1 2014-06-20 Jan Hubicka <hubicka@ucw.cz>
3 * cgraph.h (tls_model_names): New variable.
4 * print-tree.c (print_node): Simplify.
5 * varpool.c (tls_model_names): New variable.
6 (dump_varpool_node): Output tls model.
8 2014-06-20 Jan Hubicka <hubicka@ucw.cz>
10 * ipa-visibility.c (function_and_variable_visibility): Disable
11 temporarily local aliases for some targets.
13 2014-06-20 Marek Polacek <polacek@redhat.com>
15 * asan.c (pass_sanopt::execute): Handle IFN_UBSAN_BOUNDS.
16 * flag-types.h (enum sanitize_code): Add SANITIZE_BOUNDS and or it
17 into SANITIZE_UNDEFINED.
18 * doc/invoke.texi: Describe -fsanitize=bounds.
19 * gimplify.c (gimplify_call_expr): Add gimplification of internal
20 functions created in the FEs.
21 * internal-fn.c: Move "internal-fn.h" after "tree.h".
22 (expand_UBSAN_BOUNDS): New function.
23 * internal-fn.def (UBSAN_BOUNDS): New internal function.
24 * internal-fn.h: Don't define internal functions here.
25 * opts.c (common_handle_option): Add -fsanitize=bounds.
26 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_OUT_OF_BOUNDS,
27 BUILT_IN_UBSAN_HANDLE_OUT_OF_BOUNDS_ABORT): Add.
28 * tree-core.h: Define internal functions here.
29 (struct tree_base): Add ifn field.
30 * tree-pretty-print.c: Include "internal-fn.h".
31 (dump_generic_node): Handle functions without CALL_EXPR_FN.
32 * tree.c (get_callee_fndecl): Likewise.
33 (build_call_expr_internal_loc): New function.
34 * tree.def (CALL_EXPR): Update description.
35 * tree.h (CALL_EXPR_IFN): Define.
36 (build_call_expr_internal_loc): Declare.
37 * ubsan.c (get_ubsan_type_info_for_type): Return 0 for non-arithmetic
39 (ubsan_type_descriptor): Change bool parameter to enum
40 ubsan_print_style. Adjust the code. Add handling of
42 (ubsan_expand_bounds_ifn): New function.
43 (ubsan_expand_null_ifn): Adjust ubsan_type_descriptor call.
44 (ubsan_build_overflow_builtin): Likewise.
45 (instrument_bool_enum_load): Likewise.
46 (ubsan_instrument_float_cast): Likewise.
47 * ubsan.h (enum ubsan_print_style): New enum.
48 (ubsan_expand_bounds_ifn): Declare.
49 (ubsan_type_descriptor): Adjust declaration. Use a default parameter.
51 2014-06-20 Maciej W. Rozycki <macro@codesourcery.com>
53 * config/rs6000/rs6000.md: Append `DONE' to preparation
54 statements of `bswap' pattern splitters.
56 2014-06-20 Tom de Vries <tom@codesourcery.com>
58 * target.def (call_fusage_contains_non_callee_clobbers): Update
60 * doc/tm.texi: Regenerate.
62 2014-06-20 Yury Gribov <y.gribov@samsung.com>
63 Max Ostapenko <m.ostapenko@partner.samsung.com>
66 * asan.c (instrument_strlen_call): Fixed instrumentation of
69 2014-06-20 Martin Jambor <mjambor@suse.cz>
72 * ipa-prop.c (impossible_devirt_target): New function.
73 (try_make_edge_direct_virtual_call): Use it, also instead of
76 2014-06-20 Yury Gribov <y.gribov@samsung.com>
77 Max Ostapenko <m.ostapenko@partner.samsung.com>
80 * asan.c (build_check_stmt): Add condition.
82 2014-06-20 Martin Jambor <mjambor@suse.cz>
85 * cgraph.c (clone_of_p): Allow skipped_branch to deal with
88 2014-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
90 * config/aarch64/iterators.md (VCOND): Handle SI and HI modes.
92 (VCONQ): Make comment more helpful.
94 * config/aarch64/aarch64-simd.md
95 (aarch64_sqdmulh_lane<mode>):
96 Use VCOND for operands 2. Update lane checking and flipping logic.
97 (aarch64_sqrdmulh_lane<mode>): Likewise.
98 (aarch64_sq<r>dmulh_lane<mode>_internal): Likewise.
99 (aarch64_sqdmull2<mode>): Remove VCON, use VQ_HSI mode iterator.
100 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal, VD_HSI): Change mode
101 attribute of operand 3 to VCOND.
102 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal, SD_HSI): Likewise.
103 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
104 (aarch64_sqdmull_lane<mode>_internal, VD_HSI): Likewise.
105 (aarch64_sqdmull_lane<mode>_internal, SD_HSI): Likewise.
106 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
107 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>_internal, VD_HSI: New
109 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>_internal, SD_HSI): Likewise.
110 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
111 (aarch64_sqdmull_laneq<mode>_internal, VD_HSI): Likewise.
112 (aarch64_sqdmull_laneq<mode>_internal, SD_HSI): Likewise.
113 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
114 (aarch64_sqdmlal_lane<mode>): Change mode attribute of penultimate
115 operand to VCOND. Update lane flipping and bounds checking logic.
116 (aarch64_sqdmlal2_lane<mode>): Likewise.
117 (aarch64_sqdmlsl_lane<mode>): Likewise.
118 (aarch64_sqdmull_lane<mode>): Likewise.
119 (aarch64_sqdmull2_lane<mode>): Likewise.
120 (aarch64_sqdmlal_laneq<mode>):
121 Replace VCON usage with VCONQ.
122 Emit aarch64_sqdmlal_laneq<mode>_internal insn.
123 (aarch64_sqdmlal2_laneq<mode>): Emit
124 aarch64_sqdmlal2_laneq<mode>_internal insn.
125 Replace VCON with VCONQ.
126 (aarch64_sqdmlsl2_lane<mode>): Replace VCON with VCONQ.
127 (aarch64_sqdmlsl2_laneq<mode>): Likewise.
128 (aarch64_sqdmull_laneq<mode>): Emit
129 aarch64_sqdmull_laneq<mode>_internal insn.
130 Replace VCON with VCONQ.
131 (aarch64_sqdmull2_laneq<mode>): Emit
132 aarch64_sqdmull2_laneq<mode>_internal insn.
133 (aarch64_sqdmlsl_laneq<mode>): Replace VCON usage with VCONQ.
134 * config/aarch64/arm_neon.h (vqdmlal_high_lane_s16): Change type
135 of 3rd argument to int16x4_t.
136 (vqdmlalh_lane_s16): Likewise.
137 (vqdmlslh_lane_s16): Likewise.
138 (vqdmull_high_lane_s16): Likewise.
139 (vqdmullh_lane_s16): Change type of 2nd argument to int16x4_t.
140 (vqdmlal_lane_s16): Don't create temporary int16x8_t value.
141 (vqdmlsl_lane_s16): Likewise.
142 (vqdmull_lane_s16): Don't create temporary int16x8_t value.
143 (vqdmlal_high_lane_s32): Change type 3rd argument to int32x2_t.
144 (vqdmlals_lane_s32): Likewise.
145 (vqdmlsls_lane_s32): Likewise.
146 (vqdmull_high_lane_s32): Change type 2nd argument to int32x2_t.
147 (vqdmulls_lane_s32): Likewise.
148 (vqdmlal_lane_s32): Don't create temporary int32x4_t value.
149 (vqdmlsl_lane_s32): Likewise.
150 (vqdmull_lane_s32): Don't create temporary int32x4_t value.
151 (vqdmulhh_lane_s16): Change type of second argument to int16x4_t.
152 (vqrdmulhh_lane_s16): Likewise.
153 (vqdmlsl_high_lane_s16): Likewise.
154 (vqdmulhs_lane_s32): Change type of second argument to int32x2_t.
155 (vqdmlsl_high_lane_s32): Likewise.
156 (vqrdmulhs_lane_s32): Likewise.
158 2014-06-20 Tom de Vries <tom@codesourcery.com>
160 * final.c (collect_fn_hard_reg_usage): Add separate IOR_HARD_REG_SET for
161 get_call_reg_set_usage.
163 2014-06-20 Tom de Vries <tom@codesourcery.com>
165 * final.c (collect_fn_hard_reg_usage): Don't save function_used_regs if
166 it contains all call_used_regs.
168 2014-06-20 Tom de Vries <tom@codesourcery.com>
170 * final.c (collect_fn_hard_reg_usage): Add and use variable
173 2014-06-20 Jan Hubicka <hubicka@ucw.cz>
175 * cgraph.h (struct symtab_node): Add field in_init_priority_hash
176 (set_init_priority, get_init_priority, set_fini_priority,
177 get_fini_priority): New methods.
178 * tree.c (init_priority_for_decl): Remove.
179 (init_ttree): Do not initialize init priority.
180 (decl_init_priority_lookup, decl_fini_priority_lookup): Rewrite.
181 (decl_priority_info): Remove.
182 (decl_init_priority_insert): Rewrite.
183 (decl_fini_priority_insert): Rewrite.
184 * tree.h (tree_priority_map_eq, tree_priority_map_hash,
185 tree_priority_map_marked_p): Remove.
186 * lto-cgraph.c (lto_output_node, input_node): Stream init priorities.
187 * lto-streamer-out.c (hash_tree): Do not hash priorities.
188 * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields): Do
189 not output priorities.
190 (pack_ts_function_decl_value_fields): Likewise.
191 * tree-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Do
192 not input priorities.
193 (unpack_ts_function_decl_value_fields): Likewise.
194 * symtab.c (symbol_priority_map): Declare.
195 (init_priority_hash): Declare.
196 (symtab_unregister_node): Unregister from priority hash, too.
197 (symtab_node::get_init_priority, cgraph_node::get_fini_priority):
199 (symbol_priority_map_eq, symbol_priority_map_hash): New functions.
200 (symbol_priority_info): New function.
201 (symtab_node::set_init_priority, cgraph_node::set_fini_priority):
203 * tree-core.h (tree_priority_map): Remove.
205 2014-06-20 Jakub Jelinek <jakub@redhat.com>
207 * tree-ssa-math-opts.c (do_shift_rotate, find_bswap_or_nop_1): Cast
208 0xff to uint64_t before shifting it up.
210 2014-06-20 Julian Brown <julian@codesourcery.com>
211 Chung-Lin Tang <cltang@codesourcery.com>
213 * config/arm/arm.c (arm_output_mi_thunk): Fix offset for
214 TARGET_THUMB1_ONLY. Add comments.
216 2014-06-19 Tom de Vries <tom@codesourcery.com>
218 * config/aarch64/aarch64-protos.h (aarch64_emit_call_insn): Change
220 * config/aarch64/aarch64.c (aarch64_emit_call_insn): Same.
222 2014-06-19 Zhenqiang Chen <zhenqiang.chen@linaro.org>
224 * loop-invariant.c (get_inv_cost): Skip invariants, which are marked
225 as "move", from depends_on.
227 2014-06-19 Terry Guo <terry.guo@arm.com>
229 * config/arm/thumb1.md (define_split): Split 64bit constant in earlier
232 2014-06-18 Segher Boessenkool <segher@kernel.crashing.org>
234 * config/rs6000/rs6000.h (FIXED_REGISTERS): Update comment.
236 (REG_ALLOC_ORDER): Update comment. Move cr5 earlier.
238 2014-06-18 Kaz Kojima <kkojima@gcc.gnu.org>
241 * config/sh/sh.c (prepare_move_operands): Don't process TLS
242 addresses here if reload in progress or completed.
244 2014-06-18 Robert Suchanek <robert.suchanek@imgtec.com>
246 * config/mips/constraints.md ("d"): BASE_REG_CLASS replaced by
247 "TARGET_MIPS16 ? M16_REGS : GR_REGS".
248 * config/mips/mips.c (mips_regno_to_class): Update for M16_SP_REGS.
249 (mips_regno_mode_ok_for_base_p): Remove use of !strict_p for MIPS16.
250 (mips_register_priority): New function that implements the target
251 hook TARGET_REGISTER_PRIORITY.
252 (mips_spill_class): Likewise for TARGET_SPILL_CLASS.
253 (mips_lra_p): Likewise for TARGET_LRA_P.
254 (TARGET_REGISTER_PRIORITY): Define macro.
255 (TARGET_SPILL_CLASS): Likewise.
256 (TARGET_LRA_P): Likewise.
257 * config/mips/mips.h (reg_class): Add M16_SP_REGS and SPILL_REGS
259 (REG_CLASS_NAMES): Likewise.
260 (REG_CLASS_CONTENTS): Likewise.
261 (BASE_REG_CLASS): Use M16_SP_REGS.
262 * config/mips/mips.md (*mul_acc_si): Add alternative tuned for LRA.
263 New set attribute to enable alternatives depending on the register
265 (*mul_acc_si_r3900, *mul_sub_si): Likewise.
266 (*lea64): Disable pattern for MIPS16.
267 * config/mips/mips.opt (mlra): New option.
269 2014-06-18 Robert Suchanek <robert.suchanek@imgtec.com>
271 * lra-constraints.c (base_to_reg): New function.
272 (process_address): Use new function.
274 2014-06-18 Tom de Vries <tom@codesourcery.com>
276 * config/aarch64/aarch64-protos.h (aarch64_emit_call_insn): Declare.
277 * config/aarch64/aarch64.c
278 (TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS): Redefine as true.
279 (aarch64_emit_call_insn): New function.
280 (aarch64_load_symref_appropriately): Use aarch64_emit_call_insn instead
282 * config/aarch64/aarch64.md (define_expand "call_internal")
283 (define_expand "call_value_internal", define_expand "sibcall_internal")
284 (define_expand "sibcall_value_internal"): New.
285 (define_expand "call", define_expand "call_value")
286 (define_expand "sibcall", define_expand "sibcall_value"): Use internal
287 expand variant and aarch64_emit_call_insn.
289 2014-06-18 Radovan Obradovic <robradovic@mips.com>
290 Tom de Vries <tom@codesourcery.com>
292 * config/arm/arm-protos.h (arm_emit_call_insn): Add bool parameter.
293 * config/arm/arm.c (TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS):
295 (arm_emit_call_insn): Add and use sibcall parameter. Add IP and CC
296 clobbers to CALL_INSN_FUNCTION_USAGE.
297 (define_expand "sibcall_internal")
298 (define_expand "sibcall_value_internal"): New.
299 (define_expand "call", define_expand "call_value"): Add argument to
301 (define_expand "sibcall"): Use sibcall_internal and arm_emit_call_insn.
302 (define_expand "sibcall_value"): Use sibcall_value_internal and
305 2014-06-18 Charles Baylis <charles.baylis@linaro.org>
307 * config/arm/bpabi.c (__gnu_uldivmod_helper): Remove.
309 2014-06-18 Charles Baylis <charles.baylis@linaro.org>
311 * config/arm/bpabi-v6m.S (__aeabi_uldivmod): Perform division using
314 2014-06-18 Charles Baylis <charles.baylis@linaro.org>
316 * config/arm/bpabi.S (__aeabi_ldivmod, __aeabi_uldivmod,
317 push_for_divide, pop_for_divide): Use .cfi_* directives for DWARF
318 annotations. Fix DWARF information.
320 2014-06-18 Charles Baylis <charles.baylis@linaro.org>
322 * config/arm/bpabi.S (__aeabi_ldivmod): Perform division using
323 __udivmoddi4, and fixups for negative operands.
325 2014-06-18 Charles Baylis <charles.baylis@linaro.org>
327 * config/arm/bpabi.S (__aeabi_ldivmod): Optimise stack manipulation.
329 2014-06-18 Charles Baylis <charles.baylis@linaro.org>
331 * config/arm/bpabi.S (__aeabi_uldivmod): Perform division using call
334 2014-06-18 Charles Baylis <charles.baylis@linaro.org>
336 * config/arm/bpabi.S (__aeabi_uldivmod): Optimise stack pointer
339 2014-06-18 Charles Baylis <charles.baylis@linaro.org>
341 * config/arm/bpabi.S (__aeabi_uldivmod, __aeabi_ldivmod): Add comment
342 describing register usage on function entry and exit.
344 2014-06-18 Charles Baylis <charles.baylis@linaro.org>
346 * config/arm/bpabi.S (__aeabi_uldivmod): Fix whitespace.
347 (__aeabi_ldivmod): Fix whitespace.
349 2014-06-18 Andreas Schwab <schwab@suse.de>
351 * doc/md.texi (Standard Names): Use @itemx for grouped items.
352 Remove blank line after @item.
354 2014-06-18 Richard Henderson <rth@redhat.com>
357 * config/aarch64/aarch64.md (tlsdesc_small_<PTR>): Clobber CC_REGNUM.
359 2014-06-18 Charles Baylis <charles.baylis@linaro.org>
361 * config/arm/arm.c (neon_vector_mem_operand): Allow register
362 POST_MODIFY for neon loads and stores.
363 (arm_print_operand): Output post-index register for neon loads and
366 2014-06-18 Richard Biener <rguenther@suse.de>
368 * tree-ssa-dce.c (perform_tree_ssa_dce): Fixup bogus commit.
370 2014-06-18 Richard Biener <rguenther@suse.de>
372 * tree-pass.h (make_pass_dce_loop): Remove.
373 * passes.def: Replace pass_dce_loop with pass_dce.
374 * tree-ssa-dce.c (perform_tree_ssa_dce): If something
375 changed free niter estimates and reset the scev cache.
376 (tree_ssa_dce_loop, pass_data_dce_loop, pass_dce_loop,
377 make_pass_dce_loop): Remove.
378 * tree-ssa-copy.c: Include tree-ssa-loop-niter.h.
379 (fini_copy_prop): Return whether something changed. Always
380 let substitute_and_fold perform DCE and free niter estimates
381 and reset the scev cache if so.
382 (execute_copy_prop): If sth changed schedule cleanup-cfg.
383 (pass_data_copy_prop): Do not unconditionally schedule
384 cleanup-cfg or update-ssa.
386 2014-06-18 Yuri Rumyantsev <ysrumyan@gmail.com>
388 PR tree-optimization/61518
389 * tree-if-conv.c (is_cond_scalar_reduction): Add missed check that
390 reduction var is used in reduction stmt or phi-function only.
392 2014-06-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
394 * config/arm/arm_neon.h (vadd_f32): Change #ifdef to __FAST_MATH.
396 2014-06-18 Thomas Preud'homme <thomas.preudhomme@arm.com>
398 PR tree-optimization/61517
399 * tree-ssa-math-opts.c (find_bswap_or_nop_1): Adapt to return a stmt
400 whose rhs's first tree is the source expression instead of the
402 (find_bswap_or_nop): Likewise.
403 (bsap_replace): Rename stmt in cur_stmt. Pass gsi by value and src as a
404 gimple stmt whose rhs's first tree is the source. In the memory source
405 case, move the stmt to be replaced close to one of the original load to
406 avoid the problem of a store between the load and the stmt's original
408 (pass_optimize_bswap::execute): Adapt to change in bswap_replace's
411 2014-06-18 Andreas Schwab <schwab@suse.de>
413 PR rtl-optimization/54555
414 * postreload.c (move2add_use_add2_insn): Substitute
415 STRICT_LOW_PART only if it is cheaper.
417 2014-06-18 Uros Bizjak <ubizjak@gmail.com>
419 * config/i386/i386.md (*sibcall_memory): Rename from *sibcall_intern.
420 Do not use unspec as call operand. Use memory_operand instead of
421 memory_nox32_operand and add "m" operand constraint. Disable
422 pattern for TARGET_X32.
423 (*sibcall_pop_memory): Ditto.
424 (*sibcall_value_memory): Ditto.
425 (*sibcall_value_pop_memory): Ditto.
426 (sibcall peepholes): Merge SImode and DImode patterns using
427 W mode iterator. Use memory_operand instead of memory_nox32_operand.
428 Disable pattern for TARGET_X32. Check if eliminated register is
429 really dead after call insn. Generate call RTX without unspec operand.
430 (sibcall_value peepholes): Ditto.
431 (sibcall_pop peepholes): Fix call insn RTXes. Use memory_operand
432 instead of memory_nox32_operand. Check if eliminated register is
433 really dead after call insn. Generate call RTX without unspec operand.
434 (sibcall_value_pop peepholes): Ditto.
435 * config/i386/predicates.md (memory_nox32_operand): Remove predicate.
437 2014-06-18 Terry Guo <terry.guo@arm.com>
440 * config/arm/arm.c (thumb1_reorg): Move to next basic block if we
443 2014-06-18 Richard Biener <rguenther@suse.de>
445 * common.opt (fssa-phiopt): New option.
446 * opts.c (default_options_table): Enable -fssa-phiopt with -O1+
448 * tree-ssa-phiopt.c (pass_phiopt): Add gate method.
449 * doc/invoke.texi (-fssa-phiopt): Document.
451 2014-06-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
453 * genattrtab.c (n_bypassed): New variable.
454 (process_bypasses): Initialise n_bypassed.
455 Count number of bypassed reservations.
456 (make_automaton_attrs): Allocate space for bypassed reservations
457 rather than number of bypasses.
459 2014-06-18 Richard Biener <rguenther@suse.de>
461 * tree-ssa-propagate.c (replace_phi_args_in): Return whether
462 we propagated anything.
463 (substitute_and_fold_dom_walker::before_dom_children): Something
464 changed if we propagated into PHI arguments.
465 * tree-ssa-pre.c (eliminate): Always schedule cfg-cleanup if
468 2014-06-18 Evgeny Stupachenko <evstupac@gmail.com>
470 * config/i386/i386.c (ix86_reassociation_width): Add alternative for
472 * config/i386/i386.h (TARGET_VECTOR_PARALLEL_EXECUTION): New.
473 * config/i386/x86-tune.def (X86_TUNE_VECTOR_PARALLEL_EXECUTION): New.
474 * tree-vect-data-refs.c (vect_shift_permute_load_chain): New.
475 Introduces alternative way of loads group permutaions.
476 (vect_transform_grouped_load): Try alternative way of permutations.
478 2014-06-18 Jakub Jelinek <jakub@redhat.com>
480 * gimplify.c (omp_notice_variable): If n is non-NULL and no flags
481 changed in ORT_TARGET region, don't jump to do_outer.
482 (struct gimplify_adjust_omp_clauses_data): New type.
483 (gimplify_adjust_omp_clauses_1): Adjust for data being
484 a struct gimplify_adjust_omp_clauses_data pointer instead
485 of tree *. Pass pre_p as a new argument to
486 lang_hooks.decls.omp_finish_clause hook.
487 (gimplify_adjust_omp_clauses): Add pre_p argument, adjust
488 splay_tree_foreach to pass both list_p and pre_p.
489 (gimplify_omp_parallel, gimplify_omp_task, gimplify_omp_for,
490 gimplify_omp_workshare, gimplify_omp_target_update): Adjust
491 gimplify_adjust_omp_clauses callers.
492 * langhooks.c (lhd_omp_finish_clause): New function.
493 * langhooks-def.h (lhd_omp_finish_clause): New prototype.
494 (LANG_HOOKS_OMP_FINISH_CLAUSE): Define to lhd_omp_finish_clause.
495 * langhooks.h (struct lang_hooks_for_decls): Add a new
496 gimple_seq * argument to omp_finish_clause hook.
497 * omp-low.c (scan_sharing_clauses): Call scan_omp_op on
498 non-DECL_P OMP_CLAUSE_DECL if ctx->outer.
499 (scan_omp_parallel, lower_omp_for): When adding
500 _LOOPTEMP_ clause var, add it to outer ctx's decl_map as identity.
501 * tree-core.h (OMP_CLAUSE_MAP_TO_PSET): New map kind.
502 * tree-nested.c (convert_nonlocal_omp_clauses,
503 convert_local_omp_clauses): Handle various OpenMP 4.0 clauses.
504 * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE_MAP_TO_PSET.
506 2014-06-17 Andrew MacLeod <amacleod@redhat.com>
508 * tree-dfa.h (get_addr_base_and_unit_offset_1): Move from here.
509 * tree-dfa.c (get_addr_base_and_unit_offset_1): To here.
511 2014-06-17 Xinliang David Li <davidxl@google.com>
513 * tree-pretty-print.c (dump_function_header): Print cgraph uid.
514 * passes.c (pass_init_dump_file): Do not set initialize
515 flag to false unconditionally.
517 2014-06-17 Richard Biener <rguenther@suse.de>
519 * genopinit.c (main): Use vec<>::qsort method.
520 * tree-ssa-loop-niter.c (discover_iteration_bound_by_body_walk):
522 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Likewise.
524 2014-06-17 Matthew Fortune <matthew.fortune@imgtec.com>
526 * config/mips/mips-protos.h (mips_expand_fcc_reload): Remove.
527 * config/mips/mips.c (mips_expand_fcc_reload): Remove.
528 (mips_move_to_gpr_cost): Remove ST_REGS case.
529 (mips_move_from_gpr_cost): Likewise.
530 (mips_register_move_cost): Likewise.
531 (mips_secondary_reload_class): Likewise.
533 2014-06-17 Richard Biener <rguenther@suse.de>
535 * passes.def (pass_all_early_optimizations): Remove copy-prop pass.
536 (pass_all_optimizations): Move 3rd copy-prop pass from after
537 fre to before ifcombine/phiopt.
539 2014-06-17 Richard Biener <rguenther@suse.de>
541 * tree-switch-conversion.c (collect_switch_conv_info): Simplify
542 and allow all blocks to be forwarders.
544 2014-06-17 Yufeng Zhang <yufeng.zhang@arm.com>
547 * config/aarch64/aarch64.c (aarch64_layout_arg): Add new local
548 variable 'size'; calculate 'size' right in the front; use
549 'size' to compute 'nregs' (when 'allocate_ncrn != 0') and
550 pcum->aapcs_stack_words.
552 2014-06-17 Nick Clifton <nickc@redhat.com>
554 * config/msp430/msp430.md (mulhisi3): Add a NOP after the DINT.
555 (umulhi3, mulsidi3, umulsidi3): Likewise.
557 2014-06-17 Thomas Schwinge <thomas@codesourcery.com>
560 * fold-const.c (fold_checksum_tree) <TS_DECL_WITH_VIS>: Remove
561 check for section name.
563 2014-06-17 Richard Biener <rguenther@suse.de>
565 * tree-ssa-propagate.c: Include domwalk.h.
566 (substitute_and_fold): Outline main worker into a domwalker ...
567 (substitute_and_fold_dom_walker::before_dom_children): ... here.
568 Schedule stmts we can fully propagate for removal. Remove
570 (substitute_and_fold): Apply a dominator walk to perform
571 substitution. Process stmts scheduled for removal here.
573 2014-06-17 Richard Biener <rguenther@suse.de>
575 * tree-ssa-loop-im.c (determine_max_movement): Adjust cost
578 2014-06-17 Kugan Vivekanandarajah <kuganv@linaro.org>
580 * config/arm/arm.c (arm_atomic_assign_expand_fenv): call
581 default_atomic_assign_expand_fenv for !TARGET_HARD_FLOAT.
582 (arm_init_builtins) : Initialize builtins __builtins_arm_set_fpscr and
583 __builtins_arm_get_fpscr only when TARGET_HARD_FLOAT.
584 * config/arm/vfp.md (set_fpscr): Make pattern conditional on
586 (get_fpscr) : Likewise.
588 2014-06-16 Vladimir Makarov <vmakarov@redhat.com>
590 PR rtl-optimization/61325
591 * lra-constraints.c (valid_address_p): Add forward declaration.
592 (simplify_operand_subreg): Check address validity before and after
593 alter_reg of memory subreg.
595 2014-06-16 Uros Bizjak <ubizjak@gmail.com>
597 * config/i386/i386.c (decide_alg): Correctly handle
598 maximum size of stringop algorithm.
600 2014-06-16 Yury Gribov <y.gribov@samsung.com>
602 * asan.c (build_check_stmt): Fix maybe-uninitialized warning.
604 2014-06-16 Vladimir Makarov <vmakarov@redhat.com>
606 PR rtl-optimization/61522
607 * lra-assigns.c (assign_by_spills): Check null targetm.spill_class.
609 2014-06-16 Jan Hubicka <hubicka@ucw.cz>
612 * symtab.c (symtab_node::reset_section): New method.
613 * cgraph.c (cgraph_node_cannot_be_local_p_1): Accept non-local
615 * cgraph.h (reset_section): Declare.
616 * ipa-inline-analysis.c (do_estimate_growth): Check for comdat groups;
617 do not consider comdat locals.
618 * cgraphclones.c (set_new_clone_decl_and_node_flags): Get section
620 * ipa-visiblity.c (cgraph_externally_visible_p): Cleanup.
621 (update_visibility_by_resolution_info): Consider UNDEF; fix checking;
622 reset sections of symbols dragged out of the comdats.
623 (function_and_variable_visibility): Reset sections of
626 2014-06-16 Richard Biener <rguenther@suse.de>
628 PR tree-optimization/61482
629 * tree-vrp.c (adjust_range_with_scev): Avoid setting of
630 [-INF(OVF), +INF(OVF)] range.
632 2014-06-16 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
634 * config/i386/i386.c (ix86_expand_sse2_mulvxdi3): Issue
635 instructions "vpmuludq" and "vpaddq" instead of "vpmacsdql" for
636 handling 32-bit multiplication.
638 2014-06-16 Chung-Lin Tang <cltang@codesourcery.com>
641 * lra-lives.c (process_bb_lives): Skip creating copy during
642 insn scan when src/dest has constrained to same regno.
644 2014-06-15 Jan Hubicka <hubicka@ucw.cz>
646 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Check again
647 DECL_IN_CONSTANT_POOL and TREE_ASM_WRITTEN.
649 2014-06-16 Yury Gribov <y.gribov@samsung.com>
651 * asan.c (check_func): New function.
652 (maybe_create_ssa_name): Likewise.
653 (build_check_stmt_with_calls): Likewise.
654 (use_calls_p): Likewise.
655 (report_error_func): Change interface.
656 (build_check_stmt): Allow non-integer lengths; add support
658 (asan_instrument): Likewise.
659 (instrument_mem_region_access): Moved code to build_check_stmt.
660 (instrument_derefs): Likewise.
661 (instrument_strlen_call): Likewise.
662 * cfgcleanup.c (old_insns_match_p): Add support for new functions.
663 * doc/invoke.texi: Describe new parameter.
664 * params.def: Define new parameter.
665 * params.h: Likewise.
666 * sanitizer.def: Describe new builtins.
668 2014-06-16 Richard Biener <rguenther@suse.de>
670 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
671 Make all defs available at the end.
672 (eliminate): If we remove a PHI node schedule cfg-cleanup.
674 2014-06-18 Jakub Jelinek <jakub@redhat.com>
677 * config.gcc (arm*-*-linux-*): Include vxworks-dummy.h in tm_file.
679 2014-06-16 Richard Sandiford <rdsandiford@googlemail.com>
682 * auto-inc-dec.c (merge_in_block): Fix location of insn_info
683 initialization. Replace remaining use of uid.
685 2014-06-15 Jan Hubicka <hubicka@ucw.cz>
687 * c-family/c-common.c (handle_tls_model_attribute): Use
689 * c-family/c-common.c (handle_tls_model_attribute): Use
691 * cgraph.h (struct varpool_node): Add tls_model.
692 * tree.c (decl_tls_model, set_decl_tls_model): New functions.
693 * tree.h (DECL_TLS_MODEL): Update.
694 (DECL_THREAD_LOCAL_P): Check that variable is static.
695 (decl_tls_model): Declare.
696 (set_decl_tls_model): Declare.
697 * tree-emutls.c (get_emutls_init_templ_addr): First build decl and then
698 set symbol prorperties.
699 (get_emutls_init_templ_addr): Cleanup.
700 (new_emutls_decl): Update.
701 * lto-cgraph.c (lto_output_varpool_node): Stream TLS model
702 (lto_input_varpool_node): Likewise.
703 * lto-streamer-out.c (hash_tree): Likewise.
704 * tree-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Do
705 not stream DECL_TLS_MODEL.
706 * tree-profile.c (init_ic_make_global_vars): Use set_decl_tls_model.
707 * tree-core.h (tree_decl_with_vis): Remove tls_model; update comments.
709 2014-06-15 Richard Sandiford <rdsandiford@googlemail.com>
711 * df.h (DF_REF_REG_USE_P, DF_MWS_REG_USE_P): Remove null checks.
713 2014-06-15 Richard Sandiford <rdsandiford@googlemail.com>
715 * df.h (df_mw_hardreg, df_base_ref): Add a link pointer.
716 (df_insn_info): Turn defs, uses, eq_uses and mw_hardregs into linked
718 (df_scan_bb_info): Likewise artificial_defs and artificial_uses.
719 (DF_REF_NEXT_LOC, DF_MWS_NEXT): New macros.
720 (FOR_EACH_INSN_INFO_DEF, FOR_EACH_INSN_INFO_USE)
721 (FOR_EACH_INSN_INFO_EQ_USE, FOR_EACH_INSN_INFO_MW)
722 (FOR_EACH_ARTIFICIAL_USE, FOR_EACH_ARTIFICIAL_DEF)
723 (df_get_artificial_defs, df_get_artificial_uses)
724 (df_single_def, df_single_use): Update accordingly.
725 (df_refs_chain_dump): Take the first element in a linked list as
726 parameter, rather than a pointer to an array of pointers.
727 * df-core.c (df_refs_chain_dump, df_mws_dump): Likewise.
728 * df-problems.c (df_rd_bb_local_compute_process_def): Likewise.
729 (df_chain_create_bb_process_use): Likewise.
730 (df_md_bb_local_compute_process_def): Likewise.
731 * fwprop.c (process_defs, process_uses): Likewise.
732 (register_active_defs, update_uses): Likewise.
733 (forward_propagate_asm): Update for new df_ref linking.
734 * df-scan.c (df_scan_free_ref_vec, df_scan_free_mws_vec): Delete.
735 (df_null_ref_rec, df_null_mw_rec): Likewise.
736 (df_scan_free_internal): Don't free df_ref and df_mw_hardreg lists
738 (df_scan_free_bb_info): Remove check for null artificial_defs.
739 (df_install_ref_incremental): Adjust for new df_ref linking.
740 Use a single-element insertion rather than a full sort.
741 (df_ref_chain_delete_du_chain): Take the first element
742 in a linked list as parameter, rather than a pointer to an array of
744 (df_ref_chain_delete, df_mw_hardreg_chain_delete): Likewise.
745 (df_add_refs_to_table, df_refs_verify, df_mws_verify): Likewise.
746 (df_insn_info_delete): Remove check for null defs and call to
747 df_scan_free_mws_vec.
748 (df_insn_rescan): Initialize df_ref and df_mw_hardreg lists to
749 null rather than df_null_*_rec.
750 (df_insn_rescan_debug_internal): Likewise, and update null
751 checks in the same way. Remove check for null defs.
752 (df_ref_change_reg_with_loc_1): Fix choice of list for defs.
753 Move a single element rather doing a full sort.
754 (df_mw_hardreg_chain_delete_eq_uses): Adjust for new df_mw_hardreg
756 (df_notes_rescan): Likewise. Use a merge rather than a full sort.
757 Initialize df_ref and df_mw_hardreg lists to null rather than
759 (df_ref_compare): Take df_refs as parameter, transferring the
761 (df_ref_ptr_compare): ...this new function.
762 (df_sort_and_compress_refs): Update accordingly.
763 (df_mw_compare): Take df_mw_hardregs as parameter, transferring the
765 (df_mw_ptr_compare): ...this new function.
766 (df_sort_and_compress_mws): Update accordingly.
767 (df_install_refs, df_install_mws): Return a linked list rather than
768 an array of pointers.
769 (df_refs_add_to_chains): Assert that old lists are empty rather
771 (df_insn_refs_verify): Don't handle null defs speciailly.
772 * web.c (union_match_dups): Update for new df_ref linking.
774 2014-06-15 Richard Sandiford <rdsandiford@googlemail.com>
776 * df.h (df_ref_create, df_ref_remove): Delete.
777 * df-scan.c (df_ref_create, df_ref_compress_rec): Likewise.
778 (df_ref_remove): Likewise.
780 2014-06-15 Richard Sandiford <rdsandiford@googlemail.com>
782 * df.h (df_single_def, df_single_use): New functions.
783 * ira.c (find_moveable_pseudos): Use them.
785 2014-06-15 Richard Sandiford <rdsandiford@googlemail.com>
787 * df.h (FOR_EACH_INSN_INFO_MW): New macro.
788 * df-problems.c (df_note_bb_compute): Use it.
789 * regstat.c (regstat_bb_compute_ri): Likewise.
791 2014-06-15 Richard Sandiford <rdsandiford@googlemail.com>
793 * df.h (FOR_EACH_ARTIFICIAL_USE, FOR_EACH_ARTIFICIAL_DEF): New macros.
794 * cse.c (cse_extended_basic_block): Use them.
795 * dce.c (mark_artificial_use): Likewise.
796 * df-problems.c (df_rd_simulate_artificial_defs_at_top): Likewise.
797 (df_lr_bb_local_compute, df_live_bb_local_compute): Likewise.
798 (df_chain_remove_problem, df_chain_bb_dump): Likewise.
799 (df_word_lr_bb_local_compute, df_note_bb_compute): Likewise.
800 (df_simulate_initialize_backwards): Likewise.
801 (df_simulate_finalize_backwards): Likewise.
802 (df_simulate_initialize_forwards): Likewise.
803 (df_md_simulate_artificial_defs_at_top): Likewise.
804 * df-scan.c (df_reorganize_refs_by_reg_by_insn): Likewise.
805 * regrename.c (init_rename_info): Likewise.
806 * regstat.c (regstat_bb_compute_ri): Likewise.
807 (regstat_bb_compute_calls_crossed): Likewise.
809 2014-06-15 Richard Sandiford <rdsandiford@googlemail.com>
811 * df.h (DF_INSN_INFO_MWS, FOR_EACH_INSN_INFO_DEF): New macros.
812 (FOR_EACH_INSN_INFO_USE, FOR_EACH_INSN_INFO_EQ_USE): Likewise.
813 (FOR_EACH_INSN_DEF, FOR_EACH_INSN_USE, FOR_EACH_INSN_EQ_USE): Likewise.
814 * auto-inc-dec.c (find_inc, merge_in_block): Use them.
815 * combine.c (create_log_links): Likewise.
816 * compare-elim.c (find_flags_uses_in_insn): Likewise.
817 (try_eliminate_compare): Likewise.
818 * cprop.c (make_set_regs_unavailable, mark_oprs_set): Likewise.
819 * dce.c (deletable_insn_p, find_call_stack_args): Likewise.
820 (remove_reg_equal_equiv_notes_for_defs): Likewise.
821 (reset_unmarked_insns_debug_uses, mark_reg_dependencies): Likewise.
822 (word_dce_process_block, dce_process_block): Likewise.
823 * ddg.c (def_has_ccmode_p): Likewise.
824 * df-core.c (df_bb_regno_first_def_find): Likewise.
825 (df_bb_regno_last_def_find, df_find_def, df_find_use): Likewise.
826 * df-problems.c (df_rd_simulate_one_insn): Likewise.
827 (df_lr_bb_local_compute, df_live_bb_local_compute): Likewise.
828 (df_chain_remove_problem, df_chain_insn_top_dump): Likewise.
829 (df_chain_insn_bottom_dump, df_word_lr_bb_local_compute): Likewise.
830 (df_word_lr_simulate_defs, df_word_lr_simulate_uses): Likewise.
831 (df_remove_dead_eq_notes, df_note_bb_compute): Likewise.
832 (df_simulate_find_defs, df_simulate_find_uses): Likewise.
833 (df_simulate_find_noclobber_defs, df_simulate_defs): Likewise.
834 (df_simulate_uses, df_md_simulate_one_insn): Likewise.
835 * df-scan.c (df_reorganize_refs_by_reg_by_insn): Likewise.
836 * fwprop.c (local_ref_killed_between_p): Likewise.
837 (all_uses_available_at, free_load_extend): Likewise.
838 * gcse.c (update_bb_reg_pressure, calculate_bb_reg_pressure): Likewise.
839 * hw-doloop.c (scan_loop): Likewise.
840 * ifcvt.c (dead_or_predicable): Likewise.
841 * init-regs.c (initialize_uninitialized_regs): Likewise.
842 * ira-lives.c (mark_hard_reg_early_clobbers): Likewise.
843 (process_bb_node_lives): Likewise.
844 * ira.c (compute_regs_asm_clobbered, build_insn_chain): Likewise.
845 (find_moveable_pseudos): Likewise.
846 * loop-invariant.c (check_dependencies, record_uses): Likewise.
847 * recog.c (peep2_find_free_register): Likewise.
848 * ree.c (get_defs): Likewise.
849 * regstat.c (regstat_bb_compute_ri): Likewise.
850 (regstat_bb_compute_calls_crossed): Likewise.
851 * sched-deps.c (find_inc, find_mem): Likewise.
852 * sel-sched-ir.c (maybe_downgrade_id_to_use): Likewise.
853 (maybe_downgrade_id_to_use, setup_id_reg_sets): Likewise.
854 * shrink-wrap.c (requires_stack_frame_p): Likewise.
855 (prepare_shrink_wrap): Likewise.
856 * store-motion.c (compute_store_table, build_store_vectors): Likewise.
857 * web.c (union_defs, pass_web::execute): Likewise.
858 * config/i386/i386.c (increase_distance, insn_defines_reg): Likewise.
859 (insn_uses_reg_mem, ix86_ok_to_clobber_flags): Likewise.
861 2014-06-13 Vladimir Makarov <vmakarov@redhat.com>
863 * lra-assign.c (assign_by_spills): Add code to assign vector regs
864 to inheritance pseudos.
865 * config/i386/i386.c (ix86_spill_class): Add check on NO_REGS.
867 2014-06-13 Peter Bergner <bergner@vnet.ibm.com>
870 * config/rs6000/rs6000-builtin.def (BU_MISC_1): Delete.
871 (BU_MISC_2): Rename to ...
872 (BU_LDBL128_2): ... this.
873 * config/rs6000/rs6000.h (RS6000_BTM_LDBL128): New define.
874 (RS6000_BTM_COMMON): Add RS6000_BTM_LDBL128.
875 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Handle
877 (rs6000_invalid_builtin): Add long double 128-bit builtin support.
878 (rs6000_builtin_mask_names): Add RS6000_BTM_LDBL128.
879 * config/rs6000/rs6000.md (unpacktf_0): Remove define)expand.
880 (unpacktf_1): Likewise.
881 * doc/extend.texi (__builtin_longdouble_dw0): Remove documentation.
882 (__builtin_longdouble_dw1): Likewise.
883 * doc/sourcebuild.texi (longdouble128): Document.
885 2014-06-13 Jeff Law <law@redhat.com>
887 PR rtl-optimization/61094
888 PR rtl-optimization/61446
889 * ree.c (combine_reaching_defs): Get the mode for the copy from
890 the extension insn rather than the defining insn.
892 2014-06-13 Dehao Chen <dehao@google.com>
894 * dwarf2out.c (add_linkage_name): Emit more linkage name.
896 2014-06-13 Thomas Schwinge <thomas@codesourcery.com>
898 * doc/install.texi (--enable-linker-plugin-configure-flags)
899 (--enable-linker-plugin-flags): Document new flags.
901 2014-06-13 Martin Jambor <mjambor@suse.cz>
904 * ipa-devirt.c (possible_polymorphic_call_targets): Store NULL to
905 cache_token if returning early.
907 2014-06-13 Nick Clifton <nickc@redhat.com>
909 * config/rx/rx.h (JUMP_ALIGN): Return the log value if user
910 requested alignment is active.
911 (LABEL_ALIGN): Likewise.
912 (LOOP_ALIGN): Likewise.
914 2014-06-13 Richard Biener <rguenther@suse.de>
916 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
917 Rewrite to propagate the VN result into all uses where
918 possible and to remove stmts becoming dead because of that.
919 (eliminate): Generalize stmt removal handling, remove in
920 reverse dominator order to support proper debug stmt
921 generation. Update stmts before removing stmts.
922 * tree-ssa-propagate.c (propagate_tree_value): Remove bogus assert.
924 2014-06-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
926 PR tree-optimization/61375
927 * tree-ssa-math-opts.c (init_symbolic_number): Cancel optimization if
928 symbolic number cannot be represented in an uint64_t.
929 (find_bswap_or_nop_1): Likewise.
931 2014-06-12 Jan Hubicka <hubicka@ucw.cz>
933 * symtab.c (symtab_node::reset_section): New method.
934 * cgraph.c (cgraph_node_cannot_be_local_p_1): Accept non-local
936 * cgraph.h (reset_section): Declare.
937 * ipa-inline-analysis.c (do_estimate_growth): Check for comdat groups;
938 do not consider comdat locals.
939 * cgraphclones.c (set_new_clone_decl_and_node_flags): Get section
941 * ipa-visiblity.c (cgraph_externally_visible_p): Cleanup.
942 (update_visibility_by_resolution_info): Consider UNDEF; fix checking;
943 reset sections of symbols dragged out of the comdats.
944 (function_and_variable_visibility): Reset sections of
947 2014-06-12 Jan Hubicka <hubicka@ucw.cz>
949 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Reorg
950 to use symtab and decl_binds_to_current_def_p
951 * tree-vectorizer.c (increase_alignment): Increase alignment
952 of alias target, too.
954 2014-06-12 Jakub Jelinek <jakub@redhat.com>
957 * gimplify.c (struct gimplify_omp_ctx): Add distribute field.
958 (gimplify_adjust_omp_clauses): Don't or in GOVD_LASTPRIVATE
959 if outer combined construct is distribute.
960 (gimplify_omp_for): For OMP_DISTRIBUTE set
961 gimplify_omp_ctxp->distribute.
962 * omp-low.c (scan_sharing_clauses) <case OMP_CLAUSE_SHARED>: For
963 GIMPLE_OMP_TEAMS, if decl isn't global in outer context, record
964 mapping into decl map.
966 2014-06-12 Jason Merrill <jason@redhat.com>
968 * common.opt (fabi-version): Change default to 0.
970 2014-06-12 Jason Merrill <jason@redhat.com>
972 * toplev.c (process_options): Reject -fabi-version=1.
974 2014-06-12 Jeff Law <law@redhat.com>
976 PR tree-optimization/61009
977 * tree-ssa-threadedge.c (thread_through_normal_block): Correct return
978 value when we stop processing a block due to problematic PHIs.
980 2014-06-12 Alan Lawrence <alan.lawrence@arm.com>
982 * config/aarch64/arm_neon.h (vmlaq_n_f64, vmlsq_n_f64, vrsrtsq_f64,
983 vcge_p8, vcgeq_p8, vcgez_p8, vcgez_u8, vcgez_u16, vcgez_u32, vcgez_u64,
984 vcgezq_p8, vcgezq_u8, vcgezq_u16, vcgezq_u32, vcgezq_u64, vcgezd_u64,
985 vcgt_p8, vcgtq_p8, vcgtz_p8, vcgtz_u8, vcgtz_u16, vcgtz_u32, vcgtz_u64,
986 vcgtzq_p8, vcgtzq_u8, vcgtzq_u16, vcgtzq_u32, vcgtzq_u64, vcgtzd_u64,
987 vcle_p8, vcleq_p8, vclez_p8, vclez_u64, vclezq_p8, vclezd_u64, vclt_p8,
988 vcltq_p8, vcltz_p8, vcltzq_p8, vcltzd_u64): Remove functions as they
991 2014-06-10 Alan Lawrence <alan.lawrence@arm.com>
994 * config/aarch64/aarch64-modes.def: Add V1DFmode.
995 * config/aarch64/aarch64.c (aarch64_vector_mode_supported_p):
998 2014-06-12 Eric Botcazou <ebotcazou@adacore.com>
1000 * tree-core.h (DECL_NONALIASED): Use proper spelling in comment.
1002 2014-06-12 Georg-Johann Lay <avr@gjlay.de>
1005 * config/avr/avr.md (push<mode>1): Avoid (subreg(mem)) when
1006 loading from address spaces.
1008 2014-06-12 Martin Liska <mliska@suse.cz>
1011 * ipa-prop.c (ipa_make_edge_direct_to_target): Check that gimple call
1012 statement is reachable.
1014 2014-06-11 Jan Hubicka <hubicka@ucw.cz>
1016 * symtab.c (section_hash): New hash.
1017 (symtab_unregister_node): Clear section before freeing.
1018 (hash_section_hash_entry): New haser.
1019 (eq_sections): New function.
1020 (symtab_node::set_section_for_node): New method.
1021 (set_section_1): Update.
1022 (symtab_node::set_section): Take string instead of tree as parameter.
1023 (symtab_resolve_alias): Update.
1024 * cgraph.h (section_hash_entry_d): New structure.
1025 (section_hash_entry): New typedef.
1026 (cgraph_node): Change comdat_group_ to x_comdat_group,
1027 change section_ to x_section and turn into section_hash_entry;
1028 update accestors; put set_section_for_node offline.
1029 * tree.c (decl_section_name): Turn into string.
1030 (set_decl_section_name): Change parameter to be string.
1031 * tree.h (decl_section_name, set_decl_section_name): Update prototypes.
1032 * sdbout.c (sdbout_one_type): Update.
1033 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Update.
1034 * varasm.c (IN_NAMED_SECTION, get_named_section,
1035 resolve_unique_section, hot_function_section, get_named_text_section,
1036 USE_SELECT_SECTION_FOR_FUNCTIONS, default_function_rodata_section,
1037 make_decl_rtl, default_unique_section): Update.
1038 * config/c6x/c6x.c (c6x_in_small_data_p): Update.
1039 (c6x_elf_unique_section): Update.
1040 * config/nios2/nios2.c (nios2_in_small_data_p): Update.
1041 * config/pa/pa.c (pa_function_section): Update.
1042 * config/pa/pa.h (IN_NAMED_SECTION_P): Update.
1043 * config/ia64/ia64.c (ia64_in_small_data_p): Update.
1044 * config/arc/arc.c (arc_in_small_data_p): Update.
1045 * config/arm/unknown-elf.h (IN_NAMED_SECTION_P): Update.
1046 * config/mcore/mcore.c (mcore_unique_section): Update.
1047 * config/mips/mips.c (mips16_build_function_stub): Update.
1048 (mips16_build_call_stub): Update.
1049 (mips_function_rodata_section): Update.
1050 (mips_in_small_data_p): Update.
1051 * config/score/score.c (score_in_small_data_p): Update.
1052 * config/rx/rx.c (rx_in_small_data): Update.
1053 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Update.
1054 (rs6000_xcoff_asm_named_section): Update.
1055 (rs6000_xcoff_unique_section): Update.
1056 * config/frv/frv.c (frv_string_begins_with): Update.
1057 (frv_in_small_data_p): Update.
1058 * config/v850/v850.c (v850_encode_data_area): Update.
1059 * config/bfin/bfin.c (DECL_SECTION_NAME): Update.
1060 (bfin_handle_l1_data_attribute): Update.
1061 (bfin_handle_l2_attribute): Update.
1062 * config/mep/mep.c (mep_unique_section): Update.
1063 * config/microblaze/microblaze.c (microblaze_elf_in_small_data_p):
1065 * config/h8300/h8300.c (h8300_handle_eightbit_data_attribute): Update.
1066 (h8300_handle_tiny_data_attribute): Update.
1067 * config/m32r/m32r.c (m32r_in_small_data_p): Update.
1068 (m32r_in_small_data_p): Update.
1069 * config/alpha/alpha.c (alpha_in_small_data_p): Update.
1070 * config/i386/i386.c (ix86_in_large_data_p): Update.
1071 * config/i386/winnt.c (i386_pe_unique_section): Update.
1072 * config/darwin.c (darwin_function_section): Update.
1073 * config/lm32/lm32.c (lm32_in_small_data_p): Update.
1074 * tree-emutls.c (get_emutls_init_templ_addr): Update.
1075 (new_emutls_decl): Update.
1076 * lto-cgraph.c (lto_output_node, input_node, input_varpool_node,
1077 input_varpool_node): Update.
1078 (ead_string_cst): Turn to ...
1079 (read_string): ... this one.
1080 * dwarf2out.c (secname_for_decl): Update.
1081 * asan.c (asan_protect_global): Update.
1083 2014-06-11 DJ Delorie <dj@redhat.com>
1085 * config/rx/rx.h (FUNCTION_BOUNDARY): Adjust for RX100/200 4-byte
1087 * config/rx/rx.c (rx_option_override): Likewise.
1088 (rx_align_for_label): Likewise.
1090 * config/rx/rx.c (rx_max_skip_for_label): Don't skip anything if -Os.
1092 2014-06-11 Maciej W. Rozycki <macro@codesourcery.com>
1094 * config/mmix/mmix-protos.h (mmix_asm_output_source_line): Remove
1097 2014-06-11 Richard Sandiford <rdsandiford@googlemail.com>
1099 * common.md: New file.
1100 * doc/md.texi: Update description of generic, machine-independent
1102 * config/s390/constraints.md (e): Delete.
1103 * Makefile.in (md_file): Include common.md.
1104 * config/m32c/t-m32c (md_file): Likewise.
1105 * genpreds.c (general_mem): New array.
1106 (generic_constraint_letters): Remove constraints now defined by
1108 (add_constraint): Map TARGET_MEM_CONSTRAINT to general_mem.
1109 Allow the first character to be '<' or '>' as well.
1110 * genoutput.c (general_mem): New array.
1111 (indep_constraints): Remove constraints now defined by common.md.
1112 (note_constraint): Map TARGET_MEM_CONSTRAINT to general_mem.
1113 Remove special handling of 'm'.
1114 * ira-costs.c (record_reg_classes): Remove special handling of
1115 constraints now defined by common.md.
1116 * ira.c (ira_setup_alts, ira_get_dup_out_num): Likewise.
1117 * ira-lives.c (single_reg_class): Likewise.
1118 (ira_implicitly_set_insn_hard_regs): Likewise.
1119 * lra-constraints.c (reg_class_from_constraints): Likewise.
1120 (process_alt_operands, process_address, curr_insn_transform): Likewise.
1121 * postreload.c (reload_cse_simplify_operands): Likewise.
1122 * reload.c (push_secondary_reload, scratch_reload_class)
1123 (find_reloads, alternative_allows_const_pool_ref): Likewise.
1124 * reload1.c (maybe_fix_stack_asms): Likewise.
1125 * targhooks.c (default_secondary_reload): Likewise.
1126 * stmt.c (parse_output_constraint): Likewise.
1127 * recog.c (preprocess_constraints): Likewise.
1128 (constrain_operands, peep2_find_free_register): Likewise.
1129 (asm_operand_ok): Likewise, but add a comment saying why 'o'
1130 must be handled specially.
1132 2014-06-11 Richard Sandiford <rdsandiford@googlemail.com>
1134 * system.h (CONST_DOUBLE_OK_FOR_CONSTRAINT_P): Poison.
1135 * genpreds.c (have_const_dbl_constraints): Delete.
1136 (add_constraint): Don't set it.
1137 (write_tm_preds_h): Don't call CONST_DOUBLE_OK_FOR_CONSTRAINT_P.
1138 * ira-costs.c (record_reg_classes): Handle CONST_INT and CONST_DOUBLE
1139 constraints using the lookup_constraint logic.
1140 * ira-lives.c (single_reg_class): Likewise.
1141 * ira.c (ira_setup_alts): Likewise.
1142 * lra-constraints.c (process_alt_operands): Likewise.
1143 * recog.c (asm_operand_ok, constrain_operands): Likewise.
1144 * reload.c (find_reloads): Likewise.
1146 2014-06-11 Richard Sandiford <rdsandiford@googlemail.com>
1148 * genpreds.c (const_int_start, const_int_end): New variables.
1149 (choose_enum_order): Output CONST_INT constraints before memory
1151 (write_tm_preds_h): Always define insn_const_int_ok_for_constraint.
1153 * ira-costs.c (record_reg_classes): Handle CT_CONST_INT.
1154 * ira.c (ira_setup_alts): Likewise.
1155 * lra-constraints.c (process_alt_operands): Likewise.
1156 * recog.c (asm_operand_ok, preprocess_constraints): Likewise.
1157 * reload.c (find_reloads): Likewise.
1159 2014-06-11 Richard Sandiford <rdsandiford@googlemail.com>
1161 * recog.h (operand_alternative): Remove offmem_ok, nonffmem_ok,
1162 decmem_ok and incmem_ok. Reformat other bitfields for consistency.
1163 * recog.c (preprocess_constraints): Update accordingly.
1165 2014-06-11 Richard Sandiford <rdsandiford@googlemail.com>
1167 * system.h (REG_CLASS_FROM_CONSTRAINT): Poison.
1168 (REG_CLASS_FOR_CONSTRAINT, EXTRA_CONSTRAINT_STR): Likewise.
1169 (EXTRA_MEMORY_CONSTRAINT, EXTRA_ADDRESS_CONSTRAINT): Likewise.
1170 * genpreds.c (print_type_tree): New function.
1171 (write_tm_preds_h): Remove REG_CLASS_FROM_CONSTRAINT,
1172 REG_CLASS_FOR_CONSTRAINT, EXTRA_MEMORY_CONSTRAINT,
1173 EXTRA_ADDRESS_CONSTRAINT and EXTRA_CONSTRAINT_STR.
1174 Write out enum constraint_type and get_constraint_type.
1175 * lra-constraints.c (satisfies_memory_constraint_p): Take a
1176 constraint_num rather than a constraint string.
1177 (satisfies_address_constraint_p): Likewise.
1178 (reg_class_from_constraints): Avoid old constraint macros.
1179 (process_alt_operands, process_address_1): Likewise.
1180 (curr_insn_transform): Likewise.
1181 * ira-costs.c (record_reg_classes): Likewise.
1182 (record_operand_costs): Likewise.
1183 * ira-lives.c (single_reg_class): Likewise.
1184 (ira_implicitly_set_insn_hard_regs): Likewise.
1185 * ira.c (ira_setup_alts, ira_get_dup_out_num): Likewise.
1186 * postreload.c (reload_cse_simplify_operands): Likewise.
1187 * recog.c (asm_operand_ok, preprocess_constraints): Likewise.
1188 (constrain_operands, peep2_find_free_register): Likewise.
1189 * reload.c (push_secondary_reload, scratch_reload_class): Likewise.
1190 (find_reloads, alternative_allows_const_pool_ref): Likewise.
1191 * reload1.c (maybe_fix_stack_asms): Likewise.
1192 * stmt.c (parse_output_constraint, parse_input_constraint): Likewise.
1193 * targhooks.c (default_secondary_reload): Likewise.
1194 * config/m32c/m32c.c (m32c_matches_constraint_p): Avoid reference
1195 to EXTRA_CONSTRAINT_STR.
1196 * config/sparc/constraints.md (U): Likewise REG_CLASS_FROM_CONSTRAINT.
1198 2014-06-11 Richard Sandiford <rdsandiford@googlemail.com>
1200 * genpreds.c (write_constraint_satisfied_p_1): Replace with...
1201 (write_constraint_satisfied_p_array): ...this new function.
1202 (write_tm_preds_h): Replace write_constraint_satisfied_p_1 with
1204 (write_insn_preds_c): Update accordingly.
1206 2014-06-11 Richard Sandiford <rdsandiford@googlemail.com>
1208 * genpreds.c (write_lookup_constraint): Rename to...
1209 (write_lookup_constraint_1): ...this.
1210 (write_lookup_constraint_array): New function.
1211 (write_tm_preds_h): Define lookup_constraint as an inline function
1212 that uses write_lookup_constraint_array where possible.
1213 (write_insn_preds_c): Update for the changes above.
1215 2014-06-11 Richard Sandiford <rdsandiford@googlemail.com>
1217 * doc/md.texi (regclass_for_constraint): Rename to...
1218 (reg_class_for_constraint): ...this.
1219 * genpreds.c (num_constraints, enum_order, register_start)
1220 (register_end, satisfied_start, memory_start, memory_end)
1221 (address_start, address_end): New variables.
1222 (add_constraint): Count the number of constraints.
1223 (choose_enum_order): New function.
1224 (write_enum_constraint_num): Iterate over enum_order.
1225 (write_regclass_for_constraint): Rename to...
1226 (write_reg_class_for_constraint_1): ...this and update output
1228 (write_constraint_satisfied_p): Rename to...
1229 (write_constraint_satisfied_p_1): ...this and update output
1230 accordingly. Do nothing if all extra constraints are register
1232 (write_insn_extra_memory_constraint): Delete.
1233 (write_insn_extra_address_constraint): Delete.
1234 (write_range_function): New function.
1235 (write_tm_preds_h): Define constraint_satisfied_p and
1236 reg_class_for_constraint as inline functions that do a range check
1237 before calling the out-of-line function. Use write_range_function
1238 to implement insn_extra_{register,memory,address}_constraint,
1239 the first of which is new.
1240 (write_insn_preds_c): Update after above changes to write_* functions.
1241 (main): Call choose_enum_order.
1243 2014-06-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
1245 PR tree-optimization/61306
1246 * tree-ssa-math-opts.c (struct symbolic_number): Store type of
1247 expression instead of its size.
1248 (do_shift_rotate): Adapt to change in struct symbolic_number. Return
1249 false to prevent optimization when the result is unpredictable due to
1250 arithmetic right shift of signed type with highest byte is set.
1251 (verify_symbolic_number_p): Adapt to change in struct symbolic_number.
1252 (init_symbolic_number): Likewise.
1253 (find_bswap_or_nop_1): Likewise. Return NULL to prevent optimization
1254 when the result is unpredictable due to sign extension.
1256 2014-06-11 Terry Guo <terry.guo@arm.com>
1258 * config/arm/arm.md (*thumb1_adddi3): Move into new file thumb1.md.
1259 (*thumb1_addsi3): Ditto.
1260 (*thumb_subdi3): Ditto.
1261 (thumb1_subsi3_insn): Ditto.
1262 (*thumb_mulsi3): Ditto.
1263 (*thumb_mulsi3_v6): Ditto.
1264 (*thumb1_andsi3_insn): Ditto.
1265 (thumb1_bicsi3): Ditto.
1266 (*thumb1_iorsi3_insn): Ditto.
1267 (*thumb1_xorsi3_insn): Ditto.
1268 (*thumb1_ashlsi3): Ditto.
1269 (*thumb1_ashrsi3): Ditto.
1270 (*thumb1_lshrsi3): Ditto.
1271 (*thumb1_rotrsi3): Ditto.
1272 (*thumb1_negdi2): Ditto.
1273 (*thumb1_negsi2): Ditto.
1274 (*thumb1_abssi2): Ditto.
1275 (*thumb1_neg_abssi2): Ditto.
1276 (*thumb1_one_cmplsi2): Ditto.
1277 (*thumb1_zero_extendhisi2): Ditto.
1278 (*thumb1_zero_extendqisi2): Ditto.
1279 (*thumb1_zero_extendqisi2_v6): Ditto.
1280 (thumb1_extendhisi2): Ditto.
1281 (thumb1_extendqisi2): Ditto.
1282 (*thumb1_movdi_insn): Ditto.
1283 (*thumb1_movsi_insn): Ditto.
1284 (*thumb1_movhi_insn): Ditto.
1285 (thumb_movhi_clobber): Ditto.
1286 (*thumb1_movqi_insn): Ditto.
1287 (*thumb1_movhf): Ditto.
1288 (*thumb1_movsf_insn): Ditto.
1289 (*thumb_movdf_insn): Ditto.
1292 (cbranchqi4): Ditto.
1293 (cbranchsi4_insn): Ditto.
1294 (cbranchsi4_scratch): Ditto.
1295 (*negated_cbranchsi4): Ditto.
1296 (*tbit_cbranch): Ditto.
1297 (*tlobits_cbranch): Ditto.
1298 (*tstsi3_cbranch): Ditto.
1299 (*cbranchne_decr1): Ditto.
1300 (*addsi3_cbranch): Ditto.
1301 (*addsi3_cbranch_scratch): Ditto.
1302 (*thumb_cmpdi_zero): Ditto.
1303 (cstoresi_eq0_thumb1): Ditto.
1304 (cstoresi_ne0_thumb1): Ditto.
1305 (*cstoresi_eq0_thumb1_insn): Ditto.
1306 (*cstoresi_ne0_thumb1_insn): Ditto.
1307 (cstoresi_nltu_thumb1): Ditto.
1308 (cstoresi_ltu_thumb1): Ditto.
1309 (thumb1_addsi3_addgeu): Ditto.
1310 (*thumb_jump): Ditto.
1311 (*call_reg_thumb1_v5): Ditto.
1312 (*call_reg_thumb1): Ditto.
1313 (*call_value_reg_thumb1_v5): Ditto.
1314 (*call_value_reg_thumb1): Ditto.
1315 (*call_insn): Ditto.
1316 (*call_value_insn): Ditto.
1317 (thumb1_casesi_internal_pic): Ditto.
1318 (thumb1_casesi_dispatch): Ditto.
1319 (*thumb1_indirect_jump): Ditto.
1320 (prologue_thumb1_interwork): Ditto.
1321 (*epilogue_insns): Ditto.
1322 (consttable_1): Ditto.
1323 (consttable_2): Ditto.
1325 (*thumb1_tablejump): Ditto.
1326 (thumb_eh_return): Ditto.
1327 (define_peephole2): Two of them are thumb1 only and got moved into
1329 (define_split): Six of them are thumb1 only and got moved into new
1331 * config/arm/thumb1.md: New file comprised of above thumb1 only
1334 2014-06-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1336 * config.gcc (aarch64*-*-*): Add arm_acle.h to extra headers.
1337 * Makefile.in (TEXI_GCC_FILES): Add aarch64-acle-intrinsics.texi to
1339 * config/aarch64/aarch64-builtins.c (AARCH64_CRC32_BUILTINS): Define.
1340 (aarch64_crc_builtin_datum): New struct.
1341 (aarch64_crc_builtin_data): New.
1342 (aarch64_init_crc32_builtins): New function.
1343 (aarch64_init_builtins): Initialise CRC32 builtins when appropriate.
1344 (aarch64_crc32_expand_builtin): New.
1345 (aarch64_expand_builtin): Add CRC32 builtin expansion case.
1346 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define
1347 __ARM_FEATURE_CRC32 when appropriate.
1348 (TARGET_CRC32): Define.
1349 * config/aarch64/aarch64.md (UNSPEC_CRC32B, UNSPEC_CRC32H,
1350 UNSPEC_CRC32W, UNSPEC_CRC32X, UNSPEC_CRC32CB, UNSPEC_CRC32CH,
1351 UNSPEC_CRC32CW, UNSPEC_CRC32CX): New unspec values.
1352 (aarch64_<crc_variant>): New pattern.
1353 * config/aarch64/arm_acle.h: New file.
1354 * config/aarch64/iterators.md (CRC): New int iterator.
1355 (crc_variant, crc_mode): New int attributes.
1356 * doc/aarch64-acle-intrinsics.texi: New file.
1357 * doc/extend.texi (aarch64): Document aarch64 ACLE intrinsics.
1358 Include aarch64-acle-intrinsics.texi.
1360 2014-06-11 Evgeny Stupachenko <evstupac@gmail.com>
1362 * tree-vect-data-refs.c (vect_grouped_store_supported): New
1363 check for stores group of length 3.
1364 (vect_permute_store_chain): New permutations for stores group of
1366 * tree-vect-stmts.c (vect_model_store_cost): Change cost
1367 of vec_perm_shuffle for the new permutations.
1369 2014-06-11 Jan Hubicka <hubicka@ucw.cz>
1371 * ipa-visibility.c (function_and_variable_visibility): Disable virtual
1372 table rewriting temporarily on targets not supporting ONE_ONLY.
1374 2014-06-11 Richard Biener <rguenther@suse.de>
1378 2014-06-04 Richard Biener <rguenther@suse.de>
1380 * tree.h (may_be_aliased): Trust TREE_ADDRESSABLE from
1381 TREE_PUBLIC and DECL_EXTERNAL decls.
1383 2014-06-10 Jan Hubicka <hubicka@ucw.cz>
1385 * varasm.c (set_implicit_section): New function.
1386 (resolve_unique_section): Use it to set implicit section
1388 (get_named_text_section): Use symtab_get_node (decl)->implicit_section
1389 (default_function_section): Likewise.
1390 (decl_binds_to_current_def_p): Constify argument.
1391 * varasm.h (decl_binds_to_current_def_p): Update prototype.
1392 * asan.c (asan_protect_global): Use
1393 symtab_get_node (decl)->implicit_section.
1394 * symtab.c (dump_symtab_base): Dump implicit sections.
1395 (verify_symtab_base): Verify sanity of sectoins and comdats.
1396 (symtab_resolve_alias): Alias share the section of its target.
1397 (set_section_1): New function.
1398 (symtab_node::set_section): Move here, recurse to aliases.
1399 (verify_symtab): Check for duplicated symtab lists.
1400 * tree-core.h (implicit_section_name_p): Remove.
1401 * tree-vect-data-refs.c: Include varasm.h.
1402 (vect_can_force_dr_alignment_p): Fix conditional on when
1403 decl bints to current definition; use
1404 symtab_get_node (decl)->implicit_section.
1405 * cgraph.c (cgraph_make_node_local_1): Fix section set.
1406 * cgraph.h (struct symtab_node): Add implicit_section.
1407 (set_section): Rename to ...
1408 (set_section_for_node): ... this one.
1409 (set_section): Declare.
1410 * tree.h (DECL_HAS_IMPLICIT_SECTION_NAME_P): Remove.
1411 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
1412 input_overwrite_node, input_varpool_node): Stream implicit_section.
1413 * ipa.c (symtab_remove_unreachable_nodes): Do not check symtab before
1414 removal; it will fail in LTO.
1416 2014-06-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1418 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>):
1419 Change second alternative type to f_mcr.
1420 * config/aarch64/aarch64.md (*movsi_aarch64): Change 11th
1421 and 12th alternatives' types to f_mcr and f_mrc.
1422 (*movdi_aarch64): Same for 12th and 13th alternatives.
1423 (*movsf_aarch64): Change 9th alternatives' type to mov_reg.
1424 (aarch64_movtilow_tilow): Change type to fmov.
1426 2014-06-10 Jiong Wang <jiong.wang@arm.com>
1428 * config/aarch64/aarch64.c (aarch64_save_or_restore_fprs)
1429 (aarch64_save_or_restore_callee_save_registers): Fix layout.
1431 2014-06-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1433 * config/aarch64/aarch64-simd.md (aarch64_sqdmulh_lane<mode>):
1435 (aarch64_sqrdmulh_lane<mode>): Likewise.
1436 (aarch64_sq<r>dmulh_lane<mode>): Rename to...
1437 (aarch64_sq<r>dmulh_lane<mode>_internal): ...this.
1438 (aarch64_sqdmulh_laneq<mode>): New expander.
1439 (aarch64_sqrdmulh_laneq<mode>): Likewise.
1440 (aarch64_sq<r>dmulh_laneq<mode>): Rename to...
1441 (aarch64_sq<r>dmulh_laneq<mode>_internal): ...this.
1442 (aarch64_sqdmulh_lane<mode>): New expander.
1443 (aarch64_sqrdmulh_lane<mode>): Likewise.
1444 (aarch64_sq<r>dmulh_lane<mode>): Rename to...
1445 (aarch64_sq<r>dmulh_lane<mode>_internal): ...this.
1446 (aarch64_sqdmlal_lane<mode>): Add lane flip for big-endian.
1447 (aarch64_sqdmlal_laneq<mode>): Likewise.
1448 (aarch64_sqdmlsl_lane<mode>): Likewise.
1449 (aarch64_sqdmlsl_laneq<mode>): Likewise.
1450 (aarch64_sqdmlal2_lane<mode>): Likewise.
1451 (aarch64_sqdmlal2_laneq<mode>): Likewise.
1452 (aarch64_sqdmlsl2_lane<mode>): Likewise.
1453 (aarch64_sqdmlsl2_laneq<mode>): Likewise.
1454 (aarch64_sqdmull_lane<mode>): Likewise.
1455 (aarch64_sqdmull_laneq<mode>): Likewise.
1456 (aarch64_sqdmull2_lane<mode>): Likewise.
1457 (aarch64_sqdmull2_laneq<mode>): Likewise.
1459 2014-06-10 Richard Biener <rguenther@suse.de>
1461 PR tree-optimization/61438
1462 * tree-ssa-pre.c (eliminate_dom_walker): Add do_pre member.
1463 (eliminate_dom_walker::before_dom_children): Only try to inhibit
1464 insertion of IVs if running PRE.
1465 (eliminate): Adjust.
1466 (pass_pre::execute): Likewise.
1467 (pass_fre::execute): Likewise.
1469 2014-06-10 Richard Biener <rguenther@suse.de>
1472 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p):
1473 Do not use the main variant for the type comparison.
1474 (ncr_compar): Likewise.
1475 (nonoverlapping_component_refs_p): Likewise.
1477 2014-06-10 Marcus Shawcroft <marcus.shawcroft@arm.com>
1479 * config/aarch64/aarch64.c (aarch64_save_or_restore_fprs): Fix
1480 REG_CFA_RESTORE mode.
1482 2014-06-10 Evgeny Stupachenko <evstupac@gmail.com>
1484 * config/i386/i386.c (expand_vec_perm_pblendv): New.
1485 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Use
1486 expand_vec_perm_pblendv.
1488 2014-06-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1490 * doc/arm-acle-intrinsics.texi: Specify when CRC32 intrinsics are
1492 Simplify description of __crc32d and __crc32cd intrinsics.
1493 * doc/extend.texi (ARM ACLE Intrinsics): Remove comment about CRC32
1496 2014-06-10 Thomas Schwinge <thomas@codesourcery.com>
1499 * configure.ac: Use gcc_AC_CHECK_DECLS to check for strnlen prototype.
1500 * config.in: Regenerate.
1501 * configure: Likewise.
1503 2014-06-10 Jan Hubicka <hubicka@ucw.cz>
1505 * ipa-reference.c (is_proper_for_analysis): Exclude addressable
1507 (intersect_static_var_sets): Remove.
1508 (propagate): Do not prune local statics.
1510 2014-06-10 Jakub Jelinek <jakub@redhat.com>
1513 * omp-low.c (lower_rec_input_clauses) <case OMP_CLAUSE_LASTPRIVATE>:
1514 Set lastprivate_firstprivate even if omp_private_outer_ref
1515 langhook returns true.
1516 <case OMP_CLAUSE_REDUCTION>: When calling omp_clause_default_ctor
1517 langhook, call unshare_expr on new_var and call
1518 build_outer_var_ref to get the last argument.
1520 2014-06-10 Marek Polacek <polacek@redhat.com>
1523 * doc/extend.texi: Add cindex for transparent_union.
1525 2014-06-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1527 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Check return value of
1528 init_symbolic_number ().
1530 2014-05-18 John David Anglin <danglin@gcc.gnu.org>
1533 * emit-rtl.c (reset_all_used_flags): In a sequence, check that
1534 XVECEXP (pat, 0, i) is an INSN before calling reset_insn_used_flags.
1535 (verify_rtl_sharing): Likewise.
1537 2014-06-09 Marc Glisse <marc.glisse@inria.fr>
1540 * tree.c (build_qualified_type): Use a canonical type for
1543 2014-06-09 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1545 * config/arm/arm-modes.def: Remove XFmode.
1547 2014-06-09 Alan Lawrence <alan.lawrence@arm.com>
1550 * config/arm/arm_neon.h (vtrn_s8, vtrn_s16, vtrn_u8, vtrn_u16, vtrn_p8,
1551 vtrn_p16, vtrn_s32, vtrn_f32, vtrn_u32, vtrnq_s8, vtrnq_s16, vtrnq_s32,
1552 vtrnq_f32, vtrnq_u8, vtrnq_u16, vtrnq_u32, vtrnq_p8, vtrnq_p16,
1553 vzip_s8, vzip_s16, vzip_u8, vzip_u16, vzip_p8, vzip_p16, vzip_s32,
1554 vzip_f32, vzip_u32, vzipq_s8, vzipq_s16, vzipq_s32, vzipq_f32,
1555 vzipq_u8, vzipq_u16, vzipq_u32, vzipq_p8, vzipq_p16, vuzp_s8, vuzp_s16,
1556 vuzp_s32, vuzp_f32, vuzp_u8, vuzp_u16, vuzp_u32, vuzp_p8, vuzp_p16,
1557 vuzpq_s8, vuzpq_s16, vuzpq_s32, vuzpq_f32, vuzpq_u8, vuzpq_u16,
1558 vuzpq_u32, vuzpq_p8, vuzpq_p16): Correct mask for bigendian.
1560 2014-06-09 Jan Hubicka <hubicka@ucw.cz>
1562 * tree-core.h (tree_decl_with_vis): Remove section_name.
1564 2014-06-09 Kito Cheng <kito@0xlab.org>
1566 * ira.c (ira): Don't call init_caller_save if LRA enabled
1567 since LRA use its own infrastructure to handle that.
1569 2014-06-07 Jan Hubicka <hubicka@ucw.cz>
1571 * symtab.c (dump_symtab_base): Update dumping.
1572 (symtab_make_decl_local): Clear only DECL_COMDAT.
1573 * tree-vect-data-refs.c (Check that variable is static before
1574 tampering with sections.
1575 * cgraphclones.c (duplicate_thunk_for_node): Do not clear section name.
1576 (cgraph_create_virtual_clone): Likewise.
1577 * tree.c (decl_comdat_group, decl_comdat_group_id): Constify argument.
1578 (decl_section_name, set_decl_section_name): New accessors.
1579 (find_decls_types_r): Do not walk section name
1580 * tree.h (DECL_SECTION_NAME): Implement using decl_section_name.
1581 (decl_comdat_group, decl_comdat_group_id): Constify.
1582 (decl_section_name, set_decl_section_name): Update.
1583 * varpool.c (varpool_finalize_named_section_flags): Use get_section.
1584 * cgraph.c (cgraph_add_thunk): Reset node instead of rebuilding.
1585 (cgraph_make_node_local_1): Clear section and comdat group.
