Merge trunk version 209848 into gupc branch.
[official-gcc.git] / gcc / ChangeLog
blobb25f1f66eb5419ae18519750e307318a9a6910ac
1 2014-04-27  Richard Sandiford  <rdsandiford@googlemail.com>
3         * cselib.c (find_slot_memmode): Delete.
4         (cselib_hasher): Change compare_type to a struct.
5         (cselib_hasher::equal): Update accordingly.  Don't expect wrapped
6         constants.
7         (preserve_constants_and_equivs): Adjust for new compare_type.
8         (cselib_find_slot): Likewise.  Take the mode of the rtx as argument.
9         (wrap_constant): Delete.
10         (cselib_lookup_mem, cselib_lookup_1): Update calls to cselib_find_slot.
12 2014-04-26  Markus Trippelsdorf  <markus@trippelsdorf.de>
14         * doc/install.texi (Building with profile feedback): Remove
15         outdated sentence.
17 2014-04-26  Tom de Vries  <tom@codesourcery.com>
19         * config/i386/i386.md (define_expand "ldexpxf3"): Fix out-of-bounds
20         array accesses.
22 2014-04-25  Cary Coutant  <ccoutant@google.com>
24         PR debug/60929
25         * dwarf2out.c (should_move_die_to_comdat): A type definition
26         can contain a subprogram definition, but don't move it to a
27         comdat unit.
28         (clone_as_declaration): Copy DW_AT_abstract_origin attribute.
29         (generate_skeleton_bottom_up): Remove DW_AT_object_pointer attribute
30         from original DIE.
31         (clone_tree_hash): Rename to...
32         (clone_tree_partial): ...this; change callers.  Copy
33         DW_TAG_subprogram DIEs as declarations.
34         (copy_decls_walk): Don't copy children of a declaration into a
35         type unit.
37 2014-04-25  H.J. Lu  <hongjiu.lu@intel.com>
39         PR target/60969
40         * config/i386/i386.md (*movsf_internal): Set MODE to SI for
41         alternative 12.
43 2014-04-25  Jiong Wang  <jiong.wang@arm.com>
45         * config/arm/predicates.md (call_insn_operand): Add long_call check.
46         * config/arm/arm.md (sibcall, sibcall_value): Force the address to
47         reg for long_call.
48         * config/arm/arm.c (arm_function_ok_for_sibcall): Remove long_call
49         restriction.
51 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
53         * config/arm/arm.c (arm_cortex_a8_tune): Initialise
54         T16-related fields.
56 2014-04-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
58         PR tree-optimization/60930
59         * gimple-ssa-strength-reduction.c (create_mul_imm_cand):  Reject
60         creating a multiply candidate by folding two constant
61         multiplicands when the result overflows.
63 2014-04-25  Jakub Jelinek  <jakub@redhat.com>
65         PR tree-optimization/60960
66         * tree-vect-generic.c (expand_vector_operation): Only call
67         expand_vector_divmod if type's mode satisfies VECTOR_MODE_P.
69 2014-04-25  Tom de Vries  <tom@codesourcery.com>
71         * expr.c (clobber_reg_mode): New function.
72         * expr.h (clobber_reg): New function.
74 2014-04-25  Tom de Vries  <tom@codesourcery.com>
76         * rtlanal.c (find_all_hard_reg_sets): Note INSN_CALL_FUNCTION_USAGE
77         clobbers.
79 2014-04-25  Radovan Obradovic  <robradovic@mips.com>
80             Tom de Vries  <tom@codesourcery.com>
82         * rtlanal.c (find_all_hard_reg_sets): Add bool implicit parameter and
83         handle.
84         * rtl.h (find_all_hard_reg_sets): Add bool parameter.
85         * haifa-sched.c (recompute_todo_spec, check_clobbered_conditions): Add
86         new argument to find_all_hard_reg_sets call.
88 2014-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
90         * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p):
91         Use HOST_WIDE_INT_C for mask literal.
92         (aarch_rev16_shleft_mask_imm_p): Likewise.
94 2014-04-25  Eric Botcazou  <ebotcazou@adacore.com>
96         PR target/60941
97         * config/sparc/sparc.md (ashlsi3_extend): Delete.
99 2014-04-25  Marc Glisse  <marc.glisse@inria.fr>
101         PR preprocessor/56540
102         * config/i386/i386-c.c (ix86_target_macros): Define
103         __SIZEOF_FLOAT80__ and __SIZEOF_FLOAT128__.
105 2014-04-25  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
107         * configure.ac (tga_func): Remove.
108         (LIB_TLS_SPEC): Remove.
109         * configure: Regenerate.
110         * config.in: Regenerate.
111         * config/sol2.h (LIB_SPEC): Don't use LIB_TLS_SPEC.
113 2014-04-25  Richard Biener  <rguenther@suse.de>
115         PR ipa/60912
116         * tree-ssa-structalias.c (ipa_pta_execute): Compute direct
117         call stmt use/clobber sets during stmt walk instead of
118         walking the possibly incomplete set of caller edges.
120 2014-04-25  Richard Biener  <rguenther@suse.de>
122         PR ipa/60911
123         * passes.c (apply_ipa_transforms): Inline into only caller ...
124         (execute_one_pass): ... here.  Properly bring in function
125         bodies for nodes we want to apply IPA transforms to.
127 2014-04-24  Cong Hou  <congh@google.com>
129         PR tree-optimization/60896
130         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Pick up
131         all statements in PATTERN_DEF_SEQ in recognized widen-mult pattern.
132         (vect_mark_pattern_stmts): Set the def type of all statements in
133         PATTERN_DEF_SEQ as vect_internal_def.
135 2014-04-24  Michael Meissner  <meissner@linux.vnet.ibm.com>
137         * doc/extend.texi (PowerPC Built-in Functions): Document new
138         powerpc extended divide, bcd, pack/unpack 128-bit, builtin
139         functions.
140         (PowerPC AltiVec/VSX Built-in Functions): Likewise.
142         * config/rs6000/predicates.md (const_0_to_3_operand): New
143         predicate to match 0..3 integer constants.
145         * config/rs6000/rs6000-builtin.def (BU_DFP_MISC_1): Add new macros
146         to support adding miscellaneous builtin functions.
147         (BU_DFP_MISC_2): Likewise.
148         (BU_P7_MISC_1): Likewise.
149         (BU_P7_MISC_2): Likewise.
150         (BU_P8V_MISC_3): Likewise.
151         (BU_MISC_1): Likewise.
152         (BU_MISC_2): Likewise.
153         (DIVWE): Add extended divide builtin functions.
154         (DIVWEO): Likewise.
155         (DIVWEU): Likewise.
156         (DIVWEUO): Likewise.
157         (DIVDE): Likewise.
158         (DIVDEO): Likewise.
159         (DIVDEU): Likewise.
160         (DIVDEUO): Likewise.
161         (DXEX): Add decimal floating-point builtin functions.
162         (DXEXQ): Likewise.
163         (DDEDPD): Likewise.
164         (DDEDPDQ): Likewise.
165         (DENBCD): Likewise.
166         (DENBCDQ): Likewise.
167         (DIEX): Likewise.
168         (DIEXQ): Likewise.
169         (DSCLI): Likewise.
170         (DSCLIQ): Likewise.
171         (DSCRI): Likewise.
172         (DSCRIQ): Likewise.
173         (CDTBCD): Add new BCD builtin functions.
174         (CBCDTD): Likewise.
175         (ADDG6S): Likewise.
176         (BCDADD): Likewise.
177         (BCDADD_LT): Likewise.
178         (BCDADD_EQ): Likewise.
179         (BCDADD_GT): Likewise.
180         (BCDADD_OV): Likewise.
181         (BCDSUB): Likewise.
182         (BCDSUB_LT): Likewise.
183         (BCDSUB_EQ): Likewise.
184         (BCDSUB_GT): Likewise.
185         (BCDSUB_OV): Likewise.
186         (PACK_TD): Add new pack/unpack 128-bit type builtin functions.
187         (UNPACK_TD): Likewise.
188         (PACK_TF): Likewise.
189         (UNPACK_TF): Likewise.
190         (UNPACK_TF_0): Likewise.
191         (UNPACK_TF_1): Likewise.
192         (PACK_V1TI): Likewise.
193         (UNPACK_V1TI): Likewise.
195         * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
196         support for decimal floating point builtin functions.
197         (rs6000_expand_ternop_builtin): Add checks for the new builtin
198         functions that take constant arguments.
199         (rs6000_invalid_builtin): Add decimal floating point builtin
200         support.
201         (rs6000_init_builtins): Setup long double, _Decimal64, and
202         _Decimal128 types for new builtin functions.
203         (builtin_function_type): Set the unsigned flags appropriately for
204         the new builtin functions.
205         (rs6000_opt_masks): Add support for decimal floating point builtin
206         functions.
208         * config/rs6000/rs6000.h (RS6000_BTM_DFP): Add support for decimal
209         floating point builtin functions.
210         (RS6000_BTM_COMMON): Likewise.
211         (RS6000_BTI_long_double): Likewise.
212         (RS6000_BTI_dfloat64): Likewise.
213         (RS6000_BTI_dfloat128): Likewise.
214         (long_double_type_internal_node): Likewise.
215         (dfloat64_type_internal_node): Likewise.
216         (dfloat128_type_internal_node): Likewise.
218         * config/rs6000/altivec.h (UNSPEC_BCDADD): Add support for ISA
219         2.07 bcd arithmetic instructions.
220         (UNSPEC_BCDSUB): Likewise.
221         (UNSPEC_BCD_OVERFLOW): Likewise.
222         (UNSPEC_BCD_ADD_SUB): Likewise.
223         (bcd_add_sub): Likewise.
224         (BCD_TEST): Likewise.
225         (bcd<bcd_add_sub>): Likewise.
226         (bcd<bcd_add_sub>_test): Likewise.
227         (bcd<bcd_add_sub>_test2): Likewise.
228         (bcd<bcd_add_sub>_<code>): Likewise.
229         (peephole2 for combined bcd ops): Likewise.
231         * config/rs6000/dfp.md (UNSPEC_DDEDPD): Add support for new
232         decimal floating point builtin functions.
233         (UNSPEC_DENBCD): Likewise.
234         (UNSPEC_DXEX): Likewise.
235         (UNSPEC_DIEX): Likewise.
236         (UNSPEC_DSCLI): Likewise.
237         (UNSPEC_DSCRI): Likewise.
238         (D64_D128): Likewise.
239         (dfp_suffix): Likewise.
240         (dfp_ddedpd_<mode>): Likewise.
241         (dfp_denbcd_<mode>): Likewise.
242         (dfp_dxex_<mode>): Likewise.
243         (dfp_diex_<mode>): Likewise.
244         (dfp_dscli_<mode>): Likewise.
245         (dfp_dscri_<mode>): Likewise.
247         * config/rs6000/rs6000.md (UNSPEC_ADDG6S): Add support for new BCD
248         builtin functions.
249         (UNSPEC_CDTBCD): Likewise.
250         (UNSPEC_CBCDTD): Likewise.
251         (UNSPEC_DIVE): Add support for new extended divide builtin
252         functions.
253         (UNSPEC_DIVEO): Likewise.
254         (UNSPEC_DIVEU): Likewise.
255         (UNSPEC_DIVEUO): Likewise.
256         (UNSPEC_UNPACK_128BIT): Add support for new builtin functions to
257         pack/unpack 128-bit types.
258         (UNSPEC_PACK_128BIT): Likewise.
259         (idiv_ldiv): New mode attribute to set the 32/64-bit divide type.
260         (udiv<mode>3): Use idiv_ldiv mode attribute.
261         (div<mode>3): Likewise.
262         (addg6s): Add new BCD builtin functions.
263         (cdtbcd): Likewise.
264         (cbcdtd): Likewise.
265         (UNSPEC_DIV_EXTEND): Add support for new extended divide
266         instructions.
267         (div_extend): Likewise.
268         (div<div_extend>_<mode>"): Likewise.
269         (FP128_64): Add support for new builtin functions to pack/unpack
270         128-bit types.
271         (unpack<mode>): Likewise.
272         (unpacktf_0): Likewise.
273         (unpacktf_1): Likewise.
274         (unpack<mode>_dm): Likewise.
275         (unpack<mode>_nodm): Likewise.
276         (pack<mode>): Likewise.
277         (unpackv1ti): Likewise.
278         (packv1ti): Likewise.
280 2014-04-24  Vishnu K S  <Vishnu.k_s@atmel.com>
282         * gcc/config/avr/avr.c: Add comment on why -fdelete-null-pointer-checks
283         is disabled.
285 2014-04-24  Jakub Jelinek  <jakub@redhat.com>
287         * tree.h (OMP_CLAUSE_LINEAR_GIMPLE_SEQ): Define.
288         * gimplify.c (omp_is_private): Change last argument's type to int.
289         Only diagnose lastprivate if the simd argument is 1, only diagnose
290         linear if the simd argument is 2.
291         (gimplify_omp_for): Adjust omp_is_private callers.  When adding
292         lastprivate or private, add the clause to OMP_FOR_CLAUSES.  Pass
293         GOVD_EXPLICIT to omp_add_variable.  For simd with collapse == 1
294         create OMP_CLAUSE_LINEAR rather than OMP_CLAUSE_PRIVATE for var.
295         If var != decl and decl is in OMP_CLAUSE_LINEAR, gimplify decl
296         increment to OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
297         * omp-low.c (scan_sharing_clauses, lower_lastprivate_clauses): Handle
298         OMP_CLAUSE_LINEAR_GIMPLE_SEQ.
299         * tree-nested.c (convert_nonlocal_omp_clauses,
300         convert_local_omp_clauses): Handle OMP_CLAUSE_LINEAR.
302 2014-04-24  Segher Boessenkool  <segher@kernel.crashing.org>
304         PR target/60822
305         * config/m68k/m68k.md (extendplussidi): Don't allow memory for
306         operand 1.
308 2014-04-24  Dimitris Papavasiliou  <dpapavas@gmail.com>
310         * flag-types.h (enum ivar_visibility): Add.
312 2014-04-24  Trevor Saunders  <tsaunders@mozilla.com>
314         * config/sh/sh_treg_combine.c (sh_treg_combine::execute): Take
315         function * argument.
317 2014-04-24  Alan Lawrence  <alan.lawrence@arm.com>
319         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Enable for bigendian.
321 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
322             Tom de Vries  <tom@codesourcery.com>
324         * reg-notes.def (REG_NOTE (CALL_DECL)): New reg-note REG_CALL_DECL.
325         * calls.c (expand_call, emit_library_call_value_1): Add REG_CALL_DECL
326         reg-note.
327         * combine.c (distribute_notes): Handle REG_CALL_DECL reg-note.
328         * emit-rtl.c (try_split): Same.
330 2014-04-24  Radovan Obradovic  <robradovic@mips.com>
331             Tom de Vries  <tom@codesourcery.com>
333         * common.opt (fuse-caller-save): New option.
