926 * hsa-regalloc.c: Include basic-block.h.
12927 * ipa-chkp.c (chkp_produce_thunks): Update.
12928 * ipa-cp.c (struct caller_statistics): Update.
12929 (init_caller_stats): Update.
12930 (gather_caller_stats): Update.
12931 (ipcp_cloning_candidate_p): Update.
12932 (good_cloning_opportunity_p): Update.
12933 (get_info_about_necessary_edges): Update.
12934 (dump_profile_updates): Update.
12935 (update_profiling_info): Update.
12936 (update_specialized_profile): Update.
12937 (perhaps_add_new_callers): Update.
12938 (decide_about_value): Update.
12939 (ipa_cp_c_finalize): Update.
12940 * ipa-devirt.c (struct odr_type_warn_count): Update.
12941 (struct decl_warn_count): Update.
12942 (struct final_warning_record): Update.
12943 (possible_polymorphic_call_targets): Update.
12944 (ipa_devirt): Update.
12945 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12946 * ipa-icf.c (sem_function::merge): Update.
12947 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12948 * ipa-inline.c (compute_uninlined_call_time): Update.
12949 (compute_inlined_call_time): Update.
12950 (want_inline_small_function_p): Update.
12951 (want_inline_self_recursive_call_p): Update.
12952 (edge_badness): Update.
12953 (lookup_recursive_calls): Update.
12954 (recursive_inlining): Update.
12955 (inline_small_functions): Update.
12956 (dump_overall_stats): Update.
12957 (dump_inline_stats): Update.
12958 * ipa-profile.c (ipa_profile_generate_summary): Update.
12959 (ipa_propagate_frequency): Update.
12960 (ipa_profile): Update.
12961 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12962 * ipa-utils.c (ipa_merge_profiles): Update.
12963 * loop-doloop.c (doloop_modify): Update.
12964 * loop-unroll.c (report_unroll): Update.
12965 (unroll_loop_runtime_iterations): Update.
12966 * lto-cgraph.c (lto_output_edge): Update.
12967 (lto_output_node): Update.
12968 (input_node): Update.
12969 (input_edge): Update.
12970 (merge_profile_summaries): Update.
12971 * lto-streamer-in.c (input_cfg): Update.
12972 * lto-streamer-out.c (output_cfg): Update.
12973 * mcf.c (create_fixup_graph): Update.
12974 (adjust_cfg_counts): Update.
12975 (sum_edge_counts): Update.
12976 * modulo-sched.c (sms_schedule): Update.
12977 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12978 * predict.c (maybe_hot_count_p): Update.
12979 (probably_never_executed): Update.
12980 (dump_prediction): Update.
12981 (combine_predictions_for_bb): Update.
12982 (propagate_freq): Update.
12983 (handle_missing_profiles): Update.
12984 (counts_to_freqs): Update.
12985 (rebuild_frequencies): Update.
12986 (force_edge_cold): Update.
12987 * predict.h: Include profile-count.h
12988 (maybe_hot_count_p, counts_to_freqs): UPdate.
12989 * print-rtl-function.c: Do not include cfg.h
12990 * print-rtl.c: Include basic-block.h
12991 * profile-count.c: New file.
12992 * profile-count.h: New file.
12993 * profile.c (is_edge_inconsistent): Update.
12994 (correct_negative_edge_counts): Update.
12995 (is_inconsistent): Update.
12996 (set_bb_counts): Update.
12997 (read_profile_edge_counts): Update.
12998 (compute_frequency_overlap): Update.
12999 (compute_branch_probabilities): Update; Initialize and deinitialize
13001 (branch_prob): Update.
13002 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13003 (edge_gcov_count): New.
13004 (bb_gcov_count): New.
13005 * shrink-wrap.c (try_shrink_wrapping): Update.
13006 * tracer.c (better_p): Update.
13007 * trans-mem.c (expand_transaction): Update.
13008 (ipa_tm_insert_irr_call): Update.
13009 (ipa_tm_insert_gettmclone_call): Update.
13010 * tree-call-cdce.c: Update.
13011 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13012 (gimple_duplicate_sese_tail): Update.
13013 (gimple_account_profile_record): Update.
13014 (execute_fixup_cfg): Update.
13015 * tree-inline.c (copy_bb): Update.
13016 (copy_edges_for_bb): Update.
13017 (initialize_cfun): Update.
13018 (freqs_to_counts): Update.
13019 (copy_cfg_body): Update.
13020 (expand_call_inline): Update.
13021 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13022 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13023 (try_unroll_loop_completely): Update.
13024 (try_peel_loop): Update.
13025 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13026 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13027 * tree-ssa-loop-split.c (connect_loops): Update.
13028 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13029 * tree-ssa-reassoc.c (branch_fixup): Update.
13030 * tree-ssa-tail-merge.c (replace_block_by): Update.
13031 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13032 (compute_path_counts): Update.
13033 (update_profile): Update.
13034 (recompute_probabilities): Update.
13035 (update_joiner_offpath_counts): Update.
13036 (estimated_freqs_path): Update.
13037 (freqs_to_counts_path): Update.
13038 (clear_counts_path): Update.
13039 (ssa_fix_duplicate_block_edges): Update.
13040 (duplicate_thread_path): Update.
13041 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13042 (struct switch_conv_info): Update.
13043 * tree-tailcall.c (decrease_profile): Update.
13044 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13045 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13046 * value-prof.c (check_counter): Update.
13047 (gimple_divmod_fixed_value): Update.
13048 (gimple_mod_pow2): Update.
13049 (gimple_mod_subtract): Update.
13050 (gimple_ic_transform): Update.
13051 (gimple_stringop_fixed_value): Update.
13052 * value-prof.h (gimple_ic): Update.
13054 2017-06-02 Carl Love <cel@us.ibm.com>
13056 * config/rs6000/rs6000-c: Add support for built-in functions
13057 vector double vec_doublee (vector signed int);
13058 vector double vec_doublee (vector unsigned int);
13059 vector double vec_doublee (vector float);
13060 vector double vec_doubleh (vector signed int);
13061 vector double vec_doubleh (vector unsigned int);
13062 vector double vec_doubleh (vector float);
13063 vector double vec_doublel (vector signed int);
13064 vector double vec_doublel (vector unsigned int);
13065 vector double vec_doublel (vector float);
13066 vector double vec_doubleo (vector signed int);
13067 vector double vec_doubleo (vector unsigned int);
13068 vector double vec_doubleo (vector float);.
13069 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13070 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13072 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13073 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13074 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13076 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13077 vec_doublel, vec_doubleh.
13078 * doc/extend.texi: Update the built-in documentation file for the
13079 new built-in functions.
13081 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13084 * ipa-inline-analysis.c (free_growth_caches): Set
13085 edge_removal_hook_holder to NULL after removing it.
13087 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13089 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13090 comparision with zero.
13092 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13093 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13094 for early expansion of vec_min and vec_max builtins.
13095 (builtin_function_type): Add min/max unsigned variants to those
13096 identified as having unsigned arguments.
13098 2017-06-02 Olivier Hainque <hainque@adacore.com>
13100 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13102 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13104 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13105 Use VALL_F16 iterator rather than VALL.
13107 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13109 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13110 Emit CBNZ inside loop when doing a strong exchange and comparing
13111 against zero. Generate the CC flags after the loop.
13113 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13115 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13116 (dl_section_ref): New.
13117 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13118 On AIX, append an expression to subtract the size of the
13119 section length to dl_section_ref.
13121 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13123 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13124 for early expansion of vector absolute builtins.
13126 2017-06-02 Richard Biener <rguenther@suse.de>
13128 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13129 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13131 2017-06-02 Richard Biener <rguenther@suse.de>
13133 PR tree-optimization/80948
13134 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13135 stmt order as well.
13137 2017-06-02 Richard Biener <rguenther@suse.de>
13139 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13141 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13142 for inductions as relevant.
13144 2017-06-02 Richard Biener <rguenther@suse.de>
13146 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13147 (vectorizable_induction): ... this. Remove dead code.
13149 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13151 * builtins. (expand_builtin_alloca): Remove second parameter and
13152 infer its value from the first parameter instead.
13153 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13155 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13157 PR rtl-optimization/80903
13158 * loop-doloop.c (add_test): Unshare sequence.
13160 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13162 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13164 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13166 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13168 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13169 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13170 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13171 (xlogue_layout::get_stub_rtx): Make static.
13172 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13173 (xlogue_layout::compute_stub_managed_regs): Rename to...
13174 (xlogue_layout::count_stub_managed_regs): ...this.
13175 (xlogue_layout::is_stub_managed_reg): New function.
13176 (xlogue_layout::m_stub_names): Rename to...
13177 (xlogue_layout::s_stub_names): ...this, make static.
13178 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13179 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13180 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13181 xlogue_layout::s_stub_names): Instantiate statics.
13182 (stub_managed_regs): Remove.
13183 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13184 (disable_call_ms2sysv_xlogues): Rename to...
13185 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13186 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13188 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13189 change after reload_completed.
13190 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13192 (ix86_expand_prologue): Likewise.
13193 (ix86_expand_epilogue): Likewise.
13194 (ix86_expand_split_stack_prologue): Likewise.
13195 (ix86_compute_frame_layout): Remove frame parameter ...
13196 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13197 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13199 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13200 (ix86_frame): Move from here ...
13201 * config/i386/i386.h (ix86_frame): ... to here.
13202 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13203 complete ix86_frame data structure instead. Remove some_ld_name.
13205 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13207 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13208 symbols that hold a DECL_VALUE_EXPR.
13210 2017-06-01 Martin Jambor <mjambor@suse.cz>
13212 PR tree-optimization/80898
13213 * tree-sra.c (process_subtree_disqualification): Removed.
13214 (disqualify_candidate): Do not acll
13215 process_subtree_disqualification.
13216 (subtree_mark_written_and_enqueue): New function.
13217 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13218 RHS has been disqualified and re-queue LHS if necessary. Apart
13219 from that, ignore disqualified RHS.
13221 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13223 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13224 address fetch for z10 or later.
13226 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13228 * config/arc/arc.md (tst_movb): Add guard when splitting.
13230 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13232 * config/arc/arc.c (arc_can_eliminate): Test against
13233 arc_frame_pointer_needed.
13235 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13237 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13238 to prevent store reordering.
13239 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13240 (type): Add block type.
13241 (stack_tie): Define special instruction to be used in
13244 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13246 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13247 constraint. It is not valid for the pattern.
13248 (noncommutative_binary_comparison): Likewise.
13250 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13252 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13255 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13257 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13258 be used by the reg-alloc.
13260 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13262 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13263 reg-alloc when having mul64 or mul32x16 instructions.
13264 (mulsidi3): Likewise.
13265 (umulsidi3): Likewise.
13266 (mulsi32x16): New pattern.
13267 (mulsi64): Likewise.
13268 (mulsidi64): Likewise.
13269 (umulsidi64): Likewise.
13270 (MUL32x16_REG): Define.
13271 (mul64_600): Use MUL32x16_REG.
13272 (mac64_600): Likewise.
13273 (umul64_600): Likewise.
13274 (umac64_600): Likewise.
13276 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13278 * config/arc/arc.md (mulsi3_700): Make it commutative.
13280 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13282 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13284 (*sign_extendsidi2_insn): Likewise for movstosw.
13286 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13288 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13289 the type of the input discriminant value. Convert the
13290 discriminant value of signedness vary.
13292 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13294 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13295 Add to -Wall section.
13297 2017-06-01 Richard Biener <rguenther@suse.de>
13299 PR middle-end/66313
13300 * fold-const.c (fold_plusminus_mult_expr): If the factored
13301 factor may be zero use a wrapping type for the inner operation.
13302 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13303 and handle moved defs.
13304 (process_assignment): Properly guard the unary op case. Return a
13305 tri-state indicating that moving the stmt before the call may allow
13306 to continue. Pass through to_move.
13307 (find_tail_calls): Handle moving unrelated defs before
13310 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13313 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13314 splitter result in the canonical way.
13316 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13318 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13319 also for 32bit target. Update insn attributes.
13320 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13322 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13324 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13325 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13326 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13327 (_mm_maskz_min_ss): New intrinsics.
13329 2017-05-31 Martin Liska <mliska@suse.cz>
13331 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13332 Change comment style to one we normally use.
13333 (vectorizable_reduction): Likewise.
13334 (vectorizable_induction): Likewise.
13335 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13336 (vectorizable_call): Likewise.
13337 (vectorizable_simd_clone_call): Likewise.
13338 (vectorizable_conversion): Likewise.
13339 (vectorizable_assignment): Likewise.
13340 (vectorizable_shift): Likewise.
13341 (vectorizable_operation): Likewise.
13342 (vectorizable_store): Likewise.
13343 (vectorizable_load): Likewise.
13344 * tree-vectorizer.h: Likewise.
13346 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13348 * passes.c (emergency_dump_function): New.
13349 * tree-pass.h (emergency_dump_function): Declare.
13350 * plugin.c (plugins_internal_error_function): Remove.
13351 * plugin.h (plugins_internal_error_function): Remove declaration.
13352 * toplev.c (internal_error_function): New static function. Use it...
13353 (general_init): ...here.
13355 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13357 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13358 (arc_rtx_costs): Add costs for new patterns.
13359 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13360 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13362 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13364 * tree-ssa-strlen.c (get_next_strinfo): New function.
13365 (get_stridx_plus_constant): Use it.
13366 (zero_length_string): Likewise.
13367 (adjust_related_strinfos): Likewise.
13368 (adjust_last_stmt): Likewise.
13370 2017-05-31 Richard Biener <rguenther@suse.de>
13373 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13374 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13376 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13378 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13379 loop_vinfo argument and use of dependence distance vectors.
13380 Check instead whether the two references differ only in their
13381 initial value and assume that they have the same alignment if the
13382 difference is a multiple of the vector alignment.
13383 (vect_analyze_data_refs_alignment): Update call accordingly.
13385 2017-05-31 Martin Liska <mliska@suse.cz>
13388 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13390 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13392 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13393 (create_intersect_range_checks): Move from ...
13394 * tree-data-ref.c (create_intersect_range_checks_index)
13395 (create_intersect_range_checks): ... to here.
13396 (create_runtime_alias_checks): New function factored from ...
13397 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13398 here. Call above function.
13399 * tree-data-ref.h (create_runtime_alias_checks): New function.
13401 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13403 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13404 segment length for dr_b and compute it in wide_int.
13406 2017-05-31 Richard Biener <rguenther@suse.de>
13408 PR tree-optimization/80906
13409 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13410 and pass through iv_map.
13411 (copy_bb_and_scalar_dependences): Adjust.
13412 (translate_pending_phi_nodes): Likewise.
13413 (copy_loop_close_phi_args): Handle code-generating IVs instead
13416 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13418 * diagnostic-color.c (color_dict): Add "type-diff".
13419 (parse_gcc_colors): Update comment.
13420 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13421 -fdiagnostics-show-template-tree and -fno-elide-type.
13422 (GCC_COLORS): Add type-diff to example.
13424 (-fdiagnostics-show-template-tree): New.
13425 (-fno-elide-type): New.
13426 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13427 the pp_format_decoder callback. Call any m_format_postprocessor's
13429 (pretty_printer::pretty_printer): Initialize
13430 m_format_postprocessor.
13431 (pretty_printer::~pretty_printer): Delete any
13432 m_format_postprocessor.
13433 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13434 (class format_postprocessor): New class.
13435 (struct pretty_printer::format_decoder): Document the new parameters.
13436 (struct pretty_printer::m_format_postprocessor): New field.
13437 * tree-diagnostic.c (default_tree_printer): Update for new
13438 bool and const char ** params.
13439 * tree-diagnostic.h (default_tree_printer): Likewise.
13441 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13443 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13444 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13445 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13446 rs6000_gen_cell_microcode code.
13447 (rs6000_final_prescan_insn): Delete.
13448 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13449 "warn-cell-microcode" entries.
13450 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13451 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13452 throughout. Change cc_reg_not_micro_cr0_operand to
13453 cc_reg_not_cr0_operand throughout.
13454 (*extendhi<mode>2_noload): Delete.
13455 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13456 (mwarn-cell-microcode): Delete.
13457 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13458 -mgen-cell-microcode and -mwarn-cell-microcode.
13460 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13463 * config/i386/constraints.md (Yd): New constraint.
13465 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13466 and (?Yd, r) alternatives. Update insn attributes.
13467 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13468 and (?*Yd, r) alternatives. Update insn attributes.
13469 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13471 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13473 * gimplify.c (gimplify_modify_expr): Don't create a
13474 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13477 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13479 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13481 2017-05-30 Richard Biener <rguenther@suse.de>
13483 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13484 and reduc_def fields.
13485 (STMT_VINFO_REDUC_TYPE): New define.
13486 (STMT_VINFO_REDUC_DEF): Likewise.
13487 (vect_force_simple_reduction): Adjust prototype.
13488 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13489 (vect_is_simple_reduction): Remove check_reduction argument.
13490 (vect_force_simple_reduction): Adjust and set
13491 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13492 (vectorizable_reduction): Do not re-do reduction analysis
13493 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13494 * tree-parloops.c (gather_scalar_reductions): Adjust.
13496 2017-05-30 Richard Biener <rguenther@suse.de>
13498 PR middle-end/80901
13499 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13502 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13504 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13505 Introduce unknown_misalignment parameter and remove vf.
13506 (vect_peeling_hash_get_lowest_cost):
13507 Pass unknown_misalignment parameter.
13508 (vect_enhance_data_refs_alignment):
13509 Fix unsupportable data ref treatment.
13511 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13513 * tree-vect-data-refs.c (vect_get_data_access_cost):
13514 Workaround for SLP handling.
13515 (vect_enhance_data_refs_alignment):
13516 Compute costs for doing no peeling at all, compare to the best
13517 peeling costs so far and avoid peeling if cheaper.
13519 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13521 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13522 Return peeling info and set costs to zero for unlimited cost
13524 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13525 with unknown misalignment. Compute and costs for unknown
13526 misalignment, compare them to the costs for known misalignment
13527 and choose the cheapest for peeling.
13529 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13531 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13532 (vect_get_peeling_costs_all_drs): Create function.
13533 (vect_peeling_hash_get_lowest_cost):
13534 Use vect_get_peeling_costs_all_drs.
13535 (vect_peeling_supportable): Create function.
13536 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13538 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13540 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13541 DR_HAS_NEGATIVE_STEP.
13542 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13543 (vect_enhance_data_refs_alignment): Use.
13544 (vect_duplicate_ssa_name_ptr_info): Use.
13545 * tree-vectorizer.h (dr_misalignment): Use.
13546 (known_alignment_for_access_p): Use.
13548 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13551 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13553 (has_section_name): New function.
13555 2017-05-30 Martin Liska <mliska@suse.cz>
13558 * auto-profile.c (get_function_decl_from_block): Fix
13561 2017-05-30 Richard Biener <rguenther@suse.de>
13563 PR middle-end/80876
13564 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13566 2017-05-30 Martin Liska <mliska@suse.cz>
13568 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13569 * dumpfile.h (struct dump_file_info): Remove ctors.
13571 2017-05-30 Martin Liska <mliska@suse.cz>
13573 * predict.def: Fix GNU coding style.
13575 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13577 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13578 Mark 'to' argument with ATTRIBUTE_UNUSED.
13580 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13582 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13583 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13584 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13587 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13589 * doc/install.texi (Options specification): Restore entry of
13590 --enable-sjlj-exceptions.
13592 2017-05-27 Michael Eager <eager@eagercon.com>
13595 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13597 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13599 * config/microblaze/microblaze.h
13600 (FIXED_REGISTERS): Update in macro.
13601 (CALL_USED_REGISTERS): Update in macro.
13603 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13605 * doc/install.texi: Add links to macOS binary distributions.
13607 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13611 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13613 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13615 2017-05-26 Martin Liska <mliska@suse.cz>
13617 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13619 2017-05-26 Martin Liska <mliska@suse.cz>
13621 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13622 always leading ';; '.
13623 (dump_bb_info): Likewise.
13624 (brief_dump_cfg): Likewise.
13625 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13626 * dumpfile.c: Remove usage of TDF_VERBOSE.
13627 * dumpfile.h (enum dump_kind): Likewise.
13628 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13629 * print-tree.c (debug_verbose): Remove.
13630 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13631 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13632 * tree-diagnostic.c (default_tree_printer): Replace
13633 TDF_DIAGNOSTIC with TDF_SLIM.
13635 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13637 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13638 in parameter loop, rather than loop_vinfo.
13639 (create_intersect_range_checks): Ditto.
13640 (vect_create_cond_for_alias_checks): Update call to above functions.
13642 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13644 PR tree-optimization/80815
13645 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13646 for merging runtime alias checks. Handle negative DR_STEPs.
13648 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13650 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13652 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13653 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13654 out code pruning runtime alias checks.
13655 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13656 factored out from above.
13657 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13659 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13661 (prune_runtime_alias_test_list): New decalaration.
13663 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13665 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13666 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13667 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13668 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13669 (operator==, comp_dr_with_seg_len_pair): Ditto.
13670 (vect_prune_runtime_alias_test_list): Ditto.
13672 2017-05-26 Martin Liska <mliska@suse.cz>
13675 * params.def: Bound partial-inlining-entry-probability param.
13677 2017-05-26 Marek Polacek <polacek@redhat.com>
13680 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13683 2017-05-26 Richard Biener <rguenther@suse.de>
13685 PR tree-optimization/80842
13686 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13689 2017-05-26 Richard Biener <rguenther@suse.de>
13691 PR tree-optimization/80844
13692 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13694 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13696 * doc/md.texi (Machine Constraints): Update x86 family
13697 machine constraints section to match 'config/i386/constraints.md'.
13699 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13701 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13703 2017-05-25 Nathan Sidwell <nathan@acm.org>
13705 * doc/invoke.texi (--enable-languages): Update documentation.
13707 2017-05-25 Martin Liska <mliska@suse.cz>
13709 * dumpfile.c: Add TDF_FOLDING.
13710 * dumpfile.h (enum dump_kind): Likewise.
13711 * genmatch.c (dt_simplify::gen_1): Use it.
13713 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13715 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13717 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13719 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13720 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13722 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13724 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13725 * match.pd (X == C): Rewrite it here.
13726 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13727 with_certain_nonzero_bits2): New predicates.
13728 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13730 2017-05-24 Nathan Sidwell <nathan@acm.org>
13732 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13735 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13738 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13740 * config/powerpcspe: New port. Files are copied from the rs6000
13741 port, with "rs6000" in filenames replaced by "powerpcspe".
13743 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13745 PR rtl-optimization/80754
13746 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13748 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13750 * config/sparc/sparc.md (length): Return the correct value for -mflat
13751 sibcalls to match output_sibcall.
13753 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13757 * config/rs6000/rs6000.c (struct machine_function): Add new field
13759 (rs6000_get_separate_components): Init that field, use it.
13760 (rs6000_components_for_bb): Use the field.
13762 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13764 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13766 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13768 PR middle-end/80823
13769 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13771 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13774 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13775 address against address_operand predicate.
13776 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13778 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13780 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13781 parameters passed indirectly.
13783 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13785 * config/i386/i386.md (*movdi_internal): Remove SSE4
13786 alternative 18 (?r, *v). Update insn attributes.
13787 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13788 Update insn attributes.
13789 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13790 Update insn attributes.
13791 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13792 alternative 1 (r, v). Remove isa attribute.
13793 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13794 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13795 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13797 2017-05-23 Tom de Vries <tom@codesourcery.com>
13799 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13802 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13804 * cgraphunit.c (symbol_table::process_new_functions): Update.
13805 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13806 (inline_generate_summary): Rename to ...
13807 (ipa_fn_summary_generate): ... this one.
13808 (inline_read_summary): Rename to ...
13809 (ipa_fn_summary_read): ... this one.
13810 (inline_write_summary): Rename to ...
13811 (ipa_fn_summary_write): ... this one.
13812 (inline_free_summary): Rename to ...
13813 (ipa_free_fn_summary): ... this one.
13814 (pass_data_local_fn_summary, pass_local_fn_summary,
13815 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13816 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13817 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13818 make_pass_ipa_fn_summary): New.
13819 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13820 inline_write_summary, inline_free_summary): Remove.
13821 (ipa_free_fn_summary) : New.
13822 * ipa-inline.c (ipa_inline): Update.
13823 (pass_ipa_inline): Do not generate summaries.
13824 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13826 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13827 and add pass_ipa_fn_summary.
13828 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13830 (make_pass_inline_parameters): Remove.
13832 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13834 * omp-low.c (struct omp_context): Remove "default_kind" member.
13837 * omp-offload.c (execute_oacc_device_lower): Remove the
13838 parallelism dimensions function attributes for unparallelized
13839 OpenACC kernels constructs.
13841 2017-05-23 Martin Liska <mliska@suse.cz>
13843 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13845 (cgraph_edge::make_speculative): Likewise.
13846 (cgraph_edge::resolve_speculation): Likewise.
13847 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13848 (cgraph_node::dump): Likewise.
13849 * cgraph.h: Likewise.
13850 * cgraphunit.c (analyze_functions): Likewise.
13851 (symbol_table::compile): Likewise.
13852 * ipa-cp.c (print_all_lattices): Likewise.
13853 (determine_versionability): Likewise.
13854 (initialize_node_lattices): Likewise.
13855 (ipcp_verify_propagated_values): Likewise.
13856 (estimate_local_effects): Likewise.
13857 (update_profiling_info): Likewise.
13858 (create_specialized_node): Likewise.
13859 (perhaps_add_new_callers): Likewise.
13860 (decide_about_value): Likewise.
13861 (decide_whether_version_node): Likewise.
13862 (identify_dead_nodes): Likewise.
13863 (ipcp_store_bits_results): Likewise.
13864 * ipa-devirt.c (dump_targets): Likewise.
13865 (ipa_devirt): Likewise.
13866 * ipa-icf.c (sem_item::dump): Likewise.
13867 (sem_function::equals): Likewise.
13868 (sem_variable::equals): Likewise.
13869 (sem_item_optimizer::read_section): Likewise.
13870 (sem_item_optimizer::execute): Likewise.
13871 (congruence_class::dump): Likewise.
13872 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13873 (dump_inline_summary): Likewise.
13874 (estimate_node_size_and_time): Likewise.
13875 (inline_analyze_function): Likewise.
13876 * ipa-inline-transform.c (inline_call): Likewise.
13877 * ipa-inline.c (report_inline_failed_reason): Likewise.
13878 (want_early_inline_function_p): Likewise.
13879 (edge_badness): Likewise.
13880 (update_edge_key): Likewise.
13881 (inline_small_functions): Likewise.
13882 * ipa-profile.c (ipa_profile): Likewise.
13883 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13884 (ipa_make_edge_direct_to_target): Likewise.
13885 (remove_described_reference): Likewise.
13886 (ipa_impossible_devirt_target): Likewise.
13887 (propagate_controlled_uses): Likewise.
13888 (ipa_print_node_params): Likewise.
13889 (ipcp_transform_function): Likewise.
13890 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13891 (propagate_pure_const): Likewise.
13892 * ipa-reference.c (generate_summary): Likewise.
13893 (read_write_all_from_decl): Likewise.
13894 (propagate): Likewise.
13895 (ipa_reference_read_optimization_summary): Likewise.
13896 * ipa-utils.c (ipa_merge_profiles): Likewise.
13897 * ipa.c (walk_polymorphic_call_targets): Likewise.
13898 (symbol_table::remove_unreachable_nodes): Likewise.
13899 (ipa_single_use): Likewise.
13900 * passes.c (execute_todo): Likewise.
13901 * predict.c (drop_profile): Likewise.
13902 * symtab.c (symtab_node::get_dump_name): New function.
13903 (symtab_node::dump_name): Likewise.
13904 (symtab_node::dump_asm_name): Likewise.
13905 (symtab_node::dump_references): Likewise.
13906 (symtab_node::dump_referring): Likewise.
13907 (symtab_node::dump_base): Likewise.
13908 (symtab_node::debug_symtab): Likewise.
13909 * tree-sra.c (convert_callers_for_node): Likewise.
13910 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13911 * value-prof.c (init_node_map): Likewise.
13913 2017-05-23 Martin Liska <mliska@suse.cz>
13915 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13916 and symtab_node::debug_symtab to symbol_table::debug.
13917 * cgraphunit.c (analyze_functions): Use the renamed function.
13918 (symbol_table::compile): Likewise.
13919 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13920 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13921 * passes.c (execute_todo): Likewise.
13922 * symtab.c (symbol_table::dump): New function.
13923 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13925 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13927 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13928 that nonconst implies exec.
13930 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13932 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13933 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13934 (inline_edge_summary_vec): Turn into ...
13935 (ipa_call_summaries): ... this one.
13936 (redirect_to_unreachable, edge_set_predicate,
13937 evaluate_properties_for_edge, inline_summary_alloc,
13938 reset_ipa_call_summary, reset_inline_summary,
13939 inline_summary_t::duplicate): Update.
13940 (inline_edge_duplication_hook): Turn to ...
13941 (ipa_call_summary_t::duplicate): ... this one.
13942 (inline_edge_removal_hook): Turn to ...
13943 (ipa_call_summary_t::remove): ... this one.
13944 (dump_inline_edge_summary): Turn to ...
13945 (dump_ipa_call_summary): ... this one.
13946 (estimate_function_body_sizes): Update.
13947 (inline_update_callee_summaries): Update.
13948 (remap_edge_change_prob): Update.
13949 (remap_edge_summaries): Update.
13950 (inline_merge_summary): Update.
13951 (do_estimate_edge_time): Update.
13952 (inline_generate_summary): Update.
13953 (inline_read_section): Update.
13954 (inline_read_summary): Update.
13955 (inline_free_summary): Update.
13956 * ipa-inline.c (can_inline_edge_p): Update.
13957 (compute_inlined_call_time): Update.
13958 (want_inline_small_function_p): Update.
13959 (edge_badness): Update.
13960 (early_inliner): Update.
13961 * ipa-inline.h (inline_edge_summary): Turn to ...
13962 (ipa_call_summary): ... this one.
13963 (ipa_call_summary_t): New class.
13964 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13965 (ipa_call_summaries): New.
13966 (inline_edge_summary): Remove.
13967 (estimate_edge_growth): Update.
13968 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13969 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13970 * ipa-split.c (execute_split_functions): Update.
13971 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13973 2017-05-23 Tom de Vries <tom@codesourcery.com>
13975 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13976 attributes): Document rdrand effective target.
13978 2017-05-23 Tom de Vries <tom@codesourcery.com>
13980 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13981 attributes): Sort alphabetically.
13983 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13985 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13987 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13990 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13991 V2DF/V2DI splat into two separate patterns, one that handles
13992 registers, and the other that only handles memory. Drop support
13993 for splatting from a GPR on ISA 2.07 and then splitting the
13994 splat into direct move and splat.
13995 (vsx_splat_<mode>_reg): Likewise.
13996 (vsx_splat_<mode>_mem): Likewise.
13998 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14000 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14002 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14004 PR middle-end/80809
14005 * omp-low.c (finish_taskreg_remap): New function.
14006 (finish_taskreg_scan): If unit size of ctx->record_type
14007 is non-constant, unshare the size expression and replace
14008 decls in it with possible outer var refs.
14010 PR middle-end/80809
14011 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14012 GOVD_SHARED rather than GOVD_PRIVATE with it.
14013 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14014 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14016 PR middle-end/80853
14017 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14018 as last argument to build_outer_var_ref for pointer bases of array
14019 section reductions.
14021 2017-05-19 Martin Sebor <msebor@redhat.com>
14023 * print-tree.c (print_node): Print DECL_READ_P flag.
14025 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14027 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14028 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14029 * cgraph.c: Likewise.
14030 * cgraphunit.c: Likewise.
14031 * gengtype.c: Likewise.
14032 * ipa-cp.c: Likewise.
14033 * ipa-devirt.c: Likewise.
14034 * ipa-icf.c: Likewise.
14035 * ipa-predicate.c: Likewise.
14036 * ipa-profile.c: Likewise.
14037 * ipa-prop.c: Likewise.
14038 * ipa-split.c: Likewise.
14040 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14041 edge_predicate_pool, dump_inline_hints,
14042 inline_summary::account_size_time, redirect_to_unreachable,
14043 edge_set_predicate, set_hint_predicate,
14044 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14045 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14046 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14047 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14048 ipa_call_summary_t::remove, initialize_growth_caches,
14049 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14050 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14051 mark_modified, unmodified_parm_1, unmodified_parm,
14052 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14053 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14054 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14055 will_be_nonconstant_predicate, record_modified_bb_info,
14056 get_minimal_bb, record_modified, param_change_prob,
14057 phi_result_unknown_predicate, predicate_for_phi_result,
14058 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14059 estimate_function_body_sizes, compute_inline_parameters,
14060 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14061 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14062 inline_update_callee_summaries, remap_edge_change_prob,
14063 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14064 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14065 inline_analyze_function, inline_summary_t::insert,
14066 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14067 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14068 inline_free_summary): Move to ipa-fnsummary.h
14069 (predicate_t): Remove.
14070 * ipa-fnsummary.c: New file.
14071 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14072 (enum inline_hints_vals, inline_hints, agg_position_info,
14073 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14074 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14075 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14076 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14077 inline_read_summary, inline_write_summary, inline_free_summary,
14078 inline_analyze_function, initialize_inline_failed,
14079 inline_merge_summary, inline_update_overall_summary,
14080 compute_inline_parameters): Move to ipa-fnsummary.h
14081 * ipa-fnsummary.h: New file.
14082 * ipa-inline-transform.h: Include ipa-inline.h.
14083 * ipa-inline.c: LIkewise.
14085 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14087 * ipa-inline.c (edge_badness): Use inlined_time instead of
14088 inline_summaries->get.
14090 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14092 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14094 2017-05-22 Nathan Sidwell <nathan@acm.org>
14096 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14097 (fdump-lang): Document 'raw' option.
14098 * dumpfile.h (TDI_tu): Delete.
14099 * dumpfile.c (dump_files): Remove translation-unit.
14100 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14102 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14104 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14105 command option from $(AWK) call.
14106 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14107 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14108 [FORMAT]: Remove handling of variable.
14109 * config/avr/t-multilib: Regenerate.
14111 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14113 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14115 (dump_inline_summary): Do not print self_time.
14116 (estimate_function_body_sizes): Do not set self_time.
14117 (compute_inline_parameters): Likewise.
14118 (inline_read_section, inline_write_summary): Do not stream self_time.
14119 * ipa-inline.h (inline_summary): Drop self_time.
14121 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14123 * ipa-inline-analysis.c (account_size_time): Rename to ...
14124 (inline_summary::account_size_time): ... this one.
14125 (reset_ipa_call_summary): Turn to ...
14126 (ipa_call_summary::reset): ... this one.
14127 (reset_inline_summary): Turn to ...
14128 (inline_summary::reset): ... this one.
14129 (inline_summary_t::remove): Update.
14130 (inline_summary_t::duplicate): Update.
14131 (ipa_call_summary_t::remove): Update.
14132 (dump_inline_summary): Update.
14133 (estimate_function_body_sizes): Update.
14134 (compute_inline_parameters): Update.
14135 (estimate_node_size_and_time): Update.
14136 (inline_merge_summary): Update.
14137 (inline_update_overall_summary): Update.
14138 (inline_read_section): Update.
14139 (inline_write_summary): Update.
14140 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14141 add account_size_time and reset member functions.
14142 (ipa_call_summary): Add reset function.
14143 * ipa-predicate.h (predicate::operator &): Constify.
14145 2017-05-22 Richard Biener <rguenther@suse.de>
14147 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14149 2017-05-19 Jason Merrill <jason@redhat.com>
14151 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14153 2017-05-19 Marek Polacek <polacek@redhat.com>
14156 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14157 TYPE_OVERFLOW_WRAPS checks.
14159 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14161 * tree-core.h (enum omp_clause_default_kind): Add
14162 "OMP_CLAUSE_DEFAULT_PRESENT".
14163 * tree-pretty-print.c (dump_omp_clause): Handle it.
14164 * gimplify.c (enum gimplify_omp_var_data): Add
14165 "GOVD_MAP_FORCE_PRESENT".
14166 (gimplify_adjust_omp_clauses_1): Map it to
14167 "GOMP_MAP_FORCE_PRESENT".
14168 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14170 * gimplify.c (oacc_default_clause): Clarify.
14172 2017-05-19 Nathan Sidwell <nathan@acm.org>
14174 LANG_HOOK_REGISTER_DUMPS
14175 * toplev.c (general_init): Call register dump lang hook.
14176 * doc/invoke.texi: Document -fdump-lang option family.
14177 * dumpfile.c (dump_files): Remove class dump here.
14178 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14179 * dumpfile.h (tree_dump_index): Remove TDI_class.
14180 * langhooks-def.h (lhd_register_dumps): Declare.
14181 (LANG_HOOKS_REGISTER_DUMPS): Define.
14182 (LANG_HOOKS_INITIALIZER): Add it.
14183 * langhooks.c (lhd_register_dumps): Define.
14184 * langhooks.h (struct lang_hooks): Add register_dumps.
14186 2017-05-19 Nathan Sidwell <nathan@acm.org>
14188 * context.h (context::set_passes): New.
14189 * context.c (context::context): Do not create pass manager.
14190 * toplev.c (general_init): Create pass manager here.
14192 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14194 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14195 use this splitter if two add or or instructions would also work for
14196 the constant we want to generate.
14198 2017-05-19 Richard Biener <rguenther@suse.de>
14201 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14202 predicate evaluation.
14204 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14206 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14208 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14211 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14213 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14214 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14215 (inline_edge_summary_vec): Turn into ...
14216 (ipa_call_summaries): ... this one.
14217 (redirect_to_unreachable, edge_set_predicate,
14218 evaluate_properties_for_edge, inline_summary_alloc,
14219 reset_ipa_call_summary, reset_inline_summary,
14220 inline_summary_t::duplicate): Update.
14221 (inline_edge_duplication_hook): Turn to ...
14222 (ipa_call_summary_t::duplicate): ... this one.
14223 (inline_edge_removal_hook): Turn to ...
14224 (ipa_call_summary_t::remove): ... this one.
14225 (dump_inline_edge_summary): Turn to ...
14226 (dump_ipa_call_summary): ... this one.
14227 (estimate_function_body_sizes): Update.
14228 (inline_update_callee_summaries): Update.
14229 (remap_edge_change_prob): Update.
14230 (remap_edge_summaries): Update.
14231 (inline_merge_summary): Update.
14232 (do_estimate_edge_time): Update.
14233 (inline_generate_summary): Update.
14234 (inline_read_section): Update.
14235 (inline_read_summary): Update.
14236 (inline_free_summary): Update.
14237 * ipa-inline.c (can_inline_edge_p): Update.
14238 (compute_inlined_call_time): Update.
14239 (want_inline_small_function_p): Update.
14240 (edge_badness): Update.
14241 (early_inliner): Update.
14242 * ipa-inline.h (inline_edge_summary): Turn to ...
14243 (ipa_call_summary): ... this one.
14244 (ipa_call_summary_t): New class.
14245 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14246 (ipa_call_summaries): New.
14247 (inline_edge_summary): Remove.
14248 (estimate_edge_growth): Update.
14249 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14250 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14251 * ipa-split.c (execute_split_functions): Update.
14252 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14254 2017-05-19 Richard Biener <rguenther@suse.de>
14256 PR middle-end/80764
14257 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14259 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14261 * config/rs6000/rs6000.c (struct machine_function): Add field
14262 fpr_is_wrapped_separately.
14263 (rs6000_get_separate_components): Use 64 components. Handle the
14264 new FPR components.
14265 (rs6000_components_for_bb): Handle the FPR components.
14266 (rs6000_emit_prologue_components): Handle the FPR components.
14267 (rs6000_emit_epilogue_components): Handle the FPR components.
14268 (rs6000_set_handled_components): Handle the FPR components.
14269 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14270 that are already separately shrink-wrapped.
14271 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14272 that are already separately shrink-wrapped.
14274 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14277 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14280 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14281 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14282 cases where the register allocator uses a move and an offsettable
14283 memory operation to/from a FPR register on ISA 2.06/2.07.
14284 (define_peephole2 for Altivec d-form store): Likewise.
14286 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14289 * config/i386/mmx.md (*mov<mode>_internal): Enable
14290 alternatives 11, 12, 13 and 14 also for 32bit targets.
14291 Remove alternatives 15, 16, 17 and 18.
14292 * config/i386/sse.md (vec_concatv2di): Change
14293 alternative (!x, *y) to (x, ?!*Yn).
14295 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14297 * dumpfile.h (enum dump_kind): Remove stray comma.
14299 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14301 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14302 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14303 predicate::num_conditions
14304 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14305 (CHANGED): turn into predicate::changed.
14306 (agg_position_info): Move to ipa-predicate.h
14307 (add_condition, predicate::add_clause, predicate::operator &=,
14308 predicate::or_with, predicate::evaluate, predicate::probability,
14309 dump_condition, dump_clause, predicate::dump,
14310 predicate::remap_after_duplication, predicate::remap_after_inlining,
14311 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14312 (evaluate_conditions_for_known_args): Update.
14313 (set_cond_stmt_execution_predicate): Update.
14314 * ipa-inline.h: Include ipa-predicate.h
14315 (condition, inline_param_summary, conditions, agg_position_info,
14316 predicate): Move to ipa-predicate.h
14317 * ipa-predicate.c: New file.
14318 * ipa-predicate.h: New file.
14320 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14322 * final.c (leaf_function_p): Check we are not in a sequence.
14324 2017-05-18 Martin Liska <mliska@suse.cz>
14326 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14327 * dumpfile.c (dump_register): Use new enum dump_kind.
14328 (get_dump_file_name): Likewise.
14329 (dump_enable_all): Likewise.
14330 (dump_switch_p_1): Likewise.
14331 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14332 * dumpfile.h (enum dump_kind): New enum type.
14333 (struct dump_file_info): Create constructor and
14334 format fields and comments.
14335 * passes.c (pass_manager::register_one_dump_file):
14337 * statistics.c (statistics_early_init): Likewise.
14338 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14339 TDF_TREE with TDF_SLIM.
14340 (gather_memory_references_ref): Likewise.
14342 2017-05-18 Martin Liska <mliska@suse.cz>
14344 * vec.h (struct vnull): Use it.
14346 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14348 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14349 (true_predicate, false_predicate, true_predicate_p,
14350 false_predicate_p): Remove.
14351 (single_cond_predicate, not_inlined_predicate): Turn to member function
14353 (add_condition): Update.
14354 (add_clause): Turn to...
14355 (predicate::add_clause): ... this one; update; allow passing NULL
14357 (and_predicates): Turn to ...
