1 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
6 * config/arm/arm.c (output_move_neon): Likewise.
7 (arm_attr_length_move_neon): Likewise.
8 (neon_split_vcombine): Likewise.
9 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
10 (c6x_mark_reg_written): Likewise.
11 (c6x_dwarf_register_span): Likewise.
12 * config/i386/i386.c (ix86_save_reg): Likewise.
13 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
14 (rws_access_reg): Likewise.
15 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16 * mode-switching.c (create_pre_exit): Likewise.
17 * ree.c (combine_reaching_defs): Likewise.
18 (add_removable_extension): Likewise.
19 * regcprop.c (find_oldest_value_reg): Likewise.
20 (copyprop_hardreg_forward_1): Likewise.
21 * reload.c (reload_inner_reg_of_subreg): Likewise.
22 (push_reload): Likewise.
23 (combine_reloads): Likewise.
24 (find_dummy_reload): Likewise.
25 (reload_adjust_reg_for_mode): Likewise.
26 * reload1.c (find_reload_regs): Likewise.
27 (forget_old_reloads_1): Likewise.
28 (reload_reg_free_for_value_p): Likewise.
29 (reload_adjust_reg_for_temp): Likewise.
30 (emit_reload_insns): Likewise.
31 (delete_output_reload): Likewise.
32 * sel-sched.c (choose_best_reg_1): Likewise.
33 (choose_best_pseudo_reg): Likewise.
35 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
36 Alan Hayward <alan.hayward@arm.com>
37 David Sherwood <david.sherwood@arm.com>
39 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
40 * target.def (slow_unaligned_access): New hook.
41 * targhooks.h (default_slow_unaligned_access): Declare.
42 * targhooks.c (default_slow_unaligned_access): New function.
43 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
44 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
45 * doc/tm.texi: Regenerate.
46 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
47 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
48 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
50 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
51 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
53 (rs6000_slow_unaligned_access): New function.
54 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
55 (expand_block_compare): Likewise.
56 (expand_strn_compare): Likewise.
57 (rs6000_rtx_costs): Likewise.
58 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
59 (riscv_slow_unaligned_access): Likewise.
60 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
61 (riscv_slow_unaligned_access_p): ...this and make static.
62 (riscv_option_override): Update accordingly.
63 (riscv_slow_unaligned_access): New function.
64 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
65 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
66 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
67 (rs6000_slow_unaligned_access): New function.
68 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
69 (rs6000_rtx_costs): Likewise.
70 * config/rs6000/rs6000-string.c (expand_block_compare)
71 (expand_strn_compare): Use targetm.slow_unaligned_access instead
72 of SLOW_UNALIGNED_ACCESS.
73 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
74 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
75 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
76 of SLOW_UNALIGNED_ACCESS.
77 * expmed.c (simple_mem_bitfield_p): Likewise.
78 * expr.c (alignment_for_piecewise_move): Likewise.
79 (emit_group_load_1): Likewise.
80 (emit_group_store): Likewise.
81 (copy_blkmode_from_reg): Likewise.
82 (emit_push_insn): Likewise.
83 (expand_assignment): Likewise.
84 (store_field): Likewise.
85 (expand_expr_real_1): Likewise.
86 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
87 * lra-constraints.c (simplify_operand_subreg): Likewise.
88 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
89 * gimple-ssa-store-merging.c: Likewise in block comment at start
91 * tree-ssa-strlen.c: Include target.h.
92 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
93 of SLOW_UNALIGNED_ACCESS.
94 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
96 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
98 PR rtl-optimization/82185
99 * expmed.c (emit_store_flag_int): Only test tem if it has been
102 2017-09-12 Richard Biener <rguenther@suse.de>
105 * match.pd ((FTYPE) N CMP CST): Fix typo.
107 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
109 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
111 (mips_near_type_p): Add 'short_call' attribute as a synonym
113 * doc/extend.texi (short_call): Document new function attribute.
115 2017-09-12 Jakub Jelinek <jakub@redhat.com>
118 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
119 assertion check that in the condition.
120 (get_atomic_generic_size): Likewise. Before testing if parameter
121 has pointer type, if it has array type, call for C++
122 default_conversion to perform array-to-pointer conversion.
124 2017-09-12 Richard Biener <rguenther@suse.de>
126 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
127 for operations we cannot scalarize.
129 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
131 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
132 vectors heap vectors. Clean up comments.
133 Make visited_bbs a reference.
134 (profitable_jump_thread_path): Make GC
135 vectors heap vectors. Clean up comments.
137 (convert_and_register_jump_thread_path): Make GC vectors heap
139 (check_subpath_and_update_thread_path): Same. Clean up comments.
140 Make visited_bbs a reference.
141 (handle_phi): Abstract common code to to
142 register_jump_thread_path_if_profitable.
143 Rename VAR_BB to DEF_BB.
145 Make GC vectors heap vectors.
146 Make visited_bbs a reference.
147 (handle_assignment): Same.
148 (register_jump_thread_path_if_profitable): New.
149 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
151 Make GC vectors heap vectors. Clean up comments.
152 Make visited_bbs a reference.
153 (find_jump_threads_backwards): Make visited_bbs live in the stack.
154 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
157 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
160 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
161 words of E_DImode object are reachable by xtensa_uimm8x4 access.
163 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
165 Revert r251800 and r251799.
167 2017-09-11 Martin Jambor <mjambor@suse.cz>
170 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
171 arguments in advance.
172 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
173 use it to find predecessor edges.
174 (naive_outof_ssa): Collect vector of predecessors.
176 2017-09-08 Jason Merrill <jason@redhat.com>
178 PR c++/70029 - ICE with ref-qualifier and -flto
179 * langhooks.h (struct lang_hooks_for_types): Add
180 copy_lang_qualifiers.
181 * attribs.c (build_type_attribute_qual_variant): Use it.
182 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
184 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
185 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
187 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
190 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
191 (*mulsi3_sp64): New instruction.
192 (mulsi3): New expander.
194 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
196 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
198 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
200 * sancov.c: Include memmodel.h.
202 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
205 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
208 2017-09-07 Carl Love <cel@us.ibm.com>
210 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
211 the sldi instruction.
213 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
215 * sancov.c: Include tm_p.h.
217 2017-09-07 Jakub Jelinek <jakub@redhat.com>
220 * output.h (switch_to_other_text_partition): New declaration.
221 * varasm.c (switch_to_other_text_partition): New function.
222 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
223 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
224 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
225 to the other text partition before emitting LCL label and switch back
226 after emitting the word after it.
228 2017-09-07 Richard Biener <rguenther@suse.de>
230 * passes.def (pass_split_crit_edges): Remove instance before PRE.
231 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
232 critical edges here, after loop init.
233 (pass_data_pre): Remove PROP_no_crit_edges flags.
234 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
235 for valueization of call args to avoid leaking VN_TOP.
236 (visit_use): Assert we do not visit default defs.
237 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
238 Use error_mark_node to more easily detect leaking VN_TOP.
239 All default-defs are varying, not VN_TOP. Mark them visited.
240 (run_scc_vn): Make code match comment.
242 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
244 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
245 OPTION_MASK_FLOAT128_KEYWORD.
246 (POWERPC_MASKS): Likewise.
247 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
248 support for the -mfloat128-type option, and make -mfloat128
249 default on PowerPC Linux systems. Define or undefine
250 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
251 Define __float128 to be __ieee128 if IEEE 128-bit support is
252 enabled, or undefine it.
253 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
254 Delete defining __FLOAT128_TYPE__.
255 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
256 -mfloat128-type option and make -mfloat128 default on PowerPC
258 (TARGET_FLOAT128_TYPE): Likewise.
259 (-mfloat128-type): Likewise.
260 * config/rs6000/rs6000.c (rs6000_option_override_internal):
261 Delete the -mfloat128-type option and make -mfloat128 default on
262 PowerPC Linux systems. Always use __ieee128 to be the keyword for
263 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
264 128-bit floating point is enabled. Change tests from using
265 -mfloat128-type to -mfloat128.
266 (rs6000_mangle_type): Use the correct mangling for the __float128
267 type even if normal long double is restricted to 64-bits.
268 (floatn_mode): Enable the _Float128 type by default on VSX Linux
270 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
271 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
272 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
274 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
275 documentation for -mfloat128.
277 2017-09-06 Olivier Hainque <hainque@adacore.com>
279 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
281 2017-09-06 Wish Wu <wishwu007@gmail.com>
282 Jakub Jelinek <jakub@redhat.com>
284 * asan.c (initialize_sanitizer_builtins): Add
285 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
286 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
287 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
288 BT_FN_VOID_UINT64_PTR variables.
289 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
290 (BT_FN_VOID_UINT16_UINT16): Likewise.
291 (BT_FN_VOID_UINT32_UINT32): Likewise.
292 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
293 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
294 (BT_FN_VOID_UINT64_PTR): Likewise.
295 * common.opt (flag_sanitize_coverage): New variable.
296 (fsanitize-coverage=trace-pc): Remove.
297 (fsanitize-coverage=): Add.
298 * flag-types.h (enum sanitize_coverage_code): New enum.
299 * fold-const.c (fold_range_test): Disable non-short-circuit
300 optimization if flag_sanitize_coverage.
301 (fold_truth_andor): Likewise.
302 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
303 * opts.c (COVERAGE_SANITIZER_OPT): Define.
304 (coverage_sanitizer_opts): New array.
305 (get_closest_sanitizer_option): Add OPTS argument, handle also
306 OPT_fsanitize_coverage_.
307 (parse_sanitizer_options): Adjusted to also handle
308 OPT_fsanitize_coverage_.
309 (common_handle_option): Add OPT_fsanitize_coverage_.
310 * sancov.c (instrument_comparison, instrument_switch): New function.
311 (sancov_pass): Add trace-cmp support.
312 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
313 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
314 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
315 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
316 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
317 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
318 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
319 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
320 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
321 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
323 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
325 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
326 error. Only quit immediately if parsing is complete.
327 (BEGIN): Initialize fatal_err and parse_done.
328 (begin fpu, end fpu): Check number of arguments.
329 (begin arch, end arch): Likewise.
330 (begin cpu, end cpu): Likewise.
331 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
332 (optalias): Likewise.
334 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
336 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
337 * config/arm/arm-isa.h: Delete. Move definitions to ...
338 * arm-cpus.in: ... here. Use new feature and fgroup values.
339 * config/arm/arm.c (arm_option_override): Use lower case for feature
341 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
342 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
343 * config/arm/parsecpu.awk (END): Add new command 'isa'.
345 (print_isa_bits_for): New function.
346 (gen_isa): New function.
347 (gen_comm_data): Use print_isa_bits_for.
348 (define feature): New keyword.
349 (define fgroup): New keyword.
350 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
351 (arm-isa.h): Add rule to generate file.
352 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
353 case for feature bit names.
355 2017-09-06 Richard Biener <rguenther@suse.de>
357 * tree-ssa-pre.c (NECESSARY): Remove.
358 (create_expression_by_pieces): Do not touch pass-local flags.
359 (insert_into_preds_of_block): Likewise.
360 (do_pre_regular_insertion): Likewise.
361 (eliminate_insert): Likewise.
362 (eliminate_dom_walker::before_dom_children): Likewise.
363 (fini_eliminate): Do not look at inserted_exprs.
364 (mark_operand_necessary): Remove.
365 (remove_dead_inserted_code): Replace with simple work-list
366 algorithm based on inserted_exprs and SSA uses.
367 (pass_pre::execute): Re-order fini_eliminate and
368 remove_dead_inserted_code.
370 2017-09-06 Olivier Hainque <hainque@adacore.com>
372 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
373 for VxWorks 7. Adjust surrounding comments.
375 2017-09-06 Richard Biener <rguenther@suse.de>
377 * gimple-ssa-strength-reduction.c
378 (find_candidates_dom_walker::before_dom_children): Also allow
381 2017-09-06 Richard Biener <rguenther@suse.de>
383 PR tree-optimization/82108
384 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
385 for gap in the non-permutation SLP case.
387 2017-09-06 Martin Jambor <mjambor@suse.cz>
389 PR tree-optimization/82078
390 * tree-sra.c (sort_and_splice_var_accesses): Move call to
391 add_access_to_work_queue...
392 (build_accesses_from_assign): ...here.
393 (propagate_all_subaccesses): Make sure racc is the group
394 representative, if there is one.
396 2017-09-06 Jakub Jelinek <jakub@redhat.com>
399 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
402 2017-09-06 Richard Biener <rguenther@suse.de>
404 * gimple-ssa-strength-reduction.c
405 (find_candidates_dom_walker::before_doom_children): Use a
406 type and not a mode check.
408 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
411 * config/arm/predicates.md (arm_general_adddi_operand): Create new
413 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
415 2017-09-05 Jeff Law <law@redhat.com>
417 PR tree-optimization/64910
418 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
419 cases where we have 3 or more operands.
421 2017-09-05 Jakub Jelinek <jakub@redhat.com>
424 * omp-low.c (lower_omp_for): Recompute tree invariant if
425 gimple_omp_for_initial/final is ADDR_EXPR.
428 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
429 into gimple val before gimplification fo the COND_EXPR.
431 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
433 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
434 REGION_COPY argument.
435 (thread_through_all_blocks): Remove unused argument to
436 duplicate_thread_path.
438 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
439 Alan Hayward <alan.hayward@arm.com>
440 David Sherwood <david.sherwood@arm.com>
442 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
443 Take a scalar_mode rather than a machine_mode.
444 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
445 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
446 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
447 (aarch64_gen_adjusted_ldpstp): Likewise.
448 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
450 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
451 Alan Hayward <alan.hayward@arm.com>
452 David Sherwood <david.sherwood@arm.com>
454 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
455 Take a scalar_int_mode instead of a machine_mode.
456 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
457 (aarch64_output_scalar_simd_mov_immediate): Likewise.
458 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
459 (aarch64_simd_attr_length_rglist): Delete.
460 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
461 a scalar_int_mode instead of a machine_mode.
462 (aarch64_add_offset): Likewise.
463 (aarch64_internal_mov_immediate): Likewise
464 (aarch64_add_constant_internal): Likewise.
465 (aarch64_add_constant): Likewise.
466 (aarch64_movw_imm): Likewise.
467 (aarch64_rtx_arith_op_extract_p): Likewise.
468 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
469 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
470 Remove assert that the mode isn't a vector.
471 (aarch64_output_scalar_simd_mov_immediate): Likewise.
472 (aarch64_expand_mov_immediate): Update calls after above changes.
473 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
474 (aarch64_and_bitmask_imm): Check for scalar integer modes.
475 (aarch64_move_imm): Likewise.
476 (aarch64_can_const_movi_rtx_p): Likewise.
477 (aarch64_strip_extend): Likewise.
478 (aarch64_extr_rtx_p): Likewise.
479 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
480 a CONST_INT when the mode parameter is VOIDmode.
481 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
483 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
485 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
486 * stor-layout.c (bitwise_mode_for_mode): Likewise.
487 (bitwise_type_for_mode): Update accordingly.
489 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
491 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
492 * stor-layout.c (mode_for_size_tree): Likewise.
493 (mode_for_array): Update accordingly.
494 (layout_decl): Likewise.
495 (compute_record_mode): Likewise. Only set the mode once.
497 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
499 * target.def (get_mask_mode): Change return type to opt_mode.
501 * doc/tm.texi: Regenerate.
502 * targhooks.h (default_get_mask_mode): Return an opt_mode.
503 * targhooks.c (default_get_mask_mode): Likewise.
504 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
505 * optabs-query.c (can_vec_mask_load_store_p): Update use of
506 targetm.get_mask_mode.
507 * tree.c (build_truth_vector_type): Likewise.
509 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
511 * machmode.h (mode_for_vector): Return an opt_mode.
512 * stor-layout.c (mode_for_vector): Likewise.
513 (mode_for_int_vector): Update accordingly.
514 (layout_type): Likewise.
515 * config/i386/i386.c (emit_memmov): Likewise.
516 (ix86_expand_set_or_movmem): Likewise.
517 (ix86_expand_vector_init): Likewise.
518 (ix86_get_mask_mode): Likewise.
519 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
521 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
522 * expmed.c (extract_bit_field_1): Likewise.
523 * expr.c (expand_expr_real_2): Likewise.
524 * optabs-query.c (can_vec_perm_p): Likewise.
525 (can_vec_mask_load_store_p): Likewise.
526 * optabs.c (expand_vec_perm): Likewise.
527 * targhooks.c (default_get_mask_mode): Likewise.
528 * tree-vect-stmts.c (vectorizable_store): Likewise.
529 (vectorizable_load): Likewise.
530 (get_vectype_for_scalar_type_and_size): Likewise.
532 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
534 * machmode.h (mode_for_int_vector): New function.
535 * stor-layout.c (mode_for_int_vector): Likewise.
536 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
537 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
538 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
539 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
540 (s390_expand_vcond): Likewise.
542 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
544 * machmode.h (opt_machine_mode): New type.
545 (opt_mode<T>): Allow construction from anything that can be
547 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
548 (mode_for_size): Return an opt_machine_mode.
549 * stor-layout.c (mode_for_size): Likewise.
550 (mode_for_size_tree): Update call accordingly.
551 (bitwise_mode_for_mode): Likewise.
552 (make_fract_type): Likewise.
553 (make_accum_type): Likewise.
554 * caller-save.c (replace_reg_with_saved_mem): Update call
556 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
557 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
558 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
559 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
560 * expmed.c (extract_bit_field_1): Likewise.
561 * reload.c (get_secondary_mem): Likewise.
562 * varasm.c (assemble_integer): Likewise.
563 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
566 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
568 * machmode.h (decimal_float_mode_for_size): New function.
569 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
570 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
571 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
572 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
573 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
574 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
576 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
578 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
579 (get_builtin_sync_mode): Likewise.
580 (expand_ifn_atomic_compare_exchange): Likewise.
581 (expand_builtin_atomic_clear): Likewise.
582 (expand_builtin_atomic_test_and_set): Likewise.
583 (fold_builtin_atomic_always_lock_free): Likewise.
584 * calls.c (compute_argument_addresses): Likewise.
585 (emit_library_call_value_1): Likewise.
586 (store_one_arg): Likewise.
587 * combine.c (combine_instructions): Likewise.
588 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
589 * config/arm/arm.c (arm_function_value): Likewise.
590 (aapcs_allocate_return_reg): Likewise.
591 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
592 * config/i386/i386.c (construct_container): Likewise.
593 (ix86_gimplify_va_arg): Likewise.
594 (ix86_expand_sse_cmp): Likewise.
595 (emit_memmov): Likewise.
596 (emit_memset): Likewise.
597 (expand_small_movmem_or_setmem): Likewise.
598 (ix86_expand_pextr): Likewise.
599 (ix86_expand_pinsr): Likewise.
600 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
601 * config/microblaze/microblaze.c (microblaze_block_move_straight):
603 * config/mips/mips.c (mips_function_value_1) Likewise.
604 (mips_block_move_straight): Likewise.
605 (mips_expand_ins_as_unaligned_store): Likewise.
606 * config/powerpcspe/powerpcspe.c
607 (rs6000_darwin64_record_arg_advance_flush): Likewise.
608 (rs6000_darwin64_record_arg_flush): Likewise.
609 * config/rs6000/rs6000.c
610 (rs6000_darwin64_record_arg_advance_flush): Likewise.
611 (rs6000_darwin64_record_arg_flush): Likewise.
612 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
613 (sparc_function_value_1): Likewise.
614 * config/spu/spu.c (adjust_operand): Likewise.
615 (spu_emit_branch_or_set): Likewise.
616 (arith_immediate_p): Likewise.
617 * emit-rtl.c (gen_lowpart_common): Likewise.
618 * expr.c (expand_expr_real_1): Likewise.
619 * function.c (assign_parm_setup_block): Likewise.
620 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
621 * reload1.c (alter_reg): Likewise.
622 * stor-layout.c (mode_for_vector): Likewise.
623 (layout_type): Likewise.
625 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
627 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
628 (spu_convert_move): Likewise.
629 * lower-subreg.c (resolve_simple_move): Likewise.
631 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
634 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
635 define_insn to a define_expand.
636 (altivec_vsum2sws_direct): New define_insn.
637 (altivec_vsumsws): Convert from a define_insn to a define_expand.
639 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
641 * config/arm/arm.c (arm_option_params_internal): Improve setting of
644 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
649 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
650 realign stack if stack alignment needed is less than incoming
653 2017-09-05 Marek Polacek <polacek@redhat.com>
656 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
659 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
661 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
663 2017-09-05 Richard Biener <rguenther@suse.de>
665 PR tree-optimization/82084
666 * fold-const.c (can_native_encode_string_p): Handle wide characters.
668 2017-09-05 Richard Biener <rguenther@suse.de>
670 PR tree-optimization/82102
671 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
673 2017-09-05 Martin Liska <mliska@suse.cz>
675 PR tree-optimization/82032
676 * tree-cfg.c (generate_range_test): New function.
677 * tree-cfg.h (generate_range_test): Declared here.
678 * tree-cfgcleanup.c (convert_single_case_switch): New function.
679 (cleanup_control_expr_graph): Use it.
680 * tree-switch-conversion.c (try_switch_expansion): Remove
682 (emit_case_nodes): Use generate_range_test.
684 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
687 * config/i386/i386.md (*<btsc><mode>_mask): Add
688 TARGET_USE_BT to insn constraint.
689 (*btr<mode>_mask): Ditto.
691 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
693 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
694 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
696 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
699 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
700 TARGET_NEON and TARGET_IWMMXT.
701 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
702 TARGET_NEON and TARGET_IWMMXT.
703 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
705 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
707 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
708 (ix86_rewrite_tls_address): Ditto.
709 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
710 (ix86_rewrite_tls_address_1): Ditto.
711 (ix86_rewrite_tls_address): Ditto.
712 * config/i386/predicates.md (tls_address_pattern): New predicate.
713 * config/i386/i386.md (TLS address splitter): New splitter.
715 2017-09-04 Richard Biener <rguenther@suse.de>
717 PR tree-optimization/82084
718 * fold-const.h (can_native_encode_string_p): Declare.
719 * fold-const.c (can_native_encode_string_p): Factor out from ...
720 (native_encode_string): ... here.
721 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
722 vectorizing stores from constants we later cannot handle.
724 2017-09-04 Marek Polacek <polacek@redhat.com>
727 * doc/invoke.texi: Update -Wtautological-compare documentation.
729 2017-09-04 Jeff Law <law@redhat.com>
731 PR tree-optimization/64910
732 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
733 swap the first and last operand if the last is a constant.
735 2017-09-04 Marek Polacek <polacek@redhat.com>
738 * convert.c (do_narrow): When sanitizing signed integer overflows,
739 bail out for signed types.
740 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
742 2017-09-04 Richard Biener <rguenther@suse.de>
744 PR tree-optimization/82060
745 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
746 Move devirtualization after stmt folding and before EH/AB/noreturn
747 cleanup to get the stmt refs canonicalized. Use a bool instead
748 of gimple_modified_p since that doesn't work for NOPs. Schedule
749 NOPs generated by folding for removal.
751 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
752 Alan Hayward <alan.hayward@arm.com>
753 David Sherwood <david.sherwood@arm.com>
755 * coretypes.h (pad_direction): New enum.
756 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
757 (FUNCTION_ARG_PADDING): Likewise.
758 * target.def (function_arg_padding): New hook.
759 * targhooks.h (default_function_arg_padding): Declare.
760 * targhooks.c (default_function_arg_padding): New function.
761 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
762 (TARGET_FUNCTION_ARG_PADDING): ...this.
763 * doc/tm.texi: Regenerate.
764 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
765 instead of direction.
766 (compute_argument_addresses): Likewise.
767 (load_register_parameters): Likewise.
768 (emit_library_call_value_1): Likewise.
769 (store_one_arg): Use targetm.calls.function_arg_padding instead
770 of FUNCTION_ARG_PADDING.
771 (must_pass_in_stack_var_size_or_pad): Likewise.
772 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
773 (emit_group_store): Likewise.
774 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
775 instead of FUNCTION_ARG_PADDING.
776 (emit_push_insn): Likewise, and propagate enum change throughout
778 * function.h (direction): Delete.
779 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
781 * function.c (assign_parm_find_stack_rtl): Likewise.
782 (assign_parm_setup_block_p): Likewise.
783 (assign_parm_setup_block): Likewise.
784 (gimplify_parameters): Likewise.
785 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
786 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
788 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
789 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
790 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
791 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
792 (aarch64_function_arg_padding): ...this new function.
793 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
794 (TARGET_FUNCTION_ARG_PADDING): Redefine.
795 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
796 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
797 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
798 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
799 (arm_pad_arg_upward): Replace with...
800 (arm_function_arg_padding): ...this new function.
801 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
803 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
804 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
805 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
806 (ia64_hpux_function_arg_padding): Replace with...
807 (ia64_function_arg_padding): ...this new function. Use pad_direction
808 instead of direction. Check for TARGET_HPUX.
809 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
810 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
811 (iq2000_function_arg_padding): New function.
812 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
813 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
814 (mips_function_arg_padding): ...this new function.
815 (mips_pad_reg_upward): Update accordingly.
816 (TARGET_FUNCTION_ARG_PADDING): Redefine.
817 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
818 targetm.calls.function_arg_padding.
819 (FUNCTION_ARG_PADDING): Delete.
820 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
821 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
822 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
823 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
824 (nios2_block_reg_padding): Return pad_direction instead of direction.
825 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
826 instead of direction.
827 (nios2_function_arg_padding): Likewise. Make static.
828 (TARGET_FUNCTION_ARG_PADDING): Redefine.
829 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
830 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
831 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
832 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
833 (pa_function_arg_padding): Make static. Return pad_direction instead
835 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
836 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
837 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
838 instead of direction. Use targetm.calls.function_arg_padding.
839 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
840 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
841 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
842 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
843 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
845 (function_arg_padding): Rename to...
846 (rs6000_function_arg_padding): ...this. Make static. Return
847 pad_direction instead of direction.
848 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
849 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
850 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
851 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
852 instead of direction. Use targetm.calls.function_arg_padding.
853 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
854 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
855 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
856 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
857 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
858 (function_arg_padding): Rename to...
859 (rs6000_function_arg_padding): ...this. Make static. Return
860 pad_direction instead of direction.
861 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
862 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
863 * config/s390/s390.c (s390_function_arg_padding): New function.
864 (TARGET_FUNCTION_ARG_PADDING): Redefine.
865 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
866 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
867 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
868 (function_arg_padding): Rename to...
869 (sparc_function_arg_padding): ...this. Make static. Return
870 pad_direction instead of direction.
871 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
872 * config/spu/spu.c (spu_function_arg_padding): New function.
873 (TARGET_FUNCTION_ARG_PADDING): Redefine.
874 * system.h (FUNCTION_ARG_PADDING): Poison.
876 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
877 Alan Hayward <alan.hayward@arm.com>
878 David Sherwood <david.sherwood@arm.com>
880 * target.def (modes_tieable_p): New hook.
881 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
882 (TARGET_MODES_TIEABLE_P): ...this.
883 * doc/tm.texi.in: Regenerate.
884 * hooks.h (hook_bool_mode_mode_true): Declare.
885 * hooks.c (hook_bool_mode_mode_true): New function.
886 * combine.c (subst): Use targetm.modes_tieable_p instead of
888 * dse.c (find_shift_sequence): Likewise.
889 * expmed.c (extract_low_bits): Likewise.
890 * lower-subreg.c: Include target.h.
891 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
893 * rtlanal.c (rtx_cost): Likewise.
894 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
895 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
896 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
897 (TARGET_MODES_TIEABLE_P): Redefine.
898 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
899 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
900 (TARGET_MODES_TIEABLE_P): Redefine.
901 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
902 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
903 (arc_modes_tieable_p): New function.
904 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
905 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
906 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
907 (arm_modes_tieable_p): Make static.
908 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
909 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
910 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
911 (TARGET_MODES_TIEABLE_P): Redefine.
912 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
913 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
914 (TARGET_MODES_TIEABLE_P): Redefine.
915 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
916 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
917 (cr16_modes_tieable_p): New function.
918 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
919 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
920 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
921 (TRULY_NOOP_TRUNCATION): Update comment.
922 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
923 (TRULY_NOOP_TRUNCATION): Update comment.
924 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
925 (frv_modes_tieable_p): New function.
926 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
927 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
928 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
929 (TARGET_MODES_TIEABLE_P): Redefine.
930 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
931 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
932 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
933 (TARGET_MODES_TIEABLE_P): Redefine.
934 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
935 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
936 (ia64_modes_tieable_p): New function.
937 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
938 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
939 (iq2000_modes_tieable_p): New function.
940 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
941 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
942 (lm32_modes_tieable_p): New function.
943 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
944 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
945 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
946 (TARGET_MODES_TIEABLE_P): Redefine.
947 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
948 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
949 (m32r_modes_tieable_p): New function.
950 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
951 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
952 (m68k_modes_tieable_p): New function.
953 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
954 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
955 (mcore_modes_tieable_p): New function.
956 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
957 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
959 (TARGET_MODES_TIEABLE_P): Redefine.
960 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
961 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
962 * config/mips/mips.c (mips_modes_tieable_p): Make static.
963 (TARGET_MODES_TIEABLE_P): Redefine.
964 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
965 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
966 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
967 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
968 (mn10300_modes_tieable_p): ...this and make static.
969 (TARGET_MODES_TIEABLE_P): Redefine.
970 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
971 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
972 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
973 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
974 (msp430_modes_tieable_p): Make static.
975 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
976 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
977 (TARGET_MODES_TIEABLE_P): Redefine.
978 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
979 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
980 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
981 (TARGET_MODES_TIEABLE_P): Redefine.
982 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
983 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
984 * config/pa/pa.c (pa_modes_tieable_p): Make static.
985 (TARGET_MODES_TIEABLE_P): Redefine.
986 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
987 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
988 (pdp11_modes_tieable_p): New function.
989 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
990 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
991 (rs6000_modes_tieable_p): New function.
992 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
993 * config/powerpcspe/powerpcspe.md: Update comment.
994 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
995 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
996 (TARGET_MODES_TIEABLE_P): Redefine.
997 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
998 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
999 (rl78_modes_tieable_p): New function.
1000 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1001 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1002 (rs6000_modes_tieable_p): New function.
1003 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1004 * config/rs6000/rs6000.md: Update comment.
1005 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1006 * config/rx/rx.c (rx_modes_tieable_p): New function.
1007 (TARGET_MODES_TIEABLE_P): Redefine.
1008 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1009 * config/s390/s390.c (s390_modes_tieable_p): New function.
1010 (TARGET_MODES_TIEABLE_P): Redefine.
1011 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1012 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1013 (sh_modes_tieable_p): New function.
1014 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1015 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1016 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1017 (sparc_modes_tieable_p): Make static.
1018 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1019 * config/spu/spu.c (spu_modes_tieable_p): New function.
1020 (TARGET_MODES_TIEABLE_P): Redefine.
1021 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1022 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1023 (TARGET_MODES_TIEABLE_P): Redefine.
1024 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1025 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1026 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1027 * config/v850/v850.c (v850_modes_tieable_p): New function.
1028 (TARGET_MODES_TIEABLE_P): Redefine.
1029 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1030 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1031 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1032 (visium_modes_tieable_p): New function.
1033 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1034 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1035 (xtensa_modes_tieable_p): New function.
1036 * system.h (MODES_TIEABLE_P): Poison.
1038 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1039 Alan Hayward <alan.hayward@arm.com>
1040 David Sherwood <david.sherwood@arm.com>
1042 * target.def (hard_regno_mode_ok): New hook.
1043 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1044 (TARGET_HARD_REGNO_MODE_OK): ...this.
1045 * doc/tm.texi.in: Regenerate.
1046 * hooks.h (hook_bool_uint_mode_true): Declare.
1047 * hooks.c (hook_bool_uint_mode_true): New function.
1048 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1050 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1051 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1052 instead of HARD_REGNO_MODE_OK.
1053 * caller-save.c: Include target.h.
1054 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1056 * combine.c (can_combine_p): Likewise.
1057 (combinable_i3pat): Likewise.
1058 (can_change_dest_mode): Likewise.
1059 * expr.c (init_expr_target): Likewise.
1060 (convert_move): Likewise.
1061 (convert_modes): Likewise.
1062 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1063 (setup_prohibited_mode_move_regs): Likewise.
1064 * ira.h (target_ira): Likewise.
1065 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1066 * lra-constraints.c (process_alt_operands): Likewise.
1067 (split_reg): Likewise.
1068 * recog.c (peep2_find_free_register): Likewise.
1069 * ree.c (combine_reaching_defs): Likewise.
1070 * regcprop.c (maybe_mode_change): Likewise.
1071 * reginfo.c (init_reg_sets_1): Likewise.
1072 (choose_hard_reg_mode): Likewise.
1073 (simplifiable_subregs): Likewise.
1074 * regrename.c (check_new_reg_p): Likewise.
1075 * reload.c (find_valid_class): Likewise.
1076 (find_valid_class_1): Likewise.
1077 (reload_inner_reg_of_subreg): Likewise.
1078 (push_reload): Likewise.
1079 (combine_reloads): Likewise.
1080 (find_dummy_reload): Likewise.
1081 (find_reloads): Likewise.
1082 * reload1.c (find_reg): Likewise.
1083 (set_reload_reg): Likewise.
1084 (allocate_reload_reg): Likewise.
1085 (choose_reload_regs): Likewise.
1086 (reload_adjust_reg_for_temp): Likewise.
1087 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1088 (simplify_subreg_regno): Likewise.
1089 * sel-sched.c (init_regs_for_mode): Likewise.
1090 * varasm.c (make_decl_rtl): Likewise.
1091 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1092 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1094 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1095 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1096 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1097 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1098 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1099 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1100 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1101 (arc_mode_class): Delete.
1102 (HARD_REGNO_MODE_OK): Delete.
1103 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1104 (arc_hard_regno_mode_ok): Rename old array to...
1105 (arc_hard_regno_mode_ok_modes): ...this.
1106 (arc_conditional_register_usage): Update accordingly.
1107 (arc_mode_class): Make static.
1108 (arc_hard_regno_mode_ok): New function.
1109 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1110 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1111 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1112 (arm_hard_regno_mode_ok): Make static.
1113 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1115 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1116 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1117 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1119 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1120 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1121 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1122 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1123 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1124 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1125 * config/bfin/predicates.md (valid_reg_operand): Use
1126 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1127 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1128 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1129 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1130 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1131 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1132 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1133 (cr16_hard_regno_mode_ok): Make static and return a bool.
1134 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1135 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1136 (cris_hard_regno_mode_ok): New function.
1137 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1138 (epiphany_mode_class): Delete.
1139 (HARD_REGNO_MODE_OK): Delete.
1140 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1141 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1142 (hard_regno_mode_ok): Rename to...
1143 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1144 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1145 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1147 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1148 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1149 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1150 (frv_hard_regno_mode_ok): Make static and return a bool.
1151 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1153 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1154 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1155 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1156 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1158 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1159 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1160 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1161 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1163 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1164 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1165 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1166 (ia64_hard_regno_mode_ok): New function.
1167 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1168 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1169 (iq2000_hard_regno_mode_ok): New function.
1170 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1171 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1172 (lm32_hard_regno_mode_ok): New function.
1173 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1174 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1175 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1176 instead of HARD_REGNO_MODE_OK.
1177 (m32c_hard_regno_ok): Rename to...
1178 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1179 (m32c_cannot_change_mode_class): Update accordingly.
1180 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1181 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1182 (m32r_mode_class): Delete.
1183 (HARD_REGNO_MODE_OK): Delete.
1184 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1185 (m32r_hard_regno_mode_ok): Rename to...
1186 (m32r_hard_regno_modes): ...this.
1187 (m32r_mode_class): Make static.
1188 (m32r_hard_regno_mode_ok): New function.
1189 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1190 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1191 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1192 (m68k_hard_regno_mode_ok): Make static.
1193 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1194 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1195 (mcore_hard_regno_mode_ok): New function.
1196 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1197 (HARD_REGNO_MODE_OK): Delete.
1198 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1200 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1201 (microblaze_hard_regno_mode_ok): New function.
1202 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1203 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1204 (mips_hard_regno_mode_ok): Delete.
1205 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1206 (mips_hard_regno_mode_ok_p): ...this and make static.
1207 (mips_hard_regno_mode_ok_p): Rename to...
1208 (mips_hard_regno_mode_ok_uncached): ...this.
1209 (mips_hard_regno_mode_ok): New function.
1210 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1211 of HARD_REGNO_MODE_OK.
1212 (mips_option_override): Update after above name changes.
1213 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1214 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1215 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1216 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1217 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1218 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1219 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1220 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1221 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1222 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1223 (msp430_hard_regno_mode_ok): Make static and return a bool.
1224 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1225 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1226 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1228 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1229 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1230 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1231 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1232 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1233 (PA_HARD_REGNO_MODE_OK): ...this
1234 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1235 (PA_HARD_REGNO_MODE_OK): ...this.
1236 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1237 (pa_hard_regno_mode_ok): New function.
1238 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1239 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1240 (pdp11_hard_regno_mode_ok): New function.
1241 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1242 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1244 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1246 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1247 (rs6000_hard_regno_mode_ok): Rename to...
1248 (rs6000_hard_regno_mode_ok_uncached): ...this.
1249 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1250 (rs6000_hard_regno_mode_ok): New function.
1251 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1252 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1253 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1254 (riscv_hard_regno_mode_ok): ...this and make static.
1255 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1256 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1257 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1258 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1259 (rl78_hard_regno_mode_ok): Make static and return bool.
1260 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1261 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1263 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1264 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1265 (rs6000_hard_regno_mode_ok): Rename to...
1266 (rs6000_hard_regno_mode_ok_uncached): ...this.
1267 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1268 (rs6000_hard_regno_mode_ok): New function.
1269 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1270 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1271 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1272 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1273 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1274 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1275 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1276 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1277 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1278 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1279 (sh_hard_regno_mode_ok): Make static.
1280 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1281 instead of HARD_REGNO_MODE_OK.
1282 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1283 (sparc_mode_class): Delete.
1284 (HARD_REGNO_MODE_OK): Delete.
1285 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1286 (hard_regno_mode_classes): Make static.
1287 (sparc_mode_class): Likewise.
1288 (sparc_hard_regno_mode_ok): New function.
1289 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1290 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1291 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1293 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1294 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1295 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1296 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1297 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1298 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1299 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1300 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1301 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1302 (visium_hard_regno_mode_ok): New function.
1303 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1304 instead of HARD_REGNO_MODE_OK.
1305 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1306 (HARD_REGNO_MODE_OK): Delete.
1307 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1308 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1309 (xtensa_option_override): Update accordingly.
1310 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1311 (xtensa_hard_regno_mode_ok): New function.
1312 * system.h (HARD_REGNO_MODE_OK): Poison.
1314 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1315 Alan Hayward <alan.hayward@arm.com>
1316 David Sherwood <david.sherwood@arm.com>
1318 * target.def (hard_regno_call_part_clobbered): New hook.
1319 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1320 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1321 * doc/tm.texi: Regenerate.
1322 * hooks.h (hook_bool_uint_mode_false): Declare.
1323 * hooks.c (hook_bool_uint_mode_false): New function.
1324 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1325 * cselib.c (cselib_process_insn): Use
1326 targetm.hard_regno_call_part_clobbered instead of
1327 HARD_REGNO_CALL_PART_CLOBBERED.
1328 * ira-conflicts.c (ira_build_conflicts): Likewise.
1329 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1330 * lra-constraints.c (need_for_call_save_p): Likewise.
1331 * lra-lives.c: Include target.h.
1332 (check_pseudos_live_through_calls): Use
1333 targetm.hard_regno_call_part_clobbered instead of
1334 HARD_REGNO_CALL_PART_CLOBBERED.
1335 * regcprop.c: Include target.h.
1336 (copyprop_hardreg_forward_1): Use
1337 targetm.hard_regno_call_part_clobbered instead of
1338 HARD_REGNO_CALL_PART_CLOBBERED.
1339 * reginfo.c (choose_hard_reg_mode): Likewise.
1340 * regrename.c (check_new_reg_p): Likewise.
1341 * reload.c (find_equiv_reg): Likewise.
1342 * reload1.c (emit_reload_insns): Likewise.
1343 * sched-deps.c (deps_analyze_insn): Likewise.
1344 * sel-sched.c (init_regs_for_mode): Likewise.
1345 (mark_unavailable_hard_regs): Likewise.
1346 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1347 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1348 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1350 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1351 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1352 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1354 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1356 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1357 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1358 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1360 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1361 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1362 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1364 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1365 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1367 * config/powerpcspe/powerpcspe.c
1368 (rs6000_hard_regno_call_part_clobbered): New function.
1369 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1370 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1371 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1373 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1374 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1375 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1377 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1378 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1379 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1381 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1382 Alan Hayward <alan.hayward@arm.com>
1383 David Sherwood <david.sherwood@arm.com>
1385 * rtl.h (subreg_memory_offset): Declare.
1386 * emit-rtl.c (subreg_memory_offset): New function.
1387 * expmed.c (store_bit_field_1): Use it.
1388 * expr.c (undefined_operand_subword_p): Likewise.
1389 * simplify-rtx.c (simplify_subreg): Likewise.
1391 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1393 PR rtl-optimization/57448
1396 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1397 using atomic_load pattern.
1398 (expand_atomic_store): Likewise.
1400 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1403 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1404 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1405 handling. Use replace_call_with_value with NULL instead of
1406 gsi_replace, unlink_stmt_vdef and release_defs.
1408 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1411 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1413 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1416 * rtl.h (emit_library_call_value_1): Declare.
1417 (emit_library_call): Replace declaration with a series of overloads.
1418 Remove the parameter count argument.
1419 (emit_library_call_value): Likewise.
1420 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1421 with an "rtx_mode_t *".
1422 (emit_library_call_value): Delete.
1423 (emit_library_call): Likewise.
1424 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1425 (asan_emit_allocas_unpoison): Likewise.
1426 * builtins.c (expand_builtin_powi): Likewise.
1427 (expand_asan_emit_allocas_unpoison): Likewise.
1428 * cfgexpand.c (expand_main_function): Likewise.
1429 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1430 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1431 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1432 * config/arm/arm.c (arm_trampoline_init): Likewise.
1433 (arm_call_tls_get_addr): Likewise.
1434 (arm_expand_divmod_libfunc): Likewise.
1435 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1436 (smulsi3_highpart): Likewise.
1437 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1438 (c6x_expand_compare): Likewise.
1439 (c6x_expand_movmem): Likewise.
1440 * config/frv/frv.c (frv_trampoline_init): Likewise.
1441 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1442 (ix86_expand_divmod_libfunc): Likewise.
1443 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1444 (ia64_expand_compare): Likewise.
1445 (ia64_profile_hook): Likewise.
1446 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1447 (nonlocal_goto): Likewise.
1448 (restore_stack_nonlocal): Likewise.
1449 * config/m32r/m32r.c (block_move_call): Likewise.
1450 (m32r_trampoline_init): Likewise.
1451 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1452 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1453 (m68k_call_m68k_read_tp): Likewise.
1454 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1455 (microblaze_expand_divide): Likewise.
1456 * config/mips/mips.h (mips_args): Likewise.
1457 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1458 (MIPS_ICACHE_SYNC): Likewise.
1459 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1460 (nios2_trampoline_init): Likewise.
1461 * config/pa/pa.c (hppa_tls_call): Likewise.
1462 (pa_trampoline_init): Likewise.
1463 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1464 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1465 (expand_strn_compare): Likewise.
1466 (rs6000_generate_compare): Likewise.
1467 (rs6000_expand_float128_convert): Likewise.
1468 (output_profile_hook): Likewise.
1469 (rs6000_trampoline_init): Likewise.
1470 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1471 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1472 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1473 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1474 (rs6000_generate_compare): Likewise.
1475 (rs6000_expand_float128_convert): Likewise.
1476 (output_profile_hook): Likewise.
1477 (rs6000_trampoline_init): Likewise.
1478 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1479 * config/sh/sh.c (sh_trampoline_init): Likewise.
1480 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1481 (sparc_emit_float_lib_cmp): Likewise.
1482 (sparc32_initialize_trampoline): Likewise.
1483 (sparc64_initialize_trampoline): Likewise.
1484 (sparc_profile_hook): Likewise.
1485 * config/spu/spu.c (ea_load_store): Likewise.
1486 * config/spu/spu.md (floatunssidf2): Likewise.
1487 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1488 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1489 * config/visium/visium.c (expand_block_move_4): Likewise.
1490 (expand_block_move_2): Likewise.
1491 (expand_block_move_1): Likewise.
1492 (expand_block_set_4): Likewise.
1493 (expand_block_set_2): Likewise.
1494 (expand_block_set_1): Likewise.
1495 (visium_trampoline_init): Likewise.
1496 (visium_profile_hook): Likewise.
1497 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1498 (xtensa_setup_frame_addresses): Likewise.
1499 (xtensa_trampoline_init): Likewise.
1500 * except.c (sjlj_emit_function_enter): Likewise.
1501 (sjlj_emit_function_exit): Likewise.
1502 * explow.c (allocate_dynamic_stack_space): Likewise.
1503 (probe_stack_range): Likewise.
1504 * expr.c (convert_mode_scalar): Likewise.
1505 * optabs.c (expand_binop): Likewise.
1506 (expand_twoval_binop_libfunc): Likewise.
1507 (expand_unop): Likewise.
1508 (prepare_cmp_insn): Likewise.
1509 (prepare_float_lib_cmp): Likewise.
1510 (expand_float): Likewise.
1511 (expand_fix): Likewise.
1512 (expand_fixed_convert): Likewise.
1513 (maybe_emit_sync_lock_test_and_set): Likewise.
1514 (expand_atomic_compare_and_swap): Likewise.
1515 (expand_mem_thread_fence): Likewise.
1516 (expand_atomic_fetch_op): Likewise.
1518 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1520 * doc/generic.texi (OpenACC): Adjust URL.
1521 * doc/invoke.texi (C Dialect Options): Ditto.
1523 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1525 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1526 predicate for operand 1. Add (m,<S>) constraint.
1527 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1528 Prevent memory operand 1 with register operand 2.
1530 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1532 PR rtl-optimization/82024
1533 * combine.c (try_combine): If the combination result is a PARALLEL,
1534 and we only need to retain the SET in there that would be placed
1535 at I2, check that we can place that at I3 instead, before doing so.
1537 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1540 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1542 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1543 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1546 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1547 Jeff Law <law@redhat.com>
1549 * varasm.c (bss_initializer_p): Do not put constants into .bss
1550 (categorize_decl_for_section): Handle bss_initializer_p returning
1551 false when DECL_INITIAL is NULL.
1553 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1556 * config/s390/s390.c (s390_can_inline_p): New function.
1558 2017-09-01 Jeff Law <law@redhat.com>
1560 PR tree-optimization/82052
1561 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1562 Always initialize the returned slot after a hash table miss
1563 when INSERT is true.
1565 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1567 * config/s390/s390.md (mem_signal_fence): Remove.
1568 * doc/md.texi (mem_signal_fence): Remove.
1569 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1571 * target-insns.def (mem_signal_fence): Remove.
1573 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1576 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1579 * asan.c (create_odr_indicator): Strip name encoding from assembler
1580 name before appending it after __odr_asan_.
1582 2017-09-01 Martin Liska <mliska@suse.cz>
1584 PR tree-optimization/82059
1585 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1586 frequency only when an edge is redirected.
1588 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1590 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1591 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1592 (arc_conditional_register_usage): Remove ARC600 lp_count
1594 (arc_file_start): Emit Tag_ARC_CPU_variation.
1595 (arc_can_use_doloop_p): New conditions to use ZOLs.
1596 (hwloop_fail): New function.
1597 (hwloop_optimize): Likewise.
1598 (hwloop_pattern_reg): Likewise.
1599 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1600 (arc_reorg_loops): New function, calls reorg_loops.
1601 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1602 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1604 (arc_loop_hazard): Remove function, functionality moved into
1606 (arc_hazard): Remove arc_loop_hazard call.
1607 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1608 into hwloop_optimize.
1609 (arc_label_align): Remove ZOL handling.
1610 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1611 * config/arc/arc.md (doloop_begin): Remove pattern.
1612 (doloop_begin_i): Likewise.
1613 (doloop_end_i): Likewise.
1614 (doloop_fallback): Likewise.
1615 (doloop_fallback_m): Likewise.
1616 (doloop_end): Reimplement expand.
1617 (arc_lp): New pattern for LP instruction.
1618 (loop_end): New pattern.
1619 (loop_fail): Likewise.
1620 (decrement_and_branch_until_zero): Likewise.
1621 * config/arc/arc.opt (mlpc-width): New option.
1622 * doc/invoke.texi (mlpc-width): Document option.
1624 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1626 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1627 (arc_ccfsm_advance): Fix checking for delay slots.
1628 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1630 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1632 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1634 (movhi_insn): Update store instruction constraint which are saving
1635 6-bit short immediates.
1636 (movsi_insn): Consider also short scaled load operations.
1637 (zero_extendhisi2_i): Use Usd constraint instead of T.
1638 (extendhisi2_i): Add q constraint.
1639 (arc_clzsi2): Add type and length attributes.
1640 (arc_ctzsi2): Likewise.
1641 * config/arc/constraints.md (Usc): Update constraint, the
1642 assembler can parse two relocations for a single instruction.
1644 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1646 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1647 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1649 2017-08-31 Olivier Hainque <hainque@adacore.com>
1651 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1652 match as powerpc-wrs-vxworks*.
1654 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1656 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1657 register constraint for by-element operand.
1658 (aarch64_mls_elt_merge<mode>): Likewise.
1660 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1662 * config/arc/arc.c (arc_can_follow_jump): Check for short
1665 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1667 * config.gcc: Use g.opt for arc.
1668 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1669 functionality moved to ...
1670 (legitimate_scaled_address_p): New function, ...here.
1671 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1672 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1673 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1675 (arc_override_options): Handle G option.
1676 (arc_output_pic_addr_const): Correct function definition.
1677 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1678 (arc_decl_anon_ns_mem_p): Delete.
1679 (arc_in_small_data_p): Overhaul this function to take into
1680 consideration the value given via G option.
1681 (arc_rewrite_small_data_1): Renamed and corrected old
1682 arc_rewrite_small_data function.
1683 (arc_rewrite_small_data): New function.
1684 (small_data_pattern): Don't use pic_offset_table_rtx.
1685 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1686 * config/arc/simdext.md (movmisalignv2hi): Use
1687 prepare_move_operands function.
1689 (movmisalign*): Likewise.
1690 * doc/invoke.texi (ARC options): Document -G option.
1692 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1694 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1696 * config/arc/arc.c (arc_print_operand): Output scalled address for
1697 sdata whenever is possible.
1698 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1699 load/stores are available.
1700 (compact_sda_memory_operand): Check for the alignment required by
1701 code density instructions.
1702 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1704 * config/arc/constraints.md (Usd): Update constraint.
1705 (Us0): New constraint.
1706 (Usc): Update constraint.
1708 2017-08-31 Richard Biener <rguenther@suse.de>
1711 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1714 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1716 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1717 Resize type_signature.
1719 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1720 Alan Hayward <alan.hayward@arm.com>
1721 David Sherwood <david.sherwood@arm.com>
1723 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1724 subregs whose inner modes can be stored in GPRs.
1725 (aarch64_classify_index): Likewise.
1727 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1728 Alan Hayward <alan.hayward@arm.com>
1729 David Sherwood <david.sherwood@arm.com>
1731 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1732 (V_INT_EQUIV): ...this.
1733 (v_cmp_result): Rename to...
1734 (v_int_equiv): ...this.
1735 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1736 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1737 (copysign<mode>3): Likewise.
1738 (aarch64_simd_bsl<mode>_internal): Likewise.
1739 (aarch64_simd_bsl<mode>): Likewise.
1740 (vec_cmp<mode><mode>): Likewise.
1741 (vcond<mode><mode>): Likewise.
1742 (vcond<v_cmp_mixed><mode>): Likewise.
1743 (vcondu<mode><v_cmp_mixed>): Likewise.
1744 (aarch64_cm<optab><mode>): Likewise.
1745 (aarch64_cmtst<mode>): Likewise.
1746 (aarch64_fac<optab><mode>): Likewise.
1747 (vec_perm_const<mode>): Likewise.
1748 (vcond_mask_<mode><v_cmp_result>): Rename to...
1749 (vcond_mask_<mode><v_int_equiv>): ...this.
1750 (vec_cmp<mode><v_cmp_result>): Rename to...
1751 (vec_cmp<mode><v_int_equiv>): ...this.
1753 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1754 Alan Hayward <alan.hayward@arm.com>
1755 David Sherwood <david.sherwood@arm.com>
1757 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1759 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1760 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1761 (UNSPEC_LD4_DREG): New unspecs.
1762 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1763 (aarch64_ld2<mode>_dreg_be): Replace with...
1764 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1766 (aarch64_ld3<mode>_dreg_le)
1767 (aarch64_ld3<mode>_dreg_be): Replace with...
1768 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1770 (aarch64_ld4<mode>_dreg_le)
1771 (aarch64_ld4<mode>_dreg_be): Replace with...
1772 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1775 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1777 PR tree-optimization/81987
1778 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1779 insert an initializer in a location not dominated by the stride
1782 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1784 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1785 on the entire header of the finally block in the fallthru case.
1787 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1789 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1791 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1793 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1794 rs6000_emit_move_from_cr and call renamed function.
1795 (rs6000_emit_prologue): Call renamed functions.
1796 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1797 movesi_from_cr, remove volatile CRs.
1799 2017-08-30 Jon Beniston <jon@beniston.com>
1800 Richard Biener <rguenther@suse.de>
1802 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1803 of VECTOR_MODE_P check.
1804 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1805 element vector types.
1807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1809 * df.h (df_read_modify_subreg_p): Remove in favor of...
1810 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1811 const_rtx instead of an rtx.
1812 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1813 * df-problems.c (df_word_lr_mark_ref): Likewise.
1814 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1815 (mark_pseudo_reg_dead): Likewise.
1816 (mark_ref_dead): Likewise.
1817 * reginfo.c (init_subregs_of_mode): Likewise.
1818 * sched-deps.c (sched_analyze_1): Likewise.
1819 * df-scan.c (df_def_record_1): Likewise.
1820 (df_uses_record): Likewise.
1821 (df_read_modify_subreg_p): Remove in favor of...
1822 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1823 const_rtx instead of an rtx.
1825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1826 Alan Hayward <alan.hayward@arm.com>
1827 David Sherwood <david.sherwood@arm.com>
1829 * rtl.h (partial_subreg_p): New function.
1830 * caller-save.c (save_call_clobbered_regs): Use it.
1831 * calls.c (expand_call): Likewise.
1832 * combine.c (combinable_i3pat): Likewise.
1833 (simplify_set): Likewise.
1834 (make_extraction): Likewise.
1835 (make_compound_operation_int): Likewise.
1836 (gen_lowpart_or_truncate): Likewise.
1837 (force_to_mode): Likewise.
1838 (make_field_assignment): Likewise.
1839 (reg_truncated_to_mode): Likewise.
1840 (record_truncated_value): Likewise.
1841 (move_deaths): Likewise.
1842 * cse.c (record_jump_cond): Likewise.
1843 (cse_insn): Likewise.
1844 * cselib.c (cselib_lookup_1): Likewise.
1845 * expmed.c (extract_bit_field_using_extv): Likewise.
1846 * function.c (assign_parm_setup_reg): Likewise.
1847 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1848 * ira-build.c (create_insn_allocnos): Likewise.
1849 * lra-coalesce.c (merge_pseudos): Likewise.
1850 * lra-constraints.c (match_reload): Likewise.
1851 (simplify_operand_subreg): Likewise.
1852 (curr_insn_transform): Likewise.
1853 * lra-lives.c (process_bb_lives): Likewise.
1854 * lra.c (new_insn_reg): Likewise.
1855 (lra_substitute_pseudo): Likewise.
1856 * regcprop.c (mode_change_ok): Likewise.
1857 (maybe_mode_change): Likewise.
1858 (copyprop_hardreg_forward_1): Likewise.
1859 * reload.c (push_reload): Likewise.
1860 (find_reloads): Likewise.
1861 (find_reloads_subreg_address): Likewise.
1862 * reload1.c (alter_reg): Likewise.
1863 (eliminate_regs_1): Likewise.
1864 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1866 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1868 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1869 back to if statements, including unpack.
1871 2017-08-30 Martin Liska <mliska@suse.cz>
1874 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1876 (func_checker::compare_asm_inputs_outputs): ... this function.
1877 (func_checker::compare_gimple_asm): Use the function to compare
1878 also ASM constrains.
1879 * ipa-icf-gimple.h: Rename the function.
1881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1882 Alan Hayward <alan.hayward@arm.com>
1883 David Sherwood <david.sherwood@arm.com>
1885 * coretypes.h (complex_mode): New type.
1886 * gdbhooks.py (build_pretty_printer): Handle it.
1887 * machmode.h (complex_mode): New class.
1888 (complex_mode::includes_p): New function.
1889 (is_complex_int_mode): Likewise.
1890 (is_complex_float_mode): Likewise.
1891 * genmodes.c (get_mode_class): Handle complex mode classes.
1892 * function.c (expand_function_end): Use is_complex_int_mode.
1894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1895 Alan Hayward <alan.hayward@arm.com>
1896 David Sherwood <david.sherwood@arm.com>
1898 * coretypes.h (scalar_mode_pod): New typedef.
1899 * gdbhooks.py (build_pretty_printer): Handle it.
1900 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1901 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1902 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1903 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1907 Alan Hayward <alan.hayward@arm.com>
1908 David Sherwood <david.sherwood@arm.com>
1910 * machmode.h (mode_for_vector): Take a scalar_mode instead
1912 * stor-layout.c (mode_for_vector): Likewise.
1913 * explow.c (promote_mode): Use as_a <scalar_mode>.
1914 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
1916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1917 Alan Hayward <alan.hayward@arm.com>
1918 David Sherwood <david.sherwood@arm.com>
1920 * target.def (preferred_simd_mode): Take a scalar_mode
1921 instead of a machine_mode.
1922 * targhooks.h (default_preferred_simd_mode): Likewise.
1923 * targhooks.c (default_preferred_simd_mode): Likewise.
1924 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1925 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1926 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1927 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1928 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1929 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1930 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1931 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1933 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1934 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1935 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1936 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1937 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1938 * doc/tm.texi: Regenerate.
1939 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1943 Alan Hayward <alan.hayward@arm.com>
1944 David Sherwood <david.sherwood@arm.com>
1946 * target.def (scalar_mode_supported_p): Take a scalar_mode
1947 instead of a machine_mode.
1948 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1949 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1950 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1951 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1952 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1953 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1954 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1955 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1956 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1957 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1958 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1959 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1960 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1961 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1963 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1964 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1965 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1966 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1967 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1969 * doc/tm.texi: Regenerate.
1971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1972 Alan Hayward <alan.hayward@arm.com>
1973 David Sherwood <david.sherwood@arm.com>
1975 * coretypes.h (opt_scalar_mode): New typedef.
1976 * gdbhooks.py (build_pretty_printers): Handle it.
1977 * machmode.h (mode_iterator::get_2xwider): Add overload for
1979 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1981 * expr.c (convert_mode_scalar): Likewise.
1982 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1983 * optabs.c (expand_float): Likewise.
1984 (expand_fix): Likewise.
1985 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1988 Alan Hayward <alan.hayward@arm.com>
1989 David Sherwood <david.sherwood@arm.com>
1991 * optabs.c (expand_float): Explicitly check for scalars before
1992 using a branching expansion.
1993 (expand_fix): Likewise.
1995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1996 Alan Hayward <alan.hayward@arm.com>
1997 David Sherwood <david.sherwood@arm.com>
1999 * expr.c (convert_mode): Split scalar handling out into...
2000 (convert_mode_scalar): ...this new function. Treat the modes
2003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2004 Alan Hayward <alan.hayward@arm.com>
2005 David Sherwood <david.sherwood@arm.com>
2007 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2009 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2012 Alan Hayward <alan.hayward@arm.com>
2013 David Sherwood <david.sherwood@arm.com>
2015 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2016 rather than a machine_mode.
2017 (fixed_from_string): Likewise.
2018 (fixed_convert): Likewise.
2019 (fixed_convert_from_int): Likewise.
2020 (fixed_convert_from_real): Likewise.
2021 (real_convert_from_fixed): Likewise.
2022 * fixed-value.c (fixed_from_double_int): Likewise.
2023 (fixed_from_string): Likewise.
2024 (fixed_convert): Likewise.
2025 (fixed_convert_from_int): Likewise.
2026 (fixed_convert_from_real): Likewise.
2027 (real_convert_from_fixed): Likewise.
2028 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2031 Alan Hayward <alan.hayward@arm.com>
2032 David Sherwood <david.sherwood@arm.com>
2034 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2035 of separate mode class checks. Do not allow vector modes here.
2036 (immed_wide_int_const): Use as_a <scalar_mode>.
2037 * explow.c (trunc_int_for_mode): Likewise.
2038 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2039 (wi::shwi): Likewise.
2040 (wi::min_value): Likewise.
2041 (wi::max_value): Likewise.
2042 * dwarf2out.c (loc_descriptor): Likewise.
2043 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2047 Alan Hayward <alan.hayward@arm.com>
2048 David Sherwood <david.sherwood@arm.com>
2050 * tree.h (SCALAR_TYPE_MODE): New macro.
2051 * expr.c (expand_expr_addr_expr_1): Use it.
2052 (expand_expr_real_2): Likewise.
2053 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2054 (fold_convert_const_fixed_from_int): Likewise.
2055 (fold_convert_const_fixed_from_real): Likewise.
2056 (native_encode_fixed): Likewise
2057 (native_encode_complex): Likewise
2058 (native_encode_vector): Likewise.
2059 (native_interpret_fixed): Likewise.
2060 (native_interpret_real): Likewise.
2061 (native_interpret_complex): Likewise.
2062 (native_interpret_vector): Likewise.
2063 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2064 (simd_clone_adjust_argument_types): Likewise.
2065 (simd_clone_init_simd_arrays): Likewise.
2066 (simd_clone_adjust): Likewise.
2067 * stor-layout.c (layout_type): Likewise.
2068 * tree.c (build_minus_one_cst): Likewise.
2069 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2070 * tree-inline.c (estimate_move_cost): Likewise.
2071 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2072 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2073 (vectorizable_reduction): Likewise.
2074 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2075 (vect_recog_mixed_size_cond_pattern): Likewise.
2076 (check_bool_pattern): Likewise.
2077 (adjust_bool_pattern): Likewise.
2078 (search_type_for_mask_1): Likewise.
2079 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2080 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2081 (vectorizable_load): Likewise.
2082 (vectorizable_store): Likewise.
2083 * ubsan.c (ubsan_encode_value): Likewise.
2084 * varasm.c (output_constant): Likewise.
2086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2087 Alan Hayward <alan.hayward@arm.com>
2088 David Sherwood <david.sherwood@arm.com>
2090 * coretypes.h (scalar_mode): New class.
2091 * machmode.h (scalar_mode): Likewise.
2092 (scalar_mode::includes_p): New function.
2093 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2094 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2095 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2096 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2097 * expmed.c (store_bit_field_1): Likewise.
2098 (extract_bit_field_1): Likewise.
2099 * expr.c (write_complex_part): Likewise.
2100 (read_complex_part): Likewise.
2101 (emit_move_complex_push): Likewise.
2102 (expand_expr_real_2): Likewise.
2103 * function.c (assign_parm_setup_reg): Likewise.
2104 (assign_parms_unsplit_complex): Likewise.
2105 * optabs.c (expand_binop): Likewise.
2106 * rtlanal.c (subreg_get_info): Likewise.
2107 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2108 * varasm.c (output_constant_pool_2): Likewise.
2110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2111 Alan Hayward <alan.hayward@arm.com>
2112 David Sherwood <david.sherwood@arm.com>
2114 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2116 (expmed_mult_highpart_optab): Likewise.
2117 (expmed_mult_highpart): Likewise.
2119 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2120 Alan Hayward <alan.hayward@arm.com>
2121 David Sherwood <david.sherwood@arm.com>
2123 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2124 instead of a machine_mode.
2125 (builtin_memset_read_str): Likewise.
2126 * builtins.c (c_readstr): Likewise.
2127 (builtin_memcpy_read_str): Likewise.
2128 (builtin_strncpy_read_str): Likewise.
2129 (builtin_memset_read_str): Likewise.
2130 (builtin_memset_gen_str): Likewise.
2131 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2132 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2133 instead of a machine_mode.
2134 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2136 (make_extraction): Likewise.
2137 (try_widen_shift_mode): Take and return scalar_int_modes instead
2139 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2140 a scalar_int_mode instead of a machine_mode.
2141 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2142 (avr_addr_space_pointer_mode): Likewise.
2143 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2144 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2145 (msp430_unwind_word_mode): Likewise.
2146 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2147 (spu_addr_space_pointer_mode): Likewise.
2148 (spu_addr_space_address_mode): Likewise.
2149 (spu_libgcc_cmp_return_mode): Likewise.
2150 (spu_libgcc_shift_count_mode): Likewise.
2151 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2152 (rl78_addr_space_pointer_mode): Likewise.
2153 (fl78_unwind_word_mode): Likewise.
2154 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2156 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2157 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2158 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2159 (mips_valid_pointer_mode): Likewise.
2160 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2161 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2162 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2164 (ft32_addr_space_address_mode): Likewise.
2165 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2166 scalar_int_mode instead of a machine_mode.
2167 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2169 (m32c_addr_space_address_mode): Likewise.
2170 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2171 (rs6000_eh_return_filter_mode): Likewise.
2172 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2173 (rs6000_eh_return_filter_mode): Likewise.
2174 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2175 (s390_libgcc_shift_count_mode): Likewise.
2176 (s390_unwind_word_mode): Likewise.
2177 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2179 * target.def (mode_rep_extended): Likewise.
2180 (valid_pointer_mode): Likewise.
2181 (addr_space.valid_pointer_mode): Likewise.
2182 (eh_return_filter_mode): Return a scalar_int_mode rather than
2184 (libgcc_cmp_return_mode): Likewise.
2185 (libgcc_shift_count_mode): Likewise.
2186 (unwind_word_mode): Likewise.
2187 (addr_space.pointer_mode): Likewise.
2188 (addr_space.address_mode): Likewise.
2189 * doc/tm.texi: Regenerate.
2190 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2192 (do_jump): Use scalar_int_mode for local variables.
2193 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2194 rather than a machine_mode.
2195 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2196 (scompare_loc_descriptor_wide): Likewise.
2197 (scompare_loc_descriptor_narrow): Likewise.
2198 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2200 * except.c (sjlj_emit_dispatch_table): Likewise.
2201 (expand_builtin_eh_copy_values): Likewise.
2202 * explow.c (convert_memory_address_addr_space_1): Likewise.
2203 Take a scalar_int_mode rather than a machine_mode.
2204 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2205 than a machine_mode.
2206 (memory_address_addr_space): Use scalar_int_mode for local variables.
2207 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2208 rather than a machine_mode.
2209 * expmed.c (mask_rtx): Likewise.
2210 (init_expmed_one_conv): Likewise.
2211 (expand_mult_highpart_adjust): Likewise.
2212 (extract_high_half): Likewise.
2213 (expmed_mult_highpart_optab): Likewise.
2214 (expmed_mult_highpart): Likewise.
2215 (expand_smod_pow2): Likewise.
2216 (expand_sdiv_pow2): Likewise.
2217 (emit_store_flag_int): Likewise.
2218 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2220 (extract_low_bits): Likewise.
2221 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2223 * expr.c (pieces_addr::adjust): Likewise.
2224 (can_store_by_pieces): Likewise.
2225 (store_by_pieces): Likewise.
2226 (clear_by_pieces_1): Likewise.
2227 (expand_expr_addr_expr_1): Likewise.
2228 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2229 (expand_expr_real_1): Likewise.
2230 (try_casesi): Likewise.
2231 * final.c (shorten_branches): Likewise.
2232 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2233 type of "mode" to machine_mode.
2234 * internal-fn.c (expand_arith_overflow_result_store): Take a
2235 scalar_int_mode rather than a machine_mode.
2236 (expand_mul_overflow): Use scalar_int_mode for local variables.
2237 * loop-doloop.c (doloop_modify): Likewise.
2238 (doloop_optimize): Likewise.
2239 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2240 than a machine_mode.
2241 (expand_doubleword_shift_condmove): Likewise.
2242 (expand_doubleword_shift): Likewise.
2243 (expand_doubleword_clz): Likewise.
2244 (expand_doubleword_popcount): Likewise.
2245 (expand_doubleword_parity): Likewise.
2246 (expand_absneg_bit): Use scalar_int_mode for local variables.
2247 (prepare_float_lib_cmp): Likewise.
2248 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2249 rather than a machine_mode.
2250 (convert_memory_address_addr_space): Likewise.
2251 (get_mode_bounds): Likewise.
2252 (get_address_mode): Return a scalar_int_mode rather than a
2254 * rtlanal.c (get_address_mode): Likewise.
2255 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2256 than a machine_mode.
2257 * targhooks.c (default_mode_rep_extended): Likewise.
2258 (default_valid_pointer_mode): Likewise.
2259 (default_addr_space_valid_pointer_mode): Likewise.
2260 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2261 than a machine_mode.
2262 (default_libgcc_cmp_return_mode): Likewise.
2263 (default_libgcc_shift_count_mode): Likewise.
2264 (default_unwind_word_mode): Likewise.
2265 (default_addr_space_pointer_mode): Likewise.
2266 (default_addr_space_address_mode): Likewise.
2267 * targhooks.h (default_eh_return_filter_mode): Likewise.
2268 (default_libgcc_cmp_return_mode): Likewise.
2269 (default_libgcc_shift_count_mode): Likewise.
2270 (default_unwind_word_mode): Likewise.
2271 (default_addr_space_pointer_mode): Likewise.
2272 (default_addr_space_address_mode): Likewise.
2273 (default_mode_rep_extended): Take a scalar_int_mode rather than
2275 (default_valid_pointer_mode): Likewise.
2276 (default_addr_space_valid_pointer_mode): Likewise.
2277 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2279 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2280 rather than a machine_mode.
2281 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2282 for local variables.
2283 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2284 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2285 than a machine_mode.
2287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2288 Alan Hayward <alan.hayward@arm.com>
2289 David Sherwood <david.sherwood@arm.com>
2291 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2292 the mode argument to scalar_int_mode.
2293 (do_jump_by_parts_zero_rtx): Likewise.
2294 (do_jump_by_parts_equality_rtx): Likewise.
2295 (do_jump_by_parts_greater): Take a mode argument.
2296 (do_jump_by_parts_equality): Likewise.
2297 (do_jump_1): Update calls accordingly.
2299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2300 Alan Hayward <alan.hayward@arm.com>
2301 David Sherwood <david.sherwood@arm.com>
2303 * is-a.h (safe_dyn_cast): New function.
2304 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2305 (jump_table_for_label): Likewise.
2306 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2307 instead of an rtx_insn *.
2308 (shorten_branches): Use dyn_cast instead of LABEL_P and
2309 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2310 rtx_jump_table_data::get_data_mode.
2311 (final_scan_insn): Likewise.
2313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2314 Alan Hayward <alan.hayward@arm.com>
2315 David Sherwood <david.sherwood@arm.com>
2317 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2318 trying to combine a full-register integer set with a subreg
2321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2322 Alan Hayward <alan.hayward@arm.com>
2323 David Sherwood <david.sherwood@arm.com>
2325 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2326 that is always either address_mode or pointer_mode.
2328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2329 Alan Hayward <alan.hayward@arm.com>
2330 David Sherwood <david.sherwood@arm.com>
2332 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2333 when the two are known to be equal.
2335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2336 Alan Hayward <alan.hayward@arm.com>
2337 David Sherwood <david.sherwood@arm.com>
2339 * simplify-rtx.c (simplify_const_unary_operation): Use
2340 is_a <scalar_int_mode> instead of checking for a nonzero
2341 precision. Forcibly convert op_mode to a scalar_int_mode
2342 in that case. More clearly differentiate the operand and
2343 result modes and use the former when deciding what the value
2344 of a count-bits operation should be. Use is_int_mode instead
2345 of checking for a MODE_INT. Remove redundant check for whether
2346 this mode has a zero precision.
2348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2349 Alan Hayward <alan.hayward@arm.com>
2350 David Sherwood <david.sherwood@arm.com>
2352 * optabs.c (widen_leading): Change the type of the mode argument
2353 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2354 (widen_bswap): Likewise.
2355 (expand_parity): Likewise.
2356 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2357 (expand_ffs): Likewise.
2358 (epand_unop): Check for scalar integer modes before calling the
2361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2362 Alan Hayward <alan.hayward@arm.com>
2363 David Sherwood <david.sherwood@arm.com>
2365 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2367 (expand_expr_real_1): Update call accordingly.
2369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2370 Alan Hayward <alan.hayward@arm.com>
2371 David Sherwood <david.sherwood@arm.com>
2373 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2374 value_mode arguments. Use scalar_int_mode internally.
2375 (store_bit_field_1): Rename the new integer mode from imode
2376 to op0_mode and use it instead of GET_MODE (op0). Update calls
2377 to store_split_bit_field, store_bit_field_using_insv and
2378 store_fixed_bit_field.
2379 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2380 Use scalar_int_mode internally. Use a bit count rather than a mode
2381 when calculating the largest bit size for get_best_mode.
2382 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2383 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2384 Remove assertion that OP0 has a scalar integer mode.
2385 (store_split_bit_field): Add op0_mode and value_mode arguments.
2386 Update calls to extract_fixed_bit_field.
2387 (extract_bit_field_using_extv): Add an op0_mode argument.
2388 Use scalar_int_mode internally.
2389 (extract_bit_field_1): Rename the new integer mode from imode to
2390 op0_mode and use it instead of GET_MODE (op0). Update calls to
2391 extract_split_bit_field, extract_bit_field_using_extv and
2392 extract_fixed_bit_field.
2393 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2394 to extract_split_bit_field and extract_fixed_bit_field_1.
2395 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2396 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2398 (extract_split_bit_field): Add an op0_mode argument. Update call
2399 to extract_fixed_bit_field.
2401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2402 Alan Hayward <alan.hayward@arm.com>
2403 David Sherwood <david.sherwood@arm.com>
2405 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2406 * explow.c (hard_function_value): Likewise.
2407 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2408 convert_to_mode call outside the loop.
2409 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2410 for the mode iterator. Require the mode specified by max_pieces
2412 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2414 (copy_blkmode_to_reg): Likewise.
2415 (set_storage_via_setmem): Likewise.
2416 * optabs.c (prepare_cmp_insn): Likewise.
2417 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2418 * stor-layout.c (finish_bitfield_representative): Likewise.
2420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2421 Alan Hayward <alan.hayward@arm.com>
2422 David Sherwood <david.sherwood@arm.com>
2424 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2425 * expr.c (convert_move): Use them.
2426 (convert_modes): Likewise.
2427 (store_expr_with_bounds): Likewise.
2429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2430 Alan Hayward <alan.hayward@arm.com>
2431 David Sherwood <david.sherwood@arm.com>
2433 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2434 parameter for the mode of "x". Remove the "known_x", "known_mode"
2435 and "known_ret" arguments. Change the type of the mode argument
2437 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2438 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2439 (reg_num_sign_bit_copies_for_combine): Likewise.
2440 * rtlanal.c (nonzero_bits1): Likewise.
2441 (num_sign_bit_copies1): Likewise.
2442 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2443 (reg_num_sign_bit_copies_general): Likewise.
2444 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2445 (reg_nonzero_bits_general): Likewise.
2447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2448 Alan Hayward <alan.hayward@arm.com>
2449 David Sherwood <david.sherwood@arm.com>
2451 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2452 than in subroutines. Return 1 for non-integer modes.
2453 (cached_num_sign_bit_copies): Change the type of the mode parameter
2455 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2456 classes. Handle CONST_INT_P first and then check whether X also
2457 has a scalar integer mode. Check the same thing for inner registers
2458 of a SUBREG and for values that are being extended or truncated.
2460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2461 Alan Hayward <alan.hayward@arm.com>
2462 David Sherwood <david.sherwood@arm.com>
2464 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2465 in subroutines. Return the mode mask for non-integer modes.
2466 (cached_nonzero_bits): Change the type of the mode parameter
2468 (nonzero_bits1): Likewise. Remove early exit for other mode
2469 classes. Handle CONST_INT_P first and then check whether X
2470 also has a scalar integer mode.
2472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2473 Alan Hayward <alan.hayward@arm.com>
2474 David Sherwood <david.sherwood@arm.com>
2476 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2477 Return a scalar_int_mode and assert that the size is greater than
2479 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2480 (op_by_pieces_d::op_by_pieces_d): Likewise.
2481 (op_by_pieces_d::run): Likewise.
2482 (can_store_by_pieces): Likewise.
2484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2485 Alan Hayward <alan.hayward@arm.com>
2486 David Sherwood <david.sherwood@arm.com>
2488 * combine.c (extract_left_shift): Add a mode argument and update
2490 (make_compound_operation_int): Change the type of the mode parameter
2491 to scalar_int_mode and update the call to extract_left_shift.
2493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2494 Alan Hayward <alan.hayward@arm.com>
2495 David Sherwood <david.sherwood@arm.com>
2497 * combine.c (simplify_and_const_int): Change the type of the mode
2498 parameter to scalar_int_mode.
2499 (simplify_and_const_int_1): Likewise. Update recursive call.
2501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2502 Alan Hayward <alan.hayward@arm.com>
2503 David Sherwood <david.sherwood@arm.com>
2505 * combine.c (simplify_compare_const): Check that the mode is a
2506 scalar_int_mode (rather than VOIDmode) before testing its
2508 (simplify_comparison): Move COMPARISON_P handling out of the
2509 loop and restrict the latter part of the loop to scalar_int_modes.
2510 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2511 and when considering SUBREG_REGs. Use is_int_mode instead of
2512 checking GET_MODE_CLASS against MODE_INT.
2514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2515 Alan Hayward <alan.hayward@arm.com>
2516 David Sherwood <david.sherwood@arm.com>
2518 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2519 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2520 shift_unit_mode and for modes involved in scalar shifts.
2522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2523 Alan Hayward <alan.hayward@arm.com>
2524 David Sherwood <david.sherwood@arm.com>
2526 * combine.c (force_int_to_mode): New function, split out from...
2527 (force_to_mode): ...here. Keep xmode up-to-date and use it
2528 instead of GET_MODE (x).
2530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2531 Alan Hayward <alan.hayward@arm.com>
2532 David Sherwood <david.sherwood@arm.com>
2534 * optabs-query.h (extraction_insn::struct_mode): Change type to
2535 opt_scalar_int_mode and update comment.
2536 (extraction_insn::field_mode): Change type to scalar_int_mode.
2537 (extraction_insn::pos_mode): Likewise.
2538 * combine.c (make_extraction): Update accordingly.
2539 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2540 (get_optab_extraction_insn): Likewise.
2541 * recog.c (simplify_while_replacing): Likewise.
2542 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2543 parameter to opt_scalar_int_mode.
2545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2546 Alan Hayward <alan.hayward@arm.com>
2547 David Sherwood <david.sherwood@arm.com>
2549 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2550 to a scalar_int_mode instead of a machine_mode.
2551 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2552 (get_best_mode): Return a boolean and use a pointer argument to store
2553 the selected mode. Replace the limit mode parameter with a bit limit.
2554 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2555 for the values returned by bit_field_mode_iterator::next_mode.
2556 (store_bit_field): Update call to get_best_mode.
2557 (store_fixed_bit_field): Likewise.
2558 (extract_fixed_bit_field): Likewise.
2559 * expr.c (optimize_bitfield_assignment_op): Likewise.
2560 * fold-const.c (optimize_bit_field_compare): Likewise.
2561 (fold_truth_andor_1): Likewise.
2562 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2563 Update for new type of m_mode.
2564 (get_best_mode): As above.
2566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2567 Alan Hayward <alan.hayward@arm.com>
2568 David Sherwood <david.sherwood@arm.com>
2570 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2571 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2572 (store_bit_field): Check is_a <scalar_int_mode> before calling
2573 strict_volatile_bitfield_p.
2574 (extract_bit_field): Likewise.
2576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2577 Alan Hayward <alan.hayward@arm.com>
2578 David Sherwood <david.sherwood@arm.com>
2580 * target.def (cstore_mode): Return a scalar_int_mode.
2581 * doc/tm.texi: Regenerate.
2582 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2583 * targhooks.h (default_cstore_mode): Likewise.
2584 * targhooks.c (default_cstore_mode): Likewise, using a forced
2586 * expmed.c (emit_cstore): Expect the target of the cstore to be
2589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2590 Alan Hayward <alan.hayward@arm.com>
2591 David Sherwood <david.sherwood@arm.com>
2593 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2595 (niter_desc): Likewise mode.
2596 (iv_analyze): Add a mode parameter.
2598 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2599 and change its type to scalar_int_mode.
2600 * loop-iv.c: Update commentary at head of file.
2601 (iv_constant): Pass the mode paraeter before the rtx it describes
2602 and change its type to scalar_int_mode. Remove VOIDmode handling.
2603 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2604 (iv_extend): Likewise.
2605 (shorten_into_mode): Likewise.
2606 (iv_add): Use scalar_int_mode.
2607 (iv_mult): Likewise.
2608 (iv_shift): Likewise.
2609 (canonicalize_iv_subregs): Likewise.
2610 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2611 it describes and change its mode to scalar_int_mode. Also change
2612 the type of the returned inner_mode to scalar_int_mode.
2613 (get_biv_step): Likewise, turning outer_mode from a pointer
2614 into a direct parameter. Update call to get_biv_step_1.
2615 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2616 iv_constant and get_biv_step.
2617 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2618 and change its type to scalar_int_mode. Don't initialise iv->mode
2619 to VOIDmode and remove later checks for its still being VOIDmode.
2620 Update calls to iv_analyze_op and iv_analyze_expr. Check
2621 is_a <scalar_int_mode> when changing the mode under consideration.
2622 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2623 Update call to iv_analyze_expr.
2624 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2625 inner register is not also a scalar_int_mode. Update call to
2627 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2628 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2629 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2630 separate mode class checks. Update calls to iv_analyze. Remove
2631 fix-up of VOIDmodes after iv_analyze_biv.
2632 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2633 don't have a scalar_int_mode. Update call to biv_p.
2635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2636 Alan Hayward <alan.hayward@arm.com>
2637 David Sherwood <david.sherwood@arm.com>
2639 * cfgexpand.c (convert_debug_memory_address): Use
2640 as_a <scalar_int_mode>.
2641 * combine.c (expand_compound_operation): Likewise.
2642 (make_extraction): Likewise.
2643 (change_zero_ext): Likewise.
2644 (simplify_comparison): Likewise.
2645 * cse.c (cse_insn): Likewise.
2646 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2647 (mem_loc_descriptor): Likewise.
2648 (loc_descriptor): Likewise.
2649 * expmed.c (init_expmed_one_mode): Likewise.
2650 (synth_mult): Likewise.
2651 (emit_store_flag_1): Likewise.
2652 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2653 of a comparison with size.
2654 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2655 (reduce_to_bit_field_precision): Likewise.
2656 * function.c (expand_function_end): Likewise.
2657 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2658 * loop-doloop.c (doloop_modify): Likewise.
2659 * optabs.c (expand_binop): Likewise.
2660 (expand_unop): Likewise.
2661 (expand_copysign_absneg): Likewise.
2662 (prepare_cmp_insn): Likewise.
2663 (maybe_legitimize_operand): Likewise.
2664 * recog.c (const_scalar_int_operand): Likewise.
2665 * rtlanal.c (get_address_mode): Likewise.
2666 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2667 (simplify_cond_clz_ctz): Likewise.
2668 * tree-nested.c (get_nl_goto_field): Likewise.
2669 * tree.c (build_vector_type_for_mode): Likewise.
2670 * var-tracking.c (use_narrower_mode): Likewise.
2672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2673 Alan Hayward <alan.hayward@arm.com>
2674 David Sherwood <david.sherwood@arm.com>
2676 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2677 * builtins.c (expand_builtin_signbit): Use it.
2678 * cfgexpand.c (expand_debug_expr): Likewise.
2679 * dojump.c (do_jump): Likewise.
2680 (do_compare_and_jump): Likewise.
2681 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2682 * expmed.c (make_tree): Likewise.
2683 * expr.c (expand_expr_real_2): Likewise.
2684 (expand_expr_real_1): Likewise.
2685 (try_casesi): Likewise.
2686 * fold-const-call.c (fold_const_call_ss): Likewise.
2687 * fold-const.c (unextend): Likewise.
2688 (extract_muldiv_1): Likewise.
2689 (fold_single_bit_test): Likewise.
2690 (native_encode_int): Likewise.
2691 (native_encode_string): Likewise.
2692 (native_interpret_int): Likewise.
2693 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2694 * internal-fn.c (expand_addsub_overflow): Likewise.
2695 (expand_neg_overflow): Likewise.
2696 (expand_mul_overflow): Likewise.
2697 (expand_arith_overflow): Likewise.
2698 * match.pd: Likewise.
2699 * stor-layout.c (layout_type): Likewise.
2700 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2701 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2702 * tree-ssanames.c (get_range_info): Likewise.
2703 * tree-switch-conversion.c (array_value_type) Likewise.
2704 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2705 (vect_recog_divmod_pattern): Likewise.
2706 (vect_recog_mixed_size_cond_pattern): Likewise.
2707 * tree-vrp.c (extract_range_basic): Likewise.
2708 (simplify_float_conversion_using_ranges): Likewise.
2709 * tree.c (int_fits_type_p): Likewise.
2710 * ubsan.c (instrument_bool_enum_load): Likewise.
2711 * varasm.c (mergeable_string_section): Likewise.
2712 (narrowing_initializer_constant_valid_p): Likewise.
2713 (output_constant): Likewise.
2715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2716 Alan Hayward <alan.hayward@arm.com>
2717 David Sherwood <david.sherwood@arm.com>
2719 * machmode.h (NARROWEST_INT_MODE): New macro.
2720 * expr.c (alignment_for_piecewise_move): Use it instead of
2721 GET_CLASS_NARROWEST_MODE (MODE_INT).
2722 (push_block): Likewise.
2723 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2725 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2728 Alan Hayward <alan.hayward@arm.com>
2729 David Sherwood <david.sherwood@arm.com>
2731 * postreload.c (move2add_valid_value_p): Change the type of the
2732 mode parameter to scalar_int_mode.
2733 (move2add_use_add2_insn): Add a mode parameter and use it instead
2735 (move2add_use_add3_insn): Likewise.
2736 (reload_cse_move2add): Update accordingly.
2738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2739 Alan Hayward <alan.hayward@arm.com>
2740 David Sherwood <david.sherwood@arm.com>
2742 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2744 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2745 * optabs.c (expand_unop): Likewise.
2747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2748 Alan Hayward <alan.hayward@arm.com>
2749 David Sherwood <david.sherwood@arm.com>
2751 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2752 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2753 (popcount_loc_descriptor): Likewise.
2754 (bswap_loc_descriptor): Likewise.
2755 (rotate_loc_descriptor): Likewise.
2756 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2757 calling the functions above.
2759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2760 Alan Hayward <alan.hayward@arm.com>
2761 David Sherwood <david.sherwood@arm.com>
2763 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2765 (try_combine): Likewise.
2766 (simplify_if_then_else): Likewise.
2767 * cse.c (cse_insn): Likewise.
2768 * dwarf2out.c (mem_loc_descriptor): Likewise.
2769 * emit-rtl.c (gen_lowpart_common): Likewise.
2770 * simplify-rtx.c (simplify_truncation): Likewise.
2771 (simplify_binary_operation_1): Likewise.
2772 (simplify_const_relational_operation): Likewise.
2773 (simplify_ternary_operation): Likewise.
2774 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2777 Alan Hayward <alan.hayward@arm.com>
2778 David Sherwood <david.sherwood@arm.com>
2780 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2781 * reload.c (push_reload): Likewise.
2782 (find_reloads): Likewise.
2784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2785 Alan Hayward <alan.hayward@arm.com>
2786 David Sherwood <david.sherwood@arm.com>
2788 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2789 (make_compound_operation_int): Likewise.
2790 (change_zero_ext): Likewise.
2791 * expr.c (convert_move): Likewise.
2792 (convert_modes): Likewise.
2793 * fwprop.c (forward_propagate_subreg): Likewise.
2794 * loop-iv.c (get_biv_step_1): Likewise.
2795 * optabs.c (widen_operand): Likewise.
2796 * postreload.c (move2add_valid_value_p): Likewise.
2797 * recog.c (simplify_while_replacing): Likewise.
2798 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2799 (simplify_binary_operation_1): Likewise. Remove redundant
2800 mode equality check.
2802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2803 Alan Hayward <alan.hayward@arm.com>
2804 David Sherwood <david.sherwood@arm.com>
2806 * combine.c (combine_simplify_rtx): Add checks for
2807 is_a <scalar_int_mode>.
2808 (simplify_if_then_else): Likewise.
2809 (make_field_assignment): Likewise.
2810 (simplify_comparison): Likewise.
2811 * ifcvt.c (noce_try_bitop): Likewise.
2812 * loop-invariant.c (canonicalize_address_mult): Likewise.
2813 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2816 Alan Hayward <alan.hayward@arm.com>
2817 David Sherwood <david.sherwood@arm.com>
2819 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2820 is_a <scalar_int_mode> instead of != BLKmode.
2822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2823 Alan Hayward <alan.hayward@arm.com>
2824 David Sherwood <david.sherwood@arm.com>
2826 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2827 instead of != VOIDmode.
2828 * combine.c (if_then_else_cond): Likewise.
2829 (change_zero_ext): Likewise.
2830 * dwarf2out.c (mem_loc_descriptor): Likewise.
2831 (loc_descriptor): Likewise.
2832 * rtlanal.c (canonicalize_condition): Likewise.
2833 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2836 Alan Hayward <alan.hayward@arm.com>
2837 David Sherwood <david.sherwood@arm.com>
2839 * simplify-rtx.c (simplify_binary_operation_1): Use
2840 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2843 Alan Hayward <alan.hayward@arm.com>
2844 David Sherwood <david.sherwood@arm.com>
2846 * wide-int.h (int_traits<unsigned char>) New class.
2847 (int_traits<unsigned short>) Likewise.
2848 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2849 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2851 * combine.c (set_nonzero_bits_and_sign_copies): Use
2852 is_a <scalar_int_mode>.
2853 (find_split_point): Likewise.
2854 (combine_simplify_rtx): Likewise.
2855 (simplify_logical): Likewise.
2856 (expand_compound_operation): Likewise.
2857 (expand_field_assignment): Likewise.
2858 (make_compound_operation): Likewise.
2859 (extended_count): Likewise.
