1 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
4 vectors heap vectors. Clean up comments.
5 Make visited_bbs a reference.
6 (profitable_jump_thread_path): Make GC
7 vectors heap vectors. Clean up comments.
9 (convert_and_register_jump_thread_path): Make GC vectors heap
11 (check_subpath_and_update_thread_path): Same. Clean up comments.
12 Make visited_bbs a reference.
13 (handle_phi): Abstract common code to to
14 register_jump_thread_path_if_profitable.
15 Rename VAR_BB to DEF_BB.
17 Make GC vectors heap vectors.
18 Make visited_bbs a reference.
19 (handle_assignment): Same.
20 (register_jump_thread_path_if_profitable): New.
21 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
23 Make GC vectors heap vectors. Clean up comments.
24 Make visited_bbs a reference.
25 (find_jump_threads_backwards): Make visited_bbs live in the stack.
26 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
29 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
32 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
33 words of E_DImode object are reachable by xtensa_uimm8x4 access.
35 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
37 Revert r251800 and r251799.
39 2017-09-11 Martin Jambor <mjambor@suse.cz>
42 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
44 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
45 use it to find predecessor edges.
46 (naive_outof_ssa): Collect vector of predecessors.
48 2017-09-08 Jason Merrill <jason@redhat.com>
50 PR c++/70029 - ICE with ref-qualifier and -flto
51 * langhooks.h (struct lang_hooks_for_types): Add
53 * attribs.c (build_type_attribute_qual_variant): Use it.
54 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
56 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
57 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
59 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
62 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
63 (*mulsi3_sp64): New instruction.
64 (mulsi3): New expander.
66 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
68 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
70 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
72 * sancov.c: Include memmodel.h.
74 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
77 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
80 2017-09-07 Carl Love <cel@us.ibm.com>
82 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
85 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
87 * sancov.c: Include tm_p.h.
89 2017-09-07 Jakub Jelinek <jakub@redhat.com>
92 * output.h (switch_to_other_text_partition): New declaration.
93 * varasm.c (switch_to_other_text_partition): New function.
94 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
95 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
96 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
97 to the other text partition before emitting LCL label and switch back
98 after emitting the word after it.
100 2017-09-07 Richard Biener <rguenther@suse.de>
102 * passes.def (pass_split_crit_edges): Remove instance before PRE.
103 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
104 critical edges here, after loop init.
105 (pass_data_pre): Remove PROP_no_crit_edges flags.
106 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
107 for valueization of call args to avoid leaking VN_TOP.
108 (visit_use): Assert we do not visit default defs.
109 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
110 Use error_mark_node to more easily detect leaking VN_TOP.
111 All default-defs are varying, not VN_TOP. Mark them visited.
112 (run_scc_vn): Make code match comment.
114 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
116 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
117 OPTION_MASK_FLOAT128_KEYWORD.
118 (POWERPC_MASKS): Likewise.
119 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
120 support for the -mfloat128-type option, and make -mfloat128
121 default on PowerPC Linux systems. Define or undefine
122 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
123 Define __float128 to be __ieee128 if IEEE 128-bit support is
124 enabled, or undefine it.
125 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
126 Delete defining __FLOAT128_TYPE__.
127 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
128 -mfloat128-type option and make -mfloat128 default on PowerPC
130 (TARGET_FLOAT128_TYPE): Likewise.
131 (-mfloat128-type): Likewise.
132 * config/rs6000/rs6000.c (rs6000_option_override_internal):
133 Delete the -mfloat128-type option and make -mfloat128 default on
134 PowerPC Linux systems. Always use __ieee128 to be the keyword for
135 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
136 128-bit floating point is enabled. Change tests from using
137 -mfloat128-type to -mfloat128.
138 (rs6000_mangle_type): Use the correct mangling for the __float128
139 type even if normal long double is restricted to 64-bits.
140 (floatn_mode): Enable the _Float128 type by default on VSX Linux
142 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
143 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
144 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
146 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
147 documentation for -mfloat128.
149 2017-09-06 Olivier Hainque <hainque@adacore.com>
151 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
153 2017-09-06 Wish Wu <wishwu007@gmail.com>
154 Jakub Jelinek <jakub@redhat.com>
156 * asan.c (initialize_sanitizer_builtins): Add
157 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
158 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
159 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
160 BT_FN_VOID_UINT64_PTR variables.
161 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
162 (BT_FN_VOID_UINT16_UINT16): Likewise.
163 (BT_FN_VOID_UINT32_UINT32): Likewise.
164 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
165 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
166 (BT_FN_VOID_UINT64_PTR): Likewise.
167 * common.opt (flag_sanitize_coverage): New variable.
168 (fsanitize-coverage=trace-pc): Remove.
169 (fsanitize-coverage=): Add.
170 * flag-types.h (enum sanitize_coverage_code): New enum.
171 * fold-const.c (fold_range_test): Disable non-short-circuit
172 optimization if flag_sanitize_coverage.
173 (fold_truth_andor): Likewise.
174 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
175 * opts.c (COVERAGE_SANITIZER_OPT): Define.
176 (coverage_sanitizer_opts): New array.
177 (get_closest_sanitizer_option): Add OPTS argument, handle also
178 OPT_fsanitize_coverage_.
179 (parse_sanitizer_options): Adjusted to also handle
180 OPT_fsanitize_coverage_.
181 (common_handle_option): Add OPT_fsanitize_coverage_.
182 * sancov.c (instrument_comparison, instrument_switch): New function.
183 (sancov_pass): Add trace-cmp support.
184 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
185 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
186 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
187 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
188 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
189 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
190 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
191 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
192 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
193 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
195 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
197 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
198 error. Only quit immediately if parsing is complete.
199 (BEGIN): Initialize fatal_err and parse_done.
200 (begin fpu, end fpu): Check number of arguments.
201 (begin arch, end arch): Likewise.
202 (begin cpu, end cpu): Likewise.
203 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
204 (optalias): Likewise.
206 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
208 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
209 * config/arm/arm-isa.h: Delete. Move definitions to ...
210 * arm-cpus.in: ... here. Use new feature and fgroup values.
211 * config/arm/arm.c (arm_option_override): Use lower case for feature
213 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
214 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
215 * config/arm/parsecpu.awk (END): Add new command 'isa'.
217 (print_isa_bits_for): New function.
218 (gen_isa): New function.
219 (gen_comm_data): Use print_isa_bits_for.
220 (define feature): New keyword.
221 (define fgroup): New keyword.
222 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
223 (arm-isa.h): Add rule to generate file.
224 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
225 case for feature bit names.
227 2017-09-06 Richard Biener <rguenther@suse.de>
229 * tree-ssa-pre.c (NECESSARY): Remove.
230 (create_expression_by_pieces): Do not touch pass-local flags.
231 (insert_into_preds_of_block): Likewise.
232 (do_pre_regular_insertion): Likewise.
233 (eliminate_insert): Likewise.
234 (eliminate_dom_walker::before_dom_children): Likewise.
235 (fini_eliminate): Do not look at inserted_exprs.
236 (mark_operand_necessary): Remove.
237 (remove_dead_inserted_code): Replace with simple work-list
238 algorithm based on inserted_exprs and SSA uses.
239 (pass_pre::execute): Re-order fini_eliminate and
240 remove_dead_inserted_code.
242 2017-09-06 Olivier Hainque <hainque@adacore.com>
244 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
245 for VxWorks 7. Adjust surrounding comments.
247 2017-09-06 Richard Biener <rguenther@suse.de>
249 * gimple-ssa-strength-reduction.c
250 (find_candidates_dom_walker::before_dom_children): Also allow
253 2017-09-06 Richard Biener <rguenther@suse.de>
255 PR tree-optimization/82108
256 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
257 for gap in the non-permutation SLP case.
259 2017-09-06 Martin Jambor <mjambor@suse.cz>
261 PR tree-optimization/82078
262 * tree-sra.c (sort_and_splice_var_accesses): Move call to
263 add_access_to_work_queue...
264 (build_accesses_from_assign): ...here.
265 (propagate_all_subaccesses): Make sure racc is the group
266 representative, if there is one.
268 2017-09-06 Jakub Jelinek <jakub@redhat.com>
271 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
274 2017-09-06 Richard Biener <rguenther@suse.de>
276 * gimple-ssa-strength-reduction.c
277 (find_candidates_dom_walker::before_doom_children): Use a
278 type and not a mode check.
280 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
283 * config/arm/predicates.md (arm_general_adddi_operand): Create new
285 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
287 2017-09-05 Jeff Law <law@redhat.com>
289 PR tree-optimization/64910
290 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
291 cases where we have 3 or more operands.
293 2017-09-05 Jakub Jelinek <jakub@redhat.com>
296 * omp-low.c (lower_omp_for): Recompute tree invariant if
297 gimple_omp_for_initial/final is ADDR_EXPR.
300 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
301 into gimple val before gimplification fo the COND_EXPR.
303 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
305 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
306 REGION_COPY argument.
307 (thread_through_all_blocks): Remove unused argument to
308 duplicate_thread_path.
310 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
311 Alan Hayward <alan.hayward@arm.com>
312 David Sherwood <david.sherwood@arm.com>
314 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
315 Take a scalar_mode rather than a machine_mode.
316 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
317 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
318 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
319 (aarch64_gen_adjusted_ldpstp): Likewise.
320 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
322 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
323 Alan Hayward <alan.hayward@arm.com>
324 David Sherwood <david.sherwood@arm.com>
326 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
327 Take a scalar_int_mode instead of a machine_mode.
328 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
329 (aarch64_output_scalar_simd_mov_immediate): Likewise.
330 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
331 (aarch64_simd_attr_length_rglist): Delete.
332 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
333 a scalar_int_mode instead of a machine_mode.
334 (aarch64_add_offset): Likewise.
335 (aarch64_internal_mov_immediate): Likewise
336 (aarch64_add_constant_internal): Likewise.
337 (aarch64_add_constant): Likewise.
338 (aarch64_movw_imm): Likewise.
339 (aarch64_rtx_arith_op_extract_p): Likewise.
340 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
341 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
342 Remove assert that the mode isn't a vector.
343 (aarch64_output_scalar_simd_mov_immediate): Likewise.
344 (aarch64_expand_mov_immediate): Update calls after above changes.
345 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
346 (aarch64_and_bitmask_imm): Check for scalar integer modes.
347 (aarch64_move_imm): Likewise.
348 (aarch64_can_const_movi_rtx_p): Likewise.
349 (aarch64_strip_extend): Likewise.
350 (aarch64_extr_rtx_p): Likewise.
351 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
352 a CONST_INT when the mode parameter is VOIDmode.
353 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
355 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
357 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
358 * stor-layout.c (bitwise_mode_for_mode): Likewise.
359 (bitwise_type_for_mode): Update accordingly.
361 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
363 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
364 * stor-layout.c (mode_for_size_tree): Likewise.
365 (mode_for_array): Update accordingly.
366 (layout_decl): Likewise.
367 (compute_record_mode): Likewise. Only set the mode once.
369 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
371 * target.def (get_mask_mode): Change return type to opt_mode.
373 * doc/tm.texi: Regenerate.
374 * targhooks.h (default_get_mask_mode): Return an opt_mode.
375 * targhooks.c (default_get_mask_mode): Likewise.
376 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
377 * optabs-query.c (can_vec_mask_load_store_p): Update use of
378 targetm.get_mask_mode.
379 * tree.c (build_truth_vector_type): Likewise.
381 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
383 * machmode.h (mode_for_vector): Return an opt_mode.
384 * stor-layout.c (mode_for_vector): Likewise.
385 (mode_for_int_vector): Update accordingly.
386 (layout_type): Likewise.
387 * config/i386/i386.c (emit_memmov): Likewise.
388 (ix86_expand_set_or_movmem): Likewise.
389 (ix86_expand_vector_init): Likewise.
390 (ix86_get_mask_mode): Likewise.
391 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
393 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
394 * expmed.c (extract_bit_field_1): Likewise.
395 * expr.c (expand_expr_real_2): Likewise.
396 * optabs-query.c (can_vec_perm_p): Likewise.
397 (can_vec_mask_load_store_p): Likewise.
398 * optabs.c (expand_vec_perm): Likewise.
399 * targhooks.c (default_get_mask_mode): Likewise.
400 * tree-vect-stmts.c (vectorizable_store): Likewise.
401 (vectorizable_load): Likewise.
402 (get_vectype_for_scalar_type_and_size): Likewise.
404 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
406 * machmode.h (mode_for_int_vector): New function.
407 * stor-layout.c (mode_for_int_vector): Likewise.
408 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
409 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
410 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
411 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
412 (s390_expand_vcond): Likewise.
414 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
416 * machmode.h (opt_machine_mode): New type.
417 (opt_mode<T>): Allow construction from anything that can be
419 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
420 (mode_for_size): Return an opt_machine_mode.
421 * stor-layout.c (mode_for_size): Likewise.
422 (mode_for_size_tree): Update call accordingly.
423 (bitwise_mode_for_mode): Likewise.
424 (make_fract_type): Likewise.
425 (make_accum_type): Likewise.
426 * caller-save.c (replace_reg_with_saved_mem): Update call
428 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
429 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
430 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
431 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
432 * expmed.c (extract_bit_field_1): Likewise.
433 * reload.c (get_secondary_mem): Likewise.
434 * varasm.c (assemble_integer): Likewise.
435 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
438 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
440 * machmode.h (decimal_float_mode_for_size): New function.
441 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
442 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
443 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
444 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
445 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
446 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
448 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
450 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
451 (get_builtin_sync_mode): Likewise.
452 (expand_ifn_atomic_compare_exchange): Likewise.
453 (expand_builtin_atomic_clear): Likewise.
454 (expand_builtin_atomic_test_and_set): Likewise.
455 (fold_builtin_atomic_always_lock_free): Likewise.
456 * calls.c (compute_argument_addresses): Likewise.
457 (emit_library_call_value_1): Likewise.
458 (store_one_arg): Likewise.
459 * combine.c (combine_instructions): Likewise.
460 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
461 * config/arm/arm.c (arm_function_value): Likewise.
462 (aapcs_allocate_return_reg): Likewise.
463 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
464 * config/i386/i386.c (construct_container): Likewise.
465 (ix86_gimplify_va_arg): Likewise.
466 (ix86_expand_sse_cmp): Likewise.
467 (emit_memmov): Likewise.
468 (emit_memset): Likewise.
469 (expand_small_movmem_or_setmem): Likewise.
470 (ix86_expand_pextr): Likewise.
471 (ix86_expand_pinsr): Likewise.
472 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
473 * config/microblaze/microblaze.c (microblaze_block_move_straight):
475 * config/mips/mips.c (mips_function_value_1) Likewise.
476 (mips_block_move_straight): Likewise.
477 (mips_expand_ins_as_unaligned_store): Likewise.
478 * config/powerpcspe/powerpcspe.c
479 (rs6000_darwin64_record_arg_advance_flush): Likewise.
480 (rs6000_darwin64_record_arg_flush): Likewise.
481 * config/rs6000/rs6000.c
482 (rs6000_darwin64_record_arg_advance_flush): Likewise.
483 (rs6000_darwin64_record_arg_flush): Likewise.
484 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
485 (sparc_function_value_1): Likewise.
486 * config/spu/spu.c (adjust_operand): Likewise.
487 (spu_emit_branch_or_set): Likewise.
488 (arith_immediate_p): Likewise.
489 * emit-rtl.c (gen_lowpart_common): Likewise.
490 * expr.c (expand_expr_real_1): Likewise.
491 * function.c (assign_parm_setup_block): Likewise.
492 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
493 * reload1.c (alter_reg): Likewise.
494 * stor-layout.c (mode_for_vector): Likewise.
495 (layout_type): Likewise.
497 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
499 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
500 (spu_convert_move): Likewise.
501 * lower-subreg.c (resolve_simple_move): Likewise.
503 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
506 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
507 define_insn to a define_expand.
508 (altivec_vsum2sws_direct): New define_insn.
509 (altivec_vsumsws): Convert from a define_insn to a define_expand.
511 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
513 * config/arm/arm.c (arm_option_params_internal): Improve setting of
516 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
521 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
522 realign stack if stack alignment needed is less than incoming
525 2017-09-05 Marek Polacek <polacek@redhat.com>
528 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
531 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
533 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
535 2017-09-05 Richard Biener <rguenther@suse.de>
537 PR tree-optimization/82084
538 * fold-const.c (can_native_encode_string_p): Handle wide characters.
540 2017-09-05 Richard Biener <rguenther@suse.de>
542 PR tree-optimization/82102
543 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
545 2017-09-05 Martin Liska <mliska@suse.cz>
547 PR tree-optimization/82032
548 * tree-cfg.c (generate_range_test): New function.
549 * tree-cfg.h (generate_range_test): Declared here.
550 * tree-cfgcleanup.c (convert_single_case_switch): New function.
551 (cleanup_control_expr_graph): Use it.
552 * tree-switch-conversion.c (try_switch_expansion): Remove
554 (emit_case_nodes): Use generate_range_test.
556 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
559 * config/i386/i386.md (*<btsc><mode>_mask): Add
560 TARGET_USE_BT to insn constraint.
561 (*btr<mode>_mask): Ditto.
563 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
565 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
566 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
568 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
571 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
572 TARGET_NEON and TARGET_IWMMXT.
573 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
574 TARGET_NEON and TARGET_IWMMXT.
575 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
577 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
579 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
580 (ix86_rewrite_tls_address): Ditto.
581 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
582 (ix86_rewrite_tls_address_1): Ditto.
583 (ix86_rewrite_tls_address): Ditto.
584 * config/i386/predicates.md (tls_address_pattern): New predicate.
585 * config/i386/i386.md (TLS address splitter): New splitter.
587 2017-09-04 Richard Biener <rguenther@suse.de>
589 PR tree-optimization/82084
590 * fold-const.h (can_native_encode_string_p): Declare.
591 * fold-const.c (can_native_encode_string_p): Factor out from ...
592 (native_encode_string): ... here.
593 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
594 vectorizing stores from constants we later cannot handle.
596 2017-09-04 Marek Polacek <polacek@redhat.com>
599 * doc/invoke.texi: Update -Wtautological-compare documentation.
601 2017-09-04 Jeff Law <law@redhat.com>
603 PR tree-optimization/64910
604 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
605 swap the first and last operand if the last is a constant.
607 2017-09-04 Marek Polacek <polacek@redhat.com>
610 * convert.c (do_narrow): When sanitizing signed integer overflows,
611 bail out for signed types.
612 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
614 2017-09-04 Richard Biener <rguenther@suse.de>
616 PR tree-optimization/82060
617 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
618 Move devirtualization after stmt folding and before EH/AB/noreturn
619 cleanup to get the stmt refs canonicalized. Use a bool instead
620 of gimple_modified_p since that doesn't work for NOPs. Schedule
621 NOPs generated by folding for removal.
623 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
624 Alan Hayward <alan.hayward@arm.com>
625 David Sherwood <david.sherwood@arm.com>
627 * coretypes.h (pad_direction): New enum.
628 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
629 (FUNCTION_ARG_PADDING): Likewise.
630 * target.def (function_arg_padding): New hook.
631 * targhooks.h (default_function_arg_padding): Declare.
632 * targhooks.c (default_function_arg_padding): New function.
633 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
634 (TARGET_FUNCTION_ARG_PADDING): ...this.
635 * doc/tm.texi: Regenerate.
636 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
637 instead of direction.
638 (compute_argument_addresses): Likewise.
639 (load_register_parameters): Likewise.
640 (emit_library_call_value_1): Likewise.
641 (store_one_arg): Use targetm.calls.function_arg_padding instead
642 of FUNCTION_ARG_PADDING.
643 (must_pass_in_stack_var_size_or_pad): Likewise.
644 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
645 (emit_group_store): Likewise.
646 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
647 instead of FUNCTION_ARG_PADDING.
648 (emit_push_insn): Likewise, and propagate enum change throughout
650 * function.h (direction): Delete.
651 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
653 * function.c (assign_parm_find_stack_rtl): Likewise.
654 (assign_parm_setup_block_p): Likewise.
655 (assign_parm_setup_block): Likewise.
656 (gimplify_parameters): Likewise.
657 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
658 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
660 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
661 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
662 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
663 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
664 (aarch64_function_arg_padding): ...this new function.
665 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
666 (TARGET_FUNCTION_ARG_PADDING): Redefine.
667 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
668 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
669 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
670 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
671 (arm_pad_arg_upward): Replace with...
672 (arm_function_arg_padding): ...this new function.
673 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
675 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
676 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
677 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
678 (ia64_hpux_function_arg_padding): Replace with...
679 (ia64_function_arg_padding): ...this new function. Use pad_direction
680 instead of direction. Check for TARGET_HPUX.
681 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
682 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
683 (iq2000_function_arg_padding): New function.
684 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
685 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
686 (mips_function_arg_padding): ...this new function.
687 (mips_pad_reg_upward): Update accordingly.
688 (TARGET_FUNCTION_ARG_PADDING): Redefine.
689 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
690 targetm.calls.function_arg_padding.
691 (FUNCTION_ARG_PADDING): Delete.
692 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
693 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
694 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
695 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
696 (nios2_block_reg_padding): Return pad_direction instead of direction.
697 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
698 instead of direction.
699 (nios2_function_arg_padding): Likewise. Make static.
700 (TARGET_FUNCTION_ARG_PADDING): Redefine.
701 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
702 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
703 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
704 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
705 (pa_function_arg_padding): Make static. Return pad_direction instead
707 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
708 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
709 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
710 instead of direction. Use targetm.calls.function_arg_padding.
711 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
712 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
713 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
714 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
715 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
717 (function_arg_padding): Rename to...
718 (rs6000_function_arg_padding): ...this. Make static. Return
719 pad_direction instead of direction.
720 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
721 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
722 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
723 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
724 instead of direction. Use targetm.calls.function_arg_padding.
725 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
726 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
727 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
728 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
729 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
730 (function_arg_padding): Rename to...
731 (rs6000_function_arg_padding): ...this. Make static. Return
732 pad_direction instead of direction.
733 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
734 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
735 * config/s390/s390.c (s390_function_arg_padding): New function.
736 (TARGET_FUNCTION_ARG_PADDING): Redefine.
737 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
738 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
739 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
740 (function_arg_padding): Rename to...
741 (sparc_function_arg_padding): ...this. Make static. Return
742 pad_direction instead of direction.
743 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
744 * config/spu/spu.c (spu_function_arg_padding): New function.
745 (TARGET_FUNCTION_ARG_PADDING): Redefine.
746 * system.h (FUNCTION_ARG_PADDING): Poison.
748 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
749 Alan Hayward <alan.hayward@arm.com>
750 David Sherwood <david.sherwood@arm.com>
752 * target.def (modes_tieable_p): New hook.
753 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
754 (TARGET_MODES_TIEABLE_P): ...this.
755 * doc/tm.texi.in: Regenerate.
756 * hooks.h (hook_bool_mode_mode_true): Declare.
757 * hooks.c (hook_bool_mode_mode_true): New function.
758 * combine.c (subst): Use targetm.modes_tieable_p instead of
760 * dse.c (find_shift_sequence): Likewise.
761 * expmed.c (extract_low_bits): Likewise.
762 * lower-subreg.c: Include target.h.
763 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
765 * rtlanal.c (rtx_cost): Likewise.
766 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
767 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
768 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
769 (TARGET_MODES_TIEABLE_P): Redefine.
770 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
771 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
772 (TARGET_MODES_TIEABLE_P): Redefine.
773 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
774 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
775 (arc_modes_tieable_p): New function.
776 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
777 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
778 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
779 (arm_modes_tieable_p): Make static.
780 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
781 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
782 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
783 (TARGET_MODES_TIEABLE_P): Redefine.
784 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
785 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
786 (TARGET_MODES_TIEABLE_P): Redefine.
787 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
788 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
789 (cr16_modes_tieable_p): New function.
790 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
791 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
792 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
793 (TRULY_NOOP_TRUNCATION): Update comment.
794 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
795 (TRULY_NOOP_TRUNCATION): Update comment.
796 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
797 (frv_modes_tieable_p): New function.
798 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
799 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
800 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
801 (TARGET_MODES_TIEABLE_P): Redefine.
802 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
803 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
804 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
805 (TARGET_MODES_TIEABLE_P): Redefine.
806 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
807 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
808 (ia64_modes_tieable_p): New function.
809 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
810 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
811 (iq2000_modes_tieable_p): New function.
812 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
813 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
814 (lm32_modes_tieable_p): New function.
815 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
816 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
817 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
818 (TARGET_MODES_TIEABLE_P): Redefine.
819 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
820 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
821 (m32r_modes_tieable_p): New function.
822 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
823 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
824 (m68k_modes_tieable_p): New function.
825 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
826 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
827 (mcore_modes_tieable_p): New function.
828 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
829 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
831 (TARGET_MODES_TIEABLE_P): Redefine.
832 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
833 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
834 * config/mips/mips.c (mips_modes_tieable_p): Make static.
835 (TARGET_MODES_TIEABLE_P): Redefine.
836 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
837 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
838 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
839 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
840 (mn10300_modes_tieable_p): ...this and make static.
841 (TARGET_MODES_TIEABLE_P): Redefine.
842 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
843 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
844 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
845 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
846 (msp430_modes_tieable_p): Make static.
847 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
848 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
849 (TARGET_MODES_TIEABLE_P): Redefine.
850 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
851 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
852 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
853 (TARGET_MODES_TIEABLE_P): Redefine.
854 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
855 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
856 * config/pa/pa.c (pa_modes_tieable_p): Make static.
857 (TARGET_MODES_TIEABLE_P): Redefine.
858 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
859 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
860 (pdp11_modes_tieable_p): New function.
861 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
862 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
863 (rs6000_modes_tieable_p): New function.
864 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
865 * config/powerpcspe/powerpcspe.md: Update comment.
866 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
867 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
868 (TARGET_MODES_TIEABLE_P): Redefine.
869 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
870 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
871 (rl78_modes_tieable_p): New function.
872 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
873 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
874 (rs6000_modes_tieable_p): New function.
875 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
876 * config/rs6000/rs6000.md: Update comment.
877 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
878 * config/rx/rx.c (rx_modes_tieable_p): New function.
879 (TARGET_MODES_TIEABLE_P): Redefine.
880 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
881 * config/s390/s390.c (s390_modes_tieable_p): New function.
882 (TARGET_MODES_TIEABLE_P): Redefine.
883 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
884 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
885 (sh_modes_tieable_p): New function.
886 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
887 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
888 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
889 (sparc_modes_tieable_p): Make static.
890 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
891 * config/spu/spu.c (spu_modes_tieable_p): New function.
892 (TARGET_MODES_TIEABLE_P): Redefine.
893 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
894 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
895 (TARGET_MODES_TIEABLE_P): Redefine.
896 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
897 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
898 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
899 * config/v850/v850.c (v850_modes_tieable_p): New function.
900 (TARGET_MODES_TIEABLE_P): Redefine.
901 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
902 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
903 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
904 (visium_modes_tieable_p): New function.
905 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
906 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
907 (xtensa_modes_tieable_p): New function.
908 * system.h (MODES_TIEABLE_P): Poison.
910 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
911 Alan Hayward <alan.hayward@arm.com>
912 David Sherwood <david.sherwood@arm.com>
914 * target.def (hard_regno_mode_ok): New hook.
915 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
916 (TARGET_HARD_REGNO_MODE_OK): ...this.
917 * doc/tm.texi.in: Regenerate.
918 * hooks.h (hook_bool_uint_mode_true): Declare.
919 * hooks.c (hook_bool_uint_mode_true): New function.
920 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
922 * genpreds.c (write_insn_preds_c): Add an include of target.h.
923 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
924 instead of HARD_REGNO_MODE_OK.
925 * caller-save.c: Include target.h.
926 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
928 * combine.c (can_combine_p): Likewise.
929 (combinable_i3pat): Likewise.
930 (can_change_dest_mode): Likewise.
931 * expr.c (init_expr_target): Likewise.
932 (convert_move): Likewise.
933 (convert_modes): Likewise.
934 * ira.c (setup_prohibited_class_mode_regs): Likewise.
935 (setup_prohibited_mode_move_regs): Likewise.
936 * ira.h (target_ira): Likewise.
937 * lra-assigns.c (find_hard_regno_for_1): Likewise.
938 * lra-constraints.c (process_alt_operands): Likewise.
939 (split_reg): Likewise.
940 * recog.c (peep2_find_free_register): Likewise.
941 * ree.c (combine_reaching_defs): Likewise.
942 * regcprop.c (maybe_mode_change): Likewise.
943 * reginfo.c (init_reg_sets_1): Likewise.
944 (choose_hard_reg_mode): Likewise.
945 (simplifiable_subregs): Likewise.
946 * regrename.c (check_new_reg_p): Likewise.
947 * reload.c (find_valid_class): Likewise.
948 (find_valid_class_1): Likewise.
949 (reload_inner_reg_of_subreg): Likewise.
950 (push_reload): Likewise.
951 (combine_reloads): Likewise.
952 (find_dummy_reload): Likewise.
953 (find_reloads): Likewise.
954 * reload1.c (find_reg): Likewise.
955 (set_reload_reg): Likewise.
956 (allocate_reload_reg): Likewise.
957 (choose_reload_regs): Likewise.
958 (reload_adjust_reg_for_temp): Likewise.
959 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
960 (simplify_subreg_regno): Likewise.
961 * sel-sched.c (init_regs_for_mode): Likewise.
962 * varasm.c (make_decl_rtl): Likewise.
963 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
964 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
966 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
967 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
968 (TARGET_HARD_REGNO_MODE_OK): Redefine.
969 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
970 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
971 (TARGET_HARD_REGNO_MODE_OK): Redefine.
972 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
973 (arc_mode_class): Delete.
974 (HARD_REGNO_MODE_OK): Delete.
975 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
976 (arc_hard_regno_mode_ok): Rename old array to...
977 (arc_hard_regno_mode_ok_modes): ...this.
978 (arc_conditional_register_usage): Update accordingly.
979 (arc_mode_class): Make static.
980 (arc_hard_regno_mode_ok): New function.
981 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
982 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
983 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
984 (arm_hard_regno_mode_ok): Make static.
985 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
987 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
988 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
989 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
991 (TARGET_HARD_REGNO_MODE_OK): Redefine.
992 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
993 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
994 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
995 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
996 (TARGET_HARD_REGNO_MODE_OK): Redefine.
997 * config/bfin/predicates.md (valid_reg_operand): Use
998 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
999 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1000 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1002 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1003 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1004 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1005 (cr16_hard_regno_mode_ok): Make static and return a bool.
1006 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1007 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1008 (cris_hard_regno_mode_ok): New function.
1009 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1010 (epiphany_mode_class): Delete.
1011 (HARD_REGNO_MODE_OK): Delete.
1012 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1013 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1014 (hard_regno_mode_ok): Rename to...
1015 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1016 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1017 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1019 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1020 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1021 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1022 (frv_hard_regno_mode_ok): Make static and return a bool.
1023 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1025 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1026 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1027 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1028 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1030 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1031 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1032 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1033 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1035 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1036 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1037 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1038 (ia64_hard_regno_mode_ok): New function.
1039 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1040 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1041 (iq2000_hard_regno_mode_ok): New function.
1042 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1043 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1044 (lm32_hard_regno_mode_ok): New function.
1045 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1046 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1047 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1048 instead of HARD_REGNO_MODE_OK.
1049 (m32c_hard_regno_ok): Rename to...
1050 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1051 (m32c_cannot_change_mode_class): Update accordingly.
1052 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1053 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1054 (m32r_mode_class): Delete.
1055 (HARD_REGNO_MODE_OK): Delete.
1056 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1057 (m32r_hard_regno_mode_ok): Rename to...
1058 (m32r_hard_regno_modes): ...this.
1059 (m32r_mode_class): Make static.
1060 (m32r_hard_regno_mode_ok): New function.
1061 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1062 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1063 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1064 (m68k_hard_regno_mode_ok): Make static.
1065 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1066 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1067 (mcore_hard_regno_mode_ok): New function.
1068 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1069 (HARD_REGNO_MODE_OK): Delete.
1070 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1072 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1073 (microblaze_hard_regno_mode_ok): New function.
1074 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1075 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1076 (mips_hard_regno_mode_ok): Delete.
1077 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1078 (mips_hard_regno_mode_ok_p): ...this and make static.
1079 (mips_hard_regno_mode_ok_p): Rename to...
1080 (mips_hard_regno_mode_ok_uncached): ...this.
1081 (mips_hard_regno_mode_ok): New function.
1082 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1083 of HARD_REGNO_MODE_OK.
1084 (mips_option_override): Update after above name changes.
1085 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1086 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1087 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1088 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1089 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1090 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1091 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1092 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1093 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1094 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1095 (msp430_hard_regno_mode_ok): Make static and return a bool.
1096 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1097 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1098 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1100 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1101 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1102 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1103 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1104 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1105 (PA_HARD_REGNO_MODE_OK): ...this
1106 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1107 (PA_HARD_REGNO_MODE_OK): ...this.
1108 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1109 (pa_hard_regno_mode_ok): New function.
1110 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1111 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1112 (pdp11_hard_regno_mode_ok): New function.
1113 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1114 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1116 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1118 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1119 (rs6000_hard_regno_mode_ok): Rename to...
1120 (rs6000_hard_regno_mode_ok_uncached): ...this.
1121 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1122 (rs6000_hard_regno_mode_ok): New function.
1123 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1124 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1125 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1126 (riscv_hard_regno_mode_ok): ...this and make static.
1127 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1128 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1129 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1130 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1131 (rl78_hard_regno_mode_ok): Make static and return bool.
1132 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1133 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1135 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1136 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1137 (rs6000_hard_regno_mode_ok): Rename to...
1138 (rs6000_hard_regno_mode_ok_uncached): ...this.
1139 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1140 (rs6000_hard_regno_mode_ok): New function.
1141 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1142 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1143 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1144 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1145 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1146 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1147 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1148 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1149 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1150 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1151 (sh_hard_regno_mode_ok): Make static.
1152 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1153 instead of HARD_REGNO_MODE_OK.
1154 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1155 (sparc_mode_class): Delete.
1156 (HARD_REGNO_MODE_OK): Delete.
1157 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1158 (hard_regno_mode_classes): Make static.
1159 (sparc_mode_class): Likewise.
1160 (sparc_hard_regno_mode_ok): New function.
1161 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1162 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1163 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1165 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1166 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1167 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1168 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1169 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1170 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1171 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1172 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1173 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1174 (visium_hard_regno_mode_ok): New function.
1175 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1176 instead of HARD_REGNO_MODE_OK.
1177 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1178 (HARD_REGNO_MODE_OK): Delete.
1179 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1180 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1181 (xtensa_option_override): Update accordingly.
1182 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1183 (xtensa_hard_regno_mode_ok): New function.
1184 * system.h (HARD_REGNO_MODE_OK): Poison.
1186 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1187 Alan Hayward <alan.hayward@arm.com>
1188 David Sherwood <david.sherwood@arm.com>
1190 * target.def (hard_regno_call_part_clobbered): New hook.
1191 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1192 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1193 * doc/tm.texi: Regenerate.
1194 * hooks.h (hook_bool_uint_mode_false): Declare.
1195 * hooks.c (hook_bool_uint_mode_false): New function.
1196 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1197 * cselib.c (cselib_process_insn): Use
1198 targetm.hard_regno_call_part_clobbered instead of
1199 HARD_REGNO_CALL_PART_CLOBBERED.
1200 * ira-conflicts.c (ira_build_conflicts): Likewise.
1201 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1202 * lra-constraints.c (need_for_call_save_p): Likewise.
1203 * lra-lives.c: Include target.h.
1204 (check_pseudos_live_through_calls): Use
1205 targetm.hard_regno_call_part_clobbered instead of
1206 HARD_REGNO_CALL_PART_CLOBBERED.
1207 * regcprop.c: Include target.h.
1208 (copyprop_hardreg_forward_1): Use
1209 targetm.hard_regno_call_part_clobbered instead of
1210 HARD_REGNO_CALL_PART_CLOBBERED.
1211 * reginfo.c (choose_hard_reg_mode): Likewise.
1212 * regrename.c (check_new_reg_p): Likewise.
1213 * reload.c (find_equiv_reg): Likewise.
1214 * reload1.c (emit_reload_insns): Likewise.
1215 * sched-deps.c (deps_analyze_insn): Likewise.
1216 * sel-sched.c (init_regs_for_mode): Likewise.
1217 (mark_unavailable_hard_regs): Likewise.
1218 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1219 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1220 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1222 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1223 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1224 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1226 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1228 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1229 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1230 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1232 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1233 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1234 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1236 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1237 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1239 * config/powerpcspe/powerpcspe.c
1240 (rs6000_hard_regno_call_part_clobbered): New function.
1241 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1242 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1243 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1245 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1246 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1247 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1249 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1250 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1251 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1253 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1254 Alan Hayward <alan.hayward@arm.com>
1255 David Sherwood <david.sherwood@arm.com>
1257 * rtl.h (subreg_memory_offset): Declare.
1258 * emit-rtl.c (subreg_memory_offset): New function.
1259 * expmed.c (store_bit_field_1): Use it.
1260 * expr.c (undefined_operand_subword_p): Likewise.
1261 * simplify-rtx.c (simplify_subreg): Likewise.
1263 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1265 PR rtl-optimization/57448
1268 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1269 using atomic_load pattern.
1270 (expand_atomic_store): Likewise.
1272 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1275 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1276 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1277 handling. Use replace_call_with_value with NULL instead of
1278 gsi_replace, unlink_stmt_vdef and release_defs.
1280 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1283 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1285 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1288 * rtl.h (emit_library_call_value_1): Declare.
1289 (emit_library_call): Replace declaration with a series of overloads.
1290 Remove the parameter count argument.
1291 (emit_library_call_value): Likewise.
1292 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1293 with an "rtx_mode_t *".
1294 (emit_library_call_value): Delete.
1295 (emit_library_call): Likewise.
1296 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1297 (asan_emit_allocas_unpoison): Likewise.
1298 * builtins.c (expand_builtin_powi): Likewise.
1299 (expand_asan_emit_allocas_unpoison): Likewise.
1300 * cfgexpand.c (expand_main_function): Likewise.
1301 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1302 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1303 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1304 * config/arm/arm.c (arm_trampoline_init): Likewise.
1305 (arm_call_tls_get_addr): Likewise.
1306 (arm_expand_divmod_libfunc): Likewise.
1307 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1308 (smulsi3_highpart): Likewise.
1309 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1310 (c6x_expand_compare): Likewise.
1311 (c6x_expand_movmem): Likewise.
1312 * config/frv/frv.c (frv_trampoline_init): Likewise.
1313 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1314 (ix86_expand_divmod_libfunc): Likewise.
1315 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1316 (ia64_expand_compare): Likewise.
1317 (ia64_profile_hook): Likewise.
1318 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1319 (nonlocal_goto): Likewise.
1320 (restore_stack_nonlocal): Likewise.
1321 * config/m32r/m32r.c (block_move_call): Likewise.
1322 (m32r_trampoline_init): Likewise.
1323 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1324 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1325 (m68k_call_m68k_read_tp): Likewise.
1326 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1327 (microblaze_expand_divide): Likewise.
1328 * config/mips/mips.h (mips_args): Likewise.
1329 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1330 (MIPS_ICACHE_SYNC): Likewise.
1331 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1332 (nios2_trampoline_init): Likewise.
1333 * config/pa/pa.c (hppa_tls_call): Likewise.
1334 (pa_trampoline_init): Likewise.
1335 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1336 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1337 (expand_strn_compare): Likewise.
1338 (rs6000_generate_compare): Likewise.
1339 (rs6000_expand_float128_convert): Likewise.
1340 (output_profile_hook): Likewise.
1341 (rs6000_trampoline_init): Likewise.
1342 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1343 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1344 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1345 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1346 (rs6000_generate_compare): Likewise.
1347 (rs6000_expand_float128_convert): Likewise.
1348 (output_profile_hook): Likewise.
1349 (rs6000_trampoline_init): Likewise.
1350 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1351 * config/sh/sh.c (sh_trampoline_init): Likewise.
1352 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1353 (sparc_emit_float_lib_cmp): Likewise.
1354 (sparc32_initialize_trampoline): Likewise.
1355 (sparc64_initialize_trampoline): Likewise.
1356 (sparc_profile_hook): Likewise.
1357 * config/spu/spu.c (ea_load_store): Likewise.
1358 * config/spu/spu.md (floatunssidf2): Likewise.
1359 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1360 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1361 * config/visium/visium.c (expand_block_move_4): Likewise.
1362 (expand_block_move_2): Likewise.
1363 (expand_block_move_1): Likewise.
1364 (expand_block_set_4): Likewise.
1365 (expand_block_set_2): Likewise.
1366 (expand_block_set_1): Likewise.
1367 (visium_trampoline_init): Likewise.
1368 (visium_profile_hook): Likewise.
1369 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1370 (xtensa_setup_frame_addresses): Likewise.
1371 (xtensa_trampoline_init): Likewise.
1372 * except.c (sjlj_emit_function_enter): Likewise.
1373 (sjlj_emit_function_exit): Likewise.
1374 * explow.c (allocate_dynamic_stack_space): Likewise.
1375 (probe_stack_range): Likewise.
1376 * expr.c (convert_mode_scalar): Likewise.
1377 * optabs.c (expand_binop): Likewise.
1378 (expand_twoval_binop_libfunc): Likewise.
1379 (expand_unop): Likewise.
1380 (prepare_cmp_insn): Likewise.
1381 (prepare_float_lib_cmp): Likewise.
1382 (expand_float): Likewise.
1383 (expand_fix): Likewise.
1384 (expand_fixed_convert): Likewise.
1385 (maybe_emit_sync_lock_test_and_set): Likewise.
1386 (expand_atomic_compare_and_swap): Likewise.
1387 (expand_mem_thread_fence): Likewise.
1388 (expand_atomic_fetch_op): Likewise.
1390 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1392 * doc/generic.texi (OpenACC): Adjust URL.
1393 * doc/invoke.texi (C Dialect Options): Ditto.
1395 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1397 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1398 predicate for operand 1. Add (m,<S>) constraint.
1399 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1400 Prevent memory operand 1 with register operand 2.
1402 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1404 PR rtl-optimization/82024
1405 * combine.c (try_combine): If the combination result is a PARALLEL,
1406 and we only need to retain the SET in there that would be placed
1407 at I2, check that we can place that at I3 instead, before doing so.
1409 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1412 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1414 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1415 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1418 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1419 Jeff Law <law@redhat.com>
1421 * varasm.c (bss_initializer_p): Do not put constants into .bss
1422 (categorize_decl_for_section): Handle bss_initializer_p returning
1423 false when DECL_INITIAL is NULL.
1425 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1428 * config/s390/s390.c (s390_can_inline_p): New function.
1430 2017-09-01 Jeff Law <law@redhat.com>
1432 PR tree-optimization/82052
1433 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1434 Always initialize the returned slot after a hash table miss
1435 when INSERT is true.
1437 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1439 * config/s390/s390.md (mem_signal_fence): Remove.
1440 * doc/md.texi (mem_signal_fence): Remove.
1441 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1443 * target-insns.def (mem_signal_fence): Remove.
1445 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1448 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1451 * asan.c (create_odr_indicator): Strip name encoding from assembler
1452 name before appending it after __odr_asan_.
1454 2017-09-01 Martin Liska <mliska@suse.cz>
1456 PR tree-optimization/82059
1457 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1458 frequency only when an edge is redirected.
1460 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1462 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1463 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1464 (arc_conditional_register_usage): Remove ARC600 lp_count
1466 (arc_file_start): Emit Tag_ARC_CPU_variation.
1467 (arc_can_use_doloop_p): New conditions to use ZOLs.
1468 (hwloop_fail): New function.
1469 (hwloop_optimize): Likewise.
1470 (hwloop_pattern_reg): Likewise.
1471 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1472 (arc_reorg_loops): New function, calls reorg_loops.
1473 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1474 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1476 (arc_loop_hazard): Remove function, functionality moved into
1478 (arc_hazard): Remove arc_loop_hazard call.
1479 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1480 into hwloop_optimize.
1481 (arc_label_align): Remove ZOL handling.
1482 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1483 * config/arc/arc.md (doloop_begin): Remove pattern.
1484 (doloop_begin_i): Likewise.
1485 (doloop_end_i): Likewise.
1486 (doloop_fallback): Likewise.
1487 (doloop_fallback_m): Likewise.
1488 (doloop_end): Reimplement expand.
1489 (arc_lp): New pattern for LP instruction.
1490 (loop_end): New pattern.
1491 (loop_fail): Likewise.
1492 (decrement_and_branch_until_zero): Likewise.
1493 * config/arc/arc.opt (mlpc-width): New option.
1494 * doc/invoke.texi (mlpc-width): Document option.
1496 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1498 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1499 (arc_ccfsm_advance): Fix checking for delay slots.
1500 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1502 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1504 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1506 (movhi_insn): Update store instruction constraint which are saving
1507 6-bit short immediates.
1508 (movsi_insn): Consider also short scaled load operations.
1509 (zero_extendhisi2_i): Use Usd constraint instead of T.
1510 (extendhisi2_i): Add q constraint.
1511 (arc_clzsi2): Add type and length attributes.
1512 (arc_ctzsi2): Likewise.
1513 * config/arc/constraints.md (Usc): Update constraint, the
1514 assembler can parse two relocations for a single instruction.
1516 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1518 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1519 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1521 2017-08-31 Olivier Hainque <hainque@adacore.com>
1523 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1524 match as powerpc-wrs-vxworks*.
1526 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1528 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1529 register constraint for by-element operand.
1530 (aarch64_mls_elt_merge<mode>): Likewise.
1532 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1534 * config/arc/arc.c (arc_can_follow_jump): Check for short
1537 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1539 * config.gcc: Use g.opt for arc.
1540 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1541 functionality moved to ...
1542 (legitimate_scaled_address_p): New function, ...here.
1543 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1544 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1545 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1547 (arc_override_options): Handle G option.
1548 (arc_output_pic_addr_const): Correct function definition.
1549 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1550 (arc_decl_anon_ns_mem_p): Delete.
1551 (arc_in_small_data_p): Overhaul this function to take into
1552 consideration the value given via G option.
1553 (arc_rewrite_small_data_1): Renamed and corrected old
1554 arc_rewrite_small_data function.
1555 (arc_rewrite_small_data): New function.
1556 (small_data_pattern): Don't use pic_offset_table_rtx.
1557 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1558 * config/arc/simdext.md (movmisalignv2hi): Use
1559 prepare_move_operands function.
1561 (movmisalign*): Likewise.
1562 * doc/invoke.texi (ARC options): Document -G option.
1564 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1566 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1568 * config/arc/arc.c (arc_print_operand): Output scalled address for
1569 sdata whenever is possible.
1570 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1571 load/stores are available.
1572 (compact_sda_memory_operand): Check for the alignment required by
1573 code density instructions.
1574 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1576 * config/arc/constraints.md (Usd): Update constraint.
1577 (Us0): New constraint.
1578 (Usc): Update constraint.
1580 2017-08-31 Richard Biener <rguenther@suse.de>
1583 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1586 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1588 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1589 Resize type_signature.
1591 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1592 Alan Hayward <alan.hayward@arm.com>
1593 David Sherwood <david.sherwood@arm.com>
1595 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1596 subregs whose inner modes can be stored in GPRs.
1597 (aarch64_classify_index): Likewise.
1599 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1600 Alan Hayward <alan.hayward@arm.com>
1601 David Sherwood <david.sherwood@arm.com>
1603 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1604 (V_INT_EQUIV): ...this.
1605 (v_cmp_result): Rename to...
1606 (v_int_equiv): ...this.
1607 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1608 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1609 (copysign<mode>3): Likewise.
1610 (aarch64_simd_bsl<mode>_internal): Likewise.
1611 (aarch64_simd_bsl<mode>): Likewise.
1612 (vec_cmp<mode><mode>): Likewise.
1613 (vcond<mode><mode>): Likewise.
1614 (vcond<v_cmp_mixed><mode>): Likewise.
1615 (vcondu<mode><v_cmp_mixed>): Likewise.
1616 (aarch64_cm<optab><mode>): Likewise.
1617 (aarch64_cmtst<mode>): Likewise.
1618 (aarch64_fac<optab><mode>): Likewise.
1619 (vec_perm_const<mode>): Likewise.
1620 (vcond_mask_<mode><v_cmp_result>): Rename to...
1621 (vcond_mask_<mode><v_int_equiv>): ...this.
1622 (vec_cmp<mode><v_cmp_result>): Rename to...
1623 (vec_cmp<mode><v_int_equiv>): ...this.
1625 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1626 Alan Hayward <alan.hayward@arm.com>
1627 David Sherwood <david.sherwood@arm.com>
1629 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1631 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1632 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1633 (UNSPEC_LD4_DREG): New unspecs.
1634 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1635 (aarch64_ld2<mode>_dreg_be): Replace with...
1636 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1638 (aarch64_ld3<mode>_dreg_le)
1639 (aarch64_ld3<mode>_dreg_be): Replace with...
1640 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1642 (aarch64_ld4<mode>_dreg_le)
1643 (aarch64_ld4<mode>_dreg_be): Replace with...
1644 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1647 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1649 PR tree-optimization/81987
1650 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1651 insert an initializer in a location not dominated by the stride
1654 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1656 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1657 on the entire header of the finally block in the fallthru case.
1659 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1661 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1663 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1665 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1666 rs6000_emit_move_from_cr and call renamed function.
1667 (rs6000_emit_prologue): Call renamed functions.
1668 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1669 movesi_from_cr, remove volatile CRs.
1671 2017-08-30 Jon Beniston <jon@beniston.com>
1672 Richard Biener <rguenther@suse.de>
1674 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1675 of VECTOR_MODE_P check.
1676 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1677 element vector types.
1679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1681 * df.h (df_read_modify_subreg_p): Remove in favor of...
1682 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1683 const_rtx instead of an rtx.
1684 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1685 * df-problems.c (df_word_lr_mark_ref): Likewise.
1686 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1687 (mark_pseudo_reg_dead): Likewise.
1688 (mark_ref_dead): Likewise.
1689 * reginfo.c (init_subregs_of_mode): Likewise.
1690 * sched-deps.c (sched_analyze_1): Likewise.
1691 * df-scan.c (df_def_record_1): Likewise.
1692 (df_uses_record): Likewise.
1693 (df_read_modify_subreg_p): Remove in favor of...
1694 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1695 const_rtx instead of an rtx.
1697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1698 Alan Hayward <alan.hayward@arm.com>
1699 David Sherwood <david.sherwood@arm.com>
1701 * rtl.h (partial_subreg_p): New function.
1702 * caller-save.c (save_call_clobbered_regs): Use it.
1703 * calls.c (expand_call): Likewise.
1704 * combine.c (combinable_i3pat): Likewise.
1705 (simplify_set): Likewise.
1706 (make_extraction): Likewise.
1707 (make_compound_operation_int): Likewise.
1708 (gen_lowpart_or_truncate): Likewise.
1709 (force_to_mode): Likewise.
1710 (make_field_assignment): Likewise.
1711 (reg_truncated_to_mode): Likewise.
1712 (record_truncated_value): Likewise.
1713 (move_deaths): Likewise.
1714 * cse.c (record_jump_cond): Likewise.
1715 (cse_insn): Likewise.
1716 * cselib.c (cselib_lookup_1): Likewise.
1717 * expmed.c (extract_bit_field_using_extv): Likewise.
1718 * function.c (assign_parm_setup_reg): Likewise.
1719 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1720 * ira-build.c (create_insn_allocnos): Likewise.
1721 * lra-coalesce.c (merge_pseudos): Likewise.
1722 * lra-constraints.c (match_reload): Likewise.
1723 (simplify_operand_subreg): Likewise.
1724 (curr_insn_transform): Likewise.
1725 * lra-lives.c (process_bb_lives): Likewise.
1726 * lra.c (new_insn_reg): Likewise.
1727 (lra_substitute_pseudo): Likewise.
1728 * regcprop.c (mode_change_ok): Likewise.
1729 (maybe_mode_change): Likewise.
1730 (copyprop_hardreg_forward_1): Likewise.
1731 * reload.c (push_reload): Likewise.
1732 (find_reloads): Likewise.
1733 (find_reloads_subreg_address): Likewise.
1734 * reload1.c (alter_reg): Likewise.
1735 (eliminate_regs_1): Likewise.
1736 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1738 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1740 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1741 back to if statements, including unpack.
1743 2017-08-30 Martin Liska <mliska@suse.cz>
1746 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1748 (func_checker::compare_asm_inputs_outputs): ... this function.
1749 (func_checker::compare_gimple_asm): Use the function to compare
1750 also ASM constrains.
1751 * ipa-icf-gimple.h: Rename the function.
1753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1754 Alan Hayward <alan.hayward@arm.com>
1755 David Sherwood <david.sherwood@arm.com>
1757 * coretypes.h (complex_mode): New type.
1758 * gdbhooks.py (build_pretty_printer): Handle it.
1759 * machmode.h (complex_mode): New class.
1760 (complex_mode::includes_p): New function.
1761 (is_complex_int_mode): Likewise.
1762 (is_complex_float_mode): Likewise.
1763 * genmodes.c (get_mode_class): Handle complex mode classes.
1764 * function.c (expand_function_end): Use is_complex_int_mode.
1766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1767 Alan Hayward <alan.hayward@arm.com>
1768 David Sherwood <david.sherwood@arm.com>
1770 * coretypes.h (scalar_mode_pod): New typedef.
1771 * gdbhooks.py (build_pretty_printer): Handle it.
1772 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1773 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1774 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1775 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1779 Alan Hayward <alan.hayward@arm.com>
1780 David Sherwood <david.sherwood@arm.com>
1782 * machmode.h (mode_for_vector): Take a scalar_mode instead
1784 * stor-layout.c (mode_for_vector): Likewise.
1785 * explow.c (promote_mode): Use as_a <scalar_mode>.
1786 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
1788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1789 Alan Hayward <alan.hayward@arm.com>
1790 David Sherwood <david.sherwood@arm.com>
1792 * target.def (preferred_simd_mode): Take a scalar_mode
1793 instead of a machine_mode.
1794 * targhooks.h (default_preferred_simd_mode): Likewise.
1795 * targhooks.c (default_preferred_simd_mode): Likewise.
1796 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1797 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1798 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1799 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1800 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1801 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1802 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1803 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1805 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1806 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1807 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1808 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1809 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1810 * doc/tm.texi: Regenerate.
1811 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1815 Alan Hayward <alan.hayward@arm.com>
1816 David Sherwood <david.sherwood@arm.com>
1818 * target.def (scalar_mode_supported_p): Take a scalar_mode
1819 instead of a machine_mode.
1820 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1821 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1822 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1823 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1824 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1825 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1826 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1827 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1828 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1829 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1830 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1831 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1832 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1833 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1835 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1836 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1837 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1838 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1839 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1841 * doc/tm.texi: Regenerate.
1843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1844 Alan Hayward <alan.hayward@arm.com>
1845 David Sherwood <david.sherwood@arm.com>
1847 * coretypes.h (opt_scalar_mode): New typedef.
1848 * gdbhooks.py (build_pretty_printers): Handle it.
1849 * machmode.h (mode_iterator::get_2xwider): Add overload for
1851 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1853 * expr.c (convert_mode_scalar): Likewise.
1854 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1855 * optabs.c (expand_float): Likewise.
1856 (expand_fix): Likewise.
1857 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1860 Alan Hayward <alan.hayward@arm.com>
1861 David Sherwood <david.sherwood@arm.com>
1863 * optabs.c (expand_float): Explicitly check for scalars before
1864 using a branching expansion.
1865 (expand_fix): Likewise.
1867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1868 Alan Hayward <alan.hayward@arm.com>
1869 David Sherwood <david.sherwood@arm.com>
1871 * expr.c (convert_mode): Split scalar handling out into...
1872 (convert_mode_scalar): ...this new function. Treat the modes
1875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1876 Alan Hayward <alan.hayward@arm.com>
1877 David Sherwood <david.sherwood@arm.com>
1879 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1881 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1884 Alan Hayward <alan.hayward@arm.com>
1885 David Sherwood <david.sherwood@arm.com>
1887 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1888 rather than a machine_mode.
1889 (fixed_from_string): Likewise.
1890 (fixed_convert): Likewise.
1891 (fixed_convert_from_int): Likewise.
1892 (fixed_convert_from_real): Likewise.
1893 (real_convert_from_fixed): Likewise.
1894 * fixed-value.c (fixed_from_double_int): Likewise.
1895 (fixed_from_string): Likewise.
1896 (fixed_convert): Likewise.
1897 (fixed_convert_from_int): Likewise.
1898 (fixed_convert_from_real): Likewise.
1899 (real_convert_from_fixed): Likewise.
1900 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1903 Alan Hayward <alan.hayward@arm.com>
1904 David Sherwood <david.sherwood@arm.com>
1906 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1907 of separate mode class checks. Do not allow vector modes here.
1908 (immed_wide_int_const): Use as_a <scalar_mode>.
1909 * explow.c (trunc_int_for_mode): Likewise.
1910 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1911 (wi::shwi): Likewise.
1912 (wi::min_value): Likewise.
1913 (wi::max_value): Likewise.
1914 * dwarf2out.c (loc_descriptor): Likewise.
1915 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1919 Alan Hayward <alan.hayward@arm.com>
1920 David Sherwood <david.sherwood@arm.com>
1922 * tree.h (SCALAR_TYPE_MODE): New macro.
1923 * expr.c (expand_expr_addr_expr_1): Use it.
1924 (expand_expr_real_2): Likewise.
1925 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1926 (fold_convert_const_fixed_from_int): Likewise.
1927 (fold_convert_const_fixed_from_real): Likewise.
1928 (native_encode_fixed): Likewise
1929 (native_encode_complex): Likewise
1930 (native_encode_vector): Likewise.
1931 (native_interpret_fixed): Likewise.
1932 (native_interpret_real): Likewise.
1933 (native_interpret_complex): Likewise.
1934 (native_interpret_vector): Likewise.
1935 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1936 (simd_clone_adjust_argument_types): Likewise.
1937 (simd_clone_init_simd_arrays): Likewise.
1938 (simd_clone_adjust): Likewise.
1939 * stor-layout.c (layout_type): Likewise.
1940 * tree.c (build_minus_one_cst): Likewise.
1941 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1942 * tree-inline.c (estimate_move_cost): Likewise.
1943 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1944 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1945 (vectorizable_reduction): Likewise.
1946 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1947 (vect_recog_mixed_size_cond_pattern): Likewise.
1948 (check_bool_pattern): Likewise.
1949 (adjust_bool_pattern): Likewise.
1950 (search_type_for_mask_1): Likewise.
1951 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1952 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1953 (vectorizable_load): Likewise.
1954 (vectorizable_store): Likewise.
1955 * ubsan.c (ubsan_encode_value): Likewise.
1956 * varasm.c (output_constant): Likewise.
1958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1959 Alan Hayward <alan.hayward@arm.com>
1960 David Sherwood <david.sherwood@arm.com>
1962 * coretypes.h (scalar_mode): New class.
1963 * machmode.h (scalar_mode): Likewise.
1964 (scalar_mode::includes_p): New function.
1965 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1966 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
1967 * genmodes.c (get_mode_class): Handle remaining scalar modes.
1968 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
1969 * expmed.c (store_bit_field_1): Likewise.
1970 (extract_bit_field_1): Likewise.
1971 * expr.c (write_complex_part): Likewise.
1972 (read_complex_part): Likewise.
1973 (emit_move_complex_push): Likewise.
1974 (expand_expr_real_2): Likewise.
1975 * function.c (assign_parm_setup_reg): Likewise.
1976 (assign_parms_unsplit_complex): Likewise.
1977 * optabs.c (expand_binop): Likewise.
1978 * rtlanal.c (subreg_get_info): Likewise.
1979 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1980 * varasm.c (output_constant_pool_2): Likewise.
1982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1983 Alan Hayward <alan.hayward@arm.com>
1984 David Sherwood <david.sherwood@arm.com>
1986 * expmed.c (extract_high_half): Use scalar_int_mode and remove
1988 (expmed_mult_highpart_optab): Likewise.
1989 (expmed_mult_highpart): 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 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
1996 instead of a machine_mode.
1997 (builtin_memset_read_str): Likewise.
1998 * builtins.c (c_readstr): Likewise.
1999 (builtin_memcpy_read_str): Likewise.
2000 (builtin_strncpy_read_str): Likewise.
2001 (builtin_memset_read_str): Likewise.
2002 (builtin_memset_gen_str): Likewise.
2003 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2004 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2005 instead of a machine_mode.
2006 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2008 (make_extraction): Likewise.
2009 (try_widen_shift_mode): Take and return scalar_int_modes instead
2011 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2012 a scalar_int_mode instead of a machine_mode.
2013 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2014 (avr_addr_space_pointer_mode): Likewise.
2015 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2016 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2017 (msp430_unwind_word_mode): Likewise.
2018 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2019 (spu_addr_space_pointer_mode): Likewise.
2020 (spu_addr_space_address_mode): Likewise.
2021 (spu_libgcc_cmp_return_mode): Likewise.
2022 (spu_libgcc_shift_count_mode): Likewise.
2023 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2024 (rl78_addr_space_pointer_mode): Likewise.
2025 (fl78_unwind_word_mode): Likewise.
2026 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2028 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2029 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2030 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2031 (mips_valid_pointer_mode): Likewise.
2032 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2033 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2034 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2036 (ft32_addr_space_address_mode): Likewise.
2037 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2038 scalar_int_mode instead of a machine_mode.
2039 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2041 (m32c_addr_space_address_mode): Likewise.
2042 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2043 (rs6000_eh_return_filter_mode): Likewise.
2044 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2045 (rs6000_eh_return_filter_mode): Likewise.
2046 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2047 (s390_libgcc_shift_count_mode): Likewise.
2048 (s390_unwind_word_mode): Likewise.
2049 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2051 * target.def (mode_rep_extended): Likewise.
2052 (valid_pointer_mode): Likewise.
2053 (addr_space.valid_pointer_mode): Likewise.
2054 (eh_return_filter_mode): Return a scalar_int_mode rather than
2056 (libgcc_cmp_return_mode): Likewise.
2057 (libgcc_shift_count_mode): Likewise.
2058 (unwind_word_mode): Likewise.
2059 (addr_space.pointer_mode): Likewise.
2060 (addr_space.address_mode): Likewise.
2061 * doc/tm.texi: Regenerate.
2062 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2064 (do_jump): Use scalar_int_mode for local variables.
2065 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2066 rather than a machine_mode.
2067 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2068 (scompare_loc_descriptor_wide): Likewise.
2069 (scompare_loc_descriptor_narrow): Likewise.
2070 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2072 * except.c (sjlj_emit_dispatch_table): Likewise.
2073 (expand_builtin_eh_copy_values): Likewise.
2074 * explow.c (convert_memory_address_addr_space_1): Likewise.
2075 Take a scalar_int_mode rather than a machine_mode.
2076 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2077 than a machine_mode.
2078 (memory_address_addr_space): Use scalar_int_mode for local variables.
2079 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2080 rather than a machine_mode.
2081 * expmed.c (mask_rtx): Likewise.
2082 (init_expmed_one_conv): Likewise.
2083 (expand_mult_highpart_adjust): Likewise.
2084 (extract_high_half): Likewise.
2085 (expmed_mult_highpart_optab): Likewise.
2086 (expmed_mult_highpart): Likewise.
2087 (expand_smod_pow2): Likewise.
2088 (expand_sdiv_pow2): Likewise.
2089 (emit_store_flag_int): Likewise.
2090 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2092 (extract_low_bits): Likewise.
2093 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2095 * expr.c (pieces_addr::adjust): Likewise.
2096 (can_store_by_pieces): Likewise.
2097 (store_by_pieces): Likewise.
2098 (clear_by_pieces_1): Likewise.
2099 (expand_expr_addr_expr_1): Likewise.
2100 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2101 (expand_expr_real_1): Likewise.
2102 (try_casesi): Likewise.
2103 * final.c (shorten_branches): Likewise.
2104 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2105 type of "mode" to machine_mode.
2106 * internal-fn.c (expand_arith_overflow_result_store): Take a
2107 scalar_int_mode rather than a machine_mode.
2108 (expand_mul_overflow): Use scalar_int_mode for local variables.
2109 * loop-doloop.c (doloop_modify): Likewise.
2110 (doloop_optimize): Likewise.
2111 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2112 than a machine_mode.
2113 (expand_doubleword_shift_condmove): Likewise.
2114 (expand_doubleword_shift): Likewise.
2115 (expand_doubleword_clz): Likewise.
2116 (expand_doubleword_popcount): Likewise.
2117 (expand_doubleword_parity): Likewise.
2118 (expand_absneg_bit): Use scalar_int_mode for local variables.
2119 (prepare_float_lib_cmp): Likewise.
2120 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2121 rather than a machine_mode.
2122 (convert_memory_address_addr_space): Likewise.
2123 (get_mode_bounds): Likewise.
2124 (get_address_mode): Return a scalar_int_mode rather than a
2126 * rtlanal.c (get_address_mode): Likewise.
2127 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2128 than a machine_mode.
2129 * targhooks.c (default_mode_rep_extended): Likewise.
2130 (default_valid_pointer_mode): Likewise.
2131 (default_addr_space_valid_pointer_mode): Likewise.
2132 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2133 than a machine_mode.
2134 (default_libgcc_cmp_return_mode): Likewise.
2135 (default_libgcc_shift_count_mode): Likewise.
2136 (default_unwind_word_mode): Likewise.
2137 (default_addr_space_pointer_mode): Likewise.
2138 (default_addr_space_address_mode): Likewise.
2139 * targhooks.h (default_eh_return_filter_mode): Likewise.
2140 (default_libgcc_cmp_return_mode): Likewise.
2141 (default_libgcc_shift_count_mode): Likewise.
2142 (default_unwind_word_mode): Likewise.
2143 (default_addr_space_pointer_mode): Likewise.
2144 (default_addr_space_address_mode): Likewise.
2145 (default_mode_rep_extended): Take a scalar_int_mode rather than
2147 (default_valid_pointer_mode): Likewise.
2148 (default_addr_space_valid_pointer_mode): Likewise.
2149 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2151 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2152 rather than a machine_mode.
2153 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2154 for local variables.
2155 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2156 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2157 than a machine_mode.
2159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2160 Alan Hayward <alan.hayward@arm.com>
2161 David Sherwood <david.sherwood@arm.com>
2163 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2164 the mode argument to scalar_int_mode.
2165 (do_jump_by_parts_zero_rtx): Likewise.
2166 (do_jump_by_parts_equality_rtx): Likewise.
2167 (do_jump_by_parts_greater): Take a mode argument.
2168 (do_jump_by_parts_equality): Likewise.
2169 (do_jump_1): Update calls accordingly.
2171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2172 Alan Hayward <alan.hayward@arm.com>
2173 David Sherwood <david.sherwood@arm.com>
2175 * is-a.h (safe_dyn_cast): New function.
2176 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2177 (jump_table_for_label): Likewise.
2178 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2179 instead of an rtx_insn *.
2180 (shorten_branches): Use dyn_cast instead of LABEL_P and
2181 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2182 rtx_jump_table_data::get_data_mode.
2183 (final_scan_insn): Likewise.
2185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2186 Alan Hayward <alan.hayward@arm.com>
2187 David Sherwood <david.sherwood@arm.com>
2189 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2190 trying to combine a full-register integer set with a subreg
2193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2194 Alan Hayward <alan.hayward@arm.com>
2195 David Sherwood <david.sherwood@arm.com>
2197 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2198 that is always either address_mode or pointer_mode.
2200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2201 Alan Hayward <alan.hayward@arm.com>
2202 David Sherwood <david.sherwood@arm.com>
2204 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2205 when the two are known to be equal.
2207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2208 Alan Hayward <alan.hayward@arm.com>
2209 David Sherwood <david.sherwood@arm.com>
2211 * simplify-rtx.c (simplify_const_unary_operation): Use
2212 is_a <scalar_int_mode> instead of checking for a nonzero
2213 precision. Forcibly convert op_mode to a scalar_int_mode
2214 in that case. More clearly differentiate the operand and
2215 result modes and use the former when deciding what the value
2216 of a count-bits operation should be. Use is_int_mode instead
2217 of checking for a MODE_INT. Remove redundant check for whether
2218 this mode has a zero precision.
2220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2221 Alan Hayward <alan.hayward@arm.com>
2222 David Sherwood <david.sherwood@arm.com>
2224 * optabs.c (widen_leading): Change the type of the mode argument
2225 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2226 (widen_bswap): Likewise.
2227 (expand_parity): Likewise.
2228 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2229 (expand_ffs): Likewise.
2230 (epand_unop): Check for scalar integer modes before calling the
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 (const_scalar_mask_from_tree): Add a mode argument.
2239 (expand_expr_real_1): Update call accordingly.
2241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2242 Alan Hayward <alan.hayward@arm.com>
2243 David Sherwood <david.sherwood@arm.com>
2245 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2246 value_mode arguments. Use scalar_int_mode internally.
2247 (store_bit_field_1): Rename the new integer mode from imode
2248 to op0_mode and use it instead of GET_MODE (op0). Update calls
2249 to store_split_bit_field, store_bit_field_using_insv and
2250 store_fixed_bit_field.
2251 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2252 Use scalar_int_mode internally. Use a bit count rather than a mode
2253 when calculating the largest bit size for get_best_mode.
2254 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2255 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2256 Remove assertion that OP0 has a scalar integer mode.
2257 (store_split_bit_field): Add op0_mode and value_mode arguments.
2258 Update calls to extract_fixed_bit_field.
2259 (extract_bit_field_using_extv): Add an op0_mode argument.
2260 Use scalar_int_mode internally.
2261 (extract_bit_field_1): Rename the new integer mode from imode to
2262 op0_mode and use it instead of GET_MODE (op0). Update calls to
2263 extract_split_bit_field, extract_bit_field_using_extv and
2264 extract_fixed_bit_field.
2265 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2266 to extract_split_bit_field and extract_fixed_bit_field_1.
2267 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2268 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2270 (extract_split_bit_field): Add an op0_mode argument. Update call
2271 to extract_fixed_bit_field.
2273 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2274 Alan Hayward <alan.hayward@arm.com>
2275 David Sherwood <david.sherwood@arm.com>
2277 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2278 * explow.c (hard_function_value): Likewise.
2279 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2280 convert_to_mode call outside the loop.
2281 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2282 for the mode iterator. Require the mode specified by max_pieces
2284 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2286 (copy_blkmode_to_reg): Likewise.
2287 (set_storage_via_setmem): Likewise.
2288 * optabs.c (prepare_cmp_insn): Likewise.
2289 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2290 * stor-layout.c (finish_bitfield_representative): Likewise.
2292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2293 Alan Hayward <alan.hayward@arm.com>
2294 David Sherwood <david.sherwood@arm.com>
2296 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2297 * expr.c (convert_move): Use them.
2298 (convert_modes): Likewise.
2299 (store_expr_with_bounds): Likewise.
2301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2302 Alan Hayward <alan.hayward@arm.com>
2303 David Sherwood <david.sherwood@arm.com>
2305 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2306 parameter for the mode of "x". Remove the "known_x", "known_mode"
2307 and "known_ret" arguments. Change the type of the mode argument
2309 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2310 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2311 (reg_num_sign_bit_copies_for_combine): Likewise.
2312 * rtlanal.c (nonzero_bits1): Likewise.
2313 (num_sign_bit_copies1): Likewise.
2314 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2315 (reg_num_sign_bit_copies_general): Likewise.
2316 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2317 (reg_nonzero_bits_general): Likewise.
2319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2320 Alan Hayward <alan.hayward@arm.com>
2321 David Sherwood <david.sherwood@arm.com>
2323 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2324 than in subroutines. Return 1 for non-integer modes.
2325 (cached_num_sign_bit_copies): Change the type of the mode parameter
2327 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2328 classes. Handle CONST_INT_P first and then check whether X also
2329 has a scalar integer mode. Check the same thing for inner registers
2330 of a SUBREG and for values that are being extended or truncated.
2332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2333 Alan Hayward <alan.hayward@arm.com>
2334 David Sherwood <david.sherwood@arm.com>
2336 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2337 in subroutines. Return the mode mask for non-integer modes.
2338 (cached_nonzero_bits): Change the type of the mode parameter
2340 (nonzero_bits1): Likewise. Remove early exit for other mode
2341 classes. Handle CONST_INT_P first and then check whether X
2342 also has a scalar integer mode.
2344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2345 Alan Hayward <alan.hayward@arm.com>
2346 David Sherwood <david.sherwood@arm.com>
2348 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2349 Return a scalar_int_mode and assert that the size is greater than
2351 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2352 (op_by_pieces_d::op_by_pieces_d): Likewise.
2353 (op_by_pieces_d::run): Likewise.
2354 (can_store_by_pieces): Likewise.
2356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2357 Alan Hayward <alan.hayward@arm.com>
2358 David Sherwood <david.sherwood@arm.com>
2360 * combine.c (extract_left_shift): Add a mode argument and update
2362 (make_compound_operation_int): Change the type of the mode parameter
2363 to scalar_int_mode and update the call to extract_left_shift.
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 * combine.c (simplify_and_const_int): Change the type of the mode
2370 parameter to scalar_int_mode.
2371 (simplify_and_const_int_1): Likewise. Update recursive call.
2373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2374 Alan Hayward <alan.hayward@arm.com>
2375 David Sherwood <david.sherwood@arm.com>
2377 * combine.c (simplify_compare_const): Check that the mode is a
2378 scalar_int_mode (rather than VOIDmode) before testing its
2380 (simplify_comparison): Move COMPARISON_P handling out of the
2381 loop and restrict the latter part of the loop to scalar_int_modes.
2382 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2383 and when considering SUBREG_REGs. Use is_int_mode instead of
2384 checking GET_MODE_CLASS against MODE_INT.
2386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2387 Alan Hayward <alan.hayward@arm.com>
2388 David Sherwood <david.sherwood@arm.com>
2390 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2391 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2392 shift_unit_mode and for modes involved in scalar shifts.
2394 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2395 Alan Hayward <alan.hayward@arm.com>
2396 David Sherwood <david.sherwood@arm.com>
2398 * combine.c (force_int_to_mode): New function, split out from...
2399 (force_to_mode): ...here. Keep xmode up-to-date and use it
2400 instead of GET_MODE (x).
2402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2403 Alan Hayward <alan.hayward@arm.com>
2404 David Sherwood <david.sherwood@arm.com>
2406 * optabs-query.h (extraction_insn::struct_mode): Change type to
2407 opt_scalar_int_mode and update comment.
2408 (extraction_insn::field_mode): Change type to scalar_int_mode.
2409 (extraction_insn::pos_mode): Likewise.
2410 * combine.c (make_extraction): Update accordingly.
2411 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2412 (get_optab_extraction_insn): Likewise.
2413 * recog.c (simplify_while_replacing): Likewise.
2414 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2415 parameter to opt_scalar_int_mode.
2417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2418 Alan Hayward <alan.hayward@arm.com>
2419 David Sherwood <david.sherwood@arm.com>
2421 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2422 to a scalar_int_mode instead of a machine_mode.
2423 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2424 (get_best_mode): Return a boolean and use a pointer argument to store
2425 the selected mode. Replace the limit mode parameter with a bit limit.
2426 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2427 for the values returned by bit_field_mode_iterator::next_mode.
2428 (store_bit_field): Update call to get_best_mode.
2429 (store_fixed_bit_field): Likewise.
2430 (extract_fixed_bit_field): Likewise.
2431 * expr.c (optimize_bitfield_assignment_op): Likewise.
2432 * fold-const.c (optimize_bit_field_compare): Likewise.
2433 (fold_truth_andor_1): Likewise.
2434 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2435 Update for new type of m_mode.
2436 (get_best_mode): As above.
2438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2439 Alan Hayward <alan.hayward@arm.com>
2440 David Sherwood <david.sherwood@arm.com>
2442 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2443 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2444 (store_bit_field): Check is_a <scalar_int_mode> before calling
2445 strict_volatile_bitfield_p.
2446 (extract_bit_field): Likewise.
2448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2449 Alan Hayward <alan.hayward@arm.com>
2450 David Sherwood <david.sherwood@arm.com>
2452 * target.def (cstore_mode): Return a scalar_int_mode.
2453 * doc/tm.texi: Regenerate.
2454 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2455 * targhooks.h (default_cstore_mode): Likewise.
2456 * targhooks.c (default_cstore_mode): Likewise, using a forced
2458 * expmed.c (emit_cstore): Expect the target of the cstore to be
2461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2462 Alan Hayward <alan.hayward@arm.com>
2463 David Sherwood <david.sherwood@arm.com>
2465 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2467 (niter_desc): Likewise mode.
2468 (iv_analyze): Add a mode parameter.
2470 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2471 and change its type to scalar_int_mode.
2472 * loop-iv.c: Update commentary at head of file.
2473 (iv_constant): Pass the mode paraeter before the rtx it describes
2474 and change its type to scalar_int_mode. Remove VOIDmode handling.
2475 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2476 (iv_extend): Likewise.
2477 (shorten_into_mode): Likewise.
2478 (iv_add): Use scalar_int_mode.
2479 (iv_mult): Likewise.
2480 (iv_shift): Likewise.
2481 (canonicalize_iv_subregs): Likewise.
2482 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2483 it describes and change its mode to scalar_int_mode. Also change
2484 the type of the returned inner_mode to scalar_int_mode.
2485 (get_biv_step): Likewise, turning outer_mode from a pointer
2486 into a direct parameter. Update call to get_biv_step_1.
2487 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2488 iv_constant and get_biv_step.
2489 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2490 and change its type to scalar_int_mode. Don't initialise iv->mode
2491 to VOIDmode and remove later checks for its still being VOIDmode.
2492 Update calls to iv_analyze_op and iv_analyze_expr. Check
2493 is_a <scalar_int_mode> when changing the mode under consideration.
2494 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2495 Update call to iv_analyze_expr.
2496 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2497 inner register is not also a scalar_int_mode. Update call to
2499 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2500 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2501 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2502 separate mode class checks. Update calls to iv_analyze. Remove
2503 fix-up of VOIDmodes after iv_analyze_biv.
2504 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2505 don't have a scalar_int_mode. Update call to biv_p.
2507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2508 Alan Hayward <alan.hayward@arm.com>
2509 David Sherwood <david.sherwood@arm.com>
2511 * cfgexpand.c (convert_debug_memory_address): Use
2512 as_a <scalar_int_mode>.
2513 * combine.c (expand_compound_operation): Likewise.
2514 (make_extraction): Likewise.
2515 (change_zero_ext): Likewise.
2516 (simplify_comparison): Likewise.
2517 * cse.c (cse_insn): Likewise.
2518 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2519 (mem_loc_descriptor): Likewise.
2520 (loc_descriptor): Likewise.
2521 * expmed.c (init_expmed_one_mode): Likewise.
2522 (synth_mult): Likewise.
2523 (emit_store_flag_1): Likewise.
2524 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2525 of a comparison with size.
2526 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2527 (reduce_to_bit_field_precision): Likewise.
2528 * function.c (expand_function_end): Likewise.
2529 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2530 * loop-doloop.c (doloop_modify): Likewise.
2531 * optabs.c (expand_binop): Likewise.
2532 (expand_unop): Likewise.
2533 (expand_copysign_absneg): Likewise.
2534 (prepare_cmp_insn): Likewise.
2535 (maybe_legitimize_operand): Likewise.
2536 * recog.c (const_scalar_int_operand): Likewise.
2537 * rtlanal.c (get_address_mode): Likewise.
2538 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2539 (simplify_cond_clz_ctz): Likewise.
2540 * tree-nested.c (get_nl_goto_field): Likewise.
2541 * tree.c (build_vector_type_for_mode): Likewise.
2542 * var-tracking.c (use_narrower_mode): Likewise.
2544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2545 Alan Hayward <alan.hayward@arm.com>
2546 David Sherwood <david.sherwood@arm.com>
2548 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2549 * builtins.c (expand_builtin_signbit): Use it.
2550 * cfgexpand.c (expand_debug_expr): Likewise.
2551 * dojump.c (do_jump): Likewise.
2552 (do_compare_and_jump): Likewise.
2553 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2554 * expmed.c (make_tree): Likewise.
2555 * expr.c (expand_expr_real_2): Likewise.
2556 (expand_expr_real_1): Likewise.
2557 (try_casesi): Likewise.
2558 * fold-const-call.c (fold_const_call_ss): Likewise.
2559 * fold-const.c (unextend): Likewise.
2560 (extract_muldiv_1): Likewise.
2561 (fold_single_bit_test): Likewise.
2562 (native_encode_int): Likewise.
2563 (native_encode_string): Likewise.
2564 (native_interpret_int): Likewise.
2565 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2566 * internal-fn.c (expand_addsub_overflow): Likewise.
2567 (expand_neg_overflow): Likewise.
2568 (expand_mul_overflow): Likewise.
2569 (expand_arith_overflow): Likewise.
2570 * match.pd: Likewise.
2571 * stor-layout.c (layout_type): Likewise.
2572 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2573 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2574 * tree-ssanames.c (get_range_info): Likewise.
2575 * tree-switch-conversion.c (array_value_type) Likewise.
2576 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2577 (vect_recog_divmod_pattern): Likewise.
2578 (vect_recog_mixed_size_cond_pattern): Likewise.
2579 * tree-vrp.c (extract_range_basic): Likewise.
2580 (simplify_float_conversion_using_ranges): Likewise.
2581 * tree.c (int_fits_type_p): Likewise.
2582 * ubsan.c (instrument_bool_enum_load): Likewise.
2583 * varasm.c (mergeable_string_section): Likewise.
2584 (narrowing_initializer_constant_valid_p): Likewise.
2585 (output_constant): Likewise.
2587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2588 Alan Hayward <alan.hayward@arm.com>
2589 David Sherwood <david.sherwood@arm.com>
2591 * machmode.h (NARROWEST_INT_MODE): New macro.
2592 * expr.c (alignment_for_piecewise_move): Use it instead of
2593 GET_CLASS_NARROWEST_MODE (MODE_INT).
2594 (push_block): Likewise.
2595 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2597 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2600 Alan Hayward <alan.hayward@arm.com>
2601 David Sherwood <david.sherwood@arm.com>
2603 * postreload.c (move2add_valid_value_p): Change the type of the
2604 mode parameter to scalar_int_mode.
2605 (move2add_use_add2_insn): Add a mode parameter and use it instead
2607 (move2add_use_add3_insn): Likewise.
2608 (reload_cse_move2add): Update accordingly.
2610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2611 Alan Hayward <alan.hayward@arm.com>
2612 David Sherwood <david.sherwood@arm.com>
2614 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2616 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2617 * optabs.c (expand_unop): Likewise.
2619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2620 Alan Hayward <alan.hayward@arm.com>
2621 David Sherwood <david.sherwood@arm.com>
2623 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2624 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2625 (popcount_loc_descriptor): Likewise.
2626 (bswap_loc_descriptor): Likewise.
2627 (rotate_loc_descriptor): Likewise.
2628 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2629 calling the functions above.
2631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2632 Alan Hayward <alan.hayward@arm.com>
2633 David Sherwood <david.sherwood@arm.com>
2635 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2637 (try_combine): Likewise.
2638 (simplify_if_then_else): Likewise.
2639 * cse.c (cse_insn): Likewise.
2640 * dwarf2out.c (mem_loc_descriptor): Likewise.
2641 * emit-rtl.c (gen_lowpart_common): Likewise.
2642 * simplify-rtx.c (simplify_truncation): Likewise.
2643 (simplify_binary_operation_1): Likewise.
2644 (simplify_const_relational_operation): Likewise.
2645 (simplify_ternary_operation): Likewise.
2646 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2649 Alan Hayward <alan.hayward@arm.com>
2650 David Sherwood <david.sherwood@arm.com>
2652 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2653 * reload.c (push_reload): Likewise.
2654 (find_reloads): Likewise.
2656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2657 Alan Hayward <alan.hayward@arm.com>
2658 David Sherwood <david.sherwood@arm.com>
2660 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2661 (make_compound_operation_int): Likewise.
2662 (change_zero_ext): Likewise.
2663 * expr.c (convert_move): Likewise.
2664 (convert_modes): Likewise.
2665 * fwprop.c (forward_propagate_subreg): Likewise.
2666 * loop-iv.c (get_biv_step_1): Likewise.
2667 * optabs.c (widen_operand): Likewise.
2668 * postreload.c (move2add_valid_value_p): Likewise.
2669 * recog.c (simplify_while_replacing): Likewise.
2670 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2671 (simplify_binary_operation_1): Likewise. Remove redundant
2672 mode equality check.
2674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2675 Alan Hayward <alan.hayward@arm.com>
2676 David Sherwood <david.sherwood@arm.com>
2678 * combine.c (combine_simplify_rtx): Add checks for
2679 is_a <scalar_int_mode>.
2680 (simplify_if_then_else): Likewise.
2681 (make_field_assignment): Likewise.
2682 (simplify_comparison): Likewise.
2683 * ifcvt.c (noce_try_bitop): Likewise.
2684 * loop-invariant.c (canonicalize_address_mult): Likewise.
2685 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2688 Alan Hayward <alan.hayward@arm.com>
2689 David Sherwood <david.sherwood@arm.com>
2691 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2692 is_a <scalar_int_mode> instead of != BLKmode.
2694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2695 Alan Hayward <alan.hayward@arm.com>
2696 David Sherwood <david.sherwood@arm.com>
2698 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2699 instead of != VOIDmode.
2700 * combine.c (if_then_else_cond): Likewise.
2701 (change_zero_ext): Likewise.
2702 * dwarf2out.c (mem_loc_descriptor): Likewise.
2703 (loc_descriptor): Likewise.
2704 * rtlanal.c (canonicalize_condition): Likewise.
2705 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
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 * simplify-rtx.c (simplify_binary_operation_1): Use
2712 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2714 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2715 Alan Hayward <alan.hayward@arm.com>
2716 David Sherwood <david.sherwood@arm.com>
2718 * wide-int.h (int_traits<unsigned char>) New class.
2719 (int_traits<unsigned short>) Likewise.
2720 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2721 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2723 * combine.c (set_nonzero_bits_and_sign_copies): Use
2724 is_a <scalar_int_mode>.
2725 (find_split_point): Likewise.
2726 (combine_simplify_rtx): Likewise.
2727 (simplify_logical): Likewise.
2728 (expand_compound_operation): Likewise.
2729 (expand_field_assignment): Likewise.
2730 (make_compound_operation): Likewise.
2731 (extended_count): Likewise.
2732 (change_zero_ext): Likewise.
2733 (simplify_comparison): Likewise.
2734 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2735 (ucompare_loc_descriptor): Likewise.
2736 (minmax_loc_descriptor): Likewise.
2737 (mem_loc_descriptor): Likewise.
2738 (loc_descriptor): Likewise.
2739 * expmed.c (init_expmed_one_mode): Likewise.
2740 * lra-constraints.c (lra_constraint_offset): Likewise.
2741 * optabs.c (prepare_libcall_arg): Likewise.
2742 * postreload.c (move2add_note_store): Likewise.
2743 * reload.c (operands_match_p): Likewise.
2744 * rtl.h (load_extend_op): Likewise.
2745 * rtlhooks.c (gen_lowpart_general): Likewise.
2746 * simplify-rtx.c (simplify_truncation): Likewise.
2747 (simplify_unary_operation_1): Likewise.
2748 (simplify_binary_operation_1): Likewise.
2749 (simplify_const_binary_operation): Likewise.
2750 (simplify_const_relational_operation): Likewise.
2751 (simplify_subreg): Likewise.
2752 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2753 * var-tracking.c (adjust_mems): Likewise.
2754 (prepare_call_arguments): Likewise.
2756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2757 Alan Hayward <alan.hayward@arm.com>
2758 David Sherwood <david.sherwood@arm.com>
2760 * machmode.h (is_int_mode): New fuction.
2761 * combine.c (find_split_point): Use it.
2762 (combine_simplify_rtx): Likewise.
2763 (simplify_if_then_else): Likewise.
2764 (simplify_set): Likewise.
2765 (simplify_shift_const_1): Likewise.
2766 (simplify_comparison): Likewise.
2767 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2768 * cse.c (notreg_cost): Likewise.
2769 (cse_insn): Likewise.
2770 * cselib.c (cselib_lookup_1): Likewise.
2771 * dojump.c (do_jump_1): Likewise.
2772 (do_compare_rtx_and_jump): Likewise.
2773 * dse.c (get_call_args): Likewise.
2774 * dwarf2out.c (rtl_for_decl_init): Likewise.
2775 (native_encode_initializer): Likewise.
2776 * expmed.c (emit_store_flag_1): Likewise.
2777 (emit_store_flag): Likewise.
2778 * expr.c (convert_modes): Likewise.
2779 (store_field): Likewise.
2780 (expand_expr_real_1): Likewise.
2781 * fold-const.c (fold_read_from_constant_string): Likewise.
2782 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2783 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2784 * optabs.c (expand_binop): Likewise.
2785 (expand_unop): Likewise.
2786 (expand_abs_nojump): Likewise.
2787 (expand_one_cmpl_abs_nojump): Likewise.
2788 * simplify-rtx.c (mode_signbit_p): Likewise.
2789 (val_signbit_p): Likewise.
2790 (val_signbit_known_set_p): Likewise.
2791 (val_signbit_known_clear_p): Likewise.
2792 (simplify_relational_operation_1): Likewise.
2793 * tree.c (vector_type_mode): Likewise.
2795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2796 Alan Hayward <alan.hayward@arm.com>
2797 David Sherwood <david.sherwood@arm.com>
2799 * machmode.h (smallest_mode_for_size): Fix formatting.
2800 (smallest_int_mode_for_size): New function.
2801 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2802 instead of smallest_mode_for_size.
2803 * combine.c (make_extraction): Likewise.
2804 * config/arc/arc.c (arc_expand_movmem): Likewise.
2805 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2806 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2807 * config/s390/s390.c (s390_expand_insv): Likewise.
2808 * config/sparc/sparc.c (assign_int_registers): Likewise.
2809 * config/spu/spu.c (spu_function_value): Likewise.
2810 (spu_function_arg): Likewise.
2811 * coverage.c (get_gcov_type): Likewise.
2812 (get_gcov_unsigned_t): Likewise.
2813 * dse.c (find_shift_sequence): Likewise.
2814 * expmed.c (store_bit_field_1): Likewise.