14358 (predicate::operator &=): ... this one.
14359 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14360 (or_predicates): Turn to ...
14361 (predicate::or_with): ... this one.
14362 (evaluate_predicate): Turn to ...
14363 (predicate::evaluate): ... this one.
14364 (predicate_probability): Turn to ...
14365 (predicate::probability): ... this one.
14366 (dump_condition): Update.
14367 (dump_predicate): Turn to ...
14368 (predicate::dump): ... this one.
14369 (account_size_time): Update.
14370 (edge_set_predicate): Update.
14371 (set_hint_predicate): UPdate.
14372 (evaluate_conditions_for_known_args): Update.
14373 (evaluate_properties_for_edge): Update.
14374 (remap_predicate_after_duplication): Turn to...
14375 (predicate::remap_after_duplication): ... this one.
14376 (remap_hint_predicate_after_duplication): Update.
14377 (inline_summary_t::duplicate): UPdate.
14378 (dump_inline_edge_summary): Update.
14379 (dump_inline_summary): Update.
14380 (set_cond_stmt_execution_predicate): Update.
14381 (set_switch_stmt_execution_predicate): Update.
14382 (compute_bb_predicates): Update.
14383 (will_be_nonconstant_expr_predicate): Update.
14384 (will_be_nonconstant_predicate): Update.
14385 (phi_result_unknown_predicate): Update.
14386 (predicate_for_phi_result): Update.
14387 (array_index_predicate): Update.
14388 (estimate_function_body_sizes): Update.
14389 (estimate_node_size_and_time): Update.
14390 (estimate_ipcp_clone_size_and_time): Update.
14391 (remap_predicate): Rename to ...
14392 (predicate::remap_after_inlining): ... this one.
14393 (remap_hint_predicate): Update.
14394 (inline_merge_summary): Update.
14395 (inline_update_overall_summary): Update.
14396 (estimate_size_after_inlining): Update.
14397 (read_predicate): Rename to ...
14398 (predicate::stream_in): ... this one.
14399 (read_inline_edge_summary): Update.
14400 (write_predicate): Rename to ...
14401 (predicate::stream_out): ... this one.
14402 (write_inline_edge_summary): Update.
14403 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14404 (clause_t): Turn to uint32_t
14405 (predicate): Turn to class; implement constructor and operators
14407 (size_time_entry): Update.
14408 (inline_summary): Update.
14409 (inline_edge_summary): Update.
14411 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14413 * fold-const.c (fold_binary_loc): Move transformation...
14414 * match.pd (C - X CMP X): ... here.
14416 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14418 * config/sparc/sparc.c (sparc_option_override): Set function
14419 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14420 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14422 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14424 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14426 2017-05-18 Marek Polacek <polacek@redhat.com>
14429 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14430 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14431 gimple_assign_load_p.
14433 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14435 PR middle-end/80692
14436 * real.c (do_compare): Give decimal_do_compare preference over
14437 comparing just the signs.
14439 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14441 * doc/md.texi (Canonicalization of Instructions): Describe the
14442 canonical form of instructions that inherently set a condition
14445 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14447 PR middle-end/80775
14448 * tree-cfg.c: Move deletion of unreachable case statements to after
14449 the merging of consecutive case labels.
14451 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14453 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14454 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14455 restoring of callee-saved registers.
14457 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14459 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14460 * config/visium/visium.c (single_set_and_flags): Likewise.
14461 * config/visium/visium.md (Substitutions): Likewise.
14463 2017-05-17 Martin Liska <mliska@suse.cz>
14465 * cfg.c: Introduce dump_flags_t type and
14466 use it instead of int type.
14468 * cfghooks.c: Likewise.
14469 * cfghooks.h (struct cfg_hooks): Likewise.
14470 * cfgrtl.c: Likewise.
14471 * cfgrtl.h: Likewise.
14472 * cgraph.c (cgraph_node::get_body): Likewise.
14473 * coretypes.h: Likewise.
14474 * domwalk.c: Likewise.
14475 * domwalk.h: Likewise.
14476 * dumpfile.c (struct dump_option_value_info): Likewise.
14477 (dump_enable_all): Likewise.
14478 (dump_switch_p_1): Likewise.
14479 (opt_info_switch_p): Likewise.
14480 * dumpfile.h (enum tree_dump_index): Likewise.
14481 (struct dump_file_info): Likewise.
14482 * genemit.c: Likewise.
14483 * generic-match-head.c: Likewise.
14484 * gengtype.c (open_base_files): Likewise.
14485 * gimple-pretty-print.c: Likewise.
14486 * gimple-pretty-print.h: Likewise.
14487 * graph.c (print_graph_cfg): Likewise.
14488 * graphite-scop-detection.c (dot_all_sese): Likewise.
14489 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14490 * loop-unroll.c (report_unroll): Likewise.
14491 * passes.c (pass_manager::register_one_dump_file): Likewise.
14492 * print-tree.c: Likewise.
14493 * statistics.c: Likewise.
14494 * tree-cfg.c: Likewise.
14495 * tree-cfg.h: Likewise.
14496 * tree-dfa.c: Likewise.
14497 * tree-dfa.h: Likewise.
14498 * tree-dump.c (dump_function): Likewise.
14499 * tree-dump.h (struct dump_info): Likewise.
14500 * tree-pretty-print.c: Likewise.
14501 * tree-pretty-print.h: Likewise.
14502 * tree-ssa-live.c: Likewise.
14503 * tree-ssa-live.h: Likewise.
14504 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14505 * tree-vect-loop.c: Likewise.
14506 * tree-vect-slp.c: Likewise.
14508 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14509 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14511 PR tree-optimization/80457
14512 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14513 of all arguments to a statement as scalar_to_vec operations.
14514 (vectorizable_call): Adjust call to vect_model_simple_cost for
14516 (vectorizable_conversion): Likewise.
14517 (vectorizable_assignment): Likewise.
14518 (vectorizable_shift): Likewise.
14519 (vectorizable_operation): Likewise.
14520 (vectorizable_comparison): Likewise.
14521 (vect_is_simple_cond): Record the def types for operands.
14522 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14523 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14524 for statement argument count.
14526 2017-05-16 Carl Love <cel@us.ibm.com>
14528 * config/rs6000/rs6000-c: Add support for built-in functions
14529 vector unsigned long long vec_bperm (vector unsigned long long,
14530 vector unsigned char)
14531 vector signed long long vec_mule (vector signed int,
14533 vector unsigned long long vec_mule (vector unsigned int,
14534 vector unsigned int)
14535 vector signed long long vec_mulo (vector signed int,
14537 vector unsigned long long vec_mulo (vector unsigned int,
14538 vector unsigned int)
14539 vector signed char vec_sldw (vector signed char,
14540 vector signed char,
14542 vector unsigned char vec_sldw (vector unsigned char,
14543 vector unsigned char,
14545 vector signed short vec_sldw (vector signed short,
14546 vector signed short,
14548 vector unsigned short vec_sldw (vector unsigned short,
14549 vector unsigned short,
14551 vector signed int vec_sldw (vector signed int,
14554 vector unsigned int vec_sldw (vector unsigned int,
14555 vector unsigned int,
14557 vector signed long long vec_sldw (vector signed long long,
14558 vector signed long long,
14560 vector unsigned long long vec_sldw (vector unsigned long long,
14561 vector unsigned long long,
14563 * config/rs6000/rs6000-c: Add support for built-in functions
14564 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14565 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14566 * doc/extend.texi: Update the built-in documentation for the
14567 new built-in functions.
14569 2017-05-16 Marek Polacek <polacek@redhat.com>
14573 * tree.c (save_expr): Don't fold the expression.
14575 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14577 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14578 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14579 and (?*y,m). Update insn attributes.
14581 2017-05-16 Martin Liska <mliska@suse.cz>
14583 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14584 flags argument of print_gimple_stmt, print_gimple_expr,
14585 print_generic_stmt and print_generic_expr.
14586 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14587 * coretypes.h: Likewise.
14588 * except.c (dump_eh_tree): Likewise.
14589 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14590 * gimple-pretty-print.h: Likewise.
14591 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14592 (backprop::push_to_worklist): Likewise.
14593 (backprop::pop_from_worklist): Likewise.
14594 (backprop::process_use): Likewise.
14595 (backprop::intersect_uses): Likewise.
14596 (note_replacement): Likewise.
14597 * gimple-ssa-store-merging.c
14598 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14599 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14600 (pass_store_merging::execute): Likewise.
14601 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14602 (ssa_base_cand_dump_callback): Likewise.
14603 (dump_incr_vec): Likewise.
14604 (replace_refs): Likewise.
14605 (replace_mult_candidate): Likewise.
14606 (create_add_on_incoming_edge): Likewise.
14607 (create_phi_basis): Likewise.
14608 (insert_initializers): Likewise.
14609 (all_phi_incrs_profitable): Likewise.
14610 (introduce_cast_before_cand): Likewise.
14611 (replace_one_candidate): Likewise.
14612 * gimplify.c (gimplify_expr): Likewise.
14613 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14614 (set_rename): Likewise.
14615 (rename_uses): Likewise.
14616 (copy_loop_phi_nodes): Likewise.
14617 (add_close_phis_to_merge_points): Likewise.
14618 (copy_loop_close_phi_args): Likewise.
14619 (copy_cond_phi_args): Likewise.
14620 (graphite_copy_stmts_from_block): Likewise.
14621 (translate_pending_phi_nodes): Likewise.
14622 * graphite-poly.c (print_pdr): Likewise.
14623 (dump_gbb_cases): Likewise.
14624 (dump_gbb_conditions): Likewise.
14625 (print_scop_params): Likewise.
14626 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14627 (build_cross_bb_scalars_use): Likewise.
14628 (gather_bbs::before_dom_children): Likewise.
14629 * hsa-dump.c (dump_hsa_immed): Likewise.
14630 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14631 (get_replacement_map): Likewise.
14632 * ipa-inline-analysis.c (dump_condition): Likewise.
14633 (estimate_function_body_sizes): Likewise.
14634 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14635 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14636 * ipa-prop.c (ipa_dump_param): Likewise.
14637 (ipa_print_node_jump_functions_for_edge): Likewise.
14638 (ipa_modify_call_arguments): Likewise.
14639 (ipa_modify_expr): Likewise.
14640 (ipa_dump_param_adjustments): Likewise.
14641 (ipa_dump_agg_replacement_values): Likewise.
14642 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14643 * ipa-pure-const.c (check_stmt): Likewise.
14644 (pass_nothrow::execute): Likewise.
14645 * ipa-split.c (execute_split_functions): Likewise.
14646 * omp-offload.c (dump_oacc_loop_part): Likewise.
14647 (dump_oacc_loop): Likewise.
14648 * trans-mem.c (tm_log_emit): Likewise.
14649 (tm_memopt_accumulate_memops): Likewise.
14650 (dump_tm_memopt_set): Likewise.
14651 (dump_tm_memopt_transform): Likewise.
14652 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14653 (print_loop): Likewise.
14654 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14655 (chkp_gather_checks_info): Likewise.
14656 (chkp_get_check_result): Likewise.
14657 (chkp_remove_check_if_pass): Likewise.
14658 (chkp_use_outer_bounds_if_possible): Likewise.
14659 (chkp_reduce_bounds_lifetime): Likewise.
14660 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14661 (chkp_mark_completed_bounds): Likewise.
14662 (chkp_register_incomplete_bounds): Likewise.
14663 (chkp_mark_invalid_bounds): Likewise.
14664 (chkp_maybe_copy_and_register_bounds): Likewise.
14665 (chkp_build_returned_bound): Likewise.
14666 (chkp_get_bound_for_parm): Likewise.
14667 (chkp_build_bndldx): Likewise.
14668 (chkp_get_bounds_by_definition): Likewise.
14669 (chkp_generate_extern_var_bounds): Likewise.
14670 (chkp_get_bounds_for_decl_addr): Likewise.
14671 * tree-chrec.c (chrec_apply): Likewise.
14672 * tree-data-ref.c (dump_data_reference): Likewise.
14673 (dump_subscript): Likewise.
14674 (dump_data_dependence_relation): Likewise.
14675 (analyze_overlapping_iterations): Likewise.
14676 * tree-inline.c (expand_call_inline): Likewise.
14677 (tree_function_versioning): Likewise.
14678 * tree-into-ssa.c (dump_defs_stack): Likewise.
14679 (dump_currdefs): Likewise.
14680 (dump_names_replaced_by): Likewise.
14681 (dump_update_ssa): Likewise.
14682 (update_ssa): Likewise.
14683 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14684 * tree-parloops.c (build_new_reduction): Likewise.
14685 (try_create_reduction_list): Likewise.
14686 (ref_conflicts_with_region): Likewise.
14687 (oacc_entry_exit_ok_1): Likewise.
14688 (oacc_entry_exit_single_gang): Likewise.
14689 * tree-pretty-print.h: Likewise.
14690 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14691 (get_scalar_evolution): Likewise.
14692 (add_to_evolution): Likewise.
14693 (get_loop_exit_condition): Likewise.
14694 (analyze_evolution_in_loop): Likewise.
14695 (analyze_initial_condition): Likewise.
14696 (analyze_scalar_evolution): Likewise.
14697 (instantiate_scev): Likewise.
14698 (number_of_latch_executions): Likewise.
14699 (gather_chrec_stats): Likewise.
14700 (final_value_replacement_loop): Likewise.
14701 (scev_const_prop): Likewise.
14702 * tree-sra.c (dump_access): Likewise.
14703 (disqualify_candidate): Likewise.
14704 (create_access): Likewise.
14705 (reject): Likewise.
14706 (maybe_add_sra_candidate): Likewise.
14707 (create_access_replacement): Likewise.
14708 (analyze_access_subtree): Likewise.
14709 (analyze_all_variable_accesses): Likewise.
14710 (sra_modify_assign): Likewise.
14711 (initialize_constant_pool_replacements): Likewise.
14712 (find_param_candidates): Likewise.
14713 (decide_one_param_reduction): Likewise.
14714 (replace_removed_params_ssa_names): Likewise.
14715 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14716 * tree-ssa-copy.c (dump_copy_of): Likewise.
14717 (copy_prop_visit_cond_stmt): Likewise.
14718 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14719 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14720 (record_equivalences_from_stmt): Likewise.
14721 * tree-ssa-dse.c (compute_trims): Likewise.
14722 (delete_dead_call): Likewise.
14723 (delete_dead_assignment): Likewise.
14724 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14725 (forward_propagate_into_cond): Likewise.
14726 (pass_forwprop::execute): Likewise.
14727 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14728 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14730 (move_computations_worker): Likewise.
14731 (execute_sm): Likewise.
14732 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14733 (remove_exits_and_undefined_stmts): Likewise.
14734 (remove_redundant_iv_tests): Likewise.
14735 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14736 (adjust_iv_update_pos): Likewise.
14737 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14738 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14739 (value_replacement): Likewise.
14740 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14741 * tree-ssa-pre.c (print_pre_expr): Likewise.
14742 (get_representative_for): Likewise.
14743 (create_expression_by_pieces): Likewise.
14744 (insert_into_preds_of_block): Likewise.
14745 (eliminate_insert): Likewise.
14746 (eliminate_dom_walker::before_dom_children): Likewise.
14747 (eliminate): Likewise.
14748 (remove_dead_inserted_code): Likewise.
14749 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14750 * tree-ssa-reassoc.c (get_rank): Likewise.
14751 (eliminate_duplicate_pair): Likewise.
14752 (eliminate_plus_minus_pair): Likewise.
14753 (eliminate_not_pairs): Likewise.
14754 (undistribute_ops_list): Likewise.
14755 (eliminate_redundant_comparison): Likewise.
14756 (update_range_test): Likewise.
14757 (optimize_range_tests_var_bound): Likewise.
14758 (optimize_vec_cond_expr): Likewise.
14759 (rewrite_expr_tree): Likewise.
14760 (rewrite_expr_tree_parallel): Likewise.
14761 (linearize_expr): Likewise.
14762 (break_up_subtract): Likewise.
14763 (linearize_expr_tree): Likewise.
14764 (attempt_builtin_powi): Likewise.
14765 (attempt_builtin_copysign): Likewise.
14766 (transform_stmt_to_copy): Likewise.
14767 (transform_stmt_to_multiply): Likewise.
14768 (dump_ops_vector): Likewise.
14769 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14770 (print_scc): Likewise.
14771 (set_ssa_val_to): Likewise.
14772 (visit_reference_op_store): Likewise.
14773 (visit_use): Likewise.
14774 (sccvn_dom_walker::before_dom_children): Likewise.
14775 (run_scc_vn): Likewise.
14776 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14778 (expr_hash_elt::print): Likewise.
14779 (const_and_copies::pop_to_marker): Likewise.
14780 (const_and_copies::record_const_or_copy_raw): Likewise.
14781 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14782 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14783 (dump_predicates): Likewise.
14784 (find_uninit_use): Likewise.
14785 (warn_uninitialized_phi): Likewise.
14786 (pass_late_warn_uninitialized::execute): Likewise.
14787 * tree-ssa.c (verify_vssa): Likewise.
14788 (verify_ssa): Likewise.
14789 (maybe_optimize_var): Likewise.
14790 * tree-vrp.c (dump_value_range): Likewise.
14791 (dump_all_value_ranges): Likewise.
14792 (dump_asserts_for): Likewise.
14793 (register_edge_assert_for_2): Likewise.
14794 (vrp_visit_cond_stmt): Likewise.
14795 (vrp_visit_switch_stmt): Likewise.
14796 (vrp_visit_stmt): Likewise.
14797 (vrp_visit_phi_node): Likewise.
14798 (simplify_cond_using_ranges_1): Likewise.
14799 (fold_predicate_in): Likewise.
14800 (evrp_dom_walker::before_dom_children): Likewise.
14801 (evrp_dom_walker::push_value_range): Likewise.
14802 (evrp_dom_walker::pop_value_range): Likewise.
14803 (execute_early_vrp): Likewise.
14805 2017-05-16 Richard Biener <rguenther@suse.de>
14807 * dwarf2out.c (loc_list_from_tree_1): Do not create
14808 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14810 2017-05-16 Richard Biener <rguenther@suse.de>
14812 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14814 (note_variable_value_in_expr): If we resolved the decl ref
14815 do not push to the stack.
14817 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14819 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14820 operations in fast-math mode.
14821 (vaddq_f16): Likewise.
14822 (vmul_f16): Likewise.
14823 (vmulq_f16): Likewise.
14824 (vsub_f16): Likewise.
14825 (vsubq_f16): Likewise.
14826 * config/arm/neon.md (add<mode>3): New.
14828 (fma:<VH:mode>3): New. Also remove outdated comment.
14831 2017-05-16 Martin Liska <mliska@suse.cz>
14835 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14836 (odr_types_equivalent_p): Likewise.
14838 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14840 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14842 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14845 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14846 non-interunit SSE move alternatives with '?'.
14847 (zero-extendsidi peephole2): New peephole to skip intermediate
14848 general register in SSE zero-extend sequence.
14850 2017-05-15 Jeff Law <law@redhat.com>
14852 * reorg.c (relax_delay_slots): Create a new variable to hold
14853 the temporary target rather than clobbering TARGET_LABEL.
14855 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14856 missing argument to extract_bit_field call.
14857 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14859 2017-05-15 Martin Liska <mliska@suse.cz>
14862 * gcc.c (process_command): Do not allow empty argument of -o option.
14864 2017-05-15 Renlin Li <renlin.li@arm.com>
14866 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14867 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14868 * config/aarch64/constraints.md (Usf): Add long call check.
14869 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14870 (call_value): Likewise.
14871 (sibcall): Likewise.
14872 (sibcall_value): Likewise.
14874 (call_value_insn): New.
14875 (sibcall_insn): Update rtx pattern.
14876 (sibcall_value_insn): Likewise.
14877 (call_internal): Remove.
14878 (call_value_internal): Likewise.
14879 (sibcall_internal): Likewise.
14880 (sibcall_value_internal): Likewise.
14881 (call_reg): Likewise.
14882 (call_symbol): Likewise.
14883 (call_value_reg): Likewise.
14884 (call_value_symbol): Likewise.
14886 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14889 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14891 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14893 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14894 compatible with CCGOCmode and with CCZmode.
14896 2017-05-14 Martin Sebor <msebor@redhat.com>
14898 PR middle-end/77671
14899 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14900 (gimple_fold_builtin_snprintf): Same.
14901 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14902 (gimple_fold_builtin_snprintf): Same.
14903 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14904 of character types.
14905 (is_call_safe): New function.
14906 (try_substitute_return_value): Call it.
14907 (try_simplify_call): New function.
14908 (pass_sprintf_length::handle_gimple_call): Call it.
14910 2017-05-14 Martin Sebor <msebor@redhat.com>
14912 PR middle-end/80669
14913 * builtins.c (expand_builtin_stpncpy): Simplify.
14915 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14917 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14918 * config/i386/i386.h
14919 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14920 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14921 (struct machine_function): Add new members call_ms2sysv,
14922 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14923 (struct machine_frame_state): New fields sp_realigned and
14924 sp_realigned_offset.
14925 * config/i386/i386.c
14926 (enum xlogue_stub): New enum.
14927 (enum xlogue_stub_sets): New enum.
14928 (class xlogue_layout): New class.
14929 (struct ix86_frame): New fields stack_realign_allocate_offset,
14930 stack_realign_offset and outlined_save_offset. Modify comments to
14931 detail stack layout when using out-of-line stubs.
14932 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14933 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14934 -mcall-ms2sysv-xlogues.
14935 (stub_managed_regs): New static variable.
14936 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14937 registers managed by out-of-line stub.
14938 (disable_call_ms2sysv_xlogues): New function.
14939 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14940 m->call_ms2sysv when appropriate and compute frame layout for
14942 (sp_valid_at, fp_valid_at): New inline functions.
14943 (choose_basereg): New function.
14944 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14946 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14947 Use align parameter of choose_baseaddr to generated aligned SSE movs
14949 (pro_epilogue_adjust_stack): Modify to track
14950 machine_frame_state::sp_realigned.
14951 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14952 (ix86_nsaved_sseregs): Likewise.
14953 (ix86_emit_save_regs): Likewise.
14954 (ix86_emit_save_regs_using_mov): Likewise.
14955 (ix86_emit_save_sse_regs_using_mov): Likewise.
14956 (get_scratch_register_on_entry): Likewise.
14957 (gen_frame_set): New function.
14958 (gen_frame_load): Likewise.
14959 (gen_frame_store): Likewise.
14960 (emit_outlined_ms2sysv_save): Likewise.
14961 (emit_outlined_ms2sysv_restore): Likewise.
14962 (ix86_expand_prologue): Modify stack re-alignment code and call
14963 emit_outlined_ms2sysv_save when appropriate.
14964 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14965 parameter rtx_insn *insn, which allows the function to be used to only
14966 generate the notes.
14967 (ix86_expand_epilogue): Modify validity checks of frame and stack
14968 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14969 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14970 * config/i386/predicates.md
14971 (save_multiple): New predicate.
14972 (restore_multiple): Likewise.
14973 * config/i386/sse.md
14974 (save_multiple<mode>): New pattern.
14975 (save_multiple_realign<mode>): Likewise.
14976 (restore_multiple<mode>): Likewise.
14977 (restore_multiple_and_return<mode>): Likewise.
14978 (restore_multiple_leave_return<mode>): Likewise.
14979 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14981 2017-05-14 Julia Koval <julia.koval@intel.com>
14983 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14984 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14985 (__builtin_ia32_xsetbv): New builtins.
14986 * config/i386/i386.c (ix86_expand_special_args_builtin):
14988 (ix86_expand_builtin): Special expand for new intrinsics.
14989 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14990 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14991 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14993 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14995 * cfganal.c (inverted_post_order_compute): Change argument type
14997 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14998 * df-core.c (rest_of_handle_df_initialize): Adjust.
14999 (rest_of_handle_df_finish): Likewise.
15000 (df_analyze_1): Likewise.
15001 (df_analyze): Likewise.
15002 (loop_inverted_post_order_compute): Change argument to be a vec *.
15003 (df_analyze_loop): Adjust.
15004 (df_get_n_blocks): Likewise.
15005 (df_get_postorder): Likewise.
15006 * df.h (struct df_d): Change field to be a vec.
15007 * lcm.c (compute_laterin): Adjust.
15008 (compute_available): Likewise.
15009 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15010 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15011 * tree-ssa-pre.c (compute_antic): Likewise.
15013 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15015 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15016 (depth_first_search::depth_first_search): Change structure init
15017 function to this constructor.
15018 (depth_first_search::add_bb): Rename function to this member.
15019 (depth_first_search::execute): Likewise.
15020 (flow_dfs_compute_reverse_finish): Adjust.
15022 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15024 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15025 (longest_simple_path): Likewise.
15026 * shrink-wrap.c (spread_components): Likewise.
15027 (disqualify_problematic_components): Likewise.
15028 (emit_common_heads_for_components): Likewise.
15029 (emit_common_tails_for_components): Likewise.
15030 (insert_prologue_epilogue_for_components): Likewise.
15032 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15034 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15037 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15039 * df-core.c (df_set_blocks): Start using auto_bitmap.
15040 (df_compact_blocks): Likewise.
15041 * df-problems.c (df_rd_confluence_n): Likewise.
15042 * df-scan.c (df_insn_rescan_all): Likewise.
15043 (df_process_deferred_rescans): Likewise.
15044 (df_update_entry_block_defs): Likewise.
15045 (df_update_exit_block_uses): Likewise.
15046 (df_entry_block_bitmap_verify): Likewise.
15047 (df_exit_block_bitmap_verify): Likewise.
15048 (df_scan_verify): Likewise.
15049 * lra-constraints.c (lra_constraints): Likewise.
15050 (undo_optional_reloads): Likewise.
15051 (lra_undo_inheritance): Likewise.
15052 * lra-remat.c (calculate_gen_cands): Likewise.
15053 (do_remat): Likewise.
15054 * lra-spills.c (assign_spill_hard_regs): Likewise.
15055 (spill_pseudos): Likewise.
15056 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15057 (bitmap_set_subtract_values): Likewise.
15059 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15061 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15062 management with auto_bitmap.
15063 (fix_inter_tick): Likewise.
15064 (fix_recovery_deps): Likewise.
15065 * ira.c (add_store_equivs): Likewise.
15066 (find_moveable_pseudos): Likewise.
15067 (split_live_ranges_for_shrink_wrap): Likewise.
15068 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15069 (rtx_reuse_manager::seen_def_p): Likewise.
15070 (rtx_reuse_manager::set_seen_def): Likewise.
15071 * print-rtl.h (class rtx_reuse_manager): Likewise.
15073 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15075 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15077 (migrate_btr_def): Likewise.
15078 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15079 * df-core.c (loop_post_order_compute): Likewise.
15080 (loop_inverted_post_order_compute): Likewise.
15081 * hsa-common.h: Likewise.
15082 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15083 * init-regs.c (initialize_uninitialized_regs): Likewise.
15084 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15085 (inline_small_functions): Likewise.
15086 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15087 * ira.c (combine_and_move_insns): Likewise.
15088 (build_insn_chain): Likewise.
15089 * loop-invariant.c (find_invariants): Likewise.
15090 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15091 * predict.c (tree_predict_by_opcode): Likewise.
15092 (predict_paths_leading_to): Likewise.
15093 (predict_paths_leading_to_edge): Likewise.
15094 (estimate_loops_at_level): Likewise.
15095 (estimate_loops): Likewise.
15096 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15097 (spread_components): Likewise.
15098 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15099 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15100 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15101 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15102 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15103 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15104 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15105 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15106 (mark_threaded_blocks): Likewise.
15107 (thread_through_all_blocks): Likewise.
15108 * tree-ssa.c (verify_ssa): Likewise.
15109 (execute_update_addresses_taken): Likewise.
15110 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15112 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15114 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15116 (post_order_compute): Likewise.
15117 (inverted_post_order_compute): Likewise.
15118 (pre_and_rev_post_order_compute_fn): Likewise.
15120 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15122 * genrecog.c (int_set::int_set): Explicitly construct our
15123 auto_vec base class.
15124 * vec.h (auto_vec::auto_vec): New constructor.
15126 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15128 * bitmap.h (class auto_bitmap): New constructor taking
15129 bitmap_obstack * argument.
15131 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15133 * bitmap.h (class auto_bitmap): Change type of m_bits to
15134 bitmap_head, and adjust ctor / dtor and member operators.
15136 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15138 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15139 when returned register mode doesn't match original mode.
15141 2017-05-12 Jeff Law <law@redhat.com>
15142 Jakub Jelinek <jakub@redhat.com>
15144 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15145 we look for cc setter after the compare-elim changes.
15146 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15147 within the vector to match what compare-elim now expects.
15148 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15149 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15151 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15152 after the compare-elim changes.
15153 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15154 the vector to match what compare-elim now expects.
15155 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15156 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15157 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15158 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15159 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15161 * config/visium/visium.c (single_set_and_flags): Fix where
15162 we look for cc setter after the compare-elim changes.
15163 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15164 with the vector to match what compare-elim now expects.
15165 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15166 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15167 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15168 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15169 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15170 (neg<mode>2_insn_set_overflow): Likewise.
15172 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15174 PR middle-end/79794
15175 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15176 maybe_expand_insn call, set ops[0].target. If still set after call,
15177 set alt_rtl. Add extra arg to recursive calls.
15178 (extract_bit_field): Add alt_rtl argument. Pass to
15180 * expmed.h (extract_bit_field): Fix prototype.
15181 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15182 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15183 to extract_bit_field_calls.
15184 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15185 Pass alt_rtl to extract_bit_field calls.
15186 * calls.c (store_unaligned_arguments_into_psuedos)
15187 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15188 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15190 * optabs.h (struct expand_operand): Add target bitfield.
15192 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15194 * compare-elim.c (try_eliminate_compare): Canonicalize
15195 operation with embedded compare to
15196 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15197 (set (reg) (operation)].
15199 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15201 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15204 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15205 cost of adding a carry flag for ADC instruction.
15206 [case MINUS]: Ignore the cost of subtracting a carry flag
15207 for SBB instruction.
15209 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15211 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15213 * config/rs6000/bmiintrin.h: New file.
15214 * config/rs6000/bmi2intrin.h: New file.
15215 * config/rs6000/x86intrin.h: New file.
15217 2017-05-12 Jeff Law <law@redhat.com>
15219 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15222 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15224 PR middle-end/80707
15225 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15227 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15229 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15230 early expansion of vector divide builtins.
15231 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15232 builtins identified as having unsigned arguments.
15234 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15236 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15237 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15238 expansion of vector logical operations (and, andc, or, xor,
15241 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15243 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15244 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15246 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15248 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15249 early GIMPLE expansion of vector multiplies.
15251 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15253 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15254 TARGET_HAVE_MOVT conditional.
15255 (movt splitter): Likewise.
15257 2017-05-12 Richard Biener <rguenther@suse.de>
15259 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15260 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15261 Fold all stmts not inplace.
15263 2017-05-12 Richard Biener <rguenther@suse.de>
15265 PR tree-optimization/80713
15266 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15267 inserted_exprs bit for not removed stmts.
15269 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15271 PR middle-end/69921
15272 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15273 parallelized" attribute for parallelized OpenACC kernels.
15274 * omp-offload.c (execute_oacc_device_lower): Use it.
15276 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15277 Set "oacc kernels" attribute.
15278 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15279 parameter. Adjust all users.
15280 (oacc_fn_attrib_kernels_p): Remove function.
15281 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15282 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15283 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15284 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15285 assert "oacc kernels" attribute is set.
15287 2017-05-11 Carl Love <cel@us.ibm.com>
15289 * config/rs6000/rs6000-c: Add support for built-in functions
15290 vector unsigned char vec_popcnt (vector signed char)
15291 vector unsigned char vec_popcnt (vector unsigned char)
15292 vector unsigned short vec_popcnt (vector signed short)
15293 vector unsigned short vec_popcnt (vector unsigned short)
15294 vector unsigned int vec_popcnt (vector signed int)
15295 vector unsigned int vec_popcnt (vector unsigned int)
15296 vector unsigned long long vec_popcnt (vector signed long long)
15297 vector unsigned long long vec_popcnt (vector unsigned long long)
15298 vector signed long long vec_slo (vector signed long long,
15299 vector signed char)
15300 vector signed long long vec_slo (vector signed long long,
15301 vector unsigned char)
15302 vector unsigned long long vec_slo (vector unsigned long long,
15303 vector signed char)
15304 vector unsigned long long vec_slo (vector unsigned long long,
15305 vector unsigned char)
15306 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15307 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15308 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15309 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15310 * doc/extend.texi: Update the built-in documentation file for the
15311 new built-in functions.
15313 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15315 * attribs.h (sorted_attr_string): Move machine independent
15316 functions for target clone support from the i386 port to common
15317 code. Rename ix86_function_versions to common_function_versions.
15318 Rename make_name to make_unique_name.
15319 (common_function_versions): Likewise.
15320 (make_unique_name): Likewise.
15321 (make_dispatcher_decl): Likewise.
15322 (is_function_default_version): Likewise.
15323 * attribs.c (attr_strcmp): Likewise.
15324 (sorted_attr_string): Likewise.
15325 (common_function_versions): Likewise.
15326 (make_unique_name): Likewise.
15327 (make_dispatcher_decl): Likewise.
15328 (is_function_default_version): Likewise.
15329 * config/i386/i386.c (attr_strcmp): Likewise.
15330 (sorted_attr_string): Likewise.
15331 (ix86_function_versions): Likewise.
15332 (make_name): Likewise.
15333 (make_dispatcher_decl): Likewise.
15334 (is_function_default_version): Likewise.
15335 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15337 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15340 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15341 Account for direct move costs for vec_construct of integer
15344 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15347 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15348 (UNSPEC_STX_ATOMIC): Ditto.
15349 (loaddi_via_sse): New insn.
15350 (storedi_via_sse): Ditto.
15351 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15352 Update corresponding peephole2 patterns.
15353 (atomic_storedi_fpu): Ditto.
15355 2017-05-11 Julia Koval <julia.koval@intel.com>
15357 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15358 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15360 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15361 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15362 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15364 2017-05-11 Nathan Sidwell <nathan@acm.org>
15366 * graphite-poly.c: Include dumpfile.h.
15368 * dumpfle.h (dump_function): Declare here ...
15369 * tree-dump.h (dump_function): ... not here.
15370 * dumpfile.c: #include tree-cfg.h.
15371 (dump_function): Move here from ...
15372 * tree-dump.c (dump_function): ... here.
15373 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15374 * graphite-poly.c: Don't include tree-dump.h.
15375 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15376 * print-tree.c: Likewise.
15377 * stor-layout.c: Likewise.
15378 * tree-nested.c: Likewise.
15380 * dumpfile.c (dump_start): Use TDF_FLAGS.
15381 (dump_enable_all): Fix TDF_KIND check thinko.
15383 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15385 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15386 array entries to represent two legal parameterizations of the
15387 overloaded __builtin_cmpb function, as represented by the
15388 P6_OV_BUILTIN_CMPB constant.
15389 (altivec_resolve_overloaded_builtin): Add special case handling
15390 for the __builtin_cmpb function, as represented by the
15391 P6_OV_BUILTIN_CMPB constant.
15392 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15393 (BU_P6_64BIT_2): New macro.
15394 (BU_P6_OVERLOAD_2): New macro
15395 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15396 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15397 (CMPB): Add overload support to represent both 32-bit and 64-bit
15398 compare-bytes function.
15399 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15400 support for TARGET_CMPB.
15401 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15402 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15403 documentation of the __builtin_cmpb overloaded built-in function.
15405 2017-05-11 Richard Biener <rguenther@suse.de>
15407 PR tree-optimization/80705
15408 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15409 bases are not vectorizable.
15411 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15413 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15414 when counting register pressure.
15416 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15418 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15419 (iv_ca_more_deps): Renamed to ...
15420 (iv_ca_compare_deps): ... this.
15421 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15423 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15425 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15427 (determine_group_iv_costs): ... here.
15428 (find_inv_vars_cb): Record inv var if it's not recorded before.
15430 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15432 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15433 (get_shiftadd_cost): Ditto.
15435 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15437 * tree-ssa-address.c: Include header file.
15438 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15440 (add_to_parts): Refactor.
15441 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15442 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15445 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15447 PR tree-optimization/53090
15448 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15450 (extract_cond_operands): Detect condition with IV on both sides
15451 and return COMP_IV_EXPR_2.
15452 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15453 (rewrite_use_compare): Simplify by removing call to function
15454 extract_cond_operands.
15456 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15458 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15459 (extract_cond_operands): Detect condition comparing against non-
15460 invariant bound and return appropriate enum value.
15461 (find_interesting_uses_cond): Update use of extract_cond_operands.
15462 Handle its return value accordingly.
15463 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15465 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15467 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15468 nonlinear iv_use computation in loop invariant sensitive way.
15470 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15472 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15473 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15475 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15477 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15478 (dump_cand): Support iv_cand.inv_exprs.
15479 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15481 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15484 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15486 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15488 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15489 as local function. Include necessary header files.
15490 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15492 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15494 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15496 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15498 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15499 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15500 RSHIFT_EXPR and BIT_NOT_EXPR.
15502 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15504 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15505 (adjust_setup_cost): New parameter supporting round up adjustment.
15506 (struct address_cost_data): Delete.
15507 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15508 (split_address_cost, ptr_difference_cost): Delete.
15509 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15510 (struct ainc_cost_data): New struct.
15511 (get_address_cost_ainc): New function.
15512 (get_address_cost, get_computation_cost): Reimplement.
15513 (determine_group_iv_cost_address): Record inv_expr for all uses of
15515 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15516 (iv_ca_has_deps): Reimplemented to ...
15517 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15519 (iv_ca_extend): Call iv_ca_more_deps.
15521 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15523 * tree-ssa-address.c (struct mem_address): Move to header file.
15524 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15525 * tree-ssa-address.h (struct mem_address): Move from C file.
15526 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15528 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15530 * tree-affine.h (aff_combination_type): New interface.
15531 (aff_combination_zero_p): Remove static.
15532 (aff_combination_const_p): New interface.
15533 (aff_combination_singleton_var_p): New interfaces.
15535 2017-05-11 Richard Biener <rguenther@suse.de>
15537 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15538 Skip unreachable blocks and destinations.
15539 (eliminate): Move stmt removal and fixup ...
15540 (fini_eliminate): ... here. Skip inserted exprs.
15541 (pass_pre::execute): Move fini_pre after fini_eliminate.
15542 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15543 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15544 PRE to get rid of dead code that has invalid SSA form and
15545 split critical edges again.
15547 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15549 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15551 2017-05-11 Richard Biener <rguenther@suse.de>
15553 * passes.c (execute_function_todo): Verify loops if they are
15554 said to be up-to-date.
15555 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15556 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15558 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15561 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15562 handle calling assemble_external ourself.
15565 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15566 modes with zero size. Enhance comment.
15568 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15570 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15571 built-ins for vec_xl and vec_xst with short and char pointer
15574 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15576 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15577 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15578 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15579 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15580 (_mm_maskz_min_round_ss): New intrinsics.
15581 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15582 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15583 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15584 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15585 (__builtin_ia32_minss_mask_round): New builtins.
15586 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15587 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15588 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15590 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15591 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15593 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15596 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15598 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15599 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15600 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15601 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15602 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15603 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15604 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15605 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15606 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15607 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15608 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15609 (__builtin_ia32_mulss_mask_round): New builtins.
15610 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15611 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15612 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15614 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15615 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15617 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15620 2017-05-10 Julia Koval <julia.koval@intel.com>
15622 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15623 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15624 (_mm256_setr_m128i): New intrinsics.
15626 2017-05-10 Julia Koval <julia.koval@intel.com>
15628 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15629 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15630 (_mm_maskz_rcp14_ss): New intrinsics.
15631 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15632 (__builtin_ia32_rcp14ss_mask): New builtins.
15633 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15635 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15637 PR tree-optimization/51513
15638 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15639 (assert_unreachable_fallthru_edge_p): Use it.
15640 (group_case_labels_stmt): Likewise.
15641 * tree-cfg.h: Prototype it.
15642 * stmt.c: Include cfghooks.h and tree-cfg.h.
15643 (emit_case_dispatch_table) <gap_label>: New local variable.
15644 Use it to fill dispatch table gaps.
15645 Test for default_label before updating probabilities.
15646 (expand_case) <default_label>: Remove unneeded initialization.
15647 Test for unreachable default case statement and remove its edge.
15648 Set default_label accordingly.
15649 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15651 2017-05-10 Carl Love <cel@us.ibm.com>
15653 * config/rs6000/rs6000-c: Add support for built-in functions
15654 vector signed char vec_neg (vector signed char)
15655 vector signed short int vec_neg (vector short int)
15656 vector signed int vec_neg (vector signed int)
15657 vector signed long long vec_neg (vector signed long long)
15658 vector float vec_neg (vector float)
15659 vector double vec_neg (vector double)
15660 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15662 * config/rs6000/altivec.h: Add define for vec_neg
15663 * doc/extend.texi: Update the built-in documentation for the
15664 new built-in functions.
15666 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15668 PR tree-optimization/77644
15669 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15671 2017-05-10 Nathan Sidwell <nathan@acm.org>
15673 * dumpfile.h (TDI_lang_all): New.
15674 (TDF_KIND): New. Renumber others
15675 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15677 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15679 (get_dump_file_name): Adjust suffix generation.
15680 (dump_enable_all): Use TDF_KIND.
15681 * doc/invoke.texi (-fdump-lang-all): Document.
15683 * dumpfile.h: Tabify.
15685 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15688 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15689 Move member access before delete.
15691 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15693 * tree-inline.c (expand_call_inline): Split block at stmt
15696 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15699 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15700 are now unused after splitting mov{sf,sd}_hardfloat.
15701 (f32_lr2): Likewise.
15702 (f32_lm): Likewise.
15703 (f32_lm2): Likewise.
15704 (f32_li): Likewise.
15705 (f32_li2): Likewise.
15706 (f32_lv): Likewise.
15707 (f32_sr): Likewise.
15708 (f32_sr2): Likewise.
15709 (f32_sm): Likewise.
15710 (f32_sm2): Likewise.
15711 (f32_si): Likewise.
15712 (f32_si2): Likewise.
15713 (f32_sv): Likewise.
15714 (f32_dm): Likewise.
15715 (f32_vsx): Likewise.
15716 (f32_av): Likewise.
15717 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15718 For movsf, order stores so the VSX stores occur before the GPR
15719 store which encourages the register allocator to use a traditional
15720 FPR instead of a GPR. For movsd, order the stores so that the GPR
15721 store comes before the VSX stores to allow the power6 to work.