335 2014-04-24  Tejas Belagod  <tejas.belagod@arm.com>
337         * config/aarch64/aarch64.c (aarch64_evpc_tbl): Reverse order of elements
338         for big-endian.
340 2014-04-24  Richard Biener  <rguenther@suse.de>
342         * expr.c (expand_expr_real_1): Avoid gimple_assign_rhs_to_tree
343         during TER and instead use the sepops interface for expanding
344         non-GIMPLE_SINGLE_RHS.
346 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
348         * config/i386/sol2.h (ASM_PREFERRED_EH_DATA_FORMAT): Only redefine
349         if not HAVE_AS_IX86_DIFF_SECT_DELTA.
351 2014-04-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
353         * configure.ac (gcc_cv_as_cfi_directive): Support Solaris/x86
354         assembler 64-bit option.
355         * configure: Regenerate.
357 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
359         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Check
360         TARGET_SIMD rather than TARGET_GENERAL_REGS_ONLY.
361         (TARGET_SIMD): Take AARCH64_ISA_SIMD into account.
362         (TARGET_FLOAT): Take AARCH64_ISA_FP into account.
363         (TARGET_CRYPTO): Take TARGET_SIMD into account.
365 2014-04-24  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
367         * config/aarch64/aarch64-builtins.c
368         (aarch64_builtin_vectorized_function): Handle BUILT_IN_BSWAP16,
369         BUILT_IN_BSWAP32, BUILT_IN_BSWAP64.
370         * config/aarch64/aarch64-simd.md (bswap<mode>): New pattern.
371         * config/aarch64/aarch64-simd-builtins.def: Define vector bswap
372         builtins.
373         * config/aarch64/iterator.md (VDQHSD): New mode iterator.
374         (Vrevsuff): New mode attribute.
376 2014-04-24  Terry Guo  <terry.guo@arm.com>
378         * config/arm/arm.h (machine_function): Define variable
379         after_arm_reorg here.
380         * config/arm/arm.c (after_arm_reorg): Remove the definition.
381         (arm_split_constant): Update the way to access variable
382         after_arm_reorg.
383         (arm_reorg): Ditto.
384         (arm_output_function_epilogue): Remove the reset of after_arm_reorg.
386 2014-04-23  Tom de Vries  <tom@codesourcery.com>
388         * target-hooks-macros.h: Fix DEFHOOKPOD argument order in comment.
390 2014-04-23  David Malcolm  <dmalcolm@redhat.com>
392         * is-a.h: Update comments to reflect the following changes to the
393         "pointerness" of the API, making the template parameter match the
394         return type, allowing use of is-a.h with typedefs of pointers.
395         (is_a_helper::cast): Return a T rather then a pointer to a T, so
396         that the return type matches the parameter to the is_a_helper.
397         (as_a): Likewise.
398         (dyn_cast): Likewise.
400         * cgraph.c (cgraph_node_for_asm): Update for removal of implicit
401         pointer from the is-a.h API.
403         * cgraph.h (is_a_helper <cgraph_node>::test): Convert to...
404         (is_a_helper <cgraph_node *>::test): ...this, matching change to
405         is-a.h API.
406         (is_a_helper <varpool_node>::test): Likewise, convert to...
407         (is_a_helper <varpool_node *>::test): ...this.
409         (varpool_first_variable): Update for removal of implicit pointer
410         from the is-a.h API.
411         (varpool_next_variable): Likewise.
412         (varpool_first_static_initializer): Likewise.
413         (varpool_next_static_initializer): Likewise.
414         (varpool_first_defined_variable): Likewise.
415         (varpool_next_defined_variable): Likewise.
416         (cgraph_first_defined_function): Likewise.
417         (cgraph_next_defined_function): Likewise.
418         (cgraph_first_function): Likewise.
419         (cgraph_next_function): Likewise.
420         (cgraph_first_function_with_gimple_body): Likewise.
421         (cgraph_next_function_with_gimple_body): Likewise.
422         (cgraph_alias_target): Likewise.
423         (varpool_alias_target): Likewise.
424         (cgraph_function_or_thunk_node): Likewise.
425         (varpool_variable_node): Likewise.
426         (symtab_real_symbol_p): Likewise.
427         * cgraphunit.c (referred_to_p): Likewise.
428         (analyze_functions): Likewise.
429         (handle_alias_pairs): Likewise.
430         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
431         * gimple-ssa.h (gimple_vuse_op): Likewise.
432         (gimple_vdef_op): Likewise.
433         * gimple-streamer-in.c (input_gimple_stmt): Likewise.
434         * gimple.c (gimple_build_asm_1): Likewise.
435         (gimple_build_try): Likewise.
436         (gimple_build_resx): Likewise.
437         (gimple_build_eh_dispatch): Likewise.
438         (gimple_build_omp_for): Likewise.
439         (gimple_omp_for_set_clauses): Likewise.
441         * gimple.h (is_a_helper <gimple_statement_asm>::test): Convert to...
442         (is_a_helper <gimple_statement_asm *>::test): ...this.
443         (is_a_helper <gimple_statement_bind>::test): Convert to...
444         (is_a_helper <gimple_statement_bind *>::test): ...this.
445         (is_a_helper <gimple_statement_call>::test): Convert to...
446         (is_a_helper <gimple_statement_call *>::test): ...this.
447         (is_a_helper <gimple_statement_catch>::test): Convert to...
448         (is_a_helper <gimple_statement_catch *>::test): ...this.
449         (is_a_helper <gimple_statement_resx>::test): Convert to...
450         (is_a_helper <gimple_statement_resx *>::test): ...this.
451         (is_a_helper <gimple_statement_eh_dispatch>::test): Convert to...
452         (is_a_helper <gimple_statement_eh_dispatch *>::test): ...this.
453         (is_a_helper <gimple_statement_eh_else>::test): Convert to...
454         (is_a_helper <gimple_statement_eh_else *>::test): ...this.
455         (is_a_helper <gimple_statement_eh_filter>::test): Convert to...
456         (is_a_helper <gimple_statement_eh_filter *>::test): ...this.
457         (is_a_helper <gimple_statement_eh_mnt>::test): Convert to...
458         (is_a_helper <gimple_statement_eh_mnt *>::test): ...this.
459         (is_a_helper <gimple_statement_omp_atomic_load>::test): Convert to...
460         (is_a_helper <gimple_statement_omp_atomic_load *>::test): ...this.
461         (is_a_helper <gimple_statement_omp_atomic_store>::test): Convert to...
462         (is_a_helper <gimple_statement_omp_atomic_store *>::test): ...this.
463         (is_a_helper <gimple_statement_omp_return>::test): Convert to...
464         (is_a_helper <gimple_statement_omp_return *>::test): ...this.
465         (is_a_helper <gimple_statement_omp_continue>::test): Convert to...
466         (is_a_helper <gimple_statement_omp_continue *>::test): ...this.
467         (is_a_helper <gimple_statement_omp_critical>::test): Convert to...
468         (is_a_helper <gimple_statement_omp_critical *>::test): ...this.
469         (is_a_helper <gimple_statement_omp_for>::test): Convert to...
470         (is_a_helper <gimple_statement_omp_for *>::test): ...this.
471         (is_a_helper <gimple_statement_omp_taskreg>::test): Convert to...
472         (is_a_helper <gimple_statement_omp_taskreg *>::test): ...this.
473         (is_a_helper <gimple_statement_omp_parallel>::test): Convert to...
474         (is_a_helper <gimple_statement_omp_parallel *>::test): ...this.
475         (is_a_helper <gimple_statement_omp_target>::test): Convert to...
476         (is_a_helper <gimple_statement_omp_target *>::test): ...this.
477         (is_a_helper <gimple_statement_omp_sections>::test): Convert to...
478         (is_a_helper <gimple_statement_omp_sections *>::test): ...this.
479         (is_a_helper <gimple_statement_omp_single>::test): Convert to...
480         (is_a_helper <gimple_statement_omp_single *>::test): ...this.
481         (is_a_helper <gimple_statement_omp_teams>::test): Convert to...
482         (is_a_helper <gimple_statement_omp_teams *>::test): ...this.
483         (is_a_helper <gimple_statement_omp_task>::test): Convert to...
484         (is_a_helper <gimple_statement_omp_task *>::test): ...this.
485         (is_a_helper <gimple_statement_phi>::test): Convert to...
486         (is_a_helper <gimple_statement_phi *>::test): ...this.
487         (is_a_helper <gimple_statement_transaction>::test): Convert to...
488         (is_a_helper <gimple_statement_transaction *>::test): ...this.
489         (is_a_helper <gimple_statement_try>::test): Convert to...
490         (is_a_helper <gimple_statement_try *>::test): ...this.
491         (is_a_helper <gimple_statement_wce>::test): Convert to...
492         (is_a_helper <gimple_statement_wce *>::test): ...this.
493         (is_a_helper <const gimple_statement_asm>::test): Convert to...
494         (is_a_helper <const gimple_statement_asm *>::test): ...this.
495         (is_a_helper <const gimple_statement_bind>::test): Convert to...
496         (is_a_helper <const gimple_statement_bind *>::test): ...this.
497         (is_a_helper <const gimple_statement_call>::test): Convert to...
498         (is_a_helper <const gimple_statement_call *>::test): ...this.
499         (is_a_helper <const gimple_statement_catch>::test): Convert to...
500         (is_a_helper <const gimple_statement_catch *>::test): ...this.
501         (is_a_helper <const gimple_statement_resx>::test): Convert to...
502         (is_a_helper <const gimple_statement_resx *>::test): ...this.
503         (is_a_helper <const gimple_statement_eh_dispatch>::test):
504         Convert to...
505         (is_a_helper <const gimple_statement_eh_dispatch *>::test):
506         ...this.
507         (is_a_helper <const gimple_statement_eh_filter>::test): Convert
508         to...
509         (is_a_helper <const gimple_statement_eh_filter *>::test): ...this.
510         (is_a_helper <const gimple_statement_omp_atomic_load>::test):
511         Convert to...
512         (is_a_helper <const gimple_statement_omp_atomic_load *>::test):
513         ...this.
514         (is_a_helper <const gimple_statement_omp_atomic_store>::test):
515         Convert to...
516         (is_a_helper <const gimple_statement_omp_atomic_store *>::test):
517         ...this.
518         (is_a_helper <const gimple_statement_omp_return>::test): Convert
519         to...
520         (is_a_helper <const gimple_statement_omp_return *>::test): ...this.
521         (is_a_helper <const gimple_statement_omp_continue>::test): Convert
522         to...
523         (is_a_helper <const gimple_statement_omp_continue *>::test): ...this.
524         (is_a_helper <const gimple_statement_omp_critical>::test): Convert
525         to...
526         (is_a_helper <const gimple_statement_omp_critical *>::test): ...this.
527         (is_a_helper <const gimple_statement_omp_for>::test): Convert to...
528         (is_a_helper <const gimple_statement_omp_for *>::test): ...this.
529         (is_a_helper <const gimple_statement_omp_taskreg>::test): Convert
530         to...
531         (is_a_helper <const gimple_statement_omp_taskreg *>::test): ...this.
532         (is_a_helper <const gimple_statement_omp_parallel>::test): Convert
533         to...
534         (is_a_helper <const gimple_statement_omp_parallel *>::test): ...this.
535         (is_a_helper <const gimple_statement_omp_target>::test): Convert
536         to...
537         (is_a_helper <const gimple_statement_omp_target *>::test): ...this.
538         (is_a_helper <const gimple_statement_omp_sections>::test): Convert
539         to...
540         (is_a_helper <const gimple_statement_omp_sections *>::test): ...this.
541         (is_a_helper <const gimple_statement_omp_single>::test): Convert
542         to...
543         (is_a_helper <const gimple_statement_omp_single *>::test): ...this.
544         (is_a_helper <const gimple_statement_omp_teams>::test): Convert
545         to...
546         (is_a_helper <const gimple_statement_omp_teams *>::test): ...this.
547         (is_a_helper <const gimple_statement_omp_task>::test): Convert
548         to...
549         (is_a_helper <const gimple_statement_omp_task *>::test): ...this.
550         (is_a_helper <const gimple_statement_phi>::test): Convert to...
551         (is_a_helper <const gimple_statement_phi *>::test): ...this.
552         (is_a_helper <const gimple_statement_transaction>::test): Convert
553         to...
554         (is_a_helper <const gimple_statement_transaction *>::test): ...this.
555         (is_a_helper <const gimple_statement_with_ops>::test): Convert
556         to...
557         (is_a_helper <const gimple_statement_with_ops *>::test): ...this.
558         (is_a_helper <gimple_statement_with_ops>::test): Convert to...
559         (is_a_helper <gimple_statement_with_ops *>::test): ...this.
560         (is_a_helper <const gimple_statement_with_memory_ops>::test):
561         Convert to...
562         (is_a_helper <const gimple_statement_with_memory_ops *>::test):
563         ...this.
564         (is_a_helper <gimple_statement_with_memory_ops>::test): Convert
565         to...
566         (is_a_helper <gimple_statement_with_memory_ops *>::test): ...this.
568         (gimple_use_ops): Update for removal of implicit pointer from the
569         is-a.h API.
570         (gimple_set_use_ops): Likewise.
571         (gimple_vuse): Likewise.
572         (gimple_vdef): Likewise.
573         (gimple_vuse_ptr): Likewise.
574         (gimple_vdef_ptr): Likewise.
575         (gimple_set_vuse): Likewise.
576         (gimple_set_vdef): Likewise.
577         (gimple_omp_return_set_lhs): Likewise.
578         (gimple_omp_return_lhs): Likewise.
579         (gimple_omp_return_lhs_ptr): Likewise.
580         (gimple_call_fntype): Likewise.
581         (gimple_call_set_fntype): Likewise.
582         (gimple_call_set_internal_fn): Likewise.
583         (gimple_call_use_set): Likewise.
584         (gimple_call_clobber_set): Likewise.
585         (gimple_bind_vars): Likewise.
586         (gimple_bind_set_vars): Likewise.
587         (gimple_bind_body_ptr): Likewise.
588         (gimple_bind_set_body): Likewise.
589         (gimple_bind_add_stmt): Likewise.
590         (gimple_bind_block): Likewise.
591         (gimple_bind_set_block): Likewise.
592         (gimple_asm_ninputs): Likewise.
593         (gimple_asm_noutputs): Likewise.
594         (gimple_asm_nclobbers): Likewise.
595         (gimple_asm_nlabels): Likewise.
596         (gimple_asm_input_op): Likewise.
597         (gimple_asm_input_op_ptr): Likewise.
598         (gimple_asm_output_op): Likewise.
599         (gimple_asm_output_op_ptr): Likewise.
600         (gimple_asm_set_output_op): Likewise.
601         (gimple_asm_clobber_op): Likewise.
602         (gimple_asm_set_clobber_op): Likewise.
603         (gimple_asm_label_op): Likewise.
604         (gimple_asm_set_label_op): Likewise.
605         (gimple_asm_string): Likewise.
606         (gimple_catch_types): Likewise.