2860 (change_zero_ext): Likewise.
2861 (simplify_comparison): Likewise.
2862 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2863 (ucompare_loc_descriptor): Likewise.
2864 (minmax_loc_descriptor): Likewise.
2865 (mem_loc_descriptor): Likewise.
2866 (loc_descriptor): Likewise.
2867 * expmed.c (init_expmed_one_mode): Likewise.
2868 * lra-constraints.c (lra_constraint_offset): Likewise.
2869 * optabs.c (prepare_libcall_arg): Likewise.
2870 * postreload.c (move2add_note_store): Likewise.
2871 * reload.c (operands_match_p): Likewise.
2872 * rtl.h (load_extend_op): Likewise.
2873 * rtlhooks.c (gen_lowpart_general): Likewise.
2874 * simplify-rtx.c (simplify_truncation): Likewise.
2875 (simplify_unary_operation_1): Likewise.
2876 (simplify_binary_operation_1): Likewise.
2877 (simplify_const_binary_operation): Likewise.
2878 (simplify_const_relational_operation): Likewise.
2879 (simplify_subreg): Likewise.
2880 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2881 * var-tracking.c (adjust_mems): Likewise.
2882 (prepare_call_arguments): Likewise.
2884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2885 Alan Hayward <alan.hayward@arm.com>
2886 David Sherwood <david.sherwood@arm.com>
2888 * machmode.h (is_int_mode): New fuction.
2889 * combine.c (find_split_point): Use it.
2890 (combine_simplify_rtx): Likewise.
2891 (simplify_if_then_else): Likewise.
2892 (simplify_set): Likewise.
2893 (simplify_shift_const_1): Likewise.
2894 (simplify_comparison): Likewise.
2895 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2896 * cse.c (notreg_cost): Likewise.
2897 (cse_insn): Likewise.
2898 * cselib.c (cselib_lookup_1): Likewise.
2899 * dojump.c (do_jump_1): Likewise.
2900 (do_compare_rtx_and_jump): Likewise.
2901 * dse.c (get_call_args): Likewise.
2902 * dwarf2out.c (rtl_for_decl_init): Likewise.
2903 (native_encode_initializer): Likewise.
2904 * expmed.c (emit_store_flag_1): Likewise.
2905 (emit_store_flag): Likewise.
2906 * expr.c (convert_modes): Likewise.
2907 (store_field): Likewise.
2908 (expand_expr_real_1): Likewise.
2909 * fold-const.c (fold_read_from_constant_string): Likewise.
2910 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2911 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2912 * optabs.c (expand_binop): Likewise.
2913 (expand_unop): Likewise.
2914 (expand_abs_nojump): Likewise.
2915 (expand_one_cmpl_abs_nojump): Likewise.
2916 * simplify-rtx.c (mode_signbit_p): Likewise.
2917 (val_signbit_p): Likewise.
2918 (val_signbit_known_set_p): Likewise.
2919 (val_signbit_known_clear_p): Likewise.
2920 (simplify_relational_operation_1): Likewise.
2921 * tree.c (vector_type_mode): Likewise.
2923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2924 Alan Hayward <alan.hayward@arm.com>
2925 David Sherwood <david.sherwood@arm.com>
2927 * machmode.h (smallest_mode_for_size): Fix formatting.
2928 (smallest_int_mode_for_size): New function.
2929 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2930 instead of smallest_mode_for_size.
2931 * combine.c (make_extraction): Likewise.
2932 * config/arc/arc.c (arc_expand_movmem): Likewise.
2933 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2934 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2935 * config/s390/s390.c (s390_expand_insv): Likewise.
2936 * config/sparc/sparc.c (assign_int_registers): Likewise.
2937 * config/spu/spu.c (spu_function_value): Likewise.
2938 (spu_function_arg): Likewise.
2939 * coverage.c (get_gcov_type): Likewise.
2940 (get_gcov_unsigned_t): Likewise.
2941 * dse.c (find_shift_sequence): Likewise.
2942 * expmed.c (store_bit_field_1): Likewise.
2943 * expr.c (convert_move): Likewise.
2944 (store_field): Likewise.
2945 * internal-fn.c (expand_arith_overflow): Likewise.
2946 * optabs-query.c (get_best_extraction_insn): Likewise.
2947 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2948 * stor-layout.c (layout_type): Likewise.
2949 (initialize_sizetypes): Likewise.
2950 * targhooks.c (default_get_mask_mode): Likewise.
2951 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2954 Alan Hayward <alan.hayward@arm.com>
2955 David Sherwood <david.sherwood@arm.com>
2957 * machmode.h (opt_mode::else_blk): New function.
2958 (int_mode_for_mode): Declare.
2959 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2960 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2962 * cfgexpand.c (expand_debug_expr): Likewise.
2963 * combine.c (gen_lowpart_or_truncate): Likewise.
2964 (gen_lowpart_for_combine): Likewise.
2965 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2966 * config/avr/avr.c (avr_to_int_mode): Likewise.
2967 (avr_out_plus_1): Likewise.
2968 (avr_out_plus): Likewise.
2969 (avr_out_round): Likewise.
2970 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2971 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2972 (s390_expand_vcond): Likewise.
2973 * config/spu/spu.c (spu_split_immediate): Likewise.
2974 (spu_expand_mov): Likewise.
2975 * dse.c (get_stored_val): Likewise.
2976 * expmed.c (store_bit_field_1): Likewise.
2977 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2979 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2980 (extract_low_bits): Likewise.
2981 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2982 handling rather than repeating the check.
2983 (emit_group_store): Likewise.
2984 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2985 * optabs.c (expand_absneg_bit): Likewise.
2986 (expand_copysign_absneg): Likewise.
2987 (expand_copysign_bit): Likewise.
2988 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2989 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2990 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2991 * var-tracking.c (prepare_call_arguments): Likewise.
2992 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2993 int_mode_for_mode instead of mode_for_size.
2994 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2997 Alan Hayward <alan.hayward@arm.com>
2998 David Sherwood <david.sherwood@arm.com>
3000 * machmode.h (int_mode_for_size): New function.
3001 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3002 instead of mode_for_size.
3003 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3005 * combine.c (expand_field_assignment): Use int_mode_for_size
3006 instead of mode_for_size.
3007 (make_extraction): Likewise.
3008 (simplify_shift_const_1): Likewise.
3009 (simplify_comparison): Likewise.
3010 * dojump.c (do_jump): Likewise.
3011 * dwarf2out.c (mem_loc_descriptor): Likewise.
3012 * emit-rtl.c (init_derived_machine_modes): Likewise.
3013 * expmed.c (flip_storage_order): Likewise.
3014 (convert_extracted_bit_field): Likewise.
3015 * expr.c (copy_blkmode_from_reg): Likewise.
3016 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3017 * internal-fn.c (expand_mul_overflow): Likewise.
3018 * lower-subreg.c (simple_move): Likewise.
3019 * optabs-libfuncs.c (init_optabs): Likewise.
3020 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3021 * tree.c (vector_type_mode): Likewise.
3022 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3023 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3024 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3025 * tree-vect-stmts.c (vectorizable_load): Likewise.
3026 (vectorizable_store): Likewise.
3028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3029 Alan Hayward <alan.hayward@arm.com>
3030 David Sherwood <david.sherwood@arm.com>
3032 * coretypes.h (pod_mode): New type.
3033 (scalar_int_mode_pod): New typedef.
3034 * machmode.h (pod_mode): New class.
3035 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3036 * genmodes.c (emit_mode_int_n): Update accordingly.
3037 * lower-subreg.h (target_lower_subreg): Change type to
3038 scalar_int_mode_pod.
3039 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3040 scalar_int_mode_pod.
3042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3043 Alan Hayward <alan.hayward@arm.com>
3044 David Sherwood <david.sherwood@arm.com>
3046 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3047 machine_mode to scalar_int_mode.
3048 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3049 (rs6000_option_override_internal): Remove cast to int.
3050 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3051 machine_mode to scalar_int_mode.
3052 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3053 (rs6000_option_override_internal): Remove cast to int.
3054 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3055 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3057 * config/s390/s390.c (s390_expand_builtin): Likewise.
3058 * coretypes.h (scalar_int_mode): New type.
3059 (opt_scalar_int_mode): New typedef.
3060 * machmode.h (scalar_int_mode): New class.
3061 (scalar_int_mode::includes_p): New function.
3062 (byte_mode): Change type to scalar_int_mode.
3063 (word_mode): Likewise.
3064 (ptr_mode): Likewise.
3065 * emit-rtl.c (byte_mode): Likewise.
3066 (word_mode): Likewise.
3067 (ptr_mode): Likewise.
3068 (init_derived_machine_modes): Update accordingly.
3069 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3070 and MODE_PARTIAL_INT.
3071 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3072 opt_scalar_int_mode.
3074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3075 Alan Hayward <alan.hayward@arm.com>
3076 David Sherwood <david.sherwood@arm.com>
3078 * target.def (libgcc_floating_mode_supported_p): Take a
3080 * doc/tm.texi: Regenerate.
3081 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3083 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3084 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
3087 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3088 Alan Hayward <alan.hayward@arm.com>
3089 David Sherwood <david.sherwood@arm.com>
3091 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3092 * doc/tm.texi: Regenerate.
3093 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3094 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3095 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3096 * targhooks.h (default_floatn_mode): Likewise.
3097 * targhooks.c (default_floatn_mode): Likewise.
3098 * tree.c (build_common_tree_nodes): Update accordingly.
3100 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3101 Alan Hayward <alan.hayward@arm.com>
3102 David Sherwood <david.sherwood@arm.com>
3104 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3105 (mode_iterator::iterate_p): Likewise.
3106 (mode_iterator::get_wider): Likewise.
3107 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3110 Alan Hayward <alan.hayward@arm.com>
3111 David Sherwood <david.sherwood@arm.com>
3113 * coretypes.h (opt_scalar_float_mode): New typedef.
3114 * machmode.h (float_mode_for_size): New function.
3115 * emit-rtl.c (double_mode): Delete.
3116 (init_emit_once): Use float_mode_for_size.
3117 * stor-layout.c (layout_type): Likewise.
3118 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_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 * output.h (assemble_real): Take a scalar_float_mode.
3125 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3126 * config/arm/arm.md (consttable_4): Likewise.
3127 (consttable_8): Likewise.
3128 (consttable_16): Likewise.
3129 * config/mips/mips.md (consttable_float): Likewise.
3130 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3131 * varasm.c (assemble_real): Take a scalar_float_mode.
3132 (output_constant_pool_2): Update accordingly.
3133 (output_constant): Likewise.
3135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3136 Alan Hayward <alan.hayward@arm.com>
3137 David Sherwood <david.sherwood@arm.com>
3139 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3140 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3141 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3142 (native_encode_real): Likewise.
3143 (native_interpret_real): Likewise.
3144 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3145 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3148 Alan Hayward <alan.hayward@arm.com>
3149 David Sherwood <david.sherwood@arm.com>
3151 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3152 <scalar_float_mode>. Simplify.
3153 (gen_extend_conv_libfunc): Likewise.
3155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3156 Alan Hayward <alan.hayward@arm.com>
3157 David Sherwood <david.sherwood@arm.com>
3159 * coretypes.h (scalar_float_mode): New type.
3160 * machmode.h (mode_traits::from_int): Use machine_mode if
3161 USE_ENUM_MODES is defined.
3162 (is_a): New function.
3164 (dyn_cast): Likewise.
3165 (scalar_float_mode): New class.
3166 (scalar_float_mode::includes_p): New function.
3167 (is_float_mode): Likewise.
3168 * gdbhooks.py (MachineModePrinter): New class.
3169 (build_pretty_printer): Use it for scalar_float_mode.
3170 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3171 (format_helper::format_helper): Turn into a template.
3172 * genmodes.c (get_mode_class): New function.
3173 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3174 or machine_mode if none.
3175 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3176 as_a <scalar_float_mode>.
3177 * dwarf2out.c (mem_loc_descriptor): Likewise.
3178 (insert_float): Likewise.
3179 (add_const_value_attribute): Likewise.
3180 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3181 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3182 (expand_unop): Update accordingly.
3183 (expand_abs_nojump): Likewise.
3184 (expand_copysign_absneg): Take a scalar_float_mode.
3185 (expand_copysign_bit): Likewise.
3186 (expand_copysign): Update accordingly.
3188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3189 Alan Hayward <alan.hayward@arm.com>
3190 David Sherwood <david.sherwood@arm.com>
3192 * coretypes.h (opt_mode): New class.
3193 * machmode.h (opt_mode): Likewise.
3194 (opt_mode::else_void): New function.
3195 (opt_mode::require): Likewise.
3196 (opt_mode::exists): Likewise.
3197 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3198 (GET_MODE_2XWIDER_MODE): Likewise.
3199 (mode_iterator::get_wider): Update accordingly.
3200 (mode_iterator::get_2xwider): Likewise.
3201 (mode_iterator::get_known_wider): Likewise, turning into a template.
3202 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3203 forcing a wider mode to exist.
3204 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3205 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3206 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3207 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3208 * lower-subreg.c (init_lower_subreg): Likewise.
3209 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3210 on the final iteration.
3211 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3212 a wider mode exists before asking for a move pattern.
3213 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3214 forcing a wider mode to exist.
3215 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3216 returning false if no such mode exists.
3217 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3218 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3219 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3220 Avoid checking for a MODE_INT if we already know the mode is not a
3222 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3223 forcing a wider mode to exist.
3224 (expmed_mult_highpart_optab): Likewise.
3225 (expmed_mult_highpart): Likewise.
3226 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3228 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3229 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3230 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3231 * internal-fn.c (expand_mul_overflow): Update use of
3232 GET_MODE_2XWIDER_MODE.
3233 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3234 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3235 GET_MODE_WIDER_MODE.
3236 (convert_plusminus_to_widen): Likewise.
3237 * tree-switch-conversion.c (array_value_type): Likewise.
3238 * var-tracking.c (emit_note_insn_var_location): Likewise.
3239 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3240 Return false inside rather than outside the loop if no wider mode
3242 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3243 and GET_MODE_2XWIDER_MODE
3244 (can_compare_p): Use else_void.
3245 * gdbhooks.py (OptMachineModePrinter): New class.
3246 (build_pretty_printer): Use it for opt_mode.
3248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3249 Alan Hayward <alan.hayward@arm.com>
3250 David Sherwood <david.sherwood@arm.com>
3252 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3253 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3256 Alan Hayward <alan.hayward@arm.com>
3257 David Sherwood <david.sherwood@arm.com>
3259 * machmode.h (mode_traits): New structure.
3260 (get_narrowest_mode): New function.
3261 (mode_iterator::start): Likewise.
3262 (mode_iterator::iterate_p): Likewise.
3263 (mode_iterator::get_wider): Likewise.
3264 (mode_iterator::get_known_wider): Likewise.
3265 (mode_iterator::get_2xwider): Likewise.
3266 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3267 (FOR_EACH_MODE): Likewise.
3268 (FOR_EACH_MODE_FROM): Likewise.
3269 (FOR_EACH_MODE_UNTIL): Likewise.
3270 (FOR_EACH_WIDER_MODE): Likewise.
3271 (FOR_EACH_2XWIDER_MODE): Likewise.
3272 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3273 * combine.c (simplify_comparison): Likewise
3274 * config/i386/i386.c (type_natural_mode): Likewise.
3275 * cse.c (cse_insn): Likewise.
3276 * dse.c (find_shift_sequence): Likewise.
3277 * emit-rtl.c (init_derived_machine_modes): Likewise.
3278 (init_emit_once): Likewise.
3279 * explow.c (hard_function_value): Likewise.
3280 * expmed.c (extract_fixed_bit_field_1): Likewise.
3281 (extract_bit_field_1): Likewise.
3282 (expand_divmod): Likewise.
3283 (emit_store_flag_1): Likewise.
3284 * expr.c (init_expr_target): Likewise.
3285 (convert_move): Likewise.
3286 (alignment_for_piecewise_move): Likewise.
3287 (widest_int_mode_for_size): Likewise.
3288 (emit_block_move_via_movmem): Likewise.
3289 (copy_blkmode_to_reg): Likewise.
3290 (set_storage_via_setmem): Likewise.
3291 (compress_float_constant): Likewise.
3292 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3293 * optabs-query.c (get_best_extraction_insn): Likewise.
3294 * optabs.c (expand_binop): Likewise.
3295 (expand_twoval_unop): Likewise.
3296 (expand_twoval_binop): Likewise.
3297 (widen_leading): Likewise.
3298 (widen_bswap): Likewise.
3299 (expand_parity): Likewise.
3300 (expand_unop): Likewise.
3301 (prepare_cmp_insn): Likewise.
3302 (prepare_float_lib_cmp): Likewise.
3303 (expand_float): Likewise.
3304 (expand_fix): Likewise.
3305 (expand_sfix_optab): Likewise.
3306 * postreload.c (move2add_use_add2_insn): Likewise.
3307 * reg-stack.c (reg_to_stack): Likewise.
3308 * reginfo.c (choose_hard_reg_mode): Likewise.
3309 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3310 * stor-layout.c (mode_for_size): Likewise.
3311 (smallest_mode_for_size): Likewise.
3312 (mode_for_vector): Likewise.
3313 (finish_bitfield_representative): Likewise.
3314 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3315 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3316 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3317 * var-tracking.c (prepare_call_arguments): Likewise.
3319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3320 Alan Hayward <alan.hayward@arm.com>
3321 David Sherwood <david.sherwood@arm.com>
3323 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3324 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3325 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3326 * machmode.h (mode_size): Move earlier in file.
3327 (mode_precision): Likewise.
3328 (mode_inner): Likewise.
3329 (mode_nunits): Likewise.
3330 (mode_unit_size): Likewise.
3331 (unit_unit_precision): Likewise.
3332 (mode_wider): Likewise.
3333 (mode_2xwider): Likewise.
3334 (machine_mode): New class.
3335 (mode_to_bytes): New function.
3336 (mode_to_bits): Likewise.
3337 (mode_to_precision): Likewise.
3338 (mode_to_inner): Likewise.
3339 (mode_to_unit_size): Likewise.
3340 (mode_to_unit_precision): Likewise.
3341 (mode_to_nunits): Likewise.
3342 (GET_MODE_SIZE): Use mode_to_bytes.
3343 (GET_MODE_BITSIZE): Use mode_to_bits.
3344 (GET_MODE_PRECISION): Use mode_to_precision.
3345 (GET_MODE_INNER): Use mode_to_inner.
3346 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3347 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3348 (GET_MODE_NUNITS): Use mode_to_nunits.
3349 * system.h (ALWAYS_INLINE): New macro.
3350 * config/powerpcspe/powerpcspe-c.c
3351 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3352 int for arg1_mode and arg2_mode.
3354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3355 Alan Hayward <alan.hayward@arm.com>
3356 David Sherwood <david.sherwood@arm.com>
3358 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3359 Prefix mode names with E_ in case statements.
3360 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3361 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3362 (aarch64_split_simd_move): Likewise.
3363 (aarch64_gen_storewb_pair): Likewise.
3364 (aarch64_gen_loadwb_pair): Likewise.
3365 (aarch64_gen_store_pair): Likewise.
3366 (aarch64_gen_load_pair): Likewise.
3367 (aarch64_get_condition_code_1): Likewise.
3368 (aarch64_constant_pool_reload_icode): Likewise.
3369 (get_rsqrte_type): Likewise.
3370 (get_rsqrts_type): Likewise.
3371 (get_recpe_type): Likewise.
3372 (get_recps_type): Likewise.
3373 (aarch64_gimplify_va_arg_expr): Likewise.
3374 (aarch64_simd_container_mode): Likewise.
3375 (aarch64_emit_load_exclusive): Likewise.
3376 (aarch64_emit_store_exclusive): Likewise.
3377 (aarch64_expand_compare_and_swap): Likewise.
3378 (aarch64_gen_atomic_cas): Likewise.
3379 (aarch64_emit_bic): Likewise.
3380 (aarch64_emit_atomic_swap): Likewise.
3381 (aarch64_emit_atomic_load_op): Likewise.
3382 (aarch64_evpc_trn): Likewise.
3383 (aarch64_evpc_uzp): Likewise.
3384 (aarch64_evpc_zip): Likewise.
3385 (aarch64_evpc_ext): Likewise.
3386 (aarch64_evpc_rev): Likewise.
3387 (aarch64_evpc_dup): Likewise.
3388 (aarch64_gen_ccmp_first): Likewise.
3389 (aarch64_gen_ccmp_next): Likewise.
3390 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3391 (alpha_emit_xfloating_libcall): Likewise.
3392 (emit_insxl): Likewise.
3393 (alpha_arg_type): Likewise.
3394 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3395 (arc_preferred_simd_mode): Likewise.
3396 (arc_secondary_reload): Likewise.
3397 (get_arc_condition_code): Likewise.
3398 (arc_print_operand): Likewise.
3399 (arc_legitimate_constant_p): Likewise.
3400 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3401 * config/arc/arc.md (casesi_load): Likewise.
3402 (casesi_compact_jump): Likewise.
3403 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3404 (cc_use_register): Likewise.
3405 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3406 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3407 (arm_init_iwmmxt_builtins): Likewise.
3408 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3409 (neon_expand_vector_init): Likewise.
3410 (arm_attr_length_move_neon): Likewise.
3411 (maybe_get_arm_condition_code): Likewise.
3412 (arm_emit_vector_const): Likewise.
3413 (arm_preferred_simd_mode): Likewise.
3414 (arm_output_iwmmxt_tinsr): Likewise.
3415 (thumb1_output_casesi): Likewise.
3416 (thumb2_output_casesi): Likewise.
3417 (arm_emit_load_exclusive): Likewise.
3418 (arm_emit_store_exclusive): Likewise.
3419 (arm_expand_compare_and_swap): Likewise.
3420 (arm_evpc_neon_vuzp): Likewise.
3421 (arm_evpc_neon_vzip): Likewise.
3422 (arm_evpc_neon_vrev): Likewise.
3423 (arm_evpc_neon_vtrn): Likewise.
3424 (arm_evpc_neon_vext): Likewise.
3425 (arm_validize_comparison): Likewise.
3426 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3427 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3428 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3429 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3430 (c6x_preferred_simd_mode): Likewise.
3431 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3432 (epiphany_rtx_costs): Likewise.
3433 * config/epiphany/predicates.md (proper_comparison_operator):
3435 * config/frv/frv.c (condexec_memory_operand): Likewise.
3436 (frv_emit_move): Likewise.
3437 (output_move_single): Likewise.
3438 (output_condmove_single): Likewise.
3439 (frv_hard_regno_mode_ok): Likewise.
3440 (frv_matching_accg_mode): Likewise.
3441 * config/h8300/h8300.c (split_adds_subs): Likewise.
3442 (h8300_rtx_costs): Likewise.
3443 (h8300_print_operand): Likewise.
3444 (compute_mov_length): Likewise.
3445 (output_logical_op): Likewise.
3446 (compute_logical_op_length): Likewise.
3447 (compute_logical_op_cc): Likewise.
3448 (h8300_shift_needs_scratch_p): Likewise.
3449 (output_a_shift): Likewise.
3450 (compute_a_shift_length): Likewise.
3451 (compute_a_shift_cc): Likewise.
3452 (expand_a_rotate): Likewise.
3453 (output_a_rotate): Likewise.
3454 * config/i386/i386.c (classify_argument): Likewise.
3455 (function_arg_advance_32): Likewise.
3456 (function_arg_32): Likewise.
3457 (function_arg_64): Likewise.
3458 (function_value_64): Likewise.
3459 (ix86_gimplify_va_arg): Likewise.
3460 (ix86_legitimate_constant_p): Likewise.
3461 (put_condition_code): Likewise.
3462 (split_double_mode): Likewise.
3463 (ix86_avx256_split_vector_move_misalign): Likewise.
3464 (ix86_expand_vector_logical_operator): Likewise.
3465 (ix86_split_idivmod): Likewise.
3466 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3467 (ix86_build_const_vector): Likewise.
3468 (ix86_build_signbit_mask): Likewise.
3469 (ix86_match_ccmode): Likewise.
3470 (ix86_cc_modes_compatible): Likewise.
3471 (ix86_expand_branch): Likewise.
3472 (ix86_expand_sse_cmp): Likewise.
3473 (ix86_expand_sse_movcc): Likewise.
3474 (ix86_expand_int_sse_cmp): Likewise.
3475 (ix86_expand_vec_perm_vpermi2): Likewise.
3476 (ix86_expand_vec_perm): Likewise.
3477 (ix86_expand_sse_unpack): Likewise.
3478 (ix86_expand_int_addcc): Likewise.
3479 (ix86_split_to_parts): Likewise.
3480 (ix86_vectorize_builtin_gather): Likewise.
3481 (ix86_vectorize_builtin_scatter): Likewise.
3482 (avx_vpermilp_parallel): Likewise.
3483 (inline_memory_move_cost): Likewise.
3484 (ix86_tieable_integer_mode_p): Likewise.
3485 (x86_maybe_negate_const_int): Likewise.
3486 (ix86_expand_vector_init_duplicate): Likewise.
3487 (ix86_expand_vector_init_one_nonzero): Likewise.
3488 (ix86_expand_vector_init_one_var): Likewise.
3489 (ix86_expand_vector_init_concat): Likewise.
3490 (ix86_expand_vector_init_interleave): Likewise.
3491 (ix86_expand_vector_init_general): Likewise.
3492 (ix86_expand_vector_set): Likewise.
3493 (ix86_expand_vector_extract): Likewise.
3494 (emit_reduc_half): Likewise.
3495 (ix86_emit_i387_round): Likewise.
3496 (ix86_mangle_type): Likewise.
3497 (ix86_expand_round_sse4): Likewise.
3498 (expand_vec_perm_blend): Likewise.
3499 (canonicalize_vector_int_perm): Likewise.
3500 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3501 (expand_vec_perm_1): Likewise.
3502 (expand_vec_perm_interleave3): Likewise.
3503 (expand_vec_perm_even_odd_pack): Likewise.
3504 (expand_vec_perm_even_odd_1): Likewise.
3505 (expand_vec_perm_broadcast_1): Likewise.
3506 (ix86_vectorize_vec_perm_const_ok): Likewise.
3507 (ix86_expand_vecop_qihi): Likewise.
3508 (ix86_expand_mul_widen_hilo): Likewise.
3509 (ix86_expand_sse2_abs): Likewise.
3510 (ix86_expand_pextr): Likewise.
3511 (ix86_expand_pinsr): Likewise.
3512 (ix86_preferred_simd_mode): Likewise.
3513 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3514 * config/i386/sse.md (*andnot<mode>3): Likewise.
3515 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3516 (*<code><mode>3): Likewise.
3517 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3518 (ia64_expand_atomic_op): Likewise.
3519 (ia64_arg_type): Likewise.
3520 (ia64_mode_to_int): Likewise.
3521 (ia64_scalar_mode_supported_p): Likewise.
3522 (ia64_vector_mode_supported_p): Likewise.
3523 (expand_vec_perm_broadcast): Likewise.
3524 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3525 (iq2000_function_arg_advance): Likewise.
3526 (iq2000_function_arg): Likewise.
3527 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3528 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3529 (m68k_libcall_value): Likewise.
3530 (m68k_function_value): Likewise.
3531 (sched_attr_op_type): Likewise.
3532 * config/mcore/mcore.c (mcore_output_move): Likewise.
3533 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3535 (microblaze_function_arg): Likewise.
3536 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3537 (mips_print_operand): Likewise.
3538 (mips_mode_ok_for_mov_fmt_p): Likewise.
3539 (mips_vector_mode_supported_p): Likewise.
3540 (mips_preferred_simd_mode): Likewise.
3541 (mips_expand_vpc_loongson_even_odd): Likewise.
3542 (mips_expand_vec_unpack): Likewise.
3543 (mips_expand_vi_broadcast): Likewise.
3544 (mips_expand_vector_init): Likewise.
3545 (mips_expand_vec_reduc): Likewise.
3546 (mips_expand_msa_cmp): Likewise.
3547 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3548 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3549 (cc_flags_for_mode): Likewise.
3550 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3551 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3552 (nds32_output_casesi_pc_relative): Likewise.
3553 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3554 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3555 (nvptx_gen_unpack): Likewise.
3556 (nvptx_gen_pack): Likewise.
3557 (nvptx_gen_shuffle): Likewise.
3558 (nvptx_gen_wcast): Likewise.
3559 (nvptx_preferred_simd_mode): Likewise.
3560 * config/pa/pa.c (pa_secondary_reload): Likewise.
3561 * config/pa/predicates.md (base14_operand): Likewise.
3562 * config/powerpcspe/powerpcspe-c.c
3563 (altivec_resolve_overloaded_builtin): Likewise.
3564 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3566 (rs6000_preferred_simd_mode): Likewise.
3567 (output_vec_const_move): Likewise.
3568 (rs6000_expand_vector_extract): Likewise.
3569 (rs6000_split_vec_extract_var): Likewise.
3570 (reg_offset_addressing_ok_p): Likewise.
3571 (rs6000_legitimate_offset_address_p): Likewise.
3572 (rs6000_legitimize_address): Likewise.
3573 (rs6000_emit_set_const): Likewise.
3574 (rs6000_const_vec): Likewise.
3575 (rs6000_emit_move): Likewise.
3576 (spe_build_register_parallel): Likewise.
3577 (rs6000_darwin64_record_arg_recurse): Likewise.
3578 (swap_selector_for_mode): Likewise.
3579 (spe_init_builtins): Likewise.
3580 (paired_init_builtins): Likewise.
3581 (altivec_init_builtins): Likewise.
3582 (do_load_for_compare): Likewise.
3583 (rs6000_generate_compare): Likewise.
3584 (rs6000_expand_float128_convert): Likewise.
3585 (emit_load_locked): Likewise.
3586 (emit_store_conditional): Likewise.
3587 (rs6000_output_function_epilogue): Likewise.
3588 (rs6000_handle_altivec_attribute): Likewise.
3589 (rs6000_function_value): Likewise.
3590 (emit_fusion_gpr_load): Likewise.
3591 (emit_fusion_p9_load): Likewise.
3592 (emit_fusion_p9_store): Likewise.
3593 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3594 (fusion_gpr_mem_load): Likewise.
3595 (fusion_addis_mem_combo_load): Likewise.
3596 (fusion_addis_mem_combo_store): Likewise.
3597 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3598 (fusion_gpr_mem_load): Likewise.
3599 (fusion_addis_mem_combo_load): Likewise.
3600 (fusion_addis_mem_combo_store): Likewise.
3601 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3603 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3604 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3605 (rs6000_preferred_simd_mode): Likewise.
3606 (output_vec_const_move): Likewise.
3607 (rs6000_expand_vector_extract): Likewise.
3608 (rs6000_split_vec_extract_var): Likewise.
3609 (reg_offset_addressing_ok_p): Likewise.
3610 (rs6000_legitimate_offset_address_p): Likewise.
3611 (rs6000_legitimize_address): Likewise.
3612 (rs6000_emit_set_const): Likewise.
3613 (rs6000_const_vec): Likewise.
3614 (rs6000_emit_move): Likewise.
3615 (rs6000_darwin64_record_arg_recurse): Likewise.
3616 (swap_selector_for_mode): Likewise.
3617 (paired_init_builtins): Likewise.
3618 (altivec_init_builtins): Likewise.
3619 (rs6000_expand_float128_convert): Likewise.
3620 (emit_load_locked): Likewise.
3621 (emit_store_conditional): Likewise.
3622 (rs6000_output_function_epilogue): Likewise.
3623 (rs6000_handle_altivec_attribute): Likewise.
3624 (rs6000_function_value): Likewise.
3625 (emit_fusion_gpr_load): Likewise.
3626 (emit_fusion_p9_load): Likewise.
3627 (emit_fusion_p9_store): Likewise.
3628 * config/rx/rx.c (rx_gen_move_template): Likewise.
3629 (flags_from_mode): Likewise.
3630 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3631 (s390_slb_comparison): Likewise.
3632 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3633 (s390_vector_mode_supported_p): Likewise.
3634 (s390_cc_modes_compatible): Likewise.
3635 (s390_match_ccmode_set): Likewise.
3636 (s390_canonicalize_comparison): Likewise.
3637 (s390_emit_compare_and_swap): Likewise.
3638 (s390_branch_condition_mask): Likewise.
3639 (s390_rtx_costs): Likewise.
3640 (s390_secondary_reload): Likewise.
3641 (__SECONDARY_RELOAD_CASE): Likewise.
3642 (s390_expand_cs): Likewise.
3643 (s390_preferred_simd_mode): Likewise.
3644 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3645 * config/sh/sh.c (sh_print_operand): Likewise.
3646 (dump_table): Likewise.
3647 (sh_secondary_reload): Likewise.
3648 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3649 * config/sh/sh.md (casesi_worker_1): Likewise.
3650 (casesi_worker_2): Likewise.
3651 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3652 (fcc_comparison_operator): Likewise.
3653 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3654 (emit_soft_tfmode_cvt): Likewise.
3655 (sparc_preferred_simd_mode): Likewise.
3656 (output_cbranch): Likewise.
3657 (sparc_print_operand): Likewise.
3658 (sparc_expand_vec_perm_bmask): Likewise.
3659 (vector_init_bshuffle): Likewise.
3660 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3661 (spu_vector_mode_supported_p): Likewise.
3662 (spu_expand_insv): Likewise.
3663 (spu_emit_branch_or_set): Likewise.
3664 (spu_handle_vector_attribute): Likewise.
3665 (spu_builtin_splats): Likewise.
3666 (spu_builtin_extract): Likewise.
3667 (spu_builtin_promote): Likewise.
3668 (spu_expand_sign_extend): Likewise.
3669 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3670 (tilegx_simd_int): Likewise.
3671 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3672 (tilepro_simd_int): Likewise.
3673 * config/v850/v850.c (const_double_split): Likewise.
3674 (v850_print_operand): Likewise.
3675 (ep_memory_offset): Likewise.
3676 * config/vax/vax.c (vax_rtx_costs): Likewise.
3677 (vax_output_int_move): Likewise.
3678 (vax_output_int_add): Likewise.
3679 (vax_output_int_subtract): Likewise.
3680 * config/visium/predicates.md (visium_branch_operator): Likewise.
3681 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3682 (visium_print_operand_address): Likewise.
3683 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3684 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3685 (xtensa_expand_conditional_branch): Likewise.
3686 (xtensa_copy_incoming_a7): Likewise.
3687 (xtensa_output_literal): Likewise.
3688 * dfp.c (decimal_real_maxval): Likewise.
3689 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3692 Alan Hayward <alan.hayward@arm.com>
3693 David Sherwood <david.sherwood@arm.com>
3695 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3696 (mode_nunits_inline): Likewise.
3697 (mode_inner_inline): Likewise.
3698 (mode_unit_size_inline): Likewise.
3699 (mode_unit_precision_inline): Likewise.
3700 (emit_insn_modes_h): Likewise. Also emit a #define of the
3702 (emit_mode_wider): Add an E_ prefix to mode names.
3703 (emit_mode_complex): Likewise.
3704 (emit_mode_inner): Likewise.
3705 (emit_mode_adjustments): Likewise.
3706 (emit_mode_int_n): Likewise.
3707 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3708 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3709 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3710 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3711 (CRC32_BUILTIN, ENTRY): Likewise.
3712 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3713 (aarch64_pop_regs): Likewise.
3714 (aarch64_process_components): Likewise.
3715 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3716 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3717 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3718 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3719 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3720 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3722 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3723 (pcmpistr): Likewise.
3724 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3725 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3726 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3728 * config/rl78/rl78.c (mduc_regs): Likewise.
3729 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3730 (htm_expand_builtin): Likewise.
3731 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3732 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3733 * config/xtensa/xtensa.c (print_operand): Likewise.
3734 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3735 (NUM_MODE_VECTOR_INT): Likewise.
3736 * genoutput.c (null_operand): Likewise.
3737 (output_operand_data): Likewise.
3738 * genrecog.c (print_parameter_value): Likewise.
3739 * lra.c (debug_operand_data): Likewise.
3741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3743 * dwarf2out.c (scompare_loc_descriptor_wide)
3744 (scompare_loc_descriptor_narrow): New functions, split out from...
3745 (scompare_loc_descriptor): ...here.
3746 * expmed.c (emit_store_flag_int): New function, split out from...
3747 (emit_store_flag): ...here.
3749 2017-08-30 Richard Biener <rguenther@suse.de>
3751 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3752 (dwarf2out_early_finish): Move setting of AT_pubnames from
3753 early debug output to early finish.
3755 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3757 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3758 and -mdata-region to the assembler.
3760 2017-08-30 Richard Biener <rguenther@suse.de>
3762 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3764 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3765 (dwarf2out_early_global_decl): Only generate a DIE for the
3766 abstract origin if it doesn't already exist or is a declaration DIE.
3767 (resolve_addr): Do not add the linkage name twice when
3768 generating a stub DIE for the DW_TAG_GNU_call_site target.
3770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3772 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3773 Use machine_mode rather than int for arg1_mode.
3775 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3778 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3779 that the second argument of the built-in functions to unpack
3780 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3781 switch statement instead a lot of if statements.
3782 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3783 Allow 64-bit values to be in Altivec registers as well as
3784 traditional floating point registers.
3785 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3787 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3789 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3790 MAX_REGS_PER_ADDRESS == 1.
3792 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3794 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3795 * config/i386/i386.c (ix86_option_override_internal): Simplify
3796 setting of opts->x_flag_entry.
3798 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3799 Jakub Jelinek <jakub@redhat.com>
3800 Richard Biener <rguenther@suse.de>
3802 PR tree-optimization/81503
3803 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3804 folded constant fits in the target type; reorder tests for clarity.
3806 2017-08-29 Martin Liska <mliska@suse.cz>
3808 * passes.def: Include pass_lower_switch.
3809 * stmt.c (dump_case_nodes): Remove and move to
3810 tree-switch-conversion.
3811 (case_values_threshold): Likewise.
3812 (expand_switch_as_decision_tree_p): Likewise.
3813 (emit_case_decision_tree): Likewise.
3814 (expand_case): Likewise.
3815 (balance_case_nodes): Likewise.
3816 (node_has_low_bound): Likewise.
3817 (node_has_high_bound): Likewise.
3818 (node_is_bounded): Likewise.
3819 (emit_case_nodes): Likewise.
3820 (struct simple_case_node): New struct.
3821 (add_case_node): Remove.
3822 (emit_case_dispatch_table): Use vector instead of case_list.
3823 (reset_out_edges_aux): Remove.
3824 (compute_cases_per_edge): Likewise.
3825 (expand_case): Build list of simple_case_node.
3826 (expand_sjlj_dispatch_table): Use it.
3827 * tree-switch-conversion.c (struct case_node): Moved from
3828 stmt.c and adjusted.
3829 (emit_case_nodes): Likewise.
3830 (node_has_low_bound): Likewise.
3831 (node_has_high_bound): Likewise.
3832 (node_is_bounded): Likewise.
3833 (case_values_threshold): Likewise.
3834 (reset_out_edges_aux): Likewise.
3835 (compute_cases_per_edge): Likewise.
3836 (add_case_node): Likewise.
3837 (dump_case_nodes): Likewise.
3838 (balance_case_nodes): Likewise.
3839 (expand_switch_as_decision_tree_p): Likewise.
3840 (emit_jump): Likewise.
3841 (emit_case_decision_tree): Likewise.
3842 (try_switch_expansion): Likewise.
3843 (do_jump_if_equal): Likewise.
3844 (emit_cmp_and_jump_insns): Likewise.
3845 (fix_phi_operands_for_edge): New function.
3846 (record_phi_operand_mapping): Likewise.
3847 (class pass_lower_switch): New pass.
3848 (pass_lower_switch::execute): New function.
3849 (make_pass_lower_switch): Likewise.
3850 (conditional_probability):
3851 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3852 * tree-pass.h: Add make_pass_lower_switch.
3854 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3857 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3860 2017-08-29 Richard Biener <rguenther@suse.de>
3862 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3863 we do not add a DW_AT_inline attribute twice.
3864 (gen_subprogram_die): Remove code setting DW_AT_inline on
3865 DECL_ABSTRACT_P nodes.
3867 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3869 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3870 calls to internal functions.
3871 (gimplify_modify_expr): Likewise.
3872 * tree-call-cdce.c (use_internal_fn): Likewise.
3873 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3874 (convert_to_divmod): Set the nothrow flag.
3875 * tree-if-conv.c (predicate_mem_writes): Likewise.
3876 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3877 (vectorizable_call): Likewise.
3878 (vectorizable_store): Likewise.
3879 (vectorizable_load): Likewise.
3880 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3881 (vect_recog_mask_conversion_pattern): Likewise.
3883 2017-08-29 Martin Liska <mliska@suse.cz>
3886 * gcc.c (driver_handle_option): Add new argument.
3887 * opts-common.c (handle_option): Pass
3888 target_option_override_hook.
3889 * opts-global.c (lang_handle_option): Add new option.
3890 (set_default_handlers): Add new argument.
3891 (decode_options): Likewise.
3892 * opts.c (target_handle_option): Likewise.
3893 (common_handle_option): Call target_option_override_hook.
3894 * opts.h (struct cl_option_handler_func): Add hook for
3895 target option override.
3896 (struct cl_option_handlers): Likewise.
3897 (set_default_handlers): Add new argument.
3898 (decode_options): Likewise.
3899 (common_handle_option): Likewise.
3900 (target_handle_option): Likewise.
3901 * toplev.c (toplev::main): Pass targetm.target_option.override
3904 2017-08-29 Richard Biener <rguenther@suse.de>
3905 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3907 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3908 life to the active subtree.
3910 2017-08-28 Jeff Law <law@redhat.com>
3912 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3913 derive_equivalences.
3914 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3916 (edge_info::derive_equivalences): New private member function
3918 * tree-ssa-dom.c (class edge_info): Changed from a struct
3919 to a class. Add ctor/dtor, methods and data members.
3920 (edge_info::edge_info): Renamed from allocate_edge_info.
3921 Initialize additional members.
3922 (edge_info::~edge_info): New.
3923 (free_dom_edge_info): Delete the edge info.
3924 (record_edge_info): Use new class & associated member functions.
3925 Tighten forms for testing for edge equivalences.
3926 (record_temporary_equivalences): Iterate over the simple
3927 equivalences rather than assuming there's only one per edge.
3928 (cprop_into_successor_phis): Iterate over the simple
3929 equivalences rather than assuming there's only one per edge.
3930 (optimize_stmt): Use operand_equal_p rather than pointer
3931 equality for mini-DSE code.
3933 2017-08-28 Nathan Sidwell <nathan@acm.org>
3935 * gcc.c (execute): Fold SIGPIPE handling into switch
3936 statement. Adjust internal error message.
3938 2017-08-28 Richard Biener <rguenther@suse.de>
3941 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3942 Do nothing for removed DIEs.
3944 2017-08-28 Richard Biener <rguenther@suse.de>
3946 PR tree-optimization/81977
3947 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3950 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3953 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3954 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3955 using targetm.gen_mem_thread_fence.
3957 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3960 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3961 predicate to register_operand. Reorder operands.
3962 (*btr<mode>): Ditto.
3963 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3964 (*btr<mode>_mask): Ditto.
3966 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3968 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3969 * config/rs6000/xmmintrin.h: New file.
3970 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3972 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3975 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3976 parameter and_insn and return it.
3977 (recombine_lvx_pattern): Insert a copy to ensure availability of
3978 the base register of the copied masking operation at the point of
3979 the instruction replacement.
3980 (recombine_stvx_pattern): Likewise.
3982 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3984 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3985 undocumented switches.
3986 (-mpower9-dform-vector): Likewise.
3987 (-mpower9-dform): Likewise.
3988 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3989 comments to delete references to -mpower9-dform* switches.
3990 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3991 Delete reference to -mpower9-dform* switches, test for
3992 -mpower9-vector instead.
3993 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3994 (OTHER_P9_VECTOR_MASKS): Likewise.
3995 (POWERPC_MASKS): Likewise.
3996 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3997 tests against -mpower9-dform* to -mpower9-vector. Delete code
3998 that checked for -mpower9-dform* consistancy with other options.
3999 Add test for -mpower9-misc to enable other power9 switches.
4000 (rs6000_init_hard_regno_mode_ok): Likewise.
4001 (rs6000_option_override_internal): Likewise.
4002 (rs6000_emit_prologue): Likewise.
4003 (rs6000_emit_epilogue): Likewise.
4004 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4005 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4006 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4008 (emit_fusion_p9_store): Likewise.
4009 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4010 resetting these macros if the assembler does not support ISA 3.0
4012 (TARGET_P9_DFORM_VECTOR): Likewise.
4013 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4014 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4016 2017-08-25 Alan Modra <amodra@gmail.com>
4018 PR rtl-optimization/81747
4019 * cse.c (cse_extended_basic_block): Don't attempt to record
4020 equivalences for degenerate conditional branches that branch
4021 to their fall-through.
4023 2017-08-24 Martin Sebor <msebor@redhat.com>
4026 * gimple-fold.c (size_must_be_zero_p): New function.
4027 (gimple_fold_builtin_memory_op): Call it.
4029 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4031 * config/rs6000/mm_malloc.h: New file.
4033 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4035 PR tree-optimization/81913
4036 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4037 analysis when either IVs in condition can wrap.
4039 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4041 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4042 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4044 2017-08-24 Richard Biener <rguenther@suse.de>
4047 * targhooks.c (default_target_can_inline_p): Properly
4048 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4049 is present and always compare.
4050 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4051 infer -mfpmath=sse from TARGET_SSE_P.
4052 (ix86_can_inline_p): Properly use target_option_default_node when
4053 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4055 2017-08-24 Richard Biener <rguenther@suse.de>
4058 * dwarf2out.c (output_die): Handle flag_generate_offload like
4060 (output_comp_unit): Likewise.
4061 (gen_array_type_die): Likewise.
4062 (dwarf2out_early_finish): Likewise.
4063 (note_variable_value_in_expr): Likewise.
4064 (dwarf2out_finish): Likewise. Adjust assert.
4065 * cgraphunit.c (symbol_table::compile): Move setting of
4066 flag_generate_offload earlier ...
4067 (symbol_table::finalize_compilation_unit): ... here, before
4068 early debug finalization.
4070 2017-08-24 Richard Biener <rguenther@suse.de>
4072 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4073 and ipa-fnsummary.h.
4074 (ix86_can_inline_p): When ix86_fpmath flags do not match
4075 check whether the callee uses FP math at all.
4077 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4080 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4081 instead of TYPE_PRECISION.
4083 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4084 Alan Hayward <alan.hayward@arm.com>
4085 David Sherwood <david.sherwood@arm.com>
4087 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4089 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4091 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4093 (ix86_frame::stack_realign_allocate): New field.
4094 (struct machine_frame_state): Modify comments.
4095 (machine_frame_state::sp_realigned_fp_end): New field.
4096 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4098 (sp_valid_at): Add assertion to assure no attempt to access invalid
4099 offset of a realigned stack.
4100 (fp_valid_at): Likewise.
4101 (choose_baseaddr): Modify comments.
4102 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4103 ix86_expand_prologue.
4104 (ix86_expand_prologue): Modify stack realignment and allocation.
4105 (ix86_expand_epilogue): Modify comments.
4106 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4107 avx2_runtime, avx512f, and avx512f_runtime.
4109 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4111 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4112 (mstackrealign): Do not init to -1.
4113 * config/i386/i386.c (ix86_option_override_internal):
4114 Check opts_set, not opts when setting default value of
4115 opts->x_ix86_force_align_arg_pointer.
4117 2017-08-23 Richard Biener <rguenther@suse.de>
4119 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4120 lang_hooks.decl_printable_name.
4121 * print-rtl-function.c (print_rtx_function): Likewise.
4122 * tree-pretty-print.c (dump_function_header): Likewise.
4124 2017-08-23 Richard Biener <rguenther@suse.de>
4127 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4128 -g0 at compile-time.
4130 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4133 * doc/sourcebuild.texi (Other hardware attributes):
4136 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4139 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4140 Fix single-use check.
4142 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4144 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4146 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4148 * doc/install.texi: Modify to add more details on running selected
4151 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4153 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4154 is combined with -mabi=ms.
4155 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4158 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4160 PR tree-optimization/81488
4161 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4162 and cached_basis fields.
4163 (MAX_SPREAD): New constant.
4164 (alloc_cand_and_find_basis): Initialize new fields.
4165 (clear_visited): New function.
4166 (create_phi_basis_1): Rename from create_phi_basis, set visited
4167 and cached_basis fields.
4168 (create_phi_basis): New wrapper function.
4169 (phi_add_costs_1): Rename from phi_add_costs, add spread
4170 parameter, set visited field, short-circuit when limits reached.
4171 (phi_add_costs): New wrapper function.
4172 (record_phi_increments_1): Rename from record_phi_increments, set
4174 (record_phi_increments): New wrapper function.
4175 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4176 (phi_incr_cost): New wrapper function.
4177 (all_phi_incrs_profitable_1): Rename from
4178 all_phi_incrs_profitable, set visited field.
4179 (all_phi_incrs_profitable): New wrapper function.
4181 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4182 Alan Hayward <alan.hayward@arm.com>
4183 David Sherwood <david.sherwood@arm.com>
4185 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4186 that takes the outer and inner modes.
4187 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4188 comparison as the canonical test for a paradoxical subreg.
4189 * combine.c (simplify_set): Use paradoxical_subreg_p.
4190 (make_extraction): Likewise.
4191 (force_to_mode): Likewise.
4192 (rtx_equal_for_field_assignment_p): Likewise.
4193 (gen_lowpart_for_combine): Likewise.
4194 (simplify_comparison): Likewise.
4195 * cse.c (equiv_constant): Likewise.
4196 * expmed.c (store_bit_field_1): Likewise.
4197 * final.c (alter_subreg): Likewise.
4198 * fwprop.c (propagate_rtx): Likewise.
4199 (forward_propagate_subreg): Likewise.
4200 * ira-conflicts.c (ira_build_conflicts): Likewise.
4201 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4202 * lra-constraints.c (curr_insn_transform): Likewise.
4203 (split_reg): Likewise.
4204 * lra-eliminations.c (move_plus_up): Likewise.
4205 (lra_eliminate_regs_1): Likewise.
4206 * recog.c (general_operand): Likewise.
4207 * ree.c (combine_reaching_defs): Likewise.
4208 * reload.c (push_reload): Likewise.
4209 (find_reloads): Likewise.
4210 * reload1.c (elimination_effects): Likewise.
4211 (compute_reload_subreg_offset): Likewise.
4212 (choose_reload_regs): Likewise.
4213 * rtlanal.c (subreg_lsb_1): Likewise.
4214 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4215 (simplify_subreg): Likewise.
4216 * var-tracking.c (track_loc_p): Likewise.
4217 * emit-rtl.c (byte_lowpart_offset): Likewise.
4218 (paradoxical_subreg_p): Delete out-of-line definition.
4220 2017-08-22 Jeff Law <law@redhat.com>
4222 PR tree-optimization/81741
4223 PR tree-optimization/71947
4224 * tree-ssa-dom.c: Include tree-inline.h.
4225 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4226 equivalences if one is more expensive to compute than the other.
4227 * tree-ssa-scopedtables.h (class const_or_copies): Make
4228 record_const_or_copy_raw method private.
4229 (class avail_exprs_stack): New method simplify_binary_operation.
4230 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4231 avail_exprs_stack::simplify_binary_operation as needed.
4232 (avail_exprs_stack::simplify_binary_operation): New function.
4234 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4236 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4237 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4238 (DOT_SYMBOLS): Likewise.
4239 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4240 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4241 (RS6000_ABI_NAME): Likewise.
4242 (TARGET_CMODEL): Likewise.
4243 (TOC_SECTION_ASM_OP): Likewise.
4244 (SET_CMODEL): New macro.
4245 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4247 2017-08-22 Richard Biener <rguenther@suse.de>
4249 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4250 to allow for free-lang-data replacements similar to verify_type_variant.
4252 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4253 Alan Hayward <alan.hayward@arm.com>
4254 David Sherwood <david.sherwood@arm.com>
4256 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4257 VOIDmode for the LABEL_REF.
4259 2017-08-22 Richard Biener <rguenther@suse.de>
4261 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4263 2017-08-22 Richard Biener <rguenther@suse.de>
4265 * common.opt (feliminate-dwarf2-dups): Ignore.
4266 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4267 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4268 same_die_p_wrap, compute_section_prefix,
4269 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4270 (comdat_symbol_id, comdat_symbol_number): Likewise.
4271 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4273 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4274 (output_die): Mark unreachable path unreachable.
4275 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4276 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4277 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4278 (dwarf2out_early_finish): Likewise.
4280 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4282 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4284 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4287 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4288 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4289 (avr_attribute_table) <io, io_low, address>: Initialize
4290 .decl_required with true.
4292 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4294 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4295 undocumented debugging options.
4296 (-mvsx-scalar-double): Likewise.
4297 (-mallow-df-permute): Likewise.
4298 (-mvectorize-builtins): Likewise.
4299 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4300 (rs6000_builtin_vectorized_function): Likewise.
4301 (rs6000_builtin_md_vectorized_function): Likewise.
4302 (rs6000_opt_vars): Likewise.
4304 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4307 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4308 (*btrq_imm): Rename from *btrq.
4309 (*btcq_imm): Rename from *btcq.
4310 (btsc): New code attribute.
4311 (*<btsc><mode>): New insn pattern.
4312 (*btr<mode>): Ditto.
4313 (*<btsc><mode>_mask): New insn_and_split pattern.
4314 (*btr<mode>_mask): Ditto.
4316 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4317 Alan Hayward <alan.hayward@arm.com>
4318 David Sherwood <david.sherwood@arm.com>
4320 * function.c (pad_below): Simplify padding calculation.
4322 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4323 Alan Hayward <alan.hayward@arm.com>
4324 David Sherwood <david.sherwood@arm.com>
4326 * target.def (function_prologue): Remove frame size argument.
4327 (function_epilogue): Likewise.
4328 * doc/tm.texi: Regenerate.
4329 * final.c (final_start_function): Update call to function_prologue.
4330 (final_end_function): Update call to function_epilogue.
4331 (default_function_pro_epilogue): Remove frame size argument.
4332 * output.h (default_function_pro_epilogue): Likewise.
4333 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4334 (arm_output_function_prologue): Likewise.
4335 * config/frv/frv.c (frv_function_prologue): Likewise.
4336 (frv_function_epilogue): Likewise.
4337 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4338 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4339 (ia64_output_function_epilogue): Likewise.
4340 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4341 (m32r_output_function_epilogue): Likewise.
4342 * config/microblaze/microblaze.c (microblaze_function_prologue)
4343 (microblaze_function_epilogue): Likewise.
4344 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4345 (mips_output_function_epilogue): Likewise.
4346 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4347 (mmix_target_asm_function_epilogue): Likewise.
4348 * config/msp430/msp430.c (msp430_start_function): Likewise.
4349 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4350 (nds32_asm_function_epilogue): Likewise.
4351 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4352 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4353 (pa_output_function_epilogue): Likewise.
4354 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4355 (rs6000_output_function_epilogue): Likewise.
4356 * config/rl78/rl78.c (rl78_start_function): Likewise.
4357 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4358 (rs6000_output_function_epilogue): Likewise.
4359 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4360 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4361 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4362 (sparc_asm_function_epilogue): Likewise.
4364 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4366 * tree.h (type_has_mode_precision_p): New function.
4367 * convert.c (convert_to_integer_1): Use it.
4368 * expr.c (expand_expr_real_2): Likewise.
4369 (expand_expr_real_1): Likewise.
4370 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4371 * match.pd: Likewise.
4372 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4373 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4374 * tree-tailcall.c (process_assignment): Likewise.
4375 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4376 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4377 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4378 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4379 (vectorizable_assignment): Likewise.
4380 (vectorizable_shift): Likewise.
4381 (vectorizable_operation): Likewise.
4382 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4384 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4386 * match.pd: Add pow (C, x) simplification.
4388 2017-08-21 Richard Biener <rguenther@suse.de>
4390 PR tree-optimization/81900
4391 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4392 for blocks with abnormal predecessors.
4393 (compute_antic): Do not set visited flag prematurely.
4395 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4398 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4400 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4402 * stor-layout.h (vector_type_mode): Move to...
4403 * tree.h (vector_type_mode): ...here.
4404 * stor-layout.c (vector_type_mode): Move to...
4405 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4407 2017-08-21 Richard Biener <rguenther@suse.de>
4409 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4410 register_external_die hooks.
4411 (debug_false_tree_charstarstar_uhwistar): Declare.
4412 (debug_nothing_tree_charstar_uhwi): Likewise.
4413 * debug.c (do_nothing_debug_hooks): Adjust.
4414 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4415 (debug_nothing_tree_charstar_uhwi): Likewise.
4416 * dbxout.c (dbx_debug_hooks): Adjust.
4417 (xcoff_debug_hooks): Likewise.
4418 * sdbout.c (sdb_debug_hooks): Likewise.
4419 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4420 * dwarf2out.c (macinfo_label_base): New global.
4421 (dwarf2out_register_external_die): New function for the
4422 register_external_die hook.
4423 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4424 (dwarf2_debug_hooks): Use them.
4425 (dwarf2_lineno_debug_hooks): Adjust.
4426 (struct die_struct): Add with_offset flag.
4427 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4428 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4429 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4430 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4431 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4432 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4433 defining section names for the early LTO debug variants.
4434 (reset_indirect_string): New helper.
4435 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4436 (print_dw_val): Add support for offsetted symbol references.
4437 (get_ultimate_context): Split out from is_cxx.
4438 (is_cxx): Use get_ultimate_context.
4439 (is_fortran): Add decl overload.
4440 (compute_comp_unit_symbol): Split out worker from
4441 compute_section_prefix.
4442 (compute_section_prefix): Call compute_comp_unit_symbol and
4443 set comdat_type_p here.
4444 (output_die): Skip DIE symbol output for the LTO added one.
4445 Handle DIE symbol references with offset.
4446 (output_comp_unit): Guard section name mangling properly.
4447 For LTO debug sections emit a symbol at the section beginning
4448 which we use to refer to its DIEs.
4449 (add_abstract_origin_attribute): For DIEs registered via
4450 dwarf2out_register_external_die directly refer to the early
4451 DIE rather than indirectly through the shadow one we created.
4452 Remove obsolete call to dwarf2out_abstract_function for
4453 non-function/block origins.
4454 (gen_array_type_die): When generating early LTO debug do
4455 not emit DW_AT_string_length.
4456 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4457 late when in LTO. As suggested place a gcc_unreachable for
4458 the DECL_ABSTRACT_P case.
4459 (gen_subprogram_die): Avoid another specification DIE
4460 for early built declarations/definitions for the late LTO case.
4461 (gen_variable_die): Add type references for late duplicated VLA dies
4463 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4464 we have the abstract instance already.
4465 (process_scope_var): Adjust decl DIE contexts in LTO which
4466 first puts them in limbo.
4467 (gen_decl_die): Do not generate type DIEs late apart from
4468 types for VLAs or for decls we do not yet have a DIE. Do not
4469 call dwarf2out_abstract_function late.
4470 (dwarf2out_early_global_decl): Make sure to create DIEs
4471 for abstract instances of a decl first.
4472 (dwarf2out_late_global_decl): Adjust comment.
4473 (output_macinfo_op): With multiple macro sections use
4474 macinfo_label_base to distinguish labels.
4475 (output_macinfo): Likewise. Update macinfo_label_base.
4476 Pass in the line info label.
4477 (note_variable_value_in_expr): When generating LTO resolve
4478 all variable values here by generating DIEs as needed.
4479 (init_sections_and_labels): Add early LTO debug flag parameter
4480 and generate different sections and names if set. Add generation
4481 counter for the labels so we can have multiple of them.
4482 (reset_dies): Helper to allow DIEs to be output multiple times.
4483 (dwarf2out_finish): When outputting DIEs to the fat part of an
4484 LTO object first reset DIEs.
4485 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4486 (modified_type_die): Check for decl_ultimate_origin being self
4488 (gen_type_die_with_usage): Likewise.
4489 (gen_typedef_die): Allow decl_ultimate_origin being self.
4490 (set_decl_abstract_flags): Remove.
4491 (set_block_abstract_flags): Likewise.
4492 (dwarf2out_abstract_function): Treat the early generated DIEs
4493 as the abstract copy and only add DW_AT_inline and
4494 DW_AT_artificial here and call set_decl_origin_self.
4495 If the DIE has an abstract origin don't do anything.
4496 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4497 if we have none yet (Go fails to build one, PR78628).
4498 (variably_modified_type_p): Prevent endless recursion for Ada
4499 cyclic pointer types.
4500 * lto-streamer-in.c: Include debug.h.
4501 (dref_queue): New global.
4502 (lto_read_tree_1): Stream in DIE references.
4503 (lto_input_tree): Register DIE references.
4504 (input_function): Stream DECL_DEBUG_ARGS.
4505 * lto-streamer-out.c: Include debug.h.
4506 (lto_write_tree_1): Output DIE references.
4507 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4508 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4509 (output_function): Stream DECL_DEBUG_ARGS.
4510 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4511 Stream DECL_ABSTRACT_ORIGIN.
4512 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4513 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4514 DECL_CONTEXT for file-scope decls.
4515 * lto-streamer.h (struct dref_entry): Declare.
4516 (dref_queue): Likewise.
4517 * cfgexpand.c (pass_expand::execute): Do not call the
4518 outlining_inline_function hook here.
4519 * lto-wrapper.c (debug_obj): New global.
4520 (tool_cleanup): Unlink it if required.
4521 (debug_objcopy): New function.
4522 (run_gcc): Handle early debug sections in the IL files by
4523 extracting them to separate files, partially linkin them and
4524 feeding the result back as result to the linker.
4525 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4526 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4527 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4528 sections into a separate segment.
4529 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4531 (darwin_asm_dwarf_section): Likewise.
4532 (darwin_asm_output_dwarf_offset): Likewise.
4533 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4535 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4536 Alan Hayward <alan.hayward@arm.com>
4537 David Sherwood <david.sherwood@arm.com>
4539 * read-md.h (md_reader::record_potential_iterator_use): Replace
4540 pointer argument with an rtx and an index.
4541 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4542 (apply_mode_iterator): Likewise.
4543 (apply_code_iterator): Likewise.
4544 (apply_int_iterator): Likewise.
4545 (apply_subst_iterator): Likewise.
4546 (record_iterator_use): Likewise.
4547 (record_attribute_use): Likewise.
4548 (md_reader::record_potential_iterator_use): Likewise. Update calls
4549 to record_iterator_use and apply_iterator.
4550 (iterator_use): Replace ptr with x and index.
4551 (attribute_use): Likewise.
4552 (apply_attribute_uses): Update calls to apply_iterator.
4553 (apply_iterators): Likewise. Update initialization of iterator_use.
4554 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4555 and record_potential_iterator_use.
4556 (rtx_reader::read_rtx_operand): Likewise.
4558 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4559 Alan Hayward <alan.hayward@arm.com>
4560 David Sherwood <david.sherwood@arm.com>
4562 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4565 2017-08-21 Richard Biener <rguenther@suse.de>
4568 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4569 at struct end conservatively when comparing common bases.
4571 2017-08-21 Richard Biener <rguenther@suse.de>
4573 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4574 (mem_ref_in_stmt): Remove.
4575 (determine_max_movement): Use ref index to get at the reference.
4576 (invariantness_dom_walker::before_dom_children): Deal with
4577 lim data already initialized.
4578 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4580 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4582 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4583 (TARGET_ISA_ROUND): Ditto.
4584 (TARGET_ROUND): Ditto.
4585 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4586 * config/i386/i386.md: Ditto.
4587 * config/i386/sse.md: Ditto.
4588 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4589 with OPTION_MASK_ISA_SSE4_1.
4591 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4594 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4595 __builtin_ia32_lzcnt_u16.
4597 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4600 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4601 (rs6000_set_current_function): Rewrite function to use it.
4603 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4606 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4607 and TYPE_WARN_IF_NOT_ALIGN.
4608 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4609 (handle_warn_if_not_align): New.
4610 (place_union_field): Call handle_warn_if_not_align.
4611 (place_field): Call handle_warn_if_not_align.
4612 Copy TYPE_WARN_IF_NOT_ALIGN.
4613 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4614 (layout_type): Likewise.
4615 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4617 (tree_decl_common): Add warn_if_not_align.
4618 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4619 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4620 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4621 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4622 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4623 * doc/extend.texi: Document warn_if_not_aligned attribute.
4624 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4626 2017-08-17 Martin Liska <mliska@suse.cz>
4629 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4630 (get_data_dependence): Use it as pointer type.
4631 (distribute_loop): Likewise.
4633 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4635 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4636 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4637 (p8_vmrgow_<mode>_direct): New define_insn.
4638 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4639 handle endianness for vmrgew and vmrgow permute patterns.
4641 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4643 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4644 * config/rs6000/rs6000-cpus.def: Remove comment.
4645 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4646 (POWERPC_MASKS): Likewise.
4647 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4648 use of TARGET_VSX_TIMODE.
4649 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4650 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4651 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4652 (rs6000_option_override_internal): Remove dead code.
4653 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4654 (rs6000_legitimize_reload_address): Likewise.
4655 (rs6000_legitimate_address_p): Likewise.
4656 (rs6000_opt_masks): Delete "vsx-timode".
4657 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4658 from function comment.
4659 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4660 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4661 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4663 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4664 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4665 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4668 (define_peephole2): Likewise.
4670 2017-08-17 Martin Sebor <msebor@redhat.com>
4673 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4674 past the end of an array.
4675 (test_pp_format): Add test cases.
4677 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4679 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4680 missing ECF_NOTHROW flags.
4682 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4685 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4686 operands residing in integer registers.
4687 (*vsx_le_perm_load_<mode>): Likewise.
4688 (*vsx_le_perm_store_<mode>): Likewise.
4689 (define_peephole2): Add peepholes to optimize the above.
4691 2017-08-17 Marek Polacek <polacek@redhat.com>
4694 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4695 to mimic what shorten_compare did. Change the return type to bool.
4696 (fold_cond_expr_with_comparison): Update call to
4697 operand_equal_for_comparison_p.
4698 (fold_ternary_loc): Likewise.
4700 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4702 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4704 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4706 2017-08-17 Richard Biener <rguenther@suse.de>
4708 * tree-ssa-structalias.c (solve_graph): When propagating
4709 to successors update the graphs succ edges and avoid duplicate work.
4711 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4714 * config/i386/i386.c (ix86_option_override_internal): Save target
4715 specific options after ix86_stack_protector_guard_reg was changed.
4717 2017-08-17 Richard Biener <rguenther@suse.de>
4719 PR tree-optimization/81827
4720 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4721 (new_var_info): Initialize it conservatively.
4722 (get_call_vi): Mark register vars.
4723 (new_scalar_tmp_constraint_exp): Likewise.
4724 (handle_rhs_call): Likewise.
4725 (handle_const_call): Likewise.
4726 (create_function_info_for): Likewise.
4727 (solve_constraints): Sort varinfos to separate register from
4728 non-register vars to pack points-to solution bitmaps during
4731 2017-08-17 Marek Polacek <polacek@redhat.com>
4733 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4735 2017-08-17 Richard Biener <rguenther@suse.de>
4737 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4738 to true when overflow is undefined and we saturated the result.
4740 2017-08-17 Alan Modra <amodra@gmail.com>
4743 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4744 Don't use store multiple if only one reg needs saving.
4745 (interesting_frame_related_regno): New function.
4746 (rs6000_frame_related): Don't emit frame info for regs that
4748 (rs6000_emit_epilogue): Likewise.
4750 2017-08-16 Nathan Sidwell <nathan@acm.org>
4752 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4753 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4754 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4755 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4756 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4757 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4758 (verify_type): Adjust for TYPE_BINFO move.
4759 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4760 process TYPE_BINFO directly.
4761 (hash_tree): Likewise.
4762 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4764 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4767 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4769 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4771 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4774 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4775 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4776 Add ix86_binary_operator_ok to insn constraint.
4777 (*iordi_1_bts): Ditto.
4778 (*xordi_1_btc): Ditto.
4779 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4780 Update corresponding peephole2 pattern.
4784 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4786 PR tree-optimization/81832
4787 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4788 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4790 2017-08-16 Marek Polacek <polacek@redhat.com>
4793 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4794 regarding size_zero_node.
4796 2017-08-16 Martin Liska <mliska@suse.cz>
4799 * config.gcc: Respect previously set extra_objs in case
4802 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4804 PR tree-optimization/81835
4805 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4806 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4807 not depend on the phi.
4809 2017-08-16 Alan Modra <amodra@gmail.com>
4811 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4814 2017-08-16 Alan Modra <amodra@gmail.com>
4816 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4817 (save_reg_p): ..into this. Update all callers.
4818 (first_reg_to_save): Simplify.
4820 2017-08-16 Alan Modra <amodra@gmail.com>
4822 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4825 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4829 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4830 constant count if that count is less than 32.
4832 2017-08-15 Nathan Sidwell <nathan@acm.org>
4834 * gcc.c (execute): Emit friendlier message if inferior is killed
4835 by an external cause.
4837 2017-08-15 Richard Biener <rguenther@suse.de>
4839 PR tree-optimization/81790
4840 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4841 CONSTRUCTORs from simplifying and VN.
4843 2017-08-14 Martin Sebor <msebor@redhat.com>
4845 * builtin-attrs.def: Add comments.
4847 2017-08-14 Martin Sebor <msebor@redhat.com>
4850 * doc/extend.texi (attribute nonstring): Document new attribute.
4852 2017-08-14 Martin Sebor <msebor@redhat.com>
4855 * tree-diagnostic.c (default_tree_printer): Handle %G.
4856 * gimple-pretty-print.h (percent_G_format): Declare new function.
4857 * gimple-pretty-print.c (percent_G_format): Define.
4858 * tree-pretty-print.c (percent_K_format): Add argument.
4860 2017-08-14 Martin Sebor <msebor@redhat.com>
4862 PR translation/79998
4863 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4864 Remove a stray space.
4866 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4869 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4870 (*iordi_1_bts): Ditto.
4871 (*xordi_1_btc): Ditto.
4873 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4876 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4877 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4879 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4880 quoted strings, and make more translator-friendly.
4881 (darwin_rs6000_override_options): Likewise.
4882 (rs6000_option_override_internal): Likewise.
4883 (rs6000_return_in_memory): Fix overlong line.
4884 (init_cmulative_args): Use quoted strings, and make more
4885 translator-friendly.
4886 (rs6000_pass_by_reference): Fix overlong line.
4887 (def_builtin): Use quoted strings.
4888 (altivec_expand_predicate_builtin): Use quoted strings, and make
4889 more translator-friendly.
4890 (htm_expand_builtin): Use quoted strings.
4891 (cpu_expand_builtin): Use quoted strings, and make more
4892 translator-friendly.
4893 (altivec_expand_builtin): Likewise.
4894 (paired_expand_predicate_builtin): Likewise.
4895 (rs6000_invalid_builtin): Likewise.
4896 (builtin_function_type): Use quoted strings.
4897 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4898 more translator-friendly.
4899 (rs6000_trampoline_init): Likewise.
4900 (rs6000_handle_altivec_attribute): Likewise.
4901 (rs6000_inner_target_options): Use quoted strings.
4902 (rs6000_disable_incompatible_switches): Likewise.
4903 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4904 strings, and make more translator-friendly.
4905 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4907 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4909 PR tree-optimization/81799
4910 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4911 cond_expr to simple gimple operand.
4913 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4916 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4918 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4922 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4923 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4924 TARGET_GASISR_PROLOGUES.
4925 * config/avr/avr.c (avr_option_override): Same.
4926 (avr_pass_pre_proep::execute): Same.
4928 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4931 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4932 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4934 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4936 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4937 define_insn_and_split. Split to a memory load from 0 in
4938 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4939 using PTR mode iterator.
4940 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4941 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4942 (*add_tp_<mode>): Redefine as define_insn_and_split.
4943 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4944 address space. Merge with *add_tp_x32 using PTR mode iterator.
4945 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4946 Split to an add with a memory load from 0 in
4947 DEFAULT_TLS_SEG_REG address space.
4949 2017-08-12 Andrew Pinski <apinski@cavium.com>
4951 * config/aarch64/aarch64-option-extensions.def (rdma):
4952 Fix feature string to what Linux prints out in /proc/cpuinfo.
4954 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4957 * dwarf2out.c (modified_type_die): For C typedef types that have
4958 an ultimate origin, process the ultimate origin instead of the
4960 (gen_typedef_die): Assert that input DECLs have no ultimate
4962 (gen_type_die_with_usage): For typedef variants that have an
4963 ultimate origin, just call gen_decl_die on the original DECL.
4964 (process_scope_var): Avoid creating DIEs for local typedefs and
4965 concrete static variables.
4967 2017-08-12 Alan Modra <amodra@gmail.com>
4971 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4972 match gnu-user.h startfile.
4973 (ENDFILE_LINUX_SPEC): Similarly.
4975 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4978 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4980 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4982 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4983 * config/aarch64/aarch64.md (mov<mode>): Change.
4984 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4985 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4986 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4988 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4990 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4991 for storage order barriers.
4993 2017-08-11 Martin Liska <mliska@suse.cz>
4996 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4997 variables of void type.
4999 2017-08-11 Martin Liska <mliska@suse.cz>
5001 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5002 TARGET_SUPPORTS_ALIASES.
5003 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5004 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5005 (optimize_weakref): Likewise.
5006 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5007 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5008 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5010 2017-08-11 Martin Liska <mliska@suse.cz>
5013 * config/i386/i386.c (make_resolver_func): Do complete
5014 refactoring of the function.
5016 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5019 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5020 * config/i386/i386.c (ix86_stack_protect_guard): Use
5021 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5022 * doc/invoke.texi (x86 Options): Document
5023 -mstack-protector-guard-symbol= option.
5025 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5027 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5028 * config/i386/i386.c (ix86_split_stack_guard): New function.
5029 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5030 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5031 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5032 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5033 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5034 (split_stack_space_check): Call ix86_split_stack_guard.
5036 2017-08-10 Martin Sebor <msebor@redhat.com>
5038 * print-tree.c (print_node): Print location using the established
5040 Replace spaces with colons.
5041 (debug_raw, debug): Ditto.
5043 2017-08-10 Martin Sebor <msebor@redhat.com>
5046 * pretty-print.c (pp_format): Correct the handling of %s precision.
5048 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5051 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5053 (ix86_finalize_stack_frame_flags): This. Also clear
5054 frame_pointer_needed if -fno-omit-frame-pointer is used without
5056 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5057 with ix86_finalize_stack_frame_flags.
5058 (ix86_expand_epilogue): Likewise.
5059 (ix86_expand_split_stack_prologue): Likewise.
5060 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5062 2017-08-10 Martin Liska <mliska@suse.cz>
5065 * c-attribs.c (handle_target_attribute):
5066 Report warning for an empty string argument of target attribute.
5068 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5071 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5073 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5074 or DECL_NONLOCAL labels.
5075 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5076 or DECL_NONLOCAL labels here.
5078 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5080 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5081 to indicate when early gimple folding has been disabled.
5082 (rs6000_gimple_fold_builtin): Add debug content.
5083 (rs6000_invalid_builtin): Fix whitespace.
5084 (rs6000_expand_builtin): Fix whitespace.
5085 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5087 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5090 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5091 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5093 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5095 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5096 (qdf24xx): Likewise.
5097 * config/aarch64/aarch64.md: Include falkor.md.
5098 * config/aarch64/falkor.md: New.
5100 2017-08-09 Marek Polacek <polacek@redhat.com>
5103 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5104 * diagnostic.c (emit_diagnostic): Add a comment.
5105 (emit_diagnostic_valist): New function.
5107 2017-08-09 Marek Polacek <polacek@redhat.com>
5110 * input.c (make_location): New overload.
5111 * input.h (make_location): Declare.
5113 2017-08-08 Alan Modra <amodra@gmail.com>
5114 H.J. Lu <hongjiu.lu@intel.com>
5117 * gcc.c (NO_PIE_SPEC): Delete.
5118 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5120 (LINK_PIE_SPEC): ..to here.
5121 (LINK_COMMAND_SPEC): Support -no-pie.
5122 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5123 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5125 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5126 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5127 (ENDFILE_CRTEND_SPEC): Similarly.
5129 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5132 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5133 (mstack-protector-guard-offset=): Ditto.
5134 * config/i386/i386.c (ix86_option_override): Handle
5135 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5137 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5138 ix86_stack_protect_guard_offset variables.
5139 (TARGET_STACK_PROTECT_GUARD): Always define.
5140 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5141 and -mstack-protector-guard-offset= options.
5143 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5145 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5146 boundary case for the last candidate.
5148 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5150 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5151 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5153 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5156 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5157 * config/aarch64/aarch64-builtins.c
5158 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5159 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5160 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5162 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5163 Andrew Pinski <pinskia@gmail.com>
5166 * internal-fn.def (XORSIGN): New.
5167 * optabs.def (xorsign_optab): New.
5168 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5169 (convert_expand_mult_copysign): New.
5170 (pass_optimize_widening_mul::execute): Call
5171 convert_expand_mult_copysign.
5173 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5175 PR tree-optimization/81354
5176 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5177 Insert on edges rather than explicitly creating landing pads.
5178 (analyze_candidates_and_replace): Commit edge inserts.
5180 2017-08-08 Richard Biener <rguenther@suse.de>
5183 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5184 (expand_simple_operations): Also look through ADDR_EXPRs with
5185 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5187 2017-08-08 Richard Biener <rguenther@suse.de>
5189 PR tree-optimization/81723
5190 * tree-vect-slp.c (struct bst_traits): New hash traits.
5191 (bst_fail): New global.
5192 (vect_build_slp_tree_2): New worker, split out from ...
5193 (vect_build_slp_tree): ... this now wrapping it with using
5194 bst_fail set to cache SLP tree build fails. Properly handle
5196 (vect_analyze_slp_instance): Allocate and free bst_fail.
5198 2017-08-08 Martin Liska <mliska@suse.cz>
5201 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5202 LABEL_DECLs that can be from a different function.
5204 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5206 PR tree-optimization/81744
5207 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5208 loop's number of iterations.
5210 2017-08-08 Martin Liska <mliska@suse.cz>
5212 * asan.c: Include header files.
5213 * attribs.c (build_decl_attribute_variant): New function moved
5215 (build_type_attribute_qual_variant): Likewise.
5216 (cmp_attrib_identifiers): Likewise.
5217 (simple_cst_list_equal): Likewise.
5218 (omp_declare_simd_clauses_equal): Likewise.
5219 (attribute_value_equal): Likewise.
5220 (comp_type_attributes): Likewise.
5221 (build_type_attribute_variant): Likewise.
5222 (lookup_ident_attribute): Likewise.
5223 (remove_attribute): Likewise.
5224 (merge_attributes): Likewise.
5225 (merge_type_attributes): Likewise.
5226 (merge_decl_attributes): Likewise.
5227 (merge_dllimport_decl_attributes): Likewise.
5228 (handle_dll_attribute): Likewise.
5229 (attribute_list_equal): Likewise.
5230 (attribute_list_contained): Likewise.
5231 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5232 (lookup_attribute_by_prefix): Likewise.
5233 * bb-reorder.c: Include header files.
5234 * builtins.c: Likewise.
5235 * calls.c: Likewise.
5236 * cfgexpand.c: Likewise.
5237 * cgraph.c: Likewise.
5238 * cgraphunit.c: Likewise.
5239 * convert.c: Likewise.
5240 * dwarf2out.c: Likewise.
5241 * final.c: Likewise.
5242 * fold-const.c: Likewise.
5243 * function.c: Likewise.
5244 * gimple-expr.c: Likewise.
5245 * gimple-fold.c: Likewise.
5246 * gimple-pretty-print.c: Likewise.
5247 * gimple.c: Likewise.
5248 * gimplify.c: Likewise.
5249 * hsa-common.c: Likewise.
5250 * hsa-gen.c: Likewise.
5251 * internal-fn.c: Likewise.
5252 * ipa-chkp.c: Likewise.
5253 * ipa-cp.c: Likewise.
5254 * ipa-devirt.c: Likewise.
5255 * ipa-fnsummary.c: Likewise.
5256 * ipa-inline.c: Likewise.
5257 * ipa-visibility.c: Likewise.
5259 * lto-cgraph.c: Likewise.
5260 * omp-expand.c: Likewise.
5261 * omp-general.c: Likewise.
5262 * omp-low.c: Likewise.
5263 * omp-offload.c: Likewise.
5264 * omp-simd-clone.c: Likewise.
5265 * opts-global.c: Likewise.
5266 * passes.c: Likewise.
5267 * predict.c: Likewise.
5268 * sancov.c: Likewise.
5269 * sanopt.c: Likewise.
5270 * symtab.c: Likewise.
5271 * toplev.c: Likewise.
5272 * trans-mem.c: Likewise.
5273 * tree-chkp.c: Likewise.
5274 * tree-eh.c: Likewise.
5275 * tree-into-ssa.c: Likewise.
5276 * tree-object-size.c: Likewise.
5277 * tree-parloops.c: Likewise.
5278 * tree-profile.c: Likewise.
5279 * tree-ssa-ccp.c: Likewise.
5280 * tree-ssa-live.c: Likewise.
5281 * tree-ssa-loop.c: Likewise.
5282 * tree-ssa-sccvn.c: Likewise.
5283 * tree-ssa-structalias.c: Likewise.
5284 * tree-ssa.c: Likewise.
5285 * tree-streamer-in.c: Likewise.
5286 * tree-vectorizer.c: Likewise.
5287 * tree-vrp.c: Likewise.
5289 * ubsan.c: Likewise.
5290 * varasm.c: Likewise.
5291 * varpool.c: Likewise.
5292 * tree.c: Remove functions moved to attribs.[ch].
5294 * config/aarch64/aarch64.c: Add attrs.h header file.
5295 * config/alpha/alpha.c: Likewise.
5296 * config/arc/arc.c: Likewise.
5297 * config/arm/arm.c: Likewise.
5298 * config/avr/avr.c: Likewise.
5299 * config/bfin/bfin.c: Likewise.
5300 * config/c6x/c6x.c: Likewise.
5301 * config/cr16/cr16.c: Likewise.
5302 * config/cris/cris.c: Likewise.
5303 * config/darwin.c: Likewise.
5304 * config/epiphany/epiphany.c: Likewise.
5305 * config/fr30/fr30.c: Likewise.
5306 * config/frv/frv.c: Likewise.
5307 * config/ft32/ft32.c: Likewise.
5308 * config/h8300/h8300.c: Likewise.
5309 * config/i386/winnt.c: Likewise.
5310 * config/ia64/ia64.c: Likewise.
5311 * config/iq2000/iq2000.c: Likewise.
5312 * config/lm32/lm32.c: Likewise.
5313 * config/m32c/m32c.c: Likewise.
5314 * config/m32r/m32r.c: Likewise.
5315 * config/m68k/m68k.c: Likewise.
5316 * config/mcore/mcore.c: Likewise.
5317 * config/microblaze/microblaze.c: Likewise.
5318 * config/mips/mips.c: Likewise.
5319 * config/mmix/mmix.c: Likewise.
5320 * config/mn10300/mn10300.c: Likewise.
5321 * config/moxie/moxie.c: Likewise.
5322 * config/msp430/msp430.c: Likewise.
5323 * config/nds32/nds32-isr.c: Likewise.
5324 * config/nds32/nds32.c: Likewise.
5325 * config/nios2/nios2.c: Likewise.
5326 * config/nvptx/nvptx.c: Likewise.
5327 * config/pa/pa.c: Likewise.
5328 * config/pdp11/pdp11.c: Likewise.
5329 * config/powerpcspe/powerpcspe.c: Likewise.
5330 * config/riscv/riscv.c: Likewise.
5331 * config/rl78/rl78.c: Likewise.
5332 * config/rx/rx.c: Likewise.
5333 * config/s390/s390.c: Likewise.
5334 * config/sh/sh.c: Likewise.
5335 * config/sol2.c: Likewise.
5336 * config/sparc/sparc.c: Likewise.
5337 * config/spu/spu.c: Likewise.
5338 * config/stormy16/stormy16.c: Likewise.
5339 * config/tilegx/tilegx.c: Likewise.
5340 * config/tilepro/tilepro.c: Likewise.
5341 * config/v850/v850.c: Likewise.
5342 * config/vax/vax.c: Likewise.
5343 * config/visium/visium.c: Likewise.
5344 * config/xtensa/xtensa.c: Likewise.
5346 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5349 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5350 constraints since the -mupper-regs-* switches have been
5352 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5353 into a vector from a double word element that was extracted from
5354 another vector, and eliminate extra XXPERMDI instructions.
5355 (vsx_concat_<mode>_2): Likewise.
5356 (vsx_concat_<mode>_3): Likewise.
5357 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5358 concat to allow optimizing inserts from previous extracts.
5360 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5362 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5363 memory reference to a SSP offset in TLS address space.
5364 (ix86_print_operand) <case '@'>: Remove.
5365 (ix86_print_operand_punct_valid_p): Remove '@' code.
5366 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5368 (stack_tls_protect_set_<mode>): Remove.
5369 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5370 (stack_tls_protect_test_<mode>): Remove.
5371 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5373 2017-08-07 Olivier Hainque <hainque@adacore.com>
5376 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5378 2017-08-07 Douglas Rupp <rupp@adacore.com>
5380 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5381 variable was referenced as multidir in command.
5383 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5386 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5389 2017-08-07 Martin Liska <mliska@suse.cz>
5391 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5392 * config/rl78/rl78.c: Add include of attribs.h.
5393 * config/sh/sh.c: Likewise.
5394 * config/v850/v850.c: Likewise.
5396 2017-08-07 Tom de Vries <tom@codesourcery.com>
5399 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5401 2017-08-07 Martin Liska <mliska@suse.cz>
5403 * config/mips/mips.c: Include attribs.h.
5405 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5408 * doc/invoke.texi: Document change in behvaior for -Ofast for
5411 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5413 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5415 (aarch64_pop_regs): Likewise.
5416 (aarch64_gen_load_pair): Likewise.
5417 (aarch64_save_callee_saves): Likewise.
5418 (aarch64_restore_callee_saves): Likewise.
5420 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5422 * config/i386/i386.c: Revert the last change.
5424 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5427 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5429 (ix86_finalize_stack_frame_flags): This. Also clear
5430 frame_pointer_needed if -fno-omit-frame-pointer is used without
5432 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5433 with ix86_finalize_stack_frame_flags.
5434 (ix86_expand_epilogue): Likewise.
5435 (ix86_expand_split_stack_prologue): Likewise.
5437 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5440 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5441 to P_AES for Westmere.
5443 2017-08-07 Jonathan Yong <10walls@gmail.com>
5445 * config/i386/mingw.opt (fset-stack-executable): Removed.
5446 * config/i386/cygming.opt (fset-stack-executable): Moved
5448 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5450 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5452 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5454 2017-08-07 Marek Polacek <polacek@redhat.com>
5457 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5459 2017-08-07 Martin Liska <mliska@suse.cz>
5461 * attribs.h (canonicalize_attr_name): New function.
5462 (cmp_attribs): Move from c-format.c and adjusted.
5463 (is_attribute_p): Moved from tree.h.
5464 * tree-inline.c: Add new includes.
5465 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5466 (private_is_attribute_p): Remove.
5467 (private_lookup_attribute): Likewise.
5468 (private_lookup_attribute_by_prefix): Simplify.
5469 (remove_attribute): Use is_attribute_p.
5470 * tree.h: Remove removed declarations.
5472 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5475 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5476 instead of computing it in the function. Formatting fix.
5477 (expand_case): Don't rely on default_edge being the first edge,
5478 clear it if removing it, pass default_edge to
5479 emit_case_dispatch_table.
5480 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5483 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5485 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5486 insn in the function, emit NOP after the insn.
5488 2017-08-06 Tom de Vries <tom@codesourcery.com>
5490 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5493 2017-08-06 Tom de Vries <tom@codesourcery.com>
5495 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5498 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5500 PR tree-optimization/57371
5501 * match.pd: New pattern.
5503 2017-08-04 Marek Polacek <polacek@redhat.com>
5506 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5507 perform the computation in offset_int.
5509 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5511 PR tree-optimization/81136
5512 * tree-vectorizer.h: Include tree-hash-traits.h.
5513 (vec_base_alignments): New typedef.
5514 (vec_info): Add a base_alignments field.
5515 (vect_record_base_alignments): Declare.
5516 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5518 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5519 (create_data_ref): Add an is_conditional_in_stmt argument.
5520 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5521 the is_conditional_in_stmt field.
5522 (data_ref_loc): Add an is_conditional_in_stmt field.
5523 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5524 (find_data_references_in_stmt): Update call to create_data_ref.
5525 (graphite_find_data_references_in_stmt): Likewise.
5526 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5527 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5528 (vect_record_base_alignment): New function.
5529 (vect_record_base_alignments): Likewise.
5530 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5531 for nested statements even if we fail to compute a misalignment.
5532 Use pooled base alignments for unconditional references.
5533 (vect_find_same_alignment_drs): Compare base addresses instead
5535 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5536 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5538 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5540 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5541 Add an explicit name for the enum. Use auto_vec for slp_instances
5543 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5545 (_bb_vec_info): Add a constructor and destructor.
5546 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5547 (destroy_loop_vec_info): Delete.
5548 (vect_destroy_datarefs): Likewise.
5549 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5550 (vec_info::vec_info): New function.
5551 (vec_info::~vec_info): Likewise.
5552 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5553 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5554 destroy_loop_vec_info.
5555 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5556 (_loop_vec_info::_loop_vec_info): ...this.
5557 (destroy_loop_vec_info): Replace with...
5558 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5559 the stmt_vec_infos. Leave handling of vec_info information to its
5560 destructor. Remove explicit vector releases.
5561 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5562 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5563 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5564 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5565 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5566 (destroy_bb_vec_info): Replace with...
5567 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5568 information to its destructor.
5569 (vect_slp_analyze_bb_1): Use new and delete instead of
5570 new_bb_vec_info and destroy_bb_vec_info.
5571 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5574 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5576 * tree-data-ref.h (subscript): Add access_fn field.
5577 (data_dependence_relation): Add could_be_independent_p.
5578 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5579 (same_access_functions): Move to tree-data-ref.c.
5580 * tree-data-ref.c (ref_contains_union_access_p): New function.
5581 (access_fn_component_p): Likewise.
5582 (access_fn_components_comparable_p): Likewise.
5583 (dr_analyze_indices): Add a reference to access_fn_component_p.
5584 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5586 (constant_access_functions): Likewise.
5587 (add_other_self_distances): Likewise.
5588 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5589 (initialize_data_dependence_relation): Use XCNEW and remove
5590 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5591 of access functions that have the same type. Allow the
5592 subsequence to end with different bases in some circumstances.