15722 This is due to the power6 not having a 32-bit integer store
15723 instruction from a FPR.
15724 (movsf_hardfloat): Likewise.
15725 (movsd_hardfloat): Likewise.
15727 2017-05-09 Martin Sebor <msebor@redhat.com>
15729 PR translation/80280
15730 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15733 PR translation/80280
15734 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15735 data member added in r247778.
15736 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15738 2017-05-09 Nathan Sidwell <nathan@acm.org>
15740 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15742 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15745 2017-05-09 Marek Polacek <polacek@redhat.com>
15747 * doc/invoke.texi: Fix typo.
15749 2017-05-09 Richard Biener <rguenther@suse.de>
15751 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15752 (vrp_val_is_min): Likewise.
15753 (set_value_range_to_value): Likewise.
15754 (set_value_range_to_nonnegative): Likewise.
15755 (gimple_assign_nonzero_p): Likewise.
15756 (gimple_stmt_nonzero_p): Likewise.
15757 (vrp_int_const_binop): Likewise. Remove unreachable case.
15758 (adjust_range_with_scev): Adjust comments.
15759 (compare_range_with_value): Likewise.
15760 (extract_range_from_phi_node): Likewise.
15761 (test_for_singularity): Likewise.
15763 2017-05-09 Richard Biener <rguenther@suse.de>
15765 * tree-vrp.c (get_single_symbol): Add assert that we don't
15766 get overflowed constants as invariant part.
15767 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15768 checks. Use wi::cmp instead of recursing for integer constants.
15769 (compare_values): Just ignore whether we assumed undefined
15770 overflow instead of failing the compare.
15771 (extract_range_for_var_from_comparison_expr): Add comment before the
15772 TREE_NO_WARNING sets.
15773 (test_for_singularity): Likewise.
15774 (extract_range_from_comparison): Do not disable optimization
15775 when we assumed undefined overflow.
15776 (extract_range_basic): Remove init of unused var.
15778 2017-05-09 Richard Biener <rguenther@suse.de>
15780 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15781 (extract_range_from_multiplicative_op_1): Adjust.
15782 (extract_range_from_binary_expr_1): Use int_const_binop.
15784 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15787 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15788 rs6000_store_data_bypass_p in seven define_bypass directives and
15789 in several comments.
15790 * config/rs6000/rs6000-protos.h: Add prototype for
15791 rs6000_store_data_bypass_p function.
15792 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15793 function implements slightly different (rs6000-specific) semantics
15794 than store_data_bypass_p, returning false rather than aborting
15795 with assertion error when arguments do not satisfy the
15796 requirements of store data bypass.
15797 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15798 rs6000_store_data_bypass_p.
15800 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15802 * config/xtensa/xtensa-protos.h
15803 (xtensa_initial_elimination_offset): New declaration.
15804 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15805 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15806 macro definition, add case for FRAME_POINTER_REGNUM when
15807 FRAME_GROWS_DOWNWARD.
15808 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15809 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15810 xtensa_initial_elimination_offset.
15812 2017-05-08 Nathan Sidwell <nathan@acm.org>
15814 * doc/invoke.texi: Alphabetize -fdump options.
15816 2017-05-08 Martin Sebor <msebor@redhat.com>
15818 PR translation/80280
15819 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15821 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15823 * target.def (compute_frame_layout): New optional target hook.
15824 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15825 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15826 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15828 * reload1.c (verify_initial_elim_offsets): Likewise.
15829 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15830 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15831 (arm_get_frame_offsets): Split up into this ...
15832 (arm_compute_frame_layout): ... and this function.
15834 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15836 * config/aarch64/constraints.md (Usa): New constraint.
15837 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15839 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15841 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15842 with_multilib_list after it has been checked.
15844 2017-05-08 Richard Biener <rguenther@suse.de>
15846 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15847 (bitmap_set_subtract_values): Likewise.
15849 2017-05-08 Richard Biener <rguenther@suse.de>
15851 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15852 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15854 (gimple_stmt_nonzero_warnv_p): Rename to ...
15855 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15857 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15858 (extract_range_basic): Adjust, do not disable propagation on
15859 strict overflow sensitive simplification.
15860 (vrp_visit_cond_stmt): Likewise.
15862 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15864 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15865 body size unconditionally.
15867 2017-05-07 Jeff Law <law@redhat.com>
15870 2017-05-06 Jeff Law <law@redhat.com>
15871 PR tree-optimization/78496
15872 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15875 PR tree-optimization/78496
15876 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15877 (simplify_stmt_using_ranges): Call it.
15878 (vrp_dom_walker::before_dom_children): Extract equivalences
15879 from an ASSERT_EXPR with an equality comparison against a
15882 2017-05-06 Jeff Law <law@redhat.com>
15884 PR tree-optimization/78496
15885 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15888 PR tree-optimization/78496
15889 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15890 (simplify_stmt_using_ranges): Call it.
15891 (vrp_dom_walker::before_dom_children): Extract equivalences
15892 from an ASSERT_EXPR with an equality comparison against a
15895 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15897 * lra-constraints.c (lra_copy_reg_equiv): New function.
15898 (split_reg): Use it to copy equivalence information from the
15899 original register to the spill register.
15901 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15903 PR rtl-optimization/75964
15904 * simplify-rtx.c (simplify_const_relational_operation): Remove
15905 invalid handling of comparisons of integer ABS.
15907 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15909 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15910 initialize to zero.
15911 (init_regs): Remove declaration.
15912 (function_arg_advance_32): Initialize error_p as boolean variable.
15914 2017-05-05 Nathan Sidwell <nathan@acm.org>
15916 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15917 lines. Use for (;;).
15919 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15921 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15922 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15923 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15924 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15925 VF=2 that require versioning.
15927 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15929 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15932 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15934 * diagnostic.h (diagnostic_override_option_index): Convert from
15935 macro to inline function.
15937 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15939 * diagnostic.c (last_module_changed_p): New function.
15940 (set_last_module): New function.
15941 (diagnostic_report_current_module): Convert macro usage to
15942 the above functions.
15943 * diagnostic.h (diagnostic_context::last_module): Strengthen
15944 from const line_map * to const line_map_ordinary *.
15945 (diagnostic_last_module_changed): Delete macro.
15946 (diagnostic_set_last_module): Delete macro.
15948 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15950 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15951 with diagnostic_report_diagnostic.
15952 (diagnostic_n_impl_richloc): Likewise.
15953 * diagnostic.h (report_diagnostic): Delete macro.
15954 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15955 with diagnostic_report_diagnostic.
15956 * substring-locations.c (format_warning_va): Likewise.
15958 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15960 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15961 save/restor of format_spec. Move option-printing code to...
15962 (print_option_information): ...this new function, and
15963 reimplement by simply printing to the pretty_printer,
15964 rather than appending to the format string.
15966 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15968 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15969 handling logic into...
15970 (update_effective_level_from_pragmas): ...this new function.
15972 2017-05-04 Andrew Waterman <andrew@sifive.com>
15974 * config/riscv/riscv.opt (mstrict-align): New option.
15975 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15976 (SLOW_UNALIGNED_ACCESS): Define.
15977 (riscv_slow_unaligned_access): Declare.
15978 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15980 (riscv_slow_unaligned_access): New variable.
15981 (rocket_tune_info): Set slow_unaligned_access to true.
15982 (optimize_size_tune_info): Set slow_unaligned_access to false.
15983 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15984 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15985 (riscv_option_override): Set riscv_slow_unaligned_access.
15986 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15988 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15990 * config/riscv/riscv.md: Unify indentation.
15992 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15997 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15999 (extendsi<mode>2): Add support for doing sign extension via
16000 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16001 don't have ISA 3.0 instructions.
16002 (extendsi<mode>2 splitter): Likewise.
16003 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16004 generate the normal insns since SImode can now go in vector
16005 registers. Disallow the special UNSPECs needed for previous
16006 machines to hide SImode being used. Add new insns
16007 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16008 (fix_trunc<mode>si2_stfiwx): Likewise.
16009 (fix_trunc<mode>si2_internal): Likewise.
16010 (fixuns_trunc<mode>si2): Likewise.
16011 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16012 (fctiw<u>z_<mode>_smallint): Likewise.
16013 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16014 of floating point to 32-bit integer from doing a direct move to
16015 the GPR registers to do a store.
16016 (fctiwz_<mode>): Break long line.
16018 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16020 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16021 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16022 (addr_list, addr_offset_valid_p): New.
16023 (split_address_groups): Check offset validity with above function.
16024 (gt-tree-ssa-loop-ivopts.h): Include header file.
16026 2017-05-05 Nathan Sidwell <nathan@acm.org>
16028 * config.gcc (arm*-*-*): Add missing 'fi'.
16030 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16032 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16033 included in -fopt-info does not matter.
16034 * doc/optinfo.texi (-fopt-info): Fix description of default
16035 behavour. Explicitly say order of options included in -fopt-info
16038 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16040 * config.gcc: Allow combinations of aprofile and rmprofile values for
16041 --with-multilib-list.
16042 * config/arm/t-multilib: New file.
16043 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16044 variables. Remove setting of ISA and floating-point ABI in
16045 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16046 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16047 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16048 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16050 * config/arm/t-rmprofile: Likewise except for the matches changes.
16051 * doc/install.texi (--with-multilib-list): Document the combination of
16052 aprofile and rmprofile values and warn about pitfalls in doing that.
16054 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16056 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16057 (movdi_aarch64): Likewise.
16059 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16061 PR tree-optimization/80632
16062 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16064 (build_arrays): Initialize it for virtual phis.
16065 (fix_phi_nodes): Use it for virtual phis.
16067 PR tree-optimization/80558
16068 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16069 [x, y] op z into [x op, y op z] for op & or | if conditions
16072 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16073 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16076 * config/arm/arm.md (use_literal_pool): Remove.
16077 (64-bit immediate split): No longer takes cost into consideration
16078 if arm_disable_literal_pool is enabled.
16079 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16080 used when arm_disable_literal_pool is enabled.
16081 (arm_max_const_double_inline_cost): Remove use of
16082 arm_disable_literal_pool.
16083 (push_minipool_fix): Add assert.
16084 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16085 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16086 (no_literal_pool_sf_immediate): New.
16088 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16090 PR tree-optimization/80613
16091 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16092 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16094 2017-05-05 Richard Biener <rguenther@suse.de>
16096 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16098 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16100 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16101 of this flag from insn conditions due to removal from r247495.
16103 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16105 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16107 (arm_early_store_addr_dep_ptr): Likewise.
16108 * config/arm/aarch-common-protos.h
16109 (arm_early_load_addr_dep_ptr): Add prototype.
16110 (arm_early_store_addr_dep_ptr): Likewise.
16111 * config/arm/cortex-a53.md: Add new bypasses.
16113 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16115 * tree.c (next_type_uid): Change type to unsigned.
16116 (type_hash_canon): Decrement back next_type_uid if
16117 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16118 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16121 2017-05-04 Martin Sebor <msebor@redhat.com>
16123 * builtins.c: Fix a trivial typo in a comment.
16125 PR middle-end/79234
16126 * builtins.c (check_sizes): Adjust to handle reading past the end.
16127 Avoid printing excessive upper bound of ranges. Use %E to print
16128 tree nodes instead of converting them to %wu.
16129 (expand_builtin_memchr): New function.
16130 (compute_dest_size): Rename...
16131 (compute_objsize): ...to this.
16132 (expand_builtin_memcpy): Adjust.
16133 (expand_builtin_mempcpy): Adjust.
16134 (expand_builtin_strcat): Adjust.
16135 (expand_builtin_strcpy): Adjust.
16136 (check_strncat_sizes): Adjust.
16137 (expand_builtin_strncat): Adjust.
16138 (expand_builtin_strncpy): Adjust and simplify.
16139 (expand_builtin_memset): Adjust.
16140 (expand_builtin_bzero): Adjust.
16141 (expand_builtin_memcmp): Adjust.
16142 (expand_builtin): Handle memcmp.
16143 (maybe_emit_chk_warning): Check strncat just once.
16145 2017-05-04 Martin Sebor <msebor@redhat.com>
16147 PR preprocessor/79214
16148 PR middle-end/79222
16149 PR middle-end/79223
16150 * builtins.c (check_sizes): Add inlining context and issue
16151 warnings even when -Wno-system-headers is set.
16152 (check_strncat_sizes): Same.
16153 (expand_builtin_strncat): Same.
16154 (expand_builtin_memmove): New function.
16155 (expand_builtin_stpncpy): Same.
16156 (expand_builtin): Handle memmove and stpncpy.
16158 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16160 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16161 which is not used any more.
16163 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16165 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16167 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16169 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16170 (cortexa53_tunings): Likewise.
16171 (cortexa57_tunings): Likewise.
16172 (cortexa72_tunings): Likewise.
16173 (cortexa73_tunings): Likewise.
16175 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16177 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16178 Set loop alignment to 8.
16180 2017-05-04 Martin Sebor <msebor@redhat.com>
16182 PR translation/80280
16183 * builtins.c (expand_builtin_object_size): Add missing quoting to
16184 %D and like directives.
16185 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16186 (hsa_type_for_tree_type): Same.
16187 (verify_function_arguments): Same.
16188 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16189 * varasm.c (get_section): Same.
16192 2017-05-04 Martin Sebor <msebor@redhat.com>
16194 PR translation/80280
16195 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16197 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16199 * config/aarch64/aarch64.c (generic_addrcost_table):
16200 Change HI/TI mode setting.
16202 2017-05-04 Martin Jambor <mjambor@suse.cz>
16204 PR tree-optimization/80622
16205 * tree-sra.c (comes_initialized_p): New function.
16206 (build_accesses_from_assign): Only set write lazily when
16207 comes_initialized_p is false.
16208 (analyze_access_subtree): Use comes_initialized_p.
16209 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16210 instead of testing for PARM_DECL.
16212 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16214 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16215 constraint on operand 0 to allow more general addressing modes.
16216 Adjust output template.
16217 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16219 * config/aarch64/aarch64-protos.h
16220 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16221 * config/aarch64/constraints.md (Dp): New address constraint.
16222 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16225 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16227 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16228 update use of estimate_ipcp_clone_size_and_time.
16229 (estimate_local_effects): Update use of
16230 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16231 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16232 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16233 Return nonspecialized time.
16235 2017-05-04 Richard Biener <rguenther@suse.de>
16237 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16238 for the last VUSE which def dominates the PHI. Directly call
16240 (get_continuation_for_phi_1): Remove.
16242 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16244 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16245 to explain the use of truncating division. Cap the number of
16246 iterations to the maximum given by nb_iterations_upper_bound,
16249 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16251 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16252 * configure: Regenerate.
16253 * config.in: Regenerate.
16254 * config/i386/driver-mingw32.c: new file.
16255 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16256 * config.host: Link driver-mingw32.o on MinGW host.
16257 * doc/install.texi: Document new --enable-mingw-wildcard configure
16260 2017-05-04 Marek Polacek <polacek@redhat.com>
16262 PR tree-optimization/80612
16263 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16265 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16266 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16268 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16269 (movt splitter): Likewise.
16270 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16271 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16272 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16273 block for Thumb-1 with MOVT.
16274 (thumb2_legitimate_address_p): Move code block ...
16275 (can_avoid_literal_pool_for_label_p): ... into this new function.
16276 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16278 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16279 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16280 "M-profile targets with the MOVT instruction".
16282 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16284 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16285 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16286 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16288 2017-05-04 Martin Liska <mliska@suse.cz>
16290 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16291 variable cond_code.
16293 2017-05-04 Richard Biener <rguenther@suse.de>
16295 * tree.c (array_at_struct_end_p): Handle arrays at struct
16296 end with flexarrays more conservatively. Refactor and treat
16297 arrays of arrays or aggregates more strict. Fix
16298 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16299 * tree.c (array_at_struct_end_p): Adjust prototype.
16300 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16301 * gimple-fold.c (get_range_strlen): Likewise.
16302 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16304 2017-05-04 Richard Biener <rguenther@suse.de>
16306 PR tree-optimization/31130
16307 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16309 (supports_overflow_infinity): Likewise.
16310 (is_negative_overflow_infinity): Likewise.
16311 (is_positive_overflow_infinity): Likewise.
16312 (is_overflow_infinity): Likewise.
16313 (stmt_overflow_infinity): Likewise.
16314 (overflow_infinity_range_p): Likewise.
16315 (usable_range_p): Remove as always returning true.
16316 (make_overflow_infinity): Remove.
16317 (negative_overflow_infinity): Likewise.
16318 (positive_overflow_infinity): Likewise.
16319 (avoid_overflow_infinity): Likewise.
16320 (set_value_range): Adjust accordingly.
16321 (set_value_range_to_nonnegative): Likewise, remove now unused
16322 overflow_infinity arg.
16323 (vrp_operand_equal_p): Adjust.
16324 (update_value_range): Likewise.
16325 (range_int_cst_singleton_p): Likewise.
16326 (operand_less_p): Likewise.
16327 (compare_values_warnv): Likewise.
16328 (extract_range_for_var_from_comparison_expr): Likewise.
16329 (vrp_int_const_binop): Likewise.
16330 (zero_nonzero_bits_from_vr): Likewise.
16331 (extract_range_from_multiplicative_op_1): Likewise.
16332 (extract_range_from_binary_expr_1): Likewise.
16333 (extract_range_from_unary_expr): Likewise.
16334 (extract_range_from_comparison): Likewise.
16335 (extract_range_basic): Likewise.
16336 (adjust_range_with_scev): Likewise.
16337 (compare_ranges): Likewise.
16338 (compare_range_with_value): Likewise.
16339 (dump_value_range): Likewise.
16340 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16342 (simplify_cond_using_ranges): Adjust.
16344 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16346 * brig-builtins.def: Added a builtin for class_f64.
16347 * builtin-types.def: Added a builtin type needed by class_f64.
16349 2017-05-03 Jason Merrill <jason@redhat.com>
16351 * timevar.def: Add TV_CONSTEXPR.
16353 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16355 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16357 2017-05-03 Martin Jambor <mjambor@suse.cz>
16359 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16360 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16361 * ipa-cp.c (ipcp_propagate_stage): Do not call
16362 ipa_update_after_lto_read.
16363 * ipa-inline.c (ipa_inline): Likewise.
16365 2017-05-03 Martin Jambor <mjambor@suse.cz>
16367 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16368 tag. Added a default constructor and a destructor.
16369 (ipa_edge_args_sum_t): New class;
16370 (ipa_edge_args_sum): Declare.
16371 (ipa_edge_args_vector): Remove declaration.
16372 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16373 (ipa_free_edge_args_substructures): Remove declaration.
16374 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16375 (ipa_edge_args_info_available_for_edge_p): Likewise.
16376 * ipa-prop.c (ipa_edge_args_vector): Removed.
16377 (edge_removal_hook_holder): Likewise.
16378 (edge_duplication_hook_holder): Likewise.
16379 (ipa_edge_args_sum): New variable.
16380 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16381 ipa_edge_args_vector.
16382 (ipa_free_edge_args_substructures): Likewise.
16383 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16384 ipa_edge_args_vector.
16385 (ipa_edge_removal_hook): Turned into method
16386 ipa_edge_args_sum_t::remove.
16387 (ipa_edge_duplication_hook): Turned into method
16388 ipa_edge_args_sum_t::duplicate.
16389 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16390 registering edge hooks.
16391 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16392 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16393 ipa_edge_args_sum instead of ipa_edge_args_vector.
16394 * ipa-profile.c (ipa_profile): Likewise.
16396 2017-05-03 Martin Jambor <mjambor@suse.cz>
16398 * symbol-summary.h (function_summary): New method exists.
16399 (function_summary::symtab_removal): Deallocate through release.
16400 (call_summary): New class.
16401 (gt_ggc_mx): New overload.
16402 (gt_pch_nx): Likewise.
16403 (gt_pch_nx): Likewise.
16405 2017-05-03 Jeff Law <law@redhat.com>
16407 PR tree-optimization/78496
16408 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16409 from simplify_cond_using_ranges. Split off code to walk
16410 backwards through casts into ...
16411 (simplify_cond_using_ranges_2): New function.
16412 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16413 (execute_vrp): After identifying jump threads, call
16414 simplify_cond_using_ranges_2.
16416 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16419 * ipa-inline.h (inline_summary): Add ctor.
16420 (create_ggc): Do not use ggc_cleared_alloc.
16422 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16423 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16425 * gcc.c (handle_braces): Support escaping in switch matching
16427 * doc/invoke.texi (Spec Files): Document it.
16428 Remove superfluous @code markup in items.
16430 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16432 * diagnostic-show-locus.c (struct column_range): New struct.
16433 (get_affected_columns): New function.
16434 (get_printed_columns): New function.
16435 (struct correction): New struct.
16436 (correction::ensure_capacity): New function.
16437 (correction::ensure_terminated): New function.
16438 (struct line_corrections): New struct.
16439 (line_corrections::~line_corrections): New dtor.
16440 (line_corrections::add_hint): New function.
16441 (layout::print_trailing_fixits): Reimplement in terms of the new
16443 (selftest::test_overlapped_fixit_printing): New function.
16444 (selftest::diagnostic_show_locus_c_tests): Call it.
16446 2017-05-03 Nathan Sidwell <nathan@acm.org>
16448 Canonicalize canonical type hashing
16449 * tree.h (type_hash_canon_hash): Declare.
16450 * tree.c (type_hash_list, attribute_hash_list): Move into
16451 type_hash_canon_hash.
16452 (build_type_attribute_qual_variant): Break out hash code calc into
16453 type_hash_canon_hash.
16454 (type_hash_canon_hash): New. Generic type hash computation.
16455 (build_range_type_1, build_array_type_1, build_function_type,
16456 build_method_type_directly, build_offset_type, build_complex_type,
16457 make_vector_type): Call it.
16459 2017-05-03 Richard Biener <rguenther@suse.de>
16461 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16462 When all DRs have unknown misaligned do not always peel
16463 when there is a store but apply the same costing model as if
16464 there were only loads.
16466 2017-05-03 Richard Biener <rguenther@suse.de>
16469 PR tree-optimization/80492
16470 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16471 compare_base_decls returning dont-know properly.
16473 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16475 * config/arm/iterators.md (CCSI): New mode iterator.
16476 (arch): New mode attribute.
16477 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16478 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16479 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16480 code iterator for success result mode.
16481 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16482 the corresponding new insn generators.
16484 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16487 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16488 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16490 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16492 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16493 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16494 (DDR_A): Wrap DDR argument in brackets.
16495 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16496 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16497 (DDR_REVERSED_P): Likewise.
16499 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16501 PR tree-optimization/79472
16502 * tree-switch-conversion.c (struct switch_conv_info): Add
16503 contiguous_range and default_case_nonstandard fields.
16504 (collect_switch_conv_info): Compute contiguous_range and
16505 default_case_nonstandard fields, don't clear final_bb if
16506 contiguous_range and only the default case doesn't have the required
16508 (check_all_empty_except_final): Set default_case_nonstandard instead
16509 of failing if contiguous_range and the default case doesn't have empty
16511 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16512 and only the default case doesn't have the required constants. Skip
16514 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16515 if default_case_nonstandard.
16516 (build_constructors): Build constant 1 just once. Assert that default
16517 values aren't inserted in between cases if contiguous_range. Skip
16519 (build_arrays): Skip virtual phis.
16520 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16521 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16522 Handle virtual phis.
16523 (gen_inbound_check): Handle default_case_nonstandard case.
16524 (process_switch): Adjust check_final_bb caller. Call
16525 gather_default_values with the first non-default case instead of
16526 default case if default_case_nonstandard.
16528 2017-05-02 Nathan Sidwell <nathan@acm.org>
16530 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16531 check. Fix formatting.
16533 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16535 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16536 errors when comparing specialized and unspecialized times.
16538 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16540 * diagnostic-show-locus.c
16541 (layout::should_print_annotation_line_p): Make private.
16542 (layout::print_annotation_line): Make private.
16543 (layout::annotation_line_showed_range_p): Make private.
16544 (layout::show_ruler): Make private.
16545 (layout::print_source_line): Make private. Pass in line and
16546 line_width, rather than calling location_get_source_line. Drop
16548 (layout::print_leading_fixits): New method.
16549 (layout::print_any_fixits): Rename to...
16550 (layout::print_trailing_fixits): ...this, and make private.
16551 Don't print newline fixits.
16552 (diagnostic_show_locus): Move logic for printing one row into...
16553 (layout::print_line): ...this new function. Move the
16554 location_get_source_line call and error-handling from
16555 print_source_line to here. Call print_leading_fixits, and rename
16556 print_any_fixits to print_trailing_fixits.
16557 (selftest::test_fixit_insert_containing_newline): Update now that
16558 newlines are partially supported.
16559 (selftest::test_fixit_insert_containing_newline_2): New test.
16560 (selftest::test_fixit_replace_containing_newline): Update comments.
16561 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16562 * edit-context.c (class added_line): New class.
16563 (class edited_line): Describe newline handling in comment.
16564 (edited_line::actually_edited_p): New method.
16565 (edited_line::print_content): Delete redundant decl.
16566 (edited_line::m_predecessors): New field.
16567 (edited_file::print_content): Call edited_line::print_content.
16568 (edited_file::print_diff): Update to support newlines.
16569 (edited_file::print_diff_hunk): Likewise.
16570 (edited_file::print_run_of_changed_lines): New function.
16571 (edited_file::print_diff_line): Convert to...
16572 (print_diff_line): ...this.
16573 (edited_file::get_effective_line_count): New function.
16574 (edited_line::edited_line): Initialize new field m_predecessors.
16575 (edited_line::~edited_line): Clean up m_predecessors.
16576 (edited_line::apply_fixit): Handle newlines.
16577 (edited_line::get_effective_line_count): New function.
16578 (edited_line::print_content): New function.
16579 (edited_line::print_diff_lines): New function.
16580 (selftest::test_applying_fixits_insert_containing_newline): New
16582 (selftest::test_applying_fixits_replace_containing_newline): New
16584 (selftest::insert_line): New function.
16585 (selftest::test_applying_fixits_multiple_lines): Add example of
16587 (selftest::edit_context_c_tests): Call the new tests.
16589 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16591 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16592 parameter cand. Update dump information.
16593 (get_computation_cost): Update uses.
16595 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16597 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16598 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16599 (get_computation_at, rewrite_use_address): Update use of
16600 get_computation_aff.
16602 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16604 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16605 (get_computation): Delete.
16606 (get_computation_cost): Implement like get_computation_cost_at.
16607 Use get_computation_at.
16608 (get_computation_cost_at): Delete.
16609 (rewrite_use_nonlinear_expr): Use get_computation_at.
16610 (rewrite_use_compare, remove_unused_ivs): Ditto.
16612 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16614 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16616 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16618 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16619 (ivopts_global_cost_for_size): Rename parameter and update uses.
16620 (iv_ca_recount_cost): Update uses.
16621 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16622 candidates seperately in n_invs and n_cands.
16623 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16625 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16627 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16628 (find_inv_vars_cb): New.
16629 (find_depends): Renamed to ...
16630 (find_inv_vars): ... this.
16631 (add_candidate_1, force_var_cost): Call find_inv_vars.
16632 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16634 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16636 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16637 inv_vars. Add inv_exprs.
16638 (struct iv_cand): Rename depends_on to inv_vars.
16639 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16640 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16641 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16642 (dump_cand): Dump inv_vars.
16643 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16644 (record_invariant, find_depends, add_candidate_1): Ditto.
16645 (set_group_iv_cost, force_var_cost): Ditto.
16646 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16647 (get_computation_cost_at, get_computation_cost): Ditto.
16648 (determine_group_iv_cost_generic): Ditto.
16649 (determine_group_iv_cost_address): Ditto.
16650 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16651 (determine_group_iv_costs): Ditto.
16652 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16653 (iv_ca_set_remove_invariants): Renamed to ...
16654 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16655 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16656 (iv_ca_set_add_invariants): Renamed to ...
16657 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16658 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16659 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16660 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16661 (create_new_ivs): Remove useless dump.
16663 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16665 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16667 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16668 (iv_ca_set_no_cp, create_new_iv): Ditto.
16670 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16672 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16674 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16676 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16677 function tree_check2.
16679 2017-05-02 Martin Liska <mliska@suse.cz>
16681 * doc/gcov.texi: Add missing preposition.
16682 * gcov.c (function_info::function_info): Properly fill up
16683 all member variables.
16685 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16687 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16689 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16691 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16693 2017-05-02 Martin Liska <mliska@suse.cz>
16696 * lto-streamer-in.c (lto_read_tree_1): Remove
16697 LTO_STREAMER_DEBUG.
16698 * lto-streamer.c (struct tree_hash_entry): Likewise.
16699 (struct tree_entry_hasher): Likewise.
16700 (tree_entry_hasher::hash): Likewise.
16701 (tree_entry_hasher::equal): Likewise.
16702 (lto_streamer_init): Likewise.
16703 (lto_orig_address_map): Likewise.
16704 (lto_orig_address_get): Likewise.
16705 (lto_orig_address_remove): Likewise.
16706 * lto-streamer.h: Likewise.
16707 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16708 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16710 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16712 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16713 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16714 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16715 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16716 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16717 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16718 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16719 (mm_maskz_sub_ss): New intrinsics.
16720 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16721 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16722 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16723 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16724 (__builtin_ia32_subss_mask_round): New builtins.
16725 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16726 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16727 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16729 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16730 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16732 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16735 2017-05-02 Martin Jambor <mjambor@suse.cz>
16737 PR tree-optimization/78687
16738 * tree-sra.c (access): New field parent.
16739 (process_subtree_disqualification): New function.
16740 (disqualify_candidate): Call it.
16741 (build_accesses_from_assign): Reset write flag if creating an
16743 (build_access_subtree): Fill in parent field and also prpagate
16744 down grp_write flag.
16745 (create_artificial_child_access): New parameter set_grp_write, set
16746 grp_write to its value.
16747 (propagate_subaccesses_across_link): Also propagate grp_write flag
16749 (propagate_all_subaccesses): Push the closest parent back to work
16750 queue if add_access_to_work_queue returned true.
16752 2017-05-02 Richard Biener <rguenther@suse.de>
16754 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16755 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16756 -fstrict-overflow documentation.
16757 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16758 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16759 flag_strict_overflow.
16760 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16761 * lto-opts.c (lto_write_options): Do not stream it.
16762 * lto-wrapper.c (merge_and_complain): Do not handle it.
16763 * opts.c (default_options_table): Do not set -fstrict-overflow.
16764 (finish_options): Likewise do not clear it when sanitizing.
16765 * simplify-rtx.c (simplify_const_relational_operation): Do not
16766 test flag_strict_overflow.
16768 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16770 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16771 using enabled attribute.
16772 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16773 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16774 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16775 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16776 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16777 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16778 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16779 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16780 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16781 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16783 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16785 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16787 2017-05-02 Richard Biener <rguenther@suse.de>
16789 PR tree-optimization/80591
16791 2017-04-10 Richard Biener <rguenther@suse.de>
16793 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16796 2017-05-02 Richard Biener <rguenther@suse.de>
16798 PR tree-optimization/80549
16799 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16800 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16801 headers if they do not have a preheader.
16803 2017-05-02 Martin Liska <mliska@suse.cz>
16806 * common.opt: Fix typo.
16807 * doc/invoke.texi: Likewise.
16809 2017-05-01 Jan Beulich <jbeulich@suse.com>
16811 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16812 swapping, add (x,x,m,x,n) alternative.
16814 2017-05-01 Nathan Sidwell <nathan@acm.org>
16816 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16817 unnecessary unadjusted_alignment check.
16819 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16822 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16824 * gimplify.c (gimplify_cilk_detach): New function.
16825 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16826 * tree-core.h: Document EXPR_CILK_SPAWN.
16827 * tree.h (EXPR_CILK_SPAWN): Define.
16829 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16831 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16832 to use new fixit_hint representation, using the "replace" logic.
16833 (get_line_span_for_fixit_hint): Likewise.
16834 (layout::print_any_fixits): Likewise.
16835 (selftest::test_one_liner_many_fixits): Rename to...
16836 (selftest::test_one_liner_many_fixits_1): ...this, and update
16837 comment and expected output to reflect that the multiple fix-it
16838 hints are now consolidated into one insertion.
16839 (selftest::test_one_liner_many_fixits_2): New test.
16840 (selftest::test_diagnostic_show_locus_one_liner): Update for
16842 (selftest::test_fixit_consolidation): Update for fix-it API
16844 * diagnostic.c (print_parseable_fixits): Likewise.
16845 * edit-context.c (edited_line::m_line_events): Convert from
16846 auto_vec <line_event *> to auto_vec <line_event>.
16847 (class line_event): Convert from abstract base class to a concrete
16848 class, taking over the role of replace_event.
16849 (class insert_event): Delete.
16850 (class replace_event): Rename to class line_event. Convert to
16852 (edit_context::add_fixits): Reimplement.
16853 (edit_context::apply_insert): Delete.
16854 (edit_context::apply_replace): Rename to...
16855 (edit_context::apply_fixit): ...this. Convert to half-open range.
16856 (edited_file::apply_insert): Delete.
16857 (edited_file::apply_replace): Rename to...
16858 (edited_file::apply_fixit): ...this.
16859 (edited_line::~edited_line): Drop deletion of events.
16860 (edited_line::apply_insert): Delete.
16861 (edited_line::apply_replace): Rename to...
16862 (edited_line::apply_fixit): ...this. Convert to half-open range.
16863 Update for change to type of m_line_events.
16864 * edit-context.h (edit_context::apply_insert): Delete.
16865 (edit_context::apply_replace): Rename to...
16866 (edit_context::apply_fixit): ...this.
16868 2017-05-01 Martin Sebor <msebor@redhat.com>
16870 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16873 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16876 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16877 __get_cpuid_max returns 0.
16878 (__get_cpuid_count): Ditto.
16880 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16882 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16883 replacement expression is another instance of one of its arguments.
16885 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16888 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16889 check for stack push/pop autoinc.
16890 * config/i386/i386.c (ix86_agi_dependent): Return false
16891 if the only reason why modified_in_p returned true is that
16892 addr is SP based and set_insn is a push or pop.
16894 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16896 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16899 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16902 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16903 (account_size_time): Use two predicates - exec_pred and
16905 (evaluate_conditions_for_known_args): Compute both clause and
16907 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16908 (inline_summary_t::duplicate): Update.
16909 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16911 (compute_inline_parameters): Likewise.
16912 (estimate_edge_size_and_time): Update caluclation of time.
16913 (estimate_node_size_and_time): Compute both time and nonspecialized
16915 (estimate_ipcp_clone_size_and_time): Update.
16916 (inline_merge_summary): Update.
16917 (do_estimate_edge_time): Update.
16918 (do_estimate_edge_size): Update.
16919 (do_estimate_edge_hints): Update.
16920 (inline_read_section, inline_write_summary): Stream both new predicates.
16921 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16923 (compute_inlined_call_time): Cleanup.
16924 (big_speedup_p): Update.
16925 (edge_badness): Update.
16926 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16927 (size_time_entry): Replace predicate by exec_predicate and
16928 nonconst_predicate.
16929 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16930 (estimate_edge_time): Return also nonspec_time.
16931 (reset_edge_growth_cache): Update.
16933 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16935 PR rtl-optimization/80491
16936 * ifcvt.c (noce_process_if_block): When looking for x setter
16937 with missing else_bb, don't check only the insn right before
16938 cond_earliest, but look for the last insn that x is modified in
16939 within the same bb.
16941 PR rtl-optimization/80491
16942 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16944 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16946 PR tree-optimization/80487
16947 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16949 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16951 PR tree-optimization/79697
16952 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16953 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16954 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16956 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16957 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16959 2017-04-28 Martin Sebor <msebor@redhat.com>
16961 PR tree-optimization/80523
16962 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16963 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16965 (maybe_warn, format_directive, parse_directive): Use new functions.
16966 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16968 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16970 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16972 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16974 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16975 target_header_dir): Set correctly.
16976 * configure: Regenerated.
16977 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16978 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16979 instead of SYSTEM_HEADER_DIR.
16981 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16983 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16984 (estimate_local_effects): Likewise.
16985 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16986 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16987 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16988 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16989 do_estimate_edge_time, estimate_edge_time): Likewise.
16990 * ipa-inline-analysis.c (estimate_node_size_and_time,
16991 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16992 (estimate_time_after_inlining): Remove.
16994 2017-04-28 Martin Liska <mliska@suse.cz>
16996 * doc/gcov.texi: Enhance documentation of gcov.
16998 2017-04-28 Martin Liska <mliska@suse.cz>
17000 * doc/gcov.texi: Sort options in alphabetic order.
17001 * doc/gcov-dump.texi: Likewise.
17002 * doc/gcov-tool.texi: Likewise.
17003 * gcov.c (print_usage): Likewise.
17004 * gcov-dump.c (print_usage): Likewise.
17005 * gcov-tool.c (print_merge_usage_message): Likewise.
17006 (print_rewrite_usage_message): Likewise.
17007 (print_overlap_usage_message): Likewise.
17009 2017-04-28 Martin Liska <mliska@suse.cz>
17011 PR gcov-profile/53915
17012 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17014 2017-04-28 Martin Liska <mliska@suse.cz>
17016 PR gcov-profile/79891
17017 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17018 is marked by compiler as living on a line.
17019 (get_cycles_count): Remove usage of the union.
17020 (output_intermediate_file): Likewise.
17021 (find_source): Fix GNU coding style.
17022 (accumulate_line_counts): Remove old non-all block mode.
17023 (output_lines): Remove usage of the union.
17024 * profile.c (output_location): Include all BBs, even if
17025 belonging to a same line (and file) as a previous BB.
17027 2017-04-28 Martin Liska <mliska@suse.cz>
17029 * gcov.c (process_args): Handle new argument 'w'.
17030 (read_graph_file): Assign ID to BBs.
17031 (output_branch_count): Display BB # if verbose flag is set.
17032 (output_lines): Likewise for arcs.
17033 (print_usage): Add '--verbose' option help.
17034 * doc/gcov.texi: Document --verbose (-w) option.
17036 2017-04-28 Martin Liska <mliska@suse.cz>
17038 * gcov.c (struct block_location_info): New struct.
17039 (process_file): Fill up the new structure.
17040 (read_graph_file): Replace usage of encoding by the newly added
17042 (add_line_counts): Likewise.
17043 (accumulate_line_counts): Remove usage of the union.
17044 (function_info::function_info): New function.
17045 (function_info::~function_info): Likewise.
17046 (process_file): Call delete instead of release_function.
17047 (release_function): Release the function.
17048 (release_structures): Call delete instead of release_function.
17049 (solve_flow_graph): Replace usage of num_blocks.
17050 (find_exception_blocks): Likewise.
17051 (output_lines): Fix GNU coding style.
17053 2017-04-28 Martin Liska <mliska@suse.cz>
17056 * coverage.c (coverage_remove_note_file): New function.
17057 * coverage.h: Declare the function.
17058 * toplev.c (finalize): Clean if an error has been seen.
17060 2017-04-28 Martin Liska <mliska@suse.cz>
17062 PR gcov-profile/80031
17063 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17064 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17065 * gcov.c (read_graph_file): Read just number of blocks.
17066 * profile.c (branch_prob): Do not stream 0 flags per a basic
17069 2017-04-28 Martin Liska <mliska@suse.cz>
17071 * gcov-dump.c (tag_*): Add new argument to declarations.
17072 (dump_gcov_file): Likewise.
17073 (tag_blocks): Add and use new argument depth.
17074 (tag_arcs): Likewise.
17075 (tag_lines): Likewise.
17076 (tag_counters): Likewise.
17077 (tag_summary): Likewise.
17078 (dump_working_sets): Use depth to do a proper indentation.
17080 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17083 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17084 * symtab.c (symtab_node::debug_symtab): Move definition here.
17086 2017-04-28 Richard Biener <rguenther@suse.de>
17088 * lto-streamer.h (LTO_major_version): Bump to 7.
17090 2017-04-28 Richard Biener <rguenther@suse.de>
17092 * tree-vrp.c (assert_info): New struct.
17093 (add_assert_info): New helper.
17094 (register_edge_assert_for_2): Refactor to add asserts to a vector
17096 (register_edge_assert_for_1): Likewise.
17097 (register_edge_assert_for): Likewise.
17098 (finish_register_edge_assert_for): New helper actually registering
17099 asserts where live on edge.
17100 (find_conditional_asserts): Adjust.
17101 (find_switch_asserts): Likewise.
17102 (evrp_dom_walker::try_find_new_range): Generalize.
17103 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17105 2017-04-27 Marek Polacek <polacek@redhat.com>
17108 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17109 arg10 and arg11 to itype.
17111 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17113 * doc/extend.texi (Object Size Checking): Improve grammar.
17115 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17118 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17119 that the logic for permitting reciprocal estimates matches that
17122 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17125 * tree.c (type_cache_hasher::equal): Only compare
17126 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17127 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17128 non-aggregate element types.
17129 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17130 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17132 2017-04-27 Richard Biener <rguenther@suse.de>
17134 PR middle-end/80533
17135 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17136 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17137 keeping a reference to a trailing array.
17139 2017-04-27 Richard Biener <rguenther@suse.de>
17141 PR middle-end/80539
17142 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17143 being in loop-closed SSA form conservatively.
17144 (chrec_fold_multiply_poly_poly): Likewise.
17146 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17148 PR middle-end/79665
17149 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17150 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17152 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17155 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17156 (aarch64_function_arg_alignment): Return unsigned int again, but still
17157 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17158 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17159 Don't emit -Wpsabi note.
17160 (aarch64_function_arg_boundary): Likewise.
17161 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17164 2017-04-26 Nathan Sidwell <nathan@acm.org>
17166 * tree.h (crc32_unsigned_n): Declare.
17167 (crc32_unsigned, crc32_unsigned): Make inline.
17168 * tree.c (crc32_unsigned_bits): Replace with ...
17169 (crc32_unsigned_n): ... this.
17170 (crc32_unsigned, crc32_byte): Remove.
17171 (crc32_string): Remove unnecessary braces.
17173 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17175 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17176 * ipa-inline-analysis.c (MAX_TIME): Remove.
17177 (account_size_time): Use sreal for time.
17178 (dump_inline_summary): Update.
17179 (estimate_function_body_sizes): Update.
17180 (estimate_edge_size_and_time): Update.
17181 (estimate_calls_size_and_time): Update.
17182 (estimate_node_size_and_time): Update.
17183 (inline_merge_summary): Update.
17184 (inline_update_overall_summary): Update.
17185 (estimate_time_after_inlining): Update.
17186 (inline_read_section): Update.
17187 (inline_write_summary): Update.