2815 * expr.c (convert_move): Likewise.
2816 (store_field): Likewise.
2817 * internal-fn.c (expand_arith_overflow): Likewise.
2818 * optabs-query.c (get_best_extraction_insn): Likewise.
2819 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2820 * stor-layout.c (layout_type): Likewise.
2821 (initialize_sizetypes): Likewise.
2822 * targhooks.c (default_get_mask_mode): Likewise.
2823 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2826 Alan Hayward <alan.hayward@arm.com>
2827 David Sherwood <david.sherwood@arm.com>
2829 * machmode.h (opt_mode::else_blk): New function.
2830 (int_mode_for_mode): Declare.
2831 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2832 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2834 * cfgexpand.c (expand_debug_expr): Likewise.
2835 * combine.c (gen_lowpart_or_truncate): Likewise.
2836 (gen_lowpart_for_combine): Likewise.
2837 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2838 * config/avr/avr.c (avr_to_int_mode): Likewise.
2839 (avr_out_plus_1): Likewise.
2840 (avr_out_plus): Likewise.
2841 (avr_out_round): Likewise.
2842 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2843 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2844 (s390_expand_vcond): Likewise.
2845 * config/spu/spu.c (spu_split_immediate): Likewise.
2846 (spu_expand_mov): Likewise.
2847 * dse.c (get_stored_val): Likewise.
2848 * expmed.c (store_bit_field_1): Likewise.
2849 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2851 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2852 (extract_low_bits): Likewise.
2853 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2854 handling rather than repeating the check.
2855 (emit_group_store): Likewise.
2856 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2857 * optabs.c (expand_absneg_bit): Likewise.
2858 (expand_copysign_absneg): Likewise.
2859 (expand_copysign_bit): Likewise.
2860 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2861 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2862 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2863 * var-tracking.c (prepare_call_arguments): Likewise.
2864 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2865 int_mode_for_mode instead of mode_for_size.
2866 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2869 Alan Hayward <alan.hayward@arm.com>
2870 David Sherwood <david.sherwood@arm.com>
2872 * machmode.h (int_mode_for_size): New function.
2873 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2874 instead of mode_for_size.
2875 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2877 * combine.c (expand_field_assignment): Use int_mode_for_size
2878 instead of mode_for_size.
2879 (make_extraction): Likewise.
2880 (simplify_shift_const_1): Likewise.
2881 (simplify_comparison): Likewise.
2882 * dojump.c (do_jump): Likewise.
2883 * dwarf2out.c (mem_loc_descriptor): Likewise.
2884 * emit-rtl.c (init_derived_machine_modes): Likewise.
2885 * expmed.c (flip_storage_order): Likewise.
2886 (convert_extracted_bit_field): Likewise.
2887 * expr.c (copy_blkmode_from_reg): Likewise.
2888 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2889 * internal-fn.c (expand_mul_overflow): Likewise.
2890 * lower-subreg.c (simple_move): Likewise.
2891 * optabs-libfuncs.c (init_optabs): Likewise.
2892 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2893 * tree.c (vector_type_mode): Likewise.
2894 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2895 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2896 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2897 * tree-vect-stmts.c (vectorizable_load): Likewise.
2898 (vectorizable_store): Likewise.
2900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2901 Alan Hayward <alan.hayward@arm.com>
2902 David Sherwood <david.sherwood@arm.com>
2904 * coretypes.h (pod_mode): New type.
2905 (scalar_int_mode_pod): New typedef.
2906 * machmode.h (pod_mode): New class.
2907 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2908 * genmodes.c (emit_mode_int_n): Update accordingly.
2909 * lower-subreg.h (target_lower_subreg): Change type to
2910 scalar_int_mode_pod.
2911 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2912 scalar_int_mode_pod.
2914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2915 Alan Hayward <alan.hayward@arm.com>
2916 David Sherwood <david.sherwood@arm.com>
2918 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2919 machine_mode to scalar_int_mode.
2920 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2921 (rs6000_option_override_internal): Remove cast to int.
2922 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2923 machine_mode to scalar_int_mode.
2924 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2925 (rs6000_option_override_internal): Remove cast to int.
2926 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2927 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2929 * config/s390/s390.c (s390_expand_builtin): Likewise.
2930 * coretypes.h (scalar_int_mode): New type.
2931 (opt_scalar_int_mode): New typedef.
2932 * machmode.h (scalar_int_mode): New class.
2933 (scalar_int_mode::includes_p): New function.
2934 (byte_mode): Change type to scalar_int_mode.
2935 (word_mode): Likewise.
2936 (ptr_mode): Likewise.
2937 * emit-rtl.c (byte_mode): Likewise.
2938 (word_mode): Likewise.
2939 (ptr_mode): Likewise.
2940 (init_derived_machine_modes): Update accordingly.
2941 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2942 and MODE_PARTIAL_INT.
2943 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2944 opt_scalar_int_mode.
2946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2947 Alan Hayward <alan.hayward@arm.com>
2948 David Sherwood <david.sherwood@arm.com>
2950 * target.def (libgcc_floating_mode_supported_p): Take a
2952 * doc/tm.texi: Regenerate.
2953 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2955 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2956 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
2959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2960 Alan Hayward <alan.hayward@arm.com>
2961 David Sherwood <david.sherwood@arm.com>
2963 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2964 * doc/tm.texi: Regenerate.
2965 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2966 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
2967 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
2968 * targhooks.h (default_floatn_mode): Likewise.
2969 * targhooks.c (default_floatn_mode): Likewise.
2970 * tree.c (build_common_tree_nodes): Update accordingly.
2972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2973 Alan Hayward <alan.hayward@arm.com>
2974 David Sherwood <david.sherwood@arm.com>
2976 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
2977 (mode_iterator::iterate_p): Likewise.
2978 (mode_iterator::get_wider): Likewise.
2979 * expr.c (init_expr_target): Use opt_scalar_float_mode.
2981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2982 Alan Hayward <alan.hayward@arm.com>
2983 David Sherwood <david.sherwood@arm.com>
2985 * coretypes.h (opt_scalar_float_mode): New typedef.
2986 * machmode.h (float_mode_for_size): New function.
2987 * emit-rtl.c (double_mode): Delete.
2988 (init_emit_once): Use float_mode_for_size.
2989 * stor-layout.c (layout_type): Likewise.
2990 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
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 * output.h (assemble_real): Take a scalar_float_mode.
2997 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
2998 * config/arm/arm.md (consttable_4): Likewise.
2999 (consttable_8): Likewise.
3000 (consttable_16): Likewise.
3001 * config/mips/mips.md (consttable_float): Likewise.
3002 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3003 * varasm.c (assemble_real): Take a scalar_float_mode.
3004 (output_constant_pool_2): Update accordingly.
3005 (output_constant): Likewise.
3007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3008 Alan Hayward <alan.hayward@arm.com>
3009 David Sherwood <david.sherwood@arm.com>
3011 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3012 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3013 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3014 (native_encode_real): Likewise.
3015 (native_interpret_real): Likewise.
3016 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3017 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3020 Alan Hayward <alan.hayward@arm.com>
3021 David Sherwood <david.sherwood@arm.com>
3023 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3024 <scalar_float_mode>. Simplify.
3025 (gen_extend_conv_libfunc): Likewise.
3027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3028 Alan Hayward <alan.hayward@arm.com>
3029 David Sherwood <david.sherwood@arm.com>
3031 * coretypes.h (scalar_float_mode): New type.
3032 * machmode.h (mode_traits::from_int): Use machine_mode if
3033 USE_ENUM_MODES is defined.
3034 (is_a): New function.
3036 (dyn_cast): Likewise.
3037 (scalar_float_mode): New class.
3038 (scalar_float_mode::includes_p): New function.
3039 (is_float_mode): Likewise.
3040 * gdbhooks.py (MachineModePrinter): New class.
3041 (build_pretty_printer): Use it for scalar_float_mode.
3042 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3043 (format_helper::format_helper): Turn into a template.
3044 * genmodes.c (get_mode_class): New function.
3045 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3046 or machine_mode if none.
3047 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3048 as_a <scalar_float_mode>.
3049 * dwarf2out.c (mem_loc_descriptor): Likewise.
3050 (insert_float): Likewise.
3051 (add_const_value_attribute): Likewise.
3052 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3053 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3054 (expand_unop): Update accordingly.
3055 (expand_abs_nojump): Likewise.
3056 (expand_copysign_absneg): Take a scalar_float_mode.
3057 (expand_copysign_bit): Likewise.
3058 (expand_copysign): Update accordingly.
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 (opt_mode): New class.
3065 * machmode.h (opt_mode): Likewise.
3066 (opt_mode::else_void): New function.
3067 (opt_mode::require): Likewise.
3068 (opt_mode::exists): Likewise.
3069 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3070 (GET_MODE_2XWIDER_MODE): Likewise.
3071 (mode_iterator::get_wider): Update accordingly.
3072 (mode_iterator::get_2xwider): Likewise.
3073 (mode_iterator::get_known_wider): Likewise, turning into a template.
3074 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3075 forcing a wider mode to exist.
3076 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3077 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3078 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3079 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3080 * lower-subreg.c (init_lower_subreg): Likewise.
3081 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3082 on the final iteration.
3083 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3084 a wider mode exists before asking for a move pattern.
3085 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3086 forcing a wider mode to exist.
3087 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3088 returning false if no such mode exists.
3089 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3090 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3091 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3092 Avoid checking for a MODE_INT if we already know the mode is not a
3094 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3095 forcing a wider mode to exist.
3096 (expmed_mult_highpart_optab): Likewise.
3097 (expmed_mult_highpart): Likewise.
3098 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3100 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3101 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3102 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3103 * internal-fn.c (expand_mul_overflow): Update use of
3104 GET_MODE_2XWIDER_MODE.
3105 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3106 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3107 GET_MODE_WIDER_MODE.
3108 (convert_plusminus_to_widen): Likewise.
3109 * tree-switch-conversion.c (array_value_type): Likewise.
3110 * var-tracking.c (emit_note_insn_var_location): Likewise.
3111 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3112 Return false inside rather than outside the loop if no wider mode
3114 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3115 and GET_MODE_2XWIDER_MODE
3116 (can_compare_p): Use else_void.
3117 * gdbhooks.py (OptMachineModePrinter): New class.
3118 (build_pretty_printer): Use it for opt_mode.
3120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3121 Alan Hayward <alan.hayward@arm.com>
3122 David Sherwood <david.sherwood@arm.com>
3124 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3125 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3128 Alan Hayward <alan.hayward@arm.com>
3129 David Sherwood <david.sherwood@arm.com>
3131 * machmode.h (mode_traits): New structure.
3132 (get_narrowest_mode): New function.
3133 (mode_iterator::start): Likewise.
3134 (mode_iterator::iterate_p): Likewise.
3135 (mode_iterator::get_wider): Likewise.
3136 (mode_iterator::get_known_wider): Likewise.
3137 (mode_iterator::get_2xwider): Likewise.
3138 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3139 (FOR_EACH_MODE): Likewise.
3140 (FOR_EACH_MODE_FROM): Likewise.
3141 (FOR_EACH_MODE_UNTIL): Likewise.
3142 (FOR_EACH_WIDER_MODE): Likewise.
3143 (FOR_EACH_2XWIDER_MODE): Likewise.
3144 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3145 * combine.c (simplify_comparison): Likewise
3146 * config/i386/i386.c (type_natural_mode): Likewise.
3147 * cse.c (cse_insn): Likewise.
3148 * dse.c (find_shift_sequence): Likewise.
3149 * emit-rtl.c (init_derived_machine_modes): Likewise.
3150 (init_emit_once): Likewise.
3151 * explow.c (hard_function_value): Likewise.
3152 * expmed.c (extract_fixed_bit_field_1): Likewise.
3153 (extract_bit_field_1): Likewise.
3154 (expand_divmod): Likewise.
3155 (emit_store_flag_1): Likewise.
3156 * expr.c (init_expr_target): Likewise.
3157 (convert_move): Likewise.
3158 (alignment_for_piecewise_move): Likewise.
3159 (widest_int_mode_for_size): Likewise.
3160 (emit_block_move_via_movmem): Likewise.
3161 (copy_blkmode_to_reg): Likewise.
3162 (set_storage_via_setmem): Likewise.
3163 (compress_float_constant): Likewise.
3164 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3165 * optabs-query.c (get_best_extraction_insn): Likewise.
3166 * optabs.c (expand_binop): Likewise.
3167 (expand_twoval_unop): Likewise.
3168 (expand_twoval_binop): Likewise.
3169 (widen_leading): Likewise.
3170 (widen_bswap): Likewise.
3171 (expand_parity): Likewise.
3172 (expand_unop): Likewise.
3173 (prepare_cmp_insn): Likewise.
3174 (prepare_float_lib_cmp): Likewise.
3175 (expand_float): Likewise.
3176 (expand_fix): Likewise.
3177 (expand_sfix_optab): Likewise.
3178 * postreload.c (move2add_use_add2_insn): Likewise.
3179 * reg-stack.c (reg_to_stack): Likewise.
3180 * reginfo.c (choose_hard_reg_mode): Likewise.
3181 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3182 * stor-layout.c (mode_for_size): Likewise.
3183 (smallest_mode_for_size): Likewise.
3184 (mode_for_vector): Likewise.
3185 (finish_bitfield_representative): Likewise.
3186 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3187 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3188 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3189 * var-tracking.c (prepare_call_arguments): Likewise.
3191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3192 Alan Hayward <alan.hayward@arm.com>
3193 David Sherwood <david.sherwood@arm.com>
3195 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3196 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3197 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3198 * machmode.h (mode_size): Move earlier in file.
3199 (mode_precision): Likewise.
3200 (mode_inner): Likewise.
3201 (mode_nunits): Likewise.
3202 (mode_unit_size): Likewise.
3203 (unit_unit_precision): Likewise.
3204 (mode_wider): Likewise.
3205 (mode_2xwider): Likewise.
3206 (machine_mode): New class.
3207 (mode_to_bytes): New function.
3208 (mode_to_bits): Likewise.
3209 (mode_to_precision): Likewise.
3210 (mode_to_inner): Likewise.
3211 (mode_to_unit_size): Likewise.
3212 (mode_to_unit_precision): Likewise.
3213 (mode_to_nunits): Likewise.
3214 (GET_MODE_SIZE): Use mode_to_bytes.
3215 (GET_MODE_BITSIZE): Use mode_to_bits.
3216 (GET_MODE_PRECISION): Use mode_to_precision.
3217 (GET_MODE_INNER): Use mode_to_inner.
3218 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3219 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3220 (GET_MODE_NUNITS): Use mode_to_nunits.
3221 * system.h (ALWAYS_INLINE): New macro.
3222 * config/powerpcspe/powerpcspe-c.c
3223 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3224 int for arg1_mode and arg2_mode.
3226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3227 Alan Hayward <alan.hayward@arm.com>
3228 David Sherwood <david.sherwood@arm.com>
3230 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3231 Prefix mode names with E_ in case statements.
3232 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3233 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3234 (aarch64_split_simd_move): Likewise.
3235 (aarch64_gen_storewb_pair): Likewise.
3236 (aarch64_gen_loadwb_pair): Likewise.
3237 (aarch64_gen_store_pair): Likewise.
3238 (aarch64_gen_load_pair): Likewise.
3239 (aarch64_get_condition_code_1): Likewise.
3240 (aarch64_constant_pool_reload_icode): Likewise.
3241 (get_rsqrte_type): Likewise.
3242 (get_rsqrts_type): Likewise.
3243 (get_recpe_type): Likewise.
3244 (get_recps_type): Likewise.
3245 (aarch64_gimplify_va_arg_expr): Likewise.
3246 (aarch64_simd_container_mode): Likewise.
3247 (aarch64_emit_load_exclusive): Likewise.
3248 (aarch64_emit_store_exclusive): Likewise.
3249 (aarch64_expand_compare_and_swap): Likewise.
3250 (aarch64_gen_atomic_cas): Likewise.
3251 (aarch64_emit_bic): Likewise.
3252 (aarch64_emit_atomic_swap): Likewise.
3253 (aarch64_emit_atomic_load_op): Likewise.
3254 (aarch64_evpc_trn): Likewise.
3255 (aarch64_evpc_uzp): Likewise.
3256 (aarch64_evpc_zip): Likewise.
3257 (aarch64_evpc_ext): Likewise.
3258 (aarch64_evpc_rev): Likewise.
3259 (aarch64_evpc_dup): Likewise.
3260 (aarch64_gen_ccmp_first): Likewise.
3261 (aarch64_gen_ccmp_next): Likewise.
3262 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3263 (alpha_emit_xfloating_libcall): Likewise.
3264 (emit_insxl): Likewise.
3265 (alpha_arg_type): Likewise.
3266 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3267 (arc_preferred_simd_mode): Likewise.
3268 (arc_secondary_reload): Likewise.
3269 (get_arc_condition_code): Likewise.
3270 (arc_print_operand): Likewise.
3271 (arc_legitimate_constant_p): Likewise.
3272 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3273 * config/arc/arc.md (casesi_load): Likewise.
3274 (casesi_compact_jump): Likewise.
3275 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3276 (cc_use_register): Likewise.
3277 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3278 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3279 (arm_init_iwmmxt_builtins): Likewise.
3280 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3281 (neon_expand_vector_init): Likewise.
3282 (arm_attr_length_move_neon): Likewise.
3283 (maybe_get_arm_condition_code): Likewise.
3284 (arm_emit_vector_const): Likewise.
3285 (arm_preferred_simd_mode): Likewise.
3286 (arm_output_iwmmxt_tinsr): Likewise.
3287 (thumb1_output_casesi): Likewise.
3288 (thumb2_output_casesi): Likewise.
3289 (arm_emit_load_exclusive): Likewise.
3290 (arm_emit_store_exclusive): Likewise.
3291 (arm_expand_compare_and_swap): Likewise.
3292 (arm_evpc_neon_vuzp): Likewise.
3293 (arm_evpc_neon_vzip): Likewise.
3294 (arm_evpc_neon_vrev): Likewise.
3295 (arm_evpc_neon_vtrn): Likewise.
3296 (arm_evpc_neon_vext): Likewise.
3297 (arm_validize_comparison): Likewise.
3298 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3299 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3300 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3301 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3302 (c6x_preferred_simd_mode): Likewise.
3303 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3304 (epiphany_rtx_costs): Likewise.
3305 * config/epiphany/predicates.md (proper_comparison_operator):
3307 * config/frv/frv.c (condexec_memory_operand): Likewise.
3308 (frv_emit_move): Likewise.
3309 (output_move_single): Likewise.
3310 (output_condmove_single): Likewise.
3311 (frv_hard_regno_mode_ok): Likewise.
3312 (frv_matching_accg_mode): Likewise.
3313 * config/h8300/h8300.c (split_adds_subs): Likewise.
3314 (h8300_rtx_costs): Likewise.
3315 (h8300_print_operand): Likewise.
3316 (compute_mov_length): Likewise.
3317 (output_logical_op): Likewise.
3318 (compute_logical_op_length): Likewise.
3319 (compute_logical_op_cc): Likewise.
3320 (h8300_shift_needs_scratch_p): Likewise.
3321 (output_a_shift): Likewise.
3322 (compute_a_shift_length): Likewise.
3323 (compute_a_shift_cc): Likewise.
3324 (expand_a_rotate): Likewise.
3325 (output_a_rotate): Likewise.
3326 * config/i386/i386.c (classify_argument): Likewise.
3327 (function_arg_advance_32): Likewise.
3328 (function_arg_32): Likewise.
3329 (function_arg_64): Likewise.
3330 (function_value_64): Likewise.
3331 (ix86_gimplify_va_arg): Likewise.
3332 (ix86_legitimate_constant_p): Likewise.
3333 (put_condition_code): Likewise.
3334 (split_double_mode): Likewise.
3335 (ix86_avx256_split_vector_move_misalign): Likewise.
3336 (ix86_expand_vector_logical_operator): Likewise.
3337 (ix86_split_idivmod): Likewise.
3338 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3339 (ix86_build_const_vector): Likewise.
3340 (ix86_build_signbit_mask): Likewise.
3341 (ix86_match_ccmode): Likewise.
3342 (ix86_cc_modes_compatible): Likewise.
3343 (ix86_expand_branch): Likewise.
3344 (ix86_expand_sse_cmp): Likewise.
3345 (ix86_expand_sse_movcc): Likewise.
3346 (ix86_expand_int_sse_cmp): Likewise.
3347 (ix86_expand_vec_perm_vpermi2): Likewise.
3348 (ix86_expand_vec_perm): Likewise.
3349 (ix86_expand_sse_unpack): Likewise.
3350 (ix86_expand_int_addcc): Likewise.
3351 (ix86_split_to_parts): Likewise.
3352 (ix86_vectorize_builtin_gather): Likewise.
3353 (ix86_vectorize_builtin_scatter): Likewise.
3354 (avx_vpermilp_parallel): Likewise.
3355 (inline_memory_move_cost): Likewise.
3356 (ix86_tieable_integer_mode_p): Likewise.
3357 (x86_maybe_negate_const_int): Likewise.
3358 (ix86_expand_vector_init_duplicate): Likewise.
3359 (ix86_expand_vector_init_one_nonzero): Likewise.
3360 (ix86_expand_vector_init_one_var): Likewise.
3361 (ix86_expand_vector_init_concat): Likewise.
3362 (ix86_expand_vector_init_interleave): Likewise.
3363 (ix86_expand_vector_init_general): Likewise.
3364 (ix86_expand_vector_set): Likewise.
3365 (ix86_expand_vector_extract): Likewise.
3366 (emit_reduc_half): Likewise.
3367 (ix86_emit_i387_round): Likewise.
3368 (ix86_mangle_type): Likewise.
3369 (ix86_expand_round_sse4): Likewise.
3370 (expand_vec_perm_blend): Likewise.
3371 (canonicalize_vector_int_perm): Likewise.
3372 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3373 (expand_vec_perm_1): Likewise.
3374 (expand_vec_perm_interleave3): Likewise.
3375 (expand_vec_perm_even_odd_pack): Likewise.
3376 (expand_vec_perm_even_odd_1): Likewise.
3377 (expand_vec_perm_broadcast_1): Likewise.
3378 (ix86_vectorize_vec_perm_const_ok): Likewise.
3379 (ix86_expand_vecop_qihi): Likewise.
3380 (ix86_expand_mul_widen_hilo): Likewise.
3381 (ix86_expand_sse2_abs): Likewise.
3382 (ix86_expand_pextr): Likewise.
3383 (ix86_expand_pinsr): Likewise.
3384 (ix86_preferred_simd_mode): Likewise.
3385 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3386 * config/i386/sse.md (*andnot<mode>3): Likewise.
3387 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3388 (*<code><mode>3): Likewise.
3389 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3390 (ia64_expand_atomic_op): Likewise.
3391 (ia64_arg_type): Likewise.
3392 (ia64_mode_to_int): Likewise.
3393 (ia64_scalar_mode_supported_p): Likewise.
3394 (ia64_vector_mode_supported_p): Likewise.
3395 (expand_vec_perm_broadcast): Likewise.
3396 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3397 (iq2000_function_arg_advance): Likewise.
3398 (iq2000_function_arg): Likewise.
3399 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3400 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3401 (m68k_libcall_value): Likewise.
3402 (m68k_function_value): Likewise.
3403 (sched_attr_op_type): Likewise.
3404 * config/mcore/mcore.c (mcore_output_move): Likewise.
3405 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3407 (microblaze_function_arg): Likewise.
3408 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3409 (mips_print_operand): Likewise.
3410 (mips_mode_ok_for_mov_fmt_p): Likewise.
3411 (mips_vector_mode_supported_p): Likewise.
3412 (mips_preferred_simd_mode): Likewise.
3413 (mips_expand_vpc_loongson_even_odd): Likewise.
3414 (mips_expand_vec_unpack): Likewise.
3415 (mips_expand_vi_broadcast): Likewise.
3416 (mips_expand_vector_init): Likewise.
3417 (mips_expand_vec_reduc): Likewise.
3418 (mips_expand_msa_cmp): Likewise.
3419 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3420 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3421 (cc_flags_for_mode): Likewise.
3422 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3423 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3424 (nds32_output_casesi_pc_relative): Likewise.
3425 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3426 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3427 (nvptx_gen_unpack): Likewise.
3428 (nvptx_gen_pack): Likewise.
3429 (nvptx_gen_shuffle): Likewise.
3430 (nvptx_gen_wcast): Likewise.
3431 (nvptx_preferred_simd_mode): Likewise.
3432 * config/pa/pa.c (pa_secondary_reload): Likewise.
3433 * config/pa/predicates.md (base14_operand): Likewise.
3434 * config/powerpcspe/powerpcspe-c.c
3435 (altivec_resolve_overloaded_builtin): Likewise.
3436 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3438 (rs6000_preferred_simd_mode): Likewise.
3439 (output_vec_const_move): Likewise.
3440 (rs6000_expand_vector_extract): Likewise.
3441 (rs6000_split_vec_extract_var): Likewise.
3442 (reg_offset_addressing_ok_p): Likewise.
3443 (rs6000_legitimate_offset_address_p): Likewise.
3444 (rs6000_legitimize_address): Likewise.
3445 (rs6000_emit_set_const): Likewise.
3446 (rs6000_const_vec): Likewise.
3447 (rs6000_emit_move): Likewise.
3448 (spe_build_register_parallel): Likewise.
3449 (rs6000_darwin64_record_arg_recurse): Likewise.
3450 (swap_selector_for_mode): Likewise.
3451 (spe_init_builtins): Likewise.
3452 (paired_init_builtins): Likewise.
3453 (altivec_init_builtins): Likewise.
3454 (do_load_for_compare): Likewise.
3455 (rs6000_generate_compare): Likewise.
3456 (rs6000_expand_float128_convert): Likewise.
3457 (emit_load_locked): Likewise.
3458 (emit_store_conditional): Likewise.
3459 (rs6000_output_function_epilogue): Likewise.
3460 (rs6000_handle_altivec_attribute): Likewise.
3461 (rs6000_function_value): Likewise.
3462 (emit_fusion_gpr_load): Likewise.
3463 (emit_fusion_p9_load): Likewise.
3464 (emit_fusion_p9_store): Likewise.
3465 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3466 (fusion_gpr_mem_load): Likewise.
3467 (fusion_addis_mem_combo_load): Likewise.
3468 (fusion_addis_mem_combo_store): Likewise.
3469 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3470 (fusion_gpr_mem_load): Likewise.
3471 (fusion_addis_mem_combo_load): Likewise.
3472 (fusion_addis_mem_combo_store): Likewise.
3473 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3475 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3476 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3477 (rs6000_preferred_simd_mode): Likewise.
3478 (output_vec_const_move): Likewise.
3479 (rs6000_expand_vector_extract): Likewise.
3480 (rs6000_split_vec_extract_var): Likewise.
3481 (reg_offset_addressing_ok_p): Likewise.
3482 (rs6000_legitimate_offset_address_p): Likewise.
3483 (rs6000_legitimize_address): Likewise.
3484 (rs6000_emit_set_const): Likewise.
3485 (rs6000_const_vec): Likewise.
3486 (rs6000_emit_move): Likewise.
3487 (rs6000_darwin64_record_arg_recurse): Likewise.
3488 (swap_selector_for_mode): Likewise.
3489 (paired_init_builtins): Likewise.
3490 (altivec_init_builtins): Likewise.
3491 (rs6000_expand_float128_convert): Likewise.
3492 (emit_load_locked): Likewise.
3493 (emit_store_conditional): Likewise.
3494 (rs6000_output_function_epilogue): Likewise.
3495 (rs6000_handle_altivec_attribute): Likewise.
3496 (rs6000_function_value): Likewise.
3497 (emit_fusion_gpr_load): Likewise.
3498 (emit_fusion_p9_load): Likewise.
3499 (emit_fusion_p9_store): Likewise.
3500 * config/rx/rx.c (rx_gen_move_template): Likewise.
3501 (flags_from_mode): Likewise.
3502 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3503 (s390_slb_comparison): Likewise.
3504 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3505 (s390_vector_mode_supported_p): Likewise.
3506 (s390_cc_modes_compatible): Likewise.
3507 (s390_match_ccmode_set): Likewise.
3508 (s390_canonicalize_comparison): Likewise.
3509 (s390_emit_compare_and_swap): Likewise.
3510 (s390_branch_condition_mask): Likewise.
3511 (s390_rtx_costs): Likewise.
3512 (s390_secondary_reload): Likewise.
3513 (__SECONDARY_RELOAD_CASE): Likewise.
3514 (s390_expand_cs): Likewise.
3515 (s390_preferred_simd_mode): Likewise.
3516 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3517 * config/sh/sh.c (sh_print_operand): Likewise.
3518 (dump_table): Likewise.
3519 (sh_secondary_reload): Likewise.
3520 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3521 * config/sh/sh.md (casesi_worker_1): Likewise.
3522 (casesi_worker_2): Likewise.
3523 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3524 (fcc_comparison_operator): Likewise.
3525 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3526 (emit_soft_tfmode_cvt): Likewise.
3527 (sparc_preferred_simd_mode): Likewise.
3528 (output_cbranch): Likewise.
3529 (sparc_print_operand): Likewise.
3530 (sparc_expand_vec_perm_bmask): Likewise.
3531 (vector_init_bshuffle): Likewise.
3532 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3533 (spu_vector_mode_supported_p): Likewise.
3534 (spu_expand_insv): Likewise.
3535 (spu_emit_branch_or_set): Likewise.
3536 (spu_handle_vector_attribute): Likewise.
3537 (spu_builtin_splats): Likewise.
3538 (spu_builtin_extract): Likewise.
3539 (spu_builtin_promote): Likewise.
3540 (spu_expand_sign_extend): Likewise.
3541 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3542 (tilegx_simd_int): Likewise.
3543 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3544 (tilepro_simd_int): Likewise.
3545 * config/v850/v850.c (const_double_split): Likewise.
3546 (v850_print_operand): Likewise.
3547 (ep_memory_offset): Likewise.
3548 * config/vax/vax.c (vax_rtx_costs): Likewise.
3549 (vax_output_int_move): Likewise.
3550 (vax_output_int_add): Likewise.
3551 (vax_output_int_subtract): Likewise.
3552 * config/visium/predicates.md (visium_branch_operator): Likewise.
3553 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3554 (visium_print_operand_address): Likewise.
3555 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3556 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3557 (xtensa_expand_conditional_branch): Likewise.
3558 (xtensa_copy_incoming_a7): Likewise.
3559 (xtensa_output_literal): Likewise.
3560 * dfp.c (decimal_real_maxval): Likewise.
3561 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3564 Alan Hayward <alan.hayward@arm.com>
3565 David Sherwood <david.sherwood@arm.com>
3567 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3568 (mode_nunits_inline): Likewise.
3569 (mode_inner_inline): Likewise.
3570 (mode_unit_size_inline): Likewise.
3571 (mode_unit_precision_inline): Likewise.
3572 (emit_insn_modes_h): Likewise. Also emit a #define of the
3574 (emit_mode_wider): Add an E_ prefix to mode names.
3575 (emit_mode_complex): Likewise.
3576 (emit_mode_inner): Likewise.
3577 (emit_mode_adjustments): Likewise.
3578 (emit_mode_int_n): Likewise.
3579 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3580 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3581 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3582 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3583 (CRC32_BUILTIN, ENTRY): Likewise.
3584 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3585 (aarch64_pop_regs): Likewise.
3586 (aarch64_process_components): Likewise.
3587 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3588 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3589 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3590 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3591 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3592 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3594 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3595 (pcmpistr): Likewise.
3596 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3597 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3598 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3600 * config/rl78/rl78.c (mduc_regs): Likewise.
3601 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3602 (htm_expand_builtin): Likewise.
3603 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3604 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3605 * config/xtensa/xtensa.c (print_operand): Likewise.
3606 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3607 (NUM_MODE_VECTOR_INT): Likewise.
3608 * genoutput.c (null_operand): Likewise.
3609 (output_operand_data): Likewise.
3610 * genrecog.c (print_parameter_value): Likewise.
3611 * lra.c (debug_operand_data): Likewise.
3613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3615 * dwarf2out.c (scompare_loc_descriptor_wide)
3616 (scompare_loc_descriptor_narrow): New functions, split out from...
3617 (scompare_loc_descriptor): ...here.
3618 * expmed.c (emit_store_flag_int): New function, split out from...
3619 (emit_store_flag): ...here.
3621 2017-08-30 Richard Biener <rguenther@suse.de>
3623 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3624 (dwarf2out_early_finish): Move setting of AT_pubnames from
3625 early debug output to early finish.
3627 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3629 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3630 and -mdata-region to the assembler.
3632 2017-08-30 Richard Biener <rguenther@suse.de>
3634 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3636 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3637 (dwarf2out_early_global_decl): Only generate a DIE for the
3638 abstract origin if it doesn't already exist or is a declaration DIE.
3639 (resolve_addr): Do not add the linkage name twice when
3640 generating a stub DIE for the DW_TAG_GNU_call_site target.
3642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3644 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3645 Use machine_mode rather than int for arg1_mode.
3647 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3650 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3651 that the second argument of the built-in functions to unpack
3652 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3653 switch statement instead a lot of if statements.
3654 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3655 Allow 64-bit values to be in Altivec registers as well as
3656 traditional floating point registers.
3657 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3659 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3661 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3662 MAX_REGS_PER_ADDRESS == 1.
3664 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3666 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3667 * config/i386/i386.c (ix86_option_override_internal): Simplify
3668 setting of opts->x_flag_entry.
3670 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3671 Jakub Jelinek <jakub@redhat.com>
3672 Richard Biener <rguenther@suse.de>
3674 PR tree-optimization/81503
3675 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3676 folded constant fits in the target type; reorder tests for clarity.
3678 2017-08-29 Martin Liska <mliska@suse.cz>
3680 * passes.def: Include pass_lower_switch.
3681 * stmt.c (dump_case_nodes): Remove and move to
3682 tree-switch-conversion.
3683 (case_values_threshold): Likewise.
3684 (expand_switch_as_decision_tree_p): Likewise.
3685 (emit_case_decision_tree): Likewise.
3686 (expand_case): Likewise.
3687 (balance_case_nodes): Likewise.
3688 (node_has_low_bound): Likewise.
3689 (node_has_high_bound): Likewise.
3690 (node_is_bounded): Likewise.
3691 (emit_case_nodes): Likewise.
3692 (struct simple_case_node): New struct.
3693 (add_case_node): Remove.
3694 (emit_case_dispatch_table): Use vector instead of case_list.
3695 (reset_out_edges_aux): Remove.
3696 (compute_cases_per_edge): Likewise.
3697 (expand_case): Build list of simple_case_node.
3698 (expand_sjlj_dispatch_table): Use it.
3699 * tree-switch-conversion.c (struct case_node): Moved from
3700 stmt.c and adjusted.
3701 (emit_case_nodes): Likewise.
3702 (node_has_low_bound): Likewise.
3703 (node_has_high_bound): Likewise.
3704 (node_is_bounded): Likewise.
3705 (case_values_threshold): Likewise.
3706 (reset_out_edges_aux): Likewise.
3707 (compute_cases_per_edge): Likewise.
3708 (add_case_node): Likewise.
3709 (dump_case_nodes): Likewise.
3710 (balance_case_nodes): Likewise.
3711 (expand_switch_as_decision_tree_p): Likewise.
3712 (emit_jump): Likewise.
3713 (emit_case_decision_tree): Likewise.
3714 (try_switch_expansion): Likewise.
3715 (do_jump_if_equal): Likewise.
3716 (emit_cmp_and_jump_insns): Likewise.
3717 (fix_phi_operands_for_edge): New function.
3718 (record_phi_operand_mapping): Likewise.
3719 (class pass_lower_switch): New pass.
3720 (pass_lower_switch::execute): New function.
3721 (make_pass_lower_switch): Likewise.
3722 (conditional_probability):
3723 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3724 * tree-pass.h: Add make_pass_lower_switch.
3726 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3729 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3732 2017-08-29 Richard Biener <rguenther@suse.de>
3734 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3735 we do not add a DW_AT_inline attribute twice.
3736 (gen_subprogram_die): Remove code setting DW_AT_inline on
3737 DECL_ABSTRACT_P nodes.
3739 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3741 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3742 calls to internal functions.
3743 (gimplify_modify_expr): Likewise.
3744 * tree-call-cdce.c (use_internal_fn): Likewise.
3745 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3746 (convert_to_divmod): Set the nothrow flag.
3747 * tree-if-conv.c (predicate_mem_writes): Likewise.
3748 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3749 (vectorizable_call): Likewise.
3750 (vectorizable_store): Likewise.
3751 (vectorizable_load): Likewise.
3752 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3753 (vect_recog_mask_conversion_pattern): Likewise.
3755 2017-08-29 Martin Liska <mliska@suse.cz>
3758 * gcc.c (driver_handle_option): Add new argument.
3759 * opts-common.c (handle_option): Pass
3760 target_option_override_hook.
3761 * opts-global.c (lang_handle_option): Add new option.
3762 (set_default_handlers): Add new argument.
3763 (decode_options): Likewise.
3764 * opts.c (target_handle_option): Likewise.
3765 (common_handle_option): Call target_option_override_hook.
3766 * opts.h (struct cl_option_handler_func): Add hook for
3767 target option override.
3768 (struct cl_option_handlers): Likewise.
3769 (set_default_handlers): Add new argument.
3770 (decode_options): Likewise.
3771 (common_handle_option): Likewise.
3772 (target_handle_option): Likewise.
3773 * toplev.c (toplev::main): Pass targetm.target_option.override
3776 2017-08-29 Richard Biener <rguenther@suse.de>
3777 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3779 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3780 life to the active subtree.
3782 2017-08-28 Jeff Law <law@redhat.com>
3784 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3785 derive_equivalences.
3786 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3788 (edge_info::derive_equivalences): New private member function
3790 * tree-ssa-dom.c (class edge_info): Changed from a struct
3791 to a class. Add ctor/dtor, methods and data members.
3792 (edge_info::edge_info): Renamed from allocate_edge_info.
3793 Initialize additional members.
3794 (edge_info::~edge_info): New.
3795 (free_dom_edge_info): Delete the edge info.
3796 (record_edge_info): Use new class & associated member functions.
3797 Tighten forms for testing for edge equivalences.
3798 (record_temporary_equivalences): Iterate over the simple
3799 equivalences rather than assuming there's only one per edge.
3800 (cprop_into_successor_phis): Iterate over the simple
3801 equivalences rather than assuming there's only one per edge.
3802 (optimize_stmt): Use operand_equal_p rather than pointer
3803 equality for mini-DSE code.
3805 2017-08-28 Nathan Sidwell <nathan@acm.org>
3807 * gcc.c (execute): Fold SIGPIPE handling into switch
3808 statement. Adjust internal error message.
3810 2017-08-28 Richard Biener <rguenther@suse.de>
3813 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3814 Do nothing for removed DIEs.
3816 2017-08-28 Richard Biener <rguenther@suse.de>
3818 PR tree-optimization/81977
3819 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3822 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3825 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3826 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3827 using targetm.gen_mem_thread_fence.
3829 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3832 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3833 predicate to register_operand. Reorder operands.
3834 (*btr<mode>): Ditto.
3835 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3836 (*btr<mode>_mask): Ditto.
3838 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3840 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3841 * config/rs6000/xmmintrin.h: New file.
3842 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3844 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3847 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3848 parameter and_insn and return it.
3849 (recombine_lvx_pattern): Insert a copy to ensure availability of
3850 the base register of the copied masking operation at the point of
3851 the instruction replacement.
3852 (recombine_stvx_pattern): Likewise.
3854 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3856 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3857 undocumented switches.
3858 (-mpower9-dform-vector): Likewise.
3859 (-mpower9-dform): Likewise.
3860 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3861 comments to delete references to -mpower9-dform* switches.
3862 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3863 Delete reference to -mpower9-dform* switches, test for
3864 -mpower9-vector instead.
3865 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3866 (OTHER_P9_VECTOR_MASKS): Likewise.
3867 (POWERPC_MASKS): Likewise.
3868 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3869 tests against -mpower9-dform* to -mpower9-vector. Delete code
3870 that checked for -mpower9-dform* consistancy with other options.
3871 Add test for -mpower9-misc to enable other power9 switches.
3872 (rs6000_init_hard_regno_mode_ok): Likewise.
3873 (rs6000_option_override_internal): Likewise.
3874 (rs6000_emit_prologue): Likewise.
3875 (rs6000_emit_epilogue): Likewise.
3876 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3877 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3878 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3880 (emit_fusion_p9_store): Likewise.
3881 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3882 resetting these macros if the assembler does not support ISA 3.0
3884 (TARGET_P9_DFORM_VECTOR): Likewise.
3885 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3886 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3888 2017-08-25 Alan Modra <amodra@gmail.com>
3890 PR rtl-optimization/81747
3891 * cse.c (cse_extended_basic_block): Don't attempt to record
3892 equivalences for degenerate conditional branches that branch
3893 to their fall-through.
3895 2017-08-24 Martin Sebor <msebor@redhat.com>
3898 * gimple-fold.c (size_must_be_zero_p): New function.
3899 (gimple_fold_builtin_memory_op): Call it.
3901 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3903 * config/rs6000/mm_malloc.h: New file.
3905 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3907 PR tree-optimization/81913
3908 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3909 analysis when either IVs in condition can wrap.
3911 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3913 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3914 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3916 2017-08-24 Richard Biener <rguenther@suse.de>
3919 * targhooks.c (default_target_can_inline_p): Properly
3920 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3921 is present and always compare.
3922 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3923 infer -mfpmath=sse from TARGET_SSE_P.
3924 (ix86_can_inline_p): Properly use target_option_default_node when
3925 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3927 2017-08-24 Richard Biener <rguenther@suse.de>
3930 * dwarf2out.c (output_die): Handle flag_generate_offload like
3932 (output_comp_unit): Likewise.
3933 (gen_array_type_die): Likewise.
3934 (dwarf2out_early_finish): Likewise.
3935 (note_variable_value_in_expr): Likewise.
3936 (dwarf2out_finish): Likewise. Adjust assert.
3937 * cgraphunit.c (symbol_table::compile): Move setting of
3938 flag_generate_offload earlier ...
3939 (symbol_table::finalize_compilation_unit): ... here, before
3940 early debug finalization.
3942 2017-08-24 Richard Biener <rguenther@suse.de>
3944 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3945 and ipa-fnsummary.h.
3946 (ix86_can_inline_p): When ix86_fpmath flags do not match
3947 check whether the callee uses FP math at all.
3949 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3952 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3953 instead of TYPE_PRECISION.
3955 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3956 Alan Hayward <alan.hayward@arm.com>
3957 David Sherwood <david.sherwood@arm.com>
3959 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3961 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3963 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3965 (ix86_frame::stack_realign_allocate): New field.
3966 (struct machine_frame_state): Modify comments.
3967 (machine_frame_state::sp_realigned_fp_end): New field.
3968 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
3970 (sp_valid_at): Add assertion to assure no attempt to access invalid
3971 offset of a realigned stack.
3972 (fp_valid_at): Likewise.
3973 (choose_baseaddr): Modify comments.
3974 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
3975 ix86_expand_prologue.
3976 (ix86_expand_prologue): Modify stack realignment and allocation.
3977 (ix86_expand_epilogue): Modify comments.
3978 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
3979 avx2_runtime, avx512f, and avx512f_runtime.
3981 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
3983 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
3984 (mstackrealign): Do not init to -1.
3985 * config/i386/i386.c (ix86_option_override_internal):
3986 Check opts_set, not opts when setting default value of
3987 opts->x_ix86_force_align_arg_pointer.
3989 2017-08-23 Richard Biener <rguenther@suse.de>
3991 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
3992 lang_hooks.decl_printable_name.
3993 * print-rtl-function.c (print_rtx_function): Likewise.
3994 * tree-pretty-print.c (dump_function_header): Likewise.
3996 2017-08-23 Richard Biener <rguenther@suse.de>
3999 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4000 -g0 at compile-time.
4002 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4005 * doc/sourcebuild.texi (Other hardware attributes):
4008 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4011 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4012 Fix single-use check.
4014 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4016 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4018 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4020 * doc/install.texi: Modify to add more details on running selected
4023 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4025 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4026 is combined with -mabi=ms.
4027 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4030 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4032 PR tree-optimization/81488
4033 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4034 and cached_basis fields.
4035 (MAX_SPREAD): New constant.
4036 (alloc_cand_and_find_basis): Initialize new fields.
4037 (clear_visited): New function.
4038 (create_phi_basis_1): Rename from create_phi_basis, set visited
4039 and cached_basis fields.
4040 (create_phi_basis): New wrapper function.
4041 (phi_add_costs_1): Rename from phi_add_costs, add spread
4042 parameter, set visited field, short-circuit when limits reached.
4043 (phi_add_costs): New wrapper function.
4044 (record_phi_increments_1): Rename from record_phi_increments, set
4046 (record_phi_increments): New wrapper function.
4047 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4048 (phi_incr_cost): New wrapper function.
4049 (all_phi_incrs_profitable_1): Rename from
4050 all_phi_incrs_profitable, set visited field.
4051 (all_phi_incrs_profitable): New wrapper function.
4053 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4054 Alan Hayward <alan.hayward@arm.com>
4055 David Sherwood <david.sherwood@arm.com>
4057 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4058 that takes the outer and inner modes.
4059 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4060 comparison as the canonical test for a paradoxical subreg.
4061 * combine.c (simplify_set): Use paradoxical_subreg_p.
4062 (make_extraction): Likewise.
4063 (force_to_mode): Likewise.
4064 (rtx_equal_for_field_assignment_p): Likewise.
4065 (gen_lowpart_for_combine): Likewise.
4066 (simplify_comparison): Likewise.
4067 * cse.c (equiv_constant): Likewise.
4068 * expmed.c (store_bit_field_1): Likewise.
4069 * final.c (alter_subreg): Likewise.
4070 * fwprop.c (propagate_rtx): Likewise.
4071 (forward_propagate_subreg): Likewise.
4072 * ira-conflicts.c (ira_build_conflicts): Likewise.
4073 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4074 * lra-constraints.c (curr_insn_transform): Likewise.
4075 (split_reg): Likewise.
4076 * lra-eliminations.c (move_plus_up): Likewise.
4077 (lra_eliminate_regs_1): Likewise.
4078 * recog.c (general_operand): Likewise.
4079 * ree.c (combine_reaching_defs): Likewise.
4080 * reload.c (push_reload): Likewise.
4081 (find_reloads): Likewise.
4082 * reload1.c (elimination_effects): Likewise.
4083 (compute_reload_subreg_offset): Likewise.
4084 (choose_reload_regs): Likewise.
4085 * rtlanal.c (subreg_lsb_1): Likewise.
4086 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4087 (simplify_subreg): Likewise.
4088 * var-tracking.c (track_loc_p): Likewise.
4089 * emit-rtl.c (byte_lowpart_offset): Likewise.
4090 (paradoxical_subreg_p): Delete out-of-line definition.
4092 2017-08-22 Jeff Law <law@redhat.com>
4094 PR tree-optimization/81741
4095 PR tree-optimization/71947
4096 * tree-ssa-dom.c: Include tree-inline.h.
4097 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4098 equivalences if one is more expensive to compute than the other.
4099 * tree-ssa-scopedtables.h (class const_or_copies): Make
4100 record_const_or_copy_raw method private.
4101 (class avail_exprs_stack): New method simplify_binary_operation.
4102 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4103 avail_exprs_stack::simplify_binary_operation as needed.
4104 (avail_exprs_stack::simplify_binary_operation): New function.
4106 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4108 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4109 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4110 (DOT_SYMBOLS): Likewise.
4111 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4112 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4113 (RS6000_ABI_NAME): Likewise.
4114 (TARGET_CMODEL): Likewise.
4115 (TOC_SECTION_ASM_OP): Likewise.
4116 (SET_CMODEL): New macro.
4117 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4119 2017-08-22 Richard Biener <rguenther@suse.de>
4121 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4122 to allow for free-lang-data replacements similar to verify_type_variant.
4124 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4125 Alan Hayward <alan.hayward@arm.com>
4126 David Sherwood <david.sherwood@arm.com>
4128 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4129 VOIDmode for the LABEL_REF.
4131 2017-08-22 Richard Biener <rguenther@suse.de>
4133 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4135 2017-08-22 Richard Biener <rguenther@suse.de>
4137 * common.opt (feliminate-dwarf2-dups): Ignore.
4138 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4139 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4140 same_die_p_wrap, compute_section_prefix,
4141 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4142 (comdat_symbol_id, comdat_symbol_number): Likewise.
4143 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4145 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4146 (output_die): Mark unreachable path unreachable.
4147 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4148 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4149 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4150 (dwarf2out_early_finish): Likewise.
4152 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4154 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4156 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4159 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4160 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4161 (avr_attribute_table) <io, io_low, address>: Initialize
4162 .decl_required with true.
4164 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4166 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4167 undocumented debugging options.
4168 (-mvsx-scalar-double): Likewise.
4169 (-mallow-df-permute): Likewise.
4170 (-mvectorize-builtins): Likewise.
4171 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4172 (rs6000_builtin_vectorized_function): Likewise.
4173 (rs6000_builtin_md_vectorized_function): Likewise.
4174 (rs6000_opt_vars): Likewise.
4176 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4179 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4180 (*btrq_imm): Rename from *btrq.
4181 (*btcq_imm): Rename from *btcq.
4182 (btsc): New code attribute.
4183 (*<btsc><mode>): New insn pattern.
4184 (*btr<mode>): Ditto.
4185 (*<btsc><mode>_mask): New insn_and_split pattern.
4186 (*btr<mode>_mask): Ditto.
4188 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4189 Alan Hayward <alan.hayward@arm.com>
4190 David Sherwood <david.sherwood@arm.com>
4192 * function.c (pad_below): Simplify padding calculation.
4194 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4195 Alan Hayward <alan.hayward@arm.com>
4196 David Sherwood <david.sherwood@arm.com>
4198 * target.def (function_prologue): Remove frame size argument.
4199 (function_epilogue): Likewise.
4200 * doc/tm.texi: Regenerate.
4201 * final.c (final_start_function): Update call to function_prologue.
4202 (final_end_function): Update call to function_epilogue.
4203 (default_function_pro_epilogue): Remove frame size argument.
4204 * output.h (default_function_pro_epilogue): Likewise.
4205 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4206 (arm_output_function_prologue): Likewise.
4207 * config/frv/frv.c (frv_function_prologue): Likewise.
4208 (frv_function_epilogue): Likewise.
4209 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4210 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4211 (ia64_output_function_epilogue): Likewise.
4212 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4213 (m32r_output_function_epilogue): Likewise.
4214 * config/microblaze/microblaze.c (microblaze_function_prologue)
4215 (microblaze_function_epilogue): Likewise.
4216 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4217 (mips_output_function_epilogue): Likewise.
4218 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4219 (mmix_target_asm_function_epilogue): Likewise.
4220 * config/msp430/msp430.c (msp430_start_function): Likewise.
4221 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4222 (nds32_asm_function_epilogue): Likewise.
4223 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4224 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4225 (pa_output_function_epilogue): Likewise.
4226 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4227 (rs6000_output_function_epilogue): Likewise.
4228 * config/rl78/rl78.c (rl78_start_function): Likewise.
4229 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4230 (rs6000_output_function_epilogue): Likewise.
4231 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4232 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4233 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4234 (sparc_asm_function_epilogue): Likewise.
4236 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4238 * tree.h (type_has_mode_precision_p): New function.
4239 * convert.c (convert_to_integer_1): Use it.
4240 * expr.c (expand_expr_real_2): Likewise.
4241 (expand_expr_real_1): Likewise.
4242 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4243 * match.pd: Likewise.
4244 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4245 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4246 * tree-tailcall.c (process_assignment): Likewise.
4247 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4248 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4249 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4250 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4251 (vectorizable_assignment): Likewise.
4252 (vectorizable_shift): Likewise.
4253 (vectorizable_operation): Likewise.
4254 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4256 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4258 * match.pd: Add pow (C, x) simplification.
4260 2017-08-21 Richard Biener <rguenther@suse.de>
4262 PR tree-optimization/81900
4263 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4264 for blocks with abnormal predecessors.
4265 (compute_antic): Do not set visited flag prematurely.
4267 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4270 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4272 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4274 * stor-layout.h (vector_type_mode): Move to...
4275 * tree.h (vector_type_mode): ...here.
4276 * stor-layout.c (vector_type_mode): Move to...
4277 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4279 2017-08-21 Richard Biener <rguenther@suse.de>
4281 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4282 register_external_die hooks.
4283 (debug_false_tree_charstarstar_uhwistar): Declare.
4284 (debug_nothing_tree_charstar_uhwi): Likewise.
4285 * debug.c (do_nothing_debug_hooks): Adjust.
4286 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4287 (debug_nothing_tree_charstar_uhwi): Likewise.
4288 * dbxout.c (dbx_debug_hooks): Adjust.
4289 (xcoff_debug_hooks): Likewise.
4290 * sdbout.c (sdb_debug_hooks): Likewise.
4291 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4292 * dwarf2out.c (macinfo_label_base): New global.
4293 (dwarf2out_register_external_die): New function for the
4294 register_external_die hook.
4295 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4296 (dwarf2_debug_hooks): Use them.
4297 (dwarf2_lineno_debug_hooks): Adjust.
4298 (struct die_struct): Add with_offset flag.
4299 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4300 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4301 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4302 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4303 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4304 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4305 defining section names for the early LTO debug variants.
4306 (reset_indirect_string): New helper.
4307 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4308 (print_dw_val): Add support for offsetted symbol references.
4309 (get_ultimate_context): Split out from is_cxx.
4310 (is_cxx): Use get_ultimate_context.
4311 (is_fortran): Add decl overload.
4312 (compute_comp_unit_symbol): Split out worker from
4313 compute_section_prefix.
4314 (compute_section_prefix): Call compute_comp_unit_symbol and
4315 set comdat_type_p here.
4316 (output_die): Skip DIE symbol output for the LTO added one.
4317 Handle DIE symbol references with offset.
4318 (output_comp_unit): Guard section name mangling properly.
4319 For LTO debug sections emit a symbol at the section beginning
4320 which we use to refer to its DIEs.
4321 (add_abstract_origin_attribute): For DIEs registered via
4322 dwarf2out_register_external_die directly refer to the early
4323 DIE rather than indirectly through the shadow one we created.
4324 Remove obsolete call to dwarf2out_abstract_function for
4325 non-function/block origins.
4326 (gen_array_type_die): When generating early LTO debug do
4327 not emit DW_AT_string_length.
4328 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4329 late when in LTO. As suggested place a gcc_unreachable for
4330 the DECL_ABSTRACT_P case.
4331 (gen_subprogram_die): Avoid another specification DIE
4332 for early built declarations/definitions for the late LTO case.
4333 (gen_variable_die): Add type references for late duplicated VLA dies
4335 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4336 we have the abstract instance already.
4337 (process_scope_var): Adjust decl DIE contexts in LTO which
4338 first puts them in limbo.
4339 (gen_decl_die): Do not generate type DIEs late apart from
4340 types for VLAs or for decls we do not yet have a DIE. Do not
4341 call dwarf2out_abstract_function late.
4342 (dwarf2out_early_global_decl): Make sure to create DIEs
4343 for abstract instances of a decl first.
4344 (dwarf2out_late_global_decl): Adjust comment.
4345 (output_macinfo_op): With multiple macro sections use
4346 macinfo_label_base to distinguish labels.
4347 (output_macinfo): Likewise. Update macinfo_label_base.
4348 Pass in the line info label.
4349 (note_variable_value_in_expr): When generating LTO resolve
4350 all variable values here by generating DIEs as needed.
4351 (init_sections_and_labels): Add early LTO debug flag parameter
4352 and generate different sections and names if set. Add generation
4353 counter for the labels so we can have multiple of them.
4354 (reset_dies): Helper to allow DIEs to be output multiple times.
4355 (dwarf2out_finish): When outputting DIEs to the fat part of an
4356 LTO object first reset DIEs.
4357 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4358 (modified_type_die): Check for decl_ultimate_origin being self
4360 (gen_type_die_with_usage): Likewise.
4361 (gen_typedef_die): Allow decl_ultimate_origin being self.
4362 (set_decl_abstract_flags): Remove.
4363 (set_block_abstract_flags): Likewise.
4364 (dwarf2out_abstract_function): Treat the early generated DIEs
4365 as the abstract copy and only add DW_AT_inline and
4366 DW_AT_artificial here and call set_decl_origin_self.
4367 If the DIE has an abstract origin don't do anything.
4368 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4369 if we have none yet (Go fails to build one, PR78628).
4370 (variably_modified_type_p): Prevent endless recursion for Ada
4371 cyclic pointer types.
4372 * lto-streamer-in.c: Include debug.h.
4373 (dref_queue): New global.
4374 (lto_read_tree_1): Stream in DIE references.
4375 (lto_input_tree): Register DIE references.
4376 (input_function): Stream DECL_DEBUG_ARGS.
4377 * lto-streamer-out.c: Include debug.h.
4378 (lto_write_tree_1): Output DIE references.
4379 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4380 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4381 (output_function): Stream DECL_DEBUG_ARGS.
4382 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4383 Stream DECL_ABSTRACT_ORIGIN.
4384 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4385 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4386 DECL_CONTEXT for file-scope decls.
4387 * lto-streamer.h (struct dref_entry): Declare.
4388 (dref_queue): Likewise.
4389 * cfgexpand.c (pass_expand::execute): Do not call the
4390 outlining_inline_function hook here.
4391 * lto-wrapper.c (debug_obj): New global.
4392 (tool_cleanup): Unlink it if required.
4393 (debug_objcopy): New function.
4394 (run_gcc): Handle early debug sections in the IL files by
4395 extracting them to separate files, partially linkin them and
4396 feeding the result back as result to the linker.
4397 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4398 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4399 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4400 sections into a separate segment.
4401 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4403 (darwin_asm_dwarf_section): Likewise.
4404 (darwin_asm_output_dwarf_offset): Likewise.
4405 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4407 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4408 Alan Hayward <alan.hayward@arm.com>
4409 David Sherwood <david.sherwood@arm.com>
4411 * read-md.h (md_reader::record_potential_iterator_use): Replace
4412 pointer argument with an rtx and an index.
4413 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4414 (apply_mode_iterator): Likewise.
4415 (apply_code_iterator): Likewise.
4416 (apply_int_iterator): Likewise.
4417 (apply_subst_iterator): Likewise.
4418 (record_iterator_use): Likewise.
4419 (record_attribute_use): Likewise.
4420 (md_reader::record_potential_iterator_use): Likewise. Update calls
4421 to record_iterator_use and apply_iterator.
4422 (iterator_use): Replace ptr with x and index.
4423 (attribute_use): Likewise.
4424 (apply_attribute_uses): Update calls to apply_iterator.
4425 (apply_iterators): Likewise. Update initialization of iterator_use.
4426 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4427 and record_potential_iterator_use.
4428 (rtx_reader::read_rtx_operand): Likewise.
4430 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4431 Alan Hayward <alan.hayward@arm.com>
4432 David Sherwood <david.sherwood@arm.com>
4434 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4437 2017-08-21 Richard Biener <rguenther@suse.de>
4440 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4441 at struct end conservatively when comparing common bases.
4443 2017-08-21 Richard Biener <rguenther@suse.de>
4445 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4446 (mem_ref_in_stmt): Remove.
4447 (determine_max_movement): Use ref index to get at the reference.
4448 (invariantness_dom_walker::before_dom_children): Deal with
4449 lim data already initialized.
4450 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4452 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4454 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4455 (TARGET_ISA_ROUND): Ditto.
4456 (TARGET_ROUND): Ditto.
4457 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4458 * config/i386/i386.md: Ditto.
4459 * config/i386/sse.md: Ditto.
4460 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4461 with OPTION_MASK_ISA_SSE4_1.
4463 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4466 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4467 __builtin_ia32_lzcnt_u16.
4469 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4472 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4473 (rs6000_set_current_function): Rewrite function to use it.
4475 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4478 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4479 and TYPE_WARN_IF_NOT_ALIGN.
4480 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4481 (handle_warn_if_not_align): New.
4482 (place_union_field): Call handle_warn_if_not_align.
4483 (place_field): Call handle_warn_if_not_align.
4484 Copy TYPE_WARN_IF_NOT_ALIGN.
4485 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4486 (layout_type): Likewise.
4487 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4489 (tree_decl_common): Add warn_if_not_align.
4490 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4491 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4492 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4493 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4494 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4495 * doc/extend.texi: Document warn_if_not_aligned attribute.
4496 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4498 2017-08-17 Martin Liska <mliska@suse.cz>
4501 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4502 (get_data_dependence): Use it as pointer type.
4503 (distribute_loop): Likewise.
4505 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4507 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4508 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4509 (p8_vmrgow_<mode>_direct): New define_insn.
4510 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4511 handle endianness for vmrgew and vmrgow permute patterns.
4513 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4515 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4516 * config/rs6000/rs6000-cpus.def: Remove comment.
4517 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4518 (POWERPC_MASKS): Likewise.
4519 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4520 use of TARGET_VSX_TIMODE.
4521 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4522 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4523 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4524 (rs6000_option_override_internal): Remove dead code.
4525 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4526 (rs6000_legitimize_reload_address): Likewise.
4527 (rs6000_legitimate_address_p): Likewise.
4528 (rs6000_opt_masks): Delete "vsx-timode".
4529 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4530 from function comment.
4531 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4532 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4533 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4535 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4536 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4537 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4540 (define_peephole2): Likewise.
4542 2017-08-17 Martin Sebor <msebor@redhat.com>
4545 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4546 past the end of an array.
4547 (test_pp_format): Add test cases.
4549 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4551 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4552 missing ECF_NOTHROW flags.
4554 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4557 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4558 operands residing in integer registers.
4559 (*vsx_le_perm_load_<mode>): Likewise.
4560 (*vsx_le_perm_store_<mode>): Likewise.
4561 (define_peephole2): Add peepholes to optimize the above.
4563 2017-08-17 Marek Polacek <polacek@redhat.com>
4566 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4567 to mimic what shorten_compare did. Change the return type to bool.
4568 (fold_cond_expr_with_comparison): Update call to
4569 operand_equal_for_comparison_p.
4570 (fold_ternary_loc): Likewise.
4572 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4574 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4576 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4578 2017-08-17 Richard Biener <rguenther@suse.de>
4580 * tree-ssa-structalias.c (solve_graph): When propagating
4581 to successors update the graphs succ edges and avoid duplicate work.
4583 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4586 * config/i386/i386.c (ix86_option_override_internal): Save target
4587 specific options after ix86_stack_protector_guard_reg was changed.
4589 2017-08-17 Richard Biener <rguenther@suse.de>
4591 PR tree-optimization/81827
4592 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4593 (new_var_info): Initialize it conservatively.
4594 (get_call_vi): Mark register vars.
4595 (new_scalar_tmp_constraint_exp): Likewise.
4596 (handle_rhs_call): Likewise.
4597 (handle_const_call): Likewise.
4598 (create_function_info_for): Likewise.
4599 (solve_constraints): Sort varinfos to separate register from
4600 non-register vars to pack points-to solution bitmaps during
4603 2017-08-17 Marek Polacek <polacek@redhat.com>
4605 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4607 2017-08-17 Richard Biener <rguenther@suse.de>
4609 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4610 to true when overflow is undefined and we saturated the result.
4612 2017-08-17 Alan Modra <amodra@gmail.com>
4615 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4616 Don't use store multiple if only one reg needs saving.
4617 (interesting_frame_related_regno): New function.
4618 (rs6000_frame_related): Don't emit frame info for regs that
4620 (rs6000_emit_epilogue): Likewise.
4622 2017-08-16 Nathan Sidwell <nathan@acm.org>
4624 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4625 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4626 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4627 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4628 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4629 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4630 (verify_type): Adjust for TYPE_BINFO move.
4631 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4632 process TYPE_BINFO directly.
4633 (hash_tree): Likewise.
4634 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4636 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4639 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4641 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4643 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4646 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4647 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4648 Add ix86_binary_operator_ok to insn constraint.
4649 (*iordi_1_bts): Ditto.
4650 (*xordi_1_btc): Ditto.
4651 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4652 Update corresponding peephole2 pattern.
4656 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4658 PR tree-optimization/81832
4659 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4660 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4662 2017-08-16 Marek Polacek <polacek@redhat.com>
4665 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4666 regarding size_zero_node.
4668 2017-08-16 Martin Liska <mliska@suse.cz>
4671 * config.gcc: Respect previously set extra_objs in case
4674 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4676 PR tree-optimization/81835
4677 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4678 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4679 not depend on the phi.
4681 2017-08-16 Alan Modra <amodra@gmail.com>
4683 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4686 2017-08-16 Alan Modra <amodra@gmail.com>
4688 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4689 (save_reg_p): ..into this. Update all callers.
4690 (first_reg_to_save): Simplify.
4692 2017-08-16 Alan Modra <amodra@gmail.com>
4694 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4697 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4701 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4702 constant count if that count is less than 32.
4704 2017-08-15 Nathan Sidwell <nathan@acm.org>
4706 * gcc.c (execute): Emit friendlier message if inferior is killed
4707 by an external cause.
4709 2017-08-15 Richard Biener <rguenther@suse.de>
4711 PR tree-optimization/81790
4712 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4713 CONSTRUCTORs from simplifying and VN.
4715 2017-08-14 Martin Sebor <msebor@redhat.com>
4717 * builtin-attrs.def: Add comments.
4719 2017-08-14 Martin Sebor <msebor@redhat.com>
4722 * doc/extend.texi (attribute nonstring): Document new attribute.
4724 2017-08-14 Martin Sebor <msebor@redhat.com>
4727 * tree-diagnostic.c (default_tree_printer): Handle %G.
4728 * gimple-pretty-print.h (percent_G_format): Declare new function.
4729 * gimple-pretty-print.c (percent_G_format): Define.
4730 * tree-pretty-print.c (percent_K_format): Add argument.
4732 2017-08-14 Martin Sebor <msebor@redhat.com>
4734 PR translation/79998
4735 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4736 Remove a stray space.
4738 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4741 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4742 (*iordi_1_bts): Ditto.
4743 (*xordi_1_btc): Ditto.
4745 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4748 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4749 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4751 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4752 quoted strings, and make more translator-friendly.
4753 (darwin_rs6000_override_options): Likewise.
4754 (rs6000_option_override_internal): Likewise.
4755 (rs6000_return_in_memory): Fix overlong line.
4756 (init_cmulative_args): Use quoted strings, and make more
4757 translator-friendly.
4758 (rs6000_pass_by_reference): Fix overlong line.
4759 (def_builtin): Use quoted strings.
4760 (altivec_expand_predicate_builtin): Use quoted strings, and make
4761 more translator-friendly.
4762 (htm_expand_builtin): Use quoted strings.
4763 (cpu_expand_builtin): Use quoted strings, and make more
4764 translator-friendly.
4765 (altivec_expand_builtin): Likewise.
4766 (paired_expand_predicate_builtin): Likewise.
4767 (rs6000_invalid_builtin): Likewise.
4768 (builtin_function_type): Use quoted strings.
4769 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4770 more translator-friendly.
4771 (rs6000_trampoline_init): Likewise.
4772 (rs6000_handle_altivec_attribute): Likewise.
4773 (rs6000_inner_target_options): Use quoted strings.
4774 (rs6000_disable_incompatible_switches): Likewise.
4775 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4776 strings, and make more translator-friendly.
4777 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4779 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4781 PR tree-optimization/81799
4782 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4783 cond_expr to simple gimple operand.
4785 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4788 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4790 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4794 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4795 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4796 TARGET_GASISR_PROLOGUES.
4797 * config/avr/avr.c (avr_option_override): Same.
4798 (avr_pass_pre_proep::execute): Same.
4800 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4803 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4804 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4806 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4808 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4809 define_insn_and_split. Split to a memory load from 0 in
4810 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4811 using PTR mode iterator.
4812 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4813 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4814 (*add_tp_<mode>): Redefine as define_insn_and_split.
4815 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4816 address space. Merge with *add_tp_x32 using PTR mode iterator.
4817 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4818 Split to an add with a memory load from 0 in
4819 DEFAULT_TLS_SEG_REG address space.
4821 2017-08-12 Andrew Pinski <apinski@cavium.com>
4823 * config/aarch64/aarch64-option-extensions.def (rdma):
4824 Fix feature string to what Linux prints out in /proc/cpuinfo.
4826 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4829 * dwarf2out.c (modified_type_die): For C typedef types that have
4830 an ultimate origin, process the ultimate origin instead of the
4832 (gen_typedef_die): Assert that input DECLs have no ultimate
4834 (gen_type_die_with_usage): For typedef variants that have an
4835 ultimate origin, just call gen_decl_die on the original DECL.
4836 (process_scope_var): Avoid creating DIEs for local typedefs and
4837 concrete static variables.
4839 2017-08-12 Alan Modra <amodra@gmail.com>
4843 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4844 match gnu-user.h startfile.
4845 (ENDFILE_LINUX_SPEC): Similarly.
4847 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4850 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4852 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4854 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4855 * config/aarch64/aarch64.md (mov<mode>): Change.
4856 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4857 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4858 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4860 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4862 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4863 for storage order barriers.
4865 2017-08-11 Martin Liska <mliska@suse.cz>
4868 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4869 variables of void type.
4871 2017-08-11 Martin Liska <mliska@suse.cz>
4873 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4874 TARGET_SUPPORTS_ALIASES.
4875 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4876 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4877 (optimize_weakref): Likewise.
4878 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4879 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4880 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4882 2017-08-11 Martin Liska <mliska@suse.cz>
4885 * config/i386/i386.c (make_resolver_func): Do complete
4886 refactoring of the function.
4888 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4891 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4892 * config/i386/i386.c (ix86_stack_protect_guard): Use
4893 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4894 * doc/invoke.texi (x86 Options): Document
4895 -mstack-protector-guard-symbol= option.
4897 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4899 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4900 * config/i386/i386.c (ix86_split_stack_guard): New function.
4901 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4902 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4903 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4904 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4905 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4906 (split_stack_space_check): Call ix86_split_stack_guard.
4908 2017-08-10 Martin Sebor <msebor@redhat.com>
4910 * print-tree.c (print_node): Print location using the established
4912 Replace spaces with colons.
4913 (debug_raw, debug): Ditto.
4915 2017-08-10 Martin Sebor <msebor@redhat.com>
4918 * pretty-print.c (pp_format): Correct the handling of %s precision.
4920 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4923 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4925 (ix86_finalize_stack_frame_flags): This. Also clear
4926 frame_pointer_needed if -fno-omit-frame-pointer is used without
4928 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4929 with ix86_finalize_stack_frame_flags.
4930 (ix86_expand_epilogue): Likewise.
4931 (ix86_expand_split_stack_prologue): Likewise.
4932 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4934 2017-08-10 Martin Liska <mliska@suse.cz>
4937 * c-attribs.c (handle_target_attribute):
4938 Report warning for an empty string argument of target attribute.
4940 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4943 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4945 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4946 or DECL_NONLOCAL labels.
4947 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4948 or DECL_NONLOCAL labels here.
4950 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4952 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4953 to indicate when early gimple folding has been disabled.
4954 (rs6000_gimple_fold_builtin): Add debug content.
4955 (rs6000_invalid_builtin): Fix whitespace.
4956 (rs6000_expand_builtin): Fix whitespace.
4957 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4959 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4962 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4963 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4965 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
4967 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
4968 (qdf24xx): Likewise.
4969 * config/aarch64/aarch64.md: Include falkor.md.
4970 * config/aarch64/falkor.md: New.
4972 2017-08-09 Marek Polacek <polacek@redhat.com>
4975 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
4976 * diagnostic.c (emit_diagnostic): Add a comment.
4977 (emit_diagnostic_valist): New function.
4979 2017-08-09 Marek Polacek <polacek@redhat.com>
4982 * input.c (make_location): New overload.
4983 * input.h (make_location): Declare.
4985 2017-08-08 Alan Modra <amodra@gmail.com>
4986 H.J. Lu <hongjiu.lu@intel.com>
4989 * gcc.c (NO_PIE_SPEC): Delete.
4990 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
4992 (LINK_PIE_SPEC): ..to here.
4993 (LINK_COMMAND_SPEC): Support -no-pie.
4994 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
4995 chain of crtbegin*.o selection, update for PIE_SPEC changes and
4997 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
4998 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
4999 (ENDFILE_CRTEND_SPEC): Similarly.
5001 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5004 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5005 (mstack-protector-guard-offset=): Ditto.
5006 * config/i386/i386.c (ix86_option_override): Handle
5007 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5009 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5010 ix86_stack_protect_guard_offset variables.
5011 (TARGET_STACK_PROTECT_GUARD): Always define.
5012 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5013 and -mstack-protector-guard-offset= options.
5015 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5017 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5018 boundary case for the last candidate.
5020 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5022 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5023 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5025 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5028 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5029 * config/aarch64/aarch64-builtins.c
5030 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5031 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5032 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5034 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5035 Andrew Pinski <pinskia@gmail.com>
5038 * internal-fn.def (XORSIGN): New.
5039 * optabs.def (xorsign_optab): New.
5040 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5041 (convert_expand_mult_copysign): New.
5042 (pass_optimize_widening_mul::execute): Call
5043 convert_expand_mult_copysign.
5045 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5047 PR tree-optimization/81354
5048 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5049 Insert on edges rather than explicitly creating landing pads.
5050 (analyze_candidates_and_replace): Commit edge inserts.
5052 2017-08-08 Richard Biener <rguenther@suse.de>
5055 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5056 (expand_simple_operations): Also look through ADDR_EXPRs with
5057 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5059 2017-08-08 Richard Biener <rguenther@suse.de>
5061 PR tree-optimization/81723
5062 * tree-vect-slp.c (struct bst_traits): New hash traits.
5063 (bst_fail): New global.
5064 (vect_build_slp_tree_2): New worker, split out from ...
5065 (vect_build_slp_tree): ... this now wrapping it with using
5066 bst_fail set to cache SLP tree build fails. Properly handle
5068 (vect_analyze_slp_instance): Allocate and free bst_fail.
5070 2017-08-08 Martin Liska <mliska@suse.cz>
5073 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5074 LABEL_DECLs that can be from a different function.
5076 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5078 PR tree-optimization/81744
5079 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5080 loop's number of iterations.
5082 2017-08-08 Martin Liska <mliska@suse.cz>
5084 * asan.c: Include header files.
5085 * attribs.c (build_decl_attribute_variant): New function moved
5087 (build_type_attribute_qual_variant): Likewise.
5088 (cmp_attrib_identifiers): Likewise.
5089 (simple_cst_list_equal): Likewise.
5090 (omp_declare_simd_clauses_equal): Likewise.
5091 (attribute_value_equal): Likewise.
5092 (comp_type_attributes): Likewise.
5093 (build_type_attribute_variant): Likewise.
5094 (lookup_ident_attribute): Likewise.
5095 (remove_attribute): Likewise.
5096 (merge_attributes): Likewise.
5097 (merge_type_attributes): Likewise.
5098 (merge_decl_attributes): Likewise.
5099 (merge_dllimport_decl_attributes): Likewise.
5100 (handle_dll_attribute): Likewise.
5101 (attribute_list_equal): Likewise.
5102 (attribute_list_contained): Likewise.
5103 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5104 (lookup_attribute_by_prefix): Likewise.
5105 * bb-reorder.c: Include header files.
5106 * builtins.c: Likewise.
5107 * calls.c: Likewise.
5108 * cfgexpand.c: Likewise.
5109 * cgraph.c: Likewise.
5110 * cgraphunit.c: Likewise.
5111 * convert.c: Likewise.
5112 * dwarf2out.c: Likewise.
5113 * final.c: Likewise.
5114 * fold-const.c: Likewise.
5115 * function.c: Likewise.
5116 * gimple-expr.c: Likewise.
5117 * gimple-fold.c: Likewise.
5118 * gimple-pretty-print.c: Likewise.
5119 * gimple.c: Likewise.
5120 * gimplify.c: Likewise.
5121 * hsa-common.c: Likewise.
5122 * hsa-gen.c: Likewise.
5123 * internal-fn.c: Likewise.
5124 * ipa-chkp.c: Likewise.
5125 * ipa-cp.c: Likewise.
5126 * ipa-devirt.c: Likewise.
5127 * ipa-fnsummary.c: Likewise.
5128 * ipa-inline.c: Likewise.
5129 * ipa-visibility.c: Likewise.
5131 * lto-cgraph.c: Likewise.
5132 * omp-expand.c: Likewise.
5133 * omp-general.c: Likewise.
5134 * omp-low.c: Likewise.
5135 * omp-offload.c: Likewise.
5136 * omp-simd-clone.c: Likewise.
5137 * opts-global.c: Likewise.
5138 * passes.c: Likewise.
5139 * predict.c: Likewise.
5140 * sancov.c: Likewise.
5141 * sanopt.c: Likewise.
5142 * symtab.c: Likewise.
5143 * toplev.c: Likewise.
5144 * trans-mem.c: Likewise.
5145 * tree-chkp.c: Likewise.
5146 * tree-eh.c: Likewise.
5147 * tree-into-ssa.c: Likewise.
5148 * tree-object-size.c: Likewise.
5149 * tree-parloops.c: Likewise.
5150 * tree-profile.c: Likewise.
5151 * tree-ssa-ccp.c: Likewise.
5152 * tree-ssa-live.c: Likewise.
5153 * tree-ssa-loop.c: Likewise.
5154 * tree-ssa-sccvn.c: Likewise.
5155 * tree-ssa-structalias.c: Likewise.
5156 * tree-ssa.c: Likewise.
5157 * tree-streamer-in.c: Likewise.
5158 * tree-vectorizer.c: Likewise.
5159 * tree-vrp.c: Likewise.
5161 * ubsan.c: Likewise.
5162 * varasm.c: Likewise.
5163 * varpool.c: Likewise.
5164 * tree.c: Remove functions moved to attribs.[ch].
5166 * config/aarch64/aarch64.c: Add attrs.h header file.
5167 * config/alpha/alpha.c: Likewise.
5168 * config/arc/arc.c: Likewise.
5169 * config/arm/arm.c: Likewise.
5170 * config/avr/avr.c: Likewise.
5171 * config/bfin/bfin.c: Likewise.
5172 * config/c6x/c6x.c: Likewise.
5173 * config/cr16/cr16.c: Likewise.
5174 * config/cris/cris.c: Likewise.
5175 * config/darwin.c: Likewise.
5176 * config/epiphany/epiphany.c: Likewise.
5177 * config/fr30/fr30.c: Likewise.
5178 * config/frv/frv.c: Likewise.
5179 * config/ft32/ft32.c: Likewise.
5180 * config/h8300/h8300.c: Likewise.
5181 * config/i386/winnt.c: Likewise.
5182 * config/ia64/ia64.c: Likewise.
5183 * config/iq2000/iq2000.c: Likewise.
5184 * config/lm32/lm32.c: Likewise.
5185 * config/m32c/m32c.c: Likewise.
5186 * config/m32r/m32r.c: Likewise.
5187 * config/m68k/m68k.c: Likewise.
5188 * config/mcore/mcore.c: Likewise.
5189 * config/microblaze/microblaze.c: Likewise.
5190 * config/mips/mips.c: Likewise.
5191 * config/mmix/mmix.c: Likewise.
5192 * config/mn10300/mn10300.c: Likewise.
5193 * config/moxie/moxie.c: Likewise.
5194 * config/msp430/msp430.c: Likewise.
5195 * config/nds32/nds32-isr.c: Likewise.
5196 * config/nds32/nds32.c: Likewise.
5197 * config/nios2/nios2.c: Likewise.
5198 * config/nvptx/nvptx.c: Likewise.
5199 * config/pa/pa.c: Likewise.
5200 * config/pdp11/pdp11.c: Likewise.
5201 * config/powerpcspe/powerpcspe.c: Likewise.
5202 * config/riscv/riscv.c: Likewise.
5203 * config/rl78/rl78.c: Likewise.
5204 * config/rx/rx.c: Likewise.
5205 * config/s390/s390.c: Likewise.
5206 * config/sh/sh.c: Likewise.
5207 * config/sol2.c: Likewise.
5208 * config/sparc/sparc.c: Likewise.
5209 * config/spu/spu.c: Likewise.
5210 * config/stormy16/stormy16.c: Likewise.
5211 * config/tilegx/tilegx.c: Likewise.
5212 * config/tilepro/tilepro.c: Likewise.
5213 * config/v850/v850.c: Likewise.
5214 * config/vax/vax.c: Likewise.
5215 * config/visium/visium.c: Likewise.
5216 * config/xtensa/xtensa.c: Likewise.
5218 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5221 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5222 constraints since the -mupper-regs-* switches have been
5224 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5225 into a vector from a double word element that was extracted from
5226 another vector, and eliminate extra XXPERMDI instructions.
5227 (vsx_concat_<mode>_2): Likewise.
5228 (vsx_concat_<mode>_3): Likewise.
5229 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5230 concat to allow optimizing inserts from previous extracts.
5232 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5234 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5235 memory reference to a SSP offset in TLS address space.
5236 (ix86_print_operand) <case '@'>: Remove.
5237 (ix86_print_operand_punct_valid_p): Remove '@' code.
5238 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5240 (stack_tls_protect_set_<mode>): Remove.
5241 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5242 (stack_tls_protect_test_<mode>): Remove.
5243 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5245 2017-08-07 Olivier Hainque <hainque@adacore.com>
5248 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5250 2017-08-07 Douglas Rupp <rupp@adacore.com>
5252 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5253 variable was referenced as multidir in command.
5255 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5258 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5261 2017-08-07 Martin Liska <mliska@suse.cz>
5263 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5264 * config/rl78/rl78.c: Add include of attribs.h.
5265 * config/sh/sh.c: Likewise.
5266 * config/v850/v850.c: Likewise.
5268 2017-08-07 Tom de Vries <tom@codesourcery.com>
5271 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5273 2017-08-07 Martin Liska <mliska@suse.cz>
5275 * config/mips/mips.c: Include attribs.h.
5277 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5280 * doc/invoke.texi: Document change in behvaior for -Ofast for
5283 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5285 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5287 (aarch64_pop_regs): Likewise.
5288 (aarch64_gen_load_pair): Likewise.
5289 (aarch64_save_callee_saves): Likewise.
5290 (aarch64_restore_callee_saves): Likewise.
5292 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5294 * config/i386/i386.c: Revert the last change.
5296 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5299 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5301 (ix86_finalize_stack_frame_flags): This. Also clear
5302 frame_pointer_needed if -fno-omit-frame-pointer is used without
5304 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5305 with ix86_finalize_stack_frame_flags.
5306 (ix86_expand_epilogue): Likewise.
5307 (ix86_expand_split_stack_prologue): Likewise.
5309 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5312 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5313 to P_AES for Westmere.
5315 2017-08-07 Jonathan Yong <10walls@gmail.com>
5317 * config/i386/mingw.opt (fset-stack-executable): Removed.
5318 * config/i386/cygming.opt (fset-stack-executable): Moved
5320 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5322 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5324 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5326 2017-08-07 Marek Polacek <polacek@redhat.com>
5329 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5331 2017-08-07 Martin Liska <mliska@suse.cz>
5333 * attribs.h (canonicalize_attr_name): New function.
5334 (cmp_attribs): Move from c-format.c and adjusted.
5335 (is_attribute_p): Moved from tree.h.
5336 * tree-inline.c: Add new includes.
5337 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5338 (private_is_attribute_p): Remove.
5339 (private_lookup_attribute): Likewise.
5340 (private_lookup_attribute_by_prefix): Simplify.
5341 (remove_attribute): Use is_attribute_p.
5342 * tree.h: Remove removed declarations.
5344 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5347 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5348 instead of computing it in the function. Formatting fix.
5349 (expand_case): Don't rely on default_edge being the first edge,
5350 clear it if removing it, pass default_edge to
5351 emit_case_dispatch_table.
5352 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5355 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5357 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5358 insn in the function, emit NOP after the insn.
5360 2017-08-06 Tom de Vries <tom@codesourcery.com>
5362 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5365 2017-08-06 Tom de Vries <tom@codesourcery.com>
5367 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5370 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5372 PR tree-optimization/57371
5373 * match.pd: New pattern.
5375 2017-08-04 Marek Polacek <polacek@redhat.com>
5378 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5379 perform the computation in offset_int.
5381 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5383 PR tree-optimization/81136
5384 * tree-vectorizer.h: Include tree-hash-traits.h.
5385 (vec_base_alignments): New typedef.
5386 (vec_info): Add a base_alignments field.
5387 (vect_record_base_alignments): Declare.
5388 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5390 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5391 (create_data_ref): Add an is_conditional_in_stmt argument.
5392 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5393 the is_conditional_in_stmt field.
5394 (data_ref_loc): Add an is_conditional_in_stmt field.
5395 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5396 (find_data_references_in_stmt): Update call to create_data_ref.
5397 (graphite_find_data_references_in_stmt): Likewise.
5398 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5399 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5400 (vect_record_base_alignment): New function.
5401 (vect_record_base_alignments): Likewise.
5402 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5403 for nested statements even if we fail to compute a misalignment.
5404 Use pooled base alignments for unconditional references.
5405 (vect_find_same_alignment_drs): Compare base addresses instead
5407 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5408 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5410 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5412 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5413 Add an explicit name for the enum. Use auto_vec for slp_instances
5415 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5417 (_bb_vec_info): Add a constructor and destructor.
5418 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5419 (destroy_loop_vec_info): Delete.
5420 (vect_destroy_datarefs): Likewise.
5421 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5422 (vec_info::vec_info): New function.
5423 (vec_info::~vec_info): Likewise.
5424 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5425 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5426 destroy_loop_vec_info.
5427 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5428 (_loop_vec_info::_loop_vec_info): ...this.
5429 (destroy_loop_vec_info): Replace with...
5430 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5431 the stmt_vec_infos. Leave handling of vec_info information to its
5432 destructor. Remove explicit vector releases.
5433 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5434 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5435 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5436 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5437 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5438 (destroy_bb_vec_info): Replace with...
5439 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5440 information to its destructor.
5441 (vect_slp_analyze_bb_1): Use new and delete instead of
5442 new_bb_vec_info and destroy_bb_vec_info.
5443 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5446 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5448 * tree-data-ref.h (subscript): Add access_fn field.
5449 (data_dependence_relation): Add could_be_independent_p.
5450 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5451 (same_access_functions): Move to tree-data-ref.c.
5452 * tree-data-ref.c (ref_contains_union_access_p): New function.
5453 (access_fn_component_p): Likewise.
5454 (access_fn_components_comparable_p): Likewise.
5455 (dr_analyze_indices): Add a reference to access_fn_component_p.
5456 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5458 (constant_access_functions): Likewise.
5459 (add_other_self_distances): Likewise.
5460 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5461 (initialize_data_dependence_relation): Use XCNEW and remove
5462 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5463 of access functions that have the same type. Allow the
5464 subsequence to end with different bases in some circumstances.
5465 Record the chosen access functions in SUB_ACCESS_FN.
5466 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5467 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5468 (subscript_dependence_tester_1): Likewise dra and drb.
5469 (build_classic_dist_vector): Update calls accordingly.
5470 (subscript_dependence_tester): Likewise.
5471 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5472 DDR_COULD_BE_INDEPENDENT_P.
5473 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5474 comp_alias_ddrs instead of may_alias_ddrs.
5475 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5477 (vect_analyze_data_ref_dependence): Use it if
5478 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5479 distance vectors if that fails.
5480 (dependence_distance_ge_vf): New function.
5481 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5482 LOOP_VINFO_MAY_ALIAS_DDRS.
5484 2017-08-04 Richard Biener <rguenther@suse.de>
5487 * fold-const.c (fold_binary_loc): Properly restrict
5488 minus_var0 && minus_var1 case when associating undefined overflow
5491 2017-08-04 Tom de Vries <tom@codesourcery.com>
5493 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5495 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5497 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5498 Don't start diagnostic messages with a capital letter.
5499 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5501 (rs6000_invalid_builtin): Likewise.
5502 (rs6000_trampoline_init): Likewise.
5504 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5507 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5508 after setting changeable df flags.
5510 2017-08-03 Richard Biener <rguenther@suse.de>
5512 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5513 up if the use is in USE - X.
5515 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5517 * toplev.c (dumpfile.h): New include.
5518 (internal_error_reentered): New static function. Use it...
5519 (internal_error_function): ...here to handle reentered internal_error.
5521 2017-08-03 Richard Biener <rguenther@suse.de>
5524 * fold-const.c (split_tree): Add minus_var and minus_con
5525 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5526 here but always use minus_*.
5527 (associate_trees): Assert we never associate with MINUS_EXPR
5528 and NULL first operand. Do not recurse for PLUS_EXPR operands
5529 when associating as MINUS_EXPR either.
5530 (fold_binary_loc): Track minus_var and minus_con.
5532 2017-08-03 Tom de Vries <tom@codesourcery.com>
5535 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5536 ACCEL_COMPILER, apply finish_options on
5537 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5539 2017-08-03 Tom de Vries <tom@codesourcery.com>
5542 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5543 function_entry_patch_area_size > 0.
5545 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5548 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5549 instead of 10??LU, perform unit multiplication in wide_int,
5550 don't change alloc_object_size_limit if the limit is larger
5553 PR tree-optimization/81655
5554 PR tree-optimization/81588
5555 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5556 the case when ranges[i].low and high are 1 for unsigned type with
5560 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5561 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5563 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5565 * tree-vrp.h: Add include guard.
5567 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5570 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5571 (ud2): New insn pattern.
5572 * config/i386/i386.c (ix86_expand_epilogue):
5573 For naked functions, generate ud2 instead of trap insn.
5575 2017-08-02 Marek Polacek <polacek@redhat.com>
5578 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5580 2017-08-02 Tom de Vries <tom@codesourcery.com>
5581 Cesar Philippidis <cesar@codesourcery.com>
5583 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5584 Add missing edge probabilities.
5586 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5588 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5591 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5594 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5595 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5596 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5597 prologue_seq sequences - if any.
5599 2017-08-02 Richard Biener <rguenther@suse.de>
5601 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5602 via vectors if supported, integer extracts via punning if supported
5603 or otherwise vector extracts.
5605 2017-08-02 Richard Biener <rguenther@suse.de>
5607 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5609 (bitmap_insert_into_set): ... this.