5593 Record the chosen access functions in SUB_ACCESS_FN.
5594 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5595 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5596 (subscript_dependence_tester_1): Likewise dra and drb.
5597 (build_classic_dist_vector): Update calls accordingly.
5598 (subscript_dependence_tester): Likewise.
5599 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5600 DDR_COULD_BE_INDEPENDENT_P.
5601 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5602 comp_alias_ddrs instead of may_alias_ddrs.
5603 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5605 (vect_analyze_data_ref_dependence): Use it if
5606 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5607 distance vectors if that fails.
5608 (dependence_distance_ge_vf): New function.
5609 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5610 LOOP_VINFO_MAY_ALIAS_DDRS.
5612 2017-08-04 Richard Biener <rguenther@suse.de>
5615 * fold-const.c (fold_binary_loc): Properly restrict
5616 minus_var0 && minus_var1 case when associating undefined overflow
5619 2017-08-04 Tom de Vries <tom@codesourcery.com>
5621 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5623 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5625 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5626 Don't start diagnostic messages with a capital letter.
5627 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5629 (rs6000_invalid_builtin): Likewise.
5630 (rs6000_trampoline_init): Likewise.
5632 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5635 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5636 after setting changeable df flags.
5638 2017-08-03 Richard Biener <rguenther@suse.de>
5640 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5641 up if the use is in USE - X.
5643 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5645 * toplev.c (dumpfile.h): New include.
5646 (internal_error_reentered): New static function. Use it...
5647 (internal_error_function): ...here to handle reentered internal_error.
5649 2017-08-03 Richard Biener <rguenther@suse.de>
5652 * fold-const.c (split_tree): Add minus_var and minus_con
5653 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5654 here but always use minus_*.
5655 (associate_trees): Assert we never associate with MINUS_EXPR
5656 and NULL first operand. Do not recurse for PLUS_EXPR operands
5657 when associating as MINUS_EXPR either.
5658 (fold_binary_loc): Track minus_var and minus_con.
5660 2017-08-03 Tom de Vries <tom@codesourcery.com>
5663 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5664 ACCEL_COMPILER, apply finish_options on
5665 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5667 2017-08-03 Tom de Vries <tom@codesourcery.com>
5670 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5671 function_entry_patch_area_size > 0.
5673 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5676 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5677 instead of 10??LU, perform unit multiplication in wide_int,
5678 don't change alloc_object_size_limit if the limit is larger
5681 PR tree-optimization/81655
5682 PR tree-optimization/81588
5683 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5684 the case when ranges[i].low and high are 1 for unsigned type with
5688 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5689 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5691 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5693 * tree-vrp.h: Add include guard.
5695 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5698 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5699 (ud2): New insn pattern.
5700 * config/i386/i386.c (ix86_expand_epilogue):
5701 For naked functions, generate ud2 instead of trap insn.
5703 2017-08-02 Marek Polacek <polacek@redhat.com>
5706 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5708 2017-08-02 Tom de Vries <tom@codesourcery.com>
5709 Cesar Philippidis <cesar@codesourcery.com>
5711 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5712 Add missing edge probabilities.
5714 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5716 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5719 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5722 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5723 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5724 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5725 prologue_seq sequences - if any.
5727 2017-08-02 Richard Biener <rguenther@suse.de>
5729 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5730 via vectors if supported, integer extracts via punning if supported
5731 or otherwise vector extracts.
5733 2017-08-02 Richard Biener <rguenther@suse.de>
5735 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5737 (bitmap_insert_into_set): ... this.
5739 2017-08-02 Richard Biener <rguenther@suse.de>
5741 PR tree-optimization/81633
5743 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5745 PR tree-optimization/71752
5746 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5748 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5750 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5751 (machine_function::call_ms2sysv_pad_out): Remove field.
5752 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5753 (ix86_compute_frame_layout): Likewise.
5755 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5758 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5759 attribute with interrupt attribute.
5761 2017-08-01 Andrew Pinski <apinski@cavium.com>
5763 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5764 BIT_INSERT_EXPR's operand 1
5765 to see if the types precision matches.
5767 2017-08-01 Martin Liska <mliska@suse.cz>
5770 * builtins.c (expand_builtin_memcpy_args): Remove.
5771 (expand_builtin_memcpy): Call newly added function
5772 expand_builtin_memory_copy_args.
5773 (expand_builtin_memcpy_with_bounds): Likewise.
5774 (expand_builtin_mempcpy): Remove last argument.
5775 (expand_builtin_mempcpy_with_bounds): Likewise.
5776 (expand_builtin_memory_copy_args): New function created from
5777 expand_builtin_mempcpy_args with small modifications.
5778 (expand_builtin_mempcpy_args): Remove.
5779 (expand_builtin_stpcpy): Remove unused argument.
5780 (expand_builtin): Likewise.
5781 (expand_builtin_with_bounds): Likewise.
5783 2017-08-01 Martin Liska <mliska@suse.cz>
5786 Make mempcpy more optimal (PR middle-end/70140).
5788 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5791 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5792 __builtin_vec_cmpne verify both arguments are compatible vectors
5793 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5794 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5795 move computation of aligned to after checking the argument types.
5799 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5800 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5803 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5804 Doug Rupp <rupp@adacore.com>
5805 Olivier Hainque <hainque@adacore.com>
5807 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5808 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5810 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5811 for TARGET_OS_CPP_BUILTIN.
5812 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5813 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5815 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5816 passing required abi options to the assembler for EABI configurations.
5817 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5818 of .text.hot and .text.unlikely sections for kernel modules when
5819 using ARM style exceptions.
5820 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5821 options. Add EXTRA_CC1_SPEC.
5822 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5824 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5826 (ARM_TARGET2_DWARF_FORMAT): Define.
5827 * config/arm/t-vxworks: Adjust multilib control to removal of the
5828 Diab command line options.
5830 2017-08-01 Martin Liska <mliska@suse.cz>
5832 PR gcov-profile/81561
5833 * gcov.c (unblock): Make unblocking safe as we need to preserve
5834 index correspondence of blocks and block_lists.
5836 2017-08-01 Richard Biener <rguenther@suse.de>
5838 PR tree-optimization/81181
5839 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5840 (compute_antic): ... end of iteration here.
5842 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5844 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5845 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5846 (ftree-slp-vectorize): Likewise.
5847 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5848 can no longer be set independent of flag_tree_loop_vectorize.
5849 * omp-general.c (emp_max_vf): Likewise.
5850 * opts.c (enable_fdo_optimizations): Remove references to
5851 flag_tree_vectorize, these are now implicit.
5852 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5853 and leave it for the options machinery.
5855 2017-08-01 Martin Liska <mliska@suse.cz>
5858 * builtins.c (expand_builtin_memcpy_args): Remove.
5859 (expand_builtin_memcpy): Call newly added function
5860 expand_builtin_memory_copy_args.
5861 (expand_builtin_memcpy_with_bounds): Likewise.
5862 (expand_builtin_mempcpy): Remove last argument.
5863 (expand_builtin_mempcpy_with_bounds): Likewise.
5864 (expand_builtin_memory_copy_args): New function created from
5865 expand_builtin_mempcpy_args with small modifications.
5866 (expand_builtin_mempcpy_args): Remove.
5867 (expand_builtin_stpcpy): Remove unused argument.
5868 (expand_builtin): Likewise.
5869 (expand_builtin_with_bounds): Likewise.
5871 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5874 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5875 print "ds:" only for immediates in generic address space.
5877 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5880 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5881 (ix86_function_ok_for_sibcall): Return false for naked functions.
5883 2017-08-01 Richard Biener <rguenther@suse.de>
5885 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5886 (compute_antic): Seed worklist with exit block predecessors.
5887 * cfganal.c (dfs_find_deadend): For a cycle return the source
5888 of the edge closing it.
5890 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5892 * config/aarch64/aarch64.c
5893 (aarch64_can_const_movi_rtx_p): Move 0 check.
5895 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5897 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5898 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5900 * match.pd: Ditto in address comparison pattern.
5902 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5904 PR tree-optimization/81627
5905 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5906 closed ssa form for store-store chain.
5908 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5910 PR tree-optimization/81620
5911 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5912 for store-store chain.
5914 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5916 PR tree-optimization/81588
5917 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5918 ranges[i].in_p, invert comparison code ccode. For >/>=,
5919 swap rhs1 and rhs2 and comparison code unconditionally,
5920 for </<= don't do that. Don't swap rhs1/rhs2 again if
5921 ranges[i].in_p, instead invert comparison code ccode if
5922 opcode or oe->rank is BIT_IOR_EXPR.
5925 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5926 a direct optab to conversion optab.
5927 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5928 with GET_MODE_INNER as last argument instead of optab_handler.
5929 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5930 vector extraction if possible and optab is available.
5931 * expr.c (store_constructor): Use convert_optab_handler instead
5932 of optab_handler. Use vector initialization from smaller
5933 vectors if possible and optab is available.
5934 * tree-vect-stmts.c (vectorizable_load): Likewise.
5935 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5937 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5938 of vec_init from half-sized vectors with the same element mode.
5939 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5940 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5941 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5942 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5943 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5944 after mode in gen_vec_extract* calls.
5945 (vec_extract<mode>): Renamed to ...
5946 (vec_extract<mode><ssescalarmodelower>): ... this.
5947 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5948 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5949 element mode after mode in gen_vec_init* calls.
5950 (VEC_INIT_HALF_MODE): New mode iterator.
5951 (vec_init<mode>): Renamed to ...
5952 (vec_init<mode><ssescalarmodelower>): ... this.
5953 (vec_init<mode><ssehalfvecmodelower>): New expander.
5954 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5955 (vec_extractv2sfsf): ... this.
5956 (vec_initv2sf): Renamed to ...
5957 (vec_initv2sfsf): ... this.
5958 (vec_extractv2si): Renamed to ...
5959 (vec_extractv2sisi): ... this.
5960 (vec_initv2si): Renamed to ...
5961 (vec_initv2sisi): ... this.
5962 (vec_extractv4hi): Renamed to ...
5963 (vec_extractv4hihi): ... this.
5964 (vec_initv4hi): Renamed to ...
5965 (vec_initv4hihi): ... this.
5966 (vec_extractv8qi): Renamed to ...
5967 (vec_extractv8qiqi): ... this.
5968 (vec_initv8qi): Renamed to ...
5969 (vec_initv8qiqi): ... this.
5970 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5971 (vec_init<mode>): Renamed to ...
5972 (vec_init<mode><VEC_base_l>): ... this.
5973 (vec_extract<mode>): Renamed to ...
5974 (vec_extract<mode><VEC_base_l>): ... this.
5975 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5976 (vec_initv2sfsf): ... this.
5977 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5978 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5979 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5980 element mode after mode in gen_vec_init* calls.
5981 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5982 (vec_init<mode><Vel>): ... this.
5983 (vec_extract<mode>): Renamed to ...
5984 (vec_extract<mode><Vel>): ... this.
5985 * config/aarch64/iterators.md (Vel): New mode attribute.
5986 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5987 Add element mode after mode in gen_vec_extract* calls.
5988 * config/s390/vector.md (non_vec_l): New mode attribute.
5989 (vec_extract<mode>): Renamed to ...
5990 (vec_extract<mode><non_vec_l>): ... this.
5991 (vec_init<mode>): Renamed to ...
5992 (vec_init<mode><non_vec_l>): ... this.
5993 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5994 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5996 * config/arm/iterators.md (V_elem_l): New mode attribute.
5997 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5998 (vec_extract<mode><V_elem_l>): ... this.
5999 (vec_extractv2di): Renamed to ...
6000 (vec_extractv2didi): ... this.
6001 (vec_init<mode>): Renamed to ...
6002 (vec_init<mode><V_elem_l>): ... this.
6003 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6004 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6005 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6006 Add element mode after gen_vec_extract* calls.
6007 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6008 (vec_init<mode><unitmode>): ... this.
6009 (vec_extract<mode>): Renamed to ...
6010 (vec_extract<mode><unitmode>): ... this.
6011 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6012 (vec_init<mode><unitmode>): ... this.
6013 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6014 (vec_initv2sfsf): ... this.
6015 (vec_extractv2sf): Renamed to ...
6016 (vec_extractv2sfsf): ... this.
6017 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6018 Add element mode after gen_vec_extract* calls.
6019 * config/mips/mips.md (unitmode): New mode iterator.
6020 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6021 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6022 * config/spu/spu.md (inner_l): New mode attribute.
6023 (vec_init<mode>): Renamed to ...
6024 (vec_init<mode><inner_l>): ... this.
6025 (vec_extract<mode>): Renamed to ...
6026 (vec_extract<mode><inner_l>): ... this.
6027 * config/sparc/sparc.md (veltmode): New mode iterator.
6028 (vec_init<VMALL:mode>): Renamed to ...
6029 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6030 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6031 (vec_initv2sisi): ... this.
6032 (vec_initv2sf): Renamed to ...
6033 (vec_initv2sfsf): ... this.
6034 (vec_extractv2sf): Renamed to ...
6035 (vec_extractv2sfsf): ... this.
6036 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6037 (vec_init<mode>): Renamed to ...
6038 (vec_init<mode><VEC_base_l>): ... this.
6039 (vec_extract<mode>): Renamed to ...
6040 (vec_extract<mode><VEC_base_l>): ... this.
6041 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6042 (vec_initv2sfsf): ... this.
6043 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6044 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6045 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6046 gen_vec_init* calls.
6048 2017-08-01 Richard Biener <rguenther@suse.de>
6050 PR tree-optimization/81297
6051 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6052 TREE_OVERFLOW from INTEGER_CSTs.
6054 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6056 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6058 2017-07-31 Carl Love <cel@us.ibm.com>
6060 * config/rs6000/rs6000-c: Add support for built-in functions
6061 vector signed char vec_xl_be (signed long long, signed char *);
6062 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6063 vector signed int vec_xl_be (signed long long, signed int *);
6064 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6065 vector signed long long vec_xl_be (signed long long, signed long long *);
6066 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6067 vector signed short vec_xl_be (signed long long, signed short *);
6068 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6069 vector double vec_xl_be (signed long long, double *);
6070 vector float vec_xl_be (signed long long, float *);
6071 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6072 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6073 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6075 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6076 (altivec_expand_builtin): Add switch statement to call
6077 altivec_expand_xl_be for each builtin.
6078 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6079 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6080 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6081 __builtin_vsx_le_be_v16qi.
6082 * doc/extend.texi: Update the built-in documentation file for the
6083 new built-in functions.
6085 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6088 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6090 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6092 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6094 * config.gcc: Add z14.
6095 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6096 CPU model numbers for z13s and z14.
6097 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6099 * config/s390/s390-opts.h (enum processor_type): Rename
6100 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6101 * config/s390/s390.c (processor_table): Add field for CPU name to
6102 be passed to Binutils.
6103 (s390_asm_output_machine_for_arch): Use the new field in
6104 processor_table for Binutils.
6105 (s390_expand_builtin): Replace arch12 with z14.
6106 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6107 (s390_get_sched_attrmask): Likewise.
6108 (s390_get_unit_mask): Likewise.
6109 * config/s390/s390.opt: Add z14 to processor_type enum.
6111 2017-07-31 Martin Jambor <mjambor@suse.cz>
6114 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6115 regardless of optimization level.
6117 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6118 Martin Liska <mliska@suse.cz>
6120 * predict.def: Remove old comment and adjust probability.
6121 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6124 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6127 * config/i386/i386.c (ix86_function_naked): New function.
6128 (ix86_can_use_return_insn_p): Return false for naked functions.
6129 (ix86_expand_prologue): Skip prologue for naked functions.
6130 (ix86_expand_epilogue): Skip epilogue for naked functions
6131 and emit trap instruction.
6132 (ix86_warn_func_return): New function.
6133 (ix86_attribute_table): Add "naked" attribute specification.
6134 (TARGET_WARN_FUNC_RETURN): Define.
6135 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6137 2017-07-31 Martin Liska <mliska@suse.cz>
6139 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6140 (dump_gimple_bb_header): Always dump BB info.
6141 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6143 2017-07-31 Martin Liska <mliska@suse.cz>
6146 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6147 also with current_function_decl non-null equality.
6149 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6152 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6153 change type to the element type, instead add eltype variable and
6154 use it where we are interested in the element type.
6156 PR tree-optimization/81603
6157 * ipa-polymorphic-call.c
6158 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6159 offset arithmetic in offset_int, bail out if the resulting bit offset
6160 doesn't fit into shwi.
6162 2017-07-31 Martin Liska <mliska@suse.cz>
6164 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6165 (gimplify_save_expr): Fix comment.
6167 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6170 * config/i386/i386.c (ix86_function_arg): Update arguments for
6172 (ix86_compute_frame_layout): Set the initial stack offset to
6173 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6174 INCOMING_FRAME_SP_OFFSET.
6175 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6176 stack before exception handler returns.
6177 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6178 the 'ERROR_CODE' for exception handler.
6180 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6182 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6183 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6184 (ASM_OUTPUT_REG_POP): Ditto.
6185 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6186 instead of asm_fprintf to output pure string.
6188 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6190 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6191 to imported_module_or_decl hook.
6192 (debug_nothing_tree_tree_tree_bool): Remove.
6193 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6194 * debug.c (do_nothing_debug_hooks): Use
6195 debug_nothing_tree_tree_tree_bool_bool instead of
6196 debug_nothing_tree_tree_tree_bool.
6197 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6198 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6199 * sdbout.c (sdb_debug_hooks): Likewise.
6200 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6201 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6203 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6204 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6205 attribute, don't add anything.
6207 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6209 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6210 (fold_build2_stat_loc): Likewise.
6211 (fold_build3_stat_loc): Likewise.
6212 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6213 (fold_build1_loc): Remove macro.
6214 (fold_build2_loc): Likewise.
6215 (fold_build3_loc): Likewise.
6217 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6219 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6220 (gimple_build_debug_bind_source_stat): Likewise.
6221 * gimple.h (gimple_build_debug_bind): Remove macro.
6222 (gimple_build_debug_bind_source): Likewise.
6224 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6226 * bitmap.c (bitmap_alloc): Adjust.
6227 (bitmap_gc_alloc): Likewise.
6228 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6230 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6232 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6233 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6234 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6235 (bitmap_gc_alloc_stat): Likewise.
6236 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6238 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6240 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6241 * rtl.h (shallow_copy_rtx): Remove macro.
6243 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6245 * emit-rtl.c (gen_raw_REG): Adjust.
6246 * gengenrtl.c (gendef): Likewise.
6247 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6248 * rtl.h (rtx_alloc): Remove macro.
6250 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6252 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6253 (build_tree_list_stat): Likewise.
6254 * tree.h (build_tree_list): Remove macro.
6255 (build_tree_list_vec): Likewise.
6257 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6259 * tree.c (make_vector_stat): Remove _stat from name.
6260 (build_vector_stat): Likewise.
6261 * tree.h (make_vector_stat): Remove macro.
6262 (build_vector_stat): Likewise.
6264 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6266 * tree.h (build_var_debug_value): Remove prototype.
6268 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6270 * tree.c (tree_cons_stat): Remove _stat from name.
6271 * tree.h (tree_cons): Remove macro.
6273 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6275 * tree.c (build_vl_exp_stat): Remove _stat from name.
6276 * tree.h (build_vl_exp): Remove macro.
6278 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6280 * tree.c (build_decl_stat): Remove _stat from name.
6281 * tree.h (build_decl): Remove macro.
6283 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6285 * gimple.c (gimple_build_with_ops_stat): Adjust.
6286 (gimple_alloc_stat): Remove _stat from name.
6287 * gimple.h (gimple_alloc): Remove macro.
6289 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6291 * tree.c (make_tree_vec_stat): Remove _stat from name.
6292 (grow_tree_vec_stat): Likewise.
6293 * tree.h (make_tree_vec_stat): Adjust prototype.
6294 (grow_tree_vec_stat): Likewise.
6295 (make_tree_vec): Remove macro.
6296 (grow_tree_vec): Likewise.
6298 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6300 * fold-const.c (fold_build1_stat_loc): Adjust.
6301 (fold_build2_stat_loc): Likewise.
6302 (fold_build3_stat_loc): Likewise.
6303 * tree.c (build0_stat): Remove _stat from name.
6304 (build1_stat): Likewise.
6305 (build2_stat): Likewise.
6306 (build3_stat): Likewise.
6307 (build4_stat): Likewise.
6308 (build5_stat): Likewise.
6309 * tree.h (build1_loc): Remove macro, and rename _stat function
6311 (build2_loc): Likewise.
6312 (build3_loc): Likewise.
6313 (build4_loc): Likewise.
6314 (build5_loc): Likewise.
6316 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6318 * tree.c (make_int_cst_stat): Remove _stat from name.
6319 * tree.h (make_int_cst_stat): Adjust prototype.
6320 (make_int_cst): Remove macro.
6322 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6324 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6325 * tree.h (make_tree_binfo_stat): Adjust prototype.
6326 (make_tree_binfo): Remove.
6328 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6330 * tree.c (copy_node_stat): Rename to copy_node.
6331 (build_distinct_type_copy): Adjust.
6332 * tree.h (copy_node_stat): Adjust prototype.
6333 (copy_node): Remove macro.
6335 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6337 * tree.c (make_node_stat): rename to make_node.
6338 (build_tree_list_stat): Adjust.
6339 (build0_stat): Likewise.
6340 (build2_stat): Likewise.
6341 (build3_stat): Likewise.
6342 (build4_stat): Likewise.
6343 (build5_stat): Likewise.
6344 (build_decl_stat): Likewise.
6345 * tree.h (make_node_stat): Adjust prototype.
6346 (make_node): remove macro.
6348 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6350 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6351 (PPC_FEATURE2_SCV): Likewise.
6352 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6354 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6356 * config/aarch64/aarch64.c
6357 (aarch64_internal_mov_immediate): Add new special pattern.
6358 * config/aarch64/aarch64.md (*movdi_aarch64):
6359 Add reg/32bit const mov case.
6361 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6362 Richard Sandiford <richard.sandiford@linaro.org>
6364 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6365 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6366 Add integer and movi cases.
6367 (movi-split-hf-df-sf split, fp16): New.
6368 (enabled): Added TARGET_FP_F16INST.
6369 * config/aarch64/iterators.md (GPF_HF): New.
6370 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6372 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6374 * config/aarch64/aarch64.c
6375 (aarch64_simd_container_mode): Add prototype.
6376 (aarch64_expand_mov_immediate): Add HI support.
6377 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6378 (aarch64_can_const_movi_rtx_p): New.
6379 (aarch64_preferred_reload_class):
6380 Remove restrictions of using FP registers for certain SIMD operations.
6381 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6382 (aarch64_valid_floating_const): Add integer move validation.
6383 (aarch64_simd_imm_scalar_p): Remove.
6384 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6385 (aarch64_legitimate_constant_p): Expand list of supported cases.
6386 * config/aarch64/aarch64-protos.h
6387 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6388 (aarch64_reinterpret_float_as_int): New.
6389 (aarch64_simd_imm_scalar_p): Remove.
6390 * config/aarch64/constraints.md (Uvi): New.
6391 (Dd): Split into Ds and new Dd.
6392 * config/aarch64/aarch64.md (*movsi_aarch64):
6394 (*movdi_aarch64): Add SIMD mov case.
6396 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6398 * tree-predcom.c: (struct chain): Handle store-store chain in which
6399 stores for elimination only store loop invariant values.
6400 (execute_pred_commoning_chain): Ditto.
6401 (prepare_initializers_chain_store_elim): Ditto.
6402 (prepare_finalizers): Ditto.
6403 (is_inv_store_elimination_chain): New function.
6404 (initialize_root_vars_store_elim_1): New function.
6406 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6408 * tree-predcom.c: Revise general description of the pass.
6409 (enum chain_type): New enum type for store elimination.
6410 (struct chain): New field supporting store elimination.
6411 (struct component): Ditto.
6412 (dump_chain): Dump store-stores chain.
6413 (release_chain): Release resources.
6414 (split_data_refs_to_components): Compute and create component
6415 contains only stores for elimination.
6416 (get_chain_last_ref_at): New function.
6417 (make_invariant_chain): Initialization.
6418 (make_rooted_chain): Specify chain type in parameter and record it.
6419 (add_looparound_copies): Skip for store-stores chain.
6420 (determine_roots_comp): Compute type of chain and pass it to
6422 (initialize_root_vars_store_elim_2): New function.
6423 (finalize_eliminated_stores): New function.
6424 (remove_stmt): Handle store for elimination.
6425 (execute_pred_commoning_chain): Execute predictive commoning on
6427 (determine_unroll_factor): Skip unroll for store-stores chain.
6428 (prepare_initializers_chain_store_elim): New function.
6429 (prepare_initializers_chain): Hanlde store-store chain.
6430 (prepare_finalizers_chain, prepare_finalizers): New function.
6431 (tree_predictive_commoning_loop): Return integer value indicating
6432 if loop is unrolled or lcssa form is corrupted.
6433 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6435 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6437 * tree-predcom.c (initialize_root): Delete.
6438 (execute_pred_commoning_chain): Initialize root vars and replace
6439 reference of non-combined chain directly, rather than call above
6442 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6444 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6445 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6447 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6449 * tree-predcom.c (struct chain): New field init_seq.
6450 (release_chain): Release init_seq.
6451 (prepare_initializers_chain): Record intialization stmts in above
6453 (insert_init_seqs): New function.
6454 (tree_predictive_commoning_loop): Call insert_init_seqs.
6456 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6458 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6460 2017-07-28 Richard Biener <rguenther@suse.de>
6462 * match.pd: Remove superfluous :c.
6463 * genmatch.c (simplify::id): Add member.
6464 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6466 (current_id): New global.
6467 (dt_node::parent): Move from ...
6468 (dt_operand::parent): ... here. Add for_id member.
6469 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6470 (decision_tree::find_node): Relax order requirement when
6471 merging DT_TRUE nodes to ones inbetween the current simplify
6472 and the one we try to merge with. Add diagnostic whenever
6473 we need to enforce pattern order by not merging.
6474 (decision_tree::insert): Set current_id.
6475 (decision_tree::print_node): Dump parent node and for_id.
6476 (parser::last_id): Add member.
6477 (parser::push_simplify): Assign unique id.
6478 (parser::parser): Initialize last_id.
6480 2017-07-28 Martin Liska <mliska@suse.cz>
6483 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6484 gimple_build_debug_bind.
6486 2017-07-28 Richard Biener <rguenther@suse.de>
6488 PR tree-optimization/81502
6489 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6491 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6493 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6494 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6495 for BIT_FIELD_REF args.
6496 * fold-const.c (make_bit_field_ref): Likewise.
6497 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6499 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6502 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6503 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6504 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6505 Or it into SANITIZER_UNDEFINED.
6506 * ubsan.c: Include gimple-fold.h and varasm.h.
6507 (ubsan_expand_ptr_ifn): New function.
6508 (instrument_pointer_overflow): New function.
6509 (maybe_instrument_pointer_overflow): New function.
6510 (instrument_object_size): Formatting fix.
6511 (pass_ubsan::execute): Call instrument_pointer_overflow
6512 and maybe_instrument_pointer_overflow.
6513 * internal-fn.c (expand_UBSAN_PTR): New function.
6514 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6515 * sanitizer.def (__ubsan_handle_pointer_overflow,
6516 __ubsan_handle_pointer_overflow_abort): New builtins.
6517 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6518 * internal-fn.def (UBSAN_PTR): New internal function.
6519 * opts.c (sanitizer_opts): Add pointer-overflow.
6520 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6521 * fold-const.c (build_range_check): Compute pointer range check in
6522 integral type if pointer arithmetics would be needed. Formatting
6525 2017-07-28 Martin Liska <mliska@suse.cz>
6528 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6529 parameters that are of a variable-length.
6531 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6533 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6535 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6536 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6537 (CRT_CALL_STATIC_FUNCTION): Likewise.
6538 (ASM_DEFAULT_SPEC): New define.
6539 (ASM_SPEC32): Likewise.
6540 (ASM_SPEC64): Likewise.
6541 (ASM_SPEC_COMMON): Likewise.
6542 (ASM_SPEC): Likewise.
6543 (INVALID_64BIT): Likewise.
6544 (LINK_OS_DEFAULT_SPEC): Likewise.
6545 (LINK_OS_SPEC32): Likewise.
6546 (LINK_OS_SPEC64): Likewise.
6547 (POWERPC_LINUX): Likewise.
6548 (PTRDIFF_TYPE): Likewise.
6549 (RESTORE_FP_PREFIX): Likewise.
6550 (RESTORE_FP_SUFFIX): Likewise.
6551 (SAVE_FP_PREFIX): Likewise.
6552 (SAVE_FP_SUFFIX): Likewise.
6553 (SIZE_TYPE): Likewise.
6554 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6555 (TARGET_64BIT): Likewise.
6556 (TARGET_64BIT): Likewise.
6557 (TARGET_AIX): Likewise.
6558 (WCHAR_TYPE_SIZE): Likewise.
6559 (WCHAR_TYPE): Undefine.
6560 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6561 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6562 (CPP_OS_RTEMS_SPEC): Delete.
6563 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6564 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6566 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6568 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6570 PR tree-optimization/81578
6571 * tree-parloops.c (build_new_reduction): Bail out if
6572 reduction_code isn't one of the standard OpenMP reductions.
6573 Move the details printing after that decision.
6575 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6577 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6578 related to reload_in_progress.
6579 (splat_input_operand): Likewise.
6580 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6582 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6584 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6585 (TARGET_INSTANTIATE_DECLS): Likewise.
6586 (legitimate_indexed_address_p): Delete reload_in_progress code.
6587 (rs6000_debug_legitimate_address_p): Likewise.
6588 (rs6000_eliminate_indexed_memrefs): Likewise.
6589 (rs6000_emit_le_vsx_store): Likewise.
6590 (rs6000_emit_move_si_sf_subreg): Likewise.
6591 (rs6000_emit_move): Likewise.
6592 (register_to_reg_type): Likewise.
6593 (rs6000_pre_atomic_barrier): Likewise.
6594 (rs6000_machopic_legitimize_pic_address): Likewise.
6595 (rs6000_allocate_stack_temp): Likewise.
6596 (rs6000_address_for_fpconvert): Likewise.
6597 (rs6000_address_for_altivec): Likewise.
6598 (rs6000_secondary_memory_needed_rtx): Delete function.
6599 (rs6000_check_sdmode): Likewise.
6600 (rs6000_alloc_sdmode_stack_slot): Likewise.
6601 (rs6000_instantiate_decls): Likewise.
6602 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6603 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6604 Delete reload_in_progress.
6605 (*vec_reload_and_plus_<mptrsize>): Likewise.
6606 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6607 (vsx_div_v2di): Likewise.
6608 (vsx_udiv_v2di): Likewise.
6610 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6612 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6613 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6614 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6615 (rs6000_debug_reg_global): Delete print of LRA status.
6616 (rs6000_option_override_internal): Delete dead LRA related code.
6617 (rs6000_lra_p): Delete function.
6618 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6620 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6622 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6623 * config/riscv/rtems.h: New file.
6625 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6626 Sudakshina Das <sudi.das@arm.com>
6628 * config/aarch64/aarch64.md
6629 (define_split for and<mode>3nr_compare): Move
6630 non aarch64_logical_operand to a register.
6631 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6632 register immediate operand to a register.
6633 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6635 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6638 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6640 2017-07-27 Richard Biener <rguenther@suse.de>
6642 PR tree-optimization/81573
6643 PR tree-optimization/81494
6644 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6645 multi defuse cycle case.
6647 2017-07-27 Richard Biener <rguenther@suse.de>
6649 PR tree-optimization/81571
6650 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6653 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6655 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6656 earlier and only if MASK_FPU is set. Adjust formatting.
6658 2017-07-27 Martin Liska <mliska@suse.cz>
6660 * opt-functions.awk: Add validation of value of Init.
6661 * optc-gen.awk: Pass new argument.
6663 2017-07-27 Martin Liska <mliska@suse.cz>
6665 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6666 Fix wrong condition.
6668 2017-07-27 Martin Liska <mliska@suse.cz>
6670 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6671 BBs and edges seen by autoFDO.
6673 2017-07-27 Richard Biener <rguenther@suse.de>
6675 PR tree-optimization/81502
6676 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6677 with incompatible but same sized type.
6678 (execute_update_addresses_taken): Likewise.
6680 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6682 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6683 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6685 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6688 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6689 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6690 Change s_operand to memory_operand.
6692 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6694 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6695 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6696 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6697 Emit instructions rather than returning an expression. Handle TFmode
6698 and KFmode by casting to TImode.
6699 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6700 (rs6000_emit_le_vsx_store): Likewise.
6701 * config/rs6000/vsx.md (VSX_TI): New iterator.
6702 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6703 (*vsx_le_undo_permute_<mode>): Likewise.
6704 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6705 emit the split sequence.
6706 (*vsx_le_perm_store_<mode>): Likewise.
6708 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6710 PR tree-optimization/81555
6711 PR tree-optimization/81556
6712 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6713 if true, force CHANGED for the recursive invocation.
6714 (reassociate_bb): Remember original length of ops array, pass
6715 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6717 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6718 attributes for noipa attribute. For naked attribute use
6719 lookup_attribute first before lookup_attribute_spec.
6720 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6722 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6723 for functions with noipa attribute.
6724 (cgraph_externally_visible_p): Return true for functions with noipa
6726 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6727 for functions with noipa attribute.
6728 * doc/extend.texi: Document noipa function attribute.
6729 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6730 also for functions with noipa attribute.
6731 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6733 2017-07-26 Andrew Pinski <apinski@cavium.com>
6735 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6736 vec_unalign_load_cost and vec_unalign_store_cost.
6738 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6740 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6741 -mvsx-small-integer option.
6742 (ISA_3_0_MASKS_IEEE): Likewise.
6743 (OTHER_VSX_VECTOR_MASKS): Likewise.
6744 (POWERPC_MASKS): Likewise.
6745 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6746 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6747 code, only testing for DImode being allowed in non-VSX floating
6749 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6750 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6752 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6753 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6754 TARGET_P8_VECTOR test.
6755 (rs6000_secondary_reload_simple_move): Likewise.
6756 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6757 since TARGET_P9_VECTOR was already tested.
6758 (rs6000_opt_masks): Remove -mvsx-small-integer.
6759 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6760 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6762 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6763 test for TARGET_VEXTRACTUB was used, and that uses
6765 (p9 extract splitter): Likewise.
6766 (vsx_extract_<mode>_di_p9): Likewise.
6767 (vsx_extract_<mode>_store_p9): Likewise.
6768 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6769 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6770 the elimination of TARGET_VSX_SMALL_INTEGER.
6771 (vsx_extract_<mode>_p8): Likewise.
6772 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6773 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6774 (vsx_set_<mode>_p9): Likewise.
6775 (vsx_set_v4sf_p9): Likewise.
6776 (vsx_set_v4sf_p9_zero): Likewise.
6777 (vsx_insert_extract_v4sf_p9): Likewise.
6778 (vsx_insert_extract_v4sf_p9_2): Likewise.
6779 * config/rs6000/rs6000.md (sign extend splitter): Change
6780 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6781 (floatsi<mode>2_lfiwax_mem): Likewise.
6782 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6783 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6784 since a test for TARGET_P9_VECTOR was used.
6785 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6786 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6787 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6788 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6789 TARGET_P8_VECTOR test.
6790 (fix_trunc<mode>si2_stfiwx): Likewise.
6791 (fix_trunc<mode>si2_internal): Likewise.
6792 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6793 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6795 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6796 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6797 TARGET_P8_VECTOR test.
6798 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6799 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6800 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6802 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6803 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6804 since a test for TARGET_P9_VECTOR was used.
6805 (splitter for loading small constants): Likewise.
6807 2017-07-26 Andrew Pinski <apinski@cavium.com>
6809 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6812 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6815 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6816 (fp_valid_at): Likewise.
6818 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6820 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6821 (qdf24xx_addrcost_table): Likewise.
6822 (cortexa57_tunings): Update to use generic_branch_cost.
6823 (cortexa72_tunings): Likewise.
6824 (cortexa73_tunings): Likewise.
6825 (qdf24xx_tunings): Likewise.
6827 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6829 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6830 (thunderx2t99_branch_cost): Likewise.
6831 (cortexa35_tunings): Update to use generic_branch_cost.
6832 (cortexa53_tunings): Likewise.
6833 (cortexa57_tunings): Likewise.
6834 (cortexa72_tunings): Likewise.
6835 (cortexa73_tunings): Likewise.
6836 (thunderx2t99_tunings): Likewise.
6838 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6840 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6841 (sparc_option_override): Honour MASK_FSMULD.
6842 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6843 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6844 * config/sparc/sparc.opt (mfsmuld): New option.
6845 * doc/invoke.texi (mfsmuld): Document option.
6847 2017-07-26 Marek Polacek <polacek@redhat.com>
6850 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6852 2017-07-26 Richard Biener <rguenther@suse.de>
6854 * gimple-match-head.c (do_valueize): Return OP if valueize
6856 (get_def): New helper to get at the def stmt of a SSA name
6858 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6859 do_valueize to get at the def stmt.
6860 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6862 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6865 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6867 2017-07-26 Martin Liska <mliska@suse.cz>
6870 * function.c (expand_function_start): Make expansion of
6871 nonlocal_goto_save_area after parm_birth_insn.
6873 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6875 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6876 from all CPU target flags enable members.
6878 2017-07-26 Richard Biener <rguenther@suse.de>
6880 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6881 (decision_tree::gen): Make 'type' const.
6882 (write_predicate): Likewise.
6884 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6886 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6887 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6888 (rs6000_option_override_internal): Likewise.
6889 (rs6000_expand_vector_set): Likewise.
6890 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6891 (TARGET_UPPER_REGS_SF): Likewise.
6892 (TARGET_UPPER_REGS_DI): Likewise.
6893 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6894 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6895 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6896 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6897 (Splitters for DI constants in Altivec registers): Likewise.
6898 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6899 (vsx_set_v4sf_p9): Likewise.
6900 (vsx_set_v4sf_p9_zero): Likewise.
6901 (vsx_insert_extract_v4sf_p9): Likewise.
6902 (vsx_insert_extract_v4sf_p9_2): Likewise.
6904 2017-07-25 Carl Love <cel@us.ibm.com>
6906 * doc/extend.texi: Update the built-in documentation file for the
6907 existing built-in functions
6908 vector signed char vec_cnttz (vector signed char);
6909 vector unsigned char vec_cnttz (vector unsigned char);
6910 vector signed short vec_cnttz (vector signed short);
6911 vector unsigned short vec_cnttz (vector unsigned short);
6912 vector signed int vec_cnttz (vector signed int);
6913 vector unsigned int vec_cnttz (vector unsigned int);
6914 vector signed long long vec_cnttz (vector signed long long);
6915 vector unsigned long long vec_cnttz (vector unsigned long long);
6917 2017-07-25 Andrew Pinski <apinski@cavium.com>
6919 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6920 accesses where the use is for the first operand of a BIT_INSERT.
6922 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6925 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6926 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6928 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6930 * config/i386/gstabs.h: Delete.
6931 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6933 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6935 * config/i386/i386.c (ix86_decompose_address): Do not check for
6936 register RTX when looking at index_reg or base_reg.
6937 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6939 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6941 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6942 to update EH info here.
6944 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6946 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6948 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6950 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6952 2017-07-25 Torsten Duwe <duwe@suse.de>
6954 * common.opt: Introduce -fpatchable-function-entry
6955 command line option, and its variables function_entry_patch_area_size
6956 and function_entry_patch_area_start.
6957 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6958 including a two-value parser.
6959 * target.def (print_patchable_function_entry): New target hook.
6960 * targhooks.h (default_print_patchable_function_entry): New function.
6961 * targhooks.c (default_print_patchable_function_entry): Likewise.
6962 * toplev.c (process_options): Switch off IPA-RA if
6963 patchable function entries are being generated.
6964 * varasm.c (assemble_start_function): Look at the
6965 patchable-function-entry command line switch and current
6966 function attributes and maybe generate NOP instructions by
6967 calling the print_patchable_function_entry hook.
6968 * doc/extend.texi: Document patchable_function_entry attribute.
6969 * doc/invoke.texi: Document -fpatchable_function_entry
6970 command line option.
6971 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6973 * doc/tm.texi: Re-generate.
6975 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6978 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6979 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6981 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6983 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6985 2017-07-25 Richard Biener <rguenther@suse.de>
6987 PR tree-optimization/81455
6988 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6989 not walk in cycles when looking for guards.
6991 2017-07-25 Richard Biener <rguenther@suse.de>
6993 PR tree-optimization/81529
6994 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6995 when optimizing backedge uses.
6997 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6999 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7001 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7002 to dl_section_ref. On AIX, append an expression to subtract
7003 the size of the section length to dl_section_ref.
7005 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7007 * configure.ac: If any of the config.* scripts fail, exit 1.
7008 * configure: Regenerate.
7010 2017-07-25 Richard Biener <rguenther@suse.de>
7013 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7014 of immediate uses, be more verbose on errors.
7016 2017-07-25 Richard Biener <rguenther@suse.de>
7018 PR tree-optimization/81510
7019 * tree-vect-loop.c (vect_is_simple_reduction): When the
7020 reduction stmt is not inside the loop bail out.
7022 2017-07-25 Richard Biener <rguenther@suse.de>
7024 PR tree-optimization/81303
7025 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7026 profitability check against LOOP_VINFO_NITERSM1.
7028 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7030 * domwalk.c (cmp_bb_postorder): Simplify.
7031 (sort_bbs_postorder): New function. Use it...
7032 (dom_walker::walk): ...here to optimize common cases.
7034 2017-07-25 Martin Liska <mliska@suse.cz>
7037 * ipa-visibility.c (function_and_variable_visibility): Make the
7038 redirection just on target that supports aliasing.
7039 Fix GNU coding style.
7041 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7044 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7046 * config/arm/rtems.h: Likewise.
7047 * config/bfin/rtems.h: Likewise.
7048 * config/i386/rtemself.h: Likewise.
7049 * config/lm32/rtems.h: Likewise.
7050 * config/m32c/rtems.h: Likewise.
7051 * config/m68k/rtemself.h: Likewise.
7052 * config/microblaze/rtems.h: Likewise.
7053 * config/mips/rtems.h: Likewise.
7054 * config/moxie/rtems.h: Likewise.
7055 * config/nios2/rtems.h: Likewise.
7056 * config/powerpcspe/rtems.h: Likewise.
7057 * config/rs6000/rtems.h: Likewise.
7058 * config/rtems.h: Likewise.
7059 * config/sh/rtems.h: Likewise.
7060 * config/sh/rtemself.h: Likewise.
7061 * config/sparc/rtemself.h: Likewise.
7063 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7066 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7067 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7068 * tree-ssa-structalias.c (alias_get_name): Same.
7070 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7073 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7074 instructions if no du chain is found.
7076 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7078 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7080 2017-07-25 Richard Biener <rguenther@suse.de>
7083 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7086 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7088 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7090 (ISA_2_7_MASKS_SERVER): Likewise.
7091 (ISA_3_0_MASKS_IEEE): Likewise.
7092 (OTHER_P8_VECTOR_MASKS): Likewise.
7093 (OTHER_VSX_VECTOR_MASKS): Likewise.
7094 (POWERPC_MASKS): Likewise.
7095 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7096 duplicate list of options.
7097 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7098 explicit -mupper-regs options.
7099 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7100 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7101 alias for -mupper-regs-df.
7102 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7103 (rs6000_init_hard_regno_mode_ok): Likewise.
7104 (rs6000_option_override_internal): Likewise.
7105 (rs6000_opt_masks): Likewise.
7106 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7107 options in terms of whether -mvsx or -mpower8-vector was used.
7108 (TARGET_UPPER_REGS_DI): Likewise.
7109 (TARGET_UPPER_REGS_SF): Likewise.
7110 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7111 -mupper-regs-* options.
7113 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7115 * passes.c (emergency_dump_function): Print some empty lines and a
7116 header before the RTL dump.
7118 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7120 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7122 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7125 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7126 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7128 2017-07-24 Carl Love <cel@us.ibm.com>
7130 * config/rs6000/rs6000-c.c: Add support for built-in functions
7131 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7132 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7133 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7134 vec_extract_fp_from_shortl): Add defines for the two builtins.
7135 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7136 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7138 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7139 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7140 * doc/extend.texi: Update the built-in documentation file for the
7141 new built-in function.
7143 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7146 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7148 * doc/generic.texi: Likewise.
7149 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7150 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7152 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7154 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7155 (aarch64_mls_elt_merge<mode>): Likewise.
7157 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7159 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7160 having __cxa_atexit.
7162 2017-07-23 Michael Collison <michael.collison@arm.com>
7164 * config/arm/arm.c (arm_option_override): Deprecate
7165 use of -mstructure-size-boundary.
7166 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7167 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7169 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7172 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7173 Reduce cost estimate for direct moves.
7175 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7178 * config/i386/i386.c (ix86_option_override_internal): Disable
7179 BMI, BMI2 and TBM instructions for -m16.
7181 2017-07-21 Carl Love <cel@us.ibm.com>
7183 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7184 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7185 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7186 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7187 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7188 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7189 VMULOSW): New enum "unspec" values.
7190 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7191 altivec_vmulosw): New patterns.
7192 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7193 VMULOSW): Add definitions.
7195 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7197 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7198 (qdf24xx): Likewise.
7199 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7200 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7201 (AARCH64_FL_V8_1): Renumber.
7202 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7203 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7204 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7205 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7206 rdma to feature modifiers list.
7208 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7211 * ipa-visibility (function_and_variable_visibility): Convert
7212 recursive PLT call to direct call if appropriate.
7214 2017-07-21 Andrew Pinski <apinski@cavium.com>
7216 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7217 operand 1 to see if the types precision matches.
7218 * fold-const.c (operand_equal_p): Likewise.
7220 2017-07-21 Richard Biener <rguenther@suse.de>
7222 PR tree-optimization/81303
7223 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7224 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7225 (vect_peeling_hash_get_lowest_cost): Adjust.
7226 (vect_enhance_data_refs_alignment): Likewise. Use
7227 vect_get_peeling_costs_all_drs to compute the penalty for no
7228 peeling to match up costs.
7230 2017-07-21 Richard Biener <rguenther@suse.de>
7232 PR tree-optimization/81500
7233 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7234 we didn't identify a reduction path.
7236 2017-07-21 Tom de Vries <tom@codesourcery.com>
7237 Cesar Philippidis <cesar@codesourcery.com>
7239 PR gcov-profile/81442
7240 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7243 2017-07-21 Tom de Vries <tom@codesourcery.com>
7246 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7248 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7249 nvptx_override_options_after_change.
7251 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7253 * dwarf2out.c (output_file_names): Avoid double testing for
7256 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7258 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7260 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7262 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7264 (try_crossjump_to_edge): Do not punt on partitioned functions.
7266 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7268 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7269 Put all BBs reachable only via paths crossing cold region to cold
7271 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7273 2016-07-21 Richard Biener <rguenther@suse.de>
7275 PR tree-optimization/81303
7276 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7277 into account prologue and epilogue iterations when raising
7278 min_profitable_iters to sth at least covering one vector iteration.
7280 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7282 * config/arm/arm.c (arm_test_cpu_arch_dat):
7285 2017-07-20 Nathan Sidwell <nathan@acm.org>
7287 Remove TYPE_METHODS.
7288 * tree.h (TYPE_METHODS): Delete.
7289 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7290 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7291 (dbxout_type_methods): Scan TYPE_FIELDS.
7292 (dbxout_type): Don't check TYPE_METHODS here.
7293 * function.c (use_register_for_decl): Always ignore register for
7294 class types when not optimizing.
7295 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7296 * tree.c (free_lang_data_in_type): Stitch out member functions and
7297 templates from TYPE_FIELDS.
7298 (build_distinct_type_copy, verify_type_variant,
7299 verify_type): Member fns are on TYPE_FIELDS.
7300 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7301 * tree-pretty-print.c (dump_generic_node): Likewise.
7303 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7306 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7307 V2TImode and V4TImode.
7308 (ix86_expand_vector_extract): Likewise.
7309 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7310 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7311 (ssescalarmode): Handle V4TImode and V2TImode.
7312 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7313 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7314 (VEXTRACTI128_MODE): New mode iterator.
7315 (splitter for *vec_extractv?ti first element): New.
7316 (VEC_INIT_MODE): New mode iterator.
7317 (vec_init<mode>): Consolidate 3 expanders into one using
7318 VEC_INIT_MODE mode iterator.
7320 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7322 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7323 non_spilled_static_chain_regno_p.
7325 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7327 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7329 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7331 * bb-reorder.c (connect_traces): Allow copying of blocks within
7334 2017-07-20 Richard Biener <rguenther@suse.de>
7336 * gimple.h (gimple_phi_result): Add gphi * overload.
7337 (gimple_phi_result_ptr): Likewise.
7338 (gimple_phi_arg): Likewise. Adjust index assert to only
7339 allow actual argument accesses rather than all slots available
7341 (gimple_phi_arg_def): Add gphi * overload.
7342 * tree-phinodes.c (make_phi_node): Initialize only actual
7344 (resize_phi_node): Clear memory not covered by old node,
7345 do not initialize excess argument slots.
7346 (reserve_phi_args_for_new_edge): Initialize new argument slot
7349 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7351 PR tree-optimization/81388
7353 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7355 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7356 by removing computation of may_be_zero.
7358 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7359 Tom de Vries <tom@codesourcery.com>
7362 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7363 when gimple level profile disagrees with what RTL expander did.
7365 2017-07-20 Richard Biener <rguenther@suse.de>
7367 PR tree-optimization/61171
7368 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7369 (vect_analyze_stmt): Add slp instance parameter.
7370 (vectorizable_reduction): Likewise.
7371 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7372 (vect_is_simple_reduction): Deal with chains not detected
7373 as SLP reduction chain, specifically not properly associated
7374 chains containing a mix of plus/minus.
7375 (get_reduction_op): Remove.
7376 (get_initial_defs_for_reduction): Simplify, pass in whether
7377 this is a reduction chain, pass in the SLP node for the PHIs.
7378 (vect_create_epilog_for_reduction): Get the SLP instance as
7380 (vectorizable_reduction): Get the SLP instance as arg.
7381 During analysis remember the SLP node with the PHIs in the
7382 instance. Simplify getting at the vectorized reduction PHIs.
7383 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7384 through SLP instance.
7385 (vect_slp_analyze_operations): Likewise.
7386 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7387 (vect_transform_stmt): Likewise.
7389 2017-07-20 Tom de Vries <tom@codesourcery.com>
7391 PR tree-optimization/81489
7392 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7393 read of phi arg location to before loop that modifies phi.
7395 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7397 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7400 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7403 * except.c (execute): Fix ordering issue.
7405 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7407 PR rtl-optimization/81423
7408 * combine.c (make_compound_operation_int): Don't try to optimize
7409 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7411 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7413 PR rtl-optimization/81423
7414 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7415 with a constant that is -1 in the truncated to mode.
7417 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7419 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7420 (determine_unlikely_bbs): ... here.
7421 * predict.h (propagate_unlikely_bbs_forward): Declare.
7422 * cfgexpand.c (pass_expand::execute): Use it.
7423 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7425 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7426 propagate_unlikely_bbs_forward.
7428 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7431 * except.c (maybe_add_nop_after_section_switch): New function.
7434 2017-07-19 Tom de Vries <tom@codesourcery.com>
7436 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7438 2017-07-19 Tom de Vries <tom@codesourcery.com>
7440 * gimple.h (gimple_phi_arg): Make assert more strict.
7442 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7444 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7445 * config/rs6000/mmintrin.h: New file.
7446 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7448 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7450 PR tree-optimization/81346
7451 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7453 2017-07-19 Tom de Vries <tom@codesourcery.com>
7455 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7457 2017-07-19 Tom de Vries <tom@codesourcery.com>
7459 * config/nvptx/nvptx-modes.def: Add V2DImode.
7460 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7461 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7462 (nvptx_output_mov_insn): Handle lack of mov.b128.
7463 (nvptx_print_operand): Handle 'H' and 'L' codes.
7464 (nvptx_vector_mode_supported): Allow V2DImode.
7465 (nvptx_preferred_simd_mode): New function.
7466 (nvptx_data_alignment): New function.
7467 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7468 nvptx_preferred_simd_mode.
7469 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7471 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7473 2017-07-19 Tom de Vries <tom@codesourcery.com>
7475 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7476 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7477 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7478 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7479 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7480 (mov<VECIM>_insn): New define_insn.
7481 (define_expand "mov<VECIM>): New define_expand.
7483 2017-07-19 Tom de Vries <tom@codesourcery.com>
7485 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7487 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7489 PR tree-optimization/81346
7490 * fold-const.h (fold_div_compare, range_check_type): Declare.
7491 * fold-const.c (range_check_type): New function.
7492 (build_range_check): Use range_check_type.
7493 (fold_div_compare): No longer static, rewritten into
7494 a match.pd helper function.
7495 (fold_comparison): Don't call fold_div_compare here.
7496 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7499 2017-07-19 Nathan Sidwell <nathan@acm.org>
7501 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7502 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7503 * tree.c (find_decls_types_r, verify_type): Use
7504 TYPE_{MIN,MAX}_VALUE_RAW.
7505 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7506 (hash_tree): Likewise.
7507 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7509 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7512 2017-07-18 Tom de Vries <tom@codesourcery.com>
7515 * omp-expand.c (expand_omp_for_static_chunk): Handle
7516 equal-argument loop exit phi.
7518 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7521 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7522 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7523 operand 2 predicate.
7524 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7525 operand 2 predicate.
7526 (ror,rol -> rorx splitters): Use const_int_operand as
7527 operand 2 predicate.
7529 2017-06-18 Richard Biener <rguenther@suse.de>
7531 PR tree-optimization/81410
7532 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7533 the gap in the ! slp_perm SLP case after each group.
7535 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7538 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7541 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7544 * predict.c (set_even_probabilities): Cleanup; do not affect
7545 probabilities that are already known.
7546 (combine_predictions_for_bb): Call even when count is set.
7548 2017-07-18 Nathan Sidwell <nathan@acm.org>
7550 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7553 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7556 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7557 optimization for loop niter analysis.
7559 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7562 * config/avr/avr.c (avr_optimize_casesi): Don't use
7563 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7565 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7567 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7568 body_cost_vec from _vect_peel_extended_info.
7569 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7570 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7573 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7575 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7577 2017-07-18 Richard Biener <rguenther@suse.de>
7579 PR tree-optimization/80620
7580 PR tree-optimization/81403
7581 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7582 info when re-using a VN table entry.
7584 2017-07-18 Richard Biener <rguenther@suse.de>
7586 PR tree-optimization/81418
7587 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7588 vectype_in. Verify that with lane-reducing reduction operations
7589 we have a single def-use cycle.
7591 2017-07-17 Carl Love <cel@us.ibm.com>
7593 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7595 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7596 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7597 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7598 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7599 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7600 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7601 VMULOSW): New enum "unspec" values.
7602 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7603 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7604 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7605 altivec_vmulosw): New patterns.
7606 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7607 VMULOSW): Add definitions.
7608 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7610 * config/alpha/alpha.c: Include predict.h.
7612 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7614 * tree-vrp.c (compare_assert_loc): Fix comparison function
7615 to return predictable results.
7617 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7619 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7622 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7623 * doc/invoke.texi (mexpand-adddi): Update text.
7625 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7627 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7628 that also clobbers the CC register. The old expand code is moved
7630 (*arc_clzsi2): ... here.
7631 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7632 the CC register. The old expand code is moved to ...
7633 (arc_ctzsi2): ... here.
7635 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7637 * config/arc/arc.opt (mindexed-loads): Use initial value
7638 TARGET_INDEXED_LOADS_DEFAULT.
7639 (mauto-modify-reg): Use initial value
7640 TARGET_AUTO_MODIFY_REG_DEFAULT.
7641 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7642 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7643 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7644 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7646 2017-07-17 Martin Liska <mliska@suse.cz>
7649 * opts.c (finish_options): Do not allow -fgnu-tm
7650 w/ -fsanitize={kernel-,}address. Say sorry.
7652 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7655 * tree-loop-distribution.c (classify_partition): Only assert on
7656 numer of iterations.
7657 (merge_dep_scc_partitions): Delete prameter. Update function call.
7658 (distribute_loop): Remove code handling loop with unknown niters.
7659 (pass_loop_distribution::execute): Skip loop with unknown niters.
7661 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7664 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7665 function sort_partitions_by_post_order.
7667 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7669 PR tree-optimization/81374
7670 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7671 the max index of basic blocks, rather than number of basic blocks.
7673 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7675 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7677 (arc_legitimate_pic_operand_p): Likewise.
7678 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7680 (arc_needs_pcl_p): Likewise.
7681 (arc_legitimate_pc_offset_p): Likewise.
7682 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7683 function is also used in constrains.md.
7684 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7685 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7686 PLUS. Only return true/false in known cases, otherwise assert.
7687 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7688 is already called in arc_legitimate_constant_p.
7689 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7691 (LEGITIMATE_PIC_OPERAND_P): Use
7692 arc_raw_symbolic_reference_mentioned_p function.
7693 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7698 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7699 Andrew Burgess <andrew.burgess@embecosm.com>
7701 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7702 (arc_return_address_register): New function.
7703 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7704 (arc_handle_fndecl_attribute): Add naked attribute.
7705 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7706 (TARGET_WARN_FUNC_RETURN): Likewise.
7707 (arc_allocate_stack_slots_for_args): New function.
7708 (arc_warn_func_return): Likewise.
7709 (machine_function): Change type fn_type.
7710 (arc_compute_function_type): Consider new naked function type,
7711 change function return type.
7712 (arc_must_save_register): Adapt to handle new
7713 arc_compute_function_type's return type.
7714 (arc_expand_prologue): Likewise.
7715 (arc_expand_epilogue): Likewise.
7716 (arc_return_address_regs): Delete.
7717 (arc_return_address_register): New function.
7718 (arc_epilogue_uses): Use above function.
7719 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7720 (arc_function_type): Change encoding, add naked type.
7721 (ARC_INTERRUPT_P): Change to handle the new encoding.
7722 (ARC_FAST_INTERRUPT_P): Likewise.
7723 (ARC_NORMAL_P): Define.
7724 (ARC_NAKED_P): Likewise.
7725 (arc_compute_function_type): Delete prototype.
7726 * config/arc/arc.md (in_ret_delay_slot): Use
7727 arc_return_address_register function.
7728 (simple_return): Likewise.
7729 (p_return_i): Likewise.
7731 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7733 PR tree-optimization/81428
7734 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7735 can't be built for those types.
7737 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7739 Remove stuff dead since r239246.
7741 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7742 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7743 (avr_inform_devices): Remove dead stuff.
7745 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7747 * config/arm/arm_neon.h: Fix softp typo.
7749 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7751 PR tree-optimization/81365
7752 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7753 aggregate moves onto bb predecessor edges, make sure there are no
7754 loads that could alias the lhs in between the start of bb and the
7757 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7760 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7761 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7762 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7764 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7766 PR tree-optimization/81396
7767 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7768 (init_symbolic_number): Initialize it to 1.
7769 (perform_symbolic_merge): Add n_ops from both operands into the new
7771 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7772 without base_addr as useless if they need more than one operation.
7773 (bswap_replace): Handle !bswap case for NULL base_addr.
7775 2017-07-17 Tom de Vries <tom@codesourcery.com>
7778 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7781 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7783 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7784 conditional builtin define __FIX_LEON3FT_B2BST.
7786 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7788 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7789 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7792 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7794 PR rtl-optimization/81424
7795 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7796 to remove potential trapping from operands if -fnon-call-exceptions.
7798 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7800 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7801 profile_proability for scalling.
7802 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7804 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7806 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7808 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7810 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7811 fixpoint arithmetics.
7813 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7815 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7816 fixpoint arithmetics.
7818 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7820 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7821 fixpoint arithmetics.
7823 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7825 * profile-count.h (profile_probability::from_reg_br_prob_note,
7826 profile_probability::to_reg_br_prob_note): New functions.
7827 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7828 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7829 * predict.c (probability_reliable_p): Update.
7830 (edge_probability_reliable_p): Update.
7831 (br_prob_note_reliable_p): Update.
7832 (invert_br_probabilities): Update.
7833 (add_reg_br_prob_note): New function.
7834 (combine_predictions_for_insn): Update.
7835 * asan.c (asan_clear_shadow): Update.
7836 * cfgbuild.c (compute_outgoing_frequencies): Update.
7837 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7838 (update_br_prob_note): Update.
7839 (rtl_verify_edges): Update.
7840 (purge_dead_edges): Update.
7841 (fixup_reorder_chain): Update.
7842 * emit-rtl.c (try_split): Update.
7843 * ifcvt.c (cond_exec_process_insns): Update.
7844 (cond_exec_process_if_block): Update.
7845 (dead_or_predicable): Update.
7846 * internal-fn.c (expand_addsub_overflow): Update.
7847 (expand_neg_overflow): Update.
7848 (expand_mul_overflow): Update.
7849 * loop-doloop.c (doloop_modify): Update.
7850 * loop-unroll.c (compare_and_jump_seq): Update.
7851 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7852 * predict.h: Update.
7853 * reorg.c (mostly_true_jump): Update.
7855 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7856 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7857 * config/arc/arc.c: (emit_unlikely_jump): Update.
7858 * config/arm/arm.c: (emit_unlikely_jump): Update.
7859 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7860 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7861 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7862 (ix86_print_operand): Update.
7863 (ix86_split_fp_branch): Update.
7864 (predict_jump): Update.
7865 * config/ia64/ia64.c (ia64_print_operand): Update.
7866 * config/mmix/mmix.c (mmix_print_operand): Update.
7867 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7868 (rs6000_expand_split_stack_prologue): Update.
7869 * config/rs6000/rs6000.c: Update.
7870 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7871 (s390_expand_vec_movstr): Update.
7872 (s390_expand_cs_tdsi): Update.
7873 (s390_expand_split_stack_prologue): Update.
7874 * config/sh/sh.c (sh_print_operand): Update.
7875 (expand_cbranchsi4): Update.
7876 (expand_cbranchdi4): Update.
7877 * config/sparc/sparc.c (output_v9branch): Update.
7878 * config/spu/spu.c (get_branch_target): Update.
7879 (ea_load_store_inline): Update.
7880 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7881 * config/tilepro/tilepro.c: Update.
7883 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7885 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7886 (gimplify_save_expr): Likewise.
7888 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7890 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7892 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7894 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7896 * timevar.def (TV_IPA_FNSUMMARY): Define.
7898 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7900 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7901 to back store errata sensitive sequence from being generated.
7902 (sqrtdf2_fix): Likewise.
7904 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7906 * tree-ssa-threadupdate.c (compute_path_counts,
7907 update_joiner_offpath_counts): Use profile_probability.
7909 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7912 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7914 * config/arm/arm-c.c (arm_cpu_builtins): Define
7915 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7917 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7919 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7920 array entries to represent __ieee128 versions of the
7921 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7922 scalar_extract_sig, and scalar_insert_exp built-in functions.
7923 (altivec_resolve_overloaded_builtin): Add special case handling
7924 for the __builtin_scalar_insert_exp function, as represented by
7925 the P9V_BUILTIN_VEC_VSIEDP constant.
7926 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7927 exponent support for __ieee128 argument.
7928 (VSESQP): Add scalar extract signature support for __ieee128
7930 (VSTDCNQP): Add scalar test negative support for __ieee128
7932 (VSIEQP): Add scalar insert exponent support for __int128 argument
7933 with __ieee128 result.
7934 (VSIEQPF): Add scalar insert exponent support for __ieee128
7935 argument with __ieee128 result.
7936 (VSTDCQP): Add scalar test data class support for __ieee128
7938 (VSTDCNQP): Add overload support for scalar test negative with
7940 (VSTDCQP): Add overload support for scalar test data class
7942 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7944 (UNSPEC_VSX_SIEXPQP): New constant.
7945 (xsxexpqp): New insn for VSX scalar extract exponent quad
7947 (xsxsigqp): New insn for VSX scalar extract significand quad
7949 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7950 precision with floating point argument.
7951 (xststdcqp): New expand for VSX scalar test data class quad
7953 (xststdcnegqp): New expand for VSX scalar test negative quad
7955 (xststdcqp): New insn to match expansions for VSX scalar test data
7956 class quad precision and VSX scalar test negative quad precision.
7957 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7958 special case operand checking to enforce that second operand of
7959 VSX scalar test data class with quad precision argument is a 7-bit
7961 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7962 prototypes and descriptions of __ieee128 versions of
7963 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7964 scalar_test_data_class, and scalar_test_neg built-in functions.
7966 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7968 PR tree-optimization/81162
7969 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7970 replace a negate with an add.
7972 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7974 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7976 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7978 * config/arm/arm-c.c (arm_cpu_builtins): Define
7979 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7981 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7983 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7984 (armv8-r): Set ARM Cortex-R52 as default CPU.
7985 * config/arm/arm-tables.opt: Regenerate.
7986 * config/arm/arm-tune.md: Regenerate.
7987 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7989 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7990 extension for -mcpu=cortex-r52.
7992 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7994 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7995 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7996 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7997 (fp-armv8): Define it as FP_ARMv8 only.
7998 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7999 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8001 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8002 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8003 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8004 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8005 than TARGET_FPU_ARMV8.
8006 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8007 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8008 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8010 * config/arm/neon.md (neon_vrint): Likewise.
8011 (neon_vcvt): Likewise.
8012 (neon_<fmaxmin_op><mode>): Likewise.
8013 (<fmaxmin><mode>3): Likewise.
8014 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8015 * config/arm/predicates.md (arm_cond_move_operator): Check against
8016 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8018 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8020 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8023 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8025 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8026 loop in comment with memset.
8028 2017-07-14 Martin Liska <mliska@suse.cz>
8030 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8031 * dwarf2out.c (is_java): Remove the function.
8032 (output_pubname): Remove usage of the function.
8033 (lower_bound_default): Remove usage of DW_LANG_Java.
8034 (gen_compile_unit_die): Likewise.
8035 * gcc.c: Remove compiler defaults for .java and .zip files.
8036 * gimple-expr.c (remove_suffix): Change as there's no longer
8037 extension than 4-letter one.
8038 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8039 (gimplify_save_expr): Likewise.
8040 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8041 as it's possible even for other languages than Java.
8042 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8043 * lto-opts.c (lto_write_options): Remove reference to Java.
8044 * opts.c (strip_off_ending): Update file extension handling.
8045 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8046 * tree-eh.c (lower_resx): Likewise.
8047 * tree.c (free_lang_data_in_type): Remove dead code.
8048 (find_decls_types_r): Likewise.
8049 (build_common_builtin_nodes): Remove Java from a comment.
8050 (verify_type): Remove dead code.
8051 * varasm.c (assemble_external): Remove Java from a comment.
8053 2017-07-14 Martin Liska <mliska@suse.cz>
8055 * opts.c (finish_options): Add quotes.
8056 (common_handle_option): Likewise.
8058 2017-07-14 Martin Liska <mliska@suse.cz>
8060 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8061 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8063 * dwarf2out.c (lower_bound_default): Do not handle
8065 (gen_compile_unit_die): Likewise.
8066 * gcc.c: Remove default extension binding for GNU Pascal.
8067 * stmt.c: Remove Pascal language from a comment.
8068 * xcoffout.c: Likewise.
8070 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8073 * diagnostic-show-locus.c (fixit_cmp): New function.
8074 (layout::layout): Sort m_fixit_hints.
8075 (column_range::column_range): Assert that the values are valid.
8076 (struct char_span): New struct.
8077 (correction::overwrite): New method.
8078 (struct source_line): New struct.
8079 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8080 calls in terms of classes source_line and char_span, and
8081 correction::overwrite.
8082 (selftest::test_overlapped_fixit_printing_2): New function.
8083 (selftest::diagnostic_show_locus_c_tests): Call it.
8085 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8087 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8088 early if there is no lhs.
8090 2017-07-13 Martin Liska <mliska@suse.cz>
8092 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8093 (gen_reference_type_die): Likewise.
8094 * stor-layout.c: Remove Pascal-related comment.
8096 2017-07-13 Martin Liska <mliska@suse.cz>
8098 * opts.c (finish_options): Add quotes to error messages.
8099 (parse_sanitizer_options): Likewise.
8101 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8103 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8105 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8107 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8109 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8111 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8113 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8115 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8118 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8119 provides the hardware capability bits, define the macro
8120 __BUILTIN_CPU_SUPPORTS__.
8121 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8122 if GLIBC does not provide the hardware capability bits. Add a
8123 gcc_unreachable call if the built-in cpu function is neither
8124 __builtin_cpu_is nor __builtin_cpu_supports.
8125 (rs6000_get_function_versions_dispatcher): Change the warning
8126 that an old GLIBC is used which does not export the capability
8127 bits to be an error.
8128 * doc/extend.texi (target_clones attribute): Document the
8129 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8130 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8131 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8132 the macros defined by GCC if the newer GLIBC is available.
8134 2017-07-12 Jeff Law <law@redhat.com>
8136 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8137 remaining includes slightly.
8138 * config/riscv/riscv-builtins.c: Include profile-count.h.
8140 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8143 * config/avr/avr.c (avr_set_current_function): In diagnostic
8144 messages: Quote keywords and (parts of) identifiers.
8145 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8148 2017-07-12 Carl Love <cel@us.ibm.com>
8150 * config/rs6000/rs6000-c.c: Add support for built-in functions
8151 vector bool char vec_revb (vector bool char);
8152 vector bool short vec_revb (vector short char);
8153 vector bool int vec_revb (vector bool int);
8154 vector bool long long vec_revb (vector bool long long);
8155 * doc/extend.texi: Update the built-in documentation file for the
8156 new built-in functions.
8158 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8160 * config/s390/s390.md: Remove movcc splitter.
8162 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8164 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8165 load/store on condition.
8167 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8170 * config/avr/avr.c (avr_encode_section_info)
8171 [progmem && !TREE_READONLY]: Error if progmem object needs
8174 2017-07-11 Michael Collison <michael.collison@arm.com>
8176 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8179 2017-07-11 Carl Love <cel@us.ibm.com>
8181 * config/rs6000/rs6000-c.c: Add support for builtins
8182 vector unsigned int vec_parity_lsbb (vector signed int);
8183 vector unsigned int vec_parity_lsbb (vector unsigned int);
8184 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8185 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8186 vector unsigned long long vec_parity_lsbb (vector signed long long);
8187 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8188 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8189 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8190 * doc/extend.texi: Update the built-in documentation file for the
8191 new built-in functions.
8193 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8195 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8196 (layout::m_primary_loc): New field.
8197 (layout::layout): Initialize new field. Move location filtering
8198 logic from here to...
8199 (layout::maybe_add_location_range): ...this new method. Add
8200 support for filtering to just the lines already specified by other
8202 (layout::will_show_line_p): New method.
8203 (gcc_rich_location::add_location_if_nearby): New method.
8204 (selftest::test_add_location_if_nearby): New test function.
8205 (selftest::diagnostic_show_locus_c_tests): Call it.
8206 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8209 2017-07-11 Tom de Vries <tom@codesourcery.com>
8211 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8212 (bb_first_real_insn): New function.
8213 (nvptx_single): Add extra initialization of broadcasted condition
8216 2017-07-11 Nathan Sidwell <nathan@acm.org>
8218 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8220 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8222 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8223 Binutils doc as TEXI will mess them up.
8224 * doc/invoke.texi (AVR Options): Same here.
8226 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8228 * config/sparc/sparc.opt (mfix-ut700): New option.
8229 (mfix-gr712rc): Likewise.
8230 (sparc_fix_b2bst): New variable.
8231 * doc/invoke.texi (SPARC options): Document them.
8232 (ARM options): Fix warnings.
8233 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8234 instructions to prevent sequences that can trigger the store-store
8235 errata for certain LEON3FT processors.
8236 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8237 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8238 * config/sparc/sparc.md (fix_b2bst): New attribute.
8239 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8241 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8244 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8246 (*rsqrtsf2_sse): Ditto.
8248 (div<mode>3): Macroize insn from divdf3 and divsf3
8249 using MODEF mode iterator.
8251 2017-07-10 Martin Sebor <msebor@redhat.com>
8253 PR tree-optimization/80397
8254 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8255 instead of testing for equality to INTEGER_TYPE.
8257 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8259 * config.gcc: Remove uclibc from arc target spec.
8261 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8263 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8265 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8268 * lto-wrapper.c (remove_option): New function.
8269 (merge_and_complain): Merge PIC/PIE options more realistically.
8271 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8273 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8277 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8278 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8279 * config.in: Regenerate.
8280 * configure: Regenerate.
8281 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8282 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8283 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8284 (TARGET_GASISR_PROLOGUES): ...target mask.
8285 * common/config/avr/avr-common.c
8286 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8287 Set -mgas-isr-prologues.
8288 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8289 INSERT_PASS_BEFORE for it.
8290 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8291 * config/avr/avr.c (avr_option_override)
8292 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8293 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8294 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8295 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8296 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8298 (make_avr_pass_pre_proep): New function.
8299 (emit_push_sfr) <treg>: Add argument to function and use it
8301 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8302 and set machine->gasisr.yes.
8303 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8304 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8305 __gcc_isr.n_pushed to .L__stack_usage.
8306 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8307 (avr_asm_final_postscan_insn): ...this new static function.
8308 * config/avr/avr.h (machine_function)
8309 <is_no_gccisr, use_L__stack_usage>: New fields.
8310 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8311 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8312 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8313 (gasisr, *gasisr): New expander and insn.
8314 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8315 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8316 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8318 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8320 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8323 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8325 Move jump-tables out of .text again.
8328 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8329 (ASM_OUTPUT_ADDR_VEC): New function.
8330 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8331 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8332 INSN_ADDRESSes as asm comment.
8333 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8334 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8335 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8336 * config/avr/avr.md (*tablejump): Adjust comment.
8337 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8338 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8340 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8341 (avr_output_addr_vec): New proto.
8342 (avr_log_t) <insn_addresses>: New field.
8344 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8347 * config/i386/i386.c (ix86_function_arg_advance): Set
8348 outgoing_args_on_stack to true if there are outgoing arguments
8350 (ix86_function_arg): Likewise.
8351 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8352 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8353 * config/i386/i386.h (machine_function): Add
8354 outgoing_args_on_stack.
8356 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8358 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8360 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8362 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8364 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8365 (REAL_H): Remove $(MACHMODE_H).
8366 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8368 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8369 $(MACHMODE_H) and double-int.h.
8370 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8372 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8375 2017-07-07 Andrew Pinski <apinski@cavium.com>
8377 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8378 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8380 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8382 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8383 Add warning if GCC was not configured to link against a GLIBC that
8384 exports the hardware capability bits.
8385 (make_resolver_func): Make resolver function private and not a
8386 COMDAT function. Create the name with clone_function_name instead
8387 of make_unique_name.
8390 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8391 correct operand in doing the split.
8393 2017-07-07 Carl Love <cel@us.ibm.com>
8395 * config/rs6000/rs6000-c: Add support for built-in function
8396 vector unsigned short vec_pack_to_short_fp32 (vector float,
8398 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8399 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8400 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8401 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8402 (convert_4f32_8i16): Add define_expand.
8403 * doc/extend.texi: Update the built-in documentation file for the
8404 new built-in function.
8406 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8408 * config/sparc/m8.md: New file.
8409 * config/sparc/sparc.md: Include m8.md.
8411 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8413 * config/sparc/sparc.opt: New option -mvis4b.
8414 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8415 (sparc_option_override): Handle VIS4B.
8416 (enum sparc_builtins): Define
8417 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8418 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8419 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8420 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8421 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8422 (check_constant_argument): New function.
8423 (sparc_vis_init_builtins): Define builtins
8424 __builtin_vis_dictunpack{8,16,32},
8425 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8426 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8427 __builtin_vis_fpcmpde{8,16,32}shl and
8428 __builtin_vis_fpcmpur{8,16,32}shl.
8429 (sparc_expand_builtin): Check that the constant operands to
8430 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8431 constant and in range.
8432 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8434 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8435 (SPARC_IMM5_P): Likewise.
8436 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8437 (enabled): Handle vis4b.
8438 (UNSPEC_DICTUNPACK): New unspec.
8439 (UNSPEC_FPCMPSHL): Likewise.
8440 (UNSPEC_FPUCMPSHL): Likewise.
8441 (UNSPEC_FPCMPDESHL): Likewise.
8442 (UNSPEC_FPCMPURSHL): Likewise.
8443 (cpu_feature): New CPU feature `vis4b'.
8444 (dictunpack{8,16,32}): New insns.
8445 (FPCSMODE): New mode iterator.
8446 (fpcscond): New code iterator.
8447 (fpcsucond): Likewise.
8448 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8449 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8450 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8451 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8452 * config/sparc/constraints.md: Define constraints `q' for unsigned
8453 2-bit integer constants and `t' for unsigned 5-bit integer
8455 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8457 (imm5_operand_dictunpack16): Likewise.
8458 (imm5_operand_dictunpack32): Likewise.
8459 (imm2_operand): Likewise.
8460 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8461 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8462 ditunpack* and fpcmp*shl builtins.
8464 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8466 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8467 * config.in: Add HAVE_AS_SPARC6 define.
8468 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8470 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8472 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8473 (CPP_CPU_SPEC): Handle m8.
8474 (ASM_CPU_SPEC): Likewise.
8475 * config/sparc/sparc-opts.h (enum processor_type): Add
8477 * config/sparc/sparc.c (m8_costs): New struct.
8478 (sparc_option_override): Handle TARGET_CPU_m8.
8479 (sparc32_initialize_trampoline): Likewise.
8480 (sparc64_initialize_trampoline): Likewise.
8481 (sparc_issue_rate): Likewise.
8482 (sparc_register_move_cost): Likewise.
8483 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8484 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8485 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8486 (CPP_CPU_SPEC): Handle M8.
8487 (ASM_CPU_SPEC): Likewise.
8488 (AS_M8_FLAG): Define.
8489 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8490 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8491 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8493 * configure: Regenerate.
8494 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8497 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8499 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8501 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8502 ("*movdi_insn_sp32"): Do not set v3pipe.
8503 ("*movsi_insn"): Likewise.
8504 ("*movdi_insn_sp64"): Likewise.
8505 ("*movsf_insn"): Likewise.
8506 ("*movdf_insn_sp32"): Likewise.
8507 ("*movdf_insn_sp64"): Likewise.
8508 ("*zero_extendsidi2_insn_sp64"): Likewise.
8509 ("*sign_extendsidi2_insn"): Likewise.
8510 ("*mov<VM32:mode>_insn"): Likewise.
8511 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8512 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8513 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8514 ("<vlop:code><VL:mode>3"): Likewise.
8515 ("*not_<vlop:code><VL:mode>3"): Likewise.
8516 ("*nand<VL:mode>_vis"): Likewise.
8517 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8518 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8519 ("one_cmpl<VL:mode>2"): Likewise.
8520 ("faligndata<VM64:mode>_vis"): Likewise.
8521 ("alignaddrsi_vis"): Likewise.
8522 ("alignaddrdi_vis"): Likweise.
8523 ("alignaddrlsi_vis"): Likewise.
8524 ("alignaddrldi_vis"): Likewise.
8525 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8526 ("bmaskdi_vis"): Likewise.
8527 ("bmasksi_vis"): Likewise.
8528 ("bshuffle<VM64:mode>_vis"): Likewise.
8529 ("cmask8<P:mode>_vis"): Likewise.
8530 ("cmask16<P:mode>_vis"): Likewise.
8531 ("cmask32<P:mode>_vis"): Likewise.
8532 ("pdistn<P:mode>_vis"): Likewise.
8533 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8535 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8537 * config/sparc/sparc.md ("subtype"): New insn attribute.
8538 ("*wrgsr_sp64"): Set insn subtype.
8539 ("*rdgsr_sp64"): Likewise.
8540 ("alignaddrsi_vis"): Likewise.
8541 ("alignaddrdi_vis"): Likewise.
8542 ("alignaddrlsi_vis"): Likewise.
8543 ("alignaddrldi_vis"): Likewise.
8544 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8545 ("fexpand_vis"): Likewise.
8546 ("fpmerge_vis"): Likewise.
8547 ("faligndata<VM64:mode>_vis"): Likewise.
8548 ("bshuffle<VM64:mode>_vis"): Likewise.
8549 ("cmask8<P:mode>_vis"): Likewise.
8550 ("cmask16<P:mode>_vis"): Likewise.
8551 ("cmask32<P:mode>_vis"): Likewise.
8552 ("fchksm16_vis"): Likewise.
8553 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8554 ("fmean16_vis"): Likewise.
8555 ("fp<plusminus_insn>64_vis"): Likewise.
8556 ("<plusminus_insn>v8qi3"): Likewise.
8557 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8558 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8559 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8560 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8561 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8562 ("*movqi_insn"): Likewise.
8563 ("*movhi_insn"): Likewise.
8564 ("*movsi_insn"): Likewise.
8565 ("movsi_pic_gotdata_op"): Likewise.
8566 ("*movdi_insn_sp32"): Likewise.
8567 ("*movdi_insn_sp64"): Likewise.
8568 ("movdi_pic_gotdata_op"): Likewise.
8569 ("*movsf_insn"): Likewise.
8570 ("*movdf_insn_sp32"): Likewise.
8571 ("*movdf_insn_sp64"): Likewise.
8572 ("*zero_extendhisi2_insn"): Likewise.
8573 ("*zero_extendqihi2_insn"): Likewise.
8574 ("*zero_extendqisi2_insn"): Likewise.
8575 ("*zero_extendqidi2_insn"): Likewise.
8576 ("*zero_extendhidi2_insn"): Likewise.
8577 ("*zero_extendsidi2_insn_sp64"): Likewise.
8578 ("ldfsr"): Likewise.
8579 ("prefetch_64"): Likewise.
8580 ("prefetch_32"): Likewise.
8581 ("tie_ld32"): Likewise.
8582 ("tie_ld64"): Likewise.
8583 ("*tldo_ldub_sp32"): Likewise.
8584 ("*tldo_ldub1_sp32"): Likewise.
8585 ("*tldo_ldub2_sp32"): Likewise.
8586 ("*tldo_ldub_sp64"): Likewise.
8587 ("*tldo_ldub1_sp64"): Likewise.
8588 ("*tldo_ldub2_sp64"): Likewise.
8589 ("*tldo_ldub3_sp64"): Likewise.
8590 ("*tldo_lduh_sp32"): Likewise.
8591 ("*tldo_lduh1_sp32"): Likewise.
8592 ("*tldo_lduh_sp64"): Likewise.
8593 ("*tldo_lduh1_sp64"): Likewise.
8594 ("*tldo_lduh2_sp64"): Likewise.
8595 ("*tldo_lduw_sp32"): Likewise.
8596 ("*tldo_lduw_sp64"): Likewise.
8597 ("*tldo_lduw1_sp64"): Likewise.
8598 ("*tldo_ldx_sp64"): Likewise.
8599 ("*mov<VM32:mode>_insn"): Likewise.
8600 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8601 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8603 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8605 * config/sparc/sparc.md ("type"): New insn type viscmp.
8606 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8608 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8609 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8610 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8611 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8613 ("n7_vis_logical_11cycle"): Likewise.
8614 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8615 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8616 * config/sparc/niagara.md ("niag_vis"): Likewise.
8617 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8618 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8620 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8622 * config/sparc/sparc.md: New instruction type `bmask'.
8623 (bmaskdi_vis): Use the `bmask' type.
8624 (bmasksi_vis): Likewise.
8625 * config/sparc/ultra3.md (us3_array): Likewise.
8626 * config/sparc/niagara7.md (n7_array): Likewise.
8627 * config/sparc/niagara4.md (n4_array): Likewise.
8628 * config/sparc/niagara2.md (niag2_vis): Likewise.
8629 (niag3_vis): Likewise.
8630 * config/sparc/niagara.md (niag_vis): Likewise.
8632 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8634 * ipa-comdats.c: Remove optimize check from gate.
8635 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8636 for functions not optimized.
8637 (ipa_fn_summary_read): Skip optimize check.
8638 (ipa_fn_summary_write): Likewise.
8639 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8641 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8643 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8644 (check_callers): Give up on caller which is not optimized.
8645 (inline_small_functions): Likewise.
8646 (ipa_inline): Do not give up when not optimizing.
8647 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8648 away unoptimizes cdtors.
8649 (whole_program_function_and_variable_visibility): Do
8650 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8651 * ipa.c (process_references): Do not check optimize.
8652 (symbol_table::remove_unreachable_nodes): Update optimize check.
8653 (set_writeonly_bit): Update optimize check.
8654 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8655 (pass_ipa_single_use::gate): Remove.
8657 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8659 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8660 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8661 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8662 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8663 permute_load, permute_store, adjust_extract, adjust_splat,
8664 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8665 replace_swap_with_copy, dump_swap_insn_table,
8666 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8667 recombine_lvx_pattern, recombine_stvx_pattern,
8668 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8669 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8670 to file rs6000-p8swap.c.
8671 * config/rs6000/rs6000-p8swap.c: New file.
8672 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8673 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8674 and rs6000*-*-* targets.
8676 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8678 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8680 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8682 * lto-wrapper.c (merge_and_complain): Do not merge
8683 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8684 fsigned_zeros, ftrapping_math, fwrapv.
8685 (append_compiler_options): Do not track these options.
8686 (append_linker_options): Likewie
8688 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8690 * cgraphunit.c (cgraph_node::finalize_function): When
8691 !flag_toplevel_reorde set no_reorder flag.
8692 (varpool_node::finalize_decl): Likewise.
8693 (symbol_table::compile): Drop no toplevel reorder path.
8695 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8697 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8698 edges; zero probability is not better than uninitialized.
8700 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8702 * asan.h (asan_sanitize_allocas_p): Declare.
8703 * asan.c (asan_sanitize_allocas_p): New function.
8704 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8705 (handle_builtin_alloca): Likewise.
8706 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8707 if !asan_sanitize_allocas_p.
8708 * params.def (asan-instrument-allocas): Add new option.
8709 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8710 * opts.c (common_handle_option): Disable allocas sanitization for
8713 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8715 * asan.c: Include gimple-fold.h.
8716 (get_last_alloca_addr): New function.
8717 (handle_builtin_stackrestore): Likewise.
8718 (handle_builtin_alloca): Likewise.
8719 (asan_emit_allocas_unpoison): Likewise.
8720 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8721 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8722 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8723 (instrument_builtin_call): Pass gimple iterator to
8724 get_mem_refs_of_builtin_call.
8725 (last_alloca_addr): New global.
8726 * asan.h (asan_emit_allocas_unpoison): Declare.
8727 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8728 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8729 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8730 if function calls alloca.
8731 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8732 * gimple-fold.h (replace_call_with_value): Declare.
8733 * internal-fn.c: Include asan.h.
8734 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8735 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8737 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8739 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8740 (C_SELFTEST_FLAGS): New.
8741 (CPP_SELFTEST_FLAGS): New.
8742 (SELFTEST_DEPS): New, from deps of s-selftest.
8743 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8744 (CPP_SELFTEST_DEPS): New.
8745 (selftest): Add dependency on s-selftest-c++.
8746 (s-selftest): Rename to...
8747 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8748 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8749 than SELFTEST_FLAGS.
8750 (selftest-gdb): Rename to...
8751 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8753 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8754 (selftest-valgrind): Rename to...
8755 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8757 (selftest-valgrind): Reintroduce as an alias for
8758 selftest-c-valgrind.
8759 (s-selftest-c++): New.
8760 (selftest-c++-gdb): New.
8761 (selftest-c++-valgrind): New.
8763 2017-07-06 Olivier Hainque <hainque@adacore.com>
8765 * gcc.c (process_command): When deciding if undefined variables
8766 should be ignored when processing specs, accept "gcc -v" as well.
8768 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8770 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8771 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8773 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8775 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8776 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8777 * config/arm/arm-tables.opt: Regenerate.
8778 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8780 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8782 2017-07-06 Carl Love <cel@us.ibm.com>
8784 * ChangeLog: Clean up from mid air collision
8786 2017-07-06 Carl Love <cel@us.ibm.com>
8788 * config/rs6000/rs6000-c.c: Add support for built-in functions
8789 vector signed int vec_subc (vector signed int, vector signed int);
8790 vector signed __int128 vec_subc (vector signed __int128,
8791 vector signed __int128);
8792 vector unsigned __int128 vec_subc (vector unsigned __int128,
8793 vector unsigned __int128);
8794 vector signed int vec_sube (vector signed int, vector signed int,
8796 vector unsigned int vec_sube (vector unsigned int,
8797 vector unsigned int,
8798 vector unsigned int);
8799 vector signed __int128 vec_sube (vector signed __int128,
8800 vector signed __int128,
8801 vector signed__int128);
8802 vector unsigned __int128 vec_sube (vector unsigned __int128,
8803 vector unsigned __int128,
8804 vector unsigned __int128);
8805 vector signed int vec_subec (vector signed int, vector signed int,
8807 vector unsigned int vec_subec (vector unsigned int,
8808 vector unsigned int,
8809 vector unsigned int);
8810 vector signed __int128 vec_subec (vector signed __int128,
8811 vector signed __int128,
8812 vector signed__int128);
8813 vector unsigned __int128 vec_subec (vector unsigned __int128,
8814 vector unsigned __int128,
8815 vector unsigned __int128);
8816 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8817 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8818 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8819 BU_ALTIVEC_OVERLOAD_X definitions.
8820 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8821 * doc/extend.texi: Update the built-in documentation file for the new
8824 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8827 * diagnostic-show-locus.c (layout::layout): Use start and finish
8828 spelling location for the start and finish of each range.
8829 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8830 Add unused aspect param.
8831 * input.c (expand_location_1): Add "aspect" param, and use it
8832 to access the correct part of the location.
8833 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8835 (expand_location_to_spelling_point): Likewise.
8836 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8837 param, and pass it to expand_location_1.
8839 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8841 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8842 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8843 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8844 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8845 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8846 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8847 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8848 _mm_maskz_getmant_ss): New intrinsics.
8849 (__builtin_ia32_getexpss128_mask): Changed to ...
8850 __builtin_ia32_getexpss128_round ... this.
8851 (__builtin_ia32_getexpsd128_mask): Changed to ...
8852 __builtin_ia32_getexpsd128_round ... this.
8853 * config/i386/i386-builtin-types.def
8854 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8855 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8856 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8857 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8858 __builtin_ia32_getmantss_mask_round): New builtins.
8859 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8860 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8861 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8862 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8863 * config/i386/sse.md
8864 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8865 avx512f_sgetexp<mode><mask_scalar_name>
8866 <round_saeonly_scalar_name> ... this.
8867 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8868 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8869 vgetexp<ssescalarmodesuffix>
8870 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8871 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8872 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8873 avx512f_vgetmant<mode><mask_scalar_name>
8874 <round_saeonly_scalar_name> ... this.