17188 * ipa-inline.c (compute_uninlined_call_time): Update.
17189 (compute_inlined_call_time): Update.
17190 (recursive_inlining): Update.
17191 (inline_small_functions): Update.
17192 (dump_overall_stats): Update.
17193 * ipa-inline.h: Include sreal.h.
17194 (size_time_entry): Turn time to sreal.
17195 (inline_summary): Turn self_time nad time to sreal.
17197 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17199 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17201 (sreal::stream_out, sreal::stream_in): New.
17202 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17204 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17206 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17209 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17212 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17214 (dimode_scalar_chain::compute_convert_gain): Ditto.
17215 (dimode_scalar_chain::make_vector_copies): Ditto.
17216 (dimode_scalar_chain::convert_reg): Ditto.
17217 (dimode_scalar_chain::convert_insn): Ditto.
17218 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17219 (VI248_AVX512BW_1): New mode iterator.
17220 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17221 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17224 2017-04-25 Martin Sebor <msebor@redhat.com>
17226 PR tree-optimization/80497
17227 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17228 constants are representable in HOST_WIDE_INT.
17229 (parse_directive): Ditto.
17231 2017-04-25 Martin Sebor <msebor@redhat.com>
17234 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17235 (new_zero_array): Adjust signature.
17236 (dom_info::dom_init): Used unsigned rather that size_t.
17237 (dom_info::dom_info): Same.
17239 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17240 Jakub Jelinek <jakub@redhat.com>
17243 * config/arm/arm.c: Include gimple.h.
17244 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17245 returns negative, increment ncrn only if it returned positive.
17246 (arm_needs_doubleword_align): Return int instead of bool,
17247 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17248 members, but if there is any such non-FIELD_DECL
17249 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17250 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17251 returns negative, increment nregs only if it returned positive.
17252 (arm_setup_incoming_varargs): Likewise.
17253 (arm_function_arg_boundary): Emit -Wpsabi note if
17254 arm_needs_doubleword_align returns negative, return
17255 DOUBLEWORD_ALIGNMENT only if it returned positive.
17257 2017-04-25 Marek Polacek <polacek@redhat.com>
17260 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17261 first argument to type.
17263 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17266 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17267 type checks to test for compatibility instead of equality.
17269 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17270 Jakub Jelinek <jakub@redhat.com>
17273 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17275 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17276 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17277 the alignment computation, but return their maximum in warn_alignment.
17278 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17279 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17281 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17282 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17285 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17287 * config/arc/simdext.md (dmpyh): Fix typo.
17289 2017-04-25 Richard Biener <rguenther@suse.de>
17291 PR tree-optimization/80492
17292 * alias.c (compare_base_decls): Handle registers with asm
17293 specification conservatively.
17294 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17295 compare_base_decls returning dont-know properly.
17297 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17299 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17300 (legitimate_offset_address_p): New function.
17301 (arc_legitimate_address_p): Use above function.
17303 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17305 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17307 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17309 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17310 ACCH registers whenever they are available.
17312 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17314 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17315 double regs fix when not used.
17317 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17319 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17321 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17322 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17324 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17326 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17327 when emitting short ADD instructions.
17329 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17331 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17333 (cmpsi_cc_c_insn): Likewise.
17334 (cbranchsi4_scratch): Compute proper instruction length using
17335 compact_hreg_operand.
17336 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17338 2017-04-25 Richard Biener <rguenther@suse.de>
17340 PR middle-end/80509
17341 * passes.c (pass_manager::pass_manager): Initialize
17342 m_name_to_pass_map.
17344 2017-04-25 Richard Biener <rguenther@suse.de>
17346 PR tree-optimization/79201
17347 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17349 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17352 * config/s390/vector.md: Split MEM->GPR vector moves for
17353 non-s_operand addresses.
17355 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17358 * config/s390/predicates.md (reload_const_wide_int_operand): New
17360 * config/s390/s390.md ("movti"): Remove d/P alternative.
17361 ("movti_bigconst"): New pattern definition.
17363 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17366 * s390-protos.h (s390_expand_cs_hqi): Removed.
17367 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17368 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17369 modes as well as CCZ1mode and CCZmode.
17370 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17371 signature of s390_emit_compare_and_swap.
17372 (s390_expand_cs_hqi): Likewise, make static.
17373 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17374 compare-and-swap, in some cases.
17375 (s390_expand_cs): Wrapper function.
17376 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17378 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17379 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17380 patterns for small and large integers. Forbid symref memory operands.
17381 Move expander to s390.c. Require cc register.
17382 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17383 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17384 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17385 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17386 symref memory operands. Remove CC mode and call s390_match_ccmode
17388 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17390 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17392 * config/s390/s390.md (define_peephole2): New peephole to help
17393 combining the load-and-test pattern with volatile memory.
17395 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17397 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17398 with CCZmode for TARGET_Z196.
17400 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17402 PR rtl-optimization/80501
17403 * combine.c (make_compound_operation_int): Set subreg_code to SET
17404 even for AND with mask of the sign bit of mode.
17406 PR rtl-optimization/80500
17407 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17408 sum's initial value.
17410 2017-04-25 Julian Brown <julian@codesourcery.com>
17411 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17413 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17415 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17417 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17419 2017-04-25 Julian Brown <julian@codesourcery.com>
17420 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17422 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17423 (thunderx2t99_sha): New Reservation.
17425 2017-04-25 Julian Brown <julian@codesourcery.com>
17426 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17428 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17429 type for 1-element load.
17431 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17433 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17435 2017-04-24 Martin Jambor <mjambor@suse.cz>
17437 PR tree-optimization/80293
17438 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17439 char arrays not totally scalarizable if it is false.
17440 (analyze_all_variable_accesses): Pass correct value in the new
17441 parameter. Add a statistics counter.
17443 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17445 PR middle-end/79931
17446 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17448 2017-04-24 Richard Biener <rguenther@suse.de>
17450 PR tree-optimization/80494
17451 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17452 out for complex types.
17454 2017-04-24 Richard Biener <rguenther@suse.de>
17456 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17457 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17458 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17459 (DFS): Adjust and never fail.
17460 (sccvn_dom_walker::fail): Remove.
17461 (sccvn_dom_walker::before_dom_children): Adjust.
17462 (run_scc_vn): Likewise and never fail.
17463 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17464 (pass_fre::execute): Likewise.
17466 2017-04-24 Richard Biener <rguenther@suse.de>
17468 PR tree-optimization/79725
17469 * tree-ssa-sink.c (statement_sink_location): Return whether
17470 failure reason was zero uses. Move that check later.
17471 (sink_code_in_bb): Deal with zero uses by removing the stmt
17474 2017-04-24 Richard Biener <rguenther@suse.de>
17477 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17478 pointer-based references.
17480 2017-04-24 Richard Biener <rguenther@suse.de>
17483 * pass_manager.h (pass_manager::operator new): Remove.
17484 (pass_manager::operator delete): Likewise.
17485 * passes.c (pass_manager::operator new): Remove.
17486 (pass_manager::operator delete): Likewise.
17487 (pass_manager::pass_manager): Zero individual pass members.
17489 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17492 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17493 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17494 Check "XEXP (src, 1)" operand here.
17495 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17496 Check "XEXP (src, 1)" operand here.
17497 (dimode_scalar_chain::make_vector_copies): Detect count register
17498 of a shift instruction. Zero extend count register from QImode
17499 to DImode to satisfy vector shift pattern count operand predicate.
17500 Substitute vector shift count operand with a DImode copy.
17501 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17504 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17506 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17507 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17508 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17509 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17510 (UNSPEC_NOREX_MEM): Remove definition.
17512 2017-04-21 Richard Biener <rguenther@suse.de>
17514 PR tree-optimization/79547
17515 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17516 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17517 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17518 without any constraints.
17520 2017-04-21 Richard Biener <rguenther@suse.de>
17522 PR tree-optimization/78847
17523 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17525 2017-04-21 Richard Biener <rguenther@suse.de>
17527 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17528 (build_distinct_type_copy): Likewise.
17529 (build_variant_type_copy): Likewise.
17530 * tree.c (build_qualified_type): Pass down mem-stat info.
17531 (build_distinct_type_copy): Likewise.
17532 (build_variant_type_copy): Likewise.
17534 2017-04-21 Richard Biener <rguenther@suse.de>
17536 PR tree-optimization/80237
17537 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17539 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17540 for a simplified result.
17542 2016-04-21 Richard Biener <rguenther@suse.de>
17544 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17545 sth as strict as a simple_iv but a chrec without symbols and an
17546 operand defined in the loop we are peeling (and not some subloop).
17547 (propagate_constants_for_unrolling): Propagate all constants.
17549 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17552 * config/i386/i386.c (print_reg): Remove assert for disalowed
17553 regno values, call output_operand_lossage instead.
17555 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17558 * config/i386/constraints.md (Yc): New register constraint.
17559 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17560 Use Yc constraint for alternative 2 of operand 0. Remove
17561 preferred_for_speed attribute.
17563 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17565 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17566 lastprivate clauses in SIMT case.
17568 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17570 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17572 2017-04-20 Richard Biener <rguenther@suse.de>
17574 PR tree-optimization/57796
17575 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17576 as N scalar stores.
17577 (vect_model_load_cost): Cost gathers as N scalar loads.
17579 2017-04-20 Richard Biener <rguenther@suse.de>
17581 * ggc-page.c (ggc_allocated_p): Rename to ...
17582 (safe_lookup_page_table_entry): ... this and return the lookup
17584 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17586 2017-04-20 Richard Biener <rguenther@suse.de>
17588 PR tree-optimization/80453
17589 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17590 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17591 from the conditions.
17592 (vn_phi_eq): Pass them down.
17593 (vn_phi_lookup): Record them.
17594 (vn_phi_insert): Likewise.
17596 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17598 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17599 uninitialized variable warning to avoid buffer overrun.
17601 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17604 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17605 is suppressed for '{ 0 }' in C.
17607 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17609 * BASE-VER: Set to 8.0.0.
17611 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17613 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17614 priority .init_array and .fini_array section with SECTION_NOTYPE
17617 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17619 PR middle-end/80423
17620 * tree.h (build_array_type): Add typeless_storage default argument.
17621 * tree.c (type_cache_hasher::equal): Also compare
17622 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17623 (build_array_type): Add typeless_storage argument, set
17624 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17626 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17627 (build_array_type): Likewise. Add typeless_storage argument.
17629 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17630 Jakub Jelinek <jakub@redhat.com>
17632 PR tree-optimization/80426
17633 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17634 operation on symbolic operands, also compute the overflow for the
17635 invariant part when the operation degenerates into a negation.
17637 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17640 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17641 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17644 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17646 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17649 * config/avr/avr.c (tree.h): Include it.
17650 (cgraph.h): Include it.
17651 (avr_encode_section_info): Don't warn for uninitialized progmem
17652 variable if it's just an alias.
17654 2017-04-19 Richard Biener <rguenther@suse.de>
17657 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17658 when needed by AutoPGO.
17660 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17663 * doc/lto.texi: Remove an extra 'that'.
17665 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17667 PR rtl-optimization/80429
17668 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17669 are only used in debug insns.
17671 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17672 Vladimir Makarov <vmakarov@redhat.com>
17674 * config/sparc/predicates.md (input_operand): Add comment. Return
17675 true for any memory operand when LRA is in progress.
17676 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17678 2017-04-18 Jeff Law <law@redhat.com>
17681 * mips.md ({return,simple_return}_internal): Do not overwrite
17684 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17686 PR tree-optimization/80443
17687 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17688 instead of adding 1, subtract -1 and similarly instead of subtracting
17691 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17693 PR rtl-optimization/80357
17694 * haifa-sched.c (tmp_bitmap): New variable.
17695 (model_recompute): Handle duplicate use records.
17696 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17697 (free_global_sched_pressure_data): Free it.
17699 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17702 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17703 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17704 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17705 instead of SYSTEM_HEADER_DIR.
17707 2017-04-18 Jeff Law <law@redhat.com>
17709 PR middle-end/80422
17710 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17711 predecessors after walking up the insn chain.
17713 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17716 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17717 sizetype type into debug info.
17719 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17722 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17723 unneeded test for TARGET_UPPER_REGS_SF.
17724 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17726 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17729 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17730 instead of gsi_after_labels.
17732 2017-04-18 Jeff Law <law@redhat.com>
17734 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17738 2017-04-13 Jeff Law <law@redhat.com>
17739 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17740 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17742 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17745 * config/avr/avr.c (intl.h): Include it.
17746 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17748 2017-04-18 Martin Liska <mliska@suse.cz>
17750 PR gcov-profile/78783
17751 * gcov-tool.c (gcov_output_files): Validate that destination
17752 file is either removed by the tool or by a user.
17754 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17755 Guy Benyei <guybe@mellanox.com>
17757 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17758 block, and do not negate it, the stored id is already negative.
17760 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17762 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17764 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17767 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17768 masks of options that should be turned off if the VSX vector
17769 options are turned off.
17770 (OTHER_P8_VECTOR_MASKS): Likewise.
17771 (OTHER_VSX_VECTOR_MASKS): Likewise.
17772 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17773 rs6000_disable_incompatible_switches to validate no type switches
17775 (rs6000_incompatible_switch): New function to disallow turning on
17776 other vector options if -mno-vsx, -mno-power8-vector, or
17777 -mno-power9-vector are specified.
17779 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17781 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17783 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17785 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17786 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17787 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17788 (ARG_POINTER_CFA_OFFSET): Likewise.
17790 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17792 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17793 conditions to take advantage of various optimizations.
17795 2017-04-13 Jeff Law <law@redhat.com>
17797 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17798 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17799 (zero_extendsidi2_dext): Likewise.
17801 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17804 * fold-const.c (fold_ternary_loc): Revert
17805 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17808 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17810 PR rtl-optimization/80343
17811 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17812 new scratch pseudo.
17814 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17817 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17818 to ubsan_encode_value.
17820 2017-04-13 Jeff Law <law@redhat.com>
17822 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17823 appearing in DEBUG_INSNs.
17825 2017-04-13 Martin Liska <mliska@suse.cz>
17827 PR gcov-profile/80413
17828 * gcov-io.c (gcov_write_string): Copy to buffer just when
17829 allocated size is greater than zero.
17831 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17834 * dwarf2out.c (decls_for_scope): Ignore declarations of
17835 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17837 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17840 * ipa-visibility.c (non_local_p): Fix typos.
17841 (localize_node): When localizing symbol in same comdat group,
17842 dissolve the group only when we know external symbols are going
17844 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17846 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17848 PR tree-optimization/79390
17849 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17850 order does not result in usable sequence, retry with reversed operand
17856 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17857 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17858 op0 instead of fold_convert_loc (loc, type, arg0).
17860 2017-04-12 Jeff Law <law@redhat.com>
17862 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17863 has a delay slot in the generated code.
17865 * config/cris/cris.md (cris_preferred_reload_class): Return
17866 GENNONACR_REGS rather than GENERAL_REGS.
17868 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17871 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17872 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17873 signedness of the result type.
17875 2017-04-12 Richard Biener <rguenther@suse.de>
17876 Jeff Law <law@redhat.com>
17878 PR tree-optimization/80359
17879 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17880 trim stores to TARGET_MEM_REFs.
17882 2017-04-12 Richard Biener <rguenther@suse.de>
17884 PR tree-optimization/79390
17885 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17886 threading case even more.
17888 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17891 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17892 for quad_address_p for TImode, instead of just not indexed_address.
17894 2017-04-12 Richard Biener <rguenther@suse.de>
17895 Bernd Edlinger <bernd.edlinger@hotmail.de>
17897 PR middle-end/79671
17898 * alias.c (component_uses_parent_alias_set_from): Handle
17899 TYPE_TYPELESS_STORAGE.
17900 (get_alias_set): Likewise.
17901 * tree-core.h (tree_type_common): Add typeless_storage flag.
17902 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17903 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17904 for types containing members with TYPE_TYPELESS_STORAGE.
17905 (place_field): Likewise.
17906 (layout_type): Likewise for ARRAY_TYPE.
17907 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17908 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17909 TYPE_TYPELESS_STORAGE.
17910 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17912 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17915 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17916 first argument to type.
17918 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17922 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17923 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17924 (rs6000_expand_binop_builtin): Likewise.
17925 (rs6000_expand_ternop_builtin): Likewise; also add missing
17926 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17927 vshasigma built-ins.
17928 * doc/extend.texi: Document that vec_xxpermdi's third argument
17929 must be a constant.
17931 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17933 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17934 Use shift_const cost parameter when calculating gain of STV shifts.
17936 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17938 PR rtl-optimization/70478
17939 * lra-constraints.c (process_alt_operands): Check memory for
17940 disfavoring memory insn operand.
17942 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17944 PR middle-end/80100
17945 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17946 left shift in unsigned HOST_WIDE_INT type.
17948 PR rtl-optimization/80385
17949 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17950 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17953 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17954 if they have any depend clauses.
17956 2017-04-11 Martin Liska <mliska@suse.cz>
17959 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17960 * ipa-split.c (split_function): Create a local comdat symbol
17961 if caller is in a comdat group.
17963 2017-04-11 Martin Liska <mliska@suse.cz>
17966 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17969 2017-04-11 Martin Sebor <msebor@redhat.com>
17971 PR middle-end/80364
17972 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17973 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17975 (directive::set_width, directive::set_precision, format_character):
17977 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17980 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17983 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17984 conflict, set target->arch_name instead of target->cpu_name.
17986 2017-04-11 Richard Biener <rguenther@suse.de>
17988 PR tree-optimization/80374
17989 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17990 build_zero_cst, remove fold_convertible_p check again.
17992 2017-04-11 Martin Liska <mliska@suse.cz>
17995 * ubsan.c (instrument_object_size): Do not instrument register
17998 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18001 * config/i386/i386-builtin-types.def
18002 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18003 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18004 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18005 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18006 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18007 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18008 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18009 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18010 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18011 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18012 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18013 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18014 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18015 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18016 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18017 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18018 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18019 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18020 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18021 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18022 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18023 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18024 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18025 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18026 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18027 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18028 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18029 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18030 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18031 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18032 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18033 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18034 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18035 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18036 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18037 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18038 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18039 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18040 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18041 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18042 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18043 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18044 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18045 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18046 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18047 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18049 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18050 flag to second_arg_count, handle 4 argument function type _COUNT
18051 aliases, handle second_arg_count on second argument rather than last.
18053 2017-04-10 Jeff Law <law@redhat.com>
18055 PR tree-optimization/80374
18056 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18057 record anything if we can not convert integer_zero_node to the
18060 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18063 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18064 Enhance special handling given to the TARGET_P9_MINMAX option in
18065 relation to certain other options.
18067 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18069 PR tree-optimization/80153
18070 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18071 remove POINTER_PLUS_EXPR's base part directly, rather than through
18074 2017-04-10 Richard Biener <rguenther@suse.de>
18075 Bin Cheng <bin.cheng@arm.com>
18077 PR tree-optimization/80153
18078 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18079 the first element of pointer type aff_tree. Build result expr in
18081 (add_elt_to_tree): Convert to type unconditionally. Remove other
18082 fold_convert calls.
18083 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18084 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18086 2017-04-10 Richard Biener <rguenther@suse.de>
18088 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18091 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18093 PR rtl-optimization/70478
18094 * lra-constraints.c (curr_small_class_check): New.
18095 (update_and_check_small_class_inputs): New.
18096 (process_alt_operands): Update curr_small_class_check. Disfavor
18097 alternative insn memory operands. Check available regs for small
18100 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18103 * config/mips/mips.opt (-mvirt): Update description.
18104 * doc/invoke.texi (-mvirt): Likewise.
18106 2017-04-10 Richard Biener <rguenther@suse.de>
18108 PR middle-end/80362
18109 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18110 looking for NEGATE_EXPR in -A / -B to A / B folding.
18112 2017-04-10 Martin Liska <mliska@suse.cz>
18114 PR gcov-profile/80224
18115 * gcov.c (print_usage): Fix usage string.
18116 (get_gcov_intermediate_filename): Remove.
18117 (output_gcov_file): Use both for normal and intermediate format.
18118 (generate_results): Do not initialize special file for
18119 intermediate format.
18121 2017-04-10 Richard Biener <rguenther@suse.de>
18123 PR tree-optimization/80304
18124 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18127 2017-04-10 Nathan Sidwell <nathan@acm.org>
18130 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18131 (rs6000_init_builtins): Use it.
18133 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18135 * config/arm/arm.md (<mrc>): Add mode to SET source.
18136 (<mrrc>): Likewise.
18138 2017-04-10 Richard Biener <rguenther@suse.de>
18140 PR middle-end/80344
18141 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18143 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18146 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18147 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18148 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18149 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18150 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18151 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18152 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18153 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18154 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18155 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18156 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18157 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18158 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18159 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18160 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18161 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18162 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18163 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18164 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18165 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18166 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18167 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18168 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18170 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18172 PR rtl-optimization/70478
18173 * lra-constraints.c: Reverse the last patch.
18175 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18177 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18178 Add comment for WCHAR_T.
18180 2017-04-08 Martin Liska <mliska@suse.cz>
18183 2017-04-07 Martin Liska <mliska@suse.cz>
18186 * ipa-split.c (split_function): Add function part to a same comdat
18189 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18192 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18194 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18196 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18197 (vec_pairing): ...to this.
18198 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18199 (rs6000_sched_init): Adjust for name change.
18200 (struct rs6000_sched_context): Likewise.
18201 (rs6000_init_sched_context): Likewise.
18202 (rs6000_set_sched_context): Likewise.
18204 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18210 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18212 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18213 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18214 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18216 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18218 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18220 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18222 PR rtl-optimization/70703
18223 * ira-color.c (update_conflict_hard_regno_costs): Use
18224 int64_t instead of HOST_WIDE_INT.
18226 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18228 PR rtl-optimization/70478
18229 * lra-constraints.c (process_alt_operands): Disfavor alternative
18230 insn memory operands.
18232 2017-04-07 Jeff Law <law@redhat.com>
18234 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18235 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18237 2017-04-07 Martin Liska <mliska@suse.cz>
18240 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18241 Show error message instead of an ICE.
18243 2017-04-07 Martin Liska <mliska@suse.cz>
18246 * ipa-split.c (split_function): Add function part to a same comdat
18249 2017-04-07 Richard Biener <rguenther@suse.de>
18251 PR middle-end/80341
18252 * tree.c (get_unwidened): Also handle ! for_type case for
18254 * convert.c (do_narrow): Split out from ...
18255 (convert_to_integer_1): ... here. Do not pass final truncation
18256 type to get_unwidened for TRUNC_DIV_EXPR.
18258 2017-04-07 Richard Biener <rguenther@suse.de>
18260 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18262 (aff_combination_const): Adjust.
18263 (aff_combination_scale): Likewise.
18264 (aff_combination_add_elt): Likewise.
18265 (aff_combination_add_cst): Likewise.
18266 (aff_combination_convert): Likewise.
18267 (add_elt_to_tree): Likewise. Remove unused argument.
18268 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18270 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18272 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18274 * config/arm/arm.c (arm_default_short_enums): Use
18275 ARM_DEFAULT_SHORT_ENUMS.
18276 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18278 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18281 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18282 members with redundant out-of-class redeclaration.
18284 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18287 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18288 * config/i386/i386.md (*zero_extendsidi2):
18289 Add (?*x,*x) and (?*v,*v) alternatives.
18291 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18294 * config/i386/i386.c (ix86_expand_builtin)
18295 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18296 mode from insn data. Convert operands to insn operand mode.
18297 Copy operands that don't satisfy insn predicate to a register.
18299 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18301 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18304 2017-04-06 Richard Biener <rguenther@suse.de>
18306 PR tree-optimization/80334
18307 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18308 preserve alignment of accesses.
18310 2017-04-06 Richard Biener <rguenther@suse.de>
18312 PR tree-optimization/80262
18313 * tree-sra.c (build_ref_for_offset): Preserve address-space
18315 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18316 Drop useless address-space information on MEM_REF offsets.
18318 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18320 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18322 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18324 PR rtl-optimization/70703
18325 * ira-color.c (update_conflict_hard_regno_costs): Use
18326 HOST_WIDE_INT instead of long.
18328 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18331 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18332 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18334 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18335 for x86_64 target. Handle -m3dnowa option.
18337 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18339 PR rtl-optimization/70703
18340 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18341 (update_conflict_hard_regno_costs): Use long instead of unsigned
18342 arithmetic for cost calculation.
18344 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18345 Bernd Edlinger <bernd.edlinger@hotmail.de>
18348 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18351 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18354 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18355 ptr_mode with Pmode throughout.
18356 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18357 into probe_stack_range and use DImode.
18359 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18362 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18363 call_eh_return is true.
18365 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18367 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18368 Initialize last_match_fntype_index.
18370 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18373 * tree-nvr.c: Include internal-fn.h.
18374 (pass_return_slot::execute): Ignore internal calls without
18377 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18378 Richard Biener <rguenther@suse.de>
18381 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18382 captures used multiple times, except for the last use.
18383 * generic-match-head.c: Include gimplify.h.
18385 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18387 PR tree-optimization/79390
18388 * target.h (struct noce_if_info): Declare.
18389 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18390 * target.def (noce_conversion_profitable_p): New target hook.
18391 * ifcvt.h (struct noce_if_info): New type, moved from ...
18392 * ifcvt.c (struct noce_if_info): ... here.
18393 (noce_conversion_profitable_p): Renamed to ...
18394 (default_noce_conversion_profitable_p): ... this. No longer
18396 (noce_try_store_flag_constants, noce_try_addcc,
18397 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18398 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18399 instead of noce_conversion_profitable_p.
18400 * config/i386/i386.c: Include ifcvt.h.
18401 (ix86_option_override_internal): Don't override
18402 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18403 (ix86_noce_conversion_profitable_p): New function.
18404 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18405 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18406 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18407 * doc/tm.texi: Regenerated.
18409 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18411 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18414 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18417 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18418 instructions for small multiply cores.
18420 2017-04-04 Jeff Law <law@redhat.com>
18422 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18424 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18425 that are not set by the loop over the elements.
18427 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18430 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18431 int mode, convert_modes it to mode as unsigned, otherwise use
18432 lowpart_subreg to mode rather than SImode.
18433 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18434 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18435 Use DImode instead of SImode for the shift count operand.
18436 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18439 2017-04-04 Richard Biener <rguenther@suse.de>
18441 PR middle-end/80281
18442 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18443 arithmetic done for the negate or the plus. Simplify.
18444 (A - (-B) -> A + B): Likewise.
18445 * fold-const.c (split_tree): Make sure to not negate pointers.
18447 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18449 PR rtl-optimization/60818
18450 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18451 a compare of comparisons with the thing compared if this results
18452 in a different machine mode.
18454 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18456 * alias.c (base_alias_check): Fix typo in comment.
18457 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18458 * cgraphunit.c (symbol_table::compile): Likewise.
18459 * collect2.c (maybe_run_lto_and_relink): Likewise.
18460 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18461 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18462 * config/avr/avr.c (avr_map_op_t): Likewise.
18463 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18464 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18465 * config/epiphany/epiphany.md (movcc): Likewise.
18466 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18467 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18469 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18470 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18471 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18472 * config/sh/sh.c (sh_rtx_costs): Likewise.
18473 * fold-const.c (fold_truth_andor): Likewise.
18474 * genautomata.c (collapse_flag): Likewise.
18475 * gengtype.h (struct type::u::s): Likewise.
18476 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18477 * input.c (FORMAT_AMOUNT): Likewise.
18478 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18479 (known_aggs_to_agg_replacement_list): Likewise.
18480 * ipa-inline-analysis.c: Likewise.
18481 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18482 * ipa-polymorphic-call.c
18483 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18484 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18485 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18487 * modulo-sched.c (apply_reg_moves): Likewise.
18488 * omp-expand.c (build_omp_regions_1): Likewise.
18489 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18490 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18491 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18492 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18493 * value-prof.c: Likewise.
18494 * var-tracking.c (val_reset): Likewise.
18496 2017-04-03 Richard Biener <rguenther@suse.de>
18498 PR tree-optimization/80275
18499 * fold-const.c (split_address_to_core_and_offset): Handle
18502 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18504 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18505 descriptors is at least equal to that of functions.
18507 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18509 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18511 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18514 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18515 (mov<IMOD4:mode>): New expander.
18516 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18518 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18520 PR rtl-optimization/79405
18521 * fwprop.c (propagations_left): New variable.
18522 (forward_propagate_into): Decrement it.
18523 (fwprop_init): Initialize it.
18524 (fw_prop): If the variable has reached zero, stop propagating.
18525 (fwprop_addr): Ditto.
18527 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18530 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18531 a FUNCTION_DECL, pass it as decl instead of origin to
18534 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18536 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18539 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18542 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18543 TARGET_VSX_SMALL_INTEGER.
18545 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18547 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18548 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18550 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18552 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18553 extraction from odd-numbered MSA register.
18555 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18557 PR middle-end/80173
18558 * expmed.c (store_bit_field_1): Don't attempt to create
18559 a word subreg out of hard registers wider than word if they
18560 have HARD_REGNO_NREGS of 1 for their mode.
18562 PR middle-end/80163
18563 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18564 conversions to integer types wider than word and pointer.
18567 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18568 (rtx_equal_for_cselib_p): Pass 0 to it.
18569 * cselib.c (cselib_hasher::equal): Likewise.
18570 (rtx_equal_for_cselib_1): Add depth argument. If depth
18571 is 128, don't look up VALUE locs and punt. Increment
18572 depth in recursive calls when walking VALUE locs.
18574 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18576 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18577 (make_gcov_file_name): Use the canonical path name for generating
18579 (read_line): Fix handling of files with ascii null bytes.
18581 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18583 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18584 to initialise a vector register instead
18585 of using a const_int.
18587 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18589 PR translation/80189
18590 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18591 diagnostic messages.
18593 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18596 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18597 (dfp_diex_<mode>): Update mode of operand 1.
18598 * doc/extend.texi (dxex, dxexq): Document change to return type.
18599 (diex, diexq): Document change to argument type.
18601 2017-03-30 Martin Jambor <mjambor@suse.cz>
18604 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18605 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18606 it reflects the signature changes performed at the callee side.
18607 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18608 to cgraph_build_function_type_skip_args.
18609 (build_function_decl_skip_args): Adjust call to the above function.
18611 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18614 * config/i386/sse.md
18615 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18616 register as dest whenever it is a MEM not rtx_equal_p to the
18617 corresponding dup operand, and when forcing into reg move the
18618 reg into the memory afterwards.
18619 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18620 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18621 for the force_reg mode.
18622 (avx512vl_vextractf128<mode>): Use register as dest either
18623 always when a MEM, or when it is a MEM not rtx_equal_p to the
18624 corresponding dup operand, or even not when it is a CONST_VECTOR
18625 depending on the mode and lo vs. hi.
18626 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18628 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18629 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18630 Likewise. Require that operands[2] is even.
18631 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18632 Remove extraneous parens. Require that operands[2] is a multiple
18634 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18635 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18636 disallow memory then.
18638 2017-03-30 Richard Biener <rguenther@suse.de>
18640 PR tree-optimization/77498
18641 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18642 to non-constants over backedges.
18644 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18646 PR rtl-optimization/80233
18647 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18648 as last_combined_insn. Do not test for BARRIER_P separately.
18650 2017-03-29 Andreas Schwab <schwab@suse.de>
18653 * calls.c (prepare_call_address): Convert funexp to Pmode before
18654 copying to temp reg.
18656 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18658 PR tree-optimization/80158
18659 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18660 Handle possible future case of more than one alternate
18662 (replace_rhs_if_not_dup): Likewise.
18663 (replace_one_candidate): Likewise.
18665 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18667 PR rtl-optimization/80193
18668 * ira.c (ira): Do not check allocation for LRA.
18670 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18672 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18673 (nvptx_output_simt_exit): Declare.
18674 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18675 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18676 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18677 (nvptx_declare_function_name): ...here. Emit declaration of local
18678 memory space buffer for omp_simt_enter insn.
18679 (nvptx_output_unisimt_switch): New.
18680 (nvptx_output_softstack_switch): New.
18681 (nvptx_output_simt_enter): New.
18682 (nvptx_output_simt_exit): New.
18683 * config/nvptx/nvptx.h (struct machine_function): New fields
18684 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18685 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18686 (UNSPECV_SIMT_EXIT): Ditto.
18687 (omp_simt_enter_insn): New insn.
18688 (omp_simt_enter): New expansion.
18689 (omp_simt_exit): New insn.
18690 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18692 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18693 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18694 (expand_GOMP_SIMT_EXIT): New.
18695 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18696 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18697 (GOMP_SIMT_EXIT): Ditto.
18698 * target-insns.def (omp_simt_enter): New insn.
18699 (omp_simt_exit): Ditto.
18700 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18702 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18703 (lower_rec_input_clauses): Likewise.
18704 (lower_lastprivate_clauses): Handle SIMT privatization.
18706 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18707 (ompdevlow_adjust_simt_enter): New.
18708 (find_simtpriv_var_op): New.
18709 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18710 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18712 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18713 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18714 (copy_decl_for_dup_finish): Ditto.
18716 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18718 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18721 * config/i386/i386.c (ix86_option_override_internal): Always
18722 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18724 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18726 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18728 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18730 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18731 mark new edge's irreducible flag accordign to it.
18732 (vect_do_peeling): Check loop preheader edge's irreducible flag
18733 and pass it to function slpeel_add_loop_guard.
18735 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18737 PR tree-optimization/80218
18738 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18739 Update block frequencies and counts.
18741 2017-03-28 Richard Biener <rguenther@suse.de>
18743 PR tree-optimization/78644
18744 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18745 of a simplification result we may not use it at all.
18747 2017-03-28 Richard Biener <rguenther@suse.de>
18750 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18751 without arguments, generate default definition of a SSA name.
18753 2017-03-28 Richard Biener <rguenther@suse.de>
18755 PR middle-end/80222
18756 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18757 TYPE_REF_CAN_ALIAS_ALL references.
18758 * fold-const.c (fold_indirect_ref_1): Likewise.
18760 2017-03-28 Martin Liska <mliska@suse.cz>
18763 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18764 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18766 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18767 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18769 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18770 (EXTRA_SPECS): Define.
18771 (SUBTARGET_EXTRA_SPECS): Likewise.
18772 (SUBTARGET_CPP_SPEC): Likewise.
18773 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18774 SUBTARGET_EXTRA_SPECS.
18775 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18777 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18779 * config/arc/simdext.md (vst64_insn): Update pattern.
18780 (vld32wh_insn): Likewise.
18781 (vld32wl_insn): Likewise.
18782 (vld64_insn): Likewise.
18783 (vld32_insn): Likewise.
18785 2017-03-28 Marek Polacek <polacek@redhat.com>
18788 * fold-const.c (fold_comparison): Use protected_set_expr_location
18789 instead of SET_EXPR_LOCATION.
18791 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18793 * tree.c (add_expr): Avoid name lookup warning.
18795 2017-03-27 Jeff Law <law@redhat.com>
18797 PR tree-optimization/80216
18798 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18799 function name. Limit recursion depth.
18800 (record_temporary_equivalences): Corresponding changes.
18802 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18804 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18807 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18810 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18812 * cfgcleanup.c (reg_note_cfa_p): New array.
18813 (insns_have_identical_cfa_notes): New function.
18814 (old_insns_match_p): Don't cross-jump in between /f
18815 and non-/f instructions. If both i1 and i2 are frame related,
18816 verify all CFA notes, their order and content.
18818 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18821 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18822 HImode and SImode with zero extend to DImode to one insn.
18823 (bswap<mode>2_extenddi): Likewise.
18824 (bswapsi2_extenddi): Likewise.
18825 (bswaphi2_extendsi): Likewise.
18826 (bswaphi2): Combine bswap HImode and SImode into one insn.
18827 Separate memory insns from swapping register.
18828 (bswapsi2): Likewise.
18829 (bswap<mode>2): Likewise.
18830 (bswaphi2_internal): Delete, no longer used.
18831 (bswapsi2_internal): Likewise.
18832 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18833 store, and gpr<-gpr swap insns.
18834 (bswap<mode>2_store): Likewise.
18835 (bswaphi2_reg): Register only splitter, combine with the splitter.
18836 (bswaphi2 splitter): Likewise.
18837 (bswapsi2_reg): Likewise.
18838 (bswapsi2 splitter): Likewise.
18839 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18840 the insns into load, store, and register/register insns.
18841 (bswapdi2_ldbrx): Likewise.
18842 (bswapdi2_load): Likewise.
18843 (bswapdi2_store): Likewise.
18844 (bswapdi2_reg): Likewise.
18846 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18848 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18849 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18851 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18854 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18856 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18857 special handling for target option conflicts between dform
18858 options (-mpower9-dform, -mpower9-dform-vector,
18859 -mpower9-dform-scalar) and -mno-direct-move.
18861 2017-03-27 Richard Biener <rguenther@suse.de>
18863 PR tree-optimization/80181
18864 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18866 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18868 * config/arc/predicates.md (move_double_src_operand): Replace the
18869 call to move_double_src_operand with a call to address_operand.
18871 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18873 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18874 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18875 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18877 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18879 * config/arc/predicates.md (long_immediate_loadstore_operand):
18880 Consider scaled addresses cases.
18882 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18884 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18885 restored when in interrupt.
18886 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18887 doesn't have delay slot.
18889 2017-03-27 Richard Biener <rguenther@suse.de>
18892 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18893 inlined thunk clones.
18895 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18898 * asan.c (instrument_derefs): Copy over last operand from
18899 original COMPONENT_REF to the new COMPONENT_REF with
18900 DECL_BIT_FIELD_REPRESENTATIVE.
18901 * ubsan.c (instrument_object_size): Likewise.
18903 2017-03-27 Richard Biener <rguenther@suse.de>
18905 PR tree-optimization/80170
18906 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18907 sure DR/SCEV didnt fold in constants we do not see when looking
18908 at the reference base alignment.
18910 2017-03-27 Richard Biener <rguenther@suse.de>
18912 PR middle-end/80171
18913 * gimple-fold.c (fold_ctor_reference): Properly guard against
18914 NULL return value from canonicalize_constructor_val.
18916 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18919 * config/i386/i386.c (ix86_expand_builtin)
18920 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18921 flags reg setting and flags reg using instructions.
18922 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18923 clobbering instructions to zero extend op2.
18925 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18927 * doc/install.texi (Configuration) <--with-aix-soname>:
18928 Update link to AIX ld.
18930 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18932 PR rtl-optimization/80160
18933 PR rtl-optimization/80159
18934 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18935 reg_alternate_class into account.
18937 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18940 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18941 to consider in curr_insn_transform.
18943 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18945 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18946 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18947 and emit_mode_inner.
18949 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18951 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18952 argument to the overloaded builtin variants. Use the new flag to
18953 deprecate certain builtin variants.
18954 * config/s390/s390-builtin-types.def: Add new builtin types.
18955 * config/s390/s390-builtins.h: Support new flags field for
18956 overloaded builtins.
18957 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18958 (s390_macro_to_expand): Enable vector float data type.
18959 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18960 builtins by incrementing the __VEC__ version number.
18961 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18963 (s390_resolve_overloaded_builtin): Emit error messages depending
18964 on the builtin flags.
18965 * config/s390/s390.c (s390_expand_builtin): Support additional
18966 flags argument. Change error message to match the messages
18967 emitted in s390-c.c.
18968 * config/s390/s390.md: New UNSPEC_* constants.
18969 (op_type): Add new instruction types.
18970 * config/s390/vecintrin.h: Add new builtins and test data class
18972 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18973 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18974 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18975 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18976 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18977 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18979 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18980 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18981 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18982 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18984 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18985 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18986 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18987 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18989 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18990 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18991 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18992 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18993 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18994 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18995 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18997 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18998 ("vec_scatter_element<V_HW_4:mode>_DI")
18999 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19000 ("vec_fpint<mode>", "vflls")
19001 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19002 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19003 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19004 ("*vec_cmphe<mode>_cc"): ... these.
19006 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19007 mode constant instead of magic value.
19009 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19011 * config/s390/s390.c (s390_expand_vec_compare): Support other
19012 vector floating point modes than just V2DF.
19013 (s390_expand_vcond): Likewise.
19014 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19015 (s390_cannot_change_mode_class): Prevent mode changes between TF
19016 and V1TF in vector registers.
19017 * config/s390/s390.md (DF, SF): New mode attributes.
19018 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19019 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19020 SFmode support for VRs.
19021 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19023 (VFT, VF_HW): New mode iterators.
19024 (vw, sdx): New mode attributes.
19025 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19026 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19027 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19028 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19029 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19030 also the new vector floating point modes. Renaming to ...
19032 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19033 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19034 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19035 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19036 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19037 ("vec_unordered<mode>"): ... these.
19039 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19040 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19041 ("*vec_extendv2df"): New insn definitions.
19043 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19045 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19046 ("mulditi3_2", "*muldi3_sign"): New patterns.
19047 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19048 rename the pattern definition.
19050 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19052 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19054 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19056 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19058 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19059 instruction if possible.
19060 * config/s390/vector.md (vec_halfnumelts): New mode
19062 ("*vec_vllezlf<mode>"): New pattern.
19064 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19066 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19067 ("popcountv4si2", "popcountv2di2"): Rename to ...
19068 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19069 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19071 ("popcount<mode>2_vxe"): New pattern.
19073 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19075 * common/config/s390/s390-common.c (processor_flags_table): Add
19077 * config.gcc: Add arch12.
19078 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19079 Default to arch12 for unknown CPU model numbers.
19080 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19081 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19082 PROCESSOR_max sanity check.
19083 * config/s390/s390-opts.h (enum processor_type): Add
19085 * config/s390/s390.c (processor_table): Add arch12.
19086 (s390_expand_builtin): Add check for B_VXE flag.
19087 (s390_issue_rate): Add PROCESSOR_ARCH12.
19088 (s390_get_sched_attrmask): Likewise.
19089 (s390_get_unit_mask): Likewise.
19090 (s390_sched_score): Enable z13 scheduling for arch12.
19091 (s390_sched_reorder): Likewise.
19092 (s390_sched_variable_issue): Likewise.
19093 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19095 (s390_tune_attr): Use z13 scheduling also for arch12.
19096 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19097 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19098 (TARGET_VXE_P): New macros.
19099 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19100 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19101 * config/s390/s390.opt: Add arch12 as processor_type.
19103 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19105 * config/s390/s390.md
19106 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19107 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19108 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19110 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19111 Rename expanders to ...
19113 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19114 ("fixuns_truncdddi2_emu"): ... these.
19116 ("fixuns_trunc<mode>si2_emu"): New expander.