607         (gimple_catch_types_ptr): Likewise.
608         (gimple_catch_handler_ptr): Likewise.
609         (gimple_catch_set_types): Likewise.
610         (gimple_catch_set_handler): Likewise.
611         (gimple_eh_filter_types): Likewise.
612         (gimple_eh_filter_types_ptr): Likewise.
613         (gimple_eh_filter_failure_ptr): Likewise.
614         (gimple_eh_filter_set_types): Likewise.
615         (gimple_eh_filter_set_failure): Likewise.
616         (gimple_eh_must_not_throw_fndecl): Likewise.
617         (gimple_eh_must_not_throw_set_fndecl): Likewise.
618         (gimple_eh_else_n_body_ptr): Likewise.
619         (gimple_eh_else_e_body_ptr): Likewise.
620         (gimple_eh_else_set_n_body): Likewise.
621         (gimple_eh_else_set_e_body): Likewise.
622         (gimple_try_eval_ptr): Likewise.
623         (gimple_try_cleanup_ptr): Likewise.
624         (gimple_try_set_eval): Likewise.
625         (gimple_try_set_cleanup): Likewise.
626         (gimple_wce_cleanup_ptr): Likewise.
627         (gimple_wce_set_cleanup): Likewise.
628         (gimple_phi_capacity): Likewise.
629         (gimple_phi_num_args): Likewise.
630         (gimple_phi_result): Likewise.
631         (gimple_phi_result_ptr): Likewise.
632         (gimple_phi_set_result): Likewise.
633         (gimple_phi_arg): Likewise.
634         (gimple_phi_set_arg): Likewise.
635         (gimple_resx_region): Likewise.
636         (gimple_resx_set_region): Likewise.
637         (gimple_eh_dispatch_region): Likewise.
638         (gimple_eh_dispatch_set_region): Likewise.
639         (gimple_omp_critical_name): Likewise.
640         (gimple_omp_critical_name_ptr): Likewise.
641         (gimple_omp_critical_set_name): Likewise.
642         (gimple_omp_for_clauses): Likewise.
643         (gimple_omp_for_clauses_ptr): Likewise.
644         (gimple_omp_for_set_clauses): Likewise.
645         (gimple_omp_for_collapse): Likewise.
646         (gimple_omp_for_index): Likewise.
647         (gimple_omp_for_index_ptr): Likewise.
648         (gimple_omp_for_set_index): Likewise.
649         (gimple_omp_for_initial): Likewise.
650         (gimple_omp_for_initial_ptr): Likewise.
651         (gimple_omp_for_set_initial): Likewise.
652         (gimple_omp_for_final): Likewise.
653         (gimple_omp_for_final_ptr): Likewise.
654         (gimple_omp_for_set_final): Likewise.
655         (gimple_omp_for_incr): Likewise.
656         (gimple_omp_for_incr_ptr): Likewise.
657         (gimple_omp_for_set_incr): Likewise.
658         (gimple_omp_for_pre_body_ptr): Likewise.
659         (gimple_omp_for_set_pre_body): Likewise.
660         (gimple_omp_parallel_clauses): Likewise.
661         (gimple_omp_parallel_clauses_ptr): Likewise.
662         (gimple_omp_parallel_set_clauses): Likewise.
663         (gimple_omp_parallel_child_fn): Likewise.
664         (gimple_omp_parallel_child_fn_ptr): Likewise.
665         (gimple_omp_parallel_set_child_fn): Likewise.
666         (gimple_omp_parallel_data_arg): Likewise.
667         (gimple_omp_parallel_data_arg_ptr): Likewise.
668         (gimple_omp_parallel_set_data_arg): Likewise.
669         (gimple_omp_task_clauses): Likewise.
670         (gimple_omp_task_clauses_ptr): Likewise.
671         (gimple_omp_task_set_clauses): Likewise.
672         (gimple_omp_task_child_fn): Likewise.
673         (gimple_omp_task_child_fn_ptr): Likewise.
674         (gimple_omp_task_set_child_fn): Likewise.
675         (gimple_omp_task_data_arg): Likewise.
676         (gimple_omp_task_data_arg_ptr): Likewise.
677         (gimple_omp_task_set_data_arg): Likewise.
678         (gimple_omp_taskreg_clauses): Likewise.
679         (gimple_omp_taskreg_clauses_ptr): Likewise.
680         (gimple_omp_taskreg_set_clauses): Likewise.
681         (gimple_omp_taskreg_child_fn): Likewise.
682         (gimple_omp_taskreg_child_fn_ptr): Likewise.
683         (gimple_omp_taskreg_set_child_fn): Likewise.
684         (gimple_omp_taskreg_data_arg): Likewise.
685         (gimple_omp_taskreg_data_arg_ptr): Likewise.
686         (gimple_omp_taskreg_set_data_arg): Likewise.
687         (gimple_omp_task_copy_fn): Likewise.
688         (gimple_omp_task_copy_fn_ptr): Likewise.
689         (gimple_omp_task_set_copy_fn): Likewise.
690         (gimple_omp_task_arg_size): Likewise.
691         (gimple_omp_task_arg_size_ptr): Likewise.
692         (gimple_omp_task_set_arg_size): Likewise.
693         (gimple_omp_task_arg_align): Likewise.
694         (gimple_omp_task_arg_align_ptr): Likewise.
695         (gimple_omp_task_set_arg_align): Likewise.
696         (gimple_omp_single_clauses): Likewise.
697         (gimple_omp_single_clauses_ptr): Likewise.
698         (gimple_omp_single_set_clauses): Likewise.
699         (gimple_omp_target_clauses): Likewise.
700         (gimple_omp_target_clauses_ptr): Likewise.
701         (gimple_omp_target_set_clauses): Likewise.
702         (gimple_omp_target_child_fn): Likewise.
703         (gimple_omp_target_child_fn_ptr): Likewise.
704         (gimple_omp_target_set_child_fn): Likewise.
705         (gimple_omp_target_data_arg): Likewise.
706         (gimple_omp_target_data_arg_ptr): Likewise.
707         (gimple_omp_target_set_data_arg): Likewise.
708         (gimple_omp_teams_clauses): Likewise.
709         (gimple_omp_teams_clauses_ptr): Likewise.
710         (gimple_omp_teams_set_clauses): Likewise.
711         (gimple_omp_sections_clauses): Likewise.
712         (gimple_omp_sections_clauses_ptr): Likewise.
713         (gimple_omp_sections_set_clauses): Likewise.
714         (gimple_omp_sections_control): Likewise.
715         (gimple_omp_sections_control_ptr): Likewise.
716         (gimple_omp_sections_set_control): Likewise.
717         (gimple_omp_for_set_cond): Likewise.
718         (gimple_omp_for_cond): Likewise.
719         (gimple_omp_atomic_store_set_val): Likewise.
720         (gimple_omp_atomic_store_val): Likewise.
721         (gimple_omp_atomic_store_val_ptr): Likewise.
722         (gimple_omp_atomic_load_set_lhs): Likewise.
723         (gimple_omp_atomic_load_lhs): Likewise.
724         (gimple_omp_atomic_load_lhs_ptr): Likewise.
725         (gimple_omp_atomic_load_set_rhs): Likewise.
726         (gimple_omp_atomic_load_rhs): Likewise.
727         (gimple_omp_atomic_load_rhs_ptr): Likewise.
728         (gimple_omp_continue_control_def): Likewise.
729         (gimple_omp_continue_control_def_ptr): Likewise.
730         (gimple_omp_continue_set_control_def): Likewise.
731         (gimple_omp_continue_control_use): Likewise.
732         (gimple_omp_continue_control_use_ptr): Likewise.
733         (gimple_omp_continue_set_control_use): Likewise.
734         (gimple_transaction_body_ptr): Likewise.
735         (gimple_transaction_label): Likewise.
736         (gimple_transaction_label_ptr): Likewise.
737         (gimple_transaction_set_body): Likewise.
738         (gimple_transaction_set_label): Likewise.
740         * ipa-devirt.c (build_type_inheritance_graph): Likewise.
741         * ipa-inline-analysis.c (inline_write_summary): Likewise.
742         * ipa-ref.c (ipa_record_reference): Likewise.
743         * ipa-reference.c (analyze_function): Likewise.
744         (ipa_reference_write_optimization_summary): Likewise.
745         * ipa.c (symtab_remove_unreachable_nodes): Likewise.
746         (address_taken_from_non_vtable_p): Likewise.
747         (comdat_can_be_unshared_p_1): Likewise.
748         * lto-cgraph.c (lto_output_ref): Likewise.
749         (add_references): Likewise.
750         (compute_ltrans_boundary): Likewise.
751         (output_symtab): Likewise.
752         (input_ref): Likewise.
753         (input_cgraph_1): Likewise.
754         (output_cgraph_opt_summary): Likewise.
755         * lto-streamer-out.c (lto_output): Likewise.
756         (output_symbol_p): Likewise.
757         * lto-streamer.h (lsei_next_function_in_partition): Likewise.
758         (lsei_start_function_in_partition): Likewise.
759         (lsei_next_variable_in_partition): Likewise.
760         (lsei_start_variable_in_partition): Likewise.
761         * symtab.c (insert_to_assembler_name_hash): Likewise.
762         (unlink_from_assembler_name_hash): Likewise.
763         (symtab_unregister_node): Likewise.
764         (symtab_remove_node): Likewise.
765         (dump_symtab_node): Likewise.
766         (verify_symtab_base): Likewise.
767         (verify_symtab_node): Likewise.
768         (symtab_make_decl_local): Likewise.
769         (symtab_alias_ultimate_target): Likewise.
770         (symtab_resolve_alias): Likewise.
771         (symtab_get_symbol_partitioning_class): Likewise.
772         * tree-phinodes.c (allocate_phi_node): Likewise.
773         (reserve_phi_args_for_new_edge): Likewise.
774         (remove_phi_args): Likewise.
775         * varpool.c (varpool_node_for_asm): Likewise.
776         (varpool_remove_unreferenced_decls): Likewise.
778 2014-04-23  Jeff Law  <law@redhat.com>
780         PR tree-optimization/60902
781         * tree-ssa-threadedge.c
782         (record_temporary_equivalences_from_stmts_at_dest): Make sure to
783         invalidate outputs from statements that do not produce useful
784         outputs for threading.
786 2014-04-23 Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
788         * config/aarch64/aarch64.md (stack_protect_set, stack_protect_test)
789         (stack_protect_set_<mode>, stack_protect_test_<mode>): Add
790         machine descriptions for Stack Smashing Protector.
792 2014-04-23  Richard Earnshaw  <rearnsha@arm.com>
794         * aarch64.md (<optab>_rol<mode>3): New pattern.
795         (<optab>_rolsi3_uxtw): Likewise.
796         * aarch64.c (aarch64_strip_shift): Handle ROTATE and ROTATERT.
798 2014-04-23  James Greenhalgh  <james.greenhalgh@arm.com>
800         * config/arm/arm.c (arm_cortex_a57_tune): Initialize all fields.
801         (arm_cortex_a12_tune): Likewise.
803 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
805         * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle BSWAP.
807 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
809         * config/arm/arm.md (arm_rev16si2): New pattern.
810         (arm_rev16si2_alt): Likewise.
811         * config/arm/arm.c (arm_new_rtx_costs): Handle rev16 case.
813 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
815        * config/aarch64/aarch64.md (rev16<mode>2): New pattern.
816        (rev16<mode>2_alt): Likewise.
817        * config/aarch64/aarch64.c (aarch64_rtx_costs): Handle rev16 case.
818        * config/arm/aarch-common.c (aarch_rev16_shright_mask_imm_p): New.
819        (aarch_rev16_shleft_mask_imm_p): Likewise.
820        (aarch_rev16_p_1): Likewise.
821        (aarch_rev16_p): Likewise.
822        * config/arm/aarch-common-protos.h (aarch_rev16_p): Declare extern.
823        (aarch_rev16_shright_mask_imm_p): Likewise.
824        (aarch_rev16_shleft_mask_imm_p): Likewise.
826 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
828        * config/arm/aarch-common-protos.h (alu_cost_table): Add rev field.
829        * config/arm/aarch-cost-tables.h (generic_extra_costs): Specify
830        rev cost.
831        (cortex_a53_extra_costs): Likewise.
832        (cortex_a57_extra_costs): Likewise.
833        * config/arm/arm.c (cortexa9_extra_costs): Likewise.
834        (cortexa7_extra_costs): Likewise.
835        (cortexa8_extra_costs): Likewise.
836        (cortexa12_extra_costs): Likewise.
837        (cortexa15_extra_costs): Likewise.
838        (v7m_extra_costs): Likewise.
839        (arm_new_rtx_costs): Handle BSWAP.
841 2013-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
843        * config/arm/arm.c (cortexa8_extra_costs): New table.
844        (arm_cortex_a8_tune): New tuning struct.
845        * config/arm/arm-cores.def (cortex-a8): Use cortex_a8 tuning struct.
847 2014-04-23  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
849        * config/arm/arm.c (arm_new_rtx_costs): Handle FMA.
851 2014-04-23  Richard Biener  <rguenther@suse.de>
853         * Makefile.in (OBJS): Remove loop-unswitch.o.
854         * loop-unswitch.c: Delete.
855         * tree-pass.h (make_pass_rtl_unswitch): Remove.
856         * passes.def (pass_rtl_unswitch): Likewise.
857         * loop-init.c (gate_rtl_unswitch): Likewise.
858         (rtl_unswitch): Likewise.
859         (pass_data_rtl_unswitch): Likewise.
860         (pass_rtl_unswitch): Likewise.
861         (make_pass_rtl_unswitch): Likewise.
862         * rtl.h (reversed_condition): Likewise.
863         (compare_and_jump_seq): Likewise.
864         * loop-iv.c (reversed_condition): Move here from loop-unswitch.c
865         and make static.
866         * loop-unroll.c (compare_and_jump_seq): Likewise.
868 2014-04-23  Richard Biener  <rguenther@suse.de>
870         PR tree-optimization/60903
871         * tree-ssa-loop-im.c (analyze_memory_references): Remove
872         commented code block.
873         (execute_sm_if_changed): Properly apply IRREDUCIBLE_LOOP
874         loop flags to newly created BBs and edges.
876 2014-04-23  Nick Clifton  <nickc@redhat.com>
878         * config/msp430/msp430.c (msp430_handle_option): Move function
879         to msp430-common.c
880         (msp430_option_override): Simplify mcu and mcpu option handling.
881         (msp430_is_f5_mcu): Rename to msp430_use_f5_series_hwmult.  Add
882         support for -mhwmult command line option.
883         (has_32bit_hwmult): Rename to use_32bit_hwmult.  Add support for
884         -mhwmult command line option.
885         (msp430_hwmult_enabled): Delete.
886         (msp43o_output_labelref): Add support for -mhwmult command line
887         option.
888         * config/msp430/msp430.md (mulhisi3, umulhisi3, mulsidi3)
889         (umulsidi3): Likewise.
890         * config/msp430/msp430.opt (mmcu): Add Report attribute.