5611 2017-08-02 Richard Biener <rguenther@suse.de>
5613 PR tree-optimization/81633
5615 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5617 PR tree-optimization/71752
5618 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5620 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5622 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5623 (machine_function::call_ms2sysv_pad_out): Remove field.
5624 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5625 (ix86_compute_frame_layout): Likewise.
5627 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5630 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5631 attribute with interrupt attribute.
5633 2017-08-01 Andrew Pinski <apinski@cavium.com>
5635 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5636 BIT_INSERT_EXPR's operand 1
5637 to see if the types precision matches.
5639 2017-08-01 Martin Liska <mliska@suse.cz>
5642 * builtins.c (expand_builtin_memcpy_args): Remove.
5643 (expand_builtin_memcpy): Call newly added function
5644 expand_builtin_memory_copy_args.
5645 (expand_builtin_memcpy_with_bounds): Likewise.
5646 (expand_builtin_mempcpy): Remove last argument.
5647 (expand_builtin_mempcpy_with_bounds): Likewise.
5648 (expand_builtin_memory_copy_args): New function created from
5649 expand_builtin_mempcpy_args with small modifications.
5650 (expand_builtin_mempcpy_args): Remove.
5651 (expand_builtin_stpcpy): Remove unused argument.
5652 (expand_builtin): Likewise.
5653 (expand_builtin_with_bounds): Likewise.
5655 2017-08-01 Martin Liska <mliska@suse.cz>
5658 Make mempcpy more optimal (PR middle-end/70140).
5660 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5663 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5664 __builtin_vec_cmpne verify both arguments are compatible vectors
5665 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5666 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5667 move computation of aligned to after checking the argument types.
5671 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5672 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5675 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5676 Doug Rupp <rupp@adacore.com>
5677 Olivier Hainque <hainque@adacore.com>
5679 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5680 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5682 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5683 for TARGET_OS_CPP_BUILTIN.
5684 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5685 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5687 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5688 passing required abi options to the assembler for EABI configurations.
5689 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5690 of .text.hot and .text.unlikely sections for kernel modules when
5691 using ARM style exceptions.
5692 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5693 options. Add EXTRA_CC1_SPEC.
5694 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5696 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5698 (ARM_TARGET2_DWARF_FORMAT): Define.
5699 * config/arm/t-vxworks: Adjust multilib control to removal of the
5700 Diab command line options.
5702 2017-08-01 Martin Liska <mliska@suse.cz>
5704 PR gcov-profile/81561
5705 * gcov.c (unblock): Make unblocking safe as we need to preserve
5706 index correspondence of blocks and block_lists.
5708 2017-08-01 Richard Biener <rguenther@suse.de>
5710 PR tree-optimization/81181
5711 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5712 (compute_antic): ... end of iteration here.
5714 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5716 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5717 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5718 (ftree-slp-vectorize): Likewise.
5719 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5720 can no longer be set independent of flag_tree_loop_vectorize.
5721 * omp-general.c (emp_max_vf): Likewise.
5722 * opts.c (enable_fdo_optimizations): Remove references to
5723 flag_tree_vectorize, these are now implicit.
5724 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5725 and leave it for the options machinery.
5727 2017-08-01 Martin Liska <mliska@suse.cz>
5730 * builtins.c (expand_builtin_memcpy_args): Remove.
5731 (expand_builtin_memcpy): Call newly added function
5732 expand_builtin_memory_copy_args.
5733 (expand_builtin_memcpy_with_bounds): Likewise.
5734 (expand_builtin_mempcpy): Remove last argument.
5735 (expand_builtin_mempcpy_with_bounds): Likewise.
5736 (expand_builtin_memory_copy_args): New function created from
5737 expand_builtin_mempcpy_args with small modifications.
5738 (expand_builtin_mempcpy_args): Remove.
5739 (expand_builtin_stpcpy): Remove unused argument.
5740 (expand_builtin): Likewise.
5741 (expand_builtin_with_bounds): Likewise.
5743 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5746 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5747 print "ds:" only for immediates in generic address space.
5749 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5752 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5753 (ix86_function_ok_for_sibcall): Return false for naked functions.
5755 2017-08-01 Richard Biener <rguenther@suse.de>
5757 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5758 (compute_antic): Seed worklist with exit block predecessors.
5759 * cfganal.c (dfs_find_deadend): For a cycle return the source
5760 of the edge closing it.
5762 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5764 * config/aarch64/aarch64.c
5765 (aarch64_can_const_movi_rtx_p): Move 0 check.
5767 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5769 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5770 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5772 * match.pd: Ditto in address comparison pattern.
5774 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5776 PR tree-optimization/81627
5777 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5778 closed ssa form for store-store chain.
5780 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5782 PR tree-optimization/81620
5783 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5784 for store-store chain.
5786 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5788 PR tree-optimization/81588
5789 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5790 ranges[i].in_p, invert comparison code ccode. For >/>=,
5791 swap rhs1 and rhs2 and comparison code unconditionally,
5792 for </<= don't do that. Don't swap rhs1/rhs2 again if
5793 ranges[i].in_p, instead invert comparison code ccode if
5794 opcode or oe->rank is BIT_IOR_EXPR.
5797 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5798 a direct optab to conversion optab.
5799 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5800 with GET_MODE_INNER as last argument instead of optab_handler.
5801 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5802 vector extraction if possible and optab is available.
5803 * expr.c (store_constructor): Use convert_optab_handler instead
5804 of optab_handler. Use vector initialization from smaller
5805 vectors if possible and optab is available.
5806 * tree-vect-stmts.c (vectorizable_load): Likewise.
5807 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5809 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5810 of vec_init from half-sized vectors with the same element mode.
5811 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5812 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5813 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5814 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5815 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5816 after mode in gen_vec_extract* calls.
5817 (vec_extract<mode>): Renamed to ...
5818 (vec_extract<mode><ssescalarmodelower>): ... this.
5819 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5820 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5821 element mode after mode in gen_vec_init* calls.
5822 (VEC_INIT_HALF_MODE): New mode iterator.
5823 (vec_init<mode>): Renamed to ...
5824 (vec_init<mode><ssescalarmodelower>): ... this.
5825 (vec_init<mode><ssehalfvecmodelower>): New expander.
5826 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5827 (vec_extractv2sfsf): ... this.
5828 (vec_initv2sf): Renamed to ...
5829 (vec_initv2sfsf): ... this.
5830 (vec_extractv2si): Renamed to ...
5831 (vec_extractv2sisi): ... this.
5832 (vec_initv2si): Renamed to ...
5833 (vec_initv2sisi): ... this.
5834 (vec_extractv4hi): Renamed to ...
5835 (vec_extractv4hihi): ... this.
5836 (vec_initv4hi): Renamed to ...
5837 (vec_initv4hihi): ... this.
5838 (vec_extractv8qi): Renamed to ...
5839 (vec_extractv8qiqi): ... this.
5840 (vec_initv8qi): Renamed to ...
5841 (vec_initv8qiqi): ... this.
5842 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5843 (vec_init<mode>): Renamed to ...
5844 (vec_init<mode><VEC_base_l>): ... this.
5845 (vec_extract<mode>): Renamed to ...
5846 (vec_extract<mode><VEC_base_l>): ... this.
5847 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5848 (vec_initv2sfsf): ... this.
5849 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5850 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5851 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5852 element mode after mode in gen_vec_init* calls.
5853 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5854 (vec_init<mode><Vel>): ... this.
5855 (vec_extract<mode>): Renamed to ...
5856 (vec_extract<mode><Vel>): ... this.
5857 * config/aarch64/iterators.md (Vel): New mode attribute.
5858 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5859 Add element mode after mode in gen_vec_extract* calls.
5860 * config/s390/vector.md (non_vec_l): New mode attribute.
5861 (vec_extract<mode>): Renamed to ...
5862 (vec_extract<mode><non_vec_l>): ... this.
5863 (vec_init<mode>): Renamed to ...
5864 (vec_init<mode><non_vec_l>): ... this.
5865 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5866 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5868 * config/arm/iterators.md (V_elem_l): New mode attribute.
5869 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5870 (vec_extract<mode><V_elem_l>): ... this.
5871 (vec_extractv2di): Renamed to ...
5872 (vec_extractv2didi): ... this.
5873 (vec_init<mode>): Renamed to ...
5874 (vec_init<mode><V_elem_l>): ... this.
5875 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5876 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5877 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5878 Add element mode after gen_vec_extract* calls.
5879 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5880 (vec_init<mode><unitmode>): ... this.
5881 (vec_extract<mode>): Renamed to ...
5882 (vec_extract<mode><unitmode>): ... this.
5883 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5884 (vec_init<mode><unitmode>): ... this.
5885 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5886 (vec_initv2sfsf): ... this.
5887 (vec_extractv2sf): Renamed to ...
5888 (vec_extractv2sfsf): ... this.
5889 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5890 Add element mode after gen_vec_extract* calls.
5891 * config/mips/mips.md (unitmode): New mode iterator.
5892 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5893 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5894 * config/spu/spu.md (inner_l): New mode attribute.
5895 (vec_init<mode>): Renamed to ...
5896 (vec_init<mode><inner_l>): ... this.
5897 (vec_extract<mode>): Renamed to ...
5898 (vec_extract<mode><inner_l>): ... this.
5899 * config/sparc/sparc.md (veltmode): New mode iterator.
5900 (vec_init<VMALL:mode>): Renamed to ...
5901 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5902 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5903 (vec_initv2sisi): ... this.
5904 (vec_initv2sf): Renamed to ...
5905 (vec_initv2sfsf): ... this.
5906 (vec_extractv2sf): Renamed to ...
5907 (vec_extractv2sfsf): ... this.
5908 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5909 (vec_init<mode>): Renamed to ...
5910 (vec_init<mode><VEC_base_l>): ... this.
5911 (vec_extract<mode>): Renamed to ...
5912 (vec_extract<mode><VEC_base_l>): ... this.
5913 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5914 (vec_initv2sfsf): ... this.
5915 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5916 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5917 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5918 gen_vec_init* calls.
5920 2017-08-01 Richard Biener <rguenther@suse.de>
5922 PR tree-optimization/81297
5923 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5924 TREE_OVERFLOW from INTEGER_CSTs.
5926 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5928 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5930 2017-07-31 Carl Love <cel@us.ibm.com>
5932 * config/rs6000/rs6000-c: Add support for built-in functions
5933 vector signed char vec_xl_be (signed long long, signed char *);
5934 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5935 vector signed int vec_xl_be (signed long long, signed int *);
5936 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5937 vector signed long long vec_xl_be (signed long long, signed long long *);
5938 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5939 vector signed short vec_xl_be (signed long long, signed short *);
5940 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5941 vector double vec_xl_be (signed long long, double *);
5942 vector float vec_xl_be (signed long long, float *);
5943 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5944 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5945 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5947 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5948 (altivec_expand_builtin): Add switch statement to call
5949 altivec_expand_xl_be for each builtin.
5950 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5951 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5952 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5953 __builtin_vsx_le_be_v16qi.
5954 * doc/extend.texi: Update the built-in documentation file for the
5955 new built-in functions.
5957 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5960 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5962 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5964 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5966 * config.gcc: Add z14.
5967 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5968 CPU model numbers for z13s and z14.
5969 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
5971 * config/s390/s390-opts.h (enum processor_type): Rename
5972 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5973 * config/s390/s390.c (processor_table): Add field for CPU name to
5974 be passed to Binutils.
5975 (s390_asm_output_machine_for_arch): Use the new field in
5976 processor_table for Binutils.
5977 (s390_expand_builtin): Replace arch12 with z14.
5978 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5979 (s390_get_sched_attrmask): Likewise.
5980 (s390_get_unit_mask): Likewise.
5981 * config/s390/s390.opt: Add z14 to processor_type enum.
5983 2017-07-31 Martin Jambor <mjambor@suse.cz>
5986 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
5987 regardless of optimization level.
5989 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
5990 Martin Liska <mliska@suse.cz>
5992 * predict.def: Remove old comment and adjust probability.
5993 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
5996 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5999 * config/i386/i386.c (ix86_function_naked): New function.
6000 (ix86_can_use_return_insn_p): Return false for naked functions.
6001 (ix86_expand_prologue): Skip prologue for naked functions.
6002 (ix86_expand_epilogue): Skip epilogue for naked functions
6003 and emit trap instruction.
6004 (ix86_warn_func_return): New function.
6005 (ix86_attribute_table): Add "naked" attribute specification.
6006 (TARGET_WARN_FUNC_RETURN): Define.
6007 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6009 2017-07-31 Martin Liska <mliska@suse.cz>
6011 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6012 (dump_gimple_bb_header): Always dump BB info.
6013 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6015 2017-07-31 Martin Liska <mliska@suse.cz>
6018 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6019 also with current_function_decl non-null equality.
6021 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6024 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6025 change type to the element type, instead add eltype variable and
6026 use it where we are interested in the element type.
6028 PR tree-optimization/81603
6029 * ipa-polymorphic-call.c
6030 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6031 offset arithmetic in offset_int, bail out if the resulting bit offset
6032 doesn't fit into shwi.
6034 2017-07-31 Martin Liska <mliska@suse.cz>
6036 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6037 (gimplify_save_expr): Fix comment.
6039 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6042 * config/i386/i386.c (ix86_function_arg): Update arguments for
6044 (ix86_compute_frame_layout): Set the initial stack offset to
6045 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6046 INCOMING_FRAME_SP_OFFSET.
6047 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6048 stack before exception handler returns.
6049 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6050 the 'ERROR_CODE' for exception handler.
6052 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6054 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6055 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6056 (ASM_OUTPUT_REG_POP): Ditto.
6057 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6058 instead of asm_fprintf to output pure string.
6060 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6062 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6063 to imported_module_or_decl hook.
6064 (debug_nothing_tree_tree_tree_bool): Remove.
6065 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6066 * debug.c (do_nothing_debug_hooks): Use
6067 debug_nothing_tree_tree_tree_bool_bool instead of
6068 debug_nothing_tree_tree_tree_bool.
6069 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6070 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6071 * sdbout.c (sdb_debug_hooks): Likewise.
6072 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6073 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6075 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6076 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6077 attribute, don't add anything.
6079 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6081 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6082 (fold_build2_stat_loc): Likewise.
6083 (fold_build3_stat_loc): Likewise.
6084 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6085 (fold_build1_loc): Remove macro.
6086 (fold_build2_loc): Likewise.
6087 (fold_build3_loc): Likewise.
6089 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6091 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6092 (gimple_build_debug_bind_source_stat): Likewise.
6093 * gimple.h (gimple_build_debug_bind): Remove macro.
6094 (gimple_build_debug_bind_source): Likewise.
6096 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6098 * bitmap.c (bitmap_alloc): Adjust.
6099 (bitmap_gc_alloc): Likewise.
6100 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6102 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6104 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6105 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6106 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6107 (bitmap_gc_alloc_stat): Likewise.
6108 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6110 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6112 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6113 * rtl.h (shallow_copy_rtx): Remove macro.
6115 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6117 * emit-rtl.c (gen_raw_REG): Adjust.
6118 * gengenrtl.c (gendef): Likewise.
6119 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6120 * rtl.h (rtx_alloc): Remove macro.
6122 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6124 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6125 (build_tree_list_stat): Likewise.
6126 * tree.h (build_tree_list): Remove macro.
6127 (build_tree_list_vec): Likewise.
6129 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6131 * tree.c (make_vector_stat): Remove _stat from name.
6132 (build_vector_stat): Likewise.
6133 * tree.h (make_vector_stat): Remove macro.
6134 (build_vector_stat): Likewise.
6136 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6138 * tree.h (build_var_debug_value): Remove prototype.
6140 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6142 * tree.c (tree_cons_stat): Remove _stat from name.
6143 * tree.h (tree_cons): Remove macro.
6145 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6147 * tree.c (build_vl_exp_stat): Remove _stat from name.
6148 * tree.h (build_vl_exp): Remove macro.
6150 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6152 * tree.c (build_decl_stat): Remove _stat from name.
6153 * tree.h (build_decl): Remove macro.
6155 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6157 * gimple.c (gimple_build_with_ops_stat): Adjust.
6158 (gimple_alloc_stat): Remove _stat from name.
6159 * gimple.h (gimple_alloc): Remove macro.
6161 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6163 * tree.c (make_tree_vec_stat): Remove _stat from name.
6164 (grow_tree_vec_stat): Likewise.
6165 * tree.h (make_tree_vec_stat): Adjust prototype.
6166 (grow_tree_vec_stat): Likewise.
6167 (make_tree_vec): Remove macro.
6168 (grow_tree_vec): Likewise.
6170 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6172 * fold-const.c (fold_build1_stat_loc): Adjust.
6173 (fold_build2_stat_loc): Likewise.
6174 (fold_build3_stat_loc): Likewise.
6175 * tree.c (build0_stat): Remove _stat from name.
6176 (build1_stat): Likewise.
6177 (build2_stat): Likewise.
6178 (build3_stat): Likewise.
6179 (build4_stat): Likewise.
6180 (build5_stat): Likewise.
6181 * tree.h (build1_loc): Remove macro, and rename _stat function
6183 (build2_loc): Likewise.
6184 (build3_loc): Likewise.
6185 (build4_loc): Likewise.
6186 (build5_loc): Likewise.
6188 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6190 * tree.c (make_int_cst_stat): Remove _stat from name.
6191 * tree.h (make_int_cst_stat): Adjust prototype.
6192 (make_int_cst): Remove macro.
6194 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6196 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6197 * tree.h (make_tree_binfo_stat): Adjust prototype.
6198 (make_tree_binfo): Remove.
6200 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6202 * tree.c (copy_node_stat): Rename to copy_node.
6203 (build_distinct_type_copy): Adjust.
6204 * tree.h (copy_node_stat): Adjust prototype.
6205 (copy_node): Remove macro.
6207 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6209 * tree.c (make_node_stat): rename to make_node.
6210 (build_tree_list_stat): Adjust.
6211 (build0_stat): Likewise.
6212 (build2_stat): Likewise.
6213 (build3_stat): Likewise.
6214 (build4_stat): Likewise.
6215 (build5_stat): Likewise.
6216 (build_decl_stat): Likewise.
6217 * tree.h (make_node_stat): Adjust prototype.
6218 (make_node): remove macro.
6220 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6222 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6223 (PPC_FEATURE2_SCV): Likewise.
6224 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6226 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6228 * config/aarch64/aarch64.c
6229 (aarch64_internal_mov_immediate): Add new special pattern.
6230 * config/aarch64/aarch64.md (*movdi_aarch64):
6231 Add reg/32bit const mov case.
6233 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6234 Richard Sandiford <richard.sandiford@linaro.org>
6236 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6237 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6238 Add integer and movi cases.
6239 (movi-split-hf-df-sf split, fp16): New.
6240 (enabled): Added TARGET_FP_F16INST.
6241 * config/aarch64/iterators.md (GPF_HF): New.
6242 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6244 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6246 * config/aarch64/aarch64.c
6247 (aarch64_simd_container_mode): Add prototype.
6248 (aarch64_expand_mov_immediate): Add HI support.
6249 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6250 (aarch64_can_const_movi_rtx_p): New.
6251 (aarch64_preferred_reload_class):
6252 Remove restrictions of using FP registers for certain SIMD operations.
6253 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6254 (aarch64_valid_floating_const): Add integer move validation.
6255 (aarch64_simd_imm_scalar_p): Remove.
6256 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6257 (aarch64_legitimate_constant_p): Expand list of supported cases.
6258 * config/aarch64/aarch64-protos.h
6259 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6260 (aarch64_reinterpret_float_as_int): New.
6261 (aarch64_simd_imm_scalar_p): Remove.
6262 * config/aarch64/constraints.md (Uvi): New.
6263 (Dd): Split into Ds and new Dd.
6264 * config/aarch64/aarch64.md (*movsi_aarch64):
6266 (*movdi_aarch64): Add SIMD mov case.
6268 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6270 * tree-predcom.c: (struct chain): Handle store-store chain in which
6271 stores for elimination only store loop invariant values.
6272 (execute_pred_commoning_chain): Ditto.
6273 (prepare_initializers_chain_store_elim): Ditto.
6274 (prepare_finalizers): Ditto.
6275 (is_inv_store_elimination_chain): New function.
6276 (initialize_root_vars_store_elim_1): New function.
6278 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6280 * tree-predcom.c: Revise general description of the pass.
6281 (enum chain_type): New enum type for store elimination.
6282 (struct chain): New field supporting store elimination.
6283 (struct component): Ditto.
6284 (dump_chain): Dump store-stores chain.
6285 (release_chain): Release resources.
6286 (split_data_refs_to_components): Compute and create component
6287 contains only stores for elimination.
6288 (get_chain_last_ref_at): New function.
6289 (make_invariant_chain): Initialization.
6290 (make_rooted_chain): Specify chain type in parameter and record it.
6291 (add_looparound_copies): Skip for store-stores chain.
6292 (determine_roots_comp): Compute type of chain and pass it to
6294 (initialize_root_vars_store_elim_2): New function.
6295 (finalize_eliminated_stores): New function.
6296 (remove_stmt): Handle store for elimination.
6297 (execute_pred_commoning_chain): Execute predictive commoning on
6299 (determine_unroll_factor): Skip unroll for store-stores chain.
6300 (prepare_initializers_chain_store_elim): New function.
6301 (prepare_initializers_chain): Hanlde store-store chain.
6302 (prepare_finalizers_chain, prepare_finalizers): New function.
6303 (tree_predictive_commoning_loop): Return integer value indicating
6304 if loop is unrolled or lcssa form is corrupted.
6305 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6307 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6309 * tree-predcom.c (initialize_root): Delete.
6310 (execute_pred_commoning_chain): Initialize root vars and replace
6311 reference of non-combined chain directly, rather than call above
6314 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6316 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6317 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6319 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6321 * tree-predcom.c (struct chain): New field init_seq.
6322 (release_chain): Release init_seq.
6323 (prepare_initializers_chain): Record intialization stmts in above
6325 (insert_init_seqs): New function.
6326 (tree_predictive_commoning_loop): Call insert_init_seqs.
6328 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6330 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6332 2017-07-28 Richard Biener <rguenther@suse.de>
6334 * match.pd: Remove superfluous :c.
6335 * genmatch.c (simplify::id): Add member.
6336 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6338 (current_id): New global.
6339 (dt_node::parent): Move from ...
6340 (dt_operand::parent): ... here. Add for_id member.
6341 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6342 (decision_tree::find_node): Relax order requirement when
6343 merging DT_TRUE nodes to ones inbetween the current simplify
6344 and the one we try to merge with. Add diagnostic whenever
6345 we need to enforce pattern order by not merging.
6346 (decision_tree::insert): Set current_id.
6347 (decision_tree::print_node): Dump parent node and for_id.
6348 (parser::last_id): Add member.
6349 (parser::push_simplify): Assign unique id.
6350 (parser::parser): Initialize last_id.
6352 2017-07-28 Martin Liska <mliska@suse.cz>
6355 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6356 gimple_build_debug_bind.
6358 2017-07-28 Richard Biener <rguenther@suse.de>
6360 PR tree-optimization/81502
6361 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6363 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6365 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6366 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6367 for BIT_FIELD_REF args.
6368 * fold-const.c (make_bit_field_ref): Likewise.
6369 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6371 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6374 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6375 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6376 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6377 Or it into SANITIZER_UNDEFINED.
6378 * ubsan.c: Include gimple-fold.h and varasm.h.
6379 (ubsan_expand_ptr_ifn): New function.
6380 (instrument_pointer_overflow): New function.
6381 (maybe_instrument_pointer_overflow): New function.
6382 (instrument_object_size): Formatting fix.
6383 (pass_ubsan::execute): Call instrument_pointer_overflow
6384 and maybe_instrument_pointer_overflow.
6385 * internal-fn.c (expand_UBSAN_PTR): New function.
6386 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6387 * sanitizer.def (__ubsan_handle_pointer_overflow,
6388 __ubsan_handle_pointer_overflow_abort): New builtins.
6389 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6390 * internal-fn.def (UBSAN_PTR): New internal function.
6391 * opts.c (sanitizer_opts): Add pointer-overflow.
6392 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6393 * fold-const.c (build_range_check): Compute pointer range check in
6394 integral type if pointer arithmetics would be needed. Formatting
6397 2017-07-28 Martin Liska <mliska@suse.cz>
6400 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6401 parameters that are of a variable-length.
6403 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6405 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6407 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6408 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6409 (CRT_CALL_STATIC_FUNCTION): Likewise.
6410 (ASM_DEFAULT_SPEC): New define.
6411 (ASM_SPEC32): Likewise.
6412 (ASM_SPEC64): Likewise.
6413 (ASM_SPEC_COMMON): Likewise.
6414 (ASM_SPEC): Likewise.
6415 (INVALID_64BIT): Likewise.
6416 (LINK_OS_DEFAULT_SPEC): Likewise.
6417 (LINK_OS_SPEC32): Likewise.
6418 (LINK_OS_SPEC64): Likewise.
6419 (POWERPC_LINUX): Likewise.
6420 (PTRDIFF_TYPE): Likewise.
6421 (RESTORE_FP_PREFIX): Likewise.
6422 (RESTORE_FP_SUFFIX): Likewise.
6423 (SAVE_FP_PREFIX): Likewise.
6424 (SAVE_FP_SUFFIX): Likewise.
6425 (SIZE_TYPE): Likewise.
6426 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6427 (TARGET_64BIT): Likewise.
6428 (TARGET_64BIT): Likewise.
6429 (TARGET_AIX): Likewise.
6430 (WCHAR_TYPE_SIZE): Likewise.
6431 (WCHAR_TYPE): Undefine.
6432 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6433 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6434 (CPP_OS_RTEMS_SPEC): Delete.
6435 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6436 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6438 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6440 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6442 PR tree-optimization/81578
6443 * tree-parloops.c (build_new_reduction): Bail out if
6444 reduction_code isn't one of the standard OpenMP reductions.
6445 Move the details printing after that decision.
6447 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6449 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6450 related to reload_in_progress.
6451 (splat_input_operand): Likewise.
6452 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6454 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6456 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6457 (TARGET_INSTANTIATE_DECLS): Likewise.
6458 (legitimate_indexed_address_p): Delete reload_in_progress code.
6459 (rs6000_debug_legitimate_address_p): Likewise.
6460 (rs6000_eliminate_indexed_memrefs): Likewise.
6461 (rs6000_emit_le_vsx_store): Likewise.
6462 (rs6000_emit_move_si_sf_subreg): Likewise.
6463 (rs6000_emit_move): Likewise.
6464 (register_to_reg_type): Likewise.
6465 (rs6000_pre_atomic_barrier): Likewise.
6466 (rs6000_machopic_legitimize_pic_address): Likewise.
6467 (rs6000_allocate_stack_temp): Likewise.
6468 (rs6000_address_for_fpconvert): Likewise.
6469 (rs6000_address_for_altivec): Likewise.
6470 (rs6000_secondary_memory_needed_rtx): Delete function.
6471 (rs6000_check_sdmode): Likewise.
6472 (rs6000_alloc_sdmode_stack_slot): Likewise.
6473 (rs6000_instantiate_decls): Likewise.
6474 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6475 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6476 Delete reload_in_progress.
6477 (*vec_reload_and_plus_<mptrsize>): Likewise.
6478 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6479 (vsx_div_v2di): Likewise.
6480 (vsx_udiv_v2di): Likewise.
6482 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6484 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6485 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6486 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6487 (rs6000_debug_reg_global): Delete print of LRA status.
6488 (rs6000_option_override_internal): Delete dead LRA related code.
6489 (rs6000_lra_p): Delete function.
6490 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6492 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6494 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6495 * config/riscv/rtems.h: New file.
6497 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6498 Sudakshina Das <sudi.das@arm.com>
6500 * config/aarch64/aarch64.md
6501 (define_split for and<mode>3nr_compare): Move
6502 non aarch64_logical_operand to a register.
6503 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6504 register immediate operand to a register.
6505 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6507 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6510 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6512 2017-07-27 Richard Biener <rguenther@suse.de>
6514 PR tree-optimization/81573
6515 PR tree-optimization/81494
6516 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6517 multi defuse cycle case.
6519 2017-07-27 Richard Biener <rguenther@suse.de>
6521 PR tree-optimization/81571
6522 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6525 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6527 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6528 earlier and only if MASK_FPU is set. Adjust formatting.
6530 2017-07-27 Martin Liska <mliska@suse.cz>
6532 * opt-functions.awk: Add validation of value of Init.
6533 * optc-gen.awk: Pass new argument.
6535 2017-07-27 Martin Liska <mliska@suse.cz>
6537 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6538 Fix wrong condition.
6540 2017-07-27 Martin Liska <mliska@suse.cz>
6542 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6543 BBs and edges seen by autoFDO.
6545 2017-07-27 Richard Biener <rguenther@suse.de>
6547 PR tree-optimization/81502
6548 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6549 with incompatible but same sized type.
6550 (execute_update_addresses_taken): Likewise.
6552 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6554 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6555 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6557 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6560 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6561 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6562 Change s_operand to memory_operand.
6564 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6566 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6567 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6568 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6569 Emit instructions rather than returning an expression. Handle TFmode
6570 and KFmode by casting to TImode.
6571 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6572 (rs6000_emit_le_vsx_store): Likewise.
6573 * config/rs6000/vsx.md (VSX_TI): New iterator.
6574 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6575 (*vsx_le_undo_permute_<mode>): Likewise.
6576 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6577 emit the split sequence.
6578 (*vsx_le_perm_store_<mode>): Likewise.
6580 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6582 PR tree-optimization/81555
6583 PR tree-optimization/81556
6584 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6585 if true, force CHANGED for the recursive invocation.
6586 (reassociate_bb): Remember original length of ops array, pass
6587 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6589 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6590 attributes for noipa attribute. For naked attribute use
6591 lookup_attribute first before lookup_attribute_spec.
6592 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6594 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6595 for functions with noipa attribute.
6596 (cgraph_externally_visible_p): Return true for functions with noipa
6598 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6599 for functions with noipa attribute.
6600 * doc/extend.texi: Document noipa function attribute.
6601 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6602 also for functions with noipa attribute.
6603 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6605 2017-07-26 Andrew Pinski <apinski@cavium.com>
6607 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6608 vec_unalign_load_cost and vec_unalign_store_cost.
6610 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6612 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6613 -mvsx-small-integer option.
6614 (ISA_3_0_MASKS_IEEE): Likewise.
6615 (OTHER_VSX_VECTOR_MASKS): Likewise.
6616 (POWERPC_MASKS): Likewise.
6617 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6618 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6619 code, only testing for DImode being allowed in non-VSX floating
6621 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6622 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6624 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6625 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6626 TARGET_P8_VECTOR test.
6627 (rs6000_secondary_reload_simple_move): Likewise.
6628 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6629 since TARGET_P9_VECTOR was already tested.
6630 (rs6000_opt_masks): Remove -mvsx-small-integer.
6631 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6632 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6634 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6635 test for TARGET_VEXTRACTUB was used, and that uses
6637 (p9 extract splitter): Likewise.
6638 (vsx_extract_<mode>_di_p9): Likewise.
6639 (vsx_extract_<mode>_store_p9): Likewise.
6640 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6641 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6642 the elimination of TARGET_VSX_SMALL_INTEGER.
6643 (vsx_extract_<mode>_p8): Likewise.
6644 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6645 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6646 (vsx_set_<mode>_p9): Likewise.
6647 (vsx_set_v4sf_p9): Likewise.
6648 (vsx_set_v4sf_p9_zero): Likewise.
6649 (vsx_insert_extract_v4sf_p9): Likewise.
6650 (vsx_insert_extract_v4sf_p9_2): Likewise.
6651 * config/rs6000/rs6000.md (sign extend splitter): Change
6652 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6653 (floatsi<mode>2_lfiwax_mem): Likewise.
6654 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6655 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6656 since a test for TARGET_P9_VECTOR was used.
6657 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6658 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6659 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6660 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6661 TARGET_P8_VECTOR test.
6662 (fix_trunc<mode>si2_stfiwx): Likewise.
6663 (fix_trunc<mode>si2_internal): Likewise.
6664 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6665 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6667 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6668 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6669 TARGET_P8_VECTOR test.
6670 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6671 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6672 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6674 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6675 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6676 since a test for TARGET_P9_VECTOR was used.
6677 (splitter for loading small constants): Likewise.
6679 2017-07-26 Andrew Pinski <apinski@cavium.com>
6681 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6684 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6687 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6688 (fp_valid_at): Likewise.
6690 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6692 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6693 (qdf24xx_addrcost_table): Likewise.
6694 (cortexa57_tunings): Update to use generic_branch_cost.
6695 (cortexa72_tunings): Likewise.
6696 (cortexa73_tunings): Likewise.
6697 (qdf24xx_tunings): Likewise.
6699 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6701 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6702 (thunderx2t99_branch_cost): Likewise.
6703 (cortexa35_tunings): Update to use generic_branch_cost.
6704 (cortexa53_tunings): Likewise.
6705 (cortexa57_tunings): Likewise.
6706 (cortexa72_tunings): Likewise.
6707 (cortexa73_tunings): Likewise.
6708 (thunderx2t99_tunings): Likewise.
6710 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6712 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6713 (sparc_option_override): Honour MASK_FSMULD.
6714 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6715 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6716 * config/sparc/sparc.opt (mfsmuld): New option.
6717 * doc/invoke.texi (mfsmuld): Document option.
6719 2017-07-26 Marek Polacek <polacek@redhat.com>
6722 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6724 2017-07-26 Richard Biener <rguenther@suse.de>
6726 * gimple-match-head.c (do_valueize): Return OP if valueize
6728 (get_def): New helper to get at the def stmt of a SSA name
6730 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6731 do_valueize to get at the def stmt.
6732 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6734 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6737 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6739 2017-07-26 Martin Liska <mliska@suse.cz>
6742 * function.c (expand_function_start): Make expansion of
6743 nonlocal_goto_save_area after parm_birth_insn.
6745 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6747 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6748 from all CPU target flags enable members.
6750 2017-07-26 Richard Biener <rguenther@suse.de>
6752 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6753 (decision_tree::gen): Make 'type' const.
6754 (write_predicate): Likewise.
6756 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6758 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6759 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6760 (rs6000_option_override_internal): Likewise.
6761 (rs6000_expand_vector_set): Likewise.
6762 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6763 (TARGET_UPPER_REGS_SF): Likewise.
6764 (TARGET_UPPER_REGS_DI): Likewise.
6765 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6766 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6767 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6768 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6769 (Splitters for DI constants in Altivec registers): Likewise.
6770 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6771 (vsx_set_v4sf_p9): Likewise.
6772 (vsx_set_v4sf_p9_zero): Likewise.
6773 (vsx_insert_extract_v4sf_p9): Likewise.
6774 (vsx_insert_extract_v4sf_p9_2): Likewise.
6776 2017-07-25 Carl Love <cel@us.ibm.com>
6778 * doc/extend.texi: Update the built-in documentation file for the
6779 existing built-in functions
6780 vector signed char vec_cnttz (vector signed char);
6781 vector unsigned char vec_cnttz (vector unsigned char);
6782 vector signed short vec_cnttz (vector signed short);
6783 vector unsigned short vec_cnttz (vector unsigned short);
6784 vector signed int vec_cnttz (vector signed int);
6785 vector unsigned int vec_cnttz (vector unsigned int);
6786 vector signed long long vec_cnttz (vector signed long long);
6787 vector unsigned long long vec_cnttz (vector unsigned long long);
6789 2017-07-25 Andrew Pinski <apinski@cavium.com>
6791 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6792 accesses where the use is for the first operand of a BIT_INSERT.
6794 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6797 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6798 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6800 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6802 * config/i386/gstabs.h: Delete.
6803 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6805 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6807 * config/i386/i386.c (ix86_decompose_address): Do not check for
6808 register RTX when looking at index_reg or base_reg.
6809 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6811 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6813 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6814 to update EH info here.
6816 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6818 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6820 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6822 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6824 2017-07-25 Torsten Duwe <duwe@suse.de>
6826 * common.opt: Introduce -fpatchable-function-entry
6827 command line option, and its variables function_entry_patch_area_size
6828 and function_entry_patch_area_start.
6829 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6830 including a two-value parser.
6831 * target.def (print_patchable_function_entry): New target hook.
6832 * targhooks.h (default_print_patchable_function_entry): New function.
6833 * targhooks.c (default_print_patchable_function_entry): Likewise.
6834 * toplev.c (process_options): Switch off IPA-RA if
6835 patchable function entries are being generated.
6836 * varasm.c (assemble_start_function): Look at the
6837 patchable-function-entry command line switch and current
6838 function attributes and maybe generate NOP instructions by
6839 calling the print_patchable_function_entry hook.
6840 * doc/extend.texi: Document patchable_function_entry attribute.
6841 * doc/invoke.texi: Document -fpatchable_function_entry
6842 command line option.
6843 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6845 * doc/tm.texi: Re-generate.
6847 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6850 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6851 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6853 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6855 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6857 2017-07-25 Richard Biener <rguenther@suse.de>
6859 PR tree-optimization/81455
6860 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6861 not walk in cycles when looking for guards.
6863 2017-07-25 Richard Biener <rguenther@suse.de>
6865 PR tree-optimization/81529
6866 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6867 when optimizing backedge uses.
6869 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6871 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6873 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6874 to dl_section_ref. On AIX, append an expression to subtract
6875 the size of the section length to dl_section_ref.
6877 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6879 * configure.ac: If any of the config.* scripts fail, exit 1.
6880 * configure: Regenerate.
6882 2017-07-25 Richard Biener <rguenther@suse.de>
6885 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6886 of immediate uses, be more verbose on errors.
6888 2017-07-25 Richard Biener <rguenther@suse.de>
6890 PR tree-optimization/81510
6891 * tree-vect-loop.c (vect_is_simple_reduction): When the
6892 reduction stmt is not inside the loop bail out.
6894 2017-07-25 Richard Biener <rguenther@suse.de>
6896 PR tree-optimization/81303
6897 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6898 profitability check against LOOP_VINFO_NITERSM1.
6900 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6902 * domwalk.c (cmp_bb_postorder): Simplify.
6903 (sort_bbs_postorder): New function. Use it...
6904 (dom_walker::walk): ...here to optimize common cases.
6906 2017-07-25 Martin Liska <mliska@suse.cz>
6909 * ipa-visibility.c (function_and_variable_visibility): Make the
6910 redirection just on target that supports aliasing.
6911 Fix GNU coding style.
6913 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6916 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6918 * config/arm/rtems.h: Likewise.
6919 * config/bfin/rtems.h: Likewise.
6920 * config/i386/rtemself.h: Likewise.
6921 * config/lm32/rtems.h: Likewise.
6922 * config/m32c/rtems.h: Likewise.
6923 * config/m68k/rtemself.h: Likewise.
6924 * config/microblaze/rtems.h: Likewise.
6925 * config/mips/rtems.h: Likewise.
6926 * config/moxie/rtems.h: Likewise.
6927 * config/nios2/rtems.h: Likewise.
6928 * config/powerpcspe/rtems.h: Likewise.
6929 * config/rs6000/rtems.h: Likewise.
6930 * config/rtems.h: Likewise.
6931 * config/sh/rtems.h: Likewise.
6932 * config/sh/rtemself.h: Likewise.
6933 * config/sparc/rtemself.h: Likewise.
6935 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6938 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6939 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6940 * tree-ssa-structalias.c (alias_get_name): Same.
6942 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6945 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6946 instructions if no du chain is found.
6948 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6950 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6952 2017-07-25 Richard Biener <rguenther@suse.de>
6955 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6958 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6960 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6962 (ISA_2_7_MASKS_SERVER): Likewise.
6963 (ISA_3_0_MASKS_IEEE): Likewise.
6964 (OTHER_P8_VECTOR_MASKS): Likewise.
6965 (OTHER_VSX_VECTOR_MASKS): Likewise.
6966 (POWERPC_MASKS): Likewise.
6967 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
6968 duplicate list of options.
6969 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
6970 explicit -mupper-regs options.
6971 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
6972 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
6973 alias for -mupper-regs-df.
6974 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6975 (rs6000_init_hard_regno_mode_ok): Likewise.
6976 (rs6000_option_override_internal): Likewise.
6977 (rs6000_opt_masks): Likewise.
6978 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
6979 options in terms of whether -mvsx or -mpower8-vector was used.
6980 (TARGET_UPPER_REGS_DI): Likewise.
6981 (TARGET_UPPER_REGS_SF): Likewise.
6982 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
6983 -mupper-regs-* options.
6985 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6987 * passes.c (emergency_dump_function): Print some empty lines and a
6988 header before the RTL dump.
6990 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6992 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
6994 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
6997 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6998 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7000 2017-07-24 Carl Love <cel@us.ibm.com>
7002 * config/rs6000/rs6000-c.c: Add support for built-in functions
7003 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7004 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7005 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7006 vec_extract_fp_from_shortl): Add defines for the two builtins.
7007 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7008 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7010 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7011 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7012 * doc/extend.texi: Update the built-in documentation file for the
7013 new built-in function.
7015 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7018 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7020 * doc/generic.texi: Likewise.
7021 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7022 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7024 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7026 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7027 (aarch64_mls_elt_merge<mode>): Likewise.
7029 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7031 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7032 having __cxa_atexit.
7034 2017-07-23 Michael Collison <michael.collison@arm.com>
7036 * config/arm/arm.c (arm_option_override): Deprecate
7037 use of -mstructure-size-boundary.
7038 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7039 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7041 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7044 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7045 Reduce cost estimate for direct moves.
7047 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7050 * config/i386/i386.c (ix86_option_override_internal): Disable
7051 BMI, BMI2 and TBM instructions for -m16.
7053 2017-07-21 Carl Love <cel@us.ibm.com>
7055 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7056 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7057 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7058 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7059 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7060 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7061 VMULOSW): New enum "unspec" values.
7062 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7063 altivec_vmulosw): New patterns.
7064 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7065 VMULOSW): Add definitions.
7067 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7069 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7070 (qdf24xx): Likewise.
7071 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7072 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7073 (AARCH64_FL_V8_1): Renumber.
7074 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7075 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7076 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7077 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7078 rdma to feature modifiers list.
7080 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7083 * ipa-visibility (function_and_variable_visibility): Convert
7084 recursive PLT call to direct call if appropriate.
7086 2017-07-21 Andrew Pinski <apinski@cavium.com>
7088 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7089 operand 1 to see if the types precision matches.
7090 * fold-const.c (operand_equal_p): Likewise.
7092 2017-07-21 Richard Biener <rguenther@suse.de>
7094 PR tree-optimization/81303
7095 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7096 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7097 (vect_peeling_hash_get_lowest_cost): Adjust.
7098 (vect_enhance_data_refs_alignment): Likewise. Use
7099 vect_get_peeling_costs_all_drs to compute the penalty for no
7100 peeling to match up costs.
7102 2017-07-21 Richard Biener <rguenther@suse.de>
7104 PR tree-optimization/81500
7105 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7106 we didn't identify a reduction path.
7108 2017-07-21 Tom de Vries <tom@codesourcery.com>
7109 Cesar Philippidis <cesar@codesourcery.com>
7111 PR gcov-profile/81442
7112 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7115 2017-07-21 Tom de Vries <tom@codesourcery.com>
7118 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7120 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7121 nvptx_override_options_after_change.
7123 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7125 * dwarf2out.c (output_file_names): Avoid double testing for
7128 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7130 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7132 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7134 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7136 (try_crossjump_to_edge): Do not punt on partitioned functions.
7138 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7140 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7141 Put all BBs reachable only via paths crossing cold region to cold
7143 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7145 2016-07-21 Richard Biener <rguenther@suse.de>
7147 PR tree-optimization/81303
7148 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7149 into account prologue and epilogue iterations when raising
7150 min_profitable_iters to sth at least covering one vector iteration.
7152 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7154 * config/arm/arm.c (arm_test_cpu_arch_dat):
7157 2017-07-20 Nathan Sidwell <nathan@acm.org>
7159 Remove TYPE_METHODS.
7160 * tree.h (TYPE_METHODS): Delete.
7161 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7162 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7163 (dbxout_type_methods): Scan TYPE_FIELDS.
7164 (dbxout_type): Don't check TYPE_METHODS here.
7165 * function.c (use_register_for_decl): Always ignore register for
7166 class types when not optimizing.
7167 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7168 * tree.c (free_lang_data_in_type): Stitch out member functions and
7169 templates from TYPE_FIELDS.
7170 (build_distinct_type_copy, verify_type_variant,
7171 verify_type): Member fns are on TYPE_FIELDS.
7172 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7173 * tree-pretty-print.c (dump_generic_node): Likewise.
7175 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7178 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7179 V2TImode and V4TImode.
7180 (ix86_expand_vector_extract): Likewise.
7181 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7182 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7183 (ssescalarmode): Handle V4TImode and V2TImode.
7184 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7185 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7186 (VEXTRACTI128_MODE): New mode iterator.
7187 (splitter for *vec_extractv?ti first element): New.
7188 (VEC_INIT_MODE): New mode iterator.
7189 (vec_init<mode>): Consolidate 3 expanders into one using
7190 VEC_INIT_MODE mode iterator.
7192 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7194 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7195 non_spilled_static_chain_regno_p.
7197 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7199 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7201 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7203 * bb-reorder.c (connect_traces): Allow copying of blocks within
7206 2017-07-20 Richard Biener <rguenther@suse.de>
7208 * gimple.h (gimple_phi_result): Add gphi * overload.
7209 (gimple_phi_result_ptr): Likewise.
7210 (gimple_phi_arg): Likewise. Adjust index assert to only
7211 allow actual argument accesses rather than all slots available
7213 (gimple_phi_arg_def): Add gphi * overload.
7214 * tree-phinodes.c (make_phi_node): Initialize only actual
7216 (resize_phi_node): Clear memory not covered by old node,
7217 do not initialize excess argument slots.
7218 (reserve_phi_args_for_new_edge): Initialize new argument slot
7221 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7223 PR tree-optimization/81388
7225 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7227 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7228 by removing computation of may_be_zero.
7230 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7231 Tom de Vries <tom@codesourcery.com>
7234 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7235 when gimple level profile disagrees with what RTL expander did.
7237 2017-07-20 Richard Biener <rguenther@suse.de>
7239 PR tree-optimization/61171
7240 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7241 (vect_analyze_stmt): Add slp instance parameter.
7242 (vectorizable_reduction): Likewise.
7243 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7244 (vect_is_simple_reduction): Deal with chains not detected
7245 as SLP reduction chain, specifically not properly associated
7246 chains containing a mix of plus/minus.
7247 (get_reduction_op): Remove.
7248 (get_initial_defs_for_reduction): Simplify, pass in whether
7249 this is a reduction chain, pass in the SLP node for the PHIs.
7250 (vect_create_epilog_for_reduction): Get the SLP instance as
7252 (vectorizable_reduction): Get the SLP instance as arg.
7253 During analysis remember the SLP node with the PHIs in the
7254 instance. Simplify getting at the vectorized reduction PHIs.
7255 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7256 through SLP instance.
7257 (vect_slp_analyze_operations): Likewise.
7258 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7259 (vect_transform_stmt): Likewise.
7261 2017-07-20 Tom de Vries <tom@codesourcery.com>
7263 PR tree-optimization/81489
7264 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7265 read of phi arg location to before loop that modifies phi.
7267 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7269 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7272 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7275 * except.c (execute): Fix ordering issue.
7277 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7279 PR rtl-optimization/81423
7280 * combine.c (make_compound_operation_int): Don't try to optimize
7281 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7283 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7285 PR rtl-optimization/81423
7286 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7287 with a constant that is -1 in the truncated to mode.
7289 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7291 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7292 (determine_unlikely_bbs): ... here.
7293 * predict.h (propagate_unlikely_bbs_forward): Declare.
7294 * cfgexpand.c (pass_expand::execute): Use it.
7295 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7297 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7298 propagate_unlikely_bbs_forward.
7300 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7303 * except.c (maybe_add_nop_after_section_switch): New function.
7306 2017-07-19 Tom de Vries <tom@codesourcery.com>
7308 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7310 2017-07-19 Tom de Vries <tom@codesourcery.com>
7312 * gimple.h (gimple_phi_arg): Make assert more strict.
7314 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7316 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7317 * config/rs6000/mmintrin.h: New file.
7318 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7320 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7322 PR tree-optimization/81346
7323 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7325 2017-07-19 Tom de Vries <tom@codesourcery.com>
7327 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7329 2017-07-19 Tom de Vries <tom@codesourcery.com>
7331 * config/nvptx/nvptx-modes.def: Add V2DImode.
7332 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7333 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7334 (nvptx_output_mov_insn): Handle lack of mov.b128.
7335 (nvptx_print_operand): Handle 'H' and 'L' codes.
7336 (nvptx_vector_mode_supported): Allow V2DImode.
7337 (nvptx_preferred_simd_mode): New function.
7338 (nvptx_data_alignment): New function.
7339 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7340 nvptx_preferred_simd_mode.
7341 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7343 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7345 2017-07-19 Tom de Vries <tom@codesourcery.com>
7347 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7348 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7349 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7350 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7351 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7352 (mov<VECIM>_insn): New define_insn.
7353 (define_expand "mov<VECIM>): New define_expand.
7355 2017-07-19 Tom de Vries <tom@codesourcery.com>
7357 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7359 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7361 PR tree-optimization/81346
7362 * fold-const.h (fold_div_compare, range_check_type): Declare.
7363 * fold-const.c (range_check_type): New function.
7364 (build_range_check): Use range_check_type.
7365 (fold_div_compare): No longer static, rewritten into
7366 a match.pd helper function.
7367 (fold_comparison): Don't call fold_div_compare here.
7368 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7371 2017-07-19 Nathan Sidwell <nathan@acm.org>
7373 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7374 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7375 * tree.c (find_decls_types_r, verify_type): Use
7376 TYPE_{MIN,MAX}_VALUE_RAW.
7377 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7378 (hash_tree): Likewise.
7379 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7381 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7384 2017-07-18 Tom de Vries <tom@codesourcery.com>
7387 * omp-expand.c (expand_omp_for_static_chunk): Handle
7388 equal-argument loop exit phi.
7390 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7393 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7394 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7395 operand 2 predicate.
7396 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7397 operand 2 predicate.
7398 (ror,rol -> rorx splitters): Use const_int_operand as
7399 operand 2 predicate.
7401 2017-06-18 Richard Biener <rguenther@suse.de>
7403 PR tree-optimization/81410
7404 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7405 the gap in the ! slp_perm SLP case after each group.
7407 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7410 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7413 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7416 * predict.c (set_even_probabilities): Cleanup; do not affect
7417 probabilities that are already known.
7418 (combine_predictions_for_bb): Call even when count is set.
7420 2017-07-18 Nathan Sidwell <nathan@acm.org>
7422 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7425 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7428 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7429 optimization for loop niter analysis.
7431 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7434 * config/avr/avr.c (avr_optimize_casesi): Don't use
7435 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7437 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7439 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7440 body_cost_vec from _vect_peel_extended_info.
7441 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7442 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7445 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7447 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7449 2017-07-18 Richard Biener <rguenther@suse.de>
7451 PR tree-optimization/80620
7452 PR tree-optimization/81403
7453 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7454 info when re-using a VN table entry.
7456 2017-07-18 Richard Biener <rguenther@suse.de>
7458 PR tree-optimization/81418
7459 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7460 vectype_in. Verify that with lane-reducing reduction operations
7461 we have a single def-use cycle.
7463 2017-07-17 Carl Love <cel@us.ibm.com>
7465 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7467 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7468 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7469 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7470 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7471 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7472 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7473 VMULOSW): New enum "unspec" values.
7474 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7475 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7476 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7477 altivec_vmulosw): New patterns.
7478 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7479 VMULOSW): Add definitions.
7480 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7482 * config/alpha/alpha.c: Include predict.h.
7484 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7486 * tree-vrp.c (compare_assert_loc): Fix comparison function
7487 to return predictable results.
7489 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7491 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7494 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7495 * doc/invoke.texi (mexpand-adddi): Update text.
7497 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7499 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7500 that also clobbers the CC register. The old expand code is moved
7502 (*arc_clzsi2): ... here.
7503 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7504 the CC register. The old expand code is moved to ...
7505 (arc_ctzsi2): ... here.
7507 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7509 * config/arc/arc.opt (mindexed-loads): Use initial value
7510 TARGET_INDEXED_LOADS_DEFAULT.
7511 (mauto-modify-reg): Use initial value
7512 TARGET_AUTO_MODIFY_REG_DEFAULT.
7513 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7514 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7515 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7516 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7518 2017-07-17 Martin Liska <mliska@suse.cz>
7521 * opts.c (finish_options): Do not allow -fgnu-tm
7522 w/ -fsanitize={kernel-,}address. Say sorry.
7524 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7527 * tree-loop-distribution.c (classify_partition): Only assert on
7528 numer of iterations.
7529 (merge_dep_scc_partitions): Delete prameter. Update function call.
7530 (distribute_loop): Remove code handling loop with unknown niters.
7531 (pass_loop_distribution::execute): Skip loop with unknown niters.
7533 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7536 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7537 function sort_partitions_by_post_order.
7539 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7541 PR tree-optimization/81374
7542 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7543 the max index of basic blocks, rather than number of basic blocks.
7545 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7547 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7549 (arc_legitimate_pic_operand_p): Likewise.
7550 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7552 (arc_needs_pcl_p): Likewise.
7553 (arc_legitimate_pc_offset_p): Likewise.
7554 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7555 function is also used in constrains.md.
7556 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7557 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7558 PLUS. Only return true/false in known cases, otherwise assert.
7559 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7560 is already called in arc_legitimate_constant_p.
7561 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7563 (LEGITIMATE_PIC_OPERAND_P): Use
7564 arc_raw_symbolic_reference_mentioned_p function.
7565 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7570 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7571 Andrew Burgess <andrew.burgess@embecosm.com>
7573 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7574 (arc_return_address_register): New function.
7575 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7576 (arc_handle_fndecl_attribute): Add naked attribute.
7577 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7578 (TARGET_WARN_FUNC_RETURN): Likewise.
7579 (arc_allocate_stack_slots_for_args): New function.
7580 (arc_warn_func_return): Likewise.
7581 (machine_function): Change type fn_type.
7582 (arc_compute_function_type): Consider new naked function type,
7583 change function return type.
7584 (arc_must_save_register): Adapt to handle new
7585 arc_compute_function_type's return type.
7586 (arc_expand_prologue): Likewise.
7587 (arc_expand_epilogue): Likewise.
7588 (arc_return_address_regs): Delete.
7589 (arc_return_address_register): New function.
7590 (arc_epilogue_uses): Use above function.
7591 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7592 (arc_function_type): Change encoding, add naked type.
7593 (ARC_INTERRUPT_P): Change to handle the new encoding.
7594 (ARC_FAST_INTERRUPT_P): Likewise.
7595 (ARC_NORMAL_P): Define.
7596 (ARC_NAKED_P): Likewise.
7597 (arc_compute_function_type): Delete prototype.
7598 * config/arc/arc.md (in_ret_delay_slot): Use
7599 arc_return_address_register function.
7600 (simple_return): Likewise.
7601 (p_return_i): Likewise.
7603 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7605 PR tree-optimization/81428
7606 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7607 can't be built for those types.
7609 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7611 Remove stuff dead since r239246.
7613 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7614 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7615 (avr_inform_devices): Remove dead stuff.
7617 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7619 * config/arm/arm_neon.h: Fix softp typo.
7621 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7623 PR tree-optimization/81365
7624 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7625 aggregate moves onto bb predecessor edges, make sure there are no
7626 loads that could alias the lhs in between the start of bb and the
7629 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7632 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7633 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7634 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7636 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7638 PR tree-optimization/81396
7639 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7640 (init_symbolic_number): Initialize it to 1.
7641 (perform_symbolic_merge): Add n_ops from both operands into the new
7643 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7644 without base_addr as useless if they need more than one operation.
7645 (bswap_replace): Handle !bswap case for NULL base_addr.
7647 2017-07-17 Tom de Vries <tom@codesourcery.com>
7650 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7653 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7655 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7656 conditional builtin define __FIX_LEON3FT_B2BST.
7658 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7660 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7661 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7664 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7666 PR rtl-optimization/81424
7667 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7668 to remove potential trapping from operands if -fnon-call-exceptions.
7670 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7672 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7673 profile_proability for scalling.
7674 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7676 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7678 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7680 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7682 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7683 fixpoint arithmetics.
7685 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7687 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7688 fixpoint arithmetics.
7690 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7692 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7693 fixpoint arithmetics.
7695 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7697 * profile-count.h (profile_probability::from_reg_br_prob_note,
7698 profile_probability::to_reg_br_prob_note): New functions.
7699 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7700 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7701 * predict.c (probability_reliable_p): Update.
7702 (edge_probability_reliable_p): Update.
7703 (br_prob_note_reliable_p): Update.
7704 (invert_br_probabilities): Update.
7705 (add_reg_br_prob_note): New function.
7706 (combine_predictions_for_insn): Update.
7707 * asan.c (asan_clear_shadow): Update.
7708 * cfgbuild.c (compute_outgoing_frequencies): Update.
7709 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7710 (update_br_prob_note): Update.
7711 (rtl_verify_edges): Update.
7712 (purge_dead_edges): Update.
7713 (fixup_reorder_chain): Update.
7714 * emit-rtl.c (try_split): Update.
7715 * ifcvt.c (cond_exec_process_insns): Update.
7716 (cond_exec_process_if_block): Update.
7717 (dead_or_predicable): Update.
7718 * internal-fn.c (expand_addsub_overflow): Update.
7719 (expand_neg_overflow): Update.
7720 (expand_mul_overflow): Update.
7721 * loop-doloop.c (doloop_modify): Update.
7722 * loop-unroll.c (compare_and_jump_seq): Update.
7723 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7724 * predict.h: Update.
7725 * reorg.c (mostly_true_jump): Update.
7727 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7728 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7729 * config/arc/arc.c: (emit_unlikely_jump): Update.
7730 * config/arm/arm.c: (emit_unlikely_jump): Update.
7731 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7732 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7733 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7734 (ix86_print_operand): Update.
7735 (ix86_split_fp_branch): Update.
7736 (predict_jump): Update.
7737 * config/ia64/ia64.c (ia64_print_operand): Update.
7738 * config/mmix/mmix.c (mmix_print_operand): Update.
7739 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7740 (rs6000_expand_split_stack_prologue): Update.
7741 * config/rs6000/rs6000.c: Update.
7742 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7743 (s390_expand_vec_movstr): Update.
7744 (s390_expand_cs_tdsi): Update.
7745 (s390_expand_split_stack_prologue): Update.
7746 * config/sh/sh.c (sh_print_operand): Update.
7747 (expand_cbranchsi4): Update.
7748 (expand_cbranchdi4): Update.
7749 * config/sparc/sparc.c (output_v9branch): Update.
7750 * config/spu/spu.c (get_branch_target): Update.
7751 (ea_load_store_inline): Update.
7752 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7753 * config/tilepro/tilepro.c: Update.
7755 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7757 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7758 (gimplify_save_expr): Likewise.
7760 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7762 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7764 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7766 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7768 * timevar.def (TV_IPA_FNSUMMARY): Define.
7770 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7772 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7773 to back store errata sensitive sequence from being generated.
7774 (sqrtdf2_fix): Likewise.
7776 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7778 * tree-ssa-threadupdate.c (compute_path_counts,
7779 update_joiner_offpath_counts): Use profile_probability.
7781 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7784 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7786 * config/arm/arm-c.c (arm_cpu_builtins): Define
7787 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7789 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7791 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7792 array entries to represent __ieee128 versions of the
7793 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7794 scalar_extract_sig, and scalar_insert_exp built-in functions.
7795 (altivec_resolve_overloaded_builtin): Add special case handling
7796 for the __builtin_scalar_insert_exp function, as represented by
7797 the P9V_BUILTIN_VEC_VSIEDP constant.
7798 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7799 exponent support for __ieee128 argument.
7800 (VSESQP): Add scalar extract signature support for __ieee128
7802 (VSTDCNQP): Add scalar test negative support for __ieee128
7804 (VSIEQP): Add scalar insert exponent support for __int128 argument
7805 with __ieee128 result.
7806 (VSIEQPF): Add scalar insert exponent support for __ieee128
7807 argument with __ieee128 result.
7808 (VSTDCQP): Add scalar test data class support for __ieee128
7810 (VSTDCNQP): Add overload support for scalar test negative with
7812 (VSTDCQP): Add overload support for scalar test data class
7814 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7816 (UNSPEC_VSX_SIEXPQP): New constant.
7817 (xsxexpqp): New insn for VSX scalar extract exponent quad
7819 (xsxsigqp): New insn for VSX scalar extract significand quad
7821 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7822 precision with floating point argument.
7823 (xststdcqp): New expand for VSX scalar test data class quad
7825 (xststdcnegqp): New expand for VSX scalar test negative quad
7827 (xststdcqp): New insn to match expansions for VSX scalar test data
7828 class quad precision and VSX scalar test negative quad precision.
7829 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7830 special case operand checking to enforce that second operand of
7831 VSX scalar test data class with quad precision argument is a 7-bit
7833 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7834 prototypes and descriptions of __ieee128 versions of
7835 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7836 scalar_test_data_class, and scalar_test_neg built-in functions.
7838 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7840 PR tree-optimization/81162
7841 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7842 replace a negate with an add.
7844 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7846 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7848 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7850 * config/arm/arm-c.c (arm_cpu_builtins): Define
7851 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7853 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7855 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7856 (armv8-r): Set ARM Cortex-R52 as default CPU.
7857 * config/arm/arm-tables.opt: Regenerate.
7858 * config/arm/arm-tune.md: Regenerate.
7859 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7861 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7862 extension for -mcpu=cortex-r52.
7864 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7866 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7867 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7868 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7869 (fp-armv8): Define it as FP_ARMv8 only.
7870 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7871 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7873 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7874 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7875 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7876 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7877 than TARGET_FPU_ARMV8.
7878 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7879 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7880 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7882 * config/arm/neon.md (neon_vrint): Likewise.
7883 (neon_vcvt): Likewise.
7884 (neon_<fmaxmin_op><mode>): Likewise.
7885 (<fmaxmin><mode>3): Likewise.
7886 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7887 * config/arm/predicates.md (arm_cond_move_operator): Check against
7888 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7890 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7892 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7895 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7897 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7898 loop in comment with memset.
7900 2017-07-14 Martin Liska <mliska@suse.cz>
7902 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7903 * dwarf2out.c (is_java): Remove the function.
7904 (output_pubname): Remove usage of the function.
7905 (lower_bound_default): Remove usage of DW_LANG_Java.
7906 (gen_compile_unit_die): Likewise.
7907 * gcc.c: Remove compiler defaults for .java and .zip files.
7908 * gimple-expr.c (remove_suffix): Change as there's no longer
7909 extension than 4-letter one.
7910 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7911 (gimplify_save_expr): Likewise.
7912 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7913 as it's possible even for other languages than Java.
7914 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7915 * lto-opts.c (lto_write_options): Remove reference to Java.
7916 * opts.c (strip_off_ending): Update file extension handling.
7917 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7918 * tree-eh.c (lower_resx): Likewise.
7919 * tree.c (free_lang_data_in_type): Remove dead code.
7920 (find_decls_types_r): Likewise.
7921 (build_common_builtin_nodes): Remove Java from a comment.
7922 (verify_type): Remove dead code.
7923 * varasm.c (assemble_external): Remove Java from a comment.
7925 2017-07-14 Martin Liska <mliska@suse.cz>
7927 * opts.c (finish_options): Add quotes.
7928 (common_handle_option): Likewise.
7930 2017-07-14 Martin Liska <mliska@suse.cz>
7932 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7933 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7935 * dwarf2out.c (lower_bound_default): Do not handle
7937 (gen_compile_unit_die): Likewise.
7938 * gcc.c: Remove default extension binding for GNU Pascal.
7939 * stmt.c: Remove Pascal language from a comment.
7940 * xcoffout.c: Likewise.
7942 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7945 * diagnostic-show-locus.c (fixit_cmp): New function.
7946 (layout::layout): Sort m_fixit_hints.
7947 (column_range::column_range): Assert that the values are valid.
7948 (struct char_span): New struct.
7949 (correction::overwrite): New method.
7950 (struct source_line): New struct.
7951 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7952 calls in terms of classes source_line and char_span, and
7953 correction::overwrite.
7954 (selftest::test_overlapped_fixit_printing_2): New function.
7955 (selftest::diagnostic_show_locus_c_tests): Call it.
7957 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7959 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7960 early if there is no lhs.
7962 2017-07-13 Martin Liska <mliska@suse.cz>
7964 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7965 (gen_reference_type_die): Likewise.
7966 * stor-layout.c: Remove Pascal-related comment.
7968 2017-07-13 Martin Liska <mliska@suse.cz>
7970 * opts.c (finish_options): Add quotes to error messages.
7971 (parse_sanitizer_options): Likewise.
7973 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
7975 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
7977 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
7979 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
7981 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7983 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
7985 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
7987 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7990 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
7991 provides the hardware capability bits, define the macro
7992 __BUILTIN_CPU_SUPPORTS__.
7993 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
7994 if GLIBC does not provide the hardware capability bits. Add a
7995 gcc_unreachable call if the built-in cpu function is neither
7996 __builtin_cpu_is nor __builtin_cpu_supports.
7997 (rs6000_get_function_versions_dispatcher): Change the warning
7998 that an old GLIBC is used which does not export the capability
7999 bits to be an error.
8000 * doc/extend.texi (target_clones attribute): Document the
8001 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8002 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8003 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8004 the macros defined by GCC if the newer GLIBC is available.
8006 2017-07-12 Jeff Law <law@redhat.com>
8008 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8009 remaining includes slightly.
8010 * config/riscv/riscv-builtins.c: Include profile-count.h.
8012 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8015 * config/avr/avr.c (avr_set_current_function): In diagnostic
8016 messages: Quote keywords and (parts of) identifiers.
8017 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8020 2017-07-12 Carl Love <cel@us.ibm.com>
8022 * config/rs6000/rs6000-c.c: Add support for built-in functions
8023 vector bool char vec_revb (vector bool char);
8024 vector bool short vec_revb (vector short char);
8025 vector bool int vec_revb (vector bool int);
8026 vector bool long long vec_revb (vector bool long long);
8027 * doc/extend.texi: Update the built-in documentation file for the
8028 new built-in functions.
8030 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8032 * config/s390/s390.md: Remove movcc splitter.
8034 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8036 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8037 load/store on condition.
8039 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8042 * config/avr/avr.c (avr_encode_section_info)
8043 [progmem && !TREE_READONLY]: Error if progmem object needs
8046 2017-07-11 Michael Collison <michael.collison@arm.com>
8048 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8051 2017-07-11 Carl Love <cel@us.ibm.com>
8053 * config/rs6000/rs6000-c.c: Add support for builtins
8054 vector unsigned int vec_parity_lsbb (vector signed int);
8055 vector unsigned int vec_parity_lsbb (vector unsigned int);
8056 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8057 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8058 vector unsigned long long vec_parity_lsbb (vector signed long long);
8059 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8060 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8061 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8062 * doc/extend.texi: Update the built-in documentation file for the
8063 new built-in functions.
8065 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8067 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8068 (layout::m_primary_loc): New field.
8069 (layout::layout): Initialize new field. Move location filtering
8070 logic from here to...
8071 (layout::maybe_add_location_range): ...this new method. Add
8072 support for filtering to just the lines already specified by other
8074 (layout::will_show_line_p): New method.
8075 (gcc_rich_location::add_location_if_nearby): New method.
8076 (selftest::test_add_location_if_nearby): New test function.
8077 (selftest::diagnostic_show_locus_c_tests): Call it.
8078 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8081 2017-07-11 Tom de Vries <tom@codesourcery.com>
8083 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8084 (bb_first_real_insn): New function.
8085 (nvptx_single): Add extra initialization of broadcasted condition
8088 2017-07-11 Nathan Sidwell <nathan@acm.org>
8090 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8092 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8094 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8095 Binutils doc as TEXI will mess them up.
8096 * doc/invoke.texi (AVR Options): Same here.
8098 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8100 * config/sparc/sparc.opt (mfix-ut700): New option.
8101 (mfix-gr712rc): Likewise.
8102 (sparc_fix_b2bst): New variable.
8103 * doc/invoke.texi (SPARC options): Document them.
8104 (ARM options): Fix warnings.
8105 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8106 instructions to prevent sequences that can trigger the store-store
8107 errata for certain LEON3FT processors.
8108 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8109 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8110 * config/sparc/sparc.md (fix_b2bst): New attribute.
8111 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8113 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8116 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8118 (*rsqrtsf2_sse): Ditto.
8120 (div<mode>3): Macroize insn from divdf3 and divsf3
8121 using MODEF mode iterator.
8123 2017-07-10 Martin Sebor <msebor@redhat.com>
8125 PR tree-optimization/80397
8126 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8127 instead of testing for equality to INTEGER_TYPE.
8129 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8131 * config.gcc: Remove uclibc from arc target spec.
8133 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8135 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8137 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8140 * lto-wrapper.c (remove_option): New function.
8141 (merge_and_complain): Merge PIC/PIE options more realistically.
8143 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8145 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8149 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8150 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8151 * config.in: Regenerate.
8152 * configure: Regenerate.
8153 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8154 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8155 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8156 (TARGET_GASISR_PROLOGUES): ...target mask.
8157 * common/config/avr/avr-common.c
8158 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8159 Set -mgas-isr-prologues.
8160 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8161 INSERT_PASS_BEFORE for it.
8162 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8163 * config/avr/avr.c (avr_option_override)
8164 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8165 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8166 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8167 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8168 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8170 (make_avr_pass_pre_proep): New function.
8171 (emit_push_sfr) <treg>: Add argument to function and use it
8173 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8174 and set machine->gasisr.yes.
8175 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8176 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8177 __gcc_isr.n_pushed to .L__stack_usage.
8178 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8179 (avr_asm_final_postscan_insn): ...this new static function.
8180 * config/avr/avr.h (machine_function)
8181 <is_no_gccisr, use_L__stack_usage>: New fields.
8182 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8183 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8184 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8185 (gasisr, *gasisr): New expander and insn.
8186 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8187 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8188 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8190 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8192 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8195 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8197 Move jump-tables out of .text again.
8200 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8201 (ASM_OUTPUT_ADDR_VEC): New function.
8202 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8203 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8204 INSN_ADDRESSes as asm comment.
8205 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8206 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8207 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8208 * config/avr/avr.md (*tablejump): Adjust comment.
8209 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8210 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8212 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8213 (avr_output_addr_vec): New proto.
8214 (avr_log_t) <insn_addresses>: New field.
8216 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8219 * config/i386/i386.c (ix86_function_arg_advance): Set
8220 outgoing_args_on_stack to true if there are outgoing arguments
8222 (ix86_function_arg): Likewise.
8223 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8224 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8225 * config/i386/i386.h (machine_function): Add
8226 outgoing_args_on_stack.
8228 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8230 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8232 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8234 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8236 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8237 (REAL_H): Remove $(MACHMODE_H).
8238 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8240 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8241 $(MACHMODE_H) and double-int.h.
8242 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8244 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8247 2017-07-07 Andrew Pinski <apinski@cavium.com>
8249 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8250 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8252 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8254 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8255 Add warning if GCC was not configured to link against a GLIBC that
8256 exports the hardware capability bits.
8257 (make_resolver_func): Make resolver function private and not a
8258 COMDAT function. Create the name with clone_function_name instead
8259 of make_unique_name.
8262 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8263 correct operand in doing the split.
8265 2017-07-07 Carl Love <cel@us.ibm.com>
8267 * config/rs6000/rs6000-c: Add support for built-in function
8268 vector unsigned short vec_pack_to_short_fp32 (vector float,
8270 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8271 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8272 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8273 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8274 (convert_4f32_8i16): Add define_expand.
8275 * doc/extend.texi: Update the built-in documentation file for the
8276 new built-in function.
8278 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8280 * config/sparc/m8.md: New file.
8281 * config/sparc/sparc.md: Include m8.md.
8283 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8285 * config/sparc/sparc.opt: New option -mvis4b.
8286 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8287 (sparc_option_override): Handle VIS4B.
8288 (enum sparc_builtins): Define
8289 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8290 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8291 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8292 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8293 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8294 (check_constant_argument): New function.
8295 (sparc_vis_init_builtins): Define builtins
8296 __builtin_vis_dictunpack{8,16,32},
8297 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8298 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8299 __builtin_vis_fpcmpde{8,16,32}shl and
8300 __builtin_vis_fpcmpur{8,16,32}shl.
8301 (sparc_expand_builtin): Check that the constant operands to
8302 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8303 constant and in range.
8304 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8306 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8307 (SPARC_IMM5_P): Likewise.
8308 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8309 (enabled): Handle vis4b.
8310 (UNSPEC_DICTUNPACK): New unspec.
8311 (UNSPEC_FPCMPSHL): Likewise.
8312 (UNSPEC_FPUCMPSHL): Likewise.
8313 (UNSPEC_FPCMPDESHL): Likewise.
8314 (UNSPEC_FPCMPURSHL): Likewise.
8315 (cpu_feature): New CPU feature `vis4b'.
8316 (dictunpack{8,16,32}): New insns.
8317 (FPCSMODE): New mode iterator.
8318 (fpcscond): New code iterator.
8319 (fpcsucond): Likewise.
8320 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8321 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8322 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8323 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8324 * config/sparc/constraints.md: Define constraints `q' for unsigned
8325 2-bit integer constants and `t' for unsigned 5-bit integer
8327 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8329 (imm5_operand_dictunpack16): Likewise.
8330 (imm5_operand_dictunpack32): Likewise.
8331 (imm2_operand): Likewise.
8332 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8333 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8334 ditunpack* and fpcmp*shl builtins.
8336 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8338 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8339 * config.in: Add HAVE_AS_SPARC6 define.
8340 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8342 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8344 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8345 (CPP_CPU_SPEC): Handle m8.
8346 (ASM_CPU_SPEC): Likewise.
8347 * config/sparc/sparc-opts.h (enum processor_type): Add
8349 * config/sparc/sparc.c (m8_costs): New struct.
8350 (sparc_option_override): Handle TARGET_CPU_m8.
8351 (sparc32_initialize_trampoline): Likewise.
8352 (sparc64_initialize_trampoline): Likewise.
8353 (sparc_issue_rate): Likewise.
8354 (sparc_register_move_cost): Likewise.
8355 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8356 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8357 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8358 (CPP_CPU_SPEC): Handle M8.
8359 (ASM_CPU_SPEC): Likewise.
8360 (AS_M8_FLAG): Define.
8361 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8362 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8363 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8365 * configure: Regenerate.
8366 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8369 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8371 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8373 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8374 ("*movdi_insn_sp32"): Do not set v3pipe.
8375 ("*movsi_insn"): Likewise.
8376 ("*movdi_insn_sp64"): Likewise.
8377 ("*movsf_insn"): Likewise.
8378 ("*movdf_insn_sp32"): Likewise.
8379 ("*movdf_insn_sp64"): Likewise.
8380 ("*zero_extendsidi2_insn_sp64"): Likewise.
8381 ("*sign_extendsidi2_insn"): Likewise.
8382 ("*mov<VM32:mode>_insn"): Likewise.
8383 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8384 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8385 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8386 ("<vlop:code><VL:mode>3"): Likewise.
8387 ("*not_<vlop:code><VL:mode>3"): Likewise.
8388 ("*nand<VL:mode>_vis"): Likewise.
8389 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8390 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8391 ("one_cmpl<VL:mode>2"): Likewise.
8392 ("faligndata<VM64:mode>_vis"): Likewise.
8393 ("alignaddrsi_vis"): Likewise.
8394 ("alignaddrdi_vis"): Likweise.
8395 ("alignaddrlsi_vis"): Likewise.
8396 ("alignaddrldi_vis"): Likewise.
8397 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8398 ("bmaskdi_vis"): Likewise.
8399 ("bmasksi_vis"): Likewise.
8400 ("bshuffle<VM64:mode>_vis"): Likewise.
8401 ("cmask8<P:mode>_vis"): Likewise.
8402 ("cmask16<P:mode>_vis"): Likewise.
8403 ("cmask32<P:mode>_vis"): Likewise.
8404 ("pdistn<P:mode>_vis"): Likewise.
8405 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8407 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8409 * config/sparc/sparc.md ("subtype"): New insn attribute.
8410 ("*wrgsr_sp64"): Set insn subtype.
8411 ("*rdgsr_sp64"): Likewise.
8412 ("alignaddrsi_vis"): Likewise.
8413 ("alignaddrdi_vis"): Likewise.
8414 ("alignaddrlsi_vis"): Likewise.
8415 ("alignaddrldi_vis"): Likewise.
8416 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8417 ("fexpand_vis"): Likewise.
8418 ("fpmerge_vis"): Likewise.
8419 ("faligndata<VM64:mode>_vis"): Likewise.
8420 ("bshuffle<VM64:mode>_vis"): Likewise.
8421 ("cmask8<P:mode>_vis"): Likewise.
8422 ("cmask16<P:mode>_vis"): Likewise.
8423 ("cmask32<P:mode>_vis"): Likewise.
8424 ("fchksm16_vis"): Likewise.
8425 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8426 ("fmean16_vis"): Likewise.
8427 ("fp<plusminus_insn>64_vis"): Likewise.
8428 ("<plusminus_insn>v8qi3"): Likewise.
8429 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8430 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8431 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8432 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8433 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8434 ("*movqi_insn"): Likewise.
8435 ("*movhi_insn"): Likewise.
8436 ("*movsi_insn"): Likewise.
8437 ("movsi_pic_gotdata_op"): Likewise.
8438 ("*movdi_insn_sp32"): Likewise.
8439 ("*movdi_insn_sp64"): Likewise.
8440 ("movdi_pic_gotdata_op"): Likewise.
8441 ("*movsf_insn"): Likewise.
8442 ("*movdf_insn_sp32"): Likewise.
8443 ("*movdf_insn_sp64"): Likewise.
8444 ("*zero_extendhisi2_insn"): Likewise.
8445 ("*zero_extendqihi2_insn"): Likewise.
8446 ("*zero_extendqisi2_insn"): Likewise.
8447 ("*zero_extendqidi2_insn"): Likewise.
8448 ("*zero_extendhidi2_insn"): Likewise.
8449 ("*zero_extendsidi2_insn_sp64"): Likewise.
8450 ("ldfsr"): Likewise.
8451 ("prefetch_64"): Likewise.
8452 ("prefetch_32"): Likewise.
8453 ("tie_ld32"): Likewise.
8454 ("tie_ld64"): Likewise.
8455 ("*tldo_ldub_sp32"): Likewise.
8456 ("*tldo_ldub1_sp32"): Likewise.
8457 ("*tldo_ldub2_sp32"): Likewise.
8458 ("*tldo_ldub_sp64"): Likewise.
8459 ("*tldo_ldub1_sp64"): Likewise.
8460 ("*tldo_ldub2_sp64"): Likewise.
8461 ("*tldo_ldub3_sp64"): Likewise.
8462 ("*tldo_lduh_sp32"): Likewise.
8463 ("*tldo_lduh1_sp32"): Likewise.
8464 ("*tldo_lduh_sp64"): Likewise.
8465 ("*tldo_lduh1_sp64"): Likewise.
8466 ("*tldo_lduh2_sp64"): Likewise.
8467 ("*tldo_lduw_sp32"): Likewise.
8468 ("*tldo_lduw_sp64"): Likewise.
8469 ("*tldo_lduw1_sp64"): Likewise.
8470 ("*tldo_ldx_sp64"): Likewise.
8471 ("*mov<VM32:mode>_insn"): Likewise.
8472 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8473 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8475 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8477 * config/sparc/sparc.md ("type"): New insn type viscmp.
8478 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8480 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8481 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8482 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8483 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8485 ("n7_vis_logical_11cycle"): Likewise.
8486 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8487 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8488 * config/sparc/niagara.md ("niag_vis"): Likewise.
8489 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8490 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8492 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8494 * config/sparc/sparc.md: New instruction type `bmask'.
8495 (bmaskdi_vis): Use the `bmask' type.
8496 (bmasksi_vis): Likewise.
8497 * config/sparc/ultra3.md (us3_array): Likewise.
8498 * config/sparc/niagara7.md (n7_array): Likewise.
8499 * config/sparc/niagara4.md (n4_array): Likewise.
8500 * config/sparc/niagara2.md (niag2_vis): Likewise.
8501 (niag3_vis): Likewise.
8502 * config/sparc/niagara.md (niag_vis): Likewise.
8504 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8506 * ipa-comdats.c: Remove optimize check from gate.
8507 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8508 for functions not optimized.
8509 (ipa_fn_summary_read): Skip optimize check.
8510 (ipa_fn_summary_write): Likewise.
8511 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8513 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8515 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8516 (check_callers): Give up on caller which is not optimized.
8517 (inline_small_functions): Likewise.
8518 (ipa_inline): Do not give up when not optimizing.
8519 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8520 away unoptimizes cdtors.
8521 (whole_program_function_and_variable_visibility): Do
8522 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8523 * ipa.c (process_references): Do not check optimize.
8524 (symbol_table::remove_unreachable_nodes): Update optimize check.
8525 (set_writeonly_bit): Update optimize check.
8526 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8527 (pass_ipa_single_use::gate): Remove.
8529 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8531 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8532 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8533 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8534 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8535 permute_load, permute_store, adjust_extract, adjust_splat,
8536 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8537 replace_swap_with_copy, dump_swap_insn_table,
8538 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8539 recombine_lvx_pattern, recombine_stvx_pattern,
8540 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8541 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8542 to file rs6000-p8swap.c.
8543 * config/rs6000/rs6000-p8swap.c: New file.
8544 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8545 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8546 and rs6000*-*-* targets.
8548 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8550 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8552 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8554 * lto-wrapper.c (merge_and_complain): Do not merge
8555 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8556 fsigned_zeros, ftrapping_math, fwrapv.
8557 (append_compiler_options): Do not track these options.
8558 (append_linker_options): Likewie
8560 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8562 * cgraphunit.c (cgraph_node::finalize_function): When
8563 !flag_toplevel_reorde set no_reorder flag.
8564 (varpool_node::finalize_decl): Likewise.
8565 (symbol_table::compile): Drop no toplevel reorder path.
8567 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8569 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8570 edges; zero probability is not better than uninitialized.
8572 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8574 * asan.h (asan_sanitize_allocas_p): Declare.
8575 * asan.c (asan_sanitize_allocas_p): New function.
8576 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8577 (handle_builtin_alloca): Likewise.
8578 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8579 if !asan_sanitize_allocas_p.
8580 * params.def (asan-instrument-allocas): Add new option.
8581 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8582 * opts.c (common_handle_option): Disable allocas sanitization for
8585 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8587 * asan.c: Include gimple-fold.h.
8588 (get_last_alloca_addr): New function.
8589 (handle_builtin_stackrestore): Likewise.
8590 (handle_builtin_alloca): Likewise.
8591 (asan_emit_allocas_unpoison): Likewise.
8592 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8593 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8594 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8595 (instrument_builtin_call): Pass gimple iterator to
8596 get_mem_refs_of_builtin_call.
8597 (last_alloca_addr): New global.
8598 * asan.h (asan_emit_allocas_unpoison): Declare.
8599 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8600 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8601 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8602 if function calls alloca.
8603 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8604 * gimple-fold.h (replace_call_with_value): Declare.
8605 * internal-fn.c: Include asan.h.
8606 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8607 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8609 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8611 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8612 (C_SELFTEST_FLAGS): New.
8613 (CPP_SELFTEST_FLAGS): New.
8614 (SELFTEST_DEPS): New, from deps of s-selftest.
8615 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8616 (CPP_SELFTEST_DEPS): New.
8617 (selftest): Add dependency on s-selftest-c++.
8618 (s-selftest): Rename to...
8619 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8620 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8621 than SELFTEST_FLAGS.
8622 (selftest-gdb): Rename to...
8623 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8625 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8626 (selftest-valgrind): Rename to...
8627 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8629 (selftest-valgrind): Reintroduce as an alias for
8630 selftest-c-valgrind.
8631 (s-selftest-c++): New.
8632 (selftest-c++-gdb): New.
8633 (selftest-c++-valgrind): New.
8635 2017-07-06 Olivier Hainque <hainque@adacore.com>
8637 * gcc.c (process_command): When deciding if undefined variables
8638 should be ignored when processing specs, accept "gcc -v" as well.
8640 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8642 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8643 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8645 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8647 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8648 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8649 * config/arm/arm-tables.opt: Regenerate.
8650 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8652 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8654 2017-07-06 Carl Love <cel@us.ibm.com>
8656 * ChangeLog: Clean up from mid air collision
8658 2017-07-06 Carl Love <cel@us.ibm.com>
8660 * config/rs6000/rs6000-c.c: Add support for built-in functions
8661 vector signed int vec_subc (vector signed int, vector signed int);
8662 vector signed __int128 vec_subc (vector signed __int128,
8663 vector signed __int128);
8664 vector unsigned __int128 vec_subc (vector unsigned __int128,
8665 vector unsigned __int128);
8666 vector signed int vec_sube (vector signed int, vector signed int,
8668 vector unsigned int vec_sube (vector unsigned int,
8669 vector unsigned int,
8670 vector unsigned int);
8671 vector signed __int128 vec_sube (vector signed __int128,
8672 vector signed __int128,
8673 vector signed__int128);
8674 vector unsigned __int128 vec_sube (vector unsigned __int128,
8675 vector unsigned __int128,
8676 vector unsigned __int128);
8677 vector signed int vec_subec (vector signed int, vector signed int,
8679 vector unsigned int vec_subec (vector unsigned int,
8680 vector unsigned int,
8681 vector unsigned int);
8682 vector signed __int128 vec_subec (vector signed __int128,
8683 vector signed __int128,
8684 vector signed__int128);
8685 vector unsigned __int128 vec_subec (vector unsigned __int128,
8686 vector unsigned __int128,
8687 vector unsigned __int128);
8688 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8689 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8690 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8691 BU_ALTIVEC_OVERLOAD_X definitions.
8692 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8693 * doc/extend.texi: Update the built-in documentation file for the new
8696 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8699 * diagnostic-show-locus.c (layout::layout): Use start and finish
8700 spelling location for the start and finish of each range.
8701 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8702 Add unused aspect param.
8703 * input.c (expand_location_1): Add "aspect" param, and use it
8704 to access the correct part of the location.
8705 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8707 (expand_location_to_spelling_point): Likewise.
8708 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8709 param, and pass it to expand_location_1.
8711 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8713 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8714 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8715 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8716 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8717 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8718 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8719 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8720 _mm_maskz_getmant_ss): New intrinsics.
8721 (__builtin_ia32_getexpss128_mask): Changed to ...
8722 __builtin_ia32_getexpss128_round ... this.
8723 (__builtin_ia32_getexpsd128_mask): Changed to ...
8724 __builtin_ia32_getexpsd128_round ... this.
8725 * config/i386/i386-builtin-types.def
8726 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8727 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8728 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8729 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8730 __builtin_ia32_getmantss_mask_round): New builtins.
8731 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8732 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8733 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8734 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8735 * config/i386/sse.md
8736 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8737 avx512f_sgetexp<mode><mask_scalar_name>
8738 <round_saeonly_scalar_name> ... this.
8739 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8740 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8741 vgetexp<ssescalarmodesuffix>
8742 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8743 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8744 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8745 avx512f_vgetmant<mode><mask_scalar_name>
8746 <round_saeonly_scalar_name> ... this.
8747 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8748 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8749 vgetmant<ssescalarmodesuffix>
8750 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8751 %0<mask_scalar_operand4>, %1, %2
8752 <round_saeonly_scalar_mask_op4>, %3} ... this.
8753 * config/i386/subst.md (mask_scalar_operand4,
8754 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8755 round_saeonly_scalar_nimm_predicate): New subst attributes.
8757 2017-07-06 Julia Koval <julia.koval@intel.com>
8759 * config/i386/i386.c (ix86_erase_embedded_rounding):
8760 Remove code for old rounding pattern.
8762 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8764 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8766 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8768 * doc/sourcebuild.texi (Test Directives, Variants of
8769 dg-require-support): Add documentation for dg-require-stack-check.
8771 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8773 * config/i386/subst.md (mask_scalar, round_scalar,
8774 round_saeonly_scalar): New meta-templates.
8775 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8776 round_scalar_mask_operand3, round_scalar_mask_op3,
8777 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8778 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8779 round_saeonly_scalar_constraint,
8780 round_saeonly_scalar_prefix): New subst attribute.
8781 * config/i386/sse.md
8782 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8783 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8784 <round_scalar_name> ... this.
8785 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8786 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8787 <round_scalar_name> ... this.
8788 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8789 <sse>_vm<code><mode>3<mask_scalar_name>
8790 <round_saeonly_scalar_name> ... this.
8791 (v<plusminus_mnemonic><ssescalarmodesuffix>
8792 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8793 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8794 v<plusminus_mnemonic><ssescalarmodesuffix>
8795 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8796 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8797 (v<multdiv_mnemonic><ssescalarmodesuffix>
8798 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8799 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8800 v<multdiv_mnemonic><ssescalarmodesuffix>
8801 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8802 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8803 (v<maxmin_float><ssescalarmodesuffix>
8804 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8805 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8806 v<maxmin_float><ssescalarmodesuffix>
8807 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8808 %0<mask_scalar_operand3>, %1, %<iptr>2
8809 <round_saeonly_scalar_mask_op3>} ... this.
8811 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8813 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8814 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8816 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8817 Alan Hayward <alan.hayward@arm.com>
8818 David Sherwood <david.sherwood@arm.com>
8820 * combine.c (simplify_if_then_else): Remove "enum" before
8822 * compare-elim.c (can_eliminate_compare): Likewise.
8823 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8825 (aarch64_lookup_simd_builtin_type): Likewise.
8826 (aarch64_simd_builtin_type): Likewise.
8827 (aarch64_init_simd_builtin_types): Likewise.
8828 (aarch64_simd_expand_args): Likewise.
8829 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8831 (aarch64_reverse_mask): Likewise.
8832 (aarch64_simd_emit_reg_reg_move): Likewise.
8833 (aarch64_gen_adjusted_ldpstp): Likewise.
8834 (aarch64_ccmp_mode_to_code): Likewise.
8835 (aarch64_operands_ok_for_ldpstp): Likewise.
8836 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8837 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8839 (aarch64_min_divisions_for_recip_mul): Likewise.
8840 (aarch64_reassociation_width): Likewise.