19118 ("*fixuns_truncdfdi2_z13"): Rename to ...
19119 ("*fixuns_truncdfdi2_vx"): ... this.
19121 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19123 * config/s390/2964.md: Remove the single element vector compare
19124 instructions which are no longer used.
19125 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19127 (s390_canonicalize_comparison): Remove handling of DFmode
19129 (s390_expand_vec_compare_scalar): Remove function.
19130 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19131 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19133 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19135 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19137 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19139 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19140 will anyway by matched by mov<mode>_64dfp.
19142 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19144 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19145 vlef/vstef. Add missing operand to vleif.
19147 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19149 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19150 pair for all vector types with 64 bit elements.
19151 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19152 * config/s390/vector.md (V_HW_64): ... here.
19153 (V_128_NOSINGLE): New mode iterator.
19154 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19155 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19156 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19157 ("*vec_load_pairv2di"): Change to ...
19158 ("*vec_load_pair<mode>"): ... this one.
19160 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19162 * config/s390/constraints.md: Add comments.
19163 (jKK): Reject element sizes > 8 bytes.
19164 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19166 * config/s390/s390.md: Add the s_operand checks formerly in
19167 s390_split_ok_p to various splitters where they are still
19169 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19170 for 128 bit vectors. Plus two splitters.
19172 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19174 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19177 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19180 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19181 error if the boundary argument is not constant.
19183 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19185 PR rtl-optimization/80112
19186 * loop-doloop.c (doloop_condition_get): Don't check condition
19187 if cmp isn't SET with IF_THEN_ELSE src.
19189 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19191 PR tree-optimization/80158
19192 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19193 replacing a candidate statement, also replace it for the
19194 candidate's alternate interpretation.
19195 (replace_rhs_if_not_dup): Likewise.
19196 (replace_one_candidate): Likewise.
19198 2017-03-24 Richard Biener <rguenther@suse.de>
19200 PR tree-optimization/80167
19201 * graphite-isl-ast-to-gimple.c
19202 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19204 (translate_isl_ast_to_gimple::get_rename): Likewise.
19206 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19208 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19209 handling of certain combinations of target options, including the
19210 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19211 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19213 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19216 * config/arm/arm.md (*load_multiple): Add reload_completed to
19217 matching condition.
19219 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19220 Richard Biener <rguenth@suse.de>
19222 PR tree-optimization/79908
19223 PR tree-optimization/80136
19224 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19225 been cast away, gimplify_and_add suffices.
19227 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19229 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19231 2017-03-23 Richard Biener <rguenther@suse.de>
19233 PR tree-optimization/80032
19234 * gimplify.c (gimple_push_cleanup): Forced unconditional
19235 cleanups still have to go to the conditional_cleanups
19238 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19240 PR tree-optimization/80072
19241 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19243 (next_operand_entry_id): Change type to unsigned int.
19244 (sort_by_operand_rank): Make sure to return the right return value
19245 even if unsigned fields are bigger than INT_MAX.
19246 (struct oecount): Change cnt and id type to unsigned int.
19247 (oecount_hasher::equal): Formatting fix.
19248 (oecount_cmp): Make sure to return the right return value
19249 even if unsigned fields are bigger than INT_MAX.
19250 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19253 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19254 TREE_READONLY on result if writing it more than once.
19257 * doc/invoke.texi (-fsanitize=thread): Document that with
19258 -fnon-call-exceptions atomics are not able to throw
19262 * tsan.c: Include tree-eh.h.
19263 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19264 maybe_clean_or_replace_eh_stmt where needed.
19265 (instrument_memory_accesses): Add cfg_changed argument.
19266 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19267 if it returned true.
19268 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19270 PR rtl-optimization/63191
19271 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19272 wrapper function, moved the whole old content into ...
19273 (ix86_delegitimize_address_1): ... this. New inline function.
19274 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19275 true as last argument instead of ix86_delegitimize_address.
19277 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19279 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19280 cortexa57_branch_cost.
19282 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19284 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19286 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19289 * doc/md.texi (Constraints): Document wA constraint.
19290 * config/rs6000/constraints.md (wA): New.
19291 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19292 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19293 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19294 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19296 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19299 * gimplify.c (is_oacc_declared): New function.
19300 (oacc_default_clause): Use it to set default flags for acc declared
19301 variables inside parallel regions.
19302 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19303 declared variables.
19304 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19305 declare attribute to any decl as necessary.
19307 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19310 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19311 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19312 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19313 (arm_arch_lpae): This.
19314 * config/arm/arm.c (arm_arch7ve): Rename into ...
19315 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19316 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19319 2017-03-22 Martin Liska <mliska@suse.cz>
19322 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19323 error message instead of an ICE.
19325 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19327 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19329 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19331 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19333 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19336 2017-03-21 Martin Sebor <msebor@redhat.com>
19338 * doc/extend.texi: Use "cannot" instead of "can't."
19339 * doc/hostconfig.texi: Same.
19340 * doc/install.texi: Same.
19341 * doc/invoke.texi: Same.
19342 * doc/loop.texi: Same.
19343 * doc/md.texi: Same.
19344 * doc/objc.texi: Same.
19345 * doc/rtl.texi: Same.
19346 * doc/tm.texi: Same.
19347 * doc/tm.texi.in: Same.
19348 * doc/trouble.texi: Same.
19350 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19353 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19354 (collect_checksum_attributes): Set it.
19355 (die_checksum_ordered): Use it.
19357 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19359 PR tree-optimization/79908
19360 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19361 change: For a VA_ARG whose LHS has been cast away, use
19362 force_gimple_operand to construct the side effects.
19364 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19366 PR translation/80001
19367 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19368 more amenable to translation.
19369 (oacc_loop_auto_partitions): Likewise.
19371 2017-03-21 Marek Polacek <polacek@redhat.com>
19372 Martin Sebor <msebor@redhat.com>
19374 PR tree-optimization/80109
19375 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19376 on INTEGRAL_TYPE_P.
19378 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19379 Segher Boessenkool <segher@kernel.crashing.org>
19382 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19383 check reg_used_between_p between insn and one of succ or succ2
19384 depending on if succ is artificial insn not inserted into insn
19387 2017-03-21 Martin Liska <mliska@suse.cz>
19389 PR gcov-profile/80081
19390 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19391 * doc/gcc.texi: Include gcov-dump stuff.
19392 * doc/gcov-dump.texi: New file.
19394 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19396 PR rtl-optimization/79150
19397 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19398 conditional jump, if the jump is the last insn of the loop.
19400 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19401 Richard Biener <rguenth@suse.de>
19403 PR tree-optimization/79908
19404 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19405 been cast away, use force_gimple_operand to construct the side
19408 2017-03-21 Martin Liska <mliska@suse.cz>
19410 PR libfortran/79956
19411 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19414 2017-03-21 Brad Spengler <spender@grsecurity.net>
19417 * plugin.c (htab_hash_plugin): New function.
19418 (add_new_plugin): Use it and adjust.
19419 (parse_plugin_arg_opt): Adjust.
19420 (init_one_plugin): Likewise.
19422 2017-03-21 Richard Biener <rguenther@suse.de>
19424 PR tree-optimization/80032
19425 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19426 if set force the cleanup to happen unconditionally.
19427 (gimplify_target_expr): Push inserted clobbers with force_uncond
19428 to avoid them being removed by control-dependent DCE.
19430 2017-03-21 Richard Biener <rguenther@suse.de>
19432 PR tree-optimization/80122
19433 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19434 va_arg_pack_len when the inlined call stmt requires pack
19436 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19438 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19441 * tsan.c (instrument_builtin_call): If the memory model argument
19442 is not a constant, assume it is valid.
19445 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19448 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19450 PR rtl-optimization/79910
19451 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19452 if its dest is used by an insn before I2 (other than the combined
19453 insns themselves, which are properly handled already).
19455 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19458 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19460 * combine.c (record_used_regs): New static function.
19461 (try_combine): Handle situations where there is an additional
19462 instruction between I2 and I3 which needs to have a LOG_LINK
19466 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19468 * combine.c (try_combine): Delete redundant i1 test. Call
19469 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19471 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19474 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19475 alternatives 13/14.
19477 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19479 PR tree-optimization/80054
19480 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19481 the optimization if a PHI or any of its arguments is not dominated
19482 by the candidate's basis. Use gphi* rather than gimple* as
19484 (replace_profitable_candidates): Clean up a gimple* variable that
19485 should be a gphi* variable.
19487 2017-03-20 Martin Sebor <msebor@redhat.com>
19490 * doc/extend.texi (attribute constructor): Document present limitation.
19492 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19495 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19496 __POWER9_VECTOR__ #ifdef control, change template definition to
19497 use Power9-specific built-in function.
19498 (vec_any_eq): Likewise.
19499 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19500 to control outcomes from this test.
19501 (vector_ae_<mode>p): For VEC_F modes, likewise.
19503 2017-03-20 Ian Lance Taylor <iant@google.com>
19505 * config/i386/i386.c (ix86_function_regparm): Save an extra
19506 register for -fsplit-stack with DECL_STATIC_CHAIN.
19508 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19511 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19512 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19514 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19516 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19518 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19521 2017-03-20 Marek Polacek <polacek@redhat.com>
19524 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19526 2017-03-20 Richard Biener <rguenther@suse.de>
19528 PR tree-optimization/80113
19529 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19530 allocate extra SSA name for PHI def.
19531 (add_close_phis_to_outer_loops): Likewise.
19532 (add_close_phis_to_merge_points): Likewise.
19533 (copy_loop_close_phi_args): Likewise.
19534 (copy_cond_phi_nodes): Likewise.
19536 2017-03-20 Martin Liska <mliska@suse.cz>
19538 PR middle-end/79753
19539 * tree-chkp.c (chkp_build_returned_bound): Do not build
19540 returned bounds for a LHS that's not a BOUNDED_P type.
19542 2017-03-20 Martin Liska <mliska@suse.cz>
19546 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19547 COMPLEX_CST and VECTOR_CST.
19549 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19552 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19553 target operand. A new splitter adds the clobber statement in case
19554 the target operand is dead anyway.
19556 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19558 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19559 to age-old versions of binutils and glibc.
19561 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19563 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19565 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19567 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19569 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19571 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19572 requirement for binutils 2.13.
19574 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19576 * combine.c (try_combine): Delete redundant i1 test. Call
19577 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19579 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19581 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19582 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19584 <riscv64-*-elf>: Re-arrange section
19585 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19586 <riscv32-*-linux>: Likewise.
19587 <riscv64-*-elf>: Likewise
19588 <riscv64-*-linux>: Likewise.
19590 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19593 * aarch64.opt(verbose-cost-dump): Fix typo.
19595 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19598 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19599 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19601 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19603 * reload.c (find_reloads): When reloading a nonoffsettable address,
19604 use RELOAD_OTHER for it and its address reloads.
19606 PR rtl-optimization/79910
19607 * combine.c (record_used_regs): New static function.
19608 (try_combine): Handle situations where there is an additional
19609 instruction between I2 and I3 which needs to have a LOG_LINK
19612 2017-03-17 Jeff Law <law@redhat.com>
19614 PR tree-optimization/71437
19615 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19616 conditional in the hash table first.
19617 (vrp_dom_walker::before_dom_children): Extract condition from
19618 ASSERT_EXPR. Record condition, its inverion and any implied
19619 conditions as well.
19621 2017-03-17 Marek Polacek <polacek@redhat.com>
19622 Markus Trippelsdorf <markus@trippelsdorf.de>
19624 PR tree-optimization/80079
19625 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19628 2017-03-17 Richard Biener <rguenther@suse.de>
19630 PR middle-end/80075
19631 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19632 Properly verify the LHS before the RHS possibly claims to be
19634 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19637 2017-03-17 Martin Jambor <mjambor@suse.cz>
19639 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19640 (List of -O2 options): Likewise.
19641 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19644 2017-03-17 Tom de Vries <tom@codesourcery.com>
19646 * gcov-dump.c (print_usage): Print bug_report_url.
19648 2017-03-17 Richard Biener <rguenther@suse.de>
19650 PR middle-end/80050
19651 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19652 (parser::peek): Likewise.
19654 2017-03-17 Richard Biener <rguenther@suse.de>
19656 PR tree-optimization/80048
19657 * sese.c (free_sese_info): Properly release rename_map and
19658 copied_bb_map elements.
19660 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19662 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19663 Add linked-list forward and backlinks. Insert on
19664 construction, remove on destruction.
19665 (class pass_store_merging): Add m_stores_head field.
19666 (pass_store_merging::terminate_and_process_all_chains):
19667 Iterate over m_stores_head list.
19668 (pass_store_merging::terminate_all_aliasing_chains):
19670 (pass_store_merging::execute): Check for debug stmts first.
19671 Push new chains onto the m_stores_head stack.
19673 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19676 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19677 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19678 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19680 2017-03-16 Jeff Law <law@redhat.com>
19682 PR tree-optimization/71437
19683 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19684 member function. Implementation moved into after_dom_children
19685 member function and into the threader's thread_outgoing_edges
19687 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19688 some code into new thread_outgoing_edges.
19689 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19690 definition. Simplify marker handling (do it here). Assume we always
19691 have the available expression and the const/copies tables.
19692 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19694 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19695 * tree-vrp.c (equiv_stack): No longer file scoped.
19696 (vrp_dom_walker): New class.
19697 (vrp_dom_walker::before_dom_children): New member function.
19698 (vrp_dom_walker::after_dom_children): Likewise.
19699 (identify_jump_threads): Setup domwalker. Use it rather than
19700 walking edges in a random order by hand. Simplify setup/finalization.
19701 (finalize_jump_threads): Remove.
19702 (vrp_finalize): Do not call identify_jump_threads here.
19703 (execute_vrp): Do it here instead and call thread_through_all_blocks
19706 PR tree-optimization/71437
19707 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19709 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19711 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19712 (dom_opt_dom_walker::thread_across_edge): Remove
19713 handle_dominating_asserts argument. All callers changed.
19714 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19715 changes. Remove calls to lhs_of_dominating_assert. Other
19716 uses of handle_dominating_asserts turn into unconditional code
19717 (simplify_control_stmt_condition_1): Likewise.
19718 (simplify_control_stmt_condition): Likewise.
19719 (thread_through_normal_block, thread_across_edge): Likewise.
19720 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19721 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19722 object if it is not an SSA_NAME.
19723 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19724 before calling into the VRP specific simplifiers.
19725 (identify_jump_threads): Remove handle_dominating_asserts
19728 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19731 * tree-diagnostic.c (default_tree_printer): No longer static.
19732 * tree-diagnostic.h (default_tree_printer): New prototype.
19734 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19736 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19737 Change ins into fmov.
19739 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19741 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19742 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19743 Use h_con constraint for operand 1.
19744 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19745 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19747 2017-03-15 Jeff Law <law@redhat.com>
19749 PR tree-optimization/71437
19750 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19751 (record_temporary_equivalences): Use it.
19753 PR tree-optimization/71437
19754 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19755 tree-ssa-scopedtables.
19756 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19757 (record_conditions, record_cond, vuse_eq): Likewise.
19758 (record_edge_info): Adjust to API tweak of record_conditions.
19759 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19760 (record_temporary_equivalences, optimize_stmt): Likewise.
19761 (eliminate_redundant_computations): Likewise.
19762 (record_equivalences_from_stmt): Likewise.
19763 * tree-ssa-scopedtables.c: Include options.h and params.h.
19764 (vuse_eq): New function, moved from tree-ssa-dom.c
19765 (build_and_record_new_cond): Likewise.
19766 (record_conditions): Likewise. Accept vector of conditions rather
19767 than edge_equivalence structure for first argument.
19768 for the first argument.
19769 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19770 from tree-ssa-dom.c.
19771 (avail_exprs_stack::record_cond): Likewise.
19772 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19773 from tree-ssa-dom.c.
19774 (avail_exprs_stack): Add new member functions lookup_avail_expr
19776 (record_conditions): Declare.
19778 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19781 * lra-constraints.c (process_alt_operands): Increase reject for
19782 reloading an input/output operand.
19784 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19787 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19788 insns to convert from signed/unsigned char/short to IEEE 128-bit
19790 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19792 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19795 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19796 subreg of inner mode for values already in registers.
19798 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19800 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19801 iteration reg is used after the loop.
19803 2017-03-14 Martin Sebor <msebor@redhat.com>
19805 PR tree-optimization/79800
19806 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19807 precision in negative-positive range.
19808 (format_floating): Call non-const overload with adjusted precision.
19810 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19813 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19816 2017-03-14 Martin Sebor <msebor@redhat.com>
19818 PR middle-end/80020
19819 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19820 * builtins.def (aligned_alloc): Use it.
19823 * Makefile.in (GTFILES): Add calls.c.
19824 * calls.c: Include "gt-calls.h".
19826 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19828 PR rtl-optimization/79728
19829 * regs.h (struct target_regs): New field
19830 x_contains_allocatable_regs_of_mode.
19831 (contains_allocatable_regs_of_mode): New macro.
19832 * reginfo.c (init_reg_sets_1): Initialize it, and change
19833 contains_reg_of_mode so it includes global regs as well.
19834 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19835 rather than contains_regs_of_mode.
19837 2017-03-14 Martin Liska <mliska@suse.cz>
19839 * doc/invoke.texi: Document options that can't be combined with
19840 -fcheck-pointer-bounds.
19842 2017-03-14 Martin Liska <mliska@suse.cz>
19844 PR middle-end/79831
19845 * doc/invoke.texi (-Wchkp): Document the option.
19847 2017-03-14 Martin Liska <mliska@suse.cz>
19849 * Makefile.in: Install gcov-dump.
19851 2017-03-14 Martin Liska <mliska@suse.cz>
19853 * multiple_target.c (expand_target_clones): Bail out for
19854 an invalid attribute.
19856 2017-03-14 Richard Biener <rguenther@suse.de>
19858 * alias.c (struct alias_set_entry): Pack properly.
19859 * cfgloop.h (struct loop): Likewise.
19860 * cse.c (struct set): Likewise.
19861 * ipa-utils.c (struct searchc_env): Likewise.
19862 * loop-invariant.c (struct invariant): Likewise.
19863 * lra-remat.c (struct cand): Likewise.
19864 * recog.c (struct change_t): Likewise.
19865 * rtl.h (struct address_info): Likewise.
19866 * symbol-summary.h (function_summary): Likewise.
19867 * tree-loop-distribution.c (struct partition): Likewise.
19868 * tree-object-size.c (struct object_size_info): Likewise.
19869 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19870 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19871 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19872 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19873 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19874 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19875 (struct _stmt_vec_info): Likewise.
19877 2017-03-14 Martin Liska <mliska@suse.cz>
19880 * multiple_target.c (create_dispatcher_calls): Check that
19881 a target can create a function dispatcher.
19883 2017-03-14 Martin Liska <mliska@suse.cz>
19886 * multiple_target.c (expand_target_clones): Drop local.local
19887 flag for default implementation.
19889 2017-03-14 Richard Biener <rguenther@suse.de>
19891 PR tree-optimization/80030
19892 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19894 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19896 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19897 gcc_fallthrough() instead of __attribute__((fallthrough));
19899 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19901 * doc/gcc.texi: Remove "up" link to (DIR).
19902 * doc/gccint.texi: Ditto.
19904 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19906 * doc/install.texi (Specific) <avr>: Remove reference to
19909 2017-03-13 Jeff Law <law@redhat.com>
19911 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19912 attribute rather than comments.
19914 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19915 match_scratch operand is highest.
19917 2017-03-13 Martin Liska <mliska@suse.cz>
19919 PR middle-end/78339
19920 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19921 is a CHKP clone, use original declaration.
19923 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19925 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19926 (arc_conditional_register_usage): Use a different allocation order
19927 when optimizing for size.
19928 * common/config/arc/arc-common.c (arc_option_optimization_table):
19929 Section anchors default on when optimizing for size.
19931 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19933 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19935 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19937 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19938 * config/arc/arc.md (cpu_facility): Add cd variant.
19939 (*movqi_insn): Add code density variant.
19940 (*movhi_insn): Likewise.
19941 (*movqi_insn): Likewise.
19942 (*addsi3_mixed): Likewise.
19943 (subsi3_insn): Likewise.
19945 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19947 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19949 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19951 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19952 expressions with MINUS and UNARY ops.
19954 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19957 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19959 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19960 between vec_select and vector argument.
19961 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19962 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19963 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19964 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19965 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19966 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19968 2017-03-13 Richard Biener <rguenther@suse.de>
19971 * params.def (vect-max-peeling-for-alignment): Fix typo.
19973 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19975 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19976 issue that only occurred with binutils below 2.18.
19978 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19980 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19981 refer to binutils 2.11/2.12 minimum.
19983 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19985 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19986 ftp.kernel.org and simplify binutils requirement.
19988 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19990 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19992 (Optimize Options): Ditto. Also remove redundancy.
19994 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19996 PR translation/79848
19997 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19999 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20000 to G_ to avoid double translation.
20002 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20004 PR translation/79923
20005 * auto-profile.c (get_combined_location): Convert leading
20006 character of diagnostics to lower case and remove trailing period.
20007 (read_profile): Likewise for various diagnostics.
20008 * config/arm/arm.c (arm_option_override): Remove trailing period
20009 from various diagnostics.
20010 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20011 (msp430_expand_delay_cycles): Likewise.
20013 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20016 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20017 full command-line argument, rather than just "str".
20018 (aarch64_validate_march): Likewise.
20019 (aarch64_validate_mtune): Likewise.
20021 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20023 PR rtl-optimization/78911
20024 * lra-assigns.c (must_not_spill_p): New function.
20025 (spill_for): Use it.
20027 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20029 PR tree-optimization/79981
20030 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20031 ATOMIC_COMPARE_EXCHANGE ifn result.
20032 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20033 IFN_ATOMIC_COMPARE_EXCHANGE.
20035 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20038 * opts.c (parse_sanitizer_options): Add missing question mark to
20039 "did you mean" message.
20041 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20043 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20045 (VMULEUH_UNS): Likewise.
20046 (VMULOUB_UNS): Likewise.
20047 (VMULOUH_UNS): Likewise.
20048 * config/rs6000/rs6000.c (builtin_function_type): Remove
20049 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20051 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20054 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20055 x with result of extra_parsing_for_operand_code_0.
20056 (function_reader::extra_parsing_for_operand_code_0): Convert
20057 return type from void to rtx, returning x. When reading
20058 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20059 larger size containing struct block_symbol.
20061 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20063 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20064 -mfloat128-hardware without -m64.
20066 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20069 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20070 entries to the case statement that marks unsigned arguments to
20071 overloaded functions.
20073 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20075 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20076 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20078 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20081 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20082 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20084 2017-03-10 Martin Liska <mliska@suse.cz>
20088 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20089 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20092 2017-03-10 Olivier Hainque <hainque@adacore.com>
20094 * tree-switch-conversion (array_value_type): Start by resetting
20095 candidate type to it's main variant.
20097 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20099 PR rtl-optimization/79909
20100 * combine.c (try_combine): Use simplify_replace_rtx on individual
20101 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20102 of the whole CALL_INSN_FUNCTION_USAGE.
20104 PR tree-optimization/79972
20105 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20106 get_range_info on SSA_NAMEs. Formatting fixes.
20108 2017-03-10 Richard Biener <rguenther@suse.de>
20109 Jakub Jelinek <jakub@redhat.com>
20111 PR tree-optimization/77975
20112 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20113 edge to be constant.
20114 (get_val_for): For constant x return it. Formatting fix.
20115 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20116 would use the same bases as the current one.
20118 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20120 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20121 instead of vec_select for V1TImode.
20122 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20124 (VSX_LE_128): Add V1TI to this mode iterator.
20125 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20126 (*vsx_le_perm_store_<mode>): Likewise.
20127 (pre-reload splitter for VSX stores): Likewise.
20128 (post-reload splitter for VSX stores): Likewise.
20129 (*vsx_xxpermdi2_le_<mode>): Likewise.
20130 (*vsx_lxvd2x2_le_<mode>): Likewise.
20131 (*vsx_stxvd2x2_le_<mode>): Likewise.
20133 2017-03-09 Michael Eager <eager@eagercon.com>
20135 Correct failures with --enable-checking=yes,rtl.
20137 * config/microblaze/microblaze.c (microblaze_expand_shift):
20138 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20139 test for const0_rtx.
20140 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20141 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20143 2017-03-09 Richard Biener <rguenther@suse.de>
20145 PR tree-optimization/79977
20146 * graphite-scop-detection.c (scop_detection::merge_sese):
20147 Handle the case of extra exits to blocks dominating the entry.
20149 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20151 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20154 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20156 PR rtl-optimization/79949
20157 * lra-constraints.c (process_alt_operands): Check memory when
20158 trying to predict a cycle. Print about the overall increase.
20160 2017-03-09 Richard Biener <rguenther@suse.de>
20162 PR middle-end/79971
20163 * gimple-expr.c (useless_type_conversion_p): Preserve
20164 TYPE_SATURATING for fixed-point types.
20166 2017-03-09 Richard Biener <rguenther@suse.de>
20169 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20170 alignment of BLKmode params.
20172 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20175 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20176 (VALL_NO_V2Q): Likewise.
20178 * config/aarch64/aarch64-simd.md
20179 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20181 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20182 VALL_NO_V2Q mode iterator.
20183 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20185 2017-03-09 Martin Liska <mliska@suse.cz>
20187 PR tree-optimization/79631
20188 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20189 tree_int_cst_sign_bit just for INTEGER constants.
20191 2017-03-09 Martin Liska <mliska@suse.cz>
20195 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20198 2017-03-09 Marek Polacek <polacek@redhat.com>
20201 * tree.c (inchash::add_expr): Handle TREE_VEC.
20203 2017-03-09 Martin Liska <mliska@suse.cz>
20206 (chkp_narrow_size_and_offset): New function.
20207 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20208 (void chkp_parse_bit_field_ref): New function.
20209 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20210 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20212 2017-03-09 Martin Liska <mliska@suse.cz>
20215 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20216 (chkp_find_bounds_1): Remove gcc_unreachable.
20218 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20221 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20222 BUILT_IN_SYNC*, determine the access type from the size suffix and
20223 always build a MEM_REF with that type. Handle forgotten
20224 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20227 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20228 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20229 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20230 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20231 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20232 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20233 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20234 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20235 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20236 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20237 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20238 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20239 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20240 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20241 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20242 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20243 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20244 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20245 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20246 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20247 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20248 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20249 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20250 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20251 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20252 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20253 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20254 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20255 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20256 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20257 definitions outside of __OPTIMIZE__ guarded section.
20260 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20261 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20262 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20263 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20266 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20268 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20269 ("vfenez<mode>"): Add missing constraints.
20271 2017-03-08 Martin Sebor <msebor@redhat.com>
20274 * config/nds32/nds32.c (nds32_option_override):
20275 Fix misspelled diagnostic.
20277 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20280 * gimplify.c (gimplify_omp_for): Replace index var in outer
20281 taskloop statement with an artificial variable and add
20282 OMP_CLAUSE_PRIVATE clause for it.
20284 2017-03-08 Richard Biener <rguenther@suse.de>
20286 PR tree-optimization/79955
20287 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20288 for accesses that are completely outside of the variable.
20290 2017-03-08 Andrew Haley <aph@redhat.com>
20292 PR tree-optimization/79943
20293 * tree-ssa-loop-split.c (compute_new_first_bound): When
20294 calculating the new upper bound, (END-BEG) should be added, not
20297 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20299 * config/avr/avr.md (setmemhi): Make sure match_dup
20300 operand number comes before match_scratch.
20302 2017-03-08 Richard Biener <rguenther@suse.de>
20304 PR tree-optimization/79920
20305 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20306 with ncopies == 1 to ...
20307 (vect_transform_slp_perm_load): ... here. Properly compute
20308 all element loads by iterating VF times over the group. Do
20309 not handle ncopies (computed in a broken way) in
20310 vect_create_mask_and_perm.
20312 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20315 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20316 is a uniform vector, use uniform_vector_p return value instead of
20317 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20319 2017-03-07 Marek Polacek <polacek@redhat.com>
20321 PR middle-end/79809
20322 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20323 (alloca_call_type): Likewise.
20325 2017-03-07 Martin Liska <mliska@suse.cz>
20327 * gcov.c (process_args): Put comment to correct location.
20329 2017-03-07 Martin Liska <mliska@suse.cz>
20331 PR middle-end/68270
20332 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20333 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20334 (chkp_narrow_bounds_for_field): Likewise.
20335 (chkp_parse_array_and_component_ref): Pass one more argument to
20338 2017-03-07 Richard Biener <rguenther@suse.de>
20340 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20343 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20345 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20346 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20348 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20351 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20352 to end of description.
20353 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20355 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20357 PR rtl-optimization/79901
20358 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20360 (*avx512f_<code><mode>3<mask_name>): ... this.
20361 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20362 iterator instead of VI8_AVX2_AVX512BW.
20364 PR rtl-optimization/79901
20365 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20366 min/max expander, expand it using expand_vec_cond_expr.
20369 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20372 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20375 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20376 to void * for PCH reasons.
20377 * dwarf2out.c (output_loc_operands, output_die): Cast
20378 v.val_vec.array to unsigned char *.
20380 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20383 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20386 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20388 PR rtl-optimization/79571
20389 * lra-constraints.c (process_alt_operands): Calculate static
20390 reject and subtract it from overall when only addresses will be
20393 2017-03-06 Julia Koval <julia.koval@intel.com>
20396 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20397 incoming stack boundary to 128 for 64-bit targets.
20399 2017-03-06 Richard Biener <rguenther@suse.de>
20401 PR tree-optimization/79894
20402 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20403 to NULL after folding it.
20405 2017-03-06 Richard Biener <rguenther@suse.de>
20407 PR tree-optimization/79824
20408 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20409 check disabling peeling for gaps.
20411 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20413 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20414 attributes): Document gettimeofday.
20416 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20418 * config/s390/s390.c (s390_option_override_internal): Set
20419 PARAM_MIN_VECT_LOOP_BOUND
20421 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20423 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20424 * config/s390/s390.md: Likewise.
20426 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20429 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20430 (<avx2_avx512>_perm<mode>): Rename to ...
20431 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20433 (<avx512>_perm<mode>_mask): Rename to ...
20434 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20436 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20437 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20438 instead of VI8F_256_512.
20439 (avx512f_perm<mode>): New define_expand.
20440 (avx512f_perm<mode>_mask): Likewise.
20441 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20442 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20444 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20446 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20447 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20449 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20451 2017-03-06 Martin Liska <mliska@suse.cz>
20454 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20455 when having a SSA NAME w/o VAR_DECL assigned to it.
20457 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20459 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20460 msa_dpsub_<su>_d): Fix MODE for vec_select.
20462 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20464 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20466 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20468 2017-03-06 Richard Biener <rguenther@suse.de>
20470 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20471 * plugin.c (register_plugin_info): Likewise.
20472 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20474 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20476 * config/i386/sse.md (sse_storehps, sse_storelps,
20477 avx_<castmode><avxsizesuffix>_<castmode>,
20478 avx512f_<castmode><avxsizesuffix>_<castmode>,
20479 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20480 in condition that at least one operand is not a MEM.
20482 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20484 PR middle-end/79805
20485 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20486 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20488 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20489 gimple_call_nothrow_p flag based on whether original builtin can throw.
20490 If it can, emit following stmts on the fallthrough edge.
20491 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20492 don't create new bb if inserting just debug stmts on the edge, try to
20493 insert them on the fallthru bb or just reset debug stmts.
20495 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20498 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20499 restore recog_data (including the operand rtxes inside it) around
20500 the call to get_insn_template.
20502 2017-03-03 Martin Sebor <msebor@redhat.com>
20504 PR tree-optimization/79699
20505 * context.c (context::~context): Free MPFR caches to avoid
20506 a memory leak on program exit.
20508 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20510 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20511 Use wide_int::ulow () instead of .elt (0).
20513 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20515 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20516 (*pushxf): Limit oF constraint to 32bit targets and add oC
20517 constraint for 64bit targets.
20518 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20519 (*pushdf): Change rmF constraint to rmC.
20521 2017-03-03 Martin Liska <mliska@suse.cz>
20523 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20524 Remove unused variable.
20526 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20529 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20530 is a memory operand, increase num_memory.
20531 (ix86_expand_args_builtin): Likewise.
20533 2017-03-03 Jan Hubicka <jh@suse.cz>
20536 * ipa-devirt.c (maybe_record_node): Properly handle
20537 __cxa_pure_virtual visibility.
20539 2017-03-03 Martin Liska <mliska@suse.cz>
20541 PR tree-optimization/79803
20542 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20544 (pass_loop_prefetch::execute): Disabled optimization if an
20545 assumption about L1 cache size is not met.
20547 2017-03-03 Martin Liska <mliska@suse.cz>
20549 PR rtl-optimization/79574
20550 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20551 (hash_scan_set): Likewise.
20552 (dump_hash_table): Likewise.
20553 (hoist_code): Likewise.
20555 2017-03-03 Richard Biener <rguenther@suse.de>
20557 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20558 in place of uhigh ().
20559 (fixed_convert_from_real): Likewise.
20561 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20564 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20566 2017-03-03 Richard Biener <rguenther@suse.de>
20568 PR middle-end/79818
20569 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20570 TYPE_OVERFLOW_UNDEFINED check.
20572 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20574 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20576 (vector_ae_<mode>_p): Likewise.
20577 (vector_nez_<mode>_p): Likewise.
20578 (vector_ne_v2di_p): Likewise.
20579 (vector_ae_v2di_p): Likewise.
20580 (vector_ne_<mode>_p): Likewise.
20581 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20583 (vsx_tsqrt<mode>2_fe): Likewise.
20585 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20588 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20590 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20592 PR rtl-optimization/79780
20593 * cprop.c (one_cprop_pass): When second and further conditional trap
20594 in a single basic block is turned into an unconditional trap, turn it
20595 into a deleted note to avoid RTL verification failures.
20597 2017-03-02 Richard Biener <rguenther@suse.de>
20599 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20601 2017-03-02 Richard Biener <rguenther@suse.de>
20603 PR tree-optimization/79345
20605 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20606 param and abort the walk, returning -1 if it is hit.
20607 (walk_aliased_vdefs): Take a limit param and pass it on.
20608 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20609 defaulting to 0 and return a signed int.
20610 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20611 (check_defs): New helper.
20612 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20613 about uninitialized memory.
20614 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20615 bogus uninitialized warning.
20616 (fixed_convert_from_real): Likewise.
20618 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20620 PR tree-optimization/66768
20621 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20622 iv_use if base object can't be determined.
20624 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20626 PR tree-optimization/79345
20627 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20628 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20629 (get_pattern_stats): Initialize it.
20630 * genemit.c (gen_expand): Verify match_scratch numbers come after
20631 match_operand/match_dup numbers.
20632 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20633 match_scratch numbers.
20634 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20636 * config/s390/s390.md (trunctdsd2): Likewise.
20638 2017-03-02 Richard Biener <rguenther@suse.de>
20640 * wide-int.h (wide_int_storage::operator=): Implement in terms
20643 2017-03-02 Richard Biener <rguenther@suse.de>
20645 PR tree-optimization/79777
20646 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20647 the to insert expression to sth existing.
20649 2017-03-01 Martin Sebor <msebor@redhat.com>
20651 PR middle-end/79692
20652 * gimple-ssa-sprintf.c
20653 (directive::known_width_and_precision): New function.
20654 (format_integer): Use it.
20655 (get_mpfr_format_length): Consider the full range of precision
20656 when computing %g output with the # flag. Set the likely byte
20657 count to 3 rather than 1 when precision is indeterminate.
20658 (format_floating): Correct the lower bound of precision.
20660 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20662 * doc/invoke.texi: Document default code model for 64-bit Linux.
20664 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20667 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20668 udiv rather than div since input pattern is unsigned.
20670 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20672 * config/i386/i386.c (print_reg): Warn for values of
20673 unsupported size in integer register.
20675 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20678 * config/rs6000/predicates.md (current_file_function_operand): Do
20679 not allow self calls to be local if the function is replaceable.
20681 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20684 * config/rs6000/altivec.h (vec_ctz and others): Change the
20685 preprocessor macro that controls conditional compilation from
20686 _ARCH_PWR9 to __POWER9_VECTOR__.
20687 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20688 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20689 control (instead of _ARCH_PWR9 control) so that template
20690 definition uses power9-specific function.
20691 (vec_any_eq): Likewise.
20692 (vec_all_ne): Change macro definition to use a power9-specific
20693 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20694 _ARCH_PWR9 control).
20695 (vec_any_eq) Likewise.
20696 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20697 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20698 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20699 support for xvcmpnedp instruction.
20700 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20701 macro expansion so that Power9 implementation of vec_all_ne does
20702 not use the AltiVec predicate framework.
20703 (VCMPNEH_P): Likewise.
20704 (VCMPNEW_P): Likewise.
20705 (VCMPNED_P): Likewise.
20706 (VCMPNEFP_P): Likewise.
20707 (VCMPNEDP_P): Likewise.
20708 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20709 implementation of vec_any_eq to not use AltiVec predicate
20711 (VCMPAEH_P): Likewise.
20712 (VCMPAEW_P): Likewise.
20713 (VCMPAED_P): Likewise.
20714 (VCMPAEFP_P): Likewise.
20715 (VCMPAEDP_P): Likewise.
20716 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20717 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20718 not use the AltiVec predicate framework.
20719 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20720 of vec_any_eq to not use AltiVec predicate framework.
20721 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20722 support for predefined __POWER9_VECTOR__ macro to indicate that
20723 Power9 instruction selection is enabled.
20724 (altivec_overloaded_builtins): Remove extraneous
20725 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20726 function argument types RS6000_BTI_bool_V16QI and
20727 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20728 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20729 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20730 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20731 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20732 Power9 for implementations of vec_cmpne. Change the signature for
20733 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20734 (representing vec_all_ne) to remove the previously described first
20735 argument of type RS6000_BTI_INTSI, as this was an artifact of
20736 reliance on the AltiVec predicate framework, which is no longer
20737 used in the implementation of these functions. Add
20738 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20739 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20740 since, unlike the AltiVec predicate framework implementation, we
20741 do not share function descriptors between vec_alle and vec_anyeq.
20742 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20743 set of modes that receive special treatment even when
20744 TARGET_P9_VECTOR is true. The special treatment emits code that
20745 does not depend on Power9 instructions.
20746 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20747 define_expand to not rely on AltiVec predicate framework.
20748 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20750 (vector_ne_v2di_p): Change this define_expand to not rely on
20751 AltiVec predicate framework.
20752 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20754 (vector_ne_<mode>_p): Change this define_expand to not rely on
20755 AltiVec predicate framework.
20756 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20758 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20759 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20760 define_insn pattern.
20761 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20762 define_insn pattern because the xvcmpne<VSs>. instruction is not
20764 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20765 instruction is not supported.
20767 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20769 * config/nvptx/nvptx.c: Include intl.h.
20771 2017-03-01 Martin Jambor <mjambor@suse.cz>
20774 * ipa-prop.h (ipa_bits): Removed field known.
20775 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20776 to pointers. Adjusted their comments to warn about their sharing.
20777 (ipcp_transformation_summary): Change bits to a vector of pointers.
20778 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20779 (ipa_get_ipa_bits_for_value): Declare.
20780 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20781 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20782 (ipa_bits_hash_table): Likewise.
20783 (ipa_vr_ggc_hash_traits): Likewise.
20784 (ipa_vr_hash_table): Likewise.
20785 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20786 being pointers and vr_known being removed.
20787 (ipa_set_jf_unknown): Likewise.
20788 (ipa_get_ipa_bits_for_value): New function.
20789 (ipa_set_jfunc_bits): Likewise.
20790 (ipa_get_value_range): New overloaded functions.
20791 (ipa_set_jfunc_vr): Likewise.
20792 (ipa_compute_jump_functions_for_edge): Use the above functions to
20793 construct bits and vr parts of jump functions.
20794 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20795 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20797 (ipcp_grow_transformations_if_necessary): Also allocate
20798 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20800 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20801 them. Fix too long lines.
20802 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20803 vr_known being removed.
20804 (ipa_read_jump_function): Use new setter functions to construct bits
20805 and vr parts of jump functions or set them to NULL.
20806 (write_ipcp_transformation_info): Adjust for bits being pointers.
20807 (read_ipcp_transformation_info): Likewise.
20808 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20810 Include gt-ipa-prop.h.
20811 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20813 (ipcp_store_bits_results): Likewise.
20814 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20815 Do not write to existing jump functions but use a temporary instead.
20817 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20820 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20821 attempt to use its first operand as BIT_FIELD_REF base.
20823 2017-03-01 Richard Biener <rguenther@suse.de>
20825 PR middle-end/79721
20826 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20827 interpolating formula in wrapping arithmetic.
20828 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20830 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20832 PR tree-optimization/79734
20833 * tree-vect-generic.c (expand_vector_condition): Optimize
20834 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20835 Handle VEC_COND_EXPR where comparison has different inner width from
20836 type's inner width.
20838 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20840 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20841 markup, and similar issues. Remove @opindex entries for things
20842 that aren't options. Add missing -mmpy-option entries.
20844 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20846 PR tree-optimization/79737
20847 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20848 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20849 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20850 instead of byte_size. Formatting fix.
20851 (shift_bytes_in_array_right): Formatting fix.
20853 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20856 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20857 condition on optimize for the leaf function test.
20859 2017-02-28 Martin Liska <mliska@suse.cz>
20862 * read-rtl-function.c (function_reader::handle_unknown_directive):
20863 Bail out when one uses -flto.
20865 2017-02-28 Martin Liska <mliska@suse.cz>
20867 * common.opt: Replace space with tabular for options of <number>
20869 * config/i386/i386.opt: Show <number> value for
20870 -mlarge-data-threshold.
20871 * opts.c (print_filtered_help): Do not display number in hexadecimal
20874 2017-02-28 Martin Liska <mliska@suse.cz>
20876 * common.opt: Fix --help=option -Q for options which are of
20879 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20881 * config/i386/i386.c (print_reg): Error out for values
20882 of 8-bit size in invalid integer register.
20884 2017-02-28 Martin Sebor <msebor@redhat.com>
20886 PR tree-optimization/79691
20887 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20889 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20892 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20893 gcc_unreachable with output_operand_lossage.
20895 2017-02-28 Richard Biener <rguenther@suse.de>
20897 PR tree-optimization/79740
20898 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20900 (visit_nary_op): Insert the nary into the hashtable if we
20901 pattern-matched sth.
20902 * tree-ssa-pre.c (eliminate_insert): Robustify.
20904 2017-02-28 Richard Biener <rguenther@suse.de>
20906 PR middle-end/79731
20907 * fold-const.c (decode_field_reference): Reject out-of-bound
20910 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20912 * config/i386/i386.c: Include intl.h.