891         (mcpu, mlarge, msmall): Likewise.
892         (mhwmult): New option.
893         * config/msp430/msp430-protos.h (msp430_hwmult_enabled): Remove
894         prototype.
895         (msp430_is_f5_mcu): Remove prototype.
896         (msp430_use_f5_series_hwmult): Add prototype.
897         * config/msp430/msp430-opts.h: New file.
898         * common/config/msp430: New directory.
899         * common/config/msp430/msp430-common.c: New file.
900         * config.gcc (msp430): Remove target_has_targetm_common.
901         * doc/invoke.texi: Document -mhwmult command line option.
903 2014-04-23  Nick Clifton  <nickc@redhat.com>
905         * config/i386/cygwin.h (ENDFILE_SPEC): Include
906         default-manifest.o if it can be found in the search path.
907         * config/i386/mingw32.h (ENDFILE_SPEC): Likewise.
909 2014-04-23  Terry Guo  <terry.guo@arm.com>
911         * config/arm/arm.h (ASM_APP_OFF): Re-define it in a cleaner way.
913 2014-04-23  Richard Biener  <rguenther@suse.de>
915         PR middle-end/60895
916         * tree-inline.c (declare_return_variable): Use mark_addressable.
918 2014-04-23  Richard Biener  <rguenther@suse.de>
920         PR middle-end/60891
921         * loop-init.c (loop_optimizer_init): Make sure to apply
922         LOOPS_MAY_HAVE_MULTIPLE_LATCHES before fixing up loops.
924 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
926         PR sanitizer/60275
927         * common.opt (fsanitize-recover, fsanitize-undefined-trap-on-error):
928         New options.
929         * gcc.c (sanitize_spec_function): Don't return "" for "undefined"
930         if flag_sanitize_undefined_trap_on_error.
931         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW_ABORT,
932         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS_ABORT,
933         BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE_ABORT,
934         BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
935         BUILT_IN_UBSAN_HANDLE_ADD_OVERFLOW_ABORT,
936         BUILT_IN_UBSAN_HANDLE_SUB_OVERFLOW_ABORT,
937         BUILT_IN_UBSAN_HANDLE_MUL_OVERFLOW_ABORT,
938         BUILT_IN_UBSAN_HANDLE_NEGATE_OVERFLOW_ABORT,
939         BUILT_IN_UBSAN_HANDLE_LOAD_INVALID_VALUE_ABORT): New builtins.
940         * ubsan.c (ubsan_instrument_unreachable): Return
941         __builtin_trap () if flag_sanitize_undefined_trap_on_error.
942         (ubsan_expand_null_ifn): Emit __builtin_trap ()
943         if flag_sanitize_undefined_trap_on_error and
944         __ubsan_handle_type_mismatch_abort if !flag_sanitize_recover.
945         (ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
946         instrument_bool_enum_load): Emit __builtin_trap () if
947         flag_sanitize_undefined_trap_on_error and
948         __builtin_handle_*_abort () if !flag_sanitize_recover.
949         * doc/invoke.texi (-fsanitize-recover,
950         -fsanitize-undefined-trap-on-error): Document.
952 2014-04-22  Christian Bruel  <christian.bruel@st.com>
954         * config/sh/sh.md (mov<mode>): Replace movQIHI.
955          Force immediates to SImode.
957 2014-04-22  Sandra Loosemore  <sandra@codesourcery.com>
959         * config/nios2/nios2.md (UNSPEC_ROUND): New.
960         (lroundsfsi2): New.
961         * config/nios2/nios2.opt (mno-custom-round, mcustom-round=): New.
962         * config/nios2/nios2-opts.h (N2FPU_ALL_CODES): Add round.
963         * config/nios2/nios2.c (N2F_NO_ERRNO): Define.
964         (nios2_fpu_insn): Add entry for round.
965         (N2FPU_NO_ERRNO_P): Define.
966         (nios2_custom_check_insns): Add check for N2F_NO_ERRNO and
967         flag_errno_math.
968         * doc/invoke.texi (Nios II Options): Document -mcustom-round.
970 2014-04-22  Richard Henderson  <rth@redhat.com>
972         * config/aarch64/aarch64 (addti3, subti3): New expanders.
973         (add<GPI>3_compare0): Remove leading * from name.
974         (add<GPI>3_carryin): Likewise.
975         (sub<GPI>3_compare0): Likewise.
976         (sub<GPI>3_carryin): Likewise.
977         (<su_optab>mulditi3): New expander.
978         (multi3): New expander.
979         (madd<GPI>): Remove leading * from name.
981 2014-04-22  Martin Jambor  <mjambor@suse.cz>
983         * cgraphclones.c (cgraph_function_versioning): Copy
984         ipa_transforms_to_apply instead of asserting it is empty.
986 2014-04-22  H.J. Lu  <hongjiu.lu@intel.com>
988         PR target/60868
989         * config/i386/i386.c (ix86_expand_set_or_movmem): Call counter_mode
990         on count_exp to get mode.
992 2014-04-22  Andrew Pinski  <apinski@cavium.com>
994         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
995         Handle TLS for ILP32.
996         * config/aarch64/aarch64.md (tlsie_small): Rename to ...
997         (tlsie_small_<mode>): this and handle PTR.
998         (tlsie_small_sidi): New pattern.
999         (tlsle_small): Change to an expand to handle ILP32.
1000         (tlsle_small_<mode>): New pattern.
1001         (tlsdesc_small): Rename to ...
1002         (tlsdesc_small_<mode>): this and handle PTR.
1004 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1006         * config/aarch64/aarch64.c (TARGET_FLAGS_REGNUM): Define.
1008 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1010         * config/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
1011         (aarch64_types_signed_unsigned_qualifiers): Qualifier added.
1012         (aarch64_types_signed_poly_qualifiers): Likewise.
1013         (aarch64_types_unsigned_signed_qualifiers): Likewise.
1014         (aarch64_types_poly_signed_qualifiers): Likewise.
1015         (TYPES_REINTERP_SS): Type macro added.
1016         (TYPES_REINTERP_SU): Likewise.
1017         (TYPES_REINTERP_SP): Likewise.
1018         (TYPES_REINTERP_US): Likewise.
1019         (TYPES_REINTERP_PS): Likewise.
1020         (aarch64_fold_builtin): New expression folding added.
1021         * config/aarch64/aarch64-simd-builtins.def (REINTERP):
1022         Declarations removed.
1023         (REINTERP_SS): Declarations added.
1024         (REINTERP_US): Likewise.
1025         (REINTERP_PS): Likewise.
1026         (REINTERP_SU): Likewise.
1027         (REINTERP_SP): Likewise.
1028         * config/aarch64/arm_neon.h (vreinterpret_p8_f64): Implemented.
1029         (vreinterpretq_p8_f64): Likewise.
1030         (vreinterpret_p16_f64): Likewise.
1031         (vreinterpretq_p16_f64): Likewise.
1032         (vreinterpret_f32_f64): Likewise.
1033         (vreinterpretq_f32_f64): Likewise.
1034         (vreinterpret_f64_f32): Likewise.
1035         (vreinterpret_f64_p8): Likewise.
1036         (vreinterpret_f64_p16): Likewise.
1037         (vreinterpret_f64_s8): Likewise.
1038         (vreinterpret_f64_s16): Likewise.
1039         (vreinterpret_f64_s32): Likewise.
1040         (vreinterpret_f64_s64): Likewise.
1041         (vreinterpret_f64_u8): Likewise.
1042         (vreinterpret_f64_u16): Likewise.
1043         (vreinterpret_f64_u32): Likewise.
1044         (vreinterpret_f64_u64): Likewise.
1045         (vreinterpretq_f64_f32): Likewise.
1046         (vreinterpretq_f64_p8): Likewise.
1047         (vreinterpretq_f64_p16): Likewise.
1048         (vreinterpretq_f64_s8): Likewise.
1049         (vreinterpretq_f64_s16): Likewise.
1050         (vreinterpretq_f64_s32): Likewise.
1051         (vreinterpretq_f64_s64): Likewise.
1052         (vreinterpretq_f64_u8): Likewise.
1053         (vreinterpretq_f64_u16): Likewise.
1054         (vreinterpretq_f64_u32): Likewise.
1055         (vreinterpretq_f64_u64): Likewise.
1056         (vreinterpret_s64_f64): Likewise.
1057         (vreinterpretq_s64_f64): Likewise.
1058         (vreinterpret_u64_f64): Likewise.
1059         (vreinterpretq_u64_f64): Likewise.
1060         (vreinterpret_s8_f64): Likewise.
1061         (vreinterpretq_s8_f64): Likewise.
1062         (vreinterpret_s16_f64): Likewise.
1063         (vreinterpretq_s16_f64): Likewise.
1064         (vreinterpret_s32_f64): Likewise.
1065         (vreinterpretq_s32_f64): Likewise.
1066         (vreinterpret_u8_f64): Likewise.
1067         (vreinterpretq_u8_f64): Likewise.
1068         (vreinterpret_u16_f64): Likewise.
1069         (vreinterpretq_u16_f64): Likewise.
1070         (vreinterpret_u32_f64): Likewise.
1071         (vreinterpretq_u32_f64): Likewise.
1073 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1075         * config/aarch64/aarch64/aarch64-builtins.c (TYPES_REINTERP): Removed.
1076         * config/aarch64/aarch64/aarch64-simd-builtins.def (REINTERP): Removed.
1077         (vreinterpret_p8_s8): Likewise.
1078         * config/aarch64/aarch64/arm_neon.h (vreinterpret_p8_s8): Uses cast.
1079         (vreinterpret_p8_s16): Likewise.
1080         (vreinterpret_p8_s32): Likewise.
1081         (vreinterpret_p8_s64): Likewise.
1082         (vreinterpret_p8_f32): Likewise.
1083         (vreinterpret_p8_u8): Likewise.
1084         (vreinterpret_p8_u16): Likewise.
1085         (vreinterpret_p8_u32): Likewise.
1086         (vreinterpret_p8_u64): Likewise.
1087         (vreinterpret_p8_p16): Likewise.
1088         (vreinterpretq_p8_s8): Likewise.
1089         (vreinterpretq_p8_s16): Likewise.
1090         (vreinterpretq_p8_s32): Likewise.
1091         (vreinterpretq_p8_s64): Likewise.
1092         (vreinterpretq_p8_f32): Likewise.
1093         (vreinterpretq_p8_u8): Likewise.
1094         (vreinterpretq_p8_u16): Likewise.
1095         (vreinterpretq_p8_u32): Likewise.
1096         (vreinterpretq_p8_u64): Likewise.
1097         (vreinterpretq_p8_p16): Likewise.
1098         (vreinterpret_p16_s8): Likewise.
1099         (vreinterpret_p16_s16): Likewise.
1100         (vreinterpret_p16_s32): Likewise.
1101         (vreinterpret_p16_s64): Likewise.
1102         (vreinterpret_p16_f32): Likewise.
1103         (vreinterpret_p16_u8): Likewise.
1104         (vreinterpret_p16_u16): Likewise.
1105         (vreinterpret_p16_u32): Likewise.
1106         (vreinterpret_p16_u64): Likewise.
1107         (vreinterpret_p16_p8): Likewise.
1108         (vreinterpretq_p16_s8): Likewise.
1109         (vreinterpretq_p16_s16): Likewise.
1110         (vreinterpretq_p16_s32): Likewise.
1111         (vreinterpretq_p16_s64): Likewise.
1112         (vreinterpretq_p16_f32): Likewise.
1113         (vreinterpretq_p16_u8): Likewise.
1114         (vreinterpretq_p16_u16): Likewise.
1115         (vreinterpretq_p16_u32): Likewise.
1116         (vreinterpretq_p16_u64): Likewise.
1117         (vreinterpretq_p16_p8): Likewise.
1118         (vreinterpret_f32_s8): Likewise.
1119         (vreinterpret_f32_s16): Likewise.
1120         (vreinterpret_f32_s32): Likewise.
1121         (vreinterpret_f32_s64): Likewise.
1122         (vreinterpret_f32_u8): Likewise.
1123         (vreinterpret_f32_u16): Likewise.
1124         (vreinterpret_f32_u32): Likewise.
1125         (vreinterpret_f32_u64): Likewise.
1126         (vreinterpret_f32_p8): Likewise.
1127         (vreinterpret_f32_p16): Likewise.
1128         (vreinterpretq_f32_s8): Likewise.
1129         (vreinterpretq_f32_s16): Likewise.
1130         (vreinterpretq_f32_s32): Likewise.
1131         (vreinterpretq_f32_s64): Likewise.
1132         (vreinterpretq_f32_u8): Likewise.
1133         (vreinterpretq_f32_u16): Likewise.
1134         (vreinterpretq_f32_u32): Likewise.
1135         (vreinterpretq_f32_u64): Likewise.
1136         (vreinterpretq_f32_p8): Likewise.
1137         (vreinterpretq_f32_p16): Likewise.
1138         (vreinterpret_s64_s8): Likewise.
1139         (vreinterpret_s64_s16): Likewise.
1140         (vreinterpret_s64_s32): Likewise.
1141         (vreinterpret_s64_f32): Likewise.
1142         (vreinterpret_s64_u8): Likewise.
1143         (vreinterpret_s64_u16): Likewise.
1144         (vreinterpret_s64_u32): Likewise.
1145         (vreinterpret_s64_u64): Likewise.
1146         (vreinterpret_s64_p8): Likewise.
1147         (vreinterpret_s64_p16): Likewise.
1148         (vreinterpretq_s64_s8): Likewise.
1149         (vreinterpretq_s64_s16): Likewise.
1150         (vreinterpretq_s64_s32): Likewise.
1151         (vreinterpretq_s64_f32): Likewise.
1152         (vreinterpretq_s64_u8): Likewise.
1153         (vreinterpretq_s64_u16): Likewise.
1154         (vreinterpretq_s64_u32): Likewise.
1155         (vreinterpretq_s64_u64): Likewise.
1156         (vreinterpretq_s64_p8): Likewise.
1157         (vreinterpretq_s64_p16): Likewise.
1158         (vreinterpret_u64_s8): Likewise.
1159         (vreinterpret_u64_s16): Likewise.
1160         (vreinterpret_u64_s32): Likewise.
1161         (vreinterpret_u64_s64): Likewise.
1162         (vreinterpret_u64_f32): Likewise.
1163         (vreinterpret_u64_u8): Likewise.
1164         (vreinterpret_u64_u16): Likewise.
1165         (vreinterpret_u64_u32): Likewise.
1166         (vreinterpret_u64_p8): Likewise.
1167         (vreinterpret_u64_p16): Likewise.
1168         (vreinterpretq_u64_s8): Likewise.
1169         (vreinterpretq_u64_s16): Likewise.
1170         (vreinterpretq_u64_s32): Likewise.
1171         (vreinterpretq_u64_s64): Likewise.
1172         (vreinterpretq_u64_f32): Likewise.
1173         (vreinterpretq_u64_u8): Likewise.