8841 (aarch64_get_condition_code_1): Likewise.
8842 (aarch64_simd_emit_reg_reg_move): Likewise.
8843 (aarch64_simd_attr_length_rglist): Likewise.
8844 (aarch64_reverse_mask): Likewise.
8845 (aarch64_operands_ok_for_ldpstp): Likewise.
8846 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8847 (aarch64_gen_adjusted_ldpstp): Likewise.
8848 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8850 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8851 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8852 (arm_lookup_simd_builtin_type): Likewise.
8853 (arm_simd_builtin_type): Likewise.
8854 (arm_init_simd_builtin_types): Likewise.
8855 (arm_expand_builtin_args): Likewise.
8856 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8857 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8858 (ft32_setup_incoming_varargs): Likewise.
8859 (ft32_function_arg): Likewise.
8860 (ft32_function_arg_advance): Likewise.
8861 (ft32_pass_by_reference): Likewise.
8862 (ft32_arg_partial_bytes): Likewise.
8863 (ft32_valid_pointer_mode): Likewise.
8864 (ft32_addr_space_pointer_mode): Likewise.
8865 (ft32_addr_space_legitimate_address_p): Likewise.
8866 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8868 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8869 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8870 (iamcu_alignment): Likewise.
8871 (canonicalize_vector_int_perm): Likewise.
8872 (ix86_noce_conversion_profitable_p): Likewise.
8873 (ix86_mpx_bound_mode): Likewise.
8874 (ix86_operands_ok_for_move_multiple): Likewise.
8875 * config/microblaze/microblaze-protos.h
8876 (microblaze_expand_conditional_branch_reg): Likewise.
8877 * config/microblaze/microblaze.c
8878 (microblaze_expand_conditional_branch_reg): Likewise.
8879 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8881 (rs6000_reassociation_width): Likewise.
8882 (rs6000_invalid_binary_op): Likewise.
8883 (fusion_p9_p): Likewise.
8884 (emit_fusion_p9_load): Likewise.
8885 (emit_fusion_p9_store): Likewise.
8886 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8888 (riscv_hard_regno_mode_ok_p): Likewise.
8889 (riscv_address_insns): Likewise.
8890 (riscv_split_symbol): Likewise.
8891 (riscv_legitimize_move): Likewise.
8892 (riscv_function_value): Likewise.
8893 (riscv_hard_regno_nregs): Likewise.
8894 (riscv_expand_builtin): Likewise.
8895 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8896 (riscv_build_integer): Likewise.
8897 (riscv_split_integer): Likewise.
8898 (riscv_legitimate_constant_p): Likewise.
8899 (riscv_cannot_force_const_mem): Likewise.
8900 (riscv_regno_mode_ok_for_base_p): Likewise.
8901 (riscv_valid_base_register_p): Likewise.
8902 (riscv_valid_offset_p): Likewise.
8903 (riscv_valid_lo_sum_p): Likewise.
8904 (riscv_classify_address): Likewise.
8905 (riscv_legitimate_address_p): Likewise.
8906 (riscv_address_insns): Likewise.
8907 (riscv_load_store_insns): Likewise.
8908 (riscv_force_binary): Likewise.
8909 (riscv_split_symbol): Likewise.
8910 (riscv_force_address): Likewise.
8911 (riscv_legitimize_address): Likewise.
8912 (riscv_move_integer): Likewise.
8913 (riscv_legitimize_const_move): Likewise.
8914 (riscv_legitimize_move): Likewise.
8915 (riscv_address_cost): Likewise.
8916 (riscv_subword): Likewise.
8917 (riscv_output_move): Likewise.
8918 (riscv_canonicalize_int_order_test): Likewise.
8919 (riscv_emit_int_order_test): Likewise.
8920 (riscv_function_arg_boundary): Likewise.
8921 (riscv_pass_mode_in_fpr_p): Likewise.
8922 (riscv_pass_fpr_single): Likewise.
8923 (riscv_pass_fpr_pair): Likewise.
8924 (riscv_get_arg_info): Likewise.
8925 (riscv_function_arg): Likewise.
8926 (riscv_function_arg_advance): Likewise.
8927 (riscv_arg_partial_bytes): Likewise.
8928 (riscv_function_value): Likewise.
8929 (riscv_pass_by_reference): Likewise.
8930 (riscv_setup_incoming_varargs): Likewise.
8931 (riscv_print_operand): Likewise.
8932 (riscv_elf_select_rtx_section): Likewise.
8933 (riscv_save_restore_reg): Likewise.
8934 (riscv_for_each_saved_reg): Likewise.
8935 (riscv_register_move_cost): Likewise.
8936 (riscv_hard_regno_mode_ok_p): Likewise.
8937 (riscv_hard_regno_nregs): Likewise.
8938 (riscv_class_max_nregs): Likewise.
8939 (riscv_memory_move_cost): Likewise.
8940 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8941 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8942 (rl78_addr_space_address_mode): Likewise.
8943 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8945 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8946 (rs6000_reassociation_width): Likewise.
8947 (rs6000_invalid_binary_op): Likewise.
8948 (fusion_p9_p): Likewise.
8949 (emit_fusion_p9_load): Likewise.
8950 (emit_fusion_p9_store): Likewise.
8951 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8952 (ok_for_simple_move_operands): Likewise.
8953 (ok_for_simple_move_strict_operands): Likewise.
8954 (ok_for_simple_arith_logic_operands): Likewise.
8955 (visium_legitimize_reload_address): Likewise.
8956 (visium_select_cc_mode): Likewise.
8957 (output_cbranch): Likewise.
8958 (visium_split_double_move): Likewise.
8959 (visium_expand_copysign): Likewise.
8960 (visium_expand_int_cstore): Likewise.
8961 (visium_expand_fp_cstore): Likewise.
8962 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8963 (visium_function_arg): Likewise.
8964 (visium_function_arg_advance): Likewise.
8965 (visium_libcall_value): Likewise.
8966 (visium_setup_incoming_varargs): Likewise.
8967 (visium_legitimate_constant_p): Likewise.
8968 (visium_legitimate_address_p): Likewise.
8969 (visium_legitimize_address): Likewise.
8970 (visium_secondary_reload): Likewise.
8971 (visium_register_move_cost): Likewise.
8972 (visium_memory_move_cost): Likewise.
8973 (prepare_move_operands): Likewise.
8974 (ok_for_simple_move_operands): Likewise.
8975 (ok_for_simple_move_strict_operands): Likewise.
8976 (ok_for_simple_arith_logic_operands): Likewise.
8977 (visium_function_value_1): Likewise.
8978 (rtx_ok_for_offset_p): Likewise.
8979 (visium_legitimize_reload_address): Likewise.
8980 (visium_split_double_move): Likewise.
8981 (visium_expand_copysign): Likewise.
8982 (visium_expand_int_cstore): Likewise.
8983 (visium_expand_fp_cstore): Likewise.
8984 (visium_split_cstore): Likewise.
8985 (visium_select_cc_mode): Likewise.
8986 (visium_split_cbranch): Likewise.
8987 (output_cbranch): Likewise.
8988 (visium_print_operand_address): Likewise.
8989 * expmed.c (flip_storage_order): Likewise.
8990 * expmed.h (emit_cstore): Likewise.
8991 (flip_storage_order): Likewise.
8992 * genrecog.c (validate_pattern): Likewise.
8993 * hsa-gen.c (gen_hsa_addr): Likewise.
8994 * internal-fn.c (expand_arith_overflow): Likewise.
8995 * ira-color.c (allocno_copy_cost_saving): Likewise.
8996 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8997 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
8998 (process_invariant_for_inheritance): Likewise.
8999 * lra-eliminations.c (move_plus_up): Likewise.
9000 * omp-low.c (lower_oacc_reductions): Likewise.
9001 * simplify-rtx.c (simplify_subreg): Likewise.
9002 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9003 (TARGET_CHKP_BOUND_MODE): Likewise..
9004 * targhooks.c (default_chkp_bound_mode): Likewise.
9005 (default_setup_incoming_vararg_bounds): Likewise.
9006 * targhooks.h (default_chkp_bound_mode): Likewise.
9007 (default_setup_incoming_vararg_bounds): Likewise.
9008 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9009 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9010 (have_whole_vector_shift): Likewise.
9011 * tree-vect-stmts.c (vectorizable_load): Likewise.
9012 * doc/tm.texi: Regenerate.
9014 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9016 Graceful degrade if Binutils PR21472 is not available.
9019 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9020 .rodata in flash test fails.
9021 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9022 * confgure: Regenerate.
9023 * config.in: Regenerate.
9024 * config/avr/avr.c (avr_asm_named_section)
9025 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9026 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9027 (avr_asm_init_sections): Same.
9029 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9031 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9032 (fma<VH:mode>4_intrinsic): Likewise.
9033 (*fmsub<VCVTF:mode>4): Likewise.
9034 (*fmsub<VH:mode>4_intrinsic): Likewise.
9036 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9039 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9040 Don't depend on "optimize > 0".
9041 (out_movhi_r_mr, out_movqi_mr_r): Same.
9042 (out_movhi_mr_r, out_movqi_r_mr): Same.
9043 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9044 io_address_operand on "optimize > 0".
9046 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9048 * tree-loop-distribution.c: Add general explanantion on the pass.
9049 (generate_loops_for_partition): Mark distributed loop.
9050 (pg_add_dependence_edges): New parameter. Handle alias data
9051 dependence specially and record it in the parameter if asked.
9052 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9053 (init_partition_graph_vertices, add_partition_graph_edge): New.
9054 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9055 (free_partition_graph_vdata, build_partition_graph): New.
9056 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9057 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9058 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9059 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9060 (version_for_distribution_p, finalize_partitions): New.
9061 (distribute_loop): Handle alias data dependence specially. Factor
9062 out loop fusion code as functions and call these functions.
9064 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9066 * tree-loop-distribution.c (classify_partition): New parameter and
9067 better handle reduction statement.
9068 (rdg_build_partitions): Revise comment.
9069 (distribute_loop): Compute statements in all partitions and pass it
9070 to classify_partition.
9072 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9074 * tree-loop-distribution.c (enum partition_type): New.
9075 (struct partition): New field type.
9076 (partition_merge_into): Add parameter. Update partition type.
9077 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9078 (build_rdg_partition_for_vertex): Compute partition type.
9079 (rdg_build_partitions): Dump partition type.
9080 (distribute_loop): Update calls to partition_merge_into.
9082 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9084 * tree-loop-distribution.c (struct ddr_hasher): New.
9085 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9087 (classify_partition): Call get_data_dependence.
9088 (pg_add_dependence_edges): Ditto.
9089 (distribute_loop): Release data dependence hash table.
9091 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9093 * tree-loop-distribution.c (ref_base_address): Delete.
9094 (similar_memory_accesses): Rename ...
9095 (share_memory_accesses): ... to this. Check if partitions access
9096 the same memory reference.
9097 (distribute_loop): Call share_memory_accesses.
9099 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9101 * tree-loop-distribution.c (struct partition): New field recording
9103 (partition_alloc, partition_free): Init and release data refs.
9104 (partition_merge_into): Merge data refs.
9105 (build_rdg_partition_for_vertex): Collect data refs for partition.
9106 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9108 (distribute_loop): Remve data refs from vertice data of partition
9111 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9113 * tree-loop-distribution.c (params.h): Include header file.
9114 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9115 (datarefs_vec): New global var.
9116 (create_rdg_vertices): Use datarefs_vec directly.
9117 (free_rdg): Don't free data references.
9118 (build_rdg): Update use. Don't free data references.
9119 (distribute_loop): Compute global variable for data references.
9120 Bail out if there are too many data references.
9122 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9124 * tree-loop-distribution.c (loop_nest): New global var.
9125 (build_rdg): Use loop directly, rather than loop nest.
9126 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9128 (distribute_loop): Compute global variable loop nest. Update use.
9130 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9132 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9133 (partition_merge_into): New parameter. Dump reason for fusion.
9134 (distribute_loop): Update use of partition_merge_into.
9136 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9138 * tree-loop-distribution.c (bb_top_order_index): New.
9139 (bb_top_order_index_size, bb_top_order_cmp): New.
9140 (stmts_from_loop): Use topological order.
9141 (pass_loop_distribution::execute): Compute and release topological
9142 order for basic blocks.
9144 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9146 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9149 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9151 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9152 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9153 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9154 * internal-fn.def (LOOP_DIST_ALIAS): New.
9155 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9156 (fold_loop_internal_call): ... this.
9157 (vect_loop_dist_alias_call): New function.
9158 (set_uid_loop_bbs): Call fold_loop_internal_call.
9159 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9162 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9165 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9166 Require dead FLAGS_REG at the beginning of a peephole.
9168 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9171 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9172 arguments in the call to __builtin_ia32_sbb_u32.
9173 (_subborrow_u64): Swap _X and _Y arguments in the call to
9174 __builtin_ia32_sbb_u64.
9176 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9179 * tree-vrp.c (compare_assert_loc): Turn into a function template
9180 with stable template parameter. Only test if a->e is NULL,
9181 !a->e == !b->e has been verified already. Use e == NULL or
9182 e != NULL instead of e or ! e tests. If stable is true, don't use
9183 iterative_hash_expr, on the other side allow a or b or both NULL
9184 and sort the NULLs last.
9185 (process_assert_insertions): Sort using compare_assert_loc<false>
9186 instead of compare_assert_loc, later sort using
9187 compare_assert_loc<true> before calling process_assert_insertions_for
9188 in a loop. Use break instead of continue once seen NULL pointer.
9190 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9192 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9193 Cortex-R7 and Cortex-R8 processors.
9195 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9197 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9198 uninitialized while src is not.
9200 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9202 * common/config/arm/arm-common.c: Adjust include path for
9204 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9205 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9206 (arm-cpu-data.h): Likewise.
9207 (arm-cpu-cdata.h): Likewise.
9208 * config/arm/arm-cpu.h: Delete.
9209 * config/arm/arm-cpu-cdata.h: Delete.
9210 * config/arm/arm-cpu-data.h: Delete.
9212 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9214 * config/arm/arm-cpus.in (cortex-a55): New.
9215 (cortex-a75): Likewise.
9216 (cortex-a75.cortex-a55): Likewise.
9217 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9219 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9220 * config/arm/arm-cpu-cdata.h: Regenerate.
9221 * config/arm/arm-cpu-data.h: Regenerate.
9222 * config/arm/arm-cpu.h: Regenerate.
9223 * config/arm/arm-tables.opt: Regenerate.
9224 * config/arm/arm-tune.md: Regenerate.
9226 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9228 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9230 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9232 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9235 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9237 PR tree-optimization/81292
9238 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9239 full_string_p, also call adjust_related_strinfos if the adjustment
9240 is simple, otherwise invalidate related strinfos.
9242 2017-07-04 Martin Liska <mliska@suse.cz>
9245 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9246 newly created variable as DECL_IGNORED_P.
9248 2017-07-04 Martin Liska <mliska@suse.cz>
9251 * ipa-inline.c (inline_small_functions):
9252 Use xstrdup_for_dump.
9254 2017-07-04 Tom de Vries <tom@codesourcery.com>
9256 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9258 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9261 * config/darwin.c (darwin_function_switched_text_sections):
9264 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9266 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9268 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9270 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9272 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9274 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9275 min_profitable_iters, and th as inclusive lower bounds.
9276 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9277 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9278 for min_profitable_iters and min_profitable_estimate.
9279 (vect_transform_loop): Treat th as an inclusive lower bound.
9280 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9282 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9285 * config/darwin.c (darwin_function_switched_text_sections):
9286 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9287 in two pieces, and suppress the use of buf.
9289 2017-07-03 Nathan Sidwell <nathan@acm.org>
9291 * hash-table.h (hash_table_mod1): Fix indentation.
9293 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9296 * predict.c (force_edge_cold): Be more careful about propagation
9298 * profile-count.h (profile_probability::guessed,
9299 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9301 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9303 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9305 * doc/invoke.texi (rcpc architecture extension): Document it.
9307 2017-07-03 Richard Biener <rguenther@suse.de>
9309 PR tree-optimization/60510
9310 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9311 the scalar reduction PHI and use it.
9312 (vectorizable_reduction): Properly guard the single_defuse_cycle
9313 path for non-SLP reduction chains where we cannot use it.
9314 Rework reduc_def/index and vector type deduction. Rework
9315 vector operand gathering during reduction op code-gen.
9316 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9317 chains dissolve the chain and leave it to non-SLP reduction
9320 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9322 * tree-data-ref.h (dr_alignment): Declare.
9323 * tree-data-ref.c (dr_alignment): New function.
9324 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9325 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9327 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9329 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9331 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9332 and base_misalignment fields.
9333 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9334 * tree-data-ref.c: Include builtins.h.
9335 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9336 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9337 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9338 * tree-vect-data-refs.c: Include tree-cfg.h.
9339 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9340 fields instead of calculating an alignment here.
9341 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9342 innermost_loop_behavior fields.
9344 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9346 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9348 (DR_STEP_ALIGNMENT): New macro.
9349 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9350 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9351 (create_data_ref): Print it.
9352 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9353 to tell whether the step preserves vector (mis)alignment.
9354 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9355 Move the check for an integer step and generalise to all INTEGER_CST.
9356 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9357 Print the outer step alignment.
9359 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9361 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9362 with offset_alignment.
9363 (DR_ALIGNED_TO): Delete.
9364 (DR_OFFSET_ALIGNMENT): New macro.
9365 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9366 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9367 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9368 (create_data_ref): Likewise.
9369 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9370 (vect_analyze_data_refs): Likewise.
9371 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9372 creating dummy innermost behavior.
9374 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9376 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9377 with a "innermost_loop_behavior *" and refeence tree.
9378 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9379 (create_data_ref): Update call accordingly.
9380 * tree-predcom.c (find_looparound_phi): Likewise.
9382 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9384 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9385 fields with dr_wrt_vec_loop.
9386 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9387 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9388 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9389 (vect_dr_behavior): New function.
9390 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9391 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9392 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9393 track whether the step preserves the misalignment.
9394 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9395 Use vect_dr_behavior.
9396 (vect_setup_realignment): Update call accordingly.
9397 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9398 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9399 call to vect_create_addr_base_for_vector_ref.
9400 (vect_create_cond_for_align_checks): Likewise.
9401 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9402 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9403 (vect_recog_mask_conversion_pattern): Likewise.
9404 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9405 (new_stmt_vec_info): Remove redundant zeroing.
9407 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9409 * common/config/arm/arm-common.c (arm_be8_option): New function.
9410 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9411 (ISA_ARMv6): Add isa_bit_be8.
9412 * config/arm/arm.h (arm_be8_option): Add prototype.
9413 (BE8_SPEC_FUNCTION): New define.
9414 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9415 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9416 (mlittle-endian): Similarly.
9417 (mbe8, mbe32): New options.
9418 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9419 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9421 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9423 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9425 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9427 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9428 (cleanup_tree_cfg_bb): Use it.
9429 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9431 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9433 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9436 * loop-doloop.c (add_test): Update profile.
9438 2017-07-03 Martin Liska <mliska@suse.cz>
9441 * sanopt.c (rewrite_usage_of_param): New function.
9442 (sanitize_rewrite_addressable_params): Likewise.
9443 (pass_sanopt::execute): Call rewrite_usage_of_param.
9445 2017-07-03 Richard Biener <rguenther@suse.de>
9447 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9448 back to using VIEW_CONVERT_EXPR.
9450 2017-07-03 Martin Liska <mliska@suse.cz>
9453 * doc/extend.texi: Document when a resolver function is
9454 generated for target_clones.
9456 2017-07-03 Martin Liska <mliska@suse.cz>
9458 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9459 and shadow memory of auto variables which are subject of
9460 use-after-scope sanitization.
9461 (asan_expand_mark_ifn): Add do set only when is_poison.
9463 2016-07-03 Richard Biener <rguenther@suse.de>
9465 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9467 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9468 (vectorizable_reduction): Perform reduction PHI creation when
9469 visiting a reduction PHI and adjust and simplify code generation
9470 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9471 (vect_transform_loop): Visit reduction PHIs.
9472 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9473 defs into the SLP tree.
9474 (vect_build_slp_tree): Reduction defs terminate the recursion.
9475 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9477 (vect_get_vec_defs_for_stmt_copy): Export.
9478 (vect_get_vec_defs): Likewise.
9479 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9481 (vect_get_vec_defs_for_stmt_copy): Declare.
9482 (vect_get_vec_defs): Likewise.
9484 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9486 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9487 parameter with a "loop" parameter and use it instead of the
9488 loop containing DR_STMT. Don't check simple_iv when doing
9489 BB analysis. Describe the two analysis modes in the comment.
9491 2017-07-03 Tom de Vries <tom@codesourcery.com>
9493 PR tree-optimization/69468
9494 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9495 (find_same_succ_bb): Handle ignore_edge_flags.
9497 2017-07-03 Tom de Vries <tom@codesourcery.com>
9499 PR tree-optimization/81192
9500 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9502 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9504 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9506 2017-07-03 Tom de Vries <tom@codesourcery.com>
9508 PR tree-optimization/81192
9509 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9510 BB_SAME_SUCC (bb) == NULL.
9512 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9514 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9517 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9519 * dumpfile.c: Include profile-count.h
9520 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9522 (insert_cond_bb): Update profile.
9523 * tree-cfg.h (insert_cond_bb): Update prototype.
9524 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9525 * tree-dump.c: Do not include tree-cfg.
9527 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9529 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9531 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9533 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9536 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9538 * tree-complex.c (expand_complex_div_wide): update profile.
9540 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9541 Alan Hayward <alan.hayward@arm.com>
9542 David Sherwood <david.sherwood@arm.com>
9544 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9545 (CORETYPES_H): New define.
9546 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9547 (insn-modes-inline.h, s-modes-inline-h): New rules.
9548 (generated_files): Add insn-modes-inline.h.
9549 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9550 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9551 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9552 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9553 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9554 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9555 (build/gencodes.o, build/genconditions.o): Likewise.
9556 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9557 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9558 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9559 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9560 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9561 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9562 * coretypes.h: Include everything up to real.h for generators.
9563 Include insn-modes.h first. Include wide-int-print.h after
9564 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9565 * machmode.h: Don't include insn-modes.h here.
9566 * function-tests.c: Remove includes of signop.h, machmode.h,
9567 double-int.h and wide-int.h.
9569 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9571 * optc-save-gen.awk: Likewise.
9572 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9573 * godump.c: Remove include of wide-int-print.h.
9574 * pretty-print.h: Likewise.
9575 * wide-int-print.cc: Likewise.
9576 * wide-int.cc: Likewise.
9577 * hash-map-tests.c: Remove include of signop.h.
9578 * hash-set-tests.c: Likewise.
9579 * rtl-tests.c: Likewise.
9580 * mkconfig.sh: Remove include of machmode.h.
9581 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9583 (emit_insn_modes_inline_h): ...this new function. Emit the code
9584 into an insn-modes-inline.h header file, adding appropriate
9585 include guards and end comments.
9586 (emit_insn_modes_c_header): Remove include of machmode.h.
9587 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9589 (main): Handle -i flag and call emit_insn_modes_inline_h when
9592 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9594 * tree-ssa-strlen.c (strinfo): Rename the length field to
9595 nonzero_chars. Add a full_string_p field.
9596 (compare_nonzero_chars, zero_length_string_p): New functions.
9597 (get_addr_stridx): Add an offset_out parameter.
9598 Use compare_nonzero_chars.
9599 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9600 (new_strinfo): Update after above changes to strinfo.
9601 (set_endptr_and_length): Set full_string_p.
9602 (get_string_length): Update after above changes to strinfo.
9603 (unshare_strinfo): Update call to new_strinfo.
9604 (maybe_invalidate): Likewise.
9605 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9606 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9607 as a uhwi instead of an shwi. Update after above changes to
9608 strinfo and new_strinfo.
9609 (zero_length_string): Assert that chainsi contains full strings.
9610 Use zero_length_string_p. Update call to new_strinfo.
9611 (adjust_related_strinfos): Update after above changes to strinfo.
9612 Copy full_string_p from origsi.
9613 (adjust_last_stmt): Use zero_length_string_p.
9614 (handle_builtin_strlen): Update after above changes to strinfo and
9615 new_strinfo. Install the lhs as the string length if the previous
9616 entry didn't describe a full string.
9617 (handle_builtin_strchr): Update after above changes to strinfo
9619 (handle_builtin_strcpy): Likewise.
9620 (handle_builtin_strcat): Likewise.
9621 (handle_builtin_malloc): Likewise.
9622 (handle_pointer_plus): Likewise.
9623 (handle_builtin_memcpy): Likewise. Track nonzero characters
9624 that aren't necessarily followed by a nul terminator.
9625 (handle_char_store): Likewise.
9627 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9629 PR tree-optimization/80769
9630 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9631 for malloc and calloc. Document the new invariant that all related
9632 strinfos have delayed lengths or none do.
9633 (verify_related_strinfos): Move earlier in file.
9634 (set_endptr_and_length): New function, split out from...
9635 (get_string_length): ...here. Also set the lengths of related
9637 (zero_length_string): Assert that chainsi has known (rather than
9639 (adjust_related_strinfos): Likewise.
9641 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9643 PR tree-optimization/81136
9644 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9645 assert that two references with the same misalignment have the same
9646 compile-time misalignment if those compile-time misalignments
9649 2017-07-01 Andi Kleen <ak@linux.intel.com>
9651 * print-tree.c (print_node): Print all attributes.
9653 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9655 * cfg.c (scale_bbs_frequencies): New function.
9656 * cfg.h (scale_bbs_frequencies): Declare it.
9657 * cfgloopanal.c (single_likely_exit): Cleanup.
9658 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9660 (scale_loop_profile): Likewise.
9661 (loop_version): Likewise.
9662 (create_empty_loop_on_edge): Update.
9663 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9664 scale_loop_frequencies, scale_loop_profile, loopify,
9665 loop_version): Update prototypes.
9666 * modulo-sched.c (sms_schedule): Update.
9667 * predict.c (unlikely_executed_edge_p): Also check probability.
9668 (probably_never_executed_edge_p): Fix typo.
9669 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9670 * tree-parloops.c (gen_parallel_loop): Update.
9671 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9672 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9673 * tree-ssa-loop-split.c (split_loop): Update.
9674 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9675 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9676 (vect_loop_versioning): Update.
9677 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9679 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9681 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9683 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9685 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9686 to keep profile consistent.
9688 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9690 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9691 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9692 * profile-count.h (max_safe_multiplier): Make unsigned.
9693 (profile_count::guessed_zero): New.
9695 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9697 * bb-reorder.c (fix_up_crossing_landing_pad,
9698 fix_crossing_conditional_branches): Use make_single_succ_edge
9699 to keep profile consistent.
9701 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9703 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9706 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9709 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9710 the right scopes, make sure cond_jump isn't preserved between multiple
9711 iterations. Search for fallthru edge whenever there are 3+ edges and
9712 use find_fallthru_edge for it.
9714 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9716 Patch by Alexander Monakov <amonakov@ispras.ru>
9717 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9718 probabilities consistently.
9720 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9722 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9723 emit_cmp_and_jump_insns.
9725 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9728 * tree-inline.c (expand_call_inline): Combine profile statuses.
9730 2017-06-30 Andrew Pinski <apinski@cavium.com>
9732 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9733 fold_stmt returned true.
9735 2017-06-30 Nathan Sidwell <nathan@acm.org>
9737 * ggc.h (empty_string): Delete.
9738 * cfgexpand.c (expand_asm_stmt): Use plain "".
9739 * optabs.c (expand_asm_memory_barrier): Likewise.
9740 * stringpool.c (empty_string): Delete.
9741 (digit_vector, digit_string): Delete.
9742 (ggc_alloc_string): Use plain "", don't optimize single digit
9743 strings. Use ggc_alloc_atomic.
9745 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9747 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9748 comparison set and one other set, use the cost of the non-comparison
9751 2017-06-30 Nathan Sidwell <nathan@acm.org>
9753 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9756 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9758 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9759 loops. Remove now unneeded calls to gimple_switch_set_label() that
9760 just set removed labels to NULL_TREE.
9762 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9764 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9765 (set_range_info): ...here. Only call set_range_info_raw if domain
9767 (set_nonzero_bits): Call set_range_info_raw.
9768 * tree-ssanames.h (set_range_info_raw): New.
9770 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9773 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9774 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9775 of nonimmediate_operand and <store_mask_constraint> instead of m
9776 for the input operand. For V8FI iterator, always split if input
9777 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9778 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9779 <store_mask_predicate> instead of register_operand and
9780 <store_mask_constraint> instead of v for the input operand. Make
9781 sure both operands aren't MEMs for if not <mask_applied>.
9783 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9785 * lto-wrapper.c (copy_file) Close both file descriptors before
9788 2017-06-30 Martin Liska <mliska@suse.cz>
9791 * multiple_target.c (create_dispatcher_calls): Make ifunc
9792 also for function that don't have calls or are not referenced.
9794 2017-06-30 Richard Biener <rguenther@suse.de>
9796 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9797 analyze the first scalar stmt. Move vector type computation
9798 for the BB case here from ...
9799 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9800 live operation processing in the SLP case properly.
9802 2017-06-30 Richard Biener <rguenther@suse.de>
9804 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9806 2017-06-30 Martin Liska <mliska@suse.cz>
9809 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9810 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9812 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9814 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9817 2017-06-30 Richard Biener <rguenther@suse.de>
9819 PR tree-optimization/81249
9820 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9821 condition reduction result to original scalar type.
9823 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9825 * profile-count.h (enum profile_quality): Fix typos and whitespace
9828 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9830 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9831 type for branch probabilities.
9833 2017-06-29 Julian Brown <julian@codesourcery.com>
9834 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9836 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9837 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9838 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9839 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9841 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9843 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9844 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9845 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9846 CC usage from generic code to here.
9847 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9848 CC usage into the target macros.
9850 2017-06-29 Maya Rashish <coypu@sdf.org>
9852 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9855 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9857 * arm/arm-builtins.c: Include profile-count.h
9858 * except.c (sjlj_emit_function_enter): Use
9859 profile_probability::unlikely.
9861 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9863 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9864 and tocrel_offset be pointer args rather than implicitly using
9866 (legitimate_constant_pool_address_p, rs6000_emit_move,
9867 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9868 tocrel_offset and use in toc_relative_expr_p call.
9869 (print_operand, print_operand_address): Use static tocrel_base_oac
9870 and tocrel_offset_oac.
9871 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9874 2017-06-29 Maya Rashish <coypu@sdf.org>
9876 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9878 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9880 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9881 objects, take into account only the alignment of 'op0' and 'mode1' if
9884 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9886 * ccmp.c (ccmp_tree_comparison_p): New function.
9887 (ccmp_candidate_p): Update to use above function.
9888 (get_compare_parts): New function.
9889 (expand_ccmp_next): Update to use new functions.
9890 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9892 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9893 take mode as argument.
9894 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9895 * expr.c (expand_expr_real_1): Pass mode as argument.
9897 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9899 * combine.c (combine_instructions): Print insns to dump_file, together
9902 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9904 * asan.c (asan_emit_stack_protection): Update.
9905 (create_cond_insert_point): Update.
9906 * auto-profile.c (afdo_propagate_circuit): Update.
9907 * basic-block.h (struct edge_def): Turn probability to
9908 profile_probability.
9909 (EDGE_FREQUENCY): Update.
9910 * bb-reorder.c (find_traces_1_round): Update.
9911 (better_edge_p): Update.
9912 (sanitize_hot_paths): Update.
9913 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9914 (make_single_succ_edge): Update.
9915 (check_bb_profile): Update.
9916 (dump_edge_info): Update.
9917 (update_bb_profile_for_threading): Update.
9918 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9919 probabilitycount to 0.
9920 * cfgbuild.c (compute_outgoing_frequencies): Update.
9921 * cfgcleanup.c (try_forward_edges): Update.
9922 (outgoing_edges_match): Update.
9923 (try_crossjump_to_edge): Update.
9924 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9925 (expand_gimple_tailcall): Update.
9926 (construct_init_block): Use make_single_succ_edge.
9927 (construct_exit_block): Use make_single_succ_edge.
9928 * cfghooks.c (verify_flow_info): Update.
9929 (redirect_edge_succ_nodup): Update.
9930 (split_edge): Update.
9931 (account_profile_record): Update.
9932 * cfgloopanal.c (single_likely_exit): Update.
9933 * cfgloopmanip.c (scale_loop_profile): Update.
9934 (set_zero_probability): Remove.
9935 (duplicate_loop_to_header_edge): Update.
9936 * cfgloopmanip.h (loop_version): Update prototype.
9937 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9938 (force_nonfallthru_and_redirect): Update.
9939 (update_br_prob_note): Update.
9940 (rtl_verify_edges): Update.
9941 (purge_dead_edges): Update.
9942 (rtl_lv_add_condition_to_bb): Update.
9943 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9944 * cgraphunit.c (init_lowered_empty_function): Update.
9945 (cgraph_node::expand_thunk): Update.
9946 * cilk-common.c: Include profile-count.h
9947 * dojump.c (inv): Remove.
9948 (jumpifnot): Update.
9949 (jumpifnot_1): Update.
9950 (do_jump_1): Update.
9952 (do_jump_by_parts_greater_rtx): Update.
9953 (do_compare_rtx_and_jump): Update.
9954 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9955 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9956 * dwarf2cfi.c: Include profile-count.h
9957 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9958 (sjlj_emit_dispatch_table): Likewise.
9959 * explow.c: Include profile-count.h
9960 * expmed.c (emit_store_flag_force): Update.
9961 (do_cmp_and_jump): Update.
9962 * expr.c (compare_by_pieces_d::generate): Update.
9963 (compare_by_pieces_d::finish_mode): Update.
9964 (emit_block_move_via_loop): Update.
9965 (store_expr_with_bounds): Update.
9966 (store_constructor): Update.
9967 (expand_expr_real_2): Update.
9968 (expand_expr_real_1): Update.
9969 * expr.h (try_casesi, try_tablejump): Update prototypes.
9970 * gimple-pretty-print.c (dump_probability): Update.
9971 (dump_profile): New.
9972 (dump_gimple_label): Update.
9973 (dump_gimple_bb_header): Update.
9974 * graph.c (draw_cfg_node_succ_edges): Update.
9975 * hsa-gen.c (convert_switch_statements): Update.
9976 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
9977 (find_if_case_1): Update.
9978 (find_if_case_2): Update.
9979 * internal-fn.c (expand_arith_overflow_result_store): Update.
9980 (expand_addsub_overflow): Update.
9981 (expand_neg_overflow): Update.
9982 (expand_mul_overflow): Update.
9983 (expand_vector_ubsan_overflow): Update.
9984 * ipa-cp.c (good_cloning_opportunity_p): Update.
9985 * ipa-split.c (split_function): Use make_single_succ_edge.
9986 * ipa-utils.c (ipa_merge_profiles): Update.
9987 * loop-doloop.c (add_test): Update.
9988 (doloop_modify): Update.
9989 * loop-unroll.c (compare_and_jump_seq): Update.
9990 (unroll_loop_runtime_iterations): Update.
9991 * lra-constraints.c (lra_inheritance): Update.
9992 * lto-streamer-in.c (input_cfg): Update.
9993 * lto-streamer-out.c (output_cfg): Update.
9994 * mcf.c (adjust_cfg_counts): Update.
9995 * modulo-sched.c (sms_schedule): Update.
9996 * omp-expand.c (expand_omp_for_init_counts): Update.
9997 (extract_omp_for_update_vars): Update.
9998 (expand_omp_ordered_sink): Update.
9999 (expand_omp_for_ordered_loops): Update.
10000 (expand_omp_for_generic): Update.
10001 (expand_omp_for_static_nochunk): Update.
10002 (expand_omp_for_static_chunk): Update.
10003 (expand_cilk_for): Update.
10004 (expand_omp_simd): Update.
10005 (expand_omp_taskloop_for_outer): Update.
10006 (expand_omp_taskloop_for_inner): Update.
10007 * omp-simd-clone.c (simd_clone_adjust): Update.
10008 * optabs.c (expand_doubleword_shift): Update.
10009 (expand_abs): Update.
10010 (emit_cmp_and_jump_insn_1): Update.
10011 (expand_compare_and_swap_loop): Update.
10012 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10013 * predict.c (predictable_edge_p): Update.
10014 (edge_probability_reliable_p): Update.
10015 (set_even_probabilities): Update.
10016 (combine_predictions_for_insn): Update.
10017 (combine_predictions_for_bb): Update.
10018 (propagate_freq): Update.
10019 (estimate_bb_frequencies): Update.
10020 (force_edge_cold): Update.
10021 * profile-count.c (profile_count::dump): Add missing space into dump.
10022 (profile_count::debug): Add newline.
10023 (profile_count::differs_from_p): Explicitly convert to unsigned.
10024 (profile_count::stream_in): Update.
10025 (profile_probability::dump): New member function.
10026 (profile_probability::debug): New member function.
10027 (profile_probability::differs_from_p): New member function.
10028 (profile_probability::differs_lot_from_p): New member function.
10029 (profile_probability::stream_in): New member function.
10030 (profile_probability::stream_out): New member function.
10031 * profile-count.h (profile_count_quality): Rename to ...
10032 (profile_quality): ... this one.
10033 (profile_probability): New.
10034 (profile_count): Update.
10035 * profile.c (compute_branch_probabilities): Update.
10036 * recog.c (peep2_attempt): Update.
10037 * sched-ebb.c (schedule_ebbs): Update.
10038 * sched-rgn.c (find_single_block_region): Update.
10039 (compute_dom_prob_ps): Update.
10040 (schedule_region): Update.
10041 * sel-sched-ir.c (compute_succs_info): Update.
10042 * stmt.c (struct case_node): Update.
10043 (do_jump_if_equal): Update.
10044 (get_outgoing_edge_probs): Update.
10045 (conditional_probability): Update.
10046 (emit_case_dispatch_table): Update.
10047 (expand_case): Update.
10048 (expand_sjlj_dispatch_table): Update.
10049 (emit_case_nodes): Update.
10050 * targhooks.c: Update.
10051 * tracer.c (better_p): Update.
10052 (find_best_successor): Update.
10053 * trans-mem.c (expand_transaction): Update.
10054 * tree-call-cdce.c: Update.
10055 * tree-cfg.c (gimple_split_edge): Upate.
10056 (move_sese_region_to_fn): Upate.
10057 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10058 * tree-eh.c (lower_resx): Upate.
10059 (cleanup_empty_eh_move_lp): Upate.
10060 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10061 * tree-inline.c (copy_edges_for_bb): Update.
10062 (copy_cfg_body): Update.
10063 * tree-parloops.c (gen_parallel_loop): Update.
10064 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10065 (gimple_gen_time_profiler): Update.
10066 * tree-ssa-dce.c (remove_dead_stmt): Update.
10067 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10068 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10069 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10070 (unloop_loops): Update.
10071 (try_peel_loop): Update.
10072 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10073 * tree-ssa-loop-split.c (connect_loops): Update.
10074 (split_loop): Update.
10075 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10076 (hoist_guard): Update.
10077 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10078 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10079 (value_replacement): Update.
10080 * tree-ssa-reassoc.c (branch_fixup): Update.
10081 * tree-ssa-tail-merge.c (replace_block_by): Update.
10082 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10083 (create_edge_and_update_destination_phis): Update.
10084 (compute_path_counts): Update.
10085 (recompute_probabilities): Update.
10086 (update_joiner_offpath_counts): Update.
10087 (freqs_to_counts_path): Update.
10088 (duplicate_thread_path): Update.
10089 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10090 (struct switch_conv_info): Update.
10091 (gen_inbound_check): Update.
10092 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10093 (vect_do_peeling): Update.
10094 (vect_loop_versioning): Update.
10095 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10096 (optimize_mask_stores): Update.
10097 * ubsan.c (ubsan_expand_null_ifn): Update.
10098 * value-prof.c (gimple_divmod_fixed_value): Update.
10099 (gimple_divmod_fixed_value_transform): Update.
10100 (gimple_mod_pow2): Update.
10101 (gimple_mod_pow2_value_transform): Update.
10102 (gimple_mod_subtract): Update.
10103 (gimple_mod_subtract_transform): Update.
10104 (gimple_ic): Update.
10105 (gimple_stringop_fixed_value): Update.
10106 (gimple_stringops_transform): Update.
10107 * value-prof.h: Update.
10109 2017-06-29 Carl Love <cel@us.ibm.com>
10111 * config/rs6000/rs6000-c.c: Add support for built-in functions
10112 vector signed int vec_signed (vector float);
10113 vector signed long long vec_signed (vector double);
10114 vector signed int vec_signed2 (vector double, vector double);
10115 vector signed int vec_signede (vector double);
10116 vector signed int vec_signedo (vector double);
10117 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10118 instruction generator.
10119 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10120 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10121 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10123 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10124 vunsignede_v2df): Add define_expands.
10125 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10126 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10127 VEC_UNSIGNEDO): Add definitions.
10128 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10129 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10130 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10131 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10132 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10133 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10134 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10135 vec_unsignede, vec_unsignedo): Add builtin defines.
10136 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10138 * doc/extend.texi: Update the built-in documentation file for the
10139 new built-in functions.
10141 2017-06-29 Richard Biener <rguenther@suse.de>
10143 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10144 reduction chains to LOOP_VINFO_REDUCTIONS.
10145 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10146 SLP reductions after processing reduction chains.
10148 2017-06-29 Nathan Sidwell <nathan@acm.org>
10150 * builtins.c (fold_builtin_FUNCTION): Use
10151 lang_hooks.decl_printable_name.
10153 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10155 PR middle-end/81194
10156 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10157 with only one label.
10158 * stmt.c (expand_case): Assert NCASES is greater than one.
10160 2017-06-29 Richard Biener <rguenther@suse.de>
10162 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10164 (group_case_labels): Likewise.
10165 (find_taken_edge): Push sanity checking on val to workers...
10166 (find_taken_edge_cond_expr): ... here
10167 (find_taken_edge_switch_expr): ... and here, handle cases
10168 with just a default label.
10169 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10170 (group_case_labels): Likewise.
10171 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10172 group_case_labels does anything cleanup the CFG again.
10174 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10176 PR tree-optimization/81196
10177 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10178 exit condition comparing two IVs.
10180 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10182 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10183 profile to the dummy entry at the end of the list of architectures.
10184 * config/arm/arm-cpu-cdata.h: Regenerated.
10186 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10187 Michael Collison <michael.collison@arm.com>
10190 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10192 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10193 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10194 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10195 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10196 of shift when the shift amount is masked with constant equal to
10197 the size of the mode.
10198 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10201 2017-06-29 Martin Liska <mliska@suse.cz>
10203 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10205 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10208 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10211 (edge_growth_cache_entry::edge_growth_cache_entry): New
10213 (reset_edge_growth_cache): Update to use constructor.
10215 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10217 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10218 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10219 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10221 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10223 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10224 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10226 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10228 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10229 (*-linux-uclibc*): Add t-uclibc tmake_file.
10230 * config/t-musl: New.
10231 * config/t-uclibc: New.
10233 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10235 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10237 (gen_comm_data): Emit architectural setting of arch_prof.
10238 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10240 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10241 (armv8-m.base, armv8-m.main): Likewise.
10242 * arm-protos.h (arm_build_target): Add profile field.
10243 (arch_option): Likewise.
10244 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10246 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10247 arm_active_target.profile.
10249 2017-06-28 Richard Biener <rguenther@suse.de>
10251 PR middle-end/81227
10252 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10253 TYPE_OVERFLOW_WRAPS.
10254 * match.pd (negate_expr_p): Likewise.
10255 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10256 fold_build2, not fold_binary.
10258 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10260 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10261 Convert memory address to Pmode.
10262 (aarch64_print_operand): Assert MEM operands are always Pmode.
10264 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10267 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10268 Remove redundant if.
10269 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10270 * config/arm/aarch-common-protos.h
10271 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10272 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10274 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10277 * multiple_target.c (create_dispatcher_calls): Set the default
10278 clone to be static, not public.
10280 2017-06-28 Richard Biener <rguenther@suse.de>
10282 * tree-vect-loop.c (vectorizable_reduction): Move special
10283 cond reduction IV var creation ...
10284 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10285 parameter. Use STMT_VINFO_VECTYPE.
10286 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10289 2017-06-28 Martin Liska <mliska@suse.cz>
10292 * ipa-visibility.c (non_local_p): Handle visibility.
10294 2017-06-28 Martin Liska <mliska@suse.cz>
10297 * common.opt: Add IntegerRange to various options.
10298 * opt-functions.awk (integer_range_info): New function.
10299 * optc-gen.awk: Add integer_range_info to cl_options struct.
10300 * opts-common.c (decode_cmdline_option): Handle
10301 CL_ERR_INT_RANGE_ARG.
10302 (cmdline_handle_error): Likewise.
10303 * opts.c (print_filtered_help): Show valid interval in
10304 when --help is provided.
10305 * opts.h (struct cl_option): Add range_min and range_max fields.
10306 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10308 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10310 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10311 (x * C EQ/NE y * C): New transformation.
10313 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10315 * genmultilib (combination_space): Accept '+' in option names.
10317 2017-06-28 Martin Liska <mliska@suse.cz>
10320 * asan.c (instrument_derefs): Bail out inner references
10321 that are hard register variables.
10323 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10326 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10327 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10329 2017-06-28 Richard Biener <rguenther@suse.de>
10331 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10332 (vect_get_slp_defs): Adjust.
10333 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10335 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10337 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10338 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10339 (vectorizable_reduction): Adjust.
10340 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10342 (vect_get_slp_defs): Likewise.
10343 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10344 (vectorizable_bswap): Adjust.
10345 (vectorizable_call): Likewise.
10346 (vectorizable_conversion): Likewise.
10347 (vectorizable_assignment): Likewise.
10348 (vectorizable_shift): Likewise.
10349 (vectorizable_operation): Likewise.
10350 (vectorizable_store): Likewise.
10351 (vectorizable_condition): Likewise.
10352 (vectorizable_comparison): Likewise.
10354 2017-06-28 Michael Collison <michael.collison@arm.com>
10357 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10359 (arm_gen_movmemqi): Removed unused variable 'i'.
10360 Convert 'for' loop into 'while' loop.
10361 (arm_expand_prologue): Remove last unnecessary set of insn.
10362 (thumb_pop): Remove unused variable 'pushed_words'.
10363 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10365 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10367 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10368 * config/s390/s390-protos.h: Add prototype of
10369 s390_rel_address_ok_p.
10370 * config/s390/s390.c (s390_got_symbol): New function.
10371 (s390_rel_address_ok_p): New function.
10372 (legitimize_pic_address): Use s390_rel_address_ok_p.
10373 (s390_load_got): Use s390_got_symbol.
10374 (s390_option_override): Issue error if
10375 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10376 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10378 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10380 2017-06-27 Andrew Pinski <apinski@cavium.com>
10382 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10383 (X * copysign (1.0, X)): New pattern.
10384 (X * copysign (1.0, -X)): New pattern.
10385 (copysign (-1.0, CST)): New pattern.
10387 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10389 * genmultilib (combination_space): Remove variable.
10390 Validate reuse rules against regular expression for any sequence
10391 of multilib options in any order.
10393 2017-06-27 Michael Collison <michael.collison@arm.com>
10395 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10396 call aarch64_split_simd_combine.
10397 * (aarch64_combine_internal<mode>): Delete pattern.
10398 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10399 Allow register and subreg operands.
10401 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10403 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10404 specific need, just fallback on defaults.
10405 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10407 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10408 Olivier Hainque <hainque@adacore.com>
10410 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10412 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10413 targets. Pick a default if no particular attempt applied.
10414 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10417 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10419 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10420 (x86_64-wrs-vxworks7): Likewise.
10422 2017-06-27 Marek Polacek <polacek@redhat.com>
10425 * ubsan.c (instrument_null): Check get_base_address's result for null.
10427 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10429 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10431 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10433 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10434 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10435 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10436 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10437 New function types.
10438 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10439 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10440 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10441 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10442 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10443 BUILT_IN_FEUPDATEENV): New builtins.
10444 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10445 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10446 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10447 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10449 (builtin_structptr_types): Adjust size.
10450 * tree.c (builtin_structptr_types): Add four entries.
10452 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10453 Olivier Hainque <hainque@adacore.com>
10455 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10456 (TLS_SYM): New local macro, forcing reference to __tls__ on
10457 link command lines for VxWorks 7 RTPs, triggering initialization
10459 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10460 OS features TLS support, true for RTPs on VxWorks 7.
10461 * config/vxworks.c (vxworks_override_options): Setup emutls
10464 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10466 * predict.c (test_prediction_value_range): Use -1U instead of -1
10467 to avoid narrowing conversion warning.
10468 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10469 to avoid narrowing conversion warning.
10470 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10472 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10474 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10476 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10477 64bit configurations.
10478 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10479 (SIZE_TYPE): Likewise.
10480 * config/vxworks.c (vxworks_emutls_var_fields): Use
10481 long_unsigned_type_node instead of unsigned_type_node as the offset
10482 field type, which is "pointer" mode in emutls.c.
10484 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10487 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10489 PR middle-end/81207
10490 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10491 gimple_vuse copying separately from gimple_vdef copying.
10493 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10495 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10498 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10499 Olivier Hainque <hainque@adacore.com>
10501 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10502 for all vxworks7 targets.
10503 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10504 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10505 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10506 variations for VX6/VX7 and 32/64bits later on in ...
10507 (VXWORKS_LIB_SPEC): Leverage new macros.
10508 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10509 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10511 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10513 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10514 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10516 2017-06-26 Carl Love <cel@us.ibm.com>
10518 * config/rs6000/rs6000-c.c: Add support for built-in functions
10519 vector bool char vec_reve (vector bool char);
10520 vector signed char vec_reve (vector signed char);
10521 vector unsigned char vec_reve (vector unsigned char);
10522 vector bool int vec_reve (vector bool int);
10523 vector signed int vec_reve (vector signed int);
10524 vector unsigned int vec_reve (vector unsigned int);
10525 vector bool long long vec_reve (vector bool long long);
10526 vector signed long long vec_reve (vector signed long long);
10527 vector unsigned long long vec_reve (vector unsigned long long);
10528 vector bool short vec_reve (vector bool short);
10529 vector signed short vec_reve (vector signed short);
10530 vector double vec_reve (vector double);
10531 vector float vec_reve (vector float);
10532 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10533 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10534 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10535 (altivec_vreve): New pattern.
10536 * config/rs6000/altivec.h (vec_reve): New define.
10537 * doc/extend.texi (vec_rev): Update the built-in documentation file
10538 for the new built-in functions.
10540 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10542 PR tree-optimization/71815
10543 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10545 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10547 (slsr_process_phi): Likewise.
10548 (replace_uncond_cands_and_profitable_phis): Don't replace a
10549 multiply candidate with a stride of 1 (copy or cast).
10550 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10552 (lowest_cost_path): Likewise.
10553 (total_savings): Likewise.
10555 2017-06-26 Richard Biener <rguenther@suse.de>
10558 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10559 Use def_builtin_pure for all gather builtins.
10561 2017-06-26 Richard Biener <rguenther@suse.de>
10563 PR tree-optimization/81203
10564 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10565 non-dominating BBs.
10567 2017-06-26 Marek Polacek <polacek@redhat.com>
10570 * doc/invoke.texi: Document -Wmultistatement-macros.
10572 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10574 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10575 arm_neon_ok_no_float_abi effective target.
10577 2017-06-26 Richard Biener <rguenther@suse.de>
10579 PR tree-optimization/80928
10580 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10581 (copy_bbs): Set BB_DUPLICATED flag early.
10582 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10584 (execute_on_shrinking_pred): Likewise.
10585 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10586 BB_DUPLICATED blocks.
10587 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10588 iterate over all PHIs considering removal of *gsi.
10590 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10592 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10595 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10597 * config/rs6000/rs6000-string.c: (expand_block_clear,
10598 do_load_for_compare, select_block_compare_mode,
10599 compute_current_alignment, expand_block_compare,
10600 expand_strncmp_align_check, expand_strn_compare,
10601 expand_block_move, rs6000_output_load_multiple)
10602 Move functions related to string/block move/compare
10603 to a separate file.
10604 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10605 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10606 for this function which is now used in two files.
10607 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10608 * config.gcc: Add rs6000-string.o to extra_objs for
10609 targets powerpc*-*-* and rs6000*-*-*.
10611 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10614 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10615 32-bit, since indexed is not valid for DImode.
10616 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10617 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10618 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10619 (define_peephole2 for Altivec d-form store): Likewise.
10622 * multiple_target.c (create_dispatcher_calls): Only create the
10623 dispatcher call if the function is the default clone of a
10624 versioned function.
10626 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10628 PR middle-end/80902
10629 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10630 a call, force the call to not be a tail call.
10632 2017-06-23 Jeff Law <law@redhat.com>
10634 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10637 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10639 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10640 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10641 handling for early expansion of vector shifts (sl,sr,sra,rl).
10642 (builtin_function_type): Add vector shift right instructions
10643 to the unsigned argument list.
10645 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10647 rtl-optimizatoin/79286
10648 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10649 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10650 trap. PIC register plus a const unspec without offset can never trap.
10652 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10654 * tree.h (builtin_structptr_type): New type.
10655 (builtin_structptr_types): Declare new array.
10656 * tree.c (builtin_structptr_types): New array.
10657 (free_lang_data, build_common_tree_nodes): Use it.
10659 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10662 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10665 2017-06-22 Matt Turner <mattst88@gmail.com>
10667 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10668 Lake models to skylake case. Assume skylake for unknown
10669 models with clflushopt.
10671 2017-06-22 Jeff Law <law@redhat.com>
10673 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10674 frame sizes that do not satisfy aarch64_uimm12_shift.
10676 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10678 * profile-count.h (apply_probability,
10679 apply_scale, probability_in): Fix checks for zero.
10681 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10683 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10684 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10686 2016-06-22 Richard Biener <rguenther@suse.de>
10688 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10689 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10690 REDUC_MAX_EXPR support.
10691 (vectorizable_reduction): Likewise.
10692 (vect_create_epilog_for_reduction): Likewise.
10694 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10696 * match.pd (A / (1 << B) -> A >> B): New.
10697 * generic-match-head.c: Include optabs-tree.h.
10698 * gimple-match-head.c: Likewise.
10699 * optabs-tree.h (target_supports_op_p): New.
10700 * optabs-tree.c (target_supports_op_p): New.
10702 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10704 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10705 $gcc_cv_ld --help output.
10706 (gcc_cv_ld_demangle): Likewise.
10707 (gcc_cv_ld_eh_frame_hdr): Likewise.
10708 (gcc_cv_ld_pie): Likewise.
10709 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10710 (gcc_cv_ld_buildid): Likewise.
10711 (gcc_cv_ld_sysroot): Likewise.
10712 (ld_bndplt_support): Likewise.
10713 (ld_pushpopstate_support): Likewise.
10714 * configure: Regenerate.
10715 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10717 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10720 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10721 operands indexes to avoid gap between operands and match_dups.
10723 2017-06-21 Andrew Pinski <apinski@cavium.com>
10725 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10726 Increment Arith_shift and Arith_shift_reg by 1.
10727 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10729 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10730 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10731 (aarch64_strip_extend): Add new argument and test for it.
10732 (aarch64_cheap_mult_shift_p): New function.
10733 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10734 add a cost if it is true.
10735 Update calls to aarch64_strip_extend.
10736 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10738 2017-06-21 Andrew Pinski <apinski@cavium.com>
10740 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10742 (thunderxt88): Likewise.
10743 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10744 (thunderx_prefetch_tune): New variable.
10745 (thunderx2t99_prefetch_tune): Update for the correct values.
10746 (thunderxt88_tunings): New variable.
10747 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10748 generic_prefetch_tune.
10749 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10751 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10753 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10754 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10755 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10756 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10757 (aarch64_atomic_cas<mode>, GPI): Likewise.
10759 2017-06-21 Martin Liska <mliska@suse.cz>
10761 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10762 statements on cold and hot labels.
10763 * predict.c (tree_estimate_probability_bb): Remove the
10764 prediction from this place.
10766 2017-06-21 Martin Liska <mliska@suse.cz>
10768 PR tree-optimization/79489
10769 * gimplify.c (maybe_add_early_return_predict_stmt): New
10771 (gimplify_return_expr): Call the function.
10772 * predict.c (tree_estimate_probability_bb): Remove handling
10774 * predict.def: Update comment about early return predictor.
10775 * gimple-predict.h (is_gimple_predict): New function.
10776 * predict.def: Change default value of early return to 66.
10777 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10779 * passes.def: Put pass_strip_predict_hints to the beginning of
10782 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10784 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10785 FUNCTION_DECL declarations.
10786 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10788 (dwaf2out_decl): Likewise.
10789 * godump.c (go_early_global_decl): Skip call to the real debug hook
10790 for FUNCTION_DECL declarations.
10791 * passes.c (rest_of_decl_compilation): Skip call to the
10792 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10793 -fdump-go-spec is passed.
10795 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10797 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10798 Reorder for compactness.
10799 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10800 (def_builtin_pure, def_builtin_pure2): New functions.
10801 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10803 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10805 * match.pd (nop_convert): New predicate.
10806 ((A +- CST1) +- CST2): Allow some NOP conversions.
10808 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10811 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10812 with ctors/dtors if GOVD_SHARED is set.
10814 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10816 * config/aarch64/aarch64.md (movti_aarch64):
10817 Emit mov rather than orr.
10818 (movtf_aarch64): Likewise.
10819 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10820 Emit mov rather than orr.
10822 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10824 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10825 Swap alternatives, make integer dup more expensive.
10827 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10829 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10830 Return true for non-tls symbols.
10832 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10834 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10835 (cortex-a75): Likewise.
10836 (cortex-a75.cortex-a55): Likewise.
10837 * config/aarch64/aarch64-tune.md: Regenerate.
10838 * doc/invoke.texi (-mtune): Document new values for -mtune.
10840 2017-06-21 Tom de Vries <tom@codesourcery.com>
10842 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10843 stack_size feature.
10844 (Effective-Target Keywords, Other attributes): Suggest using
10845 dg-add-options stack_size feature to get stack limit in stack_size
10846 effective target documentation.
10848 2017-06-21 Julian Brown <julian@codesourcery.com>
10849 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10851 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10852 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10853 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10855 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10856 attribute type list for neon_multiply.
10857 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10858 attribute type list for neon_multiply.
10859 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10860 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10861 attribute type list for neon_multiply.
10862 * config/arm/types.md (crypto_pmull): Add.
10863 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10864 attribute type list.
10866 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10868 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10871 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10873 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10874 to make sure not to dereference a NULL cost_classes_ptr pointer.
10876 2017-06-20 Carl Love <cel@us.ibm.com>
10878 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10879 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10880 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10881 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10882 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10883 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10884 VMULOSW): New enum "unspec" values.
10885 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10886 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10887 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10888 altivec_vmulosw): New patterns.
10889 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10890 VMULOSW): Add definitions.
10892 2017-06-20 Julia Koval <julia.koval@intel.com>
10894 * config/i386/i386.c: Fix rounding expand for new pattern.
10895 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10897 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10899 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10900 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10902 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10904 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10907 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10909 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10910 architecture, then by implementer ID.
10911 * config/aarch64/aarch64-tune.md: Regenerate.
10913 2017-06-20 Richard Biener <rguenther@suse.de>
10915 PR middle-end/81097
10916 * fold-const.c (split_tree): Fold to type before negating.
10918 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10920 * diagnostic-show-locus.c
10921 (selftest::test_fixit_deletion_affecting_newline): New function.
10922 (selftest::diagnostic_show_locus_c_tests): Call it.
10924 2017-06-20 Andreas Schwab <schwab@suse.de>
10927 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10930 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10932 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10933 __ARM_FEATURE_COPROC according to support.
10935 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10937 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10938 Rewritten to avoid overflow for > 32-bit pointers.
10941 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10942 by removing enum keyword.
10943 (ubsan_type_descriptor): Likewise. Formatting fix.
10946 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10947 splitter): Require TARGET_SSE2 in the condition.
10949 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10952 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10953 for doing vector set of SFmode on ISA 3.0.
10954 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10955 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10957 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10958 SFmode value into a V4SF variable that was extracted from another
10959 V4SF variable without converting the element to double precision
10960 and back to single precision vector format.
10961 (vsx_insert_extract_v4sf_p9_2): Likewise.
10963 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10965 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10966 in UWHI to avoid undefined overflow.
10969 * ubsan.h (enum ubsan_encode_value_phase): New.
10970 (ubsan_encode_value): Change second argument to
10971 enum ubsan_encode_value_phase with default value of
10972 UBSAN_ENCODE_VALUE_GENERIC.
10973 * ubsan.c (ubsan_encode_value): Change second argument to
10974 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
10975 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
10976 create_tmp_var_raw instead of create_tmp_var and use a
10978 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
10979 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
10980 ubsan_encode_value callers.
10983 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
10984 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
10985 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
10987 2017-06-19 Richard Biener <rguenther@suse.de>
10989 PR middle-end/81118
10990 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
10991 estimates if we changed anything.
10993 2017-06-19 Richard Biener <rguenther@suse.de>
10995 PR tree-optimization/80887
10996 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
10997 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
10998 simplified lookups, then reset mprts_hook.
10999 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11001 (try_to_simplify): Likewise.
11003 2017-06-19 Martin Liska <mliska@suse.cz>
11006 * gimplify.c (gimplify_switch_expr):
11007 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11009 2017-06-19 Martin Liska <mliska@suse.cz>
11011 * doc/install.texi: Document that PGO runs in 4 stages.
11013 2017-06-19 Martin Liska <mliska@suse.cz>
11016 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11017 to dispatcher function name.
11018 * multiple_target.c (replace_function_decl): New function.
11019 (create_dispatcher_calls): Redirect both edges and references.
11021 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11023 * profile-count.c (profile_count::dump): Dump quality.
11024 (profile_count::differs_from_p): Update for unsigned val.
11025 * profile-count.h (profile_count_quality): New enum.
11026 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11028 2017-06-19 Richard Biener <rguenther@suse.de>
11030 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11031 struct function as arg.
11032 (estimate_numbers_of_iterations): Export overload with loop arg.
11033 (free_numbers_of_iterations_estimates_loop): Use an overload of
11034 free_numbers_of_iterations_estimates instead.
11035 * tree-cfg.c (remove_bb): Adjust.
11036 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11037 * tree-parloops.c (gen_parallel_loop): Likewise.
11038 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11040 (tree_unroll_loops_completely): Likewise.
11041 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11042 Use an overload instead and export.
11043 (estimated_loop_iterations): Adjust.
11044 (max_loop_iterations): Likewise.
11045 (likely_max_loop_iterations): Likewise.
11046 (estimate_numbers_of_iterations): Take struct function as arg
11048 (loop_exits_before_overflow): Adjust.
11049 (free_numbers_of_iterations_estimates_loop): Use an overload.
11050 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11051 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11053 2017-06-19 Richard Biener <rguenther@suse.de>
11056 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11057 RANGE_EXPR conservatively.
11059 2017-06-16 Carl Love <cel@us.ibm.com>
11061 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11062 definitions for vec_float, vec_float2, vec_floato,
11063 vec_floate built-ins.
11064 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11065 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11067 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11068 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11069 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11070 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11071 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11072 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11073 vec_floato): Add builtin defines.
11074 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11075 Update the built-in documentation file for the new built-in
11078 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11080 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11081 (mthumb): Mark as the negative of -marm.
11083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11085 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11087 (ARM Options, -mtune): Document that this accepts the same
11088 extension options as -mcpu.
11089 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11091 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11093 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11094 permitted extensions.
11096 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11098 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11099 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11100 (armv8-m.main): Add option +nodsp.
11101 * config/arm/arm-cpu-cdata.h: Regenerated.
11103 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11105 * config/arm/t-fuchsia: New file.
11106 * config.gcc (arm*-*-fuchsia*): Use it.
11108 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11110 * config/arm/t-symbian: Rewrite for new option infrastructure.
11112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11114 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11115 (MULTILIB_REQUIRED): Likewise.
11117 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11119 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11120 (MULTILIB_RESUE): Likewise.
11121 (MULTILIB_MATCHES): Likewise.
11122 (MULTLIB_REQUIRED): Likewise.
11124 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11126 * config/arm/t-rtems: Rewrite for new option framework.
11128 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11130 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11131 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11132 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11133 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11134 * config/arm/t-multilib: ... here.
11135 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11136 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11137 armv7-a and armv8*-a when A-profile libraries have not been built.
11138 * config/arm/t-rmprofile: Rewrite.
11140 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11142 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11143 with a backslash. Remove the backslash after substituting unescaped
11145 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11149 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11150 the driver to pass through the default setting of -mfloat-abi.
11151 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11153 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11154 (all_feat_combs): New rule.
11155 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11157 * config/arm/t-aprofile: Rewrite.
11159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11161 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11162 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11163 fpu is not specified by the user/command-line.
11164 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11165 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11166 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11167 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11168 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11169 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11171 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11173 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11174 * config/arm/t-arm-elf: Rewritten.
11176 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11178 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11179 have some floating-point instructions.
11180 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11181 (TARGET_MAYBE_HARD_FLOAT): New macro.
11182 * config/arm/arm-builtins.c (arm_init_builtins): Use
11183 TARGET_MAYBE_HARD_FLOAT.
11184 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11186 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11188 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11189 (configargs.h): Include it.
11190 (arm_print_hint_for_fpu_option): New function.
11191 (arm_parse_fpu_option): New function.
11192 (candidate_extension): New class.
11193 (arm_canon_for_multilib): New function.
11194 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11195 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11196 (ARCH_CANONICAL_SPECS): New macro.
11197 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11199 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11201 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11202 are set after handling multilib fragments. Set target_cpu_default2
11205 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11207 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11209 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11210 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11212 (check_arch): Likewise.
11213 * config/arm/arm.c (arm_configure_build_target): Handle
11214 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11215 options in the default.
11217 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11219 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11220 when an option is an alias of another.
11221 * config/arm/parsecpu.awk (optalias): New parser token.
11222 (gen_comm_data): Mark non-alias options as such. Emit entries
11223 for extension aliases.
11224 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11225 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11226 (armv6kz, armv6zk, armv6t2): Likewise.
11227 (armv7): Make vfpv3-d16 an alias.
11228 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11230 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11231 Sort in canonical order.
11232 (armv8-a): Sort in canonical order.
11233 (armv8.1-a, armv8.2-a): Likewise.
11234 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11236 (cortex-a9): Sort in canonical order.
11237 * config/arm/arm.c (selftests.h): Include it.
11238 (arm_test_cpu_arch_data): New function.
11239 (arm_run_self_tests): New function.
11240 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11241 (targetm): Move declaration to the end of the file.
11242 * arm-cpu-cdata.h: Regenerated.
11244 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11246 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11247 call to target_mode_check describing the type of option passed.
11248 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11249 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11250 arm_parse_cpu_option_name to match parameters against list of
11252 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11253 arm_arch_core_flags data structure.
11254 * config/arm/arm-cpu_cdata.h: Regenerated.
11256 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11258 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11260 (arm_print_hint_for_cpu_option): Likewise.
11261 (arm_print_hint_for_arch_option): Likewise.
11262 (arm_parse_cpu_option_name): Likewise.
11263 (arm_parse_arch_option_name): Likewise.
11264 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11265 of entries in the all_fpus list.
11266 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11267 (arm_parse_cpu_option_name): Declare.
11268 (arm_parse_arch_option_name): Declare.
11269 (arm_parse_option_features): Declare.
11270 (arm_intialize_isa): Declare.
11271 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11273 (gen_comm_data): ... here. Make definitions non-static.
11274 * config/arm/arm-cpu-data.h: Regenerated.
11275 * config/arm/arm-cpu-cdata.h: Regenerated.
11277 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11279 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11280 (cpu_arch_extension): New structure.
11281 (cpu_arch_option, arch_option, cpu_option): New structures.
11282 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11283 architecture types.
11284 (gen_data): Generate new format data tables.
11285 * config/arm/arm.c (cpu_tune): New structure.
11286 (cpu_option, processors): Delete.
11287 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11288 (arm_print_hint_for_cpu_option): ... this and ...
11289 (arm_print_hint_for_arch_option): ... this.
11290 (arm_parse_arch_cpu_name): Delete. Replace with ...
11291 (arm_parse_cpu_option_name): ... this and ...
11292 (arm_parse_arch_option_name): ... this.
11293 (arm_unrecognized_feature): Change type of target parameter to
11295 (arm_parse_arch_cpu_features): Delete. Replace with ...
11296 (arm_parse_option_features): ... this.
11297 (arm_configure_build_target): Rework to use new configuration data
11299 (arm_print_tune_info): Rework for new configuration data tables.
11300 * config/arm/arm-cpu-data.h: Regenerated.
11301 * config/arm/arm-cpu.h: Regenerated.
11303 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11305 * Makefile.in (OBJS): Move sbitmap.o from here ...
11306 (OBJS-libcommon): ... to here.
11308 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11310 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11311 (ISA_ALL_CRYPTO): New macro.
11312 (ISA_ALL_SIMD): New macro
11313 (ISA_ALL_FP): New macro.
11314 * config/arm/arm.c (fpu_bitlist): Update initializer.
11315 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11317 (arm9e): Add fpu. Add option for nofp
11318 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11319 (arm926ej-s, arm1026ej-s): Likewise.
11320 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11321 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11322 neon-fp16, neon-vfpv4, nofp and nosimd.
11323 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11324 (cortex-a8): Add fpu. Add option for nofp.
11325 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11326 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11327 (cortex-r4f): Add fpu.
11328 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11329 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11331 (cortex-r8): Likewise.
11332 (cortex-m4): Add fpu. Add option for nofp.
11333 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11334 (cortex-a17.cortex-a7): Likewise.
11335 (cortex-a32): Add fpu. Add options for crypto and nofp.
11336 (cortex-a35, cortex-a53): Likewise.
11337 (cortex-a57): Add fpu. Add option for crypto.
11338 (cortex-a72, cortex-a73): Likewise.
11339 (exynos-m1): Likewise.
11340 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11341 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11342 (cortex-m33): Add fpu. Add option for nofp.
11343 * config/arm/arm-cpu-cdata.h: Regenerated
11344 * config/arm/arm-cpu-data.h: Regenerated.
11346 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11348 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11349 (armv5te, armv5tej): Likewise.
11350 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11351 (armv7): Add options fp and vfpv3-d16.
11352 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11353 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11355 (armv7ve): Likewise.
11356 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11357 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11358 (armv8-a): Add nocrypto option.
11359 (armv8.1-a, armv8.2-a): Likewise.
11360 (armv8-m.main): add options fp, fp.dp and nofp.
11362 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11364 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11366 (armv8-a+crc): Delete.
11367 (armv8.1-a): Add options simd, crypto and nofp.
11368 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11369 (armv8.2-a+fp16): Delete.
11370 (armv8-m.main): Add option dsp.
11371 (armv8-m.main+dsp): Delete.
11372 (cortex-a8): Add fpu. Add nofp option.
11373 (cortex-a9): Add fpu. Add nofp and nosimd options.
11374 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11375 link to main cpu and architecture data structures.
11376 (gen_comm_data): Only put isa attributes from the main architecture
11378 (option): New statement for architecture and CPU entries.
11379 * arm.c (struct cpu_option): New structure.
11380 (struct processors): Add entry for options.
11381 (arm_unrecognized_feature): New function.
11382 (arm_parse_arch_cpu_name): Ignore any characters after the first
11384 (arm_parse_arch_cpu_feature): New function.
11385 (arm_configure_build_target): Separate out any CPU and architecture
11386 features and parse separately. Don't error out if -mfpu=auto is
11387 used with only an architecture string.
11388 (arm_print_asm_arch_directives): New function.
11389 (arm_file_start): Call it.
11390 * config/arm/arm-cpu-cdata.h: Regenerated.
11391 * config/arm/arm-cpu-data.h: Likewise.
11392 * config/arm/arm-tables.opt: Likewise.
11394 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11396 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11397 assembler when it is not -mfpu=auto.
11399 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11401 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11402 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11403 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11404 (ASM_CPU_SPEC): Rewrite.
11405 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11406 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11407 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11408 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11409 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11410 copied string is NUL-terminated. Also strip any characters prefixed
11412 (arm_rewrite_selected_arch): New function.
11413 (arm_rewrite_march): New function.
11415 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11417 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11418 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11419 (march, mcpu, mtune): Convert to string-based options.
11420 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11421 (arm_parse_arch_cpu_name): New function.
11422 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11423 identify selected architecture or CPU.
11424 (arm_option_save): New function.
11425 (TARGET_OPTION_SAVE): Redefine.
11426 (arm_option_restore): Restore string options.
11427 (arm_option_print): Print string options.
11429 2017-06-16 Martin Sebor <msebor@redhat.com>
11431 PR tree-optimization/80933
11432 PR tree-optimization/80934
11433 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11434 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11435 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11436 (gimple_fold_builtin): Call them.
11438 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11440 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11441 as unlikely; update profile.
11443 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11445 * predict.c (force_edge_cold): Handle declaring edges impossible
11448 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11450 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11452 (try_unroll_loop_completely): Fix reporting.
11454 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11456 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11458 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11461 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11462 if given a non-constant argument for an intrinsic which requires a
11465 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11467 * profile.c (compare_freqs): New function.
11468 (branch_prob): Sort edge list.
11469 (find_spanning_tree): Assume that the list is priority sorted.
11471 2017-06-16 Richard Biener <rguenther@suse.de>
11473 PR tree-optimization/81090
11474 * passes.def (pass_record_bounds): Remove.
11475 * tree-pass.h (make_pass_record_bounds): Likewise.
11476 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11477 make_pass_record_bounds): Likewise.
11478 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11479 not free niter estimates at the beginning but at the end.
11480 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11482 2017-06-16 Richard Biener <rguenther@suse.de>
11484 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11485 initializer to workaround ICE in host GCC 4.8.
11487 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11489 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11491 (clone_inlined_nodes): Update.
11493 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11495 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11496 prefetch settings, and enable prefetching by default at -O3.
11498 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11500 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11501 Set flag_prefetch_loop_arrays according to tuning data.
11503 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11505 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11506 New tune structure.
11507 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11508 [Unrelated to main purpose of the patch] Place the pointer field last
11509 to enable type checking errors when tune structure are wrongly merged.
11510 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11511 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11512 (thunderx2t99_prefetch_tune): New tune constants.
11513 (tune_params *_tunings): Update all tunings (no functional change).
11514 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11515 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11516 from tunings structures.
11518 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11521 * ubsan.c (instrument_null): Add T argument, use it instead
11522 of computing it based on IS_LHS.
11523 (instrument_object_size): Likewise.
11524 (pass_ubsan::execute): Adjust instrument_null and
11525 instrument_object_size callers to pass gimple_get_lhs or
11526 gimple_assign_rhs1 result to it. Use instrument_null instead of
11527 calling get_base_address and instrument_mem_ref. Handle
11528 aggregate call arguments for object-size sanitization.
11530 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11532 PR tree-optimization/81089
11533 * tree-vrp.c (is_masked_range_test): Validate operands of
11536 2017-06-15 Martin Sebor <msebor@redhat.com>
11539 * dumpfile.c (dump_register): Avoid calling memset to initialize
11540 a class with a default ctor.
11541 * gcc.c (struct compiler): Remove const qualification.
11542 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11543 * hash-table.h: Ditto.
11544 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11546 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11547 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11549 * params.h (struct param_info): Make struct members non-const.
11550 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11551 with default initialization.
11552 * vec.h (vec_copy_construct, vec_default_construct): New helper
11554 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11555 with vec_copy_construct.
11556 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11557 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11558 * doc/invoke.texi (-Wclass-memaccess): Document.
11560 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11562 * emit-rtl.h (is_leaf): Update comment about local
11563 register allocator.
11565 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11568 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11569 for a variable to have a section before checking if the section has a
11571 Set section to.persistent if persistent attribute is set.
11572 Warn if .persistent attribute is used on an automatic variable.
11574 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11576 PR rtl-optimization/80474
11577 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11579 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11581 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11584 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11586 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11587 (casesi): Emit bounds checking as RTL.
11588 (casesi_internal_mips16_<mode>): Remove bounds checking.
11590 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11592 * config/xtensa/xtensa.c (xtensa_option_override): Append
11593 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11594 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11595 xtensa_doloop_hooks): Define unconditionally.
11596 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11598 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11599 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11600 for it in xtensa_option_override.
11601 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11602 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11604 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11606 * doc/cppopts.texi: Document '-' special value to -MF.
11608 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11610 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11611 (cortex_a53_fconst): Likewise.
11612 (cortex_a53_fpmul): Likewise.
11613 (cortex_a53_f_load_64): Likewise.
11614 (cortex_a53_f_load_many): Likewise.
11615 (cortex_a53_advsimd_alu): Likewise.
11616 (cortex_a53_advsimd_alu_q): Likewise.
11617 (cortex_a53_advsimd_mul): Likewise.
11618 (cortex_a53_advsimd_mul_q): Likewise.
11619 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11620 Add missing fmul, r2f_cvt and fconst cases.
11622 2017-06-14 Richard Biener <rguenther@suse.de>
11624 PR middle-end/81088
11625 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11627 (fold_binary_loc): When associating do not treat pre-existing
11628 TREE_OVERFLOW on literal constants as a reason to allow
11629 TREE_OVERFLOW on associated literal constants.
11631 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11633 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11634 (MASK_FEATURES): New macro.
11635 * config/sparc/sparc.c (sparc_option_override): Remove the special
11636 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11638 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11640 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11641 a division of 0 if non-call exceptions are enabled.
11643 2017-06-14 Andrew Pinski <apinski@cavium.com>
11644 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11647 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11648 Improve vector initialization code gen for only variable case.
11650 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11652 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11654 2017-06-14 Richard Biener <rguenther@suse.de>
11656 PR tree-optimization/81083
11657 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11660 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11662 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11663 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11664 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11665 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11666 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11667 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11669 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11671 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11672 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11674 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11676 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11678 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11680 * config/rs6000/t-rtems: Don't handle SPE.
11682 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11684 * config/rs6000/t-linux: Don't handle SPE.
11686 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11688 * config/rs6000/eabispe.h: Delete file.
11690 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11692 * config/rs6000/t-spe: Delete file.
11694 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11696 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11697 (rs6000_legitimate_offset_address_p): Return false for anything in
11698 V2SImode or V2SFmode.
11700 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11702 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11703 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11704 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11706 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11707 (rs6000_legitimate_offset_address_p): Ditto.
11708 (rs6000_emit_move): Ditto.
11709 (rs6000_init_builtins): Remove V4HI_type_node.
11711 2017-06-13 Martin Liska <mliska@suse.cz>
11714 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11715 (gate_asan): Likewise.
11716 * asan.h (asan_no_sanitize_address_p): Remove the function.
11717 (sanitize_flags_p): New function.
11718 * builtins.def: Fix coding style.
11719 * common.opt: Use renamed enum value.
11720 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11721 * doc/extend.texi: Document no_sanitize attribute.
11722 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11723 to SANITIZE_UNDEFINED_NONDEFAULT.
11724 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11725 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11726 Use sanitize_flags_p.
11727 * gimplify.c (gimplify_function_tree): Likewise.
11728 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11729 * opts.c (parse_no_sanitize_attribute): New function.
11730 (common_handle_option): Use renamed enum value.
11731 * opts.h (parse_no_sanitize_attribute): Declare.
11732 * tree.c (sanitize_flags_p): New function.
11733 * tree.h: Declared here.
11734 * tsan.c: Use sanitize_flags_p.
11735 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11736 (instrument_mem_ref): Likewise.
11737 (instrument_bool_enum_load): Likewise.
11738 (do_ubsan_in_current_function): Remove the function.
11739 (pass_ubsan::execute): Use sanitize_flags_p.
11740 * ubsan.h: Remove do_ubsan_in_current_function
11741 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11742 (dump_function_to_file): Use it here.
11744 2017-06-13 Martin Jambor <mjambor@suse.cz>
11746 PR tree-optimization/80803
11747 PR tree-optimization/81063
11748 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11749 (propagate_subaccesses_across_link): Enqueue subtree whenever
11750 necessary instead of relying on the caller.
11752 2017-06-13 Martin Jambor <mjambor@suse.cz>
11754 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11755 that have a first_link.
11756 (sort_and_splice_var_accesses): Do not check first_link before
11758 (subtree_mark_written_and_enqueue): Likewise.
11759 (propagate_all_subaccesses): Likewise and do not stop at first
11760 parent with a first_link.
11762 2017-06-13 Martin Jambor <mjambor@suse.cz>
11764 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11767 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11769 * match.pd: New pattern.
11771 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11773 * tree-vrp.c (is_masked_range_test): New function.
11774 (register_edge_assert_for): Determine ranges for
11777 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11779 PR tree-optimization/67328
11780 * fold-const.c (maskable_range_p): New function.
11781 (build_range_check): Generate bittests if possible.
11783 2017-06-13 Martin Liska <mliska@suse.cz>
11785 * gimple-pretty-print.c (dump_probability): Add new argument.
11786 (dump_edge_probability): Dump both probability and count.
11787 (dump_gimple_label): Likewise.
11788 (dump_gimple_bb_header): Likewise.
11790 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11793 * config/avr/avr-devices.c: Fix indentation.
11794 * config/avr/gen-avr-mmcu-specs.c: Dito.
11796 2017-06-13 Richard Biener <rguenther@suse.de>
11798 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11799 instead get vector type from stmt_info.
11800 (vectorizable_reduction): Adjust. Remove dead code.
11802 2017-06-13 Richard Biener <rguenther@suse.de>
11804 PR middle-end/81065
11805 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11806 case of C * (x * C2 + C3).
11807 (fold_addr_of_array_ref_difference): Properly fold index difference.
11809 2017-06-12 David S. Miller <davem@davemloft.net>
11812 * config/sparc/sparc.md (return expander): Emit frame blockage if
11813 function uses alloca.
11815 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11817 * combine.c (make_field_assignment): Check len rather than the mode
11818 precision when calling force_to_mode.
11820 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11822 Support multilibs and devices that see flash in RAM address range.
11825 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11826 (avr_mcu_t) <flash_pm_offset>: New field.
11827 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11828 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11829 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11830 (AVR_TINY_PM_OFFSET): Remove macro.
11831 * config/avr/avr.opt (-mshort-calls): New option.
11832 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11833 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11834 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11835 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11836 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11837 instead of avr_arch->have_jmp_call.
11838 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11839 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11840 avr_arch->flash_pm_offset to define.
11841 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11842 new field flash_pm_offset. Add entry for avrxmega3.
11843 (avr_texinfo): Add entry for avrxmega3.
11844 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11845 attiny212, attiny214,
11846 attiny412, attiny414, attiny416, attiny417,
11847 attiny814, attiny816, attiny817,
11848 attiny1614, attiny1616, attiny1617,
11849 attiny3214, attiny3216, attiny3217.
11850 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11851 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11852 (avr_print_operand_address) [AVR_TINY]: Same.
11853 (avr_asm_init_sections) <readonly_data_section>: Only patch
11854 callback if avr_arch->flash_pm_offset = 0.
11855 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11856 for rodata if avr_arch->flash_pm_offset != 0.
11857 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11858 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11859 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11860 (m_options): Append opt_rcall.
11861 (m_dirnames): Append dir_rcall.
11862 * config/avr/t-multilib: Regenerate.
11864 * configure.ac [target=avr]: Check whether avrxmega3 default
11865 linker description file works as needed.
11866 * configure: Regenerate.
11867 * doc/avr-mmcu.texi: Regenerate.
11868 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11869 <__AVR_ARCH__>: Document avrxmega3 and 103.
11870 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11871 <__AVR_SHORT_CALLS__>: Document it.
11872 <__AVR_PM_BASE_ADDRESS__>: Document it.
11873 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11874 (AVR Variable Attributes) <progmem>: Document this is
11875 not needed for avrxmega3.
11876 (AVR Named Address Spaces) <__flash>: Dito.
11878 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11880 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11882 2017-06-12 Doug Rupp <rupp@adacore.com>
11884 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11885 Append vxworks-stdint.h to the tm_file list.
11886 * config/vxworks-stdint.h: New file.
11888 2017-06-12 Martin Liska <mliska@suse.cz>
11890 PR tree-optimization/81041
11891 * tree-profile.c (gimple_gen_ic_func_profiler):
11892 Create an extra BB in profile-generate
11893 (gimple_gen_time_profiler): Likewise.
11895 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11897 PR tree-optimization/81003
11898 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11899 (update_range_test): Use it instead of force_gimple_operand_gsi.
11901 2017-06-12 Richard Biener <rguenther@suse.de>
11903 PR tree-optimization/81053
11904 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11905 with backedge value not defined in loop. Simplify def stmt
11908 2017-06-11 Tom de Vries <tom@codesourcery.com>
11911 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11913 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11914 nvptx_cannot_force_const_mem.
11916 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11918 * opts.c (finish_options): Move test for flag_split_stack after
11919 it has been initialized.
11921 2017-06-11 Jason Merrill <jason@redhat.com>
11923 * tree.h (id_equal): New.
11924 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11925 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11926 instead of strcmp of IDENTIFIER_POINTER.
11928 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11930 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11931 (mark_all_inlined_calls_cdtor): Fix formating.
11932 (inline_transform): Rescale profile before inlining.
11934 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11936 * cgraph.h (cgraph_edge::clone): Update prototype.
11937 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11938 (cgraph_node::create_clone): Update.
11939 (cgraph_node::create_version_clone): Update.
11940 * tree-inline.c (copy_bb): Update.
11941 (expand_call_inline): Update.
11943 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11945 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11946 factored out from ...
11947 (rs6000_emit_prologue): ... here.
11949 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11951 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11952 factored out from ...
11953 (rs6000_emit_prologue): ... here.
11955 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11957 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11959 (handle_missing_profiles): Fix computation of tp_first_run.
11960 (counts_to_freqs): Do not touch freqs when count is 0.
11962 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11964 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
11967 2017-06-10 Tom de Vries <tom@codesourcery.com>
11969 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
11970 attributes): Document signal effective target.
11972 2017-06-10 Tom de Vries <tom@codesourcery.com>
11974 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
11975 Document effective target stack_size.
11977 2017-06-09 David Malcolm <dmalcolm@redhat.com>
11979 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
11980 to the edit_context if they can be auto-applied.
11982 2017-06-9 Ian Lance Taylor <iant@golang.org>
11984 * opts.c (finish_options): If -fsplit-stack, disable implicit
11985 -forder-blocks-and-partition.
11986 * doc/invoke.texi (Optimize Options): Document that when using
11987 -fsplit-stack -forder-blocks-and-partition is not implicitly
11990 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11992 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
11993 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
11994 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
11995 * builtins.def (abort, trap, unreachable): Declare cold.
11996 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
11997 * tree-core.h (ECF_COLD): New.
11998 * tree.c (set_call_expr_flags): Handle ECF_COLD.
11999 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12001 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12003 * predict.c (unlikely_executed_stmt_p): Cleanup.
12005 2017-06-09 Richard Biener <rguenther@suse.de>
12007 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12008 model if the ref is always written to.
12010 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12012 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12014 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12016 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12019 2017-06-09 Tom de Vries <tom@codesourcery.com>
12022 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12023 "target cannot support label values" when encountering LABEL_REF.
12025 2017-06-09 Martin Liska <mliska@suse.cz>
12027 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12028 (gimple_gen_ic_func_profiler): Emit direct comparison
12029 of __gcov_indirect_call_callee with NULL.
12030 (gimple_gen_time_profiler): Change probability from
12031 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12033 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12035 * profile.c (edge_gcov_counts): Turn to pointer.
12036 (compute_branch_probabilities, compute_branch_probabilities): Update.
12037 (branch_prob): Do not clear edge_gcov_count.
12038 * profile.h (edge_gcov_counts): Turn to pointer.
12039 (edge_gcov_count): Update.
12041 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12043 * gimple.h (gimple_check_failed): Mark cold.
12045 2017-06-09 Richard Biener <rguenther@suse.de>
12047 PR tree-optimization/66623
12048 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12049 refactor check_reduction into two parts, properly computing
12050 whether we have to check reduction validity for outer loop
12053 2017-06-09 Richard Biener <rguenther@suse.de>
12055 PR tree-optimization/79483
12056 * graphite-scop-detection.c (order): New global.
12057 (get_order): Compute bb to order mapping that satisfies code
12058 generation constraints.
12059 (cmp_pbbs): New helper.
12060 (build_scops): Start domwalk at entry block, sort generated
12063 2017-06-09 Richard Biener <rguenther@suse.de>
12065 PR middle-end/81007
12066 * ipa-polymorphic-call.c
12067 (ipa_polymorphic_call_context::restrict_to_inner_class):
12068 Skip FIELD_DECLs with error_mark_node type.
12069 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12072 2017-06-09 Martin Liska <mliska@suse.cz>
12074 * predict.c (struct branch_predictor): New struct.
12075 (test_prediction_value_range): New test.
12076 (predict_c_tests): New function.
12077 * selftest-run-tests.c (selftest::run_tests): Run the function.
12078 * selftest.h: Declare new tests.
12080 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12083 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12084 gen_add3_insn did not fail.
12085 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12086 r0, construct that number in a temporary reg and add that reg to r0.
12087 If asked to put the result in r0 as well, fail.
12089 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12091 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12092 for early expansion of vec_eqv.
12094 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12096 PR middle-end/81005
12097 * ubsan.c (instrument_null): Avoid pointless code temporary.
12098 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12100 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12104 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12107 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12110 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12112 * predict.c (unlikely_executed_edge_p): Move ahead.
12113 (probably_never_executed_edge_p): Use it.
12115 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12117 PR middle-end/79988
12118 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12119 gimple_call_builtin_p call.
12121 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12123 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12124 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12125 rtl_check_failed_type2, rtl_check_failed_code1,
12126 rtl_check_failed_code2, rtl_check_failed_code_mode,
12127 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12128 rtvec_check_failed_bounds, rtl_check_failed_flag,
12129 _fatal_insn_not_found, _fatal_insn): Likewise.
12130 * tree.h (tree_contains_struct_check_failed,
12131 tree_check_failed, tree_not_check_failed,
12132 tree_class_check_failed, tree_range_check_failed,
12133 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12134 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12135 tree_operand_check_failed, omp_clause_check_failed,
12136 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12139 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12141 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12142 flag_branch_probabilities.
12143 * ipa-inline.c (edge_badness): Likewise.
12144 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12145 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12146 * predict.c (maybe_hot_frequency_p): Likewise.