1586 * cgraph.h (set_comdat_group): Sanity check.
1587 (get_section, set_section): New.
1588 * ipa-comdats.c (ipa_comdats): Use get_section.
1589 * ipa.c (ipa_discover_readonly_nonaddressable_var): Likewise.
1590 * lto-streamer-out.c: Do not follow section names.
1591 * c-family/c-common.c (handle_section_attribute): Update.
1592 * lto-cgraph.c (lto_output_node): Output section.
1593 (lto_output_varpool_node): Likewise.
1594 (read_comdat_group): Rename to ...
1595 (read_identifier): ... this one.
1596 (read_string_cst): New function.
1597 (input_node, input_varpool_node): Input section names.
1598 * tree-emutls.c (get_emutls_init_templ_addr): Update.
1599 (new_emutls_decl): Update.
1600 (secname_for_decl): Check section names only of static vars.
1601 * config/mep/mep.c (mep_unique_section): Use set_decl_section_name.
1602 * config/i386/winnt.c (i386_pe_unique_section): Likewise.
1603 * config/i386/i386.c (x86_64_elf_unique_section): Likewise.
1604 * config/c6x/c6x.c (c6x_elf_unique_section): Likewise.
1605 * config/rs6000/rs6000.c (rs6000_xcoff_unique_section): Likewise.
1606 * config/mcore/mcore.c (mcore_unique_section): Likewise.
1607 * config/mips/mips.c (mips16_build_function_stub): Likewise.
1608 * config/v850/v850.c (v850_insert_attributes): Likewise.
1609 * config/h8300/h8300.c (h8300_handle_eightbit_data_attribute):
1611 (h8300_handle_tiny_data_attribute): Likewise.
1612 * config/bfin/bfin.c (bfin_handle_l1_text_attribute): Likewise.
1613 (bfin_handle_l2_attribute): Likewise.
1615 2014-06-07 Jan Hubicka <hubicka@ucw.cz>
1617 * symtab.c (symtab_nonoverwritable_alias): Copy VIRTUAL flag;
1618 remove static initializer.
1620 2014-06-07 Jan Hubicka <hubicka@ucw.cz>
1622 * varasm.c (use_blocks_for_decl_p): Check symbol table
1623 instead of alias attribute.
1624 (place_block_symbol): Recurse on aliases.
1626 2014-06-07 Jan Hubicka <hubicka@ucw.cz>
1628 * ipa-visibility.c: Include varasm.h
1629 (can_replace_by_local_alias): Ceck decl_binds_to_current_def_p.
1631 2014-06-07 Jan Hubicka <hubicka@ucw.cz>
1633 * cgraphunit.c (assemble_thunks_and_aliases): Expand thunks before
1636 2014-06-07 Steven Bosscher <steven@gcc.gnu.org>
1638 * gcse.c (can_assign_to_reg_without_clobbers_p): Do not let pointers
1639 from test_insn into GGC space escape via SET_SRC.
1641 2014-06-07 Eric Botcazou <ebotcazou@adacore.com>
1643 * tree-ssa-tail-merge.c (same_succ_hash): Hash the static chain of a
1644 call statement, if any.
1645 (gimple_equal_p) <GIMPLE_CALL>: Compare the static chain of the call
1646 statements, if any. Tidy up.
1648 2014-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1651 * config/rs6000/vsx.md (VSX_LE): Split VSX_D into 2 separate
1652 iterators, VSX_D that handles 64-bit types, and VSX_LE that
1653 handles swapping the two 64-bit double words on little endian
1654 systems. Include V1TImode and optionally TImode in VSX_LE so that
1655 these types are properly swapped. Change all of the insns and
1656 splits that do the 64-bit swaps to use VSX_LE.
1657 (vsx_le_perm_load_<mode>): Likewise.
1658 (vsx_le_perm_store_<mode>): Likewise.
1659 (splitters for little endian memory operations): Likewise.
1660 (vsx_xxpermdi2_le_<mode>): Likewise.
1661 (vsx_lxvd2x2_le_<mode>): Likewise.
1662 (vsx_stxvd2x2_le_<mode>): Likewise.
1664 2014-06-06 Uros Bizjak <ubizjak@gmail.com>
1667 * config/i386/i386.md (*floatunssi<mode>2_i387_with_xmm): New
1668 define_insn_and_split pattern, merged from *floatunssi<mode>2_1
1669 and corresponding splitters. Zero extend general register
1670 or memory input operand to XMM temporary. Enable for
1671 TARGET_SSE2 and TARGET_INTER_UNIT_MOVES_TO_VEC only.
1672 (floatunssi<mode>2): Update expander predicate.
1674 2014-06-06 Vladimir Makarov <vmakarov@redhat.com>
1676 PR rtl-optimization/61325
1677 * lra-constraints.c (process_address_1): Check scale equal to one
1678 to prevent transformation: base + scale * index => base + new_reg.
1680 2014-06-06 Richard Biener <rguenther@suse.de>
1682 PR tree-optimization/59299
1683 * tree-ssa-sink.c (all_immediate_uses_same_place): Work on
1685 (nearest_common_dominator_of_uses): Likewise.
1686 (statement_sink_location): Adjust. Support sinking loads.
1688 2014-06-06 Martin Jambor <mjambor@suse.cz>
1690 * ipa-prop.c (get_place_in_agg_contents_list): New function.
1691 (build_agg_jump_func_from_list): Likewise.
1692 (determine_known_aggregate_parts): Renamed to
1693 determine_locally_known_aggregate_parts. Moved some functionality
1694 to the two functions above, removed bound checks.
1696 2014-06-06 James Greenhalgh <james.greenhalgh@arm.com>
1698 * config/aarch64/aarch64-protos.h (aarch64_expand_movmem): New.
1699 * config/aarch64/aarch64.c (aarch64_move_pointer): New.
1700 (aarch64_progress_pointer): Likewise.
1701 (aarch64_copy_one_part_and_move_pointers): Likewise.
1702 (aarch64_expand_movmen): Likewise.
1703 * config/aarch64/aarch64.h (MOVE_RATIO): Set low.
1704 * config/aarch64/aarch64.md (movmem<mode>): New.
1706 2014-06-06 Bingfeng Mei <bmei@broadcom.com>
1708 * targhooks.c (default_add_stmt_cost): Call target specific
1709 hook instead of default one.
1711 2014-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
1713 * ChangeLog (2014-05-23): Fix ChangeLog entry to refer to target
1714 endianness instead of host endianness.
1715 * tree-ssa-math-opts.c (find_bswap_or_nop_1): Likewise in dumps and
1718 2014-06-06 Eric Botcazou <ebotcazou@adacore.com>
1721 * function.c (instantiate_decls): Process the saved static chain.
1722 (expand_function_start): If not optimizing, save the static chain
1724 * tree-nested.c (convert_all_function_calls): Always create the static
1725 chain for nested functions if not optimizing.
1727 2014-06-06 Eric Botcazou <ebotcazou@adacore.com>
1729 * tree-cfg.c (make_edges) <GIMPLE_RETURN>: Put a location on the edge.
1731 2014-06-06 Richard Biener <rguenther@suse.de>
1733 * cfgexpand.c (expand_gimple_cond): Remove check for current_loops.
1734 (construct_init_block): Likewise.
1735 (construct_exit_block): Likewise.
1736 (pass_expand::execute): Likewise.
1737 * graphite.c (graphite_transforms): Replace check for current_loops
1738 with a check for > 1 loops.
1739 (pass_graphite_transforms::execute): Adjust.
1740 * ipa-split.c (split_function): Remove check for current_loops.
1741 * omp-low.c (expand_parallel_call): Likewise.
1742 (expand_omp_for_init_counts): Likewise.
1743 (extract_omp_for_update_vars): Likewise.
1744 (expand_omp_for_generic): Likewise.
1745 (expand_omp_sections): Likewise.
1746 (expand_omp_target): Likewise.
1747 * tracer.c (tail_duplicate): Likewise.
1748 (pass_tracer::execute): Likewise.
1749 * trans-mem.c (expand_transaction): Likewise.
1750 * tree-complex.c (expand_complex_div_wide): Likewise.
1751 * tree-eh.c (lower_resx): Likewise.
1752 (cleanup_empty_eh_merge_phis): Likewise.
1753 * tree-predcom.c (run_tree_predictive_commoning): Replace check for
1754 current_loops with a check for > 1 loops.
1755 (pass_predcom::execute): Adjust.
1756 * tree-scalar-evolution.c (scev_reset): Remove check for current_loops.
1757 * tree-ssa-copy.c (copy_prop_visit_phi_node): Likewise.
1758 * tree-ssa-dom.c (pass_phi_only_cprop::execute): Likewise.
1759 * tree-ssa-tail-merge.c (tail_merge_optimize): Likewise.
1760 * tree-ssa-threadupdate.c (thread_through_all_blocks): Likewise.
1761 * tree-switch-conversion.c (process_switch): Likewise.
1762 * tree-tailcall.c (tree_optimize_tail_calls_1): Likewise.
1763 * tree-vrp.c (vrp_visit_phi_node): Likewise.
1764 (execute_vrp): Likewise.
1765 * ubsan.c (ubsan_expand_null_ifn): Likewise.
1767 2014-06-06 Eric Botcazou <ebotcazou@adacore.com>
1769 * rtl.h (insn_location): Declare.
1770 * cfgcleanup.c (try_forward_edges): Compare the locus of locations
1771 with UNKNOWN_LOCATION.
1772 * emit-rtl.c (insn_location): New function.
1773 * final.c (notice_source_line): Check that the instruction has a
1774 location before retrieving it and use insn_location.
1775 * modulo-sched.c (loop_single_full_bb_p): Likewise.
1776 * print-rtl.c (print_rtx): Likewise.
1778 2014-06-06 Richard Biener <rguenther@suse.de>
1780 * passes.def: Move 2nd VRP pass before phi-only-cprop.
1782 2014-06-06 Christian Bruel <christian.bruel@st.com>
1784 PR tree-optimization/43934
1785 * tree-ssa-loop-im.c (determine_max_movement): Add PHI def constant
1788 2014-06-06 Richard Sandiford <rdsandiford@googlemail.com>
1790 * ira-lives.c (single_reg_class): Add missing break. Explicitly
1791 return NO_REGS for extra address and memory constraints. Handle
1792 operands that match (or are equivalent to something that matches)
1793 extra constant constraints. Ignore other non-register operands.
1795 2014-06-06 Alan Modra <amodra@gmail.com>
1798 * doc/tm.texi.in (INCOMING_REG_PARM_STACK_SPACE): Document.
1799 * doc/tm.texi: Regenerate.
1800 * function.c (INCOMING_REG_PARM_STACK_SPACE): Provide default.
1801 Use throughout in place of REG_PARM_STACK_SPACE.
1802 * config/rs6000/rs6000.c (rs6000_reg_parm_stack_space): Add
1803 "incoming" param. Pass to rs6000_function_parms_need_stack.
1804 (rs6000_function_parms_need_stack): Add "incoming" param, ignore
1805 prototype_p when incoming. Use function decl when incoming
1806 to handle K&R style functions.
1807 * config/rs6000/rs6000.h (REG_PARM_STACK_SPACE): Adjust.
1808 (INCOMING_REG_PARM_STACK_SPACE): Define.
1810 2014-06-05 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1813 * cfgexpand.c (expand_debug_expr): Use address space of nested
1814 TREE_TYPE for ADDR_EXPR and MEM_REF.
1816 2014-06-05 Jeff Law <law@redhat.com>
1818 PR tree-optimization/61289
1819 * tree-ssa-threadedge.c (invalidate_equivalences): Remove SRC_MAP and
1820 DST_MAP parameters. Invalidate by walking all the SSA_NAME_VALUES
1821 looking for those which match LHS. All callers changed.
1822 (record_temporary_equivalences_from_phis): Remove SRC_MAP and DST_MAP
1823 parameters and code which manipulated them. All callers changed.
1824 (record_temporary_equivalences_from_stmts_at_dest): Remove SRC_MAP
1825 and DST_MAP parameters. Simplify invalidation code by just calling
1826 invalidate_equivalences. All callers changed.
1827 (thread_across_edge): Simplify now that we don't need to maintain
1828 the map of equivalences to invalidate.
1830 2014-06-05 Kai Tietz <ktietz@redhat.com>
1831 Richard Henderson <rth@redhat.com>
1834 * config/i386/predicates.md (memory_nox32_operand): Add memory_operand
1835 checking for !TARGET_X32.
1836 * config/i386/i386.md (UNSPEC_PEEPSIB): New unspec constant.
1837 (sibcall_intern): New define_insn, plus required peepholes.
1838 (sibcall_pop_intern): Likewise.
1839 (sibcall_value_intern): Likewise.
1840 (sibcall_value_pop_intern): Likewise.
1842 2014-06-05 Ilya Enkovich <ilya.enkovich@intel.com>
1844 * tree-inline.c (tree_function_versioning): Check DF info existence
1845 before accessing it.
1847 2014-06-05 Marcus Shawcroft <marcus.shawcroft@arm.com>
1849 * config/aarch64/aarch64.h (aarch64_frame): Add hard_fp_offset and
1851 * config/aarch64/aarch64.c (aarch64_layout_frame): Initialize
1852 aarch64_frame hard_fp_offset and frame_size.
1853 (aarch64_expand_prologue): Use aarch64_frame hard_fp_offset and
1854 frame_size; remove original_frame_size.
1855 (aarch64_expand_epilogue, aarch64_final_eh_return_addr): Likewise.
1856 (aarch64_initial_elimination_offset): Remove frame_size and
1857 offset. Use aarch64_frame frame_size.
1859 2014-06-05 Marcus Shawcroft <marcus.shawcroft@arm.com>
1860 Jiong Wang <jiong.wang@arm.com>
1861 Renlin <renlin.li@arm.com>
1863 * config/aarch64/aarch64.c (aarch64_layout_frame): Correct
1864 initialization of R30 offset. Update offset. Iterate core
1865 regisers upto X30. Remove X29, X30 specific code.
1867 2014-06-05 Marcus Shawcroft <marcus.shawcroft@arm.com>
1868 Jiong Wang <jiong.wang@arm.com>
1870 * config/aarch64/aarch64.c (SLOT_NOT_REQUIRED, SLOT_REQUIRED): Define.
1871 (aarch64_layout_frame): Use SLOT_NOT_REQUIRED and SLOT_REQUIRED.
1872 (aarch64_register_saved_on_entry): Adjust test.
1874 2014-06-05 Marcus Shawcroft <marcus.shawcroft@arm.com>
1876 * config/aarch64/aarch64.h (machine_function): Move
1877 saved_varargs_size from here...
1878 (aarch64_frame): ... to here.
1880 * config/aarch64/aarch64.c (aarch64_expand_prologue)
1881 (aarch64_expand_epilogue, aarch64_final_eh_return_addr)
1882 (aarch64_initial_elimination_offset)
1883 (aarch64_setup_incoming_varargs): Adjust location of
1886 2014-06-05 Marcus Shawcroft <marcus.shawcroft@arm.com>
1888 * config/aarch64/aarch64.c (aarch64_expand_prologue): Update stack
1891 2014-06-05 Jaydeep Patil <Jaydeep.Patil@imgtec.com>
1892 Prachi Godbole <Prachi.Godbole@imgtec.com>
1894 * config/mips/mips-cpus.def: Add definition for p5600. Updated
1895 mips32r5 entry to use PROCESSOR_P5600.
1896 * config/mips/mips-tables.opt: Regenerate.
1897 * config/mips/mips-protos.h (mips_fmadd_bypass): Add prototype.
1898 * config/mips/mips.c (mips_fmadd_bypass): New function.
1899 (mips_rtx_cost_data): Add costs for p5600.
1900 (mips_issue_rate): Add support for p5600.
1901 (mips_multipass_dfa_lookahead): Likewise.
1902 * config/mips/mips.h (TUNE_P5600): New define.
1903 (TUNE_MACC_CHAINS): Add TUNE_P5600.
1904 (MIPS_ISA_LEVEL_SPEC): Map -march=p5600 to -mips32r5.
1905 * config/mips/mips.md: Include p5600.md.
1906 (processor): Add p5600.
1907 * config/mips/p5600.md: New file.
1909 2014-06-05 Evgeny Stupachenko <evstupac@gmail.com>
1911 * config/i386/sse.md (*ssse3_palignr<mode>_perm): New.
1912 * config/i386/predicates.md (palignr_operand): New.
1913 Indicates if permutation is suitable for palignr instruction.
1915 2014-06-05 Yuri Rumyantsev <ysrumyan@gmail.com>
1917 PR tree-optimization/61319
1918 * tree-if-conv.c (is_cond_scalar_reduction): Add missed check that
1919 stmt belongs to loop.
1921 2014-06-05 Richard Biener <rguenther@suse.de>
1923 * gimplify.c (create_tmp_from_val): Remove is_formal parameter
1924 and set DECL_GIMPLE_REG_P unconditionally if appropriate.
1925 (lookup_tmp_var): Adjust.
1926 (prepare_gimple_addressable): Unset DECL_GIMPLE_REG_P here.
1928 2014-06-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1930 * config/arm/arm.md (enabled): Disable opt_enabled attribute.
1932 2014-06-05 Marek Polacek <polacek@redhat.com>
1935 * doc/invoke.texi: Document -Wlogical-not-parentheses.
1937 2014-06-04 Tom de Vries <tom@codesourcery.com>
1939 * config/s390/s390.md ("addptrdi3", "addptrsi3"): Use INTVAL only on
1942 2014-06-04 Marc Glisse <marc.glisse@inria.fr>
1944 PR tree-optimization/61385
1945 * tree-ssa-phiopt.c (value_replacement): Punt if there are PHI nodes.
1947 2014-06-04 Bernd Schmidt <bernds@codesourcery.com>
1949 * lto-wrapper.c (fatal, fatal_perror): Remove functions. All callers
1950 changed to use fatal_error.
1951 (main): Ensure lto_wrapper_cleanup is run atexit.
1953 2014-06-04 Richard Sandiford <rdsandiford@googlemail.com>
1955 * lra-constraints.c (valid_address_p): Move earlier in file.
1956 (address_eliminator): New structure.
1957 (satisfies_memory_constraint_p): New function.
1958 (satisfies_address_constraint_p): Likewise.
1959 (process_alt_operands, process_address, curr_insn_transform): Use them.
1961 2014-06-04 Richard Sandiford <rdsandiford@googlemail.com>
1963 * lra-int.h (lra_static_insn_data): Make operand_alternative a
1965 (target_lra_int, default_target_lra_int, this_target_lra_int)
1966 (op_alt_data): Delete.
1967 * lra.h (lra_init): Delete.
1968 * lra.c (default_target_lra_int, this_target_lra_int): Delete.
1969 (init_insn_code_data_once): Remove op_alt_data handling.
1970 (finish_insn_code_data_once): Likewise.
1971 (init_op_alt_data): Delete.
1972 (get_static_insn_data): Initialize operand_alternative to null.
1973 (free_insn_recog_data): Cast operand_alternative before freeing it.
1974 (setup_operand_alternative): Take the operand_alternative as
1975 parameter and assume it isn't already cached in the static
1977 (lra_set_insn_recog_data): Update accordingly.
1979 * ira.c (ira_init): Don't call lra_init.
1980 * target-globals.h (this_target_lra_int): Declare.
1981 (target_globals): Remove lra_int.
1982 (restore_target_globals): Update accordingly.
1983 * target-globals.c: Don't include lra-int.h.
1984 (default_target_globals, save_target_globals): Remove lra_int.
1986 2014-06-04 Richard Sandiford <rdsandiford@googlemail.com>
1988 * recog.h (operand_alternative): Convert reg_class, reject,
1989 matched and matches into bitfields.
1990 (preprocess_constraints): New overload.
1991 (preprocess_insn_constraints): New function.
1992 (preprocess_constraints): Take the insn as parameter.
1993 (recog_op_alt): Change into a pointer.
1994 (target_recog): Add x_op_alt.
1995 * recog.c (asm_op_alt): New variable.
1996 (recog_op_alt): Change into a pointer.
1997 (preprocess_constraints): New overload, replacing the old function
1998 definition with one that doesn't use global state.
1999 (preprocess_insn_constraints): New function.
2000 (preprocess_constraints): Use them. Take the insn as parameter.
2001 Use asm_op_alt for asms.
2002 (recog_init): Free existing x_op_alt entries.
2003 * ira-lives.c (check_and_make_def_conflict): Make operand_alternative
2005 (make_early_clobber_and_input_conflicts): Likewise.
2006 (process_bb_node_lives): Pass the insn to process_constraints.
2007 * reg-stack.c (check_asm_stack_operands): Likewise.
2008 (subst_asm_stack_regs): Likewise.
2009 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2010 * regrename.c (build_def_use): Likewise.
2011 * sched-deps.c (sched_analyze_insn): Likewise.
2012 * sel-sched.c (get_reg_class, implicit_clobber_conflict_p): Likewise.
2013 * config/arm/arm.c (xscale_sched_adjust_cost): Likewise.
2014 (note_invalid_constants): Likewise.
2015 * config/i386/i386.c (ix86_legitimate_combined_insn): Likewise.
2016 (ix86_legitimate_combined_insn): Make operand_alternative pointer
2019 2014-06-04 Richard Sandiford <rdsandiford@googlemail.com>
2021 * recog.c (preprocess_constraints): Don't skip disabled alternatives.
2022 * ira-lives.c (check_and_make_def_conflict): Check for disabled
2024 (make_early_clobber_and_input_conflicts): Likewise.
2025 * config/i386/i386.c (ix86_legitimate_combined_insn): Likewise.
2027 2014-06-04 Richard Sandiford <rdsandiford@googlemail.com>
2029 * recog.h (alternative_class): New function.
2030 (which_op_alt): Return a const recog_op_alt.
2031 * reg-stack.c (check_asm_stack_operands): Update type accordingly.
2032 (subst_asm_stack_regs): Likewise.
2033 * config/arm/arm.c (note_invalid_constants): Likewise.
2034 * regcprop.c (copyprop_hardreg_forward_1): Likewise. Don't modify
2035 the operand_alternative; use alternative class instead.
2036 * sel-sched.c (get_reg_class): Likewise.
2037 * regrename.c (build_def_use): Likewise.
2038 (hide_operands, restore_operands, record_out_operands): Update type
2041 2014-06-04 Richard Sandiford <rdsandiford@googlemail.com>
2043 * recog.h (recog_op_alt): Convert to a flat array.
2044 (which_op_alt): New function.
2045 * recog.c (recog_op_alt): Convert to a flat array.
2046 (preprocess_constraints): Update accordingly, grouping all
2047 operands of the same alternative together, rather than the
2049 * ira-lives.c (check_and_make_def_conflict): Likewise.
2050 (make_early_clobber_and_input_conflicts): Likewise.
2051 * config/i386/i386.c (ix86_legitimate_combined_insn): Likewise.
2052 * reg-stack.c (check_asm_stack_operands): Use which_op_alt.
2053 (subst_asm_stack_regs): Likewise.
2054 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2055 * regrename.c (hide_operands, record_out_operands): Likewise.
2056 (build_def_use): Likewise.
2057 * sel-sched.c (get_reg_class): Likewise.
2058 * config/arm/arm.c (note_invalid_constants): Likewise.
2060 2014-06-04 Jason Merrill <jason@redhat.com>
2064 * gimplify.c (gimplify_arg): Non-static.
2065 * gimplify.h: Declare it.
2067 2014-06-04 Richard Biener <rguenther@suse.de>
2069 * tree.h (may_be_aliased): Trust TREE_ADDRESSABLE from
2070 TREE_PUBLIC and DECL_EXTERNAL decls.
2072 2014-06-04 Matthew Fortune <matthew.fortune@imgtec.com>
2074 * regcprop.c (copyprop_hardreg_forward_1): Account for
2075 HARD_REGNO_CALL_PART_CLOBBERED.
2077 2014-06-04 Richard Biener <rguenther@suse.de>
2079 * configure.ac: Check whether the underlying type of int64_t
2080 is long or long long.
2081 * configure: Regenerate.
2082 * config.in: Likewise.
2083 * hwint.h (HOST_WIDE_INT): Match the underlying type of int64_t.
2084 (HOST_WIDE_INT_PRINT_*): Define in terms of PRI*64.
2086 2014-06-04 Richard Biener <rguenther@suse.de>
2088 PR tree-optimization/60098
2089 * tree-ssa-dse.c (dse_possible_dead_store_p): Walk until
2091 (dse_optimize_stmt): Simplify, now that we found a kill
2094 2014-06-04 Richard Biener <rguenther@suse.de>
2096 * tree-ssa-alias.c (stmt_may_clobber_ref_p): Improve handling
2097 of accesses with non-invariant address.
2099 2014-06-04 Martin Liska <mliska@suse.cz>
2101 * cgraph.h (cgraph_make_wrapper): New function introduced.
2102 * cgraphunit.c (cgraph_make_wrapper): The function implementation.
2103 * ipa-inline.h (inline_analyze_function): The function is global.
2104 * ipa-inline-analysis.c (inline_analyze_function): Likewise.
2106 2014-06-04 Martin Liska <mliska@suse.cz>
2108 * tree.h (private_lookup_attribute_starting): New function.
2109 (lookup_attribute_starting): Likewise.
2110 * tree.c (private_lookup_attribute_starting): Likewise.
2112 2014-06-04 Martin Liska <mliska@suse.cz>
2114 * cgraph.h (expand_thunk): New argument added.
2115 (address_taken_from_non_vtable_p): New global function.
2116 * ipa-visibility.c (address_taken_from_non_vtable_p): Likewise.
2117 * cgraphclones.c (duplicate_thunk_for_node): Argument added to call.
2118 * cgraphunit.c (analyze_function): Likewise.
2119 (assemble_thunks_and_aliases): Argument added to call.
2120 (expand_thunk): New argument forces to produce GIMPLE thunk.
2122 2014-06-04 Martin Liska <mliska@suse.cz>
2124 * coverage.h (coverage_compute_cfg_checksum): Argument added.
2125 * coverage.c (coverage_compute_cfg_checksum): Likewise.
2126 * profile.c (branch_prob): Likewise.
2128 2014-06-04 Martin Jambor <mjambor@suse.cz>
2131 * ipa-pure-const.c (propagate_pure_const): Add unreachable default
2132 handler for switch on an ipa_ref_use enum.
2133 * ipa-reference.c (analyze_function): Likewise.
2135 2014-06-04 Kai Tietz <ktietz@redhat.com>
2137 * recog.c (peep2_attempt): Copy SIBLING_CALL_P flag
2138 from old call-instruction.
2140 2014-06-04 Bin Cheng <bin.cheng@arm.com>
2142 * config/aarch64/aarch64.c (aarch64_classify_address)
2143 (aarch64_legitimize_reload_address): Support full addressing modes
2145 * config/aarch64/aarch64.md (mov<mode>, movmisalign<mode>)
2146 (*aarch64_simd_mov<mode>, *aarch64_simd_mov<mode>): Relax predicates.
2148 2014-06-03 Andrew Pinski <apinski@cavium.com>
2150 * config/aarch64/aarch64.c (aarch64_if_then_else_costs): Allow non comparisons
2153 2014-06-03 Andrew Pinski <apinski@cavium.com>
2155 * config/aarch64/aarch64.c (aarch64_if_then_else_costs): New function.
2156 (aarch64_rtx_costs): Use aarch64_if_then_else_costs.
2158 2014-06-03 Kai Tietz <ktietz@redhat.com>
2160 * config/i386/i386.c (ix86_function_value_regno_p): Disallow DX_REG
2163 2014-06-03 Dehao Chen <dehao@google.com>
2165 * tree-cfg.c (gimple_merge_blocks): Only reset count when BBs are in
2168 2014-06-03 Marek Polacek <polacek@redhat.com>
2171 * doc/invoke.texi: Document -Wswitch-bool.
2172 * function.c (stack_protect_epilogue): Cast controlling expression of
2174 * gengtype.c (walk_type): Generate switch expression with its
2175 controlling expression cast to int.
2177 2014-06-03 Vishnu K S <Vishnu.k_s@atmel.com>
2179 * config/avr/avr-mcus.def: Add new avr25 devices attiny441, attiny828
2181 * config/avr/avr-tables.opt: Regenerate.
2182 * config/avr/t-multilib: Regenerate.
2183 * doc/avr-mmcu.texi: Regenerate.
2185 2014-06-03 Vishnu K S <vishnu.k_s@atmel.com>
2186 Pitchumani Sivanupandi <pitchumani.s@atmel.com>
2188 * config/avr/avr-mcus.def (ata6616c): Add new avr25 device.
2189 (ata6617c, ata664251): Add new avr35 devices.
2190 (ata6612c): Add new avr4 device.
2191 (ata6613c, ata6614q): Add new avr5 devices.
2192 * config/avr/avr-tables.opt: Regenerate.
2193 * config/avr/t-multilib: Regenerate.
2194 * doc/avr-mmcu.texi: Regenerate.
2196 2014-06-03 Alan Lawrence <alan.lawrence@arm.com>
2198 * gcc/config/aarch64/aarch64-builtins.c
2199 (aarch64_types_binop_ssu_qualifiers): New static data.
2200 (TYPES_BINOP_SSU): Define.
2201 * gcc/config/aarch64/aarch64-simd-builtins.def (suqadd, ushl, urshl,
2202 urshr_n, ushll_n): Use appropriate unsigned qualifiers.
2203 * gcc/config/aarch64/arm_neon.h (vrshl_u8, vrshl_u16, vrshl_u32,
2204 vrshl_u64, vrshlq_u8, vrshlq_u16, vrshlq_u32, vrshlq_u64, vrshld_u64,
2205 vrshr_n_u8, vrshr_n_u16, vrshr_n_u32, vrshr_n_u64, vrshrq_n_u8,
2206 vrshrq_n_u16, vrshrq_n_u32, vrshrq_n_u64, vrshrd_n_u64, vshll_n_u8,
2207 vshll_n_u16, vshll_n_u32, vuqadd_s8, vuqadd_s16, vuqadd_s32,
2208 vuqadd_s64, vuqaddq_s8, vuqaddq_s16, vuqaddq_s32, vuqaddq_s64,
2209 vuqaddb_s8, vuqaddh_s16, vuqadds_s32, vuqaddd_s64): Add signedness
2210 suffix to builtin function name, remove cast.
2211 (vshl_s8, vshl_s16, vshl_s32, vshl_s64, vshl_u8, vshl_u16, vshl_u32,
2212 vshl_u64, vshlq_s8, vshlq_s16, vshlq_s32, vshlq_s64, vshlq_u8,
2213 vshlq_u16, vshlq_u32, vshlq_u64, vshld_s64, vshld_u64): Remove cast.
2215 2014-06-03 Alan Lawrence <alan.lawrence@arm.com>
2217 * gcc/config/aarch64/aarch64-builtins.c
2218 (aarch64_types_binop_uus_qualifiers,
2219 aarch64_types_shift_to_unsigned_qualifiers,
2220 aarch64_types_unsigned_shiftacc_qualifiers): Define.
2221 * gcc/config/aarch64/aarch64-simd-builtins.def (uqshl, uqrshl, uqadd,
2222 uqsub, usqadd, usra_n, ursra_n, uqshrn_n, uqrshrn_n, usri_n, usli_n,
2223 sqshlu_n, uqshl_n): Update qualifiers.
2224 * gcc/config/aarch64/arm_neon.h (vqadd_u8, vqadd_u16, vqadd_u32,
2225 vqadd_u64, vqaddq_u8, vqaddq_u16, vqaddq_u32, vqaddq_u64, vqsub_u8,
2226 vqsub_u16, vqsub_u32, vqsub_u64, vqsubq_u8, vqsubq_u16, vqsubq_u32,
2227 vqsubq_u64, vqaddb_u8, vqaddh_u16, vqadds_u32, vqaddd_u64, vqrshl_u8,
2228 vqrshl_u16, vqrshl_u32, vqrshl_u64, vqrshlq_u8, vqrshlq_u16,
2229 vqrshlq_u32, vqrshlq_u64, vqrshlb_u8, vqrshlh_u16, vqrshls_u32,
2230 vqrshld_u64, vqrshrn_n_u16, vqrshrn_n_u32, vqrshrn_n_u64,
2231 vqrshrnh_n_u16, vqrshrns_n_u32, vqrshrnd_n_u64, vqshl_u8, vqshl_u16,
2232 vqshl_u32, vqshl_u64, vqshlq_u8, vqshlq_u16, vqshlq_u32, vqshlq_u64,
2233 vqshlb_u8, vqshlh_u16, vqshls_u32, vqshld_u64, vqshl_n_u8, vqshl_n_u16,
2234 vqshl_n_u32, vqshl_n_u64, vqshlq_n_u8, vqshlq_n_u16, vqshlq_n_u32,
2235 vqshlq_n_u64, vqshlb_n_u8, vqshlh_n_u16, vqshls_n_u32, vqshld_n_u64,
2236 vqshlu_n_s8, vqshlu_n_s16, vqshlu_n_s32, vqshlu_n_s64, vqshluq_n_s8,
2237 vqshluq_n_s16, vqshluq_n_s32, vqshluq_n_s64, vqshlub_n_s8,
2238 vqshluh_n_s16, vqshlus_n_s32, vqshlud_n_s64, vqshrn_n_u16,
2239 vqshrn_n_u32, vqshrn_n_u64, vqshrnh_n_u16, vqshrns_n_u32,
2240 vqshrnd_n_u64, vqsubb_u8, vqsubh_u16, vqsubs_u32, vqsubd_u64,
2241 vrsra_n_u8, vrsra_n_u16, vrsra_n_u32, vrsra_n_u64, vrsraq_n_u8,
2242 vrsraq_n_u16, vrsraq_n_u32, vrsraq_n_u64, vrsrad_n_u64, vsli_n_u8,
2243 vsli_n_u16, vsli_n_u32,vsli_n_u64, vsliq_n_u8, vsliq_n_u16,
2244 vsliq_n_u32, vsliq_n_u64, vslid_n_u64, vsqadd_u8, vsqadd_u16,
2245 vsqadd_u32, vsqadd_u64, vsqaddq_u8, vsqaddq_u16, vsqaddq_u32,
2246 vsqaddq_u64, vsqaddb_u8, vsqaddh_u16, vsqadds_u32, vsqaddd_u64,
2247 vsra_n_u8, vsra_n_u16, vsra_n_u32, vsra_n_u64, vsraq_n_u8,
2248 vsraq_n_u16, vsraq_n_u32, vsraq_n_u64, vsrad_n_u64, vsri_n_u8,
2249 vsri_n_u16, vsri_n_u32, vsri_n_u64, vsriq_n_u8, vsriq_n_u16,
2250 vsriq_n_u32, vsriq_n_u64, vsrid_n_u64): Remove casts.
2252 2014-06-03 Teresa Johnson <tejohnson@google.com>
2254 * tree-sra.c (modify_function): Record caller nodes after rebuild.
2256 2014-06-02 Jason Merrill <jason@redhat.com>
2259 * varpool.c (ctor_for_folding): Handle uninitialized vtables.
2261 2014-06-03 Alan Lawrence <alan.lawrence@arm.com>
2263 * config/aarch64/aarch64.c (aarch64_evpc_ext): allow and handle
2266 2014-06-03 Alan Lawrence <alan.lawrence@arm.com>
2268 * config/aarch64/aarch64-simd.md (aarch64_rev<REVERSE:rev-op><mode>):
2270 * config/aarch64/aarch64.c (aarch64_evpc_rev): New function.
2271 (aarch64_expand_vec_perm_const_1): Add call to aarch64_evpc_rev.
2272 * config/aarch64/iterators.md (REVERSE): New iterator.
2273 (UNSPEC_REV64, UNSPEC_REV32, UNSPEC_REV16): New enum elements.
2274 (rev_op): New int_attribute.
2275 * config/aarch64/arm_neon.h (vrev16_p8, vrev16_s8, vrev16_u8,
2276 vrev16q_p8, vrev16q_s8, vrev16q_u8, vrev32_p8, vrev32_p16, vrev32_s8,
2277 vrev32_s16, vrev32_u8, vrev32_u16, vrev32q_p8, vrev32q_p16, vrev32q_s8,
2278 vrev32q_s16, vrev32q_u8, vrev32q_u16, vrev64_f32, vrev64_p8,
2279 vrev64_p16, vrev64_s8, vrev64_s16, vrev64_s32, vrev64_u8, vrev64_u16,
2280 vrev64_u32, vrev64q_f32, vrev64q_p8, vrev64q_p16, vrev64q_s8,
2281 vrev64q_s16, vrev64q_s32, vrev64q_u8, vrev64q_u16, vrev64q_u32):
2282 Replace temporary __asm__ with __builtin_shuffle.
2284 2014-06-03 Andrew Bennett <andrew.bennett@imgtec.com>
2286 * config/mips/mips-cpus.def: Add mips32r3, mips32r5, mips64r3 and
2288 * config/mips/mips-tables.opt: Regenerate.
2289 * config/mips/mips.c (mips_compute_frame_info): Changed if statement
2290 to use mips_isa_rev rather than ISA_MIPS32R2.
2291 * config/mips/mips.h (ISA_MIPS32R3): New define.
2292 (ISA_MIPS32R5): New define.
2293 (ISA_MIPS64R3): New define.
2294 (ISA_MIPS64R5): New define.
2295 (TARGET_CPU_CPP_BUILTINS): Added support for ISA_MIPS32R3,
2296 ISA_MIPS32R5, ISA_MIPS64R3 and ISA_MIPS64R5.
2297 (MIPS_ISA_LEVEL_SPEC): Added support for mips32r3, mips32r5, mips64r3
2299 (MIPS_ISA_SYNCI_SPEC): Likewise.
2300 (ISA_HAS_64BIT_REGS): Added ISA_MIPS64R3 and ISA_MIPS64R5.
2301 (LINK_SPEC): Added mips32r3 and mips32r5.
2302 * config/mips/t-isa3264 (MULTILIB_MATCHES): Map mips32r3 and mips32r5
2303 to mips32r2; and mips64r3 and mips64r5 to mips64r2.
2304 * config/mips/t-mti-elf (MULTILIB_MATCHES): Likewise.
2305 * config/mips/t-mti-linux (MULTILIB_MATCHES): Likewise.
2306 * config/mips/t-sde (MULTILIB_MATCHES): Likewise.
2307 * config/mips/t-sdemtk (MULTILIB_MATCHES): New define.
2308 * doc/invoke.texi: Document mips32r3, mips32r5, mips64r3 and mips64r5.
2310 2014-06-03 Andrew Bennett <andrew.bennett@imgtec.com>
2312 * doc/invoke.texi: Document -mxpa and -mno-xpa MIPS command line
2314 * config/mips/mips.opt (mxpa): New option.
2315 * config/mips/mips.h (ASM_SPEC): Pass mxpa and mno-xpa to the
2318 2014-06-03 Martin Jambor <mjambor@suse.cz>
2321 * ipa-cp.c (cgraph_edge_brings_value_p): Handle edges leading to
2324 2014-06-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
2326 PR tree-optimization/61328
2327 * tree-ssa-math-opts.c (init_symbolic_number): Extract symbolic number
2328 initialization from find_bswap_or_nop_1.
2329 (find_bswap_or_nop_1): Test return value of find_bswap_or_nop_1 stored
2330 in source_expr2 before using the size value the function sets. Also
2331 make use of init_symbolic_number () in both the old place and
2332 find_bswap_or_nop_load () to avoid reading uninitialized memory when
2333 doing recursion in the GIMPLE_BINARY_RHS case.
2335 2014-06-03 Richard Biener <rguenther@suse.de>
2337 PR tree-optimization/61383
2338 * tree-ssa-ifcombine.c (bb_no_side_effects_p): Make sure
2341 2014-06-03 Richard Sandiford <rdsandiford@googlemail.com>
2343 * defaults.h (USE_MD_CONSTRAINTS, EXTRA_MEMORY_CONSTRAINT)
2344 (EXTRA_ADDRESS_CONSTRAINT, DEFAULT_CONSTRAINT_LEN, CONSTRAINT_LEN)
2345 (CONST_OK_FOR_CONSTRAINT_P, CONST_DOUBLE_OK_FOR_LETTER_P)
2346 (REG_CLASS_FROM_CONSTRAINT, EXTRA_CONSTRAINT_STR): Delete definitions
2348 (REG_CLASS_FROM_LETTER, CONST_OK_FOR_LETTER_P)
2349 (CONST_DOUBLE_OK_FOR_LETTER_P, EXTRA_CONSTRAINT): Move poising to...
2350 * system.h: ...here and make it unconditional.
2351 * target.def (conditional_register_usage): Mention
2352 define_register_constraint instead of old-style constraint macros.
2353 * doc/tm.texi.in: Remove documentation for old-style constraint macros.
2354 * doc/tm.texi: Regenerate.
2355 * genoutput.c: Remove USE_MD_CONSTRAINTS conditions and all code
2356 protected by !USE_MD_CONSTRAINTS.
2357 * config/frv/frv.md: Remove quote from old version of documentation.
2358 * config/frv/frv.c (frv_conditional_register_usage): Likewise.
2359 * config/m32r/m32r.c (easy_di_const, easy_df_const): Avoid mentioning
2360 CONST_DOUBLE_OK_FOR_LETTER.
2361 * config/sh/constraints.md: Likewise EXTRA_CONSTRAINT.
2363 2014-06-02 Andrew Pinski <apinski@cavium.com>
2365 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER):
2366 /lib/ld-linux32-aarch64.so.1 is used for ILP32.
2367 (LINUX_TARGET_LINK_SPEC): Update linker script for ILP32.
2368 file whose name depends on -mabi= and -mbig-endian.
2369 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES):
2370 Handle LP64 better and handle ilp32 too.
2371 (MULTILIB_OPTIONS): Delete.
2372 (MULTILIB_DIRNAMES): Delete.
2374 2014-06-02 Andrew MacLeod <amacleod@redhat.com>
2376 * expr.h: Remove prototypes of functions defined in builtins.c.
2377 * tree.h: (build_call_expr_*, build_string_literal): Add prototypes.
2378 Remove prototypes of functions defined in builtins.c.
2379 * builtins.h: Update prototype list to include all exported functions.
2380 * builtins.c: (default_libc_has_function, gnu_libc_has_function,
2381 no_c99_libc_has_function): Move to targhooks.c
2382 (build_string_literal, build_call_expr_loc_array,
2383 build_call_expr_loc_vec, build_call_expr_loc, build_call_expr): Move
2385 (expand_builtin_object_size, fold_builtin_object_size): Make static.
2386 * targhooks.c (default_libc_has_function, gnu_libc_has_function,
2387 no_c99_libc_has_function): Relocate from builtins.c.
2388 * tree.c: Include builtins.h.
2389 (build_call_expr_loc_array, build_call_expr_loc_vec,
2390 build_call_expr_loc, build_call_expr, build_string_literal): Relocate
2392 * fold-const.h (fold_fma): Move prototype to builtins.h.
2393 * realmpfr.h (do_mpc_arg2): Move prototype to builtins.h.
2394 * asan.c: Include builtins.h.
2395 * cfgexpand.c: Likewise.
2396 * convert.c: Likewise.
2397 * emit-rtl.c: Likewise.
2398 * except.c: Likewise.
2400 * fold-const.c: Likewise.
2401 * gimple-fold.c: Likewise.
2402 * gimple-ssa-strength-reduction.c: Likewise.
2403 * gimplify.c: Likewise.
2404 * ipa-inline.c: Likewise.
2405 * ipa-prop.c: Likewise.
2406 * lto-streamer-out.c: Likewise.
2408 * tree-inline.c: Likewise.
2409 * tree-object-size.c: Likewise.
2410 * tree-sra.c: Likewise.
2411 * tree-ssa-ccp.c: Likewise.
2412 * tree-ssa-forwprop.c: Likewise.
2413 * tree-ssa-loop-ivcanon.c: Likewise.
2414 * tree-ssa-loop-ivopts.c: Likewise.
2415 * tree-ssa-math-opts.c: Likewise.
2416 * tree-ssa-reassoc.c: Likewise.
2417 * tree-ssa-threadedge.c: Likewise.
2418 * tree-streamer-in.c: Likewise.
2419 * tree-vect-data-refs.c: Likewise.
2420 * tree-vect-patterns.c: Likewise.
2421 * tree-vect-stmts.c: Likewise.
2422 * config/aarch64/aarch64.c: Likewise.
2423 * config/alpha/alpha.c: Likewise.
2424 * config/arc/arc.c: Likewise.
2425 * config/arm/arm.c: Likewise.
2426 * config/avr/avr.c: Likewise.
2427 * config/bfin/bfin.c: Likewise.
2428 * config/c6x/c6x.c: Likewise.
2429 * config/cr16/cr16.c: Likewise.
2430 * config/cris/cris.c: Likewise.
2431 * config/epiphany/epiphany.c: Likewise.
2432 * config/fr30/fr30.c: Likewise.
2433 * config/frv/frv.c: Likewise.
2434 * config/h8300/h8300.c: Likewise.
2435 * config/i386/i386.c: Likewise.
2436 * config/i386/winnt.c: Likewise.
2437 * config/ia64/ia64.c: Likewise.
2438 * config/iq2000/iq2000.c: Likewise.
2439 * config/lm32/lm32.c: Likewise.
2440 * config/m32c/m32c.c: Likewise.
2441 * config/m32r/m32r.c: Likewise.
2442 * config/m68k/m68k.c: Likewise.
2443 * config/mcore/mcore.c: Likewise.
2444 * config/mep/mep.c: Likewise.
2445 * config/microblaze/microblaze.c: Likewise.
2446 * config/mips/mips.c: Likewise.
2447 * config/mmix/mmix.c: Likewise.
2448 * config/mn10300/mn10300.c: Likewise.
2449 * config/moxie/moxie.c: Likewise.
2450 * config/msp430/msp430.c: Likewise.
2451 * config/nds32/nds32.c: Likewise.
2452 * config/pa/pa.c: Likewise.
2453 * config/pdp11/pdp11.c: Likewise.
2454 * config/picochip/picochip.c: Likewise.
2455 * config/rl78/rl78.c: Likewise.
2456 * config/rs6000/rs6000.c: Likewise.
2457 * config/rx/rx.c: Likewise.
2458 * config/s390/s390.c: Likewise.
2459 * config/score/score.c: Likewise.
2460 * config/sh/sh.c: Likewise.
2461 * config/sparc/sparc.c: Likewise.
2462 * config/spu/spu.c: Likewise.
2463 * config/stormy16/stormy16.c: Likewise.
2464 * config/tilegx/tilegx.c: Likewise.
2465 * config/tilepro/tilepro.c: Likewise.
2466 * config/v850/v850.c: Likewise.
2467 * config/vax/vax.c: Likewise.
2468 * config/xtensa/xtensa.c: Likewise.
2470 2014-06-02 Jeff Law <law@redhat.com>
2472 PR rtl-optimization/61094
2473 * ree.c (combine_reaching_defs): Do not reextend an insn if it
2474 was marked as do_no_reextend. If a copy is needed to eliminate
2475 an extension, then mark it as do_not_reextend.
2477 2014-06-02 Marcus Shawcroft <marcus.shawcroft@arm.com>
2479 * config/aarch64/aarch64.md (set_fpcr): Drop ISB after FPCR write.
2481 2014-06-02 Richard Henderson <rth@redhat.com>
2484 * config/alpha/alpha.c (print_operand_address): Allow symbolic
2485 addresses inside asms. Use output_operand_lossage instead of
2488 2014-06-02 Uros Bizjak <ubizjak@gmail.com>
2491 * config/i386/i386.c (ix86_expand_vec_perm) [case V32QImode]: Use
2492 GEN_INT (-128) instead of GEN_INT (128) to set MSB of QImode constant.
2494 2014-06-02 Tom de Vries <tom@codesourcery.com>
2496 * config/aarch64/aarch64.c (aarch64_float_const_representable_p): Handle
2497 case that x has VOIDmode.
2499 2014-06-02 Bernd Schmidt <bernds@codesourcery.com>
2501 * varasm.c (copy_constant): Delete function.
2502 (build_constant_desc): Don't call it.
2504 2014-06-02 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2507 * config/arm/arm.h (TARGET_SUPPORTS_WIDE_INT): Define.
2508 * config/arm/arm.md (mov64 splitter): Replace const_double_operand
2509 with immediate_operand.
2511 2014-06-02 Andreas Schwab <schwab@suse.de>
2513 * config/ia64/ia64.c
2514 (ia64_first_cycle_multipass_dfa_lookahead_guard): Check
2515 pending_data_specs first.
2517 2014-06-02 Richard Biener <rguenther@suse.de>
2519 PR tree-optimization/61378
2520 * tree-ssa-sccvn.c (vn_reference_lookup_3): Initialize
2523 2014-06-01 Uros Bizjak <ubizjak@gmail.com>
2525 * config/i386/constraints.md (Bw): Rename from 'w'.
2526 (Bz): Rename from 'z'.
2527 * config/i386/i386.md: Change 'w' to 'Bw' and 'z' to 'Bz' globally.
2529 2014-06-01 Kai Tietz <ktietz@redhat.com>
2532 * config/i386/constrains.md (define_constrain): New 'Bs' constraint.
2533 * config/i386/i386.md (sibcall_insn_operand): Use Bs
2534 instead of m constraint.
2536 2014-05-31 Andreas Schwab <schwab@linux-m68k.org>
2538 * config/m68k/m68k.md (beq0_di, bne0_di): Make the "o" constraint
2539 a separate alternative where the scratch operand 2 is marked as
2542 2014-05-31 Kugan Vivekanandarajah <kuganv@linaro.org>
2544 * config/arm/arm.c (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): New define.
2545 (arm_builtins) : Add ARM_BUILTIN_GET_FPSCR and ARM_BUILTIN_SET_FPSCR.
2546 (bdesc_2arg) : Add description for builtins __builtins_arm_set_fpscr
2547 and __builtins_arm_get_fpscr.
2548 (arm_init_builtins) : Initialize builtins __builtins_arm_set_fpscr and
2549 __builtins_arm_get_fpscr.
2550 (arm_expand_builtin) : Expand builtins __builtins_arm_set_fpscr and
2551 __builtins_arm_ldfpscr.
2552 (arm_atomic_assign_expand_fenv): New function.
2553 * config/arm/vfp.md (set_fpscr): New pattern.
2554 (get_fpscr) : Likewise.
2555 * config/arm/unspecs.md (unspecv): Add VUNSPEC_GET_FPSCR and
2557 * doc/extend.texi (AARCH64 Built-in Functions) : Document
2558 __builtins_arm_set_fpscr, __builtins_arm_get_fpscr.
2560 2014-05-30 Jakub Jelinek <jakub@redhat.com>
2562 * asan.c (report_error_func): Add SLOW_P argument, use
2563 BUILT_IN_ASAN_*_N if set.
2564 (build_check_stmt): Likewise.
2565 (instrument_derefs): If T has insufficient alignment,
2566 force same handling as for odd sizes.
2568 * sanitizer.def (BUILT_IN_ASAN_REPORT_LOAD_N,
2569 BUILT_IN_ASAN_REPORT_STORE_N): New.
2570 * asan.c (struct asan_mem_ref): Change access_size type to
2572 (asan_mem_ref_init, asan_mem_ref_new, get_mem_refs_of_builtin_call,
2573 update_mem_ref_hash_table): Likewise.
2574 (asan_mem_ref_hasher::hash): Hash in a HWI.
2575 (report_error_func): Change size_in_bytes argument to HWI.
2576 Use *_N builtins if size_in_bytes is larger than 16 or not power of
2578 (build_shadow_mem_access): New function.
2579 (build_check_stmt): Use it. Change size_in_bytes argument to HWI.
2580 Handle size_in_bytes not power of two or larger than 16.
2581 (instrument_derefs): Don't give up if size_in_bytes is not
2582 power of two or is larger than 16.
2584 2014-05-30 Kai Tietz <ktietz@redhat.com>
2587 * config/i386/i386.c (x86_output_mi_thunk): Add memory case
2588 for sibling-tail-calls.
2589 * config/i386/i386.md (sibcall_insn_operand): Add memory-constrain
2591 * config/i386/predicates.md (sibcall_memory_operand): New predicate.
2592 (sibcall_insn_operand): Add check for sibcall_memory_operand.
2594 2014-05-30 Pitchumani Sivanupandi <pitchumani.s@atmel.com>
2596 * config/avr/avr-mcus.def: Change ATA6289 ISA to AVR4
2597 * config/avr/avr-tables.opt: Regenerate.
2598 * config/avr/t-multilib: Regenerate.
2599 * doc/avr-mmcu.texi: Regenerate.
2601 2014-05-30 Ian Lance Taylor <iant@google.com>
2603 * config/i386/xmmintrin.h (_mm_pause): Move out of scope of pragma
2606 2014-05-30 Tom de Vries <tom@codesourcery.com>
2608 * config/i386/i386.c (TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS):
2611 2014-05-30 Tom de Vries <tom@codesourcery.com>
2613 * lra-int.h (struct lra_reg): Add field actual_call_used_reg_set.
2614 * lra.c (initialize_lra_reg_info_element): Add init of
2615 actual_call_used_reg_set field.
2616 (lra): Call lra_create_live_ranges before lra_inheritance for
2618 * lra-assigns.c (lra_assign): Allow call_used_regs to cross calls for
2620 * lra-constraints.c (need_for_call_save_p): Use actual_call_used_reg_set
2621 instead of call_used_reg_set for -fuse-caller-save.
2622 * lra-lives.c (process_bb_lives): Calculate actual_call_used_reg_set.
2624 2014-05-30 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2626 * config/arm/thumb2.md (*thumb2_movhi_insn): Set type of movw
2628 * config/arm/vfp.md (*thumb2_movsi_vfp): Likewise.
2630 2014-05-30 Richard Sandiford <rdsandiford@googlemail.com>
2632 * ira.c (ira_get_dup_out_num): Check for output operands at
2633 the start of the loop. Handle cases where an included alternative
2634 follows an excluded one.
2636 2014-05-29 Mike Stump <mikestump@comcast.net>
2639 * collect2.c (maybe_run_lto_and_relink): Be sure to always run
2640 post ld passes when lto is used.
2642 2014-05-29 Vladimir Makarov <vmakarov@redhat.com>
2644 PR rtl-optimization/61325
2645 * lra-constraints.c (process_address): Rename to process_address_1.
2646 (process_address): New function.
2648 2014-05-29 Alan Lawrence <alan.lawrence@arm.com>
2650 * config/aarch64/aarch64-builtins.c (aarch64_types_binopv_qualifiers,
2651 TYPES_BINOPV): New static data.
2652 * config/aarch64/aarch64-simd-builtins.def (im_lane_bound):
2654 * config/aarch64/aarch64-simd.md (aarch64_ext,
2655 aarch64_im_lane_boundsi): New patterns.
2656 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Match
2658 (aarch64_evpc_ext): New function.
2660 * config/aarch64/iterators.md (UNSPEC_EXT): New enum element.
2662 * config/aarch64/arm_neon.h (vext_f32, vext_f64, vext_p8, vext_p16,
2663 vext_s8, vext_s16, vext_s32, vext_s64, vext_u8, vext_u16, vext_u32,
2664 vext_u64, vextq_f32, vextq_f64, vextq_p8, vextq_p16, vextq_s8,
2665 vextq_s16, vextq_s32, vextq_s64, vextq_u8, vextq_u16, vextq_u32,
2666 vextq_u64): Replace __asm with __builtin_shuffle and im_lane_boundsi.
2668 2014-05-29 Tom de Vries <tom@codesourcery.com>
2670 * rtl.h (BLOCK_SYMBOL_CHECK): Use SYMBOL_REF_FLAGS.
2672 2014-05-29 Richard Earnshaw <rearnsha@arm.com>
2673 Richard Sandiford <rdsandiford@googlemail.com>
2675 * arm/iterators.md (shiftable_ops): New code iterator.
2676 (t2_binop0, arith_shift_insn): New code attributes.
2677 * arm/predicates.md (shift_nomul_operator): New predicate.
2678 * arm/arm.md (insn_enabled): Delete.
2679 (enabled): Remove insn_enabled test.
2680 (*arith_shiftsi): Delete. Replace with ...
2681 (*<arith_shift_insn>_multsi): ... new pattern.
2682 (*<arith_shift_insn>_shiftsi): ... new pattern.
2683 * config/arm/arm.c (arm_print_operand): Handle operand format 'b'.
2685 2014-05-29 Radovan Obradovic <robradovic@mips.com>
2686 Tom de Vries <tom@codesourcery.com>
2688 * config/mips/mips.h (POST_CALL_TMP_REG): Define.
2689 * config/mips/mips.c (mips_emit_call_insn): Add POST_CALL_TMP_REG
2691 (mips_split_call): Use POST_CALL_TMP_REG.
2692 (TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS): Redefine to true.
2694 2014-05-29 Tom de Vries <tom@codesourcery.com>
2696 * final.c (collect_fn_hard_reg_usage): Guard variable declaration
2697 with #ifdef STACK_REGS.
2699 2014-05-28 Jan Hubicka <hubicka@ucw.cz>
2701 * varasm.c (get_variable_section): Walk aliases.
2702 (place_block_symbol): Walk aliases.
2704 2014-05-28 Tom de Vries <tom@codesourcery.com>
2707 2014-05-28 Tom de Vries <tom@codesourcery.com>
2709 * lra-int.h (struct lra_reg): Add field actual_call_used_reg_set.
2710 * lra.c (initialize_lra_reg_info_element): Add init of
2711 actual_call_used_reg_set field.
2712 (lra): Call lra_create_live_ranges before lra_inheritance for
2714 * lra-assigns.c (lra_assign): Allow call_used_regs to cross calls for
2716 * lra-constraints.c (need_for_call_save_p): Use
2717 actual_call_used_reg_set instead of call_used_reg_set for
2719 * lra-lives.c (process_bb_lives): Calculate actual_call_used_reg_set.
2721 2014-05-28 Richard Sandiford <rdsandiford@googlemail.com>
2723 * doc/md.texi: Document that the % constraint character must
2724 be at the beginning of the string.
2725 * genoutput.c (validate_insn_alternatives): Check that '=',
2726 '+' and '%' only appear at the beginning of a constraint.
2727 * ira.c (commutative_constraint_p): Delete.
2728 (ira_get_dup_out_num): Expect the '%' commutativity marker to be
2729 at the start of the string.
2730 * config/alpha/alpha.md (*movmemdi_1, *clrmemdi_1): Remove
2732 * config/arm/neon.md (bicdi3_neon): Likewise.
2733 * config/iq2000/iq2000.md (addsi3_internal, subsi3_internal, sgt_si)
2734 (slt_si, sltu_si): Likewise.
2735 * config/vax/vax.md (sbcdi3): Likewise.
2736 * config/h8300/h8300.md (*cmpstz): Remove duplicate '+'.
2737 * config/arc/arc.md (mulsi_600, mulsidi_600, umulsidi_600)
2738 (mul64): Move '%' to beginning of constraint.
2739 * config/arm/arm.md (*xordi3_insn): Likewise.
2740 * config/nds32/nds32.md (add<mode>3, mulsi3, andsi3, iorsi3)
2743 2014-05-28 Richard Sandiford <rdsandiford@googlemail.com>
2745 * doc/md.texi: Document the restrictions on the "enabled" attribute.
2747 2014-05-28 Jason Merrill <jason@redhat.com>
2750 * cgraph.h (symtab_node::get_comdat_group_id): New.
2751 * cgraphunit.c (analyze_functions): Call it.
2752 * symtab.c (dump_symtab_node): Likewise.
2753 * tree.c (decl_comdat_group_id): New.
2754 * tree.h: Declare it.
2755 * lto-streamer-out.c (write_symbol): Use it.
2756 * trans-mem.c (ipa_tm_create_version_alias): Likewise.
2758 2014-05-28 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
2760 PR bootstrap/PR61146
2761 * wide-int.cc: Do not include longlong.h when compiling with clang.
2763 2014-05-28 Richard Biener <rguenther@suse.de>
2765 * tree-ssa-propagate.c (add_control_edge): Print less vertical space.
2766 * tree-vrp.c (extract_range_from_ssa_name): Also copy VR_UNDEFINED.
2767 (vrp_visit_assignment_or_call): Print less vertical space.
2768 (vrp_visit_stmt): Likewise.
2769 (vrp_visit_phi_node): Likewise. For a PHI argument with
2770 VR_VARYING range consider recording it as copy.
2772 2014-05-28 Richard Biener <rguenther@suse.de>
2775 2014-05-28 Richard Biener <rguenther@suse.de>
2777 * hwint.h (HOST_WIDE_INT_PRINT_*): Define in terms of PRI*64.
2779 2014-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
2781 * expr.c (expand_assignment): Fold the bitpos in the to_rtx if
2782 sufficiently aligned and an offset is used at the same time.
2783 (expand_expr_real_1): Likewise.
2785 2014-05-28 Richard Biener <rguenther@suse.de>
2788 * fold-const.c (fold_comparison): When folding
2789 X +- C1 CMP Y +- C2 to X CMP Y +- C2 +- C1 also ensure
2790 the sign of the remaining constant operand stays the same.
2792 2014-05-28 Kaushik Phatak <kaushik.phatak@kpit.com>
2794 * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
2795 __RL78_64BIT_DOUBLES__ or __RL78_32BIT_DOUBLES__.
2796 (ASM_SPEC): Pass -m64bit-doubles or -m32bit-doubles on
2798 (DOUBLE_TYPE_SIZE): Use 64 bit if TARGET_64BIT_DOUBLES is true.
2799 * gcc/config/rl78/rl78.opt (m64bit-doubles): New option.
2800 (m32bit-doubles) Likewise.
2801 * gcc/config/rl78/t-rl78: Add 64-bit-double multilib.
2802 * doc/invoke.texi: Document -m32bit-doubles and -m64bit-doubles
2805 2014-05-28 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2807 * configure.ac ($gcc_cv_ld_clearcap): New test.
2808 * configure: Regenerate.
2809 * config.in: Regenerate.
2810 * config/sol2.opt (mclear-hwcap): New option.
2811 * config/sol2.h (LINK_CLEARCAP_SPEC): Define.
2812 * config/sol2-clearcap.map: Moved here from
2813 testsuite/gcc.target/i386/clearcap.map.
2814 * config/sol2-clearcapv2.map: Move here from
2815 gcc.target/i386/clearcapv2.map.
2816 * config/t-sol2 (install): Depend on install-clearcap-map.
2817 (install-clearcap-map): New target.
2818 * doc/invoke.texi (Option Summary, Solaris 2 Options): Document
2821 2014-05-28 Richard Biener <rguenther@suse.de>
2823 * hwint.h (*_HALF_WIDE_INT*): Move to ...
2824 * wide-int.cc (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT):
2825 ... here and remove the rest.
2826 * hwint.h (HOST_WIDE_INT_PRINT_*): Define in terms of PRI*64.
2828 2014-05-28 Richard Biener <rguenther@suse.de>
2830 PR tree-optimization/61335
2831 * tree-vrp.c (vrp_visit_phi_node): If the compare of old and
2832 new range fails, drop to varying.
2834 2014-05-28 Tom de Vries <tom@codesourcery.com>
2836 * lra-int.h (struct lra_reg): Add field actual_call_used_reg_set.
2837 * lra.c (initialize_lra_reg_info_element): Add init of
2838 actual_call_used_reg_set field.
2839 (lra): Call lra_create_live_ranges before lra_inheritance for
2841 * lra-assigns.c (lra_assign): Allow call_used_regs to cross calls for
2843 * lra-constraints.c (need_for_call_save_p): Use
2844 actual_call_used_reg_set instead of call_used_reg_set for
2846 * lra-lives.c (process_bb_lives): Calculate actual_call_used_reg_set.
2848 2014-05-28 Radovan Obradovic <robradovic@mips.com>
2849 Tom de Vries <tom@codesourcery.com>
2851 * doc/invoke.texi (@item Optimization Options): Add -fuse-caller-save
2853 (@item -fuse-caller-save): New item.
2855 2014-05-28 Radovan Obradovic <robradovic@mips.com>
2856 Tom de Vries <tom@codesourcery.com>
2858 * opts.c (default_options_table): Add OPT_LEVELS_2_PLUS entry with
2859 OPT_fuse_caller_save.
2861 2014-05-28 Radovan Obradovic <robradovic@mips.com>
2862 Tom de Vries <tom@codesourcery.com>
2864 * df-scan.c (df_get_call_refs): Use get_call_reg_set_usage.
2865 * caller-save.c (setup_save_areas, save_call_clobbered_regs): Use
2866 get_call_reg_set_usage.
2867 * resource.c (mark_set_resources, mark_target_live_regs): Use
2868 get_call_reg_set_usage.
2869 * ira-int.h (struct ira_allocno): Add crossed_calls_clobbered_regs
2871 (ALLOCNO_CROSSED_CALLS_CLOBBERED_REGS): Define.
2872 * ira-lives.c (process_bb_node_lives): Use get_call_reg_set_usage.
2873 Calculate ALLOCNO_CROSSED_CALLS_CLOBBERED_REGS.
2874 * ira-build.c (ira_create_allocno): Init
2875 ALLOCNO_CROSSED_CALLS_CLOBBERED_REGS.
2876 (create_cap_allocno, propagate_allocno_info)
2877 (propagate_some_info_from_allocno)
2878 (copy_info_to_removed_store_destinations): Handle
2879 ALLOCNO_CROSSED_CALLS_CLOBBERED_REGS.
2880 * ira-costs.c (ira_tune_allocno_costs): Use
2881 ALLOCNO_CROSSED_CALLS_CLOBBERED_REGS to adjust costs.
2883 2014-05-28 Radovan Obradovic <robradovic@mips.com>
2884 Tom de Vries <tom@codesourcery.com>
2886 * cgraph.h (struct cgraph_rtl_info): Add function_used_regs
2887 and function_used_regs_valid fields.
2888 * final.c: Move include of hard-reg-set.h to before rtl.h to declare
2889 find_all_hard_reg_sets.
2890 (collect_fn_hard_reg_usage, get_call_fndecl, get_call_cgraph_rtl_info)
2891 (get_call_reg_set_usage): New function.
2892 (rest_of_handle_final): Use collect_fn_hard_reg_usage.
2893 * regs.h (get_call_reg_set_usage): Declare.
2895 2014-05-28 Georg-Johann Lay <avr@gjlay.de>
2898 * config/dbx.h (License): Add Runtime Library Exception.
2899 * config/newlib-stdint.h (License): Same.
2900 * config/rtems.h (License): Same
2901 * config/initfini-array.h (License): Same
2902 * config/v850/v850.h (License): Same.
2903 * config/v850/v850-opts.h (License): Same
2904 * config/v850/rtems.h (License): Same.
2906 2014-05-28 Georg-Johann Lay <avr@gjlay.de>
2909 * doc/extend.texi (Local Labels): Note that label differences are
2910 not supported for AVR.
2912 2014-05-27 Andrew Pinski <apinski@cavium.com>
2914 * config/aarch64/aarch64.md (stack_protect_set_<mode>):
2915 Use <w> for the register in assembly template.
2916 (stack_protect_test): Use the mode of operands[0] for the result.
2917 (stack_protect_test_<mode>): Use <w> for the register
2918 in assembly template.
2920 2014-05-27 DJ Delorie <dj@redhat.com>
2922 * config/rx/rx.c (add_vector_labels): New.
2923 (rx_output_function_prologue): Call it.
2924 (rx_handle_func_attribute): Don't require empty arguments.
2925 (rx_handle_vector_attribute): New.
2926 (rx_attribute_table): Add "vector" attribute.
2927 * doc/extend.texi (interrupt, vector): Document new/changed
2928 RX-specific attributes.
2930 * config/rx/rx.c (rx_adjust_insn_length): Skip for non-insns.
2932 2014-05-27 Eric Botcazou <ebotcazou@adacore.com>
2934 * double-int.c (div_and_round_double) <ROUND_DIV_EXPR>: Use the proper
2935 predicate to detect a negative quotient.
2937 2014-05-27 Eric Botcazou <ebotcazou@adacore.com>
2939 * fold-const.c (fold_comparison): Clean up and extend X +- C1 CMP C2
2940 to X CMP C2 -+ C1 transformation to EQ_EXPR/NE_EXPR.
2941 Add X - Y CMP 0 to X CMP Y transformation.
2942 (fold_binary_loc) <EQ_EXPR/NE_EXPR>: Remove same transformations.
2944 2014-05-27 Segher Boessenkool <segher@kernel.crashing.org>
2946 * stmt.c (dump_case_nodes): Don't convert values to HOST_WIDE_INT
2949 2014-05-27 Steve Ellcey <sellcey@mips.com>
2951 * config/mips/mips.c: Add include of cgraph.h.
2953 2014-05-27 Richard Biener <rguenther@suse.de>
2955 * system.h (__STDC_FORMAT_MACROS): Define as very first thing.
2957 2014-05-27 Georg-Johann Lay <avr@gjlay.de>
2960 * config/arm/arm.h (License): Add note to COPYING.RUNTIME.
2961 * config/arm/arm-cores.def (License): Same.
2962 * config/arm/arm-opts.h (License): Same.
2963 * config/arm/aout.h (License): Same.
2964 * config/arm/bpabi.h (License): Same.
2965 * config/arm/elf.h (License): Same.
2966 * config/arm/linux-elf.h (License): Same.
2967 * config/arm/linux-gas.h (License): Same.
2968 * config/arm/netbsd-elf.h (License): Same.
2969 * config/arm/uclinux-eabi.h (License): Same.
2970 * config/arm/uclinux-elf.h (License): Same.
2971 * config/arm/vxworks.h (License): Same.
2973 2014-05-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2975 * config/arm/neon.md (neon_bswap<mode>): New pattern.
2976 * config/arm/arm.c (neon_itype): Add NEON_BSWAP.
2977 (arm_init_neon_builtins): Handle NEON_BSWAP.
2978 Define required type nodes.
2979 (arm_expand_neon_builtin): Handle NEON_BSWAP.
2980 (arm_builtin_vectorized_function): Handle BUILTIN_BSWAP builtins.
2981 * config/arm/arm_neon_builtins.def (bswap): Define builtins.
2982 * config/arm/iterators.md (VDQHSD): New mode iterator.
2984 2014-05-27 Richard Biener <rguenther@suse.de>
2986 * tree-vrp.c (vrp_evaluate_conditional_warnv_with_ops_using_ranges):
2987 Try using literal operands when comparing value-ranges failed.
2989 2014-05-27 Richard Sandiford <rdsandiford@googlemail.com>
2991 * ira.c (commutative_operand): Adjust for change to recog_data.
2992 [Missing from previous commit.]
2994 2014-05-27 Richard Sandiford <rdsandiford@googlemail.com>
2996 * system.h (TEST_BIT): New macro.
2997 * recog.h (alternative_mask): New type.
2998 (ALL_ALTERNATIVES, ALTERNATIVE_BIT): New macros.
2999 (recog_data_d): Replace alternative_enabled_p array with
3000 enabled_alternatives.
3001 (target_recog): New structure.
3002 (default_target_recog, this_target_recog): Declare.
3003 (get_enabled_alternatives, recog_init): Likewise.
3004 * recog.c (default_target_recog, this_target_recog): New variables.
3005 (get_enabled_alternatives): New function.
3006 (extract_insn): Use it.
3007 (recog_init): New function.
3008 (preprocess_constraints, constrain_operands): Adjust for change to
3010 * postreload.c (reload_cse_simplify_operands): Likewise.
3011 * reload.c (find_reloads): Likewise.
3012 * ira-costs.c (record_reg_classes): Likewise.
3013 * ira-lives.c (single_reg_class): Likewise. Fix bug in which
3014 all alternatives after a disabled one would be skipped.
3015 (ira_implicitly_set_insn_hard_regs): Likewise.
3016 * ira.c (ira_setup_alts): Adjust for change to recog_data.
3017 * lra-int.h (lra_insn_recog_data): Replace alternative_enabled_p
3018 with enabled_alternatives.
3019 * lra.c (free_insn_recog_data): Update accordingly.
3020 (lra_update_insn_recog_data): Likewise.
3021 (lra_set_insn_recog_data): Likewise. Use get_enabled_alternatives.
3022 * lra-constraints.c (process_alt_operands): Likewise. Handle
3023 only_alternative as part of the enabled mask.
3024 * target-globals.h (this_target_recog): Declare.
3025 (target_globals): Add a recog field.
3026 (restore_target_globals): Restore this_target_recog.
3027 * target-globals.c: Include recog.h.
3028 (default_target_globals): Initialize recog field.
3029 (save_target_globals): Likewise.
3030 * reginfo.c (reinit_regs): Call recog_init.
3031 * toplev.c (backend_init_target): Likewise.
3033 2014-05-27 Richard Sandiford <rdsandiford@googlemail.com>
3035 * gencodes.c (main): Make LAST_INSN_CODE higher than any insn code,
3036 rather than any named insn's code.
3038 2014-05-27 Georg-Johann Lay <avr@gjlay.de>
3041 * config/arm/arm-opts.h (License): Add GCC Runtime Library Exception.
3042 * config/arm/arm-cores.def (License): Same.
3044 2014-05-26 Jan Hubicka <hubicka@ucw.cz>
3046 * tree.h (decl_comdat_group): Declare.
3047 * cgraph.h (symtab_in_same_comdat_p): Move offline to ...
3048 * tree.c (decl_comdat_group): Here.
3050 2014-05-26 Richard Sandiford <r.sandiford@uk.ibm.com>
3052 PR rtl-optimization/61222
3053 * combine.c (simplify_shift_const_1): When moving a PLUS outside
3054 the shift, truncate the PLUS operand to the result mode.
3056 2014-05-26 Uros Bizjak <ubizjak@gmail.com>
3059 * config/i386/i386.c (ix86_rtx_costs)
3060 <case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF>:
3063 2014-05-26 Martin Jambor <mjambor@suse.cz>
3065 * ira.c (split_live_ranges_for_shrink_wrap): Remove bailout on
3068 2014-05-26 Richard Biener <rguenther@suse.de>
3070 * wide-int.h (wi::int_traits <long>, wi::int_traits <unsigned long>,
3071 wi::int_traits <long long>, wi::int_traits <unsigned long long>):
3072 Provide specializations.
3073 (wi::int_traits <HOST_WIDE_INT>,
3074 wi::int_traits <unsigned HOST_WIDE_INT>): Remove specializations.
3076 2014-05-26 Alan Modra <amodra@gmail.com>
3079 * config/rs6000/rs6000.c (rs6000_emit_set_const): Remove unneeded
3080 params and return a bool. Remove dead code. Update comment.
3081 Assert we have a const_int source. Remove bogus code from
3082 32-bit HWI days. Move !TARGET_POWERPC64 handling, and correct
3083 handling of constants > 2G and reg_equal note, from..
3084 (rs6000_emit_set_long_const): ..here. Remove unneeded param and
3085 return value. Update comment. If we can, use a new pseudo
3086 for intermediate calculations.
3087 * config/rs6000/rs6000-protos.h (rs6000_emit_set_const): Update
3089 * config/rs6000/rs6000.md (movsi_internal1_single+1): Update
3090 call to rs6000_emit_set_const in splitter.
3091 (movdi_internal64+2, +3): Likewise.
3093 2014-05-26 Richard Biener <rguenther@suse.de>
3095 * system.h: Define __STDC_FORMAT_MACROS before
3096 including inttypes.h.
3097 * hwint.h (HOST_WIDEST_INT, HOST_BITS_PER_WIDEST_INT,
3098 HOST_WIDEST_INT_PRINT, HOST_WIDEST_INT_PRINT_DEC,
3099 HOST_WIDEST_INT_PRINT_DEC_C, HOST_WIDEST_INT_PRINT_UNSIGNED,
3100 HOST_WIDEST_INT_PRINT_HEX, HOST_WIDEST_INT_PRINT_DOUBLE_HEX,
3101 HOST_WIDEST_INT_C): Remove.
3102 (PRId64, PRIi64, PRIo64, PRIu64, PRIx64, PRIX64): Define
3103 if C99 inttypes.h is not available.
3104 * coretypes.h (gcov_type, gcov_type_unsigned): Use [u]int64_t.
3105 * gcov-io.h (gcov_type, gcov_type_unsigned): Likewise.
3106 * gcov-io.c (gcov_histo_index): Drop non-64bit hwi case.
3107 * cfgloop.h (struct niter_desc): Use uint64_t for niter field.
3108 * bitmap.c (struct bitmap_descriptor_d): Use uint64_t for counters.
3109 (struct output_info): Likewise.
3110 (print_statistics): Adjust.
3111 (dump_bitmap_statistics): Likewise.
3112 * bt-load.c (migrate_btr_defs): Print with PRId64.
3113 * cfg.c (dump_edge_info, dump_bb_info): Likewise.
3114 (MAX_SAFE_MULTIPLIER): Adjust.
3115 * cfghooks.c (dump_bb_for_graph): Print with PRId64.
3116 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee,
3117 dump_cgraph_node): Likewise.
3118 * final.c (dump_basic_block_info): Likewise.
3119 * gcov-dump.c (tag_counters, tag_summary, dump_working_sets): Likewise.
3120 * gcov.c (format_gcov): Likewise.
3121 * ipa-cp.c (good_cloning_opportunity_p): Likewise. Use int64_t
3123 (get_clone_agg_value): Use HOST_WIDE_INT for offset.
3124 * ipa-inline.c (compute_max_insns): Use int64_t for calcuation.
3125 (inline_small_functions, dump_overall_stats, dump_inline_stats):
3126 Use PRId64 for dumping.
3127 * ipa-profile.c (dump_histogram, ipa_profile): Likewise.
3128 * ira-color.c (struct allocno_hard_regs): Use int64_t for cost.
3129 (add_allocno_hard_regs): Adjust.
3130 * loop-doloop.c (doloop_modify): Print using PRId64.
3131 * loop-iv.c (inverse): Compute in uint64_t.
3132 (determine_max_iter, iv_number_of_iterations): Likewise.