20913 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20914 instead of just cond ? "..." : "...".
20915 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20916 * coverage.c (read_counts_file): Likewise.
20917 * omp-offload.c: Include intl.h.
20918 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20919 of just cond ? "..." : "...".
20920 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20921 of just cond ? "..." : "...".
20923 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20926 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20928 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20929 'tune for' CPU name.
20930 * config/arm/arm-cpu-data.h: Regenerated.
20932 2017-02-28 Richard Biener <rguenther@suse.de>
20934 PR tree-optimization/79732
20935 * tree-inline.c (expand_call_inline): Do not shadow var.
20937 2017-02-28 Richard Biener <rguenther@suse.de>
20939 PR tree-optimization/79723
20940 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20941 address-space properly.
20943 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20945 * doc/optinfo.texi (Optimization groups): Fix option used for
20947 * doc/invoke.texi (-fopt-info): Document "omp".
20948 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20949 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20950 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20951 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20952 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20954 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20956 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20957 OPTGROUP_OMP with "omp".
20959 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20962 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20963 for arithmetic shift of unsigned V2DI.
20965 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20967 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20968 arc/linux.h headers.
20969 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20970 (LINK_SPEC): Likewise.
20971 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20972 (EXTRA_SPECS): Likewise.
20973 (STARTFILE_SPEC): Likewise.
20974 (ENDFILE_SPEC): Likewise.
20975 (LIB_SPEC): Likewise.
20976 (TARGET_SDATA_DEFAULT): Likewise.
20977 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20978 (MULTILIB_DEFAULTS): Likewise.
20979 (DWARF2_UNWIND_INFO): Likewise.
20980 * config/arc/big.h: New file.
20981 * config/arc/elf.h: Likewise.
20982 * config/arc/linux.h: Likewise.
20983 * config/arc/t-uClibc: Remove.
20985 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20987 PR tree-optimization/77536
20988 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20989 (tree_transform_and_unroll_loop): Use above function to compute the
20990 estimated niter of unrolled loop and use it when scaling profile.
20991 Also use count info rather than frequency if it's non-zero.
20992 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20993 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20994 (vect_transform_loop): Call above function.
20996 2017-02-27 Richard Biener <rguenther@suse.de>
20998 PR tree-optimization/45397
20999 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21000 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21001 (visit_nary_op): Add pattern matching for CSEing sign-changed
21002 or truncated operations with wider ones.
21004 2017-02-27 Richard Biener <rguenther@suse.de>
21006 PR tree-optimization/79690
21007 * tree-vect-stmts.c (vectorizable_store): Use vector type
21008 built from the DR with address-space.
21010 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21012 * doc/invoke.texi (Optimize Options): Refine the description
21013 of asan-use-after-return.
21015 2017-02-25 Alan Modra <amodra@gmail.com>
21017 PR rtl-optimization/79584
21018 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21019 base, not ad->base_term, the reg within base. Remove assertion
21020 that ad->base == ad->base_term. Replace gen_int_mode using
21021 bogus mode with const0_rtx.
21023 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21025 PR middle-end/79396
21026 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21027 FMA_EXPR like tcc_binary or tcc_unary.
21029 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21032 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21034 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21035 (output_loc_operands): Handle DW_OP_call_ref and
21036 DW_OP_GNU_variable_value.
21037 (struct variable_value_struct): New type.
21038 (struct variable_value_hasher): Likewise.
21039 (variable_value_hash): New variable.
21040 (string_types): Remove.
21041 (copy_loc_descr): New function.
21042 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21043 (prepend_loc_descr_to_each): New function.
21044 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21045 instead of add_loc_descr_to_each if the first argument is single
21046 location list and the second has multiple.
21047 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21048 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21049 when looking for variable value which doesn't have other location info.
21050 (loc_list_from_tree): Formatting fix.
21051 (gen_array_type_die): Simplify DW_AT_string_length handling.
21052 (adjust_string_types): Remove.
21053 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21054 string_types. Call resolve_variable_values.
21055 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21056 (resolve_addr_in_expr): Likewise. Add A argument.
21057 (copy_deref_exprloc): Remove deref argument. Adjust for the
21058 original expression being DW_OP_GNU_variable_value with optionally
21059 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21060 optionally after it.
21061 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21062 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21063 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21064 (variable_value_hasher::hash, variable_value_hasher::equal): New
21066 (resolve_variable_value_in_expr, resolve_variable_value,
21067 resolve_variable_values, note_variable_value_in_expr,
21068 note_variable_value): New functions.
21069 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21072 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21075 * opts.h (handle_generated_option): Add GENERATED_P argument.
21076 * opts-common.c (handle_option): Adjust function comment.
21077 (handle_generated_option): Add GENERATED_P argument, pass it to
21079 (control_warning_option): Pass false to handle_generated_option
21081 * opts.c (maybe_default_option): Pass true to handle_generated_option
21083 * optc-gen.awk: Likewise.
21085 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21087 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21088 a REG, look at the REG it is a SUBREG of.
21089 (splitter for cmpeqsi_t): Ditto.
21091 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21093 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21094 the special USEs with the pattern of the insn, not the insn itself.
21096 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21099 * doc/invoke.texi: Document -mload-store-pairs.
21101 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21102 Sandra Loosemore <sandra@codesourcery.com>
21104 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21105 argument isn't a CONST_INT.
21106 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21107 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21108 (nios2_validate_compare): Bypass alternate compare logic if *op2
21109 is not a CONST_INT.
21110 (ldstwm_operation_p): Return false if first_base is not a REG or
21111 if first_offset is not a CONST_INT.
21113 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21115 * config/cris/cris.md: Use correct operand in a define_peephole2.
21117 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21119 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21121 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21123 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21124 this_insn if it is an INSN or JUMP_INSN.
21125 (force_offsettable): Look at base, not at addr.
21126 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21127 on things that aren't necessarily CONST_INTs.
21129 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21131 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21132 -mfpmath=sse is the default also for x86-32 targets with SSE2
21133 instruction set when @option{-ffast-math} is enabled
21135 2017-02-24 Jeff Law <law@redhat.com>
21137 PR rtl-optimizatoin/79286
21138 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21141 2017-02-24 Richard Biener <rguenther@suse.de>
21143 PR tree-optimization/79389
21144 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21147 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21149 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21150 function comment to reflect reality.
21151 (loop_exits_before_overflow): Fix typo in function description.
21153 2017-02-24 Richard Biener <rguenther@suse.de>
21155 PR tree-optimization/79389
21156 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21157 properly that a threading opportunity exists. Detect conditional
21158 copy/constant propagation opportunities.
21160 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21162 * config/visium/visium.md (type): Add trap.
21163 (b): New mode attribute.
21164 (*btst): Rename into...
21165 (*btst<mode>): ...this and adjust.
21166 (*cbranchsi4_btst_insn): Rename into...
21167 (*cbranch<mode>4_btst_insn): ...this and adjust.
21168 (trap): New define_insn.
21170 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21172 PR tree-optimization/79389
21173 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21174 (noce_reversed_cond_code): New function.
21175 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21176 reversed_comparison_code. Formatting fix.
21177 (noce_try_store_flag): Test rev_cond != NULL in addition to
21178 reversed_comparison_code.
21179 (noce_try_store_flag_constants): Likewise.
21180 (noce_try_store_flag_mask): Likewise.
21181 (noce_try_addcc): Use rev_cond if non-NULL instead of
21182 reversed_comparison_code.
21183 (noce_try_cmove_arith): Likewise. Formatting fixes.
21184 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21185 (noce_find_if_block): Initialize rev_cond.
21186 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21187 instead of false as last argument never attempt to reverse it
21190 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21192 PR tree-optimization/79663
21193 * tree-predcom.c (combine_chains): Process refs in reverse order
21194 only for ZERO length chains, and add explaining comment.
21196 2017-02-23 Jeff Law <law@redhat.com>
21198 PR tree-optimization/79578
21199 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21200 in call to operand_equal_p.
21202 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21205 * config/i386/cpuid.h: Fix another undefined behavior.
21207 2017-02-23 Richard Biener <rguenther@suse.de>
21209 PR tree-optimization/79683
21210 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21211 vector types for data-refs.
21213 2017-02-23 Martin Liska <mliska@suse.cz>
21215 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21217 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21219 PR middle-end/79665
21220 * internal-fn.c (get_range_pos_neg): Moved to ...
21221 * tree.c (get_range_pos_neg): ... here. No longer static.
21222 * tree.h (get_range_pos_neg): New prototype.
21223 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21224 are known to be in between 0 and signed maximum inclusive, try to
21225 expand both unsigned and signed divmod and use the cheaper one from
21228 2017-02-22 Jeff Law <law@redhat.com>
21230 PR tree-optimization/79578
21231 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21232 to compare base operands.
21234 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21237 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21238 gpc_reg_operand instead of fpr_reg_operand.
21240 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21242 * config/mips/mips.c (mips_return_in_memory): Force FP
21243 vector types to be returned in memory for o32 ABI.
21245 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21247 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21248 instead of DW_TAG_member for static data member declarations and don't
21249 set no_linkage_name for static inline data members.
21250 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21253 2017-02-22 Martin Liska <mliska@suse.cz>
21255 * doc/invoke.texi: Replace inequality signs with square brackets
21258 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21260 PR tree-optimization/68644
21261 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21263 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21266 * lra-constraints.c (simplify_operand_subreg): Handle
21267 WORD_REGISTER_OPERATIONS targets.
21269 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21272 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21273 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21274 elimination by swapping fld*.
21276 2017-02-22 Richard Biener <rguenther@suse.de>
21278 PR tree-optimization/79673
21279 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21280 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21281 irrelevant address-space qualifiers and avoiding a
21282 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21284 2017-02-22 Richard Biener <rguenther@suse.de>
21286 PR tree-optimization/79666
21287 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21288 to not symbolically negate if that may introduce undefined
21291 2017-02-22 Martin Liska <mliska@suse.cz>
21294 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21295 * data-streamer-out.c (streamer_write_gcov_count_stream):
21297 * value-prof.c (stream_out_histogram_value): Make assert more
21298 precise based on type of counter.
21300 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21303 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21304 Use nonimmediate_operand instead of memory_operand for operand 1.
21305 (float-extend standard_x87sse_constant_load splitter): Ditto.
21307 2017-02-21 Jeff Law <law@redhat.com>
21309 PR tree-optimization/79621
21310 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21311 blocks with edges to themselves.
21313 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21316 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21317 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21318 Use gimple_call_builtin_p.
21321 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21322 on temporarily removed DEBUG_INSNs.
21324 PR tree-optimization/79649
21325 * tree-loop-distribution.c (classify_partition): Give up on
21326 non-generic address space loads/stores.
21328 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21330 * doc/loop.texi (Loop manipulation): Remove nonexistent
21331 tree_ssa_loop_version from the documentation.
21332 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21334 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21337 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21338 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21339 * config/rs6000/rs6000.c: Include except.h.
21340 (rs6000_expand_split_stack_prologue): Call
21341 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21343 2017-02-21 Martin Jambor <mjambor@suse.cz>
21346 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21347 have been analyzed.
21349 2017-02-21 Martin Jambor <mjambor@suse.cz>
21351 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21352 for backward compatibility only.
21353 * doc/invoke.texi (Option Summary): Remove all references to
21354 -fipa-cp-alignment.
21356 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21360 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21362 * lra-constraints.c (curr_insn_transform): Handle
21363 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21365 2017-02-21 Martin Liska <mliska@suse.cz>
21367 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21369 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21371 PR translation/79638
21372 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21374 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21377 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21378 (arm_function_ok_for_sibcall): Return false for an indirect call by
21379 descriptor if all the argument registers are used.
21380 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21381 alignment of the function.
21383 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21385 PR tree-optimization/61441
21386 * simplify-rtx.c (simplify_const_unary_operation): For
21387 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21388 the sNaN unmodified.
21390 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21392 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21393 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21394 instead of SYSTEM_HEADER_DIR.
21396 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21397 Martin Liška <mliska@suse.cz>
21399 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21400 Fix typos and grammar, use active voice, and clarify.
21402 2017-02-20 Marek Polacek <polacek@redhat.com>
21404 PR middle-end/79537
21405 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21408 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21410 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21413 * config/i386/i386.c (ix86_expand_builtin): Handle
21414 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21415 ix86_builtins_isa[fcode].isa as a requirement of those
21416 flags and any other flag in the bitmask.
21417 (ix86_init_mmx_sse_builtins): Use 0 instead of
21418 ~OPTION_MASK_ISA_64BIT as mask.
21419 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21420 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21421 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21422 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21424 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21427 * lra-constraints.c (split_reg): Check requested split mode
21428 is supported by the register.
21430 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21432 * lra-constraints.c (simplify_operand_subreg): Remove early
21435 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21438 * lra-constraints.c (curr_insn_transform): Tighten condition
21439 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21441 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21444 * lra-constraints.c (curr_insn_transform): Handle
21445 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21447 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21450 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21452 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21454 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21457 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21460 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21462 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21465 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21467 * final.c (last_columnnum, override_columnnum): New variables.
21468 (final_start_function): Set last_columnnum, pass it to begin_prologue
21469 hook and pass 0 to dwarf2out_begin_prologue.
21470 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21471 to source_line debug hook.
21472 (notice_source_line): Compute last_columnnum and for debug_column_info
21473 return true on column changes.
21474 * debug.h (struct gcc_debug_hooks): Add column argument to
21475 source_line and begin_prologue hooks.
21476 (debug_nothing_int_charstar_int_bool): Remove prototype.
21477 (debug_nothing_int_int_charstar,
21478 debug_nothing_int_int_charstar_int_bool): New prototypes.
21479 (dwarf2out_begin_prologue): Add column argument.
21480 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21481 begin_prologue hooks.
21482 (debug_nothing_int_charstar_int_bool): Remove.
21483 (debug_nothing_int_int_charstar,
21484 debug_nothing_int_int_charstar_int_bool): New functions.
21485 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21486 through to dwarf2out_source_line.
21487 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21488 (dwarf2out_source_line): Add column argument, emit it if requested.
21489 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21491 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21492 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21493 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21494 through to dwarf2out_begin_prologue.
21495 (vmsdbgout_source_line): Add column argument, pass it through to
21496 dwarf2out_source_line.
21497 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21498 dbxout_source_line caller.
21499 (dbxout_source_line): Add column argument.
21501 * common.opt (gno-column-info, gcolumn-info): New options.
21502 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21503 (check_die): Also test for multiple DW_AT_decl_column attributes.
21504 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21505 DW_AT_decl_column if requested.
21506 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21508 (gen_variable_die): Likewise.
21509 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21510 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21513 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21514 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21515 (ix86_handle_option): Handle OPT_m3dnowa.
21516 * doc/invoke.texi (-m3dnowa): Document.
21517 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21518 -m3dnowa instead of -m3dnow -march=athlon.
21521 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21522 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21524 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21527 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21528 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21529 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21530 generator for vsx_xxpermdi_<mode>_be.
21531 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21532 force big-endian semantics.
21533 (vsx_xxpermdi_<mode>_be): New define_expand with same
21534 implementation as previous version of vsx_xxpermdi_<mode>.
21536 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21538 PR tree-optimization/79327
21539 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21540 variable, its initialization and use.
21542 2017-02-17 Julia Koval <julia.koval@intel.com>
21544 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21545 (OPTION_MASK_ISA_PKU_UNSET): New.
21546 (ix86_handle_option): Handle -mrdpid.
21547 * config/i386/cpuid.h (bit_RDPID): New.
21548 * config/i386/driver-i386.c (host_detect_local_cpu):
21549 Detect RDPID feature.
21550 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21551 * config/i386/i386-c.c (ix86_target_macros_internal):
21553 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21554 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21555 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21556 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21557 * config/i386/i386.md (define_insn "rdpid"): New.
21558 * config/i386/i386.opt Add -mrdpid.
21559 * config/i386/immintrin.h (_rdpid_u32): New.
21561 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21563 PR rtl-optimization/79541
21564 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21565 instead of transforming it into USE.
21567 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21569 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21570 If HONOR_SNANS (SFmode) force the input to a register.
21571 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21572 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21573 an frsp or similar insn.
21575 2017-02-17 Martin Liska <mliska@suse.cz>
21577 PR rtl-optimization/79577
21578 * params.def (selsched-max-sched-times): Increase minimum to 1.
21580 2017-02-17 Martin Liska <mliska@suse.cz>
21582 PR rtl-optimization/79574
21583 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21585 2017-02-17 Martin Liska <mliska@suse.cz>
21587 PR tree-optimization/79529
21588 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21589 ssa_defined_default_def_p to handle cases which are implicitly
21591 * tree-ssa.c (ssa_defined_default_def_p): New function.
21592 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21593 which are implicitly defined.
21594 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21596 2017-02-17 Richard Biener <rguenther@suse.de>
21598 PR middle-end/79576
21599 * params.def (max-ssa-name-query-depth): Limit to 10.
21601 2017-02-17 Richard Biener <rguenther@suse.de>
21603 PR tree-optimization/79552
21604 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21607 2017-02-17 Richard Biener <rguenther@suse.de>
21610 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21612 2017-02-17 Marek Polacek <polacek@redhat.com>
21614 PR middle-end/79536
21615 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21616 (fold_negate_expr): New wrapper.
21618 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21620 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21621 Correct terminology and de-emphasize pre-standard behavior.
21623 2017-02-16 Alan Modra <amodra@gmail.com>
21625 PR rtl-optimization/79286
21626 * ira.c (def_dominates_uses): New function.
21627 (update_equiv_regs): Don't create an equivalence for insns that
21628 may trap where the register def does not dominate the use.
21630 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21632 PR rtl-optimization/78127
21633 * lra.c (lra): Call lra_eliminate before finish the loop after
21636 2017-02-16 Richard Biener <rguenther@suse.de>
21638 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21640 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21641 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21642 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21643 (isl_val_int_from_wi): New function.
21644 (extract_affine_gmp): Rename to ...
21645 (extract_affine_wi): ... this, take a widest_int.
21646 (extract_affine_int): Just wrap extract_affine_wi.
21647 (add_param_constraints): Use isl_val_int_from_wi.
21648 (add_loop_constraints): Likewise, and extract_affine_wi.
21650 2017-02-15 Jeff Law <law@redhat.com>
21652 PR middle-end/79521
21653 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21654 ira_init_register_move_cost_if_necessary.
21656 2017-02-15 Martin Sebor <msebor@redhat.com>
21658 PR middle-end/32003
21659 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21660 removed in a prior commit.
21662 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21664 PR tree-optimization/79347
21665 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21666 counters during peeling.
21668 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21670 * Makefile.in (site.exp): Remove "set ISLVER".
21672 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21675 * real.c (real_from_integer): Call real_convert even for decimal.
21677 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21680 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21682 2017-02-14 Andrew Pinski <apinski@cavium.com>
21684 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21685 cores and change the partno/implementer to be correct.
21686 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21687 the 'B" as the implementer.
21688 * config/aarch64/aarch64-tune.md: Regenerate.
21690 2017-02-14 Carl Love <cel@us.ibm.com>
21692 * config/rs6000/rs6000.c: Add case statement entry to make the
21693 xvcvuxdsp built-in argument unsigned.
21694 * config/rs6000/vsx.md: Fix the source and return operand types so they
21695 match the instruction definitions from the ISA document. Fix typo
21696 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21699 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21702 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21703 member early_clobber_alts.
21704 * lra-lives.c (reg_early_clobber_p): New.
21705 (process_bb_lives): Use it.
21706 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21707 (debug_operand_data): Initialize early_clobber_alts.
21708 (setup_operand_alternative): Set up early_clobber_alts.
21709 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21710 alternatives to new_insn_reg.
21711 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21713 (lra_update_insn_regno_info): Pass the new arg.
21715 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21717 PR middle-end/79505
21718 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21719 (new_oacc_loop_raw): Don't clear already cleared fields.
21722 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21723 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21724 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21726 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21729 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21731 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21734 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21735 the extra instruction to the right place to store 128-bit constant
21738 2017-02-14 Martin Sebor <msebor@redhat.com>
21740 PR middle-end/79448
21741 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21742 warning for strings of unknown length.
21744 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21746 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21748 2017-02-14 Jeff Law <law@redhat.com>
21751 * ira-costs.c (scan_one_insn): Initialize register move costs
21752 for pseudos seen in USE/CLOBBER insns.
21754 PR tree-optimization/79095
21755 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21756 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21757 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21758 if the operands are known to be not equal, then the resulting range
21760 (intersect_ranges): If the new range is ~[0,0] and the old range is
21761 wide, then prefer ~[0,0].
21762 * tree-vrp.c (overflow_comparison_p_1): New function.
21763 (overflow_comparison_p): New function.
21764 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21765 if NAME is used in an overflow test.
21766 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21767 overflow check that can be expressed as an equality test, then adjust
21768 ops to be that equality test.
21770 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21772 * config/s390/s390-builtin-types.def: Remove flags argument.
21773 * config/s390/s390.c (s390_init_builtins): Likewise.
21775 2017-02-14 Martin Liska <mliska@suse.cz>
21777 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21778 vector. Fix trailing white spaces.
21780 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21782 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21785 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21787 PR rtl-optimization/68664
21788 * config/arm/arm.c (arm_sched_can_speculate_insn):
21789 New function. Declare prototype.
21790 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21792 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21794 PR rtl-optimization/68664
21795 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21797 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21799 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21801 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21802 max skip bytes for function, loop and jump.
21804 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21806 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21807 ABS_EXPR for gimple dump.
21809 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21812 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21814 PR tree-optimization/79408
21815 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21816 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21817 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21818 also if rhs1 is INTEGER_CST.
21820 2017-02-14 Richard Biener <rguenther@suse.de>
21822 PR middle-end/79432
21823 * tree-into-ssa.c (insert_phi_nodes): When the function can
21824 have abnormal edges rewrite SSA names with broken use-def
21825 dominance out of SSA and register them for PHI insertion.
21827 2017-02-13 Martin Sebor <msebor@redhat.com>
21829 PR middle-end/79496
21830 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21831 clearing info.nowrite flag when snprintf size argument is a range.
21833 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21835 * cprop.c (cprop_jump): Add missing space in string literal.
21836 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21837 (get_constraint_for_component_ref): Likewise.
21838 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21839 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21840 * lra-constraints.c (process_alt_operands): Likewise.
21841 * ipa-inline.c (inline_small_functions): Likewise.
21842 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21843 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21844 * trans-mem.c (diagnose_tm_1_op): Likewise.
21845 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21846 (grid_parallel_clauses_gridifiable): Likewise.
21848 * config/nvptx/mkoffload.c (process): Add space in between
21851 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21852 "MOD4_SSE_REGS" and "ALL_REGS".
21854 * spellcheck.c (test_data): Add , in between "foo" and "food".
21856 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21859 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21860 boundary crossing check and subsequent code generation agree.
21862 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21864 * config/aarch64/aarch64.c (has_memory_op): Delete.
21865 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21868 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21870 PR rtl-optimization/79388
21871 PR rtl-optimization/79450
21872 * combine.c (distribute_notes): When removing TEM_INSN for which
21873 corresponding dest has last value recorded, invalidate that last
21876 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21878 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21879 of explicit '@'. Add missing assembly comment marker on branch costs
21882 2017-02-13 Nathan Sidwell <nathan@acm.org>
21884 * gengtype-lex.l (<in_struct>): Add '/'.
21886 2017-02-13 Martin Liska <mliska@suse.cz>
21889 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21891 2017-02-13 Richard Biener <rguenther@suse.de>
21893 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21895 * configure: Re-generate.
21896 * config.in: Likewise.
21897 * graphite-dependences.c: Simplify as if
21898 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21899 * graphite-isl-ast-to-gimple.c: Likewise.
21900 * graphite-optimize-isl.c: Likewise.
21901 * graphite-poly.c: Likewise.
21902 * graphite-sese-to-poly.c: Likewise.
21903 * graphite.h: Likewise.
21904 * toplev.c: Include isl/version.h and use isl_version () for
21905 printing the ISL version.
21906 * doc/install.texi: Update ISL requirement.
21908 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21910 * doc/standards.texi (Standards): Update reference to
21913 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21915 * doc/extend.texi (Named Address Spaces): sourceware.org now
21917 * doc/install.texi (Binaries): Ditto.
21920 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21922 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21923 standard terminology "stringize"/"stringizing" throughout.
21924 * doc/cppinternals.texi: Likewise.
21926 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21928 * doc/extend.texi: Fix some spelling mistakes and typos.
21929 * doc/invoke.texi: Likewise.
21931 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21934 * params.def (inline-min-speedup) Change from 10 to 8.
21936 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21938 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21941 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21944 * ipa-inline-analysis.c (get_minimal_bb): New function.
21945 (record_modified): Use it.
21946 (remap_edge_change_prob): Handle also ancestor functions.
21948 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21950 * doc/contrib.texi (Contributors): Remove broken link into
21951 the Mauve CVS repository.
21953 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21955 PR middle-end/79454
21956 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21957 result computation whenever lhs doesn't have vector mode, not
21958 just when it has BLKmode.
21960 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21962 * doc/makefile.texi (profiledbootstrap): Refer to the
21963 installation instructions only in textual form.
21965 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21968 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21970 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21972 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21973 (Specific): Update mingw-w64 reference.
21975 (Specific): Remove broken link to Renesas RX processor.
21977 2017-02-10 Richard Biener <rguenther@suse.de>
21979 * toplev.c (process_options): Do not mention obsolete graphite
21980 options when printing sorry message about missing graphite support.
21981 Mention -floop-nest-optimize.
21983 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21985 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21986 (vtst_p16): Likewise.
21987 (vtstq_p8): Likewise.
21988 (vtstq_p16): Likewise.
21990 (vtstq_p64): Likewise.
21991 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21992 (vset_lane_p64): New.
21993 (vsetq_lane_p64): New.
21995 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21997 PR tree-optimization/79411
21998 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21999 stmt operands are SSA_NAMEs used in abnormal phis.
22000 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22003 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22006 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22009 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22011 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22013 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22015 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22019 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22020 not arbitrary TREE_CONSTANT.
22023 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22024 "omp declare target link" attribute unless is_global_var.
22025 * omp-offload.c (find_link_var_op): Likewise.
22027 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22028 Chung-Lin Tang <cltang@codesourcery.com>
22030 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22032 (gimplify_adjust_omp_clauses): Don't delete TILE.
22033 (gimplify_omp_for): Deal with TILE.
22034 * internal-fn.c (expand_GOACC_TILE): New function.
22035 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22037 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22038 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22040 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22041 avoid DIV for outermost collapse var.
22042 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22043 Remove out of date comments, fix whitespace.
22044 * omp-general.c (omp_extract_for_data): Deal with tiling.
22045 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22046 adjust OLF_DIM_BASE value.
22047 (struct omp_for_data): Add tiling field.
22048 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22049 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22050 for auto loops. Remove default auto determining, moved to
22051 oacc_loop_fixed_partitions.
22052 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22053 stmts, add e_mask field.
22054 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22055 (oacc_thread_numbers): Use oacc_dim_call.
22056 (oacc_xform_tile): New.
22057 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22058 (finish_oacc_loop): Adjust for ifns vector.
22059 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22060 add case for GOACC_TILE fns.
22061 (oacc_loop_xform_loop): Delete.
22062 (oacc_loop_process): Iterate over call list directly, and add
22063 handling for GOACC_TILE fns.
22064 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22066 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22067 vector partitioning to outer loops. Assign 2 partitions to loops
22068 when available. Add TILE handling.
22069 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22070 (execite_oacc_device_lower): Process GOACC_TILE fns,
22071 ignore unknown specs.
22072 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22073 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22074 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22076 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22078 * configure.ac (ACX_BUGURL): Update.
22079 * configure: Regenerate.
22081 2017-02-09 Richard Biener <rguenther@suse.de>
22083 PR tree-optimization/69823
22084 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22085 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22087 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22089 * config/arc/arc-c.def: Add __NPS400__ definition.
22090 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22091 (TARGET_NPS400): Define.
22093 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22095 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22097 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22098 pointer, arch_info.
22099 (arc_cpu_types): Fill the arch_info field with a pointer into the
22100 arc_arch_types table.
22101 (arc_selected_cpu): Declare.
22102 * config/arc/arc.c (arc_selected_cpu): Make global.
22103 (arc_selected_arch): Delete.
22104 (arc_base_cpu): Delete.
22105 (arc_override_options): Remove references to deleted variables,
22106 update access to arch information.
22107 (ARC_OPT): Update access to arch information.
22108 (ARC_OPTX): Likewise.
22109 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22110 (TARGET_ARC600): Update access to arch information.
22111 (TARGET_ARC601): Likewise.
22112 (TARGET_ARC700): Likewise.
22113 (TARGET_EM): Likewise.
22114 (TARGET_HS): Likewise.
22115 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22118 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22121 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22122 condition/operands for integer GE/LE/GEU/LEU operations.
22124 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22126 PR translation/79397
22127 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22130 2017-02-08 Martin Jambor <mjambor@suse.cz>
22133 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22134 whether allocation happened.
22135 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22136 nothing was allocated.
22138 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22140 PR tree-optimization/79408
22141 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22142 constant, but SSA_NAME with a known integer range, use the minimum
22143 of that range instead of op1 to determine if modulo can be replaced
22144 with its first operand.
22146 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22148 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22150 2017-02-08 Richard Biener <rguenther@suse.de>
22152 PR tree-optimization/71824
22153 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22154 Check all loops contained in the merged region.
22156 2017-02-07 Andrew Pinski <apinski@cavium.com>
22158 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22160 2017-02-07 Andrew Pinski <apinski@cavium.com>
22162 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22163 (thunderxt88): Likewise.
22164 (thunderxt81): Disable LSE and change v8.1 to v8.
22165 (thunderxt83): Likewise.
22167 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22168 Richard Biener <rguenther@suse.de>
22170 PR middle-end/79399
22171 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22172 type from int to size_t.
22173 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22175 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22177 PR rtl-optimization/79386
22178 * cprop.c (bypass_conditional_jumps): Initialize
22179 bypass_last_basic_block already before splitting bbs after
22180 unconditional traps...
22181 (bypass_conditional_jumps): ... rather than here.
22184 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22185 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22186 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22187 fix -masm=intel patterns.
22189 2017-02-07 Richard Biener <rguenther@suse.de>
22191 PR tree-optimization/79256
22192 PR middle-end/79278
22193 * builtins.c (get_object_alignment_2): Use min_align_of_type
22194 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22195 and ADJUST_FIELD_ALIGN.
22197 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22199 * doc/tm.texi: Regenerate.
22200 * stor-layout.c (layout_decl): Adjust.
22201 (update_alignment_for_field): Likewise.
22202 (place_field): Likewise.
22203 (min_align_of_type): Likewise.
22204 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22205 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22206 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22207 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22208 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22209 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22210 * config/i386/i386.c (x86_field_alignment): Likewise.
22211 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22212 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22213 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22214 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22215 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22216 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22220 2017-01-30 Richard Biener <rguenther@suse.de>
22222 PR tree-optimization/79256
22223 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22224 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22227 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22229 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22230 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22231 builtins to SImode and emit a zero-extend, if necessary.
22233 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22235 * docs/invoke.texi (RISC-V Options): Alphabetize.
22237 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22239 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22242 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22244 * config/riscv/riscv.c: New file.
22245 * common/config/riscv/riscv-common.c: Likewise.
22246 * config.gcc: Likewise.
22247 * config/riscv/constraints.md: Likewise.
22248 * config/riscv/elf.h: Likewise.
22249 * config/riscv/generic.md: Likewise.
22250 * config/riscv/linux.h: Likewise.
22251 * config/riscv/multilib-generator: Likewise.
22252 * config/riscv/peephole.md: Likewise.
22253 * config/riscv/pic.md: Likewise.
22254 * config/riscv/predicates.md: Likewise.
22255 * config/riscv/riscv-builtins.c: Likewise.
22256 * config/riscv/riscv-c.c: Likewise.
22257 * config/riscv/riscv-ftypes.def: Likewise.
22258 * config/riscv/riscv-modes.def: Likewise.
22259 * config/riscv/riscv-opts.h: Likewise.
22260 * config/riscv/riscv-protos.h: Likewise.
22261 * config/riscv/riscv.h: Likewise.
22262 * config/riscv/riscv.md: Likewise.
22263 * config/riscv/riscv.opt: Likewise.
22264 * config/riscv/sync.md: Likewise.
22265 * config/riscv/t-elf-multilib: Likewise.
22266 * config/riscv/t-linux: Likewise.
22267 * config/riscv/t-linux-multilib: Likewise.
22268 * config/riscv/t-riscv: Likewise.
22269 * configure.ac: Likewise.
22270 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22271 Waterman as RISC-V maintainers.
22272 * doc/install.texi: Add RISC-V entries.
22273 * doc/invoke.texi: Add RISC-V options section.
22274 * doc/md.texi: Add RISC-V constraints section.
22275 * configure: Regenerated.
22277 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22280 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22281 false values to be constant vectors with all 0 or all 1 bits set.
22282 (vcondu<mode><mode>): Likewise.
22283 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22285 (fpmask_comparison_operator): Update comment.
22286 (vecint_comparison_operator): New predicate.
22287 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22288 vector conditionals when the true and false values are constant
22289 vectors with all 0 bits or all 1 bits set.
22291 2017-02-06 Martin Sebor <msebor@redhat.com>
22293 PR tree-optimization/79376
22294 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22296 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22298 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22299 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22300 to simplify split condition.
22302 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22304 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22305 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22308 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22310 PR rtl-optimization/68664
22311 * target.def (can_speculate_insn): New hook.
22312 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22313 * doc/tm.texi: Regenerate.
22314 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22315 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22316 (rs6000_sched_can_speculate_insn): New function.
22318 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22320 PR tree-optimization/79284
22321 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22322 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22323 vectorizable_mask_load_store, vectorizable_operation,
22324 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22325 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22326 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22327 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22328 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22329 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22330 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22331 is_gimple_assign (stmt). Replace another such test with
22332 is_gimple_assign (stmt).
22334 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22337 * config/avr/avr.c (rtl-iter.h): Include it.
22338 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22339 (avr_legitimate_combined_insn): ...and implementation.
22341 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22343 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22344 * config/s390/s390.c (s390_const_operand_ok)
22345 (s390_canonicalize_comparison, s390_extract_part)
22346 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22347 (s390_contiguous_bitmask_p, s390_rtx_costs)
22348 (legitimize_pic_address): Likewise.
22349 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22350 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22351 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22352 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22353 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22355 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22357 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22358 REGNO($0) == REGNO($1).
22360 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22362 * config/s390/linux.h(SIZE_TYPE): Add comment.
22364 2017-02-06 Julian Brown <julian@codesourcery.com>
22365 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22366 Virendra Pathak <virendra.pathak@broadcom.com>
22368 * config/aarch64/aarch64-cores.def: Change the scheduler
22370 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22371 * config/aarch64/thunderx2t99.md: New file.
22373 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22375 * doc/standards.texi (Go Language): Update link to language
22378 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22380 * tree-eh.c (lower_resx): Sanitize profile.
22381 (cleanup_empty_eh_move_lp): Likewise.
22383 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22386 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22388 * cfgloopmanip.h (loop_version): Update prototype.
22389 * modulo-sched.c (sms_schedule): Update call of loop_version.
22390 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22391 * tree-parloops.c (gen_parallel_loop): Likewise.
22392 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22393 * tree-ssa-loop-split.c (split_loop): Likewise.
22394 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22395 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22397 2017-02-05 Martin Liska <mliska@suse.cz>
22400 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22402 (print_operand_address): Initialize a struct to zero.
22404 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22406 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22407 garbage collector only in textual form.
22409 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22411 * doc/extend.texi (x86 specific memory model extensions for
22412 transactional memory): Simplify a phrase.
22414 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22417 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22418 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22419 (atomic_storedi_1): Likewise.
22421 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22423 PR tree-optimization/79338
22424 * tree-parloops.c (gather_scalar_reductions): Don't call
22425 vect_analyze_loop_form for loop->inner before destroying loop's
22428 2017-02-03 Martin Sebor <msebor@redhat.com>
22430 PR tree-optimization/79327
22431 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22432 when precision has resulted in leading zeros.
22433 (format_integer): Adjust the likely counter to assume an unknown
22434 argument that may be zero is non-zero.
22436 2017-02-03 Jason Merrill <jason@redhat.com>
22439 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22440 avoid copying non-taken branch.
22442 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22444 PR tree-optimization/79340
22445 * tree-vect-loop.c (vectorizable_reduction): Release
22446 vec_defs elements after safe_splicing them into other vectors.
22449 PR tree-optimization/79327
22450 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22451 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22453 (format_integer): Use wide_int_to_tree instead of build_int_cst
22454 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22455 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22456 of shortest and longest sequence.
22458 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22460 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22461 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22463 2017-02-03 Walter Lee <walt@tilera.com>
22466 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22467 after initial stackframe link reg save.
22468 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22470 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22473 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22474 wu for stxssp alternative.
22476 2017-02-03 Martin Sebor <msebor@redhat.com>
22478 PR tree-optimization/79352
22479 * gimple-fold.c (get_range_strlen): Add argument.
22480 (get_range_strlen): Change return type to bool.
22481 (get_maxval_strlen): Pass in a dummy argument.
22482 * gimple-fold.h (get_range_strlen): Change return type to bool.
22483 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22484 * tree.h (array_at_struct_end_p): Add argument.
22485 * tree.c (array_at_struct_end_p): Handle it.
22487 2017-02-03 Martin Liska <mliska@suse.cz>
22490 * multiple_target.c (create_dispatcher_calls): Redirect edge
22491 from a caller of a dispatcher.
22492 (expand_target_clones): Make the clones local.
22493 (ipa_target_clone): Do both target clones and resolvers.
22494 (ipa_dispatcher_calls): Remove the pass.
22495 (pass_dispatcher_calls::gate): Likewise.
22496 (make_pass_dispatcher_calls): Likewise.
22497 * passes.def (pass_target_clone): Put as very first IPA early
22500 2017-02-03 Martin Liska <mliska@suse.cz>
22502 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22503 in case of a function with ifunc attribute.
22505 2017-02-03 Martin Liska <mliska@suse.cz>
22507 * cgraph.c (cgraph_node::dump): Dump function version info.
22508 * symtab.c (symtab_node::dump_base): Add missing new line.
22510 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22512 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22513 (ifcombine_ifandif): Use it.
22515 2017-02-03 Martin Liska <mliska@suse.cz>
22517 * doc/invoke.texi: Document default value for
22518 use-after-scope-direct-emission-threshold.
22520 2017-02-03 Martin Liska <mliska@suse.cz>
22522 PR tree-optimization/79339
22523 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22524 (format_floating): Likewise.
22526 2017-02-03 Martin Liska <mliska@suse.cz>
22529 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22531 (ipa_node_params_t::remove): Likewise.
22532 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22533 initialization from removed ipa_node_params_t::insert.
22534 (ipa_node_params::~ipa_node_params): Move from removed
22535 ipa_node_params_t::release.
22536 * symbol-summary.h (symbol_summary::m_released): New member.
22537 Do not release a summary twice. Do not allow to call finalizer
22538 for types of a summary that live in GGC memory.
22540 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22542 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22545 2017-02-02 Martin Sebor <msebor@redhat.com>
22547 PR middle-end/79275
22548 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22549 (format_string): Tighten up the range of output for non-constant
22550 strings and correct the expected range for wide non-constant strings.
22552 2017-02-02 Martin Sebor <msebor@redhat.com>
22554 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22556 PR middle-end/32003
22557 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22559 (-fdump-tree-@var): Add to index and document how to come up
22560 with pass-specific option and dump file names.
22561 (-fdump-passes): Clarify where to look for output.
22563 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22565 PR middle-end/77445
22566 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22567 statistics of the analyzed path; allow threading for speed when
22568 any of BBs along the path are optimized for speed.
22570 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22572 PR middle-end/78468
22573 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22574 settings of the virtual registers.
22577 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22579 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22580 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22583 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22585 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22586 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22588 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22590 * config/s390/s390.md: Add missing comments with the expanded
22592 * config/s390/vector.md: Likewise.
22593 * config/s390/vx-builtins.md: Likewise.
22595 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22598 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22599 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22600 conditions on a single line.
22602 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22604 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22605 __S390_VX__ to __VX__.
22607 2017-02-01 Andrew Pinski <apinski@cavium.com>
22609 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22610 stmt_info to record_stmt_cost.
22611 (vect_get_known_peeling_cost): Pass stmt_info if known to
22613 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22614 cpu_vector_cost field into
22615 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22616 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22617 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22618 splitting of scalar_stmt_cost and vec_stmt_cost.
22619 (thunderx_vector_cost): Likewise.
22620 (cortexa57_vector_cost): LIkewise.
22621 (exynosm1_vector_cost): Likewise.
22622 (xgene1_vector_cost): Likewise.
22623 (thunderx2t99_vector_cost): Improve after the splitting of the two
22625 (aarch64_builtin_vectorization_cost): Update for the splitting of
22626 scalar_stmt_cost and vec_stmt_cost.
22628 2017-02-01 Torvald Riegel <triegel@redhat.com>
22629 Richard Henderson <rth@redhat.com>
22631 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22632 conditional on existance of a fast atomic load.
22633 * optabs-query.c (can_atomic_load_p): New function.
22634 * optabs-query.h (can_atomic_load_p): Declare it.
22635 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22636 no fast atomic load is available for the particular size of access.
22637 (expand_atomic_compare_and_swap): Likewise.
22638 (expand_atomic_load): Likewise.
22639 (expand_atomic_store): Likewise.
22640 (expand_atomic_fetch_op): Likewise.
22641 * testsuite/lib/target-supports.exp
22642 (check_effective_target_sync_int_128): Remove x86 because it provides
22643 no fast atomic load.
22644 (check_effective_target_sync_int_128_runtime): Likewise.
22646 2017-02-01 Richard Biener <rguenther@suse.de>
22648 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22649 (graphite_transform_loops): Provide opt-info for optimized nests.
22650 * tree-parloop.c (parallelize_loops): Provide opt-info for
22651 parallelized loops.
22653 2017-02-01 Richard Biener <rguenther@suse.de>
22655 PR middle-end/79315
22656 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22657 was not set before.
22659 2017-02-01 Richard Biener <rguenther@suse.de>
22661 PR tree-optimization/71824
22662 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22663 Verify the loops are valid in the merged SESE region.
22664 (scop_detection::can_represent_loop_1): Check analyzing the
22665 evolution of the number of iterations in the region succeeds.