12147 (probably_never_executed): Likewise.
12148 * sched-ebb.c (schedule_ebbs): Likewise.
12149 * sched-rgn.c (find_single_block_region): Likewise.
12150 * tracer.c (tail_duplicate): Likewise.
12152 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12154 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12155 longer requires x_flag_profile_use.
12157 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12159 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12160 instead of flag_reorder_blocks_and_partition.
12161 * dbxout.c (dbxout_function_end): Likewise.
12162 * dwarf2out.c (gen_subprogram_die): Likewise.
12163 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12164 * hw-doloop.c (reorg_loops): Likewise.
12165 * varasm.c (assemble_start_function,
12166 assemble_end_function): Likewise.
12167 (decide_function_section): Do not check for
12168 flag_reorder_blocks_and_partition.
12170 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12172 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12174 (chkp_get_hard_register_fake_addr_expr): Ditto.
12175 (chkp_build_addr_expr): Add check for hard reg case.
12176 (chkp_parse_array_and_component_ref): Ditto.
12177 (chkp_find_bounds_1): Ditto.
12178 (chkp_process_stmt): Don't generate bounds store for
12181 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12183 * predict.c (maybe_hot_bb_p): Do not check profile status.
12184 (maybe_hot_edge_p): Likewise.
12185 (probably_never_executed): Check for zero counts even if profile
12187 (unlikely_executed_edge_p): New function.
12188 (unlikely_executed_stmt_p): New function.
12189 (unlikely_executed_bb_p): New function.
12190 (set_even_probabilities): Use unlikely predicates.
12191 (combine_predictions_for_bb): Likewise.
12192 (predict_paths_for_bb): Likewise.
12193 (predict_paths_leading_to_edge): Likewise.
12194 (determine_unlikely_bbs): New function.
12195 (estimate_bb_frequencies): Use it.
12196 (compute_function_frequency): Use zero counts even if profile is
12198 * profile-count.h: Fix typo.
12200 2017-08-08 Julia Koval <julia.koval@intel.com>
12202 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12203 _mm512_mask_cvtsepi16_storeu_epi8,
12204 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12205 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12206 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12207 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12208 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12209 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12210 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12211 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12212 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12213 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12214 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12215 __builtin_ia32_pmovuswb256mem_mask,
12216 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12217 __builtin_ia32_pmovwb512mem_mask): New builtins.
12219 2017-08-08 Julia Koval <julia.koval@intel.com>
12221 PR target/73350,80862
12222 * config/i386/subst.md (round): Fix round pattern.
12223 * config/i386/i386.c (ix86_erase_embedded_rounding):
12224 Fix erasing rounding for the fixed pattern.
12226 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12228 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12230 2017-06-08 Martin Liska <mliska@suse.cz>
12232 PR gcov-profile/80911
12233 * gcov.c (block_info::block_info): New constructor.
12235 2017-06-07 Carl Love <cel@us.ibm.com>
12237 * config/rs6000/rs6000-c: The return type of the following
12238 built-in functions was implemented as int not long long. Fix sign
12239 of return value for the unsigned version of vec_mulo and vec_mule.
12240 vector unsigned long long vec_bperm (vector unsigned long long,
12241 vector unsigned char)
12242 vector signed long long vec_mule (vector signed int,
12244 vector unsigned long long vec_mule (vector unsigned int,
12245 vector unsigned int)
12246 vector signed long long vec_mulo (vector signed int,
12248 vector unsigned long long vec_mulo (vector unsigned int,
12249 vector unsigned int)
12250 * doc/extend.texi: Fix the documentation for the built-in
12253 2017-06-07 Carl Love <cel@us.ibm.com>
12256 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12259 2017-06-07 Carl Love <cel@us.ibm.com>
12261 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12262 support, Generate doublehv for signed int/float for BE case only.
12264 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12266 * doc/invoke.texi (mcx16): Rewrite.
12268 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12270 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12271 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12272 *mov<mode>_softfloat, and an anonymous splitter): Use
12273 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12275 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12277 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12279 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12280 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12281 (rs6000_debug_reg_global): Adjust.
12282 (rs6000_init_hard_regno_mode_ok): Adjust.
12283 (rs6000_dbx_register_number): Adjust.
12284 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12285 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12286 Remove SPE_ACC and SPEFSCR.
12287 (REG_ALLOC_ORDER): Ditto.
12288 (FRAME_POINTER_REGNUM): Change to 111.
12289 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12290 (REG_CLASS_NAMES): Ditto.
12291 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12292 (REGISTER_NAMES): Ditto.
12293 (ADDITIONAL_REG_NAMES): Ditto.
12294 (rs6000_reg_names): Ditto.
12295 * config/rs6000/rs6000.md: Renumber some register number
12298 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12300 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12302 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12303 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12305 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12306 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12307 Delete the SPE high registers.
12308 (REG_ALLOC_ORDER): Ditto.
12309 (enum reg_class): Remove SPE_HIGH_REGS.
12310 (REG_CLASS_NAMES): Ditto.
12311 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12312 (REGISTER_NAMES): Ditto.
12313 (rs6000_reg_names): Ditto.
12314 * doc/tm.texi.in: Remove SPE as example.
12315 * doc/tm.texi: Regenerate.
12317 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12319 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12320 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12321 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12322 * config/rs6000/rs6000.md (type): Remove "brinc".
12324 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12326 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12327 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12328 * config/rs6000/linuxspe.h: Delete file.
12329 * config/rs6000/rs6000.md: Don't include spe.md.
12330 * config/rs6000/spe.h: Delete file.
12331 * config/rs6000/spe.md: Delete file.
12332 * config/rs6000/t-rs6000: Remove spe.md.
12334 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12336 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12337 (reg_or_none500mem_operand): Delete.
12338 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12339 instead of reg_or_none500mem_operand.
12341 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12343 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12344 handling of SPE flags.
12345 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12347 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12349 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12351 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12352 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12353 paired_divv2sf3): Similar.
12354 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12355 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12356 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12358 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12359 Rename the paired_* instruction patterns.
12360 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12362 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12363 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12364 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12365 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12366 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12367 PAIRED_VECTOR_MODE.
12368 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12369 (spe_func_has_64bit_regs_p): Delete.
12370 (spe_expand_predicate_builtin): Delete.
12371 (spe_expand_evsel_builtin): Delete.
12372 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12373 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12374 (invalid_e500_subreg): Delete.
12375 (rs6000_legitimize_address): Always force_reg op2 as well, for
12376 paired single memory accesses.
12377 (rs6000_member_type_forces_blk): Delete.
12378 (rs6000_spe_function_arg): Delete.
12379 (rs6000_expand_unop_builtin): Delete SPE handling.
12380 (rs6000_expand_binop_builtin): Ditto.
12381 (spe_expand_stv_builtin): Delete.
12382 (bdesc_2arg_spe): Delete.
12383 (spe_expand_builtin): Delete.
12384 (spe_expand_predicate_builtin): Delete.
12385 (spe_expand_evsel_builtin): Delete.
12386 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12387 (spe_init_builtins): Delete.
12388 (spe_func_has_64bit_regs_p): Delete.
12389 (savres_routine_name): Delete "info" parameter. Adjust callers.
12390 (rs6000_emit_stack_reset): Ditto.
12391 (rs6000_dwarf_register_span): Delete.
12392 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12393 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12394 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12396 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12398 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12399 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12400 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12401 mulv2sf3, divv2sf3): Delete expanders.
12403 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12405 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12407 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12409 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12410 * config/rs6000/rs6000.c: Ditto.
12412 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12414 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12415 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12416 comparison_operator.
12418 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12420 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12421 * config/rs6000/rs6000.opt: Ditto.
12422 * config/rs6000/t-rtems: Ditto.
12424 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12426 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12427 TARGET_E500_SINGLE by 0, simplify.
12428 * config/rs6000/rs6000.c: Ditto.
12429 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12430 (spe_build_register_parallel): Delete.
12431 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12432 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12433 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12434 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12435 (E500_CONVERT): Delete.
12436 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12438 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12440 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12441 * config/rs6000/dfp.md: Ditto.
12442 (negdd2, *negdd2_fpr): Merge.
12443 (absdd2, *absdd2_fpr): Merge.
12444 (negtd2, *negtd2_fpr): Merge.
12445 (abstd2, *abstd2_fpr): Merge.
12446 * config/rs6000/e500.h: Delete file.
12447 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12448 TARGET_FPRS by 1 and simplify.
12449 * config/rs6000/rs6000-c.c: Ditto.
12450 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12451 TARGET_DF_SPE by 0.
12452 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12454 * config/rs6000/rs6000.md: Ditto.
12455 (floatdidf2, *floatdidf2_fpr): Merge.
12456 (move_from_CR_gt_bit): Delete.
12457 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12458 (E500_CR_IOR_COMPARE): Delete.
12459 (All patterns that require !TARGET_FPRS): Delete.
12460 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12462 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12464 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12466 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12468 * graphds.c (add_edge): Intitialize edge's attached data.
12469 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12470 pointer parameter. Call pointed function on each edge during
12471 graph traversing. Skip traversing the edge when the function
12473 (graphds_dfs, graphds_scc): Ditto.
12474 (for_each_edge): New parameter. Pass the new parameter to callback
12476 * graphds.h (skip_edge_callback): New function pointer type.
12477 (graphds_dfs, graphds_scc): New function pointer parameter.
12478 (graphds_edge_callback, for_each_edge): New parameter.
12480 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12482 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12483 out code checking if runtime alias check is possible to below ...
12484 Call the new function.
12485 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12486 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12488 2017-06-07 Marek Polacek <polacek@redhat.com>
12491 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12492 TYPE_OVERFLOW_WRAPS check.
12494 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12496 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12497 if versioning is required.
12498 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12499 peeling with the check for versioning.
12501 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12503 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12504 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12505 Set true to new parameter if new ssa variable is defined.
12506 (vect_gen_vector_loop_niters): Refactor. Set range information
12507 for the new vector loop bound variable.
12508 (vect_do_peeling): Ditto.
12510 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12512 * tree-affine.c (ssa.h): Include header file.
12513 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12514 has wrapping overflow behavior.
12516 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12518 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12520 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12522 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12523 (tree_to_aff_combination): ... here.
12525 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12527 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12528 reg_pressure model function.
12529 (ivopts_global_cost_for_size): Delete.
12530 (determine_set_costs, iv_ca_recount_cost): Call new model function
12531 ivopts_estimate_reg_pressure.
12533 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12535 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12536 expensive than udiv. Remove floating point cases from mod.
12538 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12540 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12541 Increase idiv cost.
12543 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12545 * config/aarch64/aarch64.md
12546 (copysignsf3): Fix mask generation.
12548 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12550 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12552 (class dump_manager): Add register_dumps method.
12553 * dumpfile.c: Include langhooks.h.
12554 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12555 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12556 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12557 (dump_manager::dump_register): Start with 512 entries instead of 32.
12558 (dump_manager::register_dumps): New method.
12559 * toplev.c (general_init): Instead of invoking register_dumps
12560 langhook, invoke register_dumps method on the dump manager.
12561 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12564 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12566 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12567 Say that # requires an associated define_split to exist, and that
12568 the define_split must be suitable for use after register allocation.
12570 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12572 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12573 (compute_outgoing_frequencies): Also initialize zero counts.
12574 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12575 around loops; preserve more of profile when nothing changes.
12577 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12579 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12581 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12582 * config/arm/arm-cpu-cdata.h: Regenerate.
12583 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12584 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12585 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12586 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12587 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12589 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12591 * config/arm/t-rmprofile: Likewise.
12592 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12594 2017-06-06 David S. Miller <davem@davemloft.net>
12597 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12598 blockage if function uses alloca.
12600 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12602 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12603 New "uid" fields to hold pretty-print IDs of group and ref.
12604 Memory references are now identified as <group_id>:<ref_id>
12605 instead of using [random] addresses.
12606 (dump_mem_details): Simplify, no functional change.
12607 (dump_mem_ref): Simplify and make output more concise.
12608 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12609 (find_or_create_group): Initialize group uid.
12610 (record_ref): Initialize ref uid. Improve debug output.
12611 (prune_group_by_reuse, should_issue_prefetch_p,)
12612 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12613 (mark_nontemporal_store, determine_loop_nest_reuse):
12614 Improve debug output.
12616 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12618 * dbgcnt.def (prefetch): New debug counter.
12619 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12620 (schedule_prefetches): Stop issueing prefetches if debug counter
12623 2017-06-06 Tom de Vries <tom@codesourcery.com>
12625 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12626 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12628 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12630 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12631 Use aarch64_reg_or_zero predicate for operand 4.
12632 (aarch64_compare_and_swap<mode> define_insn_and_split):
12633 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12634 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12636 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12638 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12639 (arm_compute_save_core_reg_mask): This.
12640 (thumb1_compute_save_reg_mask): Rename into ...
12641 (thumb1_compute_save_core_reg_mask): This.
12642 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12643 (arm_compute_frame_layout): Likewise.
12645 2017-06-06 Richard Biener <rguenther@suse.de>
12647 PR tree-optimization/80974
12648 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12649 keep or clear leaders SSA info.
12651 2017-06-06 Tom de Vries <tom@codesourcery.com>
12653 * config/nvptx/nvptx.c (split_mode_p): New function.
12654 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12656 2017-06-06 Tom de Vries <tom@codesourcery.com>
12658 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12660 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12663 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12666 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12668 * shrink-wrap.c (handle_simple_exit): Update profile.
12669 (try_shrink_wrapping): Upate profile.
12671 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12673 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12674 (tree_guess_outgoing_edge_probabilities): New.
12675 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12676 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12678 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12680 * ipa-split.c (split_function): Initialize return bb profile.
12682 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12684 * profile.c (compute_branch_probabilities): Also initialize
12685 EXIT_BLOCK profile.
12687 2017-06-06 Richard Biener <rguenther@suse.de>
12689 PR tree-optimization/80928
12690 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12691 (vect_analyze_loop_operations): Properly guard analysis for
12693 (vect_transform_loop): Likewise.
12694 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12695 (vect_model_induction_cost): Do not cost for pure SLP.
12696 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12697 of induction in inner loop vectorization.
12698 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12699 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12700 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12702 (vect_analyze_slp_cost_1): Cost induction.
12703 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12704 (vect_get_slp_vect_defs): Likewise.
12705 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12706 (vect_transform_stmt): Handle SLP reductions.
12707 * tree-vectorizer.h (vectorizable_induction): Adjust.
12709 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12711 * config/rs6000/rs6000.c (make_resolver_func): Update
12712 init_lowered_empty_function call.
12714 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12716 * doc/invoke.texi: Document the -fprofile-abs-path option.
12717 * common.opt (fprofile-abs-path): New option.
12718 * gcov-io.h (gcov_write_filename): Declare.
12719 * gcov-io.c (gcov_write_filename): New function.
12720 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12721 * profile.c (output_location): Likewise.
12723 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12725 * shring-wrap.c: Revert accidental commit.
12727 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12729 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12731 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12733 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12735 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12736 profile in callgraph edge.
12737 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12738 (apply_scale): Likewise.
12739 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12740 Also scale profile when inlining function with zero profile.
12741 (initialize_cfun): Update exit block profile even when it is zero.
12742 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12743 when profile is read.
12745 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12747 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12748 (CLONE_*): New constants to define the processors we can generate
12749 code for with the target_clone attribute.
12750 (rs6000_clone_map): New array to identify which clone processors
12751 the current program is running on.
12752 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12753 target_clone attribute.
12754 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12755 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12756 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12757 (cpu_expand_builtin): Add support for target_clone attribute.
12758 (rs6000_valid_attribute_p): Allow "default" attribute.
12759 (get_decl_name): New debug function to simplify printing the
12760 current function name in debugging statements.
12761 (rs6000_clone_priority): New functions to support the target_clone
12762 attribute, and be able to generate code to switch between ISA 2.05
12763 through ISA 3.0 (power6 through power9).
12764 (rs6000_compare_version_priority): Likewise.
12765 (rs6000_get_function_versions_dispatcher): Likewise.
12766 (make_resolver_func): Likewise.
12767 (add_condition_to_bb): Likewise.
12768 (dispatch_function_versions): Likewise.
12769 (rs6000_generate_version_dispatcher_body): Likewise.
12770 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12771 (fusion_gpr_load_p): Fix a spacing issue.
12772 * doc/extend.texi (Common Function Attributes): Document that the
12773 PowerPC supports the target_clone attribute.
12775 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12777 * config/arm/arm.h: explain F symbol found in description of ARM
12778 register allocation in its legend.
12780 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12782 * config/mips/frame-header-opt.c: Include profile-count.h.
12783 * config/riscv/riscv.c: Include profile-count.h
12785 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12787 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12789 (sm_set_flag_if_changed): Add bbs field.
12790 (execute_sm_if_changed_flag_set): Pass BBS.
12791 (execute_sm): Update.
12793 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12795 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12798 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12800 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12801 (peephole2): New peephole2 to emit the above.
12802 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12804 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12806 * config/aarch64/aarch64.c (define_peephole2 above
12807 *sub_<shift>_<mode>): New peephole.
12809 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12811 * config/i386/i386.c (make_resolver_func): Update.
12812 * Makefile.in: Add profile-count.h and profile-count.o
12813 * auto-profile.c (afdo_indirect_call): Update to new API.
12814 (afdo_set_bb_count): Update.
12815 (afdo_propagate_edge): Update.
12816 (afdo_propagate_circuit): Update.
12817 (afdo_calculate_branch_prob): Update.
12818 (afdo_annotate_cfg): Update.
12819 * basic-block.h: Include profile-count.h
12820 (struct edge_def): Turn count to profile_count.
12821 (struct basic_block_def): Likewie.
12822 (REG_BR_PROB_BASE): Move to profile-count.h
12823 (RDIV): Move to profile-count.h
12824 * bb-reorder.c (max_entry_count): Turn to profile_count.
12825 (find_traces): Update.
12826 (rotate_loop):Update.
12827 (connect_traces):Update.
12828 (sanitize_hot_paths):Update.
12829 * bt-load.c (migrate_btr_defs): Update.
12830 * cfg.c (RDIV): Remove.
12831 (init_flow): Use alloc_block.
12832 (alloc_block): Uninitialize count.
12833 (unchecked_make_edge): Uninitialize count.
12834 (check_bb_profile): Update.
12835 (dump_edge_info): Update.
12836 (dump_bb_info): Update.
12837 (update_bb_profile_for_threading): Update.
12838 (scale_bbs_frequencies_int): Update.
12839 (scale_bbs_frequencies_gcov_type): Update.
12840 (scale_bbs_frequencies_profile_count): New.
12841 * cfg.h (update_bb_profile_for_threading): Update.
12842 (scale_bbs_frequencies_profile_count): Declare.
12843 * cfgbuild.c (compute_outgoing_frequencies): Update.
12844 (find_many_sub_basic_blocks): Update.
12845 * cfgcleanup.c (try_forward_edges): Update.
12846 (try_crossjump_to_edge): Update.
12847 * cfgexpand.c (expand_gimple_tailcall): Update.
12848 (construct_exit_block): Update.
12849 * cfghooks.c (verify_flow_info): Update.
12850 (dump_bb_for_graph): Update.
12851 (split_edge): Update.
12852 (make_forwarder_block): Update.
12853 (duplicate_block): Update.
12854 (account_profile_record): Update.
12855 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12856 (get_estimated_loop_iterations): Update.
12857 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12858 (single_likely_exit): Update.
12859 * cfgloopmanip.c (scale_loop_profile): Update.
12861 (set_zero_probability): Update.
12862 (lv_adjust_loop_entry_edge): Update.
12863 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12864 (purge_dead_edges): Update.
12865 (rtl_account_profile_record): Update.
12866 * cgraph.c (cgraph_node::create): Uninitialize count.
12867 (symbol_table::create_edge): Uninitialize count.
12868 (cgraph_update_edges_for_call_stmt_node): Update.
12869 (cgraph_edge::dump_edge_flags): Update.
12870 (cgraph_node::dump): Update.
12871 (cgraph_edge::maybe_hot_p): Update.
12872 * cgraph.h: Include profile-count.h
12873 (create_clone), create_edge, create_indirect_edge): Update.
12874 (cgraph_node): Turn count to profile_count.
12875 (cgraph_edge0: Likewise.
12876 (make_speculative, clone): Update.
12877 (create_edge): Update.
12878 (init_lowered_empty_function): Update.
12879 * cgraphclones.c (cgraph_edge::clone): Update.
12880 (duplicate_thunk_for_node): Update.
12881 (cgraph_node::create_clone): Update.
12882 * cgraphunit.c (cgraph_node::analyze): Update.
12883 (cgraph_node::expand_thunk): Update.
12884 * final.c (dump_basic_block_info): Update.
12885 * gimple-streamer-in.c (input_bb): Update.
12886 * gimple-streamer-out.c (output_bb): Update.
12887 * graphite.c (print_global_statistics): Update.
12888 (print_graphite_scop_statistics): Update.
12889 * hsa-brig.c: Include basic-block.h.
12890 * hsa-dump.c: Include basic-block.h.
12891 * hsa-gen.c (T sum_slice): Update.
12892 (convert_switch_statements):Update.
12893 * hsa-regalloc.c: Include basic-block.h.
12894 * ipa-chkp.c (chkp_produce_thunks): Update.
12895 * ipa-cp.c (struct caller_statistics): Update.
12896 (init_caller_stats): Update.
12897 (gather_caller_stats): Update.
12898 (ipcp_cloning_candidate_p): Update.
12899 (good_cloning_opportunity_p): Update.
12900 (get_info_about_necessary_edges): Update.
12901 (dump_profile_updates): Update.
12902 (update_profiling_info): Update.
12903 (update_specialized_profile): Update.
12904 (perhaps_add_new_callers): Update.
12905 (decide_about_value): Update.
12906 (ipa_cp_c_finalize): Update.
12907 * ipa-devirt.c (struct odr_type_warn_count): Update.
12908 (struct decl_warn_count): Update.
12909 (struct final_warning_record): Update.
12910 (possible_polymorphic_call_targets): Update.
12911 (ipa_devirt): Update.
12912 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12913 * ipa-icf.c (sem_function::merge): Update.
12914 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12915 * ipa-inline.c (compute_uninlined_call_time): Update.
12916 (compute_inlined_call_time): Update.
12917 (want_inline_small_function_p): Update.
12918 (want_inline_self_recursive_call_p): Update.
12919 (edge_badness): Update.
12920 (lookup_recursive_calls): Update.
12921 (recursive_inlining): Update.
12922 (inline_small_functions): Update.
12923 (dump_overall_stats): Update.
12924 (dump_inline_stats): Update.
12925 * ipa-profile.c (ipa_profile_generate_summary): Update.
12926 (ipa_propagate_frequency): Update.
12927 (ipa_profile): Update.
12928 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12929 * ipa-utils.c (ipa_merge_profiles): Update.
12930 * loop-doloop.c (doloop_modify): Update.
12931 * loop-unroll.c (report_unroll): Update.
12932 (unroll_loop_runtime_iterations): Update.
12933 * lto-cgraph.c (lto_output_edge): Update.
12934 (lto_output_node): Update.
12935 (input_node): Update.
12936 (input_edge): Update.
12937 (merge_profile_summaries): Update.
12938 * lto-streamer-in.c (input_cfg): Update.
12939 * lto-streamer-out.c (output_cfg): Update.
12940 * mcf.c (create_fixup_graph): Update.
12941 (adjust_cfg_counts): Update.
12942 (sum_edge_counts): Update.
12943 * modulo-sched.c (sms_schedule): Update.
12944 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12945 * predict.c (maybe_hot_count_p): Update.
12946 (probably_never_executed): Update.
12947 (dump_prediction): Update.
12948 (combine_predictions_for_bb): Update.
12949 (propagate_freq): Update.
12950 (handle_missing_profiles): Update.
12951 (counts_to_freqs): Update.
12952 (rebuild_frequencies): Update.
12953 (force_edge_cold): Update.
12954 * predict.h: Include profile-count.h
12955 (maybe_hot_count_p, counts_to_freqs): UPdate.
12956 * print-rtl-function.c: Do not include cfg.h
12957 * print-rtl.c: Include basic-block.h
12958 * profile-count.c: New file.
12959 * profile-count.h: New file.
12960 * profile.c (is_edge_inconsistent): Update.
12961 (correct_negative_edge_counts): Update.
12962 (is_inconsistent): Update.
12963 (set_bb_counts): Update.
12964 (read_profile_edge_counts): Update.
12965 (compute_frequency_overlap): Update.
12966 (compute_branch_probabilities): Update; Initialize and deinitialize
12968 (branch_prob): Update.
12969 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
12970 (edge_gcov_count): New.
12971 (bb_gcov_count): New.
12972 * shrink-wrap.c (try_shrink_wrapping): Update.
12973 * tracer.c (better_p): Update.
12974 * trans-mem.c (expand_transaction): Update.
12975 (ipa_tm_insert_irr_call): Update.
12976 (ipa_tm_insert_gettmclone_call): Update.
12977 * tree-call-cdce.c: Update.
12978 * tree-cfg.c (gimple_duplicate_sese_region): Update.
12979 (gimple_duplicate_sese_tail): Update.
12980 (gimple_account_profile_record): Update.
12981 (execute_fixup_cfg): Update.
12982 * tree-inline.c (copy_bb): Update.
12983 (copy_edges_for_bb): Update.
12984 (initialize_cfun): Update.
12985 (freqs_to_counts): Update.
12986 (copy_cfg_body): Update.
12987 (expand_call_inline): Update.
12988 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12989 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
12990 (try_unroll_loop_completely): Update.
12991 (try_peel_loop): Update.
12992 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12993 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
12994 * tree-ssa-loop-split.c (connect_loops): Update.
12995 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
12996 * tree-ssa-reassoc.c (branch_fixup): Update.
12997 * tree-ssa-tail-merge.c (replace_block_by): Update.
12998 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
12999 (compute_path_counts): Update.
13000 (update_profile): Update.
13001 (recompute_probabilities): Update.
13002 (update_joiner_offpath_counts): Update.
13003 (estimated_freqs_path): Update.
13004 (freqs_to_counts_path): Update.
13005 (clear_counts_path): Update.
13006 (ssa_fix_duplicate_block_edges): Update.
13007 (duplicate_thread_path): Update.
13008 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13009 (struct switch_conv_info): Update.
13010 * tree-tailcall.c (decrease_profile): Update.
13011 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13012 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13013 * value-prof.c (check_counter): Update.
13014 (gimple_divmod_fixed_value): Update.
13015 (gimple_mod_pow2): Update.
13016 (gimple_mod_subtract): Update.
13017 (gimple_ic_transform): Update.
13018 (gimple_stringop_fixed_value): Update.
13019 * value-prof.h (gimple_ic): Update.
13021 2017-06-02 Carl Love <cel@us.ibm.com>
13023 * config/rs6000/rs6000-c: Add support for built-in functions
13024 vector double vec_doublee (vector signed int);
13025 vector double vec_doublee (vector unsigned int);
13026 vector double vec_doublee (vector float);
13027 vector double vec_doubleh (vector signed int);
13028 vector double vec_doubleh (vector unsigned int);
13029 vector double vec_doubleh (vector float);
13030 vector double vec_doublel (vector signed int);
13031 vector double vec_doublel (vector unsigned int);
13032 vector double vec_doublel (vector float);
13033 vector double vec_doubleo (vector signed int);
13034 vector double vec_doubleo (vector unsigned int);
13035 vector double vec_doubleo (vector float);.
13036 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13037 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13039 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13040 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13041 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13043 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13044 vec_doublel, vec_doubleh.
13045 * doc/extend.texi: Update the built-in documentation file for the
13046 new built-in functions.
13048 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13051 * ipa-inline-analysis.c (free_growth_caches): Set
13052 edge_removal_hook_holder to NULL after removing it.
13054 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13056 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13057 comparision with zero.
13059 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13060 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13061 for early expansion of vec_min and vec_max builtins.
13062 (builtin_function_type): Add min/max unsigned variants to those
13063 identified as having unsigned arguments.
13065 2017-06-02 Olivier Hainque <hainque@adacore.com>
13067 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13069 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13071 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13072 Use VALL_F16 iterator rather than VALL.
13074 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13076 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13077 Emit CBNZ inside loop when doing a strong exchange and comparing
13078 against zero. Generate the CC flags after the loop.
13080 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13082 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13083 (dl_section_ref): New.
13084 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13085 On AIX, append an expression to subtract the size of the
13086 section length to dl_section_ref.
13088 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13090 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13091 for early expansion of vector absolute builtins.
13093 2017-06-02 Richard Biener <rguenther@suse.de>
13095 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13096 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13098 2017-06-02 Richard Biener <rguenther@suse.de>
13100 PR tree-optimization/80948
13101 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13102 stmt order as well.
13104 2017-06-02 Richard Biener <rguenther@suse.de>
13106 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13108 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13109 for inductions as relevant.
13111 2017-06-02 Richard Biener <rguenther@suse.de>
13113 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13114 (vectorizable_induction): ... this. Remove dead code.
13116 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13118 * builtins. (expand_builtin_alloca): Remove second parameter and
13119 infer its value from the first parameter instead.
13120 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13122 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13124 PR rtl-optimization/80903
13125 * loop-doloop.c (add_test): Unshare sequence.
13127 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13129 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13131 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13133 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13135 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13136 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13137 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13138 (xlogue_layout::get_stub_rtx): Make static.
13139 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13140 (xlogue_layout::compute_stub_managed_regs): Rename to...
13141 (xlogue_layout::count_stub_managed_regs): ...this.
13142 (xlogue_layout::is_stub_managed_reg): New function.
13143 (xlogue_layout::m_stub_names): Rename to...
13144 (xlogue_layout::s_stub_names): ...this, make static.
13145 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13146 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13147 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13148 xlogue_layout::s_stub_names): Instantiate statics.
13149 (stub_managed_regs): Remove.
13150 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13151 (disable_call_ms2sysv_xlogues): Rename to...
13152 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13153 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13155 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13156 change after reload_completed.
13157 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13159 (ix86_expand_prologue): Likewise.
13160 (ix86_expand_epilogue): Likewise.
13161 (ix86_expand_split_stack_prologue): Likewise.
13162 (ix86_compute_frame_layout): Remove frame parameter ...
13163 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13164 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13166 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13167 (ix86_frame): Move from here ...
13168 * config/i386/i386.h (ix86_frame): ... to here.
13169 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13170 complete ix86_frame data structure instead. Remove some_ld_name.
13172 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13174 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13175 symbols that hold a DECL_VALUE_EXPR.
13177 2017-06-01 Martin Jambor <mjambor@suse.cz>
13179 PR tree-optimization/80898
13180 * tree-sra.c (process_subtree_disqualification): Removed.
13181 (disqualify_candidate): Do not acll
13182 process_subtree_disqualification.
13183 (subtree_mark_written_and_enqueue): New function.
13184 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13185 RHS has been disqualified and re-queue LHS if necessary. Apart
13186 from that, ignore disqualified RHS.
13188 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13190 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13191 address fetch for z10 or later.
13193 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13195 * config/arc/arc.md (tst_movb): Add guard when splitting.
13197 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13199 * config/arc/arc.c (arc_can_eliminate): Test against
13200 arc_frame_pointer_needed.
13202 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13204 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13205 to prevent store reordering.
13206 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13207 (type): Add block type.
13208 (stack_tie): Define special instruction to be used in
13211 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13213 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13214 constraint. It is not valid for the pattern.
13215 (noncommutative_binary_comparison): Likewise.
13217 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13219 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13222 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13224 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13225 be used by the reg-alloc.
13227 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13229 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13230 reg-alloc when having mul64 or mul32x16 instructions.
13231 (mulsidi3): Likewise.
13232 (umulsidi3): Likewise.
13233 (mulsi32x16): New pattern.
13234 (mulsi64): Likewise.
13235 (mulsidi64): Likewise.
13236 (umulsidi64): Likewise.
13237 (MUL32x16_REG): Define.
13238 (mul64_600): Use MUL32x16_REG.
13239 (mac64_600): Likewise.
13240 (umul64_600): Likewise.
13241 (umac64_600): Likewise.
13243 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13245 * config/arc/arc.md (mulsi3_700): Make it commutative.
13247 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13249 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13251 (*sign_extendsidi2_insn): Likewise for movstosw.
13253 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13255 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13256 the type of the input discriminant value. Convert the
13257 discriminant value of signedness vary.
13259 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13261 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13262 Add to -Wall section.
13264 2017-06-01 Richard Biener <rguenther@suse.de>
13266 PR middle-end/66313
13267 * fold-const.c (fold_plusminus_mult_expr): If the factored
13268 factor may be zero use a wrapping type for the inner operation.
13269 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13270 and handle moved defs.
13271 (process_assignment): Properly guard the unary op case. Return a
13272 tri-state indicating that moving the stmt before the call may allow
13273 to continue. Pass through to_move.
13274 (find_tail_calls): Handle moving unrelated defs before
13277 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13280 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13281 splitter result in the canonical way.
13283 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13285 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13286 also for 32bit target. Update insn attributes.
13287 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13289 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13291 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13292 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13293 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13294 (_mm_maskz_min_ss): New intrinsics.
13296 2017-05-31 Martin Liska <mliska@suse.cz>
13298 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13299 Change comment style to one we normally use.
13300 (vectorizable_reduction): Likewise.
13301 (vectorizable_induction): Likewise.
13302 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13303 (vectorizable_call): Likewise.
13304 (vectorizable_simd_clone_call): Likewise.
13305 (vectorizable_conversion): Likewise.
13306 (vectorizable_assignment): Likewise.
13307 (vectorizable_shift): Likewise.
13308 (vectorizable_operation): Likewise.
13309 (vectorizable_store): Likewise.
13310 (vectorizable_load): Likewise.
13311 * tree-vectorizer.h: Likewise.
13313 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13315 * passes.c (emergency_dump_function): New.
13316 * tree-pass.h (emergency_dump_function): Declare.
13317 * plugin.c (plugins_internal_error_function): Remove.
13318 * plugin.h (plugins_internal_error_function): Remove declaration.
13319 * toplev.c (internal_error_function): New static function. Use it...
13320 (general_init): ...here.
13322 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13324 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13325 (arc_rtx_costs): Add costs for new patterns.
13326 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13327 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13329 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13331 * tree-ssa-strlen.c (get_next_strinfo): New function.
13332 (get_stridx_plus_constant): Use it.
13333 (zero_length_string): Likewise.
13334 (adjust_related_strinfos): Likewise.
13335 (adjust_last_stmt): Likewise.
13337 2017-05-31 Richard Biener <rguenther@suse.de>
13340 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13341 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13343 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13345 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13346 loop_vinfo argument and use of dependence distance vectors.
13347 Check instead whether the two references differ only in their
13348 initial value and assume that they have the same alignment if the
13349 difference is a multiple of the vector alignment.
13350 (vect_analyze_data_refs_alignment): Update call accordingly.
13352 2017-05-31 Martin Liska <mliska@suse.cz>
13355 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13357 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13359 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13360 (create_intersect_range_checks): Move from ...
13361 * tree-data-ref.c (create_intersect_range_checks_index)
13362 (create_intersect_range_checks): ... to here.
13363 (create_runtime_alias_checks): New function factored from ...
13364 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13365 here. Call above function.
13366 * tree-data-ref.h (create_runtime_alias_checks): New function.
13368 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13370 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13371 segment length for dr_b and compute it in wide_int.
13373 2017-05-31 Richard Biener <rguenther@suse.de>
13375 PR tree-optimization/80906
13376 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13377 and pass through iv_map.
13378 (copy_bb_and_scalar_dependences): Adjust.
13379 (translate_pending_phi_nodes): Likewise.
13380 (copy_loop_close_phi_args): Handle code-generating IVs instead
13383 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13385 * diagnostic-color.c (color_dict): Add "type-diff".
13386 (parse_gcc_colors): Update comment.
13387 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13388 -fdiagnostics-show-template-tree and -fno-elide-type.
13389 (GCC_COLORS): Add type-diff to example.
13391 (-fdiagnostics-show-template-tree): New.
13392 (-fno-elide-type): New.
13393 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13394 the pp_format_decoder callback. Call any m_format_postprocessor's
13396 (pretty_printer::pretty_printer): Initialize
13397 m_format_postprocessor.
13398 (pretty_printer::~pretty_printer): Delete any
13399 m_format_postprocessor.
13400 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13401 (class format_postprocessor): New class.
13402 (struct pretty_printer::format_decoder): Document the new parameters.
13403 (struct pretty_printer::m_format_postprocessor): New field.
13404 * tree-diagnostic.c (default_tree_printer): Update for new
13405 bool and const char ** params.
13406 * tree-diagnostic.h (default_tree_printer): Likewise.
13408 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13410 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13411 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13412 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13413 rs6000_gen_cell_microcode code.
13414 (rs6000_final_prescan_insn): Delete.
13415 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13416 "warn-cell-microcode" entries.
13417 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13418 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13419 throughout. Change cc_reg_not_micro_cr0_operand to
13420 cc_reg_not_cr0_operand throughout.
13421 (*extendhi<mode>2_noload): Delete.
13422 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13423 (mwarn-cell-microcode): Delete.
13424 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13425 -mgen-cell-microcode and -mwarn-cell-microcode.
13427 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13430 * config/i386/constraints.md (Yd): New constraint.
13432 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13433 and (?Yd, r) alternatives. Update insn attributes.
13434 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13435 and (?*Yd, r) alternatives. Update insn attributes.
13436 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13438 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13440 * gimplify.c (gimplify_modify_expr): Don't create a
13441 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13444 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13446 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13448 2017-05-30 Richard Biener <rguenther@suse.de>
13450 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13451 and reduc_def fields.
13452 (STMT_VINFO_REDUC_TYPE): New define.
13453 (STMT_VINFO_REDUC_DEF): Likewise.
13454 (vect_force_simple_reduction): Adjust prototype.
13455 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13456 (vect_is_simple_reduction): Remove check_reduction argument.
13457 (vect_force_simple_reduction): Adjust and set
13458 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13459 (vectorizable_reduction): Do not re-do reduction analysis
13460 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13461 * tree-parloops.c (gather_scalar_reductions): Adjust.
13463 2017-05-30 Richard Biener <rguenther@suse.de>
13465 PR middle-end/80901
13466 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13469 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13471 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13472 Introduce unknown_misalignment parameter and remove vf.
13473 (vect_peeling_hash_get_lowest_cost):
13474 Pass unknown_misalignment parameter.
13475 (vect_enhance_data_refs_alignment):
13476 Fix unsupportable data ref treatment.
13478 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13480 * tree-vect-data-refs.c (vect_get_data_access_cost):
13481 Workaround for SLP handling.
13482 (vect_enhance_data_refs_alignment):
13483 Compute costs for doing no peeling at all, compare to the best
13484 peeling costs so far and avoid peeling if cheaper.
13486 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13488 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13489 Return peeling info and set costs to zero for unlimited cost
13491 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13492 with unknown misalignment. Compute and costs for unknown
13493 misalignment, compare them to the costs for known misalignment
13494 and choose the cheapest for peeling.
13496 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13498 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13499 (vect_get_peeling_costs_all_drs): Create function.
13500 (vect_peeling_hash_get_lowest_cost):
13501 Use vect_get_peeling_costs_all_drs.
13502 (vect_peeling_supportable): Create function.
13503 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13505 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13507 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13508 DR_HAS_NEGATIVE_STEP.
13509 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13510 (vect_enhance_data_refs_alignment): Use.
13511 (vect_duplicate_ssa_name_ptr_info): Use.
13512 * tree-vectorizer.h (dr_misalignment): Use.
13513 (known_alignment_for_access_p): Use.
13515 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13518 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13520 (has_section_name): New function.
13522 2017-05-30 Martin Liska <mliska@suse.cz>
13525 * auto-profile.c (get_function_decl_from_block): Fix
13528 2017-05-30 Richard Biener <rguenther@suse.de>
13530 PR middle-end/80876
13531 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13533 2017-05-30 Martin Liska <mliska@suse.cz>
13535 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13536 * dumpfile.h (struct dump_file_info): Remove ctors.
13538 2017-05-30 Martin Liska <mliska@suse.cz>
13540 * predict.def: Fix GNU coding style.
13542 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13544 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13545 Mark 'to' argument with ATTRIBUTE_UNUSED.
13547 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13549 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13550 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13551 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13554 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13556 * doc/install.texi (Options specification): Restore entry of
13557 --enable-sjlj-exceptions.
13559 2017-05-27 Michael Eager <eager@eagercon.com>
13562 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13564 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13566 * config/microblaze/microblaze.h
13567 (FIXED_REGISTERS): Update in macro.
13568 (CALL_USED_REGISTERS): Update in macro.
13570 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13572 * doc/install.texi: Add links to macOS binary distributions.
13574 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13578 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13580 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13582 2017-05-26 Martin Liska <mliska@suse.cz>
13584 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13586 2017-05-26 Martin Liska <mliska@suse.cz>
13588 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13589 always leading ';; '.
13590 (dump_bb_info): Likewise.
13591 (brief_dump_cfg): Likewise.
13592 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13593 * dumpfile.c: Remove usage of TDF_VERBOSE.
13594 * dumpfile.h (enum dump_kind): Likewise.
13595 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13596 * print-tree.c (debug_verbose): Remove.
13597 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13598 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13599 * tree-diagnostic.c (default_tree_printer): Replace
13600 TDF_DIAGNOSTIC with TDF_SLIM.
13602 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13604 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13605 in parameter loop, rather than loop_vinfo.
13606 (create_intersect_range_checks): Ditto.
13607 (vect_create_cond_for_alias_checks): Update call to above functions.
13609 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13611 PR tree-optimization/80815
13612 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13613 for merging runtime alias checks. Handle negative DR_STEPs.
13615 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13617 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13619 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13620 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13621 out code pruning runtime alias checks.
13622 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13623 factored out from above.
13624 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13626 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13628 (prune_runtime_alias_test_list): New decalaration.
13630 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13632 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13633 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13634 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13635 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13636 (operator==, comp_dr_with_seg_len_pair): Ditto.
13637 (vect_prune_runtime_alias_test_list): Ditto.
13639 2017-05-26 Martin Liska <mliska@suse.cz>
13642 * params.def: Bound partial-inlining-entry-probability param.
13644 2017-05-26 Marek Polacek <polacek@redhat.com>
13647 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13650 2017-05-26 Richard Biener <rguenther@suse.de>
13652 PR tree-optimization/80842
13653 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13656 2017-05-26 Richard Biener <rguenther@suse.de>
13658 PR tree-optimization/80844
13659 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13661 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13663 * doc/md.texi (Machine Constraints): Update x86 family
13664 machine constraints section to match 'config/i386/constraints.md'.
13666 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13668 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13670 2017-05-25 Nathan Sidwell <nathan@acm.org>
13672 * doc/invoke.texi (--enable-languages): Update documentation.
13674 2017-05-25 Martin Liska <mliska@suse.cz>
13676 * dumpfile.c: Add TDF_FOLDING.
13677 * dumpfile.h (enum dump_kind): Likewise.
13678 * genmatch.c (dt_simplify::gen_1): Use it.
13680 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13682 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13684 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13686 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13687 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13689 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13691 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13692 * match.pd (X == C): Rewrite it here.
13693 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13694 with_certain_nonzero_bits2): New predicates.
13695 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13697 2017-05-24 Nathan Sidwell <nathan@acm.org>
13699 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13702 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13705 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13707 * config/powerpcspe: New port. Files are copied from the rs6000
13708 port, with "rs6000" in filenames replaced by "powerpcspe".
13710 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13712 PR rtl-optimization/80754
13713 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13715 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13717 * config/sparc/sparc.md (length): Return the correct value for -mflat
13718 sibcalls to match output_sibcall.
13720 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13724 * config/rs6000/rs6000.c (struct machine_function): Add new field
13726 (rs6000_get_separate_components): Init that field, use it.
13727 (rs6000_components_for_bb): Use the field.
13729 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13731 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13733 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13735 PR middle-end/80823
13736 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13738 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13741 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13742 address against address_operand predicate.
13743 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13745 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13747 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13748 parameters passed indirectly.
13750 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13752 * config/i386/i386.md (*movdi_internal): Remove SSE4
13753 alternative 18 (?r, *v). Update insn attributes.
13754 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13755 Update insn attributes.
13756 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13757 Update insn attributes.
13758 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13759 alternative 1 (r, v). Remove isa attribute.
13760 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13761 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13762 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13764 2017-05-23 Tom de Vries <tom@codesourcery.com>
13766 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13769 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13771 * cgraphunit.c (symbol_table::process_new_functions): Update.
13772 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13773 (inline_generate_summary): Rename to ...
13774 (ipa_fn_summary_generate): ... this one.
13775 (inline_read_summary): Rename to ...
13776 (ipa_fn_summary_read): ... this one.
13777 (inline_write_summary): Rename to ...
13778 (ipa_fn_summary_write): ... this one.
13779 (inline_free_summary): Rename to ...
13780 (ipa_free_fn_summary): ... this one.
13781 (pass_data_local_fn_summary, pass_local_fn_summary,
13782 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13783 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13784 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13785 make_pass_ipa_fn_summary): New.
13786 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13787 inline_write_summary, inline_free_summary): Remove.
13788 (ipa_free_fn_summary) : New.
13789 * ipa-inline.c (ipa_inline): Update.
13790 (pass_ipa_inline): Do not generate summaries.
13791 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13793 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13794 and add pass_ipa_fn_summary.
13795 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13797 (make_pass_inline_parameters): Remove.
13799 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13801 * omp-low.c (struct omp_context): Remove "default_kind" member.
13804 * omp-offload.c (execute_oacc_device_lower): Remove the
13805 parallelism dimensions function attributes for unparallelized
13806 OpenACC kernels constructs.
13808 2017-05-23 Martin Liska <mliska@suse.cz>
13810 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13812 (cgraph_edge::make_speculative): Likewise.
13813 (cgraph_edge::resolve_speculation): Likewise.
13814 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13815 (cgraph_node::dump): Likewise.
13816 * cgraph.h: Likewise.
13817 * cgraphunit.c (analyze_functions): Likewise.
13818 (symbol_table::compile): Likewise.
13819 * ipa-cp.c (print_all_lattices): Likewise.
13820 (determine_versionability): Likewise.
13821 (initialize_node_lattices): Likewise.
13822 (ipcp_verify_propagated_values): Likewise.
13823 (estimate_local_effects): Likewise.
13824 (update_profiling_info): Likewise.
13825 (create_specialized_node): Likewise.
13826 (perhaps_add_new_callers): Likewise.
13827 (decide_about_value): Likewise.
13828 (decide_whether_version_node): Likewise.
13829 (identify_dead_nodes): Likewise.
13830 (ipcp_store_bits_results): Likewise.
13831 * ipa-devirt.c (dump_targets): Likewise.
13832 (ipa_devirt): Likewise.
13833 * ipa-icf.c (sem_item::dump): Likewise.
13834 (sem_function::equals): Likewise.
13835 (sem_variable::equals): Likewise.
13836 (sem_item_optimizer::read_section): Likewise.
13837 (sem_item_optimizer::execute): Likewise.
13838 (congruence_class::dump): Likewise.
13839 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13840 (dump_inline_summary): Likewise.
13841 (estimate_node_size_and_time): Likewise.
13842 (inline_analyze_function): Likewise.
13843 * ipa-inline-transform.c (inline_call): Likewise.
13844 * ipa-inline.c (report_inline_failed_reason): Likewise.
13845 (want_early_inline_function_p): Likewise.
13846 (edge_badness): Likewise.
13847 (update_edge_key): Likewise.
13848 (inline_small_functions): Likewise.
13849 * ipa-profile.c (ipa_profile): Likewise.
13850 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13851 (ipa_make_edge_direct_to_target): Likewise.
13852 (remove_described_reference): Likewise.
13853 (ipa_impossible_devirt_target): Likewise.
13854 (propagate_controlled_uses): Likewise.
13855 (ipa_print_node_params): Likewise.
13856 (ipcp_transform_function): Likewise.
13857 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13858 (propagate_pure_const): Likewise.
13859 * ipa-reference.c (generate_summary): Likewise.
13860 (read_write_all_from_decl): Likewise.
13861 (propagate): Likewise.
13862 (ipa_reference_read_optimization_summary): Likewise.
13863 * ipa-utils.c (ipa_merge_profiles): Likewise.
13864 * ipa.c (walk_polymorphic_call_targets): Likewise.
13865 (symbol_table::remove_unreachable_nodes): Likewise.
13866 (ipa_single_use): Likewise.
13867 * passes.c (execute_todo): Likewise.
13868 * predict.c (drop_profile): Likewise.
13869 * symtab.c (symtab_node::get_dump_name): New function.
13870 (symtab_node::dump_name): Likewise.
13871 (symtab_node::dump_asm_name): Likewise.
13872 (symtab_node::dump_references): Likewise.
13873 (symtab_node::dump_referring): Likewise.
13874 (symtab_node::dump_base): Likewise.
13875 (symtab_node::debug_symtab): Likewise.
13876 * tree-sra.c (convert_callers_for_node): Likewise.
13877 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13878 * value-prof.c (init_node_map): Likewise.
13880 2017-05-23 Martin Liska <mliska@suse.cz>
13882 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13883 and symtab_node::debug_symtab to symbol_table::debug.
13884 * cgraphunit.c (analyze_functions): Use the renamed function.
13885 (symbol_table::compile): Likewise.
13886 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13887 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13888 * passes.c (execute_todo): Likewise.
13889 * symtab.c (symbol_table::dump): New function.
13890 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13892 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13894 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13895 that nonconst implies exec.
13897 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13899 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13900 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13901 (inline_edge_summary_vec): Turn into ...
13902 (ipa_call_summaries): ... this one.
13903 (redirect_to_unreachable, edge_set_predicate,
13904 evaluate_properties_for_edge, inline_summary_alloc,
13905 reset_ipa_call_summary, reset_inline_summary,
13906 inline_summary_t::duplicate): Update.
13907 (inline_edge_duplication_hook): Turn to ...
13908 (ipa_call_summary_t::duplicate): ... this one.
13909 (inline_edge_removal_hook): Turn to ...
13910 (ipa_call_summary_t::remove): ... this one.
13911 (dump_inline_edge_summary): Turn to ...
13912 (dump_ipa_call_summary): ... this one.
13913 (estimate_function_body_sizes): Update.
13914 (inline_update_callee_summaries): Update.
13915 (remap_edge_change_prob): Update.
13916 (remap_edge_summaries): Update.
13917 (inline_merge_summary): Update.
13918 (do_estimate_edge_time): Update.
13919 (inline_generate_summary): Update.
13920 (inline_read_section): Update.
13921 (inline_read_summary): Update.
13922 (inline_free_summary): Update.
13923 * ipa-inline.c (can_inline_edge_p): Update.
13924 (compute_inlined_call_time): Update.
13925 (want_inline_small_function_p): Update.
13926 (edge_badness): Update.
13927 (early_inliner): Update.
13928 * ipa-inline.h (inline_edge_summary): Turn to ...
13929 (ipa_call_summary): ... this one.
13930 (ipa_call_summary_t): New class.
13931 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13932 (ipa_call_summaries): New.
13933 (inline_edge_summary): Remove.
13934 (estimate_edge_growth): Update.
13935 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13936 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13937 * ipa-split.c (execute_split_functions): Update.
13938 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13940 2017-05-23 Tom de Vries <tom@codesourcery.com>
13942 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13943 attributes): Document rdrand effective target.
13945 2017-05-23 Tom de Vries <tom@codesourcery.com>
13947 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13948 attributes): Sort alphabetically.
13950 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13952 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13954 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13957 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13958 V2DF/V2DI splat into two separate patterns, one that handles
13959 registers, and the other that only handles memory. Drop support
13960 for splatting from a GPR on ISA 2.07 and then splitting the
13961 splat into direct move and splat.
13962 (vsx_splat_<mode>_reg): Likewise.
13963 (vsx_splat_<mode>_mem): Likewise.
13965 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
13967 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
13969 2017-05-22 Jakub Jelinek <jakub@redhat.com>
13971 PR middle-end/80809
13972 * omp-low.c (finish_taskreg_remap): New function.
13973 (finish_taskreg_scan): If unit size of ctx->record_type
13974 is non-constant, unshare the size expression and replace
13975 decls in it with possible outer var refs.
13977 PR middle-end/80809
13978 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
13979 GOVD_SHARED rather than GOVD_PRIVATE with it.
13980 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
13981 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
13983 PR middle-end/80853
13984 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
13985 as last argument to build_outer_var_ref for pointer bases of array
13986 section reductions.
13988 2017-05-19 Martin Sebor <msebor@redhat.com>
13990 * print-tree.c (print_node): Print DECL_READ_P flag.
13992 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13994 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
13995 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
13996 * cgraph.c: Likewise.
13997 * cgraphunit.c: Likewise.
13998 * gengtype.c: Likewise.
13999 * ipa-cp.c: Likewise.
14000 * ipa-devirt.c: Likewise.
14001 * ipa-icf.c: Likewise.
14002 * ipa-predicate.c: Likewise.
14003 * ipa-profile.c: Likewise.
14004 * ipa-prop.c: Likewise.
14005 * ipa-split.c: Likewise.
14007 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14008 edge_predicate_pool, dump_inline_hints,
14009 inline_summary::account_size_time, redirect_to_unreachable,
14010 edge_set_predicate, set_hint_predicate,
14011 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14012 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14013 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14014 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14015 ipa_call_summary_t::remove, initialize_growth_caches,
14016 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14017 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14018 mark_modified, unmodified_parm_1, unmodified_parm,
14019 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14020 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14021 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14022 will_be_nonconstant_predicate, record_modified_bb_info,
14023 get_minimal_bb, record_modified, param_change_prob,
14024 phi_result_unknown_predicate, predicate_for_phi_result,
14025 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14026 estimate_function_body_sizes, compute_inline_parameters,
14027 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14028 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14029 inline_update_callee_summaries, remap_edge_change_prob,
14030 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14031 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14032 inline_analyze_function, inline_summary_t::insert,
14033 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14034 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14035 inline_free_summary): Move to ipa-fnsummary.h
14036 (predicate_t): Remove.
14037 * ipa-fnsummary.c: New file.
14038 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14039 (enum inline_hints_vals, inline_hints, agg_position_info,
14040 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14041 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14042 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14043 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14044 inline_read_summary, inline_write_summary, inline_free_summary,
14045 inline_analyze_function, initialize_inline_failed,
14046 inline_merge_summary, inline_update_overall_summary,
14047 compute_inline_parameters): Move to ipa-fnsummary.h
14048 * ipa-fnsummary.h: New file.
14049 * ipa-inline-transform.h: Include ipa-inline.h.
14050 * ipa-inline.c: LIkewise.
14052 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14054 * ipa-inline.c (edge_badness): Use inlined_time instead of
14055 inline_summaries->get.
14057 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14059 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14061 2017-05-22 Nathan Sidwell <nathan@acm.org>
14063 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14064 (fdump-lang): Document 'raw' option.
14065 * dumpfile.h (TDI_tu): Delete.
14066 * dumpfile.c (dump_files): Remove translation-unit.
14067 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14069 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14071 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14072 command option from $(AWK) call.
14073 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14074 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14075 [FORMAT]: Remove handling of variable.
14076 * config/avr/t-multilib: Regenerate.
14078 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14080 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14082 (dump_inline_summary): Do not print self_time.
14083 (estimate_function_body_sizes): Do not set self_time.
14084 (compute_inline_parameters): Likewise.
14085 (inline_read_section, inline_write_summary): Do not stream self_time.
14086 * ipa-inline.h (inline_summary): Drop self_time.
14088 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14090 * ipa-inline-analysis.c (account_size_time): Rename to ...
14091 (inline_summary::account_size_time): ... this one.
14092 (reset_ipa_call_summary): Turn to ...
14093 (ipa_call_summary::reset): ... this one.
14094 (reset_inline_summary): Turn to ...
14095 (inline_summary::reset): ... this one.
14096 (inline_summary_t::remove): Update.
14097 (inline_summary_t::duplicate): Update.
14098 (ipa_call_summary_t::remove): Update.
14099 (dump_inline_summary): Update.
14100 (estimate_function_body_sizes): Update.
14101 (compute_inline_parameters): Update.
14102 (estimate_node_size_and_time): Update.
14103 (inline_merge_summary): Update.
14104 (inline_update_overall_summary): Update.
14105 (inline_read_section): Update.
14106 (inline_write_summary): Update.
14107 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14108 add account_size_time and reset member functions.
14109 (ipa_call_summary): Add reset function.
14110 * ipa-predicate.h (predicate::operator &): Constify.
14112 2017-05-22 Richard Biener <rguenther@suse.de>
14114 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14116 2017-05-19 Jason Merrill <jason@redhat.com>
14118 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14120 2017-05-19 Marek Polacek <polacek@redhat.com>
14123 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14124 TYPE_OVERFLOW_WRAPS checks.
14126 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14128 * tree-core.h (enum omp_clause_default_kind): Add
14129 "OMP_CLAUSE_DEFAULT_PRESENT".
14130 * tree-pretty-print.c (dump_omp_clause): Handle it.
14131 * gimplify.c (enum gimplify_omp_var_data): Add
14132 "GOVD_MAP_FORCE_PRESENT".
14133 (gimplify_adjust_omp_clauses_1): Map it to
14134 "GOMP_MAP_FORCE_PRESENT".
14135 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14137 * gimplify.c (oacc_default_clause): Clarify.
14139 2017-05-19 Nathan Sidwell <nathan@acm.org>
14141 LANG_HOOK_REGISTER_DUMPS
14142 * toplev.c (general_init): Call register dump lang hook.
14143 * doc/invoke.texi: Document -fdump-lang option family.
14144 * dumpfile.c (dump_files): Remove class dump here.
14145 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14146 * dumpfile.h (tree_dump_index): Remove TDI_class.
14147 * langhooks-def.h (lhd_register_dumps): Declare.
14148 (LANG_HOOKS_REGISTER_DUMPS): Define.
14149 (LANG_HOOKS_INITIALIZER): Add it.
14150 * langhooks.c (lhd_register_dumps): Define.
14151 * langhooks.h (struct lang_hooks): Add register_dumps.
14153 2017-05-19 Nathan Sidwell <nathan@acm.org>
14155 * context.h (context::set_passes): New.
14156 * context.c (context::context): Do not create pass manager.
14157 * toplev.c (general_init): Create pass manager here.
14159 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14161 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14162 use this splitter if two add or or instructions would also work for
14163 the constant we want to generate.
14165 2017-05-19 Richard Biener <rguenther@suse.de>
14168 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14169 predicate evaluation.
14171 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14173 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14175 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14178 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14180 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14181 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14182 (inline_edge_summary_vec): Turn into ...
14183 (ipa_call_summaries): ... this one.
14184 (redirect_to_unreachable, edge_set_predicate,
14185 evaluate_properties_for_edge, inline_summary_alloc,
14186 reset_ipa_call_summary, reset_inline_summary,
14187 inline_summary_t::duplicate): Update.
14188 (inline_edge_duplication_hook): Turn to ...
14189 (ipa_call_summary_t::duplicate): ... this one.
14190 (inline_edge_removal_hook): Turn to ...
14191 (ipa_call_summary_t::remove): ... this one.
14192 (dump_inline_edge_summary): Turn to ...
14193 (dump_ipa_call_summary): ... this one.
14194 (estimate_function_body_sizes): Update.
14195 (inline_update_callee_summaries): Update.
14196 (remap_edge_change_prob): Update.
14197 (remap_edge_summaries): Update.
14198 (inline_merge_summary): Update.
14199 (do_estimate_edge_time): Update.
14200 (inline_generate_summary): Update.
14201 (inline_read_section): Update.
14202 (inline_read_summary): Update.
14203 (inline_free_summary): Update.
14204 * ipa-inline.c (can_inline_edge_p): Update.
14205 (compute_inlined_call_time): Update.
14206 (want_inline_small_function_p): Update.
14207 (edge_badness): Update.
14208 (early_inliner): Update.
14209 * ipa-inline.h (inline_edge_summary): Turn to ...
14210 (ipa_call_summary): ... this one.
14211 (ipa_call_summary_t): New class.
14212 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14213 (ipa_call_summaries): New.
14214 (inline_edge_summary): Remove.
14215 (estimate_edge_growth): Update.
14216 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14217 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14218 * ipa-split.c (execute_split_functions): Update.
14219 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14221 2017-05-19 Richard Biener <rguenther@suse.de>
14223 PR middle-end/80764
14224 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14226 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14228 * config/rs6000/rs6000.c (struct machine_function): Add field
14229 fpr_is_wrapped_separately.
14230 (rs6000_get_separate_components): Use 64 components. Handle the
14231 new FPR components.
14232 (rs6000_components_for_bb): Handle the FPR components.
14233 (rs6000_emit_prologue_components): Handle the FPR components.
14234 (rs6000_emit_epilogue_components): Handle the FPR components.
14235 (rs6000_set_handled_components): Handle the FPR components.
14236 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14237 that are already separately shrink-wrapped.
14238 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14239 that are already separately shrink-wrapped.
14241 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14244 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14247 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14248 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14249 cases where the register allocator uses a move and an offsettable
14250 memory operation to/from a FPR register on ISA 2.06/2.07.
14251 (define_peephole2 for Altivec d-form store): Likewise.
14253 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14256 * config/i386/mmx.md (*mov<mode>_internal): Enable
14257 alternatives 11, 12, 13 and 14 also for 32bit targets.
14258 Remove alternatives 15, 16, 17 and 18.
14259 * config/i386/sse.md (vec_concatv2di): Change
14260 alternative (!x, *y) to (x, ?!*Yn).
14262 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14264 * dumpfile.h (enum dump_kind): Remove stray comma.
14266 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14268 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14269 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14270 predicate::num_conditions
14271 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14272 (CHANGED): turn into predicate::changed.
14273 (agg_position_info): Move to ipa-predicate.h
14274 (add_condition, predicate::add_clause, predicate::operator &=,
14275 predicate::or_with, predicate::evaluate, predicate::probability,
14276 dump_condition, dump_clause, predicate::dump,
14277 predicate::remap_after_duplication, predicate::remap_after_inlining,
14278 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14279 (evaluate_conditions_for_known_args): Update.
14280 (set_cond_stmt_execution_predicate): Update.
14281 * ipa-inline.h: Include ipa-predicate.h
14282 (condition, inline_param_summary, conditions, agg_position_info,
14283 predicate): Move to ipa-predicate.h
14284 * ipa-predicate.c: New file.
14285 * ipa-predicate.h: New file.
14287 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14289 * final.c (leaf_function_p): Check we are not in a sequence.
14291 2017-05-18 Martin Liska <mliska@suse.cz>
14293 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14294 * dumpfile.c (dump_register): Use new enum dump_kind.
14295 (get_dump_file_name): Likewise.
14296 (dump_enable_all): Likewise.
14297 (dump_switch_p_1): Likewise.
14298 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14299 * dumpfile.h (enum dump_kind): New enum type.
14300 (struct dump_file_info): Create constructor and
14301 format fields and comments.
14302 * passes.c (pass_manager::register_one_dump_file):
14304 * statistics.c (statistics_early_init): Likewise.
14305 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14306 TDF_TREE with TDF_SLIM.
14307 (gather_memory_references_ref): Likewise.
14309 2017-05-18 Martin Liska <mliska@suse.cz>
14311 * vec.h (struct vnull): Use it.
14313 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14315 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14316 (true_predicate, false_predicate, true_predicate_p,
14317 false_predicate_p): Remove.
14318 (single_cond_predicate, not_inlined_predicate): Turn to member function
14320 (add_condition): Update.
14321 (add_clause): Turn to...
14322 (predicate::add_clause): ... this one; update; allow passing NULL
14324 (and_predicates): Turn to ...
14325 (predicate::operator &=): ... this one.
14326 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14327 (or_predicates): Turn to ...
14328 (predicate::or_with): ... this one.
14329 (evaluate_predicate): Turn to ...
14330 (predicate::evaluate): ... this one.
14331 (predicate_probability): Turn to ...
14332 (predicate::probability): ... this one.
14333 (dump_condition): Update.
14334 (dump_predicate): Turn to ...
14335 (predicate::dump): ... this one.
14336 (account_size_time): Update.
14337 (edge_set_predicate): Update.
14338 (set_hint_predicate): UPdate.
14339 (evaluate_conditions_for_known_args): Update.
14340 (evaluate_properties_for_edge): Update.
14341 (remap_predicate_after_duplication): Turn to...
14342 (predicate::remap_after_duplication): ... this one.
14343 (remap_hint_predicate_after_duplication): Update.
14344 (inline_summary_t::duplicate): UPdate.
14345 (dump_inline_edge_summary): Update.
14346 (dump_inline_summary): Update.
14347 (set_cond_stmt_execution_predicate): Update.
14348 (set_switch_stmt_execution_predicate): Update.
14349 (compute_bb_predicates): Update.
14350 (will_be_nonconstant_expr_predicate): Update.
14351 (will_be_nonconstant_predicate): Update.
14352 (phi_result_unknown_predicate): Update.
14353 (predicate_for_phi_result): Update.
14354 (array_index_predicate): Update.
14355 (estimate_function_body_sizes): Update.
14356 (estimate_node_size_and_time): Update.
14357 (estimate_ipcp_clone_size_and_time): Update.
14358 (remap_predicate): Rename to ...
14359 (predicate::remap_after_inlining): ... this one.
14360 (remap_hint_predicate): Update.
14361 (inline_merge_summary): Update.
14362 (inline_update_overall_summary): Update.
14363 (estimate_size_after_inlining): Update.
14364 (read_predicate): Rename to ...
14365 (predicate::stream_in): ... this one.
14366 (read_inline_edge_summary): Update.
14367 (write_predicate): Rename to ...
14368 (predicate::stream_out): ... this one.
14369 (write_inline_edge_summary): Update.
14370 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14371 (clause_t): Turn to uint32_t
14372 (predicate): Turn to class; implement constructor and operators
14374 (size_time_entry): Update.
14375 (inline_summary): Update.
14376 (inline_edge_summary): Update.
14378 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14380 * fold-const.c (fold_binary_loc): Move transformation...
14381 * match.pd (C - X CMP X): ... here.
14383 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14385 * config/sparc/sparc.c (sparc_option_override): Set function
14386 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14387 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14389 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14391 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14393 2017-05-18 Marek Polacek <polacek@redhat.com>
14396 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14397 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14398 gimple_assign_load_p.
14400 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14402 PR middle-end/80692
14403 * real.c (do_compare): Give decimal_do_compare preference over
14404 comparing just the signs.
14406 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14408 * doc/md.texi (Canonicalization of Instructions): Describe the
14409 canonical form of instructions that inherently set a condition
14412 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14414 PR middle-end/80775
14415 * tree-cfg.c: Move deletion of unreachable case statements to after
14416 the merging of consecutive case labels.
14418 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14420 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14421 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14422 restoring of callee-saved registers.
14424 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14426 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14427 * config/visium/visium.c (single_set_and_flags): Likewise.
14428 * config/visium/visium.md (Substitutions): Likewise.
14430 2017-05-17 Martin Liska <mliska@suse.cz>
14432 * cfg.c: Introduce dump_flags_t type and
14433 use it instead of int type.
14435 * cfghooks.c: Likewise.
14436 * cfghooks.h (struct cfg_hooks): Likewise.
14437 * cfgrtl.c: Likewise.
14438 * cfgrtl.h: Likewise.
14439 * cgraph.c (cgraph_node::get_body): Likewise.
14440 * coretypes.h: Likewise.
14441 * domwalk.c: Likewise.
14442 * domwalk.h: Likewise.
14443 * dumpfile.c (struct dump_option_value_info): Likewise.
14444 (dump_enable_all): Likewise.
14445 (dump_switch_p_1): Likewise.
14446 (opt_info_switch_p): Likewise.
14447 * dumpfile.h (enum tree_dump_index): Likewise.
14448 (struct dump_file_info): Likewise.
14449 * genemit.c: Likewise.
14450 * generic-match-head.c: Likewise.
14451 * gengtype.c (open_base_files): Likewise.
14452 * gimple-pretty-print.c: Likewise.
14453 * gimple-pretty-print.h: Likewise.
14454 * graph.c (print_graph_cfg): Likewise.
14455 * graphite-scop-detection.c (dot_all_sese): Likewise.
14456 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14457 * loop-unroll.c (report_unroll): Likewise.
14458 * passes.c (pass_manager::register_one_dump_file): Likewise.
14459 * print-tree.c: Likewise.
14460 * statistics.c: Likewise.
14461 * tree-cfg.c: Likewise.
14462 * tree-cfg.h: Likewise.
14463 * tree-dfa.c: Likewise.
14464 * tree-dfa.h: Likewise.
14465 * tree-dump.c (dump_function): Likewise.
14466 * tree-dump.h (struct dump_info): Likewise.
14467 * tree-pretty-print.c: Likewise.
14468 * tree-pretty-print.h: Likewise.
14469 * tree-ssa-live.c: Likewise.
14470 * tree-ssa-live.h: Likewise.
14471 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14472 * tree-vect-loop.c: Likewise.
14473 * tree-vect-slp.c: Likewise.
14475 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14476 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14478 PR tree-optimization/80457
14479 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14480 of all arguments to a statement as scalar_to_vec operations.
14481 (vectorizable_call): Adjust call to vect_model_simple_cost for
14483 (vectorizable_conversion): Likewise.
14484 (vectorizable_assignment): Likewise.
14485 (vectorizable_shift): Likewise.
14486 (vectorizable_operation): Likewise.
14487 (vectorizable_comparison): Likewise.
14488 (vect_is_simple_cond): Record the def types for operands.
14489 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14490 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14491 for statement argument count.
14493 2017-05-16 Carl Love <cel@us.ibm.com>
14495 * config/rs6000/rs6000-c: Add support for built-in functions
14496 vector unsigned long long vec_bperm (vector unsigned long long,
14497 vector unsigned char)
14498 vector signed long long vec_mule (vector signed int,
14500 vector unsigned long long vec_mule (vector unsigned int,
14501 vector unsigned int)
14502 vector signed long long vec_mulo (vector signed int,
14504 vector unsigned long long vec_mulo (vector unsigned int,
14505 vector unsigned int)
14506 vector signed char vec_sldw (vector signed char,
14507 vector signed char,
14509 vector unsigned char vec_sldw (vector unsigned char,
14510 vector unsigned char,
14512 vector signed short vec_sldw (vector signed short,
14513 vector signed short,
14515 vector unsigned short vec_sldw (vector unsigned short,
14516 vector unsigned short,
14518 vector signed int vec_sldw (vector signed int,
14521 vector unsigned int vec_sldw (vector unsigned int,
14522 vector unsigned int,
14524 vector signed long long vec_sldw (vector signed long long,
14525 vector signed long long,
14527 vector unsigned long long vec_sldw (vector unsigned long long,
14528 vector unsigned long long,
14530 * config/rs6000/rs6000-c: Add support for built-in functions
14531 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14532 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14533 * doc/extend.texi: Update the built-in documentation for the
14534 new built-in functions.
14536 2017-05-16 Marek Polacek <polacek@redhat.com>
14540 * tree.c (save_expr): Don't fold the expression.
14542 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14544 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14545 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14546 and (?*y,m). Update insn attributes.
14548 2017-05-16 Martin Liska <mliska@suse.cz>
14550 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14551 flags argument of print_gimple_stmt, print_gimple_expr,
14552 print_generic_stmt and print_generic_expr.
14553 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14554 * coretypes.h: Likewise.
14555 * except.c (dump_eh_tree): Likewise.
14556 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14557 * gimple-pretty-print.h: Likewise.
14558 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14559 (backprop::push_to_worklist): Likewise.
14560 (backprop::pop_from_worklist): Likewise.
14561 (backprop::process_use): Likewise.
14562 (backprop::intersect_uses): Likewise.
14563 (note_replacement): Likewise.
14564 * gimple-ssa-store-merging.c
14565 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14566 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14567 (pass_store_merging::execute): Likewise.
14568 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14569 (ssa_base_cand_dump_callback): Likewise.
14570 (dump_incr_vec): Likewise.
14571 (replace_refs): Likewise.
14572 (replace_mult_candidate): Likewise.
14573 (create_add_on_incoming_edge): Likewise.
14574 (create_phi_basis): Likewise.
14575 (insert_initializers): Likewise.
14576 (all_phi_incrs_profitable): Likewise.
14577 (introduce_cast_before_cand): Likewise.
14578 (replace_one_candidate): Likewise.
14579 * gimplify.c (gimplify_expr): Likewise.
14580 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14581 (set_rename): Likewise.
14582 (rename_uses): Likewise.
14583 (copy_loop_phi_nodes): Likewise.
14584 (add_close_phis_to_merge_points): Likewise.
14585 (copy_loop_close_phi_args): Likewise.
14586 (copy_cond_phi_args): Likewise.
14587 (graphite_copy_stmts_from_block): Likewise.
14588 (translate_pending_phi_nodes): Likewise.
14589 * graphite-poly.c (print_pdr): Likewise.
14590 (dump_gbb_cases): Likewise.
14591 (dump_gbb_conditions): Likewise.
14592 (print_scop_params): Likewise.
14593 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14594 (build_cross_bb_scalars_use): Likewise.
14595 (gather_bbs::before_dom_children): Likewise.
14596 * hsa-dump.c (dump_hsa_immed): Likewise.
14597 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14598 (get_replacement_map): Likewise.
14599 * ipa-inline-analysis.c (dump_condition): Likewise.
14600 (estimate_function_body_sizes): Likewise.
14601 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14602 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14603 * ipa-prop.c (ipa_dump_param): Likewise.
14604 (ipa_print_node_jump_functions_for_edge): Likewise.
14605 (ipa_modify_call_arguments): Likewise.
14606 (ipa_modify_expr): Likewise.
14607 (ipa_dump_param_adjustments): Likewise.
14608 (ipa_dump_agg_replacement_values): Likewise.
14609 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14610 * ipa-pure-const.c (check_stmt): Likewise.
14611 (pass_nothrow::execute): Likewise.
14612 * ipa-split.c (execute_split_functions): Likewise.
14613 * omp-offload.c (dump_oacc_loop_part): Likewise.
14614 (dump_oacc_loop): Likewise.
14615 * trans-mem.c (tm_log_emit): Likewise.
14616 (tm_memopt_accumulate_memops): Likewise.
14617 (dump_tm_memopt_set): Likewise.
14618 (dump_tm_memopt_transform): Likewise.
14619 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14620 (print_loop): Likewise.
14621 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14622 (chkp_gather_checks_info): Likewise.
14623 (chkp_get_check_result): Likewise.
14624 (chkp_remove_check_if_pass): Likewise.
14625 (chkp_use_outer_bounds_if_possible): Likewise.
14626 (chkp_reduce_bounds_lifetime): Likewise.
14627 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14628 (chkp_mark_completed_bounds): Likewise.
14629 (chkp_register_incomplete_bounds): Likewise.
14630 (chkp_mark_invalid_bounds): Likewise.
14631 (chkp_maybe_copy_and_register_bounds): Likewise.
14632 (chkp_build_returned_bound): Likewise.
14633 (chkp_get_bound_for_parm): Likewise.
14634 (chkp_build_bndldx): Likewise.
14635 (chkp_get_bounds_by_definition): Likewise.
14636 (chkp_generate_extern_var_bounds): Likewise.
14637 (chkp_get_bounds_for_decl_addr): Likewise.
14638 * tree-chrec.c (chrec_apply): Likewise.
14639 * tree-data-ref.c (dump_data_reference): Likewise.
14640 (dump_subscript): Likewise.
14641 (dump_data_dependence_relation): Likewise.
14642 (analyze_overlapping_iterations): Likewise.
14643 * tree-inline.c (expand_call_inline): Likewise.
14644 (tree_function_versioning): Likewise.
14645 * tree-into-ssa.c (dump_defs_stack): Likewise.
14646 (dump_currdefs): Likewise.
14647 (dump_names_replaced_by): Likewise.
14648 (dump_update_ssa): Likewise.
14649 (update_ssa): Likewise.
14650 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14651 * tree-parloops.c (build_new_reduction): Likewise.
14652 (try_create_reduction_list): Likewise.
14653 (ref_conflicts_with_region): Likewise.
14654 (oacc_entry_exit_ok_1): Likewise.
14655 (oacc_entry_exit_single_gang): Likewise.
14656 * tree-pretty-print.h: Likewise.
14657 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14658 (get_scalar_evolution): Likewise.
14659 (add_to_evolution): Likewise.
14660 (get_loop_exit_condition): Likewise.
14661 (analyze_evolution_in_loop): Likewise.
14662 (analyze_initial_condition): Likewise.
14663 (analyze_scalar_evolution): Likewise.
14664 (instantiate_scev): Likewise.
14665 (number_of_latch_executions): Likewise.
14666 (gather_chrec_stats): Likewise.
14667 (final_value_replacement_loop): Likewise.
14668 (scev_const_prop): Likewise.
14669 * tree-sra.c (dump_access): Likewise.
14670 (disqualify_candidate): Likewise.
14671 (create_access): Likewise.
14672 (reject): Likewise.
14673 (maybe_add_sra_candidate): Likewise.
14674 (create_access_replacement): Likewise.
14675 (analyze_access_subtree): Likewise.
14676 (analyze_all_variable_accesses): Likewise.
14677 (sra_modify_assign): Likewise.
14678 (initialize_constant_pool_replacements): Likewise.
14679 (find_param_candidates): Likewise.
14680 (decide_one_param_reduction): Likewise.
14681 (replace_removed_params_ssa_names): Likewise.
14682 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14683 * tree-ssa-copy.c (dump_copy_of): Likewise.
14684 (copy_prop_visit_cond_stmt): Likewise.
14685 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14686 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14687 (record_equivalences_from_stmt): Likewise.
14688 * tree-ssa-dse.c (compute_trims): Likewise.
14689 (delete_dead_call): Likewise.
14690 (delete_dead_assignment): Likewise.
14691 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14692 (forward_propagate_into_cond): Likewise.
14693 (pass_forwprop::execute): Likewise.
14694 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14695 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14697 (move_computations_worker): Likewise.
14698 (execute_sm): Likewise.
14699 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14700 (remove_exits_and_undefined_stmts): Likewise.
14701 (remove_redundant_iv_tests): Likewise.
14702 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14703 (adjust_iv_update_pos): Likewise.
14704 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14705 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14706 (value_replacement): Likewise.
14707 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14708 * tree-ssa-pre.c (print_pre_expr): Likewise.
14709 (get_representative_for): Likewise.
14710 (create_expression_by_pieces): Likewise.
14711 (insert_into_preds_of_block): Likewise.
14712 (eliminate_insert): Likewise.
14713 (eliminate_dom_walker::before_dom_children): Likewise.
14714 (eliminate): Likewise.
14715 (remove_dead_inserted_code): Likewise.
14716 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14717 * tree-ssa-reassoc.c (get_rank): Likewise.
14718 (eliminate_duplicate_pair): Likewise.
14719 (eliminate_plus_minus_pair): Likewise.
14720 (eliminate_not_pairs): Likewise.
14721 (undistribute_ops_list): Likewise.
14722 (eliminate_redundant_comparison): Likewise.
14723 (update_range_test): Likewise.
14724 (optimize_range_tests_var_bound): Likewise.
14725 (optimize_vec_cond_expr): Likewise.
14726 (rewrite_expr_tree): Likewise.
14727 (rewrite_expr_tree_parallel): Likewise.
14728 (linearize_expr): Likewise.
14729 (break_up_subtract): Likewise.
14730 (linearize_expr_tree): Likewise.
14731 (attempt_builtin_powi): Likewise.
14732 (attempt_builtin_copysign): Likewise.
14733 (transform_stmt_to_copy): Likewise.
14734 (transform_stmt_to_multiply): Likewise.
14735 (dump_ops_vector): Likewise.
14736 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14737 (print_scc): Likewise.
14738 (set_ssa_val_to): Likewise.
14739 (visit_reference_op_store): Likewise.
14740 (visit_use): Likewise.
14741 (sccvn_dom_walker::before_dom_children): Likewise.
14742 (run_scc_vn): Likewise.
14743 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14745 (expr_hash_elt::print): Likewise.
14746 (const_and_copies::pop_to_marker): Likewise.
14747 (const_and_copies::record_const_or_copy_raw): Likewise.
14748 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14749 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14750 (dump_predicates): Likewise.
14751 (find_uninit_use): Likewise.
14752 (warn_uninitialized_phi): Likewise.
14753 (pass_late_warn_uninitialized::execute): Likewise.
14754 * tree-ssa.c (verify_vssa): Likewise.
14755 (verify_ssa): Likewise.
14756 (maybe_optimize_var): Likewise.
14757 * tree-vrp.c (dump_value_range): Likewise.
14758 (dump_all_value_ranges): Likewise.
14759 (dump_asserts_for): Likewise.
14760 (register_edge_assert_for_2): Likewise.
14761 (vrp_visit_cond_stmt): Likewise.
14762 (vrp_visit_switch_stmt): Likewise.
14763 (vrp_visit_stmt): Likewise.
14764 (vrp_visit_phi_node): Likewise.
14765 (simplify_cond_using_ranges_1): Likewise.
14766 (fold_predicate_in): Likewise.
14767 (evrp_dom_walker::before_dom_children): Likewise.
14768 (evrp_dom_walker::push_value_range): Likewise.
14769 (evrp_dom_walker::pop_value_range): Likewise.
14770 (execute_early_vrp): Likewise.
14772 2017-05-16 Richard Biener <rguenther@suse.de>
14774 * dwarf2out.c (loc_list_from_tree_1): Do not create
14775 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14777 2017-05-16 Richard Biener <rguenther@suse.de>
14779 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14781 (note_variable_value_in_expr): If we resolved the decl ref
14782 do not push to the stack.
14784 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14786 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14787 operations in fast-math mode.
14788 (vaddq_f16): Likewise.
14789 (vmul_f16): Likewise.
14790 (vmulq_f16): Likewise.
14791 (vsub_f16): Likewise.
14792 (vsubq_f16): Likewise.
14793 * config/arm/neon.md (add<mode>3): New.
14795 (fma:<VH:mode>3): New. Also remove outdated comment.
14798 2017-05-16 Martin Liska <mliska@suse.cz>
14802 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14803 (odr_types_equivalent_p): Likewise.
14805 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14807 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14809 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14812 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14813 non-interunit SSE move alternatives with '?'.
14814 (zero-extendsidi peephole2): New peephole to skip intermediate
14815 general register in SSE zero-extend sequence.
14817 2017-05-15 Jeff Law <law@redhat.com>
14819 * reorg.c (relax_delay_slots): Create a new variable to hold
14820 the temporary target rather than clobbering TARGET_LABEL.
14822 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14823 missing argument to extract_bit_field call.
14824 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14826 2017-05-15 Martin Liska <mliska@suse.cz>
14829 * gcc.c (process_command): Do not allow empty argument of -o option.
14831 2017-05-15 Renlin Li <renlin.li@arm.com>
14833 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14834 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14835 * config/aarch64/constraints.md (Usf): Add long call check.
14836 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14837 (call_value): Likewise.
14838 (sibcall): Likewise.
14839 (sibcall_value): Likewise.
14841 (call_value_insn): New.
14842 (sibcall_insn): Update rtx pattern.
14843 (sibcall_value_insn): Likewise.
14844 (call_internal): Remove.
14845 (call_value_internal): Likewise.
14846 (sibcall_internal): Likewise.
14847 (sibcall_value_internal): Likewise.
14848 (call_reg): Likewise.
14849 (call_symbol): Likewise.
14850 (call_value_reg): Likewise.
14851 (call_value_symbol): Likewise.
14853 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14856 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14858 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14860 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14861 compatible with CCGOCmode and with CCZmode.
14863 2017-05-14 Martin Sebor <msebor@redhat.com>
14865 PR middle-end/77671
14866 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14867 (gimple_fold_builtin_snprintf): Same.
14868 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14869 (gimple_fold_builtin_snprintf): Same.
14870 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14871 of character types.
14872 (is_call_safe): New function.
14873 (try_substitute_return_value): Call it.
14874 (try_simplify_call): New function.
14875 (pass_sprintf_length::handle_gimple_call): Call it.
14877 2017-05-14 Martin Sebor <msebor@redhat.com>
14879 PR middle-end/80669
14880 * builtins.c (expand_builtin_stpncpy): Simplify.
14882 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14884 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14885 * config/i386/i386.h
14886 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14887 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14888 (struct machine_function): Add new members call_ms2sysv,
14889 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14890 (struct machine_frame_state): New fields sp_realigned and
14891 sp_realigned_offset.
14892 * config/i386/i386.c
14893 (enum xlogue_stub): New enum.
14894 (enum xlogue_stub_sets): New enum.
14895 (class xlogue_layout): New class.
14896 (struct ix86_frame): New fields stack_realign_allocate_offset,
14897 stack_realign_offset and outlined_save_offset. Modify comments to
14898 detail stack layout when using out-of-line stubs.
14899 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14900 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14901 -mcall-ms2sysv-xlogues.
14902 (stub_managed_regs): New static variable.
14903 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14904 registers managed by out-of-line stub.
14905 (disable_call_ms2sysv_xlogues): New function.
14906 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14907 m->call_ms2sysv when appropriate and compute frame layout for
14909 (sp_valid_at, fp_valid_at): New inline functions.
14910 (choose_basereg): New function.
14911 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14913 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14914 Use align parameter of choose_baseaddr to generated aligned SSE movs
14916 (pro_epilogue_adjust_stack): Modify to track
14917 machine_frame_state::sp_realigned.
14918 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14919 (ix86_nsaved_sseregs): Likewise.
14920 (ix86_emit_save_regs): Likewise.
14921 (ix86_emit_save_regs_using_mov): Likewise.
14922 (ix86_emit_save_sse_regs_using_mov): Likewise.
14923 (get_scratch_register_on_entry): Likewise.
14924 (gen_frame_set): New function.
14925 (gen_frame_load): Likewise.
14926 (gen_frame_store): Likewise.
14927 (emit_outlined_ms2sysv_save): Likewise.
14928 (emit_outlined_ms2sysv_restore): Likewise.
14929 (ix86_expand_prologue): Modify stack re-alignment code and call
14930 emit_outlined_ms2sysv_save when appropriate.
14931 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14932 parameter rtx_insn *insn, which allows the function to be used to only
14933 generate the notes.
14934 (ix86_expand_epilogue): Modify validity checks of frame and stack
14935 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14936 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14937 * config/i386/predicates.md
14938 (save_multiple): New predicate.
14939 (restore_multiple): Likewise.
14940 * config/i386/sse.md
14941 (save_multiple<mode>): New pattern.
14942 (save_multiple_realign<mode>): Likewise.
14943 (restore_multiple<mode>): Likewise.
14944 (restore_multiple_and_return<mode>): Likewise.
14945 (restore_multiple_leave_return<mode>): Likewise.
14946 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14948 2017-05-14 Julia Koval <julia.koval@intel.com>
14950 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14951 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14952 (__builtin_ia32_xsetbv): New builtins.
14953 * config/i386/i386.c (ix86_expand_special_args_builtin):
14955 (ix86_expand_builtin): Special expand for new intrinsics.
14956 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14957 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14958 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14960 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14962 * cfganal.c (inverted_post_order_compute): Change argument type
14964 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14965 * df-core.c (rest_of_handle_df_initialize): Adjust.
14966 (rest_of_handle_df_finish): Likewise.
14967 (df_analyze_1): Likewise.
14968 (df_analyze): Likewise.
14969 (loop_inverted_post_order_compute): Change argument to be a vec *.
14970 (df_analyze_loop): Adjust.
14971 (df_get_n_blocks): Likewise.
14972 (df_get_postorder): Likewise.
14973 * df.h (struct df_d): Change field to be a vec.
14974 * lcm.c (compute_laterin): Adjust.
14975 (compute_available): Likewise.
14976 * lra-lives.c (lra_create_live_ranges_1): Likewise.
14977 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
14978 * tree-ssa-pre.c (compute_antic): Likewise.
14980 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14982 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
14983 (depth_first_search::depth_first_search): Change structure init
14984 function to this constructor.
14985 (depth_first_search::add_bb): Rename function to this member.
14986 (depth_first_search::execute): Likewise.
14987 (flow_dfs_compute_reverse_finish): Adjust.
14989 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14991 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
14992 (longest_simple_path): Likewise.
14993 * shrink-wrap.c (spread_components): Likewise.
14994 (disqualify_problematic_components): Likewise.
14995 (emit_common_heads_for_components): Likewise.
14996 (emit_common_tails_for_components): Likewise.
14997 (insert_prologue_epilogue_for_components): Likewise.
14999 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15001 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15004 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15006 * df-core.c (df_set_blocks): Start using auto_bitmap.
15007 (df_compact_blocks): Likewise.
15008 * df-problems.c (df_rd_confluence_n): Likewise.
15009 * df-scan.c (df_insn_rescan_all): Likewise.
15010 (df_process_deferred_rescans): Likewise.
15011 (df_update_entry_block_defs): Likewise.
15012 (df_update_exit_block_uses): Likewise.
15013 (df_entry_block_bitmap_verify): Likewise.
15014 (df_exit_block_bitmap_verify): Likewise.
15015 (df_scan_verify): Likewise.
15016 * lra-constraints.c (lra_constraints): Likewise.
15017 (undo_optional_reloads): Likewise.
15018 (lra_undo_inheritance): Likewise.
15019 * lra-remat.c (calculate_gen_cands): Likewise.
15020 (do_remat): Likewise.
15021 * lra-spills.c (assign_spill_hard_regs): Likewise.
15022 (spill_pseudos): Likewise.
15023 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15024 (bitmap_set_subtract_values): Likewise.
15026 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15028 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15029 management with auto_bitmap.
15030 (fix_inter_tick): Likewise.
15031 (fix_recovery_deps): Likewise.
15032 * ira.c (add_store_equivs): Likewise.
15033 (find_moveable_pseudos): Likewise.
15034 (split_live_ranges_for_shrink_wrap): Likewise.
15035 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15036 (rtx_reuse_manager::seen_def_p): Likewise.
15037 (rtx_reuse_manager::set_seen_def): Likewise.
15038 * print-rtl.h (class rtx_reuse_manager): Likewise.
15040 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15042 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15044 (migrate_btr_def): Likewise.
15045 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15046 * df-core.c (loop_post_order_compute): Likewise.
15047 (loop_inverted_post_order_compute): Likewise.
15048 * hsa-common.h: Likewise.
15049 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15050 * init-regs.c (initialize_uninitialized_regs): Likewise.