3133 * loop-unroll.c (decide_peel_completely, decide_peel_simple):
3135 * lto-streamer-out.c (write_symbol): Use uint64_t.
3136 * mcf.c (CAP_INFINITY): Use int64_t maximum.
3137 (dump_fixup_edge, create_fixup_graph, cancel_negative_cycle,
3138 find_max_flow, adjust_cfg_counts): Use int64_t and dump with PRId64.
3139 * modulo-sched.c (const_iteration_count): Use int64_t.
3140 (sms_schedule): Dump using PRId64.
3141 * predict.c (dump_prediction): Likewise.
3142 * pretty-print.h (pp_widest_integer): Remove.
3143 * profile.c (get_working_sets, is_edge_inconsistent,
3144 is_inconsistent, read_profile_edge_counts): Dump using PRId64.
3145 * tree-pretty-print.c (pp_double_int): Remove case handling
3146 HOST_BITS_PER_DOUBLE_INT == HOST_BITS_PER_WIDEST_INT.
3147 * tree-ssa-math-opts.c (struct symbolic_number): Use uint64_t
3149 (pass_optimize_bswap::execute): Remove restriction on hosts.
3150 * tree-streamer-in.c (streamer_alloc_tree): Use HOST_WIDE_INT.
3151 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
3152 * tree.c (widest_int_cst_value): Remove.
3153 * tree.h (widest_int_cst_value): Likewise.
3154 * value-prof.c (dump_histogram_value): Print using PRId64.
3155 * gengtype.c (main): Also inject int64_t.
3156 * ggc-page.c (struct max_alignment): Use int64_t.
3157 * alloc-pool.c (struct allocation_object_def): Likewise.
3158 * ira-conflicts.c (build_conflict_bit_table): Use uint64_t
3160 * doc/tm.texi.in: Remove reference to HOST_WIDEST_INT.
3161 * doc/tm.texi: Regenerated.
3162 * gengtype-lex.l (IWORD): Handle [u]int64_t.
3163 * config/sh/sh.c (expand_cbranchdi4): Use gcov_type.
3164 * config/mmix/mmix-protos.h (mmix_intval, mmix_shiftable_wyde_value,
3165 mmix_output_register_setting): Use [u]int64_t in prototypes.
3166 * config/mmix/mmix.c (mmix_print_operand, mmix_output_register_setting,
3167 mmix_shiftable_wyde_value, mmix_output_shiftvalue_op_from_str,
3168 mmix_output_octa, mmix_output_shifted_value): Adjust.
3169 (mmix_intval): Adjust. Remove unreachable case.
3170 * config/mmix/mmix.md (*nonlocal_goto_receiver_expanded): Use int64_t.
3172 2014-05-26 Richard Biener <rguenther@suse.de>
3174 * configure.ac: Drop __int64 type check. Insist that we
3175 found uint64_t and int64_t.
3176 * hwint.h (HOST_BITS_PER___INT64): Remove.
3177 (HOST_BITS_PER_WIDE_INT): Define to 64 and remove __int64 case.
3178 (HOST_WIDE_INT_PRINT_*): Remove 32bit case.
3179 (HOST_WIDEST_INT*): Define to HOST_WIDE_INT*.
3180 (HOST_WIDEST_FAST_INT): Remove __int64 case.
3181 * vmsdbg.h (struct _DST_SRC_COMMAND): Use int64_t
3182 for dst_q_src_df_rms_cdt.
3183 * configure: Regenerate.
3184 * config.in: Likewise.
3186 2014-05-26 Michael Tautschnig <mt@debian.org>
3189 * doc/extend.texi (X86 Built-in Functions): Fix parameter lists of
3190 __builtin_ia32_vfrczs[sd] and __builtin_ia32_mpsadbw256.
3192 2014-05-26 Zhenqiang Chen <zhenqiang.chen@linaro.org>
3194 PR rtl-optimization/61278
3195 * shrink-wrap.c (move_insn_for_shrink_wrap): Check df_live.
3197 2014-05-26 Zhenqiang Chen <zhenqiang.chen@linaro.org>
3199 PR rtl-optimization/61220
3200 Part of PR rtl-optimization/61225
3201 * shrink-wrap.c (move_insn_for_shrink_wrap): Skip SP and FP adjustment
3202 insn; skip split_edge for a block with only one successor.
3204 2014-05-23 Jan Hubicka <hubicka@ucw.cz>
3206 * symtab.c (symtab_nonoverwritable_alias): Copy READONLY flag
3209 2014-05-23 Jan Hubicka <hubicka@ucw.cz>
3211 * ipa-visibility.c (can_replace_by_local_alias_in_vtable): New function.
3212 (update_vtable_references): New function.
3213 (function_and_variable_visibility): Rewrite also vtable initializers.
3214 * varpool.c (cgraph_variable_initializer_availability): Remove assert.
3216 2014-05-23 Jan Hubicka <hubicka@ucw.cz>
3218 * ggc.h (ggc_grow): New function.
3219 * ggc-none.c (ggc_grow): New function.
3220 * ggc-page.c (ggc_grow): Likewise.
3222 2014-05-23 Jan Hubicka <hubicka@ucw.cz>
3224 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
3225 address_taken_from_non_vtable_p, comdat_can_be_unshared_p_1,
3226 comdat_can_be_unshared_p, cgraph_externally_visible_p,
3227 varpool_externally_visible_p, can_replace_by_local_alias,
3228 update_visibility_by_resolution_info, function_and_variable_visibility,
3229 pass_data_ipa_function_and_variable_visibility,
3230 make_pass_ipa_function_and_variable_visibility,
3231 whole_program_function_and_variable_visibility,
3232 pass_data_ipa_whole_program_visibility,
3233 make_pass_ipa_whole_program_visibility): Move to ipa-visibility.c
3234 * cgraph.h (cgraph_local_node_p): Declare.
3235 * ipa-visibility.c: New file.
3236 * Makefile.in (OBJS): Add ipa-visiblity.o
3238 2014-05-23 Jan Hubicka <hubicka@ucw.cz>
3240 * gimple-fold.c (can_refer_decl_in_current_unit_p): Be sure
3241 that var decl is available.
3243 2014-05-23 Jan Hubicka <hubicka@ucw.cz>
3245 * tree-core.h (tree_decl_with_vis): Replace comdat_group by
3246 symtab_node pointer.
3247 * tree.c (copy_node_stat): Be sure to not copy symtab_node pointer.
3248 (find_decls_types_r): Do not walk COMDAT_GROUP.
3249 * tree.h (DECL_COMDAT_GROUP): Revamp to use decl_comdat_group.
3250 * varasm.c (make_decl_one_only): Use set_comdat_group;
3251 create node if needed.
3252 * ipa-inline-transform.c (save_inline_function_body): Update
3253 way we decl->symtab mapping.
3254 * symtab.c (symtab_hash, hash_node, eq_node
3255 symtab_insert_node_to_hashtable): Remove.
3256 (symtab_register_node): Update.
3257 (symtab_unregister_node): Update.
3258 (symtab_get_node): Reimplement as inline function.
3259 (symtab_add_to_same_comdat_group): Update.
3260 (symtab_dissolve_same_comdat_group_list): Update.
3261 (dump_symtab_base): Update.
3262 (verify_symtab_base): Update.
3263 (symtab_make_decl_local): Update.
3264 (fixup_same_cpp_alias_visibility): Update.
3265 (symtab_nonoverwritable_alias): Update.
3266 * cgraphclones.c (set_new_clone_decl_and_node_flags): Update.
3267 * ipa.c (update_visibility_by_resolution_info): UPdate.
3268 * bb-reorder.c: Include cgraph.h
3269 * lto-streamer-out.c (DFS_write_tree_body, hash_tree): Do not deal
3271 * ipa-comdats.c (set_comdat_group, ipa_comdats): Update.
3272 * cgraph.c (cgraph_get_create_node): Update.
3273 * cgraph.h (struct symtab_node): Add get_comdat_group, set_comdat_group
3275 (symtab_get_node): Make inline.
3276 (symtab_insert_node_to_hashtable): Remove.
3277 (symtab_can_be_discarded): Update.
3278 (decl_comdat_group): New function.
3279 * tree-streamer-in.c (lto_input_ts_decl_with_vis_tree_pointers):
3281 * lto-cgraph.c (lto_output_node, lto_output_varpool_node): Stream out
3283 (read_comdat_group): New function.
3284 (input_node, input_varpool_node): Use it.
3285 * trans-mem.c (ipa_tm_create_version_alias): Update code creating
3287 * mips.c (mips_start_unique_function): Likewise.
3288 (ix86_code_end): Likewise.
3289 (rs6000_code_end): Likweise.
3290 * tree-streamer-out.c (DECL_COMDAT_GROUP): Do not stream comdat group.
3292 2014-05-23 Jan Hubicka <hubicka@ucw.cz>
3294 * gengtype-state.c (fatal_reading_state): Bring offline.
3295 * optabs.c (widening_optab_handler): Bring offline.
3296 * optabs.h (widening_optab_handler): Likewise.
3297 * final.c (get_attr_length_1): Likewise.
3299 2014-05-23 Jan Hubicka <hubicka@ucw.cz>
3301 * sched-int.h (sd_iterator_cond): Manually tail recurse.
3303 2014-05-23 Segher Boessenkool <segher@kernel.crashing.org>
3305 * config/rs6000/440.md (ppc440-integer): Include shift without dot.
3306 (ppc440-compare): Include shift with dot.
3307 * config/rs6000/e300c2c3.md (ppce300c3_iu): Include shift without dot.
3308 * config/rs6000/e5500.md (e5500_sfx2): Include constant shift
3310 * config/rs6000/e6500.md (e6500_sfx): Exclude constant shift
3312 (e6500_sfx2): Include it.
3313 * config/rs6000/rs6000.md ( *zero_extend<mode>di2_internal1,
3314 *zero_extend<mode>di2_internal2, *zero_extend<mode>di2_internal3,
3315 *zero_extendsidi2_lfiwzx, andsi3_mc, andsi3_nomc,
3316 andsi3_internal0_nomc, extzvsi_internal, extzvdi_internal,
3317 *extzvdi_internal1, *extzvdi_internal2, rotlsi3, *rotlsi3_64,
3318 *rotlsi3_internal4, *rotlsi3_internal7le, *rotlsi3_internal7be,
3319 *rotlsi3_internal10le, *rotlsi3_internal10be, rlwinm,
3320 *lshiftrt_internal1le, *lshiftrt_internal1be,
3321 *lshiftrt_internal4le, *lshiftrt_internal4be, rotldi3,
3322 *rotldi3_internal4, *rotldi3_internal7le, *rotldi3_internal7be,
3323 *rotldi3_internal10le, *rotldi3_internal10be,
3324 *rotldi3_internal13le, *rotldi3_internal13be, *ashldi3_internal4,
3325 ashldi3_internal5, *ashldi3_internal6, *ashldi3_internal7,
3326 ashldi3_internal8, *ashldi3_internal9, anddi3_mc, anddi3_nomc,
3327 *anddi3_internal2_mc, *anddi3_internal3_mc, and 4 anonymous
3328 define_insns): Use type "shift" in the appropriate alternatives.
3330 2014-05-23 Segher Boessenkool <segher@kernel.crashing.org>
3332 * config/rs6000/rs6000.md (type): Add "logical". Delete
3334 (dot): Adjust comment.
3335 (andsi3_mc, *andsi3_internal2_mc, *andsi3_internal3_mc,
3336 *andsi3_internal4, *andsi3_internal5_mc, *boolsi3_internal2,
3337 *boolsi3_internal3, *boolccsi3_internal2, *boolccsi3_internal3,
3338 anddi3_mc, *anddi3_internal2_mc, *anddi3_internal3_mc,
3339 *booldi3_internal2, *booldi3_internal3, *boolcdi3_internal2,
3340 *boolcdi3_internal3, *boolccdi3_internal2, *boolccdi3_internal3,
3341 *mov<mode>_internal2, and 10 anonymous define_insns): Use "logical".
3342 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
3344 * config/rs6000/40x.md (ppc403-integer, ppc403-compare): Adjust.
3345 * config/rs6000/440.md (ppc440-integer, ppc440-compare): Adjust.
3346 * config/rs6000/476.md (ppc476-simple-integer, ppc476-compare): Adjust.
3347 * config/rs6000/603.md (ppc603-integer, ppc603-compare): Adjust.
3348 * config/rs6000/6xx.md (ppc604-integer, ppc604-compare): Adjust.
3349 * config/rs6000/7450.md (ppc7450-integer, ppc7450-compare): Adjust.
3350 * config/rs6000/7xx.md (ppc750-integer, ppc750-compare): Adjust.
3351 * config/rs6000/8540.md (ppc8540_su): Adjust.
3352 * config/rs6000/cell.md (cell-integer, cell-fast-cmp,
3353 cell-cmp-microcoded): Adjust.
3354 * config/rs6000/e300c2c3.md (ppce300c3_cmp, ppce300c3_iu): Adjust.
3355 * config/rs6000/e500mc.md (e500mc_su): Adjust.
3356 * config/rs6000/e500mc64.md (e500mc64_su, e500mc64_su2): Adjust.
3357 * config/rs6000/e5500.md (e5500_sfx, e5500_sfx2): Adjust.
3358 * config/rs6000/e6500.md (e6500_sfx, e6500_sfx2): Adjust.
3359 * config/rs6000/mpc.md (mpccore-integer, mpccore-compare): Adjust.
3360 * config/rs6000/power4.md (power4-integer, power4-cmp): Adjust.
3361 * config/rs6000/power5.md (power5-integer, power5-cmp): Adjust.
3362 * config/rs6000/power6.md (power6-integer, power6-fast-compare):
3364 * config/rs6000/power7.md (power7-integer, power7-cmp): Adjust.
3365 * config/rs6000/power8.md (power8-1cyc, power8-fast-compare):
3366 Adjust. Adjust comment.
3367 * config/rs6000/rs64.md (rs64a-integer, rs64a-compare): Adjust.
3368 * config/rs6000/titan.md (titan_fxu_adder, titan_fxu_alu): Adjust.
3370 2014-05-23 Segher Boessenkool <segher@kernel.crashing.org>
3372 * config/rs6000/rs6000.md (type): Add "add".
3373 (*add<mode>3_internal1, addsi3_high, *add<mode>3_internal2,
3374 *add<mode>3_internal3, *neg<mode>2_internal, and 5 anonymous
3375 define_insns): Use it.
3376 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
3378 * config/rs6000/40x.md (ppc403-integer, ppc403-compare): Adjust.
3379 * config/rs6000/440.md (ppc440-integer, ppc440-compare): Adjust.
3380 * config/rs6000/476.md (ppc476-simple-integer, ppc476-compare): Adjust.
3381 * config/rs6000/601.md (ppc601-integer): Adjust.
3382 * config/rs6000/603.md (ppc603-integer, ppc603-compare): Adjust.
3383 * config/rs6000/6xx.md (ppc604-integer, ppc604-compare): Adjust.
3384 * config/rs6000/7450.md (ppc7450-integer, ppc7450-compare): Adjust.
3385 * config/rs6000/7xx.md (ppc750-integer, ppc750-compare): Adjust.
3386 * config/rs6000/8540.md (ppc8540_su): Adjust.
3387 * config/rs6000/cell.md (cell-integer, cell-fast-cmp,
3388 cell-cmp-microcoded): Adjust.
3389 * config/rs6000/e300c2c3.md (ppce300c3_cmp, ppce300c3_iu): Adjust.
3390 * config/rs6000/e500mc.md (e500mc_su): Adjust.
3391 * config/rs6000/e500mc64.md (e500mc64_su, e500mc64_su2): Adjust.
3392 * config/rs6000/e5500.md (e5500_sfx, e5500_sfx2): Adjust.
3393 * config/rs6000/e6500.md (e6500_sfx, e6500_sfx2): Adjust.
3394 * config/rs6000/mpc.md (mpccore-integer, mpccore-compare): Adjust.
3395 * config/rs6000/power4.md (power4-integer, power4-cmp): Adjust.
3396 * config/rs6000/power5.md (power5-integer, power5-cmp): Adjust.
3397 * config/rs6000/power6.md (power6-integer, power6-fast-compare):
3399 * config/rs6000/power7.md (power7-integer, power7-cmp): Adjust.
3400 * config/rs6000/power8.md (power8-1cyc, power8-fast-compare): Adjust.
3401 * config/rs6000/rs64.md (rs64a-integer, rs64a-compare): Adjust.
3402 * config/rs6000/titan.md (titan_fxu_adder, titan_fxu_alu): Adjust.
3404 2014-05-23 Segher Boessenkool <segher@kernel.crashing.org>
3406 * config/rs6000/rs6000.md (type): Delete "var_shift_rotate",
3407 "delayed_compare", "var_delayed_compare".
3408 (var_shift): New attribute.
3409 (cell_micro): Adjust.
3410 (*andsi3_internal2_mc, *andsi3_internal3_mc, *andsi3_internal4,
3411 *andsi3_internal5_mc, *extzvsi_internal1, *extzvsi_internal2,
3412 rotlsi3, *rotlsi3_64, *rotlsi3_internal2, *rotlsi3_internal3,
3413 *rotlsi3_internal4, *rotlsi3_internal5, *rotlsi3_internal6,
3414 *rotlsi3_internal8le, *rotlsi3_internal8be, *rotlsi3_internal9le,
3415 *rotlsi3_internal9be, *rotlsi3_internal10le, *rotlsi3_internal10be,
3416 *rotlsi3_internal11le, *rotlsi3_internal11be, *rotlsi3_internal12le,
3417 *rotlsi3_internal12be, ashlsi3, *ashlsi3_64, lshrsi3, *lshrsi3_64,
3418 *lshiftrt_internal2le, *lshiftrt_internal2be, *lshiftrt_internal3le,
3419 *lshiftrt_internal3be, *lshiftrt_internal5le, *lshiftrt_internal5be,
3420 *lshiftrt_internal5le, *lshiftrt_internal5be, ashrsi3, *ashrsi3_64,
3421 rotldi3, *rotldi3_internal2, *rotldi3_internal3, *rotldi3_internal4,
3422 *rotldi3_internal5, *rotldi3_internal6, *rotldi3_internal7le,
3423 *rotldi3_internal7be, *rotldi3_internal8le, *rotldi3_internal8be,
3424 *rotldi3_internal9le, *rotldi3_internal9be, *rotldi3_internal10le,
3425 *rotldi3_internal10be, *rotldi3_internal11le, *rotldi3_internal11be,
3426 *rotldi3_internal12le, *rotldi3_internal12be, *rotldi3_internal13le,
3427 *rotldi3_internal13be, *rotldi3_internal14le, *rotldi3_internal14be,
3428 *rotldi3_internal15le, *rotldi3_internal15be, *ashldi3_internal1,
3429 *ashldi3_internal2, *ashldi3_internal3, *lshrdi3_internal1,
3430 *lshrdi3_internal2, *lshrdi3_internal3, *ashrdi3_internal1,
3431 *ashrdi3_internal2, *ashrdi3_internal3, *anddi3_internal2_mc,
3432 *anddi3_internal3_mc, as well as 11 anonymous define_insns): Adjust.
3433 * config/rs6000/rs6000.c (rs6000_adjust_cost, is_cracked_insn,
3434 insn_must_be_first_in_group, insn_must_be_last_in_group): Adjust.
3436 * config/rs6000/40x.md (ppc403-integer, ppc403-compare): Adjust.
3437 * config/rs6000/440.md (ppc440-integer): Adjust.
3438 * config/rs6000/476.md (ppc476-simple-integer, ppc476-compare): Adjust.
3439 * config/rs6000/601.md (ppc601-integer, ppc601-compare): Adjust.
3440 * config/rs6000/603.md (ppc603-integer, ppc603-compare): Adjust.
3441 * config/rs6000/6xx.md (ppc604-integer, ppc604-compare): Adjust.
3442 * config/rs6000/7450.md (ppc7450-integer, ppc7450-compare): Adjust.
3443 * config/rs6000/7xx.md (ppc750-integer, ppc750-compare): Adjust.
3444 * config/rs6000/8540.md (ppc8540_su): Adjust.
3445 * config/rs6000/cell.md (cell-integer, cell-fast-cmp,
3446 cell-cmp-microcoded): Adjust.
3447 * config/rs6000/e300c2c3.md (ppce300c3_cmp): Adjust.
3448 * config/rs6000/e500mc.md (e500mc_su): Adjust.
3449 * config/rs6000/e500mc64.md (e500mc64_su, e500mc64_su2,
3450 e500mc64_delayed): Adjust.
3451 * config/rs6000/e5500.md (e5500_sfx, e5500_delayed): Adjust.
3452 * config/rs6000/e6500.md (e6500_sfx, e6500_delayed): Adjust.
3453 * config/rs6000/mpc.md (mpccore-integer, mpccore-compare): Adjust.
3454 * config/rs6000/power4.md (power4-integer, power4-compare): Adjust.
3455 * config/rs6000/power5.md (power5-integer, power5-compare): Adjust.
3456 * config/rs6000/power6.md (power6-shift, power6-var-rotate,
3457 power6-delayed-compare, power6-var-delayed-compare): Adjust.
3458 * config/rs6000/power7.md (power7-integer, power7-compare): Adjust.
3459 * config/rs6000/power8.md (power8-1cyc, power8-compare): Adjust.
3461 * config/rs6000/rs64.md (rs64a-integer, rs64a-compare): Adjust.
3462 * config/rs6000/titan.md (titan_fxu_shift_and_rotate): Adjust.
3464 2014-05-23 Segher Boessenkool <segher@kernel.crashing.org>
3466 * config/rs6000/rs6000.md (type): Delete "idiv", "ldiv". Add "div".
3467 (bits): New mode_attr.
3468 (idiv_ldiv): Delete mode_attr.
3469 (udiv<mode>3, *div<mode>3, div<div_extend>_<mode>): Adjust.
3470 * config/rs6000/rs6000.c (rs6000_adjust_cost, is_cracked_insn,
3471 rs6000_adjust_priority, is_nonpipeline_insn,
3472 insn_must_be_first_in_group, insn_must_be_last_in_group): Adjust.
3474 * config/rs6000/40x.md (ppc403-idiv): Adjust.
3475 * config/rs6000/440.md (ppc440-idiv): Adjust.
3476 * config/rs6000/476.md (ppc476-idiv): Adjust.
3477 * config/rs6000/601.md (ppc601-idiv): Adjust.
3478 * config/rs6000/603.md (ppc603-idiv): Adjust.
3479 * config/rs6000/6xx.md (ppc604-idiv, ppc620-idiv, ppc630-idiv,
3480 ppc620-ldiv): Adjust.
3481 * config/rs6000/7450.md (ppc7450-idiv): Adjust.
3482 * config/rs6000/7xx.md (ppc750-idiv): Adjust.
3483 * config/rs6000/8540.md (ppc8540_divide): Adjust.
3484 * config/rs6000/a2.md (ppca2-idiv, ppca2-ldiv): Adjust.
3485 * config/rs6000/cell.md (cell-idiv, cell-ldiv): Adjust.
3486 * config/rs6000/e300c2c3.md (ppce300c3_divide): Adjust.
3487 * config/rs6000/e500mc.md (e500mc_divide): Adjust.
3488 * config/rs6000/e500mc64.md (e500mc64_divide): Adjust.
3489 * config/rs6000/e5500.md (e5500_divide, e5500_divide_d): Adjust.
3490 * config/rs6000/e6500.md (e6500_divide, e6500_divide_d): Adjust.
3491 * config/rs6000/mpc.md (mpccore-idiv): Adjust.
3492 * config/rs6000/power4.md (power4-idiv, power4-ldiv): Adjust.
3493 * config/rs6000/power5.md (power5-idiv, power5-ldiv): Adjust.
3494 * config/rs6000/power6.md (power6-idiv, power6-ldiv): Adjust.
3495 * config/rs6000/power7.md (power7-idiv, power7-ldiv): Adjust.
3496 * config/rs6000/power8.md (power8-idiv, power8-ldiv): Adjust.
3497 * config/rs6000/rs64.md (rs64a-idiv, rs64a-ldiv): Adjust.
3498 * config/rs6000/titan.md (titan_fxu_div): Adjust.
3500 2014-05-23 Segher Boessenkool <segher@kernel.crashing.org>
3502 * config/rs6000/rs6000.md (type): Delete "insert_word",
3503 "insert_dword". Add "insert".
3504 (size): Update comment.
3505 * config/rs6000/rs6000.c (rs6000_adjust_cost, is_cracked_insn,
3506 insn_must_be_first_in_group): Adjust.
3507 (insvsi_internal, *insvsi_internal1, *insvsi_internal2,
3508 *insvsi_internal3, *insvsi_internal4, *insvsi_internal5,
3509 *insvsi_internal6, insvdi_internal): Adjust.
3511 * config/rs6000/40x.md (ppc403-integer): Adjust.
3512 * config/rs6000/440.md (ppc440-integer): Adjust.
3513 * config/rs6000/476.md (ppc476-simple-integer): Adjust.
3514 * config/rs6000/601.md (ppc601-integer): Adjust.
3515 * config/rs6000/603.md (ppc603-integer): Adjust.
3516 * config/rs6000/6xx.md (ppc604-integer): Adjust.
3517 * config/rs6000/7450.md (ppc7450-integer): Adjust.
3518 * config/rs6000/7xx.md (ppc750-integer): Adjust.
3519 * config/rs6000/8540.md (ppc8540_su): Adjust.
3520 * config/rs6000/cell.md (cell-integer, cell-insert): Adjust.
3521 * config/rs6000/e300c2c3.md (ppce300c3_iu): Adjust.
3522 * config/rs6000/e500mc.md (e500mc_su): Adjust.
3523 * config/rs6000/e500mc64.md (e500mc64_su): Adjust.
3524 * config/rs6000/e5500.md (e5500_sfx): Adjust.
3525 * config/rs6000/e6500.md (e6500_sfx): Adjust.
3526 * config/rs6000/mpc.md (mpccore-integer): Adjust.
3527 * config/rs6000/power4.md (power4-integer, power4-insert): Adjust.
3528 * config/rs6000/power5.md (power5-integer, power5-insert): Adjust.
3529 * config/rs6000/power6.md (power6-insert, power6-insert-dword): Adjust.
3530 * config/rs6000/power7.md (power7-integer): Adjust.
3531 * config/rs6000/power8.md (power8-1cyc): Adjust.
3532 * config/rs6000/rs64.md (rs64a-integer): Adjust.
3533 * config/rs6000/titan.md (titan_fxu_shift_and_rotate): Adjust.
3535 2014-05-23 Segher Boessenkool <segher@kernel.crashing.org>
3537 * config/rs6000/rs6000.md (type): Add "mul". Delete "imul",
3538 "imul2", "imul3", "lmul", "imul_compare", "lmul_compare".
3539 (size): New attribute.
3540 (dot): New attribute.
3541 (cell_micro): Adjust.
3542 (mulsi3, *mulsi3_internal1, *mulsi3_internal2, mulsidi3,
3543 umulsidi3, smulsi3_highpart, umulsi3_highpart, muldi3,
3544 *muldi3_internal1, *muldi3_internal2, smuldi3_highpart,
3545 umuldi3_highpart): Adjust.
3546 * config/rs6000/rs6000.c (rs6000_adjust_cost, is_cracked_insn,
3547 rs6000_adjust_priority, is_nonpipeline_insn,
3548 insn_must_be_first_in_group, insn_must_be_last_in_group): Adjust.
3550 * config/rs6000/40x.md (ppc403-imul, ppc405-imul, ppc405-imul2,
3551 ppc405-imul3): Adjust.
3552 * config/rs6000/440.md (ppc440-imul, ppc440-imul2): Adjust.
3553 * config/rs6000/476.md (ppc476-imul): Adjust.
3554 * config/rs6000/601.md (ppc601-imul): Adjust.
3555 * config/rs6000/603.md (ppc603-imul, ppc603-imul2): Adjust.
3556 * config/rs6000/6xx.md (ppc604-imul, ppc604e-imul, ppc620-imul,
3557 ppc620-imul2, ppc620-imul3, ppc620-lmul): Adjust.
3558 * config/rs6000/7450.md (ppc7450-imul, ppc7450-imul2): Adjust.
3559 * config/rs6000/7xx.md (ppc750-imul, ppc750-imul2, ppc750-imul3):
3561 * config/rs6000/8540.md (ppc8540_multiply): Adjust.
3562 * config/rs6000/a2.md (ppca2-imul, ppca2-lmul): Adjust.
3563 * config/rs6000/cell.md (cell-lmul, cell-lmul-cmp, cell-imul23,
3565 * config/rs6000/e300c2c3.md (ppce300c3_multiply): Adjust.
3566 * config/rs6000/e500mc.md (e500mc_multiply): Adjust.
3567 * config/rs6000/e500mc64.md (e500mc64_multiply): Adjust.
3568 * config/rs6000/e5500.md (e5500_multiply, e5500_multiply_i): Adjust.
3569 * config/rs6000/e6500.md (e6500_multiply, e6500_multiply_i): Adjust.
3570 * config/rs6000/mpc.md (mpccore-imul): Adjust.
3571 * config/rs6000/power4.md (power4-lmul-cmp, power4-imul-cmp,
3572 power4-lmul, power4-imul, power4-imul3): Adjust.
3573 * config/rs6000/power5.md (power5-lmul-cmp, power5-imul-cmp,
3574 power5-lmul, power5-imul, power5-imul3): Adjust.
3575 * config/rs6000/power6.md (power6-lmul-cmp, power6-imul-cmp,
3576 power6-lmul, power6-imul, power6-imul3): Adjust.
3577 * config/rs6000/power7.md (power7-mul, power7-mul-compare): Adjust.
3578 * config/rs6000/power8.md (power8-mul, power8-mul-compare): Adjust.
3580 * config/rs6000/rs64.md (rs64a-imul, rs64a-imul2, rs64a-imul3,
3581 rs64a-lmul): Adjust.
3582 * config/rs6000/titan.md (titan_imul): Adjust.
3584 2014-05-23 Segher Boessenkool <segher@kernel.crashing.org>
3586 * config/rs6000/rs6000.md (type): Add new value "halfmul".
3587 (*macchwc, *macchw, *macchwuc, *macchwu, *machhwc, *machhw,
3588 *machhwuc, *machhwu, *maclhwc, *maclhw, *maclhwuc, *maclhwu,
3589 *nmacchwc, *nmacchw, *nmachhwc, *nmachhw, *nmaclhwc, *nmaclhw,
3590 *mulchwc, *mulchw, *mulchwuc, *mulchwu, *mulhhwc, *mulhhw,
3591 *mulhhwuc, *mulhhwu, *mullhwc, *mullhw, *mullhwuc, *mullhwu): Use it.
3592 * config/rs6000/40x.md (ppc405-imul3): Add type halfmul.
3593 * config/rs6000/440.md (ppc440-imul2): Add type halfmul.
3594 * config/rs6000/476.md (ppc476-imul): Add type halfmul.
3595 * config/rs6000/titan.md: Delete nonsensical comment.
3596 (titan_imul): Add type imul3.
3597 (titan_mulhw): Remove type imul3; add type halfmul.
3599 2014-05-23 Segher Boessenkool <segher@kernel.crashing.org>
3601 * config/rs6000/rs6000.md (type): Reorder, reformat.
3603 2014-05-23 Martin Jambor <mjambor@suse.cz>
3605 PR tree-optimization/53787
3606 * params.def (PARAM_IPA_MAX_AA_STEPS): New param.
3607 * ipa-prop.h (ipa_node_params): Rename uses_analysis_done to
3608 analysis_done, update all uses.
3609 * ipa-prop.c: Include domwalk.h
3610 (param_analysis_info): Removed.
3611 (param_aa_status): New type.
3612 (ipa_bb_info): Likewise.
3613 (func_body_info): Likewise.
3614 (ipa_get_bb_info): New function.
3615 (aa_overwalked): Likewise.
3616 (find_dominating_aa_status): Likewise.
3617 (parm_bb_aa_status_for_bb): Likewise.
3618 (parm_preserved_before_stmt_p): Changed to use new param AA info.
3619 (load_from_unmodified_param): Accept func_body_info as a parameter
3620 instead of parms_ainfo.
3621 (parm_ref_data_preserved_p): Changed to use new param AA info.
3622 (parm_ref_data_pass_through_p): Likewise.
3623 (ipa_load_from_parm_agg_1): Likewise. Update callers.
3624 (compute_complex_assign_jump_func): Changed to use new param AA info.
3625 (compute_complex_ancestor_jump_func): Likewise.
3626 (ipa_compute_jump_functions_for_edge): Likewise.
3627 (ipa_compute_jump_functions): Removed.
3628 (ipa_compute_jump_functions_for_bb): New function.
3629 (ipa_analyze_indirect_call_uses): Likewise, moved variable
3631 (ipa_analyze_virtual_call_uses): Accept func_body_info instead of node
3632 and info, moved variable declarations down.
3633 (ipa_analyze_call_uses): Accept and pass on func_body_info instead of
3635 (ipa_analyze_stmt_uses): Likewise.
3636 (ipa_analyze_params_uses): Removed.
3637 (ipa_analyze_params_uses_in_bb): New function.
3638 (ipa_analyze_controlled_uses): Likewise.
3639 (free_ipa_bb_info): Likewise.
3640 (analysis_dom_walker): New class.
3641 (ipa_analyze_node): Handle node-specific forbidden analysis,
3642 initialize and free func_body_info, use dominator walker.
3643 (ipcp_modif_dom_walker): New class.
3644 (ipcp_transform_function): Create and free func_body_info, use
3645 ipcp_modif_dom_walker, moved a lot of functionality there.
3647 2014-05-23 Marek Polacek <polacek@redhat.com>
3648 Jakub Jelinek <jakub@redhat.com>
3650 * builtins.def: Change SANITIZE_FLOAT_DIVIDE to SANITIZE_NONDEFAULT.
3651 * gcc.c (sanitize_spec_function): Likewise.
3652 * convert.c (convert_to_integer): Include "ubsan.h". Add
3653 floating-point to integer instrumentation.
3654 * doc/invoke.texi: Document -fsanitize=float-cast-overflow.
3655 * flag-types.h (enum sanitize_code): Add SANITIZE_FLOAT_CAST and
3656 SANITIZE_NONDEFAULT.
3657 * opts.c (common_handle_option): Handle -fsanitize=float-cast-overflow.
3658 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_FLOAT_CAST_OVERFLOW,
3659 BUILT_IN_UBSAN_HANDLE_FLOAT_CAST_OVERFLOW_ABORT): Add.
3660 * ubsan.c: Include "realmpfr.h" and "dfp.h".
3661 (get_ubsan_type_info_for_type): Handle REAL_TYPEs.
3662 (ubsan_type_descriptor): Set tkind to 0xffff for types other than
3663 float/double/long double.
3664 (ubsan_instrument_float_cast): New function.
3665 * ubsan.h (ubsan_instrument_float_cast): Declare.
3667 2014-05-23 Jiong Wang <jiong.wang@arm.com>
3669 * config/aarch64/predicates.md (aarch64_call_insn_operand): New
3671 * config/aarch64/constraints.md ("Ucs", "Usf"): New constraints.
3672 * config/aarch64/aarch64.md (*sibcall_insn, *sibcall_value_insn):
3673 Adjust for tailcalling through registers.
3674 * config/aarch64/aarch64.h (enum reg_class): New caller save
3676 (REG_CLASS_NAMES): Likewise.
3677 (REG_CLASS_CONTENTS): Likewise.
3678 * config/aarch64/aarch64.c (aarch64_function_ok_for_sibcall):
3679 Allow tailcalling without decls.
3681 2014-05-23 Thomas Schwinge <thomas@codesourcery.com>
3683 * gimplify.c (omp_notice_variable) <case OMP_CLAUSE_DEFAULT_NONE>:
3684 Rewrite check for ORT_PARALLEL and ORT_COMBINED_PARALLEL.
3686 * omp-low.c (expand_omp_for_static_chunk): Rename variable si to
3687 gsi, and variables v_* to v*.
3689 2014-05-23 Eric Botcazou <ebotcazou@adacore.com>
3691 * varasm.c (output_constructor_bitfield): Fix thinkos in latest change.
3693 2014-05-23 Thomas Schwinge <thomas@codesourcery.com>
3695 * gimple.h (enum gf_mask): Add and use GF_OMP_FOR_SIMD.
3696 * omp-low.c: Update accordingly.
3698 * gimple.h (enum gf_mask): Rewrite "<< 0" shift expressions used
3699 for GF_OMP_FOR_KIND_MASK, GF_OMP_FOR_KIND_FOR,
3700 GF_OMP_FOR_KIND_DISTRIBUTE, GF_OMP_FOR_KIND_SIMD,
3701 GF_OMP_FOR_KIND_CILKSIMD, GF_OMP_TARGET_KIND_MASK,
3702 GF_OMP_TARGET_KIND_REGION, GF_OMP_TARGET_KIND_DATA,
3703 GF_OMP_TARGET_KIND_UPDATE.
3705 * gimplify.c (omp_notice_variable) <case OMP_CLAUSE_DEFAULT_NONE>:
3706 Explicitly enumerate the expected region types.
3708 2014-05-23 Paul Eggert <eggert@cs.ucla.edu>
3711 * doc/extend.texi (Function Attributes): Fix __attribute__ ((malloc))
3712 documentation; the old documentation didn't clearly state the
3713 constraints on the contents of the pointed-to storage.
3715 2014-05-23 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
3717 Fix bootstrap error on ia64
3718 * config/ia64/ia64.c (ia64_first_cycle_multipass_dfa_lookahead_guard):
3719 Return default value.
3721 2014-05-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
3723 PR tree-optimization/54733
3724 * tree-ssa-math-opts.c (nop_stats): New "bswap_stats" structure.
3726 (find_bswap_or_nop_load): New.
3727 (find_bswap_1): Renamed to ...
3728 (find_bswap_or_nop_1): This. Also add support for memory source.
3729 (find_bswap): Renamed to ...
3730 (find_bswap_or_nop): This. Also add support for memory source and
3731 detection of bitwise operations equivalent to load in target
3733 (execute_optimize_bswap): Likewise. Also move its leading comment back
3734 in place and split statement transformation into ...
3735 (bswap_replace): This.
3737 2014-05-22 Vladimir Makarov <vmakarov@redhat.com>
3739 PR rtl-optimization/61215
3740 * lra-elelimination.c (lra_eliminate_regs_1): Don't use
3741 simplify_gen_subreg until final substitution.
3743 2014-05-23 Alan Modra <amodra@gmail.com>
3746 * config/rs6000/rs6000.c (mem_operand_gpr): Handle SImode.
3747 * config/rs6000/rs6000.md (extendsidi2_lfiwax, extendsidi2_nocell):
3748 Use "Y" constraint rather than "m".
3750 2014-05-23 Kugan Vivekanandarajah <kuganv@linaro.org>
3752 * config/aarch64/aarch64.c (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): New
3754 * config/aarch64/aarch64-protos.h (aarch64_atomic_assign_expand_fenv):
3755 New function declaration.
3756 * config/aarch64/aarch64-builtins.c (aarch64_builtins) : Add
3757 AARCH64_BUILTIN_GET_FPCR, AARCH64_BUILTIN_SET_FPCR.
3758 AARCH64_BUILTIN_GET_FPSR and AARCH64_BUILTIN_SET_FPSR.
3759 (aarch64_init_builtins) : Initialize builtins
3760 __builtins_aarch64_set_fpcr, __builtins_aarch64_get_fpcr.
3761 __builtins_aarch64_set_fpsr and __builtins_aarch64_get_fpsr.
3762 (aarch64_expand_builtin) : Expand builtins __builtins_aarch64_set_fpcr
3763 __builtins_aarch64_get_fpcr, __builtins_aarch64_get_fpsr,
3764 and __builtins_aarch64_set_fpsr.
3765 (aarch64_atomic_assign_expand_fenv): New function.
3766 * config/aarch64/aarch64.md (set_fpcr): New pattern.
3767 (get_fpcr) : Likewise.
3768 (set_fpsr) : Likewise.
3769 (get_fpsr) : Likewise.
3770 (unspecv): Add UNSPECV_GET_FPCR and UNSPECV_SET_FPCR, UNSPECV_GET_FPSR
3771 and UNSPECV_SET_FPSR.
3772 * doc/extend.texi (AARCH64 Built-in Functions) : Document
3773 __builtins_aarch64_set_fpcr, __builtins_aarch64_get_fpcr.
3774 __builtins_aarch64_set_fpsr and __builtins_aarch64_get_fpsr.
3776 2014-05-22 Vladimir Makarov <vmakarov@redhat.com>
3778 PR rtl-optimization/60969
3779 * ira-costs.c (record_reg_classes): Process NO_REGS for matching
3780 constraints. Set up mem cost for NO_REGS case.
3782 2014-05-22 Thomas Schwinge <thomas@codesourcery.com>
3784 * builtin-types.def: Simplify examples for DEF_FUNCTION_TYPE_*.
3786 2012-05-22 Bernd Schmidt <bernds@codesourcery.com>
3788 * config/darwin.c: Include "lto-section-names.h".
3789 (LTO_SEGMENT_NAME): Don't define.
3790 * config/i386/winnt.c: Include "lto-section-names.h".
3791 * lto-streamer.c: Include "lto-section-names.h".
3792 * lto-streamer.h (LTO_SECTION_NAME_PREFIX): Don't define.
3793 * lto-wrapper.c: Include "lto-section-names.h".
3794 (LTO_SECTION_NAME_PREFIX): Don't define.
3795 * lto-section-names.h: New file.
3796 * cgraphunit.c: Include "lto-section-names.h".
3798 2014-05-22 Peter Bergner <bergner@vnet.ibm.com>
3800 * config/rs6000/htm.md (ttest): Use correct shift value to get CR0.
3802 2014-05-22 Richard Earnshaw <rearnsha@arm.com>
3805 * arm.md (arm_cmpdi_unsigned): Fix length calculation for Thumb2.
3807 2014-05-22 Nick Clifton <nickc@redhat.com>
3809 * config/msp430/msp430.h (ASM_SPEC): Add spaces after inserted options.
3811 2014-05-22 Eric Botcazou <ebotcazou@adacore.com>
3813 * tree-ssa-forwprop.c (associate_plusminus): Extend (T)(P + A) - (T)P
3814 -> (T)A transformation to integer types.
3816 2014-05-22 Teresa Johnson <tejohnson@google.com>
3818 * gcov-io.c (gcov_position): Use gcov_nonruntime_assert.
3819 (gcov_is_error): Remove gcc_assert from IN_LIBGCOV code.
3820 (gcov_rewrite): Use gcov_nonruntime_assert.
3822 (gcov_write_words): Ditto.
3823 (gcov_write_length): Ditto.
3824 (gcov_read_words): Use gcov_nonruntime_assert, and remove
3825 gcc_assert from IN_LIBGCOV code.
3826 (gcov_read_summary): Use gcov_error to flag profile corruption.
3827 (gcov_sync): Use gcov_nonruntime_assert.
3828 (gcov_seek): Remove gcc_assert from IN_LIBGCOV code.
3829 (gcov_histo_index): Use gcov_nonruntime_assert.
3830 (static void gcov_histogram_merge): Ditto.
3831 (compute_working_sets): Ditto.
3832 * gcov-io.h (gcov_nonruntime_assert): Define.
3833 (gcov_error): Define for !IN_LIBGCOV
3835 2014-05-22 Richard Biener <rguenther@suse.de>
3837 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
3838 BUILT_IN_REALLOC like BUILT_IN_STRDUP.
3839 (call_may_clobber_ref_p_1): Handle BUILT_IN_REALLOC as allocation
3840 and deallocation site.
3841 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3842 Handle BUILT_IN_REALLOC similar to BUILT_IN_STRDUP with also
3843 passing through the incoming points-to set.
3844 (handle_lhs_call): Use flags argument instead of recomputing it.
3845 (find_func_aliases_for_call): Call handle_lhs_call with proper
3848 2014-05-22 Jakub Jelinek <jakub@redhat.com>
3850 * tree-streamer-in.c (unpack_ts_real_cst_value_fields): Make sure
3851 all padding bits in REAL_VALUE_TYPE are cleared.
3853 2014-05-22 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
3855 Cleanup and improve multipass_dfa_lookahead_guard
3856 * config/i386/i386.c (core2i7_first_cycle_multipass_filter_ready_try,)
3857 (core2i7_first_cycle_multipass_begin,)
3858 (core2i7_first_cycle_multipass_issue,)
3859 (core2i7_first_cycle_multipass_backtrack): Update signature.
3860 * config/ia64/ia64.c
3861 (ia64_first_cycle_multipass_dfa_lookahead_guard_spec): Remove.
3862 (ia64_first_cycle_multipass_dfa_lookahead_guard): Update signature.
3863 (TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC): Remove
3865 (ia64_first_cycle_multipass_dfa_lookahead_guard): Merge logic from
3866 ia64_first_cycle_multipass_dfa_lookahead_guard_spec. Update return
3868 * config/rs6000/rs6000.c (rs6000_use_sched_lookahead_guard): Update
3870 * doc/tm.texi: Regenerate.
3872 (TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC): Remove.
3873 * haifa-sched.c (ready_try): Make signed to allow negative values.
3874 (rebug_ready_list_1): Update.
3875 (choose_ready): Simplify.
3876 (sched_extend_ready_list): Update.
3878 2014-05-22 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
3880 Remove IA64 speculation tweaking flags
3881 * config/ia64/ia64.c (ia64_set_sched_flags): Delete handling of
3882 speculation tuning flags.
3883 (msched-prefer-non-data-spec-insns,)
3884 (msched-prefer-non-control-spec-insns): Obsolete options.
3885 * haifa-sched.c (choose_ready): Remove handling of
3886 PREFER_NON_CONTROL_SPEC and PREFER_NON_DATA_SPEC.
3887 * sched-int.h (enum SPEC_SCHED_FLAGS): Remove PREFER_NON_CONTROL_SPEC
3888 and PREFER_NON_DATA_SPEC.
3889 * sel-sched.c (process_spec_exprs): Remove handling of
3890 PREFER_NON_CONTROL_SPEC and PREFER_NON_DATA_SPEC.
3892 2014-05-22 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
3894 Improve scheduling debug output
3895 * haifa-sched.c (debug_ready_list): Remove unnecessary prototype.
3896 (advance_one_cycle): Update.
3897 (schedule_insn, queue_to_ready): Add debug printouts.
3898 (debug_ready_list_1): New static function.
3899 (debug_ready_list): Update.
3900 (max_issue): Add debug printouts.
3901 (dump_insn_stream): New static function.
3902 (schedule_block): Use it. Also better indent printouts.
3904 2014-05-22 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
3906 Fix sched_insn debug counter
3907 * haifa-sched.c (schedule_insn): Update.
3908 (struct haifa_saved_data): Add nonscheduled_insns_begin.
3909 (save_backtrack_point, restore_backtrack_point): Update.
3910 (first_nonscheduled_insn): New static function.
3911 (queue_to_ready, choose_ready): Use it.
3912 (schedule_block): Init nonscheduled_insns_begin.
3913 (sched_emit_insn): Update.
3916 2014-05-22 Kugan Vivekanandarajah <kuganv@linaro.org>
3918 * config/aarch64/aarch64.c (aarch64_regno_regclass) : Change CORE_REGS
3920 (aarch64_secondary_reload) : LikeWise.
3921 (aarch64_class_max_nregs) : Remove CORE_REGS.
3922 * config/aarch64/aarch64.h (enum reg_class) : Remove CORE_REGS.
3923 (REG_CLASS_NAMES) : Likewise.
3924 (REG_CLASS_CONTENTS) : LikeWise.
3925 (INDEX_REG_CLASS) : Change CORE_REGS to GENERAL_REGS.
3927 2014-05-21 Guozhi Wei <carrot@google.com>
3930 * config/aarch64/arm_neon.h (vqdmulh_n_s16): Change the last operand's
3932 (vqdmulhq_n_s16): Likewise.
3934 2014-05-21 Segher Boessenkool <segher@kernel.crashing.org>
3936 * config/rs6000/predicates.md (update_indexed_address_mem): Delete.
3938 2014-05-21 Marek Polacek <polacek@redhat.com>
3941 * ubsan.c (is_ubsan_builtin_p): Turn assert into a condition.
3943 2014-05-21 Martin Jambor <mjambor@suse.cz>
3945 * doc/invoke.texi (Optimize Options): Document parameters
3946 ipa-cp-eval-threshold, ipa-max-agg-items, ipa-cp-loop-hint-bonus and
3947 ipa-cp-array-index-hint-bonus.
3949 2014-05-21 Mark Wielaard <mjw@redhat.com>
3952 * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_type if DWARF
3953 version >= 3 or not strict DWARF.
3954 * langhooks.h (struct lang_hooks_for_types): Add
3955 enum_underlying_base_type.
3956 * langhooks.c (lhd_enum_underlying_base_type): New function.
3957 * gcc/langhooks.h (struct lang_hooks_for_types): Add
3958 enum_underlying_base_type.
3959 * langhooks-def.h (lhd_enum_underlying_base_type): New declaration.
3960 (LANG_HOOKS_ENUM_UNDERLYING_BASE_TYPE): New define.
3961 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add new lang hook.
3963 2014-05-21 Richard Biener <rguenther@suse.de>
3965 * doc/invoke.texi (-flto-partition=): Document one and none algorithms.
3967 2014-05-21 John Marino <gnugcc@marino.st>
3969 * config.gcc (*-*-dragonfly*): New target.
3970 * configure.ac: Detect dl_iterate_phdr (*freebsd*, *dragonfly*).
3971 * configure: Regenerate.
3972 * config/dragonfly-stdint.h: New.
3973 * config/dragonfly.h: New.
3974 * config/dragonfly.opt: New.
3975 * config/i386/dragonfly.h: New.
3976 * ginclude/stddef.h: Detect _PTRDIFF_T_DECLARED for DragonFly.
3978 2014-05-21 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3980 * tree.def (VOID_CST): New.
3981 * tree-core.h (TI_VOID): New.
3982 * tree.h (void_node): New.
3983 * tree.c (tree_node_structure_for_code, tree_code_size)
3984 (iterative_hash_expr): Handle VOID_CST.
3985 (build_common_tree_nodes): Initialize void_node.
3987 2014-05-21 Bernd Schmidt <bernds@codesourcery.com>
3989 * reload1.c (remove_init_insns, will_delete_init_insn_p): New static
3991 (reload, calculate_needs_all_insns, reload_as_needed): Use them.
3993 * config/bfin/bfin.c (split_load_immediate): Use gen_int_mode in a few
3996 * cfgrtl.c (cfg_layout_initialize): Weaken assert to only trigger if
3997 flag_reorder_blocks_and_partition.
3998 * hw-doloop.c (reorg_loops): Avoid reordering if that flag is set.
4000 2014-05-21 Oleg Endo <olegendo@gcc.gnu.org>
4003 * config/sh/sh.md (*addc_r_1): Rename to addc_t_r. Remove empty
4005 (*addc_r_t): Add new insn_and_split.
4007 2014-05-21 Jakub Jelinek <jakub@redhat.com>
4010 * omp-low.c (handle_simd_reference): New function.
4011 (lower_rec_input_clauses): Use it. Defer adding reference
4012 initialization even for reduction without placeholder if in simd,
4013 handle it properly later on.
4015 2014-05-20 Jan Hubicka <hubicka@ucw.cz>
4017 PR tree-optimization/60899
4018 * gimple-fold.c (can_refer_decl_in_current_unit_p): Cleanup;
4019 assume all static symbols will have definition wile parsing and
4020 check the do have definition later in compilation; check that
4021 variable referring symbol will be output before concluding that
4022 reference is safe; be conservative for referring local statics;
4023 be more precise about when comdat is output in other partition.
4025 2014-05-20 Jan Hubicka <hubicka@ucw.cz>
4028 * ipa-inline-transform.c (inline_call): Use add CALLEE_REMOVED
4030 * ipa-inline.c (inline_to_all_callers): If callee was removed; return.
4031 (ipa_inline): Loop inline_to_all_callers until no more aliases
4034 2014-05-20 Jan Hubicka <hubicka@ucw.cz>
4036 * ipa.c (ipa_discover_readonly_nonaddressable_var): Fix dumping;
4037 set writeonly flag only for vars actually written to.
4039 2014-05-20 Dehao Chen <dehao@google.com>
4041 * ipa-inline-transform.c (clone_inlined_nodes): Use min of edge count
4042 and callee count to get clone count.
4043 * tree-inline.c (expand_call_inline): Use callee count instead of bb
4046 2014-05-20 Richard Sandiford <rdsandiford@googlemail.com>
4048 PR rtl-optimization/61243
4049 * emit-rtl.c (emit_copy_of_insn_after): Copy CROSSING_JUMP_P.
4051 2014-05-20 Xinliang David Li <davidxl@google.com>
4053 * cgraphunit.c (walk_polymorphic_call_targets): Add
4054 dbgcnt and fopt-info support.
4055 * ipa-prop.c (ipa_make_edge_direct_to_target): Ditto.
4056 * ipa-devirt.c (ipa_devirt): Ditto.
4057 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children): Ditto.
4058 * ipa.c (walk_polymorphic_call_targets): Ditto.
4059 * gimple-fold.c (fold_gimple_assign): Ditto.
4060 (gimple_fold_call): Ditto.
4061 * dbgcnt.def: New counter.
4063 2014-05-20 DJ Delorie <dj@redhat.com>
4065 * config/msp430/msp430.md (split): Don't allow subregs when
4066 splitting SImode adds.
4067 (andneghi): Fix subtraction logic.
4068 * config/msp430/predicates.md (msp430_nonsubreg_or_imm_operand): New.
4070 2014-05-20 Jan Hubicka <hubicka@ucw.cz>
4072 * tree.h (DECL_ONE_ONLY): Return true only for externally visible
4074 * except.c (switch_to_exception_section, resolve_unique_section,
4075 get_named_text_section, default_function_rodata_section,
4076 align_variable, get_block_for_decl, default_section_type_flags):
4077 Use DECL_COMDAT_GROUP instead of DECL_ONE_ONLY.
4078 * symtab.c (symtab_add_to_same_comdat_group,
4079 symtab_make_decl_local, fixup_same_cpp_alias_visibility,
4080 symtab_nonoverwritable_alias, symtab_get_symbol_partitioning_class):
4082 * cgraphclones.c (cgraph_create_virtual_clone): Likewise.
4083 * bb-reorder.c (pass_partition_blocks::gate): Likewise.
4084 * config/c6x/c6x.c (c6x_elf_unique_section): Likewise.
4085 (c6x_function_in_section_p): Likewise.
4086 * config/darwin.c (machopic_select_section): Likewise.
4087 * config/arm/arm.c (arm_function_in_section_p): Likewise.
4088 * config/mips/mips.c (mips_function_rodata_section): Likewise.
4089 * config/mep/mep.c (mep_select_section): LIkewise.
4090 * config/i386/i386.c (x86_64_elf_unique_section): Likewise.
4092 2014-05-20 Eric Botcazou <ebotcazou@adacore.com>
4094 * tree-ssa-dom.c (hashable_expr_equal_p) <EXPR_CALL>: Also compare the
4095 EH region of calls to pure functions that can throw an exception.
4096 * tree-ssa-sccvn.c (vn_reference_eq): Remove duplicated test.
4097 (copy_reference_ops_from_call): Also copy the EH region of the call if
4098 it can throw an exception.
4100 2014-05-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4102 * simplify-rtx.c (simplify_binary_operation_1): Optimize case of
4103 nested VEC_SELECTs that are inverses of each other.
4105 2014-05-20 Richard Biener <rguenther@suse.de>
4107 * tree-ssa-sccvn.c (process_scc): Dump SCC here, when iterating,
4108 (extract_and_process_scc_for_name): not here.
4109 (cond_dom_walker::before_dom_children): Only process
4110 stmts that end the BB in interesting ways.
4111 (run_scc_vn): Mark param uses as visited.
4113 2014-05-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4115 * config/arm/arm.md (arith_shiftsi): Do not predicate for
4118 2014-05-20 Nick Clifton <nickc@redhat.com>
4120 * config/msp430/msp430.c (TARGET_GIMPLIFY_VA_ARG_EXPR): Define.
4121 (msp430_gimplify_va_arg_expr): New function.
4122 (msp430_print_operand): Handle (CONST (ZERO_EXTRACT)).
4124 * config/msp430/msp430.md (zero_extendpsisi2): Use + constraint on
4125 operand 0 in order to prevent confusion about the number of
4128 2014-05-20 Richard Biener <rguenther@suse.de>
4130 PR tree-optimization/61221
4131 * tree-ssa-pre.c (el_to_update): Remove.
4132 (eliminate_dom_walker::before_dom_children): Handle released
4133 VDEFs by value-numbering them to the associated VUSE. Update
4134 stmt immediately for substituted call address.
4135 (eliminate): Remove delayed stmt updating code.
4136 * tree-ssa-sccvn.c (vuse_ssa_val): New function valueizing
4137 possibly late re-numbered vuses.
4138 (vn_reference_lookup_2): Adjust.
4139 (vn_reference_lookup_pieces): Likewise.
4140 (vn_reference_lookup): Likewise.
4142 2014-05-20 Richard Biener <rguenther@suse.de>
4144 * config.gcc: Remove need_64bit_hwint.
4145 * configure.ac: Do not define NEED_64BIT_HOST_WIDE_INT.
4146 * hwint.h: Do not check NEED_64BIT_HOST_WIDE_INT but assume
4148 * config.in: Regenerate.
4149 * configure: Likewise.
4151 2014-05-19 David Wohlferd <dw@LimeGreenSocks.com>
4153 * doc/extend.texi: Create Label Attributes section,
4154 move all label attributes into it and reference it.
4156 2014-05-19 Richard Earnshaw <rearnsha@arm.com>
4158 * arm.c (thumb1_reorg): When scanning backwards skip anything
4159 that's not a proper insn.
4161 2014-05-19 Richard Biener <rguenther@suse.de>
4163 PR tree-optimization/61221
4164 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4165 Do nothing for unreachable blocks.
4166 * tree-ssa-sccvn.c (cond_dom_walker::before_dom_children):
4167 Improve unreachability detection.
4169 2014-05-19 Richard Biener <rguenther@suse.de>
4171 PR tree-optimization/61209
4172 * tree-ssa-sccvn.c (visit_phi): Avoid setting expr to VN_TOP.
4174 2014-05-19 Nick Clifton <nickc@redhat.com>
4176 * except.c (init_eh): Fix computation of builtin setjmp buffer
4177 size to allow for targets where POINTER_SIZE > BITS_PER_WORD.
4179 2014-05-19 Richard Biener <rguenther@suse.de>
4181 PR tree-optimization/61184
4182 * tree-vrp.c (is_negative_overflow_infinity): Use
4183 TREE_OVERFLOW_P and do that check first.
4184 (is_positive_overflow_infinity): Likewise.
4185 (is_overflow_infinity): Likewise.
4186 (vrp_operand_equal_p): Properly treat operands with
4187 differing overflow as not equal.
4189 2014-05-19 Bernd Schmidt <bernds@codesourcery.com>
4191 * simplify-rtx.c (simplify_unary_operation_1): Use CONST_INT_P in
4192 shift simplification where it was intended.
4194 2014-05-19 Christian Bruel <christian.bruel@st.com>
4197 * config/sh/sh.md (movsf_ie): Unset fp_mode for fmov.
4199 2014-05-19 Richard Sandiford <r.sandiford@uk.ibm.com>
4202 * config/sparc/sparc.c (sparc_fold_builtin): Use widest_int rather
4205 2014-05-19 Richard Sandiford <rdsandiford@googlemail.com>
4207 * reg-notes.def (CROSSING_JUMP): Likewise.
4208 * rtl.h (rtx_def): Update comment for jump flag.
4209 (CROSSING_JUMP_P): Define.
4210 * cfgcleanup.c (try_forward_edges, try_optimize_cfg): Use it instead
4211 of a REG_CROSSING_JUMP note.
4212 * cfghooks.c (tidy_fallthru_edges): Likewise.
4213 * cfgrtl.c (fixup_partition_crossing, rtl_verify_edges): Likewise.
4214 * emit-rtl.c (try_split): Likewise.
4215 * haifa-sched.c (sched_create_recovery_edges): Likewise.
4216 * ifcvt.c (find_if_case_1, find_if_case_2): Likewise.
4217 * jump.c (redirect_jump_2): Likewise.
4218 * reorg.c (follow_jumps, fill_slots_from_thread): Likewise.
4219 (relax_delay_slots): Likewise.
4220 * config/arc/arc.md (jump_i, cbranchsi4_scratch, *bbit): Likewise.
4221 (bbit_di): Likewise.
4222 * config/arc/arc.c (arc_reorg, arc_can_follow_jump): Likewise.
4223 * config/sh/sh.md (jump_compact): Likewise.
4224 * bb-reorder.c (rotate_loop): Likewise.
4225 (pass_duplicate_computed_gotos::execute): Likewise.
4226 (add_reg_crossing_jump_notes): Rename to...
4227 (update_crossing_jump_flags): ...this.
4228 (pass_partition_blocks::execute): Update accordingly.
4230 2014-05-19 Richard Sandiford <rdsandiford@googlemail.com>
4232 * tree.h: Remove extraneous template <>.
4234 2014-05-17 Jan Hubicka <hubicka@ucw.cz>
4236 * ipa.c (symtab_remove_unreachable_nodes): Remove
4237 symbol from comdat group if its body was eliminated.
4238 (comdat_can_be_unshared_p_1): Static symbols can always be privatized.
4239 * symtab.c (symtab_remove_from_same_comdat_group): Break out from ...
4240 (symtab_unregister_node): ... this one.
4241 (verify_symtab_base): More strict checking of comdats.
4242 * cgraph.h (symtab_remove_from_same_comdat_group): Declare.
4244 2014-05-17 Jan Hubicka <hubicka@ucw.cz>
4246 * tree-pass.h (make_pass_ipa_comdats): New pass.
4247 * timevar.def (TV_IPA_COMDATS): New timevar.
4248 * passes.def (pass_ipa_comdats): Add.
4249 * Makefile.in (OBJS): Add ipa-comdats.o
4250 * ipa-comdats.c: New file.
4252 2014-05-17 Jan Hubicka <hubicka@ucw.cz>
4254 * ipa.c (update_visibility_by_resolution_info): New function.
4255 (function_and_variable_visibility): Use it.
4257 2014-05-17 Jan Hubicka <hubicka@ucw.cz>
4259 * cgraph.h (symtab_first_defined_symbol, symtab_next_defined_symbol):
4261 (FOR_EACH_DEFINED_SYMBOL): New macro.
4262 (varpool_first_static_initializer, varpool_next_static_initializer,
4263 varpool_first_defined_variable, varpool_next_defined_variable):
4265 (symtab_in_same_comdat_p): Correctly deal with inline functions.
4267 2014-05-17 Trevor Saunders <tsaunders@mozilla.com>
4269 * ggc-page.c (ggc_handle_finalizers): Add comment.
4271 2014-05-17 Trevor Saunders <tsaunders@mozilla.com>
4273 * ggc-common.c (ggc_internal_cleared_alloc): Adjust.
4274 * ggc-none.c (ggc_internal_alloc): Assert if a finalizer is passed.
4275 (ggc_internal_cleared_alloc): Likewise.
4276 * ggc-page.c (finalizer): New class.
4277 (vec_finalizer): Likewise.
4278 (globals::finalizers): New member.
4279 (globals::vec_finalizers): Likewise.
4280 (ggc_internal_alloc): Record the finalizer if any for the block being
4282 (ggc_handle_finalizers): New function.
4283 (ggc_collect): Call ggc_handle_finalizers.
4284 * ggc.h (ggc_internal_alloc): Add arguments to allow installing a
4286 (ggc_internal_cleared_alloc): Likewise.
4287 (finalize): New function.
4288 (need_finalization_p): Likewise.
4289 (ggc_alloc): Install the type's destructor as the finalizer if it
4291 (ggc_cleared_alloc): Likewise.
4292 (ggc_vec_alloc): Likewise.
4293 (ggc_cleared_vec_alloc): Likewise.
4295 2014-05-17 Trevor Saunders <tsaunders@mozilla.com>
4297 * ggc.h (ggc_alloc_cleared_simd_clone_stat): Remove function.
4299 2014-05-17 Trevor Saunders <tsaunders@mozilla.com>
4301 * alias.c (record_alias_subset): Adjust.
4302 * bitmap.c (bitmap_element_allocate): Likewise.
4303 (bitmap_gc_alloc_stat): Likewise.
4304 * cfg.c (init_flow): Likewise.
4305 (alloc_block): Likewise.
4306 (unchecked_make_edge): Likewise.
4307 * cfgloop.c (alloc_loop): Likewise.
4308 (flow_loops_find): Likewise.
4309 (rescan_loop_exit): Likewise.
4310 * cfgrtl.c (init_rtl_bb_info): Likewise.
4311 * cgraph.c (insert_new_cgraph_node_version): Likewise.
4312 (cgraph_allocate_node): Likewise.
4313 (cgraph_create_edge_1): Likewise.
4314 (cgraph_allocate_init_indirect_info): Likewise.
4315 * cgraphclones.c (cgraph_clone_edge): Likewise.
4316 * cgraphunit.c (add_asm_node): Likewise.
4317 (init_lowered_empty_function): Likewise.
4318 * config/aarch64/aarch64.c (aarch64_init_machine_status): Likewise.
4319 * config/alpha/alpha.c (alpha_init_machine_status): Likewise.
4320 (alpha_use_linkage): Likewise.
4321 * config/arc/arc.c (arc_init_machine_status): Likewise.
4322 * config/arm/arm.c (arm_init_machine_status): Likewise.
4323 * config/avr/avr.c (avr_init_machine_status): Likewise.
4324 * config/bfin/bfin.c (bfin_init_machine_status): Likewise.
4325 * config/c6x/c6x.c (c6x_init_machine_status): Likewise.
4326 * config/cris/cris.c (cris_init_machine_status): Likewise.
4327 * config/darwin.c (machopic_indirection_name): Likewise.
4328 (darwin_build_constant_cfstring): Likewise.
4329 (darwin_enter_string_into_cfstring_table): Likewise.
4330 * config/epiphany/epiphany.c (epiphany_init_machine_status): Likewise.
4331 * config/frv/frv.c (frv_init_machine_status): Likewise.
4332 * config/i386/i386.c (get_dllimport_decl): Likewise.
4333 (ix86_init_machine_status): Likewise.
4334 (assign_386_stack_local): Likewise.
4335 * config/i386/winnt.c (i386_pe_record_external_function): Likewise.
4336 (i386_pe_maybe_record_exported_symbol): Likewise.
4337 (i386_pe_record_stub): Likewise.
4338 * config/ia64/ia64.c (ia64_init_machine_status): Likewise.
4339 * config/iq2000/iq2000.c (iq2000_init_machine_status): Likewise.
4340 * config/m32c/m32c.c (m32c_init_machine_status): Likewise.
4341 (m32c_note_pragma_address): Likewise.
4342 * config/mep/mep.c (mep_init_machine_status): Likewise.
4343 (mep_note_pragma_flag): Likewise.
4344 * config/mips/mips.c (mflip_mips16_use_mips16_p): Likewise.
4345 (mips16_local_alias): Likewise.
4346 (mips_init_machine_status): Likewise.
4347 * config/mmix/mmix.c (mmix_init_machine_status): Likewise.
4348 * config/moxie/moxie.c (moxie_init_machine_status): Likewise.
4349 * config/msp430/msp430.c (msp430_init_machine_status): Likewise.
4350 * config/nds32/nds32.c (nds32_init_machine_status): Likewise.
4351 * config/nios2/nios2.c (nios2_init_machine_status): Likewise.
4352 * config/pa/pa.c (pa_init_machine_status): Likewise.
4353 (pa_get_deferred_plabel): Likewise.
4354 * config/rl78/rl78.c (rl78_init_machine_status): Likewise.
4355 * config/rs6000/rs6000.c (builtin_function_type): Likewise.
4356 (rs6000_init_machine_status): Likewise.
4357 (output_toc): Likewise.
4358 * config/s390/s390.c (s390_init_machine_status): Likewise.
4359 * config/score/score.c (score_output_external): Likewise.
4360 * config/sparc/sparc.c (sparc_init_machine_status): Likewise.
4361 * config/spu/spu.c (spu_init_machine_status): Likewise.
4362 * config/tilegx/tilegx.c (tilegx_init_machine_status): Likewise.
4363 * config/tilepro/tilepro.c (tilepro_init_machine_status): Likewise.
4364 * config/xtensa/xtensa.c (xtensa_init_machine_status): Likewise.
4365 * coverage.c (coverage_end_function): Likewise.
4366 * dbxout.c (dbxout_init): Likewise.
4367 * doc/gty.texi: Don't mention variable_size attribute.
4368 * dwarf2cfi.c (new_cfi): Adjust.
4369 (new_cfi_row): Likewise.
4370 (copy_cfi_row): Likewise.
4371 (create_cie_data): Likewise.
4372 * dwarf2out.c (dwarf2out_alloc_current_fde): Likewise.
4373 (new_loc_descr): Likewise.
4374 (find_AT_string_in_table): Likewise.
4375 (add_addr_table_entry): Likewise.
4376 (new_die): Likewise.
4377 (add_var_loc_to_decl): Likewise.
4378 (clone_die): Likewise.
4379 (clone_as_declaration): Likewise.
4380 (break_out_comdat_types): Likewise.
4381 (new_loc_list): Likewise.
4382 (add_loc_descr_to_each): Likewise.
4383 (add_location_or_const_value_attribute): Likewise.
4384 (add_linkage_name): Likewise.
4385 (lookup_filename): Likewise.
4386 (dwarf2out_var_location): Likewise.
4387 (new_line_info_table): Likewise.
4388 (dwarf2out_init): Likewise.
4389 (mem_loc_descriptor): Likewise.
4390 (loc_descriptor): Likewise.
4391 (add_const_value_attribute): Likewise.
4392 (tree_add_const_value_attribute): Likewise.
4393 (comp_dir_string): Likewise.
4394 (dwarf2out_vms_debug_main_pointer): Likewise.
4395 (string_cst_pool_decl): Likewise.
4396 * emit-rtl.c (set_mem_attrs): Likewise.
4397 (get_reg_attrs): Likewise.
4398 (start_sequence): Likewise.
4399 (init_emit): Likewise.
4400 (init_emit_regs): Likewise.
4401 * except.c (init_eh_for_function): Likewise.
4402 (gen_eh_region): Likewise.
4403 (gen_eh_region_catch): Likewise.
4404 (gen_eh_landing_pad): Likewise.
4405 (add_call_site): Likewise.
4406 * function.c (add_frame_space): Likewise.
4407 (insert_temp_slot_address): Likewise.
4408 (assign_stack_temp_for_type): Likewise.
4409 (get_hard_reg_initial_val): Likewise.
4410 (allocate_struct_function): Likewise.
4411 (prepare_function_start): Likewise.
4412 (types_used_by_var_decl_insert): Likewise.
4413 * gengtype.c (variable_size_p): Remove function.
4414 (enum alloc_quantity): Remove enum.
4415 (write_typed_alloc_def): Remove function.
4416 (write_typed_struct_alloc_def): Likewise.
4417 (write_typed_typedef_alloc_def): Likewise.
4418 (write_typed_alloc_defns): Likewise.
4420 * ggc-common.c (ggc_cleared_alloc_htab_ignore_args): Adjust.
4421 (ggc_cleared_alloc_ptr_array_two_args): Likewise.
4422 * ggc.h (ggc_alloc): new function.
4423 (ggc_cleared_alloc): Likewise.
4424 (ggc_vec_alloc): Template on type of vector element, and remove
4425 element size argument.
4426 (ggc_cleared_vec_alloc): Likewise.
4427 * gimple.c (gimple_build_omp_for): Adjust.
4428 (gimple_copy): Likewise.
4429 * ipa-cp.c (get_replacement_map): Likewise.
4430 (find_aggregate_values_for_callers_subset): Likewise.
4431 (known_aggs_to_agg_replacement_list): Likewise.
4432 * ipa-devirt.c (get_odr_type): Likewise.
4433 * ipa-prop.c (ipa_node_duplication_hook): Likewise.
4434 (read_agg_replacement_chain): Likewise.
4435 * loop-iv.c (get_simple_loop_desc): Likewise.
4436 * lto-cgraph.c (input_node_opt_summary): Likewise.
4437 * lto-section-in.c (lto_new_in_decl_state): Likewise.
4438 * lto-streamer-in.c (lto_input_eh_catch_list): Likewise.
4439 (input_eh_region): Likewise.
4440 (input_eh_lp): Likewise.
4441 (input_cfg): Likewise.
4442 * optabs.c (set_optab_libfunc): Likewise.
4443 (init_tree_optimization_optabs): Likewise.
4444 (set_conv_libfunc): Likewise.
4445 * passes.c (do_per_function_toporder): Likewise.
4446 * rtl.h: Don't use variable_size gty attribute.
4447 * sese.c (if_region_set_false_region): Adjust.
4448 * stringpool.c (gt_pch_save_stringpool): Likewise.
4449 * target-globals.c (save_target_globals): Likewise.
4450 * toplev.c (general_init): Likewise.
4451 * trans-mem.c (record_tm_replacement): Likewise.
4452 (split_bb_make_tm_edge): Likewise.
4453 * tree-cfg.c (move_sese_region_to_fn): Likewise.
4454 * tree-data-ref.h (lambda_vector_new): Likewise.
4455 * tree-eh.c (add_stmt_to_eh_lp_fn): Likewise.
4456 * tree-iterator.c (tsi_link_before): Likewise.
4457 (tsi_link_after): Likewise.
4458 * tree-scalar-evolution.c (new_scev_info_str): Likewise.
4459 * tree-ssa-loop-niter.c (record_estimate): Likewise.
4460 * tree-ssa-operands.c (ssa_operand_alloc): Likewise.
4461 * tree-ssa-operands.h: Don't use variable_size gty attribute.
4462 * tree-ssa.c (init_tree_ssa): Adjust.
4463 * tree-ssanames.c (set_range_info): Likewise.
4464 (get_ptr_info): Likewise.
4465 (duplicate_ssa_name_ptr_info): Likewise.
4466 (duplicate_ssa_name_range_info): Likewise.
4467 * tree-streamer-in.c (unpack_ts_real_cst_value_fields): Likewise.
4468 (unpack_ts_fixed_cst_value_fields): Likewise.
4469 * tree.c (build_fixed): Likewise.
4470 (build_real): Likewise.
4471 (build_string): Likewise.
4472 (decl_priority_info): Likewise.
4473 (decl_debug_expr_insert): Likewise.
4474 (decl_value_expr_insert): Likewise.
4475 (decl_debug_args_insert): Likewise.
4476 (type_hash_add): Likewise.
4477 (build_omp_clause): Likewise.
4478 * ubsan.c (decl_for_type_insert): Likewise.
4479 * varasm.c (get_unnamed_section): Likewise.
4480 (get_noswitch_section): Likewise.
4481 (get_section): Likewise.
4482 (get_block_for_section): Likewise.
4483 (create_block_symbol): Likewise.
4484 (build_constant_desc): Likewise.
4485 (create_constant_pool): Likewise.
4486 (force_const_mem): Likewise.
4487 (record_tm_clone_pair): Likewise.
4488 * varpool.c (varpool_create_empty_node): Likewise.
4490 2014-05-17 Trevor Saunders <tsaunders@mozilla.com>
4492 * dwarf2out.c (tree_add_const_value_attribute): Call
4493 ggc_internal_cleared_alloc instead of ggc_alloc_cleared_atomic.
4494 * gengtype.c (write_typed_alloc_def): Call ggc_internal_<x>alloc
4495 instead of ggc_internal_<x>alloc_stat.
4496 * ggc-common.c (ggc_internal_cleared_alloc): Drop _stat suffix.
4497 (ggc_realloc): Likewise.
4498 * ggc-none.c (ggc_internal_alloc): Likewise.
4499 (ggc_internal_cleared_alloc): Likewise.
4500 * ggc-page.c: Likewise.
4501 * ggc.h (ggc_internal_alloc_stat): Likewise.
4502 (ggc_internal_alloc): Remove macro.
4503 (ggc_internal_cleared_alloc_stat): Drop _stat suffix.
4504 (ggc_internal_cleared_alloc): Remove macro.
4505 (GGC_RESIZEVEC): Adjust.
4506 (ggc_resizevar): Remove macro.
4507 (ggc_internal_vec_alloc_stat): Drop _stat suffix.
4508 (ggc_internal_cleared_vec_alloc_stat): Likewise.
4509 (ggc_internal_vec_cleared_alloc): Remove macro.
4510 (ggc_alloc_atomic_stat): Drop _stat suffix.
4511 (ggc_alloc_atomic): Remove macro.
4512 (ggc_alloc_cleared_atomic): Remove macro.
4513 (ggc_alloc_string_stat): Drop _stat suffix.
4514 (ggc_alloc_string): Remove macro.
4515 (ggc_alloc_rtx_def_stat): Adjust.
4516 (ggc_alloc_tree_node_stat): Likewise.
4517 (ggc_alloc_cleared_tree_node_stat): Likewise.
4518 (ggc_alloc_cleared_gimple_statement_stat): Likewise.
4519 (ggc_alloc_cleared_simd_clone_stat): Likewise.
4520 * gimple.c (gimple_build_omp_for): Likewise.
4521 (gimple_copy): Likewise.
4522 * stringpool.c (ggc_alloc_string_stat): Drop _stat suffix.
4523 * toplev.c (realloc_for_line_map): Adjust.
4524 * tree-data-ref.h (lambda_vector_new): Likewise.
4525 * tree-phinodes.c (allocate_phi_node): Likewise.
4526 * tree.c (grow_tree_vec_stat): Likewise.
4527 * vec.h (va_gc::reserve): Adjust.
4529 2014-05-17 Ajit Agarwal <ajitkum@xilinx.com>
4531 * config/microblaze/microblaze.c (break_handler): New Declaration.