1174         (vreinterpretq_u64_u16): Likewise.
1175         (vreinterpretq_u64_u32): Likewise.
1176         (vreinterpretq_u64_p8): Likewise.
1177         (vreinterpretq_u64_p16): Likewise.
1178         (vreinterpret_s8_s16): Likewise.
1179         (vreinterpret_s8_s32): Likewise.
1180         (vreinterpret_s8_s64): Likewise.
1181         (vreinterpret_s8_f32): Likewise.
1182         (vreinterpret_s8_u8): Likewise.
1183         (vreinterpret_s8_u16): Likewise.
1184         (vreinterpret_s8_u32): Likewise.
1185         (vreinterpret_s8_u64): Likewise.
1186         (vreinterpret_s8_p8): Likewise.
1187         (vreinterpret_s8_p16): Likewise.
1188         (vreinterpretq_s8_s16): Likewise.
1189         (vreinterpretq_s8_s32): Likewise.
1190         (vreinterpretq_s8_s64): Likewise.
1191         (vreinterpretq_s8_f32): Likewise.
1192         (vreinterpretq_s8_u8): Likewise.
1193         (vreinterpretq_s8_u16): Likewise.
1194         (vreinterpretq_s8_u32): Likewise.
1195         (vreinterpretq_s8_u64): Likewise.
1196         (vreinterpretq_s8_p8): Likewise.
1197         (vreinterpretq_s8_p16): Likewise.
1198         (vreinterpret_s16_s8): Likewise.
1199         (vreinterpret_s16_s32): Likewise.
1200         (vreinterpret_s16_s64): Likewise.
1201         (vreinterpret_s16_f32): Likewise.
1202         (vreinterpret_s16_u8): Likewise.
1203         (vreinterpret_s16_u16): Likewise.
1204         (vreinterpret_s16_u32): Likewise.
1205         (vreinterpret_s16_u64): Likewise.
1206         (vreinterpret_s16_p8): Likewise.
1207         (vreinterpret_s16_p16): Likewise.
1208         (vreinterpretq_s16_s8): Likewise.
1209         (vreinterpretq_s16_s32): Likewise.
1210         (vreinterpretq_s16_s64): Likewise.
1211         (vreinterpretq_s16_f32): Likewise.
1212         (vreinterpretq_s16_u8): Likewise.
1213         (vreinterpretq_s16_u16): Likewise.
1214         (vreinterpretq_s16_u32): Likewise.
1215         (vreinterpretq_s16_u64): Likewise.
1216         (vreinterpretq_s16_p8): Likewise.
1217         (vreinterpretq_s16_p16): Likewise.
1218         (vreinterpret_s32_s8): Likewise.
1219         (vreinterpret_s32_s16): Likewise.
1220         (vreinterpret_s32_s64): Likewise.
1221         (vreinterpret_s32_f32): Likewise.
1222         (vreinterpret_s32_u8): Likewise.
1223         (vreinterpret_s32_u16): Likewise.
1224         (vreinterpret_s32_u32): Likewise.
1225         (vreinterpret_s32_u64): Likewise.
1226         (vreinterpret_s32_p8): Likewise.
1227         (vreinterpret_s32_p16): Likewise.
1228         (vreinterpretq_s32_s8): Likewise.
1229         (vreinterpretq_s32_s16): Likewise.
1230         (vreinterpretq_s32_s64): Likewise.
1231         (vreinterpretq_s32_f32): Likewise.
1232         (vreinterpretq_s32_u8): Likewise.
1233         (vreinterpretq_s32_u16): Likewise.
1234         (vreinterpretq_s32_u32): Likewise.
1235         (vreinterpretq_s32_u64): Likewise.
1236         (vreinterpretq_s32_p8): Likewise.
1237         (vreinterpretq_s32_p16): Likewise.
1238         (vreinterpret_u8_s8): Likewise.
1239         (vreinterpret_u8_s16): Likewise.
1240         (vreinterpret_u8_s32): Likewise.
1241         (vreinterpret_u8_s64): Likewise.
1242         (vreinterpret_u8_f32): Likewise.
1243         (vreinterpret_u8_u16): Likewise.
1244         (vreinterpret_u8_u32): Likewise.
1245         (vreinterpret_u8_u64): Likewise.
1246         (vreinterpret_u8_p8): Likewise.
1247         (vreinterpret_u8_p16): Likewise.
1248         (vreinterpretq_u8_s8): Likewise.
1249         (vreinterpretq_u8_s16): Likewise.
1250         (vreinterpretq_u8_s32): Likewise.
1251         (vreinterpretq_u8_s64): Likewise.
1252         (vreinterpretq_u8_f32): Likewise.
1253         (vreinterpretq_u8_u16): Likewise.
1254         (vreinterpretq_u8_u32): Likewise.
1255         (vreinterpretq_u8_u64): Likewise.
1256         (vreinterpretq_u8_p8): Likewise.
1257         (vreinterpretq_u8_p16): Likewise.
1258         (vreinterpret_u16_s8): Likewise.
1259         (vreinterpret_u16_s16): Likewise.
1260         (vreinterpret_u16_s32): Likewise.
1261         (vreinterpret_u16_s64): Likewise.
1262         (vreinterpret_u16_f32): Likewise.
1263         (vreinterpret_u16_u8): Likewise.
1264         (vreinterpret_u16_u32): Likewise.
1265         (vreinterpret_u16_u64): Likewise.
1266         (vreinterpret_u16_p8): Likewise.
1267         (vreinterpret_u16_p16): Likewise.
1268         (vreinterpretq_u16_s8): Likewise.
1269         (vreinterpretq_u16_s16): Likewise.
1270         (vreinterpretq_u16_s32): Likewise.
1271         (vreinterpretq_u16_s64): Likewise.
1272         (vreinterpretq_u16_f32): Likewise.
1273         (vreinterpretq_u16_u8): Likewise.
1274         (vreinterpretq_u16_u32): Likewise.
1275         (vreinterpretq_u16_u64): Likewise.
1276         (vreinterpretq_u16_p8): Likewise.
1277         (vreinterpretq_u16_p16): Likewise.
1278         (vreinterpret_u32_s8): Likewise.
1279         (vreinterpret_u32_s16): Likewise.
1280         (vreinterpret_u32_s32): Likewise.
1281         (vreinterpret_u32_s64): Likewise.
1282         (vreinterpret_u32_f32): Likewise.
1283         (vreinterpret_u32_u8): Likewise.
1284         (vreinterpret_u32_u16): Likewise.
1285         (vreinterpret_u32_u64): Likewise.
1286         (vreinterpret_u32_p8): Likewise.
1287         (vreinterpret_u32_p16): Likewise.
1288         (vreinterpretq_u32_s8): Likewise.
1289         (vreinterpretq_u32_s16): Likewise.
1290         (vreinterpretq_u32_s32): Likewise.
1291         (vreinterpretq_u32_s64): Likewise.
1292         (vreinterpretq_u32_f32): Likewise.
1293         (vreinterpretq_u32_u8): Likewise.
1294         (vreinterpretq_u32_u16): Likewise.
1295         (vreinterpretq_u32_u64): Likewise.
1296         (vreinterpretq_u32_p8): Likewise.
1297         (vreinterpretq_u32_p16): Likewise.
1299 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1301         * gcc/config/aarch64/aarch64-simd.md (aarch64_s<optab><mode>):
1302         Pattern extended.
1303         * config/aarch64/aarch64-simd-builtins.def (sqneg): Iterator
1304         extended.
1305         (sqabs): Likewise.
1306         * config/aarch64/arm_neon.h (vqneg_s64): New intrinsic.
1307         (vqnegd_s64): Likewise.
1308         (vqabs_s64): Likewise.
1309         (vqabsd_s64): Likewise.
1311 2014-04-22  Richard Henderson  <rth@redhat.com>
1313         * config/sparc/sparc.c (sparc_init_modes): Hoist GET_MODE_SIZE
1314         computation to the top of the loop.
1316 2014-04-22  Renlin  <renlin.li@arm.com>
1317             Jiong Wang  <jiong.wang@arm.com>
1319         * config/aarch64/aarch64.h (aarch64_frame): Delete "fp_lr_offset".
1320         * config/aarch64/aarch64.c (aarch64_layout_frame)
1321         (aarch64_initial_elimination_offset): Likewise.
1323 2014-04-22  Marcus Shawcroft  <marcus.shawcroft@arm.com>
1325         * config/aarch64/aarch64.c (aarch64_initial_elimination_offset):
1326         Fix indentation.
1328 2014-04-22  Richard Sandiford  <rdsandiford@googlemail.com>
1330         * machmode.h (bitwise_mode_for_mode): Declare.
1331         * stor-layout.h (bitwise_type_for_mode): Likewise.
1332         * stor-layout.c (bitwise_mode_for_mode): New function.
1333         (bitwise_type_for_mode): Likewise.
1334         * builtins.c (fold_builtin_memory_op): Use it instead of
1335         int_mode_for_mode and build_nonstandard_integer_type.
1337 2014-04-22  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1339         * config.gcc (enable_obsolete): Remove *-*-solaris2.9*.
1340         (*-*-solaris2.[0-9] | *-*-solaris2.[0-9].*): Mark unsupported.
1341         (*-*-solaris2*): Simplify.
1342         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Likewise.
1343         (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*): Remove
1344         *-*-solaris2.9* handling.
1346         * configure.ac (gcc_cv_as_hidden): Remove test for Solaris 9/x86
1347         as bug.
1348         (gcc_cv_ld_hidden): Remove *-*-solaris2.9* handling.
1349         (ld_tls_support): Remove i?86-*-solaris2.9, sparc*-*-solaris2.9
1350         handling, simplify.
1351         (gcc_cv_as_gstabs_flag): Remove workaround for Solaris 9/x86 as bug.
1352         * configure: Regenerate.
1354         * config/i386/sol2-9.h: Remove.
1356         * doc/install.texi (Specific, i?86-*-solaris2.9): Remove.
1357         (Specific, *-*-solaris2*): Mention Solaris 9 support removal.
1358         Remove Solaris 9 references.
1360 2014-04-22  Vidya Praveen  <vidyapraveen@arm.com>
1362         * aarch64.md (float<GPI:mode><GPF:mode>2): Remove.
1363         (floatuns<GPI:mode><GPF:mode>2): Remove.
1364         (<optab><fcvt_target><GPF:mode>2): New pattern for equal width float
1365         and floatuns conversions.
1366         (<optab><fcvt_iesize><GPF:mode>2): New pattern for inequal width float
1367         and floatuns conversions.
1368         * iterators.md (fcvt_target, FCVT_TARGET): Support SF and DF modes.
1369         (w1,w2): New mode attributes for inequal width conversions.
1371 2014-04-22  Renlin Li  <Renlin.Li@arm.com>
1373         * config/aarch64/aarch64.c (aarch64_print_operand_address): Adjust
1374         the output asm format.
1376 2014-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
1378         * config/aarch64/aarch64-simd.md
1379         (aarch64_cm<optab>di): Always split.
1380         (*aarch64_cm<optab>di): New.
1381         (aarch64_cmtstdi): Always split.
1382         (*aarch64_cmtstdi): New.
1384 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1386         PR tree-optimization/60823
1387         * omp-low.c (ipa_simd_modify_function_body): Go through
1388         all SSA_NAMEs and for those refering to vector arguments
1389         which are going to be replaced adjust SSA_NAME_VAR and,
1390         if it is a default definition, change it into a non-default
1391         definition assigned at the beginning of function from new_decl.
1392         (ipa_simd_modify_stmt_ops): Rewritten.
1393         * tree-dfa.c (set_ssa_default_def): When removing default def,
1394         check for NULL loc instead of NULL *loc.
1396 2014-04-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1398         * config/arm/arm.c (arm_hard_regno_mode_ok): Loosen
1399         restrictions on core registers for DImode values in Thumb2.
1401 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1403         * config/arm/arm.md (*anddi_notdi_zesidi): New pattern.
1404         * config/arm/thumb2.md (*iordi_notdi_zesidi): New pattern.
1406 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1408         * config/arm/thumb2.md (*iordi_notdi_di): New pattern.
1409         (*iordi_notzesidi_di): Likewise.
1410         (*iordi_notsesidi_di): Likewise.
1412 2014-04-22  Ian Bolton  <ian.bolton@arm.com>
1414         * config/arm/arm-protos.h (tune_params): New struct members.
1415         * config/arm/arm.c: Initialise tune_params per processor.
1416         (thumb2_reorg): Suppress conversion from t32 to t16 when optimizing
1417         for speed, based on new tune_params.
1419 2014-04-22  Alex Velenko  <Alex.Velenko@arm.com>
1421         * config/aarch64/aarch64-builtins.c (BUILTIN_VDQF_DF): Macro
1422         added.
1423         * config/aarch64/aarch64-simd-builtins.def (frintn): Use added
1424         macro.
1425         * config/aarch64/aarch64-simd.md (<frint_pattern>): Comment
1426         corrected.
1427         * config/aarch64/aarch64.md (<frint_pattern>): Likewise.
1428         * config/aarch64/arm_neon.h (vrnd_f64): Added.
1429         (vrnda_f64): Likewise.
1430         (vrndi_f64): Likewise.
1431         (vrndm_f64): Likewise.
1432         (vrndn_f64): Likewise.
1433         (vrndp_f64): Likewise.
1434         (vrndx_f64): Likewise.
1436 2014-04-22  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1438         * config/arm/arm.c (arm_print_operand, thumb_exit): Make sure
1439         GET_MODE_SIZE argument is enum machine_mode.
1441 2014-04-22  Jakub Jelinek  <jakub@redhat.com>
1443         PR target/60910
1444         * config/sparc/sparc.c (sparc_init_modes): Pass enum machine_mode
1445         value instead of int to GET_MODE_CLASS and GET_MODE_SIZE macros.
1447 2014-04-22  Lin Zuojian  <manjian2006@gmail.com>
1449         PR middle-end/60281
1450         * asan.c (asan_emit_stack_protection): Force the base to align to
1451         appropriate bits if STRICT_ALIGNMENT.  Set shadow_mem align to
1452         appropriate bits if STRICT_ALIGNMENT.
1453         * cfgexpand.c (expand_stack_vars): Set base_align appropriately
1454         when asan is on.
1455         (expand_used_vars): Leave a space in the stack frame for alignment
1456         if STRICT_ALIGNMENT.
1458 2014-04-21  David Malcolm  <dmalcolm@redhat.com>
1460         * gimple.h (gimple_assign_single_p): Accept a const_gimple rather
1461         than a gimple.
1462         (gimple_store_p): Likewise.
1463         (gimple_assign_load_p): Likewise.
1464         (gimple_assign_cast_p): Likewise.
1465         (gimple_clobber_p): Likewise.
1467         * doc/gimple.texi (gimple_assign_cast_p): Accept a const_gimple
1468         rather than a gimple.
1469         (gimple_assign_cast_p): Likewise.