15051 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15052 (inline_small_functions): Likewise.
15053 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15054 * ira.c (combine_and_move_insns): Likewise.
15055 (build_insn_chain): Likewise.
15056 * loop-invariant.c (find_invariants): Likewise.
15057 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15058 * predict.c (tree_predict_by_opcode): Likewise.
15059 (predict_paths_leading_to): Likewise.
15060 (predict_paths_leading_to_edge): Likewise.
15061 (estimate_loops_at_level): Likewise.
15062 (estimate_loops): Likewise.
15063 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15064 (spread_components): Likewise.
15065 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15066 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15067 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15068 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15069 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15070 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15071 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15072 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15073 (mark_threaded_blocks): Likewise.
15074 (thread_through_all_blocks): Likewise.
15075 * tree-ssa.c (verify_ssa): Likewise.
15076 (execute_update_addresses_taken): Likewise.
15077 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15079 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15081 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15083 (post_order_compute): Likewise.
15084 (inverted_post_order_compute): Likewise.
15085 (pre_and_rev_post_order_compute_fn): Likewise.
15087 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15089 * genrecog.c (int_set::int_set): Explicitly construct our
15090 auto_vec base class.
15091 * vec.h (auto_vec::auto_vec): New constructor.
15093 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15095 * bitmap.h (class auto_bitmap): New constructor taking
15096 bitmap_obstack * argument.
15098 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15100 * bitmap.h (class auto_bitmap): Change type of m_bits to
15101 bitmap_head, and adjust ctor / dtor and member operators.
15103 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15105 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15106 when returned register mode doesn't match original mode.
15108 2017-05-12 Jeff Law <law@redhat.com>
15109 Jakub Jelinek <jakub@redhat.com>
15111 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15112 we look for cc setter after the compare-elim changes.
15113 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15114 within the vector to match what compare-elim now expects.
15115 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15116 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15118 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15119 after the compare-elim changes.
15120 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15121 the vector to match what compare-elim now expects.
15122 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15123 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15124 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15125 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15126 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15128 * config/visium/visium.c (single_set_and_flags): Fix where
15129 we look for cc setter after the compare-elim changes.
15130 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15131 with the vector to match what compare-elim now expects.
15132 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15133 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15134 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15135 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15136 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15137 (neg<mode>2_insn_set_overflow): Likewise.
15139 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15141 PR middle-end/79794
15142 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15143 maybe_expand_insn call, set ops[0].target. If still set after call,
15144 set alt_rtl. Add extra arg to recursive calls.
15145 (extract_bit_field): Add alt_rtl argument. Pass to
15147 * expmed.h (extract_bit_field): Fix prototype.
15148 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15149 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15150 to extract_bit_field_calls.
15151 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15152 Pass alt_rtl to extract_bit_field calls.
15153 * calls.c (store_unaligned_arguments_into_psuedos)
15154 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15155 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15157 * optabs.h (struct expand_operand): Add target bitfield.
15159 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15161 * compare-elim.c (try_eliminate_compare): Canonicalize
15162 operation with embedded compare to
15163 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15164 (set (reg) (operation)].
15166 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15168 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15171 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15172 cost of adding a carry flag for ADC instruction.
15173 [case MINUS]: Ignore the cost of subtracting a carry flag
15174 for SBB instruction.
15176 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15178 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15180 * config/rs6000/bmiintrin.h: New file.
15181 * config/rs6000/bmi2intrin.h: New file.
15182 * config/rs6000/x86intrin.h: New file.
15184 2017-05-12 Jeff Law <law@redhat.com>
15186 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15189 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15191 PR middle-end/80707
15192 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15194 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15196 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15197 early expansion of vector divide builtins.
15198 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15199 builtins identified as having unsigned arguments.
15201 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15203 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15204 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15205 expansion of vector logical operations (and, andc, or, xor,
15208 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15210 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15211 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15213 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15215 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15216 early GIMPLE expansion of vector multiplies.
15218 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15220 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15221 TARGET_HAVE_MOVT conditional.
15222 (movt splitter): Likewise.
15224 2017-05-12 Richard Biener <rguenther@suse.de>
15226 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15227 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15228 Fold all stmts not inplace.
15230 2017-05-12 Richard Biener <rguenther@suse.de>
15232 PR tree-optimization/80713
15233 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15234 inserted_exprs bit for not removed stmts.
15236 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15238 PR middle-end/69921
15239 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15240 parallelized" attribute for parallelized OpenACC kernels.
15241 * omp-offload.c (execute_oacc_device_lower): Use it.
15243 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15244 Set "oacc kernels" attribute.
15245 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15246 parameter. Adjust all users.
15247 (oacc_fn_attrib_kernels_p): Remove function.
15248 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15249 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15250 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15251 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15252 assert "oacc kernels" attribute is set.
15254 2017-05-11 Carl Love <cel@us.ibm.com>
15256 * config/rs6000/rs6000-c: Add support for built-in functions
15257 vector unsigned char vec_popcnt (vector signed char)
15258 vector unsigned char vec_popcnt (vector unsigned char)
15259 vector unsigned short vec_popcnt (vector signed short)
15260 vector unsigned short vec_popcnt (vector unsigned short)
15261 vector unsigned int vec_popcnt (vector signed int)
15262 vector unsigned int vec_popcnt (vector unsigned int)
15263 vector unsigned long long vec_popcnt (vector signed long long)
15264 vector unsigned long long vec_popcnt (vector unsigned long long)
15265 vector signed long long vec_slo (vector signed long long,
15266 vector signed char)
15267 vector signed long long vec_slo (vector signed long long,
15268 vector unsigned char)
15269 vector unsigned long long vec_slo (vector unsigned long long,
15270 vector signed char)
15271 vector unsigned long long vec_slo (vector unsigned long long,
15272 vector unsigned char)
15273 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15274 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15275 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15276 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15277 * doc/extend.texi: Update the built-in documentation file for the
15278 new built-in functions.
15280 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15282 * attribs.h (sorted_attr_string): Move machine independent
15283 functions for target clone support from the i386 port to common
15284 code. Rename ix86_function_versions to common_function_versions.
15285 Rename make_name to make_unique_name.
15286 (common_function_versions): Likewise.
15287 (make_unique_name): Likewise.
15288 (make_dispatcher_decl): Likewise.
15289 (is_function_default_version): Likewise.
15290 * attribs.c (attr_strcmp): Likewise.
15291 (sorted_attr_string): Likewise.
15292 (common_function_versions): Likewise.
15293 (make_unique_name): Likewise.
15294 (make_dispatcher_decl): Likewise.
15295 (is_function_default_version): Likewise.
15296 * config/i386/i386.c (attr_strcmp): Likewise.
15297 (sorted_attr_string): Likewise.
15298 (ix86_function_versions): Likewise.
15299 (make_name): Likewise.
15300 (make_dispatcher_decl): Likewise.
15301 (is_function_default_version): Likewise.
15302 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15304 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15307 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15308 Account for direct move costs for vec_construct of integer
15311 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15314 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15315 (UNSPEC_STX_ATOMIC): Ditto.
15316 (loaddi_via_sse): New insn.
15317 (storedi_via_sse): Ditto.
15318 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15319 Update corresponding peephole2 patterns.
15320 (atomic_storedi_fpu): Ditto.
15322 2017-05-11 Julia Koval <julia.koval@intel.com>
15324 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15325 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15327 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15328 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15329 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15331 2017-05-11 Nathan Sidwell <nathan@acm.org>
15333 * graphite-poly.c: Include dumpfile.h.
15335 * dumpfle.h (dump_function): Declare here ...
15336 * tree-dump.h (dump_function): ... not here.
15337 * dumpfile.c: #include tree-cfg.h.
15338 (dump_function): Move here from ...
15339 * tree-dump.c (dump_function): ... here.
15340 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15341 * graphite-poly.c: Don't include tree-dump.h.
15342 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15343 * print-tree.c: Likewise.
15344 * stor-layout.c: Likewise.
15345 * tree-nested.c: Likewise.
15347 * dumpfile.c (dump_start): Use TDF_FLAGS.
15348 (dump_enable_all): Fix TDF_KIND check thinko.
15350 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15352 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15353 array entries to represent two legal parameterizations of the
15354 overloaded __builtin_cmpb function, as represented by the
15355 P6_OV_BUILTIN_CMPB constant.
15356 (altivec_resolve_overloaded_builtin): Add special case handling
15357 for the __builtin_cmpb function, as represented by the
15358 P6_OV_BUILTIN_CMPB constant.
15359 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15360 (BU_P6_64BIT_2): New macro.
15361 (BU_P6_OVERLOAD_2): New macro
15362 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15363 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15364 (CMPB): Add overload support to represent both 32-bit and 64-bit
15365 compare-bytes function.
15366 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15367 support for TARGET_CMPB.
15368 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15369 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15370 documentation of the __builtin_cmpb overloaded built-in function.
15372 2017-05-11 Richard Biener <rguenther@suse.de>
15374 PR tree-optimization/80705
15375 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15376 bases are not vectorizable.
15378 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15380 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15381 when counting register pressure.
15383 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15385 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15386 (iv_ca_more_deps): Renamed to ...
15387 (iv_ca_compare_deps): ... this.
15388 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15390 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15392 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15394 (determine_group_iv_costs): ... here.
15395 (find_inv_vars_cb): Record inv var if it's not recorded before.
15397 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15399 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15400 (get_shiftadd_cost): Ditto.
15402 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15404 * tree-ssa-address.c: Include header file.
15405 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15407 (add_to_parts): Refactor.
15408 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15409 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15412 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15414 PR tree-optimization/53090
15415 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15417 (extract_cond_operands): Detect condition with IV on both sides
15418 and return COMP_IV_EXPR_2.
15419 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15420 (rewrite_use_compare): Simplify by removing call to function
15421 extract_cond_operands.
15423 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15425 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15426 (extract_cond_operands): Detect condition comparing against non-
15427 invariant bound and return appropriate enum value.
15428 (find_interesting_uses_cond): Update use of extract_cond_operands.
15429 Handle its return value accordingly.
15430 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15432 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15434 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15435 nonlinear iv_use computation in loop invariant sensitive way.
15437 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15439 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15440 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15442 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15444 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15445 (dump_cand): Support iv_cand.inv_exprs.
15446 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15448 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15451 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15453 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15455 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15456 as local function. Include necessary header files.
15457 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15459 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15461 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15463 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15465 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15466 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15467 RSHIFT_EXPR and BIT_NOT_EXPR.
15469 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15471 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15472 (adjust_setup_cost): New parameter supporting round up adjustment.
15473 (struct address_cost_data): Delete.
15474 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15475 (split_address_cost, ptr_difference_cost): Delete.
15476 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15477 (struct ainc_cost_data): New struct.
15478 (get_address_cost_ainc): New function.
15479 (get_address_cost, get_computation_cost): Reimplement.
15480 (determine_group_iv_cost_address): Record inv_expr for all uses of
15482 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15483 (iv_ca_has_deps): Reimplemented to ...
15484 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15486 (iv_ca_extend): Call iv_ca_more_deps.
15488 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15490 * tree-ssa-address.c (struct mem_address): Move to header file.
15491 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15492 * tree-ssa-address.h (struct mem_address): Move from C file.
15493 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15495 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15497 * tree-affine.h (aff_combination_type): New interface.
15498 (aff_combination_zero_p): Remove static.
15499 (aff_combination_const_p): New interface.
15500 (aff_combination_singleton_var_p): New interfaces.
15502 2017-05-11 Richard Biener <rguenther@suse.de>
15504 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15505 Skip unreachable blocks and destinations.
15506 (eliminate): Move stmt removal and fixup ...
15507 (fini_eliminate): ... here. Skip inserted exprs.
15508 (pass_pre::execute): Move fini_pre after fini_eliminate.
15509 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15510 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15511 PRE to get rid of dead code that has invalid SSA form and
15512 split critical edges again.
15514 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15516 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15518 2017-05-11 Richard Biener <rguenther@suse.de>
15520 * passes.c (execute_function_todo): Verify loops if they are
15521 said to be up-to-date.
15522 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15523 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15525 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15528 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15529 handle calling assemble_external ourself.
15532 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15533 modes with zero size. Enhance comment.
15535 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15537 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15538 built-ins for vec_xl and vec_xst with short and char pointer
15541 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15543 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15544 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15545 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15546 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15547 (_mm_maskz_min_round_ss): New intrinsics.
15548 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15549 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15550 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15551 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15552 (__builtin_ia32_minss_mask_round): New builtins.
15553 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15554 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15555 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15557 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15558 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15560 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15563 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15565 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15566 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15567 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15568 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15569 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15570 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15571 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15572 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15573 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15574 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15575 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15576 (__builtin_ia32_mulss_mask_round): New builtins.
15577 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15578 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15579 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15581 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15582 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15584 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15587 2017-05-10 Julia Koval <julia.koval@intel.com>
15589 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15590 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15591 (_mm256_setr_m128i): New intrinsics.
15593 2017-05-10 Julia Koval <julia.koval@intel.com>
15595 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15596 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15597 (_mm_maskz_rcp14_ss): New intrinsics.
15598 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15599 (__builtin_ia32_rcp14ss_mask): New builtins.
15600 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15602 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15604 PR tree-optimization/51513
15605 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15606 (assert_unreachable_fallthru_edge_p): Use it.
15607 (group_case_labels_stmt): Likewise.
15608 * tree-cfg.h: Prototype it.
15609 * stmt.c: Include cfghooks.h and tree-cfg.h.
15610 (emit_case_dispatch_table) <gap_label>: New local variable.
15611 Use it to fill dispatch table gaps.
15612 Test for default_label before updating probabilities.
15613 (expand_case) <default_label>: Remove unneeded initialization.
15614 Test for unreachable default case statement and remove its edge.
15615 Set default_label accordingly.
15616 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15618 2017-05-10 Carl Love <cel@us.ibm.com>
15620 * config/rs6000/rs6000-c: Add support for built-in functions
15621 vector signed char vec_neg (vector signed char)
15622 vector signed short int vec_neg (vector short int)
15623 vector signed int vec_neg (vector signed int)
15624 vector signed long long vec_neg (vector signed long long)
15625 vector float vec_neg (vector float)
15626 vector double vec_neg (vector double)
15627 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15629 * config/rs6000/altivec.h: Add define for vec_neg
15630 * doc/extend.texi: Update the built-in documentation for the
15631 new built-in functions.
15633 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15635 PR tree-optimization/77644
15636 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15638 2017-05-10 Nathan Sidwell <nathan@acm.org>
15640 * dumpfile.h (TDI_lang_all): New.
15641 (TDF_KIND): New. Renumber others
15642 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15644 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15646 (get_dump_file_name): Adjust suffix generation.
15647 (dump_enable_all): Use TDF_KIND.
15648 * doc/invoke.texi (-fdump-lang-all): Document.
15650 * dumpfile.h: Tabify.
15652 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15655 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15656 Move member access before delete.
15658 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15660 * tree-inline.c (expand_call_inline): Split block at stmt
15663 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15666 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15667 are now unused after splitting mov{sf,sd}_hardfloat.
15668 (f32_lr2): Likewise.
15669 (f32_lm): Likewise.
15670 (f32_lm2): Likewise.
15671 (f32_li): Likewise.
15672 (f32_li2): Likewise.
15673 (f32_lv): Likewise.
15674 (f32_sr): Likewise.
15675 (f32_sr2): Likewise.
15676 (f32_sm): Likewise.
15677 (f32_sm2): Likewise.
15678 (f32_si): Likewise.
15679 (f32_si2): Likewise.
15680 (f32_sv): Likewise.
15681 (f32_dm): Likewise.
15682 (f32_vsx): Likewise.
15683 (f32_av): Likewise.
15684 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15685 For movsf, order stores so the VSX stores occur before the GPR
15686 store which encourages the register allocator to use a traditional
15687 FPR instead of a GPR. For movsd, order the stores so that the GPR
15688 store comes before the VSX stores to allow the power6 to work.
15689 This is due to the power6 not having a 32-bit integer store
15690 instruction from a FPR.
15691 (movsf_hardfloat): Likewise.
15692 (movsd_hardfloat): Likewise.
15694 2017-05-09 Martin Sebor <msebor@redhat.com>
15696 PR translation/80280
15697 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15700 PR translation/80280
15701 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15702 data member added in r247778.
15703 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15705 2017-05-09 Nathan Sidwell <nathan@acm.org>
15707 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15709 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15712 2017-05-09 Marek Polacek <polacek@redhat.com>
15714 * doc/invoke.texi: Fix typo.
15716 2017-05-09 Richard Biener <rguenther@suse.de>
15718 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15719 (vrp_val_is_min): Likewise.
15720 (set_value_range_to_value): Likewise.
15721 (set_value_range_to_nonnegative): Likewise.
15722 (gimple_assign_nonzero_p): Likewise.
15723 (gimple_stmt_nonzero_p): Likewise.
15724 (vrp_int_const_binop): Likewise. Remove unreachable case.
15725 (adjust_range_with_scev): Adjust comments.
15726 (compare_range_with_value): Likewise.
15727 (extract_range_from_phi_node): Likewise.
15728 (test_for_singularity): Likewise.
15730 2017-05-09 Richard Biener <rguenther@suse.de>
15732 * tree-vrp.c (get_single_symbol): Add assert that we don't
15733 get overflowed constants as invariant part.
15734 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15735 checks. Use wi::cmp instead of recursing for integer constants.
15736 (compare_values): Just ignore whether we assumed undefined
15737 overflow instead of failing the compare.
15738 (extract_range_for_var_from_comparison_expr): Add comment before the
15739 TREE_NO_WARNING sets.
15740 (test_for_singularity): Likewise.
15741 (extract_range_from_comparison): Do not disable optimization
15742 when we assumed undefined overflow.
15743 (extract_range_basic): Remove init of unused var.
15745 2017-05-09 Richard Biener <rguenther@suse.de>
15747 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15748 (extract_range_from_multiplicative_op_1): Adjust.
15749 (extract_range_from_binary_expr_1): Use int_const_binop.
15751 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15754 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15755 rs6000_store_data_bypass_p in seven define_bypass directives and
15756 in several comments.
15757 * config/rs6000/rs6000-protos.h: Add prototype for
15758 rs6000_store_data_bypass_p function.
15759 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15760 function implements slightly different (rs6000-specific) semantics
15761 than store_data_bypass_p, returning false rather than aborting
15762 with assertion error when arguments do not satisfy the
15763 requirements of store data bypass.
15764 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15765 rs6000_store_data_bypass_p.
15767 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15769 * config/xtensa/xtensa-protos.h
15770 (xtensa_initial_elimination_offset): New declaration.
15771 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15772 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15773 macro definition, add case for FRAME_POINTER_REGNUM when
15774 FRAME_GROWS_DOWNWARD.
15775 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15776 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15777 xtensa_initial_elimination_offset.
15779 2017-05-08 Nathan Sidwell <nathan@acm.org>
15781 * doc/invoke.texi: Alphabetize -fdump options.
15783 2017-05-08 Martin Sebor <msebor@redhat.com>
15785 PR translation/80280
15786 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15788 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15790 * target.def (compute_frame_layout): New optional target hook.
15791 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15792 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15793 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15795 * reload1.c (verify_initial_elim_offsets): Likewise.
15796 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15797 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15798 (arm_get_frame_offsets): Split up into this ...
15799 (arm_compute_frame_layout): ... and this function.
15801 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15803 * config/aarch64/constraints.md (Usa): New constraint.
15804 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15806 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15808 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15809 with_multilib_list after it has been checked.
15811 2017-05-08 Richard Biener <rguenther@suse.de>
15813 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15814 (bitmap_set_subtract_values): Likewise.
15816 2017-05-08 Richard Biener <rguenther@suse.de>
15818 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15819 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15821 (gimple_stmt_nonzero_warnv_p): Rename to ...
15822 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15824 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15825 (extract_range_basic): Adjust, do not disable propagation on
15826 strict overflow sensitive simplification.
15827 (vrp_visit_cond_stmt): Likewise.
15829 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15831 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15832 body size unconditionally.
15834 2017-05-07 Jeff Law <law@redhat.com>
15837 2017-05-06 Jeff Law <law@redhat.com>
15838 PR tree-optimization/78496
15839 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15842 PR tree-optimization/78496
15843 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15844 (simplify_stmt_using_ranges): Call it.
15845 (vrp_dom_walker::before_dom_children): Extract equivalences
15846 from an ASSERT_EXPR with an equality comparison against a
15849 2017-05-06 Jeff Law <law@redhat.com>
15851 PR tree-optimization/78496
15852 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15855 PR tree-optimization/78496
15856 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15857 (simplify_stmt_using_ranges): Call it.
15858 (vrp_dom_walker::before_dom_children): Extract equivalences
15859 from an ASSERT_EXPR with an equality comparison against a
15862 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15864 * lra-constraints.c (lra_copy_reg_equiv): New function.
15865 (split_reg): Use it to copy equivalence information from the
15866 original register to the spill register.
15868 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15870 PR rtl-optimization/75964
15871 * simplify-rtx.c (simplify_const_relational_operation): Remove
15872 invalid handling of comparisons of integer ABS.
15874 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15876 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15877 initialize to zero.
15878 (init_regs): Remove declaration.
15879 (function_arg_advance_32): Initialize error_p as boolean variable.
15881 2017-05-05 Nathan Sidwell <nathan@acm.org>
15883 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15884 lines. Use for (;;).
15886 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15888 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15889 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15890 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15891 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15892 VF=2 that require versioning.
15894 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15896 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15899 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15901 * diagnostic.h (diagnostic_override_option_index): Convert from
15902 macro to inline function.
15904 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15906 * diagnostic.c (last_module_changed_p): New function.
15907 (set_last_module): New function.
15908 (diagnostic_report_current_module): Convert macro usage to
15909 the above functions.
15910 * diagnostic.h (diagnostic_context::last_module): Strengthen
15911 from const line_map * to const line_map_ordinary *.
15912 (diagnostic_last_module_changed): Delete macro.
15913 (diagnostic_set_last_module): Delete macro.
15915 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15917 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15918 with diagnostic_report_diagnostic.
15919 (diagnostic_n_impl_richloc): Likewise.
15920 * diagnostic.h (report_diagnostic): Delete macro.
15921 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15922 with diagnostic_report_diagnostic.
15923 * substring-locations.c (format_warning_va): Likewise.
15925 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15927 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15928 save/restor of format_spec. Move option-printing code to...
15929 (print_option_information): ...this new function, and
15930 reimplement by simply printing to the pretty_printer,
15931 rather than appending to the format string.
15933 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15935 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15936 handling logic into...
15937 (update_effective_level_from_pragmas): ...this new function.
15939 2017-05-04 Andrew Waterman <andrew@sifive.com>
15941 * config/riscv/riscv.opt (mstrict-align): New option.
15942 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15943 (SLOW_UNALIGNED_ACCESS): Define.
15944 (riscv_slow_unaligned_access): Declare.
15945 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15947 (riscv_slow_unaligned_access): New variable.
15948 (rocket_tune_info): Set slow_unaligned_access to true.
15949 (optimize_size_tune_info): Set slow_unaligned_access to false.
15950 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15951 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15952 (riscv_option_override): Set riscv_slow_unaligned_access.
15953 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15955 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15957 * config/riscv/riscv.md: Unify indentation.
15959 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15964 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15966 (extendsi<mode>2): Add support for doing sign extension via
15967 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
15968 don't have ISA 3.0 instructions.
15969 (extendsi<mode>2 splitter): Likewise.
15970 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
15971 generate the normal insns since SImode can now go in vector
15972 registers. Disallow the special UNSPECs needed for previous
15973 machines to hide SImode being used. Add new insns
15974 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
15975 (fix_trunc<mode>si2_stfiwx): Likewise.
15976 (fix_trunc<mode>si2_internal): Likewise.
15977 (fixuns_trunc<mode>si2): Likewise.
15978 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15979 (fctiw<u>z_<mode>_smallint): Likewise.
15980 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
15981 of floating point to 32-bit integer from doing a direct move to
15982 the GPR registers to do a store.
15983 (fctiwz_<mode>): Break long line.
15985 2017-05-05 Bin Cheng <bin.cheng@arm.com>
15987 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
15988 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
15989 (addr_list, addr_offset_valid_p): New.
15990 (split_address_groups): Check offset validity with above function.
15991 (gt-tree-ssa-loop-ivopts.h): Include header file.
15993 2017-05-05 Nathan Sidwell <nathan@acm.org>
15995 * config.gcc (arm*-*-*): Add missing 'fi'.
15997 2017-05-05 Steve Ellcey <sellcey@cavium.com>
15999 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16000 included in -fopt-info does not matter.
16001 * doc/optinfo.texi (-fopt-info): Fix description of default
16002 behavour. Explicitly say order of options included in -fopt-info
16005 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16007 * config.gcc: Allow combinations of aprofile and rmprofile values for
16008 --with-multilib-list.
16009 * config/arm/t-multilib: New file.
16010 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16011 variables. Remove setting of ISA and floating-point ABI in
16012 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16013 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16014 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16015 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16017 * config/arm/t-rmprofile: Likewise except for the matches changes.
16018 * doc/install.texi (--with-multilib-list): Document the combination of
16019 aprofile and rmprofile values and warn about pitfalls in doing that.
16021 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16023 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16024 (movdi_aarch64): Likewise.
16026 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16028 PR tree-optimization/80632
16029 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16031 (build_arrays): Initialize it for virtual phis.
16032 (fix_phi_nodes): Use it for virtual phis.
16034 PR tree-optimization/80558
16035 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16036 [x, y] op z into [x op, y op z] for op & or | if conditions
16039 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16040 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16043 * config/arm/arm.md (use_literal_pool): Remove.
16044 (64-bit immediate split): No longer takes cost into consideration
16045 if arm_disable_literal_pool is enabled.
16046 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16047 used when arm_disable_literal_pool is enabled.
16048 (arm_max_const_double_inline_cost): Remove use of
16049 arm_disable_literal_pool.
16050 (push_minipool_fix): Add assert.
16051 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16052 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16053 (no_literal_pool_sf_immediate): New.
16055 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16057 PR tree-optimization/80613
16058 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16059 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16061 2017-05-05 Richard Biener <rguenther@suse.de>
16063 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16065 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16067 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16068 of this flag from insn conditions due to removal from r247495.
16070 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16072 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16074 (arm_early_store_addr_dep_ptr): Likewise.
16075 * config/arm/aarch-common-protos.h
16076 (arm_early_load_addr_dep_ptr): Add prototype.
16077 (arm_early_store_addr_dep_ptr): Likewise.
16078 * config/arm/cortex-a53.md: Add new bypasses.
16080 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16082 * tree.c (next_type_uid): Change type to unsigned.
16083 (type_hash_canon): Decrement back next_type_uid if
16084 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16085 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16088 2017-05-04 Martin Sebor <msebor@redhat.com>
16090 * builtins.c: Fix a trivial typo in a comment.
16092 PR middle-end/79234
16093 * builtins.c (check_sizes): Adjust to handle reading past the end.
16094 Avoid printing excessive upper bound of ranges. Use %E to print
16095 tree nodes instead of converting them to %wu.
16096 (expand_builtin_memchr): New function.
16097 (compute_dest_size): Rename...
16098 (compute_objsize): ...to this.
16099 (expand_builtin_memcpy): Adjust.
16100 (expand_builtin_mempcpy): Adjust.
16101 (expand_builtin_strcat): Adjust.
16102 (expand_builtin_strcpy): Adjust.
16103 (check_strncat_sizes): Adjust.
16104 (expand_builtin_strncat): Adjust.
16105 (expand_builtin_strncpy): Adjust and simplify.
16106 (expand_builtin_memset): Adjust.
16107 (expand_builtin_bzero): Adjust.
16108 (expand_builtin_memcmp): Adjust.
16109 (expand_builtin): Handle memcmp.
16110 (maybe_emit_chk_warning): Check strncat just once.
16112 2017-05-04 Martin Sebor <msebor@redhat.com>
16114 PR preprocessor/79214
16115 PR middle-end/79222
16116 PR middle-end/79223
16117 * builtins.c (check_sizes): Add inlining context and issue
16118 warnings even when -Wno-system-headers is set.
16119 (check_strncat_sizes): Same.
16120 (expand_builtin_strncat): Same.
16121 (expand_builtin_memmove): New function.
16122 (expand_builtin_stpncpy): Same.
16123 (expand_builtin): Handle memmove and stpncpy.
16125 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16127 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16128 which is not used any more.
16130 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16132 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16134 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16136 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16137 (cortexa53_tunings): Likewise.
16138 (cortexa57_tunings): Likewise.
16139 (cortexa72_tunings): Likewise.
16140 (cortexa73_tunings): Likewise.
16142 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16144 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16145 Set loop alignment to 8.
16147 2017-05-04 Martin Sebor <msebor@redhat.com>
16149 PR translation/80280
16150 * builtins.c (expand_builtin_object_size): Add missing quoting to
16151 %D and like directives.
16152 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16153 (hsa_type_for_tree_type): Same.
16154 (verify_function_arguments): Same.
16155 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16156 * varasm.c (get_section): Same.
16159 2017-05-04 Martin Sebor <msebor@redhat.com>
16161 PR translation/80280
16162 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16164 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16166 * config/aarch64/aarch64.c (generic_addrcost_table):
16167 Change HI/TI mode setting.
16169 2017-05-04 Martin Jambor <mjambor@suse.cz>
16171 PR tree-optimization/80622
16172 * tree-sra.c (comes_initialized_p): New function.
16173 (build_accesses_from_assign): Only set write lazily when
16174 comes_initialized_p is false.
16175 (analyze_access_subtree): Use comes_initialized_p.
16176 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16177 instead of testing for PARM_DECL.
16179 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16181 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16182 constraint on operand 0 to allow more general addressing modes.
16183 Adjust output template.
16184 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16186 * config/aarch64/aarch64-protos.h
16187 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16188 * config/aarch64/constraints.md (Dp): New address constraint.
16189 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16192 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16194 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16195 update use of estimate_ipcp_clone_size_and_time.
16196 (estimate_local_effects): Update use of
16197 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16198 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16199 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16200 Return nonspecialized time.
16202 2017-05-04 Richard Biener <rguenther@suse.de>
16204 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16205 for the last VUSE which def dominates the PHI. Directly call
16207 (get_continuation_for_phi_1): Remove.
16209 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16211 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16212 to explain the use of truncating division. Cap the number of
16213 iterations to the maximum given by nb_iterations_upper_bound,
16216 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16218 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16219 * configure: Regenerate.
16220 * config.in: Regenerate.
16221 * config/i386/driver-mingw32.c: new file.
16222 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16223 * config.host: Link driver-mingw32.o on MinGW host.
16224 * doc/install.texi: Document new --enable-mingw-wildcard configure
16227 2017-05-04 Marek Polacek <polacek@redhat.com>
16229 PR tree-optimization/80612
16230 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16232 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16233 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16235 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16236 (movt splitter): Likewise.
16237 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16238 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16239 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16240 block for Thumb-1 with MOVT.
16241 (thumb2_legitimate_address_p): Move code block ...
16242 (can_avoid_literal_pool_for_label_p): ... into this new function.
16243 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16245 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16246 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16247 "M-profile targets with the MOVT instruction".
16249 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16251 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16252 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16253 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16255 2017-05-04 Martin Liska <mliska@suse.cz>
16257 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16258 variable cond_code.
16260 2017-05-04 Richard Biener <rguenther@suse.de>
16262 * tree.c (array_at_struct_end_p): Handle arrays at struct
16263 end with flexarrays more conservatively. Refactor and treat
16264 arrays of arrays or aggregates more strict. Fix
16265 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16266 * tree.c (array_at_struct_end_p): Adjust prototype.
16267 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16268 * gimple-fold.c (get_range_strlen): Likewise.
16269 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16271 2017-05-04 Richard Biener <rguenther@suse.de>
16273 PR tree-optimization/31130
16274 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16276 (supports_overflow_infinity): Likewise.
16277 (is_negative_overflow_infinity): Likewise.
16278 (is_positive_overflow_infinity): Likewise.
16279 (is_overflow_infinity): Likewise.
16280 (stmt_overflow_infinity): Likewise.
16281 (overflow_infinity_range_p): Likewise.
16282 (usable_range_p): Remove as always returning true.
16283 (make_overflow_infinity): Remove.
16284 (negative_overflow_infinity): Likewise.
16285 (positive_overflow_infinity): Likewise.
16286 (avoid_overflow_infinity): Likewise.
16287 (set_value_range): Adjust accordingly.
16288 (set_value_range_to_nonnegative): Likewise, remove now unused
16289 overflow_infinity arg.
16290 (vrp_operand_equal_p): Adjust.
16291 (update_value_range): Likewise.
16292 (range_int_cst_singleton_p): Likewise.
16293 (operand_less_p): Likewise.
16294 (compare_values_warnv): Likewise.
16295 (extract_range_for_var_from_comparison_expr): Likewise.
16296 (vrp_int_const_binop): Likewise.
16297 (zero_nonzero_bits_from_vr): Likewise.
16298 (extract_range_from_multiplicative_op_1): Likewise.
16299 (extract_range_from_binary_expr_1): Likewise.
16300 (extract_range_from_unary_expr): Likewise.
16301 (extract_range_from_comparison): Likewise.
16302 (extract_range_basic): Likewise.
16303 (adjust_range_with_scev): Likewise.
16304 (compare_ranges): Likewise.
16305 (compare_range_with_value): Likewise.
16306 (dump_value_range): Likewise.
16307 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16309 (simplify_cond_using_ranges): Adjust.
16311 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16313 * brig-builtins.def: Added a builtin for class_f64.
16314 * builtin-types.def: Added a builtin type needed by class_f64.
16316 2017-05-03 Jason Merrill <jason@redhat.com>
16318 * timevar.def: Add TV_CONSTEXPR.
16320 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16322 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16324 2017-05-03 Martin Jambor <mjambor@suse.cz>
16326 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16327 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16328 * ipa-cp.c (ipcp_propagate_stage): Do not call
16329 ipa_update_after_lto_read.
16330 * ipa-inline.c (ipa_inline): Likewise.
16332 2017-05-03 Martin Jambor <mjambor@suse.cz>
16334 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16335 tag. Added a default constructor and a destructor.
16336 (ipa_edge_args_sum_t): New class;
16337 (ipa_edge_args_sum): Declare.
16338 (ipa_edge_args_vector): Remove declaration.
16339 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16340 (ipa_free_edge_args_substructures): Remove declaration.
16341 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16342 (ipa_edge_args_info_available_for_edge_p): Likewise.
16343 * ipa-prop.c (ipa_edge_args_vector): Removed.
16344 (edge_removal_hook_holder): Likewise.
16345 (edge_duplication_hook_holder): Likewise.
16346 (ipa_edge_args_sum): New variable.
16347 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16348 ipa_edge_args_vector.
16349 (ipa_free_edge_args_substructures): Likewise.
16350 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16351 ipa_edge_args_vector.
16352 (ipa_edge_removal_hook): Turned into method
16353 ipa_edge_args_sum_t::remove.
16354 (ipa_edge_duplication_hook): Turned into method
16355 ipa_edge_args_sum_t::duplicate.
16356 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16357 registering edge hooks.
16358 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16359 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16360 ipa_edge_args_sum instead of ipa_edge_args_vector.
16361 * ipa-profile.c (ipa_profile): Likewise.
16363 2017-05-03 Martin Jambor <mjambor@suse.cz>
16365 * symbol-summary.h (function_summary): New method exists.
16366 (function_summary::symtab_removal): Deallocate through release.
16367 (call_summary): New class.
16368 (gt_ggc_mx): New overload.
16369 (gt_pch_nx): Likewise.
16370 (gt_pch_nx): Likewise.
16372 2017-05-03 Jeff Law <law@redhat.com>
16374 PR tree-optimization/78496
16375 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16376 from simplify_cond_using_ranges. Split off code to walk
16377 backwards through casts into ...
16378 (simplify_cond_using_ranges_2): New function.
16379 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16380 (execute_vrp): After identifying jump threads, call
16381 simplify_cond_using_ranges_2.
16383 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16386 * ipa-inline.h (inline_summary): Add ctor.
16387 (create_ggc): Do not use ggc_cleared_alloc.
16389 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16390 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16392 * gcc.c (handle_braces): Support escaping in switch matching
16394 * doc/invoke.texi (Spec Files): Document it.
16395 Remove superfluous @code markup in items.
16397 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16399 * diagnostic-show-locus.c (struct column_range): New struct.
16400 (get_affected_columns): New function.
16401 (get_printed_columns): New function.
16402 (struct correction): New struct.
16403 (correction::ensure_capacity): New function.
16404 (correction::ensure_terminated): New function.
16405 (struct line_corrections): New struct.
16406 (line_corrections::~line_corrections): New dtor.
16407 (line_corrections::add_hint): New function.
16408 (layout::print_trailing_fixits): Reimplement in terms of the new
16410 (selftest::test_overlapped_fixit_printing): New function.
16411 (selftest::diagnostic_show_locus_c_tests): Call it.
16413 2017-05-03 Nathan Sidwell <nathan@acm.org>
16415 Canonicalize canonical type hashing
16416 * tree.h (type_hash_canon_hash): Declare.
16417 * tree.c (type_hash_list, attribute_hash_list): Move into
16418 type_hash_canon_hash.
16419 (build_type_attribute_qual_variant): Break out hash code calc into
16420 type_hash_canon_hash.
16421 (type_hash_canon_hash): New. Generic type hash computation.
16422 (build_range_type_1, build_array_type_1, build_function_type,
16423 build_method_type_directly, build_offset_type, build_complex_type,
16424 make_vector_type): Call it.
16426 2017-05-03 Richard Biener <rguenther@suse.de>
16428 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16429 When all DRs have unknown misaligned do not always peel
16430 when there is a store but apply the same costing model as if
16431 there were only loads.
16433 2017-05-03 Richard Biener <rguenther@suse.de>
16436 PR tree-optimization/80492
16437 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16438 compare_base_decls returning dont-know properly.
16440 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16442 * config/arm/iterators.md (CCSI): New mode iterator.
16443 (arch): New mode attribute.
16444 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16445 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16446 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16447 code iterator for success result mode.
16448 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16449 the corresponding new insn generators.
16451 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16454 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16455 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16457 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16459 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16460 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16461 (DDR_A): Wrap DDR argument in brackets.
16462 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16463 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16464 (DDR_REVERSED_P): Likewise.
16466 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16468 PR tree-optimization/79472
16469 * tree-switch-conversion.c (struct switch_conv_info): Add
16470 contiguous_range and default_case_nonstandard fields.
16471 (collect_switch_conv_info): Compute contiguous_range and
16472 default_case_nonstandard fields, don't clear final_bb if
16473 contiguous_range and only the default case doesn't have the required
16475 (check_all_empty_except_final): Set default_case_nonstandard instead
16476 of failing if contiguous_range and the default case doesn't have empty
16478 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16479 and only the default case doesn't have the required constants. Skip
16481 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16482 if default_case_nonstandard.
16483 (build_constructors): Build constant 1 just once. Assert that default
16484 values aren't inserted in between cases if contiguous_range. Skip
16486 (build_arrays): Skip virtual phis.
16487 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16488 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16489 Handle virtual phis.
16490 (gen_inbound_check): Handle default_case_nonstandard case.
16491 (process_switch): Adjust check_final_bb caller. Call
16492 gather_default_values with the first non-default case instead of
16493 default case if default_case_nonstandard.
16495 2017-05-02 Nathan Sidwell <nathan@acm.org>
16497 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16498 check. Fix formatting.
16500 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16502 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16503 errors when comparing specialized and unspecialized times.
16505 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16507 * diagnostic-show-locus.c
16508 (layout::should_print_annotation_line_p): Make private.
16509 (layout::print_annotation_line): Make private.
16510 (layout::annotation_line_showed_range_p): Make private.
16511 (layout::show_ruler): Make private.
16512 (layout::print_source_line): Make private. Pass in line and
16513 line_width, rather than calling location_get_source_line. Drop
16515 (layout::print_leading_fixits): New method.
16516 (layout::print_any_fixits): Rename to...
16517 (layout::print_trailing_fixits): ...this, and make private.
16518 Don't print newline fixits.
16519 (diagnostic_show_locus): Move logic for printing one row into...
16520 (layout::print_line): ...this new function. Move the
16521 location_get_source_line call and error-handling from
16522 print_source_line to here. Call print_leading_fixits, and rename
16523 print_any_fixits to print_trailing_fixits.
16524 (selftest::test_fixit_insert_containing_newline): Update now that
16525 newlines are partially supported.
16526 (selftest::test_fixit_insert_containing_newline_2): New test.
16527 (selftest::test_fixit_replace_containing_newline): Update comments.
16528 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16529 * edit-context.c (class added_line): New class.
16530 (class edited_line): Describe newline handling in comment.
16531 (edited_line::actually_edited_p): New method.
16532 (edited_line::print_content): Delete redundant decl.
16533 (edited_line::m_predecessors): New field.
16534 (edited_file::print_content): Call edited_line::print_content.
16535 (edited_file::print_diff): Update to support newlines.
16536 (edited_file::print_diff_hunk): Likewise.
16537 (edited_file::print_run_of_changed_lines): New function.
16538 (edited_file::print_diff_line): Convert to...
16539 (print_diff_line): ...this.
16540 (edited_file::get_effective_line_count): New function.
16541 (edited_line::edited_line): Initialize new field m_predecessors.
16542 (edited_line::~edited_line): Clean up m_predecessors.
16543 (edited_line::apply_fixit): Handle newlines.
16544 (edited_line::get_effective_line_count): New function.
16545 (edited_line::print_content): New function.
16546 (edited_line::print_diff_lines): New function.
16547 (selftest::test_applying_fixits_insert_containing_newline): New
16549 (selftest::test_applying_fixits_replace_containing_newline): New
16551 (selftest::insert_line): New function.
16552 (selftest::test_applying_fixits_multiple_lines): Add example of
16554 (selftest::edit_context_c_tests): Call the new tests.
16556 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16558 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16559 parameter cand. Update dump information.
16560 (get_computation_cost): Update uses.
16562 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16564 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16565 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16566 (get_computation_at, rewrite_use_address): Update use of
16567 get_computation_aff.
16569 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16571 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16572 (get_computation): Delete.
16573 (get_computation_cost): Implement like get_computation_cost_at.
16574 Use get_computation_at.
16575 (get_computation_cost_at): Delete.
16576 (rewrite_use_nonlinear_expr): Use get_computation_at.
16577 (rewrite_use_compare, remove_unused_ivs): Ditto.
16579 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16581 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16583 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16585 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16586 (ivopts_global_cost_for_size): Rename parameter and update uses.
16587 (iv_ca_recount_cost): Update uses.
16588 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16589 candidates seperately in n_invs and n_cands.
16590 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16592 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16594 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16595 (find_inv_vars_cb): New.
16596 (find_depends): Renamed to ...
16597 (find_inv_vars): ... this.
16598 (add_candidate_1, force_var_cost): Call find_inv_vars.
16599 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16601 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16603 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16604 inv_vars. Add inv_exprs.
16605 (struct iv_cand): Rename depends_on to inv_vars.
16606 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16607 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16608 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16609 (dump_cand): Dump inv_vars.
16610 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16611 (record_invariant, find_depends, add_candidate_1): Ditto.
16612 (set_group_iv_cost, force_var_cost): Ditto.
16613 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16614 (get_computation_cost_at, get_computation_cost): Ditto.
16615 (determine_group_iv_cost_generic): Ditto.
16616 (determine_group_iv_cost_address): Ditto.
16617 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16618 (determine_group_iv_costs): Ditto.
16619 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16620 (iv_ca_set_remove_invariants): Renamed to ...
16621 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16622 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16623 (iv_ca_set_add_invariants): Renamed to ...
16624 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16625 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16626 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16627 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16628 (create_new_ivs): Remove useless dump.
16630 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16632 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16634 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16635 (iv_ca_set_no_cp, create_new_iv): Ditto.
16637 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16639 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16641 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16643 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16644 function tree_check2.
16646 2017-05-02 Martin Liska <mliska@suse.cz>
16648 * doc/gcov.texi: Add missing preposition.
16649 * gcov.c (function_info::function_info): Properly fill up
16650 all member variables.
16652 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16654 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16656 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16658 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16660 2017-05-02 Martin Liska <mliska@suse.cz>
16663 * lto-streamer-in.c (lto_read_tree_1): Remove
16664 LTO_STREAMER_DEBUG.
16665 * lto-streamer.c (struct tree_hash_entry): Likewise.
16666 (struct tree_entry_hasher): Likewise.
16667 (tree_entry_hasher::hash): Likewise.
16668 (tree_entry_hasher::equal): Likewise.
16669 (lto_streamer_init): Likewise.
16670 (lto_orig_address_map): Likewise.
16671 (lto_orig_address_get): Likewise.
16672 (lto_orig_address_remove): Likewise.
16673 * lto-streamer.h: Likewise.
16674 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16675 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16677 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16679 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16680 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16681 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16682 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16683 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16684 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16685 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16686 (mm_maskz_sub_ss): New intrinsics.
16687 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16688 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16689 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16690 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16691 (__builtin_ia32_subss_mask_round): New builtins.
16692 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16693 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16694 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16696 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16697 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16699 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16702 2017-05-02 Martin Jambor <mjambor@suse.cz>
16704 PR tree-optimization/78687
16705 * tree-sra.c (access): New field parent.
16706 (process_subtree_disqualification): New function.
16707 (disqualify_candidate): Call it.
16708 (build_accesses_from_assign): Reset write flag if creating an
16710 (build_access_subtree): Fill in parent field and also prpagate
16711 down grp_write flag.
16712 (create_artificial_child_access): New parameter set_grp_write, set
16713 grp_write to its value.
16714 (propagate_subaccesses_across_link): Also propagate grp_write flag
16716 (propagate_all_subaccesses): Push the closest parent back to work
16717 queue if add_access_to_work_queue returned true.
16719 2017-05-02 Richard Biener <rguenther@suse.de>
16721 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16722 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16723 -fstrict-overflow documentation.
16724 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16725 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16726 flag_strict_overflow.
16727 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16728 * lto-opts.c (lto_write_options): Do not stream it.
16729 * lto-wrapper.c (merge_and_complain): Do not handle it.
16730 * opts.c (default_options_table): Do not set -fstrict-overflow.
16731 (finish_options): Likewise do not clear it when sanitizing.
16732 * simplify-rtx.c (simplify_const_relational_operation): Do not
16733 test flag_strict_overflow.
16735 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16737 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16738 using enabled attribute.
16739 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16740 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16741 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16742 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16743 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16744 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16745 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16746 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16747 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16748 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16750 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16752 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16754 2017-05-02 Richard Biener <rguenther@suse.de>
16756 PR tree-optimization/80591
16758 2017-04-10 Richard Biener <rguenther@suse.de>
16760 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16763 2017-05-02 Richard Biener <rguenther@suse.de>
16765 PR tree-optimization/80549
16766 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16767 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16768 headers if they do not have a preheader.
16770 2017-05-02 Martin Liska <mliska@suse.cz>
16773 * common.opt: Fix typo.
16774 * doc/invoke.texi: Likewise.
16776 2017-05-01 Jan Beulich <jbeulich@suse.com>
16778 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16779 swapping, add (x,x,m,x,n) alternative.
16781 2017-05-01 Nathan Sidwell <nathan@acm.org>
16783 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16784 unnecessary unadjusted_alignment check.
16786 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16789 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16791 * gimplify.c (gimplify_cilk_detach): New function.
16792 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16793 * tree-core.h: Document EXPR_CILK_SPAWN.
16794 * tree.h (EXPR_CILK_SPAWN): Define.
16796 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16798 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16799 to use new fixit_hint representation, using the "replace" logic.
16800 (get_line_span_for_fixit_hint): Likewise.
16801 (layout::print_any_fixits): Likewise.
16802 (selftest::test_one_liner_many_fixits): Rename to...
16803 (selftest::test_one_liner_many_fixits_1): ...this, and update
16804 comment and expected output to reflect that the multiple fix-it
16805 hints are now consolidated into one insertion.
16806 (selftest::test_one_liner_many_fixits_2): New test.
16807 (selftest::test_diagnostic_show_locus_one_liner): Update for
16809 (selftest::test_fixit_consolidation): Update for fix-it API
16811 * diagnostic.c (print_parseable_fixits): Likewise.
16812 * edit-context.c (edited_line::m_line_events): Convert from
16813 auto_vec <line_event *> to auto_vec <line_event>.
16814 (class line_event): Convert from abstract base class to a concrete
16815 class, taking over the role of replace_event.
16816 (class insert_event): Delete.
16817 (class replace_event): Rename to class line_event. Convert to
16819 (edit_context::add_fixits): Reimplement.
16820 (edit_context::apply_insert): Delete.
16821 (edit_context::apply_replace): Rename to...
16822 (edit_context::apply_fixit): ...this. Convert to half-open range.
16823 (edited_file::apply_insert): Delete.
16824 (edited_file::apply_replace): Rename to...
16825 (edited_file::apply_fixit): ...this.
16826 (edited_line::~edited_line): Drop deletion of events.
16827 (edited_line::apply_insert): Delete.
16828 (edited_line::apply_replace): Rename to...
16829 (edited_line::apply_fixit): ...this. Convert to half-open range.
16830 Update for change to type of m_line_events.
16831 * edit-context.h (edit_context::apply_insert): Delete.
16832 (edit_context::apply_replace): Rename to...
16833 (edit_context::apply_fixit): ...this.
16835 2017-05-01 Martin Sebor <msebor@redhat.com>
16837 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16840 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16843 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16844 __get_cpuid_max returns 0.
16845 (__get_cpuid_count): Ditto.
16847 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16849 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16850 replacement expression is another instance of one of its arguments.
16852 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16855 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16856 check for stack push/pop autoinc.
16857 * config/i386/i386.c (ix86_agi_dependent): Return false
16858 if the only reason why modified_in_p returned true is that
16859 addr is SP based and set_insn is a push or pop.
16861 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16863 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16866 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16869 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16870 (account_size_time): Use two predicates - exec_pred and
16872 (evaluate_conditions_for_known_args): Compute both clause and
16874 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16875 (inline_summary_t::duplicate): Update.
16876 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16878 (compute_inline_parameters): Likewise.
16879 (estimate_edge_size_and_time): Update caluclation of time.
16880 (estimate_node_size_and_time): Compute both time and nonspecialized
16882 (estimate_ipcp_clone_size_and_time): Update.
16883 (inline_merge_summary): Update.
16884 (do_estimate_edge_time): Update.
16885 (do_estimate_edge_size): Update.
16886 (do_estimate_edge_hints): Update.
16887 (inline_read_section, inline_write_summary): Stream both new predicates.
16888 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16890 (compute_inlined_call_time): Cleanup.
16891 (big_speedup_p): Update.
16892 (edge_badness): Update.
16893 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16894 (size_time_entry): Replace predicate by exec_predicate and
16895 nonconst_predicate.
16896 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16897 (estimate_edge_time): Return also nonspec_time.
16898 (reset_edge_growth_cache): Update.
16900 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16902 PR rtl-optimization/80491
16903 * ifcvt.c (noce_process_if_block): When looking for x setter
16904 with missing else_bb, don't check only the insn right before
16905 cond_earliest, but look for the last insn that x is modified in
16906 within the same bb.
16908 PR rtl-optimization/80491
16909 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16911 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16913 PR tree-optimization/80487
16914 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16916 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16918 PR tree-optimization/79697
16919 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16920 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16921 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16923 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16924 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16926 2017-04-28 Martin Sebor <msebor@redhat.com>
16928 PR tree-optimization/80523
16929 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16930 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16932 (maybe_warn, format_directive, parse_directive): Use new functions.
16933 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16935 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16937 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16939 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16941 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16942 target_header_dir): Set correctly.
16943 * configure: Regenerated.
16944 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16945 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16946 instead of SYSTEM_HEADER_DIR.
16948 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16950 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16951 (estimate_local_effects): Likewise.
16952 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16953 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16954 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16955 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16956 do_estimate_edge_time, estimate_edge_time): Likewise.
16957 * ipa-inline-analysis.c (estimate_node_size_and_time,
16958 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16959 (estimate_time_after_inlining): Remove.
16961 2017-04-28 Martin Liska <mliska@suse.cz>
16963 * doc/gcov.texi: Enhance documentation of gcov.
16965 2017-04-28 Martin Liska <mliska@suse.cz>
16967 * doc/gcov.texi: Sort options in alphabetic order.
16968 * doc/gcov-dump.texi: Likewise.
16969 * doc/gcov-tool.texi: Likewise.
16970 * gcov.c (print_usage): Likewise.
16971 * gcov-dump.c (print_usage): Likewise.
16972 * gcov-tool.c (print_merge_usage_message): Likewise.
16973 (print_rewrite_usage_message): Likewise.
16974 (print_overlap_usage_message): Likewise.
16976 2017-04-28 Martin Liska <mliska@suse.cz>
16978 PR gcov-profile/53915
16979 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
16981 2017-04-28 Martin Liska <mliska@suse.cz>
16983 PR gcov-profile/79891
16984 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
16985 is marked by compiler as living on a line.
16986 (get_cycles_count): Remove usage of the union.
16987 (output_intermediate_file): Likewise.
16988 (find_source): Fix GNU coding style.
16989 (accumulate_line_counts): Remove old non-all block mode.
16990 (output_lines): Remove usage of the union.
16991 * profile.c (output_location): Include all BBs, even if
16992 belonging to a same line (and file) as a previous BB.
16994 2017-04-28 Martin Liska <mliska@suse.cz>
16996 * gcov.c (process_args): Handle new argument 'w'.
16997 (read_graph_file): Assign ID to BBs.
16998 (output_branch_count): Display BB # if verbose flag is set.
16999 (output_lines): Likewise for arcs.
17000 (print_usage): Add '--verbose' option help.
17001 * doc/gcov.texi: Document --verbose (-w) option.
17003 2017-04-28 Martin Liska <mliska@suse.cz>
17005 * gcov.c (struct block_location_info): New struct.
17006 (process_file): Fill up the new structure.
17007 (read_graph_file): Replace usage of encoding by the newly added
17009 (add_line_counts): Likewise.
17010 (accumulate_line_counts): Remove usage of the union.
17011 (function_info::function_info): New function.
17012 (function_info::~function_info): Likewise.
17013 (process_file): Call delete instead of release_function.
17014 (release_function): Release the function.
17015 (release_structures): Call delete instead of release_function.
17016 (solve_flow_graph): Replace usage of num_blocks.
17017 (find_exception_blocks): Likewise.
17018 (output_lines): Fix GNU coding style.
17020 2017-04-28 Martin Liska <mliska@suse.cz>
17023 * coverage.c (coverage_remove_note_file): New function.
17024 * coverage.h: Declare the function.
17025 * toplev.c (finalize): Clean if an error has been seen.
17027 2017-04-28 Martin Liska <mliska@suse.cz>
17029 PR gcov-profile/80031
17030 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17031 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17032 * gcov.c (read_graph_file): Read just number of blocks.
17033 * profile.c (branch_prob): Do not stream 0 flags per a basic
17036 2017-04-28 Martin Liska <mliska@suse.cz>
17038 * gcov-dump.c (tag_*): Add new argument to declarations.
17039 (dump_gcov_file): Likewise.
17040 (tag_blocks): Add and use new argument depth.
17041 (tag_arcs): Likewise.
17042 (tag_lines): Likewise.
17043 (tag_counters): Likewise.
17044 (tag_summary): Likewise.
17045 (dump_working_sets): Use depth to do a proper indentation.
17047 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17050 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17051 * symtab.c (symtab_node::debug_symtab): Move definition here.
17053 2017-04-28 Richard Biener <rguenther@suse.de>
17055 * lto-streamer.h (LTO_major_version): Bump to 7.
17057 2017-04-28 Richard Biener <rguenther@suse.de>
17059 * tree-vrp.c (assert_info): New struct.
17060 (add_assert_info): New helper.
17061 (register_edge_assert_for_2): Refactor to add asserts to a vector
17063 (register_edge_assert_for_1): Likewise.
17064 (register_edge_assert_for): Likewise.
17065 (finish_register_edge_assert_for): New helper actually registering
17066 asserts where live on edge.
17067 (find_conditional_asserts): Adjust.
17068 (find_switch_asserts): Likewise.
17069 (evrp_dom_walker::try_find_new_range): Generalize.
17070 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17072 2017-04-27 Marek Polacek <polacek@redhat.com>
17075 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17076 arg10 and arg11 to itype.
17078 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17080 * doc/extend.texi (Object Size Checking): Improve grammar.
17082 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17085 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17086 that the logic for permitting reciprocal estimates matches that
17089 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17092 * tree.c (type_cache_hasher::equal): Only compare
17093 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17094 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17095 non-aggregate element types.
17096 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17097 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17099 2017-04-27 Richard Biener <rguenther@suse.de>
17101 PR middle-end/80533
17102 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17103 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17104 keeping a reference to a trailing array.
17106 2017-04-27 Richard Biener <rguenther@suse.de>
17108 PR middle-end/80539
17109 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17110 being in loop-closed SSA form conservatively.
17111 (chrec_fold_multiply_poly_poly): Likewise.
17113 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17115 PR middle-end/79665
17116 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17117 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17119 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17122 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17123 (aarch64_function_arg_alignment): Return unsigned int again, but still
17124 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17125 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17126 Don't emit -Wpsabi note.
17127 (aarch64_function_arg_boundary): Likewise.
17128 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17131 2017-04-26 Nathan Sidwell <nathan@acm.org>
17133 * tree.h (crc32_unsigned_n): Declare.
17134 (crc32_unsigned, crc32_unsigned): Make inline.
17135 * tree.c (crc32_unsigned_bits): Replace with ...
17136 (crc32_unsigned_n): ... this.
17137 (crc32_unsigned, crc32_byte): Remove.
17138 (crc32_string): Remove unnecessary braces.
17140 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17142 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17143 * ipa-inline-analysis.c (MAX_TIME): Remove.
17144 (account_size_time): Use sreal for time.
17145 (dump_inline_summary): Update.
17146 (estimate_function_body_sizes): Update.
17147 (estimate_edge_size_and_time): Update.
17148 (estimate_calls_size_and_time): Update.
17149 (estimate_node_size_and_time): Update.
17150 (inline_merge_summary): Update.
17151 (inline_update_overall_summary): Update.
17152 (estimate_time_after_inlining): Update.
17153 (inline_read_section): Update.
17154 (inline_write_summary): Update.
17155 * ipa-inline.c (compute_uninlined_call_time): Update.
17156 (compute_inlined_call_time): Update.
17157 (recursive_inlining): Update.
17158 (inline_small_functions): Update.
17159 (dump_overall_stats): Update.
17160 * ipa-inline.h: Include sreal.h.
17161 (size_time_entry): Turn time to sreal.
17162 (inline_summary): Turn self_time nad time to sreal.
17164 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17166 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17168 (sreal::stream_out, sreal::stream_in): New.
17169 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17171 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17173 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17176 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17179 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17181 (dimode_scalar_chain::compute_convert_gain): Ditto.
17182 (dimode_scalar_chain::make_vector_copies): Ditto.
17183 (dimode_scalar_chain::convert_reg): Ditto.
17184 (dimode_scalar_chain::convert_insn): Ditto.
17185 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17186 (VI248_AVX512BW_1): New mode iterator.
17187 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17188 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17191 2017-04-25 Martin Sebor <msebor@redhat.com>
17193 PR tree-optimization/80497
17194 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17195 constants are representable in HOST_WIDE_INT.
17196 (parse_directive): Ditto.
17198 2017-04-25 Martin Sebor <msebor@redhat.com>
17201 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17202 (new_zero_array): Adjust signature.
17203 (dom_info::dom_init): Used unsigned rather that size_t.
17204 (dom_info::dom_info): Same.
17206 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17207 Jakub Jelinek <jakub@redhat.com>
17210 * config/arm/arm.c: Include gimple.h.
17211 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17212 returns negative, increment ncrn only if it returned positive.
17213 (arm_needs_doubleword_align): Return int instead of bool,
17214 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17215 members, but if there is any such non-FIELD_DECL
17216 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17217 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17218 returns negative, increment nregs only if it returned positive.
17219 (arm_setup_incoming_varargs): Likewise.
17220 (arm_function_arg_boundary): Emit -Wpsabi note if
17221 arm_needs_doubleword_align returns negative, return
17222 DOUBLEWORD_ALIGNMENT only if it returned positive.
17224 2017-04-25 Marek Polacek <polacek@redhat.com>
17227 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17228 first argument to type.
17230 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17233 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17234 type checks to test for compatibility instead of equality.
17236 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17237 Jakub Jelinek <jakub@redhat.com>
17240 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17242 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17243 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17244 the alignment computation, but return their maximum in warn_alignment.
17245 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17246 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17248 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17249 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17252 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17254 * config/arc/simdext.md (dmpyh): Fix typo.
17256 2017-04-25 Richard Biener <rguenther@suse.de>
17258 PR tree-optimization/80492
17259 * alias.c (compare_base_decls): Handle registers with asm
17260 specification conservatively.
17261 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17262 compare_base_decls returning dont-know properly.
17264 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17266 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17267 (legitimate_offset_address_p): New function.
17268 (arc_legitimate_address_p): Use above function.
17270 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17272 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17274 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17276 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17277 ACCH registers whenever they are available.
17279 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17281 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17282 double regs fix when not used.
17284 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17286 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17288 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17289 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17291 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17293 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17294 when emitting short ADD instructions.
17296 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17298 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17300 (cmpsi_cc_c_insn): Likewise.
17301 (cbranchsi4_scratch): Compute proper instruction length using
17302 compact_hreg_operand.
17303 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17305 2017-04-25 Richard Biener <rguenther@suse.de>
17307 PR middle-end/80509
17308 * passes.c (pass_manager::pass_manager): Initialize
17309 m_name_to_pass_map.
17311 2017-04-25 Richard Biener <rguenther@suse.de>
17313 PR tree-optimization/79201
17314 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17316 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17319 * config/s390/vector.md: Split MEM->GPR vector moves for
17320 non-s_operand addresses.
17322 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17325 * config/s390/predicates.md (reload_const_wide_int_operand): New
17327 * config/s390/s390.md ("movti"): Remove d/P alternative.
17328 ("movti_bigconst"): New pattern definition.
17330 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17333 * s390-protos.h (s390_expand_cs_hqi): Removed.
17334 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17335 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17336 modes as well as CCZ1mode and CCZmode.
17337 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17338 signature of s390_emit_compare_and_swap.
17339 (s390_expand_cs_hqi): Likewise, make static.
17340 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17341 compare-and-swap, in some cases.
17342 (s390_expand_cs): Wrapper function.
17343 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17345 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17346 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17347 patterns for small and large integers. Forbid symref memory operands.
17348 Move expander to s390.c. Require cc register.
17349 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17350 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17351 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17352 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17353 symref memory operands. Remove CC mode and call s390_match_ccmode
17355 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17357 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17359 * config/s390/s390.md (define_peephole2): New peephole to help
17360 combining the load-and-test pattern with volatile memory.
17362 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17364 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17365 with CCZmode for TARGET_Z196.
17367 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17369 PR rtl-optimization/80501
17370 * combine.c (make_compound_operation_int): Set subreg_code to SET
17371 even for AND with mask of the sign bit of mode.
17373 PR rtl-optimization/80500
17374 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17375 sum's initial value.
17377 2017-04-25 Julian Brown <julian@codesourcery.com>
17378 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17380 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17382 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17384 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17386 2017-04-25 Julian Brown <julian@codesourcery.com>
17387 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17389 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17390 (thunderx2t99_sha): New Reservation.
17392 2017-04-25 Julian Brown <julian@codesourcery.com>
17393 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17395 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17396 type for 1-element load.
17398 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17400 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17402 2017-04-24 Martin Jambor <mjambor@suse.cz>
17404 PR tree-optimization/80293
17405 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17406 char arrays not totally scalarizable if it is false.
17407 (analyze_all_variable_accesses): Pass correct value in the new
17408 parameter. Add a statistics counter.
17410 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17412 PR middle-end/79931
17413 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17415 2017-04-24 Richard Biener <rguenther@suse.de>
17417 PR tree-optimization/80494
17418 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17419 out for complex types.
17421 2017-04-24 Richard Biener <rguenther@suse.de>
17423 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17424 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17425 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17426 (DFS): Adjust and never fail.
17427 (sccvn_dom_walker::fail): Remove.
17428 (sccvn_dom_walker::before_dom_children): Adjust.
17429 (run_scc_vn): Likewise and never fail.
17430 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17431 (pass_fre::execute): Likewise.
17433 2017-04-24 Richard Biener <rguenther@suse.de>
17435 PR tree-optimization/79725
17436 * tree-ssa-sink.c (statement_sink_location): Return whether
17437 failure reason was zero uses. Move that check later.
17438 (sink_code_in_bb): Deal with zero uses by removing the stmt
17441 2017-04-24 Richard Biener <rguenther@suse.de>
17444 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17445 pointer-based references.
17447 2017-04-24 Richard Biener <rguenther@suse.de>
17450 * pass_manager.h (pass_manager::operator new): Remove.
17451 (pass_manager::operator delete): Likewise.
17452 * passes.c (pass_manager::operator new): Remove.
17453 (pass_manager::operator delete): Likewise.
17454 (pass_manager::pass_manager): Zero individual pass members.
17456 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17459 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17460 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17461 Check "XEXP (src, 1)" operand here.
17462 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17463 Check "XEXP (src, 1)" operand here.
17464 (dimode_scalar_chain::make_vector_copies): Detect count register
17465 of a shift instruction. Zero extend count register from QImode
17466 to DImode to satisfy vector shift pattern count operand predicate.
17467 Substitute vector shift count operand with a DImode copy.
17468 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17471 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17473 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17474 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17475 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17476 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17477 (UNSPEC_NOREX_MEM): Remove definition.
17479 2017-04-21 Richard Biener <rguenther@suse.de>
17481 PR tree-optimization/79547
17482 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17483 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17484 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17485 without any constraints.
17487 2017-04-21 Richard Biener <rguenther@suse.de>
17489 PR tree-optimization/78847
17490 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17492 2017-04-21 Richard Biener <rguenther@suse.de>
17494 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17495 (build_distinct_type_copy): Likewise.
17496 (build_variant_type_copy): Likewise.
17497 * tree.c (build_qualified_type): Pass down mem-stat info.
17498 (build_distinct_type_copy): Likewise.
17499 (build_variant_type_copy): Likewise.
17501 2017-04-21 Richard Biener <rguenther@suse.de>
17503 PR tree-optimization/80237
17504 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17506 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17507 for a simplified result.
17509 2016-04-21 Richard Biener <rguenther@suse.de>
17511 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17512 sth as strict as a simple_iv but a chrec without symbols and an
17513 operand defined in the loop we are peeling (and not some subloop).
17514 (propagate_constants_for_unrolling): Propagate all constants.
17516 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17519 * config/i386/i386.c (print_reg): Remove assert for disalowed
17520 regno values, call output_operand_lossage instead.
17522 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17525 * config/i386/constraints.md (Yc): New register constraint.
17526 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17527 Use Yc constraint for alternative 2 of operand 0. Remove
17528 preferred_for_speed attribute.
17530 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17532 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17533 lastprivate clauses in SIMT case.
17535 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17537 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17539 2017-04-20 Richard Biener <rguenther@suse.de>
17541 PR tree-optimization/57796
17542 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17543 as N scalar stores.
17544 (vect_model_load_cost): Cost gathers as N scalar loads.
17546 2017-04-20 Richard Biener <rguenther@suse.de>
17548 * ggc-page.c (ggc_allocated_p): Rename to ...
17549 (safe_lookup_page_table_entry): ... this and return the lookup
17551 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17553 2017-04-20 Richard Biener <rguenther@suse.de>
17555 PR tree-optimization/80453
17556 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17557 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17558 from the conditions.
17559 (vn_phi_eq): Pass them down.
17560 (vn_phi_lookup): Record them.
17561 (vn_phi_insert): Likewise.
17563 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17565 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17566 uninitialized variable warning to avoid buffer overrun.
17568 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17571 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17572 is suppressed for '{ 0 }' in C.
17574 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17576 * BASE-VER: Set to 8.0.0.
17578 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17580 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17581 priority .init_array and .fini_array section with SECTION_NOTYPE
17584 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17586 PR middle-end/80423
17587 * tree.h (build_array_type): Add typeless_storage default argument.
17588 * tree.c (type_cache_hasher::equal): Also compare
17589 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17590 (build_array_type): Add typeless_storage argument, set
17591 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17593 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17594 (build_array_type): Likewise. Add typeless_storage argument.
17596 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17597 Jakub Jelinek <jakub@redhat.com>
17599 PR tree-optimization/80426
17600 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17601 operation on symbolic operands, also compute the overflow for the
17602 invariant part when the operation degenerates into a negation.
17604 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17607 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17608 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17611 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17613 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17616 * config/avr/avr.c (tree.h): Include it.
17617 (cgraph.h): Include it.
17618 (avr_encode_section_info): Don't warn for uninitialized progmem
17619 variable if it's just an alias.
17621 2017-04-19 Richard Biener <rguenther@suse.de>
17624 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17625 when needed by AutoPGO.
17627 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17630 * doc/lto.texi: Remove an extra 'that'.
17632 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17634 PR rtl-optimization/80429
17635 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17636 are only used in debug insns.
17638 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17639 Vladimir Makarov <vmakarov@redhat.com>
17641 * config/sparc/predicates.md (input_operand): Add comment. Return
17642 true for any memory operand when LRA is in progress.
17643 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17645 2017-04-18 Jeff Law <law@redhat.com>
17648 * mips.md ({return,simple_return}_internal): Do not overwrite
17651 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17653 PR tree-optimization/80443
17654 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17655 instead of adding 1, subtract -1 and similarly instead of subtracting
17658 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17660 PR rtl-optimization/80357
17661 * haifa-sched.c (tmp_bitmap): New variable.
17662 (model_recompute): Handle duplicate use records.
17663 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17664 (free_global_sched_pressure_data): Free it.
17666 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17669 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17670 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17671 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17672 instead of SYSTEM_HEADER_DIR.
17674 2017-04-18 Jeff Law <law@redhat.com>
17676 PR middle-end/80422
17677 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17678 predecessors after walking up the insn chain.
17680 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17683 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17684 sizetype type into debug info.
17686 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17689 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17690 unneeded test for TARGET_UPPER_REGS_SF.
17691 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17693 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17696 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17697 instead of gsi_after_labels.
17699 2017-04-18 Jeff Law <law@redhat.com>
17701 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17705 2017-04-13 Jeff Law <law@redhat.com>
17706 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17707 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17709 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17712 * config/avr/avr.c (intl.h): Include it.
17713 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17715 2017-04-18 Martin Liska <mliska@suse.cz>
17717 PR gcov-profile/78783
17718 * gcov-tool.c (gcov_output_files): Validate that destination
17719 file is either removed by the tool or by a user.
17721 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17722 Guy Benyei <guybe@mellanox.com>
17724 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17725 block, and do not negate it, the stored id is already negative.
17727 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17729 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17731 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17734 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17735 masks of options that should be turned off if the VSX vector
17736 options are turned off.
17737 (OTHER_P8_VECTOR_MASKS): Likewise.
17738 (OTHER_VSX_VECTOR_MASKS): Likewise.
17739 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17740 rs6000_disable_incompatible_switches to validate no type switches
17742 (rs6000_incompatible_switch): New function to disallow turning on
17743 other vector options if -mno-vsx, -mno-power8-vector, or
17744 -mno-power9-vector are specified.
17746 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17748 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17750 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17752 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17753 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17754 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17755 (ARG_POINTER_CFA_OFFSET): Likewise.
17757 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17759 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17760 conditions to take advantage of various optimizations.
17762 2017-04-13 Jeff Law <law@redhat.com>
17764 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17765 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17766 (zero_extendsidi2_dext): Likewise.
17768 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17771 * fold-const.c (fold_ternary_loc): Revert
17772 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17775 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17777 PR rtl-optimization/80343
17778 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17779 new scratch pseudo.
17781 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17784 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17785 to ubsan_encode_value.
17787 2017-04-13 Jeff Law <law@redhat.com>
17789 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17790 appearing in DEBUG_INSNs.
17792 2017-04-13 Martin Liska <mliska@suse.cz>
17794 PR gcov-profile/80413
17795 * gcov-io.c (gcov_write_string): Copy to buffer just when
17796 allocated size is greater than zero.
17798 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17801 * dwarf2out.c (decls_for_scope): Ignore declarations of
17802 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17804 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17807 * ipa-visibility.c (non_local_p): Fix typos.
17808 (localize_node): When localizing symbol in same comdat group,
17809 dissolve the group only when we know external symbols are going
17811 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17813 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17815 PR tree-optimization/79390
17816 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17817 order does not result in usable sequence, retry with reversed operand
17823 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17824 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17825 op0 instead of fold_convert_loc (loc, type, arg0).
17827 2017-04-12 Jeff Law <law@redhat.com>
17829 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17830 has a delay slot in the generated code.
17832 * config/cris/cris.md (cris_preferred_reload_class): Return
17833 GENNONACR_REGS rather than GENERAL_REGS.
17835 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17838 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17839 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17840 signedness of the result type.
17842 2017-04-12 Richard Biener <rguenther@suse.de>
17843 Jeff Law <law@redhat.com>
17845 PR tree-optimization/80359
17846 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17847 trim stores to TARGET_MEM_REFs.
17849 2017-04-12 Richard Biener <rguenther@suse.de>
17851 PR tree-optimization/79390
17852 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17853 threading case even more.
17855 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17858 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17859 for quad_address_p for TImode, instead of just not indexed_address.
17861 2017-04-12 Richard Biener <rguenther@suse.de>
17862 Bernd Edlinger <bernd.edlinger@hotmail.de>
17864 PR middle-end/79671
17865 * alias.c (component_uses_parent_alias_set_from): Handle
17866 TYPE_TYPELESS_STORAGE.
17867 (get_alias_set): Likewise.
17868 * tree-core.h (tree_type_common): Add typeless_storage flag.
17869 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17870 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17871 for types containing members with TYPE_TYPELESS_STORAGE.
17872 (place_field): Likewise.
17873 (layout_type): Likewise for ARRAY_TYPE.
17874 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17875 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17876 TYPE_TYPELESS_STORAGE.
17877 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17879 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17882 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17883 first argument to type.
17885 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17889 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17890 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17891 (rs6000_expand_binop_builtin): Likewise.
17892 (rs6000_expand_ternop_builtin): Likewise; also add missing
17893 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17894 vshasigma built-ins.
17895 * doc/extend.texi: Document that vec_xxpermdi's third argument
17896 must be a constant.
17898 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17900 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17901 Use shift_const cost parameter when calculating gain of STV shifts.
17903 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17905 PR rtl-optimization/70478
17906 * lra-constraints.c (process_alt_operands): Check memory for
17907 disfavoring memory insn operand.
17909 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17911 PR middle-end/80100
17912 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17913 left shift in unsigned HOST_WIDE_INT type.
17915 PR rtl-optimization/80385
17916 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17917 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17920 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17921 if they have any depend clauses.
17923 2017-04-11 Martin Liska <mliska@suse.cz>
17926 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17927 * ipa-split.c (split_function): Create a local comdat symbol
17928 if caller is in a comdat group.
17930 2017-04-11 Martin Liska <mliska@suse.cz>
17933 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17936 2017-04-11 Martin Sebor <msebor@redhat.com>
17938 PR middle-end/80364
17939 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17940 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17942 (directive::set_width, directive::set_precision, format_character):
17944 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17947 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17950 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17951 conflict, set target->arch_name instead of target->cpu_name.
17953 2017-04-11 Richard Biener <rguenther@suse.de>
17955 PR tree-optimization/80374
17956 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17957 build_zero_cst, remove fold_convertible_p check again.
17959 2017-04-11 Martin Liska <mliska@suse.cz>
17962 * ubsan.c (instrument_object_size): Do not instrument register
17965 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17968 * config/i386/i386-builtin-types.def
17969 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
17970 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
17971 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
17972 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
17973 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
17974 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
17975 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
17976 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
17977 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
17978 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
17979 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
17980 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
17981 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
17982 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
17983 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
17984 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
17985 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
17986 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
17987 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
17988 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
17989 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
17990 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
17991 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
17992 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
17993 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
17994 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
17995 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
17996 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
17997 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
17998 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
17999 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18000 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18001 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18002 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18003 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18004 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18005 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18006 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18007 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18008 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18009 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18010 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18011 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18012 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18013 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18014 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18016 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18017 flag to second_arg_count, handle 4 argument function type _COUNT
18018 aliases, handle second_arg_count on second argument rather than last.
18020 2017-04-10 Jeff Law <law@redhat.com>
18022 PR tree-optimization/80374
18023 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18024 record anything if we can not convert integer_zero_node to the
18027 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18030 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18031 Enhance special handling given to the TARGET_P9_MINMAX option in
18032 relation to certain other options.
18034 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18036 PR tree-optimization/80153
18037 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18038 remove POINTER_PLUS_EXPR's base part directly, rather than through
18041 2017-04-10 Richard Biener <rguenther@suse.de>
18042 Bin Cheng <bin.cheng@arm.com>
18044 PR tree-optimization/80153
18045 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18046 the first element of pointer type aff_tree. Build result expr in
18048 (add_elt_to_tree): Convert to type unconditionally. Remove other
18049 fold_convert calls.
18050 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18051 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18053 2017-04-10 Richard Biener <rguenther@suse.de>
18055 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18058 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18060 PR rtl-optimization/70478
18061 * lra-constraints.c (curr_small_class_check): New.
18062 (update_and_check_small_class_inputs): New.
18063 (process_alt_operands): Update curr_small_class_check. Disfavor
18064 alternative insn memory operands. Check available regs for small
18067 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18070 * config/mips/mips.opt (-mvirt): Update description.
18071 * doc/invoke.texi (-mvirt): Likewise.
18073 2017-04-10 Richard Biener <rguenther@suse.de>
18075 PR middle-end/80362
18076 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18077 looking for NEGATE_EXPR in -A / -B to A / B folding.
18079 2017-04-10 Martin Liska <mliska@suse.cz>
18081 PR gcov-profile/80224
18082 * gcov.c (print_usage): Fix usage string.
18083 (get_gcov_intermediate_filename): Remove.
18084 (output_gcov_file): Use both for normal and intermediate format.
18085 (generate_results): Do not initialize special file for
18086 intermediate format.
18088 2017-04-10 Richard Biener <rguenther@suse.de>
18090 PR tree-optimization/80304
18091 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18094 2017-04-10 Nathan Sidwell <nathan@acm.org>
18097 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18098 (rs6000_init_builtins): Use it.
18100 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18102 * config/arm/arm.md (<mrc>): Add mode to SET source.
18103 (<mrrc>): Likewise.
18105 2017-04-10 Richard Biener <rguenther@suse.de>
18107 PR middle-end/80344
18108 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18110 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18113 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18114 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18115 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18116 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18117 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18118 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18119 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18120 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18121 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18122 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18123 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18124 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18125 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18126 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18127 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18128 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18129 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18130 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18131 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18132 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18133 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18134 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18135 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18137 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18139 PR rtl-optimization/70478
18140 * lra-constraints.c: Reverse the last patch.
18142 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18144 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18145 Add comment for WCHAR_T.
18147 2017-04-08 Martin Liska <mliska@suse.cz>
18150 2017-04-07 Martin Liska <mliska@suse.cz>
18153 * ipa-split.c (split_function): Add function part to a same comdat
18156 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18159 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18161 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18163 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18164 (vec_pairing): ...to this.
18165 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18166 (rs6000_sched_init): Adjust for name change.
18167 (struct rs6000_sched_context): Likewise.
18168 (rs6000_init_sched_context): Likewise.
18169 (rs6000_set_sched_context): Likewise.
18171 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18177 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18179 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18180 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18181 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18183 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18185 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18187 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18189 PR rtl-optimization/70703
18190 * ira-color.c (update_conflict_hard_regno_costs): Use
18191 int64_t instead of HOST_WIDE_INT.
18193 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18195 PR rtl-optimization/70478
18196 * lra-constraints.c (process_alt_operands): Disfavor alternative
18197 insn memory operands.
18199 2017-04-07 Jeff Law <law@redhat.com>
18201 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18202 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18204 2017-04-07 Martin Liska <mliska@suse.cz>
18207 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18208 Show error message instead of an ICE.
18210 2017-04-07 Martin Liska <mliska@suse.cz>
18213 * ipa-split.c (split_function): Add function part to a same comdat
18216 2017-04-07 Richard Biener <rguenther@suse.de>
18218 PR middle-end/80341
18219 * tree.c (get_unwidened): Also handle ! for_type case for
18221 * convert.c (do_narrow): Split out from ...
18222 (convert_to_integer_1): ... here. Do not pass final truncation
18223 type to get_unwidened for TRUNC_DIV_EXPR.
18225 2017-04-07 Richard Biener <rguenther@suse.de>
18227 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18229 (aff_combination_const): Adjust.
18230 (aff_combination_scale): Likewise.
18231 (aff_combination_add_elt): Likewise.
18232 (aff_combination_add_cst): Likewise.
18233 (aff_combination_convert): Likewise.
18234 (add_elt_to_tree): Likewise. Remove unused argument.
18235 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18237 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18239 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18241 * config/arm/arm.c (arm_default_short_enums): Use
18242 ARM_DEFAULT_SHORT_ENUMS.
18243 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18245 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18248 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18249 members with redundant out-of-class redeclaration.
18251 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18254 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18255 * config/i386/i386.md (*zero_extendsidi2):
18256 Add (?*x,*x) and (?*v,*v) alternatives.
18258 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18261 * config/i386/i386.c (ix86_expand_builtin)
18262 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18263 mode from insn data. Convert operands to insn operand mode.
18264 Copy operands that don't satisfy insn predicate to a register.
18266 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18268 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18271 2017-04-06 Richard Biener <rguenther@suse.de>
18273 PR tree-optimization/80334
18274 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18275 preserve alignment of accesses.
18277 2017-04-06 Richard Biener <rguenther@suse.de>
18279 PR tree-optimization/80262
18280 * tree-sra.c (build_ref_for_offset): Preserve address-space
18282 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18283 Drop useless address-space information on MEM_REF offsets.
18285 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18287 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18289 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18291 PR rtl-optimization/70703
18292 * ira-color.c (update_conflict_hard_regno_costs): Use
18293 HOST_WIDE_INT instead of long.
18295 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18298 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18299 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18301 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18302 for x86_64 target. Handle -m3dnowa option.
18304 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18306 PR rtl-optimization/70703
18307 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18308 (update_conflict_hard_regno_costs): Use long instead of unsigned
18309 arithmetic for cost calculation.
18311 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18312 Bernd Edlinger <bernd.edlinger@hotmail.de>
18315 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18318 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18321 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18322 ptr_mode with Pmode throughout.
18323 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18324 into probe_stack_range and use DImode.
18326 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18329 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18330 call_eh_return is true.
18332 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18334 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18335 Initialize last_match_fntype_index.
18337 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18340 * tree-nvr.c: Include internal-fn.h.
18341 (pass_return_slot::execute): Ignore internal calls without
18344 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18345 Richard Biener <rguenther@suse.de>
18348 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18349 captures used multiple times, except for the last use.
18350 * generic-match-head.c: Include gimplify.h.
18352 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18354 PR tree-optimization/79390
18355 * target.h (struct noce_if_info): Declare.
18356 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18357 * target.def (noce_conversion_profitable_p): New target hook.
18358 * ifcvt.h (struct noce_if_info): New type, moved from ...
18359 * ifcvt.c (struct noce_if_info): ... here.
18360 (noce_conversion_profitable_p): Renamed to ...
18361 (default_noce_conversion_profitable_p): ... this. No longer
18363 (noce_try_store_flag_constants, noce_try_addcc,
18364 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18365 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18366 instead of noce_conversion_profitable_p.
18367 * config/i386/i386.c: Include ifcvt.h.
18368 (ix86_option_override_internal): Don't override
18369 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18370 (ix86_noce_conversion_profitable_p): New function.
18371 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18372 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18373 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18374 * doc/tm.texi: Regenerated.
18376 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18378 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18381 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18384 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18385 instructions for small multiply cores.
18387 2017-04-04 Jeff Law <law@redhat.com>
18389 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18391 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18392 that are not set by the loop over the elements.
18394 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18397 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18398 int mode, convert_modes it to mode as unsigned, otherwise use
18399 lowpart_subreg to mode rather than SImode.
18400 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18401 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18402 Use DImode instead of SImode for the shift count operand.
18403 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18406 2017-04-04 Richard Biener <rguenther@suse.de>
18408 PR middle-end/80281
18409 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18410 arithmetic done for the negate or the plus. Simplify.
18411 (A - (-B) -> A + B): Likewise.
18412 * fold-const.c (split_tree): Make sure to not negate pointers.
18414 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18416 PR rtl-optimization/60818
18417 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18418 a compare of comparisons with the thing compared if this results
18419 in a different machine mode.
18421 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18423 * alias.c (base_alias_check): Fix typo in comment.
18424 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18425 * cgraphunit.c (symbol_table::compile): Likewise.
18426 * collect2.c (maybe_run_lto_and_relink): Likewise.
18427 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18428 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18429 * config/avr/avr.c (avr_map_op_t): Likewise.
18430 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18431 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18432 * config/epiphany/epiphany.md (movcc): Likewise.
18433 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18434 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18436 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18437 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18438 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18439 * config/sh/sh.c (sh_rtx_costs): Likewise.
18440 * fold-const.c (fold_truth_andor): Likewise.
18441 * genautomata.c (collapse_flag): Likewise.
18442 * gengtype.h (struct type::u::s): Likewise.
18443 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18444 * input.c (FORMAT_AMOUNT): Likewise.
18445 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18446 (known_aggs_to_agg_replacement_list): Likewise.
18447 * ipa-inline-analysis.c: Likewise.
18448 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18449 * ipa-polymorphic-call.c
18450 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18451 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18452 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18454 * modulo-sched.c (apply_reg_moves): Likewise.
18455 * omp-expand.c (build_omp_regions_1): Likewise.
18456 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18457 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18458 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18459 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18460 * value-prof.c: Likewise.
18461 * var-tracking.c (val_reset): Likewise.
18463 2017-04-03 Richard Biener <rguenther@suse.de>
18465 PR tree-optimization/80275
18466 * fold-const.c (split_address_to_core_and_offset): Handle
18469 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18471 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18472 descriptors is at least equal to that of functions.
18474 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18476 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18478 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18481 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18482 (mov<IMOD4:mode>): New expander.
18483 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18485 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18487 PR rtl-optimization/79405
18488 * fwprop.c (propagations_left): New variable.
18489 (forward_propagate_into): Decrement it.
18490 (fwprop_init): Initialize it.
18491 (fw_prop): If the variable has reached zero, stop propagating.
18492 (fwprop_addr): Ditto.
18494 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18497 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18498 a FUNCTION_DECL, pass it as decl instead of origin to
18501 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18503 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18506 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18509 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18510 TARGET_VSX_SMALL_INTEGER.
18512 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18514 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18515 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18517 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18519 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18520 extraction from odd-numbered MSA register.
18522 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18524 PR middle-end/80173
18525 * expmed.c (store_bit_field_1): Don't attempt to create
18526 a word subreg out of hard registers wider than word if they
18527 have HARD_REGNO_NREGS of 1 for their mode.
18529 PR middle-end/80163
18530 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18531 conversions to integer types wider than word and pointer.
18534 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18535 (rtx_equal_for_cselib_p): Pass 0 to it.
18536 * cselib.c (cselib_hasher::equal): Likewise.
18537 (rtx_equal_for_cselib_1): Add depth argument. If depth
18538 is 128, don't look up VALUE locs and punt. Increment
18539 depth in recursive calls when walking VALUE locs.
18541 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18543 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18544 (make_gcov_file_name): Use the canonical path name for generating
18546 (read_line): Fix handling of files with ascii null bytes.
18548 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18550 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18551 to initialise a vector register instead
18552 of using a const_int.
18554 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18556 PR translation/80189
18557 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18558 diagnostic messages.
18560 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18563 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18564 (dfp_diex_<mode>): Update mode of operand 1.
18565 * doc/extend.texi (dxex, dxexq): Document change to return type.
18566 (diex, diexq): Document change to argument type.
18568 2017-03-30 Martin Jambor <mjambor@suse.cz>
18571 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18572 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18573 it reflects the signature changes performed at the callee side.
18574 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18575 to cgraph_build_function_type_skip_args.
18576 (build_function_decl_skip_args): Adjust call to the above function.
18578 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18581 * config/i386/sse.md
18582 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18583 register as dest whenever it is a MEM not rtx_equal_p to the
18584 corresponding dup operand, and when forcing into reg move the
18585 reg into the memory afterwards.
18586 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18587 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18588 for the force_reg mode.
18589 (avx512vl_vextractf128<mode>): Use register as dest either
18590 always when a MEM, or when it is a MEM not rtx_equal_p to the
18591 corresponding dup operand, or even not when it is a CONST_VECTOR
18592 depending on the mode and lo vs. hi.
18593 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18595 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18596 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18597 Likewise. Require that operands[2] is even.
18598 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18599 Remove extraneous parens. Require that operands[2] is a multiple
18601 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18602 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18603 disallow memory then.
18605 2017-03-30 Richard Biener <rguenther@suse.de>
18607 PR tree-optimization/77498
18608 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18609 to non-constants over backedges.
18611 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18613 PR rtl-optimization/80233
18614 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18615 as last_combined_insn. Do not test for BARRIER_P separately.
18617 2017-03-29 Andreas Schwab <schwab@suse.de>
18620 * calls.c (prepare_call_address): Convert funexp to Pmode before
18621 copying to temp reg.
18623 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18625 PR tree-optimization/80158
18626 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18627 Handle possible future case of more than one alternate
18629 (replace_rhs_if_not_dup): Likewise.
18630 (replace_one_candidate): Likewise.
18632 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18634 PR rtl-optimization/80193
18635 * ira.c (ira): Do not check allocation for LRA.
18637 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18639 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18640 (nvptx_output_simt_exit): Declare.
18641 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18642 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18643 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18644 (nvptx_declare_function_name): ...here. Emit declaration of local
18645 memory space buffer for omp_simt_enter insn.
18646 (nvptx_output_unisimt_switch): New.
18647 (nvptx_output_softstack_switch): New.
18648 (nvptx_output_simt_enter): New.
18649 (nvptx_output_simt_exit): New.