4532 (microblaze_break_function_p,microblaze_is_break_handler): New.
4533 (compute_frame_size): Use microblaze_break_function_p.
4534 Add the test of break_handler.
4535 (microblaze_function_prologue) : Add the test of variable
4536 break_handler. Check the fnname by BREAK_HANDLER_NAME.
4537 (microblaze_function_epilogue) : Add the test of break_handler.
4538 (microblaze_globalize_label) : Add the test of break_handler.
4539 Check the name by BREAK_HANDLER_NAME.
4541 * config/microblaze/microblaze.h (BREAK_HANDLER_NAME): New macro
4543 * config/microblaze/microblaze.md (*<optab>,<optab>_internal): Add
4544 microblaze_is_break_handler test.
4545 (call_internal1,call_value_intern): Use microblaze_break_function_p.
4546 Use SYMBOL_REF_DECL.
4548 * config/microblaze/microblaze-protos.h
4549 (microblaze_break_function_p,microblaze_is_break_handler):
4552 * doc/extend.texi (MicroBlaze break_handler Functions): Document
4553 new MicroBlaze break_handler functions.
4555 2014-05-17 Uros Bizjak <ubizjak@gmail.com>
4557 * doc/extend.texi (Size of an asm): Move node text according
4558 to its @menu entry position.
4560 2014-05-17 Marc Glisse <marc.glisse@inria.fr>
4562 PR tree-optimization/61140
4563 PR tree-optimization/61150
4564 PR tree-optimization/61197
4565 * tree-ssa-phiopt.c (value_replacement): Punt on multiple phis.
4567 2014-05-17 Uros Bizjak <ubizjak@gmail.com>
4569 * doc/invoke.texi (free): Mention Alpha. Also enabled at -Os.
4571 2014-05-17 Richard Sandiford <r.sandiford@uk.ibm.com>
4573 * wide-int.cc: Only include longlong.h if W_TYPE_SIZE==32 or
4574 __SIZEOF_INT128__ is defined.
4576 2014-05-17 Richard Sandiford <rdsandiford@googlemail.com>
4578 * config/rs6000/rs6000.c (rs6000_real_tls_symbol_ref_p): New function.
4579 (rs6000_delegitimize_address): Use it.
4581 2014-05-17 Richard Sandiford <rdsandiford@googlemail.com>
4583 * emit-rtl.h (replace_equiv_address, replace_equiv_address_nv): Add an
4584 inplace argument. Store the new address in the original MEM when true.
4585 * emit-rtl.c (change_address_1): Likewise.
4586 (adjust_address_1, adjust_automodify_address_1, offset_address):
4588 * rtl.h (plus_constant): Add an inplace argument.
4589 * explow.c (plus_constant): Likewise. Try to reuse the original PLUS
4590 when true. Avoid generating (plus X (const_int 0)).
4591 * function.c (instantiate_virtual_regs_in_rtx): Adjust the PLUS
4592 in-place. Pass true to plus_constant.
4593 (instantiate_virtual_regs_in_insn): Pass true to replace_equiv_address.
4595 2014-05-16 Dehao Chen <dehao@google.com>
4597 * tree-cfg.c (gimple_merge_blocks): Updates bb count with max count.
4599 2014-05-16 Oleg Endo <olegendo@gcc.gnu.org>
4602 * config/sh/predicates.md (negt_reg_shl31_operand): Match additional
4604 * config/sh/sh.md (*negt_msb): Merge SH2A and non-SH2A variants.
4606 2014-05-16 Dehao Chen <dehao@google.com>
4608 * ira-int.h (REG_FREQ_FROM_EDGE_FREQ): Use
4609 optimize_function_for_size_p.
4610 * regs.h (REG_FREQ_FROM_BB): Likewise.
4612 2014-05-16 Oleg Endo <olegendo@gcc.gnu.org>
4615 * config/sh/sh.c (sh_eval_treg_value): Handle t_reg_operand and
4616 negt_reg_operand cases.
4617 * config/sh/sh.md (*cset_zero): Likewise by using cbranch_treg_value
4619 * config/sh/predicates.md (cbranch_treg_value): Simplify.
4621 2014-05-16 Oleg Endo <olegendo@gcc.gnu.org>
4623 * config/sh/sh.c (sh_option_override): Set branch cost to 2 for all
4626 2014-05-16 David Malcolm <dmalcolm@redhat.com>
4629 2014-04-29 David Malcolm <dmalcolm@redhat.com>
4631 * tree-cfg.c (dump_function_to_file): Dump the return type of
4632 functions, in a line to itself before the function body, mimicking
4633 the layout of a C function.
4635 2014-05-16 Dehao Chen <dehao@google.com>
4637 * cfghooks.c (make_forwarder_block): Use direct computation to
4638 get fall-through edge's count and frequency.
4640 2014-05-16 Benno Schulenberg <bensberg@justemail.net>
4642 * config/arc/arc.c (arc_init): Fix typo in error message.
4643 * config/i386/i386.c (ix86_expand_builtin): Likewise.
4644 (split_stack_prologue_scratch_regno): Likewise.
4645 * fortran/check.c (gfc_check_fn_rc2008): Remove duplicate
4646 word from error message.
4648 2014-05-16 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
4650 * ira-costs.c: Fix typo in comment.
4652 2014-05-16 David Wohlferd <dw@LimeGreenSocks.com>
4654 * doc/extend.texi: (Visibility Pragmas) Fix misplaced @xref
4656 2014-05-16 Jan Hubicka <hubicka@ucw.cz>
4658 * varpool.c (dump_varpool_node): Dump write-only flag.
4659 * lto-cgraph.c (lto_output_varpool_node, input_varpool_node): Stream
4661 * tree-cfg.c (execute_fixup_cfg): Remove statements setting
4662 write-only variables.
4663 * ipa.c (process_references): New function.
4664 (set_readonly_bit): New function.
4665 (set_writeonly_bit): New function.
4666 (clear_addressable_bit): New function.
4667 (ipa_discover_readonly_nonaddressable_var): Mark write only variables;
4668 fix handling of aliases.
4669 * cgraph.h (struct varpool_node): Add writeonly flag.
4671 2014-05-16 Vladimir Makarov <vmakarov@redhat.com>
4673 PR rtl-optimization/60969
4674 * ira-costs.c (record_reg_classes): Allow only memory for pseudo.
4675 Calculate costs for this case.
4677 2014-05-16 Eric Botcazou <ebotcazou@adacore.com>
4679 * fold-const (fold_unary_loc) <NON_LVALUE_EXPR>: New case.
4680 <CASE_CONVERT>: Pass arg0 instead of op0 to fold_convert_const.
4682 2014-05-16 Richard Biener <rguenther@suse.de>
4684 PR tree-optimization/61194
4685 * tree-vect-patterns.c (adjust_bool_pattern): Also handle
4686 bool patterns ending in a COND_EXPR.
4688 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4690 * config/aarch64/aarch64.c (aarch64_rtx_mult_cost): Fix FNMUL case.
4692 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4694 * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle the case
4695 where we were unable to cost an RTX.
4697 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4699 * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost SYMBOL_REF,
4702 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4703 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4705 * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost TRUNCATE.
4707 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4708 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4710 * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost FMA,
4711 FLOAT_EXTEND, FLOAT_TRUNCATE, ABS, SMAX, and SMIN.
4713 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4714 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4716 * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost comparison
4719 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4720 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4722 * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costs for
4725 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4726 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4728 * config/aarch64/aarch64.c (aarch64_rtx_arith_op_extract_p): New.
4729 (aarch64_rtx_costs): Improve costs for SIGN/ZERO_EXTRACT.
4731 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4732 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4734 * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costs for
4737 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4738 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4740 * config/aarch64/aarch64.c (aarch64_rtx_costs): Cost
4741 ZERO_EXTEND and SIGN_EXTEND better.
4743 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4744 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4746 * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve cost for
4749 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4750 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4752 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use address
4753 costs when costing loads and stores to memory.
4755 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4756 Philip Tomsich <philipp.tomsich@theobroma-systems.com>
4758 * config/aarch64/aarch64.c (aarch64_rtx_costs): Improve costing
4761 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4763 * config/aarch64/aarch64.c (aarch64_rtx_costs): Set default costs.
4765 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4766 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4768 * config/aarch64/aarch64.c (aarch64_strip_shift_or_extend): Rename
4770 (aarch64_strip_extend): ...this, don't strip shifts, check RTX is
4772 (aarch64_rtx_mult_cost): New.
4773 (aarch64_rtx_costs): Use it, refactor as appropriate.
4775 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4776 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
4778 * config/aarch64/aarch64.c (aarch64_build_constant): Conditionally
4779 emit instructions, return number of instructions which would
4781 (aarch64_add_constant): Update call to aarch64_build_constant.
4782 (aarch64_output_mi_thunk): Likewise.
4783 (aarch64_rtx_costs): Estimate cost of a CONST_INT, cost of
4786 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4788 * config/aarch64/aarch64.c (aarch64_rtx_costs_wrapper): New.
4789 (TARGET_RTX_COSTS): Call it.
4791 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4793 * config/aarch64/aarch64.c (cortexa57_addrcost_table): New.
4794 (cortexa57_vector_cost): Likewise.
4795 (cortexa57_tunings): Use them.
4797 2014-05-16 James Greenhalgh <james.greenhalgh@arm.com>
4799 * config/aarch64/aarch64-protos.h (scale_addr_mode_cost): New.
4800 (cpu_addrcost_table): Use it.
4801 * config/aarch64/aarch64.c (generic_addrcost_table): Initialize it.
4802 (aarch64_address_cost): Rewrite using aarch64_classify_address,
4805 2014-05-16 Richard Biener <rguenther@suse.de>
4807 * tree-ssa-sccvn.c: Include tree-cfg.h and domwalk.h.
4808 (set_ssa_val_to): Handle unexpected sets to VN_TOP.
4809 (visit_phi): Ignore edges marked as not executable.
4810 (class cond_dom_walker): New.
4811 (cond_dom_walker::before_dom_children): Value-number
4812 control statements and mark successor edges as not
4813 executable if possible.
4814 (run_scc_vn): First walk all control statements in
4815 dominator order, marking edges as not executable.
4816 * tree-inline.c (copy_edges_for_bb): Be not confused
4817 about random edge flags.
4819 2014-05-16 Richard Biener <rguenther@suse.de>
4821 * tree-ssa-sccvn.c (visit_use): Also constant-fold calls.
4823 2014-05-15 Peter Bergner <bergner@vnet.ibm.com>
4826 * config/rs6000/htmxlintrin.h (_HTM_TBEGIN_STARTED): New define.
4827 (__TM_simple_begin): Use it.
4828 (__TM_begin): Likewise.
4830 2014-05-15 Martin Jambor <mjambor@suse.cz>
4833 * ipa-prop.c (update_indirect_edges_after_inlining): Check
4834 type_preserved flag when the indirect edge is polymorphic.
4836 2014-05-15 Martin Jambor <mjambor@suse.cz>
4838 PR tree-optimization/61090
4839 * tree-sra.c (sra_modify_expr): Pass the current gsi to
4840 build_ref_for_model.
4842 2014-05-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4844 * config/arm/arm.c (arm_option_override): Use the SCHED_PRESSURE_MODEL
4845 enum name for PARAM_SCHED_PRESSURE_ALGORITHM.
4847 2014-05-15 Jakub Jelinek <jakub@redhat.com>
4849 PR tree-optimization/61158
4850 * fold-const.c (fold_binary_loc): If X is zero-extended and
4851 shiftc >= prec, make sure zerobits is all ones instead of
4852 invoking undefined behavior.
4854 2014-05-15 Zhenqiang Chen <zhenqiang.chen@linaro.org>
4856 * regcprop.h: New file.
4857 * regcprop.c (skip_debug_insn_p): New decl.
4858 (replace_oldest_value_reg): Check skip_debug_insn_p.
4859 (copyprop_hardreg_forward_bb_without_debug_insn): New function.
4860 * shrink-wrap.c: Include regcprop.h.
4861 (prepare_shrink_wrap): Call
4862 copyprop_hardreg_forward_bb_without_debug_insn.
4864 2014-05-15 Zhenqiang Chen <zhenqiang.chen@linaro.org>
4866 * shrink-wrap.h: Update comment.
4867 * shrink-wrap.c: Update comment.
4868 (next_block_for_reg): Rename to live_edge_for_reg.
4869 (live_edge_for_reg): Allow live_edge->dest has two predecessors.
4870 (move_insn_for_shrink_wrap): Split live_edge.
4871 (prepre_shrink_wrap): One more parameter for move_insn_for_shrink_wrap.
4873 2014-05-14 Eric Botcazou <ebotcazou@adacore.com>
4875 * config/sparc/sparc-protos.h (sparc_absnegfloat_split_legitimate):
4877 * config/sparc/sparc.c (sparc_absnegfloat_split_legitimate): Likewise.
4878 * config/sparc/sparc.md (fptype_ut699): New attribute.
4879 (in_branch_delay): Return false if -mfix-ut699 is specified and
4880 fptype_ut699 is set to single.
4881 (truncdfsf2): Add fptype_ut699 attribute.
4882 (fix_truncdfsi2): Likewise.
4883 (floatsisf2): Change fptype attribute.
4884 (fix_truncsfsi2): Likewise.
4885 (negtf2_notv9): Delete.
4886 (negtf2_v9): Likewise.
4887 (negtf2_hq): New instruction.
4888 (negtf2): New instruction and splitter.
4889 (negdf2_notv9): Rewrite.
4890 (abstf2_notv9): Delete.
4891 (abstf2_hq_v9): Likewise.
4892 (abstf2_v9): Likewise.
4893 (abstf2_hq): New instruction.
4894 (abstf2): New instruction and splitter.
4895 (absdf2_notv9): Rewrite.
4897 2014-05-14 Cary Coutant <ccoutant@google.com>
4900 * opts.c (common_handle_option): Don't special-case "-g".
4901 (set_debug_level): Default to at least level 2 with "-g".
4903 2014-05-14 DJ Delorie <dj@redhat.com>
4905 * config/msp430/msp430.c (msp430_builtin): Add
4906 MSP430_BUILTIN_DELAY_CYCLES.
4907 (msp430_init_builtins): Register void __delay_cycles(long long).
4908 (msp430_builtin_decl): Add it.
4909 (cg_magic_constant): New.
4910 (msp430_expand_delay_cycles): New.
4911 (msp430_expand_builtin): Call it.
4912 (msp430_print_operand_raw): Change integer printing from "int" to
4914 * config/msp430/msp430.md (define_constants): Add delay_cycles tags.
4915 (delay_cycles_start): New.
4916 (delay_cycles_end): New.
4917 (delay_cycles_32): New.
4918 (delay_cycles_32x): New.
4919 (delay_cycles_16): New.
4920 (delay_cycles_16x): New.
4921 (delay_cycles_2): New.
4922 (delay_cycles_1): New.
4923 * doc/extend.texi: Document __delay_cycles().
4925 2014-05-14 Sandra Loosemore <sandra@codesourcery.com>
4927 * config/nios2/nios2.md (nios2_cbranch): Fix paste-o in
4928 length attribute computation.
4930 2014-05-14 Richard Sandiford <rdsandiford@googlemail.com>
4933 * print-rtl.c (print_rtx): Suppress uids if flag_dump_unnumbered.
4935 2014-05-14 Richard Sandiford <r.sandiford@uk.ibm.com>
4938 * config/sparc/sparc.md: Fix types of low and high in DI constant
4939 splitter. Use gen_int_mode in some other splitters.
4941 2014-05-14 Martin Jambor <mjambor@suse.cz>
4944 * ipa-prop.c (ipa_modify_formal_parameters): Reset DECL_LANG_SPECIFIC.
4946 2014-05-14 James Norris <jnorris@codesourcery.com>
4948 * omp-low.c (expand_parallel_call): Remove shadow variable.
4949 (expand_omp_taskreg): Likewise.
4951 2014-05-14 Ilya Tocar <ilya.tocar@intel.com>
4953 * common/config/i386/i386-common.c
4954 (OPTION_MASK_ISA_CLFLUSHOPT_SET): Define.
4955 (OPTION_MASK_ISA_XSAVES_SET): Ditto.
4956 (OPTION_MASK_ISA_XSAVEC_SET): Ditto.
4957 (OPTION_MASK_ISA_CLFLUSHOPT_UNSET): Ditto.
4958 (OPTION_MASK_ISA_XSAVES_UNSET): Ditto.
4959 (OPTION_MASK_ISA_XSAVEC_UNSET): Ditto.
4960 (ix86_handle_option): Handle OPT_mxsavec, OPT_mxsaves, OPT_mclflushopt.
4961 * config.gcc (i[34567]86-*-*): Add clflushoptintrin.h,
4962 xsavecintrin.h, xsavesintrin.h.
4963 (x86_64-*-*): Ditto.
4964 * config/i386/clflushoptintrin.h: New.
4965 * config/i386/xsavecintrin.h: Ditto.
4966 * config/i386/xsavesintrin.h: Ditto.
4967 * config/i386/cpuid.h (bit_CLFLUSHOPT): Define.
4968 (bit_XSAVES): Ditto.
4969 (bit_XSAVES): Ditto.
4970 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
4971 -mclflushopt, -mxsavec, -mxsaves, -mno-xsaves, -mno-xsavec,
4973 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4974 OPTION_MASK_ISA_CLFLUSHOPT, OPTION_MASK_ISA_XSAVEC,
4975 OPTION_MASK_ISA_XSAVES.
4976 * config/i386/i386.c (ix86_target_string): Handle -mclflushopt,
4978 (PTA_CLFLUSHOPT) Define.
4979 (PTA_XSAVEC): Ditto.
4980 (PTA_XSAVES): Ditto.
4981 (ix86_option_override_internal): Handle new options.
4982 (ix86_valid_target_attribute_inner_p): Ditto.
4983 (ix86_builtins): Add IX86_BUILTIN_XSAVEC, IX86_BUILTIN_XSAVEC64,
4984 IX86_BUILTIN_XSAVES, IX86_BUILTIN_XRSTORS, IX86_BUILTIN_XSAVES64,
4985 IX86_BUILTIN_XRSTORS64, IX86_BUILTIN_CLFLUSHOPT.
4986 (bdesc_special_args): Add __builtin_ia32_xsaves,
4987 __builtin_ia32_xrstors, __builtin_ia32_xsavec, __builtin_ia32_xsaves64,
4988 __builtin_ia32_xrstors64, __builtin_ia32_xsavec64.
4989 (ix86_init_mmx_sse_builtins): Add __builtin_ia32_clflushopt.
4990 (ix86_expand_builtin): Handle new builtins.
4991 * config/i386/i386.h (TARGET_CLFLUSHOPT) Define.
4992 (TARGET_CLFLUSHOPT_P): Ditto.
4993 (TARGET_XSAVEC): Ditto.
4994 (TARGET_XSAVEC_P): Ditto.
4995 (TARGET_XSAVES): Ditto.
4996 (TARGET_XSAVES_P): Ditto.
4997 * config/i386/i386.md (ANY_XSAVE): Add UNSPECV_XSAVEC, UNSPECV_XSAVES.
4998 (ANY_XSAVE64)" Add UNSPECV_XSAVEC64, UNSPECV_XSAVES64.
4999 (attr xsave): Add xsavec, xsavec64, xsaves, xsaves64.
5001 (ANY_XRSTOR64): Ditto.
5003 (xrstor): Change into <xrstor>.
5004 (xrstor_rex64): Change into <xrstor>_rex64.
5005 (xrstor64): Change into <xrstor>64
5007 * config/i386/i386.opt (mclflushopt): New.
5010 * config/i386/x86intrin.h: Add clflushoptintrin.h, xsavesintrin.h,
5012 * doc/invoke.texi: Document new options.
5014 2014-05-14 Andrey Belevantsev <abel@ispras.ru>
5016 PR rtl-optimization/60866
5017 * sel-sched-ir (sel_init_new_insn): New parameter old_seqno.
5018 Default it to -1. Pass it down to init_simplejump_data.
5019 (init_simplejump_data): New parameter old_seqno. Pass it down
5020 to get_seqno_for_a_jump.
5021 (get_seqno_for_a_jump): New parameter old_seqno. Use it for
5022 initializing new jump seqno as a last resort. Add comment.
5023 (sel_redirect_edge_and_branch): Save old seqno of the conditional
5024 jump and pass it down to sel_init_new_insn.
5025 (sel_redirect_edge_and_branch_force): Likewise.
5027 2014-05-14 Georg-Johann Lay <avr@gjlay.de>
5029 * config/avr/avr.h (REG_CLASS_CONTENTS): Use unsigned suffix for
5030 shifted values to avoid build warning.
5032 2014-05-14 Eric Botcazou <ebotcazou@adacore.com>
5034 * cfgcleanup.c (try_forward_edges): Use location_t for locations.
5035 * cfgrtl.c (rtl_merge_blocks): Fix comment.
5036 (cfg_layout_merge_blocks): Likewise.
5037 * except.c (emit_to_new_bb_before): Remove prev_bb local variable.
5039 2014-05-14 Andrey Belevantsev <abel@ispras.ru>
5041 PR rtl-optimization/60901
5042 * config/i386/i386.c (ix86_dependencies_evaluation_hook): Check that
5043 bb predecessor belongs to the same scheduling region. Adjust comment.
5045 2014-05-13 Peter Bergner <bergner@vnet.ibm.com>
5047 * doc/sourcebuild.texi: (dfp_hw): Document.
5048 (p8vector_hw): Likewise.
5049 (powerpc_eabi_ok): Likewise.
5050 (powerpc_elfv2): Likewise.
5051 (powerpc_htm_ok): Likewise.
5052 (ppc_recip_hw): Likewise.
5055 2014-05-13 Cary Coutant <ccoutant@google.com>
5057 * opts.c (finish_options): Use -ggnu-pubnames with -gsplit-dwarf.
5059 2014-05-13 David Malcolm <dmalcolm@redhat.com>
5061 * gengtype-parse.c (require3): Eliminate in favor of...
5063 (require_template_declaration): Update to support optional single *
5066 * gengtype.c (get_ultimate_base_class): Add a non-const overload.
5067 (create_user_defined_type): Handle a single level of explicit
5068 pointerness within template arguments.
5069 (struct write_types_data): Add field "kind".
5070 (filter_type_name): Handle "*" character.
5071 (write_user_func_for_structure_ptr): Require a write_types_data
5072 rather than just a prefix string, so that we can look up the kind
5073 of the wtd and use it as an index into wrote_user_func_for_ptr,
5074 ensuring that such functions are written at most once. Support
5075 subclasses by invoking the marking function of the ultimate base class.
5076 (write_user_func_for_structure_body): Require a write_types_data
5077 rather than just a prefix string, so that we can pass this to
5078 write_user_func_for_structure_ptr.
5079 (write_func_for_structure): Likewise.
5080 (ggc_wtd): Add initializer of new "kind" field.
5081 (pch_wtd): Likewise.
5083 * gengtype.h (enum write_types_kinds): New.
5084 (struct type): Add field wrote_user_func_for_ptr to the "s"
5087 2014-05-13 Richard Sandiford <r.sandiford@uk.ibm.com>
5089 * fold-const.c (optimize_bit_field_compare): Use wi:: operations
5090 instead of const_binop.
5091 (fold_binary_loc): Likewise.
5093 2014-05-13 Richard Sandiford <r.sandiford@uk.ibm.com>
5095 * tree-dfa.h (get_addr_base_and_unit_offset_1): Update array index
5096 calculation to match get_ref_base_and_extent.
5098 2014-05-13 Catherine Moore <clm@codesourcery.com>
5099 Sandra Loosemore <sandra@codesourcery.com>
5101 * configure.ac: Fix assembly for explicit JALR relocation check.
5102 * configure: Regenerate.
5104 2014-05-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5106 * config/arm/arm.c (neon_itype): Remove NEON_RESULTPAIR.
5107 (arm_init_neon_builtins): Remove handling of NEON_RESULTPAIR.
5108 Remove associated type declarations and initialisations.
5109 (arm_expand_neon_builtin): Likewise.
5110 (neon_emit_pair_result_insn): Delete.
5111 * config/arm/arm_neon_builtins (vtrn, vzip, vuzp): Delete.
5112 * config/arm/neon.md (neon_vtrn<mode>): Delete.
5113 (neon_vzip<mode>): Likewise.
5114 (neon_vuzp<mode>): Likewise.
5116 2014-05-13 Richard Biener <rguenther@suse.de>
5119 * tree-inline.c (remap_gimple_stmt): Clear tail call flag,
5120 it needs revisiting whether the call still may be tail-called.
5122 2014-05-13 Richard Sandiford <rdsandiford@googlemail.com>
5124 * rtl.def (SYMBOL_REF): Remove middle "0" field.
5125 * rtl.h (block_symbol): Reduce number of fields to 2.
5126 (rtx_def): Add u2.symbol_ref_flags.
5127 (SYMBOL_REF_FLAGS): Use it.
5128 (SYMBOL_REF_DATA, SET_SYMBOL_REF_DECL, SYMBOL_REF_DECL)
5129 (SET_SYMBOL_REF_CONSTANT, SYMBOL_REF_CONSTANT): Lower index.
5130 * gengtype.c (adjust_field_rtx_def): Remove SYMBOL_REF_FLAGS handling.
5131 Lower index of SYMBOL_REF_DATA.
5132 * print-rtl.c (print_rtx): Lower index for SYMBOL_REF_DATA.
5133 Print SYMBOL_REF_FLAGS at the same time.
5134 * genattrtab.c (attr_rtx_1): Only initialize 1 "0" SYMBOL_REF field.
5136 2014-05-13 Richard Sandiford <rdsandiford@googlemail.com>
5138 * rtl.def (VAR_LOCATION): Remove "i" field.
5139 * rtl.h (rtx_def): Add u2.var_location_status.
5140 (PAT_VAR_LOCATION_STATUS): Use it.
5141 (gen_rtx_VAR_LOCATION): Declare.
5142 * gengenrtl.c (excluded_rtx): Add VAR_LOCATION.
5143 * emit-rtl.c (gen_rtx_VAR_LOCATION): New function.
5144 * var-tracking.c (emit_note_insn_var_location): Remove casts.
5146 2014-05-13 Richard Sandiford <rdsandiford@googlemail.com>
5148 * rtl.def (scratch): Fix outdated comment and remove "0" field.
5149 * gengtype.c (adjust_field_rtx_def): Update accordingly.
5151 2014-05-13 Richard Sandiford <rdsandiford@googlemail.com>
5153 * rtl.def (DEBUG_INSN, INSN, JUMP_INSN, CALL_INSN, JUMP_TABLE_DATA)
5154 (BARRIER, CODE_LABEL, NOTE): Remove first "i" field.
5155 * rtl.h (rtx_def): Add insn_uid to u2 field.
5156 (RTX_FLAG_CHECK8): Delete in favor of...
5157 (RTL_INSN_CHAIN_FLAG_CHECK): ...this new macro.
5158 (INSN_DELETED_P): Update accordingly.
5159 (INSN_UID): Use u2.insn_uid.
5160 (INSN_CHAIN_CODE_P): Define.
5161 (PREV_INSN, NEXT_INSN, BLOCK_FOR_INSN, PATTERN, INSN_LOCATION)
5162 (INSN_CODE, REG_NOTES, CALL_INSN_FUNCTION_USAGE, CODE_LABEL_NUMBER)
5163 (NOTE_DATA, NOTE_DELETED_LABEL_NAME, NOTE_BLOCK, NOTE_EH_HANDLER)
5164 (NOTE_BASIC_BLOCK, NOTE_VAR_LOCATION, NOTE_CFI, NOTE_LABEL_NUMBER)
5165 (NOTE_KIND, LABEL_NAME, LABEL_NUSES, JUMP_LABEL, LABEL_REFS): Lower
5166 indices accordingly.
5167 * print-rtl.c (print_rtx): Print INSN_UIDs before the main loop.
5168 Update indices for insn-chain rtxes.
5169 * gengtype.c (gen_rtx_next): Adjust test for insn-chain rtxes.
5170 (adjust_field_rtx_def): Lower '0' indices for all insn-chain rtxes.
5171 * emit-rtl.c (gen_label_rtx): Update gen_rtx_LABEL call.
5172 * caller-save.c (init_caller_save): Update gen_rtx_INSN calls.
5173 * combine.c (try_combine): Likewise.
5174 * ira.c (setup_prohibited_mode_move_regs): Likewise.
5176 2014-05-13 Richard Sandiford <rdsandiford@googlemail.com>
5178 * rtl.def (REG): Remove middle field.
5179 * rtl.h (rtx_def): Add orignal_regno to u2.
5180 (ORIGINAL_REGNO): Use it instead of field 1.
5181 (REG_ATTRS): Lower field index accordingly.
5182 * gengtype.c (adjust_field_rtx_def): Remove handling of
5183 ORIGINAL_REGNO. Move REG_ATTRS index down.
5184 * print-rtl.c (print_rtx): Move ORIGINAL_REGNO handling to the
5185 code that prints the REGNO.
5187 2014-05-13 Richard Sandiford <rdsandiford@googlemail.com>
5189 * print-rtl.c (print_rtx): Guard whole '0' block with ifndef
5192 2014-05-13 Richard Sandiford <rdsandiford@googlemail.com>
5194 * rtl.h (rtx_def): Mark u2 as GTY ((skip)).
5196 2014-05-13 Bin Cheng <bin.cheng@arm.com>
5198 * tree-ssa-loop-ivopts.c (contain_complex_addr_expr): New.
5199 (alloc_iv): Lower base expressions containing ADDR_EXPR.
5201 2014-05-13 Ian Bolton <ian.bolton@arm.com>
5203 * config/aarch64/aarch64-protos.h
5204 (aarch64_hard_regno_caller_save_mode): New prototype.
5205 * config/aarch64/aarch64.c (aarch64_hard_regno_caller_save_mode):
5207 * config/aarch64/aarch64.h (HARD_REGNO_CALLER_SAVE_MODE): New macro.
5209 2014-05-13 Christian Bruel <christian.bruel@st.com>
5211 * target.def (mode_switching): New hook vector.
5212 (mode_emit, mode_needed, mode_after, mode_entry): New hooks.
5213 (mode_exit, modepriority_to_mode): Likewise.
5214 * mode-switching.c (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Hookify.
5215 (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
5216 * target.h: Include tm.h and hard-reg-set.h.
5217 * doc/tm.texi.in (EMIT_MODE_SET, MODE_NEEDED, MODE_AFTER, MODE_ENTRY)
5218 (MODE_EXIT, MODE_PRIORITY_TO_MODE): Delete and hookify.
5219 * doc/tm.texi Regenerate.
5220 * config/sh/sh.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
5221 (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
5222 * config/sh/sh.c (sh_emit_mode_set, sh_mode_priority): Hookify.
5223 (sh_mode_needed, sh_mode_after, sh_mode_entry, sh_mode_exit): Likewise.
5224 * config/i386/i386.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY): Delete
5225 (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
5226 * config/i386/i386-protos.h (ix86_mode_needed, ix86_mode_after)
5227 (ix86_mode_entrym, ix86_emit_mode_set): Remove external declaration.
5228 * config/i386/i386.c (ix86_mode_needed, ix86_mode_after,
5229 (ix86_mode_exit, ix86_mode_entry, ix86_mode_priority)
5230 (ix86_emit_mode_set): Hookify.
5231 * config/epiphany/epiphany.h (MODE_NEEDED, MODE_AFTER, MODE_ENTRY):
5233 (MODE_EXIT, MODE_PRIORITY_TO_MODE, EMIT_MODE_SET): Likewise.
5234 * config/epiphany/epiphany-protos.h (epiphany_mode_needed)
5235 (emit_set_fp_mode, epiphany_mode_entry_exit, epiphany_mode_after)
5236 (epiphany_mode_priority_to_mode): Remove declaration.
5237 * config/epiphany/epiphany.c (emit_set_fp_mode): Hookify.
5238 (epiphany_mode_needed, epiphany_mode_priority_to_mode): Likewise.
5239 (epiphany_mode_entry, epiphany_mode_exit, epiphany_mode_after):
5241 (epiphany_mode_priority_to_mode): Change priority type. Hookify.
5242 (epiphany_mode_needed, epiphany_mode_entry_exit): Hookify.
5243 (epiphany_mode_after, epiphany_mode_entry, emit_set_fp_mode): Hookify.
5245 2014-05-13 Jakub Jelinek <jakub@redhat.com>
5248 * config/i386/i386.c (ix86_expand_set_or_movmem): If count_exp
5249 is const0_rtx, return immediately. Don't test count == 0 when
5252 2014-05-13 Zhenqiang Chen <zhenqiang.chen@linaro.org>
5254 * Makefile.in: add shrink-wrap.o.
5255 * config/i386/i386.c: include "shrink-wrap.h"
5256 * function.c: Likewise.
5257 (requires_stack_frame_p, next_block_for_reg,
5258 move_insn_for_shrink_wrap, prepare_shrink_wrap,
5259 dup_block_and_redirect): Move to shrink-wrap.c
5260 (thread_prologue_and_epilogue_insns): Extract three code segments
5261 as functions in shrink-wrap.c
5262 * function.h: Move #ifdef HAVE_simple_return ... #endif block to
5264 * shrink-wrap.c: New file.
5265 * shrink-wrap.h: New file.
5267 2014-05-12 David Wohlferd <dw@LimeGreenSocks.com>
5269 * doc/extend.texi: Reflect current numbers of pragmas. Remove
5270 reference to Solaris.
5272 2014-05-12 Mike Stump <mikestump@comcast.net>
5275 * genattrtab.c (filename): Add.
5276 (convert_set_attr_alternative): Improve error message.
5277 (check_defs): Restore read_md_filename for error messages.
5278 (gen_insn): Save filename.
5280 2014-05-12 Dimitris Papavasiliou <dpapavas@gmail.com>
5282 * doc/invoke.texi: Document new switches -Wno-shadow-ivar,
5283 -fno-local-ivars and -fivar-visibility.
5284 * c-family/c.opt: Make -Wshadow also implicitly enable
5287 2014-05-12 David Wohlferd <dw@LimeGreenSocks.com>
5289 * doc/tm.texi: Remove reference to deleted macro.
5290 * doc/tm.texi.in: Likewise.
5292 2014-05-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
5295 * config/avr/avr.c (avr_out_store_psi): Use correct constant
5298 2014-05-12 Georg-Johann Lay <avr@gjlay.de>
5301 * config/arm/arm.h (License): Add GCC Runtime Library Exception.
5302 * config/arm/aout.h (License): Same.
5303 * config/arm/bpabi.h (License): Same.
5304 * config/arm/elf.h (License): Same.
5305 * config/arm/linux-elf.h (License): Same.
5306 * config/arm/linux-gas.h (License): Same.
5307 * config/arm/netbsd-elf.h (License): Same.
5308 * config/arm/uclinux-eabi.h (License): Same.
5309 * config/arm/uclinux-elf.h (License): Same.
5310 * config/arm/vxworks.h (License): Same.
5312 2014-05-11 Jakub Jelinek <jakub@redhat.com>
5314 * tree.h (OMP_CLAUSE_LINEAR_STMT): Define.
5315 * tree.c (omp_clause_num_ops): Increase OMP_CLAUSE_LINEAR
5316 number of operands to 3.
5317 (walk_tree_1): Walk all operands of OMP_CLAUSE_LINEAR.
5318 * tree-nested.c (convert_nonlocal_omp_clauses,
5319 convert_local_omp_clauses): Handle OMP_CLAUSE_DEPEND.
5320 * gimplify.c (gimplify_scan_omp_clauses): Handle
5321 OMP_CLAUSE_LINEAR_STMT.
5322 * omp-low.c (lower_rec_input_clauses): Fix typo.
5323 (maybe_add_implicit_barrier_cancel, lower_omp_1): Add
5324 cast between Fortran boolean_type_node and C _Bool if
5327 2014-05-11 Richard Sandiford <rdsandiford@googlemail.com>
5329 PR tree-optimization/61136
5330 * wide-int.h (multiple_of_p): Define a version that doesn't return
5332 * fold-const.c (extract_muldiv_1): Use wi::multiple_of_p instead of an
5333 integer_zerop/const_binop pair.
5334 (multiple_of_p): Likewise, converting both operands to widest_int
5337 2014-05-09 Teresa Johnson <tejohnson@google.com>
5339 * cgraphunit.c (analyze_functions): Use correct dump file.
5341 2014-05-09 Florian Weimer <fweimer@redhat.com>
5343 * cfgexpand.c (stack_protect_decl_p): New function, extracted from
5345 (stack_protect_return_slot_p): New function.
5346 (expand_used_vars): Call stack_protect_decl_p and
5347 stack_protect_return_slot_p for -fstack-protector-strong.
5349 2014-05-09 David Wohlferd <LimeGreenSocks@yahoo.com>
5350 Andrew Haley <aph@redhat.com>
5351 Richard Sandiford <rdsandiford@googlemail.com>
5353 * doc/extend.texi: Rewrite inline asm page / re-org asm-related
5356 2014-05-09 Kenneth Zadeck <zadeck@naturalbridge.com>
5359 * fold-const.c (fold_binary_loc): Changed width of mask.
5361 2014-05-09 Georg-Johann Lay <avr@gjlay.de>
5363 * config/avr/avr-fixed.md (round<mode>3): Use -1U instead of -1 in
5364 unsigned int initializers for regno_in, regno_out.
5366 2014-05-09 Georg-Johann Lay <avr@gjlay.de>
5369 * config/avr/avr.md (cc): Add new attribute set_vzn.
5370 (addqi3, addqq3, adduqq3, subqi3, subqq3, subuqq3, negqi2) [cc]:
5371 Set cc insn attribute to set_vzn instead of set_zn for alternatives
5372 with INC, DEC or NEG.
5373 * config/avr/avr.c (avr_notice_update_cc): Handle SET_VZN.
5374 (avr_out_plus_1): ADIW sets cc0 to CC_SET_CZN.
5375 INC, DEC and ADD+ADC set cc0 to CC_CLOBBER.
5377 2014-05-09 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5380 2014-05-08 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5382 * wide-int.cc (UTItype): Define.
5383 (UDWtype): Define for appropriate W_TYPE_SIZE.
5385 2014-05-09 Richard Biener <rguenther@suse.de>
5387 * Makefile.in (GTFILES): Remove tree-ssa-propagate.c.
5388 * tree-ssa-propagate.c: Do not include gt-tree-ssa-propagate.h.
5389 (interesting_ssa_edges, varying_ssa_edges): Move out of GC space.
5390 (add_ssa_edge, process_ssa_edge_worklist, ssa_prop_init,
5391 ssa_propagate): Adjust.
5393 2014-05-08 Jeff Law <law@redhat.com>
5395 PR tree-optimization/61009
5396 * tree-ssa-threadedge.c (thread_through_normal_block): Return a
5397 tri-state rather than a boolean. When a block is too big to
5398 thread through, inform caller via negative return value.
5399 (thread_across_edge): If a block was too big for normal threading,
5400 then it's too big for a joiner too, so remove temporary equivalences
5401 and return immediately.
5403 2014-05-08 Manuel López-Ibáñez <manu@gcc.gnu.org>
5404 Matthias Klose <doko@ubuntu.com>
5407 * optc-gen.awk: Fix option handling for -Wunused-parameter.
5409 2014-05-08 Uros Bizjak <ubizjak@gmail.com>
5412 * config/i386/i386.c (PTA_HASWELL): Remove PTA_RTM.
5414 2014-05-08 Uros Bizjak <ubizjak@gmail.com>
5417 * config/alpha/alpha.c: Include gimple-iterator.h.
5418 (alpha_gimple_fold_builtin): New function. Move
5419 ALPHA_BUILTIN_UMULH folding from ...
5420 (alpha_fold_builtin): ... here.
5421 (TARGET_GIMPLE_FOLD_BUILTIN): New define.
5423 2014-05-08 Wei Mi <wmi@google.com>
5426 * config/i386/i386.c (ix86_compute_frame_layout): Update
5427 preferred_stack_boundary for call, expanded from tls descriptor.
5428 * config/i386/i386.md (*tls_global_dynamic_32_gnu): Update RTX
5429 to depend on SP register.
5430 (*tls_local_dynamic_base_32_gnu): Ditto.
5431 (*tls_local_dynamic_32_once): Ditto.
5432 (tls_global_dynamic_64_<mode>): Set
5433 ix86_tls_descriptor_calls_expanded_in_cfun.
5434 (tls_local_dynamic_base_64_<mode>): Ditto.
5435 (tls_global_dynamic_32): Set
5436 ix86_tls_descriptor_calls_expanded_in_cfun. Update RTX
5437 to depend on SP register.
5438 (tls_local_dynamic_base_32): Ditto.
5440 2014-05-08 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5442 * config/arm/arm_neon.h: Update comment.
5443 * config/arm/neon-docgen.ml: Delete.
5444 * config/arm/neon-gen.ml: Delete.
5445 * doc/arm-neon-intrinsics.texi: Update comment.
5447 2014-05-08 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5449 * config/arm/arm_neon_builtins.def (vadd, vsub): Only define the v2sf
5451 (vand, vorr, veor, vorn, vbic): Remove.
5452 * config/arm/neon.md (neon_vadd, neon_vsub, neon_vadd_unspec): Adjust
5454 (neon_vsub_unspec): Likewise.
5455 (neon_vorr, neon_vand, neon_vbic, neon_veor, neon_vorn): Remove.
5457 2014-05-08 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5459 * config/arm/arm_neon.h (vadd_s8): GNU C implementation
5460 (vadd_s16): Likewise.
5461 (vadd_s32): Likewise.
5462 (vadd_f32): Likewise.
5463 (vadd_u8): Likewise.
5464 (vadd_u16): Likewise.
5465 (vadd_u32): Likewise.
5466 (vadd_s64): Likewise.
5467 (vadd_u64): Likewise.
5468 (vaddq_s8): Likewise.
5469 (vaddq_s16): Likewise.
5470 (vaddq_s32): Likewise.
5471 (vaddq_s64): Likewise.
5472 (vaddq_f32): Likewise.
5473 (vaddq_u8): Likewise.
5474 (vaddq_u16): Likewise.
5475 (vaddq_u32): Likewise.
5476 (vaddq_u64): Likewise.
5477 (vmul_s8): Likewise.
5478 (vmul_s16): Likewise.
5479 (vmul_s32): Likewise.
5480 (vmul_f32): Likewise.
5481 (vmul_u8): Likewise.
5482 (vmul_u16): Likewise.
5483 (vmul_u32): Likewise.
5484 (vmul_p8): Likewise.
5485 (vmulq_s8): Likewise.
5486 (vmulq_s16): Likewise.
5487 (vmulq_s32): Likewise.
5488 (vmulq_f32): Likewise.
5489 (vmulq_u8): Likewise.
5490 (vmulq_u16): Likewise.
5491 (vmulq_u32): Likewise.
5492 (vsub_s8): Likewise.
5493 (vsub_s16): Likewise.
5494 (vsub_s32): Likewise.
5495 (vsub_f32): Likewise.
5496 (vsub_u8): Likewise.
5497 (vsub_u16): Likewise.
5498 (vsub_u32): Likewise.
5499 (vsub_s64): Likewise.
5500 (vsub_u64): Likewise.
5501 (vsubq_s8): Likewise.
5502 (vsubq_s16): Likewise.
5503 (vsubq_s32): Likewise.
5504 (vsubq_s64): Likewise.
5505 (vsubq_f32): Likewise.
5506 (vsubq_u8): Likewise.
5507 (vsubq_u16): Likewise.
5508 (vsubq_u32): Likewise.
5509 (vsubq_u64): Likewise.
5510 (vand_s8): Likewise.
5511 (vand_s16): Likewise.
5512 (vand_s32): Likewise.
5513 (vand_u8): Likewise.
5514 (vand_u16): Likewise.
5515 (vand_u32): Likewise.
5516 (vand_s64): Likewise.
5517 (vand_u64): Likewise.
5518 (vandq_s8): Likewise.
5519 (vandq_s16): Likewise.
5520 (vandq_s32): Likewise.
5521 (vandq_s64): Likewise.
5522 (vandq_u8): Likewise.
5523 (vandq_u16): Likewise.
5524 (vandq_u32): Likewise.
5525 (vandq_u64): Likewise.
5526 (vorr_s8): Likewise.
5527 (vorr_s16): Likewise.
5528 (vorr_s32): Likewise.
5529 (vorr_u8): Likewise.
5530 (vorr_u16): Likewise.
5531 (vorr_u32): Likewise.
5532 (vorr_s64): Likewise.
5533 (vorr_u64): Likewise.
5534 (vorrq_s8): Likewise.
5535 (vorrq_s16): Likewise.
5536 (vorrq_s32): Likewise.
5537 (vorrq_s64): Likewise.
5538 (vorrq_u8): Likewise.
5539 (vorrq_u16): Likewise.
5540 (vorrq_u32): Likewise.
5541 (vorrq_u64): Likewise.
5542 (veor_s8): Likewise.
5543 (veor_s16): Likewise.
5544 (veor_s32): Likewise.
5545 (veor_u8): Likewise.
5546 (veor_u16): Likewise.
5547 (veor_u32): Likewise.
5548 (veor_s64): Likewise.
5549 (veor_u64): Likewise.
5550 (veorq_s8): Likewise.
5551 (veorq_s16): Likewise.
5552 (veorq_s32): Likewise.
5553 (veorq_s64): Likewise.
5554 (veorq_u8): Likewise.
5555 (veorq_u16): Likewise.
5556 (veorq_u32): Likewise.
5557 (veorq_u64): Likewise.
5558 (vbic_s8): Likewise.
5559 (vbic_s16): Likewise.
5560 (vbic_s32): Likewise.
5561 (vbic_u8): Likewise.
5562 (vbic_u16): Likewise.
5563 (vbic_u32): Likewise.
5564 (vbic_s64): Likewise.
5565 (vbic_u64): Likewise.
5566 (vbicq_s8): Likewise.
5567 (vbicq_s16): Likewise.
5568 (vbicq_s32): Likewise.
5569 (vbicq_s64): Likewise.
5570 (vbicq_u8): Likewise.
5571 (vbicq_u16): Likewise.
5572 (vbicq_u32): Likewise.
5573 (vbicq_u64): Likewise.
5574 (vorn_s8): Likewise.
5575 (vorn_s16): Likewise.
5576 (vorn_s32): Likewise.
5577 (vorn_u8): Likewise.
5578 (vorn_u16): Likewise.
5579 (vorn_u32): Likewise.
5580 (vorn_s64): Likewise.
5581 (vorn_u64): Likewise.
5582 (vornq_s8): Likewise.
5583 (vornq_s16): Likewise.
5584 (vornq_s32): Likewise.
5585 (vornq_s64): Likewise.
5586 (vornq_u8): Likewise.
5587 (vornq_u16): Likewise.
5588 (vornq_u32): Likewise.
5589 (vornq_u64): Likewise.
5591 2014-05-08 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5593 * wide-int.cc (UTItype): Define.
5594 (UDWtype): Define for appropriate W_TYPE_SIZE.
5596 2014-05-08 Marc Glisse <marc.glisse@inria.fr>
5598 PR tree-optimization/59100
5599 * tree-ssa-phiopt.c: Include tree-inline.h.
5600 (neutral_element_p, absorbing_element_p): New functions.
5601 (value_replacement): Handle conditional binary operations with a
5602 neutral or absorbing element.
5604 2014-05-08 Richard Biener <rguenther@suse.de>
5606 * tree-ssa-sccvn.c (vn_get_expr_for): Valueize operands before
5607 folding the expression.
5608 (valueize_expr): Remove.
5609 (visit_reference_op_load): Do not valueize the result of
5611 (simplify_binary_expression): Likewise.
5612 (simplify_unary_expression): Likewise.
5614 2014-05-08 Richard Biener <rguenther@suse.de>
5616 * gimplify.c (gimplify_call_expr): Use saved fnptrtype for
5617 looking at TYPE_ARG_TYPES.
5619 2014-05-08 Richard Biener <rguenther@suse.de>
5621 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Remove
5622 pointer propagation special-case.
5624 2014-05-08 Bin Cheng <bin.cheng@arm.com>
5626 * tree-affine.c (tree_to_aff_combination): Handle MEM_REF for
5627 core part of address expressions.
5629 2014-05-08 Alan Modra <amodra@gmail.com>
5632 * config/rs6000/rs6000.c (expand_block_move): Allow 64-bit
5633 loads and stores when -mno-strict-align at any alignment.
5634 (expand_block_clear): Similarly. Also correct calculation of
5637 2014-05-07 Thomas Preud'homme <thomas.preudhomme@arm.com>
5640 * tree-complex.c (expand_complex_move): Keep line info when expanding
5642 * tree-ssa-uninit.c (warn_uninit): New argument. Ignore assignment
5643 of complex expression. Use new argument to display correct location
5644 for values coming from phi statement.
5645 (warn_uninitialized_vars): Adapt to new signature of warn_uninit.
5646 (warn_uninitialized_phi): Pass location of phi argument to
5648 * tree-ssa.c (ssa_undefined_value_p): For SSA_NAME initialized by a
5649 COMPLEX_EXPR, recurse on each part of the COMPLEX_EXPR.
5651 2014-05-07 Segher Boessenkool <segher@kernel.crashing.org>
5653 * config/rs6000/predicates.md (indexed_address_mem): New.
5654 * config/rs6000/rs6000.md (type): Remove load_ext, load_ext_u,
5655 load_ext_ux, load_ux, load_u, store_ux, store_u, fpload_ux, fpload_u,
5656 fpstore_ux, fpstore_u.
5657 (sign_extend, indexed, update): New.
5658 (cell_micro): Adjust.
5659 (*zero_extend<mode>di2_internal1, *zero_extendsidi2_lfiwzx,
5660 *extendsidi2_lfiwax, *extendsidi2_nocell, *extendsfdf2_fpr,
5661 *movsi_internal1, *movsi_internal1_single, *movhi_internal,
5662 *movqi_internal, *movcc_internal1, mov<mode>_hardfloat,
5663 *mov<mode>_softfloat, *mov<mode>_hardfloat32, *mov<mode>_hardfloat64,
5664 *mov<mode>_softfloat64, *movdi_internal32, *movdi_internal64,
5665 *mov<mode>_string, *ldmsi8, *ldmsi7, *ldmsi6, *ldmsi5, *ldmsi4,
5666 *ldmsi3, *stmsi8, *stmsi7, *stmsi6, *stmsi5, *stmsi4, *stmsi3,
5667 *movdi_update1, movdi_<mode>_update, movdi_<mode>_update_stack,
5668 *movsi_update1, *movsi_update2, movsi_update, movsi_update_stack,
5669 *movhi_update1, *movhi_update2, *movhi_update3, *movhi_update4,
5670 *movqi_update1, *movqi_update2, *movqi_update3, *movsf_update1,
5671 *movsf_update2, *movsf_update3, *movsf_update4, *movdf_update1,
5672 *movdf_update2, load_toc_aix_si, load_toc_aix_di, probe_stack_<mode>,
5673 *stmw, *lmw, as well as 10 anonymous patterns): Adjust.
5675 * config/rs6000/dfp.md (movsd_store, movsd_load): Adjust.
5676 * config/rs6000/vsx.md (*vsx_movti_32bit, *vsx_extract_<mode>_load,
5677 *vsx_extract_<mode>_store): Adjust.
5678 * config/rs6000/rs6000.c (rs6000_adjust_cost, is_microcoded_insn,
5679 is_cracked_insn, insn_must_be_first_in_group,
5680 insn_must_be_last_in_group): Adjust.
5682 * config/rs6000/40x.md (ppc403-load, ppc403-store, ppc405-float):
5684 * config/rs6000/440.md (ppc440-load, ppc440-store, ppc440-fpload,
5685 ppc440-fpstore): Adjust.
5686 * config/rs6000/476.md (ppc476-load, ppc476-store, ppc476-fpload,
5687 ppc476-fpstore): Adjust.
5688 * config/rs6000/601.md (ppc601-load, ppc601-store, ppc601-fpload,
5689 ppc601-fpstore): Adjust.
5690 * config/rs6000/603.md (ppc603-load, ppc603-store, ppc603-fpload):
5692 * config/rs6000/6xx.md (ppc604-load, ppc604-store, ppc604-fpload):
5694 * config/rs6000/7450.md (ppc7450-load, ppc7450-store, ppc7450-fpload,
5695 ppc7450-fpstore): Adjust.
5696 * config/rs6000/7xx.md (ppc750-load, ppc750-store): Adjust.
5697 * config/rs6000/8540.md (ppc8540_load, ppc8540_store): Adjust.
5698 * config/rs6000/a2.md (ppca2-load, ppca2-fp-load, ppca2-fp-store):
5700 * config/rs6000/cell.md (cell-load, cell-load-ux, cell-load-ext,
5701 cell-fpload, cell-fpload-update, cell-store, cell-store-update,
5702 cell-fpstore, cell-fpstore-update): Adjust.
5703 * config/rs6000/e300c2c3.md (ppce300c3_load, ppce300c3_fpload,
5704 ppce300c3_store, ppce300c3_fpstore): Adjust.
5705 * config/rs6000/e500mc.md (e500mc_load, e500mc_fpload, e500mc_store,
5706 e500mc_fpstore): Adjust.
5707 * config/rs6000/e500mc64.md (e500mc64_load, e500mc64_fpload,
5708 e500mc64_store, e500mc64_fpstore): Adjust.
5709 * config/rs6000/e5500.md (e5500_load, e5500_fpload, e5500_store,
5710 e5500_fpstore): Adjust.
5711 * config/rs6000/e6500.md (e6500_load, e6500_fpload, e6500_store,
5712 e6500_fpstore): Adjust.
5713 * config/rs6000/mpc.md (mpccore-load, mpccore-store, mpccore-fpload):
5715 * config/rs6000/power4.md (power4-load, power4-load-ext,
5716 power4-load-ext-update, power4-load-ext-update-indexed,
5717 power4-load-update-indexed, power4-load-update, power4-fpload,
5718 power4-fpload-update, power4-store, power4-store-update,
5719 power4-store-update-indexed, power4-fpstore, power4-fpstore-update):
5721 * config/rs6000/power5.md (power5-load, power5-load-ext,
5722 power5-load-ext-update, power5-load-ext-update-indexed,
5723 power5-load-update-indexed, power5-load-update, power5-fpload,
5724 power5-fpload-update, power5-store, power5-store-update,
5725 power5-store-update-indexed, power5-fpstore, power5-fpstore-update):
5727 * config/rs6000/power6.md (power6-load, power6-load-ext,
5728 power6-load-update, power6-load-update-indexed,
5729 power6-load-ext-update, power6-load-ext-update-indexed, power6-fpload,
5730 power6-fpload-update, power6-store, power6-store-update,
5731 power6-store-update-indexed, power6-fpstore, power6-fpstore-update):
5733 * config/rs6000/power7.md (power7-load, power7-load-ext,
5734 power7-load-update, power7-load-update-indexed,
5735 power7-load-ext-update, power7-load-ext-update-indexed, power7-fpload,
5736 power7-fpload-update, power7-store, power7-store-update,
5737 power7-store-update-indexed, power7-fpstore, power7-fpstore-update):
5739 * config/rs6000/power8.md (power8-load, power8-load-update,
5740 power8-load-ext, power8-load-ext-update, power8-fpload,
5741 power8-fpload-update, power8-store, power8-store-update-indexed,
5742 power8-fpstore, power8-fpstore-update): Adjust.
5743 * config/rs6000/rs64.md (rs64a-load, rs64a-store, rs64a-fpload):
5745 * config/rs6000/titan.md (titan_lsu_load, titan_lsu_fpload,
5746 titan_lsu_store, titan_lsu_fpstore): Adjust.
5747 * config/rs6000/xfpu.md (fp-load, fp-store): Adjust.
5749 2014-05-07 Oleg Endo <olegendo@gcc.gnu.org>
5752 * config/sh/sh-mem.cc (sh_expand_strlen): Use loop when emitting
5753 unrolled byte insns. Emit address increments after move insns.
5755 2014-05-07 David Malcolm <dmalcolm@redhat.com>
5757 * gimple.h (gimple_builtin_call_types_compatible_p): Accept a
5758 const_gimple, rather than a gimple.
5759 (gimple_call_builtin_p): Likewise, for the three variants.
5761 * gimple.c (gimple_builtin_call_types_compatible_p): Likewise.
5762 (gimple_call_builtin_p): Likewise, for the three variants.
5764 2014-05-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5766 PR tree-optimization/61095
5767 * tree-ssanames.c (get_nonzero_bits): Fix type extension in wi::shwi.
5769 2014-05-07 Richard Biener <rguenther@suse.de>
5771 PR tree-optimization/61034
5772 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Export.
5773 (maybe_skip_until): Use translate to take into account
5774 lattices when trying to do disambiguations.
5775 (get_continuation_for_phi_1): Likewise.
5776 (get_continuation_for_phi): Adjust for added translate arguments.
5777 (walk_non_aliased_vuses): Likewise.
5778 * tree-ssa-alias.h (get_continuation_for_phi): Adjust prototype.
5779 (walk_non_aliased_vuses): Likewise.
5780 (call_may_clobber_ref_p_1): Declare.
5781 * tree-ssa-sccvn.c (vn_reference_lookup_3): Also disambiguate against
5782 calls. Stop early if we are only supposed to disambiguate.
5783 * tree-ssa-pre.c (translate_vuse_through_block): Adjust.
5785 2014-05-07 Joern Rennecke <joern.rennecke@embecosm.com>
5787 * config/epiphany/epiphany.c (epiphany_handle_interrupt_attribute):
5788 Emit an error when the function has arguments.
5790 2014-05-07 Thomas Schwinge <thomas@codesourcery.com>
5792 * cfgloop.h (unswitch_loops): Remove.
5793 * doc/passes.texi: Remove references to loop-unswitch.c
5794 * timevar.def (TV_LOOP_UNSWITCH): Remove.
5796 2014-05-07 Evgeny Stupachenko <evstupac@gmail.com>
5798 * tree-vect-data-refs.c (vect_grouped_load_supported): New
5799 check for loads group of length 3.
5800 (vect_permute_load_chain): New permutations for loads group of
5802 * tree-vect-stmts.c (vect_model_load_cost): Change cost
5803 of vec_perm_shuffle for the new permutations.
5805 2014-05-07 Alan Lawrence <alan.lawrence@arm.com>
5807 * config/aarch64/arm_neon.h (vtrn1_f32, vtrn1_p8, vtrn1_p16, vtrn1_s8,
5808 vtrn1_s16, vtrn1_s32, vtrn1_u8, vtrn1_u16, vtrn1_u32, vtrn1q_f32,
5809 vtrn1q_f64, vtrn1q_p8, vtrn1q_p16, vtrn1q_s8, vtrn1q_s16, vtrn1q_s32,
5810 vtrn1q_s64, vtrn1q_u8, vtrn1q_u16, vtrn1q_u32, vtrn1q_u64, vtrn2_f32,
5811 vtrn2_p8, vtrn2_p16, vtrn2_s8, vtrn2_s16, vtrn2_s32, vtrn2_u8,
5812 vtrn2_u16, vtrn2_u32, vtrn2q_f32, vtrn2q_f64, vtrn2q_p8, vtrn2q_p16,
5813 vtrn2q_s8, vtrn2q_s16, vtrn2q_s32, vtrn2q_s64, vtrn2q_u8, vtrn2q_u16,
5814 vtrn2q_u32, vtrn2q_u64): Replace temporary asm with __builtin_shuffle.
5816 2014-05-07 Thomas Schwinge <thomas@codesourcery.com>
5818 * loop-unswitch.c: Delete.
5820 2014-05-07 Richard Biener <rguenther@suse.de>
5822 * config.gcc: Always set need_64bit_hwint to yes.
5824 2014-05-07 Chung-Ju Wu <jasonwucj@gmail.com>
5826 * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Have it in favor
5827 of using optimize_size.
5829 2014-05-06 Mike Stump <mikestump@comcast.net>
5831 * wide-int.h (wi::int_traits <HOST_WIDE_INT>): Always define.
5833 2014-05-06 Joseph Myers <joseph@codesourcery.com>
5835 * config/i386/sse.md (*mov<mode>_internal)
5836 (*<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>)
5837 (*<sse2_avx_avx512f>_loaddqu<mode><mask_name>)
5838 (<sse>_andnot<mode>3, <code><mode>3, *andnot<mode>3)
5839 (*<code><mode>3, *andnot<mode>3<mask_name>)
5840 (<mask_codefor><code><mode>3<mask_name>): Only consider
5841 TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL for modes of size 16.
5843 2014-05-06 Richard Sandiford <rdsandiford@googlemail.com>
5846 2014-05-03 Richard Sandiford <rdsandiford@googlemail.com>
5848 * lra-constraints.c (valid_address_p): Move earlier in file.
5849 Add a constraint argument to the address_info version.
5850 (satisfies_memory_constraint_p): New function.
5851 (satisfies_address_constraint_p): Likewise.
5852 (process_alt_operands, curr_insn_transform): Use them.
5853 (process_address): Pass the constraint to valid_address_p when
5854 checking address operands.
5856 2014-05-06 Richard Sandiford <r.sandiford@uk.ibm.com>
5858 * lto-cgraph.c (compute_ltrans_boundary): Make node variables local
5859 to their respective blocks. Fix inadvertent use of "node".
5861 2014-05-06 Richard Sandiford <rdsandiford@googlemail.com>
5863 * emit-rtl.c (init_derived_machine_modes): New functionm, split
5865 (init_emit_once): ...here.
5866 * rtl.h (init_derived_machine_modes): Declare.
5867 * toplev.c (do_compile): Call it even if no_backend.
5869 2014-05-06 Kenneth Zadeck <zadeck@naturalbridge.com>
5870 Mike Stump <mikestump@comcast.net>
5871 Richard Sandiford <rdsandiford@googlemail.com>
5872 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5874 * alias.c (ao_ref_from_mem): Use wide-int interfaces.
5875 (rtx_equal_for_memref_p): Update comment.
5876 (adjust_offset_for_component_ref): Use wide-int interfaces.
5877 * builtins.c (get_object_alignment_2): Likewise.
5878 (c_readstr): Likewise.
5879 (target_char_cast): Add comment.
5880 (determine_block_size): Use wide-int interfaces.
5881 (expand_builtin_signbit): Likewise.
5882 (fold_builtin_int_roundingfn): Likewise.
5883 (fold_builtin_bitop): Likewise.
5884 (fold_builtin_bswap): Likewise.
5885 (fold_builtin_logarithm): Use signop.
5886 (fold_builtin_pow): Likewise.
5887 (fold_builtin_memory_op): Use wide-int interfaces.
5888 (fold_builtin_object_size): Likewise.
5889 * cfgloop.c (alloc_loop): Initialize nb_iterations_upper_bound and
5890 nb_iterations_estimate.
5891 (record_niter_bound): Use wide-int interfaces.
5892 (get_estimated_loop_iterations_int): Likewise.
5893 (get_estimated_loop_iterations): Likewise.
5894 (get_max_loop_iterations): Likewise.
5895 * cfgloop.h: Include wide-int.h.
5896 (struct nb_iter_bound): Change bound to widest_int.
5897 (struct loop): Change nb_iterations_upper_bound and
5898 nb_iterations_estimate to widest_int.
5899 (record_niter_bound): Switch to use widest_int.
5900 (get_estimated_loop_iterations): Likewise.
5901 (get_max_loop_iterations): Likewise.
5902 (gcov_type_to_double_int): Rename to gcov_type_to_wide_int and
5903 update for wide-int.
5904 * cgraph.c (cgraph_add_thunk): Use wide-int interfaces.
5905 * combine.c (try_combine): Likewise.
5906 (subst): Use CONST_SCALAR_INT_P rather than CONST_INT_P.
5907 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Use wide-int
5909 (aarch64_float_const_representable_p): Likewise.
5910 * config/arc/arc.c: Include wide-int.h.
5911 (arc_can_use_doloop_p): Use wide-int interfaces.
5912 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5913 (vfp3_const_double_index): Likewise.
5914 * config/avr/avr.c (avr_out_round): Likewise.
5915 (avr_fold_builtin): Likewise.
5916 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5917 (bfin_can_use_doloop_p): Likewise.
5918 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5919 (machopic_select_rtx_section): Update to handle CONST_WIDE_INT.
5920 * config/i386/i386.c: Include wide-int.h.
5921 (ix86_data_alignment): Use wide-int interfaces.
5922 (ix86_local_alignment): Likewise.
5923 (ix86_emit_swsqrtsf): Update real_from_integer.
5924 * config/msp430/msp430.c (msp430_attr): Use wide-int interfaces.
5925 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5926 * config/rs6000/predicates.md (any_operand): Add const_wide_int.
5927 (zero_constant): Likewise.
5928 (input_operand): Likewise.
5929 (splat_input_operand): Likewise.
5930 (non_logical_cint_operand): Change const_double to const_wide_int.
5931 * config/rs6000/rs6000.c (num_insns_constant): Handle CONST_WIDE_INT.
5932 (easy_altivec_constant): Remove comment.
5933 (paired_expand_vector_init): Use CONSTANT_P.
5934 (rs6000_legitimize_address): Handle CONST_WIDE_INT.
5935 (rs6000_emit_move): Update checks.
5936 (rs6000_aggregate_candidate): Use wide-int interfaces.
5937 (rs6000_expand_ternop_builtin): Likewise.
5938 (rs6000_output_move_128bit): Handle CONST_WIDE_INT.
5939 (rs6000_assemble_integer): Likewise.
5940 (rs6000_hash_constant): Likewise.
5941 (output_toc): Likewise.
5942 (rs6000_rtx_costs): Likewise.
5943 (rs6000_emit_swrsqrt); Update call to real_from_integer.
5944 * config/rs6000/rs6000-c.c: Include wide-int.h.
5945 (altivec_resolve_overloaded_builtin): Use wide-int interfaces.
5946 * config/rs6000/rs6000.h (TARGET_SUPPORTS_WIDE_INT): New.
5947 * config/rs6000/rs6000.md: Use const_scalar_int_operand.
5948 Handle CONST_WIDE_INT.
5949 * config/sol2-c.c (solaris_pragma_align): Change low to unsigned HWI.
5950 Use tree_fits_uhwi_p.
5951 * config/sparc/sparc.c: Include wide-int.h.
5952 (sparc_fold_builtin): Use wide-int interfaces.
5953 * config/vax/vax.c: Include wide-int.h.
5954 (vax_float_literal): Use real_from_integer.
5955 * coretypes.h (struct hwivec_def): New.
5957 (const_hwivec): New.
5958 * cse.c (hash_rtx_cb): Handle CONST_WIDE_INT.
5959 (equiv_constant): Handle CONST_WIDE_INT.
5960 * cselib.c (rtx_equal_for_cselib_1): Use CASE_CONST_UNIQUE.
5961 (cselib_hash_rtx): Handle CONST_WIDE_INT.
5962 * dbxout.c (stabstr_U): Use wide-int interfaces.
5963 (dbxout_type): Update to use cst_fits_shwi_p.
5964 * defaults.h (LOG2_BITS_PER_UNIT): Define.
5965 (TARGET_SUPPORTS_WIDE_INT): Add default.
5966 * dfp.c: Include wide-int.h.
5967 (decimal_real_to_integer2): Use wide-int interfaces and rename to
5968 decimal_real_to_integer.
5969 * dfp.h (decimal_real_to_integer2): Return a wide_int and rename to
5970 decimal_real_to_integer.
5971 * doc/generic.texi (Constant expressions): Update for wide_int.
5972 * doc/rtl.texi (const_double): Likewise.
5973 (const_wide_int, CONST_WIDE_INT, CONST_WIDE_INT_VEC): New.
5974 (CONST_WIDE_INT_NUNITS, CONST_WIDE_INT_ELT): New.
5975 * doc/tm.texi.in (REAL_VALUE_TO_INT): Remove.
5976 (REAL_VALUE_FROM_INT): Remove.
5977 (TARGET_SUPPORTS_WIDE_INT): New.
5978 * doc/tm.texi: Regenerate.
5979 * dojump.c (prefer_and_bit_test): Use wide-int interfaces.
5980 * double-int.h: Include wide-int.h.
5981 (struct wi::int_traits): New.
5982 * dwarf2out.c (get_full_len): New.
5983 (dw_val_equal_p): Add case dw_val_class_wide_int.
5984 (size_of_loc_descr): Likewise.
5985 (output_loc_operands): Likewise.
5986 (insert_double): Remove.
5987 (insert_wide_int): New.
5989 (print_die): Add case dw_val_class_wide_int.
5990 (attr_checksum): Likewise.
5991 (attr_checksum_ordered): Likewise.
5992 (same_dw_val_p): Likewise.
5993 (size_of_die): Likewise.
5994 (value_format): Likewise.
5995 (output_die): Likewise.
5996 (double_int_type_size_in_bits): Rename to offset_int_type_size_in_bits.
5998 (clz_loc_descriptor): Use wide-int interfaces.
5999 (mem_loc_descriptor): Likewise. Handle CONST_WIDE_INT.
6000 (loc_descriptor): Use wide-int interfaces. Handle CONST_WIDE_INT.
6001 (round_up_to_align): Use wide-int interfaces.
6002 (field_byte_offset): Likewise.
6003 (insert_double): Rename to insert_wide_int. Use wide-int interfaces.
6004 (add_const_value_attribute): Handle CONST_WIDE_INT. Update
6005 CONST_DOUBLE handling. Use wide-int interfaces.
6006 (add_bound_info): Use tree_fits_uhwi_p. Use wide-int interfaces.
6007 (gen_enumeration_type_die): Use add_AT_wide.
6008 (hash_loc_operands): Add case dw_val_class_wide_int.
6009 (compare_loc_operands): Likewise.
6010 * dwarf2out.h: Include wide-int.h.
6011 (wide_int_ptr): New.
6012 (enum dw_val_class): Add dw_val_class_wide_int.
6013 (struct dw_val_struct): Add val_wide.
6014 * emit-rtl.c (const_wide_int_htab): New.
6015 (const_wide_int_htab_hash): New.
6016 (const_wide_int_htab_eq): New.
6017 (lookup_const_wide_int): New.
6018 (const_double_htab_hash): Use wide-int interfaces.
6019 (const_double_htab_eq): Likewise.
6020 (rtx_to_double_int): Conditionally compile for wide-int.
6021 (immed_double_int_const): Rename to immed_wide_int_const and
6022 update for wide-int.
6023 (immed_double_const): Conditionally compile for wide-int.
6024 (init_emit_once): Use wide-int interfaces.
6025 * explow.c (plus_constant): Likewise.
6026 * expmed.c (mask_rtx): Move further up file. Use wide-int interfaces.
6027 (lshift_value): Use wide-int interfaces.
6028 (expand_mult): Likewise.
6029 (choose_multiplier): Likewise.
6030 (expand_smod_pow2): Likewise.
6031 (make_tree): Likewise.
6032 * expr.c (convert_modes): Consolidate handling of constants.
6033 Use wide-int interfaces.
6034 (emit_group_load_1): Add note.
6035 (store_expr): Update comment.
6036 (get_inner_reference): Use wide-int interfaces.
6037 (expand_constructor): Update comment.
6038 (expand_expr_real_2): Use wide-int interfaces.
6039 (expand_expr_real_1): Likewise.
6040 (reduce_to_bit_field_precision): Likewise.
6041 (const_vector_from_tree): Likewise.
6042 * final.c: Include wide-int-print.h.
6043 (output_addr_const): Handle CONST_WIDE_INT. Use CONST_DOUBLE_AS_INT_P.
6044 * fixed-value.c: Include wide-int.h.
6045 (fixed_from_string): Use wide-int interfaces.
6046 (fixed_to_decimal): Likewise.
6047 (fixed_convert_from_real): Likewise.
6048 (real_convert_from_fixed): Likewise.
6049 * fold-const.h (mem_ref_offset): Return an offset_int.
6050 (div_if_zero_remainder): Remove code parameter.
6051 * fold-const.c (div_if_zero_remainder): Remove code parameter.
6052 Use wide-int interfaces.
6053 (may_negate_without_overflow_p): Use wide-int interfaces.
6054 (negate_expr_p): Likewise.
6055 (fold_negate_expr): Likewise.
6056 (int_const_binop_1): Likewise.
6057 (const_binop): Likewise.
6058 (fold_convert_const_int_from_int): Likewise.
6059 (fold_convert_const_int_from_real): Likewise.
6060 (fold_convert_const_int_from_fixed): Likewise.
6061 (fold_convert_const_fixed_from_int): Likewise.
6062 (all_ones_mask_p): Take an unsigned size. Use wide-int interfaces.
6063 (sign_bit_p): Use wide-int interfaces.
6064 (make_range_step): Likewise.
6065 (build_range_check): Likewise. Pass an integer of the correct type
6066 instead of using integer_one_node.
6067 (range_predecessor): Pass an integer of the correct type instead
6068 of using integer_one_node.
6069 (range_successor): Likewise.
6070 (merge_ranges): Likewise.
6071 (unextend): Use wide-int interfaces.
6072 (extract_muldiv_1): Likewise.
6073 (fold_div_compare): Likewise.
6074 (fold_single_bit_test): Likewise.
6075 (fold_sign_changed_comparison): Likewise.
6076 (try_move_mult_to_index): Update calls to div_if_zero_remainder.
6077 (fold_plusminus_mult_expr): Use wide-int interfaces.
6078 (native_encode_int): Likewise.
6079 (native_interpret_int): Likewise.
6080 (fold_unary_loc): Likewise.
6081 (pointer_may_wrap_p): Likewise.
6082 (size_low_cst): Likewise.
6083 (mask_with_tz): Likewise.
6084 (fold_binary_loc): Likewise.
6085 (fold_ternary_loc): Likewise.
6086 (multiple_of_p): Likewise.
6087 (tree_call_nonnegative_warnv_p): Update calls to
6088 tree_int_cst_min_precision and real_from_integer.
6089 (fold_negate_const): Use wide-int interfaces.
6090 (fold_abs_const): Likewise.
6091 (fold_relational_const): Use tree_int_cst_lt.
6092 (round_up_loc): Use wide-int interfaces.
6093 * genemit.c (gen_exp): Add CONST_WIDE_INT case.
6094 * gengenrtl.c (excluded_rtx): Add CONST_WIDE_INT case.
6095 * gengtype.c: Remove include of double-int.h.
6096 (do_typedef): Use wide-int interfaces.
6097 (open_base_files): Add wide-int.h.
6098 (main): Add offset_int and widest_int typedefs.
6099 * gengtype-lex.l: Handle "^".
6100 (CXX_KEYWORD): Add "static".
6101 * gengtype-parse.c (require3): New.
6102 (require_template_declaration): Handle constant template arguments
6103 and nested templates.
6104 * gengtype-state.c: Don't include "double-int.h".
6105 * genpreds.c (write_one_predicate_function): Update comment.
6106 (write_tm_constrs_h): Add check for hval and lval use in
6108 * genrecog.c (validate_pattern): Add CONST_WIDE_INT case.
6109 (add_to_sequence): Likewise.
6110 * gensupport.c (struct std_pred_table): Add const_scalar_int_operand
6111 and const_double_operand.
6112 * gimple.c (preprocess_case_label_vec_for_gimple): Use wide-int
6114 * gimple-fold.c (get_base_constructor): Likewise.
6115 (fold_array_ctor_reference): Likewise.
6116 (fold_nonarray_ctor_reference): Likewise.
6117 (fold_const_aggregate_ref_1): Likewise.
6118 (gimple_val_nonnegative_real_p): Likewise.
6119 (gimple_fold_indirect_ref): Likewise.
6120 * gimple-pretty-print.c (dump_ssaname_info): Likewise.
6121 * gimple-ssa-strength-reduction.c: Include wide-int-print.h.
6122 (struct slsr_cand_d): Change index to be widest_int.
6123 (struct incr_info_d): Change incr to be widest_int.
6124 (alloc_cand_and_find_basis): Use wide-int interfaces.
6125 (slsr_process_phi): Likewise.
6126 (backtrace_base_for_ref): Likewise. Return a widest_int.
6127 (restructure_reference): Take a widest_int instead of a double_int.
6128 (slsr_process_ref): Use wide-int interfaces.
6129 (create_mul_ssa_cand): Likewise.
6130 (create_mul_imm_cand): Likewise.
6131 (create_add_ssa_cand): Likewise.
6132 (create_add_imm_cand): Take a widest_int instead of a double_int.
6133 (slsr_process_add): Use wide-int interfaces.
6134 (slsr_process_cast): Likewise.
6135 (slsr_process_copy): Likewise.
6136 (dump_candidate): Likewise.
6137 (dump_incr_vec): Likewise.
6138 (replace_ref): Likewise.
6139 (cand_increment): Likewise. Return a widest_int.
6140 (cand_abs_increment): Likewise.
6141 (replace_mult_candidate): Take a widest_int instead of a double_int.
6142 (replace_unconditional_candidate): Use wide-int interfaces.
6143 (incr_vec_index): Take a widest_int instead of a double_int.
6144 (create_add_on_incoming_edge): Likewise.
6145 (create_phi_basis): Use wide-int interfaces.
6146 (replace_conditional_candidate): Likewise.
6147 (record_increment): Take a widest_int instead of a double_int.
6148 (record_phi_increments): Use wide-int interfaces.
6149 (phi_incr_cost): Take a widest_int instead of a double_int.
6150 (lowest_cost_path): Likewise.
6151 (total_savings): Likewise.
6152 (analyze_increments): Use wide-int interfaces.
6153 (ncd_with_phi): Take a widest_int instead of a double_int.
6154 (ncd_of_cand_and_phis): Likewise.
6155 (nearest_common_dominator_for_cands): Likewise.
6156 (insert_initializers): Use wide-int interfaces.
6157 (all_phi_incrs_profitable): Likewise.
6158 (replace_one_candidate): Likewise.
6159 (replace_profitable_candidates): Likewise.
6160 * godump.c: Include wide-int-print.h.
6161 (go_output_typedef): Use wide-int interfaces.