1471 2014-04-21  Michael Meissner  <meissner@linux.vnet.ibm.com>
1473         PR target/60735
1474         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64 case):
1475         If mode is DDmode and TARGET_E500_DOUBLE allow move.
1477         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print some
1478         more debug information for E500 if -mdebug=reg.
1480 2014-04-21  Uros Bizjak  <ubizjak@gmail.com>
1482         PR target/60909
1483         * config/i386/i386.c (ix86_expand_builtin)
1484         <case IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Use temporary
1485         register for target RTX.
1486         <case IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Ditto.
1488 2014-04-18  Cong Hou  <congh@google.com>
1490         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Enhance
1491         the widen-mult pattern by handling two operands with different sizes,
1492         and operands whose size is smaller than half of the result type.
1494 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
1496         * ipa-inline.h (INLINE_HINT_known_hot): New hint.
1497         * ipa-inline-analysis.c (dump_inline_hints): Dump it.
1498         (do_estimate_edge_time): Compute it.
1499         * ipa-inline.c (want_inline_small_function_p): Bypass
1500         INLINE_INSNS_AUTO/SINGLE limits for calls that are known to be hot.
1502 2014-04-18  Jan Hubicka  <hubicka@ucw.cz>
1504         * ipa-inline.c (spec_rem): New static variable.
1505         (dump_overall_stats): New function.
1506         (dump_inline_stats): New function.
1508 2014-04-18  Richard Henderson  <rth@redhat.com>
1510         * config/aarch64/aarch64.c (aarch64_register_move_cost): Pass a mode
1511         to GET_MODE_SIZE, not a reg_class_t.
1513 2014-04-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1515         * config/rs6000/vsx.md (vsx_xxmrghw_<mode>): Adjust for little-endian.
1516         (vsx_xxmrglw_<mode>): Likewise.
1518 2014-04-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
1520         PR target/60876
1521         * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Make sure
1522         GET_MODE_SIZE gets passed an enum machine_mode type and not integer.
1523         (rs6000_init_hard_regno_mode_ok): Likewise.
1525 2014-04-17  Jan Hubicka  <hubicka@ucw.cz>
1527         * ipa-inline.c (inline_small_functions): Account only non-cold
1528         functions.
1529         * doc/invoke.texi (inline-unit-growth): Update documentation.
1531 2014-04-17  Pat Haugen  <pthaugen@us.ibm.com>
1533         * config/rs6000/rs6000.md (addti3, subti3): New.
1535 2014-04-17  H.J. Lu  <hongjiu.lu@intel.com>
1537         PR target/60863
1538         * config/i386/i386.c (ix86_expand_clear): Remove outdated
1539         comment.  Check optimize_insn_for_size_p instead of
1540         optimize_insn_for_speed_p.
1542 2014-04-17  Martin Jambor  <mjambor@suse.cz>
1544         * gimple-iterator.c (gsi_start_edge): New function.
1545         * gimple-iterator.h (gsi_start_edge): Declare.
1546         * tree-sra.c (single_non_eh_succ): New function.
1547         (disqualify_ops_if_throwing_stmt): Renamed to
1548         disqualify_if_bad_bb_terminating_stmt.  Allow throwing statements
1549         having one non-EH successor BB.
1550         (sra_modify_expr): If stmt ends bb, use single non-EH successor to
1551         generate loads into replacements.
1552         (sra_modify_assign): Likewise and and also use the simple path for
1553         such statements.
1554         (sra_modify_function_body): Commit statements on edges.
1556 2014-04-17  Richard Biener  <rguenther@suse.de>
1558         PR middle-end/60849
1559         * tree-ssa-propagate.c (valid_gimple_rhs_p): Allow vector
1560         comparison results and add clarifying comment.
1562 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
1564         * genmodes.c (struct mode_data): Add need_bytesize_adj field.
1565         (blank_mode): Initialize it.
1566         (emit_mode_size_inline, emit_mode_nunits_inline,
1567         emit_mode_inner_inline): New functions.
1568         (emit_insn_modes_h): Call them and surround their output with
1569         #if GCC_VERSION >= 4001 ... #endif.
1570         * machmode.h (GET_MODE_SIZE, GET_MODE_NUNITS, GET_MODE_INNER):
1571         For GCC_VERSION >= 4001 use mode_*_inline routines instead of
1572         mode_* arrays if the argument is __builtin_constant_p.
1573         * lower-subreg.c (dump_choices): Make sure GET_MODE_SIZE argument
1574         is enum machine_mode.
1576 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1578         * passes.c (opt_pass::execute): Adjust.
1579         (pass_manager::execute_pass_mode_switching): Likewise.
1580         (early_local_passes::execute): Likewise.
1581         (execute_one_pass): Pass cfun to the pass's execute method.
1582         * tree-pass.h (opt_pass::execute): Add function * argument.
1583         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
1584         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
1585         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
1586         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1587         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
1588         config/sparc/sparc.c, cprop.c, dce.c, df-core.c, dse.c, dwarf2cfi.c,
1589         except.c, final.c, function.c, fwprop.c, gcse.c, gimple-low.c,
1590         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
1591         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
1592         ipa-inline-analysis.c, ipa-inline.c, ipa-profile.c, ipa-pure-const.c,
1593         ipa-reference.c, ipa-split.c, ipa.c, ira.c, jump.c, loop-init.c,
1594         lower-subreg.c, mode-switching.c, omp-low.c, postreload-gcse.c,
1595         postreload.c, predict.c, recog.c, ree.c, reg-stack.c, regcprop.c,
1596         reginfo.c, regrename.c, reorg.c, sched-rgn.c, stack-ptr-mod.c,
1597         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
1598         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
1599         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
1600         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-ssa-ccp.c,
1601         tree-ssa-copy.c, tree-ssa-copyrename.c, tree-ssa-dce.c,
1602         tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1603         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1604         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1605         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1606         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1607         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1608         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1609         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
1610         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
1611         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
1612         Adjust.
1614 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1616         * passes.c (opt_pass::gate): Take function * argument.
1617         (gate_all_early_local_passes): Merge into
1618         (early_local_passes::gate): this.
1619         (gate_all_early_optimizations): Merge into
1620         (all_early_optimizations::gate): this.
1621         (gate_all_optimizations): Mege into
1622         (all_optimizations::gate): this.
1623         (gate_all_optimizations_g): Merge into
1624         (all_optimizations_g::gate): this.
1625         (gate_rest_of_compilation): Mege into
1626         (rest_of_compilation::gate): this.
1627         (gate_postreload): Merge into
1628         (postreload::gate): this.
1629         (dump_one_pass): Pass cfun to the pass's gate method.
1630         (execute_ipa_summary_passes): Likewise.
1631         (execute_one_pass): Likewise.
1632         (ipa_write_summaries_2): Likewise.
1633         (ipa_write_optimization_summaries_1): Likewise.
1634         (ipa_read_summaries_1): Likewise.
1635         (ipa_read_optimization_summaries_1): Likewise.
1636         (execute_ipa_stmt_fixups): Likewise.
1637         * tree-pass.h (opt_pass::gate): Add function * argument.
1638         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c,
1639         combine-stack-adj.c, combine.c, compare-elim.c,
1640         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1641         config/rl78/rl78.c, config/sh/sh_optimize_sett_clrt.cc,
1642         config/sh/sh_treg_combine.cc, config/sparc/sparc.c, cprop.c, cse.c,
1643         dce.c, df-core.c, dse.c, dwarf2cfi.c, except.c,  fwprop.c, gcse.c,
1644         gimple-ssa-isolate-paths.c, gimple-ssa-strength-reduction.c,
1645         graphite.c, ifcvt.c, init-regs.c, ipa-cp.c, ipa-devirt.c,
1646         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
1647         loop-init.c, lower-subreg.c, mode-switching.c, modulo-sched.c,
1648         omp-low.c, postreload-gcse.c, postreload.c, predict.c, recog.c, ree.c,
1649         reg-stack.c, regcprop.c, regrename.c, reorg.c, sched-rgn.c,
1650         store-motion.c, tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c,
1651         tree-cfgcleanup.c, tree-complex.c, tree-eh.c, tree-emutls.c,
1652         tree-if-conv.c, tree-into-ssa.c, tree-loop-distribution.c,
1653         tree-nrv.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
1654         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
1655         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1656         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1657         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1658         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1659         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1660         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1661         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1662         tree-ssa.c, tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
1663         tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c, tsan.c, ubsan.c,
1664         var-tracking.c, vtable-verify.c, web.c: Adjust.
1666 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1668         * configure.ac: Check for -Woverloaded-virtual and enable it if found.
1669         * configure: Regenerate.
1671 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1673         * passes.c (dump_one_pass): don't check pass->has_gate.
1674         (execute_ipa_summary_passes): Likewise.
1675         (execute_one_pass): Likewise.
1676         (ipa_write_summaries_2): Likewise.
1677         (ipa_write_optimization_summaries_1): Likewise.
1678         (ipa_read_optimization_summaries_1): Likewise.
1679         (execute_ipa_stmt_fixups): Likewise.
1680         * tree-pass.h (pass_data::has_gate): Remove.
1681         * asan.c, auto-inc-dec.c, bb-reorder.c, bt-load.c, cfgcleanup.c,
1682         cfgexpand.c, cfgrtl.c, cgraphbuild.c, combine-stack-adj.c, combine.c,
1683         compare-elim.c, config/arc/arc.c, config/epiphany/mode-switch-use.c,
1684         config/epiphany/resolve-sw-modes.c, config/i386/i386.c,
1685         config/mips/mips.c, config/rl78/rl78.c, config/s390/s390.c,
1686         config/sh/sh_optimize_sett_clrt.cc, config/sh/sh_treg_combine.cc,
1687         config/sparc/sparc.c, cprop.c, cse.c, dce.c, df-core.c, dse.c,
1688         dwarf2cfi.c, except.c, final.c, function.c, fwprop.c, gcse.c,
1689         gimple-low.c, gimple-ssa-isolate-paths.c,
1690         gimple-ssa-strength-reduction.c, graphite.c, ifcvt.c, init-regs.c,
1691         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
1692         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c, ipa.c,
1693         ira.c, jump.c, loop-init.c, lower-subreg.c, mode-switching.c,
1694         modulo-sched.c, omp-low.c, postreload-gcse.c, postreload.c, predict.c,
1695         recog.c, ree.c, reg-stack.c, regcprop.c, reginfo.c, regrename.c,
1696         reorg.c, sched-rgn.c, stack-ptr-mod.c, store-motion.c, tracer.c,
1697         trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
1698         tree-complex.c, tree-eh.c, tree-emutls.c, tree-if-conv.c,
1699         tree-into-ssa.c, tree-loop-distribution.c, tree-nrv.c,
1700         tree-object-size.c, tree-parloops.c, tree-predcom.c, tree-profile.c,
1701         tree-sra.c, tree-ssa-ccp.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
1702         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
1703         tree-ssa-ifcombine.c, tree-ssa-loop-ch.c, tree-ssa-loop-im.c,
1704         tree-ssa-loop-ivcanon.c, tree-ssa-loop-prefetch.c,
1705         tree-ssa-loop-unswitch.c, tree-ssa-loop.c, tree-ssa-math-opts.c,
1706         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
1707         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
1708         tree-ssa-structalias.c, tree-ssa-uncprop.c, tree-ssa-uninit.c,
1709         tree-ssa.c, tree-ssanames.c, tree-stdarg.c, tree-switch-conversion.c,
1710         tree-tailcall.c, tree-vect-generic.c, tree-vectorizer.c, tree-vrp.c,
1711         tree.c, tsan.c, ubsan.c, var-tracking.c, vtable-verify.c, web.c:
1712         Adjust.
1714 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1716         * pass_manager.h (pass_manager::register_dump_files_1): Remove
1717         declaration.
1718         * passes.c (pass_manager::register_dump_files_1): Merge into
1719         (pass_manager::register_dump_files): this, and remove its handling of
1720         properties since the pass always has the properties anyway.
1721         (pass_manager::pass_manager): Adjust.
1723 2014-04-17  Trevor Saunders  <tsaunders@mozilla.com>
1725         * pass_manager.h (pass_manager::register_dump_files_1): Adjust.
1726         * passes.c (pass_manager::register_dump_files_1): Remove dead code
1727         dealing with properties.
1728         (pass_manager::register_dump_files): Adjust.
1730 2014-03-20  Mark Wielaard  <mjw@redhat.com>
1732         * dwarf2out.c (add_bound_info): If HOST_WIDE_INT is big enough,
1733         then represent the bound as normal constant value.
1735 2014-04-17  Jakub Jelinek  <jakub@redhat.com>
1737         PR target/60847
1738         Forward port from 4.8 branch
1739         2013-07-19  Kirill Yukhin  <kirill.yukhin@intel.com>
1741         * config/i386/bmiintrin.h (_blsi_u32): New.
1742         (_blsi_u64): Ditto.
1743         (_blsr_u32): Ditto.
1744         (_blsr_u64): Ditto.
1745         (_blsmsk_u32): Ditto.
1746         (_blsmsk_u64): Ditto.
1747         (_tzcnt_u32): Ditto.
1748         (_tzcnt_u64): Ditto.
1750 2014-04-17  Kito Cheng  <kito@0xlab.org>
1752         * gcc.c (used_arg): Prevent out of bound access for multilib_options.
1754 2014-04-17  Richard Biener  <rguenther@suse.de>
1756         PR middle-end/60849
1757         * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
1758         boolean results for comparisons.
1760 2014-04-17  Richard Biener  <rguenther@suse.de>
1762         PR tree-optimization/60836
1763         * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
1764         initial PHI args to be gimple values.
1766 2014-04-17   Richard Biener  <rguenther@suse.de>
1768         PR tree-optimization/60841
1769         * tree-vect-data-refs.c (vect_analyze_data_refs): Count stmts.
1770         * tree-vect-loop.c (vect_analyze_loop_2): Pass down number
1771         of stmts to SLP build.
1772         * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
1773         (vect_analyze_slp): Likewise.
1774         (vect_analyze_slp_instance): Likewise.
1775         (vect_build_slp_tree): Limit overall SLP tree growth.
1776         * tree-vectorizer.h (vect_analyze_data_refs,
1777         vect_analyze_slp): Adjust prototypes.
1779 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
1781         * config/i386/i386.c (x86_add_stmt_cost): Fix vector cost model for
1782         Silvermont.
1784 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
1786         * config/i386/x86-tune.def (TARGET_SLOW_PSHUFB): New tune definition.
1787         * config/i386/i386.h (TARGET_SLOW_PSHUFB): New tune flag.
1788         * config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
1789         for TARGET_SLOW_PSHUFB
1791 2014-04-17  Evgeny Stupachenko  <evstupac@gmail.com>
1793         * config/i386/i386.c (slm_cost): Adjust vec_to_scalar_cost.
1794         * config/i386/i386.c (intel_cost): Ditto.
1796 2014-04-17  Joey Ye  <joey.ye@arm.com>
1798         * opts.c (OPT_fif_conversion, OPT_fif_conversion2): Disable for Og.