22667 2017-01-31 Ian Lance Taylor <iant@golang.org>
22669 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22670 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22672 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22674 PR preprocessor/79210
22675 * input.c (get_substring_ranges_for_loc): Replace line_width
22676 assertion with error-handling.
22678 2017-01-31 Richard Biener <rguenther@suse.de>
22680 PR tree-optimization/77318
22681 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22682 (create_pw_aff_from_tree): Take loop parameter.
22683 (add_condition_to_pbb): Pass loop of the condition to
22684 create_pw_aff_from_tree.
22686 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22688 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22689 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22691 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22695 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22697 (convert_int_to_float128): Likewise.
22698 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22699 (convert_int_to_float128): Likewise.
22700 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22701 (UNSPEC_IEEE128_CONVERT): Likewise.
22702 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22703 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22704 Use local variables for IBM extended format.
22705 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22706 (fix_trunc<mode>si2_fprs): Likewise.
22707 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22708 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22709 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22710 to know that we can now have integers of all sizes in vector
22712 (fix<uns>_<mode>di2_hw): Likewise.
22713 (float<uns>_<mode>si2_hw): Likewise.
22714 (fix_<mode>si2_hw): Likewise.
22715 (fixuns_<mode>si2_hw): Likewise.
22716 (float<uns>_<mode>di2_hw): Likewise.
22717 (float_<mode>di2_hw): Likewise.
22718 (float_<mode>si2_hw): Likewise.
22719 (floatuns_<mode>di2_hw): Likewise.
22720 (floatuns_<mode>si2_hw): Likewise.
22721 (xscvqp<su>wz_<mode>): Delete, no longer used.
22722 (xscvqp<su>dz_<mode>): Likewise.
22723 (xscv<su>dqp_<mode>): Likewise.
22724 (ieee128_mfvsrd_64bit): Likewise.
22725 (ieee128_mfvsrd_32bit): Likewise.
22726 (ieee128_mfvsrwz): Likewise.
22727 (ieee128_mtvsrw): Likewise.
22728 (ieee128_mtvsrd_64bit): Likewise.
22729 (ieee128_mtvsrd_32bit): Likewise.
22731 2017-01-31 Martin Liska <mliska@suse.cz>
22734 * ipa-prop.c (ipa_free_all_node_params): Call release method
22735 instead of ~sumbol_summary to not to trigger double times
22738 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22740 PR tree-optimization/71691
22741 * bitmap.h (class auto_bitmap): New.
22742 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22743 is_maybe_undefined instead of ssa_undefined_value_p.
22745 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22747 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22748 __S390_ARCH_LEVEL__ to __ARCH__.
22750 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22752 PR tree-optimization/79267
22753 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22754 if should_remove_lhs_p is true.
22756 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22759 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22760 (add_alignment_attribute): New.
22761 (base_type_die): Add alignment attribute.
22762 (subrange_type_die): Likewise.
22763 (modified_type_die): Likewise.
22764 (gen_array_type_die): Likewise.
22765 (gen_descr_array_type_die: Likewise.
22766 (gen_enumeration_type_die): Likewise.
22767 (gen_subprogram_die): Likewise.
22768 (gen_variable_die): Likewise.
22769 (gen_field_die): Likewise.
22770 (gen_ptr_to_mbr_type_die): Likewise.
22771 (gen_struct_or_union_type_die): Likewise.
22772 (gen_subroutine_type_die): Likewise.
22773 (gen_typedef_die): Likewise.
22774 (base_type_cmp): Compare alignment attribute.
22776 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22779 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22780 (setb_unsigned) New pattern for setb with CCUNS.
22781 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22782 subfc./subfe sequence to avoid overflow problems. Generate a
22783 shorter sequence with cmpld/setb for power9.
22784 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22785 for generating subfc. instruction.
22786 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22787 now uses this instruction.
22789 2017-01-30 Ian Lance Taylor <iant@google.com>
22792 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22793 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22795 2017-01-30 Martin Sebor <msebor@redhat.com>
22797 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22798 Move constant to the right of a relational operator.
22799 (get_mpfr_format_length, format_character, format_string): Ditto.
22800 (should_warn_p, maybe_warn): Same.
22802 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22804 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22807 * asan.c (get_translation_unit_decl): Remove function.
22808 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22810 2017-01-30 Martin Liska <mliska@suse.cz>
22812 PR gcov-profile/79259
22813 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22814 -fprofile-generate.
22816 2017-01-30 Martin Liska <mliska@suse.cz>
22819 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22820 Initialize variables with NULL value.
22822 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22825 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22827 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22829 2017-01-30 Richard Biener <rguenther@suse.de>
22831 PR tree-optimization/79276
22832 * tree-vrp.c (process_assert_insertions): Properly adjust common
22833 when removing a duplicate.
22835 * gcc.dg/torture/pr79276.c: New testcase.
22837 2017-01-30 Richard Biener <rguenther@suse.de>
22839 PR tree-optimization/79256
22840 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22841 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22843 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22845 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22848 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22849 ("*r<noxa>sbg_<mode>_sll_bitmask")
22850 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22851 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22852 Use contiguous_bitmask_nowrap_operand.
22854 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22857 * config/rs6000/altivec.h (vec_xl): Revise #define.
22858 (vec_xst): Likewise.
22860 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22862 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22864 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22866 PR rtl-optimization/79194
22867 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22868 traps before call to bypass_conditional_jumps.
22870 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22872 PR tree-optimization/71374
22873 * lra-constraints.c (check_conflict_input_operands): New.
22874 (match_reload): Use it.
22876 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22879 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22880 account to calculate conflict_set.
22882 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22884 PR rtl-optimization/78559
22885 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22886 other_insn in combine.
22888 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22890 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22891 uint16_type_node for BT_UINT16.
22893 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22895 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22896 "RTL Tests" to menu.
22897 (GIMPLE Tests): New node.
22898 (RTL Tests): New node.
22900 2017-01-27 Richard Biener <rguenther@suse.de>
22902 PR tree-optimization/79245
22903 * tree-loop-distribution.c (distribute_loop): Apply cost
22904 modeling also to detected patterns.
22906 2017-01-27 Richard Biener <rguenther@suse.de>
22908 PR tree-optimization/71433
22909 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22910 (compare_assert_loc): New function.
22911 (process_assert_insertions): Sort and optimize assert locations
22912 to remove duplicates and push down identical assertions on
22913 edges to their destination block.
22915 2017-01-27 Richard Biener <rguenther@suse.de>
22917 PR tree-optimization/79244
22918 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22919 out SSA names even if abnormal.
22921 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22923 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22924 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22925 instead of MPFR_RNDN.
22927 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22930 * arm.c (arm_option_override): Don't call build_target_option_node
22931 until after doing all option overrides.
22932 (arm_valid_target_attribute_tree): Likewise.
22934 2017-01-27 Martin Liska <mliska@suse.cz>
22936 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22937 for {cd}tors and C++ {cd}tors.
22939 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22941 * config/s390/s390.md ("*setmem_long_and")
22942 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22944 2017-01-26 Martin Sebor <msebor@redhat.com>
22946 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22949 2017-01-26 Martin Sebor <msebor@redhat.com>
22951 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22952 HAVE_DFmode before using XFmode or DFmode.
22953 (parse_directive): Avoid using the z length modifier to avoid
22954 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22956 PR middle-end/78703
22957 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22958 to accept adjustment as an array.
22959 (get_int_range): New function.
22960 (struct directive): Make width and prec arrays.
22961 (directive::set_width, directive::set_precision): Call get_int_range.
22962 (format_integer, format_floating): Handle width and precision ranges.
22963 (format_string, parse_directive): Same.
22965 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22968 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22969 comdat_type_p set, just clone them, but keep the children in the
22973 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22974 which have direct callers with -fvar-tracking-assignments enabled
22976 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22977 inside of type units.
22979 2017-01-26 Martin Sebor <msebor@redhat.com>
22981 PR middle-end/78703
22982 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22984 (struct format_result): Replace number_chars, number_chars_min,
22985 and number_chars_max with a single member of struct result_range.
22987 (format_result::operator+=): Adjust.
22988 (struct fmtresult): Remove bounded. Handle likely and unlikely
22990 (fmtresult::adjust_for_width_or_precision): New function.
22991 (fmtresult:type_max_digits): New function.
22992 (bytes_remaining): Handle likely and unlikely counters.
22993 (min_bytes_remaining): Remove.
22994 (format_percent): Simplify.
22995 (format_integer, format_floating): Set likely and unlikely counters.
22996 (get_string_length, format_character, format_string): Same.
22997 (format_plain, should_warn_p): New function.
22998 (maybe_warn): Call should_warn_p. Update diagnostic messages
22999 and handle those for all directives, including plain strings.
23000 (format_directive): Handle likely and unlikely counters.
23001 Remove unnecessary quoting from diagnostics. Add an informational
23003 (add_bytes): Remove.
23004 (pass_sprintf_length::compute_format_length): Simplify.
23005 (try_substitute_return_value): Handle likely and unlikely counters.
23007 2017-01-26 Carl Love <cel@us.ibm.com>
23009 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23010 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23012 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23015 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23016 endianess for subregs into account.
23017 * lra-constraints.c (lra_constraints): Do risky transformations
23018 always on the first iteration.
23019 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23020 last_call_used_reg_set.
23021 (process_bb_lives): Define and use last_call_used_reg_set.
23022 * lra.c (lra): Always continue after lra_constraints on the first
23025 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23027 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23029 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23031 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23033 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23034 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23035 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23036 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23037 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23038 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23039 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23040 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23041 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23043 2017-01-26 Marek Polacek <polacek@redhat.com>
23046 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23047 for the third operand.
23049 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23051 PR middle-end/79236
23052 * omp-low.c (struct omp_context): Add simt_stmt field.
23053 (scan_omp_for): Return omp_context *.
23054 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23055 context to the _simt_ SIMD stmt.
23056 (lower_omp_for): For combined SIMD with sibling _simt_
23057 SIMD, make sure to use the same decls in _looptemp_
23058 clauses as in the sibling.
23060 2017-01-26 David Sherwood <david.sherwood@arm.com>
23062 PR middle-end/79212
23063 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23066 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23069 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23070 emit fld b; fld a; if possible.
23072 * brig-builtins.def: Update copyright years.
23073 * config/arm/arm_acle_builtins.def: Update copyright years.
23075 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23078 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23079 constraint instead of o for the stxsd instruction.
23081 2017-01-25 Carl Love <cel@us.ibm.com>
23083 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23084 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23086 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23088 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23090 2017-01-25 Richard Biener <rguenther@suse.de>
23092 PR tree-optimization/69264
23093 * target.def (vector_alignment_reachable): Improve documentation.
23094 * doc/tm.texi: Regenerate.
23095 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23097 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23098 earlier changes with respect to TYPE_USER_ALIGN.
23099 (vector_alignment_reachable_p): Likewise. Improve dumping.
23101 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23104 * config/arm/arm.md (xordi3): Force constant operand into a register
23107 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23109 * doc/invoke.texi (-fstore-merging): Correct default optimization
23110 levels at which it is enabled.
23111 (-O): Move -fstore-merging from list to...
23114 2017-01-25 Richard Biener <rguenther@suse.de>
23117 * omp-expand.c: Include debug.h.
23118 (expand_omp_taskreg): Make sure to generate early debug before
23119 outlining anything from a function.
23120 (expand_omp_target): Likewise.
23121 (grid_expand_target_grid_body): Likewise.
23123 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23126 * asan.c (get_translation_unit_decl): New function.
23127 (asan_add_global): Extract modules file name from globals
23128 TRANSLATION_UNIT_DECL name.
23130 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23133 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23134 for long calls with APCS frame and VFP.
23136 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23138 * cfg.c (original_copy_tables_initialized_p): New function.
23139 * cfg.h (original_copy_tables_initialized_p): New decl.
23140 * cfgrtl.c (relink_block_chain): Guard the call to
23141 free_original_copy_tables with a call to
23142 original_copy_tables_initialized_p.
23143 * cgraph.h (symtab_node::native_rtl_p): New decl.
23144 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23145 (symtab_node::needed_p): Don't assert for early assembly output
23146 for __RTL functions.
23147 (cgraph_node::finalize_function): Set "force_output" for __RTL
23149 (cgraph_node::analyze): Bail out early for __RTL functions.
23150 (analyze_functions): Update assertion to support __RTL functions.
23151 (cgraph_node::expand): Bail out early for __RTL functions.
23152 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23154 * function.h (struct function): Update comment for field
23156 * gimple-expr.c: Include "tree-pass.h".
23157 (gimple_has_body_p): Return false for __RTL functions.
23158 * Makefile.in (OBJS): Add run-rtl-passes.o.
23159 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23161 (gcc::pass_manager::get_clean_slate): New accessor.
23162 * passes.c: Include "insn-addr.h".
23163 (should_skip_pass_p): Add logging. Update logic for running
23164 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23165 property-provider override so it is only done for gimple passes.
23167 (skip_pass): New function.
23168 (execute_one_pass): Call skip_pass when skipping passes.
23169 * read-md.c (md_reader::read_char): Support filtering
23170 the input to a subset of line numbers.
23171 (md_reader::md_reader): Initialize fields
23172 m_first_line and m_last_line.
23173 (md_reader::read_file_fragment): New function.
23174 * read-md.h (md_reader::read_file_fragment): New decl.
23175 (md_reader::m_first_line): New field.
23176 (md_reader::m_last_line): New field.
23177 * read-rtl-function.c (function_reader::create_function): Only
23178 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23179 curr_properties. Set DECL_INITIAL to a dummy block.
23180 (read_rtl_function_body_from_file_range): New function.
23181 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23183 * run-rtl-passes.c: New file.
23184 * run-rtl-passes.h: New file.
23186 2017-01-24 Jeff Law <law@redhat.com>
23188 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23191 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23193 PR tree-optimization/79159
23194 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23195 (record_nonwrapping_iv): Improve boundary using above function if no
23196 value range information.
23198 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23199 Martin Jambor <mjambor@suse.cz>
23201 * brig-builtins.def: New file.
23202 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23203 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23204 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23205 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23206 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23207 * builtin-types.def (BT_INT8): New.
23208 (BT_INT16): Likewise.
23209 (BT_UINT8): Likewise.
23210 (BT_UINT16): Likewise.
23211 (BT_FN_ULONG): Likewise.
23212 (BT_FN_UINT_INT): Likewise.
23213 (BT_FN_UINT_ULONG): Likewise.
23214 (BT_FN_UINT_LONG): Likewise.
23215 (BT_FN_UINT_PTR): Likewise.
23216 (BT_FN_ULONG_PTR): Likewise.
23217 (BT_FN_INT8_FLOAT): Likewise.
23218 (BT_FN_INT16_FLOAT): Likewise.
23219 (BT_FN_UINT32_FLOAT): Likewise.
23220 (BT_FN_UINT16_FLOAT): Likewise.
23221 (BT_FN_UINT8_FLOAT): Likewise.
23222 (BT_FN_UINT64_FLOAT): Likewise.
23223 (BT_FN_UINT16_UINT32): Likewise.
23224 (BT_FN_UINT32_UINT16): Likewise.
23225 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23226 (BT_FN_INT_PTR_INT): Likewise.
23227 (BT_FN_UINT_PTR_UINT): Likewise.
23228 (BT_FN_LONG_PTR_LONG): Likewise.
23229 (BT_FN_ULONG_PTR_ULONG): Likewise.
23230 (BT_FN_VOID_UINT64_UINT64): Likewise.
23231 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23232 (BT_FN_INT8_INT8_INT8): Likewise.
23233 (BT_FN_INT16_INT16_INT16): Likewise.
23234 (BT_FN_INT_INT_INT): Likewise.
23235 (BT_FN_UINT_FLOAT_UINT): Likewise.
23236 (BT_FN_FLOAT_UINT_UINT): Likewise.
23237 (BT_FN_ULONG_UINT_UINT): Likewise.
23238 (BT_FN_ULONG_UINT_PTR): Likewise.
23239 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23240 (BT_FN_UINT_UINT_UINT): Likewise.
23241 (BT_FN_VOID_UINT_PTR): Likewise.
23242 (BT_FN_UINT_UINT_PTR: Likewise.
23243 (BT_FN_UINT32_UINT64_PTR): Likewise.
23244 (BT_FN_INT_INT_UINT_UINT): Likewise.
23245 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23246 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23247 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23248 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23249 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23250 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23251 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23252 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23253 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23254 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23255 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23256 * doc/frontends.texi: List BRIG FE.
23257 * doc/install.texi (Testing): Add BRIG tesring requirements.
23258 * doc/invoke.texi (Overall Options): Mention BRIG.
23259 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23261 2017-01-24 Richard Biener <rguenther@suse.de>
23263 PR translation/79208
23264 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23266 2017-01-24 Martin Jambor <mjambor@suse.cz>
23269 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23270 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23271 and known_contexts.
23273 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23275 PR middle-end/79123
23276 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23277 casts from signed to unsigned really don't have a range.
23279 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23281 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23282 GMP_RNDx for compatiblity.
23284 2017-01-24 Martin Liska <mliska@suse.cz>
23287 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23288 that would prevent us to call alloca with -1 as argument.
23290 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23292 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23293 Avoid -Wformat-security warning.
23295 2017-01-23 Andrew Pinski <apinski@cavium.com>
23297 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23300 2017-01-23 Martin Sebor <msebor@redhat.com>
23302 PR middle-end/78703
23303 * gimple-ssa-sprintf.c (warn_level): New global.
23304 (format_integer): Use it here and throughout the rest of the file.
23305 Use the same switch to compute sign as base.
23306 (maybe_warn): New function.
23307 (format_directive): Factor out warnings into maybe_warn.
23308 Add debugging output. Use warn_level.
23309 (add_bytes): Use warn_level.
23310 (pass_sprintf_length::compute_format_length): Add debugging output.
23311 (try_substitute_return_value): Same.
23312 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23314 PR middle-end/78703
23315 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23316 (struct fmtresult, format_integer, format_floating): Adjust.
23317 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23318 (get_string_length, format_string,format_directive): Same.
23319 (pass_sprintf_length::compute_format_length): Same.
23320 (try_substitute_return_value): Simplify slightly.
23322 PR middle-end/78703
23323 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23324 (fmtresult::operator+=): Outlined.
23325 (struct fmtresult): Add ctors.
23326 (struct conversion_spec): Rename...
23327 (struct directive): ...to this. Add and remove data members.
23328 (directive::set_width, directive::set_precision): New functions.
23329 (format_percent): Use fmtresult ctor.
23330 (get_width_and_precision): Remove.
23331 (format_integer): Make naming changes. Avoid computing width and
23333 (format_floating): Same. Adjust indentation.
23334 (format_character, format_none): New functions.
23335 (format_string): Moved character handling to format_character.
23336 (format_directive): Remove arguments, change return type.
23337 (parse_directive): New function.
23338 (pass_sprintf_length::compute_format_length): Move directive
23339 parsing to parse_directive.
23341 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23343 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23344 (assign_assembler_name_if_needed): ... this.
23345 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23346 (assign_assembler_name_if_needed): ... this.
23347 (free_lang_data_in_cgraph): Adjust callers.
23348 * cgraphunit.c (cgraph_node::analyze): Likewise.
23349 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23352 2017-01-23 Richard Biener <rguenther@suse.de>
23354 PR tree-optimization/79088
23355 PR tree-optimization/79188
23356 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23357 resetting loop bounds after last path deletion. Reset loop
23358 bounds of the target loop, make code match the comments.
23359 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23360 Make sure loops need no fixups.
23362 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23364 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23365 exponent support with double type for first argument.
23366 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23367 type returned by __builtin_vec_extract_sig,
23368 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23369 functions from "vector int" to "vector unsigned int" or from
23370 "vector long long int" to "vector unsigned long long int".
23371 Changed type returned by __builtin_vec_extract_exp,
23372 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23373 functions from "vector int" to "vector unsigned int" or from
23374 "vector long long int" to "vector unsigned long long int".
23375 Changed return type of __builtin_vec_test_data_class,
23376 __builtin_vec_test_data_class_sp, and
23377 __builtin_vec_test_data_class_dp from "vector int" to
23378 "vector bool int" or from "vector long long int" to "vector bool
23379 long long int" and changed second argument type from "unsigned
23380 int" to "int". Added new overloaded function forms "vector float
23381 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23382 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23383 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23384 double, vector unsigned long long int)" and "vector double
23385 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23386 long int)". Changed return type of
23387 __builtin_scalar_test_data_class and
23388 __builtin_scalar_test_data_class_sp and
23389 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23390 int" and changed second argument from "unsigned int" to "int".
23391 Changed type returned by __builtin_scalar_test_neg,
23392 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23393 from "int" to "bool int". Added new overloaded function form
23394 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23395 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23396 exponent double-precision with floating point first argument.
23397 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23398 documentation of scalar_test_data_class, scalar_test_neg,
23399 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23400 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23401 vec_test_data_class built-in functions to reflect refinements in
23402 their type signatures.
23404 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23406 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23408 (aarch64_elf_asm_destructor): Likewise.
23410 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23412 PR rtl-optimization/78634
23413 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23414 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23415 * ifcvt.c (noce_try_cmove): Add missing cost check.
23417 PR rtl-optimization/71724
23418 * combine.c (if_then_else_cond): Look for situations where it is
23419 beneficial to undo the work of one of the recursive calls.
23421 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23423 PR tree-optimization/70754
23424 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23425 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23426 combined stmt before it if not NULL.
23427 (combine_chains): Process refs reversely and compute dominance point
23430 2017-01-23 Martin Liska <mliska@suse.cz>
23432 PR tree-optimization/79196
23433 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23434 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23436 (strlen_optimize_stmt): Call the renamed function.
23438 2017-01-23 Michael Matz <matz@suse.de>
23440 PR tree-optimization/78384
23441 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23443 2017-01-23 Richard Biener <rguenther@suse.de>
23445 PR tree-optimization/79186
23446 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23447 both incoming edges before moving an assert.
23449 2017-01-23 Martin Jambor <mjambor@suse.cz>
23451 * ipa-prop.c (load_from_param_1): Removed.
23452 (load_from_unmodified_param): Bits from load_from_param_1 put back
23454 (load_from_param): Removed.
23455 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23456 with stmt. Reverted back to use of load_from_unmodified_param.
23458 2017-01-23 Martin Jambor <mjambor@suse.cz>
23461 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23462 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23463 field a pointer to garbage collected vector, mark lattices and
23464 ipcp_orig_node with GTY((skip)).
23465 (ipa_get_param_count): Adjust to descriptors being a pointer.
23466 (ipa_get_param): Likewise.
23467 (ipa_get_type): Likewise.
23468 (ipa_get_param_move_cost): Likewise.
23469 (ipa_set_param_used): Likewise.
23470 (ipa_get_controlled_uses): Likewise.
23471 (ipa_set_controlled_uses): Likewise.
23472 (ipa_is_param_used): Likewise.
23473 (ipa_node_params_t): Move into garbage collector. New methods insert
23475 (ipa_node_params_sum): Annotate wth GTY(()).
23476 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23478 (ipa_load_from_parm_agg): Adjust declaration.
23479 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23480 * ipa-profile.c (ipa_profile): Likewise.
23481 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23482 (ipa_populate_param_decls): Make descriptors parameter garbage
23484 (ipa_dump_param): Adjust to descriptors being a pointer.
23485 (ipa_alloc_node_params): Likewise.
23486 (ipa_initialize_node_params): Likewise.
23487 (load_from_param_1): Make descriptors parameter garbage collected.
23488 (load_from_unmodified_param): Likewise.
23489 (load_from_param): Likewise.
23490 (ipa_load_from_parm_agg): Likewise.
23491 (ipa_node_params::~ipa_node_params): Removed.
23492 (ipa_free_all_node_params): Remove call to delete operator.
23493 (ipa_node_params_t::insert): New.
23494 (ipa_node_params_t::remove): Likewise.
23495 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23496 copy known_csts and known_contexts vectors.
23497 (ipa_read_node_info): Adjust to descriptors being a pointer.
23498 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23500 (ipcp_transform_function): Make descriptors variable garbage
23503 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23505 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23506 * config/i386/avx512dqintrin.h: Ditto.
23507 * config/i386/avx512fintrin.h: Ditto.
23508 * config/i386/i386.c: Handle new builtins.
23509 * config/i386/i386-builtin.def: Add new builtins.
23510 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23511 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23513 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23514 Martin Liska <mliska@suse.cz>
23516 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23517 * asan.c (asan_expand_poison_ifn): Support stores and use
23518 appropriate ASAN report function.
23519 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23520 * internal-fn.def (ASAN_POISON_USE): Declare.
23521 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23522 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23523 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23524 ASAN_POISON calls w/o LHS.
23525 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23526 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23527 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23528 * gimplify.c (asan_poison_variables): Add attribute
23529 use_after_scope_memory to variables that really needs to live
23531 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23532 having the attribute.
23534 2017-01-23 Martin Liska <mliska@suse.cz>
23536 * asan.c (create_asan_shadow_var): New function.
23537 (asan_expand_poison_ifn): Likewise.
23538 * asan.h (asan_expand_poison_ifn): New declaration.
23539 * internal-fn.c (expand_ASAN_POISON): Likewise.
23540 * internal-fn.def (ASAN_POISON): New builtin.
23541 * sanopt.c (pass_sanopt::execute): Expand
23542 asan_expand_poison_ifn.
23543 * tree-inline.c (copy_decl_for_dup_finish): Make function
23545 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23546 * tree-ssa.c (is_asan_mark_p): New function.
23547 (execute_update_addresses_taken): Rewrite local variables
23548 (identified just by use-after-scope as addressable) into SSA.
23550 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23552 * doc/install.texi (Specific): opensource.apple.com uses https
23553 now. Remove trailing slash.
23555 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23557 * README.Portability: Remove note on an Irix compatibility issue.
23559 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23561 * gcov.c (INCLUDE_ALGORITHM): Define.
23562 (INCLUDE_VECTOR): Define.
23563 No longer include <vector> and <algorithm> directly.
23565 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23567 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23569 * doc/invoke.texi (Code Gen Options): Ditto.
23571 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23574 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23576 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23578 rtl-optimization/79125
23579 * cprop.c (local_cprop_pass): Handle cases where we make an
23580 unconditional trap.
23582 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23586 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23587 read from, for big endian.
23589 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23591 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23592 register pauth builtins for LP64 only.
23594 2017-01-20 Marek Polacek <polacek@redhat.com>
23597 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23600 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23602 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23604 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23605 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23606 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23608 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23611 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23612 in big-endian mode when they are not a single duplicated value.
23614 2017-01-20 Richard Biener <rguenther@suse.de>
23616 * BASE-VER: Bump to 7.0.1.
23618 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23620 * omp-low.c (omplow_simd_context): New struct. Use it...
23621 (lower_rec_simd_input_clauses): ...here and...
23622 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23623 references to idx, lane, max_vf, is_simt.
23625 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23627 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23630 2017-01-20 Martin Jambor <mjambor@suse.cz>
23632 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23633 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23635 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23636 (GTFILES): Rename hsa.c to hsa-common.c.
23637 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23638 * hsa-dump.c: Likewise.
23639 * hsa-gen.c: Likewise.
23640 * hsa-regalloc.c: Likewise.
23641 * ipa-hsa.c: Likewise.
23642 * omp-expand.c: Likewise.
23643 * omp-low.c: Likewise.
23644 * toplev.c: Likewise.
23646 2017-01-20 Marek Polacek <polacek@redhat.com>
23649 * doc/invoke.texi: Document -Wduplicated-branches.
23650 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23651 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23652 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23653 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23654 return 0 only when not OEP_LEXICOGRAPHIC.
23655 (fold_build_cleanup_point_expr): Use the expression
23656 location when building CLEANUP_POINT_EXPR.
23657 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23658 * tree.c (add_expr): Handle error_mark_node.
23660 2017-01-20 Martin Liska <mliska@suse.cz>
23663 * tree-profile.c (init_ic_make_global_vars): Do not call
23665 (gimple_init_gcov_profiler): Likewise.
23667 2017-01-20 Martin Liska <mliska@suse.cz>
23670 * cgraph.h (maybe_create_reference): Remove argument and
23672 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23674 * ipa-cp.c (create_specialized_node): Likewise.
23675 * symtab.c (symtab_node::maybe_create_reference): Handle
23676 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23678 2017-01-20 Martin Liska <mliska@suse.cz>
23680 * read-rtl-function.c (function_reader::create_function): Use
23681 build_decl instread of build_decl_stat.
23683 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23685 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23686 * config/i386/avx512dqintrin.h: Ditto.
23687 * config/i386/avx512fintrin.h: Ditto.
23688 * config/i386/i386-builtin-types.def: Add new types.
23689 * config/i386/i386.c: Handle new types.
23690 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23691 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23692 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23693 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23694 (__builtin_ia32_kshiftridi): New.
23695 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23697 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23701 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23702 define to rs6000_init_stack_protect_guard.
23703 (rs6000_init_stack_protect_guard): New function.
23705 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23706 Yunqiang Su <yunqiang.su@imgtec.com>
23708 * config.gcc (supported_defaults): Add madd4.
23709 (with_madd4): Add validation.
23710 (all_defaults): Add madd4.
23711 * config/mips/mips.opt (mmadd4): New option.
23712 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23714 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23716 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23717 (ISA_HAS_FUSED_MADD4): Likewise.
23718 * doc/invoke.texi (-mmadd4): Document the new option.
23719 * doc/install.texi (--with-madd4): Document the new option.
23721 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23723 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23724 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23725 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23726 (aarch64_init_pauth_hint_builtins): New.
23727 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23728 (aarch64_expand_builtin): Expand new builtins.
23730 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23732 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23733 * combine-stack-adj.c (no_unhandled_cfa): Handle
23734 REG_CFA_TOGGLE_RA_MANGLE.
23735 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23736 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23737 info for return address signing.
23738 (aarch64_expand_epilogue): Likewise.
23740 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23742 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23743 * config/aarch64/aarch64-protos.h
23744 (aarch64_return_address_signing_enabled): New declaration.
23745 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23747 (aarch64_expand_prologue): Sign return address before it's pushed onto
23749 (aarch64_expand_epilogue): Authenticate return address fetched from
23751 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23752 (aarch64_attributes): New function attributes for "sign-return-address".
23753 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23754 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23755 ("*do_return"): Generate combined instructions according to key index.
23756 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23757 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23759 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23760 * config/aarch64/aarch64.opt (msign-return-address=): New.
23761 * doc/extend.texi (AArch64 Function Attributes): Documents
23762 "sign-return-address=".
23763 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23765 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23767 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23768 overall option summary.
23770 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23772 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23773 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23774 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23775 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23777 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23779 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23780 -mpower9-minmax by default for -mcpu=power9.
23781 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23782 128-bit floating point.
23784 2017-01-20 Alan Modra <amodra@gmail.com>
23786 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23787 optimizing for size.
23789 2017-01-20 Alan Modra <amodra@gmail.com>
23792 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23793 for strcmp and strncmp from corresponding builtin decl.
23795 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23797 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23798 instead of i386/rtems-64.h.
23799 * config/i386/rtems-64.h: Remove.
23801 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23805 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23807 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23809 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23811 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23812 Change int to HOST_WIDE_INT.
23813 * config/aarch64/aarch64-protos.h
23814 (aarch64_simd_gen_const_vector_dup): Likewise.
23815 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23817 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23819 * langhooks-def.h (lhd_type_for_size): New decl.
23820 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23821 * langhooks.c (lhd_type_for_size): New function, taken from
23824 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23826 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23827 define_bypass for CR latency.
23828 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23829 (power9-alu2): Add define_bypass for CR latency.
23831 (power9-mul): Update insn latency.
23832 (power9-mul-compare): Update insn latency, bypass latency and remove
23835 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23837 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23839 * config/aarch64/aarch64.md
23840 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23841 aarch64_nopcrelative_literal_loads.
23842 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23844 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23846 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23847 TARGET_LOONGSON_3A.
23848 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23850 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23853 * config.gcc (supported_defaults): Add lxc1-sxc1.
23854 (with_lxc1_sxc1): Add validation.
23855 (all_defaults): Add lxc1-sxc1.
23856 * config/mips/mips.opt (mlxc1-sxc1): New option.
23857 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23859 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23860 __mips_no_lxc1_sxc1.
23861 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23862 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23863 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23865 2017-01-19 Richard Biener <rguenther@suse.de>
23867 PR tree-optimization/72488
23868 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23869 sure to restore SSA info.
23870 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23872 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23874 PR rtl-optimization/79121
23875 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23876 of the inner type when shifting an extended value.
23878 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23881 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23882 interposable aliases.
23884 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23887 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23888 Use the evmergelohi instruction.
23889 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23890 (mov_sitf_e500_subreg8_2_be): Likewise.
23891 (mov_sitf_e500_subreg12_2_le): Likewise.
23892 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23893 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23894 (mov_sitf_e500_subreg8_2_le): Likewise.
23895 (mov_sitf_e500_subreg12_2_be): Likewise.
23897 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23899 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23900 attribute from vecsimple to vecperm.
23901 (altivec_vbpermq2): Likewise.
23903 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23906 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23908 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23909 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23910 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23911 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23912 case where N arg is SIZE_MAX.
23913 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23914 (cmpstrsi): Add pattern.
23916 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23918 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23919 __builtin_vec_revb builtins.
23920 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23921 built-in functions to support generation of the ISA 3.0 XXBR<x>
23922 vector byte reverse instructions.
23923 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23924 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23925 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23926 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23927 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23928 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23929 (P9V_BUILTIN_VEC_REVB): Likewise.
23930 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23931 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23932 (p9_xxbrq_v16qi): Likewise.
23933 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23934 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23935 (p9_xxbrh_v8hi): Likewise.
23936 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23937 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23938 vec_revb built-in functions.
23940 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23942 PR rtl-optimization/78952
23943 * config/i386/i386.md (any_extract): New code iterator.
23944 (*insvqi_2): Use any_extract for source operand.
23945 (*insvqi_3): Use any_shiftrt for source operand.
23947 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23949 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23951 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23953 2017-01-18 Matthias Klose <doko@ubuntu.com>
23955 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23957 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23959 * config/rs6000/altivec.h (vec_bperm): Change #define.
23960 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23961 (altivec_vbpermq2): New define_insn.
23962 (altivec_vbpermd): Likewise.
23963 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23964 function interface.
23965 (VBPERMD): Likewise.
23966 (VBPERM): New polymorphic function interface.
23967 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23968 Add entries for P9V_BUILTIN_VEC_VBPERM.
23969 * doc/extend.texi: Add interfaces for vec_bperm.
23971 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23973 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23974 first letter of error messages.
23975 (s390_resolve_overloaded_builtin): Likewise.
23976 * config/s390/s390.c (s390_expand_builtin): Likewise.
23977 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23978 (s390_valid_target_attribute_inner_p): Likewise.
23979 * config/s390/s390.md ("tabort"): Likewise.
23981 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23983 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23984 (ISA_AVOID_DIV_HILO): New macro.
23985 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23986 (ISA_HAS_DDIV): Likewise.
23988 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23990 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23992 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23994 * doc/invoke.texi (fabi-version): Spelling fix.
23996 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23999 * doc/invoke.texi (fabi-version): Mention mangling fix for
24002 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24005 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24007 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24010 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24011 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24013 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24015 (stack_protect_test): Ditto.
24016 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24017 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24019 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24020 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24021 -mstack-protector-guard-offset=.
24022 (RS/6000 and PowerPC Options): Ditto.
24024 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24026 * config/i386/i386.h (MASK_CLASS_P): New define.
24027 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24028 there are no registers from different register sets also when
24029 mask registers are used. Update function comment.
24030 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24031 to (*k/*r) and (*k/*km) alternatives.
24033 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24035 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24036 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24037 (EH_RETURN_HANDLER_RTX): New define.
24038 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24039 Force frame pointer in EH return functions.
24040 (aarch64_expand_epilogue): Add barrier for eh_return.
24041 (aarch64_final_eh_return_addr): Remove.
24042 (aarch64_eh_return_handler_rtx): New function.
24043 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24045 (aarch64_eh_return_handler_rtx): New prototype.
24047 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24049 * config/rs6000/altivec.h (vec_rlmi): New #define.
24050 (vec_vrlnm): Likewise.
24051 (vec_rlnm): Likewise.
24052 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24053 (UNSPEC_VRLNM): Likewise.
24054 (VIlong): New mode iterator.
24055 (altivec_vrl<VI_char>mi): New define_insn.
24056 (altivec_vrl<VI_char>nm): Likewise.
24057 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24059 (VRLDNM): Likewise.
24060 (RLNM): New polymorphic function entry.
24061 (VRLWMI): New monomorphic function entry.
24062 (VRLDMI): Likewise.
24063 (RLMI): New polymorphic function entry.
24064 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24065 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24066 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24069 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24072 * dwarf2out.c (field_byte_offset): Restore the
24073 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24074 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24075 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24078 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24081 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24083 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24086 * dwarf2out.c (add_data_member_location_attribute): For constant
24087 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24088 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24089 DW_AT_byte_size attributes.
24091 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24093 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24094 after forcing to constant memory when the code model is medium.
24096 2017-01-17 Julia Koval <julia.koval@intel.com>
24099 * config/i386/avx512fintrin.h
24100 (_mm512_i32gather_ps): Change __addr type to void const*.
24101 (_mm512_mask_i32gather_ps): Ditto.
24102 (_mm512_i32gather_pd): Ditto.
24103 (_mm512_mask_i32gather_pd): Ditto.
24104 (_mm512_i64gather_ps): Ditto.
24105 (_mm512_mask_i64gather_ps): Ditto.
24106 (_mm512_i64gather_pd): Ditto.
24107 (_mm512_mask_i64gather_pd): Ditto.
24108 (_mm512_i32gather_epi32): Ditto.
24109 (_mm512_mask_i32gather_epi32): Ditto.
24110 (_mm512_i32gather_epi64): Ditto.
24111 (_mm512_mask_i32gather_epi64): Ditto.
24112 (_mm512_i64gather_epi32): Ditto.
24113 (_mm512_mask_i64gather_epi32): Ditto.
24114 (_mm512_i64gather_epi64): Ditto.
24115 (_mm512_mask_i64gather_epi64): Ditto.
24116 (_mm512_i32scatter_ps): Change __addr type to void*.
24117 (_mm512_mask_i32scatter_ps): Ditto.
24118 (_mm512_i32scatter_pd): Ditto.
24119 (_mm512_mask_i32scatter_pd): Ditto.
24120 (_mm512_i64scatter_ps): Ditto.
24121 (_mm512_mask_i64scatter_ps): Ditto.
24122 (_mm512_i64scatter_pd): Ditto.
24123 (_mm512_mask_i64scatter_pd): Ditto.
24124 (_mm512_i32scatter_epi32): Ditto.
24125 (_mm512_mask_i32scatter_epi32): Ditto.
24126 (_mm512_i32scatter_epi64): Ditto.
24127 (_mm512_mask_i32scatter_epi64): Ditto.
24128 (_mm512_i64scatter_epi32): Ditto.
24129 (_mm512_mask_i64scatter_epi32): Ditto.
24130 (_mm512_i64scatter_epi64): Ditto.
24131 (_mm512_mask_i64scatter_epi64): Ditto.
24132 * config/i386/avx512pfintrin.h
24133 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24134 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24135 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24136 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24137 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24138 (_mm512_prefetch_i32scatter_ps): Ditto.
24139 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24140 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24141 (_mm512_prefetch_i64scatter_pd): Ditto.
24142 (_mm512_prefetch_i64scatter_ps): Ditto.
24143 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24144 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24145 * config/i386/avx512vlintrin.h
24146 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24147 (_mm_mmask_i32gather_ps): Ditto.
24148 (_mm256_mmask_i32gather_pd): Ditto.
24149 (_mm_mmask_i32gather_pd): Ditto.
24150 (_mm256_mmask_i64gather_ps): Ditto.
24151 (_mm_mmask_i64gather_ps): Ditto.
24152 (_mm256_mmask_i64gather_pd): Ditto.
24153 (_mm_mmask_i64gather_pd): Ditto.
24154 (_mm256_mmask_i32gather_epi32): Ditto.
24155 (_mm_mmask_i32gather_epi32): Ditto.
24156 (_mm256_mmask_i32gather_epi64): Ditto.
24157 (_mm_mmask_i32gather_epi64): Ditto.
24158 (_mm256_mmask_i64gather_epi32): Ditto.
24159 (_mm_mmask_i64gather_epi32): Ditto.
24160 (_mm256_mmask_i64gather_epi64): Ditto.
24161 (_mm_mmask_i64gather_epi64): Ditto.
24162 (_mm256_i32scatter_ps): Change __addr type to void*.
24163 (_mm256_mask_i32scatter_ps): Ditto.
24164 (_mm_i32scatter_ps): Ditto.
24165 (_mm_mask_i32scatter_ps): Ditto.
24166 (_mm256_i32scatter_pd): Ditto.
24167 (_mm256_mask_i32scatter_pd): Ditto.
24168 (_mm_i32scatter_pd): Ditto.
24169 (_mm_mask_i32scatter_pd): Ditto.
24170 (_mm256_i64scatter_ps): Ditto.
24171 (_mm256_mask_i64scatter_ps): Ditto.
24172 (_mm_i64scatter_ps): Ditto.
24173 (_mm_mask_i64scatter_ps): Ditto.
24174 (_mm256_i64scatter_pd): Ditto.
24175 (_mm256_mask_i64scatter_pd): Ditto.
24176 (_mm_i64scatter_pd): Ditto.
24177 (_mm_mask_i64scatter_pd): Ditto.
24178 (_mm256_i32scatter_epi32): Ditto.
24179 (_mm256_mask_i32scatter_epi32): Ditto.
24180 (_mm_i32scatter_epi32): Ditto.
24181 (_mm_mask_i32scatter_epi32): Ditto.
24182 (_mm256_i32scatter_epi64): Ditto.
24183 (_mm256_mask_i32scatter_epi64): Ditto.
24184 (_mm_i32scatter_epi64): Ditto.
24185 (_mm_mask_i32scatter_epi64): Ditto.
24186 (_mm256_i64scatter_epi32): Ditto.
24187 (_mm256_mask_i64scatter_epi32): Ditto.
24188 (_mm_i64scatter_epi32): Ditto.
24189 (_mm_mask_i64scatter_epi32): Ditto.
24190 (_mm256_i64scatter_epi64): Ditto.
24191 (_mm256_mask_i64scatter_epi64): Ditto.
24192 (_mm_i64scatter_epi64): Ditto.
24193 (_mm_mask_i64scatter_epi64): Ditto.