6162 * graphite-clast-to-gimple.c (gmp_cst_to_tree): Likewise.
6163 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6164 (build_loop_iteration_domains): Likewise.
6165 * hooks.h: Include wide-int.h rather than double-int.h.
6166 (hook_bool_dint_dint_uint_bool_true): Delete.
6167 (hook_bool_wint_wint_uint_bool_true): Declare.
6168 * hooks.c (hook_bool_dint_dint_uint_bool_true): Removed.
6169 (hook_bool_wint_wint_uint_bool_true): New.
6170 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): Use wide-int
6172 (ubsan_expand_si_overflow_mul_check): Likewise.
6173 * ipa-devirt.c (get_polymorphic_call_info): Likewise.
6174 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
6175 (get_ancestor_addr_info): Likewise.
6176 (ipa_modify_call_arguments): Likewise.
6177 * loop-doloop.c (doloop_modify): Likewise.
6178 (doloop_optimize): Likewise.
6179 * loop-iv.c (iv_number_of_iterations): Likewise.
6180 * loop-unroll.c (decide_unroll_constant_iterations): Likewise.
6181 (unroll_loop_constant_iterations): Likewise.
6182 (decide_unroll_runtime_iterations): Likewise.
6183 (unroll_loop_runtime_iterations): Likewise.
6184 (decide_peel_simple): Likewise.
6185 (decide_unroll_stupid): Likewise.
6186 * lto-streamer-in.c (streamer_read_wi): Add.
6187 (input_cfg): Use wide-int interfaces.
6188 (lto_input_tree_1): Likewise.
6189 * lto-streamer-out.c (streamer_write_wi): Add.
6190 (hash_tree): Use wide-int interfaces.
6191 (output_cfg): Likewise.
6192 * Makefile.in (OBJS): Add wide-int.o and wide-int-print.o.
6193 (GTFILES): Add wide-int.h and signop.h.
6194 (TAGS): Look for .cc files too.
6195 * omp-low.c (scan_omp_1_op): Use wide-int interfaces.
6196 * optabs.c (expand_subword_shift): Likewise.
6197 (expand_doubleword_shift): Likewise.
6198 (expand_absneg_bit): Likewise.
6199 (expand_copysign_absneg): Likewise.
6200 (expand_copysign_bit): Likewise.
6201 * postreload.c (reload_cse_simplify_set): Likewise.
6202 * predict.c (predict_iv_comparison): Likewise.
6203 * pretty-print.h: Include wide-int-print.h.
6205 * print-rtl.c (print_rtx): Add CONST_WIDE_INT case.
6206 * print-tree.c: Include wide-int-print.h.
6207 (print_node_brief): Use wide-int interfaces.
6208 (print_node): Likewise.
6209 * read-rtl.c (validate_const_wide_int): New.
6210 (read_rtx_code): Add CONST_WIDE_INT case.
6211 * real.c: Include wide-int.h.
6212 (real_to_integer2): Delete.
6213 (real_to_integer): New function, returning a wide_int.
6214 (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
6215 (ten_to_ptwo): Update call to real_from_integer.
6216 (real_digit): Likewise.
6217 * real.h: Include signop.h, wide-int.h and insn-modes.h.
6218 (real_to_integer2, REAL_VALUE_FROM_INT, REAL_VALUE_FROM_UNSIGNED_INT)
6219 (REAL_VALUE_TO_INT): Delete.
6220 (real_to_integer): Declare a wide-int form.
6221 (real_from_integer): Take a wide_int rather than two HOST_WIDE_INTs.
6222 * recog.c (const_int_operand): Improve comment.
6223 (const_scalar_int_operand): New.
6224 (const_double_operand): Add a separate definition for CONST_WIDE_INT.
6225 * rtlanal.c (commutative_operand_precedence): Handle CONST_WIDE_INT.
6226 (split_double): Likewise.
6227 * rtl.c (DEF_RTL_EXPR): Handle CONST_WIDE_INT.
6228 (rtx_size): Likewise.
6229 (rtx_alloc_stat_v): New.
6230 (rtx_alloc_stat): Now calls rtx_alloc_stat_v.
6231 (cwi_output_hex): New.
6232 (iterative_hash_rtx): Handle CONST_WIDE_INT.
6233 (cwi_check_failed_bounds): New.
6234 * rtl.def (CONST_WIDE_INT): New.
6235 * rtl.h: Include <utility> and wide-int.h.
6236 (struct hwivec_def): New.
6237 (CWI_GET_NUM_ELEM): New.
6238 (CWI_PUT_NUM_ELEM): New.
6239 (struct rtx_def): Add num_elem and hwiv.
6240 (CASE_CONST_SCALAR_INT): Modify for TARGET_SUPPORTS_WIDE_INT.
6241 (CASE_CONST_UNIQUE): Likewise.
6242 (CASE_CONST_ANY): Likewise.
6243 (CONST_SCALAR_INT_P): Likewise.
6244 (CONST_WIDE_INT_P): New.
6246 (HWIVEC_CHECK): New.
6247 (cwi_check_failed_bounds): New.
6249 (HWIVEC_CHECK): New.
6250 (CONST_WIDE_INT_VEC) New.
6251 (CONST_WIDE_INT_NUNITS) New.
6252 (CONST_WIDE_INT_ELT) New.
6253 (rtx_mode_t): New type.
6254 (wi::int_traits <rtx_mode_t>): New.
6256 (wi::min_value): New.
6257 (wi::max_value): New.
6259 (const_wide_int_alloc): New.
6260 (immed_wide_int_const): New.
6261 * sched-vis.c (print_value): Handle CONST_WIDE_INT.
6262 * sel-sched-ir.c (lhs_and_rhs_separable_p): Update comment.
6263 * signop.h: New file.
6264 * simplify-rtx.c (mode_signbit_p): Handle CONST_WIDE_INT.
6265 (simplify_const_unary_operation): Use wide-int interfaces.
6266 (simplify_binary_operation_1): Likewise.
6267 (simplify_const_binary_operation): Likewise.
6268 (simplify_const_relational_operation): Likewise.
6269 (simplify_immed_subreg): Likewise.
6270 * stmt.c (expand_case): Likewise.
6271 * stor-layout.h (set_min_and_max_values_for_integral_type): Take a
6272 signop rather than a bool.
6273 * stor-layout.c (layout_type): Use wide-int interfaces.
6274 (initialize_sizetypes): Update calls to
6275 set_min_and_max_values_for_integral_type.
6276 (set_min_and_max_values_for_integral_type): Take a signop rather
6277 than a bool. Use wide-int interfaces.
6278 (fixup_signed_type): Update accordingly. Remove
6279 HOST_BITS_PER_DOUBLE_INT limit.
6280 (fixup_unsigned_type): Likewise.
6281 * system.h (STATIC_CONSTANT_P): New.
6282 (STATIC_ASSERT): New.
6283 * target.def (can_use_doloop_p): Take widest_ints rather than
6285 * target.h: Include wide-int.h rather than double-int.h.
6286 * targhooks.h (can_use_doloop_if_innermost): Take widest_ints rather
6288 * targhooks.c (default_cxx_get_cookie_size): Use tree_int_cst_lt
6289 rather than INT_CST_LT_UNSIGNED.
6290 (can_use_doloop_if_innermost): Take widest_ints rather than
6292 * tree-affine.c: Include wide-int-print.h.
6293 (double_int_ext_for_comb): Delete.
6294 (wide_int_ext_for_comb): New.
6295 (aff_combination_zero): Use wide-int interfaces.
6296 (aff_combination_const): Take a widest_int instead of a double_int.
6297 (aff_combination_elt): Use wide-int interfaces.
6298 (aff_combination_scale): Take a widest_int instead of a double_int.
6299 (aff_combination_add_elt): Likewise.
6300 (aff_combination_add_cst): Likewise.
6301 (aff_combination_add): Use wide-int interfaces.
6302 (aff_combination_convert): Likewise.
6303 (tree_to_aff_combination): Likewise.
6304 (add_elt_to_tree): Take a widest_int instead of a double_int.
6305 (aff_combination_to_tree): Use wide-int interfaces.
6306 (aff_combination_remove_elt): Likewise.
6307 (aff_combination_add_product): Take a widest_int instead of
6309 (aff_combination_mult): Use wide-int interfaces.
6310 (aff_combination_expand): Likewise.
6311 (double_int_constant_multiple_p): Delete.
6312 (wide_int_constant_multiple_p): New.
6313 (aff_combination_constant_multiple_p): Take a widest_int pointer
6314 instead of a double_int pointer.
6315 (print_aff): Use wide-int interfaces.
6316 (get_inner_reference_aff): Take a widest_int pointer
6317 instead of a double_int pointer.
6318 (aff_comb_cannot_overlap_p): Take widest_ints instead of double_ints.
6319 * tree-affine.h: Include wide-int.h.
6320 (struct aff_comb_elt): Change type of coef to widest_int.
6321 (struct affine_tree_combination): Change type of offset to widest_int.
6322 (double_int_ext_for_comb): Delete.
6323 (wide_int_ext_for_comb): New.
6324 (aff_combination_const): Use widest_int instead of double_int.
6325 (aff_combination_scale): Likewise.
6326 (aff_combination_add_elt): Likewise.
6327 (aff_combination_constant_multiple_p): Likewise.
6328 (get_inner_reference_aff): Likewise.
6329 (aff_comb_cannot_overlap_p): Likewise.
6330 (aff_combination_zero_p): Use wide-int interfaces.
6331 * tree.c: Include tree.h.
6332 (init_ttree): Use make_int_cst.
6333 (tree_code_size): Removed code for INTEGER_CST case.
6334 (tree_size): Add INTEGER_CST case.
6335 (make_node_stat): Update comment.
6336 (get_int_cst_ext_nunits, build_new_int_cst, build_int_cstu): New.
6337 (build_int_cst_type): Use wide-int interfaces.
6338 (double_int_to_tree): Likewise.
6339 (double_int_fits_to_tree_p): Delete.
6340 (force_fit_type_double): Delete.
6341 (force_fit_type): New.
6342 (int_cst_hash_hash): Use wide-int interfaces.
6343 (int_cst_hash_eq): Likewise.
6344 (build_int_cst_wide): Delete.
6345 (wide_int_to_tree): New.
6346 (cache_integer_cst): Use wide-int interfaces.
6347 (build_low_bits_mask): Likewise.
6348 (cst_and_fits_in_hwi): Likewise.
6349 (real_value_from_int_cst): Likewise.
6350 (make_int_cst_stat): New.
6351 (integer_zerop): Use wide_int interfaces.
6352 (integer_onep): Likewise.
6353 (integer_all_onesp): Likewise.
6354 (integer_pow2p): Likewise.
6355 (integer_nonzerop): Likewise.
6356 (tree_log2): Likewise.
6357 (tree_floor_log2): Likewise.
6358 (tree_ctz): Likewise.
6359 (int_size_in_bytes): Likewise.
6360 (mem_ref_offset): Return an offset_int rather than a double_int.
6361 (build_type_attribute_qual_variant): Use wide_int interfaces.
6362 (type_hash_eq): Likewise
6363 (tree_int_cst_equal): Likewise.
6364 (tree_int_cst_lt): Delete.
6365 (tree_int_cst_compare): Likewise.
6366 (tree_fits_shwi_p): Use wide_int interfaces.
6367 (tree_fits_uhwi_p): Likewise.
6368 (tree_int_cst_sign_bit): Likewise.
6369 (tree_int_cst_sgn): Likewise.
6370 (tree_int_cst_min_precision): Take a signop rather than a bool.
6371 (simple_cst_equal): Use wide_int interfaces.
6372 (compare_tree_int): Likewise.
6373 (iterative_hash_expr): Likewise.
6374 (int_fits_type_p): Likewise. Use tree_int_cst_lt rather than
6376 (get_type_static_bounds): Use wide_int interfaces.
6377 (tree_int_cst_elt_check_failed): New.
6378 (build_common_tree_nodes): Reordered to set prec before filling in
6380 (int_cst_value): Check cst_and_fits_in_hwi.
6381 (widest_int_cst_value): Use wide_int interfaces.
6382 (upper_bound_in_type): Likewise.
6383 (lower_bound_in_type): Likewise.
6384 (num_ending_zeros): Likewise.
6385 (drop_tree_overflow): Likewise.
6386 * tree-call-cdce.c (check_pow): Update call to real_from_integer.
6387 (gen_conditions_for_pow_cst_base): Likewise.
6388 * tree-cfg.c: Include wide-int.h and wide-int-print.h.
6389 (group_case_labels_stmt): Use wide-int interfaces.
6390 (verify_gimple_assign_binary): Likewise.
6391 (print_loop): Likewise.
6392 * tree-chrec.c (tree_fold_binomial): Likewise.
6393 * tree-core.h (struct tree_base): Add int_length.
6394 (struct tree_int_cst): Change rep of value.
6395 * tree-data-ref.c (dr_analyze_innermost): Use wide-int interfaces.
6396 (dr_may_alias_p): Likewise.
6397 (max_stmt_executions_tree): Likewise.
6398 * tree.def (INTEGER_CST): Update comment.
6399 * tree-dfa.c (get_ref_base_and_extent): Use wide-int interfaces.
6400 * tree-dfa.h (get_addr_base_and_unit_offset_1): Likewise.
6401 * tree-dump.c: Include wide-int.h and wide-int-print.h.
6402 (dequeue_and_dump): Use wide-int interfaces.
6403 * tree.h: Include wide-int.h.
6404 (NULL_TREE): Moved to earlier loc in file.
6405 (TREE_INT_CST_ELT_CHECK): New.
6406 (tree_int_cst_elt_check_failed): New.
6408 (TREE_INT_CST): Delete.
6409 (TREE_INT_CST_LOW): Use wide-int interfaces.
6410 (TREE_INT_CST_HIGH): Delete.
6411 (TREE_INT_CST_NUNITS): New.
6412 (TREE_INT_CST_EXT_NUNITS): Likewise.
6413 (TREE_INT_CST_OFFSET_NUNITS): Likewise.
6414 (TREE_INT_CST_ELT): Likewise.
6415 (INT_CST_LT): Delete.
6416 (tree_int_cst_elt_check): New (two forms).
6417 (type_code_size): Update comment.
6418 (make_int_cst_stat, make_int_cst): New.
6419 (tree_to_double_int): Delete.
6420 (double_int_fits_to_tree_p): Delete.
6421 (force_fit_type_double): Delete.
6422 (build_int_cstu): Replace with out-of-line function.
6423 (build_int_cst_wide): Delete.
6424 (tree_int_cst_lt): Define inline.
6425 (tree_int_cst_le): New.
6426 (tree_int_cst_compare): Define inline.
6427 (tree_int_cst_min_precision): Take a signop rather than a bool.
6428 (wi::int_traits <const_tree>): New.
6429 (wi::int_traits <tree>): New.
6430 (wi::extended_tree): New.
6431 (wi::int_traits <wi::extended_tree>): New.
6432 (wi::to_widest): New.
6433 (wi::to_offset): New.
6434 (wi::fits_to_tree_p): New.
6435 (wi::min_value): New.
6436 (wi::max_value): New.
6437 * tree-inline.c (remap_gimple_op_r): Use wide-int interfaces.
6438 (copy_tree_body_r): Likewise.
6439 * tree-object-size.c (compute_object_offset): Likewise.
6440 (addr_object_size): Likewise.
6441 * tree-predcom.c: Include wide-int-print.h.
6442 (struct dref_d): Change type of offset to widest_int.
6443 (dump_dref): Call wide-int printer.
6444 (aff_combination_dr_offset): Use wide-int interfaces.
6445 (determine_offset): Take a widest_int pointer rather than a
6447 (split_data_refs_to_components): Use wide-int interfaces.
6448 (suitable_component_p): Likewise.
6449 (order_drefs): Likewise.
6450 (add_ref_to_chain): Likewise.
6451 (valid_initializer_p): Likewise.
6452 (determine_roots_comp): Likewise.
6453 * tree-pretty-print.c: Include wide-int-print.h.
6454 (dump_generic_node): Use wide-int interfaces.
6455 * tree-sra.c (sra_ipa_modify_expr): Likewise.
6456 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6457 (move_fixed_address_to_symbol): Likewise.
6458 (move_hint_to_base): Likewise.
6459 (move_pointer_to_base): Likewise.
6460 (move_variant_to_index): Likewise.
6461 (most_expensive_mult_to_index): Likewise.
6462 (addr_to_parts): Likewise.
6463 (copy_ref_info): Likewise.
6464 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
6465 (indirect_refs_may_alias_p): Likewise.
6466 (stmt_kills_ref_p_1): Likewise.
6467 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
6468 * tree-ssa-ccp.c: Update comment at top of file. Include
6470 (struct prop_value_d): Change type of mask to widest_int.
6471 (extend_mask): New function.
6472 (dump_lattice_value): Use wide-int interfaces.
6473 (get_default_value): Likewise.
6474 (set_constant_value): Likewise.
6475 (set_value_varying): Likewise.
6476 (valid_lattice_transition): Likewise.
6477 (set_lattice_value): Likewise.
6478 (value_to_double_int): Delete.
6479 (value_to_wide_int): New.
6480 (get_value_from_alignment): Use wide-int interfaces.
6481 (get_value_for_expr): Likewise.
6482 (do_dbg_cnt): Likewise.
6483 (ccp_finalize): Likewise.
6484 (ccp_lattice_meet): Likewise.
6485 (bit_value_unop_1): Use widest_ints rather than double_ints.
6486 (bit_value_binop_1): Likewise.
6487 (bit_value_unop): Use wide-int interfaces.
6488 (bit_value_binop): Likewise.
6489 (bit_value_assume_aligned): Likewise.
6490 (evaluate_stmt): Likewise.
6491 (ccp_fold_stmt): Likewise.
6492 (visit_cond_stmt): Likewise.
6493 (ccp_visit_stmt): Likewise.
6494 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
6495 (constant_pointer_difference): Likewise.
6496 (associate_pointerplus): Likewise.
6497 (combine_conversions): Likewise.
6498 * tree-ssa-loop.h: Include wide-int.h.
6499 (struct tree_niter_desc): Change type of max to widest_int.
6500 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Use wide-int interfaces.
6501 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Likewise.
6502 (remove_redundant_iv_tests): Likewise.
6503 (canonicalize_loop_induction_variables): Likewise.
6504 * tree-ssa-loop-ivopts.c (alloc_iv): Likewise.
6505 (constant_multiple_of): Take a widest_int pointer instead of
6506 a double_int pointer.
6507 (get_computation_aff): Use wide-int interfaces.
6508 (ptr_difference_cost): Likewise.
6509 (difference_cost): Likewise.
6510 (get_loop_invariant_expr_id): Likewise.
6511 (get_computation_cost_at): Likewise.
6512 (iv_elimination_compare_lt): Likewise.
6513 (may_eliminate_iv): Likewise.
6514 * tree-ssa-loop-niter.h (estimated_loop_iterations): Use widest_int
6515 instead of double_int.
6516 (max_loop_iterations): Likewise.
6517 (max_stmt_executions): Likewise.
6518 (estimated_stmt_executions): Likewise.
6519 * tree-ssa-loop-niter.c: Include wide-int-print.h.
6520 (split_to_var_and_offset): Use wide-int interfaces.
6521 (determine_value_range): Likewise.
6522 (bound_difference_of_offsetted_base): Likewise.
6523 (bounds_add): Take a widest_int instead of a double_int.
6524 (number_of_iterations_ne_max): Use wide-int interfaces.
6525 (number_of_iterations_ne): Likewise.
6526 (number_of_iterations_lt_to_ne): Likewise.
6527 (assert_loop_rolls_lt): Likewise.
6528 (number_of_iterations_lt): Likewise.
6529 (number_of_iterations_le): Likewise.
6530 (number_of_iterations_cond): Likewise.
6531 (number_of_iterations_exit): Likewise.
6532 (finite_loop_p): Likewise.
6533 (derive_constant_upper_bound_assign): Likewise.
6534 (derive_constant_upper_bound): Return a widest_int.
6535 (derive_constant_upper_bound_ops): Likewise.
6536 (do_warn_aggressive_loop_optimizations): Use wide-int interfaces.
6537 (record_estimate): Take a widest_int rather than a double_int.
6538 (record_nonwrapping_iv): Use wide-int interfaces.
6539 (double_int_cmp): Delete.
6540 (wide_int_cmp): New.
6541 (bound_index): Take a widest_int rather than a double_int.
6542 (discover_iteration_bound_by_body_walk): Use wide-int interfaces.
6543 (maybe_lower_iteration_bound): Likewise.
6544 (estimate_numbers_of_iterations_loop): Likewise.
6545 (estimated_loop_iterations): Take a widest_int pointer than than
6546 a double_int pointer.
6547 (estimated_loop_iterations_int): Use wide-int interfaces.
6548 (max_loop_iterations): Take a widest_int pointer than than
6549 a double_int pointer.
6550 (max_loop_iterations_int): Use wide-int interfaces.
6551 (max_stmt_executions): Take a widest_int pointer than than
6552 a double_int pointer.
6553 (estimated_stmt_executions): Likewise.
6554 (n_of_executions_at_most): Use wide-int interfaces.
6555 (scev_probably_wraps_p): Likewise.
6556 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Update calls
6558 * tree-scalar-evolution.c (simplify_peeled_chrec): Use wide-int
6560 * tree-ssanames.c (set_range_info): Use wide_int_refs rather than
6561 double_ints. Adjust for trailing_wide_ints <3> representation.
6562 (set_nonzero_bits): Likewise.
6563 (get_range_info): Return wide_ints rather than double_ints.
6564 Adjust for trailing_wide_ints <3> representation.
6565 (get_nonzero_bits): Likewise.
6566 (duplicate_ssa_name_range_info): Adjust for trailing_wide_ints <3>
6568 * tree-ssanames.h (struct range_info_def): Replace min, max and
6569 nonzero_bits with a trailing_wide_ints <3>.
6570 (set_range_info): Use wide_int_refs rather than double_ints.
6571 (set_nonzero_bits): Likewise.
6572 (get_range_info): Return wide_ints rather than double_ints.
6573 (get_nonzero_bits): Likewise.
6574 * tree-ssa-phiopt.c (jump_function_from_stmt): Use wide-int interfaces.
6575 * tree-ssa-pre.c (phi_translate_1): Likewise.
6576 * tree-ssa-reassoc.c (decrement_power): Use calls to real_from_integer.
6577 (acceptable_pow_call): Likewise.
6578 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Use wide-int
6580 (vn_reference_fold_indirect): Likewise.
6581 (vn_reference_maybe_forwprop_address): Likewise.
6582 (valueize_refs_1): Likewise.
6583 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6584 * tree-ssa-uninit.c (is_value_included_in): Use wide-int interfaces,
6585 tree_int_cst_lt and tree_int_cst_le.
6586 * tree-streamer-in.c (unpack_ts_base_value_fields): Use wide-int
6588 (streamer_alloc_tree): Likewise.
6589 * tree-streamer-out.c (pack_ts_int_cst_value_fields): Likewise.
6590 (streamer_write_tree_header): Likewise.
6591 (streamer_write_integer_cst): Likewise.
6592 * tree-switch-conversion.c (emit_case_bit_tests): Likewise.
6593 (build_constructors): Likewise.
6594 (array_value_type): Likewise.
6595 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Likewise.
6596 (vect_check_gather): Likewise.
6597 * tree-vect-generic.c (build_replicated_const): Likewise.
6598 (expand_vector_divmod): Likewise.
6599 * tree-vect-loop.c (vect_transform_loop): Likewise.
6600 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
6601 (vect_do_peeling_for_alignment): Likewise.
6602 * tree-vect-patterns.c (vect_recog_divmod_pattern): Likewise.
6603 * tree-vrp.c: Include wide-int.h.
6604 (operand_less_p): Use wide-int interfaces and tree_int_cst_lt.
6605 (extract_range_from_assert): Use wide-int interfaces.
6606 (vrp_int_const_binop): Likewise.
6607 (zero_nonzero_bits_from_vr): Take wide_int pointers rather than
6608 double_int pointers.
6609 (ranges_from_anti_range): Use wide-int interfaces.
6610 (quad_int_cmp): Delete.
6611 (quad_int_pair_sort): Likewise.
6612 (extract_range_from_binary_expr_1): Use wide-int interfaces.
6613 (extract_range_from_unary_expr_1): Likewise.
6614 (adjust_range_with_scev): Likewise.
6615 (masked_increment): Take and return wide_ints rather than double_ints.
6616 (register_edge_assert_for_2): Use wide-int interfaces.
6617 (check_array_ref): Likewise.
6618 (search_for_addr_array): Likewise.
6619 (maybe_set_nonzero_bits): Likewise.
6620 (union_ranges): Pass an integer of the correct type instead of
6621 using integer_one_node.
6622 (intersect_ranges): Likewise.
6623 (simplify_truth_ops_using_ranges): Likewise.
6624 (simplify_bit_ops_using_ranges): Use wide-int interfaces.
6625 (range_fits_type_p): Likewise.
6626 (simplify_cond_using_ranges): Likewise. Take a signop rather than
6628 (simplify_conversion_using_ranges): Use wide-int interfaces.
6629 (simplify_float_conversion_using_ranges): Likewise.
6630 (vrp_finalize): Likewise.
6631 * value-prof.c (gimple_divmod_fixed_value_transform): Likewise.
6632 (gimple_stringops_transform): Likewise.
6633 * varasm.c (decode_addr_const): Likewise.
6634 (const_hash_1): Likewise.
6635 (const_rtx_hash_1): Likewise
6636 (output_constant): Likewise.
6637 (array_size_for_constructor): Likewise.
6638 (output_constructor_regular_field): Likewise.
6639 (output_constructor_bitfield): Likewise.
6640 * var-tracking.c (loc_cmp): Handle CONST_WIDE_INT.
6641 * mkconfig.sh: Include machmode.h to pick up BITS_PER_UNIT for
6643 * gencheck.c: Define BITS_PER_UNIT.
6646 * wide-int-print.cc: New.
6647 * wide-int-print.h: New.
6649 2014-05-06 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6651 * config/avr/avr.c (avr_can_eliminate): Mark unused argument.
6653 2014-05-06 Richard Biener <rguenther@suse.de>
6655 * tree-pass.h (TODO_verify_ssa, TODO_verify_flow,
6656 TODO_verify_stmts, TODO_verify_rtl_sharing): Remove.
6657 (TODO_verify_all): Adjust.
6658 * asan.c: Remove references to TODO_verify_ssa, TODO_verify_flow,
6659 TODO_verify_stmts and TODO_verify_rtl_sharing.
6660 * bb-reorder.c: Likewise.
6661 * cfgexpand.c: Likewise.
6662 * cprop.c: Likewise.
6664 * function.c: Likewise.
6665 * fwprop.c: Likewise.
6667 * gimple-ssa-isolate-paths.c: Likewise.
6668 * gimple-ssa-strength-reduction.c: Likewise.
6669 * ipa-split.c: Likewise.
6670 * loop-init.c: Likewise.
6671 * loop-unroll.c: Likewise.
6672 * lower-subreg.c: Likewise.
6673 * modulo-sched.c: Likewise.
6674 * postreload-gcse.c: Likewise.
6675 * predict.c: Likewise.
6676 * recog.c: Likewise.
6677 * sched-rgn.c: Likewise.
6678 * store-motion.c: Likewise.
6679 * tracer.c: Likewise.
6680 * trans-mem.c: Likewise.
6681 * tree-call-cdce.c: Likewise.
6682 * tree-cfg.c: Likewise.
6683 * tree-cfgcleanup.c: Likewise.
6684 * tree-complex.c: Likewise.
6685 * tree-eh.c: Likewise.
6686 * tree-emutls.c: Likewise.
6687 * tree-if-conv.c: Likewise.
6688 * tree-into-ssa.c: Likewise.
6689 * tree-loop-distribution.c: Likewise.
6690 * tree-object-size.c: Likewise.
6691 * tree-parloops.c: Likewise.
6692 * tree-pass.h: Likewise.
6693 * tree-sra.c: Likewise.
6694 * tree-ssa-ccp.c: Likewise.
6695 * tree-ssa-copy.c: Likewise.
6696 * tree-ssa-copyrename.c: Likewise.
6697 * tree-ssa-dce.c: Likewise.
6698 * tree-ssa-dom.c: Likewise.
6699 * tree-ssa-dse.c: Likewise.
6700 * tree-ssa-forwprop.c: Likewise.
6701 * tree-ssa-ifcombine.c: Likewise.
6702 * tree-ssa-loop-ch.c: Likewise.
6703 * tree-ssa-loop-ivcanon.c: Likewise.
6704 * tree-ssa-loop.c: Likewise.
6705 * tree-ssa-math-opts.c: Likewise.
6706 * tree-ssa-phiopt.c: Likewise.
6707 * tree-ssa-phiprop.c: Likewise.
6708 * tree-ssa-pre.c: Likewise.
6709 * tree-ssa-reassoc.c: Likewise.
6710 * tree-ssa-sink.c: Likewise.
6711 * tree-ssa-strlen.c: Likewise.
6712 * tree-ssa-tail-merge.c: Likewise.
6713 * tree-ssa-uncprop.c: Likewise.
6714 * tree-switch-conversion.c: Likewise.
6715 * tree-tailcall.c: Likewise.
6716 * tree-vect-generic.c: Likewise.
6717 * tree-vectorizer.c: Likewise.
6718 * tree-vrp.c: Likewise.
6720 * var-tracking.c: Likewise.
6721 * bt-load.c: Likewise.
6722 * cfgcleanup.c: Likewise.
6723 * combine-stack-adj.c: Likewise.
6724 * combine.c: Likewise.
6725 * compare-elim.c: Likewise.
6726 * config/epiphany/resolve-sw-modes.c: Likewise.
6727 * config/i386/i386.c: Likewise.
6728 * config/mips/mips.c: Likewise.
6729 * config/s390/s390.c: Likewise.
6730 * config/sh/sh_treg_combine.cc: Likewise.
6731 * config/sparc/sparc.c: Likewise.
6734 * final.c: Likewise.
6735 * ifcvt.c: Likewise.
6736 * mode-switching.c: Likewise.
6737 * passes.c: Likewise.
6738 * postreload.c: Likewise.
6740 * reg-stack.c: Likewise.
6741 * regcprop.c: Likewise.
6742 * regrename.c: Likewise.
6745 2014-05-06 Richard Biener <rguenther@suse.de>
6748 * bitmap.c (debug_bitmap): Dump to stderr, not stdout.
6749 * tree-ssa-structalias.c (dump_solution_for_var): Likewise.
6751 2014-05-05 Jan Hubicka <hubicka@ucw.cz>
6754 * ipa-devirt.c (get_class_context): Allow POD to change to non-POD.
6756 2014-05-05 Radovan Obradovic <robradovic@mips.com>
6757 Tom de Vries <tom@codesourcery.com>
6759 * target.def (call_fusage_contains_non_callee_clobbers): New
6761 * doc/tm.texi.in (@node Stack and Calling): Add Miscellaneous Register
6763 (@node Miscellaneous Register Hooks): New node.
6764 (@hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS): New hook.
6765 * doc/tm.texi: Regenerate.
6767 2014-05-05 Marek Polacek <polacek@redhat.com>
6770 * opts.c (common_handle_option): Call error_at instead of warning_at.
6772 2014-05-05 Richard Biener <rguenther@suse.de>
6774 * passes.c (execute_function_todo): Don't reset TODO_verify_ssa
6775 from last_verified if update_ssa ran. Move TODO_verify_rtl_sharing
6776 under the TODO_verify_il umbrella.
6778 2014-05-05 Richard Biener <rguenther@suse.de>
6780 * passes.c (execute_function_todo): Move TODO_verify_flow under
6781 the TODO_verify_ul umbrella.
6783 2014-05-05 Richard Biener <rguenther@suse.de>
6786 * fold-const.c (fold_binary_loc): Consistently avoid canonicalizing
6787 X & CST away from a CST that is the mask of a mode.
6789 2014-05-05 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6791 * config/picochip/picochip-protos.h (picochip_regno_nregs): Change
6792 int argument to enum machine_mode.
6793 (picochip_class_max_nregs): Ditto.
6794 * config/picochip/picochip.c (picochip_regno_nregs): Ditto.
6795 (picochip_class_max_nregs): Ditto.
6797 2014-05-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6799 * target.def: Add new target hook.
6800 * doc/tm.texi: Regenerate.
6801 * targhooks.h (default_keep_leaf_when_profiled): Add prototype.
6802 * targhooks.c (default_keep_leaf_when_profiled): New function.
6804 * config/s390/s390.c (s390_keep_leaf_when_profiled): New function.
6805 (TARGET_KEEP_LEAF_WHEN_PROFILED): Define.
6807 2014-05-05 Bin Cheng <bin.cheng@arm.com>
6809 PR tree-optimization/60363
6810 * gcc/tree-ssa-threadupdate.c (get_value_locus_in_path): New.
6811 (copy_phi_args): New parameters. Call get_value_locus_in_path.
6812 (update_destination_phis): New parameter.
6813 (create_edge_and_update_destination_phis): Ditto.
6814 (ssa_fix_duplicate_block_edges): Pass new arguments.
6815 (thread_single_edge): Ditto.
6817 2014-05-04 Peter Bergner <bergner@vnet.ibm.com>
6819 * config/rs6000/rs6000.h (RS6000_BTM_HARD_FLOAT): New define.
6820 (RS6000_BTM_COMMON): Add RS6000_BTM_HARD_FLOAT.
6821 (TARGET_EXTRA_BUILTINS): Add TARGET_HARD_FLOAT.
6822 * config/rs6000/rs6000-builtin.def (BU_MISC_1):
6823 Use RS6000_BTM_HARD_FLOAT.
6824 (BU_MISC_2): Likewise.
6825 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Handle
6826 RS6000_BTM_HARD_FLOAT.
6827 (rs6000_option_override_internal): Enforce -mhard-float if -mhard-dfp
6829 (rs6000_invalid_builtin): Add hard floating builtin support.
6830 (rs6000_expand_builtin): Relax the gcc_assert to allow the new
6831 hard float builtins.
6832 (rs6000_builtin_mask_names): Add RS6000_BTM_HARD_FLOAT.
6834 2014-05-03 Oleg Endo <olegendo@gcc.gnu.org>
6836 * config/sh/sh_optimize_sett_clrt.cc (sh_optimize_sett_clrt::execute):
6837 Add missing function* argument.
6839 2014-05-03 Richard Sandiford <rdsandiford@googlemail.com>
6841 * lra-constraints.c (valid_address_p): Move earlier in file.
6842 Add a constraint argument to the address_info version.
6843 (satisfies_memory_constraint_p): New function.
6844 (satisfies_address_constraint_p): Likewise.
6845 (process_alt_operands, curr_insn_transform): Use them.
6846 (process_address): Pass the constraint to valid_address_p when
6847 checking address operands.
6849 2014-05-03 Richard Sandiford <rdsandiford@googlemail.com>
6851 * config/mips/mips.c (mips_isa_rev): New variable.
6852 (mips_set_architecture): Set it.
6853 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Set __mips_isa_rev
6855 (ISA_HAS_MUL3, ISA_HAS_FP_CONDMOVE, ISA_HAS_8CC, ISA_HAS_FP4)
6856 (ISA_HAS_PAIRED_SINGLE, ISA_HAS_MADD_MSUB, ISA_HAS_FP_RECIP_RSQRT)
6857 (ISA_HAS_CLZ_CLO, ISA_HAS_ROR, ISA_HAS_WSBH, ISA_HAS_PREFETCH)
6858 (ISA_HAS_SEB_SEH, ISA_HAS_EXT_INS, ISA_HAS_MXHC1)
6859 (ISA_HAS_HILO_INTERLOCKS, ISA_HAS_SYNCI, MIN_FPRS_PER_FMT): Reexpress
6860 conditions in terms of mips_isa_rev.
6861 (mips_isa_rev): Declare.
6863 2014-05-03 Oleg Endo <olegendo@gcc.gnu.org>
6865 * config/sh/sh-mem.cc: Use tabs instead of spaces.
6866 (prob_unlikely, prob_likely): Make variables const.
6868 2014-05-03 Denis Chertykov <chertykov@gmail.com>
6870 * config/avr/avr.c (avr_adjust_insn_length): Handle JUMP_TABLE_DATA.
6872 2014-05-03 Oleg Endo <olegendo@gcc.gnu.org>
6874 * config/sh/sh.h (SH_ASM_SPEC): Handle m1, m2*, m3* and m4* cases.
6876 2014-05-03 Oleg Endo <olegendo@gcc.gnu.org>
6878 * config/sh/sh.h (ROUND_ADVANCE): Delete macro.
6879 (ROUND_REG, PASS_IN_REG_P): Move and rename macros to ...
6880 * config/sh/sh.c (sh_round_reg, sh_pass_in_reg_p): ... these new
6882 (sh_arg_partial_bytes, sh_function_arg, sh_function_arg_advance,
6883 sh_setup_incoming_varargs): Replace usage of PASS_IN_REG_P with
6885 Replace usage of ROUND_REG with sh_round_reg.
6886 Use CEIL instead of ROUND_ADVANCE.
6888 2014-05-03 Oleg Endo <olegendo@gcc.gnu.org>
6891 * config/sh/sh.c: Include stdlib headers before everything else.
6893 2014-05-02 Jakub Jelinek <jakub@redhat.com>
6895 * gimplify.c (gimplify_adjust_omp_clauses_1): Handle
6896 GOVD_FIRSTPRIVATE | GOVD_LASTPRIVATE.
6897 (gimplify_adjust_omp_clauses): Simd region is never
6898 directly nested in combined parallel. Instead, for linear
6899 with copyin/copyout, if in combined for simd loop, make decl
6900 firstprivate/lastprivate on OMP_FOR.
6901 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
6902 expand_omp_for_static_chunk): When setting endvar, also set
6903 fd->loop.v to the same value.
6905 2014-05-02 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
6907 * hwint.h (zext_hwi): Fix signed overflow for prec == 63.
6909 2014-05-02 Alan Lawrence <alan.lawrence@arm.com>
6911 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_1): Tidy bit-flip
6914 2014-05-02 Marek Polacek <polacek@redhat.com>
6916 * doc/invoke.texi: Describe -fsanitize=float-divide-by-zero.
6918 2014-05-02 Kito Cheng <kito@0xlab.org>
6920 * defaults.h (HONOR_REG_ALLOC_ORDER): Change HONOR_REG_ALLOC_ORDER
6921 to a C expression marco.
6922 * ira-color.c (HONOR_REG_ALLOC_ORDER) : Ditto.
6923 * config/arm/arm.h (HONOR_REG_ALLOC_ORDER): Ditto.
6924 * config/nds32/nds32.h (HONOR_REG_ALLOC_ORDER): Ditto.
6925 * doc/tm.texi (HONOR_REG_ALLOC_ORDER): Update document for
6926 HONOR_REG_ALLOC_ORDER.
6927 * doc/tm.texi.in (HONOR_REG_ALLOC_ORDER): Ditto.
6929 2014-05-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6931 * config/arc/arc.c (TARGET_LRA_P): Undef before redefine.
6933 2014-05-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6935 * config/arc/arc.c (arc_select_cc_mode): Fix typo.
6937 2014-05-01 Yuri Rumyantsev <ysrumyan@gmail.com>
6939 * tree-if-conv.c (is_cond_scalar_reduction): New function.
6940 (convert_scalar_cond_reduction): Likewise.
6941 (predicate_scalar_phi): Add recognition and transformation
6942 of simple conditioanl reduction to be vectorizable.
6944 2014-05-01 Marek Polacek <polacek@redhat.com>
6947 * doc/invoke.texi: Document -Wdiscarded-qualifiers.
6949 2014-04-30 Alan Lawrence <alan.lawrence@arm.com>
6951 * config/aarch64/arm_neon.h (vuzp1_f32, vuzp1_p8, vuzp1_p16, vuzp1_s8,
6952 vuzp1_s16, vuzp1_s32, vuzp1_u8, vuzp1_u16, vuzp1_u32, vuzp1q_f32,
6953 vuzp1q_f64, vuzp1q_p8, vuzp1q_p16, vuzp1q_s8, vuzp1q_s16, vuzp1q_s32,
6954 vuzp1q_s64, vuzp1q_u8, vuzp1q_u16, vuzp1q_u32, vuzp1q_u64, vuzp2_f32,
6955 vuzp2_p8, vuzp2_p16, vuzp2_s8, vuzp2_s16, vuzp2_s32, vuzp2_u8,
6956 vuzp2_u16, vuzp2_u32, vuzp2q_f32, vuzp2q_f64, vuzp2q_p8, vuzp2q_p16,
6957 vuzp2q_s8, vuzp2q_s16, vuzp2q_s32, vuzp2q_s64, vuzp2q_u8, vuzp2q_u16,
6958 vuzp2q_u32, vuzp2q_u64): Replace temporary asm with __builtin_shuffle.
6960 2014-04-30 Joern Rennecke <joern.rennecke@embecosm.com>
6962 * config/arc/arc.opt (mlra): Move comment above option name
6963 to avoid mis-parsing as language options.
6965 2014-04-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6967 * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Move ...
6968 * config/sol2.h: ... here.
6969 * config/sol2-10.h: Remove.
6971 * config/sol2-bi.h (WCHAR_TYPE, WCHAR_TYPE_SIZE, WINT_TYPE)
6972 (WINT_TYPE_SIZE, MULTILIB_DEFAULTS, DEF_ARCH32_SPEC)
6973 (DEF_ARCH64_SPEC, ASM_CPU_DEFAULT_SPEC, LINK_ARCH64_SPEC_BASE)
6974 (LINK_ARCH64_SPEC, ARCH_DEFAULT_EMULATION, TARGET_LD_EMULATION)
6975 (LINK_ARCH_SPEC, SUBTARGET_EXTRA_SPECS): Move ...
6976 * config/sol2.h: ... here.
6977 (SECTION_NAME_FORMAT): Don't redefine.
6978 (STARTFILE_ARCH32_SPEC): Rename to ...
6979 (STARTFILE_ARCH_SPEC): ... this.
6980 (ASM_OUTPUT_ALIGNED_COMMON): Move ...
6981 * config/sparc/sol2.h: ... here.
6982 (SECTION_NAME_FORMAT): Don't undef.
6983 * config/i386/sol2.h (ASM_CPU_DEFAULT_SPEC)
6984 (SUBTARGET_EXTRA_SPECS): Remove.
6985 * config/sparc/sol2.h (ASM_CPU_DEFAULT_SPEC): Remove.
6987 * config/i386/sol2-bi.h (TARGET_SUBTARGET_DEFAULT)
6988 (MD_STARTFILE_PREFIX): Remove.
6989 (SUBTARGET_OPTIMIZATION_OPTIONS, ASM_CPU32_DEFAULT_SPEC)
6990 (ASM_CPU64_DEFAULT_SPEC, ASM_CPU_SPEC, ASM_SPEC, DEFAULT_ARCH32_P)
6991 (ARCH64_SUBDIR, ARCH32_EMULATION, ARCH64_EMULATION)
6992 (ASM_COMMENT_START, JUMP_TABLES_IN_TEXT_SECTION)
6993 (ASM_OUTPUT_DWARF_PCREL, ASM_OUTPUT_ALIGNED_COMMON)
6994 (USE_IX86_FRAME_POINTER, USE_X86_64_FRAME_POINTER): Move ...
6995 * config/i386/sol2.h: ... here.
6996 (TARGET_SUBTARGET_DEFAULT, SIZE_TYPE, PTRDIFF_TYPE): Remove.
6997 * config/i386/sol2-bi.h: Remove.
6998 * config/sol2.h (MD_STARTFILE_PREFIX): Remove.
6999 (LINK_ARCH32_SPEC_BASE): Remove /usr/ccs/lib/libp, /usr/ccs/lib.
7001 * config/i386/t-sol2-64: Rename to ...
7002 * config/i386/t-sol2: ... this.
7003 * config/sparc/t-sol2-64: Rename to ...
7004 * config/sparc/t-sol2: ... this.
7006 * config.gcc (*-*-solaris2*): Split sol2_tm_file into
7007 sol2_tm_file_head, sol2_tm_file_tail.
7008 Include ${cpu_type}/sol2.h before sol2.h.
7010 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Include
7011 i386/x86-64.h between sol2_tm_file_head and sol2_tm_file_tail.
7012 Remove i386/sol2-bi.h, sol2-bi.h from tm_file.
7013 Reflect i386/t-sol2-64 renaming.
7014 (sparc*-*-solaris2*): Remove sol2-bi.h from tm_file.
7015 Reflect sparc/t-sol2-64 renaming.
7017 2014-04-30 Richard Biener <rguenther@suse.de>
7019 * passes.c (execute_function_todo): Move TODO_verify_stmts
7020 and TODO_verify_ssa under the TODO_verify_il umbrella.
7021 * tree-ssa.h (verify_ssa): Adjust prototype.
7022 * tree-ssa.c (verify_ssa): Add parameter to tell whether
7023 we should verify SSA operands.
7024 * tree-cfg.h (verify_gimple_in_cfg): Adjust prototype.
7025 * tree-cfg.c (verify_gimple_in_cfg): Add parameter to tell
7026 whether we should verify whether not throwing stmts have EH info.
7027 * graphite-scop-detection.c (create_sese_edges): Adjust.
7028 * tree-ssa-loop-manip.c (verify_loop_closed_ssa): Likewise.
7029 * tree-eh.c (lower_try_finally_switch): Do not add the
7030 default case label twice.
7032 2014-04-30 Marek Polacek <polacek@redhat.com>
7034 * gcc.c (sanitize_spec_function): Handle SANITIZE_FLOAT_DIVIDE.
7035 * builtins.def: Initialize builtins even for SANITIZE_FLOAT_DIVIDE.
7036 * flag-types.h (enum sanitize_code): Add SANITIZE_FLOAT_DIVIDE.
7037 * opts.c (common_handle_option): Add -fsanitize=float-divide-by-zero.
7039 2014-04-29 Alan Lawrence <alan.lawrence@arm.com>
7041 * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16, vzip1_s8,
7042 vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
7043 vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16, vzip1q_s32,
7044 vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64, vzip2_f32,
7045 vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
7046 vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
7047 vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8, vzip2q_u16,
7048 vzip2q_u32, vzip2q_u64): Replace inline __asm__ with __builtin_shuffle.
7050 2014-04-29 David Malcolm <dmalcolm@redhat.com>
7052 * tree-cfg.c (dump_function_to_file): Dump the return type of
7053 functions, in a line to itself before the function body, mimicking
7054 the layout of a C function.
7056 2014-04-29 Jakub Jelinek <jakub@redhat.com>
7058 PR tree-optimization/60971
7059 * tree-tailcall.c (process_assignment): Reject conversions which
7062 2014-04-29 James Greenhalgh <james.greenhalgh@arm.com>
7064 * calls.c (initialize_argument_information): Always treat
7065 PUSH_ARGS_REVERSED as 1, simplify code accordingly.
7066 (expand_call): Likewise.
7067 (emit_library_call_calue_1): Likewise.
7068 * expr.c (PUSH_ARGS_REVERSED): Do not define.
7069 (emit_push_insn): Always treat PUSH_ARGS_REVERSED as 1, simplify
7072 2014-04-29 Nick Clifton <nickc@redhat.com>
7074 * config/msp430/msp430.md (umulsidi): Fix typo.
7075 (mulhisi3): Enable even inside interrupt handlers.
7076 * config/msp430/msp430.c (msp430_print_operand): %O: Allow for the
7077 bigger return address pushed in large mode.
7079 2014-04-29 Nick Clifton <nickc@redhat.com>
7081 * config/arc/arc.c (arc_select_cc_mode): Fix parentheses.
7082 (arc_init_reg_tables): Use a machine_mode enum to iterate over
7084 * config/m32r/m32r.c (init_reg_tables): Likewise.
7085 * config/m32c/m32c.c (m32c_illegal_subreg_p): Use a machine_mode
7086 enum to hold the modes.
7088 2014-04-29 Richard Biener <rguenther@suse.de>
7090 * dominance.c (free_dominance_info): Add overload with
7092 (dom_info_state): Likewise.
7093 (dom_info_available_p): Likewise.
7094 * basic-block.h (free_dominance_info, dom_info_state,
7095 dom_info_available_p): Declare overloads.
7096 * passes.c (execute_function_todo): Verify that verifiers
7097 don't change dominator info state. Drop dominator info
7098 for IPA pass invocations.
7099 * cgraph.c (release_function_body): Restore asserts that
7100 dominator information is released.
7102 2014-04-29 Patrick Palka <patrick@parcs.ath.cx>
7104 * doc/invoke.texi: Fix typo.
7105 * tree-vrp.c: Fix typos.
7106 * gimple.c (infer_nonnull_range): Reorder operands of an && condition.
7108 2014-04-29 Zhenqiang Chen <zhenqiang.chen@linaro.org>
7110 * config/aarch64/aarch64.md (mov<mode>cc): New for GPF.
7112 2014-04-28 James Greenhalgh <james.greenhalgh@arm.com>
7114 * config/aarch64/aarch64-builtins.c
7115 (aarch64_types_storestruct_lane_qualifiers): New.
7116 (TYPES_STORESTRUCT_LANE): Likewise.
7117 * config/aarch64/aarch64-simd-builtins.def (st2_lane): New.
7118 (st3_lane): Likewise.
7119 (st4_lane): Likewise.
7120 * config/aarch64/aarch64-simd.md (vec_store_lanesoi_lane<mode>): New.
7121 (vec_store_lanesci_lane<mode>): Likewise.
7122 (vec_store_lanesxi_lane<mode>): Likewise.
7123 (aarch64_st2_lane<VQ:mode>): Likewise.
7124 (aarch64_st3_lane<VQ:mode>): Likewise.
7125 (aarch64_st4_lane<VQ:mode>): Likewise.
7126 * config/aarch64/aarch64.md (unspec): Add UNSPEC_ST{2,3,4}_LANE.
7127 * config/aarch64/arm_neon.h
7128 (__ST2_LANE_FUNC): Rewrite using builtins, update use points to
7129 use new macro arguments.
7130 (__ST3_LANE_FUNC): Likewise.
7131 (__ST4_LANE_FUNC): Likewise.
7132 * config/aarch64/iterators.md (V_TWO_ELEM): New.
7133 (V_THREE_ELEM): Likewise.
7134 (V_FOUR_ELEM): Likewise.
7136 2014-04-28 David Malcolm <dmalcolm@redhat.com>
7138 * doc/gimple.texi: Replace the description of the now-defunct
7139 union gimple_statement_d with a diagram showing the
7140 gimple_statement_base class hierarchy and its relationships to
7141 the GSS_ and GIMPLE_ enums.
7143 2014-04-28 James Greenhalgh <james.greenhalgh@arm.com>
7145 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): New.
7146 * config/aarch64/aarch64.c
7147 (aarch64_cannot_change_mode_class): Weaken conditions.
7148 (aarch64_modes_tieable_p): New.
7149 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Use it.
7151 2014-04-28 Pat Haugen <pthaugen@us.ibm.com>
7153 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
7154 (loadsync_<mode>): Change mode.
7155 (load_quadpti, store_quadpti): New.
7156 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
7157 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
7159 2014-04-28 Martin Jambor <mjambor@suse.cz>
7161 * tree-sra.c (sra_modify_expr): Generate new memory accesses with
7162 same alias type as the original statement.
7163 (subreplacement_assignment_data): New type.
7164 (handle_unscalarized_data_in_subtree): New type of parameter,
7165 generate new memory accesses with same alias type as the original
7167 (load_assign_lhs_subreplacements): Likewise.
7168 (sra_modify_constructor_assign): Generate new memory accesses with
7169 same alias type as the original statement.
7171 2014-04-28 Richard Biener <rguenther@suse.de>
7173 * tree-pass.h (TODO_verify_il): Define.
7174 (TODO_verify_all): Complete properly.
7175 * passes.c (execute_function_todo): Move existing loop-closed
7176 SSA verification under TODO_verify_il.
7177 (execute_one_pass): Trigger TODO_verify_il at todo-after time.
7178 * graphite-sese-to-poly.c (rewrite_cross_bb_scalar_deps):
7179 Fix tree sharing issue.
7181 2014-04-28 Richard Biener <rguenther@suse.de>
7184 * builtins.def (DEF_C11_BUILTIN): Add.
7185 (BUILT_IN_ALIGNED_ALLOC): Likewise.
7186 * coretypes.h (enum function_class): Add function_c11_misc.
7187 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
7188 BUILT_IN_ALIGNED_ALLOC like BUILT_IN_MALLOC.
7189 (call_may_clobber_ref_p_1): Likewise.
7190 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7191 (mark_all_reaching_defs_necessary_1): Likewise.
7192 (propagate_necessity): Likewise.
7193 (eliminate_unnecessary_stmts): Likewise.
7194 * tree-ssa-ccp.c (evaluate_stmt): Handle BUILT_IN_ALIGNED_ALLOC.
7196 2014-04-28 Richard Biener <rguenther@suse.de>
7198 * tree-vrp.c (vrp_var_may_overflow): Remove.
7199 (vrp_visit_phi_node): Rather than bumping to +-INF possibly
7200 with overflow immediately bump to one before that value and
7201 let iteration figure out overflow status.
7203 2014-04-28 Richard Biener <rguenther@suse.de>
7205 * configure.ac: Do valgrind header checks unconditionally.
7206 Add --enable-valgrind-annotations.
7207 * system.h: Guard valgrind header inclusion with
7208 ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING.
7209 * alloc-pool.c (pool_alloc, pool_free): Use
7210 ENABLE_VALGRIND_ANNOTATIONS instead of ENABLE_VALGRIND_CHECKING
7211 to guard possibly dead code.
7212 * config.in: Regenerated.
7213 * configure: Likewise.
7215 2014-04-28 Jeff Law <law@redhat.com>
7217 PR tree-optimization/60902
7218 * tree-ssa-threadedge.c
7219 (record_temporary_equivalences_from_stmts_at_dest): Only iterate
7220 over real defs when invalidating outputs from statements that do not
7221 produce useful outputs for threading.
7223 2014-04-28 Richard Biener <rguenther@suse.de>
7225 PR tree-optimization/60979
7226 * graphite-scop-detection.c (scopdet_basic_block_info): Reject
7227 SCOPs that end in a block with a successor with abnormal
7230 2014-04-28 Richard Biener <rguenther@suse.de>
7232 * tree-pass.h (execute_pass_list): Adjust prototype.
7233 * passes.c (pass_manager::execute_early_local_passes): Adjust.
7234 (do_per_function): Change callback signature, push all actual
7235 work to the callbals.
7236 (do_per_function_toporder): Likewise.
7237 (execute_function_dump): Adjust.
7238 (execute_function_todo): Likewise.
7239 (clear_last_verified): Likewise.
7240 (verify_curr_properties): Likewise.
7241 (update_properties_after_pass): Likewise.
7242 (execute_pass_list_1): Split out from ...
7243 (execute_pass_list): ... here. Adjust.
7244 (execute_ipa_pass_list): Likewise.
7245 * cgraphunit.c (cgraph_add_new_function): Adjust.
7246 (analyze_function): Likewise.
7247 (expand_function): Likewise.
7248 * cgraph.c (release_function_body): Free dominance info
7249 here instead of asserting it was magically freed elsewhere.
7251 2014-04-28 Eric Botcazou <ebotcazou@adacore.com>
7253 * configure.ac: Tweak GAS check for LEON instructions on SPARC.
7254 * configure: Regenerate.
7255 * config/sparc/sparc.opt (muser-mode): New option.
7256 * config/sparc/sync.md (atomic_compare_and_swap<mode>_1): Do not enable
7258 (atomic_compare_and_swap_leon3_1): New instruction for LEON3.
7259 * doc/invoke.texi (SPARC options): Document -muser-mode.
7261 2014-04-27 Richard Sandiford <rdsandiford@googlemail.com>
7263 * cselib.c (find_slot_memmode): Delete.
7264 (cselib_hasher): Change compare_type to a struct.
7265 (cselib_hasher::equal): Update accordingly. Don't expect wrapped
7267 (preserve_constants_and_equivs): Adjust for new compare_type.
7268 (cselib_find_slot): Likewise. Take the mode of the rtx as argument.
7269 (wrap_constant): Delete.
7270 (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
7272 2014-04-26 Markus Trippelsdorf <markus@trippelsdorf.de>
7274 * doc/install.texi (Building with profile feedback): Remove
7277 2014-04-26 Tom de Vries <tom@codesourcery.com>
7279 * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
7282 2014-04-25 Cary Coutant <ccoutant@google.com>
7285 * dwarf2out.c (should_move_die_to_comdat): A type definition
7286 can contain a subprogram definition, but don't move it to a
7288 (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
7289 (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
7291 (clone_tree_hash): Rename to...
7292 (clone_tree_partial): ...this; change callers. Copy
7293 DW_TAG_subprogram DIEs as declarations.
7294 (copy_decls_walk): Don't copy children of a declaration into a
7297 2014-04-25 H.J. Lu <hongjiu.lu@intel.com>
7300 * config/i386/i386.md (*movsf_internal): Set MODE to SI for
7303 2014-04-25 Jiong Wang <jiong.wang@arm.com>
7305 * config/arm/predicates.md (call_insn_operand): Add long_call check.
7306 * config/arm/arm.md (sibcall, sibcall_value): Force the address to
7308 * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
7311 2014-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7313 * config/arm/arm.c (arm_cortex_a8_tune): Initialise T16-related fields.
7315 2014-04-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7317 PR tree-optimization/60930
7318 * gimple-ssa-strength-reduction.c (create_mul_imm_cand): Reject
7319 creating a multiply candidate by folding two constant
7320 multiplicands when the result overflows.
7322 2014-04-25 Jakub Jelinek <jakub@redhat.com>
7324 PR tree-optimization/60960
7325 * tree-vect-generic.c (expand_vector_operation): Only call
7326 expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
7328 2014-04-25 Tom de Vries <tom@codesourcery.com>
7330 * expr.c (clobber_reg_mode): New function.
7331 * expr.h (clobber_reg): New function.
7333 2014-04-25 Tom de Vries <tom@codesourcery.com>
7335 * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
7338 2014-04-25 Radovan Obradovic <robradovic@mips.com>
7339 Tom de Vries <tom@codesourcery.com>
7341 * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
7343 * rtl.h (find_all_hard_reg_sets): Add bool parameter.
7344 * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
7345 new argument to find_all_hard_reg_sets call.
7347 2014-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7349 * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
7350 Use HOST_WIDE_INT_C for mask literal.
7351 (aarch_rev16_shleft_mask_imm_p): Likewise.
7353 2014-04-25 Eric Botcazou <ebotcazou@adacore.com>
7356 * config/sparc/sparc.md (ashlsi3_extend): Delete.
7358 2014-04-25 Marc Glisse <marc.glisse@inria.fr>
7360 PR preprocessor/56540
7361 * config/i386/i386-c.c (ix86_target_macros): Define
7362 __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
7364 2014-04-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7366 * configure.ac (tga_func): Remove.
7367 (LIB_TLS_SPEC): Remove.
7368 * configure: Regenerate.
7369 * config.in: Regenerate.
7370 * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
7372 2014-04-25 Richard Biener <rguenther@suse.de>
7375 * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
7376 call stmt use/clobber sets during stmt walk instead of
7377 walking the possibly incomplete set of caller edges.
7379 2014-04-25 Richard Biener <rguenther@suse.de>
7382 * passes.c (apply_ipa_transforms): Inline into only caller ...
7383 (execute_one_pass): ... here. Properly bring in function
7384 bodies for nodes we want to apply IPA transforms to.
7386 2014-04-24 Cong Hou <congh@google.com>
7388 PR tree-optimization/60896
7389 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
7390 all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
7391 (vect_mark_pattern_stmts): Set the def type of all statements in
7392 PATTERN_DEF_SEQ as vect_internal_def.
7394 2014-04-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7396 * doc/extend.texi (PowerPC Built-in Functions): Document new
7397 powerpc extended divide, bcd, pack/unpack 128-bit, builtin functions.
7398 (PowerPC AltiVec/VSX Built-in Functions): Likewise.
7400 * config/rs6000/predicates.md (const_0_to_3_operand): New
7401 predicate to match 0..3 integer constants.
7403 * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
7404 to support adding miscellaneous builtin functions.
7405 (BU_DFP_MISC_2): Likewise.
7406 (BU_P7_MISC_1): Likewise.
7407 (BU_P7_MISC_2): Likewise.
7408 (BU_P8V_MISC_3): Likewise.
7409 (BU_MISC_1): Likewise.
7410 (BU_MISC_2): Likewise.
7411 (DIVWE): Add extended divide builtin functions.
7414 (DIVWEUO): Likewise.
7418 (DIVDEUO): Likewise.
7419 (DXEX): Add decimal floating-point builtin functions.
7422 (DDEDPDQ): Likewise.
7424 (DENBCDQ): Likewise.
7431 (CDTBCD): Add new BCD builtin functions.
7435 (BCDADD_LT): Likewise.
7436 (BCDADD_EQ): Likewise.
7437 (BCDADD_GT): Likewise.
7438 (BCDADD_OV): Likewise.
7440 (BCDSUB_LT): Likewise.
7441 (BCDSUB_EQ): Likewise.
7442 (BCDSUB_GT): Likewise.
7443 (BCDSUB_OV): Likewise.
7444 (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
7445 (UNPACK_TD): Likewise.
7446 (PACK_TF): Likewise.
7447 (UNPACK_TF): Likewise.
7448 (UNPACK_TF_0): Likewise.
7449 (UNPACK_TF_1): Likewise.
7450 (PACK_V1TI): Likewise.
7451 (UNPACK_V1TI): Likewise.
7453 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7454 support for decimal floating point builtin functions.
7455 (rs6000_expand_ternop_builtin): Add checks for the new builtin
7456 functions that take constant arguments.
7457 (rs6000_invalid_builtin): Add decimal floating point builtin support.
7458 (rs6000_init_builtins): Setup long double, _Decimal64, and
7459 _Decimal128 types for new builtin functions.
7460 (builtin_function_type): Set the unsigned flags appropriately for
7461 the new builtin functions.
7462 (rs6000_opt_masks): Add support for decimal floating point builtin
7465 * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
7466 floating point builtin functions.
7467 (RS6000_BTM_COMMON): Likewise.
7468 (RS6000_BTI_long_double): Likewise.
7469 (RS6000_BTI_dfloat64): Likewise.
7470 (RS6000_BTI_dfloat128): Likewise.
7471 (long_double_type_internal_node): Likewise.
7472 (dfloat64_type_internal_node): Likewise.
7473 (dfloat128_type_internal_node): Likewise.
7475 * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
7476 2.07 bcd arithmetic instructions.
7477 (UNSPEC_BCDSUB): Likewise.
7478 (UNSPEC_BCD_OVERFLOW): Likewise.
7479 (UNSPEC_BCD_ADD_SUB): Likewise.
7480 (bcd_add_sub): Likewise.
7481 (BCD_TEST): Likewise.
7482 (bcd<bcd_add_sub>): Likewise.
7483 (bcd<bcd_add_sub>_test): Likewise.
7484 (bcd<bcd_add_sub>_test2): Likewise.
7485 (bcd<bcd_add_sub>_<code>): Likewise.
7486 (peephole2 for combined bcd ops): Likewise.
7488 * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
7489 decimal floating point builtin functions.
7490 (UNSPEC_DENBCD): Likewise.
7491 (UNSPEC_DXEX): Likewise.
7492 (UNSPEC_DIEX): Likewise.
7493 (UNSPEC_DSCLI): Likewise.
7494 (UNSPEC_DSCRI): Likewise.
7495 (D64_D128): Likewise.
7496 (dfp_suffix): Likewise.
7497 (dfp_ddedpd_<mode>): Likewise.
7498 (dfp_denbcd_<mode>): Likewise.
7499 (dfp_dxex_<mode>): Likewise.
7500 (dfp_diex_<mode>): Likewise.
7501 (dfp_dscli_<mode>): Likewise.
7502 (dfp_dscri_<mode>): Likewise.
7504 * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
7506 (UNSPEC_CDTBCD): Likewise.
7507 (UNSPEC_CBCDTD): Likewise.
7508 (UNSPEC_DIVE): Add support for new extended divide builtin functions.
7509 (UNSPEC_DIVEO): Likewise.
7510 (UNSPEC_DIVEU): Likewise.
7511 (UNSPEC_DIVEUO): Likewise.
7512 (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
7513 pack/unpack 128-bit types.
7514 (UNSPEC_PACK_128BIT): Likewise.
7515 (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
7516 (udiv<mode>3): Use idiv_ldiv mode attribute.
7517 (div<mode>3): Likewise.
7518 (addg6s): Add new BCD builtin functions.
7521 (UNSPEC_DIV_EXTEND): Add support for new extended divide instructions.
7522 (div_extend): Likewise.
7523 (div<div_extend>_<mode>"): Likewise.
7524 (FP128_64): Add support for new builtin functions to pack/unpack
7526 (unpack<mode>): Likewise.
7527 (unpacktf_0): Likewise.
7528 (unpacktf_1): Likewise.
7529 (unpack<mode>_dm): Likewise.
7530 (unpack<mode>_nodm): Likewise.
7531 (pack<mode>): Likewise.
7532 (unpackv1ti): Likewise.
7533 (packv1ti): Likewise.
7535 2014-04-24 Vishnu K S <Vishnu.k_s@atmel.com>
7537 * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
7540 2014-04-24 Jakub Jelinek <jakub@redhat.com>
7542 * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
7543 * gimplify.c (omp_is_private): Change last argument's type to int.
7544 Only diagnose lastprivate if the simd argument is 1, only diagnose
7545 linear if the simd argument is 2.
7546 (gimplify_omp_for): Adjust omp_is_private callers. When adding
7547 lastprivate or private, add the clause to OMP_FOR_CLAUSES. Pass
7548 GOVD_EXPLICIT to omp_add_variable. For simd with collapse == 1
7549 create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
7550 If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
7551 increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
7552 * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
7553 OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
7554 * tree-nested.c (convert_nonlocal_omp_clauses,
7555 convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
7557 2014-04-24 Segher Boessenkool <segher@kernel.crashing.org>
7560 * config/m68k/m68k.md (extendplussidi): Don't allow memory for
7563 2014-04-24 Dimitris Papavasiliou <dpapavas@gmail.com>
7565 * flag-types.h (enum ivar_visibility): Add.
7567 2014-04-24 Trevor Saunders <tsaunders@mozilla.com>
7569 * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
7570 function * argument.
7572 2014-04-24 Alan Lawrence <alan.lawrence@arm.com>
7574 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
7576 2014-04-24 Radovan Obradovic <robradovic@mips.com>
7577 Tom de Vries <tom@codesourcery.com>
7579 * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
7580 * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
7582 * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
7583 * emit-rtl.c (try_split): Same.
7585 2014-04-24 Radovan Obradovic <robradovic@mips.com>
7586 Tom de Vries <tom@codesourcery.com>
7588 * common.opt (fuse-caller-save): New option.
7590 2014-04-24 Tejas Belagod <tejas.belagod@arm.com>
7592 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of
7593 elements for big-endian.
7595 2014-04-24 Richard Biener <rguenther@suse.de>
7597 * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
7598 during TER and instead use the sepops interface for expanding
7599 non-GIMPLE_SINGLE_RHS.
7601 2014-04-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7603 * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
7604 if not HAVE_AS_IX86_DIFF_SECT_DELTA.
7606 2014-04-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7608 * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
7609 assembler 64-bit option.
7610 * configure: Regenerate.
7612 2014-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7614 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
7615 TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
7616 (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
7617 (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
7618 (TARGET_CRYPTO): Take TARGET_SIMD into account.
7620 2014-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7622 * config/aarch64/aarch64-builtins.c
7623 (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
7624 BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
7625 * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
7626 * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
7628 * config/aarch64/iterator.md (VDQHSD): New mode iterator.
7629 (Vrevsuff): New mode attribute.
7631 2014-04-24 Terry Guo <terry.guo@arm.com>
7633 * config/arm/arm.h (machine_function): Define variable
7634 after_arm_reorg here.
7635 * config/arm/arm.c (after_arm_reorg): Remove the definition.
7636 (arm_split_constant): Update the way to access variable
7639 (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
7641 2014-04-23 Tom de Vries <tom@codesourcery.com>
7643 * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
7645 2014-04-23 David Malcolm <dmalcolm@redhat.com>
7647 * is-a.h: Update comments to reflect the following changes to the
7648 "pointerness" of the API, making the template parameter match the
7649 return type, allowing use of is-a.h with typedefs of pointers.
7650 (is_a_helper::cast): Return a T rather then a pointer to a T, so
7651 that the return type matches the parameter to the is_a_helper.
7653 (dyn_cast): Likewise.
7655 * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
7656 pointer from the is-a.h API.
7658 * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
7659 (is_a_helper <cgraph_node *>::test): ...this, matching change to
7661 (is_a_helper <varpool_node>::test): Likewise, convert to...
7662 (is_a_helper <varpool_node *>::test): ...this.
7664 (varpool_first_variable): Update for removal of implicit pointer
7665 from the is-a.h API.
7666 (varpool_next_variable): Likewise.
7667 (varpool_first_static_initializer): Likewise.
7668 (varpool_next_static_initializer): Likewise.
7669 (varpool_first_defined_variable): Likewise.
7670 (varpool_next_defined_variable): Likewise.
7671 (cgraph_first_defined_function): Likewise.
7672 (cgraph_next_defined_function): Likewise.
7673 (cgraph_first_function): Likewise.
7674 (cgraph_next_function): Likewise.
7675 (cgraph_first_function_with_gimple_body): Likewise.
7676 (cgraph_next_function_with_gimple_body): Likewise.
7677 (cgraph_alias_target): Likewise.
7678 (varpool_alias_target): Likewise.
7679 (cgraph_function_or_thunk_node): Likewise.
7680 (varpool_variable_node): Likewise.
7681 (symtab_real_symbol_p): Likewise.
7682 * cgraphunit.c (referred_to_p): Likewise.
7683 (analyze_functions): Likewise.
7684 (handle_alias_pairs): Likewise.
7685 * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
7686 * gimple-ssa.h (gimple_vuse_op): Likewise.
7687 (gimple_vdef_op): Likewise.
7688 * gimple-streamer-in.c (input_gimple_stmt): Likewise.
7689 * gimple.c (gimple_build_asm_1): Likewise.
7690 (gimple_build_try): Likewise.
7691 (gimple_build_resx): Likewise.
7692 (gimple_build_eh_dispatch): Likewise.
7693 (gimple_build_omp_for): Likewise.
7694 (gimple_omp_for_set_clauses): Likewise.
7696 * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
7697 (is_a_helper <gimple_statement_asm *>::test): ...this.
7698 (is_a_helper <gimple_statement_bind>::test): Convert to...
7699 (is_a_helper <gimple_statement_bind *>::test): ...this.
7700 (is_a_helper <gimple_statement_call>::test): Convert to...
7701 (is_a_helper <gimple_statement_call *>::test): ...this.
7702 (is_a_helper <gimple_statement_catch>::test): Convert to...
7703 (is_a_helper <gimple_statement_catch *>::test): ...this.
7704 (is_a_helper <gimple_statement_resx>::test): Convert to...
7705 (is_a_helper <gimple_statement_resx *>::test): ...this.
7706 (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
7707 (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
7708 (is_a_helper <gimple_statement_eh_else>::test): Convert to...
7709 (is_a_helper <gimple_statement_eh_else *>::test): ...this.
7710 (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
7711 (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
7712 (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
7713 (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
7714 (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
7715 (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
7716 (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
7717 (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
7718 (is_a_helper <gimple_statement_omp_return>::test): Convert to...
7719 (is_a_helper <gimple_statement_omp_return *>::test): ...this.
7720 (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
7721 (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
7722 (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
7723 (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
7724 (is_a_helper <gimple_statement_omp_for>::test): Convert to...
7725 (is_a_helper <gimple_statement_omp_for *>::test): ...this.
7726 (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
7727 (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
7728 (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
7729 (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
7730 (is_a_helper <gimple_statement_omp_target>::test): Convert to...
7731 (is_a_helper <gimple_statement_omp_target *>::test): ...this.
7732 (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
7733 (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
7734 (is_a_helper <gimple_statement_omp_single>::test): Convert to...
7735 (is_a_helper <gimple_statement_omp_single *>::test): ...this.
7736 (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
7737 (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
7738 (is_a_helper <gimple_statement_omp_task>::test): Convert to...
7739 (is_a_helper <gimple_statement_omp_task *>::test): ...this.
7740 (is_a_helper <gimple_statement_phi>::test): Convert to...
7741 (is_a_helper <gimple_statement_phi *>::test): ...this.
7742 (is_a_helper <gimple_statement_transaction>::test): Convert to...
7743 (is_a_helper <gimple_statement_transaction *>::test): ...this.
7744 (is_a_helper <gimple_statement_try>::test): Convert to...
7745 (is_a_helper <gimple_statement_try *>::test): ...this.
7746 (is_a_helper <gimple_statement_wce>::test): Convert to...
7747 (is_a_helper <gimple_statement_wce *>::test): ...this.
7748 (is_a_helper <const gimple_statement_asm>::test): Convert to...
7749 (is_a_helper <const gimple_statement_asm *>::test): ...this.
7750 (is_a_helper <const gimple_statement_bind>::test): Convert to...
7751 (is_a_helper <const gimple_statement_bind *>::test): ...this.
7752 (is_a_helper <const gimple_statement_call>::test): Convert to...
7753 (is_a_helper <const gimple_statement_call *>::test): ...this.
7754 (is_a_helper <const gimple_statement_catch>::test): Convert to...
7755 (is_a_helper <const gimple_statement_catch *>::test): ...this.
7756 (is_a_helper <const gimple_statement_resx>::test): Convert to...
7757 (is_a_helper <const gimple_statement_resx *>::test): ...this.
7758 (is_a_helper <const gimple_statement_eh_dispatch>::test): Convert to...
7759 (is_a_helper <const gimple_statement_eh_dispatch *>::test): ...this.
7760 (is_a_helper <const gimple_statement_eh_filter>::test): Convert to...
7761 (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
7762 (is_a_helper <const gimple_statement_omp_atomic_load>::test):
7764 (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
7766 (is_a_helper <const gimple_statement_omp_atomic_store>::test):
7768 (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
7770 (is_a_helper <const gimple_statement_omp_return>::test): Convert to...
7771 (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
7772 (is_a_helper <const gimple_statement_omp_continue>::test): Convert
7774 (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
7775 (is_a_helper <const gimple_statement_omp_critical>::test): Convert
7777 (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
7778 (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
7779 (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
7780 (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert to...
7781 (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
7782 (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
7784 (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
7785 (is_a_helper <const gimple_statement_omp_target>::test): Convert to...
7786 (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
7787 (is_a_helper <const gimple_statement_omp_sections>::test): Convert
7789 (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
7790 (is_a_helper <const gimple_statement_omp_single>::test): Convert to...
7791 (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
7792 (is_a_helper <const gimple_statement_omp_teams>::test): Convert to...
7793 (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
7794 (is_a_helper <const gimple_statement_omp_task>::test): Convert to...
7795 (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
7796 (is_a_helper <const gimple_statement_phi>::test): Convert to...
7797 (is_a_helper <const gimple_statement_phi *>::test): ...this.
7798 (is_a_helper <const gimple_statement_transaction>::test): Convert to...
7799 (is_a_helper <const gimple_statement_transaction *>::test): ...this.
7800 (is_a_helper <const gimple_statement_with_ops>::test): Convert to...
7801 (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
7802 (is_a_helper <gimple_statement_with_ops>::test): Convert to...
7803 (is_a_helper <gimple_statement_with_ops *>::test): ...this.
7804 (is_a_helper <const gimple_statement_with_memory_ops>::test): Convert
7806 (is_a_helper <const gimple_statement_with_memory_ops *>::test):
7808 (is_a_helper <gimple_statement_with_memory_ops>::test): Convert to...
7809 (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
7811 (gimple_use_ops): Update for removal of implicit pointer from the
7813 (gimple_set_use_ops): Likewise.
7814 (gimple_vuse): Likewise.
7815 (gimple_vdef): Likewise.
7816 (gimple_vuse_ptr): Likewise.
7817 (gimple_vdef_ptr): Likewise.
7818 (gimple_set_vuse): Likewise.
7819 (gimple_set_vdef): Likewise.
7820 (gimple_omp_return_set_lhs): Likewise.
7821 (gimple_omp_return_lhs): Likewise.
7822 (gimple_omp_return_lhs_ptr): Likewise.
7823 (gimple_call_fntype): Likewise.
7824 (gimple_call_set_fntype): Likewise.
7825 (gimple_call_set_internal_fn): Likewise.
7826 (gimple_call_use_set): Likewise.
7827 (gimple_call_clobber_set): Likewise.
7828 (gimple_bind_vars): Likewise.
7829 (gimple_bind_set_vars): Likewise.
7830 (gimple_bind_body_ptr): Likewise.
7831 (gimple_bind_set_body): Likewise.
7832 (gimple_bind_add_stmt): Likewise.
7833 (gimple_bind_block): Likewise.
7834 (gimple_bind_set_block): Likewise.
7835 (gimple_asm_ninputs): Likewise.
7836 (gimple_asm_noutputs): Likewise.
7837 (gimple_asm_nclobbers): Likewise.
7838 (gimple_asm_nlabels): Likewise.
7839 (gimple_asm_input_op): Likewise.
7840 (gimple_asm_input_op_ptr): Likewise.
7841 (gimple_asm_output_op): Likewise.
7842 (gimple_asm_output_op_ptr): Likewise.
7843 (gimple_asm_set_output_op): Likewise.
7844 (gimple_asm_clobber_op): Likewise.
7845 (gimple_asm_set_clobber_op): Likewise.
7846 (gimple_asm_label_op): Likewise.
7847 (gimple_asm_set_label_op): Likewise.
7848 (gimple_asm_string): Likewise.
7849 (gimple_catch_types): Likewise.