18650 * config/nvptx/nvptx.h (struct machine_function): New fields
18651 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18652 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18653 (UNSPECV_SIMT_EXIT): Ditto.
18654 (omp_simt_enter_insn): New insn.
18655 (omp_simt_enter): New expansion.
18656 (omp_simt_exit): New insn.
18657 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18659 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18660 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18661 (expand_GOMP_SIMT_EXIT): New.
18662 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18663 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18664 (GOMP_SIMT_EXIT): Ditto.
18665 * target-insns.def (omp_simt_enter): New insn.
18666 (omp_simt_exit): Ditto.
18667 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18669 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18670 (lower_rec_input_clauses): Likewise.
18671 (lower_lastprivate_clauses): Handle SIMT privatization.
18673 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18674 (ompdevlow_adjust_simt_enter): New.
18675 (find_simtpriv_var_op): New.
18676 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18677 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18679 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18680 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18681 (copy_decl_for_dup_finish): Ditto.
18683 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18685 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18688 * config/i386/i386.c (ix86_option_override_internal): Always
18689 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18691 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18693 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18695 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18697 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18698 mark new edge's irreducible flag accordign to it.
18699 (vect_do_peeling): Check loop preheader edge's irreducible flag
18700 and pass it to function slpeel_add_loop_guard.
18702 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18704 PR tree-optimization/80218
18705 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18706 Update block frequencies and counts.
18708 2017-03-28 Richard Biener <rguenther@suse.de>
18710 PR tree-optimization/78644
18711 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18712 of a simplification result we may not use it at all.
18714 2017-03-28 Richard Biener <rguenther@suse.de>
18717 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18718 without arguments, generate default definition of a SSA name.
18720 2017-03-28 Richard Biener <rguenther@suse.de>
18722 PR middle-end/80222
18723 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18724 TYPE_REF_CAN_ALIAS_ALL references.
18725 * fold-const.c (fold_indirect_ref_1): Likewise.
18727 2017-03-28 Martin Liska <mliska@suse.cz>
18730 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18731 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18733 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18734 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18736 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18737 (EXTRA_SPECS): Define.
18738 (SUBTARGET_EXTRA_SPECS): Likewise.
18739 (SUBTARGET_CPP_SPEC): Likewise.
18740 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18741 SUBTARGET_EXTRA_SPECS.
18742 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18744 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18746 * config/arc/simdext.md (vst64_insn): Update pattern.
18747 (vld32wh_insn): Likewise.
18748 (vld32wl_insn): Likewise.
18749 (vld64_insn): Likewise.
18750 (vld32_insn): Likewise.
18752 2017-03-28 Marek Polacek <polacek@redhat.com>
18755 * fold-const.c (fold_comparison): Use protected_set_expr_location
18756 instead of SET_EXPR_LOCATION.
18758 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18760 * tree.c (add_expr): Avoid name lookup warning.
18762 2017-03-27 Jeff Law <law@redhat.com>
18764 PR tree-optimization/80216
18765 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18766 function name. Limit recursion depth.
18767 (record_temporary_equivalences): Corresponding changes.
18769 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18771 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18774 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18777 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18779 * cfgcleanup.c (reg_note_cfa_p): New array.
18780 (insns_have_identical_cfa_notes): New function.
18781 (old_insns_match_p): Don't cross-jump in between /f
18782 and non-/f instructions. If both i1 and i2 are frame related,
18783 verify all CFA notes, their order and content.
18785 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18788 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18789 HImode and SImode with zero extend to DImode to one insn.
18790 (bswap<mode>2_extenddi): Likewise.
18791 (bswapsi2_extenddi): Likewise.
18792 (bswaphi2_extendsi): Likewise.
18793 (bswaphi2): Combine bswap HImode and SImode into one insn.
18794 Separate memory insns from swapping register.
18795 (bswapsi2): Likewise.
18796 (bswap<mode>2): Likewise.
18797 (bswaphi2_internal): Delete, no longer used.
18798 (bswapsi2_internal): Likewise.
18799 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18800 store, and gpr<-gpr swap insns.
18801 (bswap<mode>2_store): Likewise.
18802 (bswaphi2_reg): Register only splitter, combine with the splitter.
18803 (bswaphi2 splitter): Likewise.
18804 (bswapsi2_reg): Likewise.
18805 (bswapsi2 splitter): Likewise.
18806 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18807 the insns into load, store, and register/register insns.
18808 (bswapdi2_ldbrx): Likewise.
18809 (bswapdi2_load): Likewise.
18810 (bswapdi2_store): Likewise.
18811 (bswapdi2_reg): Likewise.
18813 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18815 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18816 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18818 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18821 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18823 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18824 special handling for target option conflicts between dform
18825 options (-mpower9-dform, -mpower9-dform-vector,
18826 -mpower9-dform-scalar) and -mno-direct-move.
18828 2017-03-27 Richard Biener <rguenther@suse.de>
18830 PR tree-optimization/80181
18831 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18833 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18835 * config/arc/predicates.md (move_double_src_operand): Replace the
18836 call to move_double_src_operand with a call to address_operand.
18838 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18840 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18841 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18842 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18844 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18846 * config/arc/predicates.md (long_immediate_loadstore_operand):
18847 Consider scaled addresses cases.
18849 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18851 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18852 restored when in interrupt.
18853 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18854 doesn't have delay slot.
18856 2017-03-27 Richard Biener <rguenther@suse.de>
18859 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18860 inlined thunk clones.
18862 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18865 * asan.c (instrument_derefs): Copy over last operand from
18866 original COMPONENT_REF to the new COMPONENT_REF with
18867 DECL_BIT_FIELD_REPRESENTATIVE.
18868 * ubsan.c (instrument_object_size): Likewise.
18870 2017-03-27 Richard Biener <rguenther@suse.de>
18872 PR tree-optimization/80170
18873 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18874 sure DR/SCEV didnt fold in constants we do not see when looking
18875 at the reference base alignment.
18877 2017-03-27 Richard Biener <rguenther@suse.de>
18879 PR middle-end/80171
18880 * gimple-fold.c (fold_ctor_reference): Properly guard against
18881 NULL return value from canonicalize_constructor_val.
18883 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18886 * config/i386/i386.c (ix86_expand_builtin)
18887 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18888 flags reg setting and flags reg using instructions.
18889 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18890 clobbering instructions to zero extend op2.
18892 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18894 * doc/install.texi (Configuration) <--with-aix-soname>:
18895 Update link to AIX ld.
18897 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18899 PR rtl-optimization/80160
18900 PR rtl-optimization/80159
18901 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18902 reg_alternate_class into account.
18904 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18907 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18908 to consider in curr_insn_transform.
18910 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18912 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18913 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18914 and emit_mode_inner.
18916 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18918 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18919 argument to the overloaded builtin variants. Use the new flag to
18920 deprecate certain builtin variants.
18921 * config/s390/s390-builtin-types.def: Add new builtin types.
18922 * config/s390/s390-builtins.h: Support new flags field for
18923 overloaded builtins.
18924 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18925 (s390_macro_to_expand): Enable vector float data type.
18926 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18927 builtins by incrementing the __VEC__ version number.
18928 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18930 (s390_resolve_overloaded_builtin): Emit error messages depending
18931 on the builtin flags.
18932 * config/s390/s390.c (s390_expand_builtin): Support additional
18933 flags argument. Change error message to match the messages
18934 emitted in s390-c.c.
18935 * config/s390/s390.md: New UNSPEC_* constants.
18936 (op_type): Add new instruction types.
18937 * config/s390/vecintrin.h: Add new builtins and test data class
18939 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18940 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18941 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18942 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18943 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18944 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18946 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18947 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18948 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18949 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18951 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18952 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18953 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18954 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18956 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18957 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18958 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18959 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18960 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18961 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18962 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18964 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18965 ("vec_scatter_element<V_HW_4:mode>_DI")
18966 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
18967 ("vec_fpint<mode>", "vflls")
18968 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
18969 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
18970 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
18971 ("*vec_cmphe<mode>_cc"): ... these.
18973 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
18974 mode constant instead of magic value.
18976 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18978 * config/s390/s390.c (s390_expand_vec_compare): Support other
18979 vector floating point modes than just V2DF.
18980 (s390_expand_vcond): Likewise.
18981 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
18982 (s390_cannot_change_mode_class): Prevent mode changes between TF
18983 and V1TF in vector registers.
18984 * config/s390/s390.md (DF, SF): New mode attributes.
18985 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
18986 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
18987 SFmode support for VRs.
18988 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
18990 (VFT, VF_HW): New mode iterators.
18991 (vw, sdx): New mode attributes.
18992 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
18993 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
18994 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
18995 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
18996 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
18997 also the new vector floating point modes. Renaming to ...
18999 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19000 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19001 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19002 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19003 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19004 ("vec_unordered<mode>"): ... these.
19006 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19007 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19008 ("*vec_extendv2df"): New insn definitions.
19010 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19012 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19013 ("mulditi3_2", "*muldi3_sign"): New patterns.
19014 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19015 rename the pattern definition.
19017 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19019 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19021 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19023 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19025 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19026 instruction if possible.
19027 * config/s390/vector.md (vec_halfnumelts): New mode
19029 ("*vec_vllezlf<mode>"): New pattern.
19031 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19033 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19034 ("popcountv4si2", "popcountv2di2"): Rename to ...
19035 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19036 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19038 ("popcount<mode>2_vxe"): New pattern.
19040 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19042 * common/config/s390/s390-common.c (processor_flags_table): Add
19044 * config.gcc: Add arch12.
19045 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19046 Default to arch12 for unknown CPU model numbers.
19047 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19048 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19049 PROCESSOR_max sanity check.
19050 * config/s390/s390-opts.h (enum processor_type): Add
19052 * config/s390/s390.c (processor_table): Add arch12.
19053 (s390_expand_builtin): Add check for B_VXE flag.
19054 (s390_issue_rate): Add PROCESSOR_ARCH12.
19055 (s390_get_sched_attrmask): Likewise.
19056 (s390_get_unit_mask): Likewise.
19057 (s390_sched_score): Enable z13 scheduling for arch12.
19058 (s390_sched_reorder): Likewise.
19059 (s390_sched_variable_issue): Likewise.
19060 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19062 (s390_tune_attr): Use z13 scheduling also for arch12.
19063 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19064 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19065 (TARGET_VXE_P): New macros.
19066 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19067 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19068 * config/s390/s390.opt: Add arch12 as processor_type.
19070 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19072 * config/s390/s390.md
19073 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19074 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19075 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19077 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19078 Rename expanders to ...
19080 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19081 ("fixuns_truncdddi2_emu"): ... these.
19083 ("fixuns_trunc<mode>si2_emu"): New expander.
19085 ("*fixuns_truncdfdi2_z13"): Rename to ...
19086 ("*fixuns_truncdfdi2_vx"): ... this.
19088 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19090 * config/s390/2964.md: Remove the single element vector compare
19091 instructions which are no longer used.
19092 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19094 (s390_canonicalize_comparison): Remove handling of DFmode
19096 (s390_expand_vec_compare_scalar): Remove function.
19097 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19098 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19100 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19102 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19104 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19106 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19107 will anyway by matched by mov<mode>_64dfp.
19109 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19111 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19112 vlef/vstef. Add missing operand to vleif.
19114 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19116 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19117 pair for all vector types with 64 bit elements.
19118 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19119 * config/s390/vector.md (V_HW_64): ... here.
19120 (V_128_NOSINGLE): New mode iterator.
19121 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19122 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19123 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19124 ("*vec_load_pairv2di"): Change to ...
19125 ("*vec_load_pair<mode>"): ... this one.
19127 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19129 * config/s390/constraints.md: Add comments.
19130 (jKK): Reject element sizes > 8 bytes.
19131 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19133 * config/s390/s390.md: Add the s_operand checks formerly in
19134 s390_split_ok_p to various splitters where they are still
19136 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19137 for 128 bit vectors. Plus two splitters.
19139 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19141 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19144 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19147 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19148 error if the boundary argument is not constant.
19150 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19152 PR rtl-optimization/80112
19153 * loop-doloop.c (doloop_condition_get): Don't check condition
19154 if cmp isn't SET with IF_THEN_ELSE src.
19156 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19158 PR tree-optimization/80158
19159 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19160 replacing a candidate statement, also replace it for the
19161 candidate's alternate interpretation.
19162 (replace_rhs_if_not_dup): Likewise.
19163 (replace_one_candidate): Likewise.
19165 2017-03-24 Richard Biener <rguenther@suse.de>
19167 PR tree-optimization/80167
19168 * graphite-isl-ast-to-gimple.c
19169 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19171 (translate_isl_ast_to_gimple::get_rename): Likewise.
19173 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19175 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19176 handling of certain combinations of target options, including the
19177 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19178 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19180 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19183 * config/arm/arm.md (*load_multiple): Add reload_completed to
19184 matching condition.
19186 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19187 Richard Biener <rguenth@suse.de>
19189 PR tree-optimization/79908
19190 PR tree-optimization/80136
19191 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19192 been cast away, gimplify_and_add suffices.
19194 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19196 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19198 2017-03-23 Richard Biener <rguenther@suse.de>
19200 PR tree-optimization/80032
19201 * gimplify.c (gimple_push_cleanup): Forced unconditional
19202 cleanups still have to go to the conditional_cleanups
19205 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19207 PR tree-optimization/80072
19208 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19210 (next_operand_entry_id): Change type to unsigned int.
19211 (sort_by_operand_rank): Make sure to return the right return value
19212 even if unsigned fields are bigger than INT_MAX.
19213 (struct oecount): Change cnt and id type to unsigned int.
19214 (oecount_hasher::equal): Formatting fix.
19215 (oecount_cmp): Make sure to return the right return value
19216 even if unsigned fields are bigger than INT_MAX.
19217 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19220 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19221 TREE_READONLY on result if writing it more than once.
19224 * doc/invoke.texi (-fsanitize=thread): Document that with
19225 -fnon-call-exceptions atomics are not able to throw
19229 * tsan.c: Include tree-eh.h.
19230 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19231 maybe_clean_or_replace_eh_stmt where needed.
19232 (instrument_memory_accesses): Add cfg_changed argument.
19233 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19234 if it returned true.
19235 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19237 PR rtl-optimization/63191
19238 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19239 wrapper function, moved the whole old content into ...
19240 (ix86_delegitimize_address_1): ... this. New inline function.
19241 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19242 true as last argument instead of ix86_delegitimize_address.
19244 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19246 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19247 cortexa57_branch_cost.
19249 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19251 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19253 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19256 * doc/md.texi (Constraints): Document wA constraint.
19257 * config/rs6000/constraints.md (wA): New.
19258 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19259 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19260 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19261 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19263 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19266 * gimplify.c (is_oacc_declared): New function.
19267 (oacc_default_clause): Use it to set default flags for acc declared
19268 variables inside parallel regions.
19269 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19270 declared variables.
19271 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19272 declare attribute to any decl as necessary.
19274 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19277 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19278 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19279 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19280 (arm_arch_lpae): This.
19281 * config/arm/arm.c (arm_arch7ve): Rename into ...
19282 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19283 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19286 2017-03-22 Martin Liska <mliska@suse.cz>
19289 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19290 error message instead of an ICE.
19292 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19294 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19296 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19298 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19300 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19303 2017-03-21 Martin Sebor <msebor@redhat.com>
19305 * doc/extend.texi: Use "cannot" instead of "can't."
19306 * doc/hostconfig.texi: Same.
19307 * doc/install.texi: Same.
19308 * doc/invoke.texi: Same.
19309 * doc/loop.texi: Same.
19310 * doc/md.texi: Same.
19311 * doc/objc.texi: Same.
19312 * doc/rtl.texi: Same.
19313 * doc/tm.texi: Same.
19314 * doc/tm.texi.in: Same.
19315 * doc/trouble.texi: Same.
19317 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19320 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19321 (collect_checksum_attributes): Set it.
19322 (die_checksum_ordered): Use it.
19324 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19326 PR tree-optimization/79908
19327 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19328 change: For a VA_ARG whose LHS has been cast away, use
19329 force_gimple_operand to construct the side effects.
19331 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19333 PR translation/80001
19334 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19335 more amenable to translation.
19336 (oacc_loop_auto_partitions): Likewise.
19338 2017-03-21 Marek Polacek <polacek@redhat.com>
19339 Martin Sebor <msebor@redhat.com>
19341 PR tree-optimization/80109
19342 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19343 on INTEGRAL_TYPE_P.
19345 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19346 Segher Boessenkool <segher@kernel.crashing.org>
19349 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19350 check reg_used_between_p between insn and one of succ or succ2
19351 depending on if succ is artificial insn not inserted into insn
19354 2017-03-21 Martin Liska <mliska@suse.cz>
19356 PR gcov-profile/80081
19357 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19358 * doc/gcc.texi: Include gcov-dump stuff.
19359 * doc/gcov-dump.texi: New file.
19361 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19363 PR rtl-optimization/79150
19364 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19365 conditional jump, if the jump is the last insn of the loop.
19367 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19368 Richard Biener <rguenth@suse.de>
19370 PR tree-optimization/79908
19371 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19372 been cast away, use force_gimple_operand to construct the side
19375 2017-03-21 Martin Liska <mliska@suse.cz>
19377 PR libfortran/79956
19378 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19381 2017-03-21 Brad Spengler <spender@grsecurity.net>
19384 * plugin.c (htab_hash_plugin): New function.
19385 (add_new_plugin): Use it and adjust.
19386 (parse_plugin_arg_opt): Adjust.
19387 (init_one_plugin): Likewise.
19389 2017-03-21 Richard Biener <rguenther@suse.de>
19391 PR tree-optimization/80032
19392 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19393 if set force the cleanup to happen unconditionally.
19394 (gimplify_target_expr): Push inserted clobbers with force_uncond
19395 to avoid them being removed by control-dependent DCE.
19397 2017-03-21 Richard Biener <rguenther@suse.de>
19399 PR tree-optimization/80122
19400 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19401 va_arg_pack_len when the inlined call stmt requires pack
19403 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19405 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19408 * tsan.c (instrument_builtin_call): If the memory model argument
19409 is not a constant, assume it is valid.
19412 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19415 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19417 PR rtl-optimization/79910
19418 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19419 if its dest is used by an insn before I2 (other than the combined
19420 insns themselves, which are properly handled already).
19422 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19425 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19427 * combine.c (record_used_regs): New static function.
19428 (try_combine): Handle situations where there is an additional
19429 instruction between I2 and I3 which needs to have a LOG_LINK
19433 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19435 * combine.c (try_combine): Delete redundant i1 test. Call
19436 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19438 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19441 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19442 alternatives 13/14.
19444 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19446 PR tree-optimization/80054
19447 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19448 the optimization if a PHI or any of its arguments is not dominated
19449 by the candidate's basis. Use gphi* rather than gimple* as
19451 (replace_profitable_candidates): Clean up a gimple* variable that
19452 should be a gphi* variable.
19454 2017-03-20 Martin Sebor <msebor@redhat.com>
19457 * doc/extend.texi (attribute constructor): Document present limitation.
19459 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19462 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19463 __POWER9_VECTOR__ #ifdef control, change template definition to
19464 use Power9-specific built-in function.
19465 (vec_any_eq): Likewise.
19466 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19467 to control outcomes from this test.
19468 (vector_ae_<mode>p): For VEC_F modes, likewise.
19470 2017-03-20 Ian Lance Taylor <iant@google.com>
19472 * config/i386/i386.c (ix86_function_regparm): Save an extra
19473 register for -fsplit-stack with DECL_STATIC_CHAIN.
19475 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19478 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19479 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19481 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19483 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19485 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19488 2017-03-20 Marek Polacek <polacek@redhat.com>
19491 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19493 2017-03-20 Richard Biener <rguenther@suse.de>
19495 PR tree-optimization/80113
19496 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19497 allocate extra SSA name for PHI def.
19498 (add_close_phis_to_outer_loops): Likewise.
19499 (add_close_phis_to_merge_points): Likewise.
19500 (copy_loop_close_phi_args): Likewise.
19501 (copy_cond_phi_nodes): Likewise.
19503 2017-03-20 Martin Liska <mliska@suse.cz>
19505 PR middle-end/79753
19506 * tree-chkp.c (chkp_build_returned_bound): Do not build
19507 returned bounds for a LHS that's not a BOUNDED_P type.
19509 2017-03-20 Martin Liska <mliska@suse.cz>
19513 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19514 COMPLEX_CST and VECTOR_CST.
19516 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19519 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19520 target operand. A new splitter adds the clobber statement in case
19521 the target operand is dead anyway.
19523 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19525 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19526 to age-old versions of binutils and glibc.
19528 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19530 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19532 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19534 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19536 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19538 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19539 requirement for binutils 2.13.
19541 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19543 * combine.c (try_combine): Delete redundant i1 test. Call
19544 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19546 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19548 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19549 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19551 <riscv64-*-elf>: Re-arrange section
19552 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19553 <riscv32-*-linux>: Likewise.
19554 <riscv64-*-elf>: Likewise
19555 <riscv64-*-linux>: Likewise.
19557 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19560 * aarch64.opt(verbose-cost-dump): Fix typo.
19562 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19565 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19566 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19568 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19570 * reload.c (find_reloads): When reloading a nonoffsettable address,
19571 use RELOAD_OTHER for it and its address reloads.
19573 PR rtl-optimization/79910
19574 * combine.c (record_used_regs): New static function.
19575 (try_combine): Handle situations where there is an additional
19576 instruction between I2 and I3 which needs to have a LOG_LINK
19579 2017-03-17 Jeff Law <law@redhat.com>
19581 PR tree-optimization/71437
19582 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19583 conditional in the hash table first.
19584 (vrp_dom_walker::before_dom_children): Extract condition from
19585 ASSERT_EXPR. Record condition, its inverion and any implied
19586 conditions as well.
19588 2017-03-17 Marek Polacek <polacek@redhat.com>
19589 Markus Trippelsdorf <markus@trippelsdorf.de>
19591 PR tree-optimization/80079
19592 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19595 2017-03-17 Richard Biener <rguenther@suse.de>
19597 PR middle-end/80075
19598 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19599 Properly verify the LHS before the RHS possibly claims to be
19601 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19604 2017-03-17 Martin Jambor <mjambor@suse.cz>
19606 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19607 (List of -O2 options): Likewise.
19608 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19611 2017-03-17 Tom de Vries <tom@codesourcery.com>
19613 * gcov-dump.c (print_usage): Print bug_report_url.
19615 2017-03-17 Richard Biener <rguenther@suse.de>
19617 PR middle-end/80050
19618 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19619 (parser::peek): Likewise.
19621 2017-03-17 Richard Biener <rguenther@suse.de>
19623 PR tree-optimization/80048
19624 * sese.c (free_sese_info): Properly release rename_map and
19625 copied_bb_map elements.
19627 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19629 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19630 Add linked-list forward and backlinks. Insert on
19631 construction, remove on destruction.
19632 (class pass_store_merging): Add m_stores_head field.
19633 (pass_store_merging::terminate_and_process_all_chains):
19634 Iterate over m_stores_head list.
19635 (pass_store_merging::terminate_all_aliasing_chains):
19637 (pass_store_merging::execute): Check for debug stmts first.
19638 Push new chains onto the m_stores_head stack.
19640 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19643 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19644 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19645 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19647 2017-03-16 Jeff Law <law@redhat.com>
19649 PR tree-optimization/71437
19650 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19651 member function. Implementation moved into after_dom_children
19652 member function and into the threader's thread_outgoing_edges
19654 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19655 some code into new thread_outgoing_edges.
19656 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19657 definition. Simplify marker handling (do it here). Assume we always
19658 have the available expression and the const/copies tables.
19659 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19661 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19662 * tree-vrp.c (equiv_stack): No longer file scoped.
19663 (vrp_dom_walker): New class.
19664 (vrp_dom_walker::before_dom_children): New member function.
19665 (vrp_dom_walker::after_dom_children): Likewise.
19666 (identify_jump_threads): Setup domwalker. Use it rather than
19667 walking edges in a random order by hand. Simplify setup/finalization.
19668 (finalize_jump_threads): Remove.
19669 (vrp_finalize): Do not call identify_jump_threads here.
19670 (execute_vrp): Do it here instead and call thread_through_all_blocks
19673 PR tree-optimization/71437
19674 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19676 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19678 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19679 (dom_opt_dom_walker::thread_across_edge): Remove
19680 handle_dominating_asserts argument. All callers changed.
19681 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19682 changes. Remove calls to lhs_of_dominating_assert. Other
19683 uses of handle_dominating_asserts turn into unconditional code
19684 (simplify_control_stmt_condition_1): Likewise.
19685 (simplify_control_stmt_condition): Likewise.
19686 (thread_through_normal_block, thread_across_edge): Likewise.
19687 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19688 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19689 object if it is not an SSA_NAME.
19690 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19691 before calling into the VRP specific simplifiers.
19692 (identify_jump_threads): Remove handle_dominating_asserts
19695 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19698 * tree-diagnostic.c (default_tree_printer): No longer static.
19699 * tree-diagnostic.h (default_tree_printer): New prototype.
19701 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19703 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19704 Change ins into fmov.
19706 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19708 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19709 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19710 Use h_con constraint for operand 1.
19711 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19712 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19714 2017-03-15 Jeff Law <law@redhat.com>
19716 PR tree-optimization/71437
19717 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19718 (record_temporary_equivalences): Use it.
19720 PR tree-optimization/71437
19721 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19722 tree-ssa-scopedtables.
19723 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19724 (record_conditions, record_cond, vuse_eq): Likewise.
19725 (record_edge_info): Adjust to API tweak of record_conditions.
19726 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19727 (record_temporary_equivalences, optimize_stmt): Likewise.
19728 (eliminate_redundant_computations): Likewise.
19729 (record_equivalences_from_stmt): Likewise.
19730 * tree-ssa-scopedtables.c: Include options.h and params.h.
19731 (vuse_eq): New function, moved from tree-ssa-dom.c
19732 (build_and_record_new_cond): Likewise.
19733 (record_conditions): Likewise. Accept vector of conditions rather
19734 than edge_equivalence structure for first argument.
19735 for the first argument.
19736 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19737 from tree-ssa-dom.c.
19738 (avail_exprs_stack::record_cond): Likewise.
19739 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19740 from tree-ssa-dom.c.
19741 (avail_exprs_stack): Add new member functions lookup_avail_expr
19743 (record_conditions): Declare.
19745 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19748 * lra-constraints.c (process_alt_operands): Increase reject for
19749 reloading an input/output operand.
19751 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19754 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19755 insns to convert from signed/unsigned char/short to IEEE 128-bit
19757 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19759 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19762 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19763 subreg of inner mode for values already in registers.
19765 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19767 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19768 iteration reg is used after the loop.
19770 2017-03-14 Martin Sebor <msebor@redhat.com>
19772 PR tree-optimization/79800
19773 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19774 precision in negative-positive range.
19775 (format_floating): Call non-const overload with adjusted precision.
19777 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19780 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19783 2017-03-14 Martin Sebor <msebor@redhat.com>
19785 PR middle-end/80020
19786 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19787 * builtins.def (aligned_alloc): Use it.
19790 * Makefile.in (GTFILES): Add calls.c.
19791 * calls.c: Include "gt-calls.h".
19793 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19795 PR rtl-optimization/79728
19796 * regs.h (struct target_regs): New field
19797 x_contains_allocatable_regs_of_mode.
19798 (contains_allocatable_regs_of_mode): New macro.
19799 * reginfo.c (init_reg_sets_1): Initialize it, and change
19800 contains_reg_of_mode so it includes global regs as well.
19801 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19802 rather than contains_regs_of_mode.
19804 2017-03-14 Martin Liska <mliska@suse.cz>
19806 * doc/invoke.texi: Document options that can't be combined with
19807 -fcheck-pointer-bounds.
19809 2017-03-14 Martin Liska <mliska@suse.cz>
19811 PR middle-end/79831
19812 * doc/invoke.texi (-Wchkp): Document the option.
19814 2017-03-14 Martin Liska <mliska@suse.cz>
19816 * Makefile.in: Install gcov-dump.
19818 2017-03-14 Martin Liska <mliska@suse.cz>
19820 * multiple_target.c (expand_target_clones): Bail out for
19821 an invalid attribute.
19823 2017-03-14 Richard Biener <rguenther@suse.de>
19825 * alias.c (struct alias_set_entry): Pack properly.
19826 * cfgloop.h (struct loop): Likewise.
19827 * cse.c (struct set): Likewise.
19828 * ipa-utils.c (struct searchc_env): Likewise.
19829 * loop-invariant.c (struct invariant): Likewise.
19830 * lra-remat.c (struct cand): Likewise.
19831 * recog.c (struct change_t): Likewise.
19832 * rtl.h (struct address_info): Likewise.
19833 * symbol-summary.h (function_summary): Likewise.
19834 * tree-loop-distribution.c (struct partition): Likewise.
19835 * tree-object-size.c (struct object_size_info): Likewise.
19836 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19837 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19838 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19839 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19840 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19841 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19842 (struct _stmt_vec_info): Likewise.
19844 2017-03-14 Martin Liska <mliska@suse.cz>
19847 * multiple_target.c (create_dispatcher_calls): Check that
19848 a target can create a function dispatcher.
19850 2017-03-14 Martin Liska <mliska@suse.cz>
19853 * multiple_target.c (expand_target_clones): Drop local.local
19854 flag for default implementation.
19856 2017-03-14 Richard Biener <rguenther@suse.de>
19858 PR tree-optimization/80030
19859 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19861 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19863 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19864 gcc_fallthrough() instead of __attribute__((fallthrough));
19866 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19868 * doc/gcc.texi: Remove "up" link to (DIR).
19869 * doc/gccint.texi: Ditto.
19871 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19873 * doc/install.texi (Specific) <avr>: Remove reference to
19876 2017-03-13 Jeff Law <law@redhat.com>
19878 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19879 attribute rather than comments.
19881 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19882 match_scratch operand is highest.
19884 2017-03-13 Martin Liska <mliska@suse.cz>
19886 PR middle-end/78339
19887 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19888 is a CHKP clone, use original declaration.
19890 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19892 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19893 (arc_conditional_register_usage): Use a different allocation order
19894 when optimizing for size.
19895 * common/config/arc/arc-common.c (arc_option_optimization_table):
19896 Section anchors default on when optimizing for size.
19898 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19900 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19902 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19904 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19905 * config/arc/arc.md (cpu_facility): Add cd variant.
19906 (*movqi_insn): Add code density variant.
19907 (*movhi_insn): Likewise.
19908 (*movqi_insn): Likewise.
19909 (*addsi3_mixed): Likewise.
19910 (subsi3_insn): Likewise.
19912 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19914 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19916 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19918 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19919 expressions with MINUS and UNARY ops.
19921 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19924 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19926 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19927 between vec_select and vector argument.
19928 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19929 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19930 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19931 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19932 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19933 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19935 2017-03-13 Richard Biener <rguenther@suse.de>
19938 * params.def (vect-max-peeling-for-alignment): Fix typo.
19940 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19942 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19943 issue that only occurred with binutils below 2.18.
19945 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19947 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19948 refer to binutils 2.11/2.12 minimum.
19950 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19952 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19953 ftp.kernel.org and simplify binutils requirement.
19955 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19957 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19959 (Optimize Options): Ditto. Also remove redundancy.
19961 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19963 PR translation/79848
19964 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19966 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
19967 to G_ to avoid double translation.
19969 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19971 PR translation/79923
19972 * auto-profile.c (get_combined_location): Convert leading
19973 character of diagnostics to lower case and remove trailing period.
19974 (read_profile): Likewise for various diagnostics.
19975 * config/arm/arm.c (arm_option_override): Remove trailing period
19976 from various diagnostics.
19977 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
19978 (msp430_expand_delay_cycles): Likewise.
19980 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19983 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
19984 full command-line argument, rather than just "str".
19985 (aarch64_validate_march): Likewise.
19986 (aarch64_validate_mtune): Likewise.
19988 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
19990 PR rtl-optimization/78911
19991 * lra-assigns.c (must_not_spill_p): New function.
19992 (spill_for): Use it.
19994 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19996 PR tree-optimization/79981
19997 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
19998 ATOMIC_COMPARE_EXCHANGE ifn result.
19999 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20000 IFN_ATOMIC_COMPARE_EXCHANGE.
20002 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20005 * opts.c (parse_sanitizer_options): Add missing question mark to
20006 "did you mean" message.
20008 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20010 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20012 (VMULEUH_UNS): Likewise.
20013 (VMULOUB_UNS): Likewise.
20014 (VMULOUH_UNS): Likewise.
20015 * config/rs6000/rs6000.c (builtin_function_type): Remove
20016 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20018 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20021 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20022 x with result of extra_parsing_for_operand_code_0.
20023 (function_reader::extra_parsing_for_operand_code_0): Convert
20024 return type from void to rtx, returning x. When reading
20025 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20026 larger size containing struct block_symbol.
20028 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20030 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20031 -mfloat128-hardware without -m64.
20033 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20036 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20037 entries to the case statement that marks unsigned arguments to
20038 overloaded functions.
20040 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20042 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20043 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20045 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20048 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20049 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20051 2017-03-10 Martin Liska <mliska@suse.cz>
20055 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20056 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20059 2017-03-10 Olivier Hainque <hainque@adacore.com>
20061 * tree-switch-conversion (array_value_type): Start by resetting
20062 candidate type to it's main variant.
20064 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20066 PR rtl-optimization/79909
20067 * combine.c (try_combine): Use simplify_replace_rtx on individual
20068 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20069 of the whole CALL_INSN_FUNCTION_USAGE.
20071 PR tree-optimization/79972
20072 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20073 get_range_info on SSA_NAMEs. Formatting fixes.
20075 2017-03-10 Richard Biener <rguenther@suse.de>
20076 Jakub Jelinek <jakub@redhat.com>
20078 PR tree-optimization/77975
20079 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20080 edge to be constant.
20081 (get_val_for): For constant x return it. Formatting fix.
20082 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20083 would use the same bases as the current one.
20085 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20087 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20088 instead of vec_select for V1TImode.
20089 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20091 (VSX_LE_128): Add V1TI to this mode iterator.
20092 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20093 (*vsx_le_perm_store_<mode>): Likewise.
20094 (pre-reload splitter for VSX stores): Likewise.
20095 (post-reload splitter for VSX stores): Likewise.
20096 (*vsx_xxpermdi2_le_<mode>): Likewise.
20097 (*vsx_lxvd2x2_le_<mode>): Likewise.
20098 (*vsx_stxvd2x2_le_<mode>): Likewise.
20100 2017-03-09 Michael Eager <eager@eagercon.com>
20102 Correct failures with --enable-checking=yes,rtl.
20104 * config/microblaze/microblaze.c (microblaze_expand_shift):
20105 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20106 test for const0_rtx.
20107 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20108 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20110 2017-03-09 Richard Biener <rguenther@suse.de>
20112 PR tree-optimization/79977
20113 * graphite-scop-detection.c (scop_detection::merge_sese):
20114 Handle the case of extra exits to blocks dominating the entry.
20116 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20118 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20121 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20123 PR rtl-optimization/79949
20124 * lra-constraints.c (process_alt_operands): Check memory when
20125 trying to predict a cycle. Print about the overall increase.
20127 2017-03-09 Richard Biener <rguenther@suse.de>
20129 PR middle-end/79971
20130 * gimple-expr.c (useless_type_conversion_p): Preserve
20131 TYPE_SATURATING for fixed-point types.
20133 2017-03-09 Richard Biener <rguenther@suse.de>
20136 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20137 alignment of BLKmode params.
20139 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20142 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20143 (VALL_NO_V2Q): Likewise.
20145 * config/aarch64/aarch64-simd.md
20146 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20148 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20149 VALL_NO_V2Q mode iterator.
20150 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20152 2017-03-09 Martin Liska <mliska@suse.cz>
20154 PR tree-optimization/79631
20155 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20156 tree_int_cst_sign_bit just for INTEGER constants.
20158 2017-03-09 Martin Liska <mliska@suse.cz>
20162 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20165 2017-03-09 Marek Polacek <polacek@redhat.com>
20168 * tree.c (inchash::add_expr): Handle TREE_VEC.
20170 2017-03-09 Martin Liska <mliska@suse.cz>
20173 (chkp_narrow_size_and_offset): New function.
20174 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20175 (void chkp_parse_bit_field_ref): New function.
20176 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20177 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20179 2017-03-09 Martin Liska <mliska@suse.cz>
20182 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20183 (chkp_find_bounds_1): Remove gcc_unreachable.
20185 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20188 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20189 BUILT_IN_SYNC*, determine the access type from the size suffix and
20190 always build a MEM_REF with that type. Handle forgotten
20191 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20194 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20195 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20196 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20197 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20198 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20199 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20200 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20201 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20202 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20203 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20204 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20205 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20206 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20207 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20208 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20209 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20210 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20211 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20212 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20213 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20214 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20215 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20216 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20217 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20218 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20219 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20220 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20221 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20222 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20223 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20224 definitions outside of __OPTIMIZE__ guarded section.
20227 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20228 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20229 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20230 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20233 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20235 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20236 ("vfenez<mode>"): Add missing constraints.
20238 2017-03-08 Martin Sebor <msebor@redhat.com>
20241 * config/nds32/nds32.c (nds32_option_override):
20242 Fix misspelled diagnostic.
20244 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20247 * gimplify.c (gimplify_omp_for): Replace index var in outer
20248 taskloop statement with an artificial variable and add
20249 OMP_CLAUSE_PRIVATE clause for it.
20251 2017-03-08 Richard Biener <rguenther@suse.de>
20253 PR tree-optimization/79955
20254 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20255 for accesses that are completely outside of the variable.
20257 2017-03-08 Andrew Haley <aph@redhat.com>
20259 PR tree-optimization/79943
20260 * tree-ssa-loop-split.c (compute_new_first_bound): When
20261 calculating the new upper bound, (END-BEG) should be added, not
20264 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20266 * config/avr/avr.md (setmemhi): Make sure match_dup
20267 operand number comes before match_scratch.
20269 2017-03-08 Richard Biener <rguenther@suse.de>
20271 PR tree-optimization/79920
20272 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20273 with ncopies == 1 to ...
20274 (vect_transform_slp_perm_load): ... here. Properly compute
20275 all element loads by iterating VF times over the group. Do
20276 not handle ncopies (computed in a broken way) in
20277 vect_create_mask_and_perm.
20279 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20282 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20283 is a uniform vector, use uniform_vector_p return value instead of
20284 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20286 2017-03-07 Marek Polacek <polacek@redhat.com>
20288 PR middle-end/79809
20289 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20290 (alloca_call_type): Likewise.
20292 2017-03-07 Martin Liska <mliska@suse.cz>
20294 * gcov.c (process_args): Put comment to correct location.
20296 2017-03-07 Martin Liska <mliska@suse.cz>
20298 PR middle-end/68270
20299 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20300 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20301 (chkp_narrow_bounds_for_field): Likewise.
20302 (chkp_parse_array_and_component_ref): Pass one more argument to
20305 2017-03-07 Richard Biener <rguenther@suse.de>
20307 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20310 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20312 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20313 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20315 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20318 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20319 to end of description.
20320 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20322 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20324 PR rtl-optimization/79901
20325 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20327 (*avx512f_<code><mode>3<mask_name>): ... this.
20328 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20329 iterator instead of VI8_AVX2_AVX512BW.
20331 PR rtl-optimization/79901
20332 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20333 min/max expander, expand it using expand_vec_cond_expr.
20336 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20339 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20342 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20343 to void * for PCH reasons.
20344 * dwarf2out.c (output_loc_operands, output_die): Cast
20345 v.val_vec.array to unsigned char *.
20347 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20350 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20353 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20355 PR rtl-optimization/79571
20356 * lra-constraints.c (process_alt_operands): Calculate static
20357 reject and subtract it from overall when only addresses will be
20360 2017-03-06 Julia Koval <julia.koval@intel.com>
20363 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20364 incoming stack boundary to 128 for 64-bit targets.
20366 2017-03-06 Richard Biener <rguenther@suse.de>
20368 PR tree-optimization/79894
20369 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20370 to NULL after folding it.
20372 2017-03-06 Richard Biener <rguenther@suse.de>
20374 PR tree-optimization/79824
20375 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20376 check disabling peeling for gaps.
20378 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20380 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20381 attributes): Document gettimeofday.
20383 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20385 * config/s390/s390.c (s390_option_override_internal): Set
20386 PARAM_MIN_VECT_LOOP_BOUND
20388 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20390 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20391 * config/s390/s390.md: Likewise.
20393 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20396 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20397 (<avx2_avx512>_perm<mode>): Rename to ...
20398 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20400 (<avx512>_perm<mode>_mask): Rename to ...
20401 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20403 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20404 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20405 instead of VI8F_256_512.
20406 (avx512f_perm<mode>): New define_expand.
20407 (avx512f_perm<mode>_mask): Likewise.
20408 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20409 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20411 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20413 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20414 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20416 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20418 2017-03-06 Martin Liska <mliska@suse.cz>
20421 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20422 when having a SSA NAME w/o VAR_DECL assigned to it.
20424 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20426 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20427 msa_dpsub_<su>_d): Fix MODE for vec_select.
20429 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20431 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20433 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20435 2017-03-06 Richard Biener <rguenther@suse.de>
20437 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20438 * plugin.c (register_plugin_info): Likewise.
20439 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20441 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20443 * config/i386/sse.md (sse_storehps, sse_storelps,
20444 avx_<castmode><avxsizesuffix>_<castmode>,
20445 avx512f_<castmode><avxsizesuffix>_<castmode>,
20446 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20447 in condition that at least one operand is not a MEM.
20449 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20451 PR middle-end/79805
20452 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20453 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20455 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20456 gimple_call_nothrow_p flag based on whether original builtin can throw.
20457 If it can, emit following stmts on the fallthrough edge.
20458 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20459 don't create new bb if inserting just debug stmts on the edge, try to
20460 insert them on the fallthru bb or just reset debug stmts.
20462 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20465 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20466 restore recog_data (including the operand rtxes inside it) around
20467 the call to get_insn_template.
20469 2017-03-03 Martin Sebor <msebor@redhat.com>
20471 PR tree-optimization/79699
20472 * context.c (context::~context): Free MPFR caches to avoid
20473 a memory leak on program exit.
20475 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20477 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20478 Use wide_int::ulow () instead of .elt (0).
20480 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20482 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20483 (*pushxf): Limit oF constraint to 32bit targets and add oC
20484 constraint for 64bit targets.
20485 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20486 (*pushdf): Change rmF constraint to rmC.
20488 2017-03-03 Martin Liska <mliska@suse.cz>
20490 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20491 Remove unused variable.
20493 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20496 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20497 is a memory operand, increase num_memory.
20498 (ix86_expand_args_builtin): Likewise.
20500 2017-03-03 Jan Hubicka <jh@suse.cz>
20503 * ipa-devirt.c (maybe_record_node): Properly handle
20504 __cxa_pure_virtual visibility.
20506 2017-03-03 Martin Liska <mliska@suse.cz>
20508 PR tree-optimization/79803
20509 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20511 (pass_loop_prefetch::execute): Disabled optimization if an
20512 assumption about L1 cache size is not met.
20514 2017-03-03 Martin Liska <mliska@suse.cz>
20516 PR rtl-optimization/79574
20517 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20518 (hash_scan_set): Likewise.
20519 (dump_hash_table): Likewise.
20520 (hoist_code): Likewise.
20522 2017-03-03 Richard Biener <rguenther@suse.de>
20524 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20525 in place of uhigh ().
20526 (fixed_convert_from_real): Likewise.
20528 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20531 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20533 2017-03-03 Richard Biener <rguenther@suse.de>
20535 PR middle-end/79818
20536 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20537 TYPE_OVERFLOW_UNDEFINED check.
20539 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20541 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20543 (vector_ae_<mode>_p): Likewise.
20544 (vector_nez_<mode>_p): Likewise.
20545 (vector_ne_v2di_p): Likewise.
20546 (vector_ae_v2di_p): Likewise.
20547 (vector_ne_<mode>_p): Likewise.
20548 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20550 (vsx_tsqrt<mode>2_fe): Likewise.
20552 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20555 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20557 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20559 PR rtl-optimization/79780
20560 * cprop.c (one_cprop_pass): When second and further conditional trap
20561 in a single basic block is turned into an unconditional trap, turn it
20562 into a deleted note to avoid RTL verification failures.
20564 2017-03-02 Richard Biener <rguenther@suse.de>
20566 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20568 2017-03-02 Richard Biener <rguenther@suse.de>
20570 PR tree-optimization/79345
20572 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20573 param and abort the walk, returning -1 if it is hit.
20574 (walk_aliased_vdefs): Take a limit param and pass it on.
20575 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20576 defaulting to 0 and return a signed int.
20577 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20578 (check_defs): New helper.
20579 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20580 about uninitialized memory.
20581 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20582 bogus uninitialized warning.
20583 (fixed_convert_from_real): Likewise.
20585 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20587 PR tree-optimization/66768
20588 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20589 iv_use if base object can't be determined.
20591 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20593 PR tree-optimization/79345
20594 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20595 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20596 (get_pattern_stats): Initialize it.
20597 * genemit.c (gen_expand): Verify match_scratch numbers come after
20598 match_operand/match_dup numbers.
20599 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20600 match_scratch numbers.
20601 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20603 * config/s390/s390.md (trunctdsd2): Likewise.
20605 2017-03-02 Richard Biener <rguenther@suse.de>
20607 * wide-int.h (wide_int_storage::operator=): Implement in terms
20610 2017-03-02 Richard Biener <rguenther@suse.de>
20612 PR tree-optimization/79777
20613 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20614 the to insert expression to sth existing.
20616 2017-03-01 Martin Sebor <msebor@redhat.com>
20618 PR middle-end/79692
20619 * gimple-ssa-sprintf.c
20620 (directive::known_width_and_precision): New function.
20621 (format_integer): Use it.
20622 (get_mpfr_format_length): Consider the full range of precision
20623 when computing %g output with the # flag. Set the likely byte
20624 count to 3 rather than 1 when precision is indeterminate.
20625 (format_floating): Correct the lower bound of precision.
20627 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20629 * doc/invoke.texi: Document default code model for 64-bit Linux.
20631 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20634 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20635 udiv rather than div since input pattern is unsigned.
20637 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20639 * config/i386/i386.c (print_reg): Warn for values of
20640 unsupported size in integer register.
20642 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20645 * config/rs6000/predicates.md (current_file_function_operand): Do
20646 not allow self calls to be local if the function is replaceable.
20648 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20651 * config/rs6000/altivec.h (vec_ctz and others): Change the
20652 preprocessor macro that controls conditional compilation from
20653 _ARCH_PWR9 to __POWER9_VECTOR__.
20654 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20655 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20656 control (instead of _ARCH_PWR9 control) so that template
20657 definition uses power9-specific function.
20658 (vec_any_eq): Likewise.
20659 (vec_all_ne): Change macro definition to use a power9-specific
20660 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20661 _ARCH_PWR9 control).
20662 (vec_any_eq) Likewise.
20663 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20664 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20665 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20666 support for xvcmpnedp instruction.
20667 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20668 macro expansion so that Power9 implementation of vec_all_ne does
20669 not use the AltiVec predicate framework.
20670 (VCMPNEH_P): Likewise.
20671 (VCMPNEW_P): Likewise.
20672 (VCMPNED_P): Likewise.
20673 (VCMPNEFP_P): Likewise.
20674 (VCMPNEDP_P): Likewise.
20675 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20676 implementation of vec_any_eq to not use AltiVec predicate
20678 (VCMPAEH_P): Likewise.
20679 (VCMPAEW_P): Likewise.
20680 (VCMPAED_P): Likewise.
20681 (VCMPAEFP_P): Likewise.
20682 (VCMPAEDP_P): Likewise.
20683 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20684 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20685 not use the AltiVec predicate framework.
20686 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20687 of vec_any_eq to not use AltiVec predicate framework.
20688 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20689 support for predefined __POWER9_VECTOR__ macro to indicate that
20690 Power9 instruction selection is enabled.
20691 (altivec_overloaded_builtins): Remove extraneous
20692 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20693 function argument types RS6000_BTI_bool_V16QI and
20694 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20695 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20696 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20697 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20698 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20699 Power9 for implementations of vec_cmpne. Change the signature for
20700 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20701 (representing vec_all_ne) to remove the previously described first
20702 argument of type RS6000_BTI_INTSI, as this was an artifact of
20703 reliance on the AltiVec predicate framework, which is no longer
20704 used in the implementation of these functions. Add
20705 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20706 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20707 since, unlike the AltiVec predicate framework implementation, we
20708 do not share function descriptors between vec_alle and vec_anyeq.
20709 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20710 set of modes that receive special treatment even when
20711 TARGET_P9_VECTOR is true. The special treatment emits code that
20712 does not depend on Power9 instructions.
20713 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20714 define_expand to not rely on AltiVec predicate framework.
20715 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20717 (vector_ne_v2di_p): Change this define_expand to not rely on
20718 AltiVec predicate framework.
20719 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20721 (vector_ne_<mode>_p): Change this define_expand to not rely on
20722 AltiVec predicate framework.
20723 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20725 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20726 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20727 define_insn pattern.
20728 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20729 define_insn pattern because the xvcmpne<VSs>. instruction is not
20731 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20732 instruction is not supported.
20734 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20736 * config/nvptx/nvptx.c: Include intl.h.
20738 2017-03-01 Martin Jambor <mjambor@suse.cz>
20741 * ipa-prop.h (ipa_bits): Removed field known.
20742 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20743 to pointers. Adjusted their comments to warn about their sharing.
20744 (ipcp_transformation_summary): Change bits to a vector of pointers.
20745 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20746 (ipa_get_ipa_bits_for_value): Declare.
20747 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20748 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20749 (ipa_bits_hash_table): Likewise.
20750 (ipa_vr_ggc_hash_traits): Likewise.
20751 (ipa_vr_hash_table): Likewise.
20752 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20753 being pointers and vr_known being removed.
20754 (ipa_set_jf_unknown): Likewise.
20755 (ipa_get_ipa_bits_for_value): New function.
20756 (ipa_set_jfunc_bits): Likewise.
20757 (ipa_get_value_range): New overloaded functions.
20758 (ipa_set_jfunc_vr): Likewise.
20759 (ipa_compute_jump_functions_for_edge): Use the above functions to
20760 construct bits and vr parts of jump functions.
20761 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20762 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20764 (ipcp_grow_transformations_if_necessary): Also allocate
20765 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20767 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20768 them. Fix too long lines.
20769 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20770 vr_known being removed.
20771 (ipa_read_jump_function): Use new setter functions to construct bits
20772 and vr parts of jump functions or set them to NULL.
20773 (write_ipcp_transformation_info): Adjust for bits being pointers.
20774 (read_ipcp_transformation_info): Likewise.
20775 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20777 Include gt-ipa-prop.h.
20778 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20780 (ipcp_store_bits_results): Likewise.
20781 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20782 Do not write to existing jump functions but use a temporary instead.
20784 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20787 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20788 attempt to use its first operand as BIT_FIELD_REF base.
20790 2017-03-01 Richard Biener <rguenther@suse.de>
20792 PR middle-end/79721
20793 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20794 interpolating formula in wrapping arithmetic.
20795 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20797 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20799 PR tree-optimization/79734
20800 * tree-vect-generic.c (expand_vector_condition): Optimize
20801 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20802 Handle VEC_COND_EXPR where comparison has different inner width from
20803 type's inner width.
20805 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20807 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20808 markup, and similar issues. Remove @opindex entries for things
20809 that aren't options. Add missing -mmpy-option entries.
20811 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20813 PR tree-optimization/79737
20814 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20815 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20816 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20817 instead of byte_size. Formatting fix.
20818 (shift_bytes_in_array_right): Formatting fix.
20820 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20823 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20824 condition on optimize for the leaf function test.
20826 2017-02-28 Martin Liska <mliska@suse.cz>
20829 * read-rtl-function.c (function_reader::handle_unknown_directive):
20830 Bail out when one uses -flto.
20832 2017-02-28 Martin Liska <mliska@suse.cz>
20834 * common.opt: Replace space with tabular for options of <number>
20836 * config/i386/i386.opt: Show <number> value for
20837 -mlarge-data-threshold.
20838 * opts.c (print_filtered_help): Do not display number in hexadecimal
20841 2017-02-28 Martin Liska <mliska@suse.cz>
20843 * common.opt: Fix --help=option -Q for options which are of
20846 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20848 * config/i386/i386.c (print_reg): Error out for values
20849 of 8-bit size in invalid integer register.
20851 2017-02-28 Martin Sebor <msebor@redhat.com>
20853 PR tree-optimization/79691
20854 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20856 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20859 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20860 gcc_unreachable with output_operand_lossage.
20862 2017-02-28 Richard Biener <rguenther@suse.de>
20864 PR tree-optimization/79740
20865 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20867 (visit_nary_op): Insert the nary into the hashtable if we
20868 pattern-matched sth.
20869 * tree-ssa-pre.c (eliminate_insert): Robustify.
20871 2017-02-28 Richard Biener <rguenther@suse.de>
20873 PR middle-end/79731
20874 * fold-const.c (decode_field_reference): Reject out-of-bound
20877 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20879 * config/i386/i386.c: Include intl.h.
20880 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20881 instead of just cond ? "..." : "...".
20882 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20883 * coverage.c (read_counts_file): Likewise.
20884 * omp-offload.c: Include intl.h.
20885 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20886 of just cond ? "..." : "...".
20887 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20888 of just cond ? "..." : "...".
20890 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20893 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20895 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20896 'tune for' CPU name.
20897 * config/arm/arm-cpu-data.h: Regenerated.
20899 2017-02-28 Richard Biener <rguenther@suse.de>
20901 PR tree-optimization/79732
20902 * tree-inline.c (expand_call_inline): Do not shadow var.
20904 2017-02-28 Richard Biener <rguenther@suse.de>
20906 PR tree-optimization/79723
20907 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20908 address-space properly.
20910 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20912 * doc/optinfo.texi (Optimization groups): Fix option used for
20914 * doc/invoke.texi (-fopt-info): Document "omp".
20915 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20916 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20917 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20918 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20919 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20921 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20923 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20924 OPTGROUP_OMP with "omp".
20926 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20929 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20930 for arithmetic shift of unsigned V2DI.
20932 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20934 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20935 arc/linux.h headers.
20936 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20937 (LINK_SPEC): Likewise.
20938 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20939 (EXTRA_SPECS): Likewise.
20940 (STARTFILE_SPEC): Likewise.
20941 (ENDFILE_SPEC): Likewise.
20942 (LIB_SPEC): Likewise.
20943 (TARGET_SDATA_DEFAULT): Likewise.
20944 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20945 (MULTILIB_DEFAULTS): Likewise.
20946 (DWARF2_UNWIND_INFO): Likewise.
20947 * config/arc/big.h: New file.
20948 * config/arc/elf.h: Likewise.
20949 * config/arc/linux.h: Likewise.
20950 * config/arc/t-uClibc: Remove.
20952 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20954 PR tree-optimization/77536
20955 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20956 (tree_transform_and_unroll_loop): Use above function to compute the
20957 estimated niter of unrolled loop and use it when scaling profile.
20958 Also use count info rather than frequency if it's non-zero.
20959 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20960 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20961 (vect_transform_loop): Call above function.
20963 2017-02-27 Richard Biener <rguenther@suse.de>
20965 PR tree-optimization/45397
20966 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
20967 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
20968 (visit_nary_op): Add pattern matching for CSEing sign-changed
20969 or truncated operations with wider ones.
20971 2017-02-27 Richard Biener <rguenther@suse.de>
20973 PR tree-optimization/79690
20974 * tree-vect-stmts.c (vectorizable_store): Use vector type
20975 built from the DR with address-space.
20977 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
20979 * doc/invoke.texi (Optimize Options): Refine the description
20980 of asan-use-after-return.
20982 2017-02-25 Alan Modra <amodra@gmail.com>
20984 PR rtl-optimization/79584
20985 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
20986 base, not ad->base_term, the reg within base. Remove assertion
20987 that ad->base == ad->base_term. Replace gen_int_mode using
20988 bogus mode with const0_rtx.
20990 2017-02-25 Jakub Jelinek <jakub@redhat.com>
20992 PR middle-end/79396
20993 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
20994 FMA_EXPR like tcc_binary or tcc_unary.
20996 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
20999 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21001 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21002 (output_loc_operands): Handle DW_OP_call_ref and
21003 DW_OP_GNU_variable_value.
21004 (struct variable_value_struct): New type.
21005 (struct variable_value_hasher): Likewise.
21006 (variable_value_hash): New variable.
21007 (string_types): Remove.
21008 (copy_loc_descr): New function.
21009 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21010 (prepend_loc_descr_to_each): New function.
21011 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21012 instead of add_loc_descr_to_each if the first argument is single
21013 location list and the second has multiple.
21014 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21015 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21016 when looking for variable value which doesn't have other location info.
21017 (loc_list_from_tree): Formatting fix.
21018 (gen_array_type_die): Simplify DW_AT_string_length handling.
21019 (adjust_string_types): Remove.
21020 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21021 string_types. Call resolve_variable_values.
21022 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21023 (resolve_addr_in_expr): Likewise. Add A argument.
21024 (copy_deref_exprloc): Remove deref argument. Adjust for the
21025 original expression being DW_OP_GNU_variable_value with optionally
21026 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21027 optionally after it.
21028 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21029 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21030 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21031 (variable_value_hasher::hash, variable_value_hasher::equal): New
21033 (resolve_variable_value_in_expr, resolve_variable_value,
21034 resolve_variable_values, note_variable_value_in_expr,
21035 note_variable_value): New functions.
21036 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21039 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21042 * opts.h (handle_generated_option): Add GENERATED_P argument.
21043 * opts-common.c (handle_option): Adjust function comment.
21044 (handle_generated_option): Add GENERATED_P argument, pass it to
21046 (control_warning_option): Pass false to handle_generated_option
21048 * opts.c (maybe_default_option): Pass true to handle_generated_option
21050 * optc-gen.awk: Likewise.
21052 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21054 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21055 a REG, look at the REG it is a SUBREG of.
21056 (splitter for cmpeqsi_t): Ditto.
21058 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21060 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21061 the special USEs with the pattern of the insn, not the insn itself.
21063 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21066 * doc/invoke.texi: Document -mload-store-pairs.
21068 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21069 Sandra Loosemore <sandra@codesourcery.com>
21071 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21072 argument isn't a CONST_INT.
21073 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21074 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21075 (nios2_validate_compare): Bypass alternate compare logic if *op2
21076 is not a CONST_INT.
21077 (ldstwm_operation_p): Return false if first_base is not a REG or
21078 if first_offset is not a CONST_INT.
21080 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21082 * config/cris/cris.md: Use correct operand in a define_peephole2.
21084 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21086 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21088 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21090 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21091 this_insn if it is an INSN or JUMP_INSN.
21092 (force_offsettable): Look at base, not at addr.
21093 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21094 on things that aren't necessarily CONST_INTs.
21096 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21098 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21099 -mfpmath=sse is the default also for x86-32 targets with SSE2
21100 instruction set when @option{-ffast-math} is enabled
21102 2017-02-24 Jeff Law <law@redhat.com>
21104 PR rtl-optimizatoin/79286
21105 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21108 2017-02-24 Richard Biener <rguenther@suse.de>
21110 PR tree-optimization/79389
21111 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21114 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21116 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21117 function comment to reflect reality.
21118 (loop_exits_before_overflow): Fix typo in function description.
21120 2017-02-24 Richard Biener <rguenther@suse.de>
21122 PR tree-optimization/79389
21123 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21124 properly that a threading opportunity exists. Detect conditional
21125 copy/constant propagation opportunities.
21127 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21129 * config/visium/visium.md (type): Add trap.
21130 (b): New mode attribute.
21131 (*btst): Rename into...
21132 (*btst<mode>): ...this and adjust.
21133 (*cbranchsi4_btst_insn): Rename into...
21134 (*cbranch<mode>4_btst_insn): ...this and adjust.
21135 (trap): New define_insn.
21137 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21139 PR tree-optimization/79389
21140 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21141 (noce_reversed_cond_code): New function.
21142 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21143 reversed_comparison_code. Formatting fix.
21144 (noce_try_store_flag): Test rev_cond != NULL in addition to
21145 reversed_comparison_code.
21146 (noce_try_store_flag_constants): Likewise.
21147 (noce_try_store_flag_mask): Likewise.
21148 (noce_try_addcc): Use rev_cond if non-NULL instead of
21149 reversed_comparison_code.
21150 (noce_try_cmove_arith): Likewise. Formatting fixes.
21151 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21152 (noce_find_if_block): Initialize rev_cond.
21153 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21154 instead of false as last argument never attempt to reverse it
21157 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21159 PR tree-optimization/79663
21160 * tree-predcom.c (combine_chains): Process refs in reverse order
21161 only for ZERO length chains, and add explaining comment.
21163 2017-02-23 Jeff Law <law@redhat.com>
21165 PR tree-optimization/79578
21166 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21167 in call to operand_equal_p.
21169 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21172 * config/i386/cpuid.h: Fix another undefined behavior.
21174 2017-02-23 Richard Biener <rguenther@suse.de>
21176 PR tree-optimization/79683
21177 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21178 vector types for data-refs.
21180 2017-02-23 Martin Liska <mliska@suse.cz>
21182 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21184 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21186 PR middle-end/79665
21187 * internal-fn.c (get_range_pos_neg): Moved to ...
21188 * tree.c (get_range_pos_neg): ... here. No longer static.
21189 * tree.h (get_range_pos_neg): New prototype.
21190 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21191 are known to be in between 0 and signed maximum inclusive, try to
21192 expand both unsigned and signed divmod and use the cheaper one from
21195 2017-02-22 Jeff Law <law@redhat.com>
21197 PR tree-optimization/79578
21198 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21199 to compare base operands.
21201 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21204 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21205 gpc_reg_operand instead of fpr_reg_operand.
21207 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21209 * config/mips/mips.c (mips_return_in_memory): Force FP
21210 vector types to be returned in memory for o32 ABI.
21212 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21214 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21215 instead of DW_TAG_member for static data member declarations and don't
21216 set no_linkage_name for static inline data members.
21217 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21220 2017-02-22 Martin Liska <mliska@suse.cz>
21222 * doc/invoke.texi: Replace inequality signs with square brackets
21225 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21227 PR tree-optimization/68644
21228 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21230 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21233 * lra-constraints.c (simplify_operand_subreg): Handle
21234 WORD_REGISTER_OPERATIONS targets.
21236 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21239 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21240 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21241 elimination by swapping fld*.
21243 2017-02-22 Richard Biener <rguenther@suse.de>
21245 PR tree-optimization/79673
21246 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21247 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21248 irrelevant address-space qualifiers and avoiding a
21249 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21251 2017-02-22 Richard Biener <rguenther@suse.de>
21253 PR tree-optimization/79666
21254 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21255 to not symbolically negate if that may introduce undefined
21258 2017-02-22 Martin Liska <mliska@suse.cz>
21261 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21262 * data-streamer-out.c (streamer_write_gcov_count_stream):
21264 * value-prof.c (stream_out_histogram_value): Make assert more
21265 precise based on type of counter.
21267 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21270 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21271 Use nonimmediate_operand instead of memory_operand for operand 1.
21272 (float-extend standard_x87sse_constant_load splitter): Ditto.
21274 2017-02-21 Jeff Law <law@redhat.com>
21276 PR tree-optimization/79621
21277 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21278 blocks with edges to themselves.
21280 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21283 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21284 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21285 Use gimple_call_builtin_p.
21288 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21289 on temporarily removed DEBUG_INSNs.
21291 PR tree-optimization/79649
21292 * tree-loop-distribution.c (classify_partition): Give up on
21293 non-generic address space loads/stores.
21295 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21297 * doc/loop.texi (Loop manipulation): Remove nonexistent
21298 tree_ssa_loop_version from the documentation.
21299 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21301 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21304 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21305 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21306 * config/rs6000/rs6000.c: Include except.h.
21307 (rs6000_expand_split_stack_prologue): Call
21308 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21310 2017-02-21 Martin Jambor <mjambor@suse.cz>
21313 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21314 have been analyzed.
21316 2017-02-21 Martin Jambor <mjambor@suse.cz>
21318 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21319 for backward compatibility only.
21320 * doc/invoke.texi (Option Summary): Remove all references to
21321 -fipa-cp-alignment.
21323 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21327 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21329 * lra-constraints.c (curr_insn_transform): Handle
21330 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21332 2017-02-21 Martin Liska <mliska@suse.cz>
21334 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21336 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21338 PR translation/79638
21339 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21341 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21344 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21345 (arm_function_ok_for_sibcall): Return false for an indirect call by
21346 descriptor if all the argument registers are used.
21347 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21348 alignment of the function.
21350 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21352 PR tree-optimization/61441
21353 * simplify-rtx.c (simplify_const_unary_operation): For
21354 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21355 the sNaN unmodified.
21357 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21359 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21360 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21361 instead of SYSTEM_HEADER_DIR.
21363 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21364 Martin Liška <mliska@suse.cz>
21366 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21367 Fix typos and grammar, use active voice, and clarify.
21369 2017-02-20 Marek Polacek <polacek@redhat.com>
21371 PR middle-end/79537
21372 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21375 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21377 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21380 * config/i386/i386.c (ix86_expand_builtin): Handle
21381 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21382 ix86_builtins_isa[fcode].isa as a requirement of those
21383 flags and any other flag in the bitmask.
21384 (ix86_init_mmx_sse_builtins): Use 0 instead of
21385 ~OPTION_MASK_ISA_64BIT as mask.
21386 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21387 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21388 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21389 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21391 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21394 * lra-constraints.c (split_reg): Check requested split mode
21395 is supported by the register.
21397 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21399 * lra-constraints.c (simplify_operand_subreg): Remove early
21402 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21405 * lra-constraints.c (curr_insn_transform): Tighten condition
21406 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21408 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21411 * lra-constraints.c (curr_insn_transform): Handle
21412 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21414 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21417 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21419 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21421 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21424 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21427 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21429 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21432 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21434 * final.c (last_columnnum, override_columnnum): New variables.
21435 (final_start_function): Set last_columnnum, pass it to begin_prologue
21436 hook and pass 0 to dwarf2out_begin_prologue.
21437 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21438 to source_line debug hook.
21439 (notice_source_line): Compute last_columnnum and for debug_column_info
21440 return true on column changes.
21441 * debug.h (struct gcc_debug_hooks): Add column argument to
21442 source_line and begin_prologue hooks.
21443 (debug_nothing_int_charstar_int_bool): Remove prototype.
21444 (debug_nothing_int_int_charstar,
21445 debug_nothing_int_int_charstar_int_bool): New prototypes.
21446 (dwarf2out_begin_prologue): Add column argument.
21447 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21448 begin_prologue hooks.
21449 (debug_nothing_int_charstar_int_bool): Remove.
21450 (debug_nothing_int_int_charstar,
21451 debug_nothing_int_int_charstar_int_bool): New functions.
21452 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21453 through to dwarf2out_source_line.
21454 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21455 (dwarf2out_source_line): Add column argument, emit it if requested.
21456 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21458 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21459 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21460 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21461 through to dwarf2out_begin_prologue.
21462 (vmsdbgout_source_line): Add column argument, pass it through to
21463 dwarf2out_source_line.
21464 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21465 dbxout_source_line caller.
21466 (dbxout_source_line): Add column argument.
21468 * common.opt (gno-column-info, gcolumn-info): New options.
21469 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21470 (check_die): Also test for multiple DW_AT_decl_column attributes.
21471 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21472 DW_AT_decl_column if requested.
21473 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21475 (gen_variable_die): Likewise.
21476 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21477 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21480 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21481 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21482 (ix86_handle_option): Handle OPT_m3dnowa.
21483 * doc/invoke.texi (-m3dnowa): Document.
21484 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21485 -m3dnowa instead of -m3dnow -march=athlon.
21488 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21489 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21491 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21494 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21495 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21496 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21497 generator for vsx_xxpermdi_<mode>_be.
21498 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21499 force big-endian semantics.
21500 (vsx_xxpermdi_<mode>_be): New define_expand with same
21501 implementation as previous version of vsx_xxpermdi_<mode>.
21503 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21505 PR tree-optimization/79327
21506 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21507 variable, its initialization and use.
21509 2017-02-17 Julia Koval <julia.koval@intel.com>
21511 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21512 (OPTION_MASK_ISA_PKU_UNSET): New.
21513 (ix86_handle_option): Handle -mrdpid.
21514 * config/i386/cpuid.h (bit_RDPID): New.
21515 * config/i386/driver-i386.c (host_detect_local_cpu):
21516 Detect RDPID feature.
21517 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21518 * config/i386/i386-c.c (ix86_target_macros_internal):
21520 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21521 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21522 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21523 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21524 * config/i386/i386.md (define_insn "rdpid"): New.
21525 * config/i386/i386.opt Add -mrdpid.
21526 * config/i386/immintrin.h (_rdpid_u32): New.
21528 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21530 PR rtl-optimization/79541
21531 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21532 instead of transforming it into USE.
21534 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21536 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21537 If HONOR_SNANS (SFmode) force the input to a register.
21538 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21539 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21540 an frsp or similar insn.
21542 2017-02-17 Martin Liska <mliska@suse.cz>
21544 PR rtl-optimization/79577
21545 * params.def (selsched-max-sched-times): Increase minimum to 1.
21547 2017-02-17 Martin Liska <mliska@suse.cz>
21549 PR rtl-optimization/79574
21550 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21552 2017-02-17 Martin Liska <mliska@suse.cz>
21554 PR tree-optimization/79529
21555 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21556 ssa_defined_default_def_p to handle cases which are implicitly
21558 * tree-ssa.c (ssa_defined_default_def_p): New function.
21559 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21560 which are implicitly defined.
21561 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21563 2017-02-17 Richard Biener <rguenther@suse.de>
21565 PR middle-end/79576
21566 * params.def (max-ssa-name-query-depth): Limit to 10.
21568 2017-02-17 Richard Biener <rguenther@suse.de>
21570 PR tree-optimization/79552
21571 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21574 2017-02-17 Richard Biener <rguenther@suse.de>
21577 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21579 2017-02-17 Marek Polacek <polacek@redhat.com>
21581 PR middle-end/79536
21582 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21583 (fold_negate_expr): New wrapper.
21585 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21587 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21588 Correct terminology and de-emphasize pre-standard behavior.
21590 2017-02-16 Alan Modra <amodra@gmail.com>
21592 PR rtl-optimization/79286
21593 * ira.c (def_dominates_uses): New function.
21594 (update_equiv_regs): Don't create an equivalence for insns that
21595 may trap where the register def does not dominate the use.
21597 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21599 PR rtl-optimization/78127
21600 * lra.c (lra): Call lra_eliminate before finish the loop after
21603 2017-02-16 Richard Biener <rguenther@suse.de>
21605 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21607 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21608 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21609 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21610 (isl_val_int_from_wi): New function.
21611 (extract_affine_gmp): Rename to ...
21612 (extract_affine_wi): ... this, take a widest_int.
21613 (extract_affine_int): Just wrap extract_affine_wi.
21614 (add_param_constraints): Use isl_val_int_from_wi.
21615 (add_loop_constraints): Likewise, and extract_affine_wi.
21617 2017-02-15 Jeff Law <law@redhat.com>
21619 PR middle-end/79521
21620 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21621 ira_init_register_move_cost_if_necessary.
21623 2017-02-15 Martin Sebor <msebor@redhat.com>
21625 PR middle-end/32003
21626 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21627 removed in a prior commit.
21629 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21631 PR tree-optimization/79347
21632 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21633 counters during peeling.
21635 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21637 * Makefile.in (site.exp): Remove "set ISLVER".
21639 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21642 * real.c (real_from_integer): Call real_convert even for decimal.
21644 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21647 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21649 2017-02-14 Andrew Pinski <apinski@cavium.com>
21651 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21652 cores and change the partno/implementer to be correct.
21653 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21654 the 'B" as the implementer.
21655 * config/aarch64/aarch64-tune.md: Regenerate.
21657 2017-02-14 Carl Love <cel@us.ibm.com>
21659 * config/rs6000/rs6000.c: Add case statement entry to make the
21660 xvcvuxdsp built-in argument unsigned.
21661 * config/rs6000/vsx.md: Fix the source and return operand types so they
21662 match the instruction definitions from the ISA document. Fix typo
21663 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21666 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21669 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21670 member early_clobber_alts.
21671 * lra-lives.c (reg_early_clobber_p): New.
21672 (process_bb_lives): Use it.
21673 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21674 (debug_operand_data): Initialize early_clobber_alts.
21675 (setup_operand_alternative): Set up early_clobber_alts.
21676 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21677 alternatives to new_insn_reg.
21678 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21680 (lra_update_insn_regno_info): Pass the new arg.
21682 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21684 PR middle-end/79505
21685 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21686 (new_oacc_loop_raw): Don't clear already cleared fields.
21689 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21690 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21691 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21693 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21696 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21698 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21701 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21702 the extra instruction to the right place to store 128-bit constant
21705 2017-02-14 Martin Sebor <msebor@redhat.com>
21707 PR middle-end/79448
21708 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21709 warning for strings of unknown length.
21711 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21713 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21715 2017-02-14 Jeff Law <law@redhat.com>
21718 * ira-costs.c (scan_one_insn): Initialize register move costs
21719 for pseudos seen in USE/CLOBBER insns.
21721 PR tree-optimization/79095
21722 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21723 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21724 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21725 if the operands are known to be not equal, then the resulting range
21727 (intersect_ranges): If the new range is ~[0,0] and the old range is
21728 wide, then prefer ~[0,0].
21729 * tree-vrp.c (overflow_comparison_p_1): New function.
21730 (overflow_comparison_p): New function.
21731 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21732 if NAME is used in an overflow test.
21733 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21734 overflow check that can be expressed as an equality test, then adjust
21735 ops to be that equality test.
21737 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21739 * config/s390/s390-builtin-types.def: Remove flags argument.
21740 * config/s390/s390.c (s390_init_builtins): Likewise.
21742 2017-02-14 Martin Liska <mliska@suse.cz>
21744 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21745 vector. Fix trailing white spaces.
21747 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21749 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21752 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21754 PR rtl-optimization/68664
21755 * config/arm/arm.c (arm_sched_can_speculate_insn):
21756 New function. Declare prototype.
21757 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21759 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21761 PR rtl-optimization/68664
21762 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21764 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21766 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21768 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21769 max skip bytes for function, loop and jump.
21771 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21773 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21774 ABS_EXPR for gimple dump.
21776 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21779 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21781 PR tree-optimization/79408
21782 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21783 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21784 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21785 also if rhs1 is INTEGER_CST.
21787 2017-02-14 Richard Biener <rguenther@suse.de>
21789 PR middle-end/79432
21790 * tree-into-ssa.c (insert_phi_nodes): When the function can
21791 have abnormal edges rewrite SSA names with broken use-def
21792 dominance out of SSA and register them for PHI insertion.
21794 2017-02-13 Martin Sebor <msebor@redhat.com>
21796 PR middle-end/79496
21797 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21798 clearing info.nowrite flag when snprintf size argument is a range.
21800 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21802 * cprop.c (cprop_jump): Add missing space in string literal.
21803 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21804 (get_constraint_for_component_ref): Likewise.
21805 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21806 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21807 * lra-constraints.c (process_alt_operands): Likewise.
21808 * ipa-inline.c (inline_small_functions): Likewise.
21809 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21810 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21811 * trans-mem.c (diagnose_tm_1_op): Likewise.
21812 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21813 (grid_parallel_clauses_gridifiable): Likewise.
21815 * config/nvptx/mkoffload.c (process): Add space in between
21818 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21819 "MOD4_SSE_REGS" and "ALL_REGS".
21821 * spellcheck.c (test_data): Add , in between "foo" and "food".
21823 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21826 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21827 boundary crossing check and subsequent code generation agree.
21829 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21831 * config/aarch64/aarch64.c (has_memory_op): Delete.
21832 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21835 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21837 PR rtl-optimization/79388
21838 PR rtl-optimization/79450
21839 * combine.c (distribute_notes): When removing TEM_INSN for which
21840 corresponding dest has last value recorded, invalidate that last
21843 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21845 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21846 of explicit '@'. Add missing assembly comment marker on branch costs
21849 2017-02-13 Nathan Sidwell <nathan@acm.org>
21851 * gengtype-lex.l (<in_struct>): Add '/'.
21853 2017-02-13 Martin Liska <mliska@suse.cz>
21856 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21858 2017-02-13 Richard Biener <rguenther@suse.de>
21860 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21862 * configure: Re-generate.
21863 * config.in: Likewise.
21864 * graphite-dependences.c: Simplify as if
21865 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21866 * graphite-isl-ast-to-gimple.c: Likewise.
21867 * graphite-optimize-isl.c: Likewise.
21868 * graphite-poly.c: Likewise.
21869 * graphite-sese-to-poly.c: Likewise.
21870 * graphite.h: Likewise.
21871 * toplev.c: Include isl/version.h and use isl_version () for
21872 printing the ISL version.
21873 * doc/install.texi: Update ISL requirement.
21875 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21877 * doc/standards.texi (Standards): Update reference to
21880 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21882 * doc/extend.texi (Named Address Spaces): sourceware.org now
21884 * doc/install.texi (Binaries): Ditto.
21887 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21889 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21890 standard terminology "stringize"/"stringizing" throughout.
21891 * doc/cppinternals.texi: Likewise.
21893 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21895 * doc/extend.texi: Fix some spelling mistakes and typos.
21896 * doc/invoke.texi: Likewise.
21898 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21901 * params.def (inline-min-speedup) Change from 10 to 8.
21903 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21905 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21908 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21911 * ipa-inline-analysis.c (get_minimal_bb): New function.
21912 (record_modified): Use it.
21913 (remap_edge_change_prob): Handle also ancestor functions.
21915 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21917 * doc/contrib.texi (Contributors): Remove broken link into
21918 the Mauve CVS repository.
21920 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21922 PR middle-end/79454
21923 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21924 result computation whenever lhs doesn't have vector mode, not
21925 just when it has BLKmode.
21927 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21929 * doc/makefile.texi (profiledbootstrap): Refer to the
21930 installation instructions only in textual form.
21932 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21935 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21937 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21939 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21940 (Specific): Update mingw-w64 reference.
21942 (Specific): Remove broken link to Renesas RX processor.
21944 2017-02-10 Richard Biener <rguenther@suse.de>
21946 * toplev.c (process_options): Do not mention obsolete graphite
21947 options when printing sorry message about missing graphite support.
21948 Mention -floop-nest-optimize.
21950 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21952 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21953 (vtst_p16): Likewise.
21954 (vtstq_p8): Likewise.
21955 (vtstq_p16): Likewise.
21957 (vtstq_p64): Likewise.
21958 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21959 (vset_lane_p64): New.
21960 (vsetq_lane_p64): New.
21962 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21964 PR tree-optimization/79411
21965 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21966 stmt operands are SSA_NAMEs used in abnormal phis.
21967 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
21970 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21973 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
21976 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21978 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
21980 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21982 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
21986 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
21987 not arbitrary TREE_CONSTANT.
21990 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
21991 "omp declare target link" attribute unless is_global_var.
21992 * omp-offload.c (find_link_var_op): Likewise.
21994 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
21995 Chung-Lin Tang <cltang@codesourcery.com>
21997 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
21999 (gimplify_adjust_omp_clauses): Don't delete TILE.
22000 (gimplify_omp_for): Deal with TILE.
22001 * internal-fn.c (expand_GOACC_TILE): New function.
22002 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22004 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22005 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22007 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22008 avoid DIV for outermost collapse var.
22009 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22010 Remove out of date comments, fix whitespace.
22011 * omp-general.c (omp_extract_for_data): Deal with tiling.
22012 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22013 adjust OLF_DIM_BASE value.
22014 (struct omp_for_data): Add tiling field.
22015 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22016 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22017 for auto loops. Remove default auto determining, moved to
22018 oacc_loop_fixed_partitions.
22019 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22020 stmts, add e_mask field.
22021 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22022 (oacc_thread_numbers): Use oacc_dim_call.
22023 (oacc_xform_tile): New.
22024 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22025 (finish_oacc_loop): Adjust for ifns vector.
22026 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22027 add case for GOACC_TILE fns.
22028 (oacc_loop_xform_loop): Delete.
22029 (oacc_loop_process): Iterate over call list directly, and add
22030 handling for GOACC_TILE fns.
22031 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22033 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22034 vector partitioning to outer loops. Assign 2 partitions to loops
22035 when available. Add TILE handling.
22036 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22037 (execite_oacc_device_lower): Process GOACC_TILE fns,
22038 ignore unknown specs.
22039 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22040 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22041 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22043 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22045 * configure.ac (ACX_BUGURL): Update.
22046 * configure: Regenerate.
22048 2017-02-09 Richard Biener <rguenther@suse.de>
22050 PR tree-optimization/69823
22051 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22052 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22054 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22056 * config/arc/arc-c.def: Add __NPS400__ definition.
22057 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22058 (TARGET_NPS400): Define.
22060 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22062 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22064 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22065 pointer, arch_info.
22066 (arc_cpu_types): Fill the arch_info field with a pointer into the
22067 arc_arch_types table.
22068 (arc_selected_cpu): Declare.
22069 * config/arc/arc.c (arc_selected_cpu): Make global.
22070 (arc_selected_arch): Delete.
22071 (arc_base_cpu): Delete.
22072 (arc_override_options): Remove references to deleted variables,
22073 update access to arch information.
22074 (ARC_OPT): Update access to arch information.
22075 (ARC_OPTX): Likewise.
22076 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22077 (TARGET_ARC600): Update access to arch information.
22078 (TARGET_ARC601): Likewise.
22079 (TARGET_ARC700): Likewise.
22080 (TARGET_EM): Likewise.
22081 (TARGET_HS): Likewise.
22082 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22085 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22088 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22089 condition/operands for integer GE/LE/GEU/LEU operations.
22091 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22093 PR translation/79397
22094 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22097 2017-02-08 Martin Jambor <mjambor@suse.cz>
22100 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22101 whether allocation happened.
22102 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22103 nothing was allocated.
22105 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22107 PR tree-optimization/79408
22108 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22109 constant, but SSA_NAME with a known integer range, use the minimum
22110 of that range instead of op1 to determine if modulo can be replaced
22111 with its first operand.
22113 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22115 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22117 2017-02-08 Richard Biener <rguenther@suse.de>
22119 PR tree-optimization/71824
22120 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22121 Check all loops contained in the merged region.
22123 2017-02-07 Andrew Pinski <apinski@cavium.com>
22125 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22127 2017-02-07 Andrew Pinski <apinski@cavium.com>
22129 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22130 (thunderxt88): Likewise.
22131 (thunderxt81): Disable LSE and change v8.1 to v8.
22132 (thunderxt83): Likewise.
22134 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22135 Richard Biener <rguenther@suse.de>
22137 PR middle-end/79399
22138 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22139 type from int to size_t.
22140 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22142 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22144 PR rtl-optimization/79386
22145 * cprop.c (bypass_conditional_jumps): Initialize
22146 bypass_last_basic_block already before splitting bbs after
22147 unconditional traps...
22148 (bypass_conditional_jumps): ... rather than here.
22151 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22152 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22153 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22154 fix -masm=intel patterns.
22156 2017-02-07 Richard Biener <rguenther@suse.de>
22158 PR tree-optimization/79256
22159 PR middle-end/79278
22160 * builtins.c (get_object_alignment_2): Use min_align_of_type
22161 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22162 and ADJUST_FIELD_ALIGN.
22164 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22166 * doc/tm.texi: Regenerate.
22167 * stor-layout.c (layout_decl): Adjust.
22168 (update_alignment_for_field): Likewise.
22169 (place_field): Likewise.
22170 (min_align_of_type): Likewise.
22171 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22172 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22173 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22174 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22175 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22176 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22177 * config/i386/i386.c (x86_field_alignment): Likewise.
22178 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22179 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22180 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22181 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22182 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22183 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22187 2017-01-30 Richard Biener <rguenther@suse.de>
22189 PR tree-optimization/79256
22190 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22191 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22194 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22196 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22197 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22198 builtins to SImode and emit a zero-extend, if necessary.
22200 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22202 * docs/invoke.texi (RISC-V Options): Alphabetize.
22204 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22206 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22209 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22211 * config/riscv/riscv.c: New file.
22212 * common/config/riscv/riscv-common.c: Likewise.
22213 * config.gcc: Likewise.
22214 * config/riscv/constraints.md: Likewise.
22215 * config/riscv/elf.h: Likewise.
22216 * config/riscv/generic.md: Likewise.
22217 * config/riscv/linux.h: Likewise.
22218 * config/riscv/multilib-generator: Likewise.
22219 * config/riscv/peephole.md: Likewise.
22220 * config/riscv/pic.md: Likewise.
22221 * config/riscv/predicates.md: Likewise.
22222 * config/riscv/riscv-builtins.c: Likewise.
22223 * config/riscv/riscv-c.c: Likewise.
22224 * config/riscv/riscv-ftypes.def: Likewise.
22225 * config/riscv/riscv-modes.def: Likewise.
22226 * config/riscv/riscv-opts.h: Likewise.
22227 * config/riscv/riscv-protos.h: Likewise.
22228 * config/riscv/riscv.h: Likewise.
22229 * config/riscv/riscv.md: Likewise.
22230 * config/riscv/riscv.opt: Likewise.
22231 * config/riscv/sync.md: Likewise.
22232 * config/riscv/t-elf-multilib: Likewise.
22233 * config/riscv/t-linux: Likewise.
22234 * config/riscv/t-linux-multilib: Likewise.
22235 * config/riscv/t-riscv: Likewise.
22236 * configure.ac: Likewise.
22237 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22238 Waterman as RISC-V maintainers.
22239 * doc/install.texi: Add RISC-V entries.
22240 * doc/invoke.texi: Add RISC-V options section.
22241 * doc/md.texi: Add RISC-V constraints section.
22242 * configure: Regenerated.
22244 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22247 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22248 false values to be constant vectors with all 0 or all 1 bits set.
22249 (vcondu<mode><mode>): Likewise.
22250 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22252 (fpmask_comparison_operator): Update comment.
22253 (vecint_comparison_operator): New predicate.
22254 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22255 vector conditionals when the true and false values are constant
22256 vectors with all 0 bits or all 1 bits set.
22258 2017-02-06 Martin Sebor <msebor@redhat.com>
22260 PR tree-optimization/79376
22261 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22263 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22265 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22266 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22267 to simplify split condition.
22269 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22271 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22272 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22275 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22277 PR rtl-optimization/68664
22278 * target.def (can_speculate_insn): New hook.
22279 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22280 * doc/tm.texi: Regenerate.
22281 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22282 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22283 (rs6000_sched_can_speculate_insn): New function.
22285 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22287 PR tree-optimization/79284
22288 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22289 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22290 vectorizable_mask_load_store, vectorizable_operation,
22291 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22292 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22293 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22294 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22295 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22296 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22297 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22298 is_gimple_assign (stmt). Replace another such test with
22299 is_gimple_assign (stmt).
22301 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22304 * config/avr/avr.c (rtl-iter.h): Include it.
22305 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22306 (avr_legitimate_combined_insn): ...and implementation.
22308 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22310 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22311 * config/s390/s390.c (s390_const_operand_ok)
22312 (s390_canonicalize_comparison, s390_extract_part)
22313 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22314 (s390_contiguous_bitmask_p, s390_rtx_costs)
22315 (legitimize_pic_address): Likewise.
22316 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22317 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22318 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22319 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22320 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22322 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22324 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22325 REGNO($0) == REGNO($1).
22327 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22329 * config/s390/linux.h(SIZE_TYPE): Add comment.
22331 2017-02-06 Julian Brown <julian@codesourcery.com>
22332 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22333 Virendra Pathak <virendra.pathak@broadcom.com>
22335 * config/aarch64/aarch64-cores.def: Change the scheduler
22337 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22338 * config/aarch64/thunderx2t99.md: New file.
22340 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22342 * doc/standards.texi (Go Language): Update link to language
22345 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22347 * tree-eh.c (lower_resx): Sanitize profile.
22348 (cleanup_empty_eh_move_lp): Likewise.
22350 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22353 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22355 * cfgloopmanip.h (loop_version): Update prototype.
22356 * modulo-sched.c (sms_schedule): Update call of loop_version.
22357 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22358 * tree-parloops.c (gen_parallel_loop): Likewise.
22359 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22360 * tree-ssa-loop-split.c (split_loop): Likewise.
22361 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22362 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22364 2017-02-05 Martin Liska <mliska@suse.cz>
22367 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22369 (print_operand_address): Initialize a struct to zero.
22371 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22373 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22374 garbage collector only in textual form.
22376 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22378 * doc/extend.texi (x86 specific memory model extensions for
22379 transactional memory): Simplify a phrase.
22381 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22384 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22385 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22386 (atomic_storedi_1): Likewise.
22388 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22390 PR tree-optimization/79338
22391 * tree-parloops.c (gather_scalar_reductions): Don't call
22392 vect_analyze_loop_form for loop->inner before destroying loop's
22395 2017-02-03 Martin Sebor <msebor@redhat.com>
22397 PR tree-optimization/79327
22398 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22399 when precision has resulted in leading zeros.
22400 (format_integer): Adjust the likely counter to assume an unknown
22401 argument that may be zero is non-zero.
22403 2017-02-03 Jason Merrill <jason@redhat.com>
22406 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22407 avoid copying non-taken branch.
22409 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22411 PR tree-optimization/79340
22412 * tree-vect-loop.c (vectorizable_reduction): Release
22413 vec_defs elements after safe_splicing them into other vectors.
22416 PR tree-optimization/79327
22417 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22418 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22420 (format_integer): Use wide_int_to_tree instead of build_int_cst
22421 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22422 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22423 of shortest and longest sequence.
22425 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22427 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22428 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22430 2017-02-03 Walter Lee <walt@tilera.com>
22433 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22434 after initial stackframe link reg save.
22435 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22437 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22440 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22441 wu for stxssp alternative.
22443 2017-02-03 Martin Sebor <msebor@redhat.com>
22445 PR tree-optimization/79352
22446 * gimple-fold.c (get_range_strlen): Add argument.
22447 (get_range_strlen): Change return type to bool.
22448 (get_maxval_strlen): Pass in a dummy argument.
22449 * gimple-fold.h (get_range_strlen): Change return type to bool.
22450 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22451 * tree.h (array_at_struct_end_p): Add argument.