8875 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8876 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8877 vgetmant<ssescalarmodesuffix>
8878 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8879 %0<mask_scalar_operand4>, %1, %2
8880 <round_saeonly_scalar_mask_op4>, %3} ... this.
8881 * config/i386/subst.md (mask_scalar_operand4,
8882 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8883 round_saeonly_scalar_nimm_predicate): New subst attributes.
8885 2017-07-06 Julia Koval <julia.koval@intel.com>
8887 * config/i386/i386.c (ix86_erase_embedded_rounding):
8888 Remove code for old rounding pattern.
8890 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8892 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8894 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8896 * doc/sourcebuild.texi (Test Directives, Variants of
8897 dg-require-support): Add documentation for dg-require-stack-check.
8899 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8901 * config/i386/subst.md (mask_scalar, round_scalar,
8902 round_saeonly_scalar): New meta-templates.
8903 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8904 round_scalar_mask_operand3, round_scalar_mask_op3,
8905 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8906 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8907 round_saeonly_scalar_constraint,
8908 round_saeonly_scalar_prefix): New subst attribute.
8909 * config/i386/sse.md
8910 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8911 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8912 <round_scalar_name> ... this.
8913 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8914 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8915 <round_scalar_name> ... this.
8916 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8917 <sse>_vm<code><mode>3<mask_scalar_name>
8918 <round_saeonly_scalar_name> ... this.
8919 (v<plusminus_mnemonic><ssescalarmodesuffix>
8920 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8921 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8922 v<plusminus_mnemonic><ssescalarmodesuffix>
8923 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8924 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8925 (v<multdiv_mnemonic><ssescalarmodesuffix>
8926 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8927 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8928 v<multdiv_mnemonic><ssescalarmodesuffix>
8929 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8930 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8931 (v<maxmin_float><ssescalarmodesuffix>
8932 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8933 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8934 v<maxmin_float><ssescalarmodesuffix>
8935 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8936 %0<mask_scalar_operand3>, %1, %<iptr>2
8937 <round_saeonly_scalar_mask_op3>} ... this.
8939 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8941 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8942 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8944 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8945 Alan Hayward <alan.hayward@arm.com>
8946 David Sherwood <david.sherwood@arm.com>
8948 * combine.c (simplify_if_then_else): Remove "enum" before
8950 * compare-elim.c (can_eliminate_compare): Likewise.
8951 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8953 (aarch64_lookup_simd_builtin_type): Likewise.
8954 (aarch64_simd_builtin_type): Likewise.
8955 (aarch64_init_simd_builtin_types): Likewise.
8956 (aarch64_simd_expand_args): Likewise.
8957 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8959 (aarch64_reverse_mask): Likewise.
8960 (aarch64_simd_emit_reg_reg_move): Likewise.
8961 (aarch64_gen_adjusted_ldpstp): Likewise.
8962 (aarch64_ccmp_mode_to_code): Likewise.
8963 (aarch64_operands_ok_for_ldpstp): Likewise.
8964 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8965 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8967 (aarch64_min_divisions_for_recip_mul): Likewise.
8968 (aarch64_reassociation_width): Likewise.
8969 (aarch64_get_condition_code_1): Likewise.
8970 (aarch64_simd_emit_reg_reg_move): Likewise.
8971 (aarch64_simd_attr_length_rglist): Likewise.
8972 (aarch64_reverse_mask): Likewise.
8973 (aarch64_operands_ok_for_ldpstp): Likewise.
8974 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8975 (aarch64_gen_adjusted_ldpstp): Likewise.
8976 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8978 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8979 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8980 (arm_lookup_simd_builtin_type): Likewise.
8981 (arm_simd_builtin_type): Likewise.
8982 (arm_init_simd_builtin_types): Likewise.
8983 (arm_expand_builtin_args): Likewise.
8984 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8985 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8986 (ft32_setup_incoming_varargs): Likewise.
8987 (ft32_function_arg): Likewise.
8988 (ft32_function_arg_advance): Likewise.
8989 (ft32_pass_by_reference): Likewise.
8990 (ft32_arg_partial_bytes): Likewise.
8991 (ft32_valid_pointer_mode): Likewise.
8992 (ft32_addr_space_pointer_mode): Likewise.
8993 (ft32_addr_space_legitimate_address_p): Likewise.
8994 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8996 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8997 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8998 (iamcu_alignment): Likewise.
8999 (canonicalize_vector_int_perm): Likewise.
9000 (ix86_noce_conversion_profitable_p): Likewise.
9001 (ix86_mpx_bound_mode): Likewise.
9002 (ix86_operands_ok_for_move_multiple): Likewise.
9003 * config/microblaze/microblaze-protos.h
9004 (microblaze_expand_conditional_branch_reg): Likewise.
9005 * config/microblaze/microblaze.c
9006 (microblaze_expand_conditional_branch_reg): Likewise.
9007 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9009 (rs6000_reassociation_width): Likewise.
9010 (rs6000_invalid_binary_op): Likewise.
9011 (fusion_p9_p): Likewise.
9012 (emit_fusion_p9_load): Likewise.
9013 (emit_fusion_p9_store): Likewise.
9014 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9016 (riscv_hard_regno_mode_ok_p): Likewise.
9017 (riscv_address_insns): Likewise.
9018 (riscv_split_symbol): Likewise.
9019 (riscv_legitimize_move): Likewise.
9020 (riscv_function_value): Likewise.
9021 (riscv_hard_regno_nregs): Likewise.
9022 (riscv_expand_builtin): Likewise.
9023 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9024 (riscv_build_integer): Likewise.
9025 (riscv_split_integer): Likewise.
9026 (riscv_legitimate_constant_p): Likewise.
9027 (riscv_cannot_force_const_mem): Likewise.
9028 (riscv_regno_mode_ok_for_base_p): Likewise.
9029 (riscv_valid_base_register_p): Likewise.
9030 (riscv_valid_offset_p): Likewise.
9031 (riscv_valid_lo_sum_p): Likewise.
9032 (riscv_classify_address): Likewise.
9033 (riscv_legitimate_address_p): Likewise.
9034 (riscv_address_insns): Likewise.
9035 (riscv_load_store_insns): Likewise.
9036 (riscv_force_binary): Likewise.
9037 (riscv_split_symbol): Likewise.
9038 (riscv_force_address): Likewise.
9039 (riscv_legitimize_address): Likewise.
9040 (riscv_move_integer): Likewise.
9041 (riscv_legitimize_const_move): Likewise.
9042 (riscv_legitimize_move): Likewise.
9043 (riscv_address_cost): Likewise.
9044 (riscv_subword): Likewise.
9045 (riscv_output_move): Likewise.
9046 (riscv_canonicalize_int_order_test): Likewise.
9047 (riscv_emit_int_order_test): Likewise.
9048 (riscv_function_arg_boundary): Likewise.
9049 (riscv_pass_mode_in_fpr_p): Likewise.
9050 (riscv_pass_fpr_single): Likewise.
9051 (riscv_pass_fpr_pair): Likewise.
9052 (riscv_get_arg_info): Likewise.
9053 (riscv_function_arg): Likewise.
9054 (riscv_function_arg_advance): Likewise.
9055 (riscv_arg_partial_bytes): Likewise.
9056 (riscv_function_value): Likewise.
9057 (riscv_pass_by_reference): Likewise.
9058 (riscv_setup_incoming_varargs): Likewise.
9059 (riscv_print_operand): Likewise.
9060 (riscv_elf_select_rtx_section): Likewise.
9061 (riscv_save_restore_reg): Likewise.
9062 (riscv_for_each_saved_reg): Likewise.
9063 (riscv_register_move_cost): Likewise.
9064 (riscv_hard_regno_mode_ok_p): Likewise.
9065 (riscv_hard_regno_nregs): Likewise.
9066 (riscv_class_max_nregs): Likewise.
9067 (riscv_memory_move_cost): Likewise.
9068 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9069 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9070 (rl78_addr_space_address_mode): Likewise.
9071 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9073 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9074 (rs6000_reassociation_width): Likewise.
9075 (rs6000_invalid_binary_op): Likewise.
9076 (fusion_p9_p): Likewise.
9077 (emit_fusion_p9_load): Likewise.
9078 (emit_fusion_p9_store): Likewise.
9079 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9080 (ok_for_simple_move_operands): Likewise.
9081 (ok_for_simple_move_strict_operands): Likewise.
9082 (ok_for_simple_arith_logic_operands): Likewise.
9083 (visium_legitimize_reload_address): Likewise.
9084 (visium_select_cc_mode): Likewise.
9085 (output_cbranch): Likewise.
9086 (visium_split_double_move): Likewise.
9087 (visium_expand_copysign): Likewise.
9088 (visium_expand_int_cstore): Likewise.
9089 (visium_expand_fp_cstore): Likewise.
9090 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9091 (visium_function_arg): Likewise.
9092 (visium_function_arg_advance): Likewise.
9093 (visium_libcall_value): Likewise.
9094 (visium_setup_incoming_varargs): Likewise.
9095 (visium_legitimate_constant_p): Likewise.
9096 (visium_legitimate_address_p): Likewise.
9097 (visium_legitimize_address): Likewise.
9098 (visium_secondary_reload): Likewise.
9099 (visium_register_move_cost): Likewise.
9100 (visium_memory_move_cost): Likewise.
9101 (prepare_move_operands): Likewise.
9102 (ok_for_simple_move_operands): Likewise.
9103 (ok_for_simple_move_strict_operands): Likewise.
9104 (ok_for_simple_arith_logic_operands): Likewise.
9105 (visium_function_value_1): Likewise.
9106 (rtx_ok_for_offset_p): Likewise.
9107 (visium_legitimize_reload_address): Likewise.
9108 (visium_split_double_move): Likewise.
9109 (visium_expand_copysign): Likewise.
9110 (visium_expand_int_cstore): Likewise.
9111 (visium_expand_fp_cstore): Likewise.
9112 (visium_split_cstore): Likewise.
9113 (visium_select_cc_mode): Likewise.
9114 (visium_split_cbranch): Likewise.
9115 (output_cbranch): Likewise.
9116 (visium_print_operand_address): Likewise.
9117 * expmed.c (flip_storage_order): Likewise.
9118 * expmed.h (emit_cstore): Likewise.
9119 (flip_storage_order): Likewise.
9120 * genrecog.c (validate_pattern): Likewise.
9121 * hsa-gen.c (gen_hsa_addr): Likewise.
9122 * internal-fn.c (expand_arith_overflow): Likewise.
9123 * ira-color.c (allocno_copy_cost_saving): Likewise.
9124 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9125 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9126 (process_invariant_for_inheritance): Likewise.
9127 * lra-eliminations.c (move_plus_up): Likewise.
9128 * omp-low.c (lower_oacc_reductions): Likewise.
9129 * simplify-rtx.c (simplify_subreg): Likewise.
9130 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9131 (TARGET_CHKP_BOUND_MODE): Likewise..
9132 * targhooks.c (default_chkp_bound_mode): Likewise.
9133 (default_setup_incoming_vararg_bounds): Likewise.
9134 * targhooks.h (default_chkp_bound_mode): Likewise.
9135 (default_setup_incoming_vararg_bounds): Likewise.
9136 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9137 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9138 (have_whole_vector_shift): Likewise.
9139 * tree-vect-stmts.c (vectorizable_load): Likewise.
9140 * doc/tm.texi: Regenerate.
9142 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9144 Graceful degrade if Binutils PR21472 is not available.
9147 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9148 .rodata in flash test fails.
9149 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9150 * confgure: Regenerate.
9151 * config.in: Regenerate.
9152 * config/avr/avr.c (avr_asm_named_section)
9153 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9154 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9155 (avr_asm_init_sections): Same.
9157 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9159 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9160 (fma<VH:mode>4_intrinsic): Likewise.
9161 (*fmsub<VCVTF:mode>4): Likewise.
9162 (*fmsub<VH:mode>4_intrinsic): Likewise.
9164 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9167 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9168 Don't depend on "optimize > 0".
9169 (out_movhi_r_mr, out_movqi_mr_r): Same.
9170 (out_movhi_mr_r, out_movqi_r_mr): Same.
9171 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9172 io_address_operand on "optimize > 0".
9174 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9176 * tree-loop-distribution.c: Add general explanantion on the pass.
9177 (generate_loops_for_partition): Mark distributed loop.
9178 (pg_add_dependence_edges): New parameter. Handle alias data
9179 dependence specially and record it in the parameter if asked.
9180 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9181 (init_partition_graph_vertices, add_partition_graph_edge): New.
9182 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9183 (free_partition_graph_vdata, build_partition_graph): New.
9184 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9185 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9186 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9187 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9188 (version_for_distribution_p, finalize_partitions): New.
9189 (distribute_loop): Handle alias data dependence specially. Factor
9190 out loop fusion code as functions and call these functions.
9192 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9194 * tree-loop-distribution.c (classify_partition): New parameter and
9195 better handle reduction statement.
9196 (rdg_build_partitions): Revise comment.
9197 (distribute_loop): Compute statements in all partitions and pass it
9198 to classify_partition.
9200 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9202 * tree-loop-distribution.c (enum partition_type): New.
9203 (struct partition): New field type.
9204 (partition_merge_into): Add parameter. Update partition type.
9205 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9206 (build_rdg_partition_for_vertex): Compute partition type.
9207 (rdg_build_partitions): Dump partition type.
9208 (distribute_loop): Update calls to partition_merge_into.
9210 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9212 * tree-loop-distribution.c (struct ddr_hasher): New.
9213 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9215 (classify_partition): Call get_data_dependence.
9216 (pg_add_dependence_edges): Ditto.
9217 (distribute_loop): Release data dependence hash table.
9219 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9221 * tree-loop-distribution.c (ref_base_address): Delete.
9222 (similar_memory_accesses): Rename ...
9223 (share_memory_accesses): ... to this. Check if partitions access
9224 the same memory reference.
9225 (distribute_loop): Call share_memory_accesses.
9227 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9229 * tree-loop-distribution.c (struct partition): New field recording
9231 (partition_alloc, partition_free): Init and release data refs.
9232 (partition_merge_into): Merge data refs.
9233 (build_rdg_partition_for_vertex): Collect data refs for partition.
9234 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9236 (distribute_loop): Remve data refs from vertice data of partition
9239 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9241 * tree-loop-distribution.c (params.h): Include header file.
9242 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9243 (datarefs_vec): New global var.
9244 (create_rdg_vertices): Use datarefs_vec directly.
9245 (free_rdg): Don't free data references.
9246 (build_rdg): Update use. Don't free data references.
9247 (distribute_loop): Compute global variable for data references.
9248 Bail out if there are too many data references.
9250 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9252 * tree-loop-distribution.c (loop_nest): New global var.
9253 (build_rdg): Use loop directly, rather than loop nest.
9254 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9256 (distribute_loop): Compute global variable loop nest. Update use.
9258 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9260 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9261 (partition_merge_into): New parameter. Dump reason for fusion.
9262 (distribute_loop): Update use of partition_merge_into.
9264 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9266 * tree-loop-distribution.c (bb_top_order_index): New.
9267 (bb_top_order_index_size, bb_top_order_cmp): New.
9268 (stmts_from_loop): Use topological order.
9269 (pass_loop_distribution::execute): Compute and release topological
9270 order for basic blocks.
9272 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9274 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9277 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9279 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9280 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9281 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9282 * internal-fn.def (LOOP_DIST_ALIAS): New.
9283 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9284 (fold_loop_internal_call): ... this.
9285 (vect_loop_dist_alias_call): New function.
9286 (set_uid_loop_bbs): Call fold_loop_internal_call.
9287 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9290 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9293 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9294 Require dead FLAGS_REG at the beginning of a peephole.
9296 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9299 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9300 arguments in the call to __builtin_ia32_sbb_u32.
9301 (_subborrow_u64): Swap _X and _Y arguments in the call to
9302 __builtin_ia32_sbb_u64.
9304 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9307 * tree-vrp.c (compare_assert_loc): Turn into a function template
9308 with stable template parameter. Only test if a->e is NULL,
9309 !a->e == !b->e has been verified already. Use e == NULL or
9310 e != NULL instead of e or ! e tests. If stable is true, don't use
9311 iterative_hash_expr, on the other side allow a or b or both NULL
9312 and sort the NULLs last.
9313 (process_assert_insertions): Sort using compare_assert_loc<false>
9314 instead of compare_assert_loc, later sort using
9315 compare_assert_loc<true> before calling process_assert_insertions_for
9316 in a loop. Use break instead of continue once seen NULL pointer.
9318 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9320 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9321 Cortex-R7 and Cortex-R8 processors.
9323 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9325 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9326 uninitialized while src is not.
9328 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9330 * common/config/arm/arm-common.c: Adjust include path for
9332 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9333 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9334 (arm-cpu-data.h): Likewise.
9335 (arm-cpu-cdata.h): Likewise.
9336 * config/arm/arm-cpu.h: Delete.
9337 * config/arm/arm-cpu-cdata.h: Delete.
9338 * config/arm/arm-cpu-data.h: Delete.
9340 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9342 * config/arm/arm-cpus.in (cortex-a55): New.
9343 (cortex-a75): Likewise.
9344 (cortex-a75.cortex-a55): Likewise.
9345 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9347 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9348 * config/arm/arm-cpu-cdata.h: Regenerate.
9349 * config/arm/arm-cpu-data.h: Regenerate.
9350 * config/arm/arm-cpu.h: Regenerate.
9351 * config/arm/arm-tables.opt: Regenerate.
9352 * config/arm/arm-tune.md: Regenerate.
9354 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9356 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9358 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9360 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9363 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9365 PR tree-optimization/81292
9366 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9367 full_string_p, also call adjust_related_strinfos if the adjustment
9368 is simple, otherwise invalidate related strinfos.
9370 2017-07-04 Martin Liska <mliska@suse.cz>
9373 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9374 newly created variable as DECL_IGNORED_P.
9376 2017-07-04 Martin Liska <mliska@suse.cz>
9379 * ipa-inline.c (inline_small_functions):
9380 Use xstrdup_for_dump.
9382 2017-07-04 Tom de Vries <tom@codesourcery.com>
9384 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9386 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9389 * config/darwin.c (darwin_function_switched_text_sections):
9392 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9394 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9396 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9398 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9400 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9402 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9403 min_profitable_iters, and th as inclusive lower bounds.
9404 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9405 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9406 for min_profitable_iters and min_profitable_estimate.
9407 (vect_transform_loop): Treat th as an inclusive lower bound.
9408 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9410 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9413 * config/darwin.c (darwin_function_switched_text_sections):
9414 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9415 in two pieces, and suppress the use of buf.
9417 2017-07-03 Nathan Sidwell <nathan@acm.org>
9419 * hash-table.h (hash_table_mod1): Fix indentation.
9421 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9424 * predict.c (force_edge_cold): Be more careful about propagation
9426 * profile-count.h (profile_probability::guessed,
9427 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9429 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9431 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9433 * doc/invoke.texi (rcpc architecture extension): Document it.
9435 2017-07-03 Richard Biener <rguenther@suse.de>
9437 PR tree-optimization/60510
9438 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9439 the scalar reduction PHI and use it.
9440 (vectorizable_reduction): Properly guard the single_defuse_cycle
9441 path for non-SLP reduction chains where we cannot use it.
9442 Rework reduc_def/index and vector type deduction. Rework
9443 vector operand gathering during reduction op code-gen.
9444 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9445 chains dissolve the chain and leave it to non-SLP reduction
9448 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9450 * tree-data-ref.h (dr_alignment): Declare.
9451 * tree-data-ref.c (dr_alignment): New function.
9452 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9453 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9455 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9457 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9459 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9460 and base_misalignment fields.
9461 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9462 * tree-data-ref.c: Include builtins.h.
9463 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9464 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9465 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9466 * tree-vect-data-refs.c: Include tree-cfg.h.
9467 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9468 fields instead of calculating an alignment here.
9469 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9470 innermost_loop_behavior fields.
9472 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9474 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9476 (DR_STEP_ALIGNMENT): New macro.
9477 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9478 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9479 (create_data_ref): Print it.
9480 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9481 to tell whether the step preserves vector (mis)alignment.
9482 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9483 Move the check for an integer step and generalise to all INTEGER_CST.
9484 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9485 Print the outer step alignment.
9487 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9489 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9490 with offset_alignment.
9491 (DR_ALIGNED_TO): Delete.
9492 (DR_OFFSET_ALIGNMENT): New macro.
9493 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9494 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9495 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9496 (create_data_ref): Likewise.
9497 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9498 (vect_analyze_data_refs): Likewise.
9499 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9500 creating dummy innermost behavior.
9502 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9504 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9505 with a "innermost_loop_behavior *" and refeence tree.
9506 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9507 (create_data_ref): Update call accordingly.
9508 * tree-predcom.c (find_looparound_phi): Likewise.
9510 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9512 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9513 fields with dr_wrt_vec_loop.
9514 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9515 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9516 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9517 (vect_dr_behavior): New function.
9518 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9519 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9520 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9521 track whether the step preserves the misalignment.
9522 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9523 Use vect_dr_behavior.
9524 (vect_setup_realignment): Update call accordingly.
9525 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9526 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9527 call to vect_create_addr_base_for_vector_ref.
9528 (vect_create_cond_for_align_checks): Likewise.
9529 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9530 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9531 (vect_recog_mask_conversion_pattern): Likewise.
9532 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9533 (new_stmt_vec_info): Remove redundant zeroing.
9535 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9537 * common/config/arm/arm-common.c (arm_be8_option): New function.
9538 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9539 (ISA_ARMv6): Add isa_bit_be8.
9540 * config/arm/arm.h (arm_be8_option): Add prototype.
9541 (BE8_SPEC_FUNCTION): New define.
9542 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9543 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9544 (mlittle-endian): Similarly.
9545 (mbe8, mbe32): New options.
9546 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9547 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9549 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9551 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9553 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9555 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9556 (cleanup_tree_cfg_bb): Use it.
9557 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9559 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9561 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9564 * loop-doloop.c (add_test): Update profile.
9566 2017-07-03 Martin Liska <mliska@suse.cz>
9569 * sanopt.c (rewrite_usage_of_param): New function.
9570 (sanitize_rewrite_addressable_params): Likewise.
9571 (pass_sanopt::execute): Call rewrite_usage_of_param.
9573 2017-07-03 Richard Biener <rguenther@suse.de>
9575 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9576 back to using VIEW_CONVERT_EXPR.
9578 2017-07-03 Martin Liska <mliska@suse.cz>
9581 * doc/extend.texi: Document when a resolver function is
9582 generated for target_clones.
9584 2017-07-03 Martin Liska <mliska@suse.cz>
9586 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9587 and shadow memory of auto variables which are subject of
9588 use-after-scope sanitization.
9589 (asan_expand_mark_ifn): Add do set only when is_poison.
9591 2016-07-03 Richard Biener <rguenther@suse.de>
9593 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9595 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9596 (vectorizable_reduction): Perform reduction PHI creation when
9597 visiting a reduction PHI and adjust and simplify code generation
9598 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9599 (vect_transform_loop): Visit reduction PHIs.
9600 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9601 defs into the SLP tree.
9602 (vect_build_slp_tree): Reduction defs terminate the recursion.
9603 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9605 (vect_get_vec_defs_for_stmt_copy): Export.
9606 (vect_get_vec_defs): Likewise.
9607 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9609 (vect_get_vec_defs_for_stmt_copy): Declare.
9610 (vect_get_vec_defs): Likewise.
9612 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9614 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9615 parameter with a "loop" parameter and use it instead of the
9616 loop containing DR_STMT. Don't check simple_iv when doing
9617 BB analysis. Describe the two analysis modes in the comment.
9619 2017-07-03 Tom de Vries <tom@codesourcery.com>
9621 PR tree-optimization/69468
9622 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9623 (find_same_succ_bb): Handle ignore_edge_flags.
9625 2017-07-03 Tom de Vries <tom@codesourcery.com>
9627 PR tree-optimization/81192
9628 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9630 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9632 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9634 2017-07-03 Tom de Vries <tom@codesourcery.com>
9636 PR tree-optimization/81192
9637 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9638 BB_SAME_SUCC (bb) == NULL.
9640 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9642 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9645 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9647 * dumpfile.c: Include profile-count.h
9648 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9650 (insert_cond_bb): Update profile.
9651 * tree-cfg.h (insert_cond_bb): Update prototype.
9652 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9653 * tree-dump.c: Do not include tree-cfg.
9655 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9657 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9659 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9661 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9664 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9666 * tree-complex.c (expand_complex_div_wide): update profile.
9668 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9669 Alan Hayward <alan.hayward@arm.com>
9670 David Sherwood <david.sherwood@arm.com>
9672 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9673 (CORETYPES_H): New define.
9674 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9675 (insn-modes-inline.h, s-modes-inline-h): New rules.
9676 (generated_files): Add insn-modes-inline.h.
9677 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9678 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9679 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9680 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9681 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9682 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9683 (build/gencodes.o, build/genconditions.o): Likewise.
9684 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9685 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9686 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9687 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9688 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9689 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9690 * coretypes.h: Include everything up to real.h for generators.
9691 Include insn-modes.h first. Include wide-int-print.h after
9692 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9693 * machmode.h: Don't include insn-modes.h here.
9694 * function-tests.c: Remove includes of signop.h, machmode.h,
9695 double-int.h and wide-int.h.
9697 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9699 * optc-save-gen.awk: Likewise.
9700 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9701 * godump.c: Remove include of wide-int-print.h.
9702 * pretty-print.h: Likewise.
9703 * wide-int-print.cc: Likewise.
9704 * wide-int.cc: Likewise.
9705 * hash-map-tests.c: Remove include of signop.h.
9706 * hash-set-tests.c: Likewise.
9707 * rtl-tests.c: Likewise.
9708 * mkconfig.sh: Remove include of machmode.h.
9709 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9711 (emit_insn_modes_inline_h): ...this new function. Emit the code
9712 into an insn-modes-inline.h header file, adding appropriate
9713 include guards and end comments.
9714 (emit_insn_modes_c_header): Remove include of machmode.h.
9715 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9717 (main): Handle -i flag and call emit_insn_modes_inline_h when
9720 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9722 * tree-ssa-strlen.c (strinfo): Rename the length field to
9723 nonzero_chars. Add a full_string_p field.
9724 (compare_nonzero_chars, zero_length_string_p): New functions.
9725 (get_addr_stridx): Add an offset_out parameter.
9726 Use compare_nonzero_chars.
9727 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9728 (new_strinfo): Update after above changes to strinfo.
9729 (set_endptr_and_length): Set full_string_p.
9730 (get_string_length): Update after above changes to strinfo.
9731 (unshare_strinfo): Update call to new_strinfo.
9732 (maybe_invalidate): Likewise.
9733 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9734 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9735 as a uhwi instead of an shwi. Update after above changes to
9736 strinfo and new_strinfo.
9737 (zero_length_string): Assert that chainsi contains full strings.
9738 Use zero_length_string_p. Update call to new_strinfo.
9739 (adjust_related_strinfos): Update after above changes to strinfo.
9740 Copy full_string_p from origsi.
9741 (adjust_last_stmt): Use zero_length_string_p.
9742 (handle_builtin_strlen): Update after above changes to strinfo and
9743 new_strinfo. Install the lhs as the string length if the previous
9744 entry didn't describe a full string.
9745 (handle_builtin_strchr): Update after above changes to strinfo
9747 (handle_builtin_strcpy): Likewise.
9748 (handle_builtin_strcat): Likewise.
9749 (handle_builtin_malloc): Likewise.
9750 (handle_pointer_plus): Likewise.
9751 (handle_builtin_memcpy): Likewise. Track nonzero characters
9752 that aren't necessarily followed by a nul terminator.
9753 (handle_char_store): Likewise.
9755 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9757 PR tree-optimization/80769
9758 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9759 for malloc and calloc. Document the new invariant that all related
9760 strinfos have delayed lengths or none do.
9761 (verify_related_strinfos): Move earlier in file.
9762 (set_endptr_and_length): New function, split out from...
9763 (get_string_length): ...here. Also set the lengths of related
9765 (zero_length_string): Assert that chainsi has known (rather than
9767 (adjust_related_strinfos): Likewise.
9769 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9771 PR tree-optimization/81136
9772 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9773 assert that two references with the same misalignment have the same
9774 compile-time misalignment if those compile-time misalignments
9777 2017-07-01 Andi Kleen <ak@linux.intel.com>
9779 * print-tree.c (print_node): Print all attributes.
9781 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9783 * cfg.c (scale_bbs_frequencies): New function.
9784 * cfg.h (scale_bbs_frequencies): Declare it.
9785 * cfgloopanal.c (single_likely_exit): Cleanup.
9786 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9788 (scale_loop_profile): Likewise.
9789 (loop_version): Likewise.
9790 (create_empty_loop_on_edge): Update.
9791 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9792 scale_loop_frequencies, scale_loop_profile, loopify,
9793 loop_version): Update prototypes.
9794 * modulo-sched.c (sms_schedule): Update.
9795 * predict.c (unlikely_executed_edge_p): Also check probability.
9796 (probably_never_executed_edge_p): Fix typo.
9797 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9798 * tree-parloops.c (gen_parallel_loop): Update.
9799 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9800 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9801 * tree-ssa-loop-split.c (split_loop): Update.
9802 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9803 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9804 (vect_loop_versioning): Update.
9805 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9807 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9809 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9811 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9813 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9814 to keep profile consistent.
9816 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9818 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9819 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9820 * profile-count.h (max_safe_multiplier): Make unsigned.
9821 (profile_count::guessed_zero): New.
9823 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9825 * bb-reorder.c (fix_up_crossing_landing_pad,
9826 fix_crossing_conditional_branches): Use make_single_succ_edge
9827 to keep profile consistent.
9829 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9831 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9834 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9837 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9838 the right scopes, make sure cond_jump isn't preserved between multiple
9839 iterations. Search for fallthru edge whenever there are 3+ edges and
9840 use find_fallthru_edge for it.
9842 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9844 Patch by Alexander Monakov <amonakov@ispras.ru>
9845 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9846 probabilities consistently.
9848 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9850 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9851 emit_cmp_and_jump_insns.
9853 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9856 * tree-inline.c (expand_call_inline): Combine profile statuses.
9858 2017-06-30 Andrew Pinski <apinski@cavium.com>
9860 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9861 fold_stmt returned true.
9863 2017-06-30 Nathan Sidwell <nathan@acm.org>
9865 * ggc.h (empty_string): Delete.
9866 * cfgexpand.c (expand_asm_stmt): Use plain "".
9867 * optabs.c (expand_asm_memory_barrier): Likewise.
9868 * stringpool.c (empty_string): Delete.
9869 (digit_vector, digit_string): Delete.
9870 (ggc_alloc_string): Use plain "", don't optimize single digit
9871 strings. Use ggc_alloc_atomic.
9873 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9875 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9876 comparison set and one other set, use the cost of the non-comparison
9879 2017-06-30 Nathan Sidwell <nathan@acm.org>
9881 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9884 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9886 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9887 loops. Remove now unneeded calls to gimple_switch_set_label() that
9888 just set removed labels to NULL_TREE.
9890 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9892 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9893 (set_range_info): ...here. Only call set_range_info_raw if domain
9895 (set_nonzero_bits): Call set_range_info_raw.
9896 * tree-ssanames.h (set_range_info_raw): New.
9898 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9901 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9902 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9903 of nonimmediate_operand and <store_mask_constraint> instead of m
9904 for the input operand. For V8FI iterator, always split if input
9905 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9906 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9907 <store_mask_predicate> instead of register_operand and
9908 <store_mask_constraint> instead of v for the input operand. Make
9909 sure both operands aren't MEMs for if not <mask_applied>.
9911 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9913 * lto-wrapper.c (copy_file) Close both file descriptors before
9916 2017-06-30 Martin Liska <mliska@suse.cz>
9919 * multiple_target.c (create_dispatcher_calls): Make ifunc
9920 also for function that don't have calls or are not referenced.
9922 2017-06-30 Richard Biener <rguenther@suse.de>
9924 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9925 analyze the first scalar stmt. Move vector type computation
9926 for the BB case here from ...
9927 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9928 live operation processing in the SLP case properly.
9930 2017-06-30 Richard Biener <rguenther@suse.de>
9932 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9934 2017-06-30 Martin Liska <mliska@suse.cz>
9937 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9938 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9940 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9942 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9945 2017-06-30 Richard Biener <rguenther@suse.de>
9947 PR tree-optimization/81249
9948 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9949 condition reduction result to original scalar type.
9951 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9953 * profile-count.h (enum profile_quality): Fix typos and whitespace
9956 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9958 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9959 type for branch probabilities.
9961 2017-06-29 Julian Brown <julian@codesourcery.com>
9962 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9964 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9965 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9966 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9967 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9969 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9971 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9972 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9973 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9974 CC usage from generic code to here.
9975 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9976 CC usage into the target macros.
9978 2017-06-29 Maya Rashish <coypu@sdf.org>
9980 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9983 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9985 * arm/arm-builtins.c: Include profile-count.h
9986 * except.c (sjlj_emit_function_enter): Use
9987 profile_probability::unlikely.
9989 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9991 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9992 and tocrel_offset be pointer args rather than implicitly using
9994 (legitimate_constant_pool_address_p, rs6000_emit_move,
9995 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9996 tocrel_offset and use in toc_relative_expr_p call.
9997 (print_operand, print_operand_address): Use static tocrel_base_oac
9998 and tocrel_offset_oac.
9999 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10002 2017-06-29 Maya Rashish <coypu@sdf.org>
10004 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10006 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10008 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10009 objects, take into account only the alignment of 'op0' and 'mode1' if
10012 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10014 * ccmp.c (ccmp_tree_comparison_p): New function.
10015 (ccmp_candidate_p): Update to use above function.
10016 (get_compare_parts): New function.
10017 (expand_ccmp_next): Update to use new functions.
10018 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10020 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10021 take mode as argument.
10022 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10023 * expr.c (expand_expr_real_1): Pass mode as argument.
10025 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10027 * combine.c (combine_instructions): Print insns to dump_file, together
10030 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10032 * asan.c (asan_emit_stack_protection): Update.
10033 (create_cond_insert_point): Update.
10034 * auto-profile.c (afdo_propagate_circuit): Update.
10035 * basic-block.h (struct edge_def): Turn probability to
10036 profile_probability.
10037 (EDGE_FREQUENCY): Update.
10038 * bb-reorder.c (find_traces_1_round): Update.
10039 (better_edge_p): Update.
10040 (sanitize_hot_paths): Update.
10041 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10042 (make_single_succ_edge): Update.
10043 (check_bb_profile): Update.
10044 (dump_edge_info): Update.
10045 (update_bb_profile_for_threading): Update.
10046 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10047 probabilitycount to 0.
10048 * cfgbuild.c (compute_outgoing_frequencies): Update.
10049 * cfgcleanup.c (try_forward_edges): Update.
10050 (outgoing_edges_match): Update.
10051 (try_crossjump_to_edge): Update.
10052 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10053 (expand_gimple_tailcall): Update.
10054 (construct_init_block): Use make_single_succ_edge.
10055 (construct_exit_block): Use make_single_succ_edge.
10056 * cfghooks.c (verify_flow_info): Update.
10057 (redirect_edge_succ_nodup): Update.
10058 (split_edge): Update.
10059 (account_profile_record): Update.
10060 * cfgloopanal.c (single_likely_exit): Update.
10061 * cfgloopmanip.c (scale_loop_profile): Update.
10062 (set_zero_probability): Remove.
10063 (duplicate_loop_to_header_edge): Update.
10064 * cfgloopmanip.h (loop_version): Update prototype.
10065 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10066 (force_nonfallthru_and_redirect): Update.
10067 (update_br_prob_note): Update.
10068 (rtl_verify_edges): Update.
10069 (purge_dead_edges): Update.
10070 (rtl_lv_add_condition_to_bb): Update.
10071 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10072 * cgraphunit.c (init_lowered_empty_function): Update.
10073 (cgraph_node::expand_thunk): Update.
10074 * cilk-common.c: Include profile-count.h
10075 * dojump.c (inv): Remove.
10076 (jumpifnot): Update.
10077 (jumpifnot_1): Update.
10078 (do_jump_1): Update.
10080 (do_jump_by_parts_greater_rtx): Update.
10081 (do_compare_rtx_and_jump): Update.
10082 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10083 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10084 * dwarf2cfi.c: Include profile-count.h
10085 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10086 (sjlj_emit_dispatch_table): Likewise.
10087 * explow.c: Include profile-count.h
10088 * expmed.c (emit_store_flag_force): Update.
10089 (do_cmp_and_jump): Update.
10090 * expr.c (compare_by_pieces_d::generate): Update.
10091 (compare_by_pieces_d::finish_mode): Update.
10092 (emit_block_move_via_loop): Update.
10093 (store_expr_with_bounds): Update.
10094 (store_constructor): Update.
10095 (expand_expr_real_2): Update.
10096 (expand_expr_real_1): Update.
10097 * expr.h (try_casesi, try_tablejump): Update prototypes.
10098 * gimple-pretty-print.c (dump_probability): Update.
10099 (dump_profile): New.
10100 (dump_gimple_label): Update.
10101 (dump_gimple_bb_header): Update.
10102 * graph.c (draw_cfg_node_succ_edges): Update.
10103 * hsa-gen.c (convert_switch_statements): Update.
10104 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10105 (find_if_case_1): Update.
10106 (find_if_case_2): Update.
10107 * internal-fn.c (expand_arith_overflow_result_store): Update.
10108 (expand_addsub_overflow): Update.
10109 (expand_neg_overflow): Update.
10110 (expand_mul_overflow): Update.
10111 (expand_vector_ubsan_overflow): Update.
10112 * ipa-cp.c (good_cloning_opportunity_p): Update.
10113 * ipa-split.c (split_function): Use make_single_succ_edge.
10114 * ipa-utils.c (ipa_merge_profiles): Update.
10115 * loop-doloop.c (add_test): Update.
10116 (doloop_modify): Update.
10117 * loop-unroll.c (compare_and_jump_seq): Update.
10118 (unroll_loop_runtime_iterations): Update.
10119 * lra-constraints.c (lra_inheritance): Update.
10120 * lto-streamer-in.c (input_cfg): Update.
10121 * lto-streamer-out.c (output_cfg): Update.
10122 * mcf.c (adjust_cfg_counts): Update.
10123 * modulo-sched.c (sms_schedule): Update.
10124 * omp-expand.c (expand_omp_for_init_counts): Update.
10125 (extract_omp_for_update_vars): Update.
10126 (expand_omp_ordered_sink): Update.
10127 (expand_omp_for_ordered_loops): Update.
10128 (expand_omp_for_generic): Update.
10129 (expand_omp_for_static_nochunk): Update.
10130 (expand_omp_for_static_chunk): Update.
10131 (expand_cilk_for): Update.
10132 (expand_omp_simd): Update.
10133 (expand_omp_taskloop_for_outer): Update.
10134 (expand_omp_taskloop_for_inner): Update.
10135 * omp-simd-clone.c (simd_clone_adjust): Update.
10136 * optabs.c (expand_doubleword_shift): Update.
10137 (expand_abs): Update.
10138 (emit_cmp_and_jump_insn_1): Update.
10139 (expand_compare_and_swap_loop): Update.
10140 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10141 * predict.c (predictable_edge_p): Update.
10142 (edge_probability_reliable_p): Update.
10143 (set_even_probabilities): Update.
10144 (combine_predictions_for_insn): Update.
10145 (combine_predictions_for_bb): Update.
10146 (propagate_freq): Update.
10147 (estimate_bb_frequencies): Update.
10148 (force_edge_cold): Update.
10149 * profile-count.c (profile_count::dump): Add missing space into dump.
10150 (profile_count::debug): Add newline.
10151 (profile_count::differs_from_p): Explicitly convert to unsigned.
10152 (profile_count::stream_in): Update.
10153 (profile_probability::dump): New member function.
10154 (profile_probability::debug): New member function.
10155 (profile_probability::differs_from_p): New member function.
10156 (profile_probability::differs_lot_from_p): New member function.
10157 (profile_probability::stream_in): New member function.
10158 (profile_probability::stream_out): New member function.
10159 * profile-count.h (profile_count_quality): Rename to ...
10160 (profile_quality): ... this one.
10161 (profile_probability): New.
10162 (profile_count): Update.
10163 * profile.c (compute_branch_probabilities): Update.
10164 * recog.c (peep2_attempt): Update.
10165 * sched-ebb.c (schedule_ebbs): Update.
10166 * sched-rgn.c (find_single_block_region): Update.
10167 (compute_dom_prob_ps): Update.
10168 (schedule_region): Update.
10169 * sel-sched-ir.c (compute_succs_info): Update.
10170 * stmt.c (struct case_node): Update.
10171 (do_jump_if_equal): Update.
10172 (get_outgoing_edge_probs): Update.
10173 (conditional_probability): Update.
10174 (emit_case_dispatch_table): Update.
10175 (expand_case): Update.
10176 (expand_sjlj_dispatch_table): Update.
10177 (emit_case_nodes): Update.
10178 * targhooks.c: Update.
10179 * tracer.c (better_p): Update.
10180 (find_best_successor): Update.
10181 * trans-mem.c (expand_transaction): Update.
10182 * tree-call-cdce.c: Update.
10183 * tree-cfg.c (gimple_split_edge): Upate.
10184 (move_sese_region_to_fn): Upate.
10185 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10186 * tree-eh.c (lower_resx): Upate.
10187 (cleanup_empty_eh_move_lp): Upate.
10188 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10189 * tree-inline.c (copy_edges_for_bb): Update.
10190 (copy_cfg_body): Update.
10191 * tree-parloops.c (gen_parallel_loop): Update.
10192 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10193 (gimple_gen_time_profiler): Update.
10194 * tree-ssa-dce.c (remove_dead_stmt): Update.
10195 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10196 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10197 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10198 (unloop_loops): Update.
10199 (try_peel_loop): Update.
10200 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10201 * tree-ssa-loop-split.c (connect_loops): Update.
10202 (split_loop): Update.
10203 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10204 (hoist_guard): Update.
10205 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10206 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10207 (value_replacement): Update.
10208 * tree-ssa-reassoc.c (branch_fixup): Update.
10209 * tree-ssa-tail-merge.c (replace_block_by): Update.
10210 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10211 (create_edge_and_update_destination_phis): Update.
10212 (compute_path_counts): Update.
10213 (recompute_probabilities): Update.
10214 (update_joiner_offpath_counts): Update.
10215 (freqs_to_counts_path): Update.
10216 (duplicate_thread_path): Update.
10217 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10218 (struct switch_conv_info): Update.
10219 (gen_inbound_check): Update.
10220 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10221 (vect_do_peeling): Update.
10222 (vect_loop_versioning): Update.
10223 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10224 (optimize_mask_stores): Update.
10225 * ubsan.c (ubsan_expand_null_ifn): Update.
10226 * value-prof.c (gimple_divmod_fixed_value): Update.
10227 (gimple_divmod_fixed_value_transform): Update.
10228 (gimple_mod_pow2): Update.
10229 (gimple_mod_pow2_value_transform): Update.
10230 (gimple_mod_subtract): Update.
10231 (gimple_mod_subtract_transform): Update.
10232 (gimple_ic): Update.
10233 (gimple_stringop_fixed_value): Update.
10234 (gimple_stringops_transform): Update.
10235 * value-prof.h: Update.
10237 2017-06-29 Carl Love <cel@us.ibm.com>
10239 * config/rs6000/rs6000-c.c: Add support for built-in functions
10240 vector signed int vec_signed (vector float);
10241 vector signed long long vec_signed (vector double);
10242 vector signed int vec_signed2 (vector double, vector double);
10243 vector signed int vec_signede (vector double);
10244 vector signed int vec_signedo (vector double);
10245 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10246 instruction generator.
10247 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10248 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10249 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10251 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10252 vunsignede_v2df): Add define_expands.
10253 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10254 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10255 VEC_UNSIGNEDO): Add definitions.
10256 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10257 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10258 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10259 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10260 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10261 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10262 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10263 vec_unsignede, vec_unsignedo): Add builtin defines.
10264 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10266 * doc/extend.texi: Update the built-in documentation file for the
10267 new built-in functions.
10269 2017-06-29 Richard Biener <rguenther@suse.de>
10271 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10272 reduction chains to LOOP_VINFO_REDUCTIONS.
10273 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10274 SLP reductions after processing reduction chains.
10276 2017-06-29 Nathan Sidwell <nathan@acm.org>
10278 * builtins.c (fold_builtin_FUNCTION): Use
10279 lang_hooks.decl_printable_name.
10281 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10283 PR middle-end/81194
10284 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10285 with only one label.
10286 * stmt.c (expand_case): Assert NCASES is greater than one.
10288 2017-06-29 Richard Biener <rguenther@suse.de>
10290 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10292 (group_case_labels): Likewise.
10293 (find_taken_edge): Push sanity checking on val to workers...
10294 (find_taken_edge_cond_expr): ... here
10295 (find_taken_edge_switch_expr): ... and here, handle cases
10296 with just a default label.
10297 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10298 (group_case_labels): Likewise.
10299 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10300 group_case_labels does anything cleanup the CFG again.
10302 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10304 PR tree-optimization/81196
10305 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10306 exit condition comparing two IVs.
10308 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10310 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10311 profile to the dummy entry at the end of the list of architectures.
10312 * config/arm/arm-cpu-cdata.h: Regenerated.
10314 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10315 Michael Collison <michael.collison@arm.com>
10318 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10320 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10321 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10322 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10323 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10324 of shift when the shift amount is masked with constant equal to
10325 the size of the mode.
10326 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10329 2017-06-29 Martin Liska <mliska@suse.cz>
10331 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10333 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10336 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10339 (edge_growth_cache_entry::edge_growth_cache_entry): New
10341 (reset_edge_growth_cache): Update to use constructor.
10343 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10345 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10346 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10347 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10349 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10351 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10352 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10354 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10356 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10357 (*-linux-uclibc*): Add t-uclibc tmake_file.
10358 * config/t-musl: New.
10359 * config/t-uclibc: New.
10361 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10363 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10365 (gen_comm_data): Emit architectural setting of arch_prof.
10366 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10368 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10369 (armv8-m.base, armv8-m.main): Likewise.
10370 * arm-protos.h (arm_build_target): Add profile field.
10371 (arch_option): Likewise.
10372 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10374 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10375 arm_active_target.profile.
10377 2017-06-28 Richard Biener <rguenther@suse.de>
10379 PR middle-end/81227
10380 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10381 TYPE_OVERFLOW_WRAPS.
10382 * match.pd (negate_expr_p): Likewise.
10383 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10384 fold_build2, not fold_binary.
10386 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10388 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10389 Convert memory address to Pmode.
10390 (aarch64_print_operand): Assert MEM operands are always Pmode.
10392 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10395 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10396 Remove redundant if.
10397 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10398 * config/arm/aarch-common-protos.h
10399 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10400 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10402 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10405 * multiple_target.c (create_dispatcher_calls): Set the default
10406 clone to be static, not public.
10408 2017-06-28 Richard Biener <rguenther@suse.de>
10410 * tree-vect-loop.c (vectorizable_reduction): Move special
10411 cond reduction IV var creation ...
10412 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10413 parameter. Use STMT_VINFO_VECTYPE.
10414 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10417 2017-06-28 Martin Liska <mliska@suse.cz>
10420 * ipa-visibility.c (non_local_p): Handle visibility.
10422 2017-06-28 Martin Liska <mliska@suse.cz>
10425 * common.opt: Add IntegerRange to various options.
10426 * opt-functions.awk (integer_range_info): New function.
10427 * optc-gen.awk: Add integer_range_info to cl_options struct.
10428 * opts-common.c (decode_cmdline_option): Handle
10429 CL_ERR_INT_RANGE_ARG.
10430 (cmdline_handle_error): Likewise.
10431 * opts.c (print_filtered_help): Show valid interval in
10432 when --help is provided.
10433 * opts.h (struct cl_option): Add range_min and range_max fields.
10434 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10436 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10438 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10439 (x * C EQ/NE y * C): New transformation.
10441 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10443 * genmultilib (combination_space): Accept '+' in option names.
10445 2017-06-28 Martin Liska <mliska@suse.cz>
10448 * asan.c (instrument_derefs): Bail out inner references
10449 that are hard register variables.
10451 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10454 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10455 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10457 2017-06-28 Richard Biener <rguenther@suse.de>
10459 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10460 (vect_get_slp_defs): Adjust.
10461 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10463 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10465 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10466 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10467 (vectorizable_reduction): Adjust.
10468 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10470 (vect_get_slp_defs): Likewise.
10471 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10472 (vectorizable_bswap): Adjust.
10473 (vectorizable_call): Likewise.
10474 (vectorizable_conversion): Likewise.
10475 (vectorizable_assignment): Likewise.
10476 (vectorizable_shift): Likewise.
10477 (vectorizable_operation): Likewise.
10478 (vectorizable_store): Likewise.
10479 (vectorizable_condition): Likewise.
10480 (vectorizable_comparison): Likewise.
10482 2017-06-28 Michael Collison <michael.collison@arm.com>
10485 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10487 (arm_gen_movmemqi): Removed unused variable 'i'.
10488 Convert 'for' loop into 'while' loop.
10489 (arm_expand_prologue): Remove last unnecessary set of insn.
10490 (thumb_pop): Remove unused variable 'pushed_words'.
10491 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10493 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10495 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10496 * config/s390/s390-protos.h: Add prototype of
10497 s390_rel_address_ok_p.
10498 * config/s390/s390.c (s390_got_symbol): New function.
10499 (s390_rel_address_ok_p): New function.
10500 (legitimize_pic_address): Use s390_rel_address_ok_p.
10501 (s390_load_got): Use s390_got_symbol.
10502 (s390_option_override): Issue error if
10503 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10504 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10506 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10508 2017-06-27 Andrew Pinski <apinski@cavium.com>
10510 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10511 (X * copysign (1.0, X)): New pattern.
10512 (X * copysign (1.0, -X)): New pattern.
10513 (copysign (-1.0, CST)): New pattern.
10515 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10517 * genmultilib (combination_space): Remove variable.
10518 Validate reuse rules against regular expression for any sequence
10519 of multilib options in any order.
10521 2017-06-27 Michael Collison <michael.collison@arm.com>
10523 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10524 call aarch64_split_simd_combine.
10525 * (aarch64_combine_internal<mode>): Delete pattern.
10526 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10527 Allow register and subreg operands.
10529 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10531 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10532 specific need, just fallback on defaults.
10533 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10535 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10536 Olivier Hainque <hainque@adacore.com>
10538 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10540 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10541 targets. Pick a default if no particular attempt applied.
10542 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10545 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10547 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10548 (x86_64-wrs-vxworks7): Likewise.
10550 2017-06-27 Marek Polacek <polacek@redhat.com>
10553 * ubsan.c (instrument_null): Check get_base_address's result for null.
10555 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10557 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10559 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10561 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10562 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10563 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10564 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10565 New function types.
10566 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10567 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10568 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10569 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10570 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10571 BUILT_IN_FEUPDATEENV): New builtins.
10572 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10573 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10574 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10575 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10577 (builtin_structptr_types): Adjust size.
10578 * tree.c (builtin_structptr_types): Add four entries.
10580 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10581 Olivier Hainque <hainque@adacore.com>
10583 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10584 (TLS_SYM): New local macro, forcing reference to __tls__ on
10585 link command lines for VxWorks 7 RTPs, triggering initialization
10587 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10588 OS features TLS support, true for RTPs on VxWorks 7.
10589 * config/vxworks.c (vxworks_override_options): Setup emutls
10592 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10594 * predict.c (test_prediction_value_range): Use -1U instead of -1
10595 to avoid narrowing conversion warning.
10596 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10597 to avoid narrowing conversion warning.
10598 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10600 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10602 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10604 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10605 64bit configurations.
10606 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10607 (SIZE_TYPE): Likewise.
10608 * config/vxworks.c (vxworks_emutls_var_fields): Use
10609 long_unsigned_type_node instead of unsigned_type_node as the offset
10610 field type, which is "pointer" mode in emutls.c.
10612 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10615 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10617 PR middle-end/81207
10618 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10619 gimple_vuse copying separately from gimple_vdef copying.
10621 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10623 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10626 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10627 Olivier Hainque <hainque@adacore.com>
10629 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10630 for all vxworks7 targets.
10631 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10632 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10633 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10634 variations for VX6/VX7 and 32/64bits later on in ...
10635 (VXWORKS_LIB_SPEC): Leverage new macros.
10636 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10637 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10639 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10641 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10642 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10644 2017-06-26 Carl Love <cel@us.ibm.com>
10646 * config/rs6000/rs6000-c.c: Add support for built-in functions
10647 vector bool char vec_reve (vector bool char);
10648 vector signed char vec_reve (vector signed char);
10649 vector unsigned char vec_reve (vector unsigned char);
10650 vector bool int vec_reve (vector bool int);
10651 vector signed int vec_reve (vector signed int);
10652 vector unsigned int vec_reve (vector unsigned int);
10653 vector bool long long vec_reve (vector bool long long);
10654 vector signed long long vec_reve (vector signed long long);
10655 vector unsigned long long vec_reve (vector unsigned long long);
10656 vector bool short vec_reve (vector bool short);
10657 vector signed short vec_reve (vector signed short);
10658 vector double vec_reve (vector double);
10659 vector float vec_reve (vector float);
10660 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10661 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10662 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10663 (altivec_vreve): New pattern.
10664 * config/rs6000/altivec.h (vec_reve): New define.
10665 * doc/extend.texi (vec_rev): Update the built-in documentation file
10666 for the new built-in functions.
10668 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10670 PR tree-optimization/71815
10671 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10673 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10675 (slsr_process_phi): Likewise.
10676 (replace_uncond_cands_and_profitable_phis): Don't replace a
10677 multiply candidate with a stride of 1 (copy or cast).
10678 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10680 (lowest_cost_path): Likewise.
10681 (total_savings): Likewise.
10683 2017-06-26 Richard Biener <rguenther@suse.de>
10686 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10687 Use def_builtin_pure for all gather builtins.
10689 2017-06-26 Richard Biener <rguenther@suse.de>
10691 PR tree-optimization/81203
10692 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10693 non-dominating BBs.
10695 2017-06-26 Marek Polacek <polacek@redhat.com>
10698 * doc/invoke.texi: Document -Wmultistatement-macros.
10700 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10702 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10703 arm_neon_ok_no_float_abi effective target.
10705 2017-06-26 Richard Biener <rguenther@suse.de>
10707 PR tree-optimization/80928
10708 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10709 (copy_bbs): Set BB_DUPLICATED flag early.
10710 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10712 (execute_on_shrinking_pred): Likewise.
10713 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10714 BB_DUPLICATED blocks.
10715 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10716 iterate over all PHIs considering removal of *gsi.
10718 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10720 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10723 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10725 * config/rs6000/rs6000-string.c: (expand_block_clear,
10726 do_load_for_compare, select_block_compare_mode,
10727 compute_current_alignment, expand_block_compare,
10728 expand_strncmp_align_check, expand_strn_compare,
10729 expand_block_move, rs6000_output_load_multiple)
10730 Move functions related to string/block move/compare
10731 to a separate file.
10732 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10733 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10734 for this function which is now used in two files.
10735 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10736 * config.gcc: Add rs6000-string.o to extra_objs for
10737 targets powerpc*-*-* and rs6000*-*-*.
10739 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10742 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10743 32-bit, since indexed is not valid for DImode.
10744 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10745 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10746 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10747 (define_peephole2 for Altivec d-form store): Likewise.
10750 * multiple_target.c (create_dispatcher_calls): Only create the
10751 dispatcher call if the function is the default clone of a
10752 versioned function.
10754 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10756 PR middle-end/80902
10757 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10758 a call, force the call to not be a tail call.
10760 2017-06-23 Jeff Law <law@redhat.com>
10762 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10765 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10767 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10768 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10769 handling for early expansion of vector shifts (sl,sr,sra,rl).
10770 (builtin_function_type): Add vector shift right instructions
10771 to the unsigned argument list.
10773 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10775 rtl-optimizatoin/79286
10776 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10777 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10778 trap. PIC register plus a const unspec without offset can never trap.
10780 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10782 * tree.h (builtin_structptr_type): New type.
10783 (builtin_structptr_types): Declare new array.
10784 * tree.c (builtin_structptr_types): New array.
10785 (free_lang_data, build_common_tree_nodes): Use it.
10787 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10790 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10793 2017-06-22 Matt Turner <mattst88@gmail.com>
10795 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10796 Lake models to skylake case. Assume skylake for unknown
10797 models with clflushopt.
10799 2017-06-22 Jeff Law <law@redhat.com>
10801 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10802 frame sizes that do not satisfy aarch64_uimm12_shift.
10804 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10806 * profile-count.h (apply_probability,
10807 apply_scale, probability_in): Fix checks for zero.
10809 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10811 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10812 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10814 2016-06-22 Richard Biener <rguenther@suse.de>
10816 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10817 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10818 REDUC_MAX_EXPR support.
10819 (vectorizable_reduction): Likewise.
10820 (vect_create_epilog_for_reduction): Likewise.
10822 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10824 * match.pd (A / (1 << B) -> A >> B): New.
10825 * generic-match-head.c: Include optabs-tree.h.
10826 * gimple-match-head.c: Likewise.
10827 * optabs-tree.h (target_supports_op_p): New.
10828 * optabs-tree.c (target_supports_op_p): New.
10830 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10832 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10833 $gcc_cv_ld --help output.
10834 (gcc_cv_ld_demangle): Likewise.
10835 (gcc_cv_ld_eh_frame_hdr): Likewise.
10836 (gcc_cv_ld_pie): Likewise.
10837 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10838 (gcc_cv_ld_buildid): Likewise.
10839 (gcc_cv_ld_sysroot): Likewise.
10840 (ld_bndplt_support): Likewise.
10841 (ld_pushpopstate_support): Likewise.
10842 * configure: Regenerate.
10843 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10845 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10848 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10849 operands indexes to avoid gap between operands and match_dups.
10851 2017-06-21 Andrew Pinski <apinski@cavium.com>
10853 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10854 Increment Arith_shift and Arith_shift_reg by 1.
10855 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10857 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10858 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10859 (aarch64_strip_extend): Add new argument and test for it.
10860 (aarch64_cheap_mult_shift_p): New function.
10861 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10862 add a cost if it is true.
10863 Update calls to aarch64_strip_extend.
10864 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10866 2017-06-21 Andrew Pinski <apinski@cavium.com>
10868 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10870 (thunderxt88): Likewise.
10871 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10872 (thunderx_prefetch_tune): New variable.
10873 (thunderx2t99_prefetch_tune): Update for the correct values.
10874 (thunderxt88_tunings): New variable.
10875 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10876 generic_prefetch_tune.
10877 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10879 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10881 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10882 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10883 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10884 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10885 (aarch64_atomic_cas<mode>, GPI): Likewise.
10887 2017-06-21 Martin Liska <mliska@suse.cz>
10889 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10890 statements on cold and hot labels.
10891 * predict.c (tree_estimate_probability_bb): Remove the
10892 prediction from this place.
10894 2017-06-21 Martin Liska <mliska@suse.cz>
10896 PR tree-optimization/79489
10897 * gimplify.c (maybe_add_early_return_predict_stmt): New
10899 (gimplify_return_expr): Call the function.
10900 * predict.c (tree_estimate_probability_bb): Remove handling
10902 * predict.def: Update comment about early return predictor.
10903 * gimple-predict.h (is_gimple_predict): New function.
10904 * predict.def: Change default value of early return to 66.
10905 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10907 * passes.def: Put pass_strip_predict_hints to the beginning of
10910 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10912 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10913 FUNCTION_DECL declarations.
10914 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10916 (dwaf2out_decl): Likewise.
10917 * godump.c (go_early_global_decl): Skip call to the real debug hook
10918 for FUNCTION_DECL declarations.
10919 * passes.c (rest_of_decl_compilation): Skip call to the
10920 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10921 -fdump-go-spec is passed.
10923 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10925 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10926 Reorder for compactness.
10927 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10928 (def_builtin_pure, def_builtin_pure2): New functions.
10929 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10931 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10933 * match.pd (nop_convert): New predicate.
10934 ((A +- CST1) +- CST2): Allow some NOP conversions.
10936 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10939 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10940 with ctors/dtors if GOVD_SHARED is set.
10942 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10944 * config/aarch64/aarch64.md (movti_aarch64):
10945 Emit mov rather than orr.
10946 (movtf_aarch64): Likewise.
10947 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10948 Emit mov rather than orr.
10950 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10952 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10953 Swap alternatives, make integer dup more expensive.
10955 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10957 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10958 Return true for non-tls symbols.
10960 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10962 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10963 (cortex-a75): Likewise.
10964 (cortex-a75.cortex-a55): Likewise.
10965 * config/aarch64/aarch64-tune.md: Regenerate.
10966 * doc/invoke.texi (-mtune): Document new values for -mtune.
10968 2017-06-21 Tom de Vries <tom@codesourcery.com>
10970 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10971 stack_size feature.
10972 (Effective-Target Keywords, Other attributes): Suggest using
10973 dg-add-options stack_size feature to get stack limit in stack_size
10974 effective target documentation.
10976 2017-06-21 Julian Brown <julian@codesourcery.com>
10977 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10979 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10980 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10981 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10983 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10984 attribute type list for neon_multiply.
10985 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10986 attribute type list for neon_multiply.
10987 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10988 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10989 attribute type list for neon_multiply.
10990 * config/arm/types.md (crypto_pmull): Add.
10991 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10992 attribute type list.
10994 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10996 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10999 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11001 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11002 to make sure not to dereference a NULL cost_classes_ptr pointer.
11004 2017-06-20 Carl Love <cel@us.ibm.com>
11006 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11007 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11008 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11009 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11010 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11011 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11012 VMULOSW): New enum "unspec" values.
11013 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11014 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11015 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11016 altivec_vmulosw): New patterns.
11017 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11018 VMULOSW): Add definitions.
11020 2017-06-20 Julia Koval <julia.koval@intel.com>
11022 * config/i386/i386.c: Fix rounding expand for new pattern.
11023 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11025 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11027 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11028 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11030 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11032 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11035 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11037 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11038 architecture, then by implementer ID.
11039 * config/aarch64/aarch64-tune.md: Regenerate.
11041 2017-06-20 Richard Biener <rguenther@suse.de>
11043 PR middle-end/81097
11044 * fold-const.c (split_tree): Fold to type before negating.
11046 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11048 * diagnostic-show-locus.c
11049 (selftest::test_fixit_deletion_affecting_newline): New function.
11050 (selftest::diagnostic_show_locus_c_tests): Call it.
11052 2017-06-20 Andreas Schwab <schwab@suse.de>
11055 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11058 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11060 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11061 __ARM_FEATURE_COPROC according to support.
11063 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11065 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11066 Rewritten to avoid overflow for > 32-bit pointers.
11069 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11070 by removing enum keyword.
11071 (ubsan_type_descriptor): Likewise. Formatting fix.
11074 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11075 splitter): Require TARGET_SSE2 in the condition.
11077 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11080 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11081 for doing vector set of SFmode on ISA 3.0.
11082 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11083 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11085 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11086 SFmode value into a V4SF variable that was extracted from another
11087 V4SF variable without converting the element to double precision
11088 and back to single precision vector format.
11089 (vsx_insert_extract_v4sf_p9_2): Likewise.
11091 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11093 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11094 in UWHI to avoid undefined overflow.
11097 * ubsan.h (enum ubsan_encode_value_phase): New.
11098 (ubsan_encode_value): Change second argument to
11099 enum ubsan_encode_value_phase with default value of
11100 UBSAN_ENCODE_VALUE_GENERIC.
11101 * ubsan.c (ubsan_encode_value): Change second argument to
11102 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11103 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11104 create_tmp_var_raw instead of create_tmp_var and use a
11106 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11107 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11108 ubsan_encode_value callers.
11111 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11112 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11113 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11115 2017-06-19 Richard Biener <rguenther@suse.de>
11117 PR middle-end/81118
11118 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11119 estimates if we changed anything.
11121 2017-06-19 Richard Biener <rguenther@suse.de>
11123 PR tree-optimization/80887
11124 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11125 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11126 simplified lookups, then reset mprts_hook.
11127 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11129 (try_to_simplify): Likewise.
11131 2017-06-19 Martin Liska <mliska@suse.cz>
11134 * gimplify.c (gimplify_switch_expr):
11135 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11137 2017-06-19 Martin Liska <mliska@suse.cz>
11139 * doc/install.texi: Document that PGO runs in 4 stages.
11141 2017-06-19 Martin Liska <mliska@suse.cz>
11144 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11145 to dispatcher function name.
11146 * multiple_target.c (replace_function_decl): New function.
11147 (create_dispatcher_calls): Redirect both edges and references.
11149 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11151 * profile-count.c (profile_count::dump): Dump quality.
11152 (profile_count::differs_from_p): Update for unsigned val.
11153 * profile-count.h (profile_count_quality): New enum.
11154 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11156 2017-06-19 Richard Biener <rguenther@suse.de>
11158 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11159 struct function as arg.
11160 (estimate_numbers_of_iterations): Export overload with loop arg.
11161 (free_numbers_of_iterations_estimates_loop): Use an overload of
11162 free_numbers_of_iterations_estimates instead.
11163 * tree-cfg.c (remove_bb): Adjust.
11164 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11165 * tree-parloops.c (gen_parallel_loop): Likewise.
11166 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11168 (tree_unroll_loops_completely): Likewise.
11169 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11170 Use an overload instead and export.
11171 (estimated_loop_iterations): Adjust.
11172 (max_loop_iterations): Likewise.
11173 (likely_max_loop_iterations): Likewise.
11174 (estimate_numbers_of_iterations): Take struct function as arg
11176 (loop_exits_before_overflow): Adjust.
11177 (free_numbers_of_iterations_estimates_loop): Use an overload.
11178 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11179 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11181 2017-06-19 Richard Biener <rguenther@suse.de>
11184 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11185 RANGE_EXPR conservatively.
11187 2017-06-16 Carl Love <cel@us.ibm.com>
11189 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11190 definitions for vec_float, vec_float2, vec_floato,
11191 vec_floate built-ins.
11192 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11193 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11195 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11196 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11197 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11198 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11199 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11200 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11201 vec_floato): Add builtin defines.
11202 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11203 Update the built-in documentation file for the new built-in
11206 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11208 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11209 (mthumb): Mark as the negative of -marm.
11211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11213 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11215 (ARM Options, -mtune): Document that this accepts the same
11216 extension options as -mcpu.
11217 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11219 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11221 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11222 permitted extensions.
11224 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11226 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11227 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11228 (armv8-m.main): Add option +nodsp.
11229 * config/arm/arm-cpu-cdata.h: Regenerated.
11231 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11233 * config/arm/t-fuchsia: New file.
11234 * config.gcc (arm*-*-fuchsia*): Use it.
11236 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11238 * config/arm/t-symbian: Rewrite for new option infrastructure.
11240 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11242 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11243 (MULTILIB_REQUIRED): Likewise.
11245 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11247 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11248 (MULTILIB_RESUE): Likewise.
11249 (MULTILIB_MATCHES): Likewise.
11250 (MULTLIB_REQUIRED): Likewise.
11252 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11254 * config/arm/t-rtems: Rewrite for new option framework.
11256 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11258 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11259 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11260 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11261 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11262 * config/arm/t-multilib: ... here.
11263 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11264 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11265 armv7-a and armv8*-a when A-profile libraries have not been built.
11266 * config/arm/t-rmprofile: Rewrite.
11268 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11270 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11271 with a backslash. Remove the backslash after substituting unescaped
11273 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11275 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11277 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11278 the driver to pass through the default setting of -mfloat-abi.
11279 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11281 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11282 (all_feat_combs): New rule.
11283 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11285 * config/arm/t-aprofile: Rewrite.
11287 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11289 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11290 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11291 fpu is not specified by the user/command-line.
11292 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11293 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11294 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11295 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11296 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11297 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11299 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11301 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11302 * config/arm/t-arm-elf: Rewritten.
11304 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11306 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11307 have some floating-point instructions.
11308 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11309 (TARGET_MAYBE_HARD_FLOAT): New macro.
11310 * config/arm/arm-builtins.c (arm_init_builtins): Use
11311 TARGET_MAYBE_HARD_FLOAT.
11312 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11314 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11316 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11317 (configargs.h): Include it.
11318 (arm_print_hint_for_fpu_option): New function.
11319 (arm_parse_fpu_option): New function.
11320 (candidate_extension): New class.
11321 (arm_canon_for_multilib): New function.
11322 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11323 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11324 (ARCH_CANONICAL_SPECS): New macro.
11325 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11327 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11329 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11330 are set after handling multilib fragments. Set target_cpu_default2
11333 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11335 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11337 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11338 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11340 (check_arch): Likewise.
11341 * config/arm/arm.c (arm_configure_build_target): Handle
11342 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11343 options in the default.
11345 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11347 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11348 when an option is an alias of another.
11349 * config/arm/parsecpu.awk (optalias): New parser token.
11350 (gen_comm_data): Mark non-alias options as such. Emit entries
11351 for extension aliases.
11352 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11353 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11354 (armv6kz, armv6zk, armv6t2): Likewise.
11355 (armv7): Make vfpv3-d16 an alias.
11356 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11358 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11359 Sort in canonical order.
11360 (armv8-a): Sort in canonical order.
11361 (armv8.1-a, armv8.2-a): Likewise.
11362 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11364 (cortex-a9): Sort in canonical order.
11365 * config/arm/arm.c (selftests.h): Include it.
11366 (arm_test_cpu_arch_data): New function.
11367 (arm_run_self_tests): New function.
11368 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11369 (targetm): Move declaration to the end of the file.
11370 * arm-cpu-cdata.h: Regenerated.
11372 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11374 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11375 call to target_mode_check describing the type of option passed.
11376 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11377 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11378 arm_parse_cpu_option_name to match parameters against list of
11380 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11381 arm_arch_core_flags data structure.
11382 * config/arm/arm-cpu_cdata.h: Regenerated.
11384 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11386 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11388 (arm_print_hint_for_cpu_option): Likewise.
11389 (arm_print_hint_for_arch_option): Likewise.
11390 (arm_parse_cpu_option_name): Likewise.
11391 (arm_parse_arch_option_name): Likewise.
11392 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11393 of entries in the all_fpus list.
11394 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11395 (arm_parse_cpu_option_name): Declare.
11396 (arm_parse_arch_option_name): Declare.
11397 (arm_parse_option_features): Declare.
11398 (arm_intialize_isa): Declare.
11399 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11401 (gen_comm_data): ... here. Make definitions non-static.
11402 * config/arm/arm-cpu-data.h: Regenerated.
11403 * config/arm/arm-cpu-cdata.h: Regenerated.
11405 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11407 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11408 (cpu_arch_extension): New structure.
11409 (cpu_arch_option, arch_option, cpu_option): New structures.
11410 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11411 architecture types.
11412 (gen_data): Generate new format data tables.
11413 * config/arm/arm.c (cpu_tune): New structure.
11414 (cpu_option, processors): Delete.
11415 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11416 (arm_print_hint_for_cpu_option): ... this and ...
11417 (arm_print_hint_for_arch_option): ... this.
11418 (arm_parse_arch_cpu_name): Delete. Replace with ...
11419 (arm_parse_cpu_option_name): ... this and ...
11420 (arm_parse_arch_option_name): ... this.
11421 (arm_unrecognized_feature): Change type of target parameter to
11423 (arm_parse_arch_cpu_features): Delete. Replace with ...
11424 (arm_parse_option_features): ... this.
11425 (arm_configure_build_target): Rework to use new configuration data
11427 (arm_print_tune_info): Rework for new configuration data tables.
11428 * config/arm/arm-cpu-data.h: Regenerated.
11429 * config/arm/arm-cpu.h: Regenerated.
11431 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11433 * Makefile.in (OBJS): Move sbitmap.o from here ...
11434 (OBJS-libcommon): ... to here.
11436 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11438 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11439 (ISA_ALL_CRYPTO): New macro.
11440 (ISA_ALL_SIMD): New macro
11441 (ISA_ALL_FP): New macro.
11442 * config/arm/arm.c (fpu_bitlist): Update initializer.
11443 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11445 (arm9e): Add fpu. Add option for nofp
11446 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11447 (arm926ej-s, arm1026ej-s): Likewise.
11448 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11449 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11450 neon-fp16, neon-vfpv4, nofp and nosimd.
11451 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11452 (cortex-a8): Add fpu. Add option for nofp.
11453 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11454 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11455 (cortex-r4f): Add fpu.
11456 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11457 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11459 (cortex-r8): Likewise.
11460 (cortex-m4): Add fpu. Add option for nofp.
11461 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11462 (cortex-a17.cortex-a7): Likewise.
11463 (cortex-a32): Add fpu. Add options for crypto and nofp.
11464 (cortex-a35, cortex-a53): Likewise.
11465 (cortex-a57): Add fpu. Add option for crypto.
11466 (cortex-a72, cortex-a73): Likewise.
11467 (exynos-m1): Likewise.
11468 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11469 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11470 (cortex-m33): Add fpu. Add option for nofp.
11471 * config/arm/arm-cpu-cdata.h: Regenerated
11472 * config/arm/arm-cpu-data.h: Regenerated.
11474 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11476 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11477 (armv5te, armv5tej): Likewise.
11478 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11479 (armv7): Add options fp and vfpv3-d16.
11480 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11481 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11483 (armv7ve): Likewise.
11484 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11485 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11486 (armv8-a): Add nocrypto option.
11487 (armv8.1-a, armv8.2-a): Likewise.
11488 (armv8-m.main): add options fp, fp.dp and nofp.
11490 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11492 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11494 (armv8-a+crc): Delete.
11495 (armv8.1-a): Add options simd, crypto and nofp.
11496 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11497 (armv8.2-a+fp16): Delete.
11498 (armv8-m.main): Add option dsp.
11499 (armv8-m.main+dsp): Delete.
11500 (cortex-a8): Add fpu. Add nofp option.
11501 (cortex-a9): Add fpu. Add nofp and nosimd options.
11502 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11503 link to main cpu and architecture data structures.
11504 (gen_comm_data): Only put isa attributes from the main architecture
11506 (option): New statement for architecture and CPU entries.
11507 * arm.c (struct cpu_option): New structure.
11508 (struct processors): Add entry for options.
11509 (arm_unrecognized_feature): New function.
11510 (arm_parse_arch_cpu_name): Ignore any characters after the first
11512 (arm_parse_arch_cpu_feature): New function.
11513 (arm_configure_build_target): Separate out any CPU and architecture
11514 features and parse separately. Don't error out if -mfpu=auto is
11515 used with only an architecture string.
11516 (arm_print_asm_arch_directives): New function.
11517 (arm_file_start): Call it.
11518 * config/arm/arm-cpu-cdata.h: Regenerated.
11519 * config/arm/arm-cpu-data.h: Likewise.
11520 * config/arm/arm-tables.opt: Likewise.
11522 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11524 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11525 assembler when it is not -mfpu=auto.
11527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11529 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11530 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11531 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11532 (ASM_CPU_SPEC): Rewrite.
11533 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11534 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11535 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11536 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11537 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11538 copied string is NUL-terminated. Also strip any characters prefixed
11540 (arm_rewrite_selected_arch): New function.
11541 (arm_rewrite_march): New function.
11543 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11545 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11546 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11547 (march, mcpu, mtune): Convert to string-based options.
11548 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11549 (arm_parse_arch_cpu_name): New function.
11550 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11551 identify selected architecture or CPU.
11552 (arm_option_save): New function.
11553 (TARGET_OPTION_SAVE): Redefine.
11554 (arm_option_restore): Restore string options.
11555 (arm_option_print): Print string options.
11557 2017-06-16 Martin Sebor <msebor@redhat.com>
11559 PR tree-optimization/80933
11560 PR tree-optimization/80934
11561 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11562 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11563 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11564 (gimple_fold_builtin): Call them.
11566 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11568 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11569 as unlikely; update profile.
11571 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11573 * predict.c (force_edge_cold): Handle declaring edges impossible
11576 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11578 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11580 (try_unroll_loop_completely): Fix reporting.
11582 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11584 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11586 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11589 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11590 if given a non-constant argument for an intrinsic which requires a
11593 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11595 * profile.c (compare_freqs): New function.
11596 (branch_prob): Sort edge list.
11597 (find_spanning_tree): Assume that the list is priority sorted.
11599 2017-06-16 Richard Biener <rguenther@suse.de>
11601 PR tree-optimization/81090
11602 * passes.def (pass_record_bounds): Remove.
11603 * tree-pass.h (make_pass_record_bounds): Likewise.
11604 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11605 make_pass_record_bounds): Likewise.
11606 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11607 not free niter estimates at the beginning but at the end.
11608 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11610 2017-06-16 Richard Biener <rguenther@suse.de>
11612 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11613 initializer to workaround ICE in host GCC 4.8.
11615 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11617 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11619 (clone_inlined_nodes): Update.
11621 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11623 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11624 prefetch settings, and enable prefetching by default at -O3.
11626 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11628 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11629 Set flag_prefetch_loop_arrays according to tuning data.
11631 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11633 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11634 New tune structure.
11635 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11636 [Unrelated to main purpose of the patch] Place the pointer field last
11637 to enable type checking errors when tune structure are wrongly merged.
11638 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11639 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11640 (thunderx2t99_prefetch_tune): New tune constants.
11641 (tune_params *_tunings): Update all tunings (no functional change).
11642 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11643 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11644 from tunings structures.
11646 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11649 * ubsan.c (instrument_null): Add T argument, use it instead
11650 of computing it based on IS_LHS.
11651 (instrument_object_size): Likewise.
11652 (pass_ubsan::execute): Adjust instrument_null and
11653 instrument_object_size callers to pass gimple_get_lhs or
11654 gimple_assign_rhs1 result to it. Use instrument_null instead of
11655 calling get_base_address and instrument_mem_ref. Handle
11656 aggregate call arguments for object-size sanitization.
11658 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11660 PR tree-optimization/81089
11661 * tree-vrp.c (is_masked_range_test): Validate operands of
11664 2017-06-15 Martin Sebor <msebor@redhat.com>
11667 * dumpfile.c (dump_register): Avoid calling memset to initialize
11668 a class with a default ctor.
11669 * gcc.c (struct compiler): Remove const qualification.
11670 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11671 * hash-table.h: Ditto.
11672 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11674 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11675 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11677 * params.h (struct param_info): Make struct members non-const.
11678 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11679 with default initialization.
11680 * vec.h (vec_copy_construct, vec_default_construct): New helper
11682 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11683 with vec_copy_construct.
11684 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11685 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11686 * doc/invoke.texi (-Wclass-memaccess): Document.
11688 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11690 * emit-rtl.h (is_leaf): Update comment about local
11691 register allocator.
11693 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11696 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11697 for a variable to have a section before checking if the section has a
11699 Set section to.persistent if persistent attribute is set.
11700 Warn if .persistent attribute is used on an automatic variable.
11702 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11704 PR rtl-optimization/80474
11705 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11707 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11709 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11712 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11714 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11715 (casesi): Emit bounds checking as RTL.
11716 (casesi_internal_mips16_<mode>): Remove bounds checking.
11718 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11720 * config/xtensa/xtensa.c (xtensa_option_override): Append
11721 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11722 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11723 xtensa_doloop_hooks): Define unconditionally.
11724 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11726 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11727 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11728 for it in xtensa_option_override.
11729 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11730 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11732 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11734 * doc/cppopts.texi: Document '-' special value to -MF.
11736 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11738 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11739 (cortex_a53_fconst): Likewise.
11740 (cortex_a53_fpmul): Likewise.
11741 (cortex_a53_f_load_64): Likewise.
11742 (cortex_a53_f_load_many): Likewise.
11743 (cortex_a53_advsimd_alu): Likewise.
11744 (cortex_a53_advsimd_alu_q): Likewise.
11745 (cortex_a53_advsimd_mul): Likewise.
11746 (cortex_a53_advsimd_mul_q): Likewise.
11747 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11748 Add missing fmul, r2f_cvt and fconst cases.
11750 2017-06-14 Richard Biener <rguenther@suse.de>
11752 PR middle-end/81088
11753 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11755 (fold_binary_loc): When associating do not treat pre-existing
11756 TREE_OVERFLOW on literal constants as a reason to allow
11757 TREE_OVERFLOW on associated literal constants.
11759 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11761 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11762 (MASK_FEATURES): New macro.
11763 * config/sparc/sparc.c (sparc_option_override): Remove the special
11764 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11766 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11768 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11769 a division of 0 if non-call exceptions are enabled.
11771 2017-06-14 Andrew Pinski <apinski@cavium.com>
11772 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11775 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11776 Improve vector initialization code gen for only variable case.
11778 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11780 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11782 2017-06-14 Richard Biener <rguenther@suse.de>
11784 PR tree-optimization/81083
11785 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11788 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11790 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11791 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11792 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11793 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11794 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11795 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11797 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11799 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11800 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11802 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11804 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11806 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11808 * config/rs6000/t-rtems: Don't handle SPE.
11810 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11812 * config/rs6000/t-linux: Don't handle SPE.
11814 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11816 * config/rs6000/eabispe.h: Delete file.
11818 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11820 * config/rs6000/t-spe: Delete file.
11822 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11824 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11825 (rs6000_legitimate_offset_address_p): Return false for anything in
11826 V2SImode or V2SFmode.
11828 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11830 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11831 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11832 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11834 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11835 (rs6000_legitimate_offset_address_p): Ditto.
11836 (rs6000_emit_move): Ditto.
11837 (rs6000_init_builtins): Remove V4HI_type_node.
11839 2017-06-13 Martin Liska <mliska@suse.cz>
11842 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11843 (gate_asan): Likewise.
11844 * asan.h (asan_no_sanitize_address_p): Remove the function.
11845 (sanitize_flags_p): New function.
11846 * builtins.def: Fix coding style.
11847 * common.opt: Use renamed enum value.
11848 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11849 * doc/extend.texi: Document no_sanitize attribute.
11850 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11851 to SANITIZE_UNDEFINED_NONDEFAULT.
11852 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11853 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11854 Use sanitize_flags_p.
11855 * gimplify.c (gimplify_function_tree): Likewise.
11856 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11857 * opts.c (parse_no_sanitize_attribute): New function.
11858 (common_handle_option): Use renamed enum value.
11859 * opts.h (parse_no_sanitize_attribute): Declare.
11860 * tree.c (sanitize_flags_p): New function.
11861 * tree.h: Declared here.
11862 * tsan.c: Use sanitize_flags_p.
11863 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11864 (instrument_mem_ref): Likewise.
11865 (instrument_bool_enum_load): Likewise.
11866 (do_ubsan_in_current_function): Remove the function.
11867 (pass_ubsan::execute): Use sanitize_flags_p.
11868 * ubsan.h: Remove do_ubsan_in_current_function
11869 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11870 (dump_function_to_file): Use it here.
11872 2017-06-13 Martin Jambor <mjambor@suse.cz>
11874 PR tree-optimization/80803
11875 PR tree-optimization/81063
11876 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11877 (propagate_subaccesses_across_link): Enqueue subtree whenever
11878 necessary instead of relying on the caller.
11880 2017-06-13 Martin Jambor <mjambor@suse.cz>
11882 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11883 that have a first_link.
11884 (sort_and_splice_var_accesses): Do not check first_link before
11886 (subtree_mark_written_and_enqueue): Likewise.
11887 (propagate_all_subaccesses): Likewise and do not stop at first
11888 parent with a first_link.
11890 2017-06-13 Martin Jambor <mjambor@suse.cz>
11892 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11895 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11897 * match.pd: New pattern.
11899 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11901 * tree-vrp.c (is_masked_range_test): New function.
11902 (register_edge_assert_for): Determine ranges for
11905 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11907 PR tree-optimization/67328
11908 * fold-const.c (maskable_range_p): New function.
11909 (build_range_check): Generate bittests if possible.
11911 2017-06-13 Martin Liska <mliska@suse.cz>
11913 * gimple-pretty-print.c (dump_probability): Add new argument.
11914 (dump_edge_probability): Dump both probability and count.
11915 (dump_gimple_label): Likewise.
11916 (dump_gimple_bb_header): Likewise.
11918 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11921 * config/avr/avr-devices.c: Fix indentation.
11922 * config/avr/gen-avr-mmcu-specs.c: Dito.
11924 2017-06-13 Richard Biener <rguenther@suse.de>
11926 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11927 instead get vector type from stmt_info.
11928 (vectorizable_reduction): Adjust. Remove dead code.
11930 2017-06-13 Richard Biener <rguenther@suse.de>
11932 PR middle-end/81065
11933 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11934 case of C * (x * C2 + C3).
11935 (fold_addr_of_array_ref_difference): Properly fold index difference.
11937 2017-06-12 David S. Miller <davem@davemloft.net>
11940 * config/sparc/sparc.md (return expander): Emit frame blockage if
11941 function uses alloca.
11943 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11945 * combine.c (make_field_assignment): Check len rather than the mode
11946 precision when calling force_to_mode.
11948 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11950 Support multilibs and devices that see flash in RAM address range.
11953 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11954 (avr_mcu_t) <flash_pm_offset>: New field.
11955 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11956 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11957 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11958 (AVR_TINY_PM_OFFSET): Remove macro.
11959 * config/avr/avr.opt (-mshort-calls): New option.
11960 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11961 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11962 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11963 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11964 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11965 instead of avr_arch->have_jmp_call.
11966 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11967 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11968 avr_arch->flash_pm_offset to define.
11969 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11970 new field flash_pm_offset. Add entry for avrxmega3.
11971 (avr_texinfo): Add entry for avrxmega3.
11972 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11973 attiny212, attiny214,
11974 attiny412, attiny414, attiny416, attiny417,
11975 attiny814, attiny816, attiny817,
11976 attiny1614, attiny1616, attiny1617,
11977 attiny3214, attiny3216, attiny3217.
11978 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11979 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11980 (avr_print_operand_address) [AVR_TINY]: Same.
11981 (avr_asm_init_sections) <readonly_data_section>: Only patch
11982 callback if avr_arch->flash_pm_offset = 0.
11983 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11984 for rodata if avr_arch->flash_pm_offset != 0.
11985 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11986 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11987 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11988 (m_options): Append opt_rcall.
11989 (m_dirnames): Append dir_rcall.
11990 * config/avr/t-multilib: Regenerate.
11992 * configure.ac [target=avr]: Check whether avrxmega3 default
11993 linker description file works as needed.
11994 * configure: Regenerate.
11995 * doc/avr-mmcu.texi: Regenerate.
11996 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11997 <__AVR_ARCH__>: Document avrxmega3 and 103.
11998 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11999 <__AVR_SHORT_CALLS__>: Document it.
12000 <__AVR_PM_BASE_ADDRESS__>: Document it.
12001 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12002 (AVR Variable Attributes) <progmem>: Document this is
12003 not needed for avrxmega3.
12004 (AVR Named Address Spaces) <__flash>: Dito.
12006 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12008 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12010 2017-06-12 Doug Rupp <rupp@adacore.com>
12012 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12013 Append vxworks-stdint.h to the tm_file list.
12014 * config/vxworks-stdint.h: New file.
12016 2017-06-12 Martin Liska <mliska@suse.cz>
12018 PR tree-optimization/81041
12019 * tree-profile.c (gimple_gen_ic_func_profiler):
12020 Create an extra BB in profile-generate
12021 (gimple_gen_time_profiler): Likewise.
12023 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12025 PR tree-optimization/81003
12026 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12027 (update_range_test): Use it instead of force_gimple_operand_gsi.
12029 2017-06-12 Richard Biener <rguenther@suse.de>
12031 PR tree-optimization/81053
12032 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12033 with backedge value not defined in loop. Simplify def stmt
12036 2017-06-11 Tom de Vries <tom@codesourcery.com>
12039 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12041 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12042 nvptx_cannot_force_const_mem.
12044 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12046 * opts.c (finish_options): Move test for flag_split_stack after
12047 it has been initialized.
12049 2017-06-11 Jason Merrill <jason@redhat.com>
12051 * tree.h (id_equal): New.
12052 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12053 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12054 instead of strcmp of IDENTIFIER_POINTER.
12056 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12058 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12059 (mark_all_inlined_calls_cdtor): Fix formating.
12060 (inline_transform): Rescale profile before inlining.
12062 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12064 * cgraph.h (cgraph_edge::clone): Update prototype.
12065 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12066 (cgraph_node::create_clone): Update.
12067 (cgraph_node::create_version_clone): Update.
12068 * tree-inline.c (copy_bb): Update.
12069 (expand_call_inline): Update.
12071 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12073 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12074 factored out from ...
12075 (rs6000_emit_prologue): ... here.
12077 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12079 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12080 factored out from ...
12081 (rs6000_emit_prologue): ... here.
12083 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12085 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12087 (handle_missing_profiles): Fix computation of tp_first_run.
12088 (counts_to_freqs): Do not touch freqs when count is 0.
12090 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12092 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12095 2017-06-10 Tom de Vries <tom@codesourcery.com>
12097 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12098 attributes): Document signal effective target.
12100 2017-06-10 Tom de Vries <tom@codesourcery.com>
12102 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12103 Document effective target stack_size.
12105 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12107 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12108 to the edit_context if they can be auto-applied.
12110 2017-06-9 Ian Lance Taylor <iant@golang.org>
12112 * opts.c (finish_options): If -fsplit-stack, disable implicit
12113 -forder-blocks-and-partition.
12114 * doc/invoke.texi (Optimize Options): Document that when using
12115 -fsplit-stack -forder-blocks-and-partition is not implicitly
12118 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12120 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12121 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12122 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12123 * builtins.def (abort, trap, unreachable): Declare cold.
12124 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12125 * tree-core.h (ECF_COLD): New.
12126 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12127 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12129 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12131 * predict.c (unlikely_executed_stmt_p): Cleanup.
12133 2017-06-09 Richard Biener <rguenther@suse.de>
12135 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12136 model if the ref is always written to.
12138 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12140 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12142 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12144 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12147 2017-06-09 Tom de Vries <tom@codesourcery.com>
12150 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12151 "target cannot support label values" when encountering LABEL_REF.
12153 2017-06-09 Martin Liska <mliska@suse.cz>
12155 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12156 (gimple_gen_ic_func_profiler): Emit direct comparison
12157 of __gcov_indirect_call_callee with NULL.
12158 (gimple_gen_time_profiler): Change probability from
12159 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12161 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12163 * profile.c (edge_gcov_counts): Turn to pointer.
12164 (compute_branch_probabilities, compute_branch_probabilities): Update.
12165 (branch_prob): Do not clear edge_gcov_count.
12166 * profile.h (edge_gcov_counts): Turn to pointer.
12167 (edge_gcov_count): Update.
12169 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12171 * gimple.h (gimple_check_failed): Mark cold.