7850 (gimple_catch_types_ptr): Likewise.
7851 (gimple_catch_handler_ptr): Likewise.
7852 (gimple_catch_set_types): Likewise.
7853 (gimple_catch_set_handler): Likewise.
7854 (gimple_eh_filter_types): Likewise.
7855 (gimple_eh_filter_types_ptr): Likewise.
7856 (gimple_eh_filter_failure_ptr): Likewise.
7857 (gimple_eh_filter_set_types): Likewise.
7858 (gimple_eh_filter_set_failure): Likewise.
7859 (gimple_eh_must_not_throw_fndecl): Likewise.
7860 (gimple_eh_must_not_throw_set_fndecl): Likewise.
7861 (gimple_eh_else_n_body_ptr): Likewise.
7862 (gimple_eh_else_e_body_ptr): Likewise.
7863 (gimple_eh_else_set_n_body): Likewise.
7864 (gimple_eh_else_set_e_body): Likewise.
7865 (gimple_try_eval_ptr): Likewise.
7866 (gimple_try_cleanup_ptr): Likewise.
7867 (gimple_try_set_eval): Likewise.
7868 (gimple_try_set_cleanup): Likewise.
7869 (gimple_wce_cleanup_ptr): Likewise.
7870 (gimple_wce_set_cleanup): Likewise.
7871 (gimple_phi_capacity): Likewise.
7872 (gimple_phi_num_args): Likewise.
7873 (gimple_phi_result): Likewise.
7874 (gimple_phi_result_ptr): Likewise.
7875 (gimple_phi_set_result): Likewise.
7876 (gimple_phi_arg): Likewise.
7877 (gimple_phi_set_arg): Likewise.
7878 (gimple_resx_region): Likewise.
7879 (gimple_resx_set_region): Likewise.
7880 (gimple_eh_dispatch_region): Likewise.
7881 (gimple_eh_dispatch_set_region): Likewise.
7882 (gimple_omp_critical_name): Likewise.
7883 (gimple_omp_critical_name_ptr): Likewise.
7884 (gimple_omp_critical_set_name): Likewise.
7885 (gimple_omp_for_clauses): Likewise.
7886 (gimple_omp_for_clauses_ptr): Likewise.
7887 (gimple_omp_for_set_clauses): Likewise.
7888 (gimple_omp_for_collapse): Likewise.
7889 (gimple_omp_for_index): Likewise.
7890 (gimple_omp_for_index_ptr): Likewise.
7891 (gimple_omp_for_set_index): Likewise.
7892 (gimple_omp_for_initial): Likewise.
7893 (gimple_omp_for_initial_ptr): Likewise.
7894 (gimple_omp_for_set_initial): Likewise.
7895 (gimple_omp_for_final): Likewise.
7896 (gimple_omp_for_final_ptr): Likewise.
7897 (gimple_omp_for_set_final): Likewise.
7898 (gimple_omp_for_incr): Likewise.
7899 (gimple_omp_for_incr_ptr): Likewise.
7900 (gimple_omp_for_set_incr): Likewise.
7901 (gimple_omp_for_pre_body_ptr): Likewise.
7902 (gimple_omp_for_set_pre_body): Likewise.
7903 (gimple_omp_parallel_clauses): Likewise.
7904 (gimple_omp_parallel_clauses_ptr): Likewise.
7905 (gimple_omp_parallel_set_clauses): Likewise.
7906 (gimple_omp_parallel_child_fn): Likewise.
7907 (gimple_omp_parallel_child_fn_ptr): Likewise.
7908 (gimple_omp_parallel_set_child_fn): Likewise.
7909 (gimple_omp_parallel_data_arg): Likewise.
7910 (gimple_omp_parallel_data_arg_ptr): Likewise.
7911 (gimple_omp_parallel_set_data_arg): Likewise.
7912 (gimple_omp_task_clauses): Likewise.
7913 (gimple_omp_task_clauses_ptr): Likewise.
7914 (gimple_omp_task_set_clauses): Likewise.
7915 (gimple_omp_task_child_fn): Likewise.
7916 (gimple_omp_task_child_fn_ptr): Likewise.
7917 (gimple_omp_task_set_child_fn): Likewise.
7918 (gimple_omp_task_data_arg): Likewise.
7919 (gimple_omp_task_data_arg_ptr): Likewise.
7920 (gimple_omp_task_set_data_arg): Likewise.
7921 (gimple_omp_taskreg_clauses): Likewise.
7922 (gimple_omp_taskreg_clauses_ptr): Likewise.
7923 (gimple_omp_taskreg_set_clauses): Likewise.
7924 (gimple_omp_taskreg_child_fn): Likewise.
7925 (gimple_omp_taskreg_child_fn_ptr): Likewise.
7926 (gimple_omp_taskreg_set_child_fn): Likewise.
7927 (gimple_omp_taskreg_data_arg): Likewise.
7928 (gimple_omp_taskreg_data_arg_ptr): Likewise.
7929 (gimple_omp_taskreg_set_data_arg): Likewise.
7930 (gimple_omp_task_copy_fn): Likewise.
7931 (gimple_omp_task_copy_fn_ptr): Likewise.
7932 (gimple_omp_task_set_copy_fn): Likewise.
7933 (gimple_omp_task_arg_size): Likewise.
7934 (gimple_omp_task_arg_size_ptr): Likewise.
7935 (gimple_omp_task_set_arg_size): Likewise.
7936 (gimple_omp_task_arg_align): Likewise.
7937 (gimple_omp_task_arg_align_ptr): Likewise.
7938 (gimple_omp_task_set_arg_align): Likewise.
7939 (gimple_omp_single_clauses): Likewise.
7940 (gimple_omp_single_clauses_ptr): Likewise.
7941 (gimple_omp_single_set_clauses): Likewise.
7942 (gimple_omp_target_clauses): Likewise.
7943 (gimple_omp_target_clauses_ptr): Likewise.
7944 (gimple_omp_target_set_clauses): Likewise.
7945 (gimple_omp_target_child_fn): Likewise.
7946 (gimple_omp_target_child_fn_ptr): Likewise.
7947 (gimple_omp_target_set_child_fn): Likewise.
7948 (gimple_omp_target_data_arg): Likewise.
7949 (gimple_omp_target_data_arg_ptr): Likewise.
7950 (gimple_omp_target_set_data_arg): Likewise.
7951 (gimple_omp_teams_clauses): Likewise.
7952 (gimple_omp_teams_clauses_ptr): Likewise.
7953 (gimple_omp_teams_set_clauses): Likewise.
7954 (gimple_omp_sections_clauses): Likewise.
7955 (gimple_omp_sections_clauses_ptr): Likewise.
7956 (gimple_omp_sections_set_clauses): Likewise.
7957 (gimple_omp_sections_control): Likewise.
7958 (gimple_omp_sections_control_ptr): Likewise.
7959 (gimple_omp_sections_set_control): Likewise.
7960 (gimple_omp_for_set_cond): Likewise.
7961 (gimple_omp_for_cond): Likewise.
7962 (gimple_omp_atomic_store_set_val): Likewise.
7963 (gimple_omp_atomic_store_val): Likewise.
7964 (gimple_omp_atomic_store_val_ptr): Likewise.
7965 (gimple_omp_atomic_load_set_lhs): Likewise.
7966 (gimple_omp_atomic_load_lhs): Likewise.
7967 (gimple_omp_atomic_load_lhs_ptr): Likewise.
7968 (gimple_omp_atomic_load_set_rhs): Likewise.
7969 (gimple_omp_atomic_load_rhs): Likewise.
7970 (gimple_omp_atomic_load_rhs_ptr): Likewise.
7971 (gimple_omp_continue_control_def): Likewise.
7972 (gimple_omp_continue_control_def_ptr): Likewise.
7973 (gimple_omp_continue_set_control_def): Likewise.
7974 (gimple_omp_continue_control_use): Likewise.
7975 (gimple_omp_continue_control_use_ptr): Likewise.
7976 (gimple_omp_continue_set_control_use): Likewise.
7977 (gimple_transaction_body_ptr): Likewise.
7978 (gimple_transaction_label): Likewise.
7979 (gimple_transaction_label_ptr): Likewise.
7980 (gimple_transaction_set_body): Likewise.
7981 (gimple_transaction_set_label): Likewise.
7983 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
7984 * ipa-inline-analysis.c (inline_write_summary): Likewise.
7985 * ipa-ref.c (ipa_record_reference): Likewise.
7986 * ipa-reference.c (analyze_function): Likewise.
7987 (ipa_reference_write_optimization_summary): Likewise.
7988 * ipa.c (symtab_remove_unreachable_nodes): Likewise.
7989 (address_taken_from_non_vtable_p): Likewise.
7990 (comdat_can_be_unshared_p_1): Likewise.
7991 * lto-cgraph.c (lto_output_ref): Likewise.
7992 (add_references): Likewise.
7993 (compute_ltrans_boundary): Likewise.
7994 (output_symtab): Likewise.
7995 (input_ref): Likewise.
7996 (input_cgraph_1): Likewise.
7997 (output_cgraph_opt_summary): Likewise.
7998 * lto-streamer-out.c (lto_output): Likewise.
7999 (output_symbol_p): Likewise.
8000 * lto-streamer.h (lsei_next_function_in_partition): Likewise.
8001 (lsei_start_function_in_partition): Likewise.
8002 (lsei_next_variable_in_partition): Likewise.
8003 (lsei_start_variable_in_partition): Likewise.
8004 * symtab.c (insert_to_assembler_name_hash): Likewise.
8005 (unlink_from_assembler_name_hash): Likewise.
8006 (symtab_unregister_node): Likewise.
8007 (symtab_remove_node): Likewise.
8008 (dump_symtab_node): Likewise.
8009 (verify_symtab_base): Likewise.
8010 (verify_symtab_node): Likewise.
8011 (symtab_make_decl_local): Likewise.
8012 (symtab_alias_ultimate_target): Likewise.
8013 (symtab_resolve_alias): Likewise.
8014 (symtab_get_symbol_partitioning_class): Likewise.
8015 * tree-phinodes.c (allocate_phi_node): Likewise.
8016 (reserve_phi_args_for_new_edge): Likewise.
8017 (remove_phi_args): Likewise.
8018 * varpool.c (varpool_node_for_asm): Likewise.
8019 (varpool_remove_unreferenced_decls): Likewise.
8021 2014-04-23 Jeff Law <law@redhat.com>
8023 PR tree-optimization/60902
8024 * tree-ssa-threadedge.c
8025 (record_temporary_equivalences_from_stmts_at_dest): Make sure to
8026 invalidate outputs from statements that do not produce useful
8027 outputs for threading.
8029 2014-04-23 Venkataramanan Kumar <venkataramanan.kumar@linaro.org>
8031 * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
8032 (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
8033 machine descriptions for Stack Smashing Protector.
8035 2014-04-23 Richard Earnshaw <rearnsha@arm.com>
8037 * aarch64.md (<optab>_rol<mode>3): New pattern.
8038 (<optab>_rolsi3_uxtw): Likewise.
8039 * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
8041 2014-04-23 James Greenhalgh <james.greenhalgh@arm.com>
8043 * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
8044 (arm_cortex_a12_tune): Likewise.
8046 2014-04-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8048 * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
8050 2014-04-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8052 * config/arm/arm.md (arm_rev16si2): New pattern.
8053 (arm_rev16si2_alt): Likewise.
8054 * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
8056 2014-04-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8058 * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
8059 (rev16<mode>2_alt): Likewise.
8060 * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
8061 * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
8062 (aarch_rev16_shleft_mask_imm_p): Likewise.
8063 (aarch_rev16_p_1): Likewise.
8064 (aarch_rev16_p): Likewise.
8065 * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
8066 (aarch_rev16_shright_mask_imm_p): Likewise.
8067 (aarch_rev16_shleft_mask_imm_p): Likewise.
8069 2014-04-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8071 * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
8072 * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
8074 (cortex_a53_extra_costs): Likewise.
8075 (cortex_a57_extra_costs): Likewise.
8076 * config/arm/arm.c (cortexa9_extra_costs): Likewise.
8077 (cortexa7_extra_costs): Likewise.
8078 (cortexa8_extra_costs): Likewise.
8079 (cortexa12_extra_costs): Likewise.
8080 (cortexa15_extra_costs): Likewise.
8081 (v7m_extra_costs): Likewise.
8082 (arm_new_rtx_costs): Handle BSWAP.
8084 2013-04-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8086 * config/arm/arm.c (cortexa8_extra_costs): New table.
8087 (arm_cortex_a8_tune): New tuning struct.
8088 * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
8090 2014-04-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8092 * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
8094 2014-04-23 Richard Biener <rguenther@suse.de>
8096 * Makefile.in (OBJS): Remove loop-unswitch.o.
8097 * tree-pass.h (make_pass_rtl_unswitch): Remove.
8098 * passes.def (pass_rtl_unswitch): Likewise.
8099 * loop-init.c (gate_rtl_unswitch): Likewise.
8100 (rtl_unswitch): Likewise.
8101 (pass_data_rtl_unswitch): Likewise.
8102 (pass_rtl_unswitch): Likewise.
8103 (make_pass_rtl_unswitch): Likewise.
8104 * rtl.h (reversed_condition): Likewise.
8105 (compare_and_jump_seq): Likewise.
8106 * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
8108 * loop-unroll.c (compare_and_jump_seq): Likewise.
8110 2014-04-23 Richard Biener <rguenther@suse.de>
8112 PR tree-optimization/60903
8113 * tree-ssa-loop-im.c (analyze_memory_references): Remove
8114 commented code block.
8115 (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
8116 loop flags to newly created BBs and edges.
8118 2014-04-23 Nick Clifton <nickc@redhat.com>
8120 * config/msp430/msp430.c (msp430_handle_option): Move function
8122 (msp430_option_override): Simplify mcu and mcpu option handling.
8123 (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult. Add
8124 support for -mhwmult command line option.
8125 (has_32bit_hwmult): Rename to use_32bit_hwmult. Add support for
8126 -mhwmult command line option.
8127 (msp430_hwmult_enabled): Delete.
8128 (msp43o_output_labelref): Add support for -mhwmult command line option.
8129 * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
8130 (umulsidi3): Likewise.
8131 * config/msp430/msp430.opt (mmcu): Add Report attribute.
8132 (mcpu, mlarge, msmall): Likewise.
8133 (mhwmult): New option.
8134 * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
8136 (msp430_is_f5_mcu): Remove prototype.
8137 (msp430_use_f5_series_hwmult): Add prototype.
8138 * config/msp430/msp430-opts.h: New file.
8139 * common/config/msp430: New directory.
8140 * common/config/msp430/msp430-common.c: New file.
8141 * config.gcc (msp430): Remove target_has_targetm_common.
8142 * doc/invoke.texi: Document -mhwmult command line option.
8144 2014-04-23 Nick Clifton <nickc@redhat.com>
8146 * config/i386/cygwin.h (ENDFILE_SPEC): Include
8147 default-manifest.o if it can be found in the search path.
8148 * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
8150 2014-04-23 Terry Guo <terry.guo@arm.com>
8152 * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
8154 2014-04-23 Richard Biener <rguenther@suse.de>
8157 * tree-inline.c (declare_return_variable): Use mark_addressable.
8159 2014-04-23 Richard Biener <rguenther@suse.de>
8162 * loop-init.c (loop_optimizer_init): Make sure to apply
8163 LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
8165 2014-04-22 Jakub Jelinek <jakub@redhat.com>
8168 * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
8170 * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
8171 if flag_sanitize_undefined_trap_on_error.
8172 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
8173 BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
8174 BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
8175 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
8176 BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
8177 BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
8178 BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
8179 BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
8180 BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
8181 * ubsan.c (ubsan_instrument_unreachable): Return
8182 __builtin_trap () if flag_sanitize_undefined_trap_on_error.
8183 (ubsan_expand_null_ifn): Emit __builtin_trap ()
8184 if flag_sanitize_undefined_trap_on_error and
8185 __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
8186 (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
8187 instrument_bool_enum_load): Emit __builtin_trap () if
8188 flag_sanitize_undefined_trap_on_error and
8189 __builtin_handle_*_abort () if !flag_sanitize_recover.
8190 * doc/invoke.texi (-fsanitize-recover,
8191 -fsanitize-undefined-trap-on-error): Document.
8193 2014-04-22 Christian Bruel <christian.bruel@st.com>
8195 * config/sh/sh.md (mov<mode>): Replace movQIHI.
8196 Force immediates to SImode.
8198 2014-04-22 Sandra Loosemore <sandra@codesourcery.com>
8200 * config/nios2/nios2.md (UNSPEC_ROUND): New.
8202 * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
8203 * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
8204 * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
8205 (nios2_fpu_insn): Add entry for round.
8206 (N2FPU_NO_ERRNO_P): Define.
8207 (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
8209 * doc/invoke.texi (Nios II Options): Document -mcustom-round.
8211 2014-04-22 Richard Henderson <rth@redhat.com>
8213 * config/aarch64/aarch64 (addti3, subti3): New expanders.
8214 (add<GPI>3_compare0): Remove leading * from name.
8215 (add<GPI>3_carryin): Likewise.
8216 (sub<GPI>3_compare0): Likewise.
8217 (sub<GPI>3_carryin): Likewise.
8218 (<su_optab>mulditi3): New expander.
8219 (multi3): New expander.
8220 (madd<GPI>): Remove leading * from name.
8222 2014-04-22 Martin Jambor <mjambor@suse.cz>
8224 * cgraphclones.c (cgraph_function_versioning): Copy
8225 ipa_transforms_to_apply instead of asserting it is empty.
8227 2014-04-22 H.J. Lu <hongjiu.lu@intel.com>
8230 * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
8231 on count_exp to get mode.
8233 2014-04-22 Andrew Pinski <apinski@cavium.com>
8235 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8236 Handle TLS for ILP32.
8237 * config/aarch64/aarch64.md (tlsie_small): Rename to ...
8238 (tlsie_small_<mode>): this and handle PTR.
8239 (tlsie_small_sidi): New pattern.
8240 (tlsle_small): Change to an expand to handle ILP32.
8241 (tlsle_small_<mode>): New pattern.
8242 (tlsdesc_small): Rename to ...
8243 (tlsdesc_small_<mode>): this and handle PTR.
8245 2014-04-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8247 * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
8249 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
8251 * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
8252 (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
8253 (aarch64_types_signed_poly_qualifiers): Likewise.
8254 (aarch64_types_unsigned_signed_qualifiers): Likewise.
8255 (aarch64_types_poly_signed_qualifiers): Likewise.
8256 (TYPES_REINTERP_SS): Type macro added.
8257 (TYPES_REINTERP_SU): Likewise.
8258 (TYPES_REINTERP_SP): Likewise.
8259 (TYPES_REINTERP_US): Likewise.
8260 (TYPES_REINTERP_PS): Likewise.
8261 (aarch64_fold_builtin): New expression folding added.
8262 * config/aarch64/aarch64-simd-builtins.def (REINTERP):
8263 Declarations removed.
8264 (REINTERP_SS): Declarations added.
8265 (REINTERP_US): Likewise.
8266 (REINTERP_PS): Likewise.
8267 (REINTERP_SU): Likewise.
8268 (REINTERP_SP): Likewise.
8269 * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
8270 (vreinterpretq_p8_f64): Likewise.
8271 (vreinterpret_p16_f64): Likewise.
8272 (vreinterpretq_p16_f64): Likewise.
8273 (vreinterpret_f32_f64): Likewise.
8274 (vreinterpretq_f32_f64): Likewise.
8275 (vreinterpret_f64_f32): Likewise.
8276 (vreinterpret_f64_p8): Likewise.
8277 (vreinterpret_f64_p16): Likewise.
8278 (vreinterpret_f64_s8): Likewise.
8279 (vreinterpret_f64_s16): Likewise.
8280 (vreinterpret_f64_s32): Likewise.
8281 (vreinterpret_f64_s64): Likewise.
8282 (vreinterpret_f64_u8): Likewise.
8283 (vreinterpret_f64_u16): Likewise.
8284 (vreinterpret_f64_u32): Likewise.
8285 (vreinterpret_f64_u64): Likewise.
8286 (vreinterpretq_f64_f32): Likewise.
8287 (vreinterpretq_f64_p8): Likewise.
8288 (vreinterpretq_f64_p16): Likewise.
8289 (vreinterpretq_f64_s8): Likewise.
8290 (vreinterpretq_f64_s16): Likewise.
8291 (vreinterpretq_f64_s32): Likewise.
8292 (vreinterpretq_f64_s64): Likewise.
8293 (vreinterpretq_f64_u8): Likewise.
8294 (vreinterpretq_f64_u16): Likewise.
8295 (vreinterpretq_f64_u32): Likewise.
8296 (vreinterpretq_f64_u64): Likewise.
8297 (vreinterpret_s64_f64): Likewise.
8298 (vreinterpretq_s64_f64): Likewise.
8299 (vreinterpret_u64_f64): Likewise.
8300 (vreinterpretq_u64_f64): Likewise.
8301 (vreinterpret_s8_f64): Likewise.
8302 (vreinterpretq_s8_f64): Likewise.
8303 (vreinterpret_s16_f64): Likewise.
8304 (vreinterpretq_s16_f64): Likewise.
8305 (vreinterpret_s32_f64): Likewise.
8306 (vreinterpretq_s32_f64): Likewise.
8307 (vreinterpret_u8_f64): Likewise.
8308 (vreinterpretq_u8_f64): Likewise.
8309 (vreinterpret_u16_f64): Likewise.
8310 (vreinterpretq_u16_f64): Likewise.
8311 (vreinterpret_u32_f64): Likewise.
8312 (vreinterpretq_u32_f64): Likewise.
8314 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
8316 * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
8317 * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
8318 (vreinterpret_p8_s8): Likewise.
8319 * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
8320 (vreinterpret_p8_s16): Likewise.
8321 (vreinterpret_p8_s32): Likewise.
8322 (vreinterpret_p8_s64): Likewise.
8323 (vreinterpret_p8_f32): Likewise.
8324 (vreinterpret_p8_u8): Likewise.
8325 (vreinterpret_p8_u16): Likewise.
8326 (vreinterpret_p8_u32): Likewise.
8327 (vreinterpret_p8_u64): Likewise.
8328 (vreinterpret_p8_p16): Likewise.
8329 (vreinterpretq_p8_s8): Likewise.
8330 (vreinterpretq_p8_s16): Likewise.
8331 (vreinterpretq_p8_s32): Likewise.
8332 (vreinterpretq_p8_s64): Likewise.
8333 (vreinterpretq_p8_f32): Likewise.
8334 (vreinterpretq_p8_u8): Likewise.
8335 (vreinterpretq_p8_u16): Likewise.
8336 (vreinterpretq_p8_u32): Likewise.
8337 (vreinterpretq_p8_u64): Likewise.
8338 (vreinterpretq_p8_p16): Likewise.
8339 (vreinterpret_p16_s8): Likewise.
8340 (vreinterpret_p16_s16): Likewise.
8341 (vreinterpret_p16_s32): Likewise.
8342 (vreinterpret_p16_s64): Likewise.
8343 (vreinterpret_p16_f32): Likewise.
8344 (vreinterpret_p16_u8): Likewise.
8345 (vreinterpret_p16_u16): Likewise.
8346 (vreinterpret_p16_u32): Likewise.
8347 (vreinterpret_p16_u64): Likewise.
8348 (vreinterpret_p16_p8): Likewise.
8349 (vreinterpretq_p16_s8): Likewise.
8350 (vreinterpretq_p16_s16): Likewise.
8351 (vreinterpretq_p16_s32): Likewise.
8352 (vreinterpretq_p16_s64): Likewise.
8353 (vreinterpretq_p16_f32): Likewise.
8354 (vreinterpretq_p16_u8): Likewise.
8355 (vreinterpretq_p16_u16): Likewise.
8356 (vreinterpretq_p16_u32): Likewise.
8357 (vreinterpretq_p16_u64): Likewise.
8358 (vreinterpretq_p16_p8): Likewise.
8359 (vreinterpret_f32_s8): Likewise.
8360 (vreinterpret_f32_s16): Likewise.
8361 (vreinterpret_f32_s32): Likewise.
8362 (vreinterpret_f32_s64): Likewise.
8363 (vreinterpret_f32_u8): Likewise.
8364 (vreinterpret_f32_u16): Likewise.
8365 (vreinterpret_f32_u32): Likewise.
8366 (vreinterpret_f32_u64): Likewise.
8367 (vreinterpret_f32_p8): Likewise.
8368 (vreinterpret_f32_p16): Likewise.
8369 (vreinterpretq_f32_s8): Likewise.
8370 (vreinterpretq_f32_s16): Likewise.
8371 (vreinterpretq_f32_s32): Likewise.
8372 (vreinterpretq_f32_s64): Likewise.
8373 (vreinterpretq_f32_u8): Likewise.
8374 (vreinterpretq_f32_u16): Likewise.
8375 (vreinterpretq_f32_u32): Likewise.
8376 (vreinterpretq_f32_u64): Likewise.
8377 (vreinterpretq_f32_p8): Likewise.
8378 (vreinterpretq_f32_p16): Likewise.
8379 (vreinterpret_s64_s8): Likewise.
8380 (vreinterpret_s64_s16): Likewise.
8381 (vreinterpret_s64_s32): Likewise.
8382 (vreinterpret_s64_f32): Likewise.
8383 (vreinterpret_s64_u8): Likewise.
8384 (vreinterpret_s64_u16): Likewise.
8385 (vreinterpret_s64_u32): Likewise.
8386 (vreinterpret_s64_u64): Likewise.
8387 (vreinterpret_s64_p8): Likewise.
8388 (vreinterpret_s64_p16): Likewise.
8389 (vreinterpretq_s64_s8): Likewise.
8390 (vreinterpretq_s64_s16): Likewise.
8391 (vreinterpretq_s64_s32): Likewise.
8392 (vreinterpretq_s64_f32): Likewise.
8393 (vreinterpretq_s64_u8): Likewise.
8394 (vreinterpretq_s64_u16): Likewise.
8395 (vreinterpretq_s64_u32): Likewise.
8396 (vreinterpretq_s64_u64): Likewise.
8397 (vreinterpretq_s64_p8): Likewise.
8398 (vreinterpretq_s64_p16): Likewise.
8399 (vreinterpret_u64_s8): Likewise.
8400 (vreinterpret_u64_s16): Likewise.
8401 (vreinterpret_u64_s32): Likewise.
8402 (vreinterpret_u64_s64): Likewise.
8403 (vreinterpret_u64_f32): Likewise.
8404 (vreinterpret_u64_u8): Likewise.
8405 (vreinterpret_u64_u16): Likewise.
8406 (vreinterpret_u64_u32): Likewise.
8407 (vreinterpret_u64_p8): Likewise.
8408 (vreinterpret_u64_p16): Likewise.
8409 (vreinterpretq_u64_s8): Likewise.
8410 (vreinterpretq_u64_s16): Likewise.
8411 (vreinterpretq_u64_s32): Likewise.
8412 (vreinterpretq_u64_s64): Likewise.
8413 (vreinterpretq_u64_f32): Likewise.
8414 (vreinterpretq_u64_u8): Likewise.
8415 (vreinterpretq_u64_u16): Likewise.
8416 (vreinterpretq_u64_u32): Likewise.
8417 (vreinterpretq_u64_p8): Likewise.
8418 (vreinterpretq_u64_p16): Likewise.
8419 (vreinterpret_s8_s16): Likewise.
8420 (vreinterpret_s8_s32): Likewise.
8421 (vreinterpret_s8_s64): Likewise.
8422 (vreinterpret_s8_f32): Likewise.
8423 (vreinterpret_s8_u8): Likewise.
8424 (vreinterpret_s8_u16): Likewise.
8425 (vreinterpret_s8_u32): Likewise.
8426 (vreinterpret_s8_u64): Likewise.
8427 (vreinterpret_s8_p8): Likewise.
8428 (vreinterpret_s8_p16): Likewise.
8429 (vreinterpretq_s8_s16): Likewise.
8430 (vreinterpretq_s8_s32): Likewise.
8431 (vreinterpretq_s8_s64): Likewise.
8432 (vreinterpretq_s8_f32): Likewise.
8433 (vreinterpretq_s8_u8): Likewise.
8434 (vreinterpretq_s8_u16): Likewise.
8435 (vreinterpretq_s8_u32): Likewise.
8436 (vreinterpretq_s8_u64): Likewise.
8437 (vreinterpretq_s8_p8): Likewise.
8438 (vreinterpretq_s8_p16): Likewise.
8439 (vreinterpret_s16_s8): Likewise.
8440 (vreinterpret_s16_s32): Likewise.
8441 (vreinterpret_s16_s64): Likewise.
8442 (vreinterpret_s16_f32): Likewise.
8443 (vreinterpret_s16_u8): Likewise.
8444 (vreinterpret_s16_u16): Likewise.
8445 (vreinterpret_s16_u32): Likewise.
8446 (vreinterpret_s16_u64): Likewise.
8447 (vreinterpret_s16_p8): Likewise.
8448 (vreinterpret_s16_p16): Likewise.
8449 (vreinterpretq_s16_s8): Likewise.
8450 (vreinterpretq_s16_s32): Likewise.
8451 (vreinterpretq_s16_s64): Likewise.
8452 (vreinterpretq_s16_f32): Likewise.
8453 (vreinterpretq_s16_u8): Likewise.
8454 (vreinterpretq_s16_u16): Likewise.
8455 (vreinterpretq_s16_u32): Likewise.
8456 (vreinterpretq_s16_u64): Likewise.
8457 (vreinterpretq_s16_p8): Likewise.
8458 (vreinterpretq_s16_p16): Likewise.
8459 (vreinterpret_s32_s8): Likewise.
8460 (vreinterpret_s32_s16): Likewise.
8461 (vreinterpret_s32_s64): Likewise.
8462 (vreinterpret_s32_f32): Likewise.
8463 (vreinterpret_s32_u8): Likewise.
8464 (vreinterpret_s32_u16): Likewise.
8465 (vreinterpret_s32_u32): Likewise.
8466 (vreinterpret_s32_u64): Likewise.
8467 (vreinterpret_s32_p8): Likewise.
8468 (vreinterpret_s32_p16): Likewise.
8469 (vreinterpretq_s32_s8): Likewise.
8470 (vreinterpretq_s32_s16): Likewise.
8471 (vreinterpretq_s32_s64): Likewise.
8472 (vreinterpretq_s32_f32): Likewise.
8473 (vreinterpretq_s32_u8): Likewise.
8474 (vreinterpretq_s32_u16): Likewise.
8475 (vreinterpretq_s32_u32): Likewise.
8476 (vreinterpretq_s32_u64): Likewise.
8477 (vreinterpretq_s32_p8): Likewise.
8478 (vreinterpretq_s32_p16): Likewise.
8479 (vreinterpret_u8_s8): Likewise.
8480 (vreinterpret_u8_s16): Likewise.
8481 (vreinterpret_u8_s32): Likewise.
8482 (vreinterpret_u8_s64): Likewise.
8483 (vreinterpret_u8_f32): Likewise.
8484 (vreinterpret_u8_u16): Likewise.
8485 (vreinterpret_u8_u32): Likewise.
8486 (vreinterpret_u8_u64): Likewise.
8487 (vreinterpret_u8_p8): Likewise.
8488 (vreinterpret_u8_p16): Likewise.
8489 (vreinterpretq_u8_s8): Likewise.
8490 (vreinterpretq_u8_s16): Likewise.
8491 (vreinterpretq_u8_s32): Likewise.
8492 (vreinterpretq_u8_s64): Likewise.
8493 (vreinterpretq_u8_f32): Likewise.
8494 (vreinterpretq_u8_u16): Likewise.
8495 (vreinterpretq_u8_u32): Likewise.
8496 (vreinterpretq_u8_u64): Likewise.
8497 (vreinterpretq_u8_p8): Likewise.
8498 (vreinterpretq_u8_p16): Likewise.
8499 (vreinterpret_u16_s8): Likewise.
8500 (vreinterpret_u16_s16): Likewise.
8501 (vreinterpret_u16_s32): Likewise.
8502 (vreinterpret_u16_s64): Likewise.
8503 (vreinterpret_u16_f32): Likewise.
8504 (vreinterpret_u16_u8): Likewise.
8505 (vreinterpret_u16_u32): Likewise.
8506 (vreinterpret_u16_u64): Likewise.
8507 (vreinterpret_u16_p8): Likewise.
8508 (vreinterpret_u16_p16): Likewise.
8509 (vreinterpretq_u16_s8): Likewise.
8510 (vreinterpretq_u16_s16): Likewise.
8511 (vreinterpretq_u16_s32): Likewise.
8512 (vreinterpretq_u16_s64): Likewise.
8513 (vreinterpretq_u16_f32): Likewise.
8514 (vreinterpretq_u16_u8): Likewise.
8515 (vreinterpretq_u16_u32): Likewise.
8516 (vreinterpretq_u16_u64): Likewise.
8517 (vreinterpretq_u16_p8): Likewise.
8518 (vreinterpretq_u16_p16): Likewise.
8519 (vreinterpret_u32_s8): Likewise.
8520 (vreinterpret_u32_s16): Likewise.
8521 (vreinterpret_u32_s32): Likewise.
8522 (vreinterpret_u32_s64): Likewise.
8523 (vreinterpret_u32_f32): Likewise.
8524 (vreinterpret_u32_u8): Likewise.
8525 (vreinterpret_u32_u16): Likewise.
8526 (vreinterpret_u32_u64): Likewise.
8527 (vreinterpret_u32_p8): Likewise.
8528 (vreinterpret_u32_p16): Likewise.
8529 (vreinterpretq_u32_s8): Likewise.
8530 (vreinterpretq_u32_s16): Likewise.
8531 (vreinterpretq_u32_s32): Likewise.
8532 (vreinterpretq_u32_s64): Likewise.
8533 (vreinterpretq_u32_f32): Likewise.
8534 (vreinterpretq_u32_u8): Likewise.
8535 (vreinterpretq_u32_u16): Likewise.
8536 (vreinterpretq_u32_u64): Likewise.
8537 (vreinterpretq_u32_p8): Likewise.
8538 (vreinterpretq_u32_p16): Likewise.
8540 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
8542 * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
8544 * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator extended.
8546 * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
8547 (vqnegd_s64): Likewise.
8548 (vqabs_s64): Likewise.
8549 (vqabsd_s64): Likewise.
8551 2014-04-22 Richard Henderson <rth@redhat.com>
8553 * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
8554 computation to the top of the loop.
8556 2014-04-22 Renlin <renlin.li@arm.com>
8557 Jiong Wang <jiong.wang@arm.com>
8559 * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
8560 * config/aarch64/aarch64.c (aarch64_layout_frame)
8561 (aarch64_initial_elimination_offset): Likewise.
8563 2014-04-22 Marcus Shawcroft <marcus.shawcroft@arm.com>
8565 * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
8568 2014-04-22 Richard Sandiford <rdsandiford@googlemail.com>
8570 * machmode.h (bitwise_mode_for_mode): Declare.
8571 * stor-layout.h (bitwise_type_for_mode): Likewise.
8572 * stor-layout.c (bitwise_mode_for_mode): New function.
8573 (bitwise_type_for_mode): Likewise.
8574 * builtins.c (fold_builtin_memory_op): Use it instead of
8575 int_mode_for_mode and build_nonstandard_integer_type.
8577 2014-04-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8579 * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
8580 (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
8581 (*-*-solaris2*): Simplify.
8582 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
8583 (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
8584 *-*-solaris2.9* handling.
8586 * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
8588 (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
8589 (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
8591 (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
8592 * configure: Regenerate.
8594 * config/i386/sol2-9.h: Remove.
8596 * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
8597 (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
8598 Remove Solaris 9 references.
8600 2014-04-22 Vidya Praveen <vidyapraveen@arm.com>
8602 * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
8603 (floatuns<GPI:mode><GPF:mode>2): Remove.
8604 (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
8605 and floatuns conversions.
8606 (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
8607 and floatuns conversions.
8608 * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
8609 (w1,w2): New mode attributes for inequal width conversions.
8611 2014-04-22 Renlin Li <Renlin.Li@arm.com>
8613 * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
8614 the output asm format.
8616 2014-04-22 James Greenhalgh <james.greenhalgh@arm.com>
8618 * config/aarch64/aarch64-simd.md
8619 (aarch64_cm<optab>di): Always split.
8620 (*aarch64_cm<optab>di): New.
8621 (aarch64_cmtstdi): Always split.
8622 (*aarch64_cmtstdi): New.
8624 2014-04-22 Jakub Jelinek <jakub@redhat.com>
8626 PR tree-optimization/60823
8627 * omp-low.c (ipa_simd_modify_function_body): Go through
8628 all SSA_NAMEs and for those refering to vector arguments
8629 which are going to be replaced adjust SSA_NAME_VAR and,
8630 if it is a default definition, change it into a non-default
8631 definition assigned at the beginning of function from new_decl.
8632 (ipa_simd_modify_stmt_ops): Rewritten.
8633 * tree-dfa.c (set_ssa_default_def): When removing default def,
8634 check for NULL loc instead of NULL *loc.
8636 2014-04-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8638 * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
8639 restrictions on core registers for DImode values in Thumb2.
8641 2014-04-22 Ian Bolton <ian.bolton@arm.com>
8643 * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
8644 * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
8646 2014-04-22 Ian Bolton <ian.bolton@arm.com>
8648 * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
8649 (*iordi_notzesidi_di): Likewise.
8650 (*iordi_notsesidi_di): Likewise.
8652 2014-04-22 Ian Bolton <ian.bolton@arm.com>
8654 * config/arm/arm-protos.h (tune_params): New struct members.
8655 * config/arm/arm.c: Initialise tune_params per processor.
8656 (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
8657 for speed, based on new tune_params.
8659 2014-04-22 Alex Velenko <Alex.Velenko@arm.com>
8661 * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro added.
8662 * config/aarch64/aarch64-simd-builtins.def (frintn): Use added macro.
8663 * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment corrected.
8664 * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
8665 * config/aarch64/arm_neon.h (vrnd_f64): Added.
8666 (vrnda_f64): Likewise.
8667 (vrndi_f64): Likewise.
8668 (vrndm_f64): Likewise.
8669 (vrndn_f64): Likewise.
8670 (vrndp_f64): Likewise.
8671 (vrndx_f64): Likewise.
8673 2014-04-22 Zhenqiang Chen <zhenqiang.chen@linaro.org>
8675 * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
8676 GET_MODE_SIZE argument is enum machine_mode.
8678 2014-04-22 Jakub Jelinek <jakub@redhat.com>
8681 * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
8682 value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
8684 2014-04-22 Lin Zuojian <manjian2006@gmail.com>
8687 * asan.c (asan_emit_stack_protection): Force the base to align to
8688 appropriate bits if STRICT_ALIGNMENT. Set shadow_mem align to
8689 appropriate bits if STRICT_ALIGNMENT.
8690 * cfgexpand.c (expand_stack_vars): Set base_align appropriately
8692 (expand_used_vars): Leave a space in the stack frame for alignment
8693 if STRICT_ALIGNMENT.
8695 2014-04-21 David Malcolm <dmalcolm@redhat.com>
8697 * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
8699 (gimple_store_p): Likewise.
8700 (gimple_assign_load_p): Likewise.
8701 (gimple_assign_cast_p): Likewise.
8702 (gimple_clobber_p): Likewise.
8704 * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
8705 rather than a gimple.
8706 (gimple_assign_cast_p): Likewise.
8708 2014-04-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8711 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
8712 If mode is DDmode and TARGET_E500_DOUBLE allow move.
8714 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
8715 more debug information for E500 if -mdebug=reg.
8717 2014-04-21 Uros Bizjak <ubizjak@gmail.com>
8720 * config/i386/i386.c (ix86_expand_builtin)
8721 <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
8722 register for target RTX.
8723 <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
8725 2014-04-18 Cong Hou <congh@google.com>
8727 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
8728 the widen-mult pattern by handling two operands with different sizes,
8729 and operands whose size is smaller than half of the result type.
8731 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
8733 * ipa-inline.h (INLINE_HINT_known_hot): New hint.
8734 * ipa-inline-analysis.c (dump_inline_hints): Dump it.
8735 (do_estimate_edge_time): Compute it.
8736 * ipa-inline.c (want_inline_small_function_p): Bypass
8737 INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
8739 2014-04-18 Jan Hubicka <hubicka@ucw.cz>
8741 * ipa-inline.c (spec_rem): New static variable.
8742 (dump_overall_stats): New function.
8743 (dump_inline_stats): New function.
8745 2014-04-18 Richard Henderson <rth@redhat.com>
8747 * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
8748 to GET_MODE_SIZE, not a reg_class_t.
8750 2014-04-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8752 * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
8753 (vsx_xxmrglw_<mode>): Likewise.
8755 2014-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
8758 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
8759 GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
8760 (rs6000_init_hard_regno_mode_ok): Likewise.
8762 2014-04-17 Jan Hubicka <hubicka@ucw.cz>
8764 * ipa-inline.c (inline_small_functions): Account only non-cold
8766 * doc/invoke.texi (inline-unit-growth): Update documentation.
8768 2014-04-17 Pat Haugen <pthaugen@us.ibm.com>
8770 * config/rs6000/rs6000.md (addti3, subti3): New.
8772 2014-04-17 H.J. Lu <hongjiu.lu@intel.com>
8775 * config/i386/i386.c (ix86_expand_clear): Remove outdated
8776 comment. Check optimize_insn_for_size_p instead of
8777 optimize_insn_for_speed_p.
8779 2014-04-17 Martin Jambor <mjambor@suse.cz>
8781 * gimple-iterator.c (gsi_start_edge): New function.
8782 * gimple-iterator.h (gsi_start_edge): Declare.
8783 * tree-sra.c (single_non_eh_succ): New function.
8784 (disqualify_ops_if_throwing_stmt): Renamed to
8785 disqualify_if_bad_bb_terminating_stmt. Allow throwing statements
8786 having one non-EH successor BB.
8787 (sra_modify_expr): If stmt ends bb, use single non-EH successor to
8788 generate loads into replacements.
8789 (sra_modify_assign): Likewise and and also use the simple path for
8791 (sra_modify_function_body): Commit statements on edges.
8793 2014-04-17 Richard Biener <rguenther@suse.de>
8796 * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
8797 comparison results and add clarifying comment.
8799 2014-04-17 Jakub Jelinek <jakub@redhat.com>
8801 * genmodes.c (struct mode_data): Add need_bytesize_adj field.
8802 (blank_mode): Initialize it.
8803 (emit_mode_size_inline, emit_mode_nunits_inline,
8804 emit_mode_inner_inline): New functions.
8805 (emit_insn_modes_h): Call them and surround their output with
8806 #if GCC_VERSION >= 4001 ... #endif.
8807 * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
8808 For GCC_VERSION >= 4001 use mode_*_inline routines instead of
8809 mode_* arrays if the argument is __builtin_constant_p.
8810 * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
8811 is enum machine_mode.
8813 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
8815 * passes.c (opt_pass::execute): Adjust.
8816 (pass_manager::execute_pass_mode_switching): Likewise.
8817 (early_local_passes::execute): Likewise.
8818 (execute_one_pass): Pass cfun to the pass's execute method.
8819 * tree-pass.h (opt_pass::execute): Add function * argument.
8820 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
8821 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
8822 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
8823 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
8824 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
8825 config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
8826 except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
8827 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
8828 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
8829 ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
8830 ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
8831 lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
8832 postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
8833 reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
8834 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
8835 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
8836 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
8837 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
8838 tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
8839 tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
8840 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
8841 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
8842 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
8843 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
8844 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
8845 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
8846 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
8847 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
8848 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
8851 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
8853 * passes.c (opt_pass::gate): Take function * argument.
8854 (gate_all_early_local_passes): Merge into
8855 (early_local_passes::gate): this.
8856 (gate_all_early_optimizations): Merge into
8857 (all_early_optimizations::gate): this.
8858 (gate_all_optimizations): Mege into
8859 (all_optimizations::gate): this.
8860 (gate_all_optimizations_g): Merge into
8861 (all_optimizations_g::gate): this.
8862 (gate_rest_of_compilation): Mege into
8863 (rest_of_compilation::gate): this.
8864 (gate_postreload): Merge into
8865 (postreload::gate): this.
8866 (dump_one_pass): Pass cfun to the pass's gate method.
8867 (execute_ipa_summary_passes): Likewise.
8868 (execute_one_pass): Likewise.
8869 (ipa_write_summaries_2): Likewise.
8870 (ipa_write_optimization_summaries_1): Likewise.
8871 (ipa_read_summaries_1): Likewise.
8872 (ipa_read_optimization_summaries_1): Likewise.
8873 (execute_ipa_stmt_fixups): Likewise.
8874 * tree-pass.h (opt_pass::gate): Add function * argument.
8875 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
8876 combine-stack-adj.c, combine.c, compare-elim.c,
8877 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
8878 config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
8879 config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
8880 dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c, fwprop.c, gcse.c,
8881 gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
8882 graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
8883 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
8884 loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
8885 omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
8886 reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
8887 store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
8888 tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
8889 tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
8890 tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
8891 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
8892 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
8893 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
8894 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
8895 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
8896 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
8897 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
8898 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
8899 tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
8900 tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
8901 var-tracking.c, vtable-verify.c, web.c: Adjust.
8903 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
8905 * configure.ac: Check for -Woverloaded-virtual and enable it if found.
8906 * configure: Regenerate.
8908 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
8910 * passes.c (dump_one_pass): don't check pass->has_gate.
8911 (execute_ipa_summary_passes): Likewise.
8912 (execute_one_pass): Likewise.
8913 (ipa_write_summaries_2): Likewise.
8914 (ipa_write_optimization_summaries_1): Likewise.
8915 (ipa_read_optimization_summaries_1): Likewise.
8916 (execute_ipa_stmt_fixups): Likewise.
8917 * tree-pass.h (pass_data::has_gate): Remove.
8918 * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
8919 cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
8920 compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
8921 config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
8922 config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
8923 config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
8924 config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
8925 dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
8926 gimple-low.c, gimple-ssa-isolate-paths.c,
8927 gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
8928 ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
8929 ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
8930 ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
8931 modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
8932 recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
8933 reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
8934 trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
8935 tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
8936 tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
8937 tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
8938 tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
8939 tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
8940 tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
8941 tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
8942 tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
8943 tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
8944 tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
8945 tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
8946 tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
8947 tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
8948 tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
8951 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
8953 * pass_manager.h (pass_manager::register_dump_files_1): Remove
8955 * passes.c (pass_manager::register_dump_files_1): Merge into
8956 (pass_manager::register_dump_files): this, and remove its handling of
8957 properties since the pass always has the properties anyway.
8958 (pass_manager::pass_manager): Adjust.
8960 2014-04-17 Trevor Saunders <tsaunders@mozilla.com>
8962 * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
8963 * passes.c (pass_manager::register_dump_files_1): Remove dead code
8964 dealing with properties.
8965 (pass_manager::register_dump_files): Adjust.
8967 2014-03-20 Mark Wielaard <mjw@redhat.com>
8969 * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
8970 then represent the bound as normal constant value.
8972 2014-04-17 Jakub Jelinek <jakub@redhat.com>
8975 Forward port from 4.8 branch
8976 2013-07-19 Kirill Yukhin <kirill.yukhin@intel.com>
8978 * config/i386/bmiintrin.h (_blsi_u32): New.
8982 (_blsmsk_u32): Ditto.
8983 (_blsmsk_u64): Ditto.
8984 (_tzcnt_u32): Ditto.
8985 (_tzcnt_u64): Ditto.
8987 2014-04-17 Kito Cheng <kito@0xlab.org>
8989 * gcc.c (used_arg): Prevent out of bound access for multilib_options.
8991 2014-04-17 Richard Biener <rguenther@suse.de>
8994 * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
8995 boolean results for comparisons.
8997 2014-04-17 Richard Biener <rguenther@suse.de>
8999 PR tree-optimization/60836
9000 * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
9001 initial PHI args to be gimple values.
9003 2014-04-17 Richard Biener <rguenther@suse.de>
9005 PR tree-optimization/60841
9006 * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
9007 * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
9008 of stmts to SLP build.
9009 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
9010 (vect_analyze_slp): Likewise.
9011 (vect_analyze_slp_instance): Likewise.
9012 (vect_build_slp_tree): Limit overall SLP tree growth.
9013 * tree-vectorizer.h (vect_analyze_data_refs,
9014 vect_analyze_slp): Adjust prototypes.
9016 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
9018 * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
9021 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
9023 * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
9024 * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
9025 * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
9026 for TARGET_SLOW_PSHUFB
9028 2014-04-17 Evgeny Stupachenko <evstupac@gmail.com>
9030 * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
9031 * config/i386/i386.c (intel_cost): Ditto.
9033 2014-04-17 Joey Ye <joey.ye@arm.com>
9035 * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
9037 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
9039 * opts.c (common_handle_option): Disable -fipa-reference coorectly
9042 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
9044 * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
9045 (type_all_derivations_known_p): New predicate.
9046 (type_all_ctors_visible_p): New predicate.
9047 (type_possibly_instantiated_p): New predicate.
9048 (get_odr_type): Compute all_derivations_known.
9049 (dump_odr_type): Dump the flag.
9050 (maybe_record_type): Cleanup.
9051 (record_target_from_binfo): Add bases_to_consider array;
9052 record bases for types w/o instances and skip CXX destructor.
9053 (possible_polymorphic_call_targets_1): Add bases_to_consider
9054 and consider_construction parameters; check if type may have instance.
9055 (get_polymorphic_call_info): Set maybe_in_construction to true
9056 when we know nothing.
9057 (record_targets_from_bases): Skip CXX destructors; they are
9058 never called for types in construction.
9059 (possible_polymorphic_call_targets): Do not record target when
9060 type may not have instance.
9062 2014-04-16 Jan Hubicka <hubicka@ucw.cz>
9065 * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
9066 external aliases alive, too.
9068 2014-04-16 Andrew Pinski <apinski@cavium.com>
9070 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
9073 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
9075 * final.c (compute_alignments): Do not apply loop alignment to a block
9076 falling through to the exit.
9078 2014-04-16 Catherine Moore <clm@codesourcery.com>
9080 * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
9081 Adjust constraints for microMIPS store patterns.
9083 2014-04-16 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
9085 * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
9087 2014-04-16 Eric Botcazou <ebotcazou@adacore.com>
9089 * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
9090 (append_use): Run at -O0.
9091 (append_vdef): Likewise.
9092 * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
9093 * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
9095 2014-04-16 Jakub Jelinek <jakub@redhat.com>
9097 PR tree-optimization/60844
9098 * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
9099 (propagate_op_to_single_use, remove_visited_stmt_chain,
9100 linearize_expr, repropagate_negates, reassociate_bb): Use it
9101 instead of gsi_remove.
9103 2014-04-16 Martin Jambor <mjambor@suse.cz>
9105 * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
9106 ipa_transforms_to_apply.
9107 (cgraph_function_versioning): Assert that old_node has empty
9108 ipa_transforms_to_apply.
9109 * trans-mem.c (ipa_tm_create_version): Likewise.
9110 * tree-inline.c (tree_function_versioning): Do not duplicate
9111 ipa_transforms_to_apply.
9113 2014-04-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9116 * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
9118 Pass necessary as flags on 64-bit Solaris/x86.
9119 Use lowercase relocs for x86_64-*-*.
9120 * configure: Regenerate.
9122 2014-04-15 Jan Hubicka <jh@suse.cz>
9124 * ipa-devirt.c (referenced_from_vtable_p): New predicate.
9125 (maybe_record_node, likely_target_p): Use it.
9127 2014-04-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9130 Revert following patch
9132 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
9135 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
9136 software floating point or no floating point registers, do not
9137 allow any type in the FPRs. Eliminate a test for SPE SIMD types
9138 in GPRs that occurs after we tested for GPRs that would never be
9141 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
9142 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
9143 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
9144 specifically allow DDmode, since that does not use the SPE SIMD
9147 2014-03-21 Mark Wielaard <mjw@redhat.com>
9149 * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
9150 as unsigned or int depending on type and value used.
9152 2014-04-15 Richard Biener <rguenther@suse.de>
9154 PR rtl-optimization/56965
9155 * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
9156 * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
9158 * alias.c (true_dependence_1): Do not call
9159 nonoverlapping_component_refs_p.
9160 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
9161 nonoverlapping_component_refs_p.
9162 (indirect_refs_may_alias_p): Likewise.
9164 2014-04-15 Teresa Johnson <tejohnson@google.com>
9166 * cfg.c (dump_bb_info): Fix flags check.
9167 * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
9169 2014-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9171 PR rtl-optimization/60663
9172 * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
9175 2014-04-15 Richard Biener <rguenther@suse.de>
9177 * lto-streamer.h (LTO_major_version): Bump to 4.
9179 2014-04-15 Richard Biener <rguenther@suse.de>
9181 * common.opt (lto_partition_model): New enum.
9182 (flto-partition=): Merge separate options with a single with argument,
9183 add -flto-partition=one support.
9184 * flag-types.h (enum lto_partition_model): Declare.
9185 * opts.c (finish_options): Remove duplicate -flto-partition=
9187 * lto-wrapper.c (run_gcc): Adjust.
9189 2014-04-15 Richard Biener <rguenther@suse.de>
9191 * alias.c (ncr_compar): New function.
9192 (nonoverlapping_component_refs_p): Re-implement in O (n log n).
9194 2014-04-15 Richard Biener <rguenther@suse.de>
9196 * alias.c (record_component_aliases): Do not walk BINFOs.
9198 2014-04-15 Richard Biener <rguenther@suse.de>
9200 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
9201 Add struct function argument and adjust.
9202 (find_func_aliases_for_call): Likewise.
9203 (find_func_aliases): Likewise.
9204 (find_func_clobbers): Likewise.
9205 (intra_create_variable_infos): Likewise.
9206 (compute_points_to_sets): Likewise.
9207 (ipa_pta_execute): Adjust. Do not push/pop cfun.
9209 2014-04-15 Richard Biener <rguenther@suse.de>
9211 * tree.c (iterative_hash_expr): Use enum tree_code_class
9212 to store TREE_CODE_CLASS.
9213 (tree_block): Likewise.
9214 (tree_set_block): Likewise.
9215 * tree.h (fold_build_pointer_plus_loc): Use
9216 convert_to_ptrofftype_loc.
9218 2014-04-15 Jakub Jelinek <jakub@redhat.com>
9221 * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
9224 2014-04-15 Eric Botcazou <ebotcazou@adacore.com>
9226 * cfgloop.h (struct loop): Move force_vectorize down.
9227 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
9228 (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
9229 * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
9230 * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
9231 * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
9232 * tree-core.h (enum annot_expr_kind): Add new kind values.
9233 * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
9234 * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
9236 * tree.def (ANNOTATE_EXPR): Tweak comment.
9238 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
9240 * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
9243 2014-04-14 Paolo Carlini <paolo.carlini@oracle.com>
9245 * tree.h (TYPE_IDENTIFIER): Declare.
9246 * tree.c (subrange_type_for_debug_p): Use it.
9247 * godump.c (go_format_type): Likewise.
9248 * dwarf2out.c (is_cxx_auto, modified_type_die,
9249 gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
9250 * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
9252 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
9255 * varpool.c (varpool_remove_node): Do not alter decls when streaming.
9257 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
9259 * config/i386/i386.c (examine_argument): Return bool. Return true if
9260 parameter should be passed in memory.
9261 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
9262 (construct_container): Update calls to examine_argument.
9263 (function_arg_advance_64): Ditto.
9264 (return_in_memory_32): Merge with ix86_return_in_memory.
9265 (return_in_memory_64): Ditto.
9266 (return_in_memory_ms_64): Ditto.
9268 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
9270 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
9271 * coverage.c (coverage_compute_profile_id): Handle externally visible
9274 2014-04-14 Martin Jambor <mjambor@suse.cz>
9276 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
9277 DECL_DISREGARD_INLINE_LIMITS functions.
9279 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
9282 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
9284 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
9287 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
9288 optimize_insn_for_speed_p instead of
9289 optimize_function_for_speed_p.
9291 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
9293 * doc/invoke.texi (free): Document AArch64.
9295 2014-04-14 Richard Biener <rguenther@suse.de>
9297 PR tree-optimization/60042
9298 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
9299 (insert_into_preds_of_block): Do not prevent PHI insertion
9300 for REFERENCE exprs here ...
9301 (eliminate_dom_walker::before_dom_children): ... but prevent
9302 their use here under similar conditions when applied to the
9303 IL after PRE optimizations.
9305 2014-04-14 Richard Biener <rguenther@suse.de>
9307 * passes.def: Move early points-to after early SRA.
9309 2014-04-14 Richard Biener <rguenther@suse.de>
9311 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
9312 check for which sign-changes we allow when forwarding
9313 a converted value into a switch.
9315 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
9317 * stor-layout.c (place_field): Finalize non-constant offset for the
9320 2014-04-14 Richard Biener <rguenther@suse.de>
9322 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
9324 (expand_switch_using_bit_tests_p): Likewise.
9325 (process_switch): Compute and pass on speed_p based on the
9327 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
9328 optimize_bb_for_speed_p.
9330 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
9332 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
9333 * function.h (struct function): Rename has_force_vect_loops into
9334 has_force_vectorize_loops.
9335 * lto-streamer-in.c (input_cfg): Adjust for renaming.
9336 (input_struct_function_base): Likewise.
9337 * lto-streamer-out.c (output_cfg): Likewise.
9338 (output_struct_function_base): Likewise.
9339 * omp-low.c (expand_omp_simd): Likewise.
9340 * tree-cfg.c (move_sese_region_to_fn): Likewise.
9341 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
9342 (version_loop_for_if_conversion): Likewise.
9343 (tree_if_conversion): Likewise.
9344 (main_tree_if_conversion): Likewise.
9345 (gate_tree_if_conversion): Likewise.
9346 * tree-inline.c (copy_loops): Likewise.
9347 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
9348 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
9349 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
9350 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
9351 * tree-vectorizer.c (vectorize_loops): Likewise.
9352 * tree-vectorizer.h (unlimited_cost_model): Likewise.
9354 2014-04-14 Richard Biener <rguenther@suse.de>
9357 * lto-streamer-out.c (wrap_refs): New function.
9358 (lto_output): Wrap symbol references in global initializes in
9359 type-preserving MEM_REFs.
9361 2014-04-14 Christian Bruel <christian.bruel@st.com>
9363 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
9365 2014-04-14 Christian Bruel <christian.bruel@st.com>
9367 * config/sh/sh.md (setmemqi): New expand pattern.
9368 * config/sh/sh.h (CLEAR_RATIO): Define.
9369 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
9370 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
9372 2014-04-14 Richard Biener <rguenther@suse.de>
9375 * fold-const.c (negate_expr_p): Don't negate directional rounding
9377 (fold_negate_expr): Likewise.
9379 2014-04-14 Richard Biener <rguenther@suse.de>
9381 PR tree-optimization/59817
9382 PR tree-optimization/60453
9383 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
9384 recursion to catch all CHRECs in the scalar evolution and restrict
9385 the predicate for the remains appropriately.
9387 2014-04-12 Catherine Moore <clm@codesourcery.com>
9389 * config/mips/constraints.md: Add new register constraint "kb".
9390 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
9391 (*movhi_internal): Likewise.
9392 (*movqi_internal): Likewise.
9393 * config/mips/mips.h (M16_STORE_REGS): New register class.
9394 (REG_CLASS_NAMES): Add M16_STORE_REGS.
9395 (REG_CLASS_CONTENTS): Likewise.
9396 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
9398 2014-04-11 Tobias Burnus <burnus@net-b.de>
9401 * doc/invoke.texi (-Wformat-signedness): Document it.
9402 (Wformat=2): Mention that this enables -Wformat-signedness.
9404 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
9406 * common/config/epiphany/epiphany-common.c
9407 (epiphany_option_optimization_table): Enable section anchors by
9408 default at -O1 or higher.
9409 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
9410 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
9411 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
9412 carries no extra cost.
9413 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
9414 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
9415 * config/epiphany/predicates.md (memclob_operand): New predicate.
9416 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
9417 Use memclob_operand predicate and X constraint for operand 3.
9419 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
9421 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
9422 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
9425 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
9427 PR rtl-optimization/60651
9428 * mode-switching.c (optimize_mode_switching): Make sure to emit
9429 sets of a lower numbered entity before sets of a higher numbered
9430 entity to a mode of the same or lower priority.
9431 When creating a seginfo for a basic block that starts with a code
9432 label, move the insertion point past the code label.
9433 (new_seginfo): Document and enforce requirement that
9434 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
9435 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
9436 * doc/tm.texi: Regenerate.
9438 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
9441 * config/arc/arc.c (arc_save_restore): Fix assert typo.
9443 2013-04-11 Jakub Jelinek <jakub@redhat.com>
9445 * BASE-VER: Set to 4.10.0.
9447 2014-04-11 Tobias Burnus <burnus@net-b.de>
9450 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
9451 * doc/gcc.texi (Service): Update description in the @menu
9452 * doc/invoke.texi (Option Summary): Remove misplaced and
9455 2014-04-11 Steve Ellcey <sellcey@mips.com>
9456 Jakub Jelinek <jakub@redhat.com>
9459 * expr.c (convert_move): Use emit_store_flag_force instead of
9460 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
9463 2014-04-11 Richard Biener <rguenther@suse.de>
9466 * varasm.c (assemble_alias): Avoid endless error reporting
9467 recursion by setting TREE_ASM_WRITTEN.
9469 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9471 * config/s390/s390.md: Add a splitter for NOT rtx.
9473 2014-04-11 Jakub Jelinek <jakub@redhat.com>
9475 PR rtl-optimization/60663
9476 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
9478 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
9479 Jakub Jelinek <jakub@redhat.com>
9482 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
9483 flag from decl_node to node.
9485 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9488 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
9489 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
9490 ameliorating the cases where it can be.
9492 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
9495 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
9497 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
9498 (loadsync_<mode>): Change mode.
9499 (load_quadpti, store_quadpti): New.
9500 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
9501 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
9502 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
9504 2014-04-09 Cong Hou <congh@google.com>
9507 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
9510 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9512 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
9513 instead of vnor to exploit possible fusion opportunity in the
9515 (altivec_expand_vec_perm_const_le): Likewise.
9517 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
9519 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
9520 (loadsync_<mode>): Change mode.
9521 (load_quadpti, store_quadpti): New.
9522 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
9523 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
9525 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
9528 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
9529 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
9530 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
9532 2014-04-08 Richard Biener <rguenther@suse.de>
9535 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
9536 a 64bit widest int print double-int similar to on HWI64 hosts.
9538 2014-04-08 Richard Biener <rguenther@suse.de>
9540 PR tree-optimization/60785
9541 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
9542 default defs properly.
9544 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
9546 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
9547 (Weffc++): Likewise.
9549 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
9551 * ipa-devirt.c (maybe_record_node): When node is not recorded,
9552 set completep to false rather than true.
9554 2014-04-07 Douglas B Rupp <rupp@adacore.com>
9557 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
9558 ARM_TARGET2_DWARF_FORMAT.
9560 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
9563 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
9564 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
9567 2014-04-07 Richard Biener <rguenther@suse.de>
9569 PR tree-optimization/60766
9570 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
9571 (may_eliminate_iv): Convert cand_value_at result to desired type.
9573 2014-04-07 Jason Merrill <jason@redhat.com>
9576 * common.opt (-fno-gnu-unique): Add.
9577 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
9579 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9581 * haifa-sched.c: Fix outdated function reference and minor
9582 grammar errors in introductory comment.
9584 2014-04-07 Richard Biener <rguenther@suse.de>
9587 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
9589 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
9591 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
9594 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
9595 size accounting for thunks.
9596 (pa_asm_output_mi_thunk): Use final_start_function() and
9597 final_end_function() to output function start and end directives.
9599 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
9601 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
9602 device specific ISA/ feature information. Remove short_sp and
9603 errata_skip ds. Add avr_device_specific_features enum to have device
9605 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
9606 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
9607 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
9608 updated device specific info.
9609 * config/avr/avr-mcus.def: Merge device specific details to
9610 dev_attribute field.
9611 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
9613 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
9614 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
9615 assembler if RMW isa supported by current device.
9616 * config/avr/genmultilib.awk: Update as device info structure changed.
9617 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
9619 2014-04-04 Cong Hou <congh@google.com>
9621 PR tree-optimization/60656
9622 * tree-vect-stmts.c (supportable_widening_operation):
9623 Fix a bug that elements in a vector with vect_used_by_reduction
9624 property are incorrectly reordered when the operation on it is not
9625 consistant with the one in reduction operation.
9627 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
9629 PR rtl-optimization/60155
9630 * gcse.c (record_set_data): New function.
9631 (single_set_gcse): New function.
9632 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
9633 (hoist_code): Likewise.
9634 (get_pressure_class_and_nregs): Likewise.
9636 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
9638 * explow.c (probe_stack_range): Emit a final optimization blockage.
9640 2014-04-04 Anthony Green <green@moxielogic.com>
9642 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
9645 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
9648 * lto-cgraph.c (input_overwrite_node): Check that partitioning
9649 flags are set only during streaming.
9650 * ipa.c (process_references, walk_polymorphic_call_targets,
9651 symtab_remove_unreachable_nodes): Drop bodies of always inline
9652 after early inlining.
9653 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
9655 2014-04-04 Jakub Jelinek <jakub@redhat.com>
9656 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9659 * dwarf2out.c (const_ok_for_output_1): Reject expressions
9662 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9665 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
9667 (cortex_a53_fdivd): Likewise.
9669 2014-04-04 Martin Jambor <mjambor@suse.cz>
9672 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
9674 * cgraph.c (clone_of_p): Also return true if thunks match.
9675 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
9676 cgraph_function_or_thunk_node and an obsolete comment.
9677 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
9679 (build_function_decl_skip_args): Likewise.
9680 (set_new_clone_decl_and_node_flags): New function.
9681 (duplicate_thunk_for_node): Likewise.
9682 (redirect_edge_duplicating_thunks): Likewise.
9683 (cgraph_clone_node): New parameter args_to_skip, pass it to
9684 redirect_edge_duplicating_thunks which is called instead of
9685 cgraph_redirect_edge_callee.
9686 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
9687 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
9689 2014-04-04 Jeff Law <law@redhat.com>
9692 * config/arm/predicates.md (const_int_I_operand): New predicate.
9693 (const_int_M_operand): Similarly.
9694 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
9696 (insv_t2, extv_reg, extzv_t2): Likewise.
9697 (load_multiple_with_writeback): Similarly for const_int_I_operand.
9698 (pop_multiple_with_writeback_and_return): Likewise.
9699 (vfp_pop_multiple_with_writeback): Likewise
9701 2014-04-04 Richard Biener <rguenther@suse.de>
9704 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
9705 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
9707 * gimplify.h (gimple_add_tmp_var_fn): Declare.
9708 * gimplify.c (gimple_add_tmp_var_fn): New function.
9709 * gimple-expr.h (create_tmp_reg_fn): Declare.
9710 * gimple-expr.c (create_tmp_reg_fn): New function.
9711 * gimple-low.c (record_vars_into): Don't change cfun.
9712 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
9713 code generation without cfun.
9715 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
9718 * Makefile.in (install-driver): Fix shell scripting.
9720 2014-04-03 Cong Hou <congh@google.com>
9722 PR tree-optimization/60505
9723 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
9724 threshold of number of iterations below which no vectorization
9726 * tree-vect-loop.c (new_loop_vec_info):
9727 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
9728 * tree-vect-loop.c (vect_analyze_loop_operations):
9729 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
9730 * tree-vect-loop.c (vect_transform_loop):
9731 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
9732 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
9733 of iterations of the loop and see if we should build the epilogue.
9735 2014-04-03 Richard Biener <rguenther@suse.de>
9737 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
9738 (streamer_tree_cache_create): Adjust.
9739 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
9740 to allow optional nodes array.
9741 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
9742 (streamer_tree_cache_append): Likewise.
9743 (streamer_tree_cache_create): Create nodes array optionally
9744 as specified by parameter.
9745 * lto-streamer-out.c (create_output_block): Avoid maintaining
9746 the node array in the writer cache.
9747 (DFS_write_tree): Remove assertion.
9748 (produce_asm_for_decls): Free the out decl state hash table early.
9749 * lto-streamer-in.c (lto_data_in_create): Adjust for
9750 streamer_tree_cache_create prototype change.
9752 2014-04-03 Richard Biener <rguenther@suse.de>
9754 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
9755 set TREE_CHAIN to NULL_TREE.
9757 2014-04-03 Richard Biener <rguenther@suse.de>
9759 PR tree-optimization/60740
9760 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
9761 over all GIMPLE_COND operands.
9763 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
9765 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
9766 (Weffc++): Remove Scott's numbering, merge lists and reference
9769 2014-04-03 Nick Clifton <nickc@redhat.com>
9771 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
9774 2014-04-03 Martin Jambor <mjambor@suse.cz>
9776 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
9777 mention gcc_unreachable before failing.
9778 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
9781 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
9784 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
9785 inconsistent code and instead mark the context inconsistent.
9786 (possible_polymorphic_call_targets): For inconsistent contexts
9787 return empty complete list.
9789 2014-04-02 Anthony Green <green@moxielogic.com>
9791 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
9792 (extendqisi2, extendhisi2): Define.
9793 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
9794 (WCHAR_TYPE): Change to unsigned int.
9796 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9798 PR tree-optimization/60733
9799 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
9800 insertion point for PHI candidates to be the end of the feeding
9801 block for the PHI argument.
9803 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
9805 PR rtl-optimization/60650
9806 * lra-constraints.c (process_alt_operands): Decrease reject for
9807 earlyclobber matching.
9809 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9811 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
9813 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
9815 * config/spu/spu.c (pad_bb): Do not crash when the last
9816 insn is CODE_FOR_blockage.
9818 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
9820 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
9821 lies outside the target mode.
9823 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
9826 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
9827 software floating point or no floating point registers, do not
9828 allow any type in the FPRs. Eliminate a test for SPE SIMD types
9829 in GPRs that occurs after we tested for GPRs that would never be
9832 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
9833 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
9834 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
9835 specifically allow DDmode, since that does not use the SPE SIMD
9838 2014-04-02 Richard Biener <rguenther@suse.de>
9841 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
9842 MODE_INTs. Properly use negv_optab.
9843 (expand_abs): Likewise.
9845 2014-04-02 Richard Biener <rguenther@suse.de>
9848 * Makefile.in (install-driver): Guard extra installs with special
9851 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
9853 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
9856 2014-04-01 Richard Henderson <rth@redhat.com>
9859 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
9860 alternative enabled before register allocation.
9862 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
9864 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
9865 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
9867 (nios2_large_got_address): Remove unneeded 'sym' parameter.
9868 (nios2_got_address): Update nios2_large_got_address call site.
9869 (nios2_delegitimize_address): New function.
9870 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
9871 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
9872 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
9874 2014-04-01 Martin Husemann <martin@duskware.de>
9876 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
9879 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
9881 PR rtl-optimization/60604
9882 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
9883 check from register_operand.
9884 (register_operand): Redefine in terms of general_operand.
9885 (nonmemory_operand): Use register_operand for the non-constant cases.
9887 2014-04-01 Richard Biener <rguenther@suse.de>
9889 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
9891 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
9893 * doc/invoke.texi (mapp-regs): Clarify.
9895 2014-03-31 Ulrich Drepper <drepper@gmail.com>
9897 * config/i386/avx512fintrin.h (__v32hi): Define type.
9898 (__v64qi): Likewise.
9899 (_mm512_set1_epi8): Define.
9900 (_mm512_set1_epi16): Define.
9901 (_mm512_set4_epi32): Define.
9902 (_mm512_set4_epi64): Define.
9903 (_mm512_set4_pd): Define.
9904 (_mm512_set4_ps): Define.
9905 (_mm512_setr4_epi64): Define.
9906 (_mm512_setr4_epi32): Define.
9907 (_mm512_setr4_pd): Define.
9908 (_mm512_setr4_ps): Define.
9909 (_mm512_setzero_epi32): Define.
9911 2014-03-31 Martin Jambor <mjambor@suse.cz>
9914 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
9915 callsite_arguments_match_p. Updated all callers. Also check types of
9916 corresponding formal parameters and actual arguments.
9917 (not_all_callers_have_enough_arguments_p) Renamed to
9918 some_callers_have_mismatched_arguments_p.
9920 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
9922 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
9924 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
9927 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
9930 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
9932 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
9933 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
9935 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
9936 Use FMAMODE_NOVF512 mode iterator.
9937 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
9938 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
9939 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
9940 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
9942 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
9943 Use VF_128_256 mode iterator.
9944 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
9947 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
9949 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
9950 static chain if needed.
9952 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
9955 * lra-constraints.c (index_part_to_reg): New.
9956 (process_address): Use it.
9958 2014-03-27 Jeff Law <law@redhat.com>
9959 Jakub Jelinek <jakub@redhat.com>
9962 * expr.c (do_tablejump): Use simplify_gen_binary rather than
9963 gen_rtx_{PLUS,MULT} to build up the address expression.
9965 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
9966 creating non-canonical RTL.
9968 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
9971 * ipa-inline.c (want_inline_small_function_p): Short circuit large
9972 functions; reorganize to make cheap checks first.
9973 (inline_small_functions): Do not estimate growth when dumping;
9975 * ipa-inline.h (inline_summary): Add min_size.
9976 (growth_likely_positive): New function.
9977 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
9978 (set_cond_stmt_execution_predicate): Cleanup.
9979 (estimate_edge_size_and_time): Compute min_size.
9980 (estimate_calls_size_and_time): Likewise.
9981 (estimate_node_size_and_time): Likewise.
9982 (inline_update_overall_summary): Update min_size.
9983 (do_estimate_edge_time): Likewise.
9984 (do_estimate_edge_size): Update.
9985 (do_estimate_edge_hints): Update.
9986 (growth_likely_positive): New function.
9988 2014-03-28 Jakub Jelinek <jakub@redhat.com>
9991 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
9992 also if addr has VOIDmode.
9994 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9996 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
9997 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
9999 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
10000 instructions as well as AdvancedSIMD loads.
10002 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10004 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
10005 Use crypto_aese type.
10006 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
10007 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
10008 crypto_aese, crypto_aesmc. Move to types.md.
10009 * config/arm/types.md (crypto_aes): Split into crypto_aese,
10011 * config/arm/iterators.md (crypto_type): Likewise.
10013 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
10015 * cgraph.c: Include expr.h and tree-dfa.h.
10016 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
10019 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
10022 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
10023 regs from checking multi-reg pseudos.
10025 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10027 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
10029 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
10031 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
10032 if it would clobber the stack pointer, even temporarily.
10034 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
10036 * mode-switching.c: Make small adjustments to the top comment.
10038 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10040 * config/rs6000/constraints.md (wD constraint): New constraint to
10041 match the constant integer to get the top DImode/DFmode out of a
10042 vector in a VSX register.
10044 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
10045 match the constant integer to get the top DImode/DFmode out of a
10046 vector in a VSX register.
10048 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
10051 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10054 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
10055 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
10057 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
10058 Optimize vec_extract of 64-bit values, where the value being
10059 extracted is in the top word, where we can use scalar
10060 instructions. Add direct move and store support. Combine the big
10061 endian/little endian vector select load support into a single insn.
10062 (vsx_extract_<mode>_internal1): Likewise.
10063 (vsx_extract_<mode>_internal2): Likewise.
10064 (vsx_extract_<mode>_load): Likewise.
10065 (vsx_extract_<mode>_store): Likewise.
10066 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
10067 combined into vsx_extract_<mode>_load.
10068 (vsx_extract_<mode>_one_le): Likewise.
10070 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
10071 define the top 64-bit vector element.
10073 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
10076 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
10077 Document vec_vbpermq builtin.
10080 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
10081 enable use of xxsldwi and xxpermdi builtin functions.
10082 (vec_xxpermdi): Likewise.
10084 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
10085 Document use of vec_xxsldwi and vec_xxpermdi builtins.
10087 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
10089 PR rtl-optimization/60650
10090 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
10092 (find_spills_for): New.
10093 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
10094 Spill all pseudos on the second iteration.
10096 2014-03-27 Marek Polacek <polacek@redhat.com>
10099 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
10102 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
10104 * config/s390/s390.c (s390_can_use_return_insn): Check for
10105 call-saved FPRs on 31 bit.
10107 2014-03-27 Jakub Jelinek <jakub@redhat.com>
10109 PR middle-end/60682
10110 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
10111 if they need regimplification, just drop them instead of
10112 calling gimple_regimplify_operands on them.
10114 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
10117 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
10118 (aarch64_frame_pointer_required): Adjust logic.
10119 (aarch64_can_eliminate): Adjust logic.
10120 (aarch64_override_options_after_change): Adjust logic.
10122 2014-03-27 Dehao Chen <dehao@google.com>
10124 * ipa-inline.c (early_inliner): Update node's inline info.
10126 2014-03-26 Dehao Chen <dehao@google.com>
10128 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
10129 compiler inserted conditional jumps for NAN float check.
10131 2014-03-26 Jakub Jelinek <jakub@redhat.com>
10133 * ubsan.h (ubsan_create_data): Change second argument's type
10134 to const location_t *.
10135 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
10137 (ubsan_create_data): Change second argument to const location_t *PLOC.
10138 Create Loc field whenever PLOC is non-NULL.
10139 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
10140 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
10144 * real.c (real_to_integer2): Change type of low to UHWI.
10146 2014-03-26 Tobias Burnus <burnus@net-b.de>
10148 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
10149 (CILK_SELF_SPECS): New define.
10150 (driver_self_specs): Use it.
10152 2014-03-26 Richard Biener <rguenther@suse.de>
10154 * tree-pretty-print.c (percent_K_format): Implement special
10155 case for LTO and its stripped down BLOCK tree.
10157 2014-03-26 Jakub Jelinek <jakub@redhat.com>
10160 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
10162 * tree-vrp.c (simplify_internal_call_using_ranges): If only
10163 one range is range_int_cst_p, but not both, at least optimize
10164 addition/subtraction of 0 and multiplication by 0 or 1.