22452 * tree.c (array_at_struct_end_p): Handle it.
22454 2017-02-03 Martin Liska <mliska@suse.cz>
22457 * multiple_target.c (create_dispatcher_calls): Redirect edge
22458 from a caller of a dispatcher.
22459 (expand_target_clones): Make the clones local.
22460 (ipa_target_clone): Do both target clones and resolvers.
22461 (ipa_dispatcher_calls): Remove the pass.
22462 (pass_dispatcher_calls::gate): Likewise.
22463 (make_pass_dispatcher_calls): Likewise.
22464 * passes.def (pass_target_clone): Put as very first IPA early
22467 2017-02-03 Martin Liska <mliska@suse.cz>
22469 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22470 in case of a function with ifunc attribute.
22472 2017-02-03 Martin Liska <mliska@suse.cz>
22474 * cgraph.c (cgraph_node::dump): Dump function version info.
22475 * symtab.c (symtab_node::dump_base): Add missing new line.
22477 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22479 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22480 (ifcombine_ifandif): Use it.
22482 2017-02-03 Martin Liska <mliska@suse.cz>
22484 * doc/invoke.texi: Document default value for
22485 use-after-scope-direct-emission-threshold.
22487 2017-02-03 Martin Liska <mliska@suse.cz>
22489 PR tree-optimization/79339
22490 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22491 (format_floating): Likewise.
22493 2017-02-03 Martin Liska <mliska@suse.cz>
22496 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22498 (ipa_node_params_t::remove): Likewise.
22499 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22500 initialization from removed ipa_node_params_t::insert.
22501 (ipa_node_params::~ipa_node_params): Move from removed
22502 ipa_node_params_t::release.
22503 * symbol-summary.h (symbol_summary::m_released): New member.
22504 Do not release a summary twice. Do not allow to call finalizer
22505 for types of a summary that live in GGC memory.
22507 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22509 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22512 2017-02-02 Martin Sebor <msebor@redhat.com>
22514 PR middle-end/79275
22515 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22516 (format_string): Tighten up the range of output for non-constant
22517 strings and correct the expected range for wide non-constant strings.
22519 2017-02-02 Martin Sebor <msebor@redhat.com>
22521 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22523 PR middle-end/32003
22524 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22526 (-fdump-tree-@var): Add to index and document how to come up
22527 with pass-specific option and dump file names.
22528 (-fdump-passes): Clarify where to look for output.
22530 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22532 PR middle-end/77445
22533 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22534 statistics of the analyzed path; allow threading for speed when
22535 any of BBs along the path are optimized for speed.
22537 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22539 PR middle-end/78468
22540 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22541 settings of the virtual registers.
22544 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22546 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22547 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22550 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22552 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22553 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22555 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22557 * config/s390/s390.md: Add missing comments with the expanded
22559 * config/s390/vector.md: Likewise.
22560 * config/s390/vx-builtins.md: Likewise.
22562 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22565 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22566 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22567 conditions on a single line.
22569 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22571 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22572 __S390_VX__ to __VX__.
22574 2017-02-01 Andrew Pinski <apinski@cavium.com>
22576 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22577 stmt_info to record_stmt_cost.
22578 (vect_get_known_peeling_cost): Pass stmt_info if known to
22580 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22581 cpu_vector_cost field into
22582 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22583 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22584 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22585 splitting of scalar_stmt_cost and vec_stmt_cost.
22586 (thunderx_vector_cost): Likewise.
22587 (cortexa57_vector_cost): LIkewise.
22588 (exynosm1_vector_cost): Likewise.
22589 (xgene1_vector_cost): Likewise.
22590 (thunderx2t99_vector_cost): Improve after the splitting of the two
22592 (aarch64_builtin_vectorization_cost): Update for the splitting of
22593 scalar_stmt_cost and vec_stmt_cost.
22595 2017-02-01 Torvald Riegel <triegel@redhat.com>
22596 Richard Henderson <rth@redhat.com>
22598 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22599 conditional on existance of a fast atomic load.
22600 * optabs-query.c (can_atomic_load_p): New function.
22601 * optabs-query.h (can_atomic_load_p): Declare it.
22602 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22603 no fast atomic load is available for the particular size of access.
22604 (expand_atomic_compare_and_swap): Likewise.
22605 (expand_atomic_load): Likewise.
22606 (expand_atomic_store): Likewise.
22607 (expand_atomic_fetch_op): Likewise.
22608 * testsuite/lib/target-supports.exp
22609 (check_effective_target_sync_int_128): Remove x86 because it provides
22610 no fast atomic load.
22611 (check_effective_target_sync_int_128_runtime): Likewise.
22613 2017-02-01 Richard Biener <rguenther@suse.de>
22615 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22616 (graphite_transform_loops): Provide opt-info for optimized nests.
22617 * tree-parloop.c (parallelize_loops): Provide opt-info for
22618 parallelized loops.
22620 2017-02-01 Richard Biener <rguenther@suse.de>
22622 PR middle-end/79315
22623 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22624 was not set before.
22626 2017-02-01 Richard Biener <rguenther@suse.de>
22628 PR tree-optimization/71824
22629 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22630 Verify the loops are valid in the merged SESE region.
22631 (scop_detection::can_represent_loop_1): Check analyzing the
22632 evolution of the number of iterations in the region succeeds.
22634 2017-01-31 Ian Lance Taylor <iant@golang.org>
22636 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22637 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22639 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22641 PR preprocessor/79210
22642 * input.c (get_substring_ranges_for_loc): Replace line_width
22643 assertion with error-handling.
22645 2017-01-31 Richard Biener <rguenther@suse.de>
22647 PR tree-optimization/77318
22648 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22649 (create_pw_aff_from_tree): Take loop parameter.
22650 (add_condition_to_pbb): Pass loop of the condition to
22651 create_pw_aff_from_tree.
22653 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22655 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22656 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22658 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22662 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22664 (convert_int_to_float128): Likewise.
22665 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22666 (convert_int_to_float128): Likewise.
22667 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22668 (UNSPEC_IEEE128_CONVERT): Likewise.
22669 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22670 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22671 Use local variables for IBM extended format.
22672 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22673 (fix_trunc<mode>si2_fprs): Likewise.
22674 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22675 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22676 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22677 to know that we can now have integers of all sizes in vector
22679 (fix<uns>_<mode>di2_hw): Likewise.
22680 (float<uns>_<mode>si2_hw): Likewise.
22681 (fix_<mode>si2_hw): Likewise.
22682 (fixuns_<mode>si2_hw): Likewise.
22683 (float<uns>_<mode>di2_hw): Likewise.
22684 (float_<mode>di2_hw): Likewise.
22685 (float_<mode>si2_hw): Likewise.
22686 (floatuns_<mode>di2_hw): Likewise.
22687 (floatuns_<mode>si2_hw): Likewise.
22688 (xscvqp<su>wz_<mode>): Delete, no longer used.
22689 (xscvqp<su>dz_<mode>): Likewise.
22690 (xscv<su>dqp_<mode>): Likewise.
22691 (ieee128_mfvsrd_64bit): Likewise.
22692 (ieee128_mfvsrd_32bit): Likewise.
22693 (ieee128_mfvsrwz): Likewise.
22694 (ieee128_mtvsrw): Likewise.
22695 (ieee128_mtvsrd_64bit): Likewise.
22696 (ieee128_mtvsrd_32bit): Likewise.
22698 2017-01-31 Martin Liska <mliska@suse.cz>
22701 * ipa-prop.c (ipa_free_all_node_params): Call release method
22702 instead of ~sumbol_summary to not to trigger double times
22705 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22707 PR tree-optimization/71691
22708 * bitmap.h (class auto_bitmap): New.
22709 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22710 is_maybe_undefined instead of ssa_undefined_value_p.
22712 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22714 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22715 __S390_ARCH_LEVEL__ to __ARCH__.
22717 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22719 PR tree-optimization/79267
22720 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22721 if should_remove_lhs_p is true.
22723 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22726 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22727 (add_alignment_attribute): New.
22728 (base_type_die): Add alignment attribute.
22729 (subrange_type_die): Likewise.
22730 (modified_type_die): Likewise.
22731 (gen_array_type_die): Likewise.
22732 (gen_descr_array_type_die: Likewise.
22733 (gen_enumeration_type_die): Likewise.
22734 (gen_subprogram_die): Likewise.
22735 (gen_variable_die): Likewise.
22736 (gen_field_die): Likewise.
22737 (gen_ptr_to_mbr_type_die): Likewise.
22738 (gen_struct_or_union_type_die): Likewise.
22739 (gen_subroutine_type_die): Likewise.
22740 (gen_typedef_die): Likewise.
22741 (base_type_cmp): Compare alignment attribute.
22743 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22746 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22747 (setb_unsigned) New pattern for setb with CCUNS.
22748 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22749 subfc./subfe sequence to avoid overflow problems. Generate a
22750 shorter sequence with cmpld/setb for power9.
22751 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22752 for generating subfc. instruction.
22753 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22754 now uses this instruction.
22756 2017-01-30 Ian Lance Taylor <iant@google.com>
22759 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22760 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22762 2017-01-30 Martin Sebor <msebor@redhat.com>
22764 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22765 Move constant to the right of a relational operator.
22766 (get_mpfr_format_length, format_character, format_string): Ditto.
22767 (should_warn_p, maybe_warn): Same.
22769 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22771 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22774 * asan.c (get_translation_unit_decl): Remove function.
22775 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22777 2017-01-30 Martin Liska <mliska@suse.cz>
22779 PR gcov-profile/79259
22780 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22781 -fprofile-generate.
22783 2017-01-30 Martin Liska <mliska@suse.cz>
22786 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22787 Initialize variables with NULL value.
22789 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22792 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22794 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22796 2017-01-30 Richard Biener <rguenther@suse.de>
22798 PR tree-optimization/79276
22799 * tree-vrp.c (process_assert_insertions): Properly adjust common
22800 when removing a duplicate.
22802 * gcc.dg/torture/pr79276.c: New testcase.
22804 2017-01-30 Richard Biener <rguenther@suse.de>
22806 PR tree-optimization/79256
22807 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22808 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22810 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22812 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22815 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22816 ("*r<noxa>sbg_<mode>_sll_bitmask")
22817 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22818 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22819 Use contiguous_bitmask_nowrap_operand.
22821 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22824 * config/rs6000/altivec.h (vec_xl): Revise #define.
22825 (vec_xst): Likewise.
22827 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22829 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22831 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22833 PR rtl-optimization/79194
22834 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22835 traps before call to bypass_conditional_jumps.
22837 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22839 PR tree-optimization/71374
22840 * lra-constraints.c (check_conflict_input_operands): New.
22841 (match_reload): Use it.
22843 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22846 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22847 account to calculate conflict_set.
22849 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22851 PR rtl-optimization/78559
22852 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22853 other_insn in combine.
22855 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22857 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22858 uint16_type_node for BT_UINT16.
22860 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22862 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22863 "RTL Tests" to menu.
22864 (GIMPLE Tests): New node.
22865 (RTL Tests): New node.
22867 2017-01-27 Richard Biener <rguenther@suse.de>
22869 PR tree-optimization/79245
22870 * tree-loop-distribution.c (distribute_loop): Apply cost
22871 modeling also to detected patterns.
22873 2017-01-27 Richard Biener <rguenther@suse.de>
22875 PR tree-optimization/71433
22876 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22877 (compare_assert_loc): New function.
22878 (process_assert_insertions): Sort and optimize assert locations
22879 to remove duplicates and push down identical assertions on
22880 edges to their destination block.
22882 2017-01-27 Richard Biener <rguenther@suse.de>
22884 PR tree-optimization/79244
22885 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22886 out SSA names even if abnormal.
22888 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22890 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22891 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22892 instead of MPFR_RNDN.
22894 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22897 * arm.c (arm_option_override): Don't call build_target_option_node
22898 until after doing all option overrides.
22899 (arm_valid_target_attribute_tree): Likewise.
22901 2017-01-27 Martin Liska <mliska@suse.cz>
22903 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22904 for {cd}tors and C++ {cd}tors.
22906 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22908 * config/s390/s390.md ("*setmem_long_and")
22909 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22911 2017-01-26 Martin Sebor <msebor@redhat.com>
22913 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22916 2017-01-26 Martin Sebor <msebor@redhat.com>
22918 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22919 HAVE_DFmode before using XFmode or DFmode.
22920 (parse_directive): Avoid using the z length modifier to avoid
22921 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22923 PR middle-end/78703
22924 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22925 to accept adjustment as an array.
22926 (get_int_range): New function.
22927 (struct directive): Make width and prec arrays.
22928 (directive::set_width, directive::set_precision): Call get_int_range.
22929 (format_integer, format_floating): Handle width and precision ranges.
22930 (format_string, parse_directive): Same.
22932 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22935 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22936 comdat_type_p set, just clone them, but keep the children in the
22940 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22941 which have direct callers with -fvar-tracking-assignments enabled
22943 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22944 inside of type units.
22946 2017-01-26 Martin Sebor <msebor@redhat.com>
22948 PR middle-end/78703
22949 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22951 (struct format_result): Replace number_chars, number_chars_min,
22952 and number_chars_max with a single member of struct result_range.
22954 (format_result::operator+=): Adjust.
22955 (struct fmtresult): Remove bounded. Handle likely and unlikely
22957 (fmtresult::adjust_for_width_or_precision): New function.
22958 (fmtresult:type_max_digits): New function.
22959 (bytes_remaining): Handle likely and unlikely counters.
22960 (min_bytes_remaining): Remove.
22961 (format_percent): Simplify.
22962 (format_integer, format_floating): Set likely and unlikely counters.
22963 (get_string_length, format_character, format_string): Same.
22964 (format_plain, should_warn_p): New function.
22965 (maybe_warn): Call should_warn_p. Update diagnostic messages
22966 and handle those for all directives, including plain strings.
22967 (format_directive): Handle likely and unlikely counters.
22968 Remove unnecessary quoting from diagnostics. Add an informational
22970 (add_bytes): Remove.
22971 (pass_sprintf_length::compute_format_length): Simplify.
22972 (try_substitute_return_value): Handle likely and unlikely counters.
22974 2017-01-26 Carl Love <cel@us.ibm.com>
22976 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
22977 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
22979 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
22982 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
22983 endianess for subregs into account.
22984 * lra-constraints.c (lra_constraints): Do risky transformations
22985 always on the first iteration.
22986 * lra-lives.c (check_pseudos_live_through_calls): Add arg
22987 last_call_used_reg_set.
22988 (process_bb_lives): Define and use last_call_used_reg_set.
22989 * lra.c (lra): Always continue after lra_constraints on the first
22992 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
22994 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
22996 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
22998 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23000 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23001 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23002 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23003 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23004 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23005 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23006 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23007 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23008 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23010 2017-01-26 Marek Polacek <polacek@redhat.com>
23013 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23014 for the third operand.
23016 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23018 PR middle-end/79236
23019 * omp-low.c (struct omp_context): Add simt_stmt field.
23020 (scan_omp_for): Return omp_context *.
23021 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23022 context to the _simt_ SIMD stmt.
23023 (lower_omp_for): For combined SIMD with sibling _simt_
23024 SIMD, make sure to use the same decls in _looptemp_
23025 clauses as in the sibling.
23027 2017-01-26 David Sherwood <david.sherwood@arm.com>
23029 PR middle-end/79212
23030 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23033 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23036 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23037 emit fld b; fld a; if possible.
23039 * brig-builtins.def: Update copyright years.
23040 * config/arm/arm_acle_builtins.def: Update copyright years.
23042 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23045 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23046 constraint instead of o for the stxsd instruction.
23048 2017-01-25 Carl Love <cel@us.ibm.com>
23050 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23051 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23053 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23055 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23057 2017-01-25 Richard Biener <rguenther@suse.de>
23059 PR tree-optimization/69264
23060 * target.def (vector_alignment_reachable): Improve documentation.
23061 * doc/tm.texi: Regenerate.
23062 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23064 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23065 earlier changes with respect to TYPE_USER_ALIGN.
23066 (vector_alignment_reachable_p): Likewise. Improve dumping.
23068 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23071 * config/arm/arm.md (xordi3): Force constant operand into a register
23074 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23076 * doc/invoke.texi (-fstore-merging): Correct default optimization
23077 levels at which it is enabled.
23078 (-O): Move -fstore-merging from list to...
23081 2017-01-25 Richard Biener <rguenther@suse.de>
23084 * omp-expand.c: Include debug.h.
23085 (expand_omp_taskreg): Make sure to generate early debug before
23086 outlining anything from a function.
23087 (expand_omp_target): Likewise.
23088 (grid_expand_target_grid_body): Likewise.
23090 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23093 * asan.c (get_translation_unit_decl): New function.
23094 (asan_add_global): Extract modules file name from globals
23095 TRANSLATION_UNIT_DECL name.
23097 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23100 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23101 for long calls with APCS frame and VFP.
23103 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23105 * cfg.c (original_copy_tables_initialized_p): New function.
23106 * cfg.h (original_copy_tables_initialized_p): New decl.
23107 * cfgrtl.c (relink_block_chain): Guard the call to
23108 free_original_copy_tables with a call to
23109 original_copy_tables_initialized_p.
23110 * cgraph.h (symtab_node::native_rtl_p): New decl.
23111 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23112 (symtab_node::needed_p): Don't assert for early assembly output
23113 for __RTL functions.
23114 (cgraph_node::finalize_function): Set "force_output" for __RTL
23116 (cgraph_node::analyze): Bail out early for __RTL functions.
23117 (analyze_functions): Update assertion to support __RTL functions.
23118 (cgraph_node::expand): Bail out early for __RTL functions.
23119 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23121 * function.h (struct function): Update comment for field
23123 * gimple-expr.c: Include "tree-pass.h".
23124 (gimple_has_body_p): Return false for __RTL functions.
23125 * Makefile.in (OBJS): Add run-rtl-passes.o.
23126 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23128 (gcc::pass_manager::get_clean_slate): New accessor.
23129 * passes.c: Include "insn-addr.h".
23130 (should_skip_pass_p): Add logging. Update logic for running
23131 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23132 property-provider override so it is only done for gimple passes.
23134 (skip_pass): New function.
23135 (execute_one_pass): Call skip_pass when skipping passes.
23136 * read-md.c (md_reader::read_char): Support filtering
23137 the input to a subset of line numbers.
23138 (md_reader::md_reader): Initialize fields
23139 m_first_line and m_last_line.
23140 (md_reader::read_file_fragment): New function.
23141 * read-md.h (md_reader::read_file_fragment): New decl.
23142 (md_reader::m_first_line): New field.
23143 (md_reader::m_last_line): New field.
23144 * read-rtl-function.c (function_reader::create_function): Only
23145 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23146 curr_properties. Set DECL_INITIAL to a dummy block.
23147 (read_rtl_function_body_from_file_range): New function.
23148 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23150 * run-rtl-passes.c: New file.
23151 * run-rtl-passes.h: New file.
23153 2017-01-24 Jeff Law <law@redhat.com>
23155 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23158 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23160 PR tree-optimization/79159
23161 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23162 (record_nonwrapping_iv): Improve boundary using above function if no
23163 value range information.
23165 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23166 Martin Jambor <mjambor@suse.cz>
23168 * brig-builtins.def: New file.
23169 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23170 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23171 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23172 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23173 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23174 * builtin-types.def (BT_INT8): New.
23175 (BT_INT16): Likewise.
23176 (BT_UINT8): Likewise.
23177 (BT_UINT16): Likewise.
23178 (BT_FN_ULONG): Likewise.
23179 (BT_FN_UINT_INT): Likewise.
23180 (BT_FN_UINT_ULONG): Likewise.
23181 (BT_FN_UINT_LONG): Likewise.
23182 (BT_FN_UINT_PTR): Likewise.
23183 (BT_FN_ULONG_PTR): Likewise.
23184 (BT_FN_INT8_FLOAT): Likewise.
23185 (BT_FN_INT16_FLOAT): Likewise.
23186 (BT_FN_UINT32_FLOAT): Likewise.
23187 (BT_FN_UINT16_FLOAT): Likewise.
23188 (BT_FN_UINT8_FLOAT): Likewise.
23189 (BT_FN_UINT64_FLOAT): Likewise.
23190 (BT_FN_UINT16_UINT32): Likewise.
23191 (BT_FN_UINT32_UINT16): Likewise.
23192 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23193 (BT_FN_INT_PTR_INT): Likewise.
23194 (BT_FN_UINT_PTR_UINT): Likewise.
23195 (BT_FN_LONG_PTR_LONG): Likewise.
23196 (BT_FN_ULONG_PTR_ULONG): Likewise.
23197 (BT_FN_VOID_UINT64_UINT64): Likewise.
23198 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23199 (BT_FN_INT8_INT8_INT8): Likewise.
23200 (BT_FN_INT16_INT16_INT16): Likewise.
23201 (BT_FN_INT_INT_INT): Likewise.
23202 (BT_FN_UINT_FLOAT_UINT): Likewise.
23203 (BT_FN_FLOAT_UINT_UINT): Likewise.
23204 (BT_FN_ULONG_UINT_UINT): Likewise.
23205 (BT_FN_ULONG_UINT_PTR): Likewise.
23206 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23207 (BT_FN_UINT_UINT_UINT): Likewise.
23208 (BT_FN_VOID_UINT_PTR): Likewise.
23209 (BT_FN_UINT_UINT_PTR: Likewise.
23210 (BT_FN_UINT32_UINT64_PTR): Likewise.
23211 (BT_FN_INT_INT_UINT_UINT): Likewise.
23212 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23213 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23214 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23215 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23216 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23217 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23218 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23219 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23220 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23221 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23222 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23223 * doc/frontends.texi: List BRIG FE.
23224 * doc/install.texi (Testing): Add BRIG tesring requirements.
23225 * doc/invoke.texi (Overall Options): Mention BRIG.
23226 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23228 2017-01-24 Richard Biener <rguenther@suse.de>
23230 PR translation/79208
23231 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23233 2017-01-24 Martin Jambor <mjambor@suse.cz>
23236 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23237 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23238 and known_contexts.
23240 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23242 PR middle-end/79123
23243 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23244 casts from signed to unsigned really don't have a range.
23246 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23248 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23249 GMP_RNDx for compatiblity.
23251 2017-01-24 Martin Liska <mliska@suse.cz>
23254 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23255 that would prevent us to call alloca with -1 as argument.
23257 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23259 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23260 Avoid -Wformat-security warning.
23262 2017-01-23 Andrew Pinski <apinski@cavium.com>
23264 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23267 2017-01-23 Martin Sebor <msebor@redhat.com>
23269 PR middle-end/78703
23270 * gimple-ssa-sprintf.c (warn_level): New global.
23271 (format_integer): Use it here and throughout the rest of the file.
23272 Use the same switch to compute sign as base.
23273 (maybe_warn): New function.
23274 (format_directive): Factor out warnings into maybe_warn.
23275 Add debugging output. Use warn_level.
23276 (add_bytes): Use warn_level.
23277 (pass_sprintf_length::compute_format_length): Add debugging output.
23278 (try_substitute_return_value): Same.
23279 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23281 PR middle-end/78703
23282 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23283 (struct fmtresult, format_integer, format_floating): Adjust.
23284 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23285 (get_string_length, format_string,format_directive): Same.
23286 (pass_sprintf_length::compute_format_length): Same.
23287 (try_substitute_return_value): Simplify slightly.
23289 PR middle-end/78703
23290 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23291 (fmtresult::operator+=): Outlined.
23292 (struct fmtresult): Add ctors.
23293 (struct conversion_spec): Rename...
23294 (struct directive): ...to this. Add and remove data members.
23295 (directive::set_width, directive::set_precision): New functions.
23296 (format_percent): Use fmtresult ctor.
23297 (get_width_and_precision): Remove.
23298 (format_integer): Make naming changes. Avoid computing width and
23300 (format_floating): Same. Adjust indentation.
23301 (format_character, format_none): New functions.
23302 (format_string): Moved character handling to format_character.
23303 (format_directive): Remove arguments, change return type.
23304 (parse_directive): New function.
23305 (pass_sprintf_length::compute_format_length): Move directive
23306 parsing to parse_directive.
23308 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23310 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23311 (assign_assembler_name_if_needed): ... this.
23312 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23313 (assign_assembler_name_if_needed): ... this.
23314 (free_lang_data_in_cgraph): Adjust callers.
23315 * cgraphunit.c (cgraph_node::analyze): Likewise.
23316 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23319 2017-01-23 Richard Biener <rguenther@suse.de>
23321 PR tree-optimization/79088
23322 PR tree-optimization/79188
23323 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23324 resetting loop bounds after last path deletion. Reset loop
23325 bounds of the target loop, make code match the comments.
23326 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23327 Make sure loops need no fixups.
23329 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23331 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23332 exponent support with double type for first argument.
23333 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23334 type returned by __builtin_vec_extract_sig,
23335 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23336 functions from "vector int" to "vector unsigned int" or from
23337 "vector long long int" to "vector unsigned long long int".
23338 Changed type returned by __builtin_vec_extract_exp,
23339 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23340 functions from "vector int" to "vector unsigned int" or from
23341 "vector long long int" to "vector unsigned long long int".
23342 Changed return type of __builtin_vec_test_data_class,
23343 __builtin_vec_test_data_class_sp, and
23344 __builtin_vec_test_data_class_dp from "vector int" to
23345 "vector bool int" or from "vector long long int" to "vector bool
23346 long long int" and changed second argument type from "unsigned
23347 int" to "int". Added new overloaded function forms "vector float
23348 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23349 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23350 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23351 double, vector unsigned long long int)" and "vector double
23352 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23353 long int)". Changed return type of
23354 __builtin_scalar_test_data_class and
23355 __builtin_scalar_test_data_class_sp and
23356 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23357 int" and changed second argument from "unsigned int" to "int".
23358 Changed type returned by __builtin_scalar_test_neg,
23359 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23360 from "int" to "bool int". Added new overloaded function form
23361 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23362 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23363 exponent double-precision with floating point first argument.
23364 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23365 documentation of scalar_test_data_class, scalar_test_neg,
23366 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23367 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23368 vec_test_data_class built-in functions to reflect refinements in
23369 their type signatures.
23371 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23373 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23375 (aarch64_elf_asm_destructor): Likewise.
23377 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23379 PR rtl-optimization/78634
23380 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23381 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23382 * ifcvt.c (noce_try_cmove): Add missing cost check.
23384 PR rtl-optimization/71724
23385 * combine.c (if_then_else_cond): Look for situations where it is
23386 beneficial to undo the work of one of the recursive calls.
23388 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23390 PR tree-optimization/70754
23391 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23392 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23393 combined stmt before it if not NULL.
23394 (combine_chains): Process refs reversely and compute dominance point
23397 2017-01-23 Martin Liska <mliska@suse.cz>
23399 PR tree-optimization/79196
23400 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23401 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23403 (strlen_optimize_stmt): Call the renamed function.
23405 2017-01-23 Michael Matz <matz@suse.de>
23407 PR tree-optimization/78384
23408 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23410 2017-01-23 Richard Biener <rguenther@suse.de>
23412 PR tree-optimization/79186
23413 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23414 both incoming edges before moving an assert.
23416 2017-01-23 Martin Jambor <mjambor@suse.cz>
23418 * ipa-prop.c (load_from_param_1): Removed.
23419 (load_from_unmodified_param): Bits from load_from_param_1 put back
23421 (load_from_param): Removed.
23422 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23423 with stmt. Reverted back to use of load_from_unmodified_param.
23425 2017-01-23 Martin Jambor <mjambor@suse.cz>
23428 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23429 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23430 field a pointer to garbage collected vector, mark lattices and
23431 ipcp_orig_node with GTY((skip)).
23432 (ipa_get_param_count): Adjust to descriptors being a pointer.
23433 (ipa_get_param): Likewise.
23434 (ipa_get_type): Likewise.
23435 (ipa_get_param_move_cost): Likewise.
23436 (ipa_set_param_used): Likewise.
23437 (ipa_get_controlled_uses): Likewise.
23438 (ipa_set_controlled_uses): Likewise.
23439 (ipa_is_param_used): Likewise.
23440 (ipa_node_params_t): Move into garbage collector. New methods insert
23442 (ipa_node_params_sum): Annotate wth GTY(()).
23443 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23445 (ipa_load_from_parm_agg): Adjust declaration.
23446 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23447 * ipa-profile.c (ipa_profile): Likewise.
23448 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23449 (ipa_populate_param_decls): Make descriptors parameter garbage
23451 (ipa_dump_param): Adjust to descriptors being a pointer.
23452 (ipa_alloc_node_params): Likewise.
23453 (ipa_initialize_node_params): Likewise.
23454 (load_from_param_1): Make descriptors parameter garbage collected.
23455 (load_from_unmodified_param): Likewise.
23456 (load_from_param): Likewise.
23457 (ipa_load_from_parm_agg): Likewise.
23458 (ipa_node_params::~ipa_node_params): Removed.
23459 (ipa_free_all_node_params): Remove call to delete operator.
23460 (ipa_node_params_t::insert): New.
23461 (ipa_node_params_t::remove): Likewise.
23462 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23463 copy known_csts and known_contexts vectors.
23464 (ipa_read_node_info): Adjust to descriptors being a pointer.
23465 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23467 (ipcp_transform_function): Make descriptors variable garbage
23470 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23472 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23473 * config/i386/avx512dqintrin.h: Ditto.
23474 * config/i386/avx512fintrin.h: Ditto.
23475 * config/i386/i386.c: Handle new builtins.
23476 * config/i386/i386-builtin.def: Add new builtins.
23477 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23478 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23480 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23481 Martin Liska <mliska@suse.cz>
23483 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23484 * asan.c (asan_expand_poison_ifn): Support stores and use
23485 appropriate ASAN report function.
23486 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23487 * internal-fn.def (ASAN_POISON_USE): Declare.
23488 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23489 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23490 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23491 ASAN_POISON calls w/o LHS.
23492 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23493 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23494 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23495 * gimplify.c (asan_poison_variables): Add attribute
23496 use_after_scope_memory to variables that really needs to live
23498 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23499 having the attribute.
23501 2017-01-23 Martin Liska <mliska@suse.cz>
23503 * asan.c (create_asan_shadow_var): New function.
23504 (asan_expand_poison_ifn): Likewise.
23505 * asan.h (asan_expand_poison_ifn): New declaration.
23506 * internal-fn.c (expand_ASAN_POISON): Likewise.
23507 * internal-fn.def (ASAN_POISON): New builtin.
23508 * sanopt.c (pass_sanopt::execute): Expand
23509 asan_expand_poison_ifn.
23510 * tree-inline.c (copy_decl_for_dup_finish): Make function
23512 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23513 * tree-ssa.c (is_asan_mark_p): New function.
23514 (execute_update_addresses_taken): Rewrite local variables
23515 (identified just by use-after-scope as addressable) into SSA.
23517 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23519 * doc/install.texi (Specific): opensource.apple.com uses https
23520 now. Remove trailing slash.
23522 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23524 * README.Portability: Remove note on an Irix compatibility issue.
23526 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23528 * gcov.c (INCLUDE_ALGORITHM): Define.
23529 (INCLUDE_VECTOR): Define.
23530 No longer include <vector> and <algorithm> directly.
23532 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23534 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23536 * doc/invoke.texi (Code Gen Options): Ditto.
23538 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23541 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23543 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23545 rtl-optimization/79125
23546 * cprop.c (local_cprop_pass): Handle cases where we make an
23547 unconditional trap.
23549 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23553 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23554 read from, for big endian.
23556 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23558 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23559 register pauth builtins for LP64 only.
23561 2017-01-20 Marek Polacek <polacek@redhat.com>
23564 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23567 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23569 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23571 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23572 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23573 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23575 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23578 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23579 in big-endian mode when they are not a single duplicated value.
23581 2017-01-20 Richard Biener <rguenther@suse.de>
23583 * BASE-VER: Bump to 7.0.1.
23585 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23587 * omp-low.c (omplow_simd_context): New struct. Use it...
23588 (lower_rec_simd_input_clauses): ...here and...
23589 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23590 references to idx, lane, max_vf, is_simt.
23592 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23594 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23597 2017-01-20 Martin Jambor <mjambor@suse.cz>
23599 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23600 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23602 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23603 (GTFILES): Rename hsa.c to hsa-common.c.
23604 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23605 * hsa-dump.c: Likewise.
23606 * hsa-gen.c: Likewise.
23607 * hsa-regalloc.c: Likewise.
23608 * ipa-hsa.c: Likewise.
23609 * omp-expand.c: Likewise.
23610 * omp-low.c: Likewise.
23611 * toplev.c: Likewise.
23613 2017-01-20 Marek Polacek <polacek@redhat.com>
23616 * doc/invoke.texi: Document -Wduplicated-branches.
23617 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23618 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23619 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23620 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23621 return 0 only when not OEP_LEXICOGRAPHIC.
23622 (fold_build_cleanup_point_expr): Use the expression
23623 location when building CLEANUP_POINT_EXPR.
23624 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23625 * tree.c (add_expr): Handle error_mark_node.
23627 2017-01-20 Martin Liska <mliska@suse.cz>
23630 * tree-profile.c (init_ic_make_global_vars): Do not call
23632 (gimple_init_gcov_profiler): Likewise.
23634 2017-01-20 Martin Liska <mliska@suse.cz>
23637 * cgraph.h (maybe_create_reference): Remove argument and
23639 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23641 * ipa-cp.c (create_specialized_node): Likewise.
23642 * symtab.c (symtab_node::maybe_create_reference): Handle
23643 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23645 2017-01-20 Martin Liska <mliska@suse.cz>
23647 * read-rtl-function.c (function_reader::create_function): Use
23648 build_decl instread of build_decl_stat.
23650 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23652 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23653 * config/i386/avx512dqintrin.h: Ditto.
23654 * config/i386/avx512fintrin.h: Ditto.
23655 * config/i386/i386-builtin-types.def: Add new types.
23656 * config/i386/i386.c: Handle new types.
23657 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23658 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23659 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23660 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23661 (__builtin_ia32_kshiftridi): New.
23662 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23664 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23668 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23669 define to rs6000_init_stack_protect_guard.
23670 (rs6000_init_stack_protect_guard): New function.
23672 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23673 Yunqiang Su <yunqiang.su@imgtec.com>
23675 * config.gcc (supported_defaults): Add madd4.
23676 (with_madd4): Add validation.
23677 (all_defaults): Add madd4.
23678 * config/mips/mips.opt (mmadd4): New option.
23679 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23681 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23683 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23684 (ISA_HAS_FUSED_MADD4): Likewise.
23685 * doc/invoke.texi (-mmadd4): Document the new option.
23686 * doc/install.texi (--with-madd4): Document the new option.
23688 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23690 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23691 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23692 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23693 (aarch64_init_pauth_hint_builtins): New.
23694 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23695 (aarch64_expand_builtin): Expand new builtins.
23697 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23699 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23700 * combine-stack-adj.c (no_unhandled_cfa): Handle
23701 REG_CFA_TOGGLE_RA_MANGLE.
23702 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23703 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23704 info for return address signing.
23705 (aarch64_expand_epilogue): Likewise.
23707 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23709 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23710 * config/aarch64/aarch64-protos.h
23711 (aarch64_return_address_signing_enabled): New declaration.
23712 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23714 (aarch64_expand_prologue): Sign return address before it's pushed onto
23716 (aarch64_expand_epilogue): Authenticate return address fetched from
23718 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23719 (aarch64_attributes): New function attributes for "sign-return-address".
23720 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23721 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23722 ("*do_return"): Generate combined instructions according to key index.
23723 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23724 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23726 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23727 * config/aarch64/aarch64.opt (msign-return-address=): New.
23728 * doc/extend.texi (AArch64 Function Attributes): Documents
23729 "sign-return-address=".
23730 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23732 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23734 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23735 overall option summary.
23737 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23739 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23740 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23741 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23742 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23744 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23746 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23747 -mpower9-minmax by default for -mcpu=power9.
23748 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23749 128-bit floating point.
23751 2017-01-20 Alan Modra <amodra@gmail.com>
23753 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23754 optimizing for size.
23756 2017-01-20 Alan Modra <amodra@gmail.com>
23759 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23760 for strcmp and strncmp from corresponding builtin decl.
23762 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23764 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23765 instead of i386/rtems-64.h.
23766 * config/i386/rtems-64.h: Remove.
23768 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23772 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23774 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23776 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23778 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23779 Change int to HOST_WIDE_INT.
23780 * config/aarch64/aarch64-protos.h
23781 (aarch64_simd_gen_const_vector_dup): Likewise.
23782 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23784 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23786 * langhooks-def.h (lhd_type_for_size): New decl.
23787 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23788 * langhooks.c (lhd_type_for_size): New function, taken from
23791 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23793 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23794 define_bypass for CR latency.
23795 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23796 (power9-alu2): Add define_bypass for CR latency.
23798 (power9-mul): Update insn latency.
23799 (power9-mul-compare): Update insn latency, bypass latency and remove
23802 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23804 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23806 * config/aarch64/aarch64.md
23807 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23808 aarch64_nopcrelative_literal_loads.
23809 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23811 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23813 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23814 TARGET_LOONGSON_3A.
23815 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23817 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23820 * config.gcc (supported_defaults): Add lxc1-sxc1.
23821 (with_lxc1_sxc1): Add validation.
23822 (all_defaults): Add lxc1-sxc1.
23823 * config/mips/mips.opt (mlxc1-sxc1): New option.
23824 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23826 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23827 __mips_no_lxc1_sxc1.
23828 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23829 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23830 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23832 2017-01-19 Richard Biener <rguenther@suse.de>
23834 PR tree-optimization/72488
23835 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23836 sure to restore SSA info.
23837 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23839 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23841 PR rtl-optimization/79121
23842 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23843 of the inner type when shifting an extended value.
23845 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23848 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23849 interposable aliases.
23851 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23854 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23855 Use the evmergelohi instruction.
23856 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23857 (mov_sitf_e500_subreg8_2_be): Likewise.
23858 (mov_sitf_e500_subreg12_2_le): Likewise.
23859 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23860 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23861 (mov_sitf_e500_subreg8_2_le): Likewise.
23862 (mov_sitf_e500_subreg12_2_be): Likewise.
23864 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23866 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23867 attribute from vecsimple to vecperm.
23868 (altivec_vbpermq2): Likewise.
23870 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23873 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23875 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23876 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23877 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23878 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23879 case where N arg is SIZE_MAX.
23880 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23881 (cmpstrsi): Add pattern.
23883 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23885 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23886 __builtin_vec_revb builtins.
23887 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23888 built-in functions to support generation of the ISA 3.0 XXBR<x>
23889 vector byte reverse instructions.
23890 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23891 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23892 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23893 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23894 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23895 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23896 (P9V_BUILTIN_VEC_REVB): Likewise.
23897 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23898 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23899 (p9_xxbrq_v16qi): Likewise.
23900 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23901 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23902 (p9_xxbrh_v8hi): Likewise.
23903 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23904 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23905 vec_revb built-in functions.
23907 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23909 PR rtl-optimization/78952
23910 * config/i386/i386.md (any_extract): New code iterator.
23911 (*insvqi_2): Use any_extract for source operand.
23912 (*insvqi_3): Use any_shiftrt for source operand.
23914 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23916 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23918 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23920 2017-01-18 Matthias Klose <doko@ubuntu.com>
23922 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23924 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23926 * config/rs6000/altivec.h (vec_bperm): Change #define.
23927 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23928 (altivec_vbpermq2): New define_insn.
23929 (altivec_vbpermd): Likewise.
23930 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23931 function interface.
23932 (VBPERMD): Likewise.
23933 (VBPERM): New polymorphic function interface.
23934 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23935 Add entries for P9V_BUILTIN_VEC_VBPERM.
23936 * doc/extend.texi: Add interfaces for vec_bperm.
23938 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23940 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23941 first letter of error messages.
23942 (s390_resolve_overloaded_builtin): Likewise.
23943 * config/s390/s390.c (s390_expand_builtin): Likewise.
23944 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23945 (s390_valid_target_attribute_inner_p): Likewise.
23946 * config/s390/s390.md ("tabort"): Likewise.
23948 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23950 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23951 (ISA_AVOID_DIV_HILO): New macro.
23952 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23953 (ISA_HAS_DDIV): Likewise.
23955 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23957 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23959 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23961 * doc/invoke.texi (fabi-version): Spelling fix.
23963 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23966 * doc/invoke.texi (fabi-version): Mention mangling fix for
23969 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23972 * doc/invoke.texi (fabi-version): Document discriminator mangling.
23974 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
23977 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
23978 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
23980 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
23982 (stack_protect_test): Ditto.
23983 * config/rs6000/rs6000.opt (mstack-protector-guard=,
23984 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
23986 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
23987 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
23988 -mstack-protector-guard-offset=.
23989 (RS/6000 and PowerPC Options): Ditto.
23991 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23993 * config/i386/i386.h (MASK_CLASS_P): New define.
23994 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
23995 there are no registers from different register sets also when
23996 mask registers are used. Update function comment.
23997 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
23998 to (*k/*r) and (*k/*km) alternatives.
24000 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24002 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24003 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24004 (EH_RETURN_HANDLER_RTX): New define.
24005 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24006 Force frame pointer in EH return functions.
24007 (aarch64_expand_epilogue): Add barrier for eh_return.
24008 (aarch64_final_eh_return_addr): Remove.
24009 (aarch64_eh_return_handler_rtx): New function.
24010 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24012 (aarch64_eh_return_handler_rtx): New prototype.
24014 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24016 * config/rs6000/altivec.h (vec_rlmi): New #define.
24017 (vec_vrlnm): Likewise.
24018 (vec_rlnm): Likewise.
24019 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24020 (UNSPEC_VRLNM): Likewise.
24021 (VIlong): New mode iterator.
24022 (altivec_vrl<VI_char>mi): New define_insn.
24023 (altivec_vrl<VI_char>nm): Likewise.
24024 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24026 (VRLDNM): Likewise.
24027 (RLNM): New polymorphic function entry.
24028 (VRLWMI): New monomorphic function entry.
24029 (VRLDMI): Likewise.
24030 (RLMI): New polymorphic function entry.
24031 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24032 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24033 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24036 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24039 * dwarf2out.c (field_byte_offset): Restore the
24040 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24041 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24042 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24045 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24048 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24050 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24053 * dwarf2out.c (add_data_member_location_attribute): For constant
24054 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24055 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24056 DW_AT_byte_size attributes.
24058 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24060 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24061 after forcing to constant memory when the code model is medium.
24063 2017-01-17 Julia Koval <julia.koval@intel.com>
24066 * config/i386/avx512fintrin.h
24067 (_mm512_i32gather_ps): Change __addr type to void const*.
24068 (_mm512_mask_i32gather_ps): Ditto.
24069 (_mm512_i32gather_pd): Ditto.
24070 (_mm512_mask_i32gather_pd): Ditto.
24071 (_mm512_i64gather_ps): Ditto.
24072 (_mm512_mask_i64gather_ps): Ditto.
24073 (_mm512_i64gather_pd): Ditto.
24074 (_mm512_mask_i64gather_pd): Ditto.
24075 (_mm512_i32gather_epi32): Ditto.
24076 (_mm512_mask_i32gather_epi32): Ditto.
24077 (_mm512_i32gather_epi64): Ditto.
24078 (_mm512_mask_i32gather_epi64): Ditto.
24079 (_mm512_i64gather_epi32): Ditto.
24080 (_mm512_mask_i64gather_epi32): Ditto.
24081 (_mm512_i64gather_epi64): Ditto.
24082 (_mm512_mask_i64gather_epi64): Ditto.
24083 (_mm512_i32scatter_ps): Change __addr type to void*.
24084 (_mm512_mask_i32scatter_ps): Ditto.
24085 (_mm512_i32scatter_pd): Ditto.
24086 (_mm512_mask_i32scatter_pd): Ditto.
24087 (_mm512_i64scatter_ps): Ditto.
24088 (_mm512_mask_i64scatter_ps): Ditto.
24089 (_mm512_i64scatter_pd): Ditto.
24090 (_mm512_mask_i64scatter_pd): Ditto.
24091 (_mm512_i32scatter_epi32): Ditto.
24092 (_mm512_mask_i32scatter_epi32): Ditto.
24093 (_mm512_i32scatter_epi64): Ditto.
24094 (_mm512_mask_i32scatter_epi64): Ditto.
24095 (_mm512_i64scatter_epi32): Ditto.
24096 (_mm512_mask_i64scatter_epi32): Ditto.
24097 (_mm512_i64scatter_epi64): Ditto.
24098 (_mm512_mask_i64scatter_epi64): Ditto.
24099 * config/i386/avx512pfintrin.h
24100 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24101 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24102 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24103 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24104 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24105 (_mm512_prefetch_i32scatter_ps): Ditto.
24106 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24107 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24108 (_mm512_prefetch_i64scatter_pd): Ditto.
24109 (_mm512_prefetch_i64scatter_ps): Ditto.
24110 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24111 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24112 * config/i386/avx512vlintrin.h
24113 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24114 (_mm_mmask_i32gather_ps): Ditto.
24115 (_mm256_mmask_i32gather_pd): Ditto.
24116 (_mm_mmask_i32gather_pd): Ditto.
24117 (_mm256_mmask_i64gather_ps): Ditto.
24118 (_mm_mmask_i64gather_ps): Ditto.
24119 (_mm256_mmask_i64gather_pd): Ditto.
24120 (_mm_mmask_i64gather_pd): Ditto.
24121 (_mm256_mmask_i32gather_epi32): Ditto.
24122 (_mm_mmask_i32gather_epi32): Ditto.
24123 (_mm256_mmask_i32gather_epi64): Ditto.
24124 (_mm_mmask_i32gather_epi64): Ditto.
24125 (_mm256_mmask_i64gather_epi32): Ditto.
24126 (_mm_mmask_i64gather_epi32): Ditto.
24127 (_mm256_mmask_i64gather_epi64): Ditto.
24128 (_mm_mmask_i64gather_epi64): Ditto.
24129 (_mm256_i32scatter_ps): Change __addr type to void*.
24130 (_mm256_mask_i32scatter_ps): Ditto.
24131 (_mm_i32scatter_ps): Ditto.
24132 (_mm_mask_i32scatter_ps): Ditto.
24133 (_mm256_i32scatter_pd): Ditto.
24134 (_mm256_mask_i32scatter_pd): Ditto.
24135 (_mm_i32scatter_pd): Ditto.
24136 (_mm_mask_i32scatter_pd): Ditto.
24137 (_mm256_i64scatter_ps): Ditto.
24138 (_mm256_mask_i64scatter_ps): Ditto.
24139 (_mm_i64scatter_ps): Ditto.
24140 (_mm_mask_i64scatter_ps): Ditto.
24141 (_mm256_i64scatter_pd): Ditto.
24142 (_mm256_mask_i64scatter_pd): Ditto.
24143 (_mm_i64scatter_pd): Ditto.
24144 (_mm_mask_i64scatter_pd): Ditto.
24145 (_mm256_i32scatter_epi32): Ditto.
24146 (_mm256_mask_i32scatter_epi32): Ditto.
24147 (_mm_i32scatter_epi32): Ditto.
24148 (_mm_mask_i32scatter_epi32): Ditto.
24149 (_mm256_i32scatter_epi64): Ditto.
24150 (_mm256_mask_i32scatter_epi64): Ditto.
24151 (_mm_i32scatter_epi64): Ditto.
24152 (_mm_mask_i32scatter_epi64): Ditto.
24153 (_mm256_i64scatter_epi32): Ditto.
24154 (_mm256_mask_i64scatter_epi32): Ditto.
24155 (_mm_i64scatter_epi32): Ditto.
24156 (_mm_mask_i64scatter_epi32): Ditto.
24157 (_mm256_i64scatter_epi64): Ditto.
24158 (_mm256_mask_i64scatter_epi64): Ditto.
24159 (_mm_i64scatter_epi64): Ditto.
24160 (_mm_mask_i64scatter_epi64): Ditto.
24161 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24162 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24163 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24164 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24165 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24166 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24167 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24168 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24169 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24170 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24171 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24172 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24173 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24174 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24175 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24176 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24177 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24178 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24179 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24180 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24181 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24182 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24183 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24184 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24185 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24186 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24187 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24188 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24189 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24190 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24191 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24192 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24193 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24194 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24195 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24196 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24197 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24198 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24199 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24200 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24201 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24202 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24203 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24204 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24205 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24206 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24207 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24208 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24209 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24210 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24211 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24212 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24213 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24214 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24215 definitions accordingly.
24217 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24218 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24221 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24224 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24227 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24228 hard regs for inner regno.
24230 2017-01-17 Martin Liska <mliska@suse.cz>
24233 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24234 assumption and add comment.
24236 2017-01-17 Nathan Sidwell <nathan@acm.org>
24238 * ipa-visibility.c (localize_node): New function, broken out of ...
24239 (function_and_variable_visibility): ... here. Call it.
24241 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24243 PR middle-end/77445
24244 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24245 correctly set frequency of oudgoing edge.
24246 (duplicate_thread_path): Fix profile updating.
24248 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24251 * configure.ac: Add GCC_BASE_VER.
24252 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24253 version from BASE-VER file.
24254 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24255 (gcc.o): Depend on $(BASEVER).
24256 * common.opt (dumpfullversion): New option.
24257 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24258 * doc/invoke.texi: Document -dumpfullversion.
24259 * doc/install.texi: Document --with-gcc-major-version-only.
24260 * configure: Regenerated.
24262 2017-01-17 Richard Biener <rguenther@suse.de>
24264 PR tree-optimization/71433
24265 * tree-vrp.c (register_new_assert_for): Merge same asserts
24266 on all incoming edges.
24267 (process_assert_insertions_for): Handle insertions at the
24270 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24272 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24273 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24275 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24278 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24281 2017-01-17 Alan Modra <amodra@gmail.com>
24284 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24285 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24286 symbolic stack limit when pic.
24288 2017-01-16 Martin Sebor <msebor@redhat.com>
24290 PR tree-optimization/78608
24291 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24293 2017-01-16 Jeff Law <law@redhat.com>
24296 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24297 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24298 for several include directories that may be relative to sysroot.
24299 * config/i386/x-mingw32 (gplus_includedir): Define.
24300 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24301 (native_system_includedir): Likewise.
24302 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24303 override if TARGET_SYSTEM_ROOT is defined.
24304 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24306 PR tree-optimization/79090
24307 PR tree-optimization/33562
24308 PR tree-optimization/61912
24309 PR tree-optimization/77485
24310 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24311 and computed trims into the dump file.
24313 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24315 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24317 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24320 * gimplify.c (gimplify_init_constructor): If want_value and
24321 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24325 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24326 sequence. Formatting fixes.
24327 (doloop_optimize): Formatting fixes.
24330 * gcc.c (debug_level_greater_than_spec_func): New function.
24331 (static_spec_functions): Add debug-level-gt spec function.
24332 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24334 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24335 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24336 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24337 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24338 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24339 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24341 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24343 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24344 QImode fixups to general and mask registers only.
24346 2017-01-16 Carl Love <cel@us.ibm.com>
24348 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24349 for built-in functions
24350 vector signed char vec_nabs (vector signed char)
24351 vector signed short vec_nabs (vector signed short)
24352 vector signed int vec_nabs (vector signed int)
24353 vector signed long long vec_nabs (vector signed long long)
24354 vector float vec_nabs (vector float)
24355 vector double vec_nabs (vector double)
24356 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24358 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24359 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24360 * doc/extend.texi: Update the documentation file for the new built-in
24363 2017-01-16 Martin Sebor <msebor@redhat.com>
24365 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24368 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24370 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24371 UNSPEC_VSX__XXSPLTD to require special splat handling.
24373 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24376 * system.h: Poison strndup.
24378 2017-01-16 Alan Modra <amodra@gmail.com>
24381 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24384 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24386 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24388 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24390 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24391 call recog here. Assert that INSN_CODE (insn) is non-negative.
24393 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24396 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24398 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24399 in the currently scheduled RTL fragment.
24401 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24403 PR rtl-optimization/78751
24404 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24407 2017-01-14 Jeff Law <law@redhat.com>
24409 PR tree-optimization/79090
24410 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24411 variable length stores.
24412 (compute_trims): Delete dead assignment to *trim_tail.
24413 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24416 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24418 PR rtl-optimization/78626
24419 PR rtl-optimization/78727
24420 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24421 of a block, and split such blocks after everything else is finished.
24423 2017-01-14 Alan Modra <amodra@gmail.com>
24426 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24427 target legitimate_combined_insn.
24428 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24429 (rs6000_legitimate_combined_insn): New function.
24430 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24432 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24433 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24434 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24436 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24438 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24440 2017-01-13 Jeff Law <law@redhat.com>
24442 PR tree-optimization/33562
24443 PR tree-optimization/61912
24444 PR tree-optimization/77485
24445 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24447 (delete_dead_assignment): Likewise.
24448 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24449 statement to delete_dead_call and delete_dead_assignment.
24451 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24454 * substring-locations.c (format_warning_va): Strengthen case 1 so
24455 that both endpoints of the substring must be within the format
24456 range for just the substring to be printed.
24458 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24460 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24461 * config/i386/i386.c (ix86_target_string): Add missing options
24462 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24463 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24464 flags_other and ix86_target_other to flags2_other. Display unknown
24466 (ix86_valid_target_attribute_inner_p): Add missing options and
24467 reorder options by implied ISAs, as in ix86_target_string.
24469 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24471 * hash-table.h (hash_table::too_empty_p): New function.
24472 (hash_table::expand): Use it.
24473 (hash_table::traverse): Likewise.
24474 (hash_table::empty_slot): Use sizeof (value_type) instead of
24475 sizeof (PTR) to convert bytes to elements. Shrink the table
24476 if the current size is excessive for the current number of
24479 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24481 * ira-costs.c (record_reg_classes): Break from the inner loop
24482 early once alt_fail is known to be true. Update outer loop
24483 handling accordingly.
24485 2017-01-13 Jeff Law <law@redhat.com>
24487 * tree-ssa-dse.c (decrement_count): New function.
24488 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24489 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24490 when we know the partially dead statement is a mem* function.
24492 PR tree-optimization/61912
24493 PR tree-optimization/77485
24494 * tree-ssa-dse.c: Include expr.h.
24495 (maybe_trim_constructor_store): New function.
24496 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24498 PR tree-optimization/33562
24499 PR tree-optimization/61912
24500 PR tree-optimization/77485
24501 * doc/invoke.texi: Document new dse-max-object-size param.
24502 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24503 * tree-ssa-dse.c: Include params.h.
24504 (dse_store_status): New enum.
24505 (initialize_ao_ref_for_dse): New, partially extracted from
24507 (valid_ao_ref_for_dse, normalize_ref): New.
24508 (setup_live_bytes_from_ref, compute_trims): Likewise.
24509 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24510 (maybe_trim_partially_dead_store): Likewise.
24511 (maybe_trim_complex_store): Likewise.
24512 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24513 Track what bytes live from the original store. Return tri-state
24514 for dead, partially dead or live.
24515 (dse_dom_walker): Add constructor, destructor and new private members.
24516 (delete_dead_call, delete_dead_assignment): New extracted from
24518 (dse_optimize_stmt): Make a member of dse_dom_walker.
24519 Use initialize_ao_ref_for_dse.
24521 PR tree-optimization/33562
24522 PR tree-optimization/61912
24523 PR tree-optimization/77485
24524 * sbitmap.h (bitmap_count_bits): Prototype.
24525 (bitmap_clear_range, bitmap_set_range): Likewise.
24526 * sbitmap.c (bitmap_clear_range): New function.
24527 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24529 2017-01-13 Martin Liska <mliska@suse.cz>
24532 * function.c (set_cfun): Add new argument force.
24533 * function.h (set_cfun): Likewise.
24534 * ipa-inline-transform.c (inline_call): Use the function when
24535 strict alising from is dropped for function we inline to.
24537 2017-01-13 Richard Biener <rguenther@suse.de>
24539 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24540 for dumping GIMPLE INTEGER_CSTs.
24542 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24544 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24545 to 201112L since C++17.
24547 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24550 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24551 if -fsanitize=kernel-address is present.
24553 2017-01-13 Richard Biener <rguenther@suse.de>
24555 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24556 as _Literal ( type ) number in case usual suffixes do not
24557 preserve all information.
24559 2017-01-13 Richard Biener <rguenther@suse.de>
24561 PR tree-optimization/77283
24562 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24563 and ssa-iterators.h.
24564 (is_feasible_trace): Implement a cost model based on joiner
24567 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24570 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24571 char or short to __float128/_Float128 directly.
24573 2017-01-12 Martin Sebor <msebor@redhat.com>
24575 to -Wformat-overflow.
24576 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24577 (min_bytes_remaining): Same.
24578 (get_string_length): Same.
24579 (format_string): Same.
24580 (format_directive): Same.
24582 (pass_sprintf_length::handle_gimple_call): Same.
24584 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24586 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24587 info.nowrite calls with no lhs that can't throw. Return bool
24588 whether gsi_remove has been called or not.
24589 (pass_sprintf_length::handle_gimple_call): Return bool whether
24590 try_substitute_return_value called gsi_remove. Formatting fix.
24591 (pass_sprintf_length::execute): Don't use gsi_remove if
24592 handle_gimple_call returned true.
24595 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24596 be removed due to side-effects, don't remove following barrier nor
24597 turn the successor edge into fallthru edge.
24599 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24602 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24603 element-reversing loads and stores as not swappable.
24605 2017-01-12 Nathan Sidwell <nathan@acm.org>
24606 Nicolai Stange <nicstange@gmail.com>
24608 * combine.c (try_combine): Don't ignore result of overlap checking
24609 loop. Combine overlap & asm check into single loop.
24611 2017-01-12 Richard Biener <rguenther@suse.de>
24613 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24614 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24616 2017-01-12 Richard Biener <rguenther@suse.de>
24618 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24619 and TS_TARGET_OPTION directly derive from TS_BASE.
24620 * tree-core.h (tree_optimization_option): Derive from tree_base.
24621 (tree_target_option): Likewise.
24623 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24625 * config/i386/i386.c (memory_address_length): Increase len
24626 only when rip_relative_addr_p returns false.
24628 2017-01-11 Julia Koval <julia.koval@intel.com>
24630 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24631 (OPTION_MASK_ISA_SGX_SET): New.
24632 (ix86_handle_option): Handle OPT_msgx.
24633 * config.gcc: Added sgxintrin.h.
24634 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24635 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24636 * config/i386/i386.c (ix86_target_string): Add -msgx.
24638 (ix86_option_override_internal): Handle new options.
24639 (ix86_valid_target_attribute_inner_p): Add sgx.
24640 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24641 * config/i386/i386.opt: Add msgx.
24642 * config/i386/sgxintrin.h: New file.
24643 * config/i386/x86intrin.h: Add sgxintrin.h.
24645 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24648 * fold-const.c (maybe_nonzero_address): Return 1 for function
24650 (tree_single_nonzero_warnv_p): Don't handle function local objects
24654 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24657 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24660 * opts.c: Include "spellcheck.h"
24661 (struct string_fragment): New struct.
24662 (struct edit_distance_traits<const string_fragment &>): New
24664 (get_closest_sanitizer_option): New function.
24665 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24667 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24669 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24671 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24672 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24673 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24674 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24675 for initial die_offset if dwarf_split_debug_info.
24676 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24677 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24679 (output_skeleton_debug_sections): Formatting fix. Use
24680 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24681 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24683 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24685 * config/arm/cortex-a53.md: Add bypasses for
24686 cortex_a53_r2f_cvt.
24687 (cortex_a53_r2f): Only use for transfers.
24688 (cortex_a53_f2r): Likewise.
24689 (cortex_a53_r2f_cvt): Add reservation for conversions.
24690 (cortex_a53_f2r_cvt): Likewise.
24692 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24694 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24695 to all inlined functions, change static to extern.
24697 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24700 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24702 (arm_assemble_integer): Likewise.
24704 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24706 * config.gcc: Use new awk script to check CPU, FPU and architecture
24707 parameters for --with-... options.
24708 * config/arm/parsecpu.awk: New file
24709 * config/arm/arm-cpus.in: New file.
24710 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24712 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24714 * config/arm/t-arm: Update dependency rules.
24715 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24716 of processing .def files.
24717 * config/arm/genopt.sh: Deleted.
24718 * config/arm/gentune.sh: Deleted.
24719 * config/arm/arm-cores.def: Deleted.
24720 * config/arm/arm-arches.def: Deleted.
24721 * config/arm/arm-fpus.def: Deleted.
24722 * config/arm/arm-tune.md: Regenerated.
24723 * config/arm/arm-tables.opt: Regenerated.
24724 * config/arm/arm-cpu.h: New generated file.
24725 * config/arm/arm-cpu-data.h: New generated file.
24726 * config/arm/arm-cpu-cdata.h: New generated file.
24728 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24731 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24733 (input_varpool_node): Unpack dynamically_initialized bit.
24735 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24737 PR rtl-optimization/79032
24738 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24739 the alignment of the adjusted memory reference against that of MODE,
24740 instead of the alignment of the original memory reference.
24742 2017-01-11 Martin Jambor <mjambor@suse.cz>
24744 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24746 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24747 decorated functions.
24749 2017-01-11 Richard Biener <rguenther@suse.de>
24751 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24752 set range/nonnull info for PHI results. Do not set it on
24753 stmts marked for removal.
24755 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24757 * expr.c (store_field): In the bitfield case, fetch the return value
24758 from the registers before applying a single big-endian adjustment.
24759 Always do a final load for a BLKmode value not larger than a word.
24761 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24764 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24765 that we correctly handle column numbers greater than
24766 LINE_MAP_MAX_COLUMN_NUMBER.
24768 2017-01-10 Martin Sebor <msebor@redhat.com>
24770 PR middle-end/78245
24771 * gimple-ssa-sprintf.c (get_destination_size): Call
24772 {init,fini}object_sizes.
24773 * tree-object-size.c (addr_object_size): Adjust.
24774 (pass_through_call): Adjust.
24775 (pass_object_sizes::execute): Adjust.
24776 * tree-object-size.h (fini_object_sizes): Declare.
24778 2017-01-10 Martin Sebor <msebor@redhat.com>
24780 PR tree-optimization/78775
24781 * builtins.c (get_size_range): Move...
24782 * calls.c: ...to here.
24783 (alloc_max_size): Accept zero argument.
24784 (operand_signed_p): Remove.
24785 (maybe_warn_alloc_args_overflow): Call get_size_range.
24786 * calls.h (get_size_range): Declare.
24788 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24790 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24791 from TI's devices.csv file as of September 2016.
24792 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24794 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24796 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24797 * doc/invoke.texi: Likewise.
24798 * doc/md.texi: Likewise.
24799 * doc/objc.texi: Likewise.
24801 2017-01-10 Joshua Conner <joshconner@google.com>
24803 * config/arm/fuchsia-elf.h: New file.
24804 * config/fuchsia.h: New file.
24805 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24806 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24808 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24810 2016-01-10 Richard Biener <rguenther@suse.de>
24812 PR tree-optimization/79034
24813 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24814 Propagate out degenerate PHIs in the joiner.
24816 2017-01-10 Martin Liska <mliska@suse.cz>
24818 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24819 (sort_congruence_classes_by_decl_uid): Likewise.
24820 (sort_congruence_class_groups_by_decl_uid): Likewise.
24821 (sem_item_optimizer::merge_classes): Sort class, groups in these
24822 classes and members in the groups by DECL_UID of declarations.
24823 This would make merge operations stable.
24825 2017-01-10 Martin Liska <mliska@suse.cz>
24827 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24828 usage of m_classes_vec.
24829 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24830 (sem_item_optimizer::get_group_by_hash): Likewise.
24831 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24832 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24833 (sem_item_optimizer::verify_classes): Likewise.
24834 (sem_item_optimizer::process_cong_reduction): Likewise.
24835 (sem_item_optimizer::dump_cong_classes): Likewise.
24836 (sem_item_optimizer::merge_classes): Likewise.
24837 * ipa-icf.h (congruence_class_hash): Rename from
24838 congruence_class_group_hash. Remove declaration of m_classes_vec.
24840 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24842 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24843 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24844 * config.gcc: Add avx512vpopcntdqintrin.h.
24845 * config/i386/avx512vpopcntdqintrin.h: New.
24846 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24847 * config/i386/i386-builtin-types.def: Add new types.
24848 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24849 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24850 __builtin_ia32_vpopcountq_v8di_mask): New.
24851 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24852 __AVX512VPOPCNTDQ__.
24853 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24854 (PTA_AVX512VPOPCNTDQ): Define.
24855 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24856 TARGET_AVX512VPOPCNTDQ_P): Define.
24857 * config/i386/i386.opt: Add mavx512vpopcntdq.
24858 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24859 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24861 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24863 PR middle-end/77484
24864 * predict.def (PRED_CALL): Set to 67.
24866 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24868 * expr.c (store_field): In the bitfield case, if the value comes from
24869 a function call and is of an aggregate type returned in registers, do
24870 not modify the field mode; extract the value in all cases if the mode
24871 is BLKmode and the size is not larger than a word.
24873 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24876 * config/i386/cpuid.h: Fix undefined behavior.
24878 2017-01-04 Jeff Law <law@redhat.com>
24880 PR tree-optimization/79007
24881 PR tree-optimization/67955
24882 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24883 conservative for pt.null when flag_non_call_exceptions is on.
24885 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24887 PR translation/79019
24888 PR translation/79020
24889 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24890 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24891 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24893 * config/avr/avr.opt (maccumulate-args): Likewise.
24894 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24895 * common.opt (freport-bug): Likewise.
24896 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24897 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24898 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24899 translatable string.
24900 * config/i386/i386.c (function_value_32): Likewise.
24901 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24902 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24904 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24905 * common/config/msp430/msp430-common.c (msp430_handle_option):
24907 * symtab.c (symtab_node::verify_base): Likewise.
24908 * opts.c (set_debug_level): Likewise.
24909 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24910 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24911 missing whitespace to translatable strings.
24912 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24913 * config/sh/superh.h: Likewise.
24914 * config/i386/xopintrin.h: Likewise.
24915 * config/i386/znver1.md: Likewise.
24916 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24917 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24918 * double-int.h (struct double_int): Likewise.
24919 * double-int.c (div_and_round_double): Likewise.
24920 * wide-int.cc: Likewise.
24921 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24922 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24923 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24924 (crossjumps_occurred): ... this.
24925 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24928 PR tree-optimization/78899
24929 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24930 returning bool return struct loop *, NULL for failure and the new
24932 (versionable_outer_loop_p): Don't version outer loop if it has
24933 dont_vectorized bit set.
24934 (tree_if_conversion): When versioning outer loop, ensure
24935 tree_if_conversion is performed also on the inner loop of the
24936 non-vectorizable outer loop copy.
24937 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24938 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24939 prevent vectorization of it.
24940 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24941 the outer loop vectorization of the non-scalar version is attempted
24942 before vectorization of the inner loop in scalar version. If
24943 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24944 vectorization of its inner loop.
24945 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24946 has 2 inner loops, rename also on edges from bb whose single pred
24947 is outer_loop->header. Fix typo in function comment.
24949 2017-01-09 Martin Sebor <msebor@redhat.com>
24952 * asan.c (asan_emit_stack_protection): Increase local buffer size
24953 to avoid snprintf truncation warning.
24955 2017-01-09 Andrew Pinski <apinski@cavium.com>
24957 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24958 to reference thunderx2t99 for the tuning structure
24959 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24961 (thunderx2t99_extra_costs): This.
24962 * config/aarch64/aarch64-tune.md: Regenerate.
24963 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24964 (vulcan_addrcost_table): This.
24965 (vulcan_regmove_cost): Rename to ...
24966 (thunderx2t99_regmove_cost): This.
24967 (vulcan_vector_cost): Rename to ...
24968 (thunderx2t99_vector_cost): this.
24969 (vulcan_branch_cost): Rename to ...
24970 (thunderx2t99_branch_cost): This.
24971 (vulcan_tunings): Rename to ...
24972 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
24973 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
24975 2017-01-09 Martin Jambor <mjambor@suse.cz>
24979 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
24980 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
24981 (propagate_vr_accross_jump_function): Use the above function for all
24982 value range computations for pass-through jump functions and type
24983 converasion from explicit value range values.
24984 (ipcp_propagate_stage): Do not attempt to deduce types of formal
24985 parameters from TYPE_ARG_TYPES.
24986 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
24987 (ipa_write_node_info): Stream type of the actual argument.
24988 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
24990 2017-01-09 Martin Liska <mliska@suse.cz>
24993 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
24994 (lookup_compiler): Do not show error message with have_E.
24996 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24998 PR tree-optimization/78938
24999 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25000 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25001 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25002 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25005 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25007 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25010 2017-01-09 Richard Biener <rguenther@suse.de>
25012 PR tree-optimization/78997
25013 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25014 name condition properly.
25016 2017-01-09 Richard Biener <rguenther@suse.de>
25019 * dwarf2out.c (is_cxx): New overload with context.
25020 (is_naming_typedef_decl): Use it.
25022 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25024 * invoke.texi (Option Summary): Correct spacing in option lists
25025 and add line breaks to fix over-long lines.
25027 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25029 PR middle-end/17660
25031 * extend.texi (Common Variable Attributes): Add xref to GCC
25032 Internals manual to explain mode attribute keywords.
25034 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25037 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25038 and Preprocessor Options.
25039 (Options for Linking): Document -pthread here....
25040 (RS/6000 and PowerPC Options): ...not here.
25041 (Solaris 2 Options): ...or here.
25042 * doc/cppopts.texi: Document -pthread.
25044 2017-01-08 Martin Sebor <msebor@redhat.com>
25046 PR middle-end/77708
25047 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25048 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25049 New member functions.
25050 (format_directive): Used them.
25052 (pass_sprintf_length::handle_gimple_call): Same.
25053 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25054 to avoid truncation for any argument.
25055 (extract_affine_mul): Same.
25056 * tree.c (get_file_function_name): Same.
25058 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25060 PR middle-end/77484
25061 * predict.def (PRED_INDIR_CALL): Set to 86.
25063 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25065 PR preprocessor/54124
25066 * doc/cppopts.texi: Reformat -d subtable to list the full name
25067 of the options. Add cross-reference to the docs for the general
25068 compiler -d options.
25069 * doc/invoke.texi (Developer Options): Add cross-reference to the
25070 preprocessor-specific -d option documentation.
25072 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25074 PR preprocessor/13498
25075 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25076 redudant material, and reflect new command-line options.
25077 (System Headers): Likewise.
25079 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25081 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25082 -isystem, and -idirafter. Copy-edit.
25083 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25084 default for -ftrack-macro-expansion. Delete obsolete and
25085 badly-formatted implementation details about -fdebug-cpp output.
25086 * doc/cppwarnopts.texi: Copy-edit.
25088 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25091 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25092 that the transition from a max line width >= 1<<10 to narrower
25093 lines works correctly.
25095 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25097 * doc/options.texi (PerFunction): New.
25098 * opt-functions.awk (switch_flags): Map both Optimization and
25099 PerFunction to CL_OPTIMIZATION.
25100 * opth-gen.awk: Test for PerFunction flag along with
25102 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25103 it only when the latter is present. Skip those that don't in
25104 the hash function generator.
25105 * common.opt (fvar-tracking): Mark as PerFunction instead of
25107 (fvar-tracking-assignments): Likewise.
25108 (fvar-tracking-assignments-toggle): Likewise.
25109 (fvar-tracking-uninit): Likewise.
25111 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25113 PR translation/79018
25114 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25117 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25120 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25121 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25122 TARGET_LONG_JUMP_TABLE_OFFSETS.
25123 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25124 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25125 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25126 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25127 * config/m68k/m68k.md (tablejump expander): Likewise.
25128 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25129 TARGET_LONG_JUMP_TABLE_OFFSETS.
25130 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25131 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25133 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25134 David Holsgrove <david.holsgrove@xilinx.com>
25136 * common/config/microblaze/microblaze-common.c
25137 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25138 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25140 * config/microblaze/microblaze.c (microblaze_must_save_register)
25141 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25143 (microblaze_eh_return): New function.
25144 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25145 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25146 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25147 * config/microblaze/microblaze.md (eh_return): New pattern.
25149 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25151 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25152 GCC_DIAGNOSTIC_STRINGIFY): Define.
25154 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25156 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25158 * config/arm/arm.md (<mcrr>): New.
25160 * config/arm/arm.c (arm_arch5te): New.
25161 (arm_option_override): Set arm_arch5te.
25162 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25164 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25165 (arm_mcrr_qualifiers): ... this. New.
25166 (MRRC_QUALIFIERS): Define to...
25167 (arm_mrrc_qualifiers): ... this. New.
25168 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25170 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25171 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25172 (MRRCI, mrrc, MRRC): New.
25173 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25174 VUNSPEC_MRRC2): New.
25176 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25178 * config/arm/arm.md (<mcr>): New.
25180 * config/arm/arm.c (arm_coproc_builtin_available): Add
25181 support for mcr, mrc, mcr2 and mrc2.
25182 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25183 (arm_mcr_qualifiers): ... this. New.
25184 (MRC_QUALIFIERS): Define to ...
25185 (arm_mrc_qualifiers): ... this. New.
25186 (MCR_QUALIFIERS): Define to ...
25187 (arm_mcr_qualifiers): ... this. New.
25188 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25190 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25191 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25192 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25193 VUNSPEC_MRC2): New.
25195 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25197 * config/arm/arm.md (*ldc): New.
25201 * config/arm/arm.c (arm_coproc_builtin_available): Add
25202 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25203 (arm_coproc_ldc_stc_legitimate_address): New.
25204 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25205 'qualifier_const_pointer'.
25206 (LDC_QUALIFIERS): Define to...
25207 (arm_ldc_qualifiers): ... this. New.
25208 (STC_QUALIFIERS): Define to...
25209 (arm_stc_qualifiers): ... this. New.
25210 * config/arm/arm-protos.h
25211 (arm_coproc_ldc_stc_legitimate_address): New.
25212 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25213 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25214 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25215 stc2, stcl, stc2l): New.
25216 * config/arm/constraints.md (Uz): New.
25217 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25218 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25219 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25220 VUNSPEC_STC2L): New.
25222 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25224 * config/arm/arm.md (<cdp>): New.
25225 * config/arm/arm.c (neon_const_bounds): Rename this ...
25226 (arm_const_bounds): ... this.
25227 (arm_coproc_builtin_available): New.
25228 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25229 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25230 (CDP_QUALIFIERS): Define to...
25231 (arm_cdp_qualifiers): ... this. New.
25233 (arm_expand_builtin_args): Add case for 6 arguments.
25234 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25235 (arm_const_bounds): ... this.
25236 (arm_coproc_builtin_available): New.
25237 * config/arm/arm_acle.h (__arm_cdp): New.
25239 * config/arm/arm_acle_builtins.def (cdp): New.
25241 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25242 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25243 'arm_const_bounds'.
25244 * config/arm/types.md (coproc): New.
25245 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25246 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25247 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25248 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25250 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25252 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25253 (UBINOP_QUALIFIERS): New.
25255 (acle_builtin_data): New. Change comment.
25256 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25257 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25258 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25259 arm_acle_builtins.def.
25260 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25261 (arm_init_acle_builtins): New.
25262 (CRC32_BUILTIN): Remove.
25263 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25264 crc32cb, crc32ch and crc32cw.
25265 (arm_init_crc32_builtins): Remove.
25266 (arm_init_builtins): Use arm_init_acle_builtins rather
25267 than arm_init_crc32_builtins.
25268 (arm_expand_acle_builtin): New.
25269 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25270 * config/arm/arm_acle_builtins.def: New.
25272 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25274 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25275 (arm_builtin_datum): ... this.
25276 (arm_init_neon_builtin): Rename to ...
25277 (arm_init_builtin): ... this. Add a new parameters PREFIX
25278 and USE_SIG_IN_NAME.
25279 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25280 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25281 'arm_builtin_datum'.
25282 (arm_init_vfp_builtins): Likewise.
25283 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25284 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25285 (arm_expand_neon_args): Rename to ...
25286 (arm_expand_builtin_args): ... this. Rename builtin_arg
25287 enum values and differentiate between ARG_BUILTIN_MEMORY
25288 and ARG_BUILTIN_NEON_MEMORY.
25289 (arm_expand_neon_builtin_1): Rename to ...
25290 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25291 values, arm_expand_builtin_args and add bool parameter NEON.
25292 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25293 (arm_expand_vfp_builtin): Likewise.
25294 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25296 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25298 PR middle-end/77484
25299 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25300 * predict.c (tree_estimate_probability_bb): Reverse direction of
25301 polymorphic call predictor.
25303 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25305 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25306 (determine_pass_name_match): ...this new function and...
25307 (should_skip_pass_p): ...this new function.
25309 2017-01-06 Nathan Sidwell <nathan@acm.org>
25311 * ipa-visibility.c (function_and_variable_visibility): Reformat
25312 comments and long lines. Remove extrneous if.
25313 * symtab.c (symtab_node::make_decl_local): Fix code format.
25314 (symtab_node::set_section_for_node): Fix comment typo.
25316 2017-01-06 Martin Liska <mliska@suse.cz>
25319 * lra-constraints.c: Rename invariant to lra_invariant.
25320 * predict.c (set_even_probabilities): Initialize e to NULL.
25322 2017-01-05 Martin Sebor <msebor@redhat.com>
25324 PR tree-optimization/78910
25325 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25326 (format_integer): Correct off-by-one error in the handling
25327 of precision with negative numbers in signed conversions..
25329 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25331 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25333 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25335 PR tree-optimization/71016
25336 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25337 factor_out_conditional_conversion. Formatting fix.
25338 (factor_out_conditional_conversion): Add cond_stmt argument.
25339 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25340 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25343 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25345 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25346 read-rtl-function.o, and selftest-rtl.o.
25347 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25348 (selftest::aarch64_test_loading_full_dump): New function.
25349 (selftest::aarch64_run_selftests): New function.
25350 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25351 selftest::aarch64_run_selftests.
25352 * config/i386/i386.c
25353 (selftest::ix86_test_loading_dump_fragment_1): New function.
25354 (selftest::ix86_test_loading_call_insn): New function.
25355 (selftest::ix86_test_loading_full_dump): New function.
25356 (selftest::ix86_test_loading_unspec): New function.
25357 (selftest::ix86_run_selftests): Call the new functions.
25358 * emit-rtl.c (maybe_set_max_label_num): New function.
25359 * emit-rtl.h (maybe_set_max_label_num): New decl.
25360 * function.c (instantiate_decls): Guard call to
25361 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25362 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25364 * gensupport.c (gen_reader::gen_reader): Pass "false"
25365 for new "compact" param of rtx_reader.
25366 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25367 rather than an empty string for NULL strings.
25368 * read-md.c: Potentially include config.h rather than bconfig.h.
25369 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25370 (have_error): New global, copied from errors.c.
25371 (md_reader::read_name): Rename to...
25372 (md_reader::read_name_1): ...this, adding "out_loc" param,
25373 and converting "missing name or number" to returning false, rather
25375 (md_reader::read_name): Reimplement in terms of read_name_1.
25376 (md_reader::read_name_or_nil): New function.
25377 (md_reader::read_string): Handle "(nil)" by returning NULL.
25378 (md_reader::md_reader): Add new param "compact".
25379 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25380 (md_reader::read_file): New method.
25381 * read-md.h (md_reader::md_reader): Add new param "compact".
25382 (md_reader::read_file): New method.
25383 (md_reader::is_compact): New accessor.
25384 (md_reader::read_name): Convert return type from void to file_location.
25385 (md_reader::read_name_or_nil): New decl.
25386 (md_reader::read_name_1): New decl.
25387 (md_reader::m_compact): New field.
25388 (noop_reader::noop_reader): Pass "false" for new "compact" param
25390 (rtx_reader::rtx_reader): Add new "compact" param.
25391 (rtx_reader::read_rtx_operand): Make virtual and convert return
25392 type from void to rtx.
25393 (rtx_reader::read_until): New decl.
25394 (rtx_reader::handle_any_trailing_information): New virtual function.
25395 (rtx_reader::postprocess): New virtual function.
25396 (rtx_reader::finalize_string): New virtual function.
25397 (rtx_reader::m_in_call_function_usage): New field.
25398 (rtx_reader::m_reuse_rtx_by_id): New field.
25399 * read-rtl-function.c: New file.
25400 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25401 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25402 (selftest::verify_three_block_rtl_cfg): New decl.
25403 * read-rtl-function.h: New file.
25404 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25405 For host, include function.h, memmodel.h, and emit-rtl.h.
25406 (one_time_initialization): New function.
25407 (struct compact_insn_name): New struct.
25408 (compact_insn_names): New array.
25409 (find_code): Handle insn codes in compact dumps.
25410 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25411 (bind_subst_iter_and_attr): Likewise.
25412 (add_condition_to_string): Likewise.
25413 (add_condition_to_rtx): Likewise.
25414 (apply_attribute_uses): Likewise.
25415 (add_current_iterators): Likewise.
25416 (apply_iterators): Likewise.
25417 (initialize_iterators): Guard usage of apply_subst_iterator with
25418 #ifdef GENERATOR_FILE.
25419 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25420 (md_reader::read_mapping): Likewise.
25421 (add_define_attr_for_define_subst): Likewise.
25422 (add_define_subst_attr): Likewise.
25423 (read_subst_mapping): Likewise.
25424 (check_code_iterator): Likewise.
25425 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25427 (one_time_initialization): New function.
25428 (rtx_reader::read_until): New method.
25429 (read_flags): New function.
25430 (parse_reg_note_name): New function.
25431 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25432 Handle reuse_rtx ids.
25433 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25434 Add parsing support for RTL dumps, mirroring the special-cases in
25435 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25436 values, and calling handle_any_trailing_information.
25437 (rtx_reader::read_rtx_operand): Convert return type from void
25438 to rtx, returning return_rtx. Handle case 'e'. Call
25439 finalize_string on XSTR and XTMPL fields.
25440 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25441 "(nil)" values were omitted. Call the postprocess vfunc on the
25443 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25444 class ctor. Initialize m_in_call_function_usage. Call
25445 one_time_initialization.
25446 * rtl-tests.c (selftest::test_uncond_jump): Call
25447 set_new_first_and_last_insn.
25448 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25449 * selftest-rtl.c: New file.
25450 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25451 (selftest::get_insn_by_uid): New decl.
25452 * selftest-run-tests.c (selftest::run_tests): Call
25453 read_rtl_function_c_tests.
25454 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25455 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25458 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25460 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25461 operands in a special way. Assert that pos+len <= mode precision.
25463 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25465 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25466 3 argument Alias with unlimited for the negative form.
25467 (fno-vect-cost-model): Removed.
25469 2017-01-05 Martin Liska <mliska@suse.cz>
25471 * hsa-gen.c (gen_hsa_divmod): New function.
25472 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25474 2017-01-05 Martin Liska <mliska@suse.cz>
25477 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25480 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25482 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25483 small constant length operands.
25485 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25487 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25488 between loop iterations.
25490 2017-01-05 Martin Liska <mliska@suse.cz>
25493 * gimplify.c (gimplify_decl_expr): Compare to
25494 asan_poisoned_variables instread of checking flags.
25495 (gimplify_target_expr): Likewise.
25496 (gimplify_expr): Likewise.
25497 (gimplify_function_tree): Conditionally initialize
25498 asan_poisoned_variables.
25500 2017-01-04 Jeff Law <law@redhat.com>
25502 PR tree-optimizatin/78812
25503 * rtl.h (contains_mem_rtx_p): Prototype.
25504 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25505 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25506 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25507 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25508 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25510 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25512 * input.c (assert_char_at_range): Default-initialize actual_range.
25514 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25516 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25517 to match the caller.
25519 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25521 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25522 insns after final jump in test to emit dummy move.
25524 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25526 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25527 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25529 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25531 * multiple_target.c (create_dispatcher_calls): Init e_next.
25532 * tree-ssa-loop-split.c (split_loop): Init border.
25533 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25536 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25541 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25542 (altivec_register_operand): Do not return true if the operand
25543 contains a SUBREG mixing SImode and SFmode.
25544 (vsx_register_operand): Likewise.
25545 (vsx_reg_sfsubreg_ok): New predicate.
25546 (vfloat_operand): Do not return true if the operand contains a
25547 SUBREG mixing SImode and SFmode.
25548 (vint_operand): Likewise.
25549 (vlogical_operand): Likewise.
25550 (gpc_reg_operand): Likewise.
25551 (int_reg_operand): Likewise.
25552 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25553 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25554 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25556 (rs6000_emit_move_si_sf_subreg): New helper function.
25557 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25558 fixup SUBREGs involving SImode and SFmode.
25559 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25560 numbers for the new peephole2 optimization.
25561 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25562 the GLIBC math library that do AND/IOR/XOR operations on single
25563 precision floating point.
25564 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25565 target macros to say whether we need to avoid SUBREGs mixing
25567 (TARGET_ALLOW_SF_SUBREG): Likewise.
25568 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25569 (UNSPEC_SI_FROM_SF): Likewise.
25570 (iorxor): Change spacing.
25571 (and_ior_xor): New iterator for AND, IOR, and XOR.
25572 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25573 (movdi_from_sf_zero_ext): Likewise.
25574 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25575 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25576 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25577 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25578 (fms<mode>4): Likewise.
25579 (fnma<mode>4): Likewise.
25580 (fnms<mode>4): Likewise.
25581 (nfma<mode>4): Likewise.
25582 (nfms<mode>4): Likewise.
25584 2017-01-04 Marek Polacek <polacek@redhat.com>
25587 * doc/invoke.texi: Document -Wpointer-compare.
25589 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25591 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25594 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25595 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25598 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25601 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25602 documentation of the powerpc_popcntb_ok attribute.
25603 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25604 code to issue warning messages if a requested CPU configuration is
25605 not supported by the binary (assembler and loader) toolchain.
25606 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25607 made to define a built-in function that has been disabled.
25608 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25609 made to define a built-in function that has been disabled.
25610 (altivec_init_builtins): Add comment explaining why definition
25611 of the DST built-in functions is not preceded by an assertion
25612 check. Add assertions to prevent ICE if attempts are made to
25613 define an altivec predicate or an abs* built-in function that has
25615 (htm_init_builtins): Add comment explaining why definition of the
25616 htm built-in functions is not preceded by an assertion check.
25618 2017-01-04 Jeff Law <law@redhat.com>
25620 PR tree-optimizatin/67955
25621 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25622 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25623 the points-to solution does not include pt_null. Use DECL_PT_UID
25626 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25628 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25629 Use gen_int_mode instead of gen_lopwart for const_int operands.
25631 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25633 PR tree-optimization/71563
25634 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25635 out of range value - has low bits known to be zero.
25637 2017-01-04 Alan Modra <amodra@gmail.com>
25639 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25640 * configure: Regenerate.
25641 * config.in: Regenerate.
25643 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25646 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25647 a substring of the message, but strcmp with the whole message. Ifdef
25648 ENABLE_NLS, translate the message first using dgettext.
25650 2017-01-03 Jeff Law <law@redhat.com>
25652 PR tree-optimizatin/78856
25653 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25654 (mark_threaded_blocks): Remove code to truncate thread paths that
25655 cross multiple loop headers. Instead invalidate the cached loop
25656 iteration information and handle case of a thread path walking
25657 into an irreducible region.
25659 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25662 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25663 assertions. Add support for doing the signbit if the IEEE 128-bit
25664 floating point value is in a GPR.
25665 * config/rs6000/rs6000.md (Fsignbit): Delete.
25666 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25667 Update the length attribute if the value is in a GPR.
25668 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25669 the sign or zero extension instruction, since the value is always 0/1.
25670 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25673 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25674 extracting SImode to a GPR register so that we can generate a
25675 store, limit the vector to be in a traditional Altivec register
25676 for the vextuwrx instruction.
25678 2017-01-03 Ian Lance Taylor <iant@google.com>
25680 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25682 2017-01-03 Martin Sebor <msebor@redhat.com>
25684 PR tree-optimization/78696
25685 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25686 precision. Use MPFR for %f for greater fidelity. Correct handling
25688 (pass_sprintf_length::compute_format_length): Set width and precision
25689 specified by asrerisk to void_node for vararg functions.
25690 (try_substitute_return_value): Adjust dump output.
25692 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25694 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25696 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25698 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25699 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25700 -mlra/-mno-lra was passed to the compiler.
25702 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25704 PR rtl-optimization/65618
25705 * emit-rtl.c (try_split): Move initialization of "before" and
25706 "after" to just before the call to emit_insn_after_setloc.
25708 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25710 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25712 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25714 * dwarf2out.c (gen_enumeration_type_die): When
25715 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25717 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25719 PR tree-optimization/78965
25720 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25721 Change first argument from const call_info & to call_info &. For %n
25722 set info.nowrite to false.
25724 PR middle-end/78901
25725 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25726 possibly throwing calls.
25728 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25729 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25730 and fns handling, rather than in a separate case SSA_NAME.
25732 2017-01-02 Jeff Law <law@redhat.com>
25734 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25735 fixes for first_period and second_period variables.
25737 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25740 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25741 (*insvqi_1): New insn pattern.
25742 (*insvqi_1_mem_rex64): Ditto.
25743 (*insvqi_2): Ditto.
25744 (*insvqi_3): Rename from *insvqi.
25746 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25748 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25750 * doc/cfg.texi (Edges): Remove reference to Java.
25751 (Maintaining the CFG): Ditto.
25753 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25755 PR middle-end/77674
25756 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25757 transparent aliases.
25759 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25761 PR middle-end/77484
25762 * predict.def (PRED_CALL): Update hitrate.
25763 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25764 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25765 into direct/indirect/polymorphic variants.
25767 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25769 Update copyright years.
25771 * gcc.c (process_command): Update copyright notice dates.
25772 * gcov-dump.c (print_version): Ditto.
25773 * gcov.c (print_version): Ditto.
25774 * gcov-tool.c (print_version): Ditto.
25775 * gengtype.c (create_file): Ditto.
25776 * doc/cpp.texi: Bump @copying's copyright year.
25777 * doc/cppinternals.texi: Ditto.
25778 * doc/gcc.texi: Ditto.
25779 * doc/gccint.texi: Ditto.
25780 * doc/gcov.texi: Ditto.
25781 * doc/install.texi: Ditto.
25782 * doc/invoke.texi: Ditto.
25784 Copyright (C) 2017 Free Software Foundation, Inc.
25786 Copying and distribution of this file, with or without modification,
25787 are permitted in any medium without royalty provided the copyright
25788 notice and this notice are preserved.