24194 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24195 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24196 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24197 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24198 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24199 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24200 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24201 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24202 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24203 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24204 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24205 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24206 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24207 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24208 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24209 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24210 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24211 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24212 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24213 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24214 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24215 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24216 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24217 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24218 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24219 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24220 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24221 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24222 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24223 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24224 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24225 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24226 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24227 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24228 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24229 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24230 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24231 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24232 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24233 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24234 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24235 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24236 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24237 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24238 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24239 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24240 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24241 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24242 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24243 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24244 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24245 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24246 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24247 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24248 definitions accordingly.
24250 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24251 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24254 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24257 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24260 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24261 hard regs for inner regno.
24263 2017-01-17 Martin Liska <mliska@suse.cz>
24266 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24267 assumption and add comment.
24269 2017-01-17 Nathan Sidwell <nathan@acm.org>
24271 * ipa-visibility.c (localize_node): New function, broken out of ...
24272 (function_and_variable_visibility): ... here. Call it.
24274 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24276 PR middle-end/77445
24277 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24278 correctly set frequency of oudgoing edge.
24279 (duplicate_thread_path): Fix profile updating.
24281 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24284 * configure.ac: Add GCC_BASE_VER.
24285 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24286 version from BASE-VER file.
24287 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24288 (gcc.o): Depend on $(BASEVER).
24289 * common.opt (dumpfullversion): New option.
24290 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24291 * doc/invoke.texi: Document -dumpfullversion.
24292 * doc/install.texi: Document --with-gcc-major-version-only.
24293 * configure: Regenerated.
24295 2017-01-17 Richard Biener <rguenther@suse.de>
24297 PR tree-optimization/71433
24298 * tree-vrp.c (register_new_assert_for): Merge same asserts
24299 on all incoming edges.
24300 (process_assert_insertions_for): Handle insertions at the
24303 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24305 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24306 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24308 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24311 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24314 2017-01-17 Alan Modra <amodra@gmail.com>
24317 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24318 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24319 symbolic stack limit when pic.
24321 2017-01-16 Martin Sebor <msebor@redhat.com>
24323 PR tree-optimization/78608
24324 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24326 2017-01-16 Jeff Law <law@redhat.com>
24329 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24330 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24331 for several include directories that may be relative to sysroot.
24332 * config/i386/x-mingw32 (gplus_includedir): Define.
24333 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24334 (native_system_includedir): Likewise.
24335 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24336 override if TARGET_SYSTEM_ROOT is defined.
24337 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24339 PR tree-optimization/79090
24340 PR tree-optimization/33562
24341 PR tree-optimization/61912
24342 PR tree-optimization/77485
24343 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24344 and computed trims into the dump file.
24346 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24348 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24350 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24353 * gimplify.c (gimplify_init_constructor): If want_value and
24354 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24358 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24359 sequence. Formatting fixes.
24360 (doloop_optimize): Formatting fixes.
24363 * gcc.c (debug_level_greater_than_spec_func): New function.
24364 (static_spec_functions): Add debug-level-gt spec function.
24365 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24367 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24368 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24369 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24370 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24371 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24372 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24374 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24376 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24377 QImode fixups to general and mask registers only.
24379 2017-01-16 Carl Love <cel@us.ibm.com>
24381 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24382 for built-in functions
24383 vector signed char vec_nabs (vector signed char)
24384 vector signed short vec_nabs (vector signed short)
24385 vector signed int vec_nabs (vector signed int)
24386 vector signed long long vec_nabs (vector signed long long)
24387 vector float vec_nabs (vector float)
24388 vector double vec_nabs (vector double)
24389 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24391 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24392 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24393 * doc/extend.texi: Update the documentation file for the new built-in
24396 2017-01-16 Martin Sebor <msebor@redhat.com>
24398 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24401 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24403 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24404 UNSPEC_VSX__XXSPLTD to require special splat handling.
24406 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24409 * system.h: Poison strndup.
24411 2017-01-16 Alan Modra <amodra@gmail.com>
24414 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24417 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24419 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24421 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24423 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24424 call recog here. Assert that INSN_CODE (insn) is non-negative.
24426 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24429 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24431 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24432 in the currently scheduled RTL fragment.
24434 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24436 PR rtl-optimization/78751
24437 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24440 2017-01-14 Jeff Law <law@redhat.com>
24442 PR tree-optimization/79090
24443 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24444 variable length stores.
24445 (compute_trims): Delete dead assignment to *trim_tail.
24446 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24449 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24451 PR rtl-optimization/78626
24452 PR rtl-optimization/78727
24453 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24454 of a block, and split such blocks after everything else is finished.
24456 2017-01-14 Alan Modra <amodra@gmail.com>
24459 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24460 target legitimate_combined_insn.
24461 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24462 (rs6000_legitimate_combined_insn): New function.
24463 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24465 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24466 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24467 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24469 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24471 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24473 2017-01-13 Jeff Law <law@redhat.com>
24475 PR tree-optimization/33562
24476 PR tree-optimization/61912
24477 PR tree-optimization/77485
24478 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24480 (delete_dead_assignment): Likewise.
24481 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24482 statement to delete_dead_call and delete_dead_assignment.
24484 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24487 * substring-locations.c (format_warning_va): Strengthen case 1 so
24488 that both endpoints of the substring must be within the format
24489 range for just the substring to be printed.
24491 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24493 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24494 * config/i386/i386.c (ix86_target_string): Add missing options
24495 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24496 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24497 flags_other and ix86_target_other to flags2_other. Display unknown
24499 (ix86_valid_target_attribute_inner_p): Add missing options and
24500 reorder options by implied ISAs, as in ix86_target_string.
24502 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24504 * hash-table.h (hash_table::too_empty_p): New function.
24505 (hash_table::expand): Use it.
24506 (hash_table::traverse): Likewise.
24507 (hash_table::empty_slot): Use sizeof (value_type) instead of
24508 sizeof (PTR) to convert bytes to elements. Shrink the table
24509 if the current size is excessive for the current number of
24512 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24514 * ira-costs.c (record_reg_classes): Break from the inner loop
24515 early once alt_fail is known to be true. Update outer loop
24516 handling accordingly.
24518 2017-01-13 Jeff Law <law@redhat.com>
24520 * tree-ssa-dse.c (decrement_count): New function.
24521 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24522 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24523 when we know the partially dead statement is a mem* function.
24525 PR tree-optimization/61912
24526 PR tree-optimization/77485
24527 * tree-ssa-dse.c: Include expr.h.
24528 (maybe_trim_constructor_store): New function.
24529 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24531 PR tree-optimization/33562
24532 PR tree-optimization/61912
24533 PR tree-optimization/77485
24534 * doc/invoke.texi: Document new dse-max-object-size param.
24535 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24536 * tree-ssa-dse.c: Include params.h.
24537 (dse_store_status): New enum.
24538 (initialize_ao_ref_for_dse): New, partially extracted from
24540 (valid_ao_ref_for_dse, normalize_ref): New.
24541 (setup_live_bytes_from_ref, compute_trims): Likewise.
24542 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24543 (maybe_trim_partially_dead_store): Likewise.
24544 (maybe_trim_complex_store): Likewise.
24545 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24546 Track what bytes live from the original store. Return tri-state
24547 for dead, partially dead or live.
24548 (dse_dom_walker): Add constructor, destructor and new private members.
24549 (delete_dead_call, delete_dead_assignment): New extracted from
24551 (dse_optimize_stmt): Make a member of dse_dom_walker.
24552 Use initialize_ao_ref_for_dse.
24554 PR tree-optimization/33562
24555 PR tree-optimization/61912
24556 PR tree-optimization/77485
24557 * sbitmap.h (bitmap_count_bits): Prototype.
24558 (bitmap_clear_range, bitmap_set_range): Likewise.
24559 * sbitmap.c (bitmap_clear_range): New function.
24560 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24562 2017-01-13 Martin Liska <mliska@suse.cz>
24565 * function.c (set_cfun): Add new argument force.
24566 * function.h (set_cfun): Likewise.
24567 * ipa-inline-transform.c (inline_call): Use the function when
24568 strict alising from is dropped for function we inline to.
24570 2017-01-13 Richard Biener <rguenther@suse.de>
24572 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24573 for dumping GIMPLE INTEGER_CSTs.
24575 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24577 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24578 to 201112L since C++17.
24580 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24583 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24584 if -fsanitize=kernel-address is present.
24586 2017-01-13 Richard Biener <rguenther@suse.de>
24588 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24589 as _Literal ( type ) number in case usual suffixes do not
24590 preserve all information.
24592 2017-01-13 Richard Biener <rguenther@suse.de>
24594 PR tree-optimization/77283
24595 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24596 and ssa-iterators.h.
24597 (is_feasible_trace): Implement a cost model based on joiner
24600 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24603 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24604 char or short to __float128/_Float128 directly.
24606 2017-01-12 Martin Sebor <msebor@redhat.com>
24608 to -Wformat-overflow.
24609 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24610 (min_bytes_remaining): Same.
24611 (get_string_length): Same.
24612 (format_string): Same.
24613 (format_directive): Same.
24615 (pass_sprintf_length::handle_gimple_call): Same.
24617 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24619 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24620 info.nowrite calls with no lhs that can't throw. Return bool
24621 whether gsi_remove has been called or not.
24622 (pass_sprintf_length::handle_gimple_call): Return bool whether
24623 try_substitute_return_value called gsi_remove. Formatting fix.
24624 (pass_sprintf_length::execute): Don't use gsi_remove if
24625 handle_gimple_call returned true.
24628 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24629 be removed due to side-effects, don't remove following barrier nor
24630 turn the successor edge into fallthru edge.
24632 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24635 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24636 element-reversing loads and stores as not swappable.
24638 2017-01-12 Nathan Sidwell <nathan@acm.org>
24639 Nicolai Stange <nicstange@gmail.com>
24641 * combine.c (try_combine): Don't ignore result of overlap checking
24642 loop. Combine overlap & asm check into single loop.
24644 2017-01-12 Richard Biener <rguenther@suse.de>
24646 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24647 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24649 2017-01-12 Richard Biener <rguenther@suse.de>
24651 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24652 and TS_TARGET_OPTION directly derive from TS_BASE.
24653 * tree-core.h (tree_optimization_option): Derive from tree_base.
24654 (tree_target_option): Likewise.
24656 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24658 * config/i386/i386.c (memory_address_length): Increase len
24659 only when rip_relative_addr_p returns false.
24661 2017-01-11 Julia Koval <julia.koval@intel.com>
24663 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24664 (OPTION_MASK_ISA_SGX_SET): New.
24665 (ix86_handle_option): Handle OPT_msgx.
24666 * config.gcc: Added sgxintrin.h.
24667 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24668 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24669 * config/i386/i386.c (ix86_target_string): Add -msgx.
24671 (ix86_option_override_internal): Handle new options.
24672 (ix86_valid_target_attribute_inner_p): Add sgx.
24673 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24674 * config/i386/i386.opt: Add msgx.
24675 * config/i386/sgxintrin.h: New file.
24676 * config/i386/x86intrin.h: Add sgxintrin.h.
24678 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24681 * fold-const.c (maybe_nonzero_address): Return 1 for function
24683 (tree_single_nonzero_warnv_p): Don't handle function local objects
24687 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24690 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24693 * opts.c: Include "spellcheck.h"
24694 (struct string_fragment): New struct.
24695 (struct edit_distance_traits<const string_fragment &>): New
24697 (get_closest_sanitizer_option): New function.
24698 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24700 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24702 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24704 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24705 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24706 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24707 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24708 for initial die_offset if dwarf_split_debug_info.
24709 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24710 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24712 (output_skeleton_debug_sections): Formatting fix. Use
24713 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24714 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24716 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24718 * config/arm/cortex-a53.md: Add bypasses for
24719 cortex_a53_r2f_cvt.
24720 (cortex_a53_r2f): Only use for transfers.
24721 (cortex_a53_f2r): Likewise.
24722 (cortex_a53_r2f_cvt): Add reservation for conversions.
24723 (cortex_a53_f2r_cvt): Likewise.
24725 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24727 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24728 to all inlined functions, change static to extern.
24730 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24733 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24735 (arm_assemble_integer): Likewise.
24737 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24739 * config.gcc: Use new awk script to check CPU, FPU and architecture
24740 parameters for --with-... options.
24741 * config/arm/parsecpu.awk: New file
24742 * config/arm/arm-cpus.in: New file.
24743 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24745 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24747 * config/arm/t-arm: Update dependency rules.
24748 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24749 of processing .def files.
24750 * config/arm/genopt.sh: Deleted.
24751 * config/arm/gentune.sh: Deleted.
24752 * config/arm/arm-cores.def: Deleted.
24753 * config/arm/arm-arches.def: Deleted.
24754 * config/arm/arm-fpus.def: Deleted.
24755 * config/arm/arm-tune.md: Regenerated.
24756 * config/arm/arm-tables.opt: Regenerated.
24757 * config/arm/arm-cpu.h: New generated file.
24758 * config/arm/arm-cpu-data.h: New generated file.
24759 * config/arm/arm-cpu-cdata.h: New generated file.
24761 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24764 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24766 (input_varpool_node): Unpack dynamically_initialized bit.
24768 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24770 PR rtl-optimization/79032
24771 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24772 the alignment of the adjusted memory reference against that of MODE,
24773 instead of the alignment of the original memory reference.
24775 2017-01-11 Martin Jambor <mjambor@suse.cz>
24777 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24779 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24780 decorated functions.
24782 2017-01-11 Richard Biener <rguenther@suse.de>
24784 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24785 set range/nonnull info for PHI results. Do not set it on
24786 stmts marked for removal.
24788 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24790 * expr.c (store_field): In the bitfield case, fetch the return value
24791 from the registers before applying a single big-endian adjustment.
24792 Always do a final load for a BLKmode value not larger than a word.
24794 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24797 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24798 that we correctly handle column numbers greater than
24799 LINE_MAP_MAX_COLUMN_NUMBER.
24801 2017-01-10 Martin Sebor <msebor@redhat.com>
24803 PR middle-end/78245
24804 * gimple-ssa-sprintf.c (get_destination_size): Call
24805 {init,fini}object_sizes.
24806 * tree-object-size.c (addr_object_size): Adjust.
24807 (pass_through_call): Adjust.
24808 (pass_object_sizes::execute): Adjust.
24809 * tree-object-size.h (fini_object_sizes): Declare.
24811 2017-01-10 Martin Sebor <msebor@redhat.com>
24813 PR tree-optimization/78775
24814 * builtins.c (get_size_range): Move...
24815 * calls.c: ...to here.
24816 (alloc_max_size): Accept zero argument.
24817 (operand_signed_p): Remove.
24818 (maybe_warn_alloc_args_overflow): Call get_size_range.
24819 * calls.h (get_size_range): Declare.
24821 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24823 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24824 from TI's devices.csv file as of September 2016.
24825 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24827 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24829 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24830 * doc/invoke.texi: Likewise.
24831 * doc/md.texi: Likewise.
24832 * doc/objc.texi: Likewise.
24834 2017-01-10 Joshua Conner <joshconner@google.com>
24836 * config/arm/fuchsia-elf.h: New file.
24837 * config/fuchsia.h: New file.
24838 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24839 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24841 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24843 2016-01-10 Richard Biener <rguenther@suse.de>
24845 PR tree-optimization/79034
24846 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24847 Propagate out degenerate PHIs in the joiner.
24849 2017-01-10 Martin Liska <mliska@suse.cz>
24851 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24852 (sort_congruence_classes_by_decl_uid): Likewise.
24853 (sort_congruence_class_groups_by_decl_uid): Likewise.
24854 (sem_item_optimizer::merge_classes): Sort class, groups in these
24855 classes and members in the groups by DECL_UID of declarations.
24856 This would make merge operations stable.
24858 2017-01-10 Martin Liska <mliska@suse.cz>
24860 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24861 usage of m_classes_vec.
24862 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24863 (sem_item_optimizer::get_group_by_hash): Likewise.
24864 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24865 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24866 (sem_item_optimizer::verify_classes): Likewise.
24867 (sem_item_optimizer::process_cong_reduction): Likewise.
24868 (sem_item_optimizer::dump_cong_classes): Likewise.
24869 (sem_item_optimizer::merge_classes): Likewise.
24870 * ipa-icf.h (congruence_class_hash): Rename from
24871 congruence_class_group_hash. Remove declaration of m_classes_vec.
24873 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24875 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24876 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24877 * config.gcc: Add avx512vpopcntdqintrin.h.
24878 * config/i386/avx512vpopcntdqintrin.h: New.
24879 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24880 * config/i386/i386-builtin-types.def: Add new types.
24881 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24882 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24883 __builtin_ia32_vpopcountq_v8di_mask): New.
24884 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24885 __AVX512VPOPCNTDQ__.
24886 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24887 (PTA_AVX512VPOPCNTDQ): Define.
24888 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24889 TARGET_AVX512VPOPCNTDQ_P): Define.
24890 * config/i386/i386.opt: Add mavx512vpopcntdq.
24891 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24892 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24894 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24896 PR middle-end/77484
24897 * predict.def (PRED_CALL): Set to 67.
24899 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24901 * expr.c (store_field): In the bitfield case, if the value comes from
24902 a function call and is of an aggregate type returned in registers, do
24903 not modify the field mode; extract the value in all cases if the mode
24904 is BLKmode and the size is not larger than a word.
24906 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24909 * config/i386/cpuid.h: Fix undefined behavior.
24911 2017-01-04 Jeff Law <law@redhat.com>
24913 PR tree-optimization/79007
24914 PR tree-optimization/67955
24915 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24916 conservative for pt.null when flag_non_call_exceptions is on.
24918 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24920 PR translation/79019
24921 PR translation/79020
24922 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24923 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24924 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24926 * config/avr/avr.opt (maccumulate-args): Likewise.
24927 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24928 * common.opt (freport-bug): Likewise.
24929 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24930 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24931 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24932 translatable string.
24933 * config/i386/i386.c (function_value_32): Likewise.
24934 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24935 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24937 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24938 * common/config/msp430/msp430-common.c (msp430_handle_option):
24940 * symtab.c (symtab_node::verify_base): Likewise.
24941 * opts.c (set_debug_level): Likewise.
24942 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24943 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24944 missing whitespace to translatable strings.
24945 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24946 * config/sh/superh.h: Likewise.
24947 * config/i386/xopintrin.h: Likewise.
24948 * config/i386/znver1.md: Likewise.
24949 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24950 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24951 * double-int.h (struct double_int): Likewise.
24952 * double-int.c (div_and_round_double): Likewise.
24953 * wide-int.cc: Likewise.
24954 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24955 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24956 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24957 (crossjumps_occurred): ... this.
24958 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24961 PR tree-optimization/78899
24962 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24963 returning bool return struct loop *, NULL for failure and the new
24965 (versionable_outer_loop_p): Don't version outer loop if it has
24966 dont_vectorized bit set.
24967 (tree_if_conversion): When versioning outer loop, ensure
24968 tree_if_conversion is performed also on the inner loop of the
24969 non-vectorizable outer loop copy.
24970 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24971 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24972 prevent vectorization of it.
24973 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24974 the outer loop vectorization of the non-scalar version is attempted
24975 before vectorization of the inner loop in scalar version. If
24976 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24977 vectorization of its inner loop.
24978 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24979 has 2 inner loops, rename also on edges from bb whose single pred
24980 is outer_loop->header. Fix typo in function comment.
24982 2017-01-09 Martin Sebor <msebor@redhat.com>
24985 * asan.c (asan_emit_stack_protection): Increase local buffer size
24986 to avoid snprintf truncation warning.
24988 2017-01-09 Andrew Pinski <apinski@cavium.com>
24990 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24991 to reference thunderx2t99 for the tuning structure
24992 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24994 (thunderx2t99_extra_costs): This.
24995 * config/aarch64/aarch64-tune.md: Regenerate.
24996 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24997 (vulcan_addrcost_table): This.
24998 (vulcan_regmove_cost): Rename to ...
24999 (thunderx2t99_regmove_cost): This.
25000 (vulcan_vector_cost): Rename to ...
25001 (thunderx2t99_vector_cost): this.
25002 (vulcan_branch_cost): Rename to ...
25003 (thunderx2t99_branch_cost): This.
25004 (vulcan_tunings): Rename to ...
25005 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25006 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25008 2017-01-09 Martin Jambor <mjambor@suse.cz>
25012 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25013 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25014 (propagate_vr_accross_jump_function): Use the above function for all
25015 value range computations for pass-through jump functions and type
25016 converasion from explicit value range values.
25017 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25018 parameters from TYPE_ARG_TYPES.
25019 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25020 (ipa_write_node_info): Stream type of the actual argument.
25021 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25023 2017-01-09 Martin Liska <mliska@suse.cz>
25026 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25027 (lookup_compiler): Do not show error message with have_E.
25029 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25031 PR tree-optimization/78938
25032 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25033 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25034 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25035 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25038 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25040 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25043 2017-01-09 Richard Biener <rguenther@suse.de>
25045 PR tree-optimization/78997
25046 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25047 name condition properly.
25049 2017-01-09 Richard Biener <rguenther@suse.de>
25052 * dwarf2out.c (is_cxx): New overload with context.
25053 (is_naming_typedef_decl): Use it.
25055 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25057 * invoke.texi (Option Summary): Correct spacing in option lists
25058 and add line breaks to fix over-long lines.
25060 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25062 PR middle-end/17660
25064 * extend.texi (Common Variable Attributes): Add xref to GCC
25065 Internals manual to explain mode attribute keywords.
25067 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25070 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25071 and Preprocessor Options.
25072 (Options for Linking): Document -pthread here....
25073 (RS/6000 and PowerPC Options): ...not here.
25074 (Solaris 2 Options): ...or here.
25075 * doc/cppopts.texi: Document -pthread.
25077 2017-01-08 Martin Sebor <msebor@redhat.com>
25079 PR middle-end/77708
25080 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25081 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25082 New member functions.
25083 (format_directive): Used them.
25085 (pass_sprintf_length::handle_gimple_call): Same.
25086 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25087 to avoid truncation for any argument.
25088 (extract_affine_mul): Same.
25089 * tree.c (get_file_function_name): Same.
25091 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25093 PR middle-end/77484
25094 * predict.def (PRED_INDIR_CALL): Set to 86.
25096 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25098 PR preprocessor/54124
25099 * doc/cppopts.texi: Reformat -d subtable to list the full name
25100 of the options. Add cross-reference to the docs for the general
25101 compiler -d options.
25102 * doc/invoke.texi (Developer Options): Add cross-reference to the
25103 preprocessor-specific -d option documentation.
25105 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25107 PR preprocessor/13498
25108 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25109 redudant material, and reflect new command-line options.
25110 (System Headers): Likewise.
25112 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25114 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25115 -isystem, and -idirafter. Copy-edit.
25116 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25117 default for -ftrack-macro-expansion. Delete obsolete and
25118 badly-formatted implementation details about -fdebug-cpp output.
25119 * doc/cppwarnopts.texi: Copy-edit.
25121 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25124 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25125 that the transition from a max line width >= 1<<10 to narrower
25126 lines works correctly.
25128 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25130 * doc/options.texi (PerFunction): New.
25131 * opt-functions.awk (switch_flags): Map both Optimization and
25132 PerFunction to CL_OPTIMIZATION.
25133 * opth-gen.awk: Test for PerFunction flag along with
25135 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25136 it only when the latter is present. Skip those that don't in
25137 the hash function generator.
25138 * common.opt (fvar-tracking): Mark as PerFunction instead of
25140 (fvar-tracking-assignments): Likewise.
25141 (fvar-tracking-assignments-toggle): Likewise.
25142 (fvar-tracking-uninit): Likewise.
25144 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25146 PR translation/79018
25147 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25150 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25153 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25154 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25155 TARGET_LONG_JUMP_TABLE_OFFSETS.
25156 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25157 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25158 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25159 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25160 * config/m68k/m68k.md (tablejump expander): Likewise.
25161 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25162 TARGET_LONG_JUMP_TABLE_OFFSETS.
25163 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25164 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25166 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25167 David Holsgrove <david.holsgrove@xilinx.com>
25169 * common/config/microblaze/microblaze-common.c
25170 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25171 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25173 * config/microblaze/microblaze.c (microblaze_must_save_register)
25174 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25176 (microblaze_eh_return): New function.
25177 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25178 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25179 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25180 * config/microblaze/microblaze.md (eh_return): New pattern.
25182 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25184 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25185 GCC_DIAGNOSTIC_STRINGIFY): Define.
25187 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25189 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25191 * config/arm/arm.md (<mcrr>): New.
25193 * config/arm/arm.c (arm_arch5te): New.
25194 (arm_option_override): Set arm_arch5te.
25195 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25197 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25198 (arm_mcrr_qualifiers): ... this. New.
25199 (MRRC_QUALIFIERS): Define to...
25200 (arm_mrrc_qualifiers): ... this. New.
25201 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25203 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25204 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25205 (MRRCI, mrrc, MRRC): New.
25206 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25207 VUNSPEC_MRRC2): New.
25209 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25211 * config/arm/arm.md (<mcr>): New.
25213 * config/arm/arm.c (arm_coproc_builtin_available): Add
25214 support for mcr, mrc, mcr2 and mrc2.
25215 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25216 (arm_mcr_qualifiers): ... this. New.
25217 (MRC_QUALIFIERS): Define to ...
25218 (arm_mrc_qualifiers): ... this. New.
25219 (MCR_QUALIFIERS): Define to ...
25220 (arm_mcr_qualifiers): ... this. New.
25221 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25223 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25224 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25225 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25226 VUNSPEC_MRC2): New.
25228 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25230 * config/arm/arm.md (*ldc): New.
25234 * config/arm/arm.c (arm_coproc_builtin_available): Add
25235 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25236 (arm_coproc_ldc_stc_legitimate_address): New.
25237 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25238 'qualifier_const_pointer'.
25239 (LDC_QUALIFIERS): Define to...
25240 (arm_ldc_qualifiers): ... this. New.
25241 (STC_QUALIFIERS): Define to...
25242 (arm_stc_qualifiers): ... this. New.
25243 * config/arm/arm-protos.h
25244 (arm_coproc_ldc_stc_legitimate_address): New.
25245 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25246 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25247 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25248 stc2, stcl, stc2l): New.
25249 * config/arm/constraints.md (Uz): New.
25250 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25251 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25252 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25253 VUNSPEC_STC2L): New.
25255 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25257 * config/arm/arm.md (<cdp>): New.
25258 * config/arm/arm.c (neon_const_bounds): Rename this ...
25259 (arm_const_bounds): ... this.
25260 (arm_coproc_builtin_available): New.
25261 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25262 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25263 (CDP_QUALIFIERS): Define to...
25264 (arm_cdp_qualifiers): ... this. New.
25266 (arm_expand_builtin_args): Add case for 6 arguments.
25267 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25268 (arm_const_bounds): ... this.
25269 (arm_coproc_builtin_available): New.
25270 * config/arm/arm_acle.h (__arm_cdp): New.
25272 * config/arm/arm_acle_builtins.def (cdp): New.
25274 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25275 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25276 'arm_const_bounds'.
25277 * config/arm/types.md (coproc): New.
25278 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25279 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25280 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25281 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25283 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25285 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25286 (UBINOP_QUALIFIERS): New.
25288 (acle_builtin_data): New. Change comment.
25289 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25290 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25291 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25292 arm_acle_builtins.def.
25293 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25294 (arm_init_acle_builtins): New.
25295 (CRC32_BUILTIN): Remove.
25296 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25297 crc32cb, crc32ch and crc32cw.
25298 (arm_init_crc32_builtins): Remove.
25299 (arm_init_builtins): Use arm_init_acle_builtins rather
25300 than arm_init_crc32_builtins.
25301 (arm_expand_acle_builtin): New.
25302 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25303 * config/arm/arm_acle_builtins.def: New.
25305 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25307 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25308 (arm_builtin_datum): ... this.
25309 (arm_init_neon_builtin): Rename to ...
25310 (arm_init_builtin): ... this. Add a new parameters PREFIX
25311 and USE_SIG_IN_NAME.
25312 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25313 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25314 'arm_builtin_datum'.
25315 (arm_init_vfp_builtins): Likewise.
25316 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25317 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25318 (arm_expand_neon_args): Rename to ...
25319 (arm_expand_builtin_args): ... this. Rename builtin_arg
25320 enum values and differentiate between ARG_BUILTIN_MEMORY
25321 and ARG_BUILTIN_NEON_MEMORY.
25322 (arm_expand_neon_builtin_1): Rename to ...
25323 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25324 values, arm_expand_builtin_args and add bool parameter NEON.
25325 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25326 (arm_expand_vfp_builtin): Likewise.
25327 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25329 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25331 PR middle-end/77484
25332 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25333 * predict.c (tree_estimate_probability_bb): Reverse direction of
25334 polymorphic call predictor.
25336 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25338 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25339 (determine_pass_name_match): ...this new function and...
25340 (should_skip_pass_p): ...this new function.
25342 2017-01-06 Nathan Sidwell <nathan@acm.org>
25344 * ipa-visibility.c (function_and_variable_visibility): Reformat
25345 comments and long lines. Remove extrneous if.
25346 * symtab.c (symtab_node::make_decl_local): Fix code format.
25347 (symtab_node::set_section_for_node): Fix comment typo.
25349 2017-01-06 Martin Liska <mliska@suse.cz>
25352 * lra-constraints.c: Rename invariant to lra_invariant.
25353 * predict.c (set_even_probabilities): Initialize e to NULL.
25355 2017-01-05 Martin Sebor <msebor@redhat.com>
25357 PR tree-optimization/78910
25358 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25359 (format_integer): Correct off-by-one error in the handling
25360 of precision with negative numbers in signed conversions..
25362 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25364 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25366 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25368 PR tree-optimization/71016
25369 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25370 factor_out_conditional_conversion. Formatting fix.
25371 (factor_out_conditional_conversion): Add cond_stmt argument.
25372 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25373 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25376 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25378 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25379 read-rtl-function.o, and selftest-rtl.o.
25380 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25381 (selftest::aarch64_test_loading_full_dump): New function.
25382 (selftest::aarch64_run_selftests): New function.
25383 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25384 selftest::aarch64_run_selftests.
25385 * config/i386/i386.c
25386 (selftest::ix86_test_loading_dump_fragment_1): New function.
25387 (selftest::ix86_test_loading_call_insn): New function.
25388 (selftest::ix86_test_loading_full_dump): New function.
25389 (selftest::ix86_test_loading_unspec): New function.
25390 (selftest::ix86_run_selftests): Call the new functions.
25391 * emit-rtl.c (maybe_set_max_label_num): New function.
25392 * emit-rtl.h (maybe_set_max_label_num): New decl.
25393 * function.c (instantiate_decls): Guard call to
25394 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25395 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25397 * gensupport.c (gen_reader::gen_reader): Pass "false"
25398 for new "compact" param of rtx_reader.
25399 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25400 rather than an empty string for NULL strings.
25401 * read-md.c: Potentially include config.h rather than bconfig.h.
25402 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25403 (have_error): New global, copied from errors.c.
25404 (md_reader::read_name): Rename to...
25405 (md_reader::read_name_1): ...this, adding "out_loc" param,
25406 and converting "missing name or number" to returning false, rather
25408 (md_reader::read_name): Reimplement in terms of read_name_1.
25409 (md_reader::read_name_or_nil): New function.
25410 (md_reader::read_string): Handle "(nil)" by returning NULL.
25411 (md_reader::md_reader): Add new param "compact".
25412 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25413 (md_reader::read_file): New method.
25414 * read-md.h (md_reader::md_reader): Add new param "compact".
25415 (md_reader::read_file): New method.
25416 (md_reader::is_compact): New accessor.
25417 (md_reader::read_name): Convert return type from void to file_location.
25418 (md_reader::read_name_or_nil): New decl.
25419 (md_reader::read_name_1): New decl.
25420 (md_reader::m_compact): New field.
25421 (noop_reader::noop_reader): Pass "false" for new "compact" param
25423 (rtx_reader::rtx_reader): Add new "compact" param.
25424 (rtx_reader::read_rtx_operand): Make virtual and convert return
25425 type from void to rtx.
25426 (rtx_reader::read_until): New decl.
25427 (rtx_reader::handle_any_trailing_information): New virtual function.
25428 (rtx_reader::postprocess): New virtual function.
25429 (rtx_reader::finalize_string): New virtual function.
25430 (rtx_reader::m_in_call_function_usage): New field.
25431 (rtx_reader::m_reuse_rtx_by_id): New field.
25432 * read-rtl-function.c: New file.
25433 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25434 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25435 (selftest::verify_three_block_rtl_cfg): New decl.
25436 * read-rtl-function.h: New file.
25437 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25438 For host, include function.h, memmodel.h, and emit-rtl.h.
25439 (one_time_initialization): New function.
25440 (struct compact_insn_name): New struct.
25441 (compact_insn_names): New array.
25442 (find_code): Handle insn codes in compact dumps.
25443 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25444 (bind_subst_iter_and_attr): Likewise.
25445 (add_condition_to_string): Likewise.
25446 (add_condition_to_rtx): Likewise.
25447 (apply_attribute_uses): Likewise.
25448 (add_current_iterators): Likewise.
25449 (apply_iterators): Likewise.
25450 (initialize_iterators): Guard usage of apply_subst_iterator with
25451 #ifdef GENERATOR_FILE.
25452 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25453 (md_reader::read_mapping): Likewise.
25454 (add_define_attr_for_define_subst): Likewise.
25455 (add_define_subst_attr): Likewise.
25456 (read_subst_mapping): Likewise.
25457 (check_code_iterator): Likewise.
25458 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25460 (one_time_initialization): New function.
25461 (rtx_reader::read_until): New method.
25462 (read_flags): New function.
25463 (parse_reg_note_name): New function.
25464 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25465 Handle reuse_rtx ids.
25466 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25467 Add parsing support for RTL dumps, mirroring the special-cases in
25468 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25469 values, and calling handle_any_trailing_information.
25470 (rtx_reader::read_rtx_operand): Convert return type from void
25471 to rtx, returning return_rtx. Handle case 'e'. Call
25472 finalize_string on XSTR and XTMPL fields.
25473 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25474 "(nil)" values were omitted. Call the postprocess vfunc on the
25476 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25477 class ctor. Initialize m_in_call_function_usage. Call
25478 one_time_initialization.
25479 * rtl-tests.c (selftest::test_uncond_jump): Call
25480 set_new_first_and_last_insn.
25481 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25482 * selftest-rtl.c: New file.
25483 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25484 (selftest::get_insn_by_uid): New decl.
25485 * selftest-run-tests.c (selftest::run_tests): Call
25486 read_rtl_function_c_tests.
25487 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25488 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25491 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25493 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25494 operands in a special way. Assert that pos+len <= mode precision.
25496 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25498 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25499 3 argument Alias with unlimited for the negative form.
25500 (fno-vect-cost-model): Removed.
25502 2017-01-05 Martin Liska <mliska@suse.cz>
25504 * hsa-gen.c (gen_hsa_divmod): New function.
25505 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25507 2017-01-05 Martin Liska <mliska@suse.cz>
25510 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25513 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25515 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25516 small constant length operands.
25518 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25520 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25521 between loop iterations.
25523 2017-01-05 Martin Liska <mliska@suse.cz>
25526 * gimplify.c (gimplify_decl_expr): Compare to
25527 asan_poisoned_variables instread of checking flags.
25528 (gimplify_target_expr): Likewise.
25529 (gimplify_expr): Likewise.
25530 (gimplify_function_tree): Conditionally initialize
25531 asan_poisoned_variables.
25533 2017-01-04 Jeff Law <law@redhat.com>
25535 PR tree-optimizatin/78812
25536 * rtl.h (contains_mem_rtx_p): Prototype.
25537 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25538 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25539 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25540 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25541 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25543 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25545 * input.c (assert_char_at_range): Default-initialize actual_range.
25547 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25549 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25550 to match the caller.
25552 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25554 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25555 insns after final jump in test to emit dummy move.
25557 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25559 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25560 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25562 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25564 * multiple_target.c (create_dispatcher_calls): Init e_next.
25565 * tree-ssa-loop-split.c (split_loop): Init border.
25566 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25569 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25574 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25575 (altivec_register_operand): Do not return true if the operand
25576 contains a SUBREG mixing SImode and SFmode.
25577 (vsx_register_operand): Likewise.
25578 (vsx_reg_sfsubreg_ok): New predicate.
25579 (vfloat_operand): Do not return true if the operand contains a
25580 SUBREG mixing SImode and SFmode.
25581 (vint_operand): Likewise.
25582 (vlogical_operand): Likewise.
25583 (gpc_reg_operand): Likewise.
25584 (int_reg_operand): Likewise.
25585 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25586 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25587 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25589 (rs6000_emit_move_si_sf_subreg): New helper function.
25590 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25591 fixup SUBREGs involving SImode and SFmode.
25592 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25593 numbers for the new peephole2 optimization.
25594 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25595 the GLIBC math library that do AND/IOR/XOR operations on single
25596 precision floating point.
25597 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25598 target macros to say whether we need to avoid SUBREGs mixing
25600 (TARGET_ALLOW_SF_SUBREG): Likewise.
25601 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25602 (UNSPEC_SI_FROM_SF): Likewise.
25603 (iorxor): Change spacing.
25604 (and_ior_xor): New iterator for AND, IOR, and XOR.
25605 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25606 (movdi_from_sf_zero_ext): Likewise.
25607 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25608 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25609 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25610 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25611 (fms<mode>4): Likewise.
25612 (fnma<mode>4): Likewise.
25613 (fnms<mode>4): Likewise.
25614 (nfma<mode>4): Likewise.
25615 (nfms<mode>4): Likewise.
25617 2017-01-04 Marek Polacek <polacek@redhat.com>
25620 * doc/invoke.texi: Document -Wpointer-compare.
25622 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25624 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25627 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25628 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25631 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25634 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25635 documentation of the powerpc_popcntb_ok attribute.
25636 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25637 code to issue warning messages if a requested CPU configuration is
25638 not supported by the binary (assembler and loader) toolchain.
25639 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25640 made to define a built-in function that has been disabled.
25641 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25642 made to define a built-in function that has been disabled.
25643 (altivec_init_builtins): Add comment explaining why definition
25644 of the DST built-in functions is not preceded by an assertion
25645 check. Add assertions to prevent ICE if attempts are made to
25646 define an altivec predicate or an abs* built-in function that has
25648 (htm_init_builtins): Add comment explaining why definition of the
25649 htm built-in functions is not preceded by an assertion check.
25651 2017-01-04 Jeff Law <law@redhat.com>
25653 PR tree-optimizatin/67955
25654 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25655 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25656 the points-to solution does not include pt_null. Use DECL_PT_UID
25659 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25661 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25662 Use gen_int_mode instead of gen_lopwart for const_int operands.
25664 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25666 PR tree-optimization/71563
25667 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25668 out of range value - has low bits known to be zero.
25670 2017-01-04 Alan Modra <amodra@gmail.com>
25672 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25673 * configure: Regenerate.
25674 * config.in: Regenerate.
25676 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25679 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25680 a substring of the message, but strcmp with the whole message. Ifdef
25681 ENABLE_NLS, translate the message first using dgettext.
25683 2017-01-03 Jeff Law <law@redhat.com>
25685 PR tree-optimizatin/78856
25686 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25687 (mark_threaded_blocks): Remove code to truncate thread paths that
25688 cross multiple loop headers. Instead invalidate the cached loop
25689 iteration information and handle case of a thread path walking
25690 into an irreducible region.
25692 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25695 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25696 assertions. Add support for doing the signbit if the IEEE 128-bit
25697 floating point value is in a GPR.
25698 * config/rs6000/rs6000.md (Fsignbit): Delete.
25699 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25700 Update the length attribute if the value is in a GPR.
25701 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25702 the sign or zero extension instruction, since the value is always 0/1.
25703 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25706 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25707 extracting SImode to a GPR register so that we can generate a
25708 store, limit the vector to be in a traditional Altivec register
25709 for the vextuwrx instruction.
25711 2017-01-03 Ian Lance Taylor <iant@google.com>
25713 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25715 2017-01-03 Martin Sebor <msebor@redhat.com>
25717 PR tree-optimization/78696
25718 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25719 precision. Use MPFR for %f for greater fidelity. Correct handling
25721 (pass_sprintf_length::compute_format_length): Set width and precision
25722 specified by asrerisk to void_node for vararg functions.
25723 (try_substitute_return_value): Adjust dump output.
25725 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25727 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25729 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25731 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25732 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25733 -mlra/-mno-lra was passed to the compiler.
25735 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25737 PR rtl-optimization/65618
25738 * emit-rtl.c (try_split): Move initialization of "before" and
25739 "after" to just before the call to emit_insn_after_setloc.
25741 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25743 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25745 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25747 * dwarf2out.c (gen_enumeration_type_die): When
25748 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25750 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25752 PR tree-optimization/78965
25753 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25754 Change first argument from const call_info & to call_info &. For %n
25755 set info.nowrite to false.
25757 PR middle-end/78901
25758 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25759 possibly throwing calls.
25761 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25762 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25763 and fns handling, rather than in a separate case SSA_NAME.
25765 2017-01-02 Jeff Law <law@redhat.com>
25767 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25768 fixes for first_period and second_period variables.
25770 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25773 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25774 (*insvqi_1): New insn pattern.
25775 (*insvqi_1_mem_rex64): Ditto.
25776 (*insvqi_2): Ditto.
25777 (*insvqi_3): Rename from *insvqi.
25779 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25781 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25783 * doc/cfg.texi (Edges): Remove reference to Java.
25784 (Maintaining the CFG): Ditto.
25786 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25788 PR middle-end/77674
25789 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25790 transparent aliases.
25792 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25794 PR middle-end/77484
25795 * predict.def (PRED_CALL): Update hitrate.
25796 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25797 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25798 into direct/indirect/polymorphic variants.
25800 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25802 Update copyright years.
25804 * gcc.c (process_command): Update copyright notice dates.
25805 * gcov-dump.c (print_version): Ditto.
25806 * gcov.c (print_version): Ditto.
25807 * gcov-tool.c (print_version): Ditto.
25808 * gengtype.c (create_file): Ditto.
25809 * doc/cpp.texi: Bump @copying's copyright year.
25810 * doc/cppinternals.texi: Ditto.
25811 * doc/gcc.texi: Ditto.
25812 * doc/gccint.texi: Ditto.
25813 * doc/gcov.texi: Ditto.
25814 * doc/install.texi: Ditto.
25815 * doc/invoke.texi: Ditto.
25817 Copyright (C) 2017 Free Software Foundation, Inc.
25819 Copying and distribution of this file, with or without modification,
25820 are permitted in any medium without royalty provided the copyright
25821 notice and this notice are preserved.