10165 * gimple-fold.c (gimple_fold_call): Fold
10166 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
10167 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
10168 INTEGER_CSTs, try to fold at least x * 0 and y - y.
10170 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
10172 PR rtl-optimization/60452
10173 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
10174 <case REG>: Return 1 for invalid offsets from the frame pointer.
10176 2014-03-26 Marek Polacek <polacek@redhat.com>
10179 * doc/extend.texi (C Extensions): Mention variable-length arrays in
10182 2014-03-26 Marek Polacek <polacek@redhat.com>
10185 * doc/extend.texi (Designated Inits): Describe what happens to omitted
10188 2014-03-26 Marek Polacek <polacek@redhat.com>
10191 * ira-color.c (update_conflict_hard_regno_costs): Perform the
10192 multiplication in unsigned type.
10194 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
10196 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
10198 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
10200 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
10202 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
10205 * cif-code.def (UNREACHABLE) New code.
10206 * ipa-inline.c (inline_small_functions): Skip edges to
10207 __builtlin_unreachable.
10208 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
10209 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
10210 predicate to __bulitin_unreachable.
10211 (set_cond_stmt_execution_predicate): Fix issue when
10212 invert_tree_comparison returns ERROR_MARK.
10213 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
10214 propagate to inline clones.
10215 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
10217 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
10218 * cgraphclones.c (cgraph_clone_node): If call destination is already
10219 ureachable, do not redirect it back.
10220 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
10223 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
10225 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
10226 Do not modify inline clones.
10228 2014-03-25 Jakub Jelinek <jakub@redhat.com>
10230 * config/i386/i386.md (general_sext_operand): New mode attr.
10231 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
10232 don't generate (sign_extend (const_int)).
10233 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
10234 operands[2]. Use We constraint instead of <i> and
10235 <general_sext_operand> predicate instead of <general_operand>.
10236 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
10237 * config/i386/constraints.md (We): New constraint.
10238 * config/i386/predicates.md (x86_64_sext_operand,
10239 sext_operand): New predicates.
10241 2014-03-25 Martin Jambor <mjambor@suse.cz>
10244 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
10245 inconsistent devirtualizations to __builtin_unreachable.
10247 2014-03-25 Marek Polacek <polacek@redhat.com>
10250 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
10252 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
10254 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
10255 order of elements for big-endian.
10257 2014-03-25 Richard Biener <rguenther@suse.de>
10259 PR middle-end/60635
10260 * gimplify-me.c (gimple_regimplify_operands): Update the
10263 2014-03-25 Martin Jambor <mjambor@suse.cz>
10266 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
10267 (lto_output_varpool_node): Likewise.
10268 (input_overwrite_node): Likewise.
10269 (input_varpool_node): Likewise.
10271 2014-03-25 Richard Biener <rguenther@suse.de>
10273 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
10274 (run_gcc): Likewise.
10276 2014-03-25 Jakub Jelinek <jakub@redhat.com>
10278 * combine.c (simplify_compare_const): Add MODE argument.
10279 Handle mode_width 0 as very large mode_width.
10280 (try_combine, simplify_comparison): Adjust callers.
10282 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
10283 type to avoid signed integer overflow.
10284 * explow.c (plus_constant): Likewise.
10286 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
10288 * doc/generic.texi: Correct typos.
10290 2014-03-24 Tobias Burnus <burnus@net-b.de>
10292 * doc/invoke.texi (-flto): Expand section about
10293 using static libraries with LTO.
10295 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
10297 PR rtl-optimization/60501
10298 * optabs.def (addptr3_optab): New optab.
10299 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
10300 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
10301 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
10303 * lra.c (emit_add3_insn): Use the addptr pattern if available.
10305 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
10307 2014-03-24 Ulrich Drepper <drepper@gmail.com>
10309 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
10312 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
10313 (_mm256_undefined_ps): Define.
10314 (_mm256_undefined_pd): Define.
10315 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
10316 (_mm_undefined_pd): Define.
10317 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
10318 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
10319 (_mm512_undefined_ps): Define.
10320 (_mm512_undefined_pd): Define.
10321 Use _mm*_undefined_*.
10322 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
10324 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
10326 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
10327 (lshr_simd): DI mode added.
10328 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
10329 (aarch64_ushr_simddi): Likewise.
10330 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
10331 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
10332 (vshrd_n_u64): Likewise.
10334 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10336 * Makefile.in (s-macro_list): Depend on cc1.
10338 2014-03-23 Teresa Johnson <tejohnson@google.com>
10340 * ipa-utils.c (ipa_print_order): Use specified dump file.
10342 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
10344 PR rtl-optimization/60601
10345 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
10347 * gcc.c (eval_spec_function): Initialize save_growing_value.
10349 2014-03-22 Jakub Jelinek <jakub@redhat.com>
10352 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
10353 code == MINUS_EXPR, never swap op0 with op1.
10355 * toplev.c (init_local_tick): Avoid signed integer multiplication
10357 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
10358 shift by first operand's bitsize.
10360 2014-03-21 Jakub Jelinek <jakub@redhat.com>
10363 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
10364 redefine to 1 or 0.
10365 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
10366 TARGET_ISA_64BIT_P(x).
10368 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10370 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
10371 pattern for vector nor instead of subtract from splat(-1).
10372 (altivec_expand_vec_perm_const_le): Likewise.
10374 2014-03-21 Richard Henderson <rth@twiddle.net>
10377 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
10378 related insns after epilogue_completed.
10380 2014-03-21 Martin Jambor <mjambor@suse.cz>
10383 * cgraph.h (symtab_node): New flag body_removed.
10384 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
10385 when removing bodies.
10386 * symtab.c (dump_symtab_base): Dump body_removed flag.
10387 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
10388 had their bodies removed.
10390 2014-03-21 Martin Jambor <mjambor@suse.cz>
10393 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
10396 2014-03-21 Richard Biener <rguenther@suse.de>
10398 PR tree-optimization/60577
10399 * tree-core.h (struct tree_base): Document nothrow_flag use
10400 in DECL_NONALIASED.
10401 * tree.h (DECL_NONALIASED): New.
10402 (may_be_aliased): Adjust.
10403 * coverage.c (build_var): Set DECL_NONALIASED.
10405 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
10407 * expr.c (expand_expr_real_1): Remove outdated comment.
10409 2014-03-20 Jakub Jelinek <jakub@redhat.com>
10411 PR middle-end/60597
10412 * ira.c (adjust_cleared_regs): Call copy_rtx on
10413 *reg_equiv[REGNO (loc)].src_p before passing it to
10414 simplify_replace_fn_rtx.
10417 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
10418 into CONST, put pic register as first operand of PLUS. Use
10419 gen_const_mem for both 32-bit and 64-bit PIC got loads.
10421 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10423 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
10425 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
10427 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
10428 around for store forwarding issue in the FPU on the UT699.
10429 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
10430 loads and operations if -mfix-ut699 is specified.
10431 (divtf3_hq): Tweak attribute.
10432 (sqrttf2_hq): Likewise.
10434 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
10436 * calls.c (store_one_arg): Remove incorrect const qualification on the
10437 type of the temporary.
10438 * cfgexpand.c (expand_return): Likewise.
10439 * expr.c (expand_constructor): Likewise.
10440 (expand_expr_real_1): Likewise.
10442 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
10444 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
10447 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
10450 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
10452 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
10454 * config/arm/aarch-common-protos.h
10455 (alu_cost_table): Fix spelling of "extend".
10456 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
10458 2014-03-19 Richard Biener <rguenther@suse.de>
10460 PR middle-end/60553
10461 * tree-core.h (tree_type_common): Re-order pointer members
10462 to reduce recursion depth during GC walks.
10464 2014-03-19 Marek Polacek <polacek@redhat.com>
10467 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
10468 before accessing it.
10470 2014-03-19 Richard Biener <rguenther@suse.de>
10473 * lto-streamer-in.c (input_function): In WPA stage do not drop
10476 2014-03-19 Jakub Jelinek <jakub@redhat.com>
10478 PR tree-optimization/60559
10479 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
10480 with build_zero_cst assignment.
10482 2014-03-18 Kai Tietz <ktietz@redhat.com>
10484 PR rtl-optimization/56356
10485 * sdbout.c (sdbout_parms): Verify that parms'
10486 incoming argument is valid.
10487 (sdbout_reg_parms): Likewise.
10489 2014-03-18 Richard Henderson <rth@redhat.com>
10492 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
10493 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
10494 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
10496 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
10498 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
10499 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
10500 Italicize plugin event names in description. Explain that
10501 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
10502 Remind that no GCC functions should be called after PLUGIN_FINISH.
10503 Explain what pragmas with expansion are.
10505 2014-03-18 Martin Liska <mliska@suse.cz>
10507 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
10508 gimple call statement is update.
10509 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
10510 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
10512 2014-03-18 Jakub Jelinek <jakub@redhat.com>
10515 * ubsan.c (ubsan_instrument_unreachable): Call
10516 initialize_sanitizer_builtins.
10517 (ubsan_pass): Likewise.
10520 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
10521 varpool_finalize_decl instead of rest_of_decl_compilation.
10523 2014-03-18 Richard Biener <rguenther@suse.de>
10525 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
10526 by using bitmap_and_compl instead of bitmap_and_compl_into.
10527 (df_rd_transfer_function): Likewise.
10529 2014-03-18 Richard Biener <rguenther@suse.de>
10531 * doc/lto.texi (fresolution): Fix typo.
10533 2014-03-18 Richard Biener <rguenther@suse.de>
10535 * doc/invoke.texi (flto): Update for changes in 4.9.
10537 2014-03-18 Richard Biener <rguenther@suse.de>
10539 * doc/loop.texi: Remove section on the removed lambda framework.
10540 Update loop docs with recent changes in preserving loop structure.
10542 2014-03-18 Richard Biener <rguenther@suse.de>
10544 * doc/lto.texi (-fresolution): Document.
10546 2014-03-18 Richard Biener <rguenther@suse.de>
10548 * doc/contrib.texi: Adjust my name.
10550 2014-03-18 Jakub Jelinek <jakub@redhat.com>
10553 * internal-fn.c: Include diagnostic-core.h.
10554 (expand_BUILTIN_EXPECT): New function.
10555 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
10556 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
10557 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
10558 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
10559 IFN_BUILTIN_EXPECT.
10560 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
10561 Revert 3 argument __builtin_expect code.
10562 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
10563 * gimple-fold.c (gimple_fold_call): Likewise.
10564 * tree.h (fold_builtin_expect): New prototype.
10565 * builtins.c (build_builtin_expect_predicate): Add predictor
10566 argument, if non-NULL, create 3 argument __builtin_expect.
10567 (fold_builtin_expect): No longer static. Add ARG2 argument,
10568 pass it through to build_builtin_expect_predicate.
10569 (fold_builtin_2): Adjust caller.
10570 (fold_builtin_3): Handle BUILT_IN_EXPECT.
10571 * internal-fn.def (BUILTIN_EXPECT): New.
10573 2014-03-18 Tobias Burnus <burnus@net-b.de>
10576 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
10577 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
10578 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
10580 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
10583 * predict.c (combine_predictions_for_bb): Fix up formatting.
10584 (expr_expected_value_1, expr_expected_value): Add predictor argument,
10585 fill what it points to if non-NULL.
10586 (tree_predict_by_opcode): Adjust caller, use the predictor.
10587 * predict.def (PRED_COMPARE_AND_SWAP): Add.
10589 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
10591 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
10592 proper constant for the store mode.
10594 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
10596 * symtab.c (change_decl_assembler_name): Fix transparent alias
10597 chain construction.
10599 2014-03-16 Renlin Li <Renlin.Li@arm.com>
10601 * config/aarch64/aarch64.c: Correct the comments about the
10602 aarch64 stack layout.
10604 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
10606 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
10607 check for GF_OMP_FOR_KIND_FOR.
10609 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
10611 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
10612 ymm and zmm register names.
10614 2014-03-17 Jakub Jelinek <jakub@redhat.com>
10617 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
10618 note creation for the 2010-08-31 changes.
10620 2014-03-17 Marek Polacek <polacek@redhat.com>
10622 PR middle-end/60534
10623 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
10624 as -fno-tree-loop-vectorize.
10625 (expand_omp_simd): Likewise.
10627 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
10629 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
10630 (eligible_for_call_delay): New prototype.
10631 * config/sparc/sparc.c (tls_call_delay): Rename into...
10632 (eligible_for_call_delay): ...this. Return false if the instruction
10633 cannot be put in the delay slot of a branch.
10634 (eligible_for_restore_insn): Simplify.
10635 (eligible_for_return_delay): Return false if the instruction cannot be
10636 put in the delay slot of a branch and simplify.
10637 (eligible_for_sibcall_delay): Return false if the instruction cannot be
10638 put in the delay slot of a branch.
10639 * config/sparc/sparc.md (fix_ut699): New attribute.
10640 (tls_call_delay): Delete.
10641 (in_call_delay): Reimplement.
10642 (eligible_for_sibcall_delay): Rename into...
10643 (in_sibcall_delay): ...this.
10644 (eligible_for_return_delay): Rename into...
10645 (in_return_delay): ...this.
10646 (in_branch_delay): Reimplement.
10647 (in_uncond_branch_delay): Delete.
10648 (in_annul_branch_delay): Delete.
10650 2014-03-14 Richard Henderson <rth@redhat.com>
10653 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
10654 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
10655 (*floathi<X87MODEF>2_i387_with_temp): Remove.
10656 (floathi splitters): Remove.
10657 (float<SWI48x>xf2): New pattern.
10658 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
10659 code that tried to handle DImode for 32-bit, but which was excluded
10660 by the pattern's condition. Drop allocation of stack temporary.
10661 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
10662 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
10663 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
10664 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
10665 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
10666 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
10667 (*float<SWI48><MODEF>2_sse_interunit): Remove.
10668 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
10669 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
10670 (*float<SWI48x><X87MODEF>2_i387): Remove.
10671 (all float _with_temp splitters): Remove.
10672 (*float<SWI48x><MODEF>2_i387): New pattern.
10673 (*float<SWI48><MODEF>2_sse): New pattern.
10674 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
10675 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
10677 2014-03-14 Jakub Jelinek <jakub@redhat.com>
10678 Marek Polacek <polacek@redhat.com>
10680 PR middle-end/60484
10681 * common.opt (dump_base_name_prefixed): New Variable.
10682 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
10683 if x_dump_base_name_prefixed is already set, set it at the end.
10685 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
10687 PR rtl-optimization/60508
10688 * lra-constraints.c (get_reload_reg): Add new parameter
10690 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
10691 Pass the new parameter values.
10693 2014-03-14 Richard Biener <rguenther@suse.de>
10695 * common.opt: Revert unintented changes from r205065.
10696 * opts.c: Likewise.
10698 2014-03-14 Richard Biener <rguenther@suse.de>
10700 PR middle-end/60518
10701 * cfghooks.c (split_block): Properly adjust all loops the
10702 block was a latch of.
10704 2014-03-14 Martin Jambor <mjambor@suse.cz>
10707 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
10710 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
10713 * config/avr/avr.c (avr_set_current_function): Pass function name
10714 through default_strip_name_encoding before sanity checking instead
10715 of skipping the first char of the assembler name.
10717 2014-03-13 Richard Henderson <rth@redhat.com>
10720 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
10721 (ix86_force_to_memory, ix86_free_from_memory): Remove.
10722 * config/i386/i386-protos.h: Likewise.
10723 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
10724 in the expander instead of a splitter.
10725 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
10726 any possibility of requiring a memory.
10727 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
10728 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
10729 (fp branch splitters): Update for ix86_split_fp_branch.
10730 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
10731 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
10732 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
10733 (*fop_<MODEF>_2_i387): Remove f/r alternative.
10734 (*fop_<MODEF>_3_i387): Likewise.
10735 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
10736 (splitters for the fop_* register patterns): Remove.
10737 (fscalexf4_i387): Rename from *fscalexf4_i387.
10738 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
10740 2014-03-13 Jakub Jelinek <jakub@redhat.com>
10742 PR tree-optimization/59779
10743 * tree-dfa.c (get_ref_base_and_extent): Use double_int
10744 type for bitsize and maxsize instead of HOST_WIDE_INT.
10746 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
10748 PR rtl-optimization/57320
10749 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
10750 the CFG after thread_prologue_and_epilogue_insns.
10752 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
10754 PR rtl-optimization/57189
10755 * lra-constraints.c (process_alt_operands): Disfavor spilling
10758 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
10760 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
10762 2014-03-13 Jakub Jelinek <jakub@redhat.com>
10764 PR tree-optimization/59025
10765 PR middle-end/60418
10766 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
10767 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
10769 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
10772 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
10773 calls of avr_out_plus_1.
10775 2014-03-13 Bin Cheng <bin.cheng@arm.com>
10777 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
10778 BB's single pred and update the father loop's latch info later.
10780 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10782 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
10786 (VEC_base): Likewise.
10787 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
10788 registers, we need to swap double words in little endian mode.
10790 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
10791 to be a container mode for 128-bit integer operations added in ISA
10792 2.07. Unlike TImode and PTImode, the preferred register set is
10793 the Altivec/VMX registers for the 128-bit operations.
10795 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
10797 (rs6000_split_128bit_ok_p): Likewise.
10799 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
10800 macros for creating ISA 2.07 normal and overloaded builtin
10801 functions with 3 arguments.
10802 (BU_P8V_OVERLOAD_3): Likewise.
10803 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
10804 for use as overloaded functions.
10805 (VPERM_1TI_UNS): Likewise.
10806 (VSEL_1TI): Likewise.
10807 (VSEL_1TI_UNS): Likewise.
10808 (ST_INTERNAL_1ti): Likewise.
10809 (LD_INTERNAL_1ti): Likewise.
10810 (XXSEL_1TI): Likewise.
10811 (XXSEL_1TI_UNS): Likewise.
10812 (VPERM_1TI): Likewise.
10813 (VPERM_1TI_UNS): Likewise.
10814 (XXPERMDI_1TI): Likewise.
10815 (SET_1TI): Likewise.
10816 (LXVD2X_V1TI): Likewise.
10817 (STXVD2X_V1TI): Likewise.
10818 (VEC_INIT_V1TI): Likewise.
10819 (VEC_SET_V1TI): Likewise.
10820 (VEC_EXT_V1TI): Likewise.
10821 (EQV_V1TI): Likewise.
10822 (NAND_V1TI): Likewise.
10823 (ORC_V1TI): Likewise.
10824 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
10825 added in ISA 2.07. Add both normal 'altivec' builtins, and the
10826 overloaded builtin.
10827 (VADDUQM): Likewise.
10828 (VSUBCUQ): Likewise.
10829 (VADDEUQM): Likewise.
10830 (VADDECUQ): Likewise.
10831 (VSUBEUQM): Likewise.
10832 (VSUBECUQ): Likewise.
10834 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
10835 __int128_t and __uint128_t types.
10836 (__uint128_type): Likewise.
10837 (altivec_categorize_keyword): Add support for vector __int128_t,
10838 vector __uint128_t, vector __int128, and vector unsigned __int128
10839 as a container type for TImode operations that need to be done in
10840 VSX/Altivec registers.
10841 (rs6000_macro_to_expand): Likewise.
10842 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
10843 to support 128-bit integer instructions vaddcuq, vadduqm,
10844 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
10845 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
10847 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
10848 for V1TImode, and set up preferences to use VSX/Altivec registers.
10849 Setup VSX reload handlers.
10850 (rs6000_debug_reg_global): Likewise.
10851 (rs6000_init_hard_regno_mode_ok): Likewise.
10852 (rs6000_preferred_simd_mode): Likewise.
10853 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
10854 (easy_altivec_constant): Likewise.
10855 (output_vec_const_move): Likewise.
10856 (rs6000_expand_vector_set): Convert V1TImode set and extract to
10858 (rs6000_expand_vector_extract): Likewise.
10859 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
10861 (rs6000_const_vec): Add support for V1TImode.
10862 (rs6000_emit_le_vsx_load): Swap double words when loading or
10863 storing TImode/V1TImode.
10864 (rs6000_emit_le_vsx_store): Likewise.
10865 (rs6000_emit_le_vsx_move): Likewise.
10866 (rs6000_emit_move): Add support for V1TImode.
10867 (altivec_expand_ld_builtin): Likewise.
10868 (altivec_expand_st_builtin): Likewise.
10869 (altivec_expand_vec_init_builtin): Likewise.
10870 (altivec_expand_builtin): Likewise.
10871 (rs6000_init_builtins): Add support for V1TImode type. Add
10872 support for ISA 2.07 128-bit integer builtins. Define type names
10873 for the VSX/Altivec vector types.
10874 (altivec_init_builtins): Add support for overloaded vector
10875 functions with V1TImode type.
10876 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
10877 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
10879 (rs6000_split_128bit_ok_p): Likewise.
10880 (rs6000_handle_altivec_attribute): Create V1TImode from vector
10881 __int128_t and vector __uint128_t.
10883 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
10884 and mode attributes.
10886 (VSX_M2): Likewise.
10891 (VS_scalar): Likewise.
10892 (VS_double): Likewise.
10893 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
10895 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
10896 we support the ISA 2.07 128-bit integer arithmetic instructions.
10897 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
10898 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
10899 and TImode types for use with the builtin functions.
10900 (V1TI_type_node): Likewise.
10901 (unsigned_V1TI_type_node): Likewise.
10902 (intTI_type_internal_node): Likewise.
10903 (uintTI_type_internal_node): Likewise.
10905 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
10906 128-bit builtin functions.
10907 (UNSPEC_VADDEUQM): Likewise.
10908 (UNSPEC_VADDECUQ): Likewise.
10909 (UNSPEC_VSUBCUQ): Likewise.
10910 (UNSPEC_VSUBEUQM): Likewise.
10911 (UNSPEC_VSUBECUQ): Likewise.
10912 (VM): Add V1TImode to vector mode iterators.
10914 (VI_unit): Likewise.
10915 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
10916 (altivec_vaddcuq): Likewise.
10917 (altivec_vsubuqm): Likewise.
10918 (altivec_vsubcuq): Likewise.
10919 (altivec_vaddeuqm): Likewise.
10920 (altivec_vaddecuq): Likewise.
10921 (altivec_vsubeuqm): Likewise.
10922 (altivec_vsubecuq): Likewise.
10924 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
10926 (BOOL_128): Likewise.
10927 (BOOL_REGS_OUTPUT): Likewise.
10928 (BOOL_REGS_OP1): Likewise.
10929 (BOOL_REGS_OP2): Likewise.
10930 (BOOL_REGS_UNARY): Likewise.
10931 (BOOL_REGS_AND_CR0): Likewise.
10933 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
10934 128-bit integer builtin support.
10935 (vec_vadduqm): Likewise.
10936 (vec_vaddecuq): Likewise.
10937 (vec_vaddeuqm): Likewise.
10938 (vec_vsubecuq): Likewise.
10939 (vec_vsubeuqm): Likewise.
10940 (vec_vsubcuq): Likewise.
10941 (vec_vsubuqm): Likewise.
10943 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
10944 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
10945 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
10946 128-bit integer add/subtract to ISA 2.07.
10948 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
10950 * config/arc/arc.c (arc_predicate_delay_insns):
10951 Fix third argument passed to conditionalize_nonjump.
10953 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
10955 * config/aarch64/aarch64-builtins.c
10956 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
10957 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
10958 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
10959 instead of __builtin_lfloor.
10960 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
10962 2014-03-12 Jakub Jelinek <jakub@redhat.com>
10964 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
10965 (tree_ssa_ifcombine_bb_1): New function.
10966 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
10967 is an empty forwarder block to then_bb or vice versa and then_bb
10968 and else_bb are effectively swapped.
10970 2014-03-12 Christian Bruel <christian.bruel@st.com>
10973 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
10974 REG_CFA_DEF_CFA note.
10975 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
10976 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
10978 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
10980 PR tree-optimization/60454
10981 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
10983 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10985 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
10986 Do not define target_cpu_default2 to generic.
10987 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
10988 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
10989 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
10991 2014-03-12 Jakub Jelinek <jakub@redhat.com>
10992 Marc Glisse <marc.glisse@inria.fr>
10994 PR tree-optimization/60502
10995 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
10996 instead of build_low_bits_mask.
10998 2014-03-12 Jakub Jelinek <jakub@redhat.com>
11000 PR middle-end/60482
11001 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
11002 if there are multiple uses, but op doesn't live on E edge.
11003 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
11004 clobber stmts before __builtin_unreachable.
11006 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
11008 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
11009 hard_frame_pointer_rtx.
11010 * cse.c (cse_insn): Remove volatile check.
11011 * cselib.c (cselib_process_insn): Likewise.
11012 * dse.c (scan_insn): Likewise.
11014 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
11016 * config/arc/arc.c (conditionalize_nonjump): New function,
11018 (arc_ifcvt): ... this.
11019 (arc_predicate_delay_insns): Use it.
11021 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
11023 * config/arc/predicates.md (extend_operand): During/after reload,
11024 allow const_int_operand.
11025 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
11026 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
11027 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
11029 (umulsi3_highpart_i): Likewise.
11031 2014-03-11 Richard Biener <rguenther@suse.de>
11033 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
11034 Add asserts to guard possible wrong-code bugs.
11036 2014-03-11 Richard Biener <rguenther@suse.de>
11038 PR tree-optimization/60429
11039 PR tree-optimization/60485
11040 * tree-ssa-structalias.c (set_union_with_increment): Properly
11041 take into account all fields that overlap the shifted vars.
11042 (do_sd_constraint): Likewise.
11043 (do_ds_constraint): Likewise.
11044 (get_constraint_for_ptr_offset): Likewise.
11046 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
11048 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
11049 (nios2_compute_frame_layout):
11050 Add calculation of cfun->machine->fp_save_offset.
11051 (nios2_expand_prologue): Correct setting of frame pointer register
11053 (nios2_expand_epilogue): Update recovery of stack pointer from
11054 frame pointer accordingly.
11055 (nios2_initial_elimination_offset): Update calculation of offset
11056 for eliminating to HARD_FRAME_POINTER_REGNUM.
11058 2014-03-10 Jakub Jelinek <jakub@redhat.com>
11061 * ipa.c (symtab_remove_unreachable_nodes): Don't call
11062 cgraph_get_create_node on VAR_DECLs.
11064 2014-03-10 Richard Biener <rguenther@suse.de>
11066 PR middle-end/60474
11067 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
11069 2014-03-08 Douglas B Rupp <rupp@gnat.com>
11071 * config/vms/vms.opt (vms_float_format): New variable.
11073 2014-03-08 Tobias Burnus <burnus@net-b.de>
11075 * doc/invoke.texi (-fcilkplus): Update implementation status.
11077 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
11078 Richard Biener <rguenther@suse.de>
11080 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
11081 consistently accross all TUs.
11082 (run_gcc): Enable -fshort-double automatically at link at link-time
11083 and disallow override.
11085 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
11088 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
11089 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
11090 if they can't be used.
11092 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11094 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
11095 for Solaris 11/x86 ld.
11096 * configure: Regenerate.
11098 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11100 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
11101 (LIB_TLS_SPEC): Save as ld_tls_libs.
11102 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
11103 (HAVE_AS_IX86_TLSLDM): New test.
11104 * configure, config.in: Regenerate.
11105 * config/i386/i386.c (legitimize_tls_address): Fall back to
11106 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
11107 cannot support TLS_MODEL_LOCAL_DYNAMIC.
11108 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
11109 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
11111 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
11113 * common.opt (fira-loop-pressure): Mark as optimization.
11115 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
11117 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
11118 an OpenMP mappable type.
11120 2014-03-06 Matthias Klose <doko@ubuntu.com>
11122 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
11123 MULTILIB_OSDIRNAMES is not defined.
11125 2014-03-06 Jakub Jelinek <jakub@redhat.com>
11126 Meador Inge <meadori@codesourcery.com>
11129 * config/arm/arm.c (arm_tls_symbol_p): Remove.
11130 (arm_legitimize_address): Call legitimize_tls_address for any
11131 arm_tls_referenced_p expression, handle constant addend. Call it
11132 before testing for !TARGET_ARM.
11133 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
11135 2014-03-06 Richard Biener <rguenther@suse.de>
11137 PR middle-end/60445
11141 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
11143 * tree-streamer.c (record_common_node): Assert we don't record
11144 nodes with type double.
11145 (preload_common_node): Skip type double, complex double and double
11146 pointer since it is now frontend dependent due to fshort-double option.
11148 2014-03-06 Richard Biener <rguenther@suse.de>
11150 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
11151 or -fno-lto is specified and the linker has full plugin support.
11152 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
11153 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
11154 * lto-wrapper.c (merge_and_complain): Merge compile-time
11155 optimization levels.
11156 (run_gcc): And pass it through to the link options.
11158 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
11162 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
11164 * cselib.c (remove_useless_values): Skip to avoid quadratic
11165 behavior if the condition moved from...
11166 (cselib_process_insn): ... here holds.
11168 2014-03-05 Jakub Jelinek <jakub@redhat.com>
11171 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
11172 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
11175 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
11176 (TM_H): Add x86-tune.def.
11178 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11180 * config/aarch64/aarch64.c (generic_tunings):
11181 Use cortexa57_extra_costs.
11183 2014-03-05 Jakub Jelinek <jakub@redhat.com>
11186 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
11187 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
11188 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
11191 2014-03-04 Heiher <r@hev.cc>
11193 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
11194 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
11196 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
11198 * config/i386/predicates.md (const2356_operand): Change to ...
11199 (const2367_operand): ... this.
11200 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
11202 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
11203 (*avx512pf_scatterpf<mode>sf): Ditto.
11204 (avx512pf_scatterpf<mode>df): Ditto.
11205 (*avx512pf_scatterpf<mode>df_mask): Ditto.
11206 (*avx512pf_scatterpf<mode>df): Ditto.
11207 * config/i386/i386.c (ix86_expand_builtin): Update
11208 incorrect hint operand error message.
11210 2014-03-04 Richard Biener <rguenther@suse.de>
11212 * lto-section-in.c (lto_get_section_data): Fix const cast.
11214 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
11216 * tree-streamer.c (record_common_node): Assert we don't record
11217 nodes with type double.
11218 (preload_common_node): Skip type double, complex double and double
11219 pointer since it is now frontend dependent due to fshort-double option.
11221 2014-03-04 Richard Biener <rguenther@suse.de>
11224 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
11225 (lto_input_toplevel_asms): Likewise.
11226 * lto-section-in.c (lto_get_section_data): Instead do it here
11229 2014-03-04 Richard Biener <rguenther@suse.de>
11231 PR tree-optimization/60382
11232 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
11233 dead PHIs a reduction.
11235 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
11237 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
11239 (_mm_prefetch): Move out of GCC target("sse") pragma.
11240 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
11241 GCC target("prfchw") pragma.
11242 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
11244 * config/i386/i386.c (ix86_option_override_internal): Enable
11245 -mprfchw with -mprefetchwt1.
11247 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
11249 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
11252 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
11254 * opts.h (CL_PCH_IGNORE): Define.
11255 * targhooks.c (option_affects_pch_p):
11256 Return false for options that have CL_PCH_IGNORE set.
11257 * opt-functions.awk: Process PchIgnore.
11258 * doc/options.texi: Document PchIgnore.
11260 * config/arc/arc.opt (misize): Add PchIgnore property.
11262 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11264 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
11265 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
11266 constraint on constants to permit them being loaded into
11267 GENERAL_REGS or BASE_REGS.
11269 2014-03-03 Nick Clifton <nickc@redhat.com>
11271 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
11272 anti-cacnonical alternatives.
11273 (negandhi3_real): New pattern.
11274 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
11276 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
11278 * config/avr/avr-mcus.def: Remove atxmega16x1.
11279 * config/avr/avr-tables.opt: Regenerate.
11280 * config/avr/t-multilib: Regenerate.
11281 * doc/avr-mmcu.texi: Regenerate.
11283 2014-03-03 Tobias Grosser <tobias@grosser.es>
11284 Mircea Namolaru <mircea.namolaru@inria.fr>
11286 PR tree-optimization/58028
11287 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
11290 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11292 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
11293 not handled by recognizers.
11295 2014-03-03 Jakub Jelinek <jakub@redhat.com>
11297 PR middle-end/60175
11298 * function.c (expand_function_end): Don't emit
11299 clobber_return_register sequence if clobber_after is a BARRIER.
11300 * cfgexpand.c (construct_exit_block): Append instructions before
11301 return_label to prev_bb.
11303 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11305 * config/rs6000/constraints.md: Document reserved use of "wc".
11307 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
11310 * ipa.c (function_and_variable_visibility): When dissolving comdat
11311 group, also set all symbols to local.
11313 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
11318 2013-12-14 Jan Hubicka <jh@suse.cz>
11319 PR middle-end/58477
11320 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
11322 2014-03-02 Jon Beniston <jon@beniston.com>
11328 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
11329 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
11330 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
11331 (simple_return, *simple_return): New patterns
11332 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
11333 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
11335 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
11337 * dwarf2out.c (gen_subprogram_die): Tidy.
11339 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
11342 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
11343 (*mov_t_msb_neg_negc): ... this new insn.
11345 2014-02-28 Jason Merrill <jason@redhat.com>
11348 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
11351 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
11354 * dwarf2out.c (decltype_auto_die): New static.
11355 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
11356 (gen_type_die_with_usage): Handle 'decltype(auto)'.
11357 (is_cxx_auto): Likewise.
11359 2014-02-28 Ian Bolton <ian.bolton@arm.com>
11361 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
11362 we are not using general regs only.
11364 2014-02-28 Richard Biener <rguenther@suse.de>
11367 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
11368 previous fix and only allow to remove trivial pre-headers
11369 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
11370 (remove_forwarder_block): Properly update the latch of a loop.
11372 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
11375 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
11376 (cselib_preserved_hash_table): New.
11377 (preserve_constants_and_equivs): Move preserved vals to it.
11378 (cselib_find_slot): Look it up first.
11379 (cselib_init): Initialize it.
11380 (cselib_finish): Release it.
11381 (dump_cselib_table): Dump it.
11383 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
11386 * cselib.c (remove_useless_values): Skip to avoid quadratic
11387 behavior if the condition moved from...
11388 (cselib_process_insn): ... here holds.
11390 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
11393 * var-tracking.c (vt_initialize): Apply the same condition to
11394 preserve the CFA base value.
11396 2014-02-28 Joey Ye <joey.ye@arm.com>
11399 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
11400 if reload in progress or completed.
11402 2014-02-28 Tobias Burnus <burnus@net-b.de>
11404 PR middle-end/60147
11405 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
11408 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
11410 * doc/tm.texi.in (Condition Code Status): Update documention for
11411 relative locations of cc0-setter and cc0-user.
11413 2014-02-27 Jeff Law <law@redhat.com>
11415 PR rtl-optimization/52714
11416 * combine.c (try_combine): When splitting an unrecognized PARALLEL
11417 into two independent simple sets, if I3 is a jump, ensure the
11418 pattern we place into I3 is a (set (pc) ...).
11420 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
11421 Jeff Law <law@redhat.com>
11423 PR rtl-optimization/49847
11424 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
11425 are in different blocks.
11426 * doc/tm.texi (Condition Code Status): Update documention for
11427 relative locations of cc0-setter and cc0-user.
11429 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
11432 * lra.c (lra_emit_add): Check SUBREG too.
11434 2014-02-27 Andreas Schwab <schwab@suse.de>
11436 * config/m68k/m68k.c (m68k_option_override): Disable
11437 -flive-range-shrinkage for classic m68k.
11438 (m68k_override_options_after_change): Likewise.
11440 2014-02-27 Marek Polacek <polacek@redhat.com>
11442 PR middle-end/59223
11443 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
11444 -Wmaybe-uninitialized.
11446 2014-02-27 Alan Modra <amodra@gmail.com>
11449 * reload1.c (emit_input_reload_insns): When reload_override_in,
11450 set old to rl->in_reg when rl->in_reg is a subreg.
11452 2014-02-26 Richard Biener <rguenther@suse.de>
11455 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
11457 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
11459 * common/config/i386/predicates.md (const1256_operand): Remove.
11460 (const2356_operand): New.
11461 (const_1_to_2_operand): Remove.
11462 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
11463 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
11464 (*avx512pf_gatherpf<mode>sf): Ditto.
11465 (avx512pf_gatherpf<mode>df): Ditto.
11466 (*avx512pf_gatherpf<mode>df_mask): Ditto.
11467 (*avx512pf_gatherpf<mode>df): Ditto.
11468 (avx512pf_scatterpf<mode>sf): Ditto.
11469 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
11470 (*avx512pf_scatterpf<mode>sf): Ditto.
11471 (avx512pf_scatterpf<mode>df): Ditto.
11472 (*avx512pf_scatterpf<mode>df_mask): Ditto.
11473 (*avx512pf_scatterpf<mode>df): Ditto.
11474 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
11476 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
11478 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
11479 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
11480 (_mm512_mask_testn_epi64_mask): Move to ...
11481 * config/i386/avx512cdintrin.h: Here.
11482 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
11483 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
11484 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
11485 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
11486 TARGET_AVX512F from TARGET_AVX512CD.
11488 2014-02-26 Richard Biener <rguenther@suse.de>
11491 * ipa.c (walk_polymorphic_call_targets): Properly guard
11492 call to inline_update_overall_summary.
11494 2014-02-26 Bin Cheng <bin.cheng@arm.com>
11497 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
11498 and latches only if requested. Fix latch if it is removed.
11499 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
11500 LOOPS_HAVE_PREHEADERS.
11502 2014-02-25 Andrew Pinski <apinski@cavium.com>
11504 * builtins.c (expand_builtin_thread_pointer): Create a new target
11505 when the target is NULL.
11507 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
11509 PR rtl-optimization/60317
11510 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
11511 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
11512 * lra-assigns.c: Include params.h.
11513 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
11514 other reload pseudos considerations.
11516 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11518 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
11519 to use canonical form for nor<mode>3.
11521 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11524 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
11527 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
11529 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
11530 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
11531 (ix86_handle_option): Handle OPT_mprefetchwt1.
11532 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
11533 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11535 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11536 OPTION_MASK_ISA_PREFETCHWT1.
11537 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
11538 (PTA_PREFETCHWT1): New.
11539 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
11540 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
11541 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
11542 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
11543 (*prefetch_avx512pf_<mode>_: Change into ...
11544 (*prefetch_prefetchwt1_<mode>: This.
11545 * config/i386/i386.opt (mprefetchwt1): New.
11546 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
11547 (_mm_prefetch): Handle intent to write.
11548 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
11550 2014-02-25 Richard Biener <rguenther@suse.de>
11552 PR middle-end/60291
11553 * emit-rtl.c (mem_attrs_htab): Remove.
11554 (mem_attrs_htab_hash): Likewise.
11555 (mem_attrs_htab_eq): Likewise.
11556 (set_mem_attrs): Always allocate new mem-attrs when something changed.
11557 (init_emit_once): Do not allocate mem_attrs_htab.
11559 2014-02-25 Richard Biener <rguenther@suse.de>
11562 * lto-opts.c (lto_write_options): Output non-explicit conservative
11563 -fwrapv, -fno-trapv and -fno-strict-overflow.
11564 * lto-wrapper.c (merge_and_complain): Handle merging those options.
11565 (run_gcc): And pass them through.
11567 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
11569 * sel-sched.c (calculate_new_fences): New parameter ptime.
11570 Calculate it as a maximum over all fence cycles.
11571 (sel_sched_region_2): Adjust the call to calculate_new_fences.
11572 Print the final schedule timing when sched_verbose.
11574 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
11576 PR rtl-optimization/60292
11577 * sel-sched.c (fill_vec_av_set): Do not reset target availability
11578 bit fot the fence instruction.
11580 2014-02-24 Alangi Derick <alangiderick@gmail.com>
11582 * calls.h: Fix typo in comment.
11584 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
11586 * config/pa/pa.c (pa_output_move_double): Don't valididate when
11587 adjusting offsetable addresses.
11589 2014-02-24 Guozhi Wei <carrot@google.com>
11591 * sparseset.h (sparseset_pop): Fix the wrong index.
11593 2014-02-24 Walter Lee <walt@tilera.com>
11595 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
11596 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
11598 * common/config/tilegx/tilegx-common.c
11599 (TARGET_DEFAULT_TARGET_FLAGS): Define.
11600 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
11601 (LINK_SPEC): Ditto.
11602 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
11603 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
11604 (tilegx_gimplify_va_arg_expr): Handle big endian.
11605 (tilegx_expand_unaligned_load): Ditto.
11606 (tilegx_expand_unaligned_store): Ditto.
11607 (TARGET_RETURN_IN_MSB): New.
11608 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
11609 (TARGET_ENDIAN_DEFAULT): New.
11610 (TARGET_BIG_ENDIAN): Handle big endian.
11611 (BYTES_BIG_ENDIAN): Ditto.
11612 (WORDS_BIG_ENDIAN): Ditto.
11613 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
11614 (ENDIAN_SPEC): New.
11615 (EXTRA_SPECS): New.
11616 * config/tilegx/tilegx.md (extv): Handle big endian.
11618 (insn_st<n>): Ditto.
11619 (insn_st<n>_add<bitsuffix>): Ditto.
11620 (insn_stnt<n>): Ditto.
11621 (insn_stnt<n>_add<bitsuffix>):Ditto.
11622 (vec_interleave_highv8qi): Handle big endian.
11623 (vec_interleave_highv8qi_be): New.
11624 (vec_interleave_highv8qi_le): New.
11625 (insn_v1int_h): Handle big endian.
11626 (vec_interleave_lowv8qi): Handle big endian.
11627 (vec_interleave_lowv8qi_be): New.
11628 (vec_interleave_lowv8qi_le): New.
11629 (insn_v1int_l): Handle big endian.
11630 (vec_interleave_highv4hi): Handle big endian.
11631 (vec_interleave_highv4hi_be): New.
11632 (vec_interleave_highv4hi_le): New.
11633 (insn_v2int_h): Handle big endian.
11634 (vec_interleave_lowv4hi): Handle big endian.
11635 (vec_interleave_lowv4hi_be): New.
11636 (vec_interleave_lowv4hi_le): New.
11637 (insn_v2int_l): Handle big endian.
11638 (vec_interleave_highv2si): Handle big endian.
11639 (vec_interleave_highv2si_be): New.
11640 (vec_interleave_highv2si_le): New.
11641 (insn_v4int_h): Handle big endian.
11642 (vec_interleave_lowv2si): Handle big endian.
11643 (vec_interleave_lowv2si_be): New.
11644 (vec_interleave_lowv2si_le): New.
11645 (insn_v4int_l): Handle big endian.
11646 * config/tilegx/tilegx.opt (mbig-endian): New option.
11647 (mlittle-endian): New option.
11648 * doc/install.texi: Document tilegxbe-linux.
11649 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
11651 2014-02-24 Martin Jambor <mjambor@suse.cz>
11654 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
11655 there are no parameter descriptors.
11657 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
11659 PR rtl-optimization/60268
11660 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
11661 initialization to ...
11662 (sched_rgn_init): ... here.
11663 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
11665 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
11667 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
11670 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
11672 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
11675 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
11677 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
11678 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
11680 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
11682 * config/microblaze/predicates.md: Add cmp_op predicate.
11683 * config/microblaze/microblaze.md: Add branch_compare instruction
11684 which uses cmp_op predicate and emits cmp insn before branch.
11685 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
11686 to microblaze_expand_conditional_branch and consolidate logic.
11687 (microblaze_expand_conditional_branch): emit branch_compare
11688 insn instead of handling cmp op separate from branch insn.
11690 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11692 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
11695 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11697 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
11698 define_insn with define_expand and new define_insn
11699 *altivec_lve<VI_char>x_internal.
11700 (altivec_stve<VI_char>x): Replace define_insn with define_expand
11701 and new define_insn *altivec_stve<VI_char>x_internal.
11702 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
11704 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
11706 (altivec_expand_stvex_be): New function.
11708 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
11710 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
11711 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
11712 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
11713 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
11715 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
11718 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
11719 instead of emit_move_insn.
11721 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11723 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
11724 vspltw with vsldoi.
11725 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
11726 gen_altivec_vsumsws.
11728 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11730 * config/rs6000/altivec.md (altivec_lvxl): Rename as
11731 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
11732 (altivec_lvxl_<mode>): New define_expand incorporating
11733 -maltivec=be semantics where needed.
11734 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
11735 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
11736 semantics where needed.
11737 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
11738 (altivec_stvx_<mode>): New define_expand incorporating
11739 -maltivec=be semantics where needed.
11740 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
11741 VM2 iterator instead of V4SI.
11742 (altivec_stvxl_<mode>): New define_expand incorporating
11743 -maltivec=be semantics where needed.
11744 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
11745 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
11746 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
11747 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
11748 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
11749 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
11750 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
11751 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
11752 ALTIVEC_BUILTIN_STVXL.
11753 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
11754 (altivec_expand_stvx_be): Likewise.
11755 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
11756 (altivec_expand_lvx_be): Likewise.
11757 (altivec_expand_stvx_be): Likewise.
11758 (altivec_expand_builtin): Add cases for
11759 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
11760 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
11761 (altivec_init_builtins): Add definitions for
11762 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
11763 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
11765 2014-02-21 Catherine Moore <clm@codesourcery.com>
11767 * doc/invoke.texi (mvirt, mno-virt): Document.
11768 * config/mips/mips.opt (mvirt): New option.
11769 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
11771 2014-02-21 Richard Biener <rguenther@suse.de>
11773 PR tree-optimization/60276
11774 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
11775 (STMT_VINFO_MIN_NEG_DIST): New macro.
11776 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
11777 STMT_VINFO_MIN_NEG_DIST.
11778 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
11779 made for negative dependence distances still hold.
11781 2014-02-21 Richard Biener <rguenther@suse.de>
11783 PR middle-end/60291
11784 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
11785 DECL_INITIAL for globals not in the current function context.
11787 2014-02-21 Jakub Jelinek <jakub@redhat.com>
11789 PR tree-optimization/56490
11790 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
11791 * tree-ssa-uninit.c: Include params.h.
11792 (compute_control_dep_chain): Add num_calls argument, return false
11793 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
11794 num_calls to recursive call.
11795 (find_predicates): Change dep_chain into normal array,
11796 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
11797 variable and adjust compute_control_dep_chain caller.
11798 (find_def_preds): Likewise.
11800 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
11802 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
11803 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
11805 2014-02-21 Nick Clifton <nickc@redhat.com>
11807 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
11808 (pushhi1): Likewise.
11809 (popqi1): Add mode to pre_dec.
11810 (pophi1): Likewise.
11812 2014-02-21 Jakub Jelinek <jakub@redhat.com>
11814 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
11815 mode for mask of V8SFmode permutation.
11817 2014-02-20 Richard Henderson <rth@redhat.com>
11820 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
11821 a new pseudo for OLDVAL.
11823 2014-02-20 Jakub Jelinek <jakub@redhat.com>
11826 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
11827 gen_reg_rtx if d->testing_p.
11828 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
11829 if d->testing_p and we will certainly return true.
11830 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
11833 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
11835 * emit-rtl.c (gen_reg_rtx): Assert that
11836 crtl->emit.regno_pointer_align_length is non-zero.
11838 2014-02-20 Richard Henderson <rth@redhat.com>
11841 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
11842 on failure the store back into EXPECT.
11844 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
11845 Sandra Loosemore <sandra@codesourcery.com>
11847 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
11848 * config/nios2/nios2.c (nios2_function_profiler): Add
11849 -fPIC (flag_pic == 2) support.
11850 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
11851 (nios2_large_offset_p): New function.
11852 (nios2_unspec_reloc_p): Move up position, update to use
11853 nios2_large_offset_p.
11854 (nios2_unspec_address): Remove function.
11855 (nios2_unspec_offset): New function.
11856 (nios2_large_got_address): New function.
11857 (nios2_got_address): Add large offset support.
11858 (nios2_legitimize_tls_address): Update usage of removed and new
11860 (nios2_symbol_binds_local_p): New function.
11861 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
11862 (nios2_legitimize_address): Update to use nios2_large_offset_p.
11863 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
11864 (nios2_print_operand): Merge H/L processing, add hiadj/lo
11865 processing for (const (unspec ...)).
11866 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
11868 2014-02-20 Richard Biener <rguenther@suse.de>
11870 * tree-cfg.c (replace_uses_by): Mark altered BBs before
11871 doing the substitution.
11872 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
11874 2014-02-20 Martin Jambor <mjambor@suse.cz>
11877 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
11878 info when checking whether lattices are bottom.
11880 2014-02-20 Richard Biener <rguenther@suse.de>
11882 PR middle-end/60221
11883 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
11886 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
11889 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
11890 parameter specifying the scaling.
11891 (inline_call): Update.
11892 (want_inline_recursively): Guard division by zero.
11893 (recursive_inlining): Update.
11894 * ipa-inline.h (clone_inlined_nodes): Update.
11896 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
11899 * config/i386/i386.c (classify_argument): Pass structures of size
11900 64 bytes or less in register.
11902 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
11903 Kirill Yukhin <kirill.yukhin@intel.com>
11905 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
11906 (_mm_rcp28_round_ss): Ditto.
11907 (_mm_rsqrt28_round_sd): Ditto.
11908 (_mm_rsqrt28_round_ss): Ditto.
11909 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
11910 (_mm_rcp14_round_ss): Ditto.
11911 (_mm_rsqrt14_round_sd): Ditto.
11912 (_mm_rsqrt14_round_ss): Ditto.
11913 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
11914 the first input operand, get rid of match_dup.
11915 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
11917 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
11919 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
11920 operand as the first input operand, set type attribute.
11921 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
11922 Set type attribute.
11923 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
11924 operand as the first input operand, set type attribute.
11926 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11928 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
11931 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
11934 * config/i386/i386.c (construct_container): Remove TFmode check
11935 for X86_64_INTEGER_CLASS.
11937 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
11940 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
11941 only when -Wpsabi is enabled.
11943 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
11946 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
11947 passing arrays in registers are the same as for structs, so remove the
11948 special case for them.
11950 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
11952 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
11953 destination type, extract only the valid bits if the source type is not
11954 integral and has a different mode.
11956 2014-02-19 Richard Biener <rguenther@suse.de>
11959 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
11962 2014-02-19 Richard Biener <rguenther@suse.de>
11965 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
11966 (ipa_modify_call_arguments): Emit an argument load explicitely and
11967 preserve virtual SSA form there and for the replacement call.
11968 Do not update SSA form nor free dominance info.
11970 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
11972 * ipa.c (function_and_variable_visibility): Also clear WEAK
11973 flag when disolving COMDAT_GROUP.
11975 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
11977 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
11978 * ipa-prop.c (ipa_set_jf_known_type): Return early when
11979 not devirtualizing.
11980 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
11981 do more sanity checks.
11982 (detect_type_change): Return true when giving up early.
11983 (compute_complex_assign_jump_func): Fix type parameter of
11984 ipa_set_ancestor_jf.
11985 (compute_complex_ancestor_jump_func): Likewise.
11986 (update_jump_functions_after_inlining): Fix updating of
11988 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
11990 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
11992 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
11993 inline clones when edge disappears.
11995 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11998 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
11999 Split 64-bit moves into 2 patterns. Do not allow the use of
12000 direct move for TDmode in little endian, since the decimal value
12001 has little endian bytes within a word, but the 64-bit pieces are
12002 ordered in a big endian fashion, and normal subreg's of TDmode are
12004 (mov<mode>_64bit_dm): Likewise.
12005 (movtd_64bit_nodm): Likewise.
12007 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
12009 PR tree-optimization/60174
12010 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
12011 statement of an SSA_NAME that occurs in an abnormal PHI node.
12013 2014-02-18 Jakub Jelinek <jakub@redhat.com>
12016 * final.c (SEEN_BB): Remove.
12017 (SEEN_NOTE, SEEN_EMITTED): Renumber.
12018 (final_scan_insn): Don't force_source_line on second
12019 NOTE_INSN_BASIC_BLOCK.
12021 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
12024 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
12025 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
12026 (type_natural_mode): Warn ABI change when %zmm register is not
12027 available for AVX512F vector value passing.
12029 2014-02-18 Kai Tietz <ktietz@redhat.com>
12032 * config/i386/i386.c (ix86_expand_prologue): Use value in
12033 rax register as displacement when restoring %r10 or %rax.
12034 Fix wrong offset when restoring both registers.
12036 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
12038 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
12039 assertion with conditional return.
12041 2014-02-18 Jakub Jelinek <jakub@redhat.com>
12042 Uros Bizjak <ubizjak@gmail.com>
12045 * config/i386/driver-i386.c (host_detect_local_cpu): If
12046 YMM state is not saved by the OS, also clear has_f16c. Move
12047 CPUID 0x80000001 handling before YMM state saving checking.
12049 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
12051 PR rtl-optimization/58960
12052 * haifa-sched.c (alloc_global_sched_pressure_data): New,
12053 factored out from ...
12054 (sched_init): ... here.
12055 (free_global_sched_pressure_data): New, factored out from ...
12056 (sched_finish): ... here.
12057 * sched-int.h (free_global_sched_pressure_data): Declare.
12058 * sched-rgn.c (nr_regions_initial): New static global.
12059 (haifa_find_rgns): Initialize it.
12060 (schedule_region): Disable sched-pressure for the newly
12063 2014-02-17 Richard Biener <rguenther@suse.de>
12065 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
12066 release SSA defs of pattern stmts.
12068 2014-02-17 Richard Biener <rguenther@suse.de>
12070 * tree-inline.c (expand_call_inline): Release the virtual
12071 operand defined by the call we are about to inline.
12073 2014-02-17 Richard Biener <rguenther@suse.de>
12075 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
12077 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
12078 Ilya Tocar <ilya.tocar@intel.com>
12080 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
12081 arguments order in builtin.
12082 (_mm512_permutexvar_epi64): Ditto.
12083 (_mm512_mask_permutexvar_epi64): Ditto
12084 (_mm512_maskz_permutexvar_epi32): Ditto
12085 (_mm512_permutexvar_epi32): Ditto
12086 (_mm512_mask_permutexvar_epi32): Ditto
12088 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12090 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
12091 (p8_vmrgow): Likewise.
12093 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12095 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
12098 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
12101 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
12102 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
12103 into 64-bit and 32-bit moves. On 64-bit moves, add support for
12104 using direct move instructions on ISA 2.07. Also adjust
12105 instruction length for 64-bit.
12106 (mov<mode>_64bit, TFmode/TDmode): Likewise.
12107 (mov<mode>_32bit, TFmode/TDmode): Likewise.
12109 2014-02-15 Alan Modra <amodra@gmail.com>
12113 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
12114 find_replacement on parts of insn rtl that might be reloaded.
12116 2014-02-15 Richard Biener <rguenther@suse.de>
12118 PR tree-optimization/60183
12119 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
12120 (tree_ssa_phiprop): Calculate and free post-dominators.
12122 2014-02-14 Jeff Law <law@redhat.com>
12124 PR rtl-optimization/60131
12125 * ree.c (get_extended_src_reg): New function.
12126 (combine_reaching_defs): Use it rather than assuming location of REG.
12127 (find_and_remove_re): Verify first operand of extension is
12128 a REG before adding the insns to the copy list.
12130 2014-02-14 Roland McGrath <mcgrathr@google.com>
12132 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
12133 * configure: Regenerated.
12134 * config.in: Regenerated.
12135 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
12136 instead of ASM_SHORT.
12138 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
12139 Richard Earnshaw <rearnsha@arm.com>
12141 PR rtl-optimization/59535
12142 * lra-constraints.c (process_alt_operands): Encourage alternative
12143 when unassigned pseudo class is superset of the alternative class.
12144 (inherit_reload_reg): Don't inherit when optimizing for code size.
12145 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
12146 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
12147 modes not less than 4 for Thumb1.
12149 2014-02-14 Kyle McMartin <kyle@redhat.com>
12152 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
12154 2014-02-14 Richard Biener <rguenther@suse.de>
12156 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
12157 (get_frame_arg): Drop the assert with langhook types_compatible_p.
12158 Do not strip INDIRECT_REFs.
12160 2014-02-14 Richard Biener <rguenther@suse.de>
12163 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
12164 DECL_FUNCTION_SPECIFIC_TARGET.
12165 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
12166 * tree-streamer-out.c (pack_ts_target_option): Remove.
12167 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
12168 (write_ts_function_decl_tree_pointers): Do not stream
12169 DECL_FUNCTION_SPECIFIC_TARGET.
12170 * tree-streamer-in.c (unpack_ts_target_option): Remove.
12171 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
12172 (lto_input_ts_function_decl_tree_pointers): Do not stream
12173 DECL_FUNCTION_SPECIFIC_TARGET.
12175 2014-02-14 Jakub Jelinek <jakub@redhat.com>
12177 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
12178 (get_initial_def_for_induction, vectorizable_induction): Ignore
12179 debug stmts when looking for exit_phi.
12180 (vectorizable_live_operation): Fix up condition.
12182 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
12184 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
12185 nreverse() because it changes the content of original tree list.
12187 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
12189 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
12190 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
12192 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
12194 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
12195 GNU coding standards.
12197 2014-02-13 Jakub Jelinek <jakub@redhat.com>
12200 * dwarf2out.c (gen_subprogram_die): Don't call
12201 add_calling_convention_attribute if subr_die is old_die.
12203 2014-02-13 Sharad Singhai <singhai@google.com>
12205 * doc/optinfo.texi: Fix order of nodes.
12207 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
12209 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
12210 operands[2], not operands[3].
12212 2014-02-13 Richard Biener <rguenther@suse.de>
12215 * doc/install.texi (ISL): Update recommended version to 0.12.2,
12216 mention the possibility of an in-tree build.
12217 (CLooG): Update recommended version to 0.18.1, mention the
12218 possibility of an in-tree build and clarify that the ISL
12219 bundled with CLooG does not work.
12221 2014-02-13 Jakub Jelinek <jakub@redhat.com>
12224 * expr.c (compress_float_constant): If x is a hard register,
12225 extend into a pseudo and then move to x.
12227 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
12229 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
12230 caused by bad second argument to warning_at() with -mhotpatch and
12231 nested functions (e.g. with gfortran).
12233 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
12235 * opts.c (option_name): Remove "enabled by default" rider.
12237 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
12239 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
12241 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
12242 Uros Bizjak <ubizjak@gmail.com>
12245 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
12246 * configure: Regenerated.
12248 2014-02-12 Richard Biener <rguenther@suse.de>
12250 * vec.c (vec_prefix::calculate_allocation): Move as
12251 inline variant to vec.h.
12252 (vec_prefix::calculate_allocation_1): New out-of-line version.
12253 * vec.h (vec_prefix::calculate_allocation_1): Declare.
12254 (vec_prefix::m_has_auto_buf): Rename to ...
12255 (vec_prefix::m_using_auto_storage): ... this.
12256 (vec_prefix::calculate_allocation): Inline the easy cases
12257 and dispatch to calculate_allocation_1 which doesn't need the
12259 (va_heap::reserve): Use gcc_checking_assert.
12260 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
12261 m_using_auto_storage.
12262 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
12264 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
12265 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
12266 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
12268 2014-02-12 Richard Biener <rguenther@suse.de>
12270 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
12271 when we found a dependence.
12273 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
12275 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
12277 (maybe_fold_stmt): ... into this new function.
12278 * omp-low.c (lower_omp): Update comment.
12280 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
12283 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
12286 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
12288 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
12289 identifiers in comments.
12290 (cortexa53_extra_costs): Likewise.
12291 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
12292 (cortexa7_extra_costs): Likewise.
12293 (cortexa12_extra_costs): Likewise.
12294 (cortexa15_extra_costs): Likewise.
12295 (v7m_extra_costs): Likewise.
12297 2014-02-12 Richard Biener <rguenther@suse.de>
12299 PR middle-end/60092
12300 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
12301 of posix_memalign being successful.
12302 (lower_stmt): Restrict lowering of posix_memalign to when
12303 -ftree-bit-ccp is enabled.
12305 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
12307 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
12309 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
12311 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
12313 PR rtl-optimization/60116
12314 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
12315 other_insn once the combination has been validated.
12317 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
12320 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
12322 * ipa-devirt.c: Include demangle.h
12323 (odr_violation_reported): New static variable.
12324 (add_type_duplicate): Update odr_violations.
12325 (maybe_record_node): Add completep parameter; update it.
12326 (record_target_from_binfo): Add COMPLETEP parameter;
12327 update it as needed.
12328 (possible_polymorphic_call_targets_1): Likewise.
12329 (struct polymorphic_call_target_d): Add nonconstruction_targets;
12330 rename FINAL to COMPLETE.
12331 (record_targets_from_bases): Sanity check we found the binfo;
12332 fix COMPLETEP updating.
12333 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
12334 parameter, fix computing of COMPLETEP.
12335 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
12336 at LTO time do demangling.
12337 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
12338 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
12340 (gimple_get_virt_method_for_binfo): Likewise.
12341 * gimple-fold.h (gimple_get_virt_method_for_binfo,
12342 gimple_get_virt_method_for_vtable): Update prototypes.
12344 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
12347 * genautomata.c (add_presence_absence): Fix typo with
12348 {final_}presence_list.
12350 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12353 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
12354 for VSX/Altivec vectors that land in GPR registers.
12356 2014-02-11 Richard Henderson <rth@redhat.com>
12357 Jakub Jelinek <jakub@redhat.com>
12360 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
12361 around drhs if type conversion to lacc->type is not useless.
12363 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12365 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
12367 (cortex-a57.cortex-a53): Likewise.
12368 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
12370 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12372 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
12375 2014-02-11 Renlin Li <Renlin.Li@arm.com>
12377 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
12378 add_options_for_arm_vfp3.
12380 2014-02-11 Jeff Law <law@redhat.com>
12382 PR middle-end/54041
12383 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
12384 object with an undesirable mode.
12386 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12389 * config/i386/sol2-9.h: New file.
12390 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
12391 *-*-solaris2.9*): Use it.
12393 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
12395 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
12396 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
12398 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
12400 * config/microblaze/microblaze.c: Extend mcpu version format
12402 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
12404 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
12406 2014-02-10 Richard Henderson <rth@redhat.com>
12409 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
12410 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
12411 ms-abi vs -mno-accumulate-outgoing-args.
12412 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
12413 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
12416 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
12418 PR middle-end/60080
12419 * cfgexpand.c (expand_asm_operands): Attach source location to
12420 ASM_INPUT rtx objects.
12421 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
12423 2014-02-10 Nick Clifton <nickc@redhat.com>
12425 * config/mn10300/mn10300.c (popcount): New function.
12426 (mn10300_expand_prologue): Include saved registers in stack usage
12429 2014-02-10 Jeff Law <law@redhat.com>
12431 PR middle-end/52306
12432 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
12433 when changing the SET_DEST of a prior insn to avoid an input reload.
12435 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
12437 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
12438 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
12439 -mcall-openbsd, or -mcall-linux.
12440 (CC1_ENDIAN_BIG_SPEC): Remove.
12441 (CC1_ENDIAN_LITTLE_SPEC): Remove.
12442 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
12443 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
12444 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
12445 and %cc1_endian_default.
12446 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
12448 2014-02-10 Richard Biener <rguenther@suse.de>
12450 PR tree-optimization/60115
12451 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
12452 MEM_REF handling. Properly verify that the accesses are not
12453 out of the objects bound.
12455 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12457 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
12460 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
12462 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
12463 proper constants and fix formatting.
12464 (possible_polymorphic_call_targets): Fix formatting.
12466 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
12467 Ilya Tocar <ilya.tocar@intel.com>
12469 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
12470 (_mm512_loadu_epi32): Renamed into...
12471 (_mm512_loadu_si512): This.
12472 (_mm512_storeu_epi32): Renamed into...
12473 (_mm512_storeu_si512): This.
12474 (_mm512_maskz_ceil_ps): Removed.
12475 (_mm512_maskz_ceil_pd): Ditto.
12476 (_mm512_maskz_floor_ps): Ditto.
12477 (_mm512_maskz_floor_pd): Ditto.
12478 (_mm512_floor_round_ps): Ditto.
12479 (_mm512_floor_round_pd): Ditto.
12480 (_mm512_ceil_round_ps): Ditto.
12481 (_mm512_ceil_round_pd): Ditto.
12482 (_mm512_mask_floor_round_ps): Ditto.
12483 (_mm512_mask_floor_round_pd): Ditto.
12484 (_mm512_mask_ceil_round_ps): Ditto.
12485 (_mm512_mask_ceil_round_pd): Ditto.
12486 (_mm512_maskz_floor_round_ps): Ditto.
12487 (_mm512_maskz_floor_round_pd): Ditto.
12488 (_mm512_maskz_ceil_round_ps): Ditto.
12489 (_mm512_maskz_ceil_round_pd): Ditto.
12490 (_mm512_expand_pd): Ditto.
12491 (_mm512_expand_ps): Ditto.
12492 * config/i386/i386.c (ix86_builtins): Remove
12493 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
12494 (bdesc_args): Ditto.
12495 * config/i386/predicates.md (const1256_operand): New.
12496 (const_1_to_2_operand): Ditto.
12497 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
12498 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
12499 (*avx512pf_gatherpf<mode>sf): Ditto.
12500 (avx512pf_gatherpf<mode>df): Ditto.
12501 (*avx512pf_gatherpf<mode>df_mask): Ditto.
12502 (*avx512pf_gatherpf<mode>df): Ditto.
12503 (avx512pf_scatterpf<mode>sf): Ditto.
12504 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
12505 (*avx512pf_scatterpf<mode>sf): Ditto.
12506 (avx512pf_scatterpf<mode>df): Ditto.
12507 (*avx512pf_scatterpf<mode>df_mask): Ditto.
12508 (*avx512pf_scatterpf<mode>df): Ditto.
12509 (avx512f_expand<mode>): Removed.
12510 (<shift_insn><mode>3<mask_name>): Change predicate type.
12512 2014-02-08 Jakub Jelinek <jakub@redhat.com>
12514 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
12515 not at the end of datarefs vector use ordered_remove to avoid
12516 reordering datarefs vector.
12519 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
12520 mark local addressable non-static vars as GOVD_PRIVATE
12521 instead of GOVD_LOCAL.
12522 * omp-low.c (lower_omp_for): Move gimple_bind_vars
12523 and BLOCK_VARS of gimple_bind_block to new_stmt rather
12526 PR middle-end/60092
12527 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
12528 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
12529 assume_aligned or alloc_align attributes.
12530 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
12531 arguments. Handle also assume_aligned and alloc_align attributes.
12532 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
12533 calls to functions with assume_aligned or alloc_align attributes.
12534 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
12536 2014-02-08 Terry Guo <terry.guo@arm.com>
12538 * doc/invoke.texi: Document ARM -march=armv7e-m.
12540 2014-02-08 Jakub Jelinek <jakub@redhat.com>
12542 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
12543 flag on __cilkrts_rethrow builtin.
12546 * ipa-cp.c (determine_versionability): Fail at -O0
12547 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
12548 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
12551 2014-02-04 Jakub Jelinek <jakub@redhat.com>
12554 * tree-inline.c (copy_forbidden): Fail for
12555 __attribute__((optimize (0))) functions.
12557 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
12559 * varpool.c: Include pointer-set.h.
12560 (varpool_remove_unreferenced_decls): Variables in other partitions
12561 will not be output; be however careful to not lose information
12562 about partitioning.
12564 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
12566 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
12567 lookup in the vtable constructor.
12569 2014-02-07 Jeff Law <law@redhat.com>
12572 * config/m68k/m68k.md (ashldi_extsi): Turn into a
12573 define_insn_and_split.
12575 * ipa-inline.c (inline_small_functions): Fix typos.
12577 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
12579 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
12580 (s390_can_use_return_insn): Declare.
12581 * config/s390/s390.h (EPILOGUE_USES): Define.
12582 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
12584 (s390_chunkify_start): Handle return JUMP_LABELs.
12585 (s390_early_mach): Emit a main_pool instruction on the entry edge.
12586 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
12587 (s390_can_use_return_insn): New functions.
12588 (s390_fix_long_loop_prediction): Handle conditional returns.
12589 (TARGET_SET_UP_BY_PROLOGUE): Define.
12590 * config/s390/s390.md (ANY_RETURN): New code iterator.
12591 (*creturn, *csimple_return, return, simple_return): New patterns.
12593 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
12595 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
12596 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
12597 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
12598 REG_CFA_RESTORE list when deciding not to restore a register.
12600 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
12602 * config/s390/s390.c: Include tree-pass.h and context.h.
12603 (s390_early_mach): New function, split out from...
12604 (s390_emit_prologue): ...here.
12605 (pass_data_s390_early_mach): New pass structure.
12606 (pass_s390_early_mach): New class.
12607 (s390_option_override): Create and register early_mach pass.
12608 Move to end of file.
12610 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
12612 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
12613 to match for the exit block.
12615 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12617 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
12618 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
12619 Reject misaligned operands.
12621 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12623 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
12625 2014-02-07 Richard Biener <rguenther@suse.de>
12627 PR middle-end/60092
12628 * gimple-low.c (lower_builtin_posix_memalign): New function.
12629 (lower_stmt): Call it to lower posix_memalign in a way
12630 to make alignment info accessible.
12632 2014-02-07 Jakub Jelinek <jakub@redhat.com>
12635 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
12636 __builtin_setjmp_receiver.
12638 2014-02-07 Richard Biener <rguenther@suse.de>
12640 PR middle-end/60092
12641 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
12642 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
12643 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
12644 Handle BUILT_IN_POSIX_MEMALIGN.
12645 (find_func_clobbers): Likewise.
12646 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
12647 (call_may_clobber_ref_p_1): Likewise.
12649 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
12652 * ipa-devirt.c (record_target_from_binfo): Remove overactive
12655 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
12658 * lto-cgraph.c (lto_output_node): Use
12659 symtab_get_symbol_partitioning_class.
12660 (lto_output_varpool_node): likewise.
12661 (symtab_get_symbol_partitioning_class): Move here from
12662 lto/lto-partition.c
12663 * cgraph.h (symbol_partitioning_class): Likewise.
12664 (symtab_get_symbol_partitioning_class): Declare.
12666 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
12668 * ggc.h (ggc_internal_cleared_alloc): New macro.
12669 * vec.h (vec_safe_copy): Handle memory stats.
12670 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
12671 * target-globals.c (save_target_globals): Likewise.
12673 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
12676 * expr.c (emit_move_resolve_push): Export; be bit more selective
12677 on when to clear alias set.
12678 * expr.h (emit_move_resolve_push): Declare.
12679 * function.h (struct function): Add tail_call_marked.
12680 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
12681 * config/i386/i386-protos.h (ix86_expand_push): Remove.
12682 * config/i386/i386.md (TImode move expander): De not call
12684 (FP push expanders): Preserve memory attributes.
12685 * config/i386/sse.md (push<mode>1): Remove.
12686 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
12687 (ix86_expand_push): Remove.
12688 * config/i386/mmx.md (push<mode>1): Remove.
12690 2014-02-06 Jakub Jelinek <jakub@redhat.com>
12692 PR rtl-optimization/60030
12693 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
12694 lopart with paradoxical subreg before shifting it up by hprec.
12696 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12698 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
12699 Remove extra newline at end of file.
12700 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
12701 (arm_issue_rate): Handle cortexa57.
12702 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
12703 (cortex-a57.cortex-a53): Likewise.
12705 2014-02-06 Jakub Jelinek <jakub@redhat.com>
12708 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
12709 don't record in REG_FRAME_RELATED_EXPR registers not set in that
12711 (arm_expand_prologue): Adjust all callers.
12712 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
12713 info, registers also at the lowest numbered registers side. Use
12714 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
12718 * var-tracking.c (adjust_mems): Before adding a SET to
12719 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
12721 2014-02-06 Alan Modra <amodra@gmail.com>
12724 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
12725 change SDmode to DDmode when lra_in_progress.
12727 2014-02-06 Jakub Jelinek <jakub@redhat.com>
12729 PR middle-end/59150
12730 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
12731 free_data_ref on the dr first, and before goto again also set dr
12732 to the next dr. For simd_lane_access, free old datarefs[i] before
12733 overwriting it. For get_vectype_for_scalar_type failure, don't
12734 free_data_ref if simd_lane_access.
12736 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
12739 * tree.h (opts_for_fn): New inline function.
12740 (opt_for_fn): Define.
12741 * config/i386/i386.c (ix86_function_regparm): Use
12742 opt_for_fn (decl, optimize) instead of optimize.
12744 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
12746 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
12747 for SYMBOL_REF in large memory model.
12749 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12751 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
12752 and crypto support.
12753 (cortex-a57): Likewise.
12754 (cortex-a57.cortex-a53): Likewise.
12756 2014-02-06 Yury Gribov <y.gribov@samsung.com>
12757 Kugan Vivekanandarajah <kuganv@linaro.org>
12759 * config/arm/arm.c (arm_vector_alignment_reachable): Check
12761 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
12763 2014-02-06 Richard Biener <rguenther@suse.de>
12765 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
12766 set_loop_copy and initialize_original_copy_tables.
12768 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
12770 * config/aarch64/aarch64-simd.md
12771 (aarch64_ashr_simddi): Change QI to SI.
12773 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
12774 Jakub Jelinek <jakub@redhat.com>
12776 PR middle-end/60013
12777 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
12780 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12782 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
12783 CODE_FOR_altivec_vpku[hw]um to
12784 CODE_FOR_altivec_vpku[hw]um_direct.
12785 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
12786 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
12787 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
12788 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
12790 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12792 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
12793 generation for -maltivec=be.
12794 (altivec_vsumsws): Simplify redundant test.
12796 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12798 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
12799 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
12800 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
12801 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
12802 gen_altivec_vpkuwum.
12803 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
12805 (altivec_vpks<VI_char>ss): Likewise.
12806 (altivec_vpks<VI_char>us): Likewise.
12807 (altivec_vpku<VI_char>us): Likewise.
12808 (altivec_vpku<VI_char>um): Likewise.
12809 (altivec_vpku<VI_char>um_direct): New (copy of
12810 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
12812 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
12813 target is little endian and -maltivec=be is not specified.
12814 (*altivec_vupkhs<VU_char>_direct): New (copy of
12815 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
12816 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
12817 target is little endian and -maltivec=be is not specified.
12818 (*altivec_vupkls<VU_char>_direct): New (copy of
12819 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
12820 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
12821 little endian and -maltivec=be is not specified.
12822 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
12823 little endian and -maltivec=be is not specified.
12825 2014-02-05 Richard Henderson <rth@redhat.com>
12828 * combine-stack-adj.c: Revert r206943.
12829 * sched-int.h (struct deps_desc): Add last_args_size.
12830 * sched-deps.c (init_deps): Initialize it.
12831 (sched_analyze_insn): Add OUTPUT dependencies between insns that
12832 contain REG_ARGS_SIZE notes.
12834 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
12836 * lto-cgraph.c (asm_nodes_output): Make global.
12837 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
12838 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
12839 (driver_handle_option): Handle OPT_fwpa.
12841 2014-02-05 Jakub Jelinek <jakub@redhat.com>
12844 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
12845 a comment typo and formatting issue. If odr_hash hasn't been
12846 created, return vNULL and set *completep to false.
12848 PR middle-end/57499
12849 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
12850 bb with no successors.
12852 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
12855 * doc/invoke.texi (-march): Clarify documentation for ARM.
12856 (-mtune): Likewise.
12859 2014-02-05 Richard Biener <rguenther@suse.de>
12861 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
12862 when not vectorizing because of too many alias checks.
12863 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
12864 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
12866 2014-02-05 Nick Clifton <nickc@redhat.com>
12868 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
12869 accept extended registers in any mode when compiling for the MN10300.
12871 2014-02-05 Yury Gribov <y.gribov@samsung.com>
12873 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
12874 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
12875 sanitization attributes.
12876 (can_inline_edge_p): Likewise.
12877 (sanitize_attrs_match_for_inline_p): New function.
12879 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
12881 * ipa-prop.c (detect_type_change): Shor circuit testing of
12882 type changes on THIS pointer.
12884 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
12887 * config/pa/pa.c (legitimize_tls_address): Return original address
12888 if not passed a SYMBOL_REF rtx.
12889 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
12891 (pa_emit_move_sequence): Simplify TLS source operands.
12892 (pa_legitimate_constant_p): Reject all TLS constants.
12893 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
12894 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
12896 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
12898 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
12899 groups when we know they are controlled by LTO.
12900 * varasm.c (default_binds_local_p_1): If object is in other partition,
12901 it will be resolved locally.
12903 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12905 * config/host-linux.c (linux_gt_pch_use_address): Don't
12906 use SSIZE_MAX because it is not always defined.
12908 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
12911 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
12912 threshold for pseudo splitting.
12913 (update_ebb_live_info): Process call argument hard registers and
12914 hard registers from insn definition too.
12915 (max_small_class_regs_num): New constant.
12916 (inherit_in_ebb): Update live hard regs through EBBs. Update
12917 reloads_num only for small register classes. Don't split for
12920 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
12923 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
12926 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
12928 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
12929 visibility is safe.
12931 2014-02-04 Marek Polacek <polacek@redhat.com>
12933 * gdbinit.in (pel): Define.
12935 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12937 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
12940 2014-02-04 Richard Biener <rguenther@suse.de>
12943 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
12944 in function context local.
12945 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
12946 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
12947 similar to LTO_imported_decl_ref.
12949 2014-02-04 Jakub Jelinek <jakub@redhat.com>
12951 PR tree-optimization/60002
12952 * cgraphclones.c (build_function_decl_skip_args): Clear
12953 DECL_LANG_SPECIFIC.
12955 PR tree-optimization/60023
12956 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
12957 false to gsi_replace.
12958 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
12959 has been in some EH region and vec_stmt could throw, add
12960 vec_stmt into the same EH region.
12961 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
12962 has no lhs, ignore it.
12963 * internal-fn.c (expand_MASK_LOAD): Likewise.