12173 2017-06-09 Richard Biener <rguenther@suse.de>
12175 PR tree-optimization/66623
12176 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12177 refactor check_reduction into two parts, properly computing
12178 whether we have to check reduction validity for outer loop
12181 2017-06-09 Richard Biener <rguenther@suse.de>
12183 PR tree-optimization/79483
12184 * graphite-scop-detection.c (order): New global.
12185 (get_order): Compute bb to order mapping that satisfies code
12186 generation constraints.
12187 (cmp_pbbs): New helper.
12188 (build_scops): Start domwalk at entry block, sort generated
12191 2017-06-09 Richard Biener <rguenther@suse.de>
12193 PR middle-end/81007
12194 * ipa-polymorphic-call.c
12195 (ipa_polymorphic_call_context::restrict_to_inner_class):
12196 Skip FIELD_DECLs with error_mark_node type.
12197 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12200 2017-06-09 Martin Liska <mliska@suse.cz>
12202 * predict.c (struct branch_predictor): New struct.
12203 (test_prediction_value_range): New test.
12204 (predict_c_tests): New function.
12205 * selftest-run-tests.c (selftest::run_tests): Run the function.
12206 * selftest.h: Declare new tests.
12208 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12211 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12212 gen_add3_insn did not fail.
12213 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12214 r0, construct that number in a temporary reg and add that reg to r0.
12215 If asked to put the result in r0 as well, fail.
12217 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12219 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12220 for early expansion of vec_eqv.
12222 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12224 PR middle-end/81005
12225 * ubsan.c (instrument_null): Avoid pointless code temporary.
12226 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12228 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12232 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12235 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12238 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12240 * predict.c (unlikely_executed_edge_p): Move ahead.
12241 (probably_never_executed_edge_p): Use it.
12243 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12245 PR middle-end/79988
12246 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12247 gimple_call_builtin_p call.
12249 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12251 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12252 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12253 rtl_check_failed_type2, rtl_check_failed_code1,
12254 rtl_check_failed_code2, rtl_check_failed_code_mode,
12255 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12256 rtvec_check_failed_bounds, rtl_check_failed_flag,
12257 _fatal_insn_not_found, _fatal_insn): Likewise.
12258 * tree.h (tree_contains_struct_check_failed,
12259 tree_check_failed, tree_not_check_failed,
12260 tree_class_check_failed, tree_range_check_failed,
12261 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12262 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12263 tree_operand_check_failed, omp_clause_check_failed,
12264 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12267 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12269 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12270 flag_branch_probabilities.
12271 * ipa-inline.c (edge_badness): Likewise.
12272 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12273 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12274 * predict.c (maybe_hot_frequency_p): Likewise.
12275 (probably_never_executed): Likewise.
12276 * sched-ebb.c (schedule_ebbs): Likewise.
12277 * sched-rgn.c (find_single_block_region): Likewise.
12278 * tracer.c (tail_duplicate): Likewise.
12280 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12282 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12283 longer requires x_flag_profile_use.
12285 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12287 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12288 instead of flag_reorder_blocks_and_partition.
12289 * dbxout.c (dbxout_function_end): Likewise.
12290 * dwarf2out.c (gen_subprogram_die): Likewise.
12291 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12292 * hw-doloop.c (reorg_loops): Likewise.
12293 * varasm.c (assemble_start_function,
12294 assemble_end_function): Likewise.
12295 (decide_function_section): Do not check for
12296 flag_reorder_blocks_and_partition.
12298 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12300 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12302 (chkp_get_hard_register_fake_addr_expr): Ditto.
12303 (chkp_build_addr_expr): Add check for hard reg case.
12304 (chkp_parse_array_and_component_ref): Ditto.
12305 (chkp_find_bounds_1): Ditto.
12306 (chkp_process_stmt): Don't generate bounds store for
12309 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12311 * predict.c (maybe_hot_bb_p): Do not check profile status.
12312 (maybe_hot_edge_p): Likewise.
12313 (probably_never_executed): Check for zero counts even if profile
12315 (unlikely_executed_edge_p): New function.
12316 (unlikely_executed_stmt_p): New function.
12317 (unlikely_executed_bb_p): New function.
12318 (set_even_probabilities): Use unlikely predicates.
12319 (combine_predictions_for_bb): Likewise.
12320 (predict_paths_for_bb): Likewise.
12321 (predict_paths_leading_to_edge): Likewise.
12322 (determine_unlikely_bbs): New function.
12323 (estimate_bb_frequencies): Use it.
12324 (compute_function_frequency): Use zero counts even if profile is
12326 * profile-count.h: Fix typo.
12328 2017-08-08 Julia Koval <julia.koval@intel.com>
12330 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12331 _mm512_mask_cvtsepi16_storeu_epi8,
12332 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12333 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12334 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12335 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12336 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12337 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12338 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12339 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12340 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12341 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12342 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12343 __builtin_ia32_pmovuswb256mem_mask,
12344 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12345 __builtin_ia32_pmovwb512mem_mask): New builtins.
12347 2017-08-08 Julia Koval <julia.koval@intel.com>
12349 PR target/73350,80862
12350 * config/i386/subst.md (round): Fix round pattern.
12351 * config/i386/i386.c (ix86_erase_embedded_rounding):
12352 Fix erasing rounding for the fixed pattern.
12354 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12356 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12358 2017-06-08 Martin Liska <mliska@suse.cz>
12360 PR gcov-profile/80911
12361 * gcov.c (block_info::block_info): New constructor.
12363 2017-06-07 Carl Love <cel@us.ibm.com>
12365 * config/rs6000/rs6000-c: The return type of the following
12366 built-in functions was implemented as int not long long. Fix sign
12367 of return value for the unsigned version of vec_mulo and vec_mule.
12368 vector unsigned long long vec_bperm (vector unsigned long long,
12369 vector unsigned char)
12370 vector signed long long vec_mule (vector signed int,
12372 vector unsigned long long vec_mule (vector unsigned int,
12373 vector unsigned int)
12374 vector signed long long vec_mulo (vector signed int,
12376 vector unsigned long long vec_mulo (vector unsigned int,
12377 vector unsigned int)
12378 * doc/extend.texi: Fix the documentation for the built-in
12381 2017-06-07 Carl Love <cel@us.ibm.com>
12384 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12387 2017-06-07 Carl Love <cel@us.ibm.com>
12389 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12390 support, Generate doublehv for signed int/float for BE case only.
12392 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12394 * doc/invoke.texi (mcx16): Rewrite.
12396 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12398 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12399 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12400 *mov<mode>_softfloat, and an anonymous splitter): Use
12401 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12403 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12405 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12407 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12408 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12409 (rs6000_debug_reg_global): Adjust.
12410 (rs6000_init_hard_regno_mode_ok): Adjust.
12411 (rs6000_dbx_register_number): Adjust.
12412 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12413 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12414 Remove SPE_ACC and SPEFSCR.
12415 (REG_ALLOC_ORDER): Ditto.
12416 (FRAME_POINTER_REGNUM): Change to 111.
12417 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12418 (REG_CLASS_NAMES): Ditto.
12419 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12420 (REGISTER_NAMES): Ditto.
12421 (ADDITIONAL_REG_NAMES): Ditto.
12422 (rs6000_reg_names): Ditto.
12423 * config/rs6000/rs6000.md: Renumber some register number
12426 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12428 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12430 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12431 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12433 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12434 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12435 Delete the SPE high registers.
12436 (REG_ALLOC_ORDER): Ditto.
12437 (enum reg_class): Remove SPE_HIGH_REGS.
12438 (REG_CLASS_NAMES): Ditto.
12439 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12440 (REGISTER_NAMES): Ditto.
12441 (rs6000_reg_names): Ditto.
12442 * doc/tm.texi.in: Remove SPE as example.
12443 * doc/tm.texi: Regenerate.
12445 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12447 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12448 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12449 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12450 * config/rs6000/rs6000.md (type): Remove "brinc".
12452 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12454 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12455 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12456 * config/rs6000/linuxspe.h: Delete file.
12457 * config/rs6000/rs6000.md: Don't include spe.md.
12458 * config/rs6000/spe.h: Delete file.
12459 * config/rs6000/spe.md: Delete file.
12460 * config/rs6000/t-rs6000: Remove spe.md.
12462 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12464 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12465 (reg_or_none500mem_operand): Delete.
12466 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12467 instead of reg_or_none500mem_operand.
12469 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12471 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12472 handling of SPE flags.
12473 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12475 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12477 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12479 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12480 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12481 paired_divv2sf3): Similar.
12482 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12483 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12484 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12486 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12487 Rename the paired_* instruction patterns.
12488 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12490 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12491 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12492 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12493 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12494 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12495 PAIRED_VECTOR_MODE.
12496 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12497 (spe_func_has_64bit_regs_p): Delete.
12498 (spe_expand_predicate_builtin): Delete.
12499 (spe_expand_evsel_builtin): Delete.
12500 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12501 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12502 (invalid_e500_subreg): Delete.
12503 (rs6000_legitimize_address): Always force_reg op2 as well, for
12504 paired single memory accesses.
12505 (rs6000_member_type_forces_blk): Delete.
12506 (rs6000_spe_function_arg): Delete.
12507 (rs6000_expand_unop_builtin): Delete SPE handling.
12508 (rs6000_expand_binop_builtin): Ditto.
12509 (spe_expand_stv_builtin): Delete.
12510 (bdesc_2arg_spe): Delete.
12511 (spe_expand_builtin): Delete.
12512 (spe_expand_predicate_builtin): Delete.
12513 (spe_expand_evsel_builtin): Delete.
12514 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12515 (spe_init_builtins): Delete.
12516 (spe_func_has_64bit_regs_p): Delete.
12517 (savres_routine_name): Delete "info" parameter. Adjust callers.
12518 (rs6000_emit_stack_reset): Ditto.
12519 (rs6000_dwarf_register_span): Delete.
12520 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12521 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12522 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12524 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12526 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12527 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12528 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12529 mulv2sf3, divv2sf3): Delete expanders.
12531 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12533 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12535 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12537 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12538 * config/rs6000/rs6000.c: Ditto.
12540 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12542 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12543 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12544 comparison_operator.
12546 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12548 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12549 * config/rs6000/rs6000.opt: Ditto.
12550 * config/rs6000/t-rtems: Ditto.
12552 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12554 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12555 TARGET_E500_SINGLE by 0, simplify.
12556 * config/rs6000/rs6000.c: Ditto.
12557 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12558 (spe_build_register_parallel): Delete.
12559 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12560 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12561 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12562 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12563 (E500_CONVERT): Delete.
12564 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12566 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12568 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12569 * config/rs6000/dfp.md: Ditto.
12570 (negdd2, *negdd2_fpr): Merge.
12571 (absdd2, *absdd2_fpr): Merge.
12572 (negtd2, *negtd2_fpr): Merge.
12573 (abstd2, *abstd2_fpr): Merge.
12574 * config/rs6000/e500.h: Delete file.
12575 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12576 TARGET_FPRS by 1 and simplify.
12577 * config/rs6000/rs6000-c.c: Ditto.
12578 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12579 TARGET_DF_SPE by 0.
12580 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12582 * config/rs6000/rs6000.md: Ditto.
12583 (floatdidf2, *floatdidf2_fpr): Merge.
12584 (move_from_CR_gt_bit): Delete.
12585 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12586 (E500_CR_IOR_COMPARE): Delete.
12587 (All patterns that require !TARGET_FPRS): Delete.
12588 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12590 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12592 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12594 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12596 * graphds.c (add_edge): Intitialize edge's attached data.
12597 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12598 pointer parameter. Call pointed function on each edge during
12599 graph traversing. Skip traversing the edge when the function
12601 (graphds_dfs, graphds_scc): Ditto.
12602 (for_each_edge): New parameter. Pass the new parameter to callback
12604 * graphds.h (skip_edge_callback): New function pointer type.
12605 (graphds_dfs, graphds_scc): New function pointer parameter.
12606 (graphds_edge_callback, for_each_edge): New parameter.
12608 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12610 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12611 out code checking if runtime alias check is possible to below ...
12612 Call the new function.
12613 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12614 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12616 2017-06-07 Marek Polacek <polacek@redhat.com>
12619 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12620 TYPE_OVERFLOW_WRAPS check.
12622 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12624 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12625 if versioning is required.
12626 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12627 peeling with the check for versioning.
12629 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12631 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12632 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12633 Set true to new parameter if new ssa variable is defined.
12634 (vect_gen_vector_loop_niters): Refactor. Set range information
12635 for the new vector loop bound variable.
12636 (vect_do_peeling): Ditto.
12638 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12640 * tree-affine.c (ssa.h): Include header file.
12641 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12642 has wrapping overflow behavior.
12644 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12646 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12648 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12650 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12651 (tree_to_aff_combination): ... here.
12653 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12655 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12656 reg_pressure model function.
12657 (ivopts_global_cost_for_size): Delete.
12658 (determine_set_costs, iv_ca_recount_cost): Call new model function
12659 ivopts_estimate_reg_pressure.
12661 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12663 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12664 expensive than udiv. Remove floating point cases from mod.
12666 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12668 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12669 Increase idiv cost.
12671 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12673 * config/aarch64/aarch64.md
12674 (copysignsf3): Fix mask generation.
12676 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12678 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12680 (class dump_manager): Add register_dumps method.
12681 * dumpfile.c: Include langhooks.h.
12682 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12683 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12684 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12685 (dump_manager::dump_register): Start with 512 entries instead of 32.
12686 (dump_manager::register_dumps): New method.
12687 * toplev.c (general_init): Instead of invoking register_dumps
12688 langhook, invoke register_dumps method on the dump manager.
12689 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12692 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12694 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12695 Say that # requires an associated define_split to exist, and that
12696 the define_split must be suitable for use after register allocation.
12698 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12700 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12701 (compute_outgoing_frequencies): Also initialize zero counts.
12702 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12703 around loops; preserve more of profile when nothing changes.
12705 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12707 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12709 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12710 * config/arm/arm-cpu-cdata.h: Regenerate.
12711 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12712 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12713 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12714 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12715 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12717 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12719 * config/arm/t-rmprofile: Likewise.
12720 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12722 2017-06-06 David S. Miller <davem@davemloft.net>
12725 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12726 blockage if function uses alloca.
12728 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12730 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12731 New "uid" fields to hold pretty-print IDs of group and ref.
12732 Memory references are now identified as <group_id>:<ref_id>
12733 instead of using [random] addresses.
12734 (dump_mem_details): Simplify, no functional change.
12735 (dump_mem_ref): Simplify and make output more concise.
12736 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12737 (find_or_create_group): Initialize group uid.
12738 (record_ref): Initialize ref uid. Improve debug output.
12739 (prune_group_by_reuse, should_issue_prefetch_p,)
12740 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12741 (mark_nontemporal_store, determine_loop_nest_reuse):
12742 Improve debug output.
12744 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12746 * dbgcnt.def (prefetch): New debug counter.
12747 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12748 (schedule_prefetches): Stop issueing prefetches if debug counter
12751 2017-06-06 Tom de Vries <tom@codesourcery.com>
12753 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12754 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12756 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12758 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12759 Use aarch64_reg_or_zero predicate for operand 4.
12760 (aarch64_compare_and_swap<mode> define_insn_and_split):
12761 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12762 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12764 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12766 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12767 (arm_compute_save_core_reg_mask): This.
12768 (thumb1_compute_save_reg_mask): Rename into ...
12769 (thumb1_compute_save_core_reg_mask): This.
12770 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12771 (arm_compute_frame_layout): Likewise.
12773 2017-06-06 Richard Biener <rguenther@suse.de>
12775 PR tree-optimization/80974
12776 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12777 keep or clear leaders SSA info.
12779 2017-06-06 Tom de Vries <tom@codesourcery.com>
12781 * config/nvptx/nvptx.c (split_mode_p): New function.
12782 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12784 2017-06-06 Tom de Vries <tom@codesourcery.com>
12786 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12788 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12791 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12794 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12796 * shrink-wrap.c (handle_simple_exit): Update profile.
12797 (try_shrink_wrapping): Upate profile.
12799 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12801 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12802 (tree_guess_outgoing_edge_probabilities): New.
12803 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12804 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12806 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12808 * ipa-split.c (split_function): Initialize return bb profile.
12810 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12812 * profile.c (compute_branch_probabilities): Also initialize
12813 EXIT_BLOCK profile.
12815 2017-06-06 Richard Biener <rguenther@suse.de>
12817 PR tree-optimization/80928
12818 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12819 (vect_analyze_loop_operations): Properly guard analysis for
12821 (vect_transform_loop): Likewise.
12822 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12823 (vect_model_induction_cost): Do not cost for pure SLP.
12824 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12825 of induction in inner loop vectorization.
12826 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12827 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12828 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12830 (vect_analyze_slp_cost_1): Cost induction.
12831 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12832 (vect_get_slp_vect_defs): Likewise.
12833 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12834 (vect_transform_stmt): Handle SLP reductions.
12835 * tree-vectorizer.h (vectorizable_induction): Adjust.
12837 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12839 * config/rs6000/rs6000.c (make_resolver_func): Update
12840 init_lowered_empty_function call.
12842 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12844 * doc/invoke.texi: Document the -fprofile-abs-path option.
12845 * common.opt (fprofile-abs-path): New option.
12846 * gcov-io.h (gcov_write_filename): Declare.
12847 * gcov-io.c (gcov_write_filename): New function.
12848 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12849 * profile.c (output_location): Likewise.
12851 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12853 * shring-wrap.c: Revert accidental commit.
12855 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12857 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12859 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12861 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12863 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12864 profile in callgraph edge.
12865 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12866 (apply_scale): Likewise.
12867 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12868 Also scale profile when inlining function with zero profile.
12869 (initialize_cfun): Update exit block profile even when it is zero.
12870 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12871 when profile is read.
12873 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12875 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12876 (CLONE_*): New constants to define the processors we can generate
12877 code for with the target_clone attribute.
12878 (rs6000_clone_map): New array to identify which clone processors
12879 the current program is running on.
12880 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12881 target_clone attribute.
12882 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12883 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12884 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12885 (cpu_expand_builtin): Add support for target_clone attribute.
12886 (rs6000_valid_attribute_p): Allow "default" attribute.
12887 (get_decl_name): New debug function to simplify printing the
12888 current function name in debugging statements.
12889 (rs6000_clone_priority): New functions to support the target_clone
12890 attribute, and be able to generate code to switch between ISA 2.05
12891 through ISA 3.0 (power6 through power9).
12892 (rs6000_compare_version_priority): Likewise.
12893 (rs6000_get_function_versions_dispatcher): Likewise.
12894 (make_resolver_func): Likewise.
12895 (add_condition_to_bb): Likewise.
12896 (dispatch_function_versions): Likewise.
12897 (rs6000_generate_version_dispatcher_body): Likewise.
12898 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12899 (fusion_gpr_load_p): Fix a spacing issue.
12900 * doc/extend.texi (Common Function Attributes): Document that the
12901 PowerPC supports the target_clone attribute.
12903 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12905 * config/arm/arm.h: explain F symbol found in description of ARM
12906 register allocation in its legend.
12908 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12910 * config/mips/frame-header-opt.c: Include profile-count.h.
12911 * config/riscv/riscv.c: Include profile-count.h
12913 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12915 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12917 (sm_set_flag_if_changed): Add bbs field.
12918 (execute_sm_if_changed_flag_set): Pass BBS.
12919 (execute_sm): Update.
12921 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12923 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12926 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12928 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12929 (peephole2): New peephole2 to emit the above.
12930 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12932 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12934 * config/aarch64/aarch64.c (define_peephole2 above
12935 *sub_<shift>_<mode>): New peephole.
12937 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12939 * config/i386/i386.c (make_resolver_func): Update.
12940 * Makefile.in: Add profile-count.h and profile-count.o
12941 * auto-profile.c (afdo_indirect_call): Update to new API.
12942 (afdo_set_bb_count): Update.
12943 (afdo_propagate_edge): Update.
12944 (afdo_propagate_circuit): Update.
12945 (afdo_calculate_branch_prob): Update.
12946 (afdo_annotate_cfg): Update.
12947 * basic-block.h: Include profile-count.h
12948 (struct edge_def): Turn count to profile_count.
12949 (struct basic_block_def): Likewie.
12950 (REG_BR_PROB_BASE): Move to profile-count.h
12951 (RDIV): Move to profile-count.h
12952 * bb-reorder.c (max_entry_count): Turn to profile_count.
12953 (find_traces): Update.
12954 (rotate_loop):Update.
12955 (connect_traces):Update.
12956 (sanitize_hot_paths):Update.
12957 * bt-load.c (migrate_btr_defs): Update.
12958 * cfg.c (RDIV): Remove.
12959 (init_flow): Use alloc_block.
12960 (alloc_block): Uninitialize count.
12961 (unchecked_make_edge): Uninitialize count.
12962 (check_bb_profile): Update.
12963 (dump_edge_info): Update.
12964 (dump_bb_info): Update.
12965 (update_bb_profile_for_threading): Update.
12966 (scale_bbs_frequencies_int): Update.
12967 (scale_bbs_frequencies_gcov_type): Update.
12968 (scale_bbs_frequencies_profile_count): New.
12969 * cfg.h (update_bb_profile_for_threading): Update.
12970 (scale_bbs_frequencies_profile_count): Declare.
12971 * cfgbuild.c (compute_outgoing_frequencies): Update.
12972 (find_many_sub_basic_blocks): Update.
12973 * cfgcleanup.c (try_forward_edges): Update.
12974 (try_crossjump_to_edge): Update.
12975 * cfgexpand.c (expand_gimple_tailcall): Update.
12976 (construct_exit_block): Update.
12977 * cfghooks.c (verify_flow_info): Update.
12978 (dump_bb_for_graph): Update.
12979 (split_edge): Update.
12980 (make_forwarder_block): Update.
12981 (duplicate_block): Update.
12982 (account_profile_record): Update.
12983 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12984 (get_estimated_loop_iterations): Update.
12985 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12986 (single_likely_exit): Update.
12987 * cfgloopmanip.c (scale_loop_profile): Update.
12989 (set_zero_probability): Update.
12990 (lv_adjust_loop_entry_edge): Update.
12991 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12992 (purge_dead_edges): Update.
12993 (rtl_account_profile_record): Update.
12994 * cgraph.c (cgraph_node::create): Uninitialize count.
12995 (symbol_table::create_edge): Uninitialize count.
12996 (cgraph_update_edges_for_call_stmt_node): Update.
12997 (cgraph_edge::dump_edge_flags): Update.
12998 (cgraph_node::dump): Update.
12999 (cgraph_edge::maybe_hot_p): Update.
13000 * cgraph.h: Include profile-count.h
13001 (create_clone), create_edge, create_indirect_edge): Update.
13002 (cgraph_node): Turn count to profile_count.
13003 (cgraph_edge0: Likewise.
13004 (make_speculative, clone): Update.
13005 (create_edge): Update.
13006 (init_lowered_empty_function): Update.
13007 * cgraphclones.c (cgraph_edge::clone): Update.
13008 (duplicate_thunk_for_node): Update.
13009 (cgraph_node::create_clone): Update.
13010 * cgraphunit.c (cgraph_node::analyze): Update.
13011 (cgraph_node::expand_thunk): Update.
13012 * final.c (dump_basic_block_info): Update.
13013 * gimple-streamer-in.c (input_bb): Update.
13014 * gimple-streamer-out.c (output_bb): Update.
13015 * graphite.c (print_global_statistics): Update.
13016 (print_graphite_scop_statistics): Update.
13017 * hsa-brig.c: Include basic-block.h.
13018 * hsa-dump.c: Include basic-block.h.
13019 * hsa-gen.c (T sum_slice): Update.
13020 (convert_switch_statements):Update.
13021 * hsa-regalloc.c: Include basic-block.h.
13022 * ipa-chkp.c (chkp_produce_thunks): Update.
13023 * ipa-cp.c (struct caller_statistics): Update.
13024 (init_caller_stats): Update.
13025 (gather_caller_stats): Update.
13026 (ipcp_cloning_candidate_p): Update.
13027 (good_cloning_opportunity_p): Update.
13028 (get_info_about_necessary_edges): Update.
13029 (dump_profile_updates): Update.
13030 (update_profiling_info): Update.
13031 (update_specialized_profile): Update.
13032 (perhaps_add_new_callers): Update.
13033 (decide_about_value): Update.
13034 (ipa_cp_c_finalize): Update.
13035 * ipa-devirt.c (struct odr_type_warn_count): Update.
13036 (struct decl_warn_count): Update.
13037 (struct final_warning_record): Update.
13038 (possible_polymorphic_call_targets): Update.
13039 (ipa_devirt): Update.
13040 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13041 * ipa-icf.c (sem_function::merge): Update.
13042 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13043 * ipa-inline.c (compute_uninlined_call_time): Update.
13044 (compute_inlined_call_time): Update.
13045 (want_inline_small_function_p): Update.
13046 (want_inline_self_recursive_call_p): Update.
13047 (edge_badness): Update.
13048 (lookup_recursive_calls): Update.
13049 (recursive_inlining): Update.
13050 (inline_small_functions): Update.
13051 (dump_overall_stats): Update.
13052 (dump_inline_stats): Update.
13053 * ipa-profile.c (ipa_profile_generate_summary): Update.
13054 (ipa_propagate_frequency): Update.
13055 (ipa_profile): Update.
13056 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13057 * ipa-utils.c (ipa_merge_profiles): Update.
13058 * loop-doloop.c (doloop_modify): Update.
13059 * loop-unroll.c (report_unroll): Update.
13060 (unroll_loop_runtime_iterations): Update.
13061 * lto-cgraph.c (lto_output_edge): Update.
13062 (lto_output_node): Update.
13063 (input_node): Update.
13064 (input_edge): Update.
13065 (merge_profile_summaries): Update.
13066 * lto-streamer-in.c (input_cfg): Update.
13067 * lto-streamer-out.c (output_cfg): Update.
13068 * mcf.c (create_fixup_graph): Update.
13069 (adjust_cfg_counts): Update.
13070 (sum_edge_counts): Update.
13071 * modulo-sched.c (sms_schedule): Update.
13072 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13073 * predict.c (maybe_hot_count_p): Update.
13074 (probably_never_executed): Update.
13075 (dump_prediction): Update.
13076 (combine_predictions_for_bb): Update.
13077 (propagate_freq): Update.
13078 (handle_missing_profiles): Update.
13079 (counts_to_freqs): Update.
13080 (rebuild_frequencies): Update.
13081 (force_edge_cold): Update.
13082 * predict.h: Include profile-count.h
13083 (maybe_hot_count_p, counts_to_freqs): UPdate.
13084 * print-rtl-function.c: Do not include cfg.h
13085 * print-rtl.c: Include basic-block.h
13086 * profile-count.c: New file.
13087 * profile-count.h: New file.
13088 * profile.c (is_edge_inconsistent): Update.
13089 (correct_negative_edge_counts): Update.
13090 (is_inconsistent): Update.
13091 (set_bb_counts): Update.
13092 (read_profile_edge_counts): Update.
13093 (compute_frequency_overlap): Update.
13094 (compute_branch_probabilities): Update; Initialize and deinitialize
13096 (branch_prob): Update.
13097 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13098 (edge_gcov_count): New.
13099 (bb_gcov_count): New.
13100 * shrink-wrap.c (try_shrink_wrapping): Update.
13101 * tracer.c (better_p): Update.
13102 * trans-mem.c (expand_transaction): Update.
13103 (ipa_tm_insert_irr_call): Update.
13104 (ipa_tm_insert_gettmclone_call): Update.
13105 * tree-call-cdce.c: Update.
13106 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13107 (gimple_duplicate_sese_tail): Update.
13108 (gimple_account_profile_record): Update.
13109 (execute_fixup_cfg): Update.
13110 * tree-inline.c (copy_bb): Update.
13111 (copy_edges_for_bb): Update.
13112 (initialize_cfun): Update.
13113 (freqs_to_counts): Update.
13114 (copy_cfg_body): Update.
13115 (expand_call_inline): Update.
13116 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13117 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13118 (try_unroll_loop_completely): Update.
13119 (try_peel_loop): Update.
13120 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13121 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13122 * tree-ssa-loop-split.c (connect_loops): Update.
13123 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13124 * tree-ssa-reassoc.c (branch_fixup): Update.
13125 * tree-ssa-tail-merge.c (replace_block_by): Update.
13126 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13127 (compute_path_counts): Update.
13128 (update_profile): Update.
13129 (recompute_probabilities): Update.
13130 (update_joiner_offpath_counts): Update.
13131 (estimated_freqs_path): Update.
13132 (freqs_to_counts_path): Update.
13133 (clear_counts_path): Update.
13134 (ssa_fix_duplicate_block_edges): Update.
13135 (duplicate_thread_path): Update.
13136 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13137 (struct switch_conv_info): Update.
13138 * tree-tailcall.c (decrease_profile): Update.
13139 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13140 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13141 * value-prof.c (check_counter): Update.
13142 (gimple_divmod_fixed_value): Update.
13143 (gimple_mod_pow2): Update.
13144 (gimple_mod_subtract): Update.
13145 (gimple_ic_transform): Update.
13146 (gimple_stringop_fixed_value): Update.
13147 * value-prof.h (gimple_ic): Update.
13149 2017-06-02 Carl Love <cel@us.ibm.com>
13151 * config/rs6000/rs6000-c: Add support for built-in functions
13152 vector double vec_doublee (vector signed int);
13153 vector double vec_doublee (vector unsigned int);
13154 vector double vec_doublee (vector float);
13155 vector double vec_doubleh (vector signed int);
13156 vector double vec_doubleh (vector unsigned int);
13157 vector double vec_doubleh (vector float);
13158 vector double vec_doublel (vector signed int);
13159 vector double vec_doublel (vector unsigned int);
13160 vector double vec_doublel (vector float);
13161 vector double vec_doubleo (vector signed int);
13162 vector double vec_doubleo (vector unsigned int);
13163 vector double vec_doubleo (vector float);.
13164 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13165 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13167 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13168 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13169 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13171 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13172 vec_doublel, vec_doubleh.
13173 * doc/extend.texi: Update the built-in documentation file for the
13174 new built-in functions.
13176 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13179 * ipa-inline-analysis.c (free_growth_caches): Set
13180 edge_removal_hook_holder to NULL after removing it.
13182 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13184 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13185 comparision with zero.
13187 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13188 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13189 for early expansion of vec_min and vec_max builtins.
13190 (builtin_function_type): Add min/max unsigned variants to those
13191 identified as having unsigned arguments.
13193 2017-06-02 Olivier Hainque <hainque@adacore.com>
13195 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13197 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13199 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13200 Use VALL_F16 iterator rather than VALL.
13202 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13204 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13205 Emit CBNZ inside loop when doing a strong exchange and comparing
13206 against zero. Generate the CC flags after the loop.
13208 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13210 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13211 (dl_section_ref): New.
13212 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13213 On AIX, append an expression to subtract the size of the
13214 section length to dl_section_ref.
13216 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13218 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13219 for early expansion of vector absolute builtins.
13221 2017-06-02 Richard Biener <rguenther@suse.de>
13223 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13224 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13226 2017-06-02 Richard Biener <rguenther@suse.de>
13228 PR tree-optimization/80948
13229 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13230 stmt order as well.
13232 2017-06-02 Richard Biener <rguenther@suse.de>
13234 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13236 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13237 for inductions as relevant.
13239 2017-06-02 Richard Biener <rguenther@suse.de>
13241 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13242 (vectorizable_induction): ... this. Remove dead code.
13244 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13246 * builtins. (expand_builtin_alloca): Remove second parameter and
13247 infer its value from the first parameter instead.
13248 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13250 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13252 PR rtl-optimization/80903
13253 * loop-doloop.c (add_test): Unshare sequence.
13255 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13257 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13259 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13261 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13263 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13264 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13265 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13266 (xlogue_layout::get_stub_rtx): Make static.
13267 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13268 (xlogue_layout::compute_stub_managed_regs): Rename to...
13269 (xlogue_layout::count_stub_managed_regs): ...this.
13270 (xlogue_layout::is_stub_managed_reg): New function.
13271 (xlogue_layout::m_stub_names): Rename to...
13272 (xlogue_layout::s_stub_names): ...this, make static.
13273 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13274 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13275 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13276 xlogue_layout::s_stub_names): Instantiate statics.
13277 (stub_managed_regs): Remove.
13278 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13279 (disable_call_ms2sysv_xlogues): Rename to...
13280 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13281 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13283 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13284 change after reload_completed.
13285 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13287 (ix86_expand_prologue): Likewise.
13288 (ix86_expand_epilogue): Likewise.
13289 (ix86_expand_split_stack_prologue): Likewise.
13290 (ix86_compute_frame_layout): Remove frame parameter ...
13291 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13292 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13294 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13295 (ix86_frame): Move from here ...
13296 * config/i386/i386.h (ix86_frame): ... to here.
13297 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13298 complete ix86_frame data structure instead. Remove some_ld_name.
13300 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13302 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13303 symbols that hold a DECL_VALUE_EXPR.
13305 2017-06-01 Martin Jambor <mjambor@suse.cz>
13307 PR tree-optimization/80898
13308 * tree-sra.c (process_subtree_disqualification): Removed.
13309 (disqualify_candidate): Do not acll
13310 process_subtree_disqualification.
13311 (subtree_mark_written_and_enqueue): New function.
13312 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13313 RHS has been disqualified and re-queue LHS if necessary. Apart
13314 from that, ignore disqualified RHS.
13316 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13318 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13319 address fetch for z10 or later.
13321 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13323 * config/arc/arc.md (tst_movb): Add guard when splitting.
13325 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13327 * config/arc/arc.c (arc_can_eliminate): Test against
13328 arc_frame_pointer_needed.
13330 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13332 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13333 to prevent store reordering.
13334 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13335 (type): Add block type.
13336 (stack_tie): Define special instruction to be used in
13339 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13341 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13342 constraint. It is not valid for the pattern.
13343 (noncommutative_binary_comparison): Likewise.
13345 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13347 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13350 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13352 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13353 be used by the reg-alloc.
13355 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13357 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13358 reg-alloc when having mul64 or mul32x16 instructions.
13359 (mulsidi3): Likewise.
13360 (umulsidi3): Likewise.
13361 (mulsi32x16): New pattern.
13362 (mulsi64): Likewise.
13363 (mulsidi64): Likewise.
13364 (umulsidi64): Likewise.
13365 (MUL32x16_REG): Define.
13366 (mul64_600): Use MUL32x16_REG.
13367 (mac64_600): Likewise.
13368 (umul64_600): Likewise.
13369 (umac64_600): Likewise.
13371 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13373 * config/arc/arc.md (mulsi3_700): Make it commutative.
13375 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13377 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13379 (*sign_extendsidi2_insn): Likewise for movstosw.
13381 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13383 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13384 the type of the input discriminant value. Convert the
13385 discriminant value of signedness vary.
13387 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13389 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13390 Add to -Wall section.
13392 2017-06-01 Richard Biener <rguenther@suse.de>
13394 PR middle-end/66313
13395 * fold-const.c (fold_plusminus_mult_expr): If the factored
13396 factor may be zero use a wrapping type for the inner operation.
13397 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13398 and handle moved defs.
13399 (process_assignment): Properly guard the unary op case. Return a
13400 tri-state indicating that moving the stmt before the call may allow
13401 to continue. Pass through to_move.
13402 (find_tail_calls): Handle moving unrelated defs before
13405 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13408 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13409 splitter result in the canonical way.
13411 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13413 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13414 also for 32bit target. Update insn attributes.
13415 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13417 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13419 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13420 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13421 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13422 (_mm_maskz_min_ss): New intrinsics.
13424 2017-05-31 Martin Liska <mliska@suse.cz>
13426 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13427 Change comment style to one we normally use.
13428 (vectorizable_reduction): Likewise.
13429 (vectorizable_induction): Likewise.
13430 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13431 (vectorizable_call): Likewise.
13432 (vectorizable_simd_clone_call): Likewise.
13433 (vectorizable_conversion): Likewise.
13434 (vectorizable_assignment): Likewise.
13435 (vectorizable_shift): Likewise.
13436 (vectorizable_operation): Likewise.
13437 (vectorizable_store): Likewise.
13438 (vectorizable_load): Likewise.
13439 * tree-vectorizer.h: Likewise.
13441 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13443 * passes.c (emergency_dump_function): New.
13444 * tree-pass.h (emergency_dump_function): Declare.
13445 * plugin.c (plugins_internal_error_function): Remove.
13446 * plugin.h (plugins_internal_error_function): Remove declaration.
13447 * toplev.c (internal_error_function): New static function. Use it...
13448 (general_init): ...here.
13450 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13452 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13453 (arc_rtx_costs): Add costs for new patterns.
13454 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13455 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13457 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13459 * tree-ssa-strlen.c (get_next_strinfo): New function.
13460 (get_stridx_plus_constant): Use it.
13461 (zero_length_string): Likewise.
13462 (adjust_related_strinfos): Likewise.
13463 (adjust_last_stmt): Likewise.
13465 2017-05-31 Richard Biener <rguenther@suse.de>
13468 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13469 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13471 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13473 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13474 loop_vinfo argument and use of dependence distance vectors.
13475 Check instead whether the two references differ only in their
13476 initial value and assume that they have the same alignment if the
13477 difference is a multiple of the vector alignment.
13478 (vect_analyze_data_refs_alignment): Update call accordingly.
13480 2017-05-31 Martin Liska <mliska@suse.cz>
13483 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13485 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13487 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13488 (create_intersect_range_checks): Move from ...
13489 * tree-data-ref.c (create_intersect_range_checks_index)
13490 (create_intersect_range_checks): ... to here.
13491 (create_runtime_alias_checks): New function factored from ...
13492 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13493 here. Call above function.
13494 * tree-data-ref.h (create_runtime_alias_checks): New function.
13496 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13498 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13499 segment length for dr_b and compute it in wide_int.
13501 2017-05-31 Richard Biener <rguenther@suse.de>
13503 PR tree-optimization/80906
13504 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13505 and pass through iv_map.
13506 (copy_bb_and_scalar_dependences): Adjust.
13507 (translate_pending_phi_nodes): Likewise.
13508 (copy_loop_close_phi_args): Handle code-generating IVs instead
13511 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13513 * diagnostic-color.c (color_dict): Add "type-diff".
13514 (parse_gcc_colors): Update comment.
13515 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13516 -fdiagnostics-show-template-tree and -fno-elide-type.
13517 (GCC_COLORS): Add type-diff to example.
13519 (-fdiagnostics-show-template-tree): New.
13520 (-fno-elide-type): New.
13521 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13522 the pp_format_decoder callback. Call any m_format_postprocessor's
13524 (pretty_printer::pretty_printer): Initialize
13525 m_format_postprocessor.
13526 (pretty_printer::~pretty_printer): Delete any
13527 m_format_postprocessor.
13528 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13529 (class format_postprocessor): New class.
13530 (struct pretty_printer::format_decoder): Document the new parameters.
13531 (struct pretty_printer::m_format_postprocessor): New field.
13532 * tree-diagnostic.c (default_tree_printer): Update for new
13533 bool and const char ** params.
13534 * tree-diagnostic.h (default_tree_printer): Likewise.
13536 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13538 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13539 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13540 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13541 rs6000_gen_cell_microcode code.
13542 (rs6000_final_prescan_insn): Delete.
13543 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13544 "warn-cell-microcode" entries.
13545 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13546 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13547 throughout. Change cc_reg_not_micro_cr0_operand to
13548 cc_reg_not_cr0_operand throughout.
13549 (*extendhi<mode>2_noload): Delete.
13550 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13551 (mwarn-cell-microcode): Delete.
13552 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13553 -mgen-cell-microcode and -mwarn-cell-microcode.
13555 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13558 * config/i386/constraints.md (Yd): New constraint.
13560 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13561 and (?Yd, r) alternatives. Update insn attributes.
13562 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13563 and (?*Yd, r) alternatives. Update insn attributes.
13564 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13566 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13568 * gimplify.c (gimplify_modify_expr): Don't create a
13569 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13572 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13574 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13576 2017-05-30 Richard Biener <rguenther@suse.de>
13578 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13579 and reduc_def fields.
13580 (STMT_VINFO_REDUC_TYPE): New define.
13581 (STMT_VINFO_REDUC_DEF): Likewise.
13582 (vect_force_simple_reduction): Adjust prototype.
13583 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13584 (vect_is_simple_reduction): Remove check_reduction argument.
13585 (vect_force_simple_reduction): Adjust and set
13586 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13587 (vectorizable_reduction): Do not re-do reduction analysis
13588 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13589 * tree-parloops.c (gather_scalar_reductions): Adjust.
13591 2017-05-30 Richard Biener <rguenther@suse.de>
13593 PR middle-end/80901
13594 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13597 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13599 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13600 Introduce unknown_misalignment parameter and remove vf.
13601 (vect_peeling_hash_get_lowest_cost):
13602 Pass unknown_misalignment parameter.
13603 (vect_enhance_data_refs_alignment):
13604 Fix unsupportable data ref treatment.
13606 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13608 * tree-vect-data-refs.c (vect_get_data_access_cost):
13609 Workaround for SLP handling.
13610 (vect_enhance_data_refs_alignment):
13611 Compute costs for doing no peeling at all, compare to the best
13612 peeling costs so far and avoid peeling if cheaper.
13614 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13616 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13617 Return peeling info and set costs to zero for unlimited cost
13619 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13620 with unknown misalignment. Compute and costs for unknown
13621 misalignment, compare them to the costs for known misalignment
13622 and choose the cheapest for peeling.
13624 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13626 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13627 (vect_get_peeling_costs_all_drs): Create function.
13628 (vect_peeling_hash_get_lowest_cost):
13629 Use vect_get_peeling_costs_all_drs.
13630 (vect_peeling_supportable): Create function.
13631 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13633 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13635 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13636 DR_HAS_NEGATIVE_STEP.
13637 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13638 (vect_enhance_data_refs_alignment): Use.
13639 (vect_duplicate_ssa_name_ptr_info): Use.
13640 * tree-vectorizer.h (dr_misalignment): Use.
13641 (known_alignment_for_access_p): Use.
13643 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13646 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13648 (has_section_name): New function.
13650 2017-05-30 Martin Liska <mliska@suse.cz>
13653 * auto-profile.c (get_function_decl_from_block): Fix
13656 2017-05-30 Richard Biener <rguenther@suse.de>
13658 PR middle-end/80876
13659 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13661 2017-05-30 Martin Liska <mliska@suse.cz>
13663 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13664 * dumpfile.h (struct dump_file_info): Remove ctors.
13666 2017-05-30 Martin Liska <mliska@suse.cz>
13668 * predict.def: Fix GNU coding style.
13670 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13672 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13673 Mark 'to' argument with ATTRIBUTE_UNUSED.
13675 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13677 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13678 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13679 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13682 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13684 * doc/install.texi (Options specification): Restore entry of
13685 --enable-sjlj-exceptions.
13687 2017-05-27 Michael Eager <eager@eagercon.com>
13690 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13692 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13694 * config/microblaze/microblaze.h
13695 (FIXED_REGISTERS): Update in macro.
13696 (CALL_USED_REGISTERS): Update in macro.
13698 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13700 * doc/install.texi: Add links to macOS binary distributions.
13702 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13706 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13708 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13710 2017-05-26 Martin Liska <mliska@suse.cz>
13712 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13714 2017-05-26 Martin Liska <mliska@suse.cz>
13716 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13717 always leading ';; '.
13718 (dump_bb_info): Likewise.
13719 (brief_dump_cfg): Likewise.
13720 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13721 * dumpfile.c: Remove usage of TDF_VERBOSE.
13722 * dumpfile.h (enum dump_kind): Likewise.
13723 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13724 * print-tree.c (debug_verbose): Remove.
13725 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13726 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13727 * tree-diagnostic.c (default_tree_printer): Replace
13728 TDF_DIAGNOSTIC with TDF_SLIM.
13730 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13732 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13733 in parameter loop, rather than loop_vinfo.
13734 (create_intersect_range_checks): Ditto.
13735 (vect_create_cond_for_alias_checks): Update call to above functions.
13737 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13739 PR tree-optimization/80815
13740 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13741 for merging runtime alias checks. Handle negative DR_STEPs.
13743 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13745 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13747 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13748 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13749 out code pruning runtime alias checks.
13750 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13751 factored out from above.
13752 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13754 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13756 (prune_runtime_alias_test_list): New decalaration.
13758 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13760 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13761 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13762 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13763 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13764 (operator==, comp_dr_with_seg_len_pair): Ditto.
13765 (vect_prune_runtime_alias_test_list): Ditto.
13767 2017-05-26 Martin Liska <mliska@suse.cz>
13770 * params.def: Bound partial-inlining-entry-probability param.
13772 2017-05-26 Marek Polacek <polacek@redhat.com>
13775 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13778 2017-05-26 Richard Biener <rguenther@suse.de>
13780 PR tree-optimization/80842
13781 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13784 2017-05-26 Richard Biener <rguenther@suse.de>
13786 PR tree-optimization/80844
13787 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13789 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13791 * doc/md.texi (Machine Constraints): Update x86 family
13792 machine constraints section to match 'config/i386/constraints.md'.
13794 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13796 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13798 2017-05-25 Nathan Sidwell <nathan@acm.org>
13800 * doc/invoke.texi (--enable-languages): Update documentation.
13802 2017-05-25 Martin Liska <mliska@suse.cz>
13804 * dumpfile.c: Add TDF_FOLDING.
13805 * dumpfile.h (enum dump_kind): Likewise.
13806 * genmatch.c (dt_simplify::gen_1): Use it.
13808 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13810 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13812 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13814 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13815 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13817 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13819 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13820 * match.pd (X == C): Rewrite it here.
13821 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13822 with_certain_nonzero_bits2): New predicates.
13823 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13825 2017-05-24 Nathan Sidwell <nathan@acm.org>
13827 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13830 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13833 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13835 * config/powerpcspe: New port. Files are copied from the rs6000
13836 port, with "rs6000" in filenames replaced by "powerpcspe".
13838 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13840 PR rtl-optimization/80754
13841 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13843 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13845 * config/sparc/sparc.md (length): Return the correct value for -mflat
13846 sibcalls to match output_sibcall.
13848 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13852 * config/rs6000/rs6000.c (struct machine_function): Add new field
13854 (rs6000_get_separate_components): Init that field, use it.
13855 (rs6000_components_for_bb): Use the field.
13857 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13859 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13861 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13863 PR middle-end/80823
13864 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13866 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13869 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13870 address against address_operand predicate.
13871 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13873 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13875 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13876 parameters passed indirectly.
13878 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13880 * config/i386/i386.md (*movdi_internal): Remove SSE4
13881 alternative 18 (?r, *v). Update insn attributes.
13882 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13883 Update insn attributes.
13884 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13885 Update insn attributes.
13886 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13887 alternative 1 (r, v). Remove isa attribute.
13888 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13889 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13890 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13892 2017-05-23 Tom de Vries <tom@codesourcery.com>
13894 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13897 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13899 * cgraphunit.c (symbol_table::process_new_functions): Update.
13900 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13901 (inline_generate_summary): Rename to ...
13902 (ipa_fn_summary_generate): ... this one.
13903 (inline_read_summary): Rename to ...
13904 (ipa_fn_summary_read): ... this one.
13905 (inline_write_summary): Rename to ...
13906 (ipa_fn_summary_write): ... this one.
13907 (inline_free_summary): Rename to ...
13908 (ipa_free_fn_summary): ... this one.
13909 (pass_data_local_fn_summary, pass_local_fn_summary,
13910 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13911 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13912 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13913 make_pass_ipa_fn_summary): New.
13914 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13915 inline_write_summary, inline_free_summary): Remove.
13916 (ipa_free_fn_summary) : New.
13917 * ipa-inline.c (ipa_inline): Update.
13918 (pass_ipa_inline): Do not generate summaries.
13919 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13921 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13922 and add pass_ipa_fn_summary.
13923 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13925 (make_pass_inline_parameters): Remove.
13927 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13929 * omp-low.c (struct omp_context): Remove "default_kind" member.
13932 * omp-offload.c (execute_oacc_device_lower): Remove the
13933 parallelism dimensions function attributes for unparallelized
13934 OpenACC kernels constructs.
13936 2017-05-23 Martin Liska <mliska@suse.cz>
13938 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13940 (cgraph_edge::make_speculative): Likewise.
13941 (cgraph_edge::resolve_speculation): Likewise.
13942 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13943 (cgraph_node::dump): Likewise.
13944 * cgraph.h: Likewise.
13945 * cgraphunit.c (analyze_functions): Likewise.
13946 (symbol_table::compile): Likewise.
13947 * ipa-cp.c (print_all_lattices): Likewise.
13948 (determine_versionability): Likewise.
13949 (initialize_node_lattices): Likewise.
13950 (ipcp_verify_propagated_values): Likewise.
13951 (estimate_local_effects): Likewise.
13952 (update_profiling_info): Likewise.
13953 (create_specialized_node): Likewise.
13954 (perhaps_add_new_callers): Likewise.
13955 (decide_about_value): Likewise.
13956 (decide_whether_version_node): Likewise.
13957 (identify_dead_nodes): Likewise.
13958 (ipcp_store_bits_results): Likewise.
13959 * ipa-devirt.c (dump_targets): Likewise.
13960 (ipa_devirt): Likewise.
13961 * ipa-icf.c (sem_item::dump): Likewise.
13962 (sem_function::equals): Likewise.
13963 (sem_variable::equals): Likewise.
13964 (sem_item_optimizer::read_section): Likewise.
13965 (sem_item_optimizer::execute): Likewise.
13966 (congruence_class::dump): Likewise.
13967 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13968 (dump_inline_summary): Likewise.
13969 (estimate_node_size_and_time): Likewise.
13970 (inline_analyze_function): Likewise.
13971 * ipa-inline-transform.c (inline_call): Likewise.
13972 * ipa-inline.c (report_inline_failed_reason): Likewise.
13973 (want_early_inline_function_p): Likewise.
13974 (edge_badness): Likewise.
13975 (update_edge_key): Likewise.
13976 (inline_small_functions): Likewise.
13977 * ipa-profile.c (ipa_profile): Likewise.
13978 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13979 (ipa_make_edge_direct_to_target): Likewise.
13980 (remove_described_reference): Likewise.
13981 (ipa_impossible_devirt_target): Likewise.
13982 (propagate_controlled_uses): Likewise.
13983 (ipa_print_node_params): Likewise.
13984 (ipcp_transform_function): Likewise.
13985 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13986 (propagate_pure_const): Likewise.
13987 * ipa-reference.c (generate_summary): Likewise.
13988 (read_write_all_from_decl): Likewise.
13989 (propagate): Likewise.
13990 (ipa_reference_read_optimization_summary): Likewise.
13991 * ipa-utils.c (ipa_merge_profiles): Likewise.
13992 * ipa.c (walk_polymorphic_call_targets): Likewise.
13993 (symbol_table::remove_unreachable_nodes): Likewise.
13994 (ipa_single_use): Likewise.
13995 * passes.c (execute_todo): Likewise.
13996 * predict.c (drop_profile): Likewise.
13997 * symtab.c (symtab_node::get_dump_name): New function.
13998 (symtab_node::dump_name): Likewise.
13999 (symtab_node::dump_asm_name): Likewise.
14000 (symtab_node::dump_references): Likewise.
14001 (symtab_node::dump_referring): Likewise.
14002 (symtab_node::dump_base): Likewise.
14003 (symtab_node::debug_symtab): Likewise.
14004 * tree-sra.c (convert_callers_for_node): Likewise.
14005 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14006 * value-prof.c (init_node_map): Likewise.
14008 2017-05-23 Martin Liska <mliska@suse.cz>
14010 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14011 and symtab_node::debug_symtab to symbol_table::debug.
14012 * cgraphunit.c (analyze_functions): Use the renamed function.
14013 (symbol_table::compile): Likewise.
14014 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14015 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14016 * passes.c (execute_todo): Likewise.
14017 * symtab.c (symbol_table::dump): New function.
14018 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14020 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14022 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14023 that nonconst implies exec.
14025 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14027 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14028 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14029 (inline_edge_summary_vec): Turn into ...
14030 (ipa_call_summaries): ... this one.
14031 (redirect_to_unreachable, edge_set_predicate,
14032 evaluate_properties_for_edge, inline_summary_alloc,
14033 reset_ipa_call_summary, reset_inline_summary,
14034 inline_summary_t::duplicate): Update.
14035 (inline_edge_duplication_hook): Turn to ...
14036 (ipa_call_summary_t::duplicate): ... this one.
14037 (inline_edge_removal_hook): Turn to ...
14038 (ipa_call_summary_t::remove): ... this one.
14039 (dump_inline_edge_summary): Turn to ...
14040 (dump_ipa_call_summary): ... this one.
14041 (estimate_function_body_sizes): Update.
14042 (inline_update_callee_summaries): Update.
14043 (remap_edge_change_prob): Update.
14044 (remap_edge_summaries): Update.
14045 (inline_merge_summary): Update.
14046 (do_estimate_edge_time): Update.
14047 (inline_generate_summary): Update.
14048 (inline_read_section): Update.
14049 (inline_read_summary): Update.
14050 (inline_free_summary): Update.
14051 * ipa-inline.c (can_inline_edge_p): Update.
14052 (compute_inlined_call_time): Update.
14053 (want_inline_small_function_p): Update.
14054 (edge_badness): Update.
14055 (early_inliner): Update.
14056 * ipa-inline.h (inline_edge_summary): Turn to ...
14057 (ipa_call_summary): ... this one.
14058 (ipa_call_summary_t): New class.
14059 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14060 (ipa_call_summaries): New.
14061 (inline_edge_summary): Remove.
14062 (estimate_edge_growth): Update.
14063 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14064 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14065 * ipa-split.c (execute_split_functions): Update.
14066 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14068 2017-05-23 Tom de Vries <tom@codesourcery.com>
14070 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14071 attributes): Document rdrand effective target.
14073 2017-05-23 Tom de Vries <tom@codesourcery.com>
14075 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14076 attributes): Sort alphabetically.
14078 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14080 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14082 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14085 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14086 V2DF/V2DI splat into two separate patterns, one that handles
14087 registers, and the other that only handles memory. Drop support
14088 for splatting from a GPR on ISA 2.07 and then splitting the
14089 splat into direct move and splat.
14090 (vsx_splat_<mode>_reg): Likewise.
14091 (vsx_splat_<mode>_mem): Likewise.
14093 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14095 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14097 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14099 PR middle-end/80809
14100 * omp-low.c (finish_taskreg_remap): New function.
14101 (finish_taskreg_scan): If unit size of ctx->record_type
14102 is non-constant, unshare the size expression and replace
14103 decls in it with possible outer var refs.
14105 PR middle-end/80809
14106 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14107 GOVD_SHARED rather than GOVD_PRIVATE with it.
14108 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14109 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14111 PR middle-end/80853
14112 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14113 as last argument to build_outer_var_ref for pointer bases of array
14114 section reductions.
14116 2017-05-19 Martin Sebor <msebor@redhat.com>
14118 * print-tree.c (print_node): Print DECL_READ_P flag.
14120 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14122 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14123 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14124 * cgraph.c: Likewise.
14125 * cgraphunit.c: Likewise.
14126 * gengtype.c: Likewise.
14127 * ipa-cp.c: Likewise.
14128 * ipa-devirt.c: Likewise.
14129 * ipa-icf.c: Likewise.
14130 * ipa-predicate.c: Likewise.
14131 * ipa-profile.c: Likewise.
14132 * ipa-prop.c: Likewise.
14133 * ipa-split.c: Likewise.
14135 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14136 edge_predicate_pool, dump_inline_hints,
14137 inline_summary::account_size_time, redirect_to_unreachable,
14138 edge_set_predicate, set_hint_predicate,
14139 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14140 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14141 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14142 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14143 ipa_call_summary_t::remove, initialize_growth_caches,
14144 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14145 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14146 mark_modified, unmodified_parm_1, unmodified_parm,
14147 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14148 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14149 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14150 will_be_nonconstant_predicate, record_modified_bb_info,
14151 get_minimal_bb, record_modified, param_change_prob,
14152 phi_result_unknown_predicate, predicate_for_phi_result,
14153 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14154 estimate_function_body_sizes, compute_inline_parameters,
14155 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14156 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14157 inline_update_callee_summaries, remap_edge_change_prob,
14158 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14159 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14160 inline_analyze_function, inline_summary_t::insert,
14161 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14162 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14163 inline_free_summary): Move to ipa-fnsummary.h
14164 (predicate_t): Remove.
14165 * ipa-fnsummary.c: New file.
14166 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14167 (enum inline_hints_vals, inline_hints, agg_position_info,
14168 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14169 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14170 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14171 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14172 inline_read_summary, inline_write_summary, inline_free_summary,
14173 inline_analyze_function, initialize_inline_failed,
14174 inline_merge_summary, inline_update_overall_summary,
14175 compute_inline_parameters): Move to ipa-fnsummary.h
14176 * ipa-fnsummary.h: New file.
14177 * ipa-inline-transform.h: Include ipa-inline.h.
14178 * ipa-inline.c: LIkewise.
14180 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14182 * ipa-inline.c (edge_badness): Use inlined_time instead of
14183 inline_summaries->get.
14185 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14187 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14189 2017-05-22 Nathan Sidwell <nathan@acm.org>
14191 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14192 (fdump-lang): Document 'raw' option.
14193 * dumpfile.h (TDI_tu): Delete.
14194 * dumpfile.c (dump_files): Remove translation-unit.
14195 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14197 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14199 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14200 command option from $(AWK) call.
14201 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14202 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14203 [FORMAT]: Remove handling of variable.
14204 * config/avr/t-multilib: Regenerate.
14206 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14208 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14210 (dump_inline_summary): Do not print self_time.
14211 (estimate_function_body_sizes): Do not set self_time.
14212 (compute_inline_parameters): Likewise.
14213 (inline_read_section, inline_write_summary): Do not stream self_time.
14214 * ipa-inline.h (inline_summary): Drop self_time.
14216 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14218 * ipa-inline-analysis.c (account_size_time): Rename to ...
14219 (inline_summary::account_size_time): ... this one.
14220 (reset_ipa_call_summary): Turn to ...
14221 (ipa_call_summary::reset): ... this one.
14222 (reset_inline_summary): Turn to ...
14223 (inline_summary::reset): ... this one.
14224 (inline_summary_t::remove): Update.
14225 (inline_summary_t::duplicate): Update.
14226 (ipa_call_summary_t::remove): Update.
14227 (dump_inline_summary): Update.
14228 (estimate_function_body_sizes): Update.
14229 (compute_inline_parameters): Update.
14230 (estimate_node_size_and_time): Update.
14231 (inline_merge_summary): Update.
14232 (inline_update_overall_summary): Update.
14233 (inline_read_section): Update.
14234 (inline_write_summary): Update.
14235 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14236 add account_size_time and reset member functions.
14237 (ipa_call_summary): Add reset function.
14238 * ipa-predicate.h (predicate::operator &): Constify.
14240 2017-05-22 Richard Biener <rguenther@suse.de>
14242 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14244 2017-05-19 Jason Merrill <jason@redhat.com>
14246 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14248 2017-05-19 Marek Polacek <polacek@redhat.com>
14251 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14252 TYPE_OVERFLOW_WRAPS checks.
14254 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14256 * tree-core.h (enum omp_clause_default_kind): Add
14257 "OMP_CLAUSE_DEFAULT_PRESENT".
14258 * tree-pretty-print.c (dump_omp_clause): Handle it.
14259 * gimplify.c (enum gimplify_omp_var_data): Add
14260 "GOVD_MAP_FORCE_PRESENT".
14261 (gimplify_adjust_omp_clauses_1): Map it to
14262 "GOMP_MAP_FORCE_PRESENT".
14263 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14265 * gimplify.c (oacc_default_clause): Clarify.
14267 2017-05-19 Nathan Sidwell <nathan@acm.org>
14269 LANG_HOOK_REGISTER_DUMPS
14270 * toplev.c (general_init): Call register dump lang hook.
14271 * doc/invoke.texi: Document -fdump-lang option family.
14272 * dumpfile.c (dump_files): Remove class dump here.
14273 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14274 * dumpfile.h (tree_dump_index): Remove TDI_class.
14275 * langhooks-def.h (lhd_register_dumps): Declare.
14276 (LANG_HOOKS_REGISTER_DUMPS): Define.
14277 (LANG_HOOKS_INITIALIZER): Add it.
14278 * langhooks.c (lhd_register_dumps): Define.
14279 * langhooks.h (struct lang_hooks): Add register_dumps.
14281 2017-05-19 Nathan Sidwell <nathan@acm.org>
14283 * context.h (context::set_passes): New.
14284 * context.c (context::context): Do not create pass manager.
14285 * toplev.c (general_init): Create pass manager here.
14287 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14289 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14290 use this splitter if two add or or instructions would also work for
14291 the constant we want to generate.
14293 2017-05-19 Richard Biener <rguenther@suse.de>
14296 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14297 predicate evaluation.
14299 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14301 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14303 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14306 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14308 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14309 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14310 (inline_edge_summary_vec): Turn into ...
14311 (ipa_call_summaries): ... this one.
14312 (redirect_to_unreachable, edge_set_predicate,
14313 evaluate_properties_for_edge, inline_summary_alloc,
14314 reset_ipa_call_summary, reset_inline_summary,
14315 inline_summary_t::duplicate): Update.
14316 (inline_edge_duplication_hook): Turn to ...
14317 (ipa_call_summary_t::duplicate): ... this one.
14318 (inline_edge_removal_hook): Turn to ...
14319 (ipa_call_summary_t::remove): ... this one.
14320 (dump_inline_edge_summary): Turn to ...
14321 (dump_ipa_call_summary): ... this one.
14322 (estimate_function_body_sizes): Update.
14323 (inline_update_callee_summaries): Update.
14324 (remap_edge_change_prob): Update.
14325 (remap_edge_summaries): Update.
14326 (inline_merge_summary): Update.
14327 (do_estimate_edge_time): Update.
14328 (inline_generate_summary): Update.
14329 (inline_read_section): Update.
14330 (inline_read_summary): Update.
14331 (inline_free_summary): Update.
14332 * ipa-inline.c (can_inline_edge_p): Update.
14333 (compute_inlined_call_time): Update.
14334 (want_inline_small_function_p): Update.
14335 (edge_badness): Update.
14336 (early_inliner): Update.
14337 * ipa-inline.h (inline_edge_summary): Turn to ...
14338 (ipa_call_summary): ... this one.
14339 (ipa_call_summary_t): New class.
14340 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14341 (ipa_call_summaries): New.
14342 (inline_edge_summary): Remove.
14343 (estimate_edge_growth): Update.
14344 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14345 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14346 * ipa-split.c (execute_split_functions): Update.
14347 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14349 2017-05-19 Richard Biener <rguenther@suse.de>
14351 PR middle-end/80764
14352 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14354 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14356 * config/rs6000/rs6000.c (struct machine_function): Add field
14357 fpr_is_wrapped_separately.
14358 (rs6000_get_separate_components): Use 64 components. Handle the
14359 new FPR components.
14360 (rs6000_components_for_bb): Handle the FPR components.
14361 (rs6000_emit_prologue_components): Handle the FPR components.
14362 (rs6000_emit_epilogue_components): Handle the FPR components.
14363 (rs6000_set_handled_components): Handle the FPR components.
14364 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14365 that are already separately shrink-wrapped.
14366 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14367 that are already separately shrink-wrapped.
14369 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14372 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14375 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14376 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14377 cases where the register allocator uses a move and an offsettable
14378 memory operation to/from a FPR register on ISA 2.06/2.07.
14379 (define_peephole2 for Altivec d-form store): Likewise.
14381 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14384 * config/i386/mmx.md (*mov<mode>_internal): Enable
14385 alternatives 11, 12, 13 and 14 also for 32bit targets.
14386 Remove alternatives 15, 16, 17 and 18.
14387 * config/i386/sse.md (vec_concatv2di): Change
14388 alternative (!x, *y) to (x, ?!*Yn).
14390 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14392 * dumpfile.h (enum dump_kind): Remove stray comma.
14394 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14396 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14397 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14398 predicate::num_conditions
14399 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14400 (CHANGED): turn into predicate::changed.
14401 (agg_position_info): Move to ipa-predicate.h
14402 (add_condition, predicate::add_clause, predicate::operator &=,
14403 predicate::or_with, predicate::evaluate, predicate::probability,
14404 dump_condition, dump_clause, predicate::dump,
14405 predicate::remap_after_duplication, predicate::remap_after_inlining,
14406 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14407 (evaluate_conditions_for_known_args): Update.
14408 (set_cond_stmt_execution_predicate): Update.
14409 * ipa-inline.h: Include ipa-predicate.h
14410 (condition, inline_param_summary, conditions, agg_position_info,
14411 predicate): Move to ipa-predicate.h
14412 * ipa-predicate.c: New file.
14413 * ipa-predicate.h: New file.
14415 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14417 * final.c (leaf_function_p): Check we are not in a sequence.
14419 2017-05-18 Martin Liska <mliska@suse.cz>
14421 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14422 * dumpfile.c (dump_register): Use new enum dump_kind.
14423 (get_dump_file_name): Likewise.
14424 (dump_enable_all): Likewise.
14425 (dump_switch_p_1): Likewise.
14426 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14427 * dumpfile.h (enum dump_kind): New enum type.
14428 (struct dump_file_info): Create constructor and
14429 format fields and comments.
14430 * passes.c (pass_manager::register_one_dump_file):
14432 * statistics.c (statistics_early_init): Likewise.
14433 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14434 TDF_TREE with TDF_SLIM.
14435 (gather_memory_references_ref): Likewise.
14437 2017-05-18 Martin Liska <mliska@suse.cz>
14439 * vec.h (struct vnull): Use it.
14441 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14443 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14444 (true_predicate, false_predicate, true_predicate_p,
14445 false_predicate_p): Remove.
14446 (single_cond_predicate, not_inlined_predicate): Turn to member function
14448 (add_condition): Update.
14449 (add_clause): Turn to...
14450 (predicate::add_clause): ... this one; update; allow passing NULL
14452 (and_predicates): Turn to ...
14453 (predicate::operator &=): ... this one.
14454 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14455 (or_predicates): Turn to ...
14456 (predicate::or_with): ... this one.
14457 (evaluate_predicate): Turn to ...
14458 (predicate::evaluate): ... this one.
14459 (predicate_probability): Turn to ...
14460 (predicate::probability): ... this one.
14461 (dump_condition): Update.
14462 (dump_predicate): Turn to ...
14463 (predicate::dump): ... this one.
14464 (account_size_time): Update.
14465 (edge_set_predicate): Update.
14466 (set_hint_predicate): UPdate.
14467 (evaluate_conditions_for_known_args): Update.
14468 (evaluate_properties_for_edge): Update.
14469 (remap_predicate_after_duplication): Turn to...
14470 (predicate::remap_after_duplication): ... this one.
14471 (remap_hint_predicate_after_duplication): Update.
14472 (inline_summary_t::duplicate): UPdate.
14473 (dump_inline_edge_summary): Update.
14474 (dump_inline_summary): Update.
14475 (set_cond_stmt_execution_predicate): Update.
14476 (set_switch_stmt_execution_predicate): Update.
14477 (compute_bb_predicates): Update.
14478 (will_be_nonconstant_expr_predicate): Update.
14479 (will_be_nonconstant_predicate): Update.
14480 (phi_result_unknown_predicate): Update.
14481 (predicate_for_phi_result): Update.
14482 (array_index_predicate): Update.
14483 (estimate_function_body_sizes): Update.
14484 (estimate_node_size_and_time): Update.
14485 (estimate_ipcp_clone_size_and_time): Update.
14486 (remap_predicate): Rename to ...
14487 (predicate::remap_after_inlining): ... this one.
14488 (remap_hint_predicate): Update.
14489 (inline_merge_summary): Update.
14490 (inline_update_overall_summary): Update.
14491 (estimate_size_after_inlining): Update.
14492 (read_predicate): Rename to ...
14493 (predicate::stream_in): ... this one.
14494 (read_inline_edge_summary): Update.
14495 (write_predicate): Rename to ...
14496 (predicate::stream_out): ... this one.
14497 (write_inline_edge_summary): Update.
14498 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14499 (clause_t): Turn to uint32_t
14500 (predicate): Turn to class; implement constructor and operators
14502 (size_time_entry): Update.
14503 (inline_summary): Update.
14504 (inline_edge_summary): Update.
14506 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14508 * fold-const.c (fold_binary_loc): Move transformation...
14509 * match.pd (C - X CMP X): ... here.
14511 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14513 * config/sparc/sparc.c (sparc_option_override): Set function
14514 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14515 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14517 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14519 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14521 2017-05-18 Marek Polacek <polacek@redhat.com>
14524 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14525 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14526 gimple_assign_load_p.
14528 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14530 PR middle-end/80692
14531 * real.c (do_compare): Give decimal_do_compare preference over
14532 comparing just the signs.
14534 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14536 * doc/md.texi (Canonicalization of Instructions): Describe the
14537 canonical form of instructions that inherently set a condition
14540 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14542 PR middle-end/80775
14543 * tree-cfg.c: Move deletion of unreachable case statements to after
14544 the merging of consecutive case labels.
14546 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14548 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14549 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14550 restoring of callee-saved registers.
14552 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14554 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14555 * config/visium/visium.c (single_set_and_flags): Likewise.
14556 * config/visium/visium.md (Substitutions): Likewise.
14558 2017-05-17 Martin Liska <mliska@suse.cz>
14560 * cfg.c: Introduce dump_flags_t type and
14561 use it instead of int type.
14563 * cfghooks.c: Likewise.
14564 * cfghooks.h (struct cfg_hooks): Likewise.
14565 * cfgrtl.c: Likewise.
14566 * cfgrtl.h: Likewise.
14567 * cgraph.c (cgraph_node::get_body): Likewise.
14568 * coretypes.h: Likewise.
14569 * domwalk.c: Likewise.
14570 * domwalk.h: Likewise.
14571 * dumpfile.c (struct dump_option_value_info): Likewise.
14572 (dump_enable_all): Likewise.
14573 (dump_switch_p_1): Likewise.
14574 (opt_info_switch_p): Likewise.
14575 * dumpfile.h (enum tree_dump_index): Likewise.
14576 (struct dump_file_info): Likewise.
14577 * genemit.c: Likewise.
14578 * generic-match-head.c: Likewise.
14579 * gengtype.c (open_base_files): Likewise.
14580 * gimple-pretty-print.c: Likewise.
14581 * gimple-pretty-print.h: Likewise.
14582 * graph.c (print_graph_cfg): Likewise.
14583 * graphite-scop-detection.c (dot_all_sese): Likewise.
14584 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14585 * loop-unroll.c (report_unroll): Likewise.
14586 * passes.c (pass_manager::register_one_dump_file): Likewise.
14587 * print-tree.c: Likewise.
14588 * statistics.c: Likewise.
14589 * tree-cfg.c: Likewise.
14590 * tree-cfg.h: Likewise.
14591 * tree-dfa.c: Likewise.
14592 * tree-dfa.h: Likewise.
14593 * tree-dump.c (dump_function): Likewise.
14594 * tree-dump.h (struct dump_info): Likewise.
14595 * tree-pretty-print.c: Likewise.
14596 * tree-pretty-print.h: Likewise.
14597 * tree-ssa-live.c: Likewise.
14598 * tree-ssa-live.h: Likewise.
14599 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14600 * tree-vect-loop.c: Likewise.
14601 * tree-vect-slp.c: Likewise.
14603 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14604 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14606 PR tree-optimization/80457
14607 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14608 of all arguments to a statement as scalar_to_vec operations.
14609 (vectorizable_call): Adjust call to vect_model_simple_cost for
14611 (vectorizable_conversion): Likewise.
14612 (vectorizable_assignment): Likewise.
14613 (vectorizable_shift): Likewise.
14614 (vectorizable_operation): Likewise.
14615 (vectorizable_comparison): Likewise.
14616 (vect_is_simple_cond): Record the def types for operands.
14617 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14618 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14619 for statement argument count.
14621 2017-05-16 Carl Love <cel@us.ibm.com>
14623 * config/rs6000/rs6000-c: Add support for built-in functions
14624 vector unsigned long long vec_bperm (vector unsigned long long,
14625 vector unsigned char)
14626 vector signed long long vec_mule (vector signed int,
14628 vector unsigned long long vec_mule (vector unsigned int,
14629 vector unsigned int)
14630 vector signed long long vec_mulo (vector signed int,
14632 vector unsigned long long vec_mulo (vector unsigned int,
14633 vector unsigned int)
14634 vector signed char vec_sldw (vector signed char,
14635 vector signed char,
14637 vector unsigned char vec_sldw (vector unsigned char,
14638 vector unsigned char,
14640 vector signed short vec_sldw (vector signed short,
14641 vector signed short,
14643 vector unsigned short vec_sldw (vector unsigned short,
14644 vector unsigned short,
14646 vector signed int vec_sldw (vector signed int,
14649 vector unsigned int vec_sldw (vector unsigned int,
14650 vector unsigned int,
14652 vector signed long long vec_sldw (vector signed long long,
14653 vector signed long long,
14655 vector unsigned long long vec_sldw (vector unsigned long long,
14656 vector unsigned long long,
14658 * config/rs6000/rs6000-c: Add support for built-in functions
14659 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14660 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14661 * doc/extend.texi: Update the built-in documentation for the
14662 new built-in functions.
14664 2017-05-16 Marek Polacek <polacek@redhat.com>
14668 * tree.c (save_expr): Don't fold the expression.
14670 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14672 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14673 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14674 and (?*y,m). Update insn attributes.
14676 2017-05-16 Martin Liska <mliska@suse.cz>
14678 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14679 flags argument of print_gimple_stmt, print_gimple_expr,
14680 print_generic_stmt and print_generic_expr.
14681 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14682 * coretypes.h: Likewise.
14683 * except.c (dump_eh_tree): Likewise.
14684 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14685 * gimple-pretty-print.h: Likewise.
14686 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14687 (backprop::push_to_worklist): Likewise.
14688 (backprop::pop_from_worklist): Likewise.
14689 (backprop::process_use): Likewise.
14690 (backprop::intersect_uses): Likewise.
14691 (note_replacement): Likewise.
14692 * gimple-ssa-store-merging.c
14693 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14694 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14695 (pass_store_merging::execute): Likewise.
14696 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14697 (ssa_base_cand_dump_callback): Likewise.
14698 (dump_incr_vec): Likewise.
14699 (replace_refs): Likewise.
14700 (replace_mult_candidate): Likewise.
14701 (create_add_on_incoming_edge): Likewise.
14702 (create_phi_basis): Likewise.
14703 (insert_initializers): Likewise.
14704 (all_phi_incrs_profitable): Likewise.
14705 (introduce_cast_before_cand): Likewise.
14706 (replace_one_candidate): Likewise.
14707 * gimplify.c (gimplify_expr): Likewise.
14708 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14709 (set_rename): Likewise.
14710 (rename_uses): Likewise.
14711 (copy_loop_phi_nodes): Likewise.
14712 (add_close_phis_to_merge_points): Likewise.
14713 (copy_loop_close_phi_args): Likewise.
14714 (copy_cond_phi_args): Likewise.
14715 (graphite_copy_stmts_from_block): Likewise.
14716 (translate_pending_phi_nodes): Likewise.
14717 * graphite-poly.c (print_pdr): Likewise.
14718 (dump_gbb_cases): Likewise.
14719 (dump_gbb_conditions): Likewise.
14720 (print_scop_params): Likewise.
14721 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14722 (build_cross_bb_scalars_use): Likewise.
14723 (gather_bbs::before_dom_children): Likewise.
14724 * hsa-dump.c (dump_hsa_immed): Likewise.
14725 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14726 (get_replacement_map): Likewise.
14727 * ipa-inline-analysis.c (dump_condition): Likewise.
14728 (estimate_function_body_sizes): Likewise.
14729 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14730 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14731 * ipa-prop.c (ipa_dump_param): Likewise.
14732 (ipa_print_node_jump_functions_for_edge): Likewise.
14733 (ipa_modify_call_arguments): Likewise.
14734 (ipa_modify_expr): Likewise.
14735 (ipa_dump_param_adjustments): Likewise.
14736 (ipa_dump_agg_replacement_values): Likewise.
14737 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14738 * ipa-pure-const.c (check_stmt): Likewise.
14739 (pass_nothrow::execute): Likewise.
14740 * ipa-split.c (execute_split_functions): Likewise.
14741 * omp-offload.c (dump_oacc_loop_part): Likewise.
14742 (dump_oacc_loop): Likewise.
14743 * trans-mem.c (tm_log_emit): Likewise.
14744 (tm_memopt_accumulate_memops): Likewise.
14745 (dump_tm_memopt_set): Likewise.
14746 (dump_tm_memopt_transform): Likewise.
14747 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14748 (print_loop): Likewise.
14749 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14750 (chkp_gather_checks_info): Likewise.
14751 (chkp_get_check_result): Likewise.
14752 (chkp_remove_check_if_pass): Likewise.
14753 (chkp_use_outer_bounds_if_possible): Likewise.
14754 (chkp_reduce_bounds_lifetime): Likewise.
14755 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14756 (chkp_mark_completed_bounds): Likewise.
14757 (chkp_register_incomplete_bounds): Likewise.
14758 (chkp_mark_invalid_bounds): Likewise.
14759 (chkp_maybe_copy_and_register_bounds): Likewise.
14760 (chkp_build_returned_bound): Likewise.
14761 (chkp_get_bound_for_parm): Likewise.
14762 (chkp_build_bndldx): Likewise.
14763 (chkp_get_bounds_by_definition): Likewise.
14764 (chkp_generate_extern_var_bounds): Likewise.
14765 (chkp_get_bounds_for_decl_addr): Likewise.
14766 * tree-chrec.c (chrec_apply): Likewise.
14767 * tree-data-ref.c (dump_data_reference): Likewise.
14768 (dump_subscript): Likewise.
14769 (dump_data_dependence_relation): Likewise.
14770 (analyze_overlapping_iterations): Likewise.
14771 * tree-inline.c (expand_call_inline): Likewise.
14772 (tree_function_versioning): Likewise.
14773 * tree-into-ssa.c (dump_defs_stack): Likewise.
14774 (dump_currdefs): Likewise.
14775 (dump_names_replaced_by): Likewise.
14776 (dump_update_ssa): Likewise.
14777 (update_ssa): Likewise.
14778 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14779 * tree-parloops.c (build_new_reduction): Likewise.
14780 (try_create_reduction_list): Likewise.
14781 (ref_conflicts_with_region): Likewise.
14782 (oacc_entry_exit_ok_1): Likewise.
14783 (oacc_entry_exit_single_gang): Likewise.
14784 * tree-pretty-print.h: Likewise.
14785 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14786 (get_scalar_evolution): Likewise.
14787 (add_to_evolution): Likewise.
14788 (get_loop_exit_condition): Likewise.
14789 (analyze_evolution_in_loop): Likewise.
14790 (analyze_initial_condition): Likewise.
14791 (analyze_scalar_evolution): Likewise.
14792 (instantiate_scev): Likewise.
14793 (number_of_latch_executions): Likewise.
14794 (gather_chrec_stats): Likewise.
14795 (final_value_replacement_loop): Likewise.
14796 (scev_const_prop): Likewise.
14797 * tree-sra.c (dump_access): Likewise.
14798 (disqualify_candidate): Likewise.
14799 (create_access): Likewise.
14800 (reject): Likewise.
14801 (maybe_add_sra_candidate): Likewise.
14802 (create_access_replacement): Likewise.
14803 (analyze_access_subtree): Likewise.
14804 (analyze_all_variable_accesses): Likewise.
14805 (sra_modify_assign): Likewise.
14806 (initialize_constant_pool_replacements): Likewise.
14807 (find_param_candidates): Likewise.
14808 (decide_one_param_reduction): Likewise.
14809 (replace_removed_params_ssa_names): Likewise.
14810 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14811 * tree-ssa-copy.c (dump_copy_of): Likewise.
14812 (copy_prop_visit_cond_stmt): Likewise.
14813 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14814 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14815 (record_equivalences_from_stmt): Likewise.
14816 * tree-ssa-dse.c (compute_trims): Likewise.
14817 (delete_dead_call): Likewise.
14818 (delete_dead_assignment): Likewise.
14819 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14820 (forward_propagate_into_cond): Likewise.
14821 (pass_forwprop::execute): Likewise.
14822 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14823 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14825 (move_computations_worker): Likewise.
14826 (execute_sm): Likewise.
14827 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14828 (remove_exits_and_undefined_stmts): Likewise.
14829 (remove_redundant_iv_tests): Likewise.
14830 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14831 (adjust_iv_update_pos): Likewise.
14832 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14833 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14834 (value_replacement): Likewise.
14835 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14836 * tree-ssa-pre.c (print_pre_expr): Likewise.
14837 (get_representative_for): Likewise.
14838 (create_expression_by_pieces): Likewise.
14839 (insert_into_preds_of_block): Likewise.
14840 (eliminate_insert): Likewise.
14841 (eliminate_dom_walker::before_dom_children): Likewise.
14842 (eliminate): Likewise.
14843 (remove_dead_inserted_code): Likewise.
14844 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14845 * tree-ssa-reassoc.c (get_rank): Likewise.
14846 (eliminate_duplicate_pair): Likewise.
14847 (eliminate_plus_minus_pair): Likewise.
14848 (eliminate_not_pairs): Likewise.
14849 (undistribute_ops_list): Likewise.
14850 (eliminate_redundant_comparison): Likewise.
14851 (update_range_test): Likewise.
14852 (optimize_range_tests_var_bound): Likewise.
14853 (optimize_vec_cond_expr): Likewise.
14854 (rewrite_expr_tree): Likewise.
14855 (rewrite_expr_tree_parallel): Likewise.
14856 (linearize_expr): Likewise.
14857 (break_up_subtract): Likewise.
14858 (linearize_expr_tree): Likewise.
14859 (attempt_builtin_powi): Likewise.
14860 (attempt_builtin_copysign): Likewise.
14861 (transform_stmt_to_copy): Likewise.
14862 (transform_stmt_to_multiply): Likewise.
14863 (dump_ops_vector): Likewise.
14864 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14865 (print_scc): Likewise.
14866 (set_ssa_val_to): Likewise.
14867 (visit_reference_op_store): Likewise.
14868 (visit_use): Likewise.
14869 (sccvn_dom_walker::before_dom_children): Likewise.
14870 (run_scc_vn): Likewise.
14871 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14873 (expr_hash_elt::print): Likewise.
14874 (const_and_copies::pop_to_marker): Likewise.
14875 (const_and_copies::record_const_or_copy_raw): Likewise.
14876 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14877 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14878 (dump_predicates): Likewise.
14879 (find_uninit_use): Likewise.
14880 (warn_uninitialized_phi): Likewise.
14881 (pass_late_warn_uninitialized::execute): Likewise.
14882 * tree-ssa.c (verify_vssa): Likewise.
14883 (verify_ssa): Likewise.
14884 (maybe_optimize_var): Likewise.
14885 * tree-vrp.c (dump_value_range): Likewise.
14886 (dump_all_value_ranges): Likewise.
14887 (dump_asserts_for): Likewise.
14888 (register_edge_assert_for_2): Likewise.
14889 (vrp_visit_cond_stmt): Likewise.
14890 (vrp_visit_switch_stmt): Likewise.
14891 (vrp_visit_stmt): Likewise.
14892 (vrp_visit_phi_node): Likewise.
14893 (simplify_cond_using_ranges_1): Likewise.
14894 (fold_predicate_in): Likewise.
14895 (evrp_dom_walker::before_dom_children): Likewise.
14896 (evrp_dom_walker::push_value_range): Likewise.
14897 (evrp_dom_walker::pop_value_range): Likewise.
14898 (execute_early_vrp): Likewise.
14900 2017-05-16 Richard Biener <rguenther@suse.de>
14902 * dwarf2out.c (loc_list_from_tree_1): Do not create
14903 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14905 2017-05-16 Richard Biener <rguenther@suse.de>
14907 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14909 (note_variable_value_in_expr): If we resolved the decl ref
14910 do not push to the stack.
14912 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14914 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14915 operations in fast-math mode.
14916 (vaddq_f16): Likewise.
14917 (vmul_f16): Likewise.
14918 (vmulq_f16): Likewise.
14919 (vsub_f16): Likewise.
14920 (vsubq_f16): Likewise.
14921 * config/arm/neon.md (add<mode>3): New.
14923 (fma:<VH:mode>3): New. Also remove outdated comment.
14926 2017-05-16 Martin Liska <mliska@suse.cz>
14930 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14931 (odr_types_equivalent_p): Likewise.
14933 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14935 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14937 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14940 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14941 non-interunit SSE move alternatives with '?'.
14942 (zero-extendsidi peephole2): New peephole to skip intermediate
14943 general register in SSE zero-extend sequence.
14945 2017-05-15 Jeff Law <law@redhat.com>
14947 * reorg.c (relax_delay_slots): Create a new variable to hold
14948 the temporary target rather than clobbering TARGET_LABEL.
14950 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14951 missing argument to extract_bit_field call.
14952 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14954 2017-05-15 Martin Liska <mliska@suse.cz>
14957 * gcc.c (process_command): Do not allow empty argument of -o option.
14959 2017-05-15 Renlin Li <renlin.li@arm.com>
14961 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14962 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14963 * config/aarch64/constraints.md (Usf): Add long call check.
14964 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14965 (call_value): Likewise.
14966 (sibcall): Likewise.
14967 (sibcall_value): Likewise.
14969 (call_value_insn): New.
14970 (sibcall_insn): Update rtx pattern.
14971 (sibcall_value_insn): Likewise.
14972 (call_internal): Remove.
14973 (call_value_internal): Likewise.
14974 (sibcall_internal): Likewise.
14975 (sibcall_value_internal): Likewise.
14976 (call_reg): Likewise.
14977 (call_symbol): Likewise.
14978 (call_value_reg): Likewise.
14979 (call_value_symbol): Likewise.
14981 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14984 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14986 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14988 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14989 compatible with CCGOCmode and with CCZmode.
14991 2017-05-14 Martin Sebor <msebor@redhat.com>
14993 PR middle-end/77671
14994 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14995 (gimple_fold_builtin_snprintf): Same.
14996 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14997 (gimple_fold_builtin_snprintf): Same.
14998 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14999 of character types.
15000 (is_call_safe): New function.
15001 (try_substitute_return_value): Call it.
15002 (try_simplify_call): New function.
15003 (pass_sprintf_length::handle_gimple_call): Call it.
15005 2017-05-14 Martin Sebor <msebor@redhat.com>
15007 PR middle-end/80669
15008 * builtins.c (expand_builtin_stpncpy): Simplify.
15010 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15012 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15013 * config/i386/i386.h
15014 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15015 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15016 (struct machine_function): Add new members call_ms2sysv,
15017 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15018 (struct machine_frame_state): New fields sp_realigned and
15019 sp_realigned_offset.
15020 * config/i386/i386.c
15021 (enum xlogue_stub): New enum.
15022 (enum xlogue_stub_sets): New enum.
15023 (class xlogue_layout): New class.
15024 (struct ix86_frame): New fields stack_realign_allocate_offset,
15025 stack_realign_offset and outlined_save_offset. Modify comments to
15026 detail stack layout when using out-of-line stubs.
15027 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15028 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15029 -mcall-ms2sysv-xlogues.
15030 (stub_managed_regs): New static variable.
15031 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15032 registers managed by out-of-line stub.
15033 (disable_call_ms2sysv_xlogues): New function.
15034 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15035 m->call_ms2sysv when appropriate and compute frame layout for
15037 (sp_valid_at, fp_valid_at): New inline functions.
15038 (choose_basereg): New function.
15039 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15041 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15042 Use align parameter of choose_baseaddr to generated aligned SSE movs
15044 (pro_epilogue_adjust_stack): Modify to track
15045 machine_frame_state::sp_realigned.
15046 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15047 (ix86_nsaved_sseregs): Likewise.
15048 (ix86_emit_save_regs): Likewise.
15049 (ix86_emit_save_regs_using_mov): Likewise.
15050 (ix86_emit_save_sse_regs_using_mov): Likewise.
15051 (get_scratch_register_on_entry): Likewise.
15052 (gen_frame_set): New function.
15053 (gen_frame_load): Likewise.
15054 (gen_frame_store): Likewise.
15055 (emit_outlined_ms2sysv_save): Likewise.
15056 (emit_outlined_ms2sysv_restore): Likewise.
15057 (ix86_expand_prologue): Modify stack re-alignment code and call
15058 emit_outlined_ms2sysv_save when appropriate.
15059 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15060 parameter rtx_insn *insn, which allows the function to be used to only
15061 generate the notes.
15062 (ix86_expand_epilogue): Modify validity checks of frame and stack
15063 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15064 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15065 * config/i386/predicates.md
15066 (save_multiple): New predicate.
15067 (restore_multiple): Likewise.
15068 * config/i386/sse.md
15069 (save_multiple<mode>): New pattern.
15070 (save_multiple_realign<mode>): Likewise.
15071 (restore_multiple<mode>): Likewise.
15072 (restore_multiple_and_return<mode>): Likewise.
15073 (restore_multiple_leave_return<mode>): Likewise.
15074 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15076 2017-05-14 Julia Koval <julia.koval@intel.com>
15078 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15079 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15080 (__builtin_ia32_xsetbv): New builtins.
15081 * config/i386/i386.c (ix86_expand_special_args_builtin):
15083 (ix86_expand_builtin): Special expand for new intrinsics.
15084 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15085 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15086 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15088 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15090 * cfganal.c (inverted_post_order_compute): Change argument type
15092 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15093 * df-core.c (rest_of_handle_df_initialize): Adjust.
15094 (rest_of_handle_df_finish): Likewise.
15095 (df_analyze_1): Likewise.
15096 (df_analyze): Likewise.
15097 (loop_inverted_post_order_compute): Change argument to be a vec *.
15098 (df_analyze_loop): Adjust.
15099 (df_get_n_blocks): Likewise.
15100 (df_get_postorder): Likewise.
15101 * df.h (struct df_d): Change field to be a vec.
15102 * lcm.c (compute_laterin): Adjust.
15103 (compute_available): Likewise.
15104 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15105 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15106 * tree-ssa-pre.c (compute_antic): Likewise.
15108 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15110 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15111 (depth_first_search::depth_first_search): Change structure init
15112 function to this constructor.
15113 (depth_first_search::add_bb): Rename function to this member.
15114 (depth_first_search::execute): Likewise.
15115 (flow_dfs_compute_reverse_finish): Adjust.
15117 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15119 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15120 (longest_simple_path): Likewise.
15121 * shrink-wrap.c (spread_components): Likewise.
15122 (disqualify_problematic_components): Likewise.
15123 (emit_common_heads_for_components): Likewise.
15124 (emit_common_tails_for_components): Likewise.
15125 (insert_prologue_epilogue_for_components): Likewise.