1800 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
1802         * opts.c (common_handle_option): Disable -fipa-reference coorectly
1803         with -fuse-profile.
1805 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
1807         * ipa-devirt.c (odr_type_d): Add field all_derivations_known.
1808         (type_all_derivations_known_p): New predicate.
1809         (type_all_ctors_visible_p): New predicate.
1810         (type_possibly_instantiated_p): New predicate.
1811         (get_odr_type): Compute all_derivations_known.
1812         (dump_odr_type): Dump the flag.
1813         (maybe_record_type): Cleanup.
1814         (record_target_from_binfo): Add bases_to_consider array;
1815         record bases for types w/o instances and skip CXX destructor.
1816         (possible_polymorphic_call_targets_1): Add bases_to_consider
1817         and consider_construction parameters; check if type may have instance.
1818         (get_polymorphic_call_info): Set maybe_in_construction to true
1819         when we know nothing.
1820         (record_targets_from_bases): Skip CXX destructors; they are
1821         never called for types in construction.
1822         (possible_polymorphic_call_targets): Do not record target when
1823         type may not have instance.
1825 2014-04-16  Jan Hubicka  <hubicka@ucw.cz>
1827         PR ipa/60854
1828         * ipa.c (symtab_remove_unreachable_nodes): Mark targets of
1829         external aliases alive, too.
1831 2014-04-16  Andrew  Pinski  <apinski@cavium.com>
1833         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Change aarch64 ilp32
1834         definition.
1836 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
1838         * final.c (compute_alignments): Do not apply loop alignment to a block
1839         falling through to the exit.
1841 2014-04-16  Catherine Moore  <clm@codesourcery.com>
1843         * mips.md (*mov<mode>_internal, *movhi_internal, *movqi_internal):
1844         Adjust constraints for microMIPS store patterns.
1846 2014-04-16  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
1848         * config/avr/avr-mcus.def: Correct typo for atxmega256a3bu macro.
1850 2014-04-16  Eric Botcazou  <ebotcazou@adacore.com>
1852         * tree-ssa-operands.c (create_vop_var): Set DECL_IGNORED_P.
1853         (append_use): Run at -O0.
1854         (append_vdef): Likewise.
1855         * tree-ssa-ter.c (ter_is_replaceable_p): Do not special-case -O0.
1856         * tree-ssa-uninit.c (warn_uninitialized_vars): Remove obsolete comment.
1858 2014-04-16  Jakub Jelinek  <jakub@redhat.com>
1860         PR tree-optimization/60844
1861         * tree-ssa-reassoc.c (reassoc_remove_stmt): New function.
1862         (propagate_op_to_single_use, remove_visited_stmt_chain,
1863         linearize_expr, repropagate_negates, reassociate_bb): Use it
1864         instead of gsi_remove.
1866 2014-04-16  Martin Jambor  <mjambor@suse.cz>
1868         * cgraphclones.c (cgraph_create_virtual_clone): Duplicate
1869         ipa_transforms_to_apply.
1870         (cgraph_function_versioning): Assert that old_node has empty
1871         ipa_transforms_to_apply.
1872         * trans-mem.c (ipa_tm_create_version): Likewise.
1873         * tree-inline.c (tree_function_versioning): Do not duplicate
1874         ipa_transforms_to_apply.
1876 2014-04-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1878         PR target/60817
1879         * configure.ac (set_have_as_tls): Merge i[34567]86-*-* and
1880         x86_64-*-* cases.
1881         Pass necessary as flags on 64-bit Solaris/x86.
1882         Use lowercase relocs for x86_64-*-*.
1883         * configure: Regenerate.
1885 2014-04-15  Jan Hubicka  <jh@suse.cz>
1887         * ipa-devirt.c (referenced_from_vtable_p): New predicate.
1888         (maybe_record_node, likely_target_p): Use it.
1890 2014-04-15  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1892         PR target/60839
1893         Revert following patch
1895         2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
1897         PR target/60735
1898         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
1899         software floating point or no floating point registers, do not
1900         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
1901         in GPRs that occurs after we tested for GPRs that would never be
1902         true.
1904         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
1905         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
1906         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
1907         specifically allow DDmode, since that does not use the SPE SIMD
1908         instructions.
1910 2014-03-21  Mark Wielaard  <mjw@redhat.com>
1912         * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
1913         as unsigned or int depending on type and value used.
1915 2014-04-15  Richard Biener  <rguenther@suse.de>
1917         PR rtl-optimization/56965
1918         * alias.c (ncr_compar, nonoverlapping_component_refs_p): Move ...
1919         * tree-ssa-alias.c (ncr_compar, nonoverlapping_component_refs_p):
1920         ... here.
1921         * alias.c (true_dependence_1): Do not call
1922         nonoverlapping_component_refs_p.
1923         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Call
1924         nonoverlapping_component_refs_p.
1925         (indirect_refs_may_alias_p): Likewise.
1927 2014-04-15  Teresa Johnson  <tejohnson@google.com>
1929         * cfg.c (dump_bb_info): Fix flags check.
1930         * tree-cfg.c (remove_bb): Only dump TDF_BLOCKS when removing.
1932 2014-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1934         PR rtl-optimization/60663
1935         * config/arm/arm.c (arm_new_rtx_costs): Improve ASM_OPERANDS case,
1936         avoid 0 cost.
1938 2014-04-15  Richard Biener  <rguenther@suse.de>
1940         * lto-streamer.h (LTO_major_version): Bump to 4.
1942 2014-04-15  Richard Biener  <rguenther@suse.de>
1944         * common.opt (lto_partition_model): New enum.
1945         (flto-partition=): Merge separate options with a single with argument,
1946         add -flto-partition=one support.
1947         * flag-types.h (enum lto_partition_model): Declare.
1948         * opts.c (finish_options): Remove duplicate -flto-partition=
1949         option check.
1950         * lto-wrapper.c (run_gcc): Adjust.
1952 2014-04-15  Richard Biener  <rguenther@suse.de>
1954         * alias.c (ncr_compar): New function.
1955         (nonoverlapping_component_refs_p): Re-implement in O (n log n).
1957 2014-04-15  Richard Biener  <rguenther@suse.de>
1959         * alias.c (record_component_aliases): Do not walk BINFOs.
1961 2014-04-15  Richard Biener  <rguenther@suse.de>
1963         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1964         Add struct function argument and adjust.
1965         (find_func_aliases_for_call): Likewise.
1966         (find_func_aliases): Likewise.
1967         (find_func_clobbers): Likewise.
1968         (intra_create_variable_infos): Likewise.
1969         (compute_points_to_sets): Likewise.
1970         (ipa_pta_execute): Adjust.  Do not push/pop cfun.
1972 2014-04-15  Richard Biener  <rguenther@suse.de>
1974         * tree.c (iterative_hash_expr): Use enum tree_code_class
1975         to store TREE_CODE_CLASS.
1976         (tree_block): Likewise.
1977         (tree_set_block): Likewise.
1978         * tree.h (fold_build_pointer_plus_loc): Use
1979         convert_to_ptrofftype_loc.
1981 2014-04-15  Jakub Jelinek  <jakub@redhat.com>
1983         PR plugins/59335
1984         * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
1985         added in 4.9.
1987 2014-04-15  Eric Botcazou  <ebotcazou@adacore.com>
1989         * cfgloop.h (struct loop): Move force_vectorize down.
1990         * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
1991         (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
1992         * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
1993         * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
1994         * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
1995         * tree-core.h (enum annot_expr_kind): Add new kind values.
1996         * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
1997         * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
1998         kinds.
1999         * tree.def (ANNOTATE_EXPR): Tweak comment.
2001 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
2003         * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
2004         cxa_pure_virtual).
2006 2014-04-14  Paolo Carlini  <paolo.carlini@oracle.com>
2008         * tree.h (TYPE_IDENTIFIER): Declare.
2009         * tree.c (subrange_type_for_debug_p): Use it.
2010         * godump.c (go_format_type): Likewise.
2011         * dwarf2out.c (is_cxx_auto, modified_type_die,
2012         gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
2013         * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
2015 2014-04-14  Jan Hubicka   <hubicka@ucw.cz>
2017         PR lto/60820
2018         * varpool.c (varpool_remove_node): Do not alter decls when streaming.
2020 2014-04-14  Uros Bizjak  <ubizjak@gmail.com>
2022         * config/i386/i386.c (examine_argument): Return bool.  Return true if
2023         parameter should be passed in memory.
2024         <case X86_64_COMPLEX_X87_CLASS>: Adjust.
2025         (construct_container): Update calls to examine_argument.
2026         (function_arg_advance_64): Ditto.
2027         (return_in_memory_32): Merge with ix86_return_in_memory.
2028         (return_in_memory_64): Ditto.
2029         (return_in_memory_ms_64): Ditto.
2031 2014-04-14  Jan Hubicka  <hubicka@ucw.cz>
2033         * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
2034         * coverage.c (coverage_compute_profile_id): Handle externally visible
2035         symbols.
2037 2014-04-14  Martin Jambor  <mjambor@suse.cz>
2039         * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
2040         DECL_DISREGARD_INLINE_LIMITS functions.
2042 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
2044         PR target/60827
2045         * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
2047 2014-04-14  H.J. Lu  <hongjiu.lu@intel.com>
2049         PR target/60827
2050         * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
2051         optimize_insn_for_speed_p instead of
2052         optimize_function_for_speed_p.
2054 2014-04-14  Yufeng Zhang  <yufeng.zhang@arm.com>
2056         * doc/invoke.texi (free): Document AArch64.
2058 2014-04-14  Richard Biener  <rguenther@suse.de>
2060         PR tree-optimization/60042
2061         * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
2062         (insert_into_preds_of_block): Do not prevent PHI insertion
2063         for REFERENCE exprs here ...
2064         (eliminate_dom_walker::before_dom_children): ... but prevent
2065         their use here under similar conditions when applied to the
2066         IL after PRE optimizations.
2068 2014-04-14  Richard Biener  <rguenther@suse.de>
2070         * passes.def: Move early points-to after early SRA.
2072 2014-04-14  Richard Biener  <rguenther@suse.de>
2074         * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
2075         check for which sign-changes we allow when forwarding
2076         a converted value into a switch.
2078 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
2080         * stor-layout.c (place_field): Finalize non-constant offset for the
2081         field, if any.
2083 2014-04-14  Richard Biener  <rguenther@suse.de>
2085         * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
2086         as argument.
2087         (expand_switch_using_bit_tests_p): Likewise.
2088         (process_switch): Compute and pass on speed_p based on the
2089         switch stmt.
2090         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
2091         optimize_bb_for_speed_p.
2093 2014-04-14  Eric Botcazou  <ebotcazou@adacore.com>
2095         * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
2096         * function.h (struct function): Rename has_force_vect_loops into
2097         has_force_vectorize_loops.
2098         * lto-streamer-in.c (input_cfg): Adjust for renaming.
2099         (input_struct_function_base): Likewise.
2100         * lto-streamer-out.c (output_cfg): Likewise.
2101         (output_struct_function_base): Likewise.
2102         * omp-low.c (expand_omp_simd): Likewise.
2103         * tree-cfg.c (move_sese_region_to_fn): Likewise.
2104         * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2105         (version_loop_for_if_conversion): Likewise.
2106         (tree_if_conversion): Likewise.
2107         (main_tree_if_conversion): Likewise.
2108         (gate_tree_if_conversion): Likewise.
2109         * tree-inline.c (copy_loops): Likewise.
2110         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
2111         * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
2112         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
2113         * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
2114         * tree-vectorizer.c (vectorize_loops): Likewise.
2115         * tree-vectorizer.h (unlimited_cost_model): Likewise.
2117 2014-04-14  Richard Biener  <rguenther@suse.de>
2119         PR lto/60720
2120         * lto-streamer-out.c (wrap_refs): New function.
2121         (lto_output): Wrap symbol references in global initializes in
2122         type-preserving MEM_REFs.
2124 2014-04-14  Christian Bruel  <christian.bruel@st.com>
2126         * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
2128 2014-04-14  Christian Bruel  <christian.bruel@st.com>
2130         * config/sh/sh.md (setmemqi): New expand pattern.
2131         * config/sh/sh.h (CLEAR_RATIO): Define.
2132         * config/sh/sh-mem.cc (sh_expand_setmem): Define.
2133         * config/sh/sh-protos.h (sh_expand_setmem): Declare.
2135 2014-04-14  Richard Biener  <rguenther@suse.de>
2137         PR middle-end/55022
2138         * fold-const.c (negate_expr_p): Don't negate directional rounding
2139         division.
2140         (fold_negate_expr): Likewise.
2142 2014-04-14  Richard Biener  <rguenther@suse.de>
2144         PR tree-optimization/59817
2145         PR tree-optimization/60453
2146         * graphite-scop-detection.c (graphite_can_represent_scev): Complete
2147         recursion to catch all CHRECs in the scalar evolution and restrict
2148         the predicate for the remains appropriately.
2150 2014-04-12  Catherine Moore  <clm@codesourcery.com>
2152         * config/mips/constraints.md: Add new register constraint "kb".
2153         * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
2154         (*movhi_internal): Likewise.
2155         (*movqi_internal): Likewise.
2156         * config/mips/mips.h (M16_STORE_REGS): New register class.
2157         (REG_CLASS_NAMES): Add M16_STORE_REGS.
2158         (REG_CLASS_CONTENTS): Likewise.
2159         * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
2161 2014-04-11  Tobias Burnus  <burnus@net-b.de>
2163         PR c/60194
2164         * doc/invoke.texi (-Wformat-signedness): Document it.
2165         (Wformat=2): Mention that this enables -Wformat-signedness.
2167 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2169         * common/config/epiphany/epiphany-common.c
2170         (epiphany_option_optimization_table): Enable section anchors by
2171         default at -O1 or higher.
2172         * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
2173         (TARGET_MIN_ANCHOR_OFFSET): Likewise.
2174         (epiphany_rtx_costs) <SET>: For binary operators, the set as such
2175         carries no extra cost.
2176         (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
2177         * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
2178         * config/epiphany/predicates.md (memclob_operand): New predicate.
2179         * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
2180         Use memclob_operand predicate and X constraint for operand 3.
2182 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2184         * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
2185         with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
2186         its operands.
2188 2014-04-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2190         PR rtl-optimization/60651
2191         * mode-switching.c (optimize_mode_switching): Make sure to emit
2192         sets of a lower numbered entity before sets of a higher numbered
2193         entity to a mode of the same or lower priority.
2194         When creating a seginfo for a basic block that starts with a code
2195         label, move the insertion point past the code label.
2196         (new_seginfo): Document and enforce requirement that
2197         NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
2198         * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
2199         * doc/tm.texi: Regenerate.
2201 2014-01-11  Joern Rennecke  <joern.rennecke@embecosm.com>
2203         PR target/60811
2204         * config/arc/arc.c (arc_save_restore): Fix assert typo.
2206 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
2208         * BASE-VER: Set to 4.10.0.