12966 * tree-inline.c (copy_forbidden): Fail for
12967 __attribute__((optimize (0))) functions.
12970 * omp-low.c (simd_clone_struct_copy): If from->inbranch
12971 is set, copy one less argument.
12972 (expand_simd_clones): Don't subtract clone_info->inbranch
12973 from simd_clone_struct_alloc argument.
12975 PR rtl-optimization/57915
12976 * recog.c (simplify_while_replacing): If all unary/binary/relational
12977 operation arguments are constant, attempt to simplify those.
12979 PR middle-end/59261
12980 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
12981 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
12983 2014-02-04 Richard Biener <rguenther@suse.de>
12985 PR tree-optimization/60012
12986 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
12987 TBAA disambiguation to all DDRs.
12989 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12992 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
12993 (LINK_SPEC): Use it for -shared, -shared-libgcc.
12995 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
12998 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
13000 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
13002 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
13003 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
13005 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
13008 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
13009 to figure out targets of polymorphic calls with known decl.
13010 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
13011 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
13012 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
13013 (get_polymorphic_call_info): ... here.
13014 (get_polymorphic_call_info_from_invariant): New function.
13016 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
13018 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
13019 lookup via vtable pointer; check for type consistency
13020 and turn inconsitent facts into UNREACHABLE.
13021 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
13022 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
13023 type inconsistent querries; return UNREACHABLE instead.
13025 2014-02-03 Richard Henderson <rth@twiddle.net>
13028 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
13029 already processed this node.
13030 (normalize_one_pred_1): Pass along mark_set.
13031 (normalize_one_pred): Create and destroy a pointer_set_t.
13032 (normalize_one_pred_chain): Likewise.
13034 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
13036 PR gcov-profile/58602
13037 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
13039 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
13042 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
13043 -fno-devirtualize; try to devirtualize by the knowledge of
13044 virtual table pointer given by aggregate propagation.
13045 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
13046 (ipa_print_node_jump_functions): Dump also offset that
13047 is relevant for polymorphic calls.
13048 (determine_known_aggregate_parts): Add arg_type parameter; use it
13049 instead of determining the type from pointer type.
13050 (ipa_compute_jump_functions_for_edge): Update call of
13051 determine_known_aggregate_parts.
13052 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
13053 (gimple_get_virt_method_for_binfo): ... here; simplify using
13054 vtable_pointer_value_to_vtable.
13055 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
13056 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
13057 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
13058 (vtable_pointer_value_to_vtable): Break out from ...; handle also
13060 (vtable_pointer_value_to_binfo): ... here.
13061 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
13063 2014-02-03 Teresa Johnson <tejohnson@google.com>
13065 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
13066 redef of outer loop index variable.
13068 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
13072 * doc/extend.texi (Function Attributes): Typo.
13074 2014-02-03 Cong Hou <congh@google.com>
13076 PR tree-optimization/60000
13077 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
13078 if the vectorized statement is a store. A store statement can only
13079 appear at the end of pattern statements.
13081 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
13083 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
13084 (ix86_option_override_internal): Default long double to 64-bit for
13085 32-bit Bionic and to 128-bit for 64-bit Bionic.
13087 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
13088 TARGET_LONG_DOUBLE_128 is true.
13089 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
13091 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
13092 (mlong-double-64): Negate -mlong-double-128.
13093 (mlong-double-128): New option.
13095 * config/i386/i386-c.c (ix86_target_macros): Define
13096 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
13098 * doc/invoke.texi: Document -mlong-double-128.
13100 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
13102 PR rtl-optimization/60024
13103 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
13105 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
13107 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
13109 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
13111 PR rtl-optimization/57662
13112 * sel-sched.c (code_motion_path_driver): Do not mark already not
13113 existing blocks in the visiting bitmap.
13115 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
13117 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
13118 on the insn being emitted.
13120 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
13121 Will Deacon <will.deacon@arm.com>
13123 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
13125 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13127 * config/arm/arm-tables.opt: Regenerate.
13129 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13131 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
13132 for vector types other than V16QImode.
13133 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
13134 define_expand, and call altivec_expand_vec_perm_le when producing
13135 code with little endian element order.
13136 (*altivec_vperm_<mode>_internal): New insn having previous
13137 behavior of altivec_vperm_<mode>.
13138 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
13139 altivec_expand_vec_perm_le when producing code with little endian
13141 (*altivec_vperm_<mode>_uns_internal): New insn having previous
13142 behavior of altivec_vperm_<mode>_uns.
13144 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13146 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
13147 (altivec_vsumsws): Add handling for -maltivec=be with a little
13149 (altivec_vsumsws_direct): New.
13150 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
13151 gen_altivec_vsumsws.
13153 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
13155 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
13156 vtable_pointer_value_to_binfo): New functions.
13157 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
13158 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
13160 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
13162 * config/nios2/nios2.md (load_got_register): Initialize GOT
13163 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
13164 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
13166 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
13168 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
13169 preserverd by passthrough, do not propagate the type.
13171 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
13173 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
13174 (mips_atomic_assign_expand_fenv): New function.
13175 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
13177 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
13179 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
13180 (__builtin_mips_set_fcsr): Likewise.
13181 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
13182 MIPS_USI_FTYPE_VOID.
13183 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
13184 (mips16_expand_set_fcsr): Likewise.
13185 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
13186 (mips16_set_fcsr_stub): Likewise.
13187 (mips16_get_fcsr_one_only_stub): New class.
13188 (mips16_set_fcsr_one_only_stub): Likewise.
13189 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
13190 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
13191 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
13192 (hard_float): New availability predicate.
13193 (mips_builtins): Add get_fcsr and set_fcsr.
13194 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
13195 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
13196 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
13197 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
13198 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
13201 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
13203 * config/mips/mips.c (mips_one_only_stub): New class.
13204 (mips_need_mips16_rdhwr_p): Replace with...
13205 (mips16_rdhwr_stub): ...this new variable.
13206 (mips16_stub_call_address): New function.
13207 (mips16_rdhwr_one_only_stub): New class.
13208 (mips_expand_thread_pointer): Use mips16_stub_call_address.
13209 (mips_output_mips16_rdhwr): Delete.
13210 (mips_finish_stub): New function.
13211 (mips_code_end): Use it to handle rdhwr stubs.
13213 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
13216 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
13217 when calculating size of integer atomic types.
13219 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
13221 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
13223 2014-02-01 Jakub Jelinek <jakub@redhat.com>
13225 PR tree-optimization/60003
13226 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
13227 * profile.c (branch_prob): Use gimple_call_builtin_p
13228 to check for BUILT_IN_SETJMP_RECEIVER.
13229 * tree-inline.c (copy_bb): Call notice_special_calls.
13231 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
13234 * lra-constraints.c (process_alt_operands): Update reload_sum only
13237 2014-01-31 Richard Henderson <rth@redhat.com>
13239 PR middle-end/60004
13240 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
13241 until after else_eh is processed.
13243 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
13245 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
13246 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
13247 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
13248 in smmintrin.h, remove them.
13249 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
13250 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
13251 * config/i386/i386.md (ROUND_SAE): Fix value.
13252 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
13253 (const48_operand): New.
13254 * config/i386/subst.md (round), (round_expand): Use
13255 const_4_or_8_to_11_operand.
13256 (round_saeonly), (round_saeonly_expand): Use const48_operand.
13258 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
13260 * config/i386/constraints.md (Yk): Swap meaning with k.
13261 * config/i386/i386.md (movhi_internal): Change Yk to k.
13262 (movqi_internal): Ditto.
13263 (*k<logic><mode>): Ditto.
13266 (kandn<mode>): Ditto.
13267 (*<code>hi_1): Ditto.
13268 (*<code>qi_1): Ditto.
13269 (kxnor<mode>): Ditto.
13270 (kortestzhi): Ditto.
13271 (kortestchi): Ditto.
13273 (*one_cmplhi2_1): Ditto.
13274 (*one_cmplqi2_1): Ditto.
13275 * config/i386/sse.md (): Change k to Yk.
13276 (avx512f_load<mode>_mask): Ditto.
13277 (avx512f_blendm<mode>): Ditto.
13278 (avx512f_store<mode>_mask): Ditto.
13279 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
13280 (avx512f_storedqu<mode>_mask): Ditto.
13281 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
13283 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
13284 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
13285 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
13286 (avx512f_maskcmp<mode>3): Ditto.
13287 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
13288 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
13289 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
13290 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
13291 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
13292 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
13293 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
13294 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
13295 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
13296 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
13297 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
13298 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
13299 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
13300 (vec_extract_lo_<mode>_maskm): Ditto.
13301 (vec_extract_hi_<mode>_maskm): Ditto.
13302 (avx512f_vternlog<mode>_mask): Ditto.
13303 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
13304 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
13305 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
13306 (avx512f_<code>v8div16qi2_mask): Ditto.
13307 (avx512f_<code>v8div16qi2_mask_store): Ditto.
13308 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
13309 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
13310 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
13311 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
13312 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
13313 (*avx512pf_gatherpf<mode>df_mask): Ditto.
13314 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
13315 (*avx512pf_scatterpf<mode>df_mask): Ditto.
13316 (avx512cd_maskb_vec_dupv8di): Ditto.
13317 (avx512cd_maskw_vec_dupv16si): Ditto.
13318 (avx512f_vpermi2var<mode>3_maskz): Ditto.
13319 (avx512f_vpermi2var<mode>3_mask): Ditto.
13320 (avx512f_vpermi2var<mode>3_mask): Ditto.
13321 (avx512f_vpermt2var<mode>3_maskz): Ditto.
13322 (*avx512f_gathersi<mode>): Ditto.
13323 (*avx512f_gathersi<mode>_2): Ditto.
13324 (*avx512f_gatherdi<mode>): Ditto.
13325 (*avx512f_gatherdi<mode>_2): Ditto.
13326 (*avx512f_scattersi<mode>): Ditto.
13327 (*avx512f_scatterdi<mode>): Ditto.
13328 (avx512f_compress<mode>_mask): Ditto.
13329 (avx512f_compressstore<mode>_mask): Ditto.
13330 (avx512f_expand<mode>_mask): Ditto.
13331 * config/i386/subst.md (mask): Change k to Yk.
13332 (mask_scalar_merge): Ditto.
13335 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
13337 * doc/extend.texi (Vector Extensions): Document ?: in C++.
13339 2014-01-31 Richard Biener <rguenther@suse.de>
13341 PR middle-end/59990
13342 * builtins.c (fold_builtin_memory_op): Make sure to not
13343 use a floating-point mode or a boolean or enumeral type for
13344 the copy operation.
13346 2014-01-30 DJ Delorie <dj@redhat.com>
13348 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
13349 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
13350 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
13351 whenever main() has an epilogue.
13353 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13355 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
13356 unused variable "field".
13357 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
13358 (vsx_mergeh_<mode>): Likewise.
13359 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
13360 (altivec_vmrghh): Likewise.
13361 (altivec_vmrghw): Likewise.
13362 (altivec_vmrglb): Likewise.
13363 (altivec_vmrglh): Likewise.
13364 (altivec_vmrglw): Likewise.
13365 (altivec_vspltb): Add missing uses.
13366 (altivec_vsplth): Likewise.
13367 (altivec_vspltw): Likewise.
13368 (altivec_vspltsf): Likewise.
13370 2014-01-30 Jakub Jelinek <jakub@redhat.com>
13373 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
13374 frame related instructions.
13376 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
13378 PR rtl-optimization/59959
13379 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
13380 any reload of register whose subreg is invalid.
13382 2014-01-30 Jakub Jelinek <jakub@redhat.com>
13384 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
13385 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
13386 Add missing return type - void.
13388 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13390 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
13391 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
13392 remove element index adjustment for endian (now handled in vsx.md
13394 (altivec_expand_vec_perm_const): Use
13395 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
13396 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
13397 (vsx_xxspltw_<mode>): Adjust element index for little endian.
13398 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
13399 define_expand and a new define_insn *altivec_vspltb_internal;
13400 adjust for -maltivec=be on a little endian target.
13401 (altivec_vspltb_direct): New.
13402 (altivec_vsplth): Divide into a define_expand and a new
13403 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
13404 little endian target.
13405 (altivec_vsplth_direct): New.
13406 (altivec_vspltw): Divide into a define_expand and a new
13407 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
13408 little endian target.
13409 (altivec_vspltw_direct): New.
13410 (altivec_vspltsf): Divide into a define_expand and a new
13411 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
13412 a little endian target.
13414 2014-01-30 Richard Biener <rguenther@suse.de>
13416 PR tree-optimization/59993
13417 * tree-ssa-forwprop.c (associate_pointerplus): Check we
13418 can propagate form the earlier stmt and avoid the transform
13419 when the intermediate result is needed.
13421 2014-01-30 Alangi Derick <alangiderick@gmail.com>
13423 * README.Portability: Fix typo.
13425 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
13427 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
13428 comparison_operator with ordered_comparison_operator.
13430 2014-01-30 Nick Clifton <nickc@redhat.com>
13432 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
13433 Rename to mn10300_store_multiple_regs.
13434 * config/mn10300/mn10300.c: Likewise.
13435 * config/mn10300/mn10300.md (store_movm): Fix typo: call
13436 store_multiple_regs.
13437 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
13438 Call mn10300_store_multiple_regs.
13440 2014-01-30 Nick Clifton <nickc@redhat.com>
13441 DJ Delorie <dj@redhat.com>
13443 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
13444 %fp 2 to keep registers after it properly word-aligned.
13445 (rl78_alloc_physical_registers_umul): Handle the case where both
13446 input operands are the same.
13448 2014-01-30 Richard Biener <rguenther@suse.de>
13450 PR tree-optimization/59903
13451 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
13454 2014-01-30 Jason Merrill <jason@redhat.com>
13457 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
13460 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
13462 2014-01-30 Richard Biener <rguenther@suse.de>
13464 PR tree-optimization/59951
13465 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
13467 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
13470 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
13471 SFmode to DFmode case.
13473 2014-01-29 DJ Delorie <dj@redhat.com>
13475 * config/msp430/msp430.opt (-minrt): New.
13476 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
13478 (ENDFILE_SPEC): Likewise.
13480 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
13482 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
13483 (estimate_function_body_sizes): Use it.
13485 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
13488 * dwarf2out.c (is_cxx_auto): New.
13489 (is_base_type): Use it.
13490 (gen_type_die_with_usage): Likewise.
13492 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13494 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
13495 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
13496 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
13497 -maltivec=be with LE targets.
13498 (vsx_mergeh_<mode>): Likewise.
13499 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
13500 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
13501 (altivec_vmrghb): Replace with define_expand and new
13502 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
13503 (altivec_vmrghb_direct): New define_insn.
13504 (altivec_vmrghh): Replace with define_expand and new
13505 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
13506 (altivec_vmrghh_direct): New define_insn.
13507 (altivec_vmrghw): Replace with define_expand and new
13508 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
13509 (altivec_vmrghw_direct): New define_insn.
13510 (*altivec_vmrghsf): Adjust for endianness.
13511 (altivec_vmrglb): Replace with define_expand and new
13512 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
13513 (altivec_vmrglb_direct): New define_insn.
13514 (altivec_vmrglh): Replace with define_expand and new
13515 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
13516 (altivec_vmrglh_direct): New define_insn.
13517 (altivec_vmrglw): Replace with define_expand and new
13518 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
13519 (altivec_vmrglw_direct): New define_insn.
13520 (*altivec_vmrglsf): Adjust for endianness.
13521 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
13522 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
13523 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
13524 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
13525 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
13526 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
13527 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
13528 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
13530 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
13532 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
13533 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
13536 2014-01-29 Richard Biener <rguenther@suse.de>
13538 PR tree-optimization/58742
13539 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
13540 associate_pointerplus_align.
13541 (associate_pointerplus_diff): New function.
13542 (associate_pointerplus): Likewise. Call associate_pointerplus_align
13543 and associate_pointerplus_diff.
13545 2014-01-29 Richard Biener <rguenther@suse.de>
13547 * lto-streamer.h (LTO_major_version): Bump to 3.
13548 (LTO_minor_version): Reset to 0.
13550 2014-01-29 Renlin Li <Renlin.Li@arm.com>
13552 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
13553 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
13554 (arm_file_start): Generate correct asm header for armv7ve.
13555 * config/arm/bpabi.h: Add multilib support for armv7ve.
13556 * config/arm/driver-arm.c: Change the architectures of cortex-a7
13557 and cortex-a15 to armv7ve.
13558 * config/arm/t-aprofile: Add multilib support for armv7ve.
13559 * doc/invoke.texi: Document -march=armv7ve.
13561 2014-01-29 Richard Biener <rguenther@suse.de>
13563 PR tree-optimization/58742
13564 * tree-ssa-forwprop.c (associate_plusminus): Return true
13565 if we changed sth, defer EH cleanup to ...
13566 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
13567 (simplify_mult): New function.
13569 2014-01-29 Jakub Jelinek <jakub@redhat.com>
13571 PR middle-end/59917
13572 PR tree-optimization/59920
13573 * tree.c (build_common_builtin_nodes): Remove
13574 __builtin_setjmp_dispatcher initialization.
13575 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
13576 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
13577 instead of gsi_after_labels + manually skipping debug stmts.
13578 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
13579 ignore bbs with IFN_ABNORMAL_DISPATCHER.
13580 * tree-inline.c (copy_edges_for_bb): Remove
13581 can_make_abnormal_goto argument, instead add abnormal_goto_dest
13582 argument. Ignore computed_goto_p stmts. Don't call
13583 make_abnormal_goto_edges. If a call might need abnormal edges
13584 for non-local gotos, see if it already has an edge to
13585 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
13586 with true argument, don't do anything then, otherwise add
13587 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
13588 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
13590 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
13591 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
13592 (lower_stmt): Don't set data->calls_builtin_setjmp.
13593 (lower_builtin_setjmp): Adjust comment.
13594 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
13595 * tree-cfg.c (found_computed_goto): Remove.
13596 (factor_computed_gotos): Remove.
13597 (make_goto_expr_edges): Return bool, true for computed gotos.
13598 Don't call make_abnormal_goto_edges.
13599 (build_gimple_cfg): Don't set found_computed_goto, don't call
13600 factor_computed_gotos.
13601 (computed_goto_p): No longer static.
13602 (make_blocks): Don't set found_computed_goto.
13603 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
13604 (make_edges): If make_goto_expr_edges returns true, push bb
13605 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
13606 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
13607 vector. Record mapping between bbs and OpenMP regions if there
13608 are any, adjust make_gimple_omp_edges caller. Call
13609 handle_abnormal_edges.
13610 (make_abnormal_goto_edges): Remove.
13611 * tree-cfg.h (make_abnormal_goto_edges): Remove.
13612 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
13613 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
13614 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
13615 * internal-fn.def (ABNORMAL_DISPATCHER): New.
13616 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
13617 filling *region also set *region_idx to (*region)->entry->index.
13620 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
13621 For REGs set ORIGINAL_REGNO.
13623 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
13625 * doc/md.texi: Mention that a target shouldn't implement
13626 vec_widen_(s|u)mul_even/odd pair if it is less efficient
13629 2014-01-29 Jakub Jelinek <jakub@redhat.com>
13631 PR tree-optimization/59594
13632 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
13633 a copy of the datarefs vector rather than the vector itself.
13635 2014-01-28 Jason Merrill <jason@redhat.com>
13638 * dwarf2out.c (auto_die): New static.
13639 (gen_type_die_with_usage): Handle C++1y 'auto'.
13640 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
13643 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
13646 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
13647 (SPEC_X32): Likewise.
13648 (SPEC_64): Likewise.
13649 * config/i386/i386.c (ix86_option_override_internal): Turn off
13650 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
13652 (x86_file_start): Output .code16gcc for TARGET_16BIT.
13653 * config/i386/i386.h (TARGET_16BIT): New macro.
13654 (TARGET_16BIT_P): Likewise.
13655 * config/i386/i386.opt: Add m16.
13656 * doc/invoke.texi: Document -m16.
13658 2014-01-28 Jakub Jelinek <jakub@redhat.com>
13660 PR preprocessor/59935
13661 * input.c (location_get_source_line): Bail out on when line number
13662 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
13664 2014-01-28 Richard Biener <rguenther@suse.de>
13666 PR tree-optimization/58742
13667 * tree-ssa-forwprop.c (associate_plusminus): Handle
13668 pointer subtraction of the form (T)(P + A) - (T)P.
13670 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13672 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
13675 2014-01-28 Richard Biener <rguenther@suse.de>
13678 2014-01-28 Richard Biener <rguenther@suse.de>
13680 PR rtl-optimization/45364
13681 PR rtl-optimization/59890
13682 * var-tracking.c (local_get_addr_clear_given_value): Handle
13683 already cleared slot.
13684 (val_reset): Handle not allocated local_get_addr_cache.
13685 (vt_find_locations): Use post-order on the inverted CFG.
13687 2014-01-28 Richard Biener <rguenther@suse.de>
13689 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
13691 2014-01-28 Richard Biener <rguenther@suse.de>
13693 PR rtl-optimization/45364
13694 PR rtl-optimization/59890
13695 * var-tracking.c (local_get_addr_clear_given_value): Handle
13696 already cleared slot.
13697 (val_reset): Handle not allocated local_get_addr_cache.
13698 (vt_find_locations): Use post-order on the inverted CFG.
13700 2014-01-28 Alan Modra <amodra@gmail.com>
13702 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
13703 * configure.ac <recursive call for build != host>: Define
13704 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
13705 and LD_FOR_BUILD too.
13706 * configure: Regenerate.
13708 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
13710 * config/i386/i386.c (get_builtin_code_for_version): Separate
13711 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
13712 Broadwell from Haswell.
13714 2014-01-27 Steve Ellcey <sellcey@mips.com>
13716 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
13717 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
13718 * config/mips/mips.c (mips_option_override): Change setting
13720 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
13721 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
13722 Change from Mask to Var.
13724 2014-01-27 Jeff Law <law@redhat.com>
13726 * ipa-inline.c (inline_small_functions): Fix typo.
13728 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
13730 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
13731 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
13732 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
13733 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
13734 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
13735 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
13736 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
13737 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
13738 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
13739 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
13740 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
13741 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
13742 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
13743 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
13744 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
13745 (_mm512_storeu_epi64): Ditto.
13746 (_mm512_cmpge_epi32_mask): Ditto.
13747 (_mm512_cmpge_epu32_mask): Ditto.
13748 (_mm512_cmpge_epi64_mask): Ditto.
13749 (_mm512_cmpge_epu64_mask): Ditto.
13750 (_mm512_cmple_epi32_mask): Ditto.
13751 (_mm512_cmple_epu32_mask): Ditto.
13752 (_mm512_cmple_epi64_mask): Ditto.
13753 (_mm512_cmple_epu64_mask): Ditto.
13754 (_mm512_cmplt_epi32_mask): Ditto.
13755 (_mm512_cmplt_epu32_mask): Ditto.
13756 (_mm512_cmplt_epi64_mask): Ditto.
13757 (_mm512_cmplt_epu64_mask): Ditto.
13758 (_mm512_cmpneq_epi32_mask): Ditto.
13759 (_mm512_cmpneq_epu32_mask): Ditto.
13760 (_mm512_cmpneq_epi64_mask): Ditto.
13761 (_mm512_cmpneq_epu64_mask): Ditto.
13762 (_mm512_expand_pd): Ditto.
13763 (_mm512_expand_ps): Ditto.
13764 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
13765 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
13766 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
13767 * config/i386/i386.c (ix86_builtins): Add
13768 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
13769 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
13770 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
13771 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
13772 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
13773 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
13774 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
13775 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
13776 IX86_BUILTIN_PMOVUSQW512_MEM.
13777 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
13778 __builtin_ia32_pmovsqd512mem_mask,
13779 __builtin_ia32_pmovqd512mem_mask,
13780 __builtin_ia32_pmovusqw512mem_mask,
13781 __builtin_ia32_pmovsqw512mem_mask,
13782 __builtin_ia32_pmovqw512mem_mask,
13783 __builtin_ia32_pmovusdw512mem_mask,
13784 __builtin_ia32_pmovsdw512mem_mask,
13785 __builtin_ia32_pmovdw512mem_mask,
13786 __builtin_ia32_pmovqb512mem_mask,
13787 __builtin_ia32_pmovusqb512mem_mask,
13788 __builtin_ia32_pmovsqb512mem_mask,
13789 __builtin_ia32_pmovusdb512mem_mask,
13790 __builtin_ia32_pmovsdb512mem_mask,
13791 __builtin_ia32_pmovdb512mem_mask.
13792 (bdesc_args): Add __builtin_ia32_expanddf512,
13793 __builtin_ia32_expandsf512.
13794 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
13795 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
13796 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
13797 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
13798 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
13799 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
13800 (avx512f_<code>v8div16qi2_mask_store): This.
13801 (avx512f_expand<mode>): New.
13803 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
13805 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
13807 (_mm512_mask_prefetch_i64gather_pd): Ditto.
13808 (_mm512_prefetch_i32scatter_pd): Ditto.
13809 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
13810 (_mm512_prefetch_i64scatter_pd): Ditto.
13811 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
13812 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
13813 (_mm512_mask_prefetch_i64gather_ps): Ditto.
13814 (_mm512_prefetch_i32scatter_ps): Ditto.
13815 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
13816 (_mm512_prefetch_i64scatter_ps): Ditto.
13817 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
13818 * config/i386/i386-builtin-types.def: Define
13819 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
13820 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
13821 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
13822 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
13823 IX86_BUILTIN_SCATTERPFQPD.
13824 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
13825 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
13826 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
13827 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
13828 __builtin_ia32_scatterpfqps.
13829 (ix86_expand_builtin): Expand new built-ins.
13830 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
13831 fix memory access data type.
13832 (*avx512pf_gatherpf<mode>_mask): Ditto.
13833 (*avx512pf_gatherpf<mode>): Ditto.
13834 (avx512pf_scatterpf<mode>): Ditto.
13835 (*avx512pf_scatterpf<mode>_mask): Ditto.
13836 (*avx512pf_scatterpf<mode>): Ditto.
13837 (GATHER_SCATTER_SF_MEM_MODE): New.
13838 (avx512pf_gatherpf<mode>df): Ditto.
13839 (*avx512pf_gatherpf<mode>df_mask): Ditto.
13840 (*avx512pf_scatterpf<mode>df): Ditto.
13842 2014-01-27 Jakub Jelinek <jakub@redhat.com>
13845 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
13846 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
13849 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
13851 * common/config/arm/arm-common.c
13852 (arm_rewrite_mcpu): Handle multiple names.
13854 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
13856 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
13858 * gimple-builder.h (create_gimple_tmp): Delete.
13860 2014-01-27 Christian Bruel <christian.bruel@st.com>
13862 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
13865 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
13867 * config/pa/pa.md (call): Generate indirect long calls to non-local
13868 functions when outputing 32-bit code.
13869 (call_value): Likewise except for special call to buggy powf function.
13871 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
13872 portable runtime and PIC indirect calls.
13873 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
13874 and PIC call sequences. Use ldo instead of blr to set return register
13875 in PIC call sequence.
13877 2014-01-25 Walter Lee <walt@tilera.com>
13879 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
13880 avoid clobbering a live register.
13882 2014-01-25 Walter Lee <walt@tilera.com>
13884 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
13885 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
13886 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
13887 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
13889 2014-01-25 Walter Lee <walt@tilera.com>
13891 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
13892 arguments on even registers.
13893 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
13895 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
13896 (BIGGEST_ALIGNMENT): Ditto.
13897 (BIGGEST_FIELD_ALIGNMENT): Ditto.
13899 2014-01-25 Walter Lee <walt@tilera.com>
13901 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
13902 insns before bundling.
13903 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
13905 2014-01-25 Walter Lee <walt@tilera.com>
13907 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
13908 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
13909 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
13911 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
13913 * config/mips/constraints.md (kl): Delete.
13914 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
13915 define expands, using...
13916 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
13917 instructions for MIPS16.
13918 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
13919 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
13921 2014-01-25 Walter Lee <walt@tilera.com>
13923 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
13927 2014-01-25 Walter Lee <walt@tilera.com>
13929 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
13930 (TARGET_EXPAND_TO_RTL_HOOK): Define.
13932 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
13934 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
13937 2014-01-25 Jakub Jelinek <jakub@redhat.com>
13939 * print-rtl.c (in_call_function_usage): New var.
13940 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
13941 EXPR_LIST mode as mode and not as reg note name.
13943 PR middle-end/59561
13944 * cfgloopmanip.c (copy_loop_info): If
13945 loop->warned_aggressive_loop_optimizations, make sure
13946 the flag is set in target loop too.
13948 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
13950 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
13952 * builtins.def: Likewise.
13953 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
13954 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
13955 * ira.c (ira_setup_eliminable_regset): Likewise.
13956 * omp-low.c (gate_expand_omp): Likewise.
13957 (execute_lower_omp): Likewise.
13958 (diagnose_sb_0): Likewise.
13959 (gate_diagnose_omp_blocks): Likewise.
13960 (simd_clone_clauses_extract): Likewise.
13963 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13965 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
13966 correction for little endian...
13967 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
13970 2014-01-24 Jeff Law <law@redhat.com>
13972 PR tree-optimization/59919
13973 * tree-vrp.c (find_assert_locations_1): Do not register asserts
13974 for non-returning calls.
13976 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
13978 * common/config/aarch64/aarch64-common.c
13979 (aarch64_rewrite_mcpu): Handle multiple names.
13980 * config/aarch64/aarch64.h
13981 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
13983 2014-01-24 Dodji Seketeli <dodji@redhat.com>
13985 * input.c (add_file_to_cache_tab): Handle the case where fopen
13988 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
13991 * config/i386/i386.md (pushsf splitter): Get stack adjustment
13992 from push operand if code of push isn't PRE_DEC.
13994 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13997 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
13998 -mquad-memory-atomic. Update -mquad-memory documentation to say
13999 it is only used for non-atomic loads/stores.
14001 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
14002 -mquad-memory or -mquad-memory-atomic switches.
14004 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
14005 -mquad-memory-atomic to ISA 2.07 support.
14007 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
14008 to separate support of normal quad word memory operations (ldq, stq)
14009 from the atomic quad word memory operations.
14011 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
14012 support to separate non-atomic quad word operations from atomic
14013 quad word operations. Disable non-atomic quad word operations in
14014 little endian mode so that we don't have to swap words after the
14015 load and before the store.
14016 (quad_load_store_p): Add comment about atomic quad word support.
14017 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
14018 options printed with -mdebug=reg.
14020 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
14021 -mquad-memory-atomic as the test for whether we have quad word
14022 atomic instructions.
14023 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
14024 or -mp8-vector are used, allow byte/half-word atomic operations.
14026 * config/rs6000/sync.md (load_lockedti): Insure that the address
14027 is a proper indexed or indirect address for the lqarx instruction.
14028 On little endian systems, swap the hi/lo registers after the lqarx
14030 (load_lockedpti): Use indexed_or_indirect_operand predicate to
14031 insure the address is valid for the lqarx instruction.
14032 (store_conditionalti): Insure that the address is a proper indexed
14033 or indirect address for the stqcrx. instruction. On little endian
14034 systems, swap the hi/lo registers before doing the stqcrx.
14036 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
14037 insure the address is valid for the stqcrx. instruction.
14039 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
14040 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
14041 type of quad memory support is available.
14043 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
14045 PR regression/59915
14046 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
14047 there is a danger of looping.
14049 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
14051 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
14052 force flag_ira_loop_pressure if set via command line.
14054 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
14056 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
14057 (ashr_simd): New builtin handling DI mode.
14058 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
14059 (aarch64_sshr_simddi): New match pattern.
14060 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
14061 (vshrd_n_s64): Likewise.
14062 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
14064 2014-01-23 Nick Clifton <nickc@redhat.com>
14066 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
14067 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
14068 favour of mcu specific scripts.
14069 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
14072 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
14073 Alex Velenko <Alex.Velenko@arm.com>
14075 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
14076 (vaddv_s16): Likewise.
14077 (vaddv_s32): Likewise.
14078 (vaddv_u8): Likewise.
14079 (vaddv_u16): Likewise.
14080 (vaddv_u32): Likewise.
14081 (vaddvq_s8): Likewise.
14082 (vaddvq_s16): Likewise.
14083 (vaddvq_s32): Likewise.
14084 (vaddvq_s64): Likewise.
14085 (vaddvq_u8): Likewise.
14086 (vaddvq_u16): Likewise.
14087 (vaddvq_u32): Likewise.
14088 (vaddvq_u64): Likewise.
14089 (vaddv_f32): Likewise.
14090 (vaddvq_f32): Likewise.
14091 (vaddvq_f64): Likewise.
14092 (vmaxv_f32): Likewise.
14093 (vmaxv_s8): Likewise.
14094 (vmaxv_s16): Likewise.
14095 (vmaxv_s32): Likewise.
14096 (vmaxv_u8): Likewise.
14097 (vmaxv_u16): Likewise.
14098 (vmaxv_u32): Likewise.
14099 (vmaxvq_f32): Likewise.
14100 (vmaxvq_f64): Likewise.
14101 (vmaxvq_s8): Likewise.
14102 (vmaxvq_s16): Likewise.
14103 (vmaxvq_s32): Likewise.
14104 (vmaxvq_u8): Likewise.
14105 (vmaxvq_u16): Likewise.
14106 (vmaxvq_u32): Likewise.
14107 (vmaxnmv_f32): Likewise.
14108 (vmaxnmvq_f32): Likewise.
14109 (vmaxnmvq_f64): Likewise.
14110 (vminv_f32): Likewise.
14111 (vminv_s8): Likewise.
14112 (vminv_s16): Likewise.
14113 (vminv_s32): Likewise.
14114 (vminv_u8): Likewise.
14115 (vminv_u16): Likewise.
14116 (vminv_u32): Likewise.
14117 (vminvq_f32): Likewise.
14118 (vminvq_f64): Likewise.
14119 (vminvq_s8): Likewise.
14120 (vminvq_s16): Likewise.
14121 (vminvq_s32): Likewise.
14122 (vminvq_u8): Likewise.
14123 (vminvq_u16): Likewise.
14124 (vminvq_u32): Likewise.
14125 (vminnmv_f32): Likewise.
14126 (vminnmvq_f32): Likewise.
14127 (vminnmvq_f64): Likewise.
14129 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
14131 * config/aarch64/aarch64-simd.md
14132 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
14133 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
14134 (*aarch64_mul3_elt<mode>): Likewise.
14135 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
14136 (*aarch64_mul3_elt_to_64v2df): Likewise.
14137 (*aarch64_mla_elt<mode>): Likewise.
14138 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
14139 (*aarch64_mls_elt<mode>): Likewise.
14140 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
14141 (*aarch64_fma4_elt<mode>): Likewise.
14142 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
14143 (*aarch64_fma4_elt_to_64v2df): Likewise.
14144 (*aarch64_fnma4_elt<mode>): Likewise.
14145 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
14146 (*aarch64_fnma4_elt_to_64v2df): Likewise.
14147 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
14148 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
14149 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
14150 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
14151 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
14152 (aarch64_sqdmull_lane<mode>_internal): Likewise.
14153 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
14155 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
14157 * config/aarch64/aarch64-simd.md
14158 (aarch64_be_checked_get_lane<mode>): New define_expand.
14159 * config/aarch64/aarch64-simd-builtins.def
14160 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
14161 New builtin definition.
14162 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
14163 Use new safe be builtin.
14165 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
14167 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
14169 (aarch64_be_st1<mode>): Likewise.
14170 (aarch_ld1<VALL:mode>): Define_expand modified.
14171 (aarch_st1<VALL:mode>): Likewise.
14172 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
14173 (UNSPEC_ST1): Likewise.
14175 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
14177 * config/microblaze/microblaze.md: Add trap insn and attribute
14179 2014-01-23 Dodji Seketeli <dodji@redhat.com>
14181 PR preprocessor/58580
14182 * input.h (location_get_source_line): Take an additional line_size
14184 (void diagnostics_file_cache_fini): Declare new function.
14185 * input.c (struct fcache): New type.
14186 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
14187 New static constants.
14188 (diagnostic_file_cache_init, total_lines_num)
14189 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
14190 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
14191 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
14192 (get_next_line, read_next_line, goto_next_line, read_line_num):
14193 New static function definitions.
14194 (diagnostic_file_cache_fini): New function.
14195 (location_get_source_line): Take an additional output line_len
14196 parameter. Re-write using lookup_or_add_file_to_cache_tab and
14198 * diagnostic.c (diagnostic_finish): Call
14199 diagnostic_file_cache_fini.
14200 (adjust_line): Take an additional input parameter for the length
14201 of the line, rather than calculating it with strlen.
14202 (diagnostic_show_locus): Adjust the use of
14203 location_get_source_line and adjust_line with respect to their new
14204 signature. While displaying a line now, do not stop at the first
14205 null byte. Rather, display the zero byte as a space and keep
14206 going until we reach the size of the line.
14207 * Makefile.in: Add vec.o to OBJS-libcommon
14209 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
14210 Ilya Tocar <ilya.tocar@intel.com>
14212 * config/i386/avx512fintrin.h (_mm512_kmov): New.
14213 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
14214 (__builtin_ia32_kmov16): Ditto.
14215 * config/i386/i386.md (UNSPEC_KMOV): New.
14218 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
14220 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
14221 (_mm512_storeu_si512): Ditto.
14223 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
14226 * rtl.h (get_referenced_operands): Declare.
14227 * recog.c (get_referenced_operands): New function.
14228 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
14229 operands have been referenced when recording LO_SUM references.
14231 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
14233 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
14235 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
14237 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
14238 Enable for generic and recent AMD targets.
14240 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
14242 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
14243 ARG_SIZE note when adjustment was eliminated.
14245 2014-01-22 Jeff Law <law@redhat.com>
14247 PR tree-optimization/59597
14248 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
14249 in file. Accept new argument REGISTERING and use it to modify
14250 dump output appropriately.
14251 (register_jump_thread): Corresponding changes.
14252 (mark_threaded_blocks): Reinstate code to cancel unprofitable
14253 thread paths involving joiner blocks. Add code to dump cancelled
14254 jump threading paths.
14256 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
14258 PR rtl-optimization/59477
14259 * lra-constraints.c (inherit_in_ebb): Process call for living hard
14260 regs. Update reloads_num and potential_reload_hard_regs for all insns.
14262 2014-01-22 Tom Tromey <tromey@redhat.com>
14264 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
14266 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
14268 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
14270 PR rtl-optimization/59896
14271 * lra-constraints.c (process_alt_operands): Check unused note for
14272 matched operands of insn with no output reloads.
14274 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
14276 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
14277 (mips_move_from_gpr_cost): Likewise.
14279 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
14281 PR rtl-optimization/59858
14282 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
14283 ira_class_hard_regs_num.
14284 (process_alt_operands): Increase reject for dying matched operand.
14286 2014-01-21 Jakub Jelinek <jakub@redhat.com>
14289 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
14290 smaller than size, perform several stores or loads and stores
14291 at dst + count - size to store or copy all of size bytes, rather
14292 than just last modesize bytes.
14294 2014-01-20 DJ Delorie <dj@redhat.com>
14296 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
14297 that CLOBBERs are REGs before propogating their values.
14299 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
14301 PR middle-end/59789
14302 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
14303 (cgraph_inline_failed_type): New function.
14304 * cgraph.h (DEFCIFCODE): Add type.
14305 (cgraph_inline_failed_type_t): New enum.
14306 (cgraph_inline_failed_type): New prototype.
14307 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
14308 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
14309 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
14310 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
14311 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
14312 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
14313 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
14314 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
14315 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
14316 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
14317 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
14318 OPTIMIZATION_MISMATCH.
14319 * tree-inline.c (expand_call_inline): Emit errors during
14320 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
14322 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
14325 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
14326 mode attribute in insn output.
14328 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
14330 * output.h (output_constant): Delete.
14331 * varasm.c (output_constant): Make private.
14333 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
14335 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
14337 2014-01-20 Jakub Jelinek <jakub@redhat.com>
14339 PR middle-end/59860
14340 * tree.h (fold_builtin_strcat): New prototype.
14341 * builtins.c (fold_builtin_strcat): No longer static. Add len
14342 argument, if non-NULL, don't call c_strlen. Optimize
14343 directly into __builtin_memcpy instead of __builtin_strcpy.
14344 (fold_builtin_2): Adjust fold_builtin_strcat caller.
14345 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
14347 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
14349 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
14350 for SImode_address_operand operands, having only a REG argument.
14352 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
14354 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
14355 loader name using mbig-endian.
14356 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
14358 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
14360 * doc/invoke.texi (-march): Clarify documentation for AArch64.
14361 (-mtune): Likewise.
14364 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
14366 * config/aarch64/aarch64-protos.h
14367 (aarch64_cannot_change_mode_class_ptr): Declare.
14368 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
14369 aarch64_cannot_change_mode_class_ptr): New.
14370 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
14371 backend hook aarch64_cannot_change_mode_class.
14373 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
14375 * common/config/aarch64/aarch64-common.c
14376 (aarch64_handle_option): Don't handle any option order logic here.
14377 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
14378 selected_cpu, warn on architecture version mismatch.
14379 (aarch64_override_options): Fix parsing order for option strings.
14381 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
14382 Iain Sandoe <iain@codesourcery.com>
14385 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
14386 warning. Amend comment to reflect current functionality.
14388 2014-01-20 Richard Biener <rguenther@suse.de>
14390 PR middle-end/59860
14391 * builtins.c (fold_builtin_strcat): Remove case better handled
14392 by tree-ssa-strlen.c.
14394 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
14396 * config/aarch64/aarch64.opt
14397 (mcpu, march, mtune): Make case-insensitive.
14399 2014-01-20 Jakub Jelinek <jakub@redhat.com>
14402 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
14403 if operands[1] is a REG or ZERO_EXTEND of a REG.
14405 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
14407 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
14409 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
14411 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
14412 long non-pic millicode calls.
14414 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
14416 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
14418 2014-01-19 Kito Cheng <kito@0xlab.org>
14420 * builtins.c (expand_movstr): Check movstr expand done or fail.
14422 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
14423 H.J. Lu <hongjiu.lu@intel.com>
14426 * config/i386/i386.md (*lea<mode>): Zero-extend return register
14427 to DImode for zero-extended addresses.
14429 2014-01-19 Jakub Jelinek <jakub@redhat.com>
14431 PR rtl-optimization/57763
14432 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
14433 on the new indirect jump_insn and increment LABEL_NUSES (label).
14435 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
14439 * config.gcc (x86_archs): New variable.
14440 (x86_64_archs): Likewise.
14441 (x86_cpus): Likewise.
14442 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
14443 --with-arch/--with-cpu= options.
14444 Support --with-arch=/--with-cpu={nehalem,westmere,
14445 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
14447 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
14449 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
14450 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
14452 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
14454 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
14456 2014-01-18 Jakub Jelinek <jakub@redhat.com>
14459 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
14460 clear cpp_get_options (parse_in)->warn_unused_macros for
14461 ix86_target_macros_internal with cpp_define.
14463 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
14465 * jump.c (delete_related_insns): Keep (use (insn))s.
14466 * reorg.c (redundant_insn): Check for barriers too.
14468 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
14470 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
14472 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
14474 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
14475 call to $$dyncall when TARGET_LONG_CALLS is true.
14477 2014-01-17 Jeff Law <law@redhat.com>
14479 * ree.c (combine_set_extension): Temporarily disable test for
14480 changing number of hard registers.
14482 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
14484 PR middle-end/58125
14485 * ipa-inline-analysis.c (inline_free_summary):
14486 Do not free summary of aliases.
14488 2014-01-17 Jakub Jelinek <jakub@redhat.com>
14490 PR middle-end/59706
14491 * gimplify.c (gimplify_expr): Use create_tmp_var
14492 instead of create_tmp_var_raw. If cond doesn't have
14493 integral type, don't add the IFN_ANNOTATE builtin at all.
14495 2014-01-17 Martin Jambor <mjambor@suse.cz>
14498 * ipa-cp.c (prev_edge_clone): New variable.
14499 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
14500 Also resize prev_edge_clone vector.
14501 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
14502 (ipcp_edge_removal_hook): New function.
14503 (ipcp_driver): Register ipcp_edge_removal_hook.
14505 2014-01-17 Andrew Pinski <apinski@cavium.com>
14506 Steve Ellcey <sellcey@mips.com>
14509 * config/mips/mips.c (mips_print_operand): Check operand mode instead
14512 2014-01-17 Jeff Law <law@redhat.com>
14514 PR middle-end/57904
14515 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
14516 so that pass_ccp runs first.
14518 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
14520 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
14521 (ix86_adjust_cost): Use !TARGET_XXX.
14522 (do_reorder_for_imul): Likewise.
14523 (swap_top_of_ready_list): Likewise.
14524 (ix86_sched_reorder): Likewise.
14526 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
14528 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14529 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
14530 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
14531 (intel_memset): New. Duplicate slm_memset.
14532 (intel_cost): New. Duplicate slm_cost.
14533 (m_INTEL): New macro.
14534 (processor_target_table): Add "intel".
14535 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
14536 with PROCESSOR_INTEL for "intel".
14537 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
14538 PROCESSOR_SILVERMONT.
14539 (ix86_issue_rate): Likewise.
14540 (ix86_adjust_cost): Likewise.
14541 (ia32_multipass_dfa_lookahead): Likewise.
14542 (swap_top_of_ready_list): Likewise.
14543 (ix86_sched_reorder): Likewise.
14544 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
14545 instead of TARGET_OPT_AGU.
14546 * config/i386/i386.h (TARGET_INTEL): New.
14547 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
14548 (processor_type): Add PROCESSOR_INTEL.
14549 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
14550 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
14552 2014-01-17 Marek Polacek <polacek@redhat.com>
14555 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
14558 2014-01-17 Richard Biener <rguenther@suse.de>
14560 PR tree-optimization/46590
14561 * opts.c (default_options_table): Add entries for
14562 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
14563 all enabled at -O1 but not for -Og.
14564 * common.opt (fbranch-count-reg): Remove Init(1).
14565 (fmove-loop-invariants): Likewise.
14566 (ftree-pta): Likewise.
14568 2014-01-17 Jakub Jelinek <jakub@redhat.com>
14570 * config/i386/i386.c (ix86_data_alignment): For compatibility with
14571 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
14572 decls to at least the GCC 4.8 used alignments.
14575 * tree-nested.c (convert_nonlocal_reference_stmt,
14576 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
14577 of GIMPLE_BIND stmts, adjust associated decls.
14579 2014-01-17 Richard Biener <rguenther@suse.de>
14581 PR tree-optimization/46590
14582 * vec.h (vec<>::bseach): New member function implementing
14583 binary search according to C89 bsearch.
14584 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
14585 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
14586 bitmap pointer again. Make accesses_in_loop a flat array.
14587 (mem_ref_obstack): New global.
14588 (outermost_indep_loop): Adjust for mem_ref->stored changes.
14589 (mark_ref_stored): Likewise.
14590 (ref_indep_loop_p_2): Likewise.
14591 (set_ref_stored_in_loop): New helper function.
14592 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
14593 (memref_free): Adjust.
14594 (record_mem_ref_loc): Simplify.
14595 (gather_mem_refs_stmt): Adjust.
14596 (sort_locs_in_loop_postorder_cmp): New function.
14597 (analyze_memory_references): Sort accesses_in_loop after
14598 loop postorder number.
14599 (find_ref_loc_in_loop_cmp): New function.
14600 (for_all_locs_in_loop): Find relevant cluster of locs in
14601 accesses_in_loop and iterate without recursion.
14602 (execute_sm): Avoid uninit warning.
14603 (struct ref_always_accessed): Simplify.
14604 (ref_always_accessed::operator ()): Likewise.
14605 (ref_always_accessed_p): Likewise.
14606 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
14607 loop postorder numbers here.
14608 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
14611 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
14614 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
14615 on decls for which assemble_alias has been called.
14617 2014-01-17 Nick Clifton <nickc@redhat.com>
14619 * config/msp430/msp430.opt: (mcpu): New option.
14620 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
14621 (msp430_option_override): Parse target_cpu. If the MCU name
14622 matches a generic string, clear target_mcu.
14623 (msp430_attr): Allow numeric interrupt values up to 63.
14624 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
14625 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
14627 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
14629 * config/msp430/msp430.md (popm): Use %J rather than %I.
14630 (addsi3): Use msp430_nonimmediate_operand for operand 2.
14631 (addhi_cy_i): Use immediate_operand for operand 2.
14632 * doc/invoke.texi: Document -mcpu option.
14634 2014-01-17 Richard Biener <rguenther@suse.de>
14636 PR rtl-optimization/38518
14637 * df.h (df_analyze_loop): Declare.
14638 * df-core.c: Include cfgloop.h.
14639 (df_analyze_1): Split out main part of df_analyze.
14640 (df_analyze): Adjust.
14641 (loop_inverted_post_order_compute): New function.
14642 (loop_post_order_compute): Likewise.
14643 (df_analyze_loop): New function avoiding whole-function
14644 postorder computes.
14645 * loop-invariant.c (find_defs): Use df_analyze_loop.
14646 (find_invariants): Adjust.
14647 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
14649 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
14651 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
14652 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
14654 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
14656 * ipa-ref.c (ipa_remove_stmt_references): Fix references
14657 traversal when removing references.
14659 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
14662 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
14664 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
14666 PR middle-end/56791
14667 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
14668 pushing a reload for an autoinc when we had previously reloaded an
14669 inner part of the address.
14671 2014-01-16 Jakub Jelinek <jakub@redhat.com>
14673 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
14675 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
14676 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
14677 when not giving up or versioning for alias only because of
14679 (vect_analyze_data_ref_dependences): Set to true.
14680 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
14682 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
14683 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
14686 PR middle-end/58344
14687 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
14690 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
14691 operand 0 predicate for gathers, use a new pseudo as subtarget.
14693 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
14695 PR middle-end/59609
14696 * lra-constraints.c (process_alt_operands): Add printing debug info.
14697 Check absence of input/output reloads for matched operands too.
14699 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
14701 PR rtl-optimization/59835
14702 * ira.c (ira_init_register_move_cost): Increase cost for
14705 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
14707 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
14709 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
14712 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
14713 non-register objects. Use gen_(high/low)part more consistently.
14716 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
14719 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
14720 endian support, remove tests for WORDS_BIG_ENDIAN.
14721 (p8_mfvsrd_3_<mode>): Likewise.
14722 (reload_gpr_from_vsx<mode>): Likewise.
14723 (reload_gpr_from_vsxsf): Likewise.
14724 (p8_mfvsrd_4_disf): Likewise.
14726 2014-01-16 Richard Biener <rguenther@suse.de>
14728 PR rtl-optimization/46590
14729 * lcm.c (compute_antinout_edge): Use postorder iteration.
14730 (compute_laterin): Use inverted postorder iteration.
14732 2014-01-16 Nick Clifton <nickc@redhat.com>
14734 PR middle-end/28865
14735 * varasm.c (output_constant): Return the number of bytes actually
14737 (output_constructor_array_range): Update the field size with the
14738 number of bytes emitted by output_constant.
14739 (output_constructor_regular_field): Likewise. Also do not
14740 complain if the total number of bytes emitted is now greater
14741 than the expected fieldpos.
14742 * output.h (output_constant): Update prototype and descriptive comment.
14744 2014-01-16 Marek Polacek <polacek@redhat.com>
14746 PR middle-end/59827
14747 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
14748 it is error_mark_node.
14750 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
14752 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
14753 VALID_AVX256_REG_OR_OI_MODE.
14755 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
14757 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
14758 current procedure should be profiled.
14760 2014-01-15 Andrew Pinski <apinski@cavium.com>
14762 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
14763 of moving from/to the STACK_REG register class.
14765 2014-01-15 Richard Henderson <rth@redhat.com>
14768 * reginfo.c (global_regs_decl): Globalize.
14769 * rtl.h (global_regs_decl): Declare.
14770 * ira.c (do_reload): Diagnose frame_pointer_needed and it
14771 reserved via global_regs.
14773 2014-01-15 Teresa Johnson <tejohnson@google.com>
14775 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
14777 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
14779 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
14780 and vmulosh rather than call gen_vec_widen_smult_*.
14781 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
14782 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
14783 (vec_widen_smult_even_v16qi): Likewise.
14784 (vec_widen_umult_even_v8hi): Likewise.
14785 (vec_widen_smult_even_v8hi): Likewise.
14786 (vec_widen_umult_odd_v16qi): Likewise.
14787 (vec_widen_smult_odd_v16qi): Likewise.
14788 (vec_widen_umult_odd_v8hi): Likewise.
14789 (vec_widen_smult_odd_v8hi): Likewise.
14790 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
14791 vmuloub rather than call gen_vec_widen_umult_*.
14792 (vec_widen_umult_lo_v16qi): Likewise.
14793 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
14794 vmulosb rather than call gen_vec_widen_smult_*.
14795 (vec_widen_smult_lo_v16qi): Likewise.
14796 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
14797 rather than call gen_vec_widen_umult_*.
14798 (vec_widen_umult_lo_v8hi): Likewise.
14799 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
14800 rather than call gen_vec_widen_smult_*.
14801 (vec_widen_smult_lo_v8hi): Likewise.
14803 2014-01-15 Jeff Law <law@redhat.com>
14805 PR tree-optimization/59747
14806 * ree.c (find_and_remove_re): Properly handle case where a second
14807 eliminated extension requires widening a copy created for elimination
14808 of a prior extension.
14809 (combine_set_extension): Ensure that the number of hard regs needed
14810 for a destination register does not change when we widen it.
14812 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
14814 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
14815 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
14816 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
14817 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
14818 (avr-*-rtems*): Likewise.
14819 (bfin*-rtems*): Likewise.
14820 (moxie-*-rtems*): Likewise.
14821 (h8300-*-rtems*): Likewise.
14822 (i[34567]86-*-rtems*): Likewise.
14823 (lm32-*-rtems*): Likewise.
14824 (m32r-*-rtems*): Likewise.
14825 (m68k-*-rtems*): Likewise.
14826 (microblaze*-*-rtems*): Likewise.
14827 (mips*-*-rtems*): Likewise.
14828 (powerpc-*-rtems*): Likewise.
14829 (sh-*-rtems*): Likewise.
14830 (sparc-*-rtems*): Likewise.
14831 (sparc64-*-rtems*): Likewise.
14832 (v850-*-rtems*): Likewise.
14833 (m32c-*-rtems*): Likewise.
14835 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
14837 PR rtl-optimization/59511
14838 * ira.c (ira_init_register_move_cost): Use memory costs for some
14839 cases of register move cost calculations.
14840 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
14841 instead of BB frequency.
14842 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
14843 * lra-assigns.c (find_hard_regno_for): Ditto.
14845 2014-01-15 Richard Biener <rguenther@suse.de>
14847 PR tree-optimization/59822
14848 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
14849 (vectorizable_load): Use it to hoist defs of uses of invariant
14850 loads out of the loop.
14852 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
14853 Kugan Vivekanandarajah <kuganv@linaro.org>
14856 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
14859 2014-01-15 Richard Biener <rguenther@suse.de>
14861 PR rtl-optimization/59802
14862 * lcm.c (compute_available): Use inverted postorder to seed
14863 the initial worklist.
14865 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
14868 * config/s390/s390.c (s390_preferred_reload_class): Don't return
14869 ADDR_REGS for invalid symrefs in non-PIC code.
14871 2014-01-15 Jakub Jelinek <jakub@redhat.com>
14874 * builtins.c (determine_block_size): Initialize *probable_max_size
14875 even if len_rtx is CONST_INT.
14877 2014-01-14 Andrew Pinski <apinski@cavium.com>
14879 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
14880 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
14881 (cortexa53_tunings): Likewise.
14882 (aarch64_sched_issue_rate): New function.
14883 (TARGET_SCHED_ISSUE_RATE): Define.
14885 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
14887 * ira-costs.c (find_costs_and_classes): Add missed
14888 ira_init_register_move_cost_if_necessary.
14890 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
14893 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
14895 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
14898 * config/i386/i386.c (type_natural_mode): Add a bool parameter
14899 to indicate if type is used for function return value. Warn ABI
14900 change if the vector mode isn't available for function return value.
14901 (ix86_function_arg_advance): Pass false to type_natural_mode.
14902 (ix86_function_arg): Likewise.
14903 (ix86_gimplify_va_arg): Likewise.
14904 (function_arg_32): Don't warn ABI change.
14905 (ix86_function_value): Pass true to type_natural_mode.
14906 (ix86_return_in_memory): Likewise.
14907 (ix86_struct_value_rtx): Removed.
14908 (TARGET_STRUCT_VALUE_RTX): Likewise.
14910 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
14912 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
14913 converting a conditional jump into a conditional return.
14915 2014-01-14 Richard Biener <rguenther@suse.de>
14917 PR tree-optimization/58921
14918 PR tree-optimization/59006
14919 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
14920 hoisting invariant stmts.
14921 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
14922 invariant loads on the preheader edge if possible.
14924 2014-01-14 Joey Ye <joey.ye@arm.com>
14926 * doc/plugin.texi (Building GCC plugins): Update to C++.
14928 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
14930 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
14931 (_mm_rcp28_round_ss): Ditto.
14932 (_mm_rsqrt28_round_sd): Ditto.
14933 (_mm_rsqrt28_round_ss): Ditto.
14934 (_mm_rcp28_sd): Ditto.
14935 (_mm_rcp28_ss): Ditto.
14936 (_mm_rsqrt28_sd): Ditto.
14937 (_mm_rsqrt28_ss): Ditto.
14938 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
14939 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
14940 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
14941 (IX86_BUILTIN_RCP28SD): Ditto.
14942 (IX86_BUILTIN_RCP28SS): Ditto.
14943 (IX86_BUILTIN_RSQRT28SD): Ditto.
14944 (IX86_BUILTIN_RSQRT28SS): Ditto.
14945 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
14946 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
14947 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
14948 (ix86_expand_special_args_builtin): Expand new FTYPE.
14949 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
14950 (srcp14<mode>): Make insn unary.
14951 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
14952 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
14953 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
14954 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
14955 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
14956 Fix rounding: make it SAE only.
14957 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
14959 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
14961 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
14962 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
14963 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
14964 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
14965 (round_saeonly_mask_scalar_operand4): Ditto.
14966 (round_saeonly_mask_scalar_op3): Ditto.
14967 (round_saeonly_mask_scalar_op4): Ditto.
14969 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14971 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14972 Implement -maltivec=be for vec_insert and vec_extract.
14974 2014-01-10 DJ Delorie <dj@redhat.com>
14976 * config/msp430/msp430.md (call_internal): Don't allow memory
14977 references with SP as the base register.
14978 (call_value_internal): Likewise.
14979 * config/msp430/constraints.md (Yc): New. For memory references
14980 that don't use SP as a base register.
14982 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
14983 "an integer without a # prefix"
14984 * config/msp430/msp430.md (epilogue_helper): Use it.
14986 2014-01-13 Jakub Jelinek <jakub@redhat.com>
14989 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
14990 AVX512F gather builtins.
14991 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
14992 on gather decls with INTEGER_TYPE masktype.
14993 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
14994 directly into the builtin rather than hoisting it before loop.
14996 PR tree-optimization/59387
14997 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
14998 (scev_const_prop): If folded_casts and type has undefined overflow,
14999 use force_gimple_operand instead of force_gimple_operand_gsi and
15000 for each added stmt if it is assign with
15001 arith_code_with_undefined_signed_overflow, call
15002 rewrite_to_defined_overflow.
15003 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
15004 gimple-fold.h instead.
15005 (arith_code_with_undefined_signed_overflow,
15006 rewrite_to_defined_overflow): Moved to ...
15007 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
15008 rewrite_to_defined_overflow): ... here. No longer static.
15009 Include gimplify-me.h.
15010 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
15011 rewrite_to_defined_overflow): New prototypes.
15013 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15015 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
15017 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
15019 * builtins.c (get_object_alignment_2): Minor tweak.
15020 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
15022 2014-01-13 Christian Bruel <christian.bruel@st.com>
15024 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
15025 optimized non constant lengths.
15027 2014-01-13 Jakub Jelinek <jakub@redhat.com>
15030 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
15031 load as __atomic_load_N if possible.
15033 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
15035 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
15037 (rs6000_expand_builtin): Adjust call.
15039 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
15042 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
15043 * config/rs6000/rs6000.c: Include target-globals.h.
15044 (rs6000_set_current_function): Instead of doing target_reinit
15045 unconditionally, use save_target_globals_default_opts and
15046 restore_target_globals.
15048 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
15050 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
15051 (rs6000_expand_builtin): Handle mffs and mtfsf.
15052 (rs6000_init_builtins): Define mffs and mtfsf.
15053 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
15054 (rs6000_mffs): New pattern.
15055 (rs6000_mtfsf): New pattern.
15057 2014-01-11 Bin Cheng <bin.cheng@arm.com>
15059 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
15060 Start narrowing with START. Apply candidate-use pair
15061 and check overall cost in narrowing.
15062 (iv_ca_prune): Pass new argument.
15064 2014-01-10 Jeff Law <law@redhat.com>
15066 PR middle-end/59743
15067 * ree.c (combine_reaching_defs): Ensure the defining statement
15068 occurs before the extension when optimizing extensions with
15069 different source and destination hard registers.
15071 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
15074 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
15075 vtables into the type inheritance graph.
15077 2014-01-10 Jakub Jelinek <jakub@redhat.com>
15079 PR rtl-optimization/59754
15080 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
15081 modes in the REGNO != REGNO case.
15083 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15085 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
15087 2014-01-10 Jakub Jelinek <jakub@redhat.com>
15089 PR tree-optimization/59745
15090 * tree-predcom.c (tree_predictive_commoning_loop): Call
15091 free_affine_expand_cache if giving up because components is NULL.
15093 * target-globals.c (save_target_globals): Allocate < 4KB structs using
15094 GC in payload of target_globals struct instead of allocating them on
15095 the heap and the larger structs separately using GC.
15096 * target-globals.h (struct target_globals): Make regs, hard_regs,
15097 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
15098 of GTY((skip)) and change type to void *.
15099 (reset_target_globals): Cast loads from those fields to corresponding
15102 2014-01-10 Steve Ellcey <sellcey@mips.com>
15105 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
15106 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
15107 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
15109 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
15112 * aarch64-modes.def (CC_Zmode): New flags mode.
15113 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
15114 represents an equality.
15115 (aarch64_get_condition_code): Handle CC_Zmode.
15116 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
15118 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
15120 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
15121 extraction in good case.
15123 2014-01-10 Richard Biener <rguenther@suse.de>
15125 PR tree-optimization/59374
15126 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
15127 checking after SLP discovery. Mark stmts not participating
15128 in any SLP instance properly.
15130 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15132 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
15133 when handling a SET rtx.
15135 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15137 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
15138 (cortex-a57): Likewise.
15139 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
15141 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15143 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
15144 non-iwmmxt builtins.
15146 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
15150 * ipa-devirt.c record_target_from_binfo): Take as argument
15151 stack of binfos and lookup matching one for virtual inheritance.
15152 (possible_polymorphic_call_targets_1): Update.
15154 2014-01-10 Huacai Chen <chenhc@lemote.com>
15156 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
15157 kernel strings for Loongson-2E/2F/3A.
15159 2014-01-10 Jakub Jelinek <jakub@redhat.com>
15161 PR middle-end/59670
15162 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
15163 is_gimple_call before calling gimple_call_internal_p.
15165 2014-01-09 Steve Ellcey <sellcey@mips.com>
15167 * Makefile.in (TREE_FLOW_H): Remove.
15168 (TREE_SSA_H): Add file names from tree-flow.h.
15169 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
15170 * tree.h: Remove tree-flow.h reference.
15171 * hash-table.h: Remove tree-flow.h reference.
15172 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
15173 reference with tree-ssa-loop.h.
15175 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15177 * doc/invoke.texi: Add -maltivec={be,le} options, and document
15178 default element-order behavior for -maltivec.
15179 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
15180 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
15181 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
15182 when targeting big endian, at least for now.
15183 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
15185 2014-01-09 Jakub Jelinek <jakub@redhat.com>
15187 PR middle-end/47735
15188 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
15189 var satisfies use_register_for_decl, just take into account type
15190 alignment, rather than decl alignment.
15192 PR tree-optimization/59622
15193 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
15194 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
15195 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
15196 Don't devirtualize for inplace at all. For targets.length () == 1,
15197 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
15199 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
15201 * config/i386/i386.md (cpu): Remove the unused btver1.
15203 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
15205 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
15207 2014-01-09 Jakub Jelinek <jakub@redhat.com>
15210 * tree-core.h (struct target_globals): New forward declaration.
15211 (struct tree_target_option): Add globals field.
15212 * tree.h (TREE_TARGET_GLOBALS): Define.
15213 (prepare_target_option_nodes_for_pch): New prototype.
15214 * target-globals.h (struct target_globals): Define even if
15215 !SWITCHABLE_TARGET.
15216 * tree.c (prepare_target_option_node_for_pch,
15217 prepare_target_option_nodes_for_pch): New functions.
15218 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
15219 * config/i386/i386.c: Include target-globals.h.
15220 (ix86_set_current_function): Instead of doing target_reinit
15221 unconditionally, use save_target_globals_default_opts and
15222 restore_target_globals.
15224 2014-01-09 Richard Biener <rguenther@suse.de>
15226 PR tree-optimization/59715
15227 * tree-cfg.h (split_critical_edges): Declare.
15228 * tree-cfg.c (split_critical_edges): Export.
15229 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
15231 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
15233 * cfgexpand.c (expand_stack_vars): Optionally disable
15234 asan stack protection.
15235 (expand_used_vars): Likewise.
15236 (partition_stack_vars): Likewise.
15237 * asan.c (asan_emit_stack_protection): Optionally disable
15238 after return stack usage.
15239 (instrument_derefs): Optionally disable memory access instrumentation.
15240 (instrument_builtin_call): Likewise.
15241 (instrument_strlen_call): Likewise.
15242 (asan_protect_global): Optionally disable global variables protection.
15243 * doc/invoke.texi: Added doc for new options.
15244 * params.def: Added new options.
15245 * params.h: Likewise.
15247 2014-01-09 Jakub Jelinek <jakub@redhat.com>
15249 PR rtl-optimization/59724
15250 * ifcvt.c (cond_exec_process_if_block): Don't call
15251 flow_find_head_matching_sequence with 0 longest_match.
15252 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
15253 non-active insns if !stop_after.
15254 (try_head_merge_bb): Revert 2014-01-07 changes.
15256 2014-01-08 Jeff Law <law@redhat.com>
15258 * ree.c (get_sub_rtx): New function, extracted from...
15259 (merge_def_and_ext): Here.
15260 (combine_reaching_defs): Use get_sub_rtx.
15262 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
15264 * cgraph.h (varpool_variable_node): Do not choke on null node.
15266 2014-01-08 Catherine Moore <clm@codesourcery.com>
15268 * config/mips/mips.md (simple_return): Attempt to use JRC
15270 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
15272 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
15274 PR rtl-optimization/59137
15275 * reorg.c (steal_delay_list_from_target): Call update_block for
15277 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
15279 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15281 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
15282 two duplicate entries.
15284 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
15287 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
15289 * config/mips/mips.c (mips_truncated_op_cost): New function.
15290 (mips_rtx_costs): Adjust test for BADDU.
15291 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
15293 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
15295 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
15296 (*baddu_si): ...this new pattern.
15298 2014-01-08 Jakub Jelinek <jakub@redhat.com>
15301 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
15303 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15305 PR middle-end/57748
15306 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
15308 (expand_expr, expand_normal): Adjust.
15309 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
15310 inner_reference_p. Use inner_reference_p to expand inner references.
15311 (store_expr): Adjust.
15312 * cfgexpand.c (expand_call_stmt): Adjust.
15314 2014-01-08 Rong Xu <xur@google.com>
15316 * gcov-io.c (gcov_var): Move from gcov-io.h.
15317 (gcov_position): Ditto.
15318 (gcov_is_error): Ditto.
15319 (gcov_rewrite): Ditto.
15320 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
15321 only part to libgcc/libgcov.h.
15323 2014-01-08 Marek Polacek <polacek@redhat.com>
15325 PR middle-end/59669
15326 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
15328 2014-01-08 Marek Polacek <polacek@redhat.com>
15331 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
15333 2014-01-08 Jakub Jelinek <jakub@redhat.com>
15335 PR rtl-optimization/59649
15336 * stor-layout.c (get_mode_bounds): For BImode return
15337 0 and STORE_FLAG_VALUE.
15339 2014-01-08 Richard Biener <rguenther@suse.de>
15341 PR middle-end/59630
15342 * gimple.h (is_gimple_builtin_call): Remove.
15343 (gimple_builtin_call_types_compatible_p): New.
15344 (gimple_call_builtin_p): New overload.
15345 * gimple.c (is_gimple_builtin_call): Remove.
15346 (validate_call): Rename to ...
15347 (gimple_builtin_call_types_compatible_p): ... this and export. Also
15348 check return types.
15349 (validate_type): New static function.
15350 (gimple_call_builtin_p): New overload and adjust.
15351 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
15352 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
15353 (gimple_fold_stmt_to_constant_1): Likewise.
15354 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
15356 2014-01-08 Richard Biener <rguenther@suse.de>
15358 PR middle-end/59471
15359 * gimplify.c (gimplify_expr): Gimplify register-register type
15360 VIEW_CONVERT_EXPRs to separate stmts.
15362 2014-01-07 Jeff Law <law@redhat.com>
15364 PR middle-end/53623
15365 * ree.c (combine_set_extension): Handle case where source
15366 and destination registers in an extension insn are different.
15367 (combine_reaching_defs): Allow source and destination registers
15368 in extension to be different under limited circumstances.
15369 (add_removable_extension): Remove restriction that the
15370 source and destination registers in the extension are the same.
15371 (find_and_remove_re): Emit a copy from the extension's
15372 destination to its source after the defining insn if
15373 the source and destination registers are different.
15375 PR middle-end/59285
15376 * ifcvt.c (merge_if_block): If we are merging a block with more than
15377 one successor with a block with no successors, remove any BARRIER
15378 after the second block.
15380 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
15382 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
15384 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
15387 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
15388 for 14-bit register offsets when INT14_OK_STRICT is false.
15390 2014-01-07 Roland Stigge <stigge@antcom.de>
15391 Michael Meissner <meissner@linux.vnet.ibm.com>
15394 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
15395 Only check TFmode for SPE constants. Don't check TImode or TDmode.
15397 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
15399 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
15402 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
15404 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
15405 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
15406 rtx is const0_rtx or not.
15408 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
15411 * target-globals.c (save_target_globals): Remove this_fn_optab
15413 * toplev.c: Include optabs.h.
15414 (target_reinit): Temporarily restore the global options if another
15415 set of options are in force.
15417 2014-01-07 Jakub Jelinek <jakub@redhat.com>
15419 PR rtl-optimization/58668
15420 * cfgcleanup.c (flow_find_cross_jump): Don't count
15421 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
15422 to determine what is counted.
15423 (flow_find_head_matching_sequence): Use active_insn_p to determine
15425 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
15427 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
15428 determine what is counted.
15430 PR tree-optimization/59643
15431 * tree-predcom.c (split_data_refs_to_components): If one dr is
15432 read and one write, determine_offset fails and the write isn't
15433 in the bad component, just put the read into the bad component.
15435 2014-01-07 Mike Stump <mikestump@comcast.net>
15436 Jakub Jelinek <jakub@redhat.com>
15439 * tree-core.h (struct tree_optimization_option): Change optabs
15440 type from unsigned char * to void *.
15441 * optabs.c (init_tree_optimization_optabs): Adjust
15442 TREE_OPTIMIZATION_OPTABS initialization.
15444 2014-01-06 Jakub Jelinek <jakub@redhat.com>
15447 * config/i386/i386.h (struct machine_function): Add
15448 no_drap_save_restore field.
15449 * config/i386/i386.c (ix86_save_reg): Use
15450 !cfun->machine->no_drap_save_restore instead of
15451 crtl->stack_realign_needed.
15452 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
15453 this function clears frame_pointer_needed. Set
15454 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
15455 and DRAP reg is needed.
15457 2014-01-06 Marek Polacek <polacek@redhat.com>
15460 * doc/implement-c.texi: Mention that other integer types are
15461 permitted as bit-field types in strictly conforming mode.
15463 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
15465 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
15466 is newly allocated.
15468 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
15470 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
15472 2014-01-06 Martin Jambor <mjambor@suse.cz>
15475 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
15477 * ipa-prop.c (ipa_print_node_params): Fix indentation.
15479 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
15483 * var-tracking.c (add_stores): Preserve the value of the source even if
15484 we don't record the store.
15486 2014-01-06 Terry Guo <terry.guo@arm.com>
15488 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
15490 2014-01-05 Iain Sandoe <iain@codesourcery.com>
15493 * config/darwin.c (darwin_function_section): Adjust return values to
15494 correspond to optimisation changes made in r206070.
15496 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
15498 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
15499 from prefetch_block tune setting.
15500 (nocona_cost): Correct size of prefetch block to 64.
15502 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
15504 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
15505 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
15506 used to save the static chain register in the computation of the offset
15507 from which the FP registers need to be restored.
15509 2014-01-04 Jakub Jelinek <jakub@redhat.com>
15511 PR tree-optimization/59519
15512 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
15513 ICE if get_current_def (current_new_name) is already non-NULL, as long
15514 as it is a phi result of some other phi in *new_exit_bb that has
15517 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
15518 or vmovdqu* for misaligned_operand.
15519 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
15520 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
15521 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
15522 aligned_mem for AVX512F masked aligned load and store builtins and for
15523 non-temporal moves.
15525 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
15527 PR tree-optimization/59651
15528 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
15529 Address range for negative step should be added by TYPE_SIZE_UNIT.
15531 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
15533 * config/m68k/m68k.c (handle_move_double): Handle pushes with
15534 overlapping registers also for registers other than the stack pointer.
15536 2014-01-03 Marek Polacek <polacek@redhat.com>
15539 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
15542 2014-01-03 Jakub Jelinek <jakub@redhat.com>
15545 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
15548 * config/i386/i386.md (MODE_SIZE): New mode attribute.
15549 (push splitter): Use <P:MODE_SIZE> instead of
15550 GET_MODE_SIZE (<P:MODE>mode).
15551 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
15552 (mov -1, reg peephole2): Likewise.
15553 * config/i386/sse.md (*mov<mode>_internal,
15554 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
15555 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
15556 *<code><mode>3, *andnot<mode>3<mask_name>,
15557 <mask_codefor><code><mode>3<mask_name>): Likewise.
15558 * config/i386/subst.md (mask_mode512bit_condition,
15559 sd_mask_mode512bit_condition): Likewise.
15561 2014-01-02 Xinliang David Li <davidxl@google.com>
15563 PR tree-optimization/59303
15564 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
15565 (dump_predicates): Better output format.
15566 (pred_equal_p): New function.
15567 (is_neq_relop_p): Ditto.
15568 (is_neq_zero_form_p): Ditto.
15569 (pred_expr_equal_p): Ditto.
15570 (pred_neg_p): Ditto.
15571 (simplify_pred): Ditto.
15572 (simplify_preds_2): Ditto.
15573 (simplify_preds_3): Ditto.
15574 (simplify_preds_4): Ditto.
15575 (simplify_preds): Ditto.
15576 (push_pred): Ditto.
15577 (push_to_worklist): Ditto.
15578 (get_pred_info_from_cmp): Ditto.
15579 (is_degenerated_phi): Ditto.
15580 (normalize_one_pred_1): Ditto.
15581 (normalize_one_pred): Ditto.
15582 (normalize_one_pred_chain): Ditto.
15583 (normalize_preds): Ditto.
15584 (normalize_cond_1): Remove function.
15585 (normalize_cond): Ditto.
15586 (is_gcond_subset_of): Ditto.
15587 (is_subset_of_any): Ditto.
15588 (is_or_set_subset_of): Ditto.
15589 (is_and_set_subset_of): Ditto.
15590 (is_norm_cond_subset_of): Ditto.
15591 (pred_chain_length_cmp): Ditto.
15592 (convert_control_dep_chain_into_preds): Type change.
15593 (find_predicates): Ditto.
15594 (find_def_preds): Ditto.
15595 (destroy_predicates_vecs): Ditto.
15596 (find_matching_predicates_in_rest_chains): Ditto.
15597 (use_pred_not_overlap_with_undef_path_pred): Ditto.
15598 (is_pred_expr_subset): Ditto.
15599 (is_pred_chain_subset_of): Ditto.
15600 (is_included_in): Ditto.
15601 (is_superset_of): Ditto.
15603 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
15605 Update copyright years.
15607 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
15609 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
15610 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
15611 config/arc/arc.md, config/arc/arc.opt,
15612 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
15613 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
15614 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
15615 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
15616 config/linux-protos.h, config/linux.c, config/winnt-c.c,
15617 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
15618 vtable-verify.c, vtable-verify.h: Use the standard form for the
15621 2014-01-02 Tobias Burnus <burnus@net-b.de>
15623 * gcc.c (process_command): Update copyright notice dates.
15624 * gcov-dump.c: Ditto.
15626 * doc/cpp.texi: Bump @copying's copyright year.
15627 * doc/cppinternals.texi: Ditto.
15628 * doc/gcc.texi: Ditto.
15629 * doc/gccint.texi: Ditto.
15630 * doc/gcov.texi: Ditto.
15631 * doc/install.texi: Ditto.
15632 * doc/invoke.texi: Ditto.
15634 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
15636 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
15638 2014-01-01 Jakub Jelinek <jakub@redhat.com>
15640 * config/i386/sse.md (*mov<mode>_internal): Guard
15641 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
15643 PR rtl-optimization/59647
15644 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
15645 new_rtx into UNSIGNED_FLOAT rtxes.
15647 Copyright (C) 2014 Free Software Foundation, Inc.
15649 Copying and distribution of this file, with or without modification,
15650 are permitted in any medium without royalty provided the copyright
15651 notice and this notice are preserved.