15127 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15129 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15132 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15134 * df-core.c (df_set_blocks): Start using auto_bitmap.
15135 (df_compact_blocks): Likewise.
15136 * df-problems.c (df_rd_confluence_n): Likewise.
15137 * df-scan.c (df_insn_rescan_all): Likewise.
15138 (df_process_deferred_rescans): Likewise.
15139 (df_update_entry_block_defs): Likewise.
15140 (df_update_exit_block_uses): Likewise.
15141 (df_entry_block_bitmap_verify): Likewise.
15142 (df_exit_block_bitmap_verify): Likewise.
15143 (df_scan_verify): Likewise.
15144 * lra-constraints.c (lra_constraints): Likewise.
15145 (undo_optional_reloads): Likewise.
15146 (lra_undo_inheritance): Likewise.
15147 * lra-remat.c (calculate_gen_cands): Likewise.
15148 (do_remat): Likewise.
15149 * lra-spills.c (assign_spill_hard_regs): Likewise.
15150 (spill_pseudos): Likewise.
15151 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15152 (bitmap_set_subtract_values): Likewise.
15154 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15156 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15157 management with auto_bitmap.
15158 (fix_inter_tick): Likewise.
15159 (fix_recovery_deps): Likewise.
15160 * ira.c (add_store_equivs): Likewise.
15161 (find_moveable_pseudos): Likewise.
15162 (split_live_ranges_for_shrink_wrap): Likewise.
15163 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15164 (rtx_reuse_manager::seen_def_p): Likewise.
15165 (rtx_reuse_manager::set_seen_def): Likewise.
15166 * print-rtl.h (class rtx_reuse_manager): Likewise.
15168 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15170 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15172 (migrate_btr_def): Likewise.
15173 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15174 * df-core.c (loop_post_order_compute): Likewise.
15175 (loop_inverted_post_order_compute): Likewise.
15176 * hsa-common.h: Likewise.
15177 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15178 * init-regs.c (initialize_uninitialized_regs): Likewise.
15179 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15180 (inline_small_functions): Likewise.
15181 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15182 * ira.c (combine_and_move_insns): Likewise.
15183 (build_insn_chain): Likewise.
15184 * loop-invariant.c (find_invariants): Likewise.
15185 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15186 * predict.c (tree_predict_by_opcode): Likewise.
15187 (predict_paths_leading_to): Likewise.
15188 (predict_paths_leading_to_edge): Likewise.
15189 (estimate_loops_at_level): Likewise.
15190 (estimate_loops): Likewise.
15191 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15192 (spread_components): Likewise.
15193 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15194 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15195 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15196 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15197 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15198 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15199 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15200 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15201 (mark_threaded_blocks): Likewise.
15202 (thread_through_all_blocks): Likewise.
15203 * tree-ssa.c (verify_ssa): Likewise.
15204 (execute_update_addresses_taken): Likewise.
15205 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15207 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15209 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15211 (post_order_compute): Likewise.
15212 (inverted_post_order_compute): Likewise.
15213 (pre_and_rev_post_order_compute_fn): Likewise.
15215 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15217 * genrecog.c (int_set::int_set): Explicitly construct our
15218 auto_vec base class.
15219 * vec.h (auto_vec::auto_vec): New constructor.
15221 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15223 * bitmap.h (class auto_bitmap): New constructor taking
15224 bitmap_obstack * argument.
15226 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15228 * bitmap.h (class auto_bitmap): Change type of m_bits to
15229 bitmap_head, and adjust ctor / dtor and member operators.
15231 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15233 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15234 when returned register mode doesn't match original mode.
15236 2017-05-12 Jeff Law <law@redhat.com>
15237 Jakub Jelinek <jakub@redhat.com>
15239 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15240 we look for cc setter after the compare-elim changes.
15241 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15242 within the vector to match what compare-elim now expects.
15243 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15244 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15246 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15247 after the compare-elim changes.
15248 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15249 the vector to match what compare-elim now expects.
15250 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15251 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15252 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15253 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15254 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15256 * config/visium/visium.c (single_set_and_flags): Fix where
15257 we look for cc setter after the compare-elim changes.
15258 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15259 with the vector to match what compare-elim now expects.
15260 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15261 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15262 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15263 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15264 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15265 (neg<mode>2_insn_set_overflow): Likewise.
15267 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15269 PR middle-end/79794
15270 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15271 maybe_expand_insn call, set ops[0].target. If still set after call,
15272 set alt_rtl. Add extra arg to recursive calls.
15273 (extract_bit_field): Add alt_rtl argument. Pass to
15275 * expmed.h (extract_bit_field): Fix prototype.
15276 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15277 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15278 to extract_bit_field_calls.
15279 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15280 Pass alt_rtl to extract_bit_field calls.
15281 * calls.c (store_unaligned_arguments_into_psuedos)
15282 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15283 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15285 * optabs.h (struct expand_operand): Add target bitfield.
15287 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15289 * compare-elim.c (try_eliminate_compare): Canonicalize
15290 operation with embedded compare to
15291 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15292 (set (reg) (operation)].
15294 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15296 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15299 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15300 cost of adding a carry flag for ADC instruction.
15301 [case MINUS]: Ignore the cost of subtracting a carry flag
15302 for SBB instruction.
15304 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15306 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15308 * config/rs6000/bmiintrin.h: New file.
15309 * config/rs6000/bmi2intrin.h: New file.
15310 * config/rs6000/x86intrin.h: New file.
15312 2017-05-12 Jeff Law <law@redhat.com>
15314 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15317 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15319 PR middle-end/80707
15320 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15322 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15324 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15325 early expansion of vector divide builtins.
15326 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15327 builtins identified as having unsigned arguments.
15329 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15331 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15332 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15333 expansion of vector logical operations (and, andc, or, xor,
15336 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15338 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15339 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15341 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15343 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15344 early GIMPLE expansion of vector multiplies.
15346 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15348 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15349 TARGET_HAVE_MOVT conditional.
15350 (movt splitter): Likewise.
15352 2017-05-12 Richard Biener <rguenther@suse.de>
15354 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15355 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15356 Fold all stmts not inplace.
15358 2017-05-12 Richard Biener <rguenther@suse.de>
15360 PR tree-optimization/80713
15361 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15362 inserted_exprs bit for not removed stmts.
15364 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15366 PR middle-end/69921
15367 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15368 parallelized" attribute for parallelized OpenACC kernels.
15369 * omp-offload.c (execute_oacc_device_lower): Use it.
15371 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15372 Set "oacc kernels" attribute.
15373 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15374 parameter. Adjust all users.
15375 (oacc_fn_attrib_kernels_p): Remove function.
15376 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15377 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15378 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15379 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15380 assert "oacc kernels" attribute is set.
15382 2017-05-11 Carl Love <cel@us.ibm.com>
15384 * config/rs6000/rs6000-c: Add support for built-in functions
15385 vector unsigned char vec_popcnt (vector signed char)
15386 vector unsigned char vec_popcnt (vector unsigned char)
15387 vector unsigned short vec_popcnt (vector signed short)
15388 vector unsigned short vec_popcnt (vector unsigned short)
15389 vector unsigned int vec_popcnt (vector signed int)
15390 vector unsigned int vec_popcnt (vector unsigned int)
15391 vector unsigned long long vec_popcnt (vector signed long long)
15392 vector unsigned long long vec_popcnt (vector unsigned long long)
15393 vector signed long long vec_slo (vector signed long long,
15394 vector signed char)
15395 vector signed long long vec_slo (vector signed long long,
15396 vector unsigned char)
15397 vector unsigned long long vec_slo (vector unsigned long long,
15398 vector signed char)
15399 vector unsigned long long vec_slo (vector unsigned long long,
15400 vector unsigned char)
15401 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15402 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15403 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15404 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15405 * doc/extend.texi: Update the built-in documentation file for the
15406 new built-in functions.
15408 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15410 * attribs.h (sorted_attr_string): Move machine independent
15411 functions for target clone support from the i386 port to common
15412 code. Rename ix86_function_versions to common_function_versions.
15413 Rename make_name to make_unique_name.
15414 (common_function_versions): Likewise.
15415 (make_unique_name): Likewise.
15416 (make_dispatcher_decl): Likewise.
15417 (is_function_default_version): Likewise.
15418 * attribs.c (attr_strcmp): Likewise.
15419 (sorted_attr_string): Likewise.
15420 (common_function_versions): Likewise.
15421 (make_unique_name): Likewise.
15422 (make_dispatcher_decl): Likewise.
15423 (is_function_default_version): Likewise.
15424 * config/i386/i386.c (attr_strcmp): Likewise.
15425 (sorted_attr_string): Likewise.
15426 (ix86_function_versions): Likewise.
15427 (make_name): Likewise.
15428 (make_dispatcher_decl): Likewise.
15429 (is_function_default_version): Likewise.
15430 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15432 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15435 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15436 Account for direct move costs for vec_construct of integer
15439 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15442 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15443 (UNSPEC_STX_ATOMIC): Ditto.
15444 (loaddi_via_sse): New insn.
15445 (storedi_via_sse): Ditto.
15446 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15447 Update corresponding peephole2 patterns.
15448 (atomic_storedi_fpu): Ditto.
15450 2017-05-11 Julia Koval <julia.koval@intel.com>
15452 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15453 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15455 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15456 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15457 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15459 2017-05-11 Nathan Sidwell <nathan@acm.org>
15461 * graphite-poly.c: Include dumpfile.h.
15463 * dumpfle.h (dump_function): Declare here ...
15464 * tree-dump.h (dump_function): ... not here.
15465 * dumpfile.c: #include tree-cfg.h.
15466 (dump_function): Move here from ...
15467 * tree-dump.c (dump_function): ... here.
15468 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15469 * graphite-poly.c: Don't include tree-dump.h.
15470 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15471 * print-tree.c: Likewise.
15472 * stor-layout.c: Likewise.
15473 * tree-nested.c: Likewise.
15475 * dumpfile.c (dump_start): Use TDF_FLAGS.
15476 (dump_enable_all): Fix TDF_KIND check thinko.
15478 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15480 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15481 array entries to represent two legal parameterizations of the
15482 overloaded __builtin_cmpb function, as represented by the
15483 P6_OV_BUILTIN_CMPB constant.
15484 (altivec_resolve_overloaded_builtin): Add special case handling
15485 for the __builtin_cmpb function, as represented by the
15486 P6_OV_BUILTIN_CMPB constant.
15487 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15488 (BU_P6_64BIT_2): New macro.
15489 (BU_P6_OVERLOAD_2): New macro
15490 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15491 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15492 (CMPB): Add overload support to represent both 32-bit and 64-bit
15493 compare-bytes function.
15494 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15495 support for TARGET_CMPB.
15496 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15497 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15498 documentation of the __builtin_cmpb overloaded built-in function.
15500 2017-05-11 Richard Biener <rguenther@suse.de>
15502 PR tree-optimization/80705
15503 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15504 bases are not vectorizable.
15506 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15508 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15509 when counting register pressure.
15511 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15513 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15514 (iv_ca_more_deps): Renamed to ...
15515 (iv_ca_compare_deps): ... this.
15516 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15518 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15520 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15522 (determine_group_iv_costs): ... here.
15523 (find_inv_vars_cb): Record inv var if it's not recorded before.
15525 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15527 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15528 (get_shiftadd_cost): Ditto.
15530 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15532 * tree-ssa-address.c: Include header file.
15533 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15535 (add_to_parts): Refactor.
15536 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15537 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15540 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15542 PR tree-optimization/53090
15543 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15545 (extract_cond_operands): Detect condition with IV on both sides
15546 and return COMP_IV_EXPR_2.
15547 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15548 (rewrite_use_compare): Simplify by removing call to function
15549 extract_cond_operands.
15551 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15553 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15554 (extract_cond_operands): Detect condition comparing against non-
15555 invariant bound and return appropriate enum value.
15556 (find_interesting_uses_cond): Update use of extract_cond_operands.
15557 Handle its return value accordingly.
15558 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15560 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15562 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15563 nonlinear iv_use computation in loop invariant sensitive way.
15565 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15567 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15568 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15570 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15572 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15573 (dump_cand): Support iv_cand.inv_exprs.
15574 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15576 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15579 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15581 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15583 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15584 as local function. Include necessary header files.
15585 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15587 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15589 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15591 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15593 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15594 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15595 RSHIFT_EXPR and BIT_NOT_EXPR.
15597 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15599 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15600 (adjust_setup_cost): New parameter supporting round up adjustment.
15601 (struct address_cost_data): Delete.
15602 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15603 (split_address_cost, ptr_difference_cost): Delete.
15604 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15605 (struct ainc_cost_data): New struct.
15606 (get_address_cost_ainc): New function.
15607 (get_address_cost, get_computation_cost): Reimplement.
15608 (determine_group_iv_cost_address): Record inv_expr for all uses of
15610 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15611 (iv_ca_has_deps): Reimplemented to ...
15612 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15614 (iv_ca_extend): Call iv_ca_more_deps.
15616 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15618 * tree-ssa-address.c (struct mem_address): Move to header file.
15619 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15620 * tree-ssa-address.h (struct mem_address): Move from C file.
15621 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15623 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15625 * tree-affine.h (aff_combination_type): New interface.
15626 (aff_combination_zero_p): Remove static.
15627 (aff_combination_const_p): New interface.
15628 (aff_combination_singleton_var_p): New interfaces.
15630 2017-05-11 Richard Biener <rguenther@suse.de>
15632 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15633 Skip unreachable blocks and destinations.
15634 (eliminate): Move stmt removal and fixup ...
15635 (fini_eliminate): ... here. Skip inserted exprs.
15636 (pass_pre::execute): Move fini_pre after fini_eliminate.
15637 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15638 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15639 PRE to get rid of dead code that has invalid SSA form and
15640 split critical edges again.
15642 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15644 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15646 2017-05-11 Richard Biener <rguenther@suse.de>
15648 * passes.c (execute_function_todo): Verify loops if they are
15649 said to be up-to-date.
15650 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15651 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15653 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15656 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15657 handle calling assemble_external ourself.
15660 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15661 modes with zero size. Enhance comment.
15663 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15665 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15666 built-ins for vec_xl and vec_xst with short and char pointer
15669 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15671 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15672 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15673 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15674 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15675 (_mm_maskz_min_round_ss): New intrinsics.
15676 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15677 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15678 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15679 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15680 (__builtin_ia32_minss_mask_round): New builtins.
15681 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15682 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15683 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15685 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15686 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15688 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15691 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15693 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15694 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15695 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15696 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15697 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15698 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15699 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15700 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15701 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15702 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15703 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15704 (__builtin_ia32_mulss_mask_round): New builtins.
15705 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15706 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15707 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15709 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15710 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15712 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15715 2017-05-10 Julia Koval <julia.koval@intel.com>
15717 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15718 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15719 (_mm256_setr_m128i): New intrinsics.
15721 2017-05-10 Julia Koval <julia.koval@intel.com>
15723 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15724 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15725 (_mm_maskz_rcp14_ss): New intrinsics.
15726 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15727 (__builtin_ia32_rcp14ss_mask): New builtins.
15728 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15730 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15732 PR tree-optimization/51513
15733 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15734 (assert_unreachable_fallthru_edge_p): Use it.
15735 (group_case_labels_stmt): Likewise.
15736 * tree-cfg.h: Prototype it.
15737 * stmt.c: Include cfghooks.h and tree-cfg.h.
15738 (emit_case_dispatch_table) <gap_label>: New local variable.
15739 Use it to fill dispatch table gaps.
15740 Test for default_label before updating probabilities.
15741 (expand_case) <default_label>: Remove unneeded initialization.
15742 Test for unreachable default case statement and remove its edge.
15743 Set default_label accordingly.
15744 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15746 2017-05-10 Carl Love <cel@us.ibm.com>
15748 * config/rs6000/rs6000-c: Add support for built-in functions
15749 vector signed char vec_neg (vector signed char)
15750 vector signed short int vec_neg (vector short int)
15751 vector signed int vec_neg (vector signed int)
15752 vector signed long long vec_neg (vector signed long long)
15753 vector float vec_neg (vector float)
15754 vector double vec_neg (vector double)
15755 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15757 * config/rs6000/altivec.h: Add define for vec_neg
15758 * doc/extend.texi: Update the built-in documentation for the
15759 new built-in functions.
15761 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15763 PR tree-optimization/77644
15764 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15766 2017-05-10 Nathan Sidwell <nathan@acm.org>
15768 * dumpfile.h (TDI_lang_all): New.
15769 (TDF_KIND): New. Renumber others
15770 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15772 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15774 (get_dump_file_name): Adjust suffix generation.
15775 (dump_enable_all): Use TDF_KIND.
15776 * doc/invoke.texi (-fdump-lang-all): Document.
15778 * dumpfile.h: Tabify.
15780 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15783 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15784 Move member access before delete.
15786 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15788 * tree-inline.c (expand_call_inline): Split block at stmt
15791 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15794 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15795 are now unused after splitting mov{sf,sd}_hardfloat.
15796 (f32_lr2): Likewise.
15797 (f32_lm): Likewise.
15798 (f32_lm2): Likewise.
15799 (f32_li): Likewise.
15800 (f32_li2): Likewise.
15801 (f32_lv): Likewise.
15802 (f32_sr): Likewise.
15803 (f32_sr2): Likewise.
15804 (f32_sm): Likewise.
15805 (f32_sm2): Likewise.
15806 (f32_si): Likewise.
15807 (f32_si2): Likewise.
15808 (f32_sv): Likewise.
15809 (f32_dm): Likewise.
15810 (f32_vsx): Likewise.
15811 (f32_av): Likewise.
15812 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15813 For movsf, order stores so the VSX stores occur before the GPR
15814 store which encourages the register allocator to use a traditional
15815 FPR instead of a GPR. For movsd, order the stores so that the GPR
15816 store comes before the VSX stores to allow the power6 to work.
15817 This is due to the power6 not having a 32-bit integer store
15818 instruction from a FPR.
15819 (movsf_hardfloat): Likewise.
15820 (movsd_hardfloat): Likewise.
15822 2017-05-09 Martin Sebor <msebor@redhat.com>
15824 PR translation/80280
15825 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15828 PR translation/80280
15829 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15830 data member added in r247778.
15831 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15833 2017-05-09 Nathan Sidwell <nathan@acm.org>
15835 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15837 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15840 2017-05-09 Marek Polacek <polacek@redhat.com>
15842 * doc/invoke.texi: Fix typo.
15844 2017-05-09 Richard Biener <rguenther@suse.de>
15846 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15847 (vrp_val_is_min): Likewise.
15848 (set_value_range_to_value): Likewise.
15849 (set_value_range_to_nonnegative): Likewise.
15850 (gimple_assign_nonzero_p): Likewise.
15851 (gimple_stmt_nonzero_p): Likewise.
15852 (vrp_int_const_binop): Likewise. Remove unreachable case.
15853 (adjust_range_with_scev): Adjust comments.
15854 (compare_range_with_value): Likewise.
15855 (extract_range_from_phi_node): Likewise.
15856 (test_for_singularity): Likewise.
15858 2017-05-09 Richard Biener <rguenther@suse.de>
15860 * tree-vrp.c (get_single_symbol): Add assert that we don't
15861 get overflowed constants as invariant part.
15862 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15863 checks. Use wi::cmp instead of recursing for integer constants.
15864 (compare_values): Just ignore whether we assumed undefined
15865 overflow instead of failing the compare.
15866 (extract_range_for_var_from_comparison_expr): Add comment before the
15867 TREE_NO_WARNING sets.
15868 (test_for_singularity): Likewise.
15869 (extract_range_from_comparison): Do not disable optimization
15870 when we assumed undefined overflow.
15871 (extract_range_basic): Remove init of unused var.
15873 2017-05-09 Richard Biener <rguenther@suse.de>
15875 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15876 (extract_range_from_multiplicative_op_1): Adjust.
15877 (extract_range_from_binary_expr_1): Use int_const_binop.
15879 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15882 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15883 rs6000_store_data_bypass_p in seven define_bypass directives and
15884 in several comments.
15885 * config/rs6000/rs6000-protos.h: Add prototype for
15886 rs6000_store_data_bypass_p function.
15887 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15888 function implements slightly different (rs6000-specific) semantics
15889 than store_data_bypass_p, returning false rather than aborting
15890 with assertion error when arguments do not satisfy the
15891 requirements of store data bypass.
15892 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15893 rs6000_store_data_bypass_p.
15895 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15897 * config/xtensa/xtensa-protos.h
15898 (xtensa_initial_elimination_offset): New declaration.
15899 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15900 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15901 macro definition, add case for FRAME_POINTER_REGNUM when
15902 FRAME_GROWS_DOWNWARD.
15903 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15904 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15905 xtensa_initial_elimination_offset.
15907 2017-05-08 Nathan Sidwell <nathan@acm.org>
15909 * doc/invoke.texi: Alphabetize -fdump options.
15911 2017-05-08 Martin Sebor <msebor@redhat.com>
15913 PR translation/80280
15914 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15916 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15918 * target.def (compute_frame_layout): New optional target hook.
15919 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15920 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15921 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15923 * reload1.c (verify_initial_elim_offsets): Likewise.
15924 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15925 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15926 (arm_get_frame_offsets): Split up into this ...
15927 (arm_compute_frame_layout): ... and this function.
15929 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15931 * config/aarch64/constraints.md (Usa): New constraint.
15932 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15934 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15936 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15937 with_multilib_list after it has been checked.
15939 2017-05-08 Richard Biener <rguenther@suse.de>
15941 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15942 (bitmap_set_subtract_values): Likewise.
15944 2017-05-08 Richard Biener <rguenther@suse.de>
15946 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15947 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15949 (gimple_stmt_nonzero_warnv_p): Rename to ...
15950 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15952 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15953 (extract_range_basic): Adjust, do not disable propagation on
15954 strict overflow sensitive simplification.
15955 (vrp_visit_cond_stmt): Likewise.
15957 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15959 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15960 body size unconditionally.
15962 2017-05-07 Jeff Law <law@redhat.com>
15965 2017-05-06 Jeff Law <law@redhat.com>
15966 PR tree-optimization/78496
15967 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15970 PR tree-optimization/78496
15971 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15972 (simplify_stmt_using_ranges): Call it.
15973 (vrp_dom_walker::before_dom_children): Extract equivalences
15974 from an ASSERT_EXPR with an equality comparison against a
15977 2017-05-06 Jeff Law <law@redhat.com>
15979 PR tree-optimization/78496
15980 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15983 PR tree-optimization/78496
15984 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15985 (simplify_stmt_using_ranges): Call it.
15986 (vrp_dom_walker::before_dom_children): Extract equivalences
15987 from an ASSERT_EXPR with an equality comparison against a
15990 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15992 * lra-constraints.c (lra_copy_reg_equiv): New function.
15993 (split_reg): Use it to copy equivalence information from the
15994 original register to the spill register.
15996 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15998 PR rtl-optimization/75964
15999 * simplify-rtx.c (simplify_const_relational_operation): Remove
16000 invalid handling of comparisons of integer ABS.
16002 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16004 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16005 initialize to zero.
16006 (init_regs): Remove declaration.
16007 (function_arg_advance_32): Initialize error_p as boolean variable.
16009 2017-05-05 Nathan Sidwell <nathan@acm.org>
16011 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16012 lines. Use for (;;).
16014 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16016 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16017 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16018 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16019 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16020 VF=2 that require versioning.
16022 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16024 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16027 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16029 * diagnostic.h (diagnostic_override_option_index): Convert from
16030 macro to inline function.
16032 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16034 * diagnostic.c (last_module_changed_p): New function.
16035 (set_last_module): New function.
16036 (diagnostic_report_current_module): Convert macro usage to
16037 the above functions.
16038 * diagnostic.h (diagnostic_context::last_module): Strengthen
16039 from const line_map * to const line_map_ordinary *.
16040 (diagnostic_last_module_changed): Delete macro.
16041 (diagnostic_set_last_module): Delete macro.
16043 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16045 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16046 with diagnostic_report_diagnostic.
16047 (diagnostic_n_impl_richloc): Likewise.
16048 * diagnostic.h (report_diagnostic): Delete macro.
16049 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16050 with diagnostic_report_diagnostic.
16051 * substring-locations.c (format_warning_va): Likewise.
16053 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16055 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16056 save/restor of format_spec. Move option-printing code to...
16057 (print_option_information): ...this new function, and
16058 reimplement by simply printing to the pretty_printer,
16059 rather than appending to the format string.
16061 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16063 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16064 handling logic into...
16065 (update_effective_level_from_pragmas): ...this new function.
16067 2017-05-04 Andrew Waterman <andrew@sifive.com>
16069 * config/riscv/riscv.opt (mstrict-align): New option.
16070 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16071 (SLOW_UNALIGNED_ACCESS): Define.
16072 (riscv_slow_unaligned_access): Declare.
16073 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16075 (riscv_slow_unaligned_access): New variable.
16076 (rocket_tune_info): Set slow_unaligned_access to true.
16077 (optimize_size_tune_info): Set slow_unaligned_access to false.
16078 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16079 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16080 (riscv_option_override): Set riscv_slow_unaligned_access.
16081 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16083 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16085 * config/riscv/riscv.md: Unify indentation.
16087 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16092 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16094 (extendsi<mode>2): Add support for doing sign extension via
16095 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16096 don't have ISA 3.0 instructions.
16097 (extendsi<mode>2 splitter): Likewise.
16098 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16099 generate the normal insns since SImode can now go in vector
16100 registers. Disallow the special UNSPECs needed for previous
16101 machines to hide SImode being used. Add new insns
16102 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16103 (fix_trunc<mode>si2_stfiwx): Likewise.
16104 (fix_trunc<mode>si2_internal): Likewise.
16105 (fixuns_trunc<mode>si2): Likewise.
16106 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16107 (fctiw<u>z_<mode>_smallint): Likewise.
16108 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16109 of floating point to 32-bit integer from doing a direct move to
16110 the GPR registers to do a store.
16111 (fctiwz_<mode>): Break long line.
16113 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16115 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16116 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16117 (addr_list, addr_offset_valid_p): New.
16118 (split_address_groups): Check offset validity with above function.
16119 (gt-tree-ssa-loop-ivopts.h): Include header file.
16121 2017-05-05 Nathan Sidwell <nathan@acm.org>
16123 * config.gcc (arm*-*-*): Add missing 'fi'.
16125 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16127 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16128 included in -fopt-info does not matter.
16129 * doc/optinfo.texi (-fopt-info): Fix description of default
16130 behavour. Explicitly say order of options included in -fopt-info
16133 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16135 * config.gcc: Allow combinations of aprofile and rmprofile values for
16136 --with-multilib-list.
16137 * config/arm/t-multilib: New file.
16138 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16139 variables. Remove setting of ISA and floating-point ABI in
16140 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16141 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16142 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16143 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16145 * config/arm/t-rmprofile: Likewise except for the matches changes.
16146 * doc/install.texi (--with-multilib-list): Document the combination of
16147 aprofile and rmprofile values and warn about pitfalls in doing that.
16149 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16151 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16152 (movdi_aarch64): Likewise.
16154 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16156 PR tree-optimization/80632
16157 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16159 (build_arrays): Initialize it for virtual phis.
16160 (fix_phi_nodes): Use it for virtual phis.
16162 PR tree-optimization/80558
16163 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16164 [x, y] op z into [x op, y op z] for op & or | if conditions
16167 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16168 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16171 * config/arm/arm.md (use_literal_pool): Remove.
16172 (64-bit immediate split): No longer takes cost into consideration
16173 if arm_disable_literal_pool is enabled.
16174 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16175 used when arm_disable_literal_pool is enabled.
16176 (arm_max_const_double_inline_cost): Remove use of
16177 arm_disable_literal_pool.
16178 (push_minipool_fix): Add assert.
16179 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16180 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16181 (no_literal_pool_sf_immediate): New.
16183 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16185 PR tree-optimization/80613
16186 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16187 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16189 2017-05-05 Richard Biener <rguenther@suse.de>
16191 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16193 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16195 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16196 of this flag from insn conditions due to removal from r247495.
16198 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16200 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16202 (arm_early_store_addr_dep_ptr): Likewise.
16203 * config/arm/aarch-common-protos.h
16204 (arm_early_load_addr_dep_ptr): Add prototype.
16205 (arm_early_store_addr_dep_ptr): Likewise.
16206 * config/arm/cortex-a53.md: Add new bypasses.
16208 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16210 * tree.c (next_type_uid): Change type to unsigned.
16211 (type_hash_canon): Decrement back next_type_uid if
16212 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16213 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16216 2017-05-04 Martin Sebor <msebor@redhat.com>
16218 * builtins.c: Fix a trivial typo in a comment.
16220 PR middle-end/79234
16221 * builtins.c (check_sizes): Adjust to handle reading past the end.
16222 Avoid printing excessive upper bound of ranges. Use %E to print
16223 tree nodes instead of converting them to %wu.
16224 (expand_builtin_memchr): New function.
16225 (compute_dest_size): Rename...
16226 (compute_objsize): ...to this.
16227 (expand_builtin_memcpy): Adjust.
16228 (expand_builtin_mempcpy): Adjust.
16229 (expand_builtin_strcat): Adjust.
16230 (expand_builtin_strcpy): Adjust.
16231 (check_strncat_sizes): Adjust.
16232 (expand_builtin_strncat): Adjust.
16233 (expand_builtin_strncpy): Adjust and simplify.
16234 (expand_builtin_memset): Adjust.
16235 (expand_builtin_bzero): Adjust.
16236 (expand_builtin_memcmp): Adjust.
16237 (expand_builtin): Handle memcmp.
16238 (maybe_emit_chk_warning): Check strncat just once.
16240 2017-05-04 Martin Sebor <msebor@redhat.com>
16242 PR preprocessor/79214
16243 PR middle-end/79222
16244 PR middle-end/79223
16245 * builtins.c (check_sizes): Add inlining context and issue
16246 warnings even when -Wno-system-headers is set.
16247 (check_strncat_sizes): Same.
16248 (expand_builtin_strncat): Same.
16249 (expand_builtin_memmove): New function.
16250 (expand_builtin_stpncpy): Same.
16251 (expand_builtin): Handle memmove and stpncpy.
16253 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16255 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16256 which is not used any more.
16258 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16260 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16262 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16264 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16265 (cortexa53_tunings): Likewise.
16266 (cortexa57_tunings): Likewise.
16267 (cortexa72_tunings): Likewise.
16268 (cortexa73_tunings): Likewise.
16270 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16272 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16273 Set loop alignment to 8.
16275 2017-05-04 Martin Sebor <msebor@redhat.com>
16277 PR translation/80280
16278 * builtins.c (expand_builtin_object_size): Add missing quoting to
16279 %D and like directives.
16280 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16281 (hsa_type_for_tree_type): Same.
16282 (verify_function_arguments): Same.
16283 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16284 * varasm.c (get_section): Same.
16287 2017-05-04 Martin Sebor <msebor@redhat.com>
16289 PR translation/80280
16290 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16292 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16294 * config/aarch64/aarch64.c (generic_addrcost_table):
16295 Change HI/TI mode setting.
16297 2017-05-04 Martin Jambor <mjambor@suse.cz>
16299 PR tree-optimization/80622
16300 * tree-sra.c (comes_initialized_p): New function.
16301 (build_accesses_from_assign): Only set write lazily when
16302 comes_initialized_p is false.
16303 (analyze_access_subtree): Use comes_initialized_p.
16304 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16305 instead of testing for PARM_DECL.
16307 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16309 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16310 constraint on operand 0 to allow more general addressing modes.
16311 Adjust output template.
16312 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16314 * config/aarch64/aarch64-protos.h
16315 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16316 * config/aarch64/constraints.md (Dp): New address constraint.
16317 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16320 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16322 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16323 update use of estimate_ipcp_clone_size_and_time.
16324 (estimate_local_effects): Update use of
16325 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16326 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16327 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16328 Return nonspecialized time.
16330 2017-05-04 Richard Biener <rguenther@suse.de>
16332 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16333 for the last VUSE which def dominates the PHI. Directly call
16335 (get_continuation_for_phi_1): Remove.
16337 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16339 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16340 to explain the use of truncating division. Cap the number of
16341 iterations to the maximum given by nb_iterations_upper_bound,
16344 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16346 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16347 * configure: Regenerate.
16348 * config.in: Regenerate.
16349 * config/i386/driver-mingw32.c: new file.
16350 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16351 * config.host: Link driver-mingw32.o on MinGW host.
16352 * doc/install.texi: Document new --enable-mingw-wildcard configure
16355 2017-05-04 Marek Polacek <polacek@redhat.com>
16357 PR tree-optimization/80612
16358 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16360 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16361 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16363 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16364 (movt splitter): Likewise.
16365 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16366 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16367 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16368 block for Thumb-1 with MOVT.
16369 (thumb2_legitimate_address_p): Move code block ...
16370 (can_avoid_literal_pool_for_label_p): ... into this new function.
16371 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16373 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16374 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16375 "M-profile targets with the MOVT instruction".
16377 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16379 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16380 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16381 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16383 2017-05-04 Martin Liska <mliska@suse.cz>
16385 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16386 variable cond_code.
16388 2017-05-04 Richard Biener <rguenther@suse.de>
16390 * tree.c (array_at_struct_end_p): Handle arrays at struct
16391 end with flexarrays more conservatively. Refactor and treat
16392 arrays of arrays or aggregates more strict. Fix
16393 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16394 * tree.c (array_at_struct_end_p): Adjust prototype.
16395 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16396 * gimple-fold.c (get_range_strlen): Likewise.
16397 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16399 2017-05-04 Richard Biener <rguenther@suse.de>
16401 PR tree-optimization/31130
16402 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16404 (supports_overflow_infinity): Likewise.
16405 (is_negative_overflow_infinity): Likewise.
16406 (is_positive_overflow_infinity): Likewise.
16407 (is_overflow_infinity): Likewise.
16408 (stmt_overflow_infinity): Likewise.
16409 (overflow_infinity_range_p): Likewise.
16410 (usable_range_p): Remove as always returning true.
16411 (make_overflow_infinity): Remove.
16412 (negative_overflow_infinity): Likewise.
16413 (positive_overflow_infinity): Likewise.
16414 (avoid_overflow_infinity): Likewise.
16415 (set_value_range): Adjust accordingly.
16416 (set_value_range_to_nonnegative): Likewise, remove now unused
16417 overflow_infinity arg.
16418 (vrp_operand_equal_p): Adjust.
16419 (update_value_range): Likewise.
16420 (range_int_cst_singleton_p): Likewise.
16421 (operand_less_p): Likewise.
16422 (compare_values_warnv): Likewise.
16423 (extract_range_for_var_from_comparison_expr): Likewise.
16424 (vrp_int_const_binop): Likewise.
16425 (zero_nonzero_bits_from_vr): Likewise.
16426 (extract_range_from_multiplicative_op_1): Likewise.
16427 (extract_range_from_binary_expr_1): Likewise.
16428 (extract_range_from_unary_expr): Likewise.
16429 (extract_range_from_comparison): Likewise.
16430 (extract_range_basic): Likewise.
16431 (adjust_range_with_scev): Likewise.
16432 (compare_ranges): Likewise.
16433 (compare_range_with_value): Likewise.
16434 (dump_value_range): Likewise.
16435 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16437 (simplify_cond_using_ranges): Adjust.
16439 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16441 * brig-builtins.def: Added a builtin for class_f64.
16442 * builtin-types.def: Added a builtin type needed by class_f64.
16444 2017-05-03 Jason Merrill <jason@redhat.com>
16446 * timevar.def: Add TV_CONSTEXPR.
16448 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16450 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16452 2017-05-03 Martin Jambor <mjambor@suse.cz>
16454 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16455 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16456 * ipa-cp.c (ipcp_propagate_stage): Do not call
16457 ipa_update_after_lto_read.
16458 * ipa-inline.c (ipa_inline): Likewise.
16460 2017-05-03 Martin Jambor <mjambor@suse.cz>
16462 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16463 tag. Added a default constructor and a destructor.
16464 (ipa_edge_args_sum_t): New class;
16465 (ipa_edge_args_sum): Declare.
16466 (ipa_edge_args_vector): Remove declaration.
16467 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16468 (ipa_free_edge_args_substructures): Remove declaration.
16469 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16470 (ipa_edge_args_info_available_for_edge_p): Likewise.
16471 * ipa-prop.c (ipa_edge_args_vector): Removed.
16472 (edge_removal_hook_holder): Likewise.
16473 (edge_duplication_hook_holder): Likewise.
16474 (ipa_edge_args_sum): New variable.
16475 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16476 ipa_edge_args_vector.
16477 (ipa_free_edge_args_substructures): Likewise.
16478 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16479 ipa_edge_args_vector.
16480 (ipa_edge_removal_hook): Turned into method
16481 ipa_edge_args_sum_t::remove.
16482 (ipa_edge_duplication_hook): Turned into method
16483 ipa_edge_args_sum_t::duplicate.
16484 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16485 registering edge hooks.
16486 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16487 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16488 ipa_edge_args_sum instead of ipa_edge_args_vector.
16489 * ipa-profile.c (ipa_profile): Likewise.
16491 2017-05-03 Martin Jambor <mjambor@suse.cz>
16493 * symbol-summary.h (function_summary): New method exists.
16494 (function_summary::symtab_removal): Deallocate through release.
16495 (call_summary): New class.
16496 (gt_ggc_mx): New overload.
16497 (gt_pch_nx): Likewise.
16498 (gt_pch_nx): Likewise.
16500 2017-05-03 Jeff Law <law@redhat.com>
16502 PR tree-optimization/78496
16503 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16504 from simplify_cond_using_ranges. Split off code to walk
16505 backwards through casts into ...
16506 (simplify_cond_using_ranges_2): New function.
16507 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16508 (execute_vrp): After identifying jump threads, call
16509 simplify_cond_using_ranges_2.
16511 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16514 * ipa-inline.h (inline_summary): Add ctor.
16515 (create_ggc): Do not use ggc_cleared_alloc.
16517 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16518 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16520 * gcc.c (handle_braces): Support escaping in switch matching
16522 * doc/invoke.texi (Spec Files): Document it.
16523 Remove superfluous @code markup in items.
16525 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16527 * diagnostic-show-locus.c (struct column_range): New struct.
16528 (get_affected_columns): New function.
16529 (get_printed_columns): New function.
16530 (struct correction): New struct.
16531 (correction::ensure_capacity): New function.
16532 (correction::ensure_terminated): New function.
16533 (struct line_corrections): New struct.
16534 (line_corrections::~line_corrections): New dtor.
16535 (line_corrections::add_hint): New function.
16536 (layout::print_trailing_fixits): Reimplement in terms of the new
16538 (selftest::test_overlapped_fixit_printing): New function.
16539 (selftest::diagnostic_show_locus_c_tests): Call it.
16541 2017-05-03 Nathan Sidwell <nathan@acm.org>
16543 Canonicalize canonical type hashing
16544 * tree.h (type_hash_canon_hash): Declare.
16545 * tree.c (type_hash_list, attribute_hash_list): Move into
16546 type_hash_canon_hash.
16547 (build_type_attribute_qual_variant): Break out hash code calc into
16548 type_hash_canon_hash.
16549 (type_hash_canon_hash): New. Generic type hash computation.
16550 (build_range_type_1, build_array_type_1, build_function_type,
16551 build_method_type_directly, build_offset_type, build_complex_type,
16552 make_vector_type): Call it.
16554 2017-05-03 Richard Biener <rguenther@suse.de>
16556 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16557 When all DRs have unknown misaligned do not always peel
16558 when there is a store but apply the same costing model as if
16559 there were only loads.
16561 2017-05-03 Richard Biener <rguenther@suse.de>
16564 PR tree-optimization/80492
16565 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16566 compare_base_decls returning dont-know properly.
16568 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16570 * config/arm/iterators.md (CCSI): New mode iterator.
16571 (arch): New mode attribute.
16572 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16573 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16574 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16575 code iterator for success result mode.
16576 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16577 the corresponding new insn generators.
16579 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16582 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16583 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16585 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16587 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16588 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16589 (DDR_A): Wrap DDR argument in brackets.
16590 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16591 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16592 (DDR_REVERSED_P): Likewise.
16594 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16596 PR tree-optimization/79472
16597 * tree-switch-conversion.c (struct switch_conv_info): Add
16598 contiguous_range and default_case_nonstandard fields.
16599 (collect_switch_conv_info): Compute contiguous_range and
16600 default_case_nonstandard fields, don't clear final_bb if
16601 contiguous_range and only the default case doesn't have the required
16603 (check_all_empty_except_final): Set default_case_nonstandard instead
16604 of failing if contiguous_range and the default case doesn't have empty
16606 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16607 and only the default case doesn't have the required constants. Skip
16609 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16610 if default_case_nonstandard.
16611 (build_constructors): Build constant 1 just once. Assert that default
16612 values aren't inserted in between cases if contiguous_range. Skip
16614 (build_arrays): Skip virtual phis.
16615 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16616 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16617 Handle virtual phis.
16618 (gen_inbound_check): Handle default_case_nonstandard case.
16619 (process_switch): Adjust check_final_bb caller. Call
16620 gather_default_values with the first non-default case instead of
16621 default case if default_case_nonstandard.
16623 2017-05-02 Nathan Sidwell <nathan@acm.org>
16625 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16626 check. Fix formatting.
16628 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16630 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16631 errors when comparing specialized and unspecialized times.
16633 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16635 * diagnostic-show-locus.c
16636 (layout::should_print_annotation_line_p): Make private.
16637 (layout::print_annotation_line): Make private.
16638 (layout::annotation_line_showed_range_p): Make private.
16639 (layout::show_ruler): Make private.
16640 (layout::print_source_line): Make private. Pass in line and
16641 line_width, rather than calling location_get_source_line. Drop
16643 (layout::print_leading_fixits): New method.
16644 (layout::print_any_fixits): Rename to...
16645 (layout::print_trailing_fixits): ...this, and make private.
16646 Don't print newline fixits.
16647 (diagnostic_show_locus): Move logic for printing one row into...
16648 (layout::print_line): ...this new function. Move the
16649 location_get_source_line call and error-handling from
16650 print_source_line to here. Call print_leading_fixits, and rename
16651 print_any_fixits to print_trailing_fixits.
16652 (selftest::test_fixit_insert_containing_newline): Update now that
16653 newlines are partially supported.
16654 (selftest::test_fixit_insert_containing_newline_2): New test.
16655 (selftest::test_fixit_replace_containing_newline): Update comments.
16656 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16657 * edit-context.c (class added_line): New class.
16658 (class edited_line): Describe newline handling in comment.
16659 (edited_line::actually_edited_p): New method.
16660 (edited_line::print_content): Delete redundant decl.
16661 (edited_line::m_predecessors): New field.
16662 (edited_file::print_content): Call edited_line::print_content.
16663 (edited_file::print_diff): Update to support newlines.
16664 (edited_file::print_diff_hunk): Likewise.
16665 (edited_file::print_run_of_changed_lines): New function.
16666 (edited_file::print_diff_line): Convert to...
16667 (print_diff_line): ...this.
16668 (edited_file::get_effective_line_count): New function.
16669 (edited_line::edited_line): Initialize new field m_predecessors.
16670 (edited_line::~edited_line): Clean up m_predecessors.
16671 (edited_line::apply_fixit): Handle newlines.
16672 (edited_line::get_effective_line_count): New function.
16673 (edited_line::print_content): New function.
16674 (edited_line::print_diff_lines): New function.
16675 (selftest::test_applying_fixits_insert_containing_newline): New
16677 (selftest::test_applying_fixits_replace_containing_newline): New
16679 (selftest::insert_line): New function.
16680 (selftest::test_applying_fixits_multiple_lines): Add example of
16682 (selftest::edit_context_c_tests): Call the new tests.
16684 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16686 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16687 parameter cand. Update dump information.
16688 (get_computation_cost): Update uses.
16690 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16692 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16693 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16694 (get_computation_at, rewrite_use_address): Update use of
16695 get_computation_aff.
16697 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16699 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16700 (get_computation): Delete.
16701 (get_computation_cost): Implement like get_computation_cost_at.
16702 Use get_computation_at.
16703 (get_computation_cost_at): Delete.
16704 (rewrite_use_nonlinear_expr): Use get_computation_at.
16705 (rewrite_use_compare, remove_unused_ivs): Ditto.
16707 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16709 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16711 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16713 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16714 (ivopts_global_cost_for_size): Rename parameter and update uses.
16715 (iv_ca_recount_cost): Update uses.
16716 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16717 candidates seperately in n_invs and n_cands.
16718 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16720 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16722 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16723 (find_inv_vars_cb): New.
16724 (find_depends): Renamed to ...
16725 (find_inv_vars): ... this.
16726 (add_candidate_1, force_var_cost): Call find_inv_vars.
16727 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16729 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16731 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16732 inv_vars. Add inv_exprs.
16733 (struct iv_cand): Rename depends_on to inv_vars.
16734 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16735 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16736 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16737 (dump_cand): Dump inv_vars.
16738 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16739 (record_invariant, find_depends, add_candidate_1): Ditto.
16740 (set_group_iv_cost, force_var_cost): Ditto.
16741 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16742 (get_computation_cost_at, get_computation_cost): Ditto.
16743 (determine_group_iv_cost_generic): Ditto.
16744 (determine_group_iv_cost_address): Ditto.
16745 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16746 (determine_group_iv_costs): Ditto.
16747 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16748 (iv_ca_set_remove_invariants): Renamed to ...
16749 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16750 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16751 (iv_ca_set_add_invariants): Renamed to ...
16752 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16753 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16754 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16755 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16756 (create_new_ivs): Remove useless dump.
16758 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16760 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16762 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16763 (iv_ca_set_no_cp, create_new_iv): Ditto.
16765 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16767 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16769 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16771 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16772 function tree_check2.
16774 2017-05-02 Martin Liska <mliska@suse.cz>
16776 * doc/gcov.texi: Add missing preposition.
16777 * gcov.c (function_info::function_info): Properly fill up
16778 all member variables.
16780 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16782 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16784 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16786 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16788 2017-05-02 Martin Liska <mliska@suse.cz>
16791 * lto-streamer-in.c (lto_read_tree_1): Remove
16792 LTO_STREAMER_DEBUG.
16793 * lto-streamer.c (struct tree_hash_entry): Likewise.
16794 (struct tree_entry_hasher): Likewise.
16795 (tree_entry_hasher::hash): Likewise.
16796 (tree_entry_hasher::equal): Likewise.
16797 (lto_streamer_init): Likewise.
16798 (lto_orig_address_map): Likewise.
16799 (lto_orig_address_get): Likewise.
16800 (lto_orig_address_remove): Likewise.
16801 * lto-streamer.h: Likewise.
16802 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16803 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16805 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16807 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16808 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16809 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16810 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16811 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16812 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16813 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16814 (mm_maskz_sub_ss): New intrinsics.
16815 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16816 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16817 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16818 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16819 (__builtin_ia32_subss_mask_round): New builtins.
16820 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16821 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16822 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16824 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16825 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16827 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16830 2017-05-02 Martin Jambor <mjambor@suse.cz>
16832 PR tree-optimization/78687
16833 * tree-sra.c (access): New field parent.
16834 (process_subtree_disqualification): New function.
16835 (disqualify_candidate): Call it.
16836 (build_accesses_from_assign): Reset write flag if creating an
16838 (build_access_subtree): Fill in parent field and also prpagate
16839 down grp_write flag.
16840 (create_artificial_child_access): New parameter set_grp_write, set
16841 grp_write to its value.
16842 (propagate_subaccesses_across_link): Also propagate grp_write flag
16844 (propagate_all_subaccesses): Push the closest parent back to work
16845 queue if add_access_to_work_queue returned true.
16847 2017-05-02 Richard Biener <rguenther@suse.de>
16849 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16850 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16851 -fstrict-overflow documentation.
16852 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16853 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16854 flag_strict_overflow.
16855 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16856 * lto-opts.c (lto_write_options): Do not stream it.
16857 * lto-wrapper.c (merge_and_complain): Do not handle it.
16858 * opts.c (default_options_table): Do not set -fstrict-overflow.
16859 (finish_options): Likewise do not clear it when sanitizing.
16860 * simplify-rtx.c (simplify_const_relational_operation): Do not
16861 test flag_strict_overflow.
16863 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16865 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16866 using enabled attribute.
16867 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16868 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16869 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16870 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16871 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16872 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16873 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16874 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16875 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16876 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16878 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16880 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16882 2017-05-02 Richard Biener <rguenther@suse.de>
16884 PR tree-optimization/80591
16886 2017-04-10 Richard Biener <rguenther@suse.de>
16888 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16891 2017-05-02 Richard Biener <rguenther@suse.de>
16893 PR tree-optimization/80549
16894 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16895 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16896 headers if they do not have a preheader.
16898 2017-05-02 Martin Liska <mliska@suse.cz>
16901 * common.opt: Fix typo.
16902 * doc/invoke.texi: Likewise.
16904 2017-05-01 Jan Beulich <jbeulich@suse.com>
16906 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16907 swapping, add (x,x,m,x,n) alternative.
16909 2017-05-01 Nathan Sidwell <nathan@acm.org>
16911 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16912 unnecessary unadjusted_alignment check.
16914 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16917 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16919 * gimplify.c (gimplify_cilk_detach): New function.
16920 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16921 * tree-core.h: Document EXPR_CILK_SPAWN.
16922 * tree.h (EXPR_CILK_SPAWN): Define.
16924 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16926 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16927 to use new fixit_hint representation, using the "replace" logic.
16928 (get_line_span_for_fixit_hint): Likewise.
16929 (layout::print_any_fixits): Likewise.
16930 (selftest::test_one_liner_many_fixits): Rename to...
16931 (selftest::test_one_liner_many_fixits_1): ...this, and update
16932 comment and expected output to reflect that the multiple fix-it
16933 hints are now consolidated into one insertion.
16934 (selftest::test_one_liner_many_fixits_2): New test.
16935 (selftest::test_diagnostic_show_locus_one_liner): Update for
16937 (selftest::test_fixit_consolidation): Update for fix-it API
16939 * diagnostic.c (print_parseable_fixits): Likewise.
16940 * edit-context.c (edited_line::m_line_events): Convert from
16941 auto_vec <line_event *> to auto_vec <line_event>.
16942 (class line_event): Convert from abstract base class to a concrete
16943 class, taking over the role of replace_event.
16944 (class insert_event): Delete.
16945 (class replace_event): Rename to class line_event. Convert to
16947 (edit_context::add_fixits): Reimplement.
16948 (edit_context::apply_insert): Delete.
16949 (edit_context::apply_replace): Rename to...
16950 (edit_context::apply_fixit): ...this. Convert to half-open range.
16951 (edited_file::apply_insert): Delete.
16952 (edited_file::apply_replace): Rename to...
16953 (edited_file::apply_fixit): ...this.
16954 (edited_line::~edited_line): Drop deletion of events.
16955 (edited_line::apply_insert): Delete.
16956 (edited_line::apply_replace): Rename to...
16957 (edited_line::apply_fixit): ...this. Convert to half-open range.
16958 Update for change to type of m_line_events.
16959 * edit-context.h (edit_context::apply_insert): Delete.
16960 (edit_context::apply_replace): Rename to...
16961 (edit_context::apply_fixit): ...this.
16963 2017-05-01 Martin Sebor <msebor@redhat.com>
16965 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16968 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16971 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16972 __get_cpuid_max returns 0.
16973 (__get_cpuid_count): Ditto.
16975 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16977 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16978 replacement expression is another instance of one of its arguments.
16980 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16983 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16984 check for stack push/pop autoinc.
16985 * config/i386/i386.c (ix86_agi_dependent): Return false
16986 if the only reason why modified_in_p returned true is that
16987 addr is SP based and set_insn is a push or pop.
16989 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16991 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16994 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16997 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16998 (account_size_time): Use two predicates - exec_pred and
17000 (evaluate_conditions_for_known_args): Compute both clause and
17002 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17003 (inline_summary_t::duplicate): Update.
17004 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17006 (compute_inline_parameters): Likewise.
17007 (estimate_edge_size_and_time): Update caluclation of time.
17008 (estimate_node_size_and_time): Compute both time and nonspecialized
17010 (estimate_ipcp_clone_size_and_time): Update.
17011 (inline_merge_summary): Update.
17012 (do_estimate_edge_time): Update.
17013 (do_estimate_edge_size): Update.
17014 (do_estimate_edge_hints): Update.
17015 (inline_read_section, inline_write_summary): Stream both new predicates.
17016 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17018 (compute_inlined_call_time): Cleanup.
17019 (big_speedup_p): Update.
17020 (edge_badness): Update.
17021 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17022 (size_time_entry): Replace predicate by exec_predicate and
17023 nonconst_predicate.
17024 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17025 (estimate_edge_time): Return also nonspec_time.
17026 (reset_edge_growth_cache): Update.
17028 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17030 PR rtl-optimization/80491
17031 * ifcvt.c (noce_process_if_block): When looking for x setter
17032 with missing else_bb, don't check only the insn right before
17033 cond_earliest, but look for the last insn that x is modified in
17034 within the same bb.
17036 PR rtl-optimization/80491
17037 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17039 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17041 PR tree-optimization/80487
17042 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17044 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17046 PR tree-optimization/79697
17047 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17048 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17049 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17051 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17052 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17054 2017-04-28 Martin Sebor <msebor@redhat.com>
17056 PR tree-optimization/80523
17057 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17058 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17060 (maybe_warn, format_directive, parse_directive): Use new functions.
17061 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17063 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17065 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17067 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17069 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17070 target_header_dir): Set correctly.
17071 * configure: Regenerated.
17072 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17073 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17074 instead of SYSTEM_HEADER_DIR.
17076 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17078 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17079 (estimate_local_effects): Likewise.
17080 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17081 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17082 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17083 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17084 do_estimate_edge_time, estimate_edge_time): Likewise.
17085 * ipa-inline-analysis.c (estimate_node_size_and_time,
17086 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17087 (estimate_time_after_inlining): Remove.
17089 2017-04-28 Martin Liska <mliska@suse.cz>
17091 * doc/gcov.texi: Enhance documentation of gcov.
17093 2017-04-28 Martin Liska <mliska@suse.cz>
17095 * doc/gcov.texi: Sort options in alphabetic order.
17096 * doc/gcov-dump.texi: Likewise.
17097 * doc/gcov-tool.texi: Likewise.
17098 * gcov.c (print_usage): Likewise.
17099 * gcov-dump.c (print_usage): Likewise.
17100 * gcov-tool.c (print_merge_usage_message): Likewise.
17101 (print_rewrite_usage_message): Likewise.
17102 (print_overlap_usage_message): Likewise.
17104 2017-04-28 Martin Liska <mliska@suse.cz>
17106 PR gcov-profile/53915
17107 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17109 2017-04-28 Martin Liska <mliska@suse.cz>
17111 PR gcov-profile/79891
17112 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17113 is marked by compiler as living on a line.
17114 (get_cycles_count): Remove usage of the union.
17115 (output_intermediate_file): Likewise.
17116 (find_source): Fix GNU coding style.
17117 (accumulate_line_counts): Remove old non-all block mode.
17118 (output_lines): Remove usage of the union.
17119 * profile.c (output_location): Include all BBs, even if
17120 belonging to a same line (and file) as a previous BB.
17122 2017-04-28 Martin Liska <mliska@suse.cz>
17124 * gcov.c (process_args): Handle new argument 'w'.
17125 (read_graph_file): Assign ID to BBs.
17126 (output_branch_count): Display BB # if verbose flag is set.
17127 (output_lines): Likewise for arcs.
17128 (print_usage): Add '--verbose' option help.
17129 * doc/gcov.texi: Document --verbose (-w) option.
17131 2017-04-28 Martin Liska <mliska@suse.cz>
17133 * gcov.c (struct block_location_info): New struct.
17134 (process_file): Fill up the new structure.
17135 (read_graph_file): Replace usage of encoding by the newly added
17137 (add_line_counts): Likewise.
17138 (accumulate_line_counts): Remove usage of the union.
17139 (function_info::function_info): New function.
17140 (function_info::~function_info): Likewise.
17141 (process_file): Call delete instead of release_function.
17142 (release_function): Release the function.
17143 (release_structures): Call delete instead of release_function.
17144 (solve_flow_graph): Replace usage of num_blocks.
17145 (find_exception_blocks): Likewise.
17146 (output_lines): Fix GNU coding style.
17148 2017-04-28 Martin Liska <mliska@suse.cz>
17151 * coverage.c (coverage_remove_note_file): New function.
17152 * coverage.h: Declare the function.
17153 * toplev.c (finalize): Clean if an error has been seen.
17155 2017-04-28 Martin Liska <mliska@suse.cz>
17157 PR gcov-profile/80031
17158 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17159 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17160 * gcov.c (read_graph_file): Read just number of blocks.
17161 * profile.c (branch_prob): Do not stream 0 flags per a basic
17164 2017-04-28 Martin Liska <mliska@suse.cz>
17166 * gcov-dump.c (tag_*): Add new argument to declarations.
17167 (dump_gcov_file): Likewise.
17168 (tag_blocks): Add and use new argument depth.
17169 (tag_arcs): Likewise.
17170 (tag_lines): Likewise.
17171 (tag_counters): Likewise.
17172 (tag_summary): Likewise.
17173 (dump_working_sets): Use depth to do a proper indentation.
17175 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17178 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17179 * symtab.c (symtab_node::debug_symtab): Move definition here.
17181 2017-04-28 Richard Biener <rguenther@suse.de>
17183 * lto-streamer.h (LTO_major_version): Bump to 7.
17185 2017-04-28 Richard Biener <rguenther@suse.de>
17187 * tree-vrp.c (assert_info): New struct.
17188 (add_assert_info): New helper.
17189 (register_edge_assert_for_2): Refactor to add asserts to a vector
17191 (register_edge_assert_for_1): Likewise.
17192 (register_edge_assert_for): Likewise.
17193 (finish_register_edge_assert_for): New helper actually registering
17194 asserts where live on edge.
17195 (find_conditional_asserts): Adjust.
17196 (find_switch_asserts): Likewise.
17197 (evrp_dom_walker::try_find_new_range): Generalize.
17198 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17200 2017-04-27 Marek Polacek <polacek@redhat.com>
17203 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17204 arg10 and arg11 to itype.
17206 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17208 * doc/extend.texi (Object Size Checking): Improve grammar.
17210 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17213 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17214 that the logic for permitting reciprocal estimates matches that
17217 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17220 * tree.c (type_cache_hasher::equal): Only compare
17221 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17222 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17223 non-aggregate element types.
17224 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17225 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17227 2017-04-27 Richard Biener <rguenther@suse.de>
17229 PR middle-end/80533
17230 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17231 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17232 keeping a reference to a trailing array.
17234 2017-04-27 Richard Biener <rguenther@suse.de>
17236 PR middle-end/80539
17237 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17238 being in loop-closed SSA form conservatively.
17239 (chrec_fold_multiply_poly_poly): Likewise.
17241 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17243 PR middle-end/79665
17244 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17245 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17247 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17250 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17251 (aarch64_function_arg_alignment): Return unsigned int again, but still
17252 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17253 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17254 Don't emit -Wpsabi note.
17255 (aarch64_function_arg_boundary): Likewise.
17256 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17259 2017-04-26 Nathan Sidwell <nathan@acm.org>
17261 * tree.h (crc32_unsigned_n): Declare.
17262 (crc32_unsigned, crc32_unsigned): Make inline.
17263 * tree.c (crc32_unsigned_bits): Replace with ...
17264 (crc32_unsigned_n): ... this.
17265 (crc32_unsigned, crc32_byte): Remove.
17266 (crc32_string): Remove unnecessary braces.
17268 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17270 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17271 * ipa-inline-analysis.c (MAX_TIME): Remove.
17272 (account_size_time): Use sreal for time.
17273 (dump_inline_summary): Update.
17274 (estimate_function_body_sizes): Update.
17275 (estimate_edge_size_and_time): Update.
17276 (estimate_calls_size_and_time): Update.
17277 (estimate_node_size_and_time): Update.
17278 (inline_merge_summary): Update.
17279 (inline_update_overall_summary): Update.
17280 (estimate_time_after_inlining): Update.
17281 (inline_read_section): Update.
17282 (inline_write_summary): Update.
17283 * ipa-inline.c (compute_uninlined_call_time): Update.
17284 (compute_inlined_call_time): Update.
17285 (recursive_inlining): Update.
17286 (inline_small_functions): Update.
17287 (dump_overall_stats): Update.
17288 * ipa-inline.h: Include sreal.h.
17289 (size_time_entry): Turn time to sreal.
17290 (inline_summary): Turn self_time nad time to sreal.
17292 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17294 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17296 (sreal::stream_out, sreal::stream_in): New.
17297 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17299 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17301 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17304 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17307 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17309 (dimode_scalar_chain::compute_convert_gain): Ditto.
17310 (dimode_scalar_chain::make_vector_copies): Ditto.
17311 (dimode_scalar_chain::convert_reg): Ditto.
17312 (dimode_scalar_chain::convert_insn): Ditto.
17313 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17314 (VI248_AVX512BW_1): New mode iterator.
17315 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17316 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17319 2017-04-25 Martin Sebor <msebor@redhat.com>
17321 PR tree-optimization/80497
17322 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17323 constants are representable in HOST_WIDE_INT.
17324 (parse_directive): Ditto.
17326 2017-04-25 Martin Sebor <msebor@redhat.com>
17329 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17330 (new_zero_array): Adjust signature.
17331 (dom_info::dom_init): Used unsigned rather that size_t.
17332 (dom_info::dom_info): Same.
17334 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17335 Jakub Jelinek <jakub@redhat.com>
17338 * config/arm/arm.c: Include gimple.h.
17339 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17340 returns negative, increment ncrn only if it returned positive.
17341 (arm_needs_doubleword_align): Return int instead of bool,
17342 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17343 members, but if there is any such non-FIELD_DECL
17344 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17345 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17346 returns negative, increment nregs only if it returned positive.
17347 (arm_setup_incoming_varargs): Likewise.
17348 (arm_function_arg_boundary): Emit -Wpsabi note if
17349 arm_needs_doubleword_align returns negative, return
17350 DOUBLEWORD_ALIGNMENT only if it returned positive.
17352 2017-04-25 Marek Polacek <polacek@redhat.com>
17355 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17356 first argument to type.
17358 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17361 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17362 type checks to test for compatibility instead of equality.
17364 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17365 Jakub Jelinek <jakub@redhat.com>
17368 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17370 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17371 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17372 the alignment computation, but return their maximum in warn_alignment.
17373 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17374 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17376 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17377 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17380 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17382 * config/arc/simdext.md (dmpyh): Fix typo.
17384 2017-04-25 Richard Biener <rguenther@suse.de>
17386 PR tree-optimization/80492
17387 * alias.c (compare_base_decls): Handle registers with asm
17388 specification conservatively.
17389 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17390 compare_base_decls returning dont-know properly.
17392 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17394 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17395 (legitimate_offset_address_p): New function.
17396 (arc_legitimate_address_p): Use above function.
17398 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17400 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17402 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17404 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17405 ACCH registers whenever they are available.
17407 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17409 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17410 double regs fix when not used.
17412 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17414 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17416 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17417 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17419 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17421 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17422 when emitting short ADD instructions.
17424 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17426 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17428 (cmpsi_cc_c_insn): Likewise.
17429 (cbranchsi4_scratch): Compute proper instruction length using
17430 compact_hreg_operand.
17431 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17433 2017-04-25 Richard Biener <rguenther@suse.de>
17435 PR middle-end/80509
17436 * passes.c (pass_manager::pass_manager): Initialize
17437 m_name_to_pass_map.
17439 2017-04-25 Richard Biener <rguenther@suse.de>
17441 PR tree-optimization/79201
17442 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17444 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17447 * config/s390/vector.md: Split MEM->GPR vector moves for
17448 non-s_operand addresses.
17450 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17453 * config/s390/predicates.md (reload_const_wide_int_operand): New
17455 * config/s390/s390.md ("movti"): Remove d/P alternative.
17456 ("movti_bigconst"): New pattern definition.
17458 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17461 * s390-protos.h (s390_expand_cs_hqi): Removed.
17462 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17463 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17464 modes as well as CCZ1mode and CCZmode.
17465 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17466 signature of s390_emit_compare_and_swap.
17467 (s390_expand_cs_hqi): Likewise, make static.
17468 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17469 compare-and-swap, in some cases.
17470 (s390_expand_cs): Wrapper function.
17471 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17473 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17474 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17475 patterns for small and large integers. Forbid symref memory operands.
17476 Move expander to s390.c. Require cc register.
17477 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17478 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17479 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17480 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17481 symref memory operands. Remove CC mode and call s390_match_ccmode
17483 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17485 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17487 * config/s390/s390.md (define_peephole2): New peephole to help
17488 combining the load-and-test pattern with volatile memory.
17490 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17492 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17493 with CCZmode for TARGET_Z196.
17495 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17497 PR rtl-optimization/80501
17498 * combine.c (make_compound_operation_int): Set subreg_code to SET
17499 even for AND with mask of the sign bit of mode.
17501 PR rtl-optimization/80500
17502 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17503 sum's initial value.
17505 2017-04-25 Julian Brown <julian@codesourcery.com>
17506 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17508 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17510 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17512 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17514 2017-04-25 Julian Brown <julian@codesourcery.com>
17515 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17517 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17518 (thunderx2t99_sha): New Reservation.
17520 2017-04-25 Julian Brown <julian@codesourcery.com>
17521 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17523 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17524 type for 1-element load.
17526 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17528 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17530 2017-04-24 Martin Jambor <mjambor@suse.cz>
17532 PR tree-optimization/80293
17533 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17534 char arrays not totally scalarizable if it is false.
17535 (analyze_all_variable_accesses): Pass correct value in the new
17536 parameter. Add a statistics counter.
17538 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17540 PR middle-end/79931
17541 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17543 2017-04-24 Richard Biener <rguenther@suse.de>
17545 PR tree-optimization/80494
17546 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17547 out for complex types.
17549 2017-04-24 Richard Biener <rguenther@suse.de>
17551 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17552 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17553 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17554 (DFS): Adjust and never fail.
17555 (sccvn_dom_walker::fail): Remove.
17556 (sccvn_dom_walker::before_dom_children): Adjust.
17557 (run_scc_vn): Likewise and never fail.
17558 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17559 (pass_fre::execute): Likewise.
17561 2017-04-24 Richard Biener <rguenther@suse.de>
17563 PR tree-optimization/79725
17564 * tree-ssa-sink.c (statement_sink_location): Return whether
17565 failure reason was zero uses. Move that check later.
17566 (sink_code_in_bb): Deal with zero uses by removing the stmt
17569 2017-04-24 Richard Biener <rguenther@suse.de>
17572 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17573 pointer-based references.
17575 2017-04-24 Richard Biener <rguenther@suse.de>
17578 * pass_manager.h (pass_manager::operator new): Remove.
17579 (pass_manager::operator delete): Likewise.
17580 * passes.c (pass_manager::operator new): Remove.
17581 (pass_manager::operator delete): Likewise.
17582 (pass_manager::pass_manager): Zero individual pass members.
17584 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17587 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17588 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17589 Check "XEXP (src, 1)" operand here.
17590 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17591 Check "XEXP (src, 1)" operand here.
17592 (dimode_scalar_chain::make_vector_copies): Detect count register
17593 of a shift instruction. Zero extend count register from QImode
17594 to DImode to satisfy vector shift pattern count operand predicate.
17595 Substitute vector shift count operand with a DImode copy.
17596 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17599 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17601 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17602 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17603 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17604 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17605 (UNSPEC_NOREX_MEM): Remove definition.
17607 2017-04-21 Richard Biener <rguenther@suse.de>
17609 PR tree-optimization/79547
17610 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17611 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17612 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17613 without any constraints.
17615 2017-04-21 Richard Biener <rguenther@suse.de>
17617 PR tree-optimization/78847
17618 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17620 2017-04-21 Richard Biener <rguenther@suse.de>
17622 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17623 (build_distinct_type_copy): Likewise.
17624 (build_variant_type_copy): Likewise.
17625 * tree.c (build_qualified_type): Pass down mem-stat info.
17626 (build_distinct_type_copy): Likewise.
17627 (build_variant_type_copy): Likewise.
17629 2017-04-21 Richard Biener <rguenther@suse.de>
17631 PR tree-optimization/80237
17632 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17634 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17635 for a simplified result.
17637 2016-04-21 Richard Biener <rguenther@suse.de>
17639 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17640 sth as strict as a simple_iv but a chrec without symbols and an
17641 operand defined in the loop we are peeling (and not some subloop).
17642 (propagate_constants_for_unrolling): Propagate all constants.
17644 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17647 * config/i386/i386.c (print_reg): Remove assert for disalowed
17648 regno values, call output_operand_lossage instead.
17650 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17653 * config/i386/constraints.md (Yc): New register constraint.
17654 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17655 Use Yc constraint for alternative 2 of operand 0. Remove
17656 preferred_for_speed attribute.
17658 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17660 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17661 lastprivate clauses in SIMT case.
17663 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17665 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17667 2017-04-20 Richard Biener <rguenther@suse.de>
17669 PR tree-optimization/57796
17670 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17671 as N scalar stores.
17672 (vect_model_load_cost): Cost gathers as N scalar loads.
17674 2017-04-20 Richard Biener <rguenther@suse.de>
17676 * ggc-page.c (ggc_allocated_p): Rename to ...
17677 (safe_lookup_page_table_entry): ... this and return the lookup
17679 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17681 2017-04-20 Richard Biener <rguenther@suse.de>
17683 PR tree-optimization/80453
17684 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17685 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17686 from the conditions.
17687 (vn_phi_eq): Pass them down.
17688 (vn_phi_lookup): Record them.
17689 (vn_phi_insert): Likewise.
17691 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17693 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17694 uninitialized variable warning to avoid buffer overrun.
17696 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17699 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17700 is suppressed for '{ 0 }' in C.
17702 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17704 * BASE-VER: Set to 8.0.0.
17706 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17708 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17709 priority .init_array and .fini_array section with SECTION_NOTYPE
17712 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17714 PR middle-end/80423
17715 * tree.h (build_array_type): Add typeless_storage default argument.
17716 * tree.c (type_cache_hasher::equal): Also compare
17717 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17718 (build_array_type): Add typeless_storage argument, set
17719 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17721 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17722 (build_array_type): Likewise. Add typeless_storage argument.
17724 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17725 Jakub Jelinek <jakub@redhat.com>
17727 PR tree-optimization/80426
17728 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17729 operation on symbolic operands, also compute the overflow for the
17730 invariant part when the operation degenerates into a negation.
17732 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17735 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17736 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17739 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17741 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17744 * config/avr/avr.c (tree.h): Include it.
17745 (cgraph.h): Include it.
17746 (avr_encode_section_info): Don't warn for uninitialized progmem
17747 variable if it's just an alias.
17749 2017-04-19 Richard Biener <rguenther@suse.de>
17752 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17753 when needed by AutoPGO.
17755 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17758 * doc/lto.texi: Remove an extra 'that'.
17760 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17762 PR rtl-optimization/80429
17763 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17764 are only used in debug insns.
17766 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17767 Vladimir Makarov <vmakarov@redhat.com>
17769 * config/sparc/predicates.md (input_operand): Add comment. Return
17770 true for any memory operand when LRA is in progress.
17771 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17773 2017-04-18 Jeff Law <law@redhat.com>
17776 * mips.md ({return,simple_return}_internal): Do not overwrite
17779 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17781 PR tree-optimization/80443
17782 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17783 instead of adding 1, subtract -1 and similarly instead of subtracting
17786 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17788 PR rtl-optimization/80357
17789 * haifa-sched.c (tmp_bitmap): New variable.
17790 (model_recompute): Handle duplicate use records.
17791 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17792 (free_global_sched_pressure_data): Free it.
17794 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17797 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17798 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17799 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17800 instead of SYSTEM_HEADER_DIR.
17802 2017-04-18 Jeff Law <law@redhat.com>
17804 PR middle-end/80422
17805 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17806 predecessors after walking up the insn chain.
17808 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17811 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17812 sizetype type into debug info.
17814 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17817 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17818 unneeded test for TARGET_UPPER_REGS_SF.
17819 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17821 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17824 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17825 instead of gsi_after_labels.
17827 2017-04-18 Jeff Law <law@redhat.com>
17829 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17833 2017-04-13 Jeff Law <law@redhat.com>
17834 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17835 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17837 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17840 * config/avr/avr.c (intl.h): Include it.
17841 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17843 2017-04-18 Martin Liska <mliska@suse.cz>
17845 PR gcov-profile/78783
17846 * gcov-tool.c (gcov_output_files): Validate that destination
17847 file is either removed by the tool or by a user.
17849 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17850 Guy Benyei <guybe@mellanox.com>
17852 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17853 block, and do not negate it, the stored id is already negative.
17855 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17857 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17859 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17862 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17863 masks of options that should be turned off if the VSX vector
17864 options are turned off.
17865 (OTHER_P8_VECTOR_MASKS): Likewise.
17866 (OTHER_VSX_VECTOR_MASKS): Likewise.
17867 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17868 rs6000_disable_incompatible_switches to validate no type switches
17870 (rs6000_incompatible_switch): New function to disallow turning on
17871 other vector options if -mno-vsx, -mno-power8-vector, or
17872 -mno-power9-vector are specified.
17874 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17876 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17878 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17880 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17881 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17882 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17883 (ARG_POINTER_CFA_OFFSET): Likewise.
17885 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17887 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17888 conditions to take advantage of various optimizations.
17890 2017-04-13 Jeff Law <law@redhat.com>
17892 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17893 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17894 (zero_extendsidi2_dext): Likewise.
17896 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17899 * fold-const.c (fold_ternary_loc): Revert
17900 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17903 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17905 PR rtl-optimization/80343
17906 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17907 new scratch pseudo.
17909 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17912 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17913 to ubsan_encode_value.
17915 2017-04-13 Jeff Law <law@redhat.com>
17917 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17918 appearing in DEBUG_INSNs.
17920 2017-04-13 Martin Liska <mliska@suse.cz>
17922 PR gcov-profile/80413
17923 * gcov-io.c (gcov_write_string): Copy to buffer just when
17924 allocated size is greater than zero.
17926 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17929 * dwarf2out.c (decls_for_scope): Ignore declarations of
17930 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17932 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17935 * ipa-visibility.c (non_local_p): Fix typos.
17936 (localize_node): When localizing symbol in same comdat group,
17937 dissolve the group only when we know external symbols are going
17939 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17941 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17943 PR tree-optimization/79390
17944 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17945 order does not result in usable sequence, retry with reversed operand
17951 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17952 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17953 op0 instead of fold_convert_loc (loc, type, arg0).
17955 2017-04-12 Jeff Law <law@redhat.com>
17957 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17958 has a delay slot in the generated code.
17960 * config/cris/cris.md (cris_preferred_reload_class): Return
17961 GENNONACR_REGS rather than GENERAL_REGS.
17963 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17966 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17967 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17968 signedness of the result type.
17970 2017-04-12 Richard Biener <rguenther@suse.de>
17971 Jeff Law <law@redhat.com>
17973 PR tree-optimization/80359
17974 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17975 trim stores to TARGET_MEM_REFs.
17977 2017-04-12 Richard Biener <rguenther@suse.de>
17979 PR tree-optimization/79390
17980 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17981 threading case even more.
17983 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17986 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17987 for quad_address_p for TImode, instead of just not indexed_address.
17989 2017-04-12 Richard Biener <rguenther@suse.de>
17990 Bernd Edlinger <bernd.edlinger@hotmail.de>
17992 PR middle-end/79671
17993 * alias.c (component_uses_parent_alias_set_from): Handle
17994 TYPE_TYPELESS_STORAGE.
17995 (get_alias_set): Likewise.
17996 * tree-core.h (tree_type_common): Add typeless_storage flag.
17997 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17998 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17999 for types containing members with TYPE_TYPELESS_STORAGE.
18000 (place_field): Likewise.
18001 (layout_type): Likewise for ARRAY_TYPE.
18002 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18003 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18004 TYPE_TYPELESS_STORAGE.
18005 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18007 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18010 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18011 first argument to type.
18013 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18017 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18018 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18019 (rs6000_expand_binop_builtin): Likewise.
18020 (rs6000_expand_ternop_builtin): Likewise; also add missing
18021 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18022 vshasigma built-ins.
18023 * doc/extend.texi: Document that vec_xxpermdi's third argument
18024 must be a constant.
18026 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18028 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18029 Use shift_const cost parameter when calculating gain of STV shifts.
18031 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18033 PR rtl-optimization/70478
18034 * lra-constraints.c (process_alt_operands): Check memory for
18035 disfavoring memory insn operand.
18037 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18039 PR middle-end/80100
18040 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18041 left shift in unsigned HOST_WIDE_INT type.
18043 PR rtl-optimization/80385
18044 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18045 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18048 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18049 if they have any depend clauses.
18051 2017-04-11 Martin Liska <mliska@suse.cz>
18054 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18055 * ipa-split.c (split_function): Create a local comdat symbol
18056 if caller is in a comdat group.
18058 2017-04-11 Martin Liska <mliska@suse.cz>
18061 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18064 2017-04-11 Martin Sebor <msebor@redhat.com>
18066 PR middle-end/80364
18067 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18068 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18070 (directive::set_width, directive::set_precision, format_character):
18072 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18075 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18078 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18079 conflict, set target->arch_name instead of target->cpu_name.
18081 2017-04-11 Richard Biener <rguenther@suse.de>
18083 PR tree-optimization/80374
18084 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18085 build_zero_cst, remove fold_convertible_p check again.
18087 2017-04-11 Martin Liska <mliska@suse.cz>
18090 * ubsan.c (instrument_object_size): Do not instrument register
18093 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18096 * config/i386/i386-builtin-types.def
18097 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18098 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18099 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18100 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18101 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18102 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18103 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18104 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18105 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18106 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18107 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18108 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18109 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18110 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18111 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18112 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18113 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18114 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18115 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18116 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18117 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18118 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18119 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18120 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18121 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18122 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18123 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18124 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18125 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18126 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18127 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18128 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18129 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18130 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18131 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18132 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18133 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18134 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18135 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18136 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18137 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18138 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18139 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18140 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18141 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18142 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18144 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18145 flag to second_arg_count, handle 4 argument function type _COUNT
18146 aliases, handle second_arg_count on second argument rather than last.
18148 2017-04-10 Jeff Law <law@redhat.com>
18150 PR tree-optimization/80374
18151 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18152 record anything if we can not convert integer_zero_node to the
18155 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18158 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18159 Enhance special handling given to the TARGET_P9_MINMAX option in
18160 relation to certain other options.
18162 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18164 PR tree-optimization/80153
18165 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18166 remove POINTER_PLUS_EXPR's base part directly, rather than through
18169 2017-04-10 Richard Biener <rguenther@suse.de>
18170 Bin Cheng <bin.cheng@arm.com>
18172 PR tree-optimization/80153
18173 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18174 the first element of pointer type aff_tree. Build result expr in
18176 (add_elt_to_tree): Convert to type unconditionally. Remove other
18177 fold_convert calls.
18178 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18179 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18181 2017-04-10 Richard Biener <rguenther@suse.de>
18183 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18186 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18188 PR rtl-optimization/70478
18189 * lra-constraints.c (curr_small_class_check): New.
18190 (update_and_check_small_class_inputs): New.
18191 (process_alt_operands): Update curr_small_class_check. Disfavor
18192 alternative insn memory operands. Check available regs for small
18195 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18198 * config/mips/mips.opt (-mvirt): Update description.
18199 * doc/invoke.texi (-mvirt): Likewise.
18201 2017-04-10 Richard Biener <rguenther@suse.de>
18203 PR middle-end/80362
18204 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18205 looking for NEGATE_EXPR in -A / -B to A / B folding.
18207 2017-04-10 Martin Liska <mliska@suse.cz>
18209 PR gcov-profile/80224
18210 * gcov.c (print_usage): Fix usage string.
18211 (get_gcov_intermediate_filename): Remove.
18212 (output_gcov_file): Use both for normal and intermediate format.
18213 (generate_results): Do not initialize special file for
18214 intermediate format.
18216 2017-04-10 Richard Biener <rguenther@suse.de>
18218 PR tree-optimization/80304
18219 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18222 2017-04-10 Nathan Sidwell <nathan@acm.org>
18225 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18226 (rs6000_init_builtins): Use it.
18228 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18230 * config/arm/arm.md (<mrc>): Add mode to SET source.
18231 (<mrrc>): Likewise.
18233 2017-04-10 Richard Biener <rguenther@suse.de>
18235 PR middle-end/80344
18236 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18238 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18241 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18242 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18243 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18244 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18245 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18246 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18247 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18248 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18249 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18250 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18251 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18252 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18253 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18254 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18255 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18256 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18257 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18258 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18259 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18260 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18261 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18262 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18263 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18265 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18267 PR rtl-optimization/70478
18268 * lra-constraints.c: Reverse the last patch.
18270 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18272 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18273 Add comment for WCHAR_T.
18275 2017-04-08 Martin Liska <mliska@suse.cz>
18278 2017-04-07 Martin Liska <mliska@suse.cz>
18281 * ipa-split.c (split_function): Add function part to a same comdat
18284 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18287 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18289 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18291 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18292 (vec_pairing): ...to this.
18293 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18294 (rs6000_sched_init): Adjust for name change.
18295 (struct rs6000_sched_context): Likewise.
18296 (rs6000_init_sched_context): Likewise.
18297 (rs6000_set_sched_context): Likewise.
18299 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18305 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18307 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18308 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18309 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18311 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18313 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18315 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18317 PR rtl-optimization/70703
18318 * ira-color.c (update_conflict_hard_regno_costs): Use
18319 int64_t instead of HOST_WIDE_INT.
18321 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18323 PR rtl-optimization/70478
18324 * lra-constraints.c (process_alt_operands): Disfavor alternative
18325 insn memory operands.
18327 2017-04-07 Jeff Law <law@redhat.com>
18329 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18330 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18332 2017-04-07 Martin Liska <mliska@suse.cz>
18335 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18336 Show error message instead of an ICE.
18338 2017-04-07 Martin Liska <mliska@suse.cz>
18341 * ipa-split.c (split_function): Add function part to a same comdat
18344 2017-04-07 Richard Biener <rguenther@suse.de>
18346 PR middle-end/80341
18347 * tree.c (get_unwidened): Also handle ! for_type case for
18349 * convert.c (do_narrow): Split out from ...
18350 (convert_to_integer_1): ... here. Do not pass final truncation
18351 type to get_unwidened for TRUNC_DIV_EXPR.
18353 2017-04-07 Richard Biener <rguenther@suse.de>
18355 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18357 (aff_combination_const): Adjust.
18358 (aff_combination_scale): Likewise.
18359 (aff_combination_add_elt): Likewise.
18360 (aff_combination_add_cst): Likewise.
18361 (aff_combination_convert): Likewise.
18362 (add_elt_to_tree): Likewise. Remove unused argument.
18363 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18365 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18367 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18369 * config/arm/arm.c (arm_default_short_enums): Use
18370 ARM_DEFAULT_SHORT_ENUMS.
18371 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18373 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18376 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18377 members with redundant out-of-class redeclaration.
18379 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18382 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18383 * config/i386/i386.md (*zero_extendsidi2):
18384 Add (?*x,*x) and (?*v,*v) alternatives.
18386 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18389 * config/i386/i386.c (ix86_expand_builtin)
18390 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18391 mode from insn data. Convert operands to insn operand mode.
18392 Copy operands that don't satisfy insn predicate to a register.
18394 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18396 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18399 2017-04-06 Richard Biener <rguenther@suse.de>
18401 PR tree-optimization/80334
18402 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18403 preserve alignment of accesses.
18405 2017-04-06 Richard Biener <rguenther@suse.de>
18407 PR tree-optimization/80262
18408 * tree-sra.c (build_ref_for_offset): Preserve address-space
18410 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18411 Drop useless address-space information on MEM_REF offsets.
18413 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18415 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18417 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18419 PR rtl-optimization/70703
18420 * ira-color.c (update_conflict_hard_regno_costs): Use
18421 HOST_WIDE_INT instead of long.
18423 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18426 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18427 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18429 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18430 for x86_64 target. Handle -m3dnowa option.
18432 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18434 PR rtl-optimization/70703
18435 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18436 (update_conflict_hard_regno_costs): Use long instead of unsigned
18437 arithmetic for cost calculation.
18439 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18440 Bernd Edlinger <bernd.edlinger@hotmail.de>
18443 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18446 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18449 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18450 ptr_mode with Pmode throughout.
18451 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18452 into probe_stack_range and use DImode.
18454 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18457 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18458 call_eh_return is true.
18460 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18462 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18463 Initialize last_match_fntype_index.
18465 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18468 * tree-nvr.c: Include internal-fn.h.
18469 (pass_return_slot::execute): Ignore internal calls without
18472 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18473 Richard Biener <rguenther@suse.de>
18476 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18477 captures used multiple times, except for the last use.
18478 * generic-match-head.c: Include gimplify.h.
18480 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18482 PR tree-optimization/79390
18483 * target.h (struct noce_if_info): Declare.
18484 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18485 * target.def (noce_conversion_profitable_p): New target hook.
18486 * ifcvt.h (struct noce_if_info): New type, moved from ...
18487 * ifcvt.c (struct noce_if_info): ... here.
18488 (noce_conversion_profitable_p): Renamed to ...
18489 (default_noce_conversion_profitable_p): ... this. No longer
18491 (noce_try_store_flag_constants, noce_try_addcc,
18492 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18493 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18494 instead of noce_conversion_profitable_p.
18495 * config/i386/i386.c: Include ifcvt.h.
18496 (ix86_option_override_internal): Don't override
18497 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18498 (ix86_noce_conversion_profitable_p): New function.
18499 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18500 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18501 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18502 * doc/tm.texi: Regenerated.
18504 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18506 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18509 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18512 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18513 instructions for small multiply cores.
18515 2017-04-04 Jeff Law <law@redhat.com>
18517 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18519 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18520 that are not set by the loop over the elements.
18522 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18525 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18526 int mode, convert_modes it to mode as unsigned, otherwise use
18527 lowpart_subreg to mode rather than SImode.
18528 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18529 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18530 Use DImode instead of SImode for the shift count operand.
18531 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18534 2017-04-04 Richard Biener <rguenther@suse.de>
18536 PR middle-end/80281
18537 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18538 arithmetic done for the negate or the plus. Simplify.
18539 (A - (-B) -> A + B): Likewise.
18540 * fold-const.c (split_tree): Make sure to not negate pointers.
18542 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18544 PR rtl-optimization/60818
18545 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18546 a compare of comparisons with the thing compared if this results
18547 in a different machine mode.
18549 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18551 * alias.c (base_alias_check): Fix typo in comment.
18552 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18553 * cgraphunit.c (symbol_table::compile): Likewise.
18554 * collect2.c (maybe_run_lto_and_relink): Likewise.
18555 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18556 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18557 * config/avr/avr.c (avr_map_op_t): Likewise.
18558 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18559 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18560 * config/epiphany/epiphany.md (movcc): Likewise.
18561 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18562 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18564 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18565 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18566 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18567 * config/sh/sh.c (sh_rtx_costs): Likewise.
18568 * fold-const.c (fold_truth_andor): Likewise.
18569 * genautomata.c (collapse_flag): Likewise.
18570 * gengtype.h (struct type::u::s): Likewise.
18571 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18572 * input.c (FORMAT_AMOUNT): Likewise.
18573 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18574 (known_aggs_to_agg_replacement_list): Likewise.
18575 * ipa-inline-analysis.c: Likewise.
18576 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18577 * ipa-polymorphic-call.c
18578 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18579 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18580 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18582 * modulo-sched.c (apply_reg_moves): Likewise.
18583 * omp-expand.c (build_omp_regions_1): Likewise.
18584 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18585 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18586 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18587 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18588 * value-prof.c: Likewise.
18589 * var-tracking.c (val_reset): Likewise.
18591 2017-04-03 Richard Biener <rguenther@suse.de>
18593 PR tree-optimization/80275
18594 * fold-const.c (split_address_to_core_and_offset): Handle
18597 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18599 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18600 descriptors is at least equal to that of functions.
18602 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18604 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18606 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18609 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18610 (mov<IMOD4:mode>): New expander.
18611 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18613 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18615 PR rtl-optimization/79405
18616 * fwprop.c (propagations_left): New variable.
18617 (forward_propagate_into): Decrement it.
18618 (fwprop_init): Initialize it.
18619 (fw_prop): If the variable has reached zero, stop propagating.
18620 (fwprop_addr): Ditto.
18622 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18625 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18626 a FUNCTION_DECL, pass it as decl instead of origin to
18629 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18631 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18634 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18637 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18638 TARGET_VSX_SMALL_INTEGER.
18640 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18642 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18643 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18645 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18647 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18648 extraction from odd-numbered MSA register.
18650 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18652 PR middle-end/80173
18653 * expmed.c (store_bit_field_1): Don't attempt to create
18654 a word subreg out of hard registers wider than word if they
18655 have HARD_REGNO_NREGS of 1 for their mode.
18657 PR middle-end/80163
18658 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18659 conversions to integer types wider than word and pointer.
18662 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18663 (rtx_equal_for_cselib_p): Pass 0 to it.
18664 * cselib.c (cselib_hasher::equal): Likewise.
18665 (rtx_equal_for_cselib_1): Add depth argument. If depth
18666 is 128, don't look up VALUE locs and punt. Increment
18667 depth in recursive calls when walking VALUE locs.
18669 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18671 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18672 (make_gcov_file_name): Use the canonical path name for generating
18674 (read_line): Fix handling of files with ascii null bytes.
18676 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18678 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18679 to initialise a vector register instead
18680 of using a const_int.
18682 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18684 PR translation/80189
18685 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18686 diagnostic messages.
18688 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18691 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18692 (dfp_diex_<mode>): Update mode of operand 1.
18693 * doc/extend.texi (dxex, dxexq): Document change to return type.
18694 (diex, diexq): Document change to argument type.
18696 2017-03-30 Martin Jambor <mjambor@suse.cz>
18699 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18700 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18701 it reflects the signature changes performed at the callee side.
18702 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18703 to cgraph_build_function_type_skip_args.
18704 (build_function_decl_skip_args): Adjust call to the above function.
18706 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18709 * config/i386/sse.md
18710 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18711 register as dest whenever it is a MEM not rtx_equal_p to the
18712 corresponding dup operand, and when forcing into reg move the
18713 reg into the memory afterwards.
18714 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18715 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18716 for the force_reg mode.
18717 (avx512vl_vextractf128<mode>): Use register as dest either
18718 always when a MEM, or when it is a MEM not rtx_equal_p to the
18719 corresponding dup operand, or even not when it is a CONST_VECTOR
18720 depending on the mode and lo vs. hi.
18721 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18723 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18724 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18725 Likewise. Require that operands[2] is even.
18726 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18727 Remove extraneous parens. Require that operands[2] is a multiple
18729 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18730 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18731 disallow memory then.
18733 2017-03-30 Richard Biener <rguenther@suse.de>
18735 PR tree-optimization/77498
18736 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18737 to non-constants over backedges.
18739 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18741 PR rtl-optimization/80233
18742 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18743 as last_combined_insn. Do not test for BARRIER_P separately.
18745 2017-03-29 Andreas Schwab <schwab@suse.de>
18748 * calls.c (prepare_call_address): Convert funexp to Pmode before
18749 copying to temp reg.
18751 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18753 PR tree-optimization/80158
18754 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18755 Handle possible future case of more than one alternate
18757 (replace_rhs_if_not_dup): Likewise.
18758 (replace_one_candidate): Likewise.
18760 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18762 PR rtl-optimization/80193
18763 * ira.c (ira): Do not check allocation for LRA.
18765 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18767 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18768 (nvptx_output_simt_exit): Declare.
18769 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18770 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18771 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18772 (nvptx_declare_function_name): ...here. Emit declaration of local
18773 memory space buffer for omp_simt_enter insn.
18774 (nvptx_output_unisimt_switch): New.
18775 (nvptx_output_softstack_switch): New.
18776 (nvptx_output_simt_enter): New.
18777 (nvptx_output_simt_exit): New.
18778 * config/nvptx/nvptx.h (struct machine_function): New fields
18779 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18780 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18781 (UNSPECV_SIMT_EXIT): Ditto.
18782 (omp_simt_enter_insn): New insn.
18783 (omp_simt_enter): New expansion.
18784 (omp_simt_exit): New insn.
18785 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18787 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18788 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18789 (expand_GOMP_SIMT_EXIT): New.
18790 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18791 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18792 (GOMP_SIMT_EXIT): Ditto.
18793 * target-insns.def (omp_simt_enter): New insn.
18794 (omp_simt_exit): Ditto.
18795 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18797 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18798 (lower_rec_input_clauses): Likewise.
18799 (lower_lastprivate_clauses): Handle SIMT privatization.
18801 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18802 (ompdevlow_adjust_simt_enter): New.
18803 (find_simtpriv_var_op): New.
18804 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18805 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18807 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18808 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18809 (copy_decl_for_dup_finish): Ditto.
18811 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18813 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18816 * config/i386/i386.c (ix86_option_override_internal): Always
18817 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18819 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18821 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18823 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18825 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18826 mark new edge's irreducible flag accordign to it.
18827 (vect_do_peeling): Check loop preheader edge's irreducible flag
18828 and pass it to function slpeel_add_loop_guard.
18830 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18832 PR tree-optimization/80218
18833 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18834 Update block frequencies and counts.
18836 2017-03-28 Richard Biener <rguenther@suse.de>
18838 PR tree-optimization/78644
18839 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18840 of a simplification result we may not use it at all.
18842 2017-03-28 Richard Biener <rguenther@suse.de>
18845 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18846 without arguments, generate default definition of a SSA name.
18848 2017-03-28 Richard Biener <rguenther@suse.de>
18850 PR middle-end/80222
18851 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18852 TYPE_REF_CAN_ALIAS_ALL references.
18853 * fold-const.c (fold_indirect_ref_1): Likewise.
18855 2017-03-28 Martin Liska <mliska@suse.cz>
18858 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18859 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18861 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18862 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18864 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18865 (EXTRA_SPECS): Define.
18866 (SUBTARGET_EXTRA_SPECS): Likewise.
18867 (SUBTARGET_CPP_SPEC): Likewise.
18868 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18869 SUBTARGET_EXTRA_SPECS.
18870 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18872 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18874 * config/arc/simdext.md (vst64_insn): Update pattern.
18875 (vld32wh_insn): Likewise.
18876 (vld32wl_insn): Likewise.
18877 (vld64_insn): Likewise.
18878 (vld32_insn): Likewise.
18880 2017-03-28 Marek Polacek <polacek@redhat.com>
18883 * fold-const.c (fold_comparison): Use protected_set_expr_location
18884 instead of SET_EXPR_LOCATION.
18886 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18888 * tree.c (add_expr): Avoid name lookup warning.
18890 2017-03-27 Jeff Law <law@redhat.com>
18892 PR tree-optimization/80216
18893 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18894 function name. Limit recursion depth.
18895 (record_temporary_equivalences): Corresponding changes.
18897 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18899 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18902 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18905 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18907 * cfgcleanup.c (reg_note_cfa_p): New array.
18908 (insns_have_identical_cfa_notes): New function.
18909 (old_insns_match_p): Don't cross-jump in between /f
18910 and non-/f instructions. If both i1 and i2 are frame related,
18911 verify all CFA notes, their order and content.
18913 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18916 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18917 HImode and SImode with zero extend to DImode to one insn.
18918 (bswap<mode>2_extenddi): Likewise.
18919 (bswapsi2_extenddi): Likewise.
18920 (bswaphi2_extendsi): Likewise.
18921 (bswaphi2): Combine bswap HImode and SImode into one insn.
18922 Separate memory insns from swapping register.
18923 (bswapsi2): Likewise.
18924 (bswap<mode>2): Likewise.
18925 (bswaphi2_internal): Delete, no longer used.
18926 (bswapsi2_internal): Likewise.
18927 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18928 store, and gpr<-gpr swap insns.
18929 (bswap<mode>2_store): Likewise.
18930 (bswaphi2_reg): Register only splitter, combine with the splitter.
18931 (bswaphi2 splitter): Likewise.
18932 (bswapsi2_reg): Likewise.
18933 (bswapsi2 splitter): Likewise.
18934 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18935 the insns into load, store, and register/register insns.
18936 (bswapdi2_ldbrx): Likewise.
18937 (bswapdi2_load): Likewise.
18938 (bswapdi2_store): Likewise.
18939 (bswapdi2_reg): Likewise.
18941 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18943 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18944 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18946 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18949 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18951 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18952 special handling for target option conflicts between dform
18953 options (-mpower9-dform, -mpower9-dform-vector,
18954 -mpower9-dform-scalar) and -mno-direct-move.
18956 2017-03-27 Richard Biener <rguenther@suse.de>
18958 PR tree-optimization/80181
18959 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18961 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18963 * config/arc/predicates.md (move_double_src_operand): Replace the
18964 call to move_double_src_operand with a call to address_operand.
18966 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18968 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18969 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18970 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18972 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18974 * config/arc/predicates.md (long_immediate_loadstore_operand):
18975 Consider scaled addresses cases.
18977 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18979 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18980 restored when in interrupt.
18981 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18982 doesn't have delay slot.
18984 2017-03-27 Richard Biener <rguenther@suse.de>
18987 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18988 inlined thunk clones.
18990 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18993 * asan.c (instrument_derefs): Copy over last operand from
18994 original COMPONENT_REF to the new COMPONENT_REF with
18995 DECL_BIT_FIELD_REPRESENTATIVE.
18996 * ubsan.c (instrument_object_size): Likewise.
18998 2017-03-27 Richard Biener <rguenther@suse.de>
19000 PR tree-optimization/80170
19001 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19002 sure DR/SCEV didnt fold in constants we do not see when looking
19003 at the reference base alignment.
19005 2017-03-27 Richard Biener <rguenther@suse.de>
19007 PR middle-end/80171
19008 * gimple-fold.c (fold_ctor_reference): Properly guard against
19009 NULL return value from canonicalize_constructor_val.
19011 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19014 * config/i386/i386.c (ix86_expand_builtin)
19015 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19016 flags reg setting and flags reg using instructions.
19017 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19018 clobbering instructions to zero extend op2.
19020 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19022 * doc/install.texi (Configuration) <--with-aix-soname>:
19023 Update link to AIX ld.
19025 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19027 PR rtl-optimization/80160
19028 PR rtl-optimization/80159
19029 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19030 reg_alternate_class into account.
19032 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19035 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19036 to consider in curr_insn_transform.
19038 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19040 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19041 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19042 and emit_mode_inner.
19044 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19046 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19047 argument to the overloaded builtin variants. Use the new flag to
19048 deprecate certain builtin variants.
19049 * config/s390/s390-builtin-types.def: Add new builtin types.
19050 * config/s390/s390-builtins.h: Support new flags field for
19051 overloaded builtins.
19052 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19053 (s390_macro_to_expand): Enable vector float data type.
19054 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19055 builtins by incrementing the __VEC__ version number.
19056 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19058 (s390_resolve_overloaded_builtin): Emit error messages depending
19059 on the builtin flags.
19060 * config/s390/s390.c (s390_expand_builtin): Support additional
19061 flags argument. Change error message to match the messages
19062 emitted in s390-c.c.
19063 * config/s390/s390.md: New UNSPEC_* constants.
19064 (op_type): Add new instruction types.
19065 * config/s390/vecintrin.h: Add new builtins and test data class
19067 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19068 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19069 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19070 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19071 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19072 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19074 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19075 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19076 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19077 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19079 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19080 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19081 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19082 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19084 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19085 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19086 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19087 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19088 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19089 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19090 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19092 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19093 ("vec_scatter_element<V_HW_4:mode>_DI")
19094 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19095 ("vec_fpint<mode>", "vflls")
19096 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19097 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19098 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19099 ("*vec_cmphe<mode>_cc"): ... these.
19101 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19102 mode constant instead of magic value.
19104 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19106 * config/s390/s390.c (s390_expand_vec_compare): Support other
19107 vector floating point modes than just V2DF.
19108 (s390_expand_vcond): Likewise.
19109 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19110 (s390_cannot_change_mode_class): Prevent mode changes between TF
19111 and V1TF in vector registers.
19112 * config/s390/s390.md (DF, SF): New mode attributes.
19113 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19114 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19115 SFmode support for VRs.
19116 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19118 (VFT, VF_HW): New mode iterators.
19119 (vw, sdx): New mode attributes.
19120 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19121 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19122 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19123 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19124 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19125 also the new vector floating point modes. Renaming to ...
19127 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19128 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19129 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19130 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19131 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19132 ("vec_unordered<mode>"): ... these.
19134 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19135 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19136 ("*vec_extendv2df"): New insn definitions.
19138 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19140 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19141 ("mulditi3_2", "*muldi3_sign"): New patterns.
19142 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19143 rename the pattern definition.
19145 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19147 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19149 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19151 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19153 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19154 instruction if possible.
19155 * config/s390/vector.md (vec_halfnumelts): New mode
19157 ("*vec_vllezlf<mode>"): New pattern.
19159 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19161 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19162 ("popcountv4si2", "popcountv2di2"): Rename to ...
19163 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19164 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19166 ("popcount<mode>2_vxe"): New pattern.
19168 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19170 * common/config/s390/s390-common.c (processor_flags_table): Add
19172 * config.gcc: Add arch12.
19173 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19174 Default to arch12 for unknown CPU model numbers.
19175 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19176 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19177 PROCESSOR_max sanity check.
19178 * config/s390/s390-opts.h (enum processor_type): Add
19180 * config/s390/s390.c (processor_table): Add arch12.
19181 (s390_expand_builtin): Add check for B_VXE flag.
19182 (s390_issue_rate): Add PROCESSOR_ARCH12.
19183 (s390_get_sched_attrmask): Likewise.
19184 (s390_get_unit_mask): Likewise.
19185 (s390_sched_score): Enable z13 scheduling for arch12.
19186 (s390_sched_reorder): Likewise.
19187 (s390_sched_variable_issue): Likewise.
19188 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19190 (s390_tune_attr): Use z13 scheduling also for arch12.
19191 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19192 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19193 (TARGET_VXE_P): New macros.
19194 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19195 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19196 * config/s390/s390.opt: Add arch12 as processor_type.
19198 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19200 * config/s390/s390.md
19201 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19202 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19203 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19205 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19206 Rename expanders to ...
19208 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19209 ("fixuns_truncdddi2_emu"): ... these.
19211 ("fixuns_trunc<mode>si2_emu"): New expander.
19213 ("*fixuns_truncdfdi2_z13"): Rename to ...
19214 ("*fixuns_truncdfdi2_vx"): ... this.
19216 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19218 * config/s390/2964.md: Remove the single element vector compare
19219 instructions which are no longer used.
19220 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19222 (s390_canonicalize_comparison): Remove handling of DFmode
19224 (s390_expand_vec_compare_scalar): Remove function.
19225 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19226 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19228 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19230 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19232 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19234 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19235 will anyway by matched by mov<mode>_64dfp.
19237 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19239 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19240 vlef/vstef. Add missing operand to vleif.
19242 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19244 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19245 pair for all vector types with 64 bit elements.
19246 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19247 * config/s390/vector.md (V_HW_64): ... here.
19248 (V_128_NOSINGLE): New mode iterator.
19249 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19250 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19251 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19252 ("*vec_load_pairv2di"): Change to ...
19253 ("*vec_load_pair<mode>"): ... this one.
19255 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19257 * config/s390/constraints.md: Add comments.
19258 (jKK): Reject element sizes > 8 bytes.
19259 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19261 * config/s390/s390.md: Add the s_operand checks formerly in
19262 s390_split_ok_p to various splitters where they are still
19264 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19265 for 128 bit vectors. Plus two splitters.
19267 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19269 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19272 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19275 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19276 error if the boundary argument is not constant.
19278 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19280 PR rtl-optimization/80112
19281 * loop-doloop.c (doloop_condition_get): Don't check condition
19282 if cmp isn't SET with IF_THEN_ELSE src.
19284 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19286 PR tree-optimization/80158
19287 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19288 replacing a candidate statement, also replace it for the
19289 candidate's alternate interpretation.
19290 (replace_rhs_if_not_dup): Likewise.
19291 (replace_one_candidate): Likewise.
19293 2017-03-24 Richard Biener <rguenther@suse.de>
19295 PR tree-optimization/80167
19296 * graphite-isl-ast-to-gimple.c
19297 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19299 (translate_isl_ast_to_gimple::get_rename): Likewise.
19301 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19303 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19304 handling of certain combinations of target options, including the
19305 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19306 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19308 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19311 * config/arm/arm.md (*load_multiple): Add reload_completed to
19312 matching condition.
19314 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19315 Richard Biener <rguenth@suse.de>
19317 PR tree-optimization/79908
19318 PR tree-optimization/80136
19319 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19320 been cast away, gimplify_and_add suffices.
19322 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19324 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19326 2017-03-23 Richard Biener <rguenther@suse.de>
19328 PR tree-optimization/80032
19329 * gimplify.c (gimple_push_cleanup): Forced unconditional
19330 cleanups still have to go to the conditional_cleanups
19333 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19335 PR tree-optimization/80072
19336 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19338 (next_operand_entry_id): Change type to unsigned int.
19339 (sort_by_operand_rank): Make sure to return the right return value
19340 even if unsigned fields are bigger than INT_MAX.
19341 (struct oecount): Change cnt and id type to unsigned int.
19342 (oecount_hasher::equal): Formatting fix.
19343 (oecount_cmp): Make sure to return the right return value
19344 even if unsigned fields are bigger than INT_MAX.
19345 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19348 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19349 TREE_READONLY on result if writing it more than once.
19352 * doc/invoke.texi (-fsanitize=thread): Document that with
19353 -fnon-call-exceptions atomics are not able to throw
19357 * tsan.c: Include tree-eh.h.
19358 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19359 maybe_clean_or_replace_eh_stmt where needed.
19360 (instrument_memory_accesses): Add cfg_changed argument.
19361 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19362 if it returned true.
19363 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19365 PR rtl-optimization/63191
19366 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19367 wrapper function, moved the whole old content into ...
19368 (ix86_delegitimize_address_1): ... this. New inline function.
19369 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19370 true as last argument instead of ix86_delegitimize_address.
19372 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19374 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19375 cortexa57_branch_cost.
19377 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19379 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19381 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19384 * doc/md.texi (Constraints): Document wA constraint.
19385 * config/rs6000/constraints.md (wA): New.
19386 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19387 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19388 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19389 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19391 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19394 * gimplify.c (is_oacc_declared): New function.
19395 (oacc_default_clause): Use it to set default flags for acc declared
19396 variables inside parallel regions.
19397 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19398 declared variables.
19399 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19400 declare attribute to any decl as necessary.
19402 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19405 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19406 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19407 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19408 (arm_arch_lpae): This.
19409 * config/arm/arm.c (arm_arch7ve): Rename into ...
19410 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19411 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19414 2017-03-22 Martin Liska <mliska@suse.cz>
19417 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19418 error message instead of an ICE.
19420 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19422 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19424 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19426 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19428 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19431 2017-03-21 Martin Sebor <msebor@redhat.com>
19433 * doc/extend.texi: Use "cannot" instead of "can't."
19434 * doc/hostconfig.texi: Same.
19435 * doc/install.texi: Same.
19436 * doc/invoke.texi: Same.
19437 * doc/loop.texi: Same.
19438 * doc/md.texi: Same.
19439 * doc/objc.texi: Same.
19440 * doc/rtl.texi: Same.
19441 * doc/tm.texi: Same.
19442 * doc/tm.texi.in: Same.
19443 * doc/trouble.texi: Same.
19445 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19448 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19449 (collect_checksum_attributes): Set it.
19450 (die_checksum_ordered): Use it.
19452 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19454 PR tree-optimization/79908
19455 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19456 change: For a VA_ARG whose LHS has been cast away, use
19457 force_gimple_operand to construct the side effects.
19459 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19461 PR translation/80001
19462 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19463 more amenable to translation.
19464 (oacc_loop_auto_partitions): Likewise.
19466 2017-03-21 Marek Polacek <polacek@redhat.com>
19467 Martin Sebor <msebor@redhat.com>
19469 PR tree-optimization/80109
19470 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19471 on INTEGRAL_TYPE_P.
19473 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19474 Segher Boessenkool <segher@kernel.crashing.org>
19477 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19478 check reg_used_between_p between insn and one of succ or succ2
19479 depending on if succ is artificial insn not inserted into insn
19482 2017-03-21 Martin Liska <mliska@suse.cz>
19484 PR gcov-profile/80081
19485 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19486 * doc/gcc.texi: Include gcov-dump stuff.
19487 * doc/gcov-dump.texi: New file.
19489 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19491 PR rtl-optimization/79150
19492 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19493 conditional jump, if the jump is the last insn of the loop.
19495 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19496 Richard Biener <rguenth@suse.de>
19498 PR tree-optimization/79908
19499 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19500 been cast away, use force_gimple_operand to construct the side
19503 2017-03-21 Martin Liska <mliska@suse.cz>
19505 PR libfortran/79956
19506 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19509 2017-03-21 Brad Spengler <spender@grsecurity.net>
19512 * plugin.c (htab_hash_plugin): New function.
19513 (add_new_plugin): Use it and adjust.
19514 (parse_plugin_arg_opt): Adjust.
19515 (init_one_plugin): Likewise.
19517 2017-03-21 Richard Biener <rguenther@suse.de>
19519 PR tree-optimization/80032
19520 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19521 if set force the cleanup to happen unconditionally.
19522 (gimplify_target_expr): Push inserted clobbers with force_uncond
19523 to avoid them being removed by control-dependent DCE.
19525 2017-03-21 Richard Biener <rguenther@suse.de>
19527 PR tree-optimization/80122
19528 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19529 va_arg_pack_len when the inlined call stmt requires pack
19531 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19533 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19536 * tsan.c (instrument_builtin_call): If the memory model argument
19537 is not a constant, assume it is valid.
19540 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19543 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19545 PR rtl-optimization/79910
19546 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19547 if its dest is used by an insn before I2 (other than the combined
19548 insns themselves, which are properly handled already).
19550 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19553 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19555 * combine.c (record_used_regs): New static function.
19556 (try_combine): Handle situations where there is an additional
19557 instruction between I2 and I3 which needs to have a LOG_LINK
19561 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19563 * combine.c (try_combine): Delete redundant i1 test. Call
19564 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19566 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19569 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19570 alternatives 13/14.
19572 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19574 PR tree-optimization/80054
19575 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19576 the optimization if a PHI or any of its arguments is not dominated
19577 by the candidate's basis. Use gphi* rather than gimple* as
19579 (replace_profitable_candidates): Clean up a gimple* variable that
19580 should be a gphi* variable.
19582 2017-03-20 Martin Sebor <msebor@redhat.com>
19585 * doc/extend.texi (attribute constructor): Document present limitation.
19587 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19590 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19591 __POWER9_VECTOR__ #ifdef control, change template definition to
19592 use Power9-specific built-in function.
19593 (vec_any_eq): Likewise.
19594 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19595 to control outcomes from this test.
19596 (vector_ae_<mode>p): For VEC_F modes, likewise.
19598 2017-03-20 Ian Lance Taylor <iant@google.com>
19600 * config/i386/i386.c (ix86_function_regparm): Save an extra
19601 register for -fsplit-stack with DECL_STATIC_CHAIN.
19603 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19606 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19607 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19609 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19611 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19613 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19616 2017-03-20 Marek Polacek <polacek@redhat.com>
19619 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19621 2017-03-20 Richard Biener <rguenther@suse.de>
19623 PR tree-optimization/80113
19624 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19625 allocate extra SSA name for PHI def.
19626 (add_close_phis_to_outer_loops): Likewise.
19627 (add_close_phis_to_merge_points): Likewise.
19628 (copy_loop_close_phi_args): Likewise.
19629 (copy_cond_phi_nodes): Likewise.
19631 2017-03-20 Martin Liska <mliska@suse.cz>
19633 PR middle-end/79753
19634 * tree-chkp.c (chkp_build_returned_bound): Do not build
19635 returned bounds for a LHS that's not a BOUNDED_P type.
19637 2017-03-20 Martin Liska <mliska@suse.cz>
19641 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19642 COMPLEX_CST and VECTOR_CST.
19644 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19647 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19648 target operand. A new splitter adds the clobber statement in case
19649 the target operand is dead anyway.
19651 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19653 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19654 to age-old versions of binutils and glibc.
19656 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19658 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19660 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19662 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19664 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19666 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19667 requirement for binutils 2.13.
19669 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19671 * combine.c (try_combine): Delete redundant i1 test. Call
19672 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19674 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19676 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19677 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19679 <riscv64-*-elf>: Re-arrange section
19680 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19681 <riscv32-*-linux>: Likewise.
19682 <riscv64-*-elf>: Likewise
19683 <riscv64-*-linux>: Likewise.
19685 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19688 * aarch64.opt(verbose-cost-dump): Fix typo.
19690 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19693 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19694 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19696 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19698 * reload.c (find_reloads): When reloading a nonoffsettable address,
19699 use RELOAD_OTHER for it and its address reloads.
19701 PR rtl-optimization/79910
19702 * combine.c (record_used_regs): New static function.
19703 (try_combine): Handle situations where there is an additional
19704 instruction between I2 and I3 which needs to have a LOG_LINK
19707 2017-03-17 Jeff Law <law@redhat.com>
19709 PR tree-optimization/71437
19710 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19711 conditional in the hash table first.
19712 (vrp_dom_walker::before_dom_children): Extract condition from
19713 ASSERT_EXPR. Record condition, its inverion and any implied
19714 conditions as well.
19716 2017-03-17 Marek Polacek <polacek@redhat.com>
19717 Markus Trippelsdorf <markus@trippelsdorf.de>
19719 PR tree-optimization/80079
19720 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19723 2017-03-17 Richard Biener <rguenther@suse.de>
19725 PR middle-end/80075
19726 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19727 Properly verify the LHS before the RHS possibly claims to be
19729 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19732 2017-03-17 Martin Jambor <mjambor@suse.cz>
19734 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19735 (List of -O2 options): Likewise.
19736 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19739 2017-03-17 Tom de Vries <tom@codesourcery.com>
19741 * gcov-dump.c (print_usage): Print bug_report_url.
19743 2017-03-17 Richard Biener <rguenther@suse.de>
19745 PR middle-end/80050
19746 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19747 (parser::peek): Likewise.
19749 2017-03-17 Richard Biener <rguenther@suse.de>
19751 PR tree-optimization/80048
19752 * sese.c (free_sese_info): Properly release rename_map and
19753 copied_bb_map elements.
19755 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19757 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19758 Add linked-list forward and backlinks. Insert on
19759 construction, remove on destruction.
19760 (class pass_store_merging): Add m_stores_head field.
19761 (pass_store_merging::terminate_and_process_all_chains):
19762 Iterate over m_stores_head list.
19763 (pass_store_merging::terminate_all_aliasing_chains):
19765 (pass_store_merging::execute): Check for debug stmts first.
19766 Push new chains onto the m_stores_head stack.
19768 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19771 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19772 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19773 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19775 2017-03-16 Jeff Law <law@redhat.com>
19777 PR tree-optimization/71437
19778 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19779 member function. Implementation moved into after_dom_children
19780 member function and into the threader's thread_outgoing_edges
19782 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19783 some code into new thread_outgoing_edges.
19784 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19785 definition. Simplify marker handling (do it here). Assume we always
19786 have the available expression and the const/copies tables.
19787 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19789 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19790 * tree-vrp.c (equiv_stack): No longer file scoped.
19791 (vrp_dom_walker): New class.
19792 (vrp_dom_walker::before_dom_children): New member function.
19793 (vrp_dom_walker::after_dom_children): Likewise.
19794 (identify_jump_threads): Setup domwalker. Use it rather than
19795 walking edges in a random order by hand. Simplify setup/finalization.
19796 (finalize_jump_threads): Remove.
19797 (vrp_finalize): Do not call identify_jump_threads here.
19798 (execute_vrp): Do it here instead and call thread_through_all_blocks
19801 PR tree-optimization/71437
19802 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19804 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19806 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19807 (dom_opt_dom_walker::thread_across_edge): Remove
19808 handle_dominating_asserts argument. All callers changed.
19809 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19810 changes. Remove calls to lhs_of_dominating_assert. Other
19811 uses of handle_dominating_asserts turn into unconditional code
19812 (simplify_control_stmt_condition_1): Likewise.
19813 (simplify_control_stmt_condition): Likewise.
19814 (thread_through_normal_block, thread_across_edge): Likewise.
19815 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19816 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19817 object if it is not an SSA_NAME.
19818 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19819 before calling into the VRP specific simplifiers.
19820 (identify_jump_threads): Remove handle_dominating_asserts
19823 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19826 * tree-diagnostic.c (default_tree_printer): No longer static.
19827 * tree-diagnostic.h (default_tree_printer): New prototype.
19829 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19831 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19832 Change ins into fmov.
19834 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19836 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19837 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19838 Use h_con constraint for operand 1.
19839 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19840 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19842 2017-03-15 Jeff Law <law@redhat.com>
19844 PR tree-optimization/71437
19845 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19846 (record_temporary_equivalences): Use it.
19848 PR tree-optimization/71437
19849 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19850 tree-ssa-scopedtables.
19851 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19852 (record_conditions, record_cond, vuse_eq): Likewise.
19853 (record_edge_info): Adjust to API tweak of record_conditions.
19854 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19855 (record_temporary_equivalences, optimize_stmt): Likewise.
19856 (eliminate_redundant_computations): Likewise.
19857 (record_equivalences_from_stmt): Likewise.
19858 * tree-ssa-scopedtables.c: Include options.h and params.h.
19859 (vuse_eq): New function, moved from tree-ssa-dom.c
19860 (build_and_record_new_cond): Likewise.
19861 (record_conditions): Likewise. Accept vector of conditions rather
19862 than edge_equivalence structure for first argument.
19863 for the first argument.
19864 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19865 from tree-ssa-dom.c.
19866 (avail_exprs_stack::record_cond): Likewise.
19867 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19868 from tree-ssa-dom.c.
19869 (avail_exprs_stack): Add new member functions lookup_avail_expr
19871 (record_conditions): Declare.
19873 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19876 * lra-constraints.c (process_alt_operands): Increase reject for
19877 reloading an input/output operand.
19879 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19882 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19883 insns to convert from signed/unsigned char/short to IEEE 128-bit
19885 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19887 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19890 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19891 subreg of inner mode for values already in registers.
19893 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19895 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19896 iteration reg is used after the loop.
19898 2017-03-14 Martin Sebor <msebor@redhat.com>
19900 PR tree-optimization/79800
19901 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19902 precision in negative-positive range.
19903 (format_floating): Call non-const overload with adjusted precision.
19905 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19908 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19911 2017-03-14 Martin Sebor <msebor@redhat.com>
19913 PR middle-end/80020
19914 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19915 * builtins.def (aligned_alloc): Use it.
19918 * Makefile.in (GTFILES): Add calls.c.
19919 * calls.c: Include "gt-calls.h".
19921 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19923 PR rtl-optimization/79728
19924 * regs.h (struct target_regs): New field
19925 x_contains_allocatable_regs_of_mode.
19926 (contains_allocatable_regs_of_mode): New macro.
19927 * reginfo.c (init_reg_sets_1): Initialize it, and change
19928 contains_reg_of_mode so it includes global regs as well.
19929 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19930 rather than contains_regs_of_mode.
19932 2017-03-14 Martin Liska <mliska@suse.cz>
19934 * doc/invoke.texi: Document options that can't be combined with
19935 -fcheck-pointer-bounds.
19937 2017-03-14 Martin Liska <mliska@suse.cz>
19939 PR middle-end/79831
19940 * doc/invoke.texi (-Wchkp): Document the option.
19942 2017-03-14 Martin Liska <mliska@suse.cz>
19944 * Makefile.in: Install gcov-dump.
19946 2017-03-14 Martin Liska <mliska@suse.cz>
19948 * multiple_target.c (expand_target_clones): Bail out for
19949 an invalid attribute.
19951 2017-03-14 Richard Biener <rguenther@suse.de>
19953 * alias.c (struct alias_set_entry): Pack properly.
19954 * cfgloop.h (struct loop): Likewise.
19955 * cse.c (struct set): Likewise.
19956 * ipa-utils.c (struct searchc_env): Likewise.
19957 * loop-invariant.c (struct invariant): Likewise.
19958 * lra-remat.c (struct cand): Likewise.
19959 * recog.c (struct change_t): Likewise.
19960 * rtl.h (struct address_info): Likewise.
19961 * symbol-summary.h (function_summary): Likewise.
19962 * tree-loop-distribution.c (struct partition): Likewise.
19963 * tree-object-size.c (struct object_size_info): Likewise.
19964 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19965 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19966 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19967 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19968 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19969 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19970 (struct _stmt_vec_info): Likewise.
19972 2017-03-14 Martin Liska <mliska@suse.cz>
19975 * multiple_target.c (create_dispatcher_calls): Check that
19976 a target can create a function dispatcher.
19978 2017-03-14 Martin Liska <mliska@suse.cz>
19981 * multiple_target.c (expand_target_clones): Drop local.local
19982 flag for default implementation.
19984 2017-03-14 Richard Biener <rguenther@suse.de>
19986 PR tree-optimization/80030
19987 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19989 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19991 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19992 gcc_fallthrough() instead of __attribute__((fallthrough));
19994 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19996 * doc/gcc.texi: Remove "up" link to (DIR).
19997 * doc/gccint.texi: Ditto.
19999 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20001 * doc/install.texi (Specific) <avr>: Remove reference to
20004 2017-03-13 Jeff Law <law@redhat.com>
20006 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20007 attribute rather than comments.
20009 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20010 match_scratch operand is highest.
20012 2017-03-13 Martin Liska <mliska@suse.cz>
20014 PR middle-end/78339
20015 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20016 is a CHKP clone, use original declaration.
20018 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20020 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20021 (arc_conditional_register_usage): Use a different allocation order
20022 when optimizing for size.
20023 * common/config/arc/arc-common.c (arc_option_optimization_table):
20024 Section anchors default on when optimizing for size.
20026 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20028 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20030 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20032 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20033 * config/arc/arc.md (cpu_facility): Add cd variant.
20034 (*movqi_insn): Add code density variant.
20035 (*movhi_insn): Likewise.
20036 (*movqi_insn): Likewise.
20037 (*addsi3_mixed): Likewise.
20038 (subsi3_insn): Likewise.
20040 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20042 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20044 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20046 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20047 expressions with MINUS and UNARY ops.
20049 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20052 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20054 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20055 between vec_select and vector argument.
20056 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20057 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20058 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20059 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20060 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20061 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20063 2017-03-13 Richard Biener <rguenther@suse.de>
20066 * params.def (vect-max-peeling-for-alignment): Fix typo.
20068 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20070 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20071 issue that only occurred with binutils below 2.18.
20073 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20075 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20076 refer to binutils 2.11/2.12 minimum.
20078 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20080 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20081 ftp.kernel.org and simplify binutils requirement.
20083 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20085 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20087 (Optimize Options): Ditto. Also remove redundancy.
20089 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20091 PR translation/79848
20092 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20094 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20095 to G_ to avoid double translation.
20097 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20099 PR translation/79923
20100 * auto-profile.c (get_combined_location): Convert leading
20101 character of diagnostics to lower case and remove trailing period.
20102 (read_profile): Likewise for various diagnostics.
20103 * config/arm/arm.c (arm_option_override): Remove trailing period
20104 from various diagnostics.
20105 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20106 (msp430_expand_delay_cycles): Likewise.
20108 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20111 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20112 full command-line argument, rather than just "str".
20113 (aarch64_validate_march): Likewise.
20114 (aarch64_validate_mtune): Likewise.
20116 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20118 PR rtl-optimization/78911
20119 * lra-assigns.c (must_not_spill_p): New function.
20120 (spill_for): Use it.
20122 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20124 PR tree-optimization/79981
20125 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20126 ATOMIC_COMPARE_EXCHANGE ifn result.
20127 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20128 IFN_ATOMIC_COMPARE_EXCHANGE.
20130 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20133 * opts.c (parse_sanitizer_options): Add missing question mark to
20134 "did you mean" message.
20136 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20138 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20140 (VMULEUH_UNS): Likewise.
20141 (VMULOUB_UNS): Likewise.
20142 (VMULOUH_UNS): Likewise.
20143 * config/rs6000/rs6000.c (builtin_function_type): Remove
20144 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20146 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20149 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20150 x with result of extra_parsing_for_operand_code_0.
20151 (function_reader::extra_parsing_for_operand_code_0): Convert
20152 return type from void to rtx, returning x. When reading
20153 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20154 larger size containing struct block_symbol.
20156 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20158 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20159 -mfloat128-hardware without -m64.
20161 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20164 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20165 entries to the case statement that marks unsigned arguments to
20166 overloaded functions.
20168 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20170 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20171 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20173 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20176 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20177 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20179 2017-03-10 Martin Liska <mliska@suse.cz>
20183 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20184 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20187 2017-03-10 Olivier Hainque <hainque@adacore.com>
20189 * tree-switch-conversion (array_value_type): Start by resetting
20190 candidate type to it's main variant.
20192 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20194 PR rtl-optimization/79909
20195 * combine.c (try_combine): Use simplify_replace_rtx on individual
20196 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20197 of the whole CALL_INSN_FUNCTION_USAGE.
20199 PR tree-optimization/79972
20200 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20201 get_range_info on SSA_NAMEs. Formatting fixes.
20203 2017-03-10 Richard Biener <rguenther@suse.de>
20204 Jakub Jelinek <jakub@redhat.com>
20206 PR tree-optimization/77975
20207 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20208 edge to be constant.
20209 (get_val_for): For constant x return it. Formatting fix.
20210 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20211 would use the same bases as the current one.
20213 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20215 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20216 instead of vec_select for V1TImode.
20217 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20219 (VSX_LE_128): Add V1TI to this mode iterator.
20220 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20221 (*vsx_le_perm_store_<mode>): Likewise.
20222 (pre-reload splitter for VSX stores): Likewise.
20223 (post-reload splitter for VSX stores): Likewise.
20224 (*vsx_xxpermdi2_le_<mode>): Likewise.
20225 (*vsx_lxvd2x2_le_<mode>): Likewise.
20226 (*vsx_stxvd2x2_le_<mode>): Likewise.
20228 2017-03-09 Michael Eager <eager@eagercon.com>
20230 Correct failures with --enable-checking=yes,rtl.
20232 * config/microblaze/microblaze.c (microblaze_expand_shift):
20233 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20234 test for const0_rtx.
20235 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20236 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20238 2017-03-09 Richard Biener <rguenther@suse.de>
20240 PR tree-optimization/79977
20241 * graphite-scop-detection.c (scop_detection::merge_sese):
20242 Handle the case of extra exits to blocks dominating the entry.
20244 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20246 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20249 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20251 PR rtl-optimization/79949
20252 * lra-constraints.c (process_alt_operands): Check memory when
20253 trying to predict a cycle. Print about the overall increase.
20255 2017-03-09 Richard Biener <rguenther@suse.de>
20257 PR middle-end/79971
20258 * gimple-expr.c (useless_type_conversion_p): Preserve
20259 TYPE_SATURATING for fixed-point types.
20261 2017-03-09 Richard Biener <rguenther@suse.de>
20264 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20265 alignment of BLKmode params.
20267 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20270 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20271 (VALL_NO_V2Q): Likewise.
20273 * config/aarch64/aarch64-simd.md
20274 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20276 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20277 VALL_NO_V2Q mode iterator.
20278 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20280 2017-03-09 Martin Liska <mliska@suse.cz>
20282 PR tree-optimization/79631
20283 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20284 tree_int_cst_sign_bit just for INTEGER constants.
20286 2017-03-09 Martin Liska <mliska@suse.cz>
20290 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20293 2017-03-09 Marek Polacek <polacek@redhat.com>
20296 * tree.c (inchash::add_expr): Handle TREE_VEC.
20298 2017-03-09 Martin Liska <mliska@suse.cz>
20301 (chkp_narrow_size_and_offset): New function.
20302 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20303 (void chkp_parse_bit_field_ref): New function.
20304 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20305 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20307 2017-03-09 Martin Liska <mliska@suse.cz>
20310 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20311 (chkp_find_bounds_1): Remove gcc_unreachable.
20313 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20316 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20317 BUILT_IN_SYNC*, determine the access type from the size suffix and
20318 always build a MEM_REF with that type. Handle forgotten
20319 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20322 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20323 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20324 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20325 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20326 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20327 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20328 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20329 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20330 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20331 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20332 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20333 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20334 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20335 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20336 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20337 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20338 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20339 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20340 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20341 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20342 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20343 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20344 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20345 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20346 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20347 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20348 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20349 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20350 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20351 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20352 definitions outside of __OPTIMIZE__ guarded section.
20355 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20356 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20357 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20358 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20361 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20363 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20364 ("vfenez<mode>"): Add missing constraints.
20366 2017-03-08 Martin Sebor <msebor@redhat.com>
20369 * config/nds32/nds32.c (nds32_option_override):
20370 Fix misspelled diagnostic.
20372 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20375 * gimplify.c (gimplify_omp_for): Replace index var in outer
20376 taskloop statement with an artificial variable and add
20377 OMP_CLAUSE_PRIVATE clause for it.
20379 2017-03-08 Richard Biener <rguenther@suse.de>
20381 PR tree-optimization/79955
20382 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20383 for accesses that are completely outside of the variable.
20385 2017-03-08 Andrew Haley <aph@redhat.com>
20387 PR tree-optimization/79943
20388 * tree-ssa-loop-split.c (compute_new_first_bound): When
20389 calculating the new upper bound, (END-BEG) should be added, not
20392 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20394 * config/avr/avr.md (setmemhi): Make sure match_dup
20395 operand number comes before match_scratch.
20397 2017-03-08 Richard Biener <rguenther@suse.de>
20399 PR tree-optimization/79920
20400 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20401 with ncopies == 1 to ...
20402 (vect_transform_slp_perm_load): ... here. Properly compute
20403 all element loads by iterating VF times over the group. Do
20404 not handle ncopies (computed in a broken way) in
20405 vect_create_mask_and_perm.
20407 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20410 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20411 is a uniform vector, use uniform_vector_p return value instead of
20412 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20414 2017-03-07 Marek Polacek <polacek@redhat.com>
20416 PR middle-end/79809
20417 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20418 (alloca_call_type): Likewise.
20420 2017-03-07 Martin Liska <mliska@suse.cz>
20422 * gcov.c (process_args): Put comment to correct location.
20424 2017-03-07 Martin Liska <mliska@suse.cz>
20426 PR middle-end/68270
20427 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20428 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20429 (chkp_narrow_bounds_for_field): Likewise.
20430 (chkp_parse_array_and_component_ref): Pass one more argument to
20433 2017-03-07 Richard Biener <rguenther@suse.de>
20435 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20438 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20440 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20441 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20443 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20446 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20447 to end of description.
20448 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20450 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20452 PR rtl-optimization/79901
20453 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20455 (*avx512f_<code><mode>3<mask_name>): ... this.
20456 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20457 iterator instead of VI8_AVX2_AVX512BW.
20459 PR rtl-optimization/79901
20460 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20461 min/max expander, expand it using expand_vec_cond_expr.
20464 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20467 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20470 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20471 to void * for PCH reasons.
20472 * dwarf2out.c (output_loc_operands, output_die): Cast
20473 v.val_vec.array to unsigned char *.
20475 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20478 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20481 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20483 PR rtl-optimization/79571
20484 * lra-constraints.c (process_alt_operands): Calculate static
20485 reject and subtract it from overall when only addresses will be
20488 2017-03-06 Julia Koval <julia.koval@intel.com>
20491 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20492 incoming stack boundary to 128 for 64-bit targets.
20494 2017-03-06 Richard Biener <rguenther@suse.de>
20496 PR tree-optimization/79894
20497 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20498 to NULL after folding it.
20500 2017-03-06 Richard Biener <rguenther@suse.de>
20502 PR tree-optimization/79824
20503 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20504 check disabling peeling for gaps.
20506 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20508 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20509 attributes): Document gettimeofday.
20511 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20513 * config/s390/s390.c (s390_option_override_internal): Set
20514 PARAM_MIN_VECT_LOOP_BOUND
20516 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20518 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20519 * config/s390/s390.md: Likewise.
20521 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20524 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20525 (<avx2_avx512>_perm<mode>): Rename to ...
20526 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20528 (<avx512>_perm<mode>_mask): Rename to ...
20529 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20531 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20532 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20533 instead of VI8F_256_512.
20534 (avx512f_perm<mode>): New define_expand.
20535 (avx512f_perm<mode>_mask): Likewise.
20536 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20537 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20539 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20541 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20542 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20544 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20546 2017-03-06 Martin Liska <mliska@suse.cz>
20549 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20550 when having a SSA NAME w/o VAR_DECL assigned to it.
20552 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20554 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20555 msa_dpsub_<su>_d): Fix MODE for vec_select.
20557 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20559 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20561 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20563 2017-03-06 Richard Biener <rguenther@suse.de>
20565 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20566 * plugin.c (register_plugin_info): Likewise.
20567 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20569 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20571 * config/i386/sse.md (sse_storehps, sse_storelps,
20572 avx_<castmode><avxsizesuffix>_<castmode>,
20573 avx512f_<castmode><avxsizesuffix>_<castmode>,
20574 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20575 in condition that at least one operand is not a MEM.
20577 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20579 PR middle-end/79805
20580 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20581 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20583 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20584 gimple_call_nothrow_p flag based on whether original builtin can throw.
20585 If it can, emit following stmts on the fallthrough edge.
20586 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20587 don't create new bb if inserting just debug stmts on the edge, try to
20588 insert them on the fallthru bb or just reset debug stmts.
20590 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20593 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20594 restore recog_data (including the operand rtxes inside it) around
20595 the call to get_insn_template.
20597 2017-03-03 Martin Sebor <msebor@redhat.com>
20599 PR tree-optimization/79699
20600 * context.c (context::~context): Free MPFR caches to avoid
20601 a memory leak on program exit.
20603 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20605 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20606 Use wide_int::ulow () instead of .elt (0).
20608 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20610 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20611 (*pushxf): Limit oF constraint to 32bit targets and add oC
20612 constraint for 64bit targets.
20613 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20614 (*pushdf): Change rmF constraint to rmC.
20616 2017-03-03 Martin Liska <mliska@suse.cz>
20618 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20619 Remove unused variable.
20621 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20624 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20625 is a memory operand, increase num_memory.
20626 (ix86_expand_args_builtin): Likewise.
20628 2017-03-03 Jan Hubicka <jh@suse.cz>
20631 * ipa-devirt.c (maybe_record_node): Properly handle
20632 __cxa_pure_virtual visibility.
20634 2017-03-03 Martin Liska <mliska@suse.cz>
20636 PR tree-optimization/79803
20637 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20639 (pass_loop_prefetch::execute): Disabled optimization if an
20640 assumption about L1 cache size is not met.
20642 2017-03-03 Martin Liska <mliska@suse.cz>
20644 PR rtl-optimization/79574
20645 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20646 (hash_scan_set): Likewise.
20647 (dump_hash_table): Likewise.
20648 (hoist_code): Likewise.
20650 2017-03-03 Richard Biener <rguenther@suse.de>
20652 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20653 in place of uhigh ().
20654 (fixed_convert_from_real): Likewise.
20656 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20659 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20661 2017-03-03 Richard Biener <rguenther@suse.de>
20663 PR middle-end/79818
20664 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20665 TYPE_OVERFLOW_UNDEFINED check.
20667 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20669 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20671 (vector_ae_<mode>_p): Likewise.
20672 (vector_nez_<mode>_p): Likewise.
20673 (vector_ne_v2di_p): Likewise.
20674 (vector_ae_v2di_p): Likewise.
20675 (vector_ne_<mode>_p): Likewise.
20676 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20678 (vsx_tsqrt<mode>2_fe): Likewise.
20680 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20683 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20685 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20687 PR rtl-optimization/79780
20688 * cprop.c (one_cprop_pass): When second and further conditional trap
20689 in a single basic block is turned into an unconditional trap, turn it
20690 into a deleted note to avoid RTL verification failures.
20692 2017-03-02 Richard Biener <rguenther@suse.de>
20694 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20696 2017-03-02 Richard Biener <rguenther@suse.de>
20698 PR tree-optimization/79345
20700 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20701 param and abort the walk, returning -1 if it is hit.
20702 (walk_aliased_vdefs): Take a limit param and pass it on.
20703 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20704 defaulting to 0 and return a signed int.
20705 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20706 (check_defs): New helper.
20707 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20708 about uninitialized memory.
20709 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20710 bogus uninitialized warning.
20711 (fixed_convert_from_real): Likewise.
20713 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20715 PR tree-optimization/66768
20716 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20717 iv_use if base object can't be determined.
20719 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20721 PR tree-optimization/79345
20722 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20723 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20724 (get_pattern_stats): Initialize it.
20725 * genemit.c (gen_expand): Verify match_scratch numbers come after
20726 match_operand/match_dup numbers.
20727 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20728 match_scratch numbers.
20729 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20731 * config/s390/s390.md (trunctdsd2): Likewise.
20733 2017-03-02 Richard Biener <rguenther@suse.de>
20735 * wide-int.h (wide_int_storage::operator=): Implement in terms
20738 2017-03-02 Richard Biener <rguenther@suse.de>
20740 PR tree-optimization/79777
20741 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20742 the to insert expression to sth existing.
20744 2017-03-01 Martin Sebor <msebor@redhat.com>
20746 PR middle-end/79692
20747 * gimple-ssa-sprintf.c
20748 (directive::known_width_and_precision): New function.
20749 (format_integer): Use it.
20750 (get_mpfr_format_length): Consider the full range of precision
20751 when computing %g output with the # flag. Set the likely byte
20752 count to 3 rather than 1 when precision is indeterminate.
20753 (format_floating): Correct the lower bound of precision.
20755 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20757 * doc/invoke.texi: Document default code model for 64-bit Linux.
20759 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20762 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20763 udiv rather than div since input pattern is unsigned.
20765 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20767 * config/i386/i386.c (print_reg): Warn for values of
20768 unsupported size in integer register.
20770 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20773 * config/rs6000/predicates.md (current_file_function_operand): Do
20774 not allow self calls to be local if the function is replaceable.
20776 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20779 * config/rs6000/altivec.h (vec_ctz and others): Change the
20780 preprocessor macro that controls conditional compilation from
20781 _ARCH_PWR9 to __POWER9_VECTOR__.
20782 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20783 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20784 control (instead of _ARCH_PWR9 control) so that template
20785 definition uses power9-specific function.
20786 (vec_any_eq): Likewise.
20787 (vec_all_ne): Change macro definition to use a power9-specific
20788 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20789 _ARCH_PWR9 control).
20790 (vec_any_eq) Likewise.
20791 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20792 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20793 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20794 support for xvcmpnedp instruction.
20795 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20796 macro expansion so that Power9 implementation of vec_all_ne does
20797 not use the AltiVec predicate framework.
20798 (VCMPNEH_P): Likewise.
20799 (VCMPNEW_P): Likewise.
20800 (VCMPNED_P): Likewise.
20801 (VCMPNEFP_P): Likewise.
20802 (VCMPNEDP_P): Likewise.
20803 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20804 implementation of vec_any_eq to not use AltiVec predicate
20806 (VCMPAEH_P): Likewise.
20807 (VCMPAEW_P): Likewise.
20808 (VCMPAED_P): Likewise.
20809 (VCMPAEFP_P): Likewise.
20810 (VCMPAEDP_P): Likewise.
20811 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20812 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20813 not use the AltiVec predicate framework.
20814 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20815 of vec_any_eq to not use AltiVec predicate framework.
20816 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20817 support for predefined __POWER9_VECTOR__ macro to indicate that
20818 Power9 instruction selection is enabled.
20819 (altivec_overloaded_builtins): Remove extraneous
20820 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20821 function argument types RS6000_BTI_bool_V16QI and
20822 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20823 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20824 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20825 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20826 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20827 Power9 for implementations of vec_cmpne. Change the signature for
20828 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20829 (representing vec_all_ne) to remove the previously described first
20830 argument of type RS6000_BTI_INTSI, as this was an artifact of
20831 reliance on the AltiVec predicate framework, which is no longer
20832 used in the implementation of these functions. Add
20833 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20834 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20835 since, unlike the AltiVec predicate framework implementation, we
20836 do not share function descriptors between vec_alle and vec_anyeq.
20837 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20838 set of modes that receive special treatment even when
20839 TARGET_P9_VECTOR is true. The special treatment emits code that
20840 does not depend on Power9 instructions.
20841 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20842 define_expand to not rely on AltiVec predicate framework.
20843 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20845 (vector_ne_v2di_p): Change this define_expand to not rely on
20846 AltiVec predicate framework.
20847 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20849 (vector_ne_<mode>_p): Change this define_expand to not rely on
20850 AltiVec predicate framework.
20851 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20853 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20854 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20855 define_insn pattern.
20856 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20857 define_insn pattern because the xvcmpne<VSs>. instruction is not
20859 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20860 instruction is not supported.
20862 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20864 * config/nvptx/nvptx.c: Include intl.h.
20866 2017-03-01 Martin Jambor <mjambor@suse.cz>
20869 * ipa-prop.h (ipa_bits): Removed field known.
20870 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20871 to pointers. Adjusted their comments to warn about their sharing.
20872 (ipcp_transformation_summary): Change bits to a vector of pointers.
20873 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20874 (ipa_get_ipa_bits_for_value): Declare.
20875 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20876 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20877 (ipa_bits_hash_table): Likewise.
20878 (ipa_vr_ggc_hash_traits): Likewise.
20879 (ipa_vr_hash_table): Likewise.
20880 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20881 being pointers and vr_known being removed.
20882 (ipa_set_jf_unknown): Likewise.
20883 (ipa_get_ipa_bits_for_value): New function.
20884 (ipa_set_jfunc_bits): Likewise.
20885 (ipa_get_value_range): New overloaded functions.
20886 (ipa_set_jfunc_vr): Likewise.
20887 (ipa_compute_jump_functions_for_edge): Use the above functions to
20888 construct bits and vr parts of jump functions.
20889 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20890 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20892 (ipcp_grow_transformations_if_necessary): Also allocate
20893 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20895 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20896 them. Fix too long lines.
20897 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20898 vr_known being removed.
20899 (ipa_read_jump_function): Use new setter functions to construct bits
20900 and vr parts of jump functions or set them to NULL.
20901 (write_ipcp_transformation_info): Adjust for bits being pointers.
20902 (read_ipcp_transformation_info): Likewise.
20903 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20905 Include gt-ipa-prop.h.
20906 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20908 (ipcp_store_bits_results): Likewise.
20909 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20910 Do not write to existing jump functions but use a temporary instead.
20912 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20915 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20916 attempt to use its first operand as BIT_FIELD_REF base.
20918 2017-03-01 Richard Biener <rguenther@suse.de>
20920 PR middle-end/79721
20921 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20922 interpolating formula in wrapping arithmetic.
20923 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20925 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20927 PR tree-optimization/79734
20928 * tree-vect-generic.c (expand_vector_condition): Optimize
20929 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20930 Handle VEC_COND_EXPR where comparison has different inner width from
20931 type's inner width.
20933 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20935 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20936 markup, and similar issues. Remove @opindex entries for things
20937 that aren't options. Add missing -mmpy-option entries.
20939 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20941 PR tree-optimization/79737
20942 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20943 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20944 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20945 instead of byte_size. Formatting fix.
20946 (shift_bytes_in_array_right): Formatting fix.
20948 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20951 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20952 condition on optimize for the leaf function test.
20954 2017-02-28 Martin Liska <mliska@suse.cz>
20957 * read-rtl-function.c (function_reader::handle_unknown_directive):
20958 Bail out when one uses -flto.
20960 2017-02-28 Martin Liska <mliska@suse.cz>
20962 * common.opt: Replace space with tabular for options of <number>
20964 * config/i386/i386.opt: Show <number> value for
20965 -mlarge-data-threshold.
20966 * opts.c (print_filtered_help): Do not display number in hexadecimal
20969 2017-02-28 Martin Liska <mliska@suse.cz>
20971 * common.opt: Fix --help=option -Q for options which are of
20974 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20976 * config/i386/i386.c (print_reg): Error out for values
20977 of 8-bit size in invalid integer register.
20979 2017-02-28 Martin Sebor <msebor@redhat.com>
20981 PR tree-optimization/79691
20982 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20984 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20987 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20988 gcc_unreachable with output_operand_lossage.
20990 2017-02-28 Richard Biener <rguenther@suse.de>
20992 PR tree-optimization/79740
20993 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20995 (visit_nary_op): Insert the nary into the hashtable if we
20996 pattern-matched sth.
20997 * tree-ssa-pre.c (eliminate_insert): Robustify.
20999 2017-02-28 Richard Biener <rguenther@suse.de>
21001 PR middle-end/79731
21002 * fold-const.c (decode_field_reference): Reject out-of-bound
21005 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21007 * config/i386/i386.c: Include intl.h.
21008 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21009 instead of just cond ? "..." : "...".
21010 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21011 * coverage.c (read_counts_file): Likewise.
21012 * omp-offload.c: Include intl.h.
21013 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21014 of just cond ? "..." : "...".
21015 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21016 of just cond ? "..." : "...".
21018 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21021 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21023 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21024 'tune for' CPU name.
21025 * config/arm/arm-cpu-data.h: Regenerated.
21027 2017-02-28 Richard Biener <rguenther@suse.de>
21029 PR tree-optimization/79732
21030 * tree-inline.c (expand_call_inline): Do not shadow var.
21032 2017-02-28 Richard Biener <rguenther@suse.de>
21034 PR tree-optimization/79723
21035 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21036 address-space properly.
21038 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21040 * doc/optinfo.texi (Optimization groups): Fix option used for
21042 * doc/invoke.texi (-fopt-info): Document "omp".
21043 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21044 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21045 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21046 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21047 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21049 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21051 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21052 OPTGROUP_OMP with "omp".
21054 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21057 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21058 for arithmetic shift of unsigned V2DI.
21060 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21062 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21063 arc/linux.h headers.
21064 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21065 (LINK_SPEC): Likewise.
21066 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21067 (EXTRA_SPECS): Likewise.
21068 (STARTFILE_SPEC): Likewise.
21069 (ENDFILE_SPEC): Likewise.
21070 (LIB_SPEC): Likewise.
21071 (TARGET_SDATA_DEFAULT): Likewise.
21072 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21073 (MULTILIB_DEFAULTS): Likewise.
21074 (DWARF2_UNWIND_INFO): Likewise.
21075 * config/arc/big.h: New file.
21076 * config/arc/elf.h: Likewise.
21077 * config/arc/linux.h: Likewise.
21078 * config/arc/t-uClibc: Remove.
21080 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21082 PR tree-optimization/77536
21083 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21084 (tree_transform_and_unroll_loop): Use above function to compute the
21085 estimated niter of unrolled loop and use it when scaling profile.
21086 Also use count info rather than frequency if it's non-zero.
21087 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21088 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21089 (vect_transform_loop): Call above function.
21091 2017-02-27 Richard Biener <rguenther@suse.de>
21093 PR tree-optimization/45397
21094 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21095 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21096 (visit_nary_op): Add pattern matching for CSEing sign-changed
21097 or truncated operations with wider ones.
21099 2017-02-27 Richard Biener <rguenther@suse.de>
21101 PR tree-optimization/79690
21102 * tree-vect-stmts.c (vectorizable_store): Use vector type
21103 built from the DR with address-space.
21105 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21107 * doc/invoke.texi (Optimize Options): Refine the description
21108 of asan-use-after-return.
21110 2017-02-25 Alan Modra <amodra@gmail.com>
21112 PR rtl-optimization/79584
21113 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21114 base, not ad->base_term, the reg within base. Remove assertion
21115 that ad->base == ad->base_term. Replace gen_int_mode using
21116 bogus mode with const0_rtx.
21118 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21120 PR middle-end/79396
21121 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21122 FMA_EXPR like tcc_binary or tcc_unary.
21124 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21127 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21129 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21130 (output_loc_operands): Handle DW_OP_call_ref and
21131 DW_OP_GNU_variable_value.
21132 (struct variable_value_struct): New type.
21133 (struct variable_value_hasher): Likewise.
21134 (variable_value_hash): New variable.
21135 (string_types): Remove.
21136 (copy_loc_descr): New function.
21137 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21138 (prepend_loc_descr_to_each): New function.
21139 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21140 instead of add_loc_descr_to_each if the first argument is single
21141 location list and the second has multiple.
21142 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21143 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21144 when looking for variable value which doesn't have other location info.
21145 (loc_list_from_tree): Formatting fix.
21146 (gen_array_type_die): Simplify DW_AT_string_length handling.
21147 (adjust_string_types): Remove.
21148 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21149 string_types. Call resolve_variable_values.
21150 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21151 (resolve_addr_in_expr): Likewise. Add A argument.
21152 (copy_deref_exprloc): Remove deref argument. Adjust for the
21153 original expression being DW_OP_GNU_variable_value with optionally
21154 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21155 optionally after it.
21156 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21157 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21158 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21159 (variable_value_hasher::hash, variable_value_hasher::equal): New
21161 (resolve_variable_value_in_expr, resolve_variable_value,
21162 resolve_variable_values, note_variable_value_in_expr,
21163 note_variable_value): New functions.
21164 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21167 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21170 * opts.h (handle_generated_option): Add GENERATED_P argument.
21171 * opts-common.c (handle_option): Adjust function comment.
21172 (handle_generated_option): Add GENERATED_P argument, pass it to
21174 (control_warning_option): Pass false to handle_generated_option
21176 * opts.c (maybe_default_option): Pass true to handle_generated_option
21178 * optc-gen.awk: Likewise.
21180 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21182 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21183 a REG, look at the REG it is a SUBREG of.
21184 (splitter for cmpeqsi_t): Ditto.
21186 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21188 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21189 the special USEs with the pattern of the insn, not the insn itself.
21191 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21194 * doc/invoke.texi: Document -mload-store-pairs.
21196 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21197 Sandra Loosemore <sandra@codesourcery.com>
21199 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21200 argument isn't a CONST_INT.
21201 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21202 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21203 (nios2_validate_compare): Bypass alternate compare logic if *op2
21204 is not a CONST_INT.
21205 (ldstwm_operation_p): Return false if first_base is not a REG or
21206 if first_offset is not a CONST_INT.
21208 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21210 * config/cris/cris.md: Use correct operand in a define_peephole2.
21212 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21214 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21216 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21218 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21219 this_insn if it is an INSN or JUMP_INSN.
21220 (force_offsettable): Look at base, not at addr.
21221 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21222 on things that aren't necessarily CONST_INTs.
21224 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21226 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21227 -mfpmath=sse is the default also for x86-32 targets with SSE2
21228 instruction set when @option{-ffast-math} is enabled
21230 2017-02-24 Jeff Law <law@redhat.com>
21232 PR rtl-optimizatoin/79286
21233 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21236 2017-02-24 Richard Biener <rguenther@suse.de>
21238 PR tree-optimization/79389
21239 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21242 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21244 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21245 function comment to reflect reality.
21246 (loop_exits_before_overflow): Fix typo in function description.
21248 2017-02-24 Richard Biener <rguenther@suse.de>
21250 PR tree-optimization/79389
21251 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21252 properly that a threading opportunity exists. Detect conditional
21253 copy/constant propagation opportunities.
21255 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21257 * config/visium/visium.md (type): Add trap.
21258 (b): New mode attribute.
21259 (*btst): Rename into...
21260 (*btst<mode>): ...this and adjust.
21261 (*cbranchsi4_btst_insn): Rename into...
21262 (*cbranch<mode>4_btst_insn): ...this and adjust.
21263 (trap): New define_insn.
21265 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21267 PR tree-optimization/79389
21268 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21269 (noce_reversed_cond_code): New function.
21270 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21271 reversed_comparison_code. Formatting fix.
21272 (noce_try_store_flag): Test rev_cond != NULL in addition to
21273 reversed_comparison_code.
21274 (noce_try_store_flag_constants): Likewise.
21275 (noce_try_store_flag_mask): Likewise.
21276 (noce_try_addcc): Use rev_cond if non-NULL instead of
21277 reversed_comparison_code.
21278 (noce_try_cmove_arith): Likewise. Formatting fixes.
21279 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21280 (noce_find_if_block): Initialize rev_cond.
21281 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21282 instead of false as last argument never attempt to reverse it
21285 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21287 PR tree-optimization/79663
21288 * tree-predcom.c (combine_chains): Process refs in reverse order
21289 only for ZERO length chains, and add explaining comment.
21291 2017-02-23 Jeff Law <law@redhat.com>
21293 PR tree-optimization/79578
21294 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21295 in call to operand_equal_p.
21297 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21300 * config/i386/cpuid.h: Fix another undefined behavior.
21302 2017-02-23 Richard Biener <rguenther@suse.de>
21304 PR tree-optimization/79683
21305 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21306 vector types for data-refs.
21308 2017-02-23 Martin Liska <mliska@suse.cz>
21310 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21312 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21314 PR middle-end/79665
21315 * internal-fn.c (get_range_pos_neg): Moved to ...
21316 * tree.c (get_range_pos_neg): ... here. No longer static.
21317 * tree.h (get_range_pos_neg): New prototype.
21318 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21319 are known to be in between 0 and signed maximum inclusive, try to
21320 expand both unsigned and signed divmod and use the cheaper one from
21323 2017-02-22 Jeff Law <law@redhat.com>
21325 PR tree-optimization/79578
21326 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21327 to compare base operands.
21329 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21332 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21333 gpc_reg_operand instead of fpr_reg_operand.
21335 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21337 * config/mips/mips.c (mips_return_in_memory): Force FP
21338 vector types to be returned in memory for o32 ABI.
21340 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21342 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21343 instead of DW_TAG_member for static data member declarations and don't
21344 set no_linkage_name for static inline data members.
21345 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21348 2017-02-22 Martin Liska <mliska@suse.cz>
21350 * doc/invoke.texi: Replace inequality signs with square brackets
21353 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21355 PR tree-optimization/68644
21356 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21358 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21361 * lra-constraints.c (simplify_operand_subreg): Handle
21362 WORD_REGISTER_OPERATIONS targets.
21364 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21367 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21368 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21369 elimination by swapping fld*.
21371 2017-02-22 Richard Biener <rguenther@suse.de>
21373 PR tree-optimization/79673
21374 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21375 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21376 irrelevant address-space qualifiers and avoiding a
21377 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21379 2017-02-22 Richard Biener <rguenther@suse.de>
21381 PR tree-optimization/79666
21382 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21383 to not symbolically negate if that may introduce undefined
21386 2017-02-22 Martin Liska <mliska@suse.cz>
21389 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21390 * data-streamer-out.c (streamer_write_gcov_count_stream):
21392 * value-prof.c (stream_out_histogram_value): Make assert more
21393 precise based on type of counter.
21395 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21398 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21399 Use nonimmediate_operand instead of memory_operand for operand 1.
21400 (float-extend standard_x87sse_constant_load splitter): Ditto.
21402 2017-02-21 Jeff Law <law@redhat.com>
21404 PR tree-optimization/79621
21405 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21406 blocks with edges to themselves.
21408 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21411 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21412 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21413 Use gimple_call_builtin_p.
21416 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21417 on temporarily removed DEBUG_INSNs.
21419 PR tree-optimization/79649
21420 * tree-loop-distribution.c (classify_partition): Give up on
21421 non-generic address space loads/stores.
21423 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21425 * doc/loop.texi (Loop manipulation): Remove nonexistent
21426 tree_ssa_loop_version from the documentation.
21427 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21429 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21432 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21433 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21434 * config/rs6000/rs6000.c: Include except.h.
21435 (rs6000_expand_split_stack_prologue): Call
21436 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21438 2017-02-21 Martin Jambor <mjambor@suse.cz>
21441 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21442 have been analyzed.
21444 2017-02-21 Martin Jambor <mjambor@suse.cz>
21446 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21447 for backward compatibility only.
21448 * doc/invoke.texi (Option Summary): Remove all references to
21449 -fipa-cp-alignment.
21451 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21455 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21457 * lra-constraints.c (curr_insn_transform): Handle
21458 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21460 2017-02-21 Martin Liska <mliska@suse.cz>
21462 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21464 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21466 PR translation/79638
21467 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21469 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21472 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21473 (arm_function_ok_for_sibcall): Return false for an indirect call by
21474 descriptor if all the argument registers are used.
21475 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21476 alignment of the function.
21478 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21480 PR tree-optimization/61441
21481 * simplify-rtx.c (simplify_const_unary_operation): For
21482 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21483 the sNaN unmodified.
21485 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21487 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21488 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21489 instead of SYSTEM_HEADER_DIR.
21491 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21492 Martin Liška <mliska@suse.cz>
21494 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21495 Fix typos and grammar, use active voice, and clarify.
21497 2017-02-20 Marek Polacek <polacek@redhat.com>
21499 PR middle-end/79537
21500 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21503 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21505 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21508 * config/i386/i386.c (ix86_expand_builtin): Handle
21509 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21510 ix86_builtins_isa[fcode].isa as a requirement of those
21511 flags and any other flag in the bitmask.
21512 (ix86_init_mmx_sse_builtins): Use 0 instead of
21513 ~OPTION_MASK_ISA_64BIT as mask.
21514 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21515 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21516 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21517 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21519 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21522 * lra-constraints.c (split_reg): Check requested split mode
21523 is supported by the register.
21525 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21527 * lra-constraints.c (simplify_operand_subreg): Remove early
21530 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21533 * lra-constraints.c (curr_insn_transform): Tighten condition
21534 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21536 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21539 * lra-constraints.c (curr_insn_transform): Handle
21540 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21542 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21545 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21547 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21549 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21552 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21555 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21557 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21560 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21562 * final.c (last_columnnum, override_columnnum): New variables.
21563 (final_start_function): Set last_columnnum, pass it to begin_prologue
21564 hook and pass 0 to dwarf2out_begin_prologue.
21565 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21566 to source_line debug hook.
21567 (notice_source_line): Compute last_columnnum and for debug_column_info
21568 return true on column changes.
21569 * debug.h (struct gcc_debug_hooks): Add column argument to
21570 source_line and begin_prologue hooks.
21571 (debug_nothing_int_charstar_int_bool): Remove prototype.
21572 (debug_nothing_int_int_charstar,
21573 debug_nothing_int_int_charstar_int_bool): New prototypes.
21574 (dwarf2out_begin_prologue): Add column argument.
21575 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21576 begin_prologue hooks.
21577 (debug_nothing_int_charstar_int_bool): Remove.
21578 (debug_nothing_int_int_charstar,
21579 debug_nothing_int_int_charstar_int_bool): New functions.
21580 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21581 through to dwarf2out_source_line.
21582 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21583 (dwarf2out_source_line): Add column argument, emit it if requested.
21584 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21586 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21587 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21588 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21589 through to dwarf2out_begin_prologue.
21590 (vmsdbgout_source_line): Add column argument, pass it through to
21591 dwarf2out_source_line.
21592 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21593 dbxout_source_line caller.
21594 (dbxout_source_line): Add column argument.
21596 * common.opt (gno-column-info, gcolumn-info): New options.
21597 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21598 (check_die): Also test for multiple DW_AT_decl_column attributes.
21599 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21600 DW_AT_decl_column if requested.
21601 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21603 (gen_variable_die): Likewise.
21604 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21605 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21608 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21609 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21610 (ix86_handle_option): Handle OPT_m3dnowa.
21611 * doc/invoke.texi (-m3dnowa): Document.
21612 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21613 -m3dnowa instead of -m3dnow -march=athlon.
21616 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21617 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21619 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21622 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21623 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21624 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21625 generator for vsx_xxpermdi_<mode>_be.
21626 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21627 force big-endian semantics.
21628 (vsx_xxpermdi_<mode>_be): New define_expand with same
21629 implementation as previous version of vsx_xxpermdi_<mode>.
21631 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21633 PR tree-optimization/79327
21634 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21635 variable, its initialization and use.
21637 2017-02-17 Julia Koval <julia.koval@intel.com>
21639 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21640 (OPTION_MASK_ISA_PKU_UNSET): New.
21641 (ix86_handle_option): Handle -mrdpid.
21642 * config/i386/cpuid.h (bit_RDPID): New.
21643 * config/i386/driver-i386.c (host_detect_local_cpu):
21644 Detect RDPID feature.
21645 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21646 * config/i386/i386-c.c (ix86_target_macros_internal):
21648 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21649 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21650 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21651 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21652 * config/i386/i386.md (define_insn "rdpid"): New.
21653 * config/i386/i386.opt Add -mrdpid.
21654 * config/i386/immintrin.h (_rdpid_u32): New.
21656 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21658 PR rtl-optimization/79541
21659 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21660 instead of transforming it into USE.
21662 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21664 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21665 If HONOR_SNANS (SFmode) force the input to a register.
21666 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21667 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21668 an frsp or similar insn.
21670 2017-02-17 Martin Liska <mliska@suse.cz>
21672 PR rtl-optimization/79577
21673 * params.def (selsched-max-sched-times): Increase minimum to 1.
21675 2017-02-17 Martin Liska <mliska@suse.cz>
21677 PR rtl-optimization/79574
21678 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21680 2017-02-17 Martin Liska <mliska@suse.cz>
21682 PR tree-optimization/79529
21683 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21684 ssa_defined_default_def_p to handle cases which are implicitly
21686 * tree-ssa.c (ssa_defined_default_def_p): New function.
21687 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21688 which are implicitly defined.
21689 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21691 2017-02-17 Richard Biener <rguenther@suse.de>
21693 PR middle-end/79576
21694 * params.def (max-ssa-name-query-depth): Limit to 10.
21696 2017-02-17 Richard Biener <rguenther@suse.de>
21698 PR tree-optimization/79552
21699 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21702 2017-02-17 Richard Biener <rguenther@suse.de>
21705 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21707 2017-02-17 Marek Polacek <polacek@redhat.com>
21709 PR middle-end/79536
21710 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21711 (fold_negate_expr): New wrapper.
21713 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21715 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21716 Correct terminology and de-emphasize pre-standard behavior.
21718 2017-02-16 Alan Modra <amodra@gmail.com>
21720 PR rtl-optimization/79286
21721 * ira.c (def_dominates_uses): New function.
21722 (update_equiv_regs): Don't create an equivalence for insns that
21723 may trap where the register def does not dominate the use.
21725 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21727 PR rtl-optimization/78127
21728 * lra.c (lra): Call lra_eliminate before finish the loop after
21731 2017-02-16 Richard Biener <rguenther@suse.de>
21733 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21735 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21736 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21737 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21738 (isl_val_int_from_wi): New function.
21739 (extract_affine_gmp): Rename to ...
21740 (extract_affine_wi): ... this, take a widest_int.
21741 (extract_affine_int): Just wrap extract_affine_wi.
21742 (add_param_constraints): Use isl_val_int_from_wi.
21743 (add_loop_constraints): Likewise, and extract_affine_wi.
21745 2017-02-15 Jeff Law <law@redhat.com>
21747 PR middle-end/79521
21748 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21749 ira_init_register_move_cost_if_necessary.
21751 2017-02-15 Martin Sebor <msebor@redhat.com>
21753 PR middle-end/32003
21754 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21755 removed in a prior commit.
21757 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21759 PR tree-optimization/79347
21760 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21761 counters during peeling.
21763 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21765 * Makefile.in (site.exp): Remove "set ISLVER".
21767 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21770 * real.c (real_from_integer): Call real_convert even for decimal.
21772 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21775 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21777 2017-02-14 Andrew Pinski <apinski@cavium.com>
21779 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21780 cores and change the partno/implementer to be correct.
21781 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21782 the 'B" as the implementer.
21783 * config/aarch64/aarch64-tune.md: Regenerate.
21785 2017-02-14 Carl Love <cel@us.ibm.com>
21787 * config/rs6000/rs6000.c: Add case statement entry to make the
21788 xvcvuxdsp built-in argument unsigned.
21789 * config/rs6000/vsx.md: Fix the source and return operand types so they
21790 match the instruction definitions from the ISA document. Fix typo
21791 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21794 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21797 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21798 member early_clobber_alts.
21799 * lra-lives.c (reg_early_clobber_p): New.
21800 (process_bb_lives): Use it.
21801 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21802 (debug_operand_data): Initialize early_clobber_alts.
21803 (setup_operand_alternative): Set up early_clobber_alts.
21804 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21805 alternatives to new_insn_reg.
21806 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21808 (lra_update_insn_regno_info): Pass the new arg.
21810 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21812 PR middle-end/79505
21813 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21814 (new_oacc_loop_raw): Don't clear already cleared fields.
21817 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21818 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21819 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21821 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21824 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21826 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21829 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21830 the extra instruction to the right place to store 128-bit constant
21833 2017-02-14 Martin Sebor <msebor@redhat.com>
21835 PR middle-end/79448
21836 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21837 warning for strings of unknown length.
21839 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21841 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21843 2017-02-14 Jeff Law <law@redhat.com>
21846 * ira-costs.c (scan_one_insn): Initialize register move costs
21847 for pseudos seen in USE/CLOBBER insns.
21849 PR tree-optimization/79095
21850 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21851 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21852 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21853 if the operands are known to be not equal, then the resulting range
21855 (intersect_ranges): If the new range is ~[0,0] and the old range is
21856 wide, then prefer ~[0,0].
21857 * tree-vrp.c (overflow_comparison_p_1): New function.
21858 (overflow_comparison_p): New function.
21859 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21860 if NAME is used in an overflow test.
21861 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21862 overflow check that can be expressed as an equality test, then adjust
21863 ops to be that equality test.
21865 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21867 * config/s390/s390-builtin-types.def: Remove flags argument.
21868 * config/s390/s390.c (s390_init_builtins): Likewise.
21870 2017-02-14 Martin Liska <mliska@suse.cz>
21872 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21873 vector. Fix trailing white spaces.
21875 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21877 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21880 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21882 PR rtl-optimization/68664
21883 * config/arm/arm.c (arm_sched_can_speculate_insn):
21884 New function. Declare prototype.
21885 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21887 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21889 PR rtl-optimization/68664
21890 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21892 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21894 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21896 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21897 max skip bytes for function, loop and jump.
21899 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21901 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21902 ABS_EXPR for gimple dump.
21904 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21907 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21909 PR tree-optimization/79408
21910 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21911 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21912 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21913 also if rhs1 is INTEGER_CST.
21915 2017-02-14 Richard Biener <rguenther@suse.de>
21917 PR middle-end/79432
21918 * tree-into-ssa.c (insert_phi_nodes): When the function can
21919 have abnormal edges rewrite SSA names with broken use-def
21920 dominance out of SSA and register them for PHI insertion.
21922 2017-02-13 Martin Sebor <msebor@redhat.com>
21924 PR middle-end/79496
21925 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21926 clearing info.nowrite flag when snprintf size argument is a range.
21928 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21930 * cprop.c (cprop_jump): Add missing space in string literal.
21931 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21932 (get_constraint_for_component_ref): Likewise.
21933 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21934 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21935 * lra-constraints.c (process_alt_operands): Likewise.
21936 * ipa-inline.c (inline_small_functions): Likewise.
21937 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21938 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21939 * trans-mem.c (diagnose_tm_1_op): Likewise.
21940 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21941 (grid_parallel_clauses_gridifiable): Likewise.
21943 * config/nvptx/mkoffload.c (process): Add space in between
21946 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21947 "MOD4_SSE_REGS" and "ALL_REGS".
21949 * spellcheck.c (test_data): Add , in between "foo" and "food".
21951 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21954 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21955 boundary crossing check and subsequent code generation agree.
21957 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21959 * config/aarch64/aarch64.c (has_memory_op): Delete.
21960 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21963 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21965 PR rtl-optimization/79388
21966 PR rtl-optimization/79450
21967 * combine.c (distribute_notes): When removing TEM_INSN for which
21968 corresponding dest has last value recorded, invalidate that last
21971 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21973 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21974 of explicit '@'. Add missing assembly comment marker on branch costs
21977 2017-02-13 Nathan Sidwell <nathan@acm.org>
21979 * gengtype-lex.l (<in_struct>): Add '/'.
21981 2017-02-13 Martin Liska <mliska@suse.cz>
21984 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21986 2017-02-13 Richard Biener <rguenther@suse.de>
21988 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21990 * configure: Re-generate.
21991 * config.in: Likewise.
21992 * graphite-dependences.c: Simplify as if
21993 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21994 * graphite-isl-ast-to-gimple.c: Likewise.
21995 * graphite-optimize-isl.c: Likewise.
21996 * graphite-poly.c: Likewise.
21997 * graphite-sese-to-poly.c: Likewise.
21998 * graphite.h: Likewise.
21999 * toplev.c: Include isl/version.h and use isl_version () for
22000 printing the ISL version.
22001 * doc/install.texi: Update ISL requirement.
22003 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22005 * doc/standards.texi (Standards): Update reference to
22008 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22010 * doc/extend.texi (Named Address Spaces): sourceware.org now
22012 * doc/install.texi (Binaries): Ditto.
22015 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22017 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22018 standard terminology "stringize"/"stringizing" throughout.
22019 * doc/cppinternals.texi: Likewise.
22021 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22023 * doc/extend.texi: Fix some spelling mistakes and typos.
22024 * doc/invoke.texi: Likewise.
22026 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22029 * params.def (inline-min-speedup) Change from 10 to 8.
22031 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22033 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22036 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22039 * ipa-inline-analysis.c (get_minimal_bb): New function.
22040 (record_modified): Use it.
22041 (remap_edge_change_prob): Handle also ancestor functions.
22043 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22045 * doc/contrib.texi (Contributors): Remove broken link into
22046 the Mauve CVS repository.
22048 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22050 PR middle-end/79454
22051 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22052 result computation whenever lhs doesn't have vector mode, not
22053 just when it has BLKmode.
22055 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22057 * doc/makefile.texi (profiledbootstrap): Refer to the
22058 installation instructions only in textual form.
22060 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22063 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22065 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22067 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22068 (Specific): Update mingw-w64 reference.
22070 (Specific): Remove broken link to Renesas RX processor.
22072 2017-02-10 Richard Biener <rguenther@suse.de>
22074 * toplev.c (process_options): Do not mention obsolete graphite
22075 options when printing sorry message about missing graphite support.
22076 Mention -floop-nest-optimize.
22078 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22080 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22081 (vtst_p16): Likewise.
22082 (vtstq_p8): Likewise.
22083 (vtstq_p16): Likewise.
22085 (vtstq_p64): Likewise.
22086 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22087 (vset_lane_p64): New.
22088 (vsetq_lane_p64): New.
22090 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22092 PR tree-optimization/79411
22093 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22094 stmt operands are SSA_NAMEs used in abnormal phis.
22095 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22098 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22101 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22104 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22106 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22108 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22110 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22114 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22115 not arbitrary TREE_CONSTANT.
22118 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22119 "omp declare target link" attribute unless is_global_var.
22120 * omp-offload.c (find_link_var_op): Likewise.
22122 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22123 Chung-Lin Tang <cltang@codesourcery.com>
22125 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22127 (gimplify_adjust_omp_clauses): Don't delete TILE.
22128 (gimplify_omp_for): Deal with TILE.
22129 * internal-fn.c (expand_GOACC_TILE): New function.
22130 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22132 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22133 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22135 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22136 avoid DIV for outermost collapse var.
22137 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22138 Remove out of date comments, fix whitespace.
22139 * omp-general.c (omp_extract_for_data): Deal with tiling.
22140 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22141 adjust OLF_DIM_BASE value.
22142 (struct omp_for_data): Add tiling field.
22143 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22144 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22145 for auto loops. Remove default auto determining, moved to
22146 oacc_loop_fixed_partitions.
22147 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22148 stmts, add e_mask field.
22149 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22150 (oacc_thread_numbers): Use oacc_dim_call.
22151 (oacc_xform_tile): New.
22152 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22153 (finish_oacc_loop): Adjust for ifns vector.
22154 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22155 add case for GOACC_TILE fns.
22156 (oacc_loop_xform_loop): Delete.
22157 (oacc_loop_process): Iterate over call list directly, and add
22158 handling for GOACC_TILE fns.
22159 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22161 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22162 vector partitioning to outer loops. Assign 2 partitions to loops
22163 when available. Add TILE handling.
22164 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22165 (execite_oacc_device_lower): Process GOACC_TILE fns,
22166 ignore unknown specs.
22167 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22168 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22169 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22171 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22173 * configure.ac (ACX_BUGURL): Update.
22174 * configure: Regenerate.
22176 2017-02-09 Richard Biener <rguenther@suse.de>
22178 PR tree-optimization/69823
22179 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22180 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22182 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22184 * config/arc/arc-c.def: Add __NPS400__ definition.
22185 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22186 (TARGET_NPS400): Define.
22188 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22190 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22192 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22193 pointer, arch_info.
22194 (arc_cpu_types): Fill the arch_info field with a pointer into the
22195 arc_arch_types table.
22196 (arc_selected_cpu): Declare.
22197 * config/arc/arc.c (arc_selected_cpu): Make global.
22198 (arc_selected_arch): Delete.
22199 (arc_base_cpu): Delete.
22200 (arc_override_options): Remove references to deleted variables,
22201 update access to arch information.
22202 (ARC_OPT): Update access to arch information.
22203 (ARC_OPTX): Likewise.
22204 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22205 (TARGET_ARC600): Update access to arch information.
22206 (TARGET_ARC601): Likewise.
22207 (TARGET_ARC700): Likewise.
22208 (TARGET_EM): Likewise.
22209 (TARGET_HS): Likewise.
22210 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22213 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22216 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22217 condition/operands for integer GE/LE/GEU/LEU operations.
22219 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22221 PR translation/79397
22222 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22225 2017-02-08 Martin Jambor <mjambor@suse.cz>
22228 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22229 whether allocation happened.
22230 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22231 nothing was allocated.
22233 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22235 PR tree-optimization/79408
22236 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22237 constant, but SSA_NAME with a known integer range, use the minimum
22238 of that range instead of op1 to determine if modulo can be replaced
22239 with its first operand.
22241 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22243 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22245 2017-02-08 Richard Biener <rguenther@suse.de>
22247 PR tree-optimization/71824
22248 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22249 Check all loops contained in the merged region.
22251 2017-02-07 Andrew Pinski <apinski@cavium.com>
22253 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22255 2017-02-07 Andrew Pinski <apinski@cavium.com>
22257 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22258 (thunderxt88): Likewise.
22259 (thunderxt81): Disable LSE and change v8.1 to v8.
22260 (thunderxt83): Likewise.
22262 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22263 Richard Biener <rguenther@suse.de>
22265 PR middle-end/79399
22266 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22267 type from int to size_t.
22268 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22270 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22272 PR rtl-optimization/79386
22273 * cprop.c (bypass_conditional_jumps): Initialize
22274 bypass_last_basic_block already before splitting bbs after
22275 unconditional traps...
22276 (bypass_conditional_jumps): ... rather than here.
22279 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22280 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22281 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22282 fix -masm=intel patterns.
22284 2017-02-07 Richard Biener <rguenther@suse.de>
22286 PR tree-optimization/79256
22287 PR middle-end/79278
22288 * builtins.c (get_object_alignment_2): Use min_align_of_type
22289 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22290 and ADJUST_FIELD_ALIGN.
22292 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22294 * doc/tm.texi: Regenerate.
22295 * stor-layout.c (layout_decl): Adjust.
22296 (update_alignment_for_field): Likewise.
22297 (place_field): Likewise.
22298 (min_align_of_type): Likewise.
22299 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22300 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22301 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22302 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22303 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22304 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22305 * config/i386/i386.c (x86_field_alignment): Likewise.
22306 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22307 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22308 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22309 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22310 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22311 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22315 2017-01-30 Richard Biener <rguenther@suse.de>
22317 PR tree-optimization/79256
22318 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22319 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22322 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22324 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22325 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22326 builtins to SImode and emit a zero-extend, if necessary.
22328 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22330 * docs/invoke.texi (RISC-V Options): Alphabetize.
22332 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22334 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22337 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22339 * config/riscv/riscv.c: New file.
22340 * common/config/riscv/riscv-common.c: Likewise.
22341 * config.gcc: Likewise.
22342 * config/riscv/constraints.md: Likewise.
22343 * config/riscv/elf.h: Likewise.
22344 * config/riscv/generic.md: Likewise.
22345 * config/riscv/linux.h: Likewise.
22346 * config/riscv/multilib-generator: Likewise.
22347 * config/riscv/peephole.md: Likewise.
22348 * config/riscv/pic.md: Likewise.
22349 * config/riscv/predicates.md: Likewise.
22350 * config/riscv/riscv-builtins.c: Likewise.
22351 * config/riscv/riscv-c.c: Likewise.
22352 * config/riscv/riscv-ftypes.def: Likewise.
22353 * config/riscv/riscv-modes.def: Likewise.
22354 * config/riscv/riscv-opts.h: Likewise.
22355 * config/riscv/riscv-protos.h: Likewise.
22356 * config/riscv/riscv.h: Likewise.
22357 * config/riscv/riscv.md: Likewise.
22358 * config/riscv/riscv.opt: Likewise.
22359 * config/riscv/sync.md: Likewise.
22360 * config/riscv/t-elf-multilib: Likewise.
22361 * config/riscv/t-linux: Likewise.
22362 * config/riscv/t-linux-multilib: Likewise.
22363 * config/riscv/t-riscv: Likewise.
22364 * configure.ac: Likewise.
22365 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22366 Waterman as RISC-V maintainers.
22367 * doc/install.texi: Add RISC-V entries.
22368 * doc/invoke.texi: Add RISC-V options section.
22369 * doc/md.texi: Add RISC-V constraints section.
22370 * configure: Regenerated.
22372 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22375 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22376 false values to be constant vectors with all 0 or all 1 bits set.
22377 (vcondu<mode><mode>): Likewise.
22378 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22380 (fpmask_comparison_operator): Update comment.
22381 (vecint_comparison_operator): New predicate.
22382 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22383 vector conditionals when the true and false values are constant
22384 vectors with all 0 bits or all 1 bits set.
22386 2017-02-06 Martin Sebor <msebor@redhat.com>
22388 PR tree-optimization/79376
22389 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22391 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22393 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22394 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22395 to simplify split condition.
22397 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22399 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22400 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22403 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22405 PR rtl-optimization/68664
22406 * target.def (can_speculate_insn): New hook.
22407 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22408 * doc/tm.texi: Regenerate.
22409 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22410 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22411 (rs6000_sched_can_speculate_insn): New function.
22413 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22415 PR tree-optimization/79284
22416 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22417 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22418 vectorizable_mask_load_store, vectorizable_operation,
22419 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22420 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22421 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22422 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22423 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22424 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22425 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22426 is_gimple_assign (stmt). Replace another such test with
22427 is_gimple_assign (stmt).
22429 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22432 * config/avr/avr.c (rtl-iter.h): Include it.
22433 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22434 (avr_legitimate_combined_insn): ...and implementation.
22436 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22438 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22439 * config/s390/s390.c (s390_const_operand_ok)
22440 (s390_canonicalize_comparison, s390_extract_part)
22441 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22442 (s390_contiguous_bitmask_p, s390_rtx_costs)
22443 (legitimize_pic_address): Likewise.
22444 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22445 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22446 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22447 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22448 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22450 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22452 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22453 REGNO($0) == REGNO($1).
22455 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22457 * config/s390/linux.h(SIZE_TYPE): Add comment.
22459 2017-02-06 Julian Brown <julian@codesourcery.com>
22460 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22461 Virendra Pathak <virendra.pathak@broadcom.com>
22463 * config/aarch64/aarch64-cores.def: Change the scheduler
22465 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22466 * config/aarch64/thunderx2t99.md: New file.
22468 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22470 * doc/standards.texi (Go Language): Update link to language
22473 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22475 * tree-eh.c (lower_resx): Sanitize profile.
22476 (cleanup_empty_eh_move_lp): Likewise.
22478 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22481 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22483 * cfgloopmanip.h (loop_version): Update prototype.
22484 * modulo-sched.c (sms_schedule): Update call of loop_version.
22485 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22486 * tree-parloops.c (gen_parallel_loop): Likewise.
22487 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22488 * tree-ssa-loop-split.c (split_loop): Likewise.
22489 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22490 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22492 2017-02-05 Martin Liska <mliska@suse.cz>
22495 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22497 (print_operand_address): Initialize a struct to zero.
22499 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22501 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22502 garbage collector only in textual form.
22504 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22506 * doc/extend.texi (x86 specific memory model extensions for
22507 transactional memory): Simplify a phrase.
22509 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22512 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22513 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22514 (atomic_storedi_1): Likewise.
22516 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22518 PR tree-optimization/79338
22519 * tree-parloops.c (gather_scalar_reductions): Don't call
22520 vect_analyze_loop_form for loop->inner before destroying loop's
22523 2017-02-03 Martin Sebor <msebor@redhat.com>
22525 PR tree-optimization/79327
22526 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22527 when precision has resulted in leading zeros.
22528 (format_integer): Adjust the likely counter to assume an unknown
22529 argument that may be zero is non-zero.
22531 2017-02-03 Jason Merrill <jason@redhat.com>
22534 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22535 avoid copying non-taken branch.
22537 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22539 PR tree-optimization/79340
22540 * tree-vect-loop.c (vectorizable_reduction): Release
22541 vec_defs elements after safe_splicing them into other vectors.
22544 PR tree-optimization/79327
22545 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22546 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22548 (format_integer): Use wide_int_to_tree instead of build_int_cst
22549 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22550 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22551 of shortest and longest sequence.
22553 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22555 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22556 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22558 2017-02-03 Walter Lee <walt@tilera.com>
22561 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22562 after initial stackframe link reg save.
22563 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22565 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22568 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22569 wu for stxssp alternative.
22571 2017-02-03 Martin Sebor <msebor@redhat.com>
22573 PR tree-optimization/79352
22574 * gimple-fold.c (get_range_strlen): Add argument.
22575 (get_range_strlen): Change return type to bool.
22576 (get_maxval_strlen): Pass in a dummy argument.
22577 * gimple-fold.h (get_range_strlen): Change return type to bool.
22578 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22579 * tree.h (array_at_struct_end_p): Add argument.
22580 * tree.c (array_at_struct_end_p): Handle it.
22582 2017-02-03 Martin Liska <mliska@suse.cz>
22585 * multiple_target.c (create_dispatcher_calls): Redirect edge
22586 from a caller of a dispatcher.
22587 (expand_target_clones): Make the clones local.
22588 (ipa_target_clone): Do both target clones and resolvers.
22589 (ipa_dispatcher_calls): Remove the pass.
22590 (pass_dispatcher_calls::gate): Likewise.
22591 (make_pass_dispatcher_calls): Likewise.
22592 * passes.def (pass_target_clone): Put as very first IPA early
22595 2017-02-03 Martin Liska <mliska@suse.cz>
22597 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22598 in case of a function with ifunc attribute.
22600 2017-02-03 Martin Liska <mliska@suse.cz>
22602 * cgraph.c (cgraph_node::dump): Dump function version info.
22603 * symtab.c (symtab_node::dump_base): Add missing new line.
22605 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22607 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22608 (ifcombine_ifandif): Use it.
22610 2017-02-03 Martin Liska <mliska@suse.cz>
22612 * doc/invoke.texi: Document default value for
22613 use-after-scope-direct-emission-threshold.
22615 2017-02-03 Martin Liska <mliska@suse.cz>
22617 PR tree-optimization/79339
22618 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22619 (format_floating): Likewise.
22621 2017-02-03 Martin Liska <mliska@suse.cz>
22624 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22626 (ipa_node_params_t::remove): Likewise.
22627 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22628 initialization from removed ipa_node_params_t::insert.
22629 (ipa_node_params::~ipa_node_params): Move from removed
22630 ipa_node_params_t::release.
22631 * symbol-summary.h (symbol_summary::m_released): New member.
22632 Do not release a summary twice. Do not allow to call finalizer
22633 for types of a summary that live in GGC memory.
22635 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22637 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22640 2017-02-02 Martin Sebor <msebor@redhat.com>
22642 PR middle-end/79275
22643 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22644 (format_string): Tighten up the range of output for non-constant
22645 strings and correct the expected range for wide non-constant strings.
22647 2017-02-02 Martin Sebor <msebor@redhat.com>
22649 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22651 PR middle-end/32003
22652 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22654 (-fdump-tree-@var): Add to index and document how to come up
22655 with pass-specific option and dump file names.
22656 (-fdump-passes): Clarify where to look for output.
22658 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22660 PR middle-end/77445
22661 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22662 statistics of the analyzed path; allow threading for speed when
22663 any of BBs along the path are optimized for speed.
22665 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22667 PR middle-end/78468
22668 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22669 settings of the virtual registers.
22672 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22674 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22675 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22678 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22680 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22681 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22683 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22685 * config/s390/s390.md: Add missing comments with the expanded
22687 * config/s390/vector.md: Likewise.
22688 * config/s390/vx-builtins.md: Likewise.
22690 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22693 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22694 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22695 conditions on a single line.
22697 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22699 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22700 __S390_VX__ to __VX__.
22702 2017-02-01 Andrew Pinski <apinski@cavium.com>
22704 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22705 stmt_info to record_stmt_cost.
22706 (vect_get_known_peeling_cost): Pass stmt_info if known to
22708 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22709 cpu_vector_cost field into
22710 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22711 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22712 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22713 splitting of scalar_stmt_cost and vec_stmt_cost.
22714 (thunderx_vector_cost): Likewise.
22715 (cortexa57_vector_cost): LIkewise.
22716 (exynosm1_vector_cost): Likewise.
22717 (xgene1_vector_cost): Likewise.
22718 (thunderx2t99_vector_cost): Improve after the splitting of the two
22720 (aarch64_builtin_vectorization_cost): Update for the splitting of
22721 scalar_stmt_cost and vec_stmt_cost.
22723 2017-02-01 Torvald Riegel <triegel@redhat.com>
22724 Richard Henderson <rth@redhat.com>
22726 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22727 conditional on existance of a fast atomic load.
22728 * optabs-query.c (can_atomic_load_p): New function.
22729 * optabs-query.h (can_atomic_load_p): Declare it.
22730 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22731 no fast atomic load is available for the particular size of access.
22732 (expand_atomic_compare_and_swap): Likewise.
22733 (expand_atomic_load): Likewise.
22734 (expand_atomic_store): Likewise.
22735 (expand_atomic_fetch_op): Likewise.
22736 * testsuite/lib/target-supports.exp
22737 (check_effective_target_sync_int_128): Remove x86 because it provides
22738 no fast atomic load.
22739 (check_effective_target_sync_int_128_runtime): Likewise.
22741 2017-02-01 Richard Biener <rguenther@suse.de>
22743 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22744 (graphite_transform_loops): Provide opt-info for optimized nests.
22745 * tree-parloop.c (parallelize_loops): Provide opt-info for
22746 parallelized loops.
22748 2017-02-01 Richard Biener <rguenther@suse.de>
22750 PR middle-end/79315
22751 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22752 was not set before.
22754 2017-02-01 Richard Biener <rguenther@suse.de>
22756 PR tree-optimization/71824
22757 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22758 Verify the loops are valid in the merged SESE region.
22759 (scop_detection::can_represent_loop_1): Check analyzing the
22760 evolution of the number of iterations in the region succeeds.
22762 2017-01-31 Ian Lance Taylor <iant@golang.org>
22764 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22765 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22767 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22769 PR preprocessor/79210
22770 * input.c (get_substring_ranges_for_loc): Replace line_width
22771 assertion with error-handling.
22773 2017-01-31 Richard Biener <rguenther@suse.de>
22775 PR tree-optimization/77318
22776 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22777 (create_pw_aff_from_tree): Take loop parameter.
22778 (add_condition_to_pbb): Pass loop of the condition to
22779 create_pw_aff_from_tree.
22781 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22783 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22784 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22786 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22790 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22792 (convert_int_to_float128): Likewise.
22793 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22794 (convert_int_to_float128): Likewise.
22795 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22796 (UNSPEC_IEEE128_CONVERT): Likewise.
22797 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22798 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22799 Use local variables for IBM extended format.
22800 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22801 (fix_trunc<mode>si2_fprs): Likewise.
22802 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22803 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22804 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22805 to know that we can now have integers of all sizes in vector
22807 (fix<uns>_<mode>di2_hw): Likewise.
22808 (float<uns>_<mode>si2_hw): Likewise.
22809 (fix_<mode>si2_hw): Likewise.
22810 (fixuns_<mode>si2_hw): Likewise.
22811 (float<uns>_<mode>di2_hw): Likewise.
22812 (float_<mode>di2_hw): Likewise.
22813 (float_<mode>si2_hw): Likewise.
22814 (floatuns_<mode>di2_hw): Likewise.
22815 (floatuns_<mode>si2_hw): Likewise.
22816 (xscvqp<su>wz_<mode>): Delete, no longer used.
22817 (xscvqp<su>dz_<mode>): Likewise.
22818 (xscv<su>dqp_<mode>): Likewise.
22819 (ieee128_mfvsrd_64bit): Likewise.
22820 (ieee128_mfvsrd_32bit): Likewise.
22821 (ieee128_mfvsrwz): Likewise.
22822 (ieee128_mtvsrw): Likewise.
22823 (ieee128_mtvsrd_64bit): Likewise.
22824 (ieee128_mtvsrd_32bit): Likewise.
22826 2017-01-31 Martin Liska <mliska@suse.cz>
22829 * ipa-prop.c (ipa_free_all_node_params): Call release method
22830 instead of ~sumbol_summary to not to trigger double times
22833 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22835 PR tree-optimization/71691
22836 * bitmap.h (class auto_bitmap): New.
22837 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22838 is_maybe_undefined instead of ssa_undefined_value_p.
22840 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22842 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22843 __S390_ARCH_LEVEL__ to __ARCH__.
22845 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22847 PR tree-optimization/79267
22848 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22849 if should_remove_lhs_p is true.
22851 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22854 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22855 (add_alignment_attribute): New.
22856 (base_type_die): Add alignment attribute.
22857 (subrange_type_die): Likewise.
22858 (modified_type_die): Likewise.
22859 (gen_array_type_die): Likewise.
22860 (gen_descr_array_type_die: Likewise.
22861 (gen_enumeration_type_die): Likewise.
22862 (gen_subprogram_die): Likewise.
22863 (gen_variable_die): Likewise.
22864 (gen_field_die): Likewise.
22865 (gen_ptr_to_mbr_type_die): Likewise.
22866 (gen_struct_or_union_type_die): Likewise.
22867 (gen_subroutine_type_die): Likewise.
22868 (gen_typedef_die): Likewise.
22869 (base_type_cmp): Compare alignment attribute.
22871 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22874 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22875 (setb_unsigned) New pattern for setb with CCUNS.
22876 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22877 subfc./subfe sequence to avoid overflow problems. Generate a
22878 shorter sequence with cmpld/setb for power9.
22879 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22880 for generating subfc. instruction.
22881 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22882 now uses this instruction.
22884 2017-01-30 Ian Lance Taylor <iant@google.com>
22887 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22888 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22890 2017-01-30 Martin Sebor <msebor@redhat.com>
22892 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22893 Move constant to the right of a relational operator.
22894 (get_mpfr_format_length, format_character, format_string): Ditto.
22895 (should_warn_p, maybe_warn): Same.
22897 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22899 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22902 * asan.c (get_translation_unit_decl): Remove function.
22903 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22905 2017-01-30 Martin Liska <mliska@suse.cz>
22907 PR gcov-profile/79259
22908 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22909 -fprofile-generate.
22911 2017-01-30 Martin Liska <mliska@suse.cz>
22914 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22915 Initialize variables with NULL value.
22917 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22920 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22922 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22924 2017-01-30 Richard Biener <rguenther@suse.de>
22926 PR tree-optimization/79276
22927 * tree-vrp.c (process_assert_insertions): Properly adjust common
22928 when removing a duplicate.
22930 * gcc.dg/torture/pr79276.c: New testcase.
22932 2017-01-30 Richard Biener <rguenther@suse.de>
22934 PR tree-optimization/79256
22935 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22936 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22938 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22940 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22943 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22944 ("*r<noxa>sbg_<mode>_sll_bitmask")
22945 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22946 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22947 Use contiguous_bitmask_nowrap_operand.
22949 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22952 * config/rs6000/altivec.h (vec_xl): Revise #define.
22953 (vec_xst): Likewise.
22955 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22957 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22959 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22961 PR rtl-optimization/79194
22962 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22963 traps before call to bypass_conditional_jumps.
22965 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22967 PR tree-optimization/71374
22968 * lra-constraints.c (check_conflict_input_operands): New.
22969 (match_reload): Use it.
22971 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22974 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22975 account to calculate conflict_set.
22977 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22979 PR rtl-optimization/78559
22980 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22981 other_insn in combine.
22983 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22985 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22986 uint16_type_node for BT_UINT16.
22988 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22990 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22991 "RTL Tests" to menu.
22992 (GIMPLE Tests): New node.
22993 (RTL Tests): New node.
22995 2017-01-27 Richard Biener <rguenther@suse.de>
22997 PR tree-optimization/79245
22998 * tree-loop-distribution.c (distribute_loop): Apply cost
22999 modeling also to detected patterns.
23001 2017-01-27 Richard Biener <rguenther@suse.de>
23003 PR tree-optimization/71433
23004 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23005 (compare_assert_loc): New function.
23006 (process_assert_insertions): Sort and optimize assert locations
23007 to remove duplicates and push down identical assertions on
23008 edges to their destination block.
23010 2017-01-27 Richard Biener <rguenther@suse.de>
23012 PR tree-optimization/79244
23013 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23014 out SSA names even if abnormal.
23016 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23018 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23019 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23020 instead of MPFR_RNDN.
23022 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23025 * arm.c (arm_option_override): Don't call build_target_option_node
23026 until after doing all option overrides.
23027 (arm_valid_target_attribute_tree): Likewise.
23029 2017-01-27 Martin Liska <mliska@suse.cz>
23031 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23032 for {cd}tors and C++ {cd}tors.
23034 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23036 * config/s390/s390.md ("*setmem_long_and")
23037 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23039 2017-01-26 Martin Sebor <msebor@redhat.com>
23041 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23044 2017-01-26 Martin Sebor <msebor@redhat.com>
23046 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23047 HAVE_DFmode before using XFmode or DFmode.
23048 (parse_directive): Avoid using the z length modifier to avoid
23049 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23051 PR middle-end/78703
23052 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23053 to accept adjustment as an array.
23054 (get_int_range): New function.
23055 (struct directive): Make width and prec arrays.
23056 (directive::set_width, directive::set_precision): Call get_int_range.
23057 (format_integer, format_floating): Handle width and precision ranges.
23058 (format_string, parse_directive): Same.
23060 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23063 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23064 comdat_type_p set, just clone them, but keep the children in the
23068 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23069 which have direct callers with -fvar-tracking-assignments enabled
23071 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23072 inside of type units.
23074 2017-01-26 Martin Sebor <msebor@redhat.com>
23076 PR middle-end/78703
23077 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23079 (struct format_result): Replace number_chars, number_chars_min,
23080 and number_chars_max with a single member of struct result_range.
23082 (format_result::operator+=): Adjust.
23083 (struct fmtresult): Remove bounded. Handle likely and unlikely
23085 (fmtresult::adjust_for_width_or_precision): New function.
23086 (fmtresult:type_max_digits): New function.
23087 (bytes_remaining): Handle likely and unlikely counters.
23088 (min_bytes_remaining): Remove.
23089 (format_percent): Simplify.
23090 (format_integer, format_floating): Set likely and unlikely counters.
23091 (get_string_length, format_character, format_string): Same.
23092 (format_plain, should_warn_p): New function.
23093 (maybe_warn): Call should_warn_p. Update diagnostic messages
23094 and handle those for all directives, including plain strings.
23095 (format_directive): Handle likely and unlikely counters.
23096 Remove unnecessary quoting from diagnostics. Add an informational
23098 (add_bytes): Remove.
23099 (pass_sprintf_length::compute_format_length): Simplify.
23100 (try_substitute_return_value): Handle likely and unlikely counters.
23102 2017-01-26 Carl Love <cel@us.ibm.com>
23104 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23105 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23107 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23110 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23111 endianess for subregs into account.
23112 * lra-constraints.c (lra_constraints): Do risky transformations
23113 always on the first iteration.
23114 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23115 last_call_used_reg_set.
23116 (process_bb_lives): Define and use last_call_used_reg_set.
23117 * lra.c (lra): Always continue after lra_constraints on the first
23120 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23122 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23124 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23126 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23128 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23129 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23130 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23131 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23132 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23133 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23134 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23135 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23136 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23138 2017-01-26 Marek Polacek <polacek@redhat.com>
23141 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23142 for the third operand.
23144 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23146 PR middle-end/79236
23147 * omp-low.c (struct omp_context): Add simt_stmt field.
23148 (scan_omp_for): Return omp_context *.
23149 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23150 context to the _simt_ SIMD stmt.
23151 (lower_omp_for): For combined SIMD with sibling _simt_
23152 SIMD, make sure to use the same decls in _looptemp_
23153 clauses as in the sibling.
23155 2017-01-26 David Sherwood <david.sherwood@arm.com>
23157 PR middle-end/79212
23158 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23161 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23164 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23165 emit fld b; fld a; if possible.
23167 * brig-builtins.def: Update copyright years.
23168 * config/arm/arm_acle_builtins.def: Update copyright years.
23170 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23173 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23174 constraint instead of o for the stxsd instruction.
23176 2017-01-25 Carl Love <cel@us.ibm.com>
23178 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23179 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23181 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23183 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23185 2017-01-25 Richard Biener <rguenther@suse.de>
23187 PR tree-optimization/69264
23188 * target.def (vector_alignment_reachable): Improve documentation.
23189 * doc/tm.texi: Regenerate.
23190 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23192 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23193 earlier changes with respect to TYPE_USER_ALIGN.
23194 (vector_alignment_reachable_p): Likewise. Improve dumping.
23196 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23199 * config/arm/arm.md (xordi3): Force constant operand into a register
23202 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23204 * doc/invoke.texi (-fstore-merging): Correct default optimization
23205 levels at which it is enabled.
23206 (-O): Move -fstore-merging from list to...
23209 2017-01-25 Richard Biener <rguenther@suse.de>
23212 * omp-expand.c: Include debug.h.
23213 (expand_omp_taskreg): Make sure to generate early debug before
23214 outlining anything from a function.
23215 (expand_omp_target): Likewise.
23216 (grid_expand_target_grid_body): Likewise.
23218 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23221 * asan.c (get_translation_unit_decl): New function.
23222 (asan_add_global): Extract modules file name from globals
23223 TRANSLATION_UNIT_DECL name.
23225 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23228 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23229 for long calls with APCS frame and VFP.
23231 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23233 * cfg.c (original_copy_tables_initialized_p): New function.
23234 * cfg.h (original_copy_tables_initialized_p): New decl.
23235 * cfgrtl.c (relink_block_chain): Guard the call to
23236 free_original_copy_tables with a call to
23237 original_copy_tables_initialized_p.
23238 * cgraph.h (symtab_node::native_rtl_p): New decl.
23239 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23240 (symtab_node::needed_p): Don't assert for early assembly output
23241 for __RTL functions.
23242 (cgraph_node::finalize_function): Set "force_output" for __RTL
23244 (cgraph_node::analyze): Bail out early for __RTL functions.
23245 (analyze_functions): Update assertion to support __RTL functions.
23246 (cgraph_node::expand): Bail out early for __RTL functions.
23247 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23249 * function.h (struct function): Update comment for field
23251 * gimple-expr.c: Include "tree-pass.h".
23252 (gimple_has_body_p): Return false for __RTL functions.
23253 * Makefile.in (OBJS): Add run-rtl-passes.o.
23254 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23256 (gcc::pass_manager::get_clean_slate): New accessor.
23257 * passes.c: Include "insn-addr.h".
23258 (should_skip_pass_p): Add logging. Update logic for running
23259 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23260 property-provider override so it is only done for gimple passes.
23262 (skip_pass): New function.
23263 (execute_one_pass): Call skip_pass when skipping passes.
23264 * read-md.c (md_reader::read_char): Support filtering
23265 the input to a subset of line numbers.
23266 (md_reader::md_reader): Initialize fields
23267 m_first_line and m_last_line.
23268 (md_reader::read_file_fragment): New function.
23269 * read-md.h (md_reader::read_file_fragment): New decl.
23270 (md_reader::m_first_line): New field.
23271 (md_reader::m_last_line): New field.
23272 * read-rtl-function.c (function_reader::create_function): Only
23273 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23274 curr_properties. Set DECL_INITIAL to a dummy block.
23275 (read_rtl_function_body_from_file_range): New function.
23276 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23278 * run-rtl-passes.c: New file.
23279 * run-rtl-passes.h: New file.
23281 2017-01-24 Jeff Law <law@redhat.com>
23283 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23286 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23288 PR tree-optimization/79159
23289 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23290 (record_nonwrapping_iv): Improve boundary using above function if no
23291 value range information.
23293 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23294 Martin Jambor <mjambor@suse.cz>
23296 * brig-builtins.def: New file.
23297 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23298 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23299 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23300 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23301 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23302 * builtin-types.def (BT_INT8): New.
23303 (BT_INT16): Likewise.
23304 (BT_UINT8): Likewise.
23305 (BT_UINT16): Likewise.
23306 (BT_FN_ULONG): Likewise.
23307 (BT_FN_UINT_INT): Likewise.
23308 (BT_FN_UINT_ULONG): Likewise.
23309 (BT_FN_UINT_LONG): Likewise.
23310 (BT_FN_UINT_PTR): Likewise.
23311 (BT_FN_ULONG_PTR): Likewise.
23312 (BT_FN_INT8_FLOAT): Likewise.
23313 (BT_FN_INT16_FLOAT): Likewise.
23314 (BT_FN_UINT32_FLOAT): Likewise.
23315 (BT_FN_UINT16_FLOAT): Likewise.
23316 (BT_FN_UINT8_FLOAT): Likewise.
23317 (BT_FN_UINT64_FLOAT): Likewise.
23318 (BT_FN_UINT16_UINT32): Likewise.
23319 (BT_FN_UINT32_UINT16): Likewise.
23320 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23321 (BT_FN_INT_PTR_INT): Likewise.
23322 (BT_FN_UINT_PTR_UINT): Likewise.
23323 (BT_FN_LONG_PTR_LONG): Likewise.
23324 (BT_FN_ULONG_PTR_ULONG): Likewise.
23325 (BT_FN_VOID_UINT64_UINT64): Likewise.
23326 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23327 (BT_FN_INT8_INT8_INT8): Likewise.
23328 (BT_FN_INT16_INT16_INT16): Likewise.
23329 (BT_FN_INT_INT_INT): Likewise.
23330 (BT_FN_UINT_FLOAT_UINT): Likewise.
23331 (BT_FN_FLOAT_UINT_UINT): Likewise.
23332 (BT_FN_ULONG_UINT_UINT): Likewise.
23333 (BT_FN_ULONG_UINT_PTR): Likewise.
23334 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23335 (BT_FN_UINT_UINT_UINT): Likewise.
23336 (BT_FN_VOID_UINT_PTR): Likewise.
23337 (BT_FN_UINT_UINT_PTR: Likewise.
23338 (BT_FN_UINT32_UINT64_PTR): Likewise.
23339 (BT_FN_INT_INT_UINT_UINT): Likewise.
23340 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23341 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23342 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23343 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23344 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23345 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23346 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23347 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23348 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23349 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23350 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23351 * doc/frontends.texi: List BRIG FE.
23352 * doc/install.texi (Testing): Add BRIG tesring requirements.
23353 * doc/invoke.texi (Overall Options): Mention BRIG.
23354 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23356 2017-01-24 Richard Biener <rguenther@suse.de>
23358 PR translation/79208
23359 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23361 2017-01-24 Martin Jambor <mjambor@suse.cz>
23364 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23365 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23366 and known_contexts.
23368 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23370 PR middle-end/79123
23371 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23372 casts from signed to unsigned really don't have a range.
23374 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23376 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23377 GMP_RNDx for compatiblity.
23379 2017-01-24 Martin Liska <mliska@suse.cz>
23382 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23383 that would prevent us to call alloca with -1 as argument.
23385 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23387 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23388 Avoid -Wformat-security warning.
23390 2017-01-23 Andrew Pinski <apinski@cavium.com>
23392 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23395 2017-01-23 Martin Sebor <msebor@redhat.com>
23397 PR middle-end/78703
23398 * gimple-ssa-sprintf.c (warn_level): New global.
23399 (format_integer): Use it here and throughout the rest of the file.
23400 Use the same switch to compute sign as base.
23401 (maybe_warn): New function.
23402 (format_directive): Factor out warnings into maybe_warn.
23403 Add debugging output. Use warn_level.
23404 (add_bytes): Use warn_level.
23405 (pass_sprintf_length::compute_format_length): Add debugging output.
23406 (try_substitute_return_value): Same.
23407 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23409 PR middle-end/78703
23410 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23411 (struct fmtresult, format_integer, format_floating): Adjust.
23412 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23413 (get_string_length, format_string,format_directive): Same.
23414 (pass_sprintf_length::compute_format_length): Same.
23415 (try_substitute_return_value): Simplify slightly.
23417 PR middle-end/78703
23418 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23419 (fmtresult::operator+=): Outlined.
23420 (struct fmtresult): Add ctors.
23421 (struct conversion_spec): Rename...
23422 (struct directive): ...to this. Add and remove data members.
23423 (directive::set_width, directive::set_precision): New functions.
23424 (format_percent): Use fmtresult ctor.
23425 (get_width_and_precision): Remove.
23426 (format_integer): Make naming changes. Avoid computing width and
23428 (format_floating): Same. Adjust indentation.
23429 (format_character, format_none): New functions.
23430 (format_string): Moved character handling to format_character.
23431 (format_directive): Remove arguments, change return type.
23432 (parse_directive): New function.
23433 (pass_sprintf_length::compute_format_length): Move directive
23434 parsing to parse_directive.
23436 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23438 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23439 (assign_assembler_name_if_needed): ... this.
23440 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23441 (assign_assembler_name_if_needed): ... this.
23442 (free_lang_data_in_cgraph): Adjust callers.
23443 * cgraphunit.c (cgraph_node::analyze): Likewise.
23444 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23447 2017-01-23 Richard Biener <rguenther@suse.de>
23449 PR tree-optimization/79088
23450 PR tree-optimization/79188
23451 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23452 resetting loop bounds after last path deletion. Reset loop
23453 bounds of the target loop, make code match the comments.
23454 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23455 Make sure loops need no fixups.
23457 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23459 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23460 exponent support with double type for first argument.
23461 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23462 type returned by __builtin_vec_extract_sig,
23463 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23464 functions from "vector int" to "vector unsigned int" or from
23465 "vector long long int" to "vector unsigned long long int".
23466 Changed type returned by __builtin_vec_extract_exp,
23467 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23468 functions from "vector int" to "vector unsigned int" or from
23469 "vector long long int" to "vector unsigned long long int".
23470 Changed return type of __builtin_vec_test_data_class,
23471 __builtin_vec_test_data_class_sp, and
23472 __builtin_vec_test_data_class_dp from "vector int" to
23473 "vector bool int" or from "vector long long int" to "vector bool
23474 long long int" and changed second argument type from "unsigned
23475 int" to "int". Added new overloaded function forms "vector float
23476 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23477 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23478 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23479 double, vector unsigned long long int)" and "vector double
23480 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23481 long int)". Changed return type of
23482 __builtin_scalar_test_data_class and
23483 __builtin_scalar_test_data_class_sp and
23484 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23485 int" and changed second argument from "unsigned int" to "int".
23486 Changed type returned by __builtin_scalar_test_neg,
23487 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23488 from "int" to "bool int". Added new overloaded function form
23489 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23490 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23491 exponent double-precision with floating point first argument.
23492 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23493 documentation of scalar_test_data_class, scalar_test_neg,
23494 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23495 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23496 vec_test_data_class built-in functions to reflect refinements in
23497 their type signatures.
23499 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23501 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23503 (aarch64_elf_asm_destructor): Likewise.
23505 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23507 PR rtl-optimization/78634
23508 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23509 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23510 * ifcvt.c (noce_try_cmove): Add missing cost check.
23512 PR rtl-optimization/71724
23513 * combine.c (if_then_else_cond): Look for situations where it is
23514 beneficial to undo the work of one of the recursive calls.
23516 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23518 PR tree-optimization/70754
23519 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23520 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23521 combined stmt before it if not NULL.
23522 (combine_chains): Process refs reversely and compute dominance point
23525 2017-01-23 Martin Liska <mliska@suse.cz>
23527 PR tree-optimization/79196
23528 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23529 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23531 (strlen_optimize_stmt): Call the renamed function.
23533 2017-01-23 Michael Matz <matz@suse.de>
23535 PR tree-optimization/78384
23536 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23538 2017-01-23 Richard Biener <rguenther@suse.de>
23540 PR tree-optimization/79186
23541 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23542 both incoming edges before moving an assert.
23544 2017-01-23 Martin Jambor <mjambor@suse.cz>
23546 * ipa-prop.c (load_from_param_1): Removed.
23547 (load_from_unmodified_param): Bits from load_from_param_1 put back
23549 (load_from_param): Removed.
23550 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23551 with stmt. Reverted back to use of load_from_unmodified_param.
23553 2017-01-23 Martin Jambor <mjambor@suse.cz>
23556 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23557 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23558 field a pointer to garbage collected vector, mark lattices and
23559 ipcp_orig_node with GTY((skip)).
23560 (ipa_get_param_count): Adjust to descriptors being a pointer.
23561 (ipa_get_param): Likewise.
23562 (ipa_get_type): Likewise.
23563 (ipa_get_param_move_cost): Likewise.
23564 (ipa_set_param_used): Likewise.
23565 (ipa_get_controlled_uses): Likewise.
23566 (ipa_set_controlled_uses): Likewise.
23567 (ipa_is_param_used): Likewise.
23568 (ipa_node_params_t): Move into garbage collector. New methods insert
23570 (ipa_node_params_sum): Annotate wth GTY(()).
23571 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23573 (ipa_load_from_parm_agg): Adjust declaration.
23574 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23575 * ipa-profile.c (ipa_profile): Likewise.
23576 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23577 (ipa_populate_param_decls): Make descriptors parameter garbage
23579 (ipa_dump_param): Adjust to descriptors being a pointer.
23580 (ipa_alloc_node_params): Likewise.
23581 (ipa_initialize_node_params): Likewise.
23582 (load_from_param_1): Make descriptors parameter garbage collected.
23583 (load_from_unmodified_param): Likewise.
23584 (load_from_param): Likewise.
23585 (ipa_load_from_parm_agg): Likewise.
23586 (ipa_node_params::~ipa_node_params): Removed.
23587 (ipa_free_all_node_params): Remove call to delete operator.
23588 (ipa_node_params_t::insert): New.
23589 (ipa_node_params_t::remove): Likewise.
23590 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23591 copy known_csts and known_contexts vectors.
23592 (ipa_read_node_info): Adjust to descriptors being a pointer.
23593 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23595 (ipcp_transform_function): Make descriptors variable garbage
23598 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23600 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23601 * config/i386/avx512dqintrin.h: Ditto.
23602 * config/i386/avx512fintrin.h: Ditto.
23603 * config/i386/i386.c: Handle new builtins.
23604 * config/i386/i386-builtin.def: Add new builtins.
23605 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23606 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23608 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23609 Martin Liska <mliska@suse.cz>
23611 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23612 * asan.c (asan_expand_poison_ifn): Support stores and use
23613 appropriate ASAN report function.
23614 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23615 * internal-fn.def (ASAN_POISON_USE): Declare.
23616 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23617 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23618 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23619 ASAN_POISON calls w/o LHS.
23620 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23621 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23622 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23623 * gimplify.c (asan_poison_variables): Add attribute
23624 use_after_scope_memory to variables that really needs to live
23626 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23627 having the attribute.
23629 2017-01-23 Martin Liska <mliska@suse.cz>
23631 * asan.c (create_asan_shadow_var): New function.
23632 (asan_expand_poison_ifn): Likewise.
23633 * asan.h (asan_expand_poison_ifn): New declaration.
23634 * internal-fn.c (expand_ASAN_POISON): Likewise.
23635 * internal-fn.def (ASAN_POISON): New builtin.
23636 * sanopt.c (pass_sanopt::execute): Expand
23637 asan_expand_poison_ifn.
23638 * tree-inline.c (copy_decl_for_dup_finish): Make function
23640 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23641 * tree-ssa.c (is_asan_mark_p): New function.
23642 (execute_update_addresses_taken): Rewrite local variables
23643 (identified just by use-after-scope as addressable) into SSA.
23645 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23647 * doc/install.texi (Specific): opensource.apple.com uses https
23648 now. Remove trailing slash.
23650 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23652 * README.Portability: Remove note on an Irix compatibility issue.
23654 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23656 * gcov.c (INCLUDE_ALGORITHM): Define.
23657 (INCLUDE_VECTOR): Define.
23658 No longer include <vector> and <algorithm> directly.
23660 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23662 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23664 * doc/invoke.texi (Code Gen Options): Ditto.
23666 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23669 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23671 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23673 rtl-optimization/79125
23674 * cprop.c (local_cprop_pass): Handle cases where we make an
23675 unconditional trap.
23677 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23681 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23682 read from, for big endian.
23684 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23686 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23687 register pauth builtins for LP64 only.
23689 2017-01-20 Marek Polacek <polacek@redhat.com>
23692 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23695 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23697 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23699 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23700 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23701 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23703 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23706 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23707 in big-endian mode when they are not a single duplicated value.
23709 2017-01-20 Richard Biener <rguenther@suse.de>
23711 * BASE-VER: Bump to 7.0.1.
23713 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23715 * omp-low.c (omplow_simd_context): New struct. Use it...
23716 (lower_rec_simd_input_clauses): ...here and...
23717 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23718 references to idx, lane, max_vf, is_simt.
23720 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23722 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23725 2017-01-20 Martin Jambor <mjambor@suse.cz>
23727 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23728 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23730 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23731 (GTFILES): Rename hsa.c to hsa-common.c.
23732 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23733 * hsa-dump.c: Likewise.
23734 * hsa-gen.c: Likewise.
23735 * hsa-regalloc.c: Likewise.
23736 * ipa-hsa.c: Likewise.
23737 * omp-expand.c: Likewise.
23738 * omp-low.c: Likewise.
23739 * toplev.c: Likewise.
23741 2017-01-20 Marek Polacek <polacek@redhat.com>
23744 * doc/invoke.texi: Document -Wduplicated-branches.
23745 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23746 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23747 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23748 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23749 return 0 only when not OEP_LEXICOGRAPHIC.
23750 (fold_build_cleanup_point_expr): Use the expression
23751 location when building CLEANUP_POINT_EXPR.
23752 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23753 * tree.c (add_expr): Handle error_mark_node.
23755 2017-01-20 Martin Liska <mliska@suse.cz>
23758 * tree-profile.c (init_ic_make_global_vars): Do not call
23760 (gimple_init_gcov_profiler): Likewise.
23762 2017-01-20 Martin Liska <mliska@suse.cz>
23765 * cgraph.h (maybe_create_reference): Remove argument and
23767 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23769 * ipa-cp.c (create_specialized_node): Likewise.
23770 * symtab.c (symtab_node::maybe_create_reference): Handle
23771 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23773 2017-01-20 Martin Liska <mliska@suse.cz>
23775 * read-rtl-function.c (function_reader::create_function): Use
23776 build_decl instread of build_decl_stat.
23778 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23780 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23781 * config/i386/avx512dqintrin.h: Ditto.
23782 * config/i386/avx512fintrin.h: Ditto.
23783 * config/i386/i386-builtin-types.def: Add new types.
23784 * config/i386/i386.c: Handle new types.
23785 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23786 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23787 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23788 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23789 (__builtin_ia32_kshiftridi): New.
23790 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23792 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23796 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23797 define to rs6000_init_stack_protect_guard.
23798 (rs6000_init_stack_protect_guard): New function.
23800 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23801 Yunqiang Su <yunqiang.su@imgtec.com>
23803 * config.gcc (supported_defaults): Add madd4.
23804 (with_madd4): Add validation.
23805 (all_defaults): Add madd4.
23806 * config/mips/mips.opt (mmadd4): New option.
23807 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23809 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23811 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23812 (ISA_HAS_FUSED_MADD4): Likewise.
23813 * doc/invoke.texi (-mmadd4): Document the new option.
23814 * doc/install.texi (--with-madd4): Document the new option.
23816 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23818 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23819 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23820 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23821 (aarch64_init_pauth_hint_builtins): New.
23822 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23823 (aarch64_expand_builtin): Expand new builtins.
23825 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23827 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23828 * combine-stack-adj.c (no_unhandled_cfa): Handle
23829 REG_CFA_TOGGLE_RA_MANGLE.
23830 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23831 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23832 info for return address signing.
23833 (aarch64_expand_epilogue): Likewise.
23835 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23837 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23838 * config/aarch64/aarch64-protos.h
23839 (aarch64_return_address_signing_enabled): New declaration.
23840 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23842 (aarch64_expand_prologue): Sign return address before it's pushed onto
23844 (aarch64_expand_epilogue): Authenticate return address fetched from
23846 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23847 (aarch64_attributes): New function attributes for "sign-return-address".
23848 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23849 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23850 ("*do_return"): Generate combined instructions according to key index.
23851 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23852 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23854 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23855 * config/aarch64/aarch64.opt (msign-return-address=): New.
23856 * doc/extend.texi (AArch64 Function Attributes): Documents
23857 "sign-return-address=".
23858 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23860 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23862 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23863 overall option summary.
23865 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23867 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23868 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23869 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23870 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23872 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23874 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23875 -mpower9-minmax by default for -mcpu=power9.
23876 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23877 128-bit floating point.
23879 2017-01-20 Alan Modra <amodra@gmail.com>
23881 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23882 optimizing for size.
23884 2017-01-20 Alan Modra <amodra@gmail.com>
23887 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23888 for strcmp and strncmp from corresponding builtin decl.
23890 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23892 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23893 instead of i386/rtems-64.h.
23894 * config/i386/rtems-64.h: Remove.
23896 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23900 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23902 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23904 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23906 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23907 Change int to HOST_WIDE_INT.
23908 * config/aarch64/aarch64-protos.h
23909 (aarch64_simd_gen_const_vector_dup): Likewise.
23910 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23912 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23914 * langhooks-def.h (lhd_type_for_size): New decl.
23915 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23916 * langhooks.c (lhd_type_for_size): New function, taken from
23919 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23921 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23922 define_bypass for CR latency.
23923 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23924 (power9-alu2): Add define_bypass for CR latency.
23926 (power9-mul): Update insn latency.
23927 (power9-mul-compare): Update insn latency, bypass latency and remove
23930 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23932 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23934 * config/aarch64/aarch64.md
23935 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23936 aarch64_nopcrelative_literal_loads.
23937 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23939 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23941 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23942 TARGET_LOONGSON_3A.
23943 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23945 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23948 * config.gcc (supported_defaults): Add lxc1-sxc1.
23949 (with_lxc1_sxc1): Add validation.
23950 (all_defaults): Add lxc1-sxc1.
23951 * config/mips/mips.opt (mlxc1-sxc1): New option.
23952 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23954 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23955 __mips_no_lxc1_sxc1.
23956 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23957 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23958 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23960 2017-01-19 Richard Biener <rguenther@suse.de>
23962 PR tree-optimization/72488
23963 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23964 sure to restore SSA info.
23965 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23967 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23969 PR rtl-optimization/79121
23970 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23971 of the inner type when shifting an extended value.
23973 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23976 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23977 interposable aliases.
23979 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23982 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23983 Use the evmergelohi instruction.
23984 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23985 (mov_sitf_e500_subreg8_2_be): Likewise.
23986 (mov_sitf_e500_subreg12_2_le): Likewise.
23987 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23988 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23989 (mov_sitf_e500_subreg8_2_le): Likewise.
23990 (mov_sitf_e500_subreg12_2_be): Likewise.
23992 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23994 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23995 attribute from vecsimple to vecperm.
23996 (altivec_vbpermq2): Likewise.
23998 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24001 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24003 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24004 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24005 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24006 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24007 case where N arg is SIZE_MAX.
24008 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24009 (cmpstrsi): Add pattern.
24011 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24013 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24014 __builtin_vec_revb builtins.
24015 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24016 built-in functions to support generation of the ISA 3.0 XXBR<x>
24017 vector byte reverse instructions.
24018 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24019 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24020 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24021 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24022 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24023 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24024 (P9V_BUILTIN_VEC_REVB): Likewise.
24025 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24026 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24027 (p9_xxbrq_v16qi): Likewise.
24028 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24029 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24030 (p9_xxbrh_v8hi): Likewise.
24031 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24032 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24033 vec_revb built-in functions.
24035 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24037 PR rtl-optimization/78952
24038 * config/i386/i386.md (any_extract): New code iterator.
24039 (*insvqi_2): Use any_extract for source operand.
24040 (*insvqi_3): Use any_shiftrt for source operand.
24042 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24044 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24046 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24048 2017-01-18 Matthias Klose <doko@ubuntu.com>
24050 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24052 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24054 * config/rs6000/altivec.h (vec_bperm): Change #define.
24055 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24056 (altivec_vbpermq2): New define_insn.
24057 (altivec_vbpermd): Likewise.
24058 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24059 function interface.
24060 (VBPERMD): Likewise.
24061 (VBPERM): New polymorphic function interface.
24062 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24063 Add entries for P9V_BUILTIN_VEC_VBPERM.
24064 * doc/extend.texi: Add interfaces for vec_bperm.
24066 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24068 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24069 first letter of error messages.
24070 (s390_resolve_overloaded_builtin): Likewise.
24071 * config/s390/s390.c (s390_expand_builtin): Likewise.
24072 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24073 (s390_valid_target_attribute_inner_p): Likewise.
24074 * config/s390/s390.md ("tabort"): Likewise.
24076 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24078 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24079 (ISA_AVOID_DIV_HILO): New macro.
24080 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24081 (ISA_HAS_DDIV): Likewise.
24083 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24085 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24087 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24089 * doc/invoke.texi (fabi-version): Spelling fix.
24091 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24094 * doc/invoke.texi (fabi-version): Mention mangling fix for
24097 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24100 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24102 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24105 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24106 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24108 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24110 (stack_protect_test): Ditto.
24111 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24112 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24114 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24115 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24116 -mstack-protector-guard-offset=.
24117 (RS/6000 and PowerPC Options): Ditto.
24119 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24121 * config/i386/i386.h (MASK_CLASS_P): New define.
24122 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24123 there are no registers from different register sets also when
24124 mask registers are used. Update function comment.
24125 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24126 to (*k/*r) and (*k/*km) alternatives.
24128 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24130 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24131 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24132 (EH_RETURN_HANDLER_RTX): New define.
24133 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24134 Force frame pointer in EH return functions.
24135 (aarch64_expand_epilogue): Add barrier for eh_return.
24136 (aarch64_final_eh_return_addr): Remove.
24137 (aarch64_eh_return_handler_rtx): New function.
24138 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24140 (aarch64_eh_return_handler_rtx): New prototype.
24142 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24144 * config/rs6000/altivec.h (vec_rlmi): New #define.
24145 (vec_vrlnm): Likewise.
24146 (vec_rlnm): Likewise.
24147 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24148 (UNSPEC_VRLNM): Likewise.
24149 (VIlong): New mode iterator.
24150 (altivec_vrl<VI_char>mi): New define_insn.
24151 (altivec_vrl<VI_char>nm): Likewise.
24152 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24154 (VRLDNM): Likewise.
24155 (RLNM): New polymorphic function entry.
24156 (VRLWMI): New monomorphic function entry.
24157 (VRLDMI): Likewise.
24158 (RLMI): New polymorphic function entry.
24159 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24160 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24161 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24164 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24167 * dwarf2out.c (field_byte_offset): Restore the
24168 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24169 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24170 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24173 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24176 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24178 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24181 * dwarf2out.c (add_data_member_location_attribute): For constant
24182 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24183 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24184 DW_AT_byte_size attributes.
24186 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24188 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24189 after forcing to constant memory when the code model is medium.
24191 2017-01-17 Julia Koval <julia.koval@intel.com>
24194 * config/i386/avx512fintrin.h
24195 (_mm512_i32gather_ps): Change __addr type to void const*.
24196 (_mm512_mask_i32gather_ps): Ditto.
24197 (_mm512_i32gather_pd): Ditto.
24198 (_mm512_mask_i32gather_pd): Ditto.
24199 (_mm512_i64gather_ps): Ditto.
24200 (_mm512_mask_i64gather_ps): Ditto.
24201 (_mm512_i64gather_pd): Ditto.
24202 (_mm512_mask_i64gather_pd): Ditto.
24203 (_mm512_i32gather_epi32): Ditto.
24204 (_mm512_mask_i32gather_epi32): Ditto.
24205 (_mm512_i32gather_epi64): Ditto.
24206 (_mm512_mask_i32gather_epi64): Ditto.
24207 (_mm512_i64gather_epi32): Ditto.
24208 (_mm512_mask_i64gather_epi32): Ditto.
24209 (_mm512_i64gather_epi64): Ditto.
24210 (_mm512_mask_i64gather_epi64): Ditto.
24211 (_mm512_i32scatter_ps): Change __addr type to void*.
24212 (_mm512_mask_i32scatter_ps): Ditto.
24213 (_mm512_i32scatter_pd): Ditto.
24214 (_mm512_mask_i32scatter_pd): Ditto.
24215 (_mm512_i64scatter_ps): Ditto.
24216 (_mm512_mask_i64scatter_ps): Ditto.
24217 (_mm512_i64scatter_pd): Ditto.
24218 (_mm512_mask_i64scatter_pd): Ditto.
24219 (_mm512_i32scatter_epi32): Ditto.
24220 (_mm512_mask_i32scatter_epi32): Ditto.
24221 (_mm512_i32scatter_epi64): Ditto.
24222 (_mm512_mask_i32scatter_epi64): Ditto.
24223 (_mm512_i64scatter_epi32): Ditto.
24224 (_mm512_mask_i64scatter_epi32): Ditto.
24225 (_mm512_i64scatter_epi64): Ditto.
24226 (_mm512_mask_i64scatter_epi64): Ditto.
24227 * config/i386/avx512pfintrin.h
24228 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24229 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24230 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24231 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24232 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24233 (_mm512_prefetch_i32scatter_ps): Ditto.
24234 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24235 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24236 (_mm512_prefetch_i64scatter_pd): Ditto.
24237 (_mm512_prefetch_i64scatter_ps): Ditto.
24238 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24239 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24240 * config/i386/avx512vlintrin.h
24241 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24242 (_mm_mmask_i32gather_ps): Ditto.
24243 (_mm256_mmask_i32gather_pd): Ditto.
24244 (_mm_mmask_i32gather_pd): Ditto.
24245 (_mm256_mmask_i64gather_ps): Ditto.
24246 (_mm_mmask_i64gather_ps): Ditto.
24247 (_mm256_mmask_i64gather_pd): Ditto.
24248 (_mm_mmask_i64gather_pd): Ditto.
24249 (_mm256_mmask_i32gather_epi32): Ditto.
24250 (_mm_mmask_i32gather_epi32): Ditto.
24251 (_mm256_mmask_i32gather_epi64): Ditto.
24252 (_mm_mmask_i32gather_epi64): Ditto.
24253 (_mm256_mmask_i64gather_epi32): Ditto.
24254 (_mm_mmask_i64gather_epi32): Ditto.
24255 (_mm256_mmask_i64gather_epi64): Ditto.
24256 (_mm_mmask_i64gather_epi64): Ditto.
24257 (_mm256_i32scatter_ps): Change __addr type to void*.
24258 (_mm256_mask_i32scatter_ps): Ditto.
24259 (_mm_i32scatter_ps): Ditto.
24260 (_mm_mask_i32scatter_ps): Ditto.
24261 (_mm256_i32scatter_pd): Ditto.
24262 (_mm256_mask_i32scatter_pd): Ditto.
24263 (_mm_i32scatter_pd): Ditto.
24264 (_mm_mask_i32scatter_pd): Ditto.
24265 (_mm256_i64scatter_ps): Ditto.
24266 (_mm256_mask_i64scatter_ps): Ditto.
24267 (_mm_i64scatter_ps): Ditto.
24268 (_mm_mask_i64scatter_ps): Ditto.
24269 (_mm256_i64scatter_pd): Ditto.
24270 (_mm256_mask_i64scatter_pd): Ditto.
24271 (_mm_i64scatter_pd): Ditto.
24272 (_mm_mask_i64scatter_pd): Ditto.
24273 (_mm256_i32scatter_epi32): Ditto.
24274 (_mm256_mask_i32scatter_epi32): Ditto.
24275 (_mm_i32scatter_epi32): Ditto.
24276 (_mm_mask_i32scatter_epi32): Ditto.
24277 (_mm256_i32scatter_epi64): Ditto.
24278 (_mm256_mask_i32scatter_epi64): Ditto.
24279 (_mm_i32scatter_epi64): Ditto.
24280 (_mm_mask_i32scatter_epi64): Ditto.
24281 (_mm256_i64scatter_epi32): Ditto.
24282 (_mm256_mask_i64scatter_epi32): Ditto.
24283 (_mm_i64scatter_epi32): Ditto.
24284 (_mm_mask_i64scatter_epi32): Ditto.
24285 (_mm256_i64scatter_epi64): Ditto.
24286 (_mm256_mask_i64scatter_epi64): Ditto.
24287 (_mm_i64scatter_epi64): Ditto.
24288 (_mm_mask_i64scatter_epi64): Ditto.
24289 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24290 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24291 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24292 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24293 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24294 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24295 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24296 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24297 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24298 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24299 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24300 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24301 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24302 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24303 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24304 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24305 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24306 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24307 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24308 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24309 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24310 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24311 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24312 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24313 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24314 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24315 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24316 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24317 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24318 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24319 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24320 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24321 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24322 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24323 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24324 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24325 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24326 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24327 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24328 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24329 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24330 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24331 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24332 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24333 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24334 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24335 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24336 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24337 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24338 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24339 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24340 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24341 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24342 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24343 definitions accordingly.
24345 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24346 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24349 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24352 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24355 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24356 hard regs for inner regno.
24358 2017-01-17 Martin Liska <mliska@suse.cz>
24361 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24362 assumption and add comment.
24364 2017-01-17 Nathan Sidwell <nathan@acm.org>
24366 * ipa-visibility.c (localize_node): New function, broken out of ...
24367 (function_and_variable_visibility): ... here. Call it.
24369 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24371 PR middle-end/77445
24372 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24373 correctly set frequency of oudgoing edge.
24374 (duplicate_thread_path): Fix profile updating.
24376 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24379 * configure.ac: Add GCC_BASE_VER.
24380 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24381 version from BASE-VER file.
24382 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24383 (gcc.o): Depend on $(BASEVER).
24384 * common.opt (dumpfullversion): New option.
24385 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24386 * doc/invoke.texi: Document -dumpfullversion.
24387 * doc/install.texi: Document --with-gcc-major-version-only.
24388 * configure: Regenerated.
24390 2017-01-17 Richard Biener <rguenther@suse.de>
24392 PR tree-optimization/71433
24393 * tree-vrp.c (register_new_assert_for): Merge same asserts
24394 on all incoming edges.
24395 (process_assert_insertions_for): Handle insertions at the
24398 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24400 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24401 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24403 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24406 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24409 2017-01-17 Alan Modra <amodra@gmail.com>
24412 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24413 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24414 symbolic stack limit when pic.
24416 2017-01-16 Martin Sebor <msebor@redhat.com>
24418 PR tree-optimization/78608
24419 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24421 2017-01-16 Jeff Law <law@redhat.com>
24424 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24425 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24426 for several include directories that may be relative to sysroot.
24427 * config/i386/x-mingw32 (gplus_includedir): Define.
24428 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24429 (native_system_includedir): Likewise.
24430 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24431 override if TARGET_SYSTEM_ROOT is defined.
24432 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24434 PR tree-optimization/79090
24435 PR tree-optimization/33562
24436 PR tree-optimization/61912
24437 PR tree-optimization/77485
24438 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24439 and computed trims into the dump file.
24441 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24443 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24445 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24448 * gimplify.c (gimplify_init_constructor): If want_value and
24449 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24453 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24454 sequence. Formatting fixes.
24455 (doloop_optimize): Formatting fixes.
24458 * gcc.c (debug_level_greater_than_spec_func): New function.
24459 (static_spec_functions): Add debug-level-gt spec function.
24460 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24462 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24463 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24464 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24465 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24466 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24467 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24469 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24471 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24472 QImode fixups to general and mask registers only.
24474 2017-01-16 Carl Love <cel@us.ibm.com>
24476 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24477 for built-in functions
24478 vector signed char vec_nabs (vector signed char)
24479 vector signed short vec_nabs (vector signed short)
24480 vector signed int vec_nabs (vector signed int)
24481 vector signed long long vec_nabs (vector signed long long)
24482 vector float vec_nabs (vector float)
24483 vector double vec_nabs (vector double)
24484 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24486 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24487 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24488 * doc/extend.texi: Update the documentation file for the new built-in
24491 2017-01-16 Martin Sebor <msebor@redhat.com>
24493 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24496 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24498 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24499 UNSPEC_VSX__XXSPLTD to require special splat handling.
24501 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24504 * system.h: Poison strndup.
24506 2017-01-16 Alan Modra <amodra@gmail.com>
24509 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24512 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24514 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24516 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24518 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24519 call recog here. Assert that INSN_CODE (insn) is non-negative.
24521 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24524 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24526 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24527 in the currently scheduled RTL fragment.
24529 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24531 PR rtl-optimization/78751
24532 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24535 2017-01-14 Jeff Law <law@redhat.com>
24537 PR tree-optimization/79090
24538 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24539 variable length stores.
24540 (compute_trims): Delete dead assignment to *trim_tail.
24541 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24544 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24546 PR rtl-optimization/78626
24547 PR rtl-optimization/78727
24548 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24549 of a block, and split such blocks after everything else is finished.
24551 2017-01-14 Alan Modra <amodra@gmail.com>
24554 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24555 target legitimate_combined_insn.
24556 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24557 (rs6000_legitimate_combined_insn): New function.
24558 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24560 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24561 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24562 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24564 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24566 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24568 2017-01-13 Jeff Law <law@redhat.com>
24570 PR tree-optimization/33562
24571 PR tree-optimization/61912
24572 PR tree-optimization/77485
24573 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24575 (delete_dead_assignment): Likewise.
24576 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24577 statement to delete_dead_call and delete_dead_assignment.
24579 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24582 * substring-locations.c (format_warning_va): Strengthen case 1 so
24583 that both endpoints of the substring must be within the format
24584 range for just the substring to be printed.
24586 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24588 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24589 * config/i386/i386.c (ix86_target_string): Add missing options
24590 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24591 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24592 flags_other and ix86_target_other to flags2_other. Display unknown
24594 (ix86_valid_target_attribute_inner_p): Add missing options and
24595 reorder options by implied ISAs, as in ix86_target_string.
24597 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24599 * hash-table.h (hash_table::too_empty_p): New function.
24600 (hash_table::expand): Use it.
24601 (hash_table::traverse): Likewise.
24602 (hash_table::empty_slot): Use sizeof (value_type) instead of
24603 sizeof (PTR) to convert bytes to elements. Shrink the table
24604 if the current size is excessive for the current number of
24607 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24609 * ira-costs.c (record_reg_classes): Break from the inner loop
24610 early once alt_fail is known to be true. Update outer loop
24611 handling accordingly.
24613 2017-01-13 Jeff Law <law@redhat.com>
24615 * tree-ssa-dse.c (decrement_count): New function.
24616 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24617 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24618 when we know the partially dead statement is a mem* function.
24620 PR tree-optimization/61912
24621 PR tree-optimization/77485
24622 * tree-ssa-dse.c: Include expr.h.
24623 (maybe_trim_constructor_store): New function.
24624 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24626 PR tree-optimization/33562
24627 PR tree-optimization/61912
24628 PR tree-optimization/77485
24629 * doc/invoke.texi: Document new dse-max-object-size param.
24630 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24631 * tree-ssa-dse.c: Include params.h.
24632 (dse_store_status): New enum.
24633 (initialize_ao_ref_for_dse): New, partially extracted from
24635 (valid_ao_ref_for_dse, normalize_ref): New.
24636 (setup_live_bytes_from_ref, compute_trims): Likewise.
24637 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24638 (maybe_trim_partially_dead_store): Likewise.
24639 (maybe_trim_complex_store): Likewise.
24640 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24641 Track what bytes live from the original store. Return tri-state
24642 for dead, partially dead or live.
24643 (dse_dom_walker): Add constructor, destructor and new private members.
24644 (delete_dead_call, delete_dead_assignment): New extracted from
24646 (dse_optimize_stmt): Make a member of dse_dom_walker.
24647 Use initialize_ao_ref_for_dse.
24649 PR tree-optimization/33562
24650 PR tree-optimization/61912
24651 PR tree-optimization/77485
24652 * sbitmap.h (bitmap_count_bits): Prototype.
24653 (bitmap_clear_range, bitmap_set_range): Likewise.
24654 * sbitmap.c (bitmap_clear_range): New function.
24655 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24657 2017-01-13 Martin Liska <mliska@suse.cz>
24660 * function.c (set_cfun): Add new argument force.
24661 * function.h (set_cfun): Likewise.
24662 * ipa-inline-transform.c (inline_call): Use the function when
24663 strict alising from is dropped for function we inline to.
24665 2017-01-13 Richard Biener <rguenther@suse.de>
24667 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24668 for dumping GIMPLE INTEGER_CSTs.
24670 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24672 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24673 to 201112L since C++17.
24675 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24678 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24679 if -fsanitize=kernel-address is present.
24681 2017-01-13 Richard Biener <rguenther@suse.de>
24683 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24684 as _Literal ( type ) number in case usual suffixes do not
24685 preserve all information.
24687 2017-01-13 Richard Biener <rguenther@suse.de>
24689 PR tree-optimization/77283
24690 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24691 and ssa-iterators.h.
24692 (is_feasible_trace): Implement a cost model based on joiner
24695 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24698 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24699 char or short to __float128/_Float128 directly.
24701 2017-01-12 Martin Sebor <msebor@redhat.com>
24703 to -Wformat-overflow.
24704 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24705 (min_bytes_remaining): Same.
24706 (get_string_length): Same.
24707 (format_string): Same.
24708 (format_directive): Same.
24710 (pass_sprintf_length::handle_gimple_call): Same.
24712 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24714 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24715 info.nowrite calls with no lhs that can't throw. Return bool
24716 whether gsi_remove has been called or not.
24717 (pass_sprintf_length::handle_gimple_call): Return bool whether
24718 try_substitute_return_value called gsi_remove. Formatting fix.
24719 (pass_sprintf_length::execute): Don't use gsi_remove if
24720 handle_gimple_call returned true.
24723 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24724 be removed due to side-effects, don't remove following barrier nor
24725 turn the successor edge into fallthru edge.
24727 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24730 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24731 element-reversing loads and stores as not swappable.
24733 2017-01-12 Nathan Sidwell <nathan@acm.org>
24734 Nicolai Stange <nicstange@gmail.com>
24736 * combine.c (try_combine): Don't ignore result of overlap checking
24737 loop. Combine overlap & asm check into single loop.
24739 2017-01-12 Richard Biener <rguenther@suse.de>
24741 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24742 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24744 2017-01-12 Richard Biener <rguenther@suse.de>
24746 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24747 and TS_TARGET_OPTION directly derive from TS_BASE.
24748 * tree-core.h (tree_optimization_option): Derive from tree_base.
24749 (tree_target_option): Likewise.
24751 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24753 * config/i386/i386.c (memory_address_length): Increase len
24754 only when rip_relative_addr_p returns false.
24756 2017-01-11 Julia Koval <julia.koval@intel.com>
24758 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24759 (OPTION_MASK_ISA_SGX_SET): New.
24760 (ix86_handle_option): Handle OPT_msgx.
24761 * config.gcc: Added sgxintrin.h.
24762 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24763 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24764 * config/i386/i386.c (ix86_target_string): Add -msgx.
24766 (ix86_option_override_internal): Handle new options.
24767 (ix86_valid_target_attribute_inner_p): Add sgx.
24768 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24769 * config/i386/i386.opt: Add msgx.
24770 * config/i386/sgxintrin.h: New file.
24771 * config/i386/x86intrin.h: Add sgxintrin.h.
24773 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24776 * fold-const.c (maybe_nonzero_address): Return 1 for function
24778 (tree_single_nonzero_warnv_p): Don't handle function local objects
24782 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24785 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24788 * opts.c: Include "spellcheck.h"
24789 (struct string_fragment): New struct.
24790 (struct edit_distance_traits<const string_fragment &>): New
24792 (get_closest_sanitizer_option): New function.
24793 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24795 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24797 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24799 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24800 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24801 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24802 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24803 for initial die_offset if dwarf_split_debug_info.
24804 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24805 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24807 (output_skeleton_debug_sections): Formatting fix. Use
24808 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24809 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24811 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24813 * config/arm/cortex-a53.md: Add bypasses for
24814 cortex_a53_r2f_cvt.
24815 (cortex_a53_r2f): Only use for transfers.
24816 (cortex_a53_f2r): Likewise.
24817 (cortex_a53_r2f_cvt): Add reservation for conversions.
24818 (cortex_a53_f2r_cvt): Likewise.
24820 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24822 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24823 to all inlined functions, change static to extern.
24825 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24828 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24830 (arm_assemble_integer): Likewise.
24832 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24834 * config.gcc: Use new awk script to check CPU, FPU and architecture
24835 parameters for --with-... options.
24836 * config/arm/parsecpu.awk: New file
24837 * config/arm/arm-cpus.in: New file.
24838 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24840 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24842 * config/arm/t-arm: Update dependency rules.
24843 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24844 of processing .def files.
24845 * config/arm/genopt.sh: Deleted.
24846 * config/arm/gentune.sh: Deleted.
24847 * config/arm/arm-cores.def: Deleted.
24848 * config/arm/arm-arches.def: Deleted.
24849 * config/arm/arm-fpus.def: Deleted.
24850 * config/arm/arm-tune.md: Regenerated.
24851 * config/arm/arm-tables.opt: Regenerated.
24852 * config/arm/arm-cpu.h: New generated file.
24853 * config/arm/arm-cpu-data.h: New generated file.
24854 * config/arm/arm-cpu-cdata.h: New generated file.
24856 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24859 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24861 (input_varpool_node): Unpack dynamically_initialized bit.
24863 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24865 PR rtl-optimization/79032
24866 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24867 the alignment of the adjusted memory reference against that of MODE,
24868 instead of the alignment of the original memory reference.
24870 2017-01-11 Martin Jambor <mjambor@suse.cz>
24872 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24874 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24875 decorated functions.
24877 2017-01-11 Richard Biener <rguenther@suse.de>
24879 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24880 set range/nonnull info for PHI results. Do not set it on
24881 stmts marked for removal.
24883 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24885 * expr.c (store_field): In the bitfield case, fetch the return value
24886 from the registers before applying a single big-endian adjustment.
24887 Always do a final load for a BLKmode value not larger than a word.
24889 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24892 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24893 that we correctly handle column numbers greater than
24894 LINE_MAP_MAX_COLUMN_NUMBER.
24896 2017-01-10 Martin Sebor <msebor@redhat.com>
24898 PR middle-end/78245
24899 * gimple-ssa-sprintf.c (get_destination_size): Call
24900 {init,fini}object_sizes.
24901 * tree-object-size.c (addr_object_size): Adjust.
24902 (pass_through_call): Adjust.
24903 (pass_object_sizes::execute): Adjust.
24904 * tree-object-size.h (fini_object_sizes): Declare.
24906 2017-01-10 Martin Sebor <msebor@redhat.com>
24908 PR tree-optimization/78775
24909 * builtins.c (get_size_range): Move...
24910 * calls.c: ...to here.
24911 (alloc_max_size): Accept zero argument.
24912 (operand_signed_p): Remove.
24913 (maybe_warn_alloc_args_overflow): Call get_size_range.
24914 * calls.h (get_size_range): Declare.
24916 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24918 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24919 from TI's devices.csv file as of September 2016.
24920 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24922 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24924 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24925 * doc/invoke.texi: Likewise.
24926 * doc/md.texi: Likewise.
24927 * doc/objc.texi: Likewise.
24929 2017-01-10 Joshua Conner <joshconner@google.com>
24931 * config/arm/fuchsia-elf.h: New file.
24932 * config/fuchsia.h: New file.
24933 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24934 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24936 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24938 2016-01-10 Richard Biener <rguenther@suse.de>
24940 PR tree-optimization/79034
24941 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24942 Propagate out degenerate PHIs in the joiner.
24944 2017-01-10 Martin Liska <mliska@suse.cz>
24946 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24947 (sort_congruence_classes_by_decl_uid): Likewise.
24948 (sort_congruence_class_groups_by_decl_uid): Likewise.
24949 (sem_item_optimizer::merge_classes): Sort class, groups in these
24950 classes and members in the groups by DECL_UID of declarations.
24951 This would make merge operations stable.
24953 2017-01-10 Martin Liska <mliska@suse.cz>
24955 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24956 usage of m_classes_vec.
24957 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24958 (sem_item_optimizer::get_group_by_hash): Likewise.
24959 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24960 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24961 (sem_item_optimizer::verify_classes): Likewise.
24962 (sem_item_optimizer::process_cong_reduction): Likewise.
24963 (sem_item_optimizer::dump_cong_classes): Likewise.
24964 (sem_item_optimizer::merge_classes): Likewise.
24965 * ipa-icf.h (congruence_class_hash): Rename from
24966 congruence_class_group_hash. Remove declaration of m_classes_vec.
24968 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24970 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24971 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24972 * config.gcc: Add avx512vpopcntdqintrin.h.
24973 * config/i386/avx512vpopcntdqintrin.h: New.
24974 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24975 * config/i386/i386-builtin-types.def: Add new types.
24976 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24977 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24978 __builtin_ia32_vpopcountq_v8di_mask): New.
24979 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24980 __AVX512VPOPCNTDQ__.
24981 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24982 (PTA_AVX512VPOPCNTDQ): Define.
24983 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24984 TARGET_AVX512VPOPCNTDQ_P): Define.
24985 * config/i386/i386.opt: Add mavx512vpopcntdq.
24986 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24987 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24989 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24991 PR middle-end/77484
24992 * predict.def (PRED_CALL): Set to 67.
24994 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24996 * expr.c (store_field): In the bitfield case, if the value comes from
24997 a function call and is of an aggregate type returned in registers, do
24998 not modify the field mode; extract the value in all cases if the mode
24999 is BLKmode and the size is not larger than a word.
25001 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25004 * config/i386/cpuid.h: Fix undefined behavior.
25006 2017-01-04 Jeff Law <law@redhat.com>
25008 PR tree-optimization/79007
25009 PR tree-optimization/67955
25010 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25011 conservative for pt.null when flag_non_call_exceptions is on.
25013 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25015 PR translation/79019
25016 PR translation/79020
25017 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25018 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25019 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25021 * config/avr/avr.opt (maccumulate-args): Likewise.
25022 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25023 * common.opt (freport-bug): Likewise.
25024 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25025 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25026 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25027 translatable string.
25028 * config/i386/i386.c (function_value_32): Likewise.
25029 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25030 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25032 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25033 * common/config/msp430/msp430-common.c (msp430_handle_option):
25035 * symtab.c (symtab_node::verify_base): Likewise.
25036 * opts.c (set_debug_level): Likewise.
25037 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25038 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25039 missing whitespace to translatable strings.
25040 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25041 * config/sh/superh.h: Likewise.
25042 * config/i386/xopintrin.h: Likewise.
25043 * config/i386/znver1.md: Likewise.
25044 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25045 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25046 * double-int.h (struct double_int): Likewise.
25047 * double-int.c (div_and_round_double): Likewise.
25048 * wide-int.cc: Likewise.
25049 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25050 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25051 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25052 (crossjumps_occurred): ... this.
25053 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25056 PR tree-optimization/78899
25057 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25058 returning bool return struct loop *, NULL for failure and the new
25060 (versionable_outer_loop_p): Don't version outer loop if it has
25061 dont_vectorized bit set.
25062 (tree_if_conversion): When versioning outer loop, ensure
25063 tree_if_conversion is performed also on the inner loop of the
25064 non-vectorizable outer loop copy.
25065 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25066 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25067 prevent vectorization of it.
25068 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25069 the outer loop vectorization of the non-scalar version is attempted
25070 before vectorization of the inner loop in scalar version. If
25071 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25072 vectorization of its inner loop.
25073 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25074 has 2 inner loops, rename also on edges from bb whose single pred
25075 is outer_loop->header. Fix typo in function comment.
25077 2017-01-09 Martin Sebor <msebor@redhat.com>
25080 * asan.c (asan_emit_stack_protection): Increase local buffer size
25081 to avoid snprintf truncation warning.
25083 2017-01-09 Andrew Pinski <apinski@cavium.com>
25085 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25086 to reference thunderx2t99 for the tuning structure
25087 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25089 (thunderx2t99_extra_costs): This.
25090 * config/aarch64/aarch64-tune.md: Regenerate.
25091 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25092 (vulcan_addrcost_table): This.
25093 (vulcan_regmove_cost): Rename to ...
25094 (thunderx2t99_regmove_cost): This.
25095 (vulcan_vector_cost): Rename to ...
25096 (thunderx2t99_vector_cost): this.
25097 (vulcan_branch_cost): Rename to ...
25098 (thunderx2t99_branch_cost): This.
25099 (vulcan_tunings): Rename to ...
25100 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25101 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25103 2017-01-09 Martin Jambor <mjambor@suse.cz>
25107 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25108 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25109 (propagate_vr_accross_jump_function): Use the above function for all
25110 value range computations for pass-through jump functions and type
25111 converasion from explicit value range values.
25112 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25113 parameters from TYPE_ARG_TYPES.
25114 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25115 (ipa_write_node_info): Stream type of the actual argument.
25116 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25118 2017-01-09 Martin Liska <mliska@suse.cz>
25121 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25122 (lookup_compiler): Do not show error message with have_E.
25124 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25126 PR tree-optimization/78938
25127 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25128 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25129 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25130 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25133 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25135 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25138 2017-01-09 Richard Biener <rguenther@suse.de>
25140 PR tree-optimization/78997
25141 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25142 name condition properly.
25144 2017-01-09 Richard Biener <rguenther@suse.de>
25147 * dwarf2out.c (is_cxx): New overload with context.
25148 (is_naming_typedef_decl): Use it.
25150 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25152 * invoke.texi (Option Summary): Correct spacing in option lists
25153 and add line breaks to fix over-long lines.
25155 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25157 PR middle-end/17660
25159 * extend.texi (Common Variable Attributes): Add xref to GCC
25160 Internals manual to explain mode attribute keywords.
25162 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25165 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25166 and Preprocessor Options.
25167 (Options for Linking): Document -pthread here....
25168 (RS/6000 and PowerPC Options): ...not here.
25169 (Solaris 2 Options): ...or here.
25170 * doc/cppopts.texi: Document -pthread.
25172 2017-01-08 Martin Sebor <msebor@redhat.com>
25174 PR middle-end/77708
25175 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25176 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25177 New member functions.
25178 (format_directive): Used them.
25180 (pass_sprintf_length::handle_gimple_call): Same.
25181 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25182 to avoid truncation for any argument.
25183 (extract_affine_mul): Same.
25184 * tree.c (get_file_function_name): Same.
25186 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25188 PR middle-end/77484
25189 * predict.def (PRED_INDIR_CALL): Set to 86.
25191 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25193 PR preprocessor/54124
25194 * doc/cppopts.texi: Reformat -d subtable to list the full name
25195 of the options. Add cross-reference to the docs for the general
25196 compiler -d options.
25197 * doc/invoke.texi (Developer Options): Add cross-reference to the
25198 preprocessor-specific -d option documentation.
25200 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25202 PR preprocessor/13498
25203 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25204 redudant material, and reflect new command-line options.
25205 (System Headers): Likewise.
25207 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25209 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25210 -isystem, and -idirafter. Copy-edit.
25211 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25212 default for -ftrack-macro-expansion. Delete obsolete and
25213 badly-formatted implementation details about -fdebug-cpp output.
25214 * doc/cppwarnopts.texi: Copy-edit.
25216 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25219 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25220 that the transition from a max line width >= 1<<10 to narrower
25221 lines works correctly.
25223 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25225 * doc/options.texi (PerFunction): New.
25226 * opt-functions.awk (switch_flags): Map both Optimization and
25227 PerFunction to CL_OPTIMIZATION.
25228 * opth-gen.awk: Test for PerFunction flag along with
25230 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25231 it only when the latter is present. Skip those that don't in
25232 the hash function generator.
25233 * common.opt (fvar-tracking): Mark as PerFunction instead of
25235 (fvar-tracking-assignments): Likewise.
25236 (fvar-tracking-assignments-toggle): Likewise.
25237 (fvar-tracking-uninit): Likewise.
25239 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25241 PR translation/79018
25242 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25245 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25248 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25249 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25250 TARGET_LONG_JUMP_TABLE_OFFSETS.
25251 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25252 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25253 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25254 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25255 * config/m68k/m68k.md (tablejump expander): Likewise.
25256 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25257 TARGET_LONG_JUMP_TABLE_OFFSETS.
25258 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25259 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25261 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25262 David Holsgrove <david.holsgrove@xilinx.com>
25264 * common/config/microblaze/microblaze-common.c
25265 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25266 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25268 * config/microblaze/microblaze.c (microblaze_must_save_register)
25269 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25271 (microblaze_eh_return): New function.
25272 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25273 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25274 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25275 * config/microblaze/microblaze.md (eh_return): New pattern.
25277 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25279 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25280 GCC_DIAGNOSTIC_STRINGIFY): Define.
25282 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25284 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25286 * config/arm/arm.md (<mcrr>): New.
25288 * config/arm/arm.c (arm_arch5te): New.
25289 (arm_option_override): Set arm_arch5te.
25290 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25292 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25293 (arm_mcrr_qualifiers): ... this. New.
25294 (MRRC_QUALIFIERS): Define to...
25295 (arm_mrrc_qualifiers): ... this. New.
25296 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25298 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25299 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25300 (MRRCI, mrrc, MRRC): New.
25301 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25302 VUNSPEC_MRRC2): New.
25304 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25306 * config/arm/arm.md (<mcr>): New.
25308 * config/arm/arm.c (arm_coproc_builtin_available): Add
25309 support for mcr, mrc, mcr2 and mrc2.
25310 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25311 (arm_mcr_qualifiers): ... this. New.
25312 (MRC_QUALIFIERS): Define to ...
25313 (arm_mrc_qualifiers): ... this. New.
25314 (MCR_QUALIFIERS): Define to ...
25315 (arm_mcr_qualifiers): ... this. New.
25316 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25318 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25319 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25320 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25321 VUNSPEC_MRC2): New.
25323 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25325 * config/arm/arm.md (*ldc): New.
25329 * config/arm/arm.c (arm_coproc_builtin_available): Add
25330 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25331 (arm_coproc_ldc_stc_legitimate_address): New.
25332 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25333 'qualifier_const_pointer'.
25334 (LDC_QUALIFIERS): Define to...
25335 (arm_ldc_qualifiers): ... this. New.
25336 (STC_QUALIFIERS): Define to...
25337 (arm_stc_qualifiers): ... this. New.
25338 * config/arm/arm-protos.h
25339 (arm_coproc_ldc_stc_legitimate_address): New.
25340 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25341 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25342 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25343 stc2, stcl, stc2l): New.
25344 * config/arm/constraints.md (Uz): New.
25345 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25346 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25347 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25348 VUNSPEC_STC2L): New.
25350 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25352 * config/arm/arm.md (<cdp>): New.
25353 * config/arm/arm.c (neon_const_bounds): Rename this ...
25354 (arm_const_bounds): ... this.
25355 (arm_coproc_builtin_available): New.
25356 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25357 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25358 (CDP_QUALIFIERS): Define to...
25359 (arm_cdp_qualifiers): ... this. New.
25361 (arm_expand_builtin_args): Add case for 6 arguments.
25362 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25363 (arm_const_bounds): ... this.
25364 (arm_coproc_builtin_available): New.
25365 * config/arm/arm_acle.h (__arm_cdp): New.
25367 * config/arm/arm_acle_builtins.def (cdp): New.
25369 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25370 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25371 'arm_const_bounds'.
25372 * config/arm/types.md (coproc): New.
25373 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25374 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25375 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25376 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25378 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25380 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25381 (UBINOP_QUALIFIERS): New.
25383 (acle_builtin_data): New. Change comment.
25384 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25385 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25386 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25387 arm_acle_builtins.def.
25388 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25389 (arm_init_acle_builtins): New.
25390 (CRC32_BUILTIN): Remove.
25391 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25392 crc32cb, crc32ch and crc32cw.
25393 (arm_init_crc32_builtins): Remove.
25394 (arm_init_builtins): Use arm_init_acle_builtins rather
25395 than arm_init_crc32_builtins.
25396 (arm_expand_acle_builtin): New.
25397 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25398 * config/arm/arm_acle_builtins.def: New.
25400 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25402 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25403 (arm_builtin_datum): ... this.
25404 (arm_init_neon_builtin): Rename to ...
25405 (arm_init_builtin): ... this. Add a new parameters PREFIX
25406 and USE_SIG_IN_NAME.
25407 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25408 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25409 'arm_builtin_datum'.
25410 (arm_init_vfp_builtins): Likewise.
25411 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25412 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25413 (arm_expand_neon_args): Rename to ...
25414 (arm_expand_builtin_args): ... this. Rename builtin_arg
25415 enum values and differentiate between ARG_BUILTIN_MEMORY
25416 and ARG_BUILTIN_NEON_MEMORY.
25417 (arm_expand_neon_builtin_1): Rename to ...
25418 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25419 values, arm_expand_builtin_args and add bool parameter NEON.
25420 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25421 (arm_expand_vfp_builtin): Likewise.
25422 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25424 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25426 PR middle-end/77484
25427 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25428 * predict.c (tree_estimate_probability_bb): Reverse direction of
25429 polymorphic call predictor.
25431 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25433 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25434 (determine_pass_name_match): ...this new function and...
25435 (should_skip_pass_p): ...this new function.
25437 2017-01-06 Nathan Sidwell <nathan@acm.org>
25439 * ipa-visibility.c (function_and_variable_visibility): Reformat
25440 comments and long lines. Remove extrneous if.
25441 * symtab.c (symtab_node::make_decl_local): Fix code format.
25442 (symtab_node::set_section_for_node): Fix comment typo.
25444 2017-01-06 Martin Liska <mliska@suse.cz>
25447 * lra-constraints.c: Rename invariant to lra_invariant.
25448 * predict.c (set_even_probabilities): Initialize e to NULL.
25450 2017-01-05 Martin Sebor <msebor@redhat.com>
25452 PR tree-optimization/78910
25453 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25454 (format_integer): Correct off-by-one error in the handling
25455 of precision with negative numbers in signed conversions..
25457 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25459 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25461 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25463 PR tree-optimization/71016
25464 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25465 factor_out_conditional_conversion. Formatting fix.
25466 (factor_out_conditional_conversion): Add cond_stmt argument.
25467 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25468 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25471 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25473 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25474 read-rtl-function.o, and selftest-rtl.o.
25475 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25476 (selftest::aarch64_test_loading_full_dump): New function.
25477 (selftest::aarch64_run_selftests): New function.
25478 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25479 selftest::aarch64_run_selftests.
25480 * config/i386/i386.c
25481 (selftest::ix86_test_loading_dump_fragment_1): New function.
25482 (selftest::ix86_test_loading_call_insn): New function.
25483 (selftest::ix86_test_loading_full_dump): New function.
25484 (selftest::ix86_test_loading_unspec): New function.
25485 (selftest::ix86_run_selftests): Call the new functions.
25486 * emit-rtl.c (maybe_set_max_label_num): New function.
25487 * emit-rtl.h (maybe_set_max_label_num): New decl.
25488 * function.c (instantiate_decls): Guard call to
25489 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25490 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25492 * gensupport.c (gen_reader::gen_reader): Pass "false"
25493 for new "compact" param of rtx_reader.
25494 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25495 rather than an empty string for NULL strings.
25496 * read-md.c: Potentially include config.h rather than bconfig.h.
25497 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25498 (have_error): New global, copied from errors.c.
25499 (md_reader::read_name): Rename to...
25500 (md_reader::read_name_1): ...this, adding "out_loc" param,
25501 and converting "missing name or number" to returning false, rather
25503 (md_reader::read_name): Reimplement in terms of read_name_1.
25504 (md_reader::read_name_or_nil): New function.
25505 (md_reader::read_string): Handle "(nil)" by returning NULL.
25506 (md_reader::md_reader): Add new param "compact".
25507 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25508 (md_reader::read_file): New method.
25509 * read-md.h (md_reader::md_reader): Add new param "compact".
25510 (md_reader::read_file): New method.
25511 (md_reader::is_compact): New accessor.
25512 (md_reader::read_name): Convert return type from void to file_location.
25513 (md_reader::read_name_or_nil): New decl.
25514 (md_reader::read_name_1): New decl.
25515 (md_reader::m_compact): New field.
25516 (noop_reader::noop_reader): Pass "false" for new "compact" param
25518 (rtx_reader::rtx_reader): Add new "compact" param.
25519 (rtx_reader::read_rtx_operand): Make virtual and convert return
25520 type from void to rtx.
25521 (rtx_reader::read_until): New decl.
25522 (rtx_reader::handle_any_trailing_information): New virtual function.
25523 (rtx_reader::postprocess): New virtual function.
25524 (rtx_reader::finalize_string): New virtual function.
25525 (rtx_reader::m_in_call_function_usage): New field.
25526 (rtx_reader::m_reuse_rtx_by_id): New field.
25527 * read-rtl-function.c: New file.
25528 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25529 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25530 (selftest::verify_three_block_rtl_cfg): New decl.
25531 * read-rtl-function.h: New file.
25532 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25533 For host, include function.h, memmodel.h, and emit-rtl.h.
25534 (one_time_initialization): New function.
25535 (struct compact_insn_name): New struct.
25536 (compact_insn_names): New array.
25537 (find_code): Handle insn codes in compact dumps.
25538 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25539 (bind_subst_iter_and_attr): Likewise.
25540 (add_condition_to_string): Likewise.
25541 (add_condition_to_rtx): Likewise.
25542 (apply_attribute_uses): Likewise.
25543 (add_current_iterators): Likewise.
25544 (apply_iterators): Likewise.
25545 (initialize_iterators): Guard usage of apply_subst_iterator with
25546 #ifdef GENERATOR_FILE.
25547 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25548 (md_reader::read_mapping): Likewise.
25549 (add_define_attr_for_define_subst): Likewise.
25550 (add_define_subst_attr): Likewise.
25551 (read_subst_mapping): Likewise.
25552 (check_code_iterator): Likewise.
25553 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25555 (one_time_initialization): New function.
25556 (rtx_reader::read_until): New method.
25557 (read_flags): New function.
25558 (parse_reg_note_name): New function.
25559 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25560 Handle reuse_rtx ids.
25561 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25562 Add parsing support for RTL dumps, mirroring the special-cases in
25563 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25564 values, and calling handle_any_trailing_information.
25565 (rtx_reader::read_rtx_operand): Convert return type from void
25566 to rtx, returning return_rtx. Handle case 'e'. Call
25567 finalize_string on XSTR and XTMPL fields.
25568 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25569 "(nil)" values were omitted. Call the postprocess vfunc on the
25571 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25572 class ctor. Initialize m_in_call_function_usage. Call
25573 one_time_initialization.
25574 * rtl-tests.c (selftest::test_uncond_jump): Call
25575 set_new_first_and_last_insn.
25576 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25577 * selftest-rtl.c: New file.
25578 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25579 (selftest::get_insn_by_uid): New decl.
25580 * selftest-run-tests.c (selftest::run_tests): Call
25581 read_rtl_function_c_tests.
25582 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25583 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25586 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25588 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25589 operands in a special way. Assert that pos+len <= mode precision.
25591 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25593 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25594 3 argument Alias with unlimited for the negative form.
25595 (fno-vect-cost-model): Removed.
25597 2017-01-05 Martin Liska <mliska@suse.cz>
25599 * hsa-gen.c (gen_hsa_divmod): New function.
25600 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25602 2017-01-05 Martin Liska <mliska@suse.cz>
25605 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25608 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25610 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25611 small constant length operands.
25613 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25615 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25616 between loop iterations.
25618 2017-01-05 Martin Liska <mliska@suse.cz>
25621 * gimplify.c (gimplify_decl_expr): Compare to
25622 asan_poisoned_variables instread of checking flags.
25623 (gimplify_target_expr): Likewise.
25624 (gimplify_expr): Likewise.
25625 (gimplify_function_tree): Conditionally initialize
25626 asan_poisoned_variables.
25628 2017-01-04 Jeff Law <law@redhat.com>
25630 PR tree-optimizatin/78812
25631 * rtl.h (contains_mem_rtx_p): Prototype.
25632 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25633 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25634 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25635 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25636 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25638 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25640 * input.c (assert_char_at_range): Default-initialize actual_range.
25642 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25644 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25645 to match the caller.
25647 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25649 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25650 insns after final jump in test to emit dummy move.
25652 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25654 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25655 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25657 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25659 * multiple_target.c (create_dispatcher_calls): Init e_next.
25660 * tree-ssa-loop-split.c (split_loop): Init border.
25661 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25664 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25669 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25670 (altivec_register_operand): Do not return true if the operand
25671 contains a SUBREG mixing SImode and SFmode.
25672 (vsx_register_operand): Likewise.
25673 (vsx_reg_sfsubreg_ok): New predicate.
25674 (vfloat_operand): Do not return true if the operand contains a
25675 SUBREG mixing SImode and SFmode.
25676 (vint_operand): Likewise.
25677 (vlogical_operand): Likewise.
25678 (gpc_reg_operand): Likewise.
25679 (int_reg_operand): Likewise.
25680 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25681 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25682 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25684 (rs6000_emit_move_si_sf_subreg): New helper function.
25685 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25686 fixup SUBREGs involving SImode and SFmode.
25687 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25688 numbers for the new peephole2 optimization.
25689 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25690 the GLIBC math library that do AND/IOR/XOR operations on single
25691 precision floating point.
25692 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25693 target macros to say whether we need to avoid SUBREGs mixing
25695 (TARGET_ALLOW_SF_SUBREG): Likewise.
25696 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25697 (UNSPEC_SI_FROM_SF): Likewise.
25698 (iorxor): Change spacing.
25699 (and_ior_xor): New iterator for AND, IOR, and XOR.
25700 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25701 (movdi_from_sf_zero_ext): Likewise.
25702 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25703 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25704 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25705 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25706 (fms<mode>4): Likewise.
25707 (fnma<mode>4): Likewise.
25708 (fnms<mode>4): Likewise.
25709 (nfma<mode>4): Likewise.
25710 (nfms<mode>4): Likewise.
25712 2017-01-04 Marek Polacek <polacek@redhat.com>
25715 * doc/invoke.texi: Document -Wpointer-compare.
25717 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25719 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25722 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25723 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25726 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25729 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25730 documentation of the powerpc_popcntb_ok attribute.
25731 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25732 code to issue warning messages if a requested CPU configuration is
25733 not supported by the binary (assembler and loader) toolchain.
25734 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25735 made to define a built-in function that has been disabled.
25736 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25737 made to define a built-in function that has been disabled.
25738 (altivec_init_builtins): Add comment explaining why definition
25739 of the DST built-in functions is not preceded by an assertion
25740 check. Add assertions to prevent ICE if attempts are made to
25741 define an altivec predicate or an abs* built-in function that has
25743 (htm_init_builtins): Add comment explaining why definition of the
25744 htm built-in functions is not preceded by an assertion check.
25746 2017-01-04 Jeff Law <law@redhat.com>
25748 PR tree-optimizatin/67955
25749 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25750 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25751 the points-to solution does not include pt_null. Use DECL_PT_UID
25754 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25756 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25757 Use gen_int_mode instead of gen_lopwart for const_int operands.
25759 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25761 PR tree-optimization/71563
25762 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25763 out of range value - has low bits known to be zero.
25765 2017-01-04 Alan Modra <amodra@gmail.com>
25767 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25768 * configure: Regenerate.
25769 * config.in: Regenerate.
25771 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25774 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25775 a substring of the message, but strcmp with the whole message. Ifdef
25776 ENABLE_NLS, translate the message first using dgettext.
25778 2017-01-03 Jeff Law <law@redhat.com>
25780 PR tree-optimizatin/78856
25781 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25782 (mark_threaded_blocks): Remove code to truncate thread paths that
25783 cross multiple loop headers. Instead invalidate the cached loop
25784 iteration information and handle case of a thread path walking
25785 into an irreducible region.
25787 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25790 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25791 assertions. Add support for doing the signbit if the IEEE 128-bit
25792 floating point value is in a GPR.
25793 * config/rs6000/rs6000.md (Fsignbit): Delete.
25794 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25795 Update the length attribute if the value is in a GPR.
25796 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25797 the sign or zero extension instruction, since the value is always 0/1.
25798 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25801 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25802 extracting SImode to a GPR register so that we can generate a
25803 store, limit the vector to be in a traditional Altivec register
25804 for the vextuwrx instruction.
25806 2017-01-03 Ian Lance Taylor <iant@google.com>
25808 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25810 2017-01-03 Martin Sebor <msebor@redhat.com>
25812 PR tree-optimization/78696
25813 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25814 precision. Use MPFR for %f for greater fidelity. Correct handling
25816 (pass_sprintf_length::compute_format_length): Set width and precision
25817 specified by asrerisk to void_node for vararg functions.
25818 (try_substitute_return_value): Adjust dump output.
25820 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25822 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25824 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25826 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25827 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25828 -mlra/-mno-lra was passed to the compiler.
25830 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25832 PR rtl-optimization/65618
25833 * emit-rtl.c (try_split): Move initialization of "before" and
25834 "after" to just before the call to emit_insn_after_setloc.
25836 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25838 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25840 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25842 * dwarf2out.c (gen_enumeration_type_die): When
25843 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25845 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25847 PR tree-optimization/78965
25848 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25849 Change first argument from const call_info & to call_info &. For %n
25850 set info.nowrite to false.
25852 PR middle-end/78901
25853 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25854 possibly throwing calls.
25856 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25857 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25858 and fns handling, rather than in a separate case SSA_NAME.
25860 2017-01-02 Jeff Law <law@redhat.com>
25862 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25863 fixes for first_period and second_period variables.
25865 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25868 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25869 (*insvqi_1): New insn pattern.
25870 (*insvqi_1_mem_rex64): Ditto.
25871 (*insvqi_2): Ditto.
25872 (*insvqi_3): Rename from *insvqi.
25874 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25876 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25878 * doc/cfg.texi (Edges): Remove reference to Java.
25879 (Maintaining the CFG): Ditto.
25881 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25883 PR middle-end/77674
25884 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25885 transparent aliases.
25887 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25889 PR middle-end/77484
25890 * predict.def (PRED_CALL): Update hitrate.
25891 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25892 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25893 into direct/indirect/polymorphic variants.
25895 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25897 Update copyright years.
25899 * gcc.c (process_command): Update copyright notice dates.
25900 * gcov-dump.c (print_version): Ditto.
25901 * gcov.c (print_version): Ditto.
25902 * gcov-tool.c (print_version): Ditto.
25903 * gengtype.c (create_file): Ditto.
25904 * doc/cpp.texi: Bump @copying's copyright year.
25905 * doc/cppinternals.texi: Ditto.
25906 * doc/gcc.texi: Ditto.
25907 * doc/gccint.texi: Ditto.
25908 * doc/gcov.texi: Ditto.
25909 * doc/install.texi: Ditto.
25910 * doc/invoke.texi: Ditto.
25912 Copyright (C) 2017 Free Software Foundation, Inc.
25914 Copying and distribution of this file, with or without modification,
25915 are permitted in any medium without royalty provided the copyright
25916 notice and this notice are preserved.