2210 2014-04-11  Tobias Burnus  <burnus@net-b.de>
2212         PR other/59055
2213         * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
2214         * doc/gcc.texi (Service): Update description in the @menu
2215         * doc/invoke.texi (Option Summary): Remove misplaced and
2216         duplicated @menu.
2218 2014-04-11  Steve Ellcey  <sellcey@mips.com>
2219             Jakub Jelinek  <jakub@redhat.com>
2221         PR middle-end/60556
2222         * expr.c (convert_move): Use emit_store_flag_force instead of
2223         emit_store_flag.  Pass lowpart_mode instead of VOIDmode as 5th
2224         argument to it.
2226 2014-04-11  Richard Biener  <rguenther@suse.de>
2228         PR middle-end/60797
2229         * varasm.c (assemble_alias): Avoid endless error reporting
2230         recursion by setting TREE_ASM_WRITTEN.
2232 2014-04-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2234         * config/s390/s390.md: Add a splitter for NOT rtx.
2236 2014-04-11  Jakub Jelinek  <jakub@redhat.com>
2238         PR rtl-optimization/60663
2239         * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
2241 2014-04-10  Jan Hubicka  <hubicka@ucw.cz>
2242             Jakub Jelinek  <jakub@redhat.com>
2244         PR lto/60567
2245         * ipa.c (function_and_variable_visibility): Copy forced_by_abi
2246         flag from decl_node to node.
2248 2014-04-10  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2250         PR debug/60655
2251         * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
2252         (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
2253         ameliorating the cases where it can be.
2255 2014-04-09  David Edelsohn  <dje.gcc@gmail.com>
2257         Revert
2258         2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
2260         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2261         (loadsync_<mode>): Change mode.
2262         (load_quadpti, store_quadpti): New.
2263         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2264         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2265         * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
2267 2014-04-09  Cong Hou  <congh@google.com>
2269         PR testsuite/60773
2270         * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
2271         documentation.
2273 2014-04-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2275         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
2276         instead of vnor to exploit possible fusion opportunity in the
2277         future.
2278         (altivec_expand_vec_perm_const_le): Likewise.
2280 2014-04-08  Pat Haugen  <pthaugen@us.ibm.com>
2282         * config/rs6000/sync.md (AINT mode_iterator): Move definition.
2283         (loadsync_<mode>): Change mode.
2284         (load_quadpti, store_quadpti): New.
2285         (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
2286         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
2288 2014-04-08  Richard Sandiford  <rdsandiford@googlemail.com>
2290         PR target/60763
2291         * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
2292         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
2293         Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
2295 2014-04-08  Richard Biener  <rguenther@suse.de>
2297         PR middle-end/60706
2298         * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
2299         a 64bit widest int print double-int similar to on HWI64 hosts.
2301 2014-04-08  Richard Biener  <rguenther@suse.de>
2303         PR tree-optimization/60785
2304         * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
2305         default defs properly.
2307 2014-04-08  Nathan Sidwell  <nathan@codesourcery.com>
2309         * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
2310         (Weffc++): Likewise.
2312 2014-04-07  Jan Hubicka  <hubcika@ucw.cz>
2314         * ipa-devirt.c (maybe_record_node): When node is not recorded,
2315         set completep to false rather than true.
2317 2014-04-07  Douglas B Rupp  <rupp@adacore.com>
2319         PR target/60504
2320         * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
2321         ARM_TARGET2_DWARF_FORMAT.
2323 2014-04-07  Charles Baylis  <charles.baylis@linaro.org>
2325         PR target/60609
2326         * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
2327         (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
2328         ADDR_DIFF_VEC.
2330 2014-04-07  Richard Biener  <rguenther@suse.de>
2332         PR tree-optimization/60766
2333         * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
2334         (may_eliminate_iv): Convert cand_value_at result to desired type.
2336 2014-04-07  Jason Merrill  <jason@redhat.com>
2338         PR c++/60731
2339         * common.opt (-fno-gnu-unique): Add.
2340         * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
2342 2014-04-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2344         * haifa-sched.c: Fix outdated function reference and minor
2345         grammar errors in introductory comment.
2347 2014-04-07  Richard Biener  <rguenther@suse.de>
2349         PR middle-end/60750
2350         * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
2351         for noreturn calls.
2352         * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
2354 2014-04-06  John David Anglin  <danglin@gcc.gnu.org>
2356         PR debug/55794
2357         * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
2358         size accounting for thunks.
2359         (pa_asm_output_mi_thunk): Use final_start_function() and
2360         final_end_function() to output function start and end directives.
2362 2014-04-05  Pitchumani Sivanupandi  <Pitchumani.S@atmel.com>
2364         * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
2365         device specific ISA/ feature information. Remove short_sp and
2366         errata_skip ds.  Add avr_device_specific_features enum to have device
2367         specific info.
2368         * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
2369         errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
2370         * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
2371         updated device specific info.
2372         * config/avr/avr-mcus.def: Merge device specific details to
2373         dev_attribute field.
2374         * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
2375         errata_skip.
2376         * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
2377         * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
2378         assembler if RMW isa supported by current device.
2379         * config/avr/genmultilib.awk: Update as device info structure changed.
2380         * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
2382 2014-04-04  Cong Hou  <congh@google.com>
2384         PR tree-optimization/60656
2385         * tree-vect-stmts.c (supportable_widening_operation):
2386         Fix a bug that elements in a vector with vect_used_by_reduction
2387         property are incorrectly reordered when the operation on it is not
2388         consistant with the one in reduction operation.
2390 2014-04-04  John David Anglin  <danglin@gcc.gnu.org>
2392         PR rtl-optimization/60155
2393         * gcse.c (record_set_data): New function.
2394         (single_set_gcse): New function.
2395         (gcse_emit_move_after): Use single_set_gcse instead of single_set.
2396         (hoist_code): Likewise.
2397         (get_pressure_class_and_nregs): Likewise.
2399 2014-04-04  Eric Botcazou  <ebotcazou@adacore.com>
2401         * explow.c (probe_stack_range): Emit a final optimization blockage.
2403 2014-04-04  Anthony Green  <green@moxielogic.com>
2405         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
2406         typos.
2408 2014-04-04  Jan Hubicka  <hubicka@ucw.cz>
2410         PR ipa/59626
2411         * lto-cgraph.c (input_overwrite_node): Check that partitioning
2412         flags are set only during streaming.
2413         * ipa.c (process_references, walk_polymorphic_call_targets,
2414         symtab_remove_unreachable_nodes): Drop bodies of always inline
2415         after early inlining.
2416         (symtab_remove_unreachable_nodes): Remove always_inline attribute.
2418 2014-04-04  Jakub Jelinek  <jakub@redhat.com>
2419         Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2421         PR debug/60655
2422         * dwarf2out.c (const_ok_for_output_1): Reject expressions
2423         containing a NOT.
2425 2014-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2427         PR bootstrap/60743
2428         * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
2429         duration.
2430         (cortex_a53_fdivd): Likewise.
2432 2014-04-04  Martin Jambor  <mjambor@suse.cz>
2434         PR ipa/60640
2435         * cgraph.h (cgraph_clone_node): New parameter added to declaration.
2436         Adjust all callers.
2437         * cgraph.c (clone_of_p): Also return true if thunks match.
2438         (verify_edge_corresponds_to_fndecl): Removed extraneous call to
2439         cgraph_function_or_thunk_node and an obsolete comment.
2440         * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
2441         file.
2442         (build_function_decl_skip_args): Likewise.
2443         (set_new_clone_decl_and_node_flags): New function.
2444         (duplicate_thunk_for_node): Likewise.
2445         (redirect_edge_duplicating_thunks): Likewise.
2446         (cgraph_clone_node): New parameter args_to_skip, pass it to
2447         redirect_edge_duplicating_thunks which is called instead of
2448         cgraph_redirect_edge_callee.
2449         (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
2450         moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
2452 2014-04-04  Jeff Law  <law@redhat.com>
2454         PR target/60657
2455         * config/arm/predicates.md (const_int_I_operand): New predicate.
2456         (const_int_M_operand): Similarly.
2457         * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
2458         const_int_operand.
2459         (insv_t2, extv_reg, extzv_t2): Likewise.
2460         (load_multiple_with_writeback): Similarly for const_int_I_operand.
2461         (pop_multiple_with_writeback_and_return): Likewise.
2462         (vfp_pop_multiple_with_writeback): Likewise
2464 2014-04-04  Richard Biener  <rguenther@suse.de>
2466         PR ipa/60746
2467         * tree-ssanames.c (make_ssa_name_fn): Fix assert.
2468         * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
2469         non-GIMPLE_LABELs.
2470         * gimplify.h (gimple_add_tmp_var_fn): Declare.
2471         * gimplify.c (gimple_add_tmp_var_fn): New function.
2472         * gimple-expr.h (create_tmp_reg_fn): Declare.
2473         * gimple-expr.c (create_tmp_reg_fn): New function.
2474         * gimple-low.c (record_vars_into): Don't change cfun.
2475         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
2476         code generation without cfun.
2478 2014-04-04  Thomas Schwinge  <thomas@codesourcery.com>
2480         PR bootstrap/60719
2481         * Makefile.in (install-driver): Fix shell scripting.
2483 2014-04-03  Cong Hou  <congh@google.com>
2485         PR tree-optimization/60505
2486         * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
2487         threshold of number of iterations below which no vectorization
2488         will be done.
2489         * tree-vect-loop.c (new_loop_vec_info):
2490         Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
2491         * tree-vect-loop.c (vect_analyze_loop_operations):
2492         Set LOOP_VINFO_COST_MODEL_THRESHOLD.
2493         * tree-vect-loop.c (vect_transform_loop):
2494         Use LOOP_VINFO_COST_MODEL_THRESHOLD.
2495         * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
2496         of iterations of the loop and see if we should build the epilogue.
2498 2014-04-03  Richard Biener  <rguenther@suse.de>
2500         * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
2501         (streamer_tree_cache_create): Adjust.
2502         * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
2503         to allow optional nodes array.
2504         (streamer_tree_cache_insert_1): Use next_idx to assign idx.
2505         (streamer_tree_cache_append): Likewise.
2506         (streamer_tree_cache_create): Create nodes array optionally
2507         as specified by parameter.
2508         * lto-streamer-out.c (create_output_block): Avoid maintaining
2509         the node array in the writer cache.
2510         (DFS_write_tree): Remove assertion.
2511         (produce_asm_for_decls): Free the out decl state hash table early.
2512         * lto-streamer-in.c (lto_data_in_create): Adjust for
2513         streamer_tree_cache_create prototype change.
2515 2014-04-03  Richard Biener  <rguenther@suse.de>
2517         * tree-streamer-out.c (streamer_write_chain): Do not temporarily
2518         set TREE_CHAIN to NULL_TREE.
2520 2014-04-03  Richard Biener  <rguenther@suse.de>
2522         PR tree-optimization/60740
2523         * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
2524         over all GIMPLE_COND operands.
2526 2014-04-03  Nathan Sidwell  <nathan@codesourcery.com>
2528         * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
2529         (Weffc++): Remove Scott's numbering, merge lists and reference
2530         Wnon-virtual-dtor.
2532 2014-04-03  Nick Clifton  <nickc@redhat.com>
2534         * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
2535         properly.
2537 2014-04-03  Martin Jambor  <mjambor@suse.cz>
2539         * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
2540         mention gcc_unreachable before failing.
2541         * ipa.c (symtab_remove_unreachable_nodes): Also print order of
2542         removed symbols.
2544 2014-04-02  Jan Hubicka  <hubicka@ucw.cz>
2546         PR ipa/60659
2547         * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
2548         inconsistent code and instead mark the context inconsistent.
2549         (possible_polymorphic_call_targets): For inconsistent contexts
2550         return empty complete list.
2552 2014-04-02  Anthony Green  <green@moxielogic.com>
2554         * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
2555         (extendqisi2, extendhisi2): Define.
2556         * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
2557         (WCHAR_TYPE): Change to unsigned int.
2559 2014-04-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2561         PR tree-optimization/60733
2562         * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
2563         insertion point for PHI candidates to be the end of the feeding
2564         block for the PHI argument.
2566 2014-04-02  Vladimir Makarov  <vmakarov@redhat.com>
2568         PR rtl-optimization/60650
2569         * lra-constraints.c (process_alt_operands): Decrease reject for
2570         earlyclobber matching.
2572 2014-04-02  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2574         * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
2576 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2578         * config/spu/spu.c (pad_bb): Do not crash when the last
2579         insn is CODE_FOR_blockage.
2581 2014-04-02  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2583         * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
2584         lies outside the target mode.
2586 2014-04-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
2588         PR target/60735
2589         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
2590         software floating point or no floating point registers, do not
2591         allow any type in the FPRs.  Eliminate a test for SPE SIMD types
2592         in GPRs that occurs after we tested for GPRs that would never be
2593         true.
2595         * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
2596         Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
2597         since the FMOVE64 type is DFmode/DDmode.  If TARGET_E500_DOUBLE,
2598         specifically allow DDmode, since that does not use the SPE SIMD
2599         instructions.
2601 2014-04-02  Richard Biener  <rguenther@suse.de>
2603         PR middle-end/60729
2604         * optabs.c (expand_abs_nojump): Honor flag_trapv only for
2605         MODE_INTs.  Properly use negv_optab.
2606         (expand_abs): Likewise.
2608 2014-04-02  Richard Biener  <rguenther@suse.de>
2610         PR bootstrap/60719
2611         * Makefile.in (install-driver): Guard extra installs with special
2612         names properly.
2614 2014-04-01  Michael Meissner  <meissner@linux.vnet.ibm.com>
2616         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2617         Document vec_vgbbd.
2619 2014-04-01  Richard Henderson  <rth@redhat.com>
2621         PR target/60704
2622         * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
2623         alternative enabled before register allocation.
2625 2014-04-01  Chung-Lin Tang  <cltang@codesourcery.com>
2627         * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
2628         * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
2629         typo.
2630         (nios2_large_got_address): Remove unneeded 'sym' parameter.
2631         (nios2_got_address): Update nios2_large_got_address call site.
2632         (nios2_delegitimize_address): New function.
2633         (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
2634         * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
2635         (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
2637 2014-04-01  Martin Husemann  <martin@duskware.de>
2639         * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
2640         for -mabi=32.
2642 2014-04-01  Richard Sandiford  <rdsandiford@googlemail.com>
2644         PR rtl-optimization/60604
2645         * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
2646         check from register_operand.
2647         (register_operand): Redefine in terms of general_operand.
2648         (nonmemory_operand): Use register_operand for the non-constant cases.
2650 2014-04-01  Richard Biener  <rguenther@suse.de>
2652         * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
2654 2014-04-01  Sebastian Huber  <sebastian.huber@embedded-brains.de>
2656         * doc/invoke.texi (mapp-regs): Clarify.
2658 2014-03-31  Ulrich Drepper  <drepper@gmail.com>
2660         * config/i386/avx512fintrin.h (__v32hi): Define type.
2661         (__v64qi): Likewise.
2662         (_mm512_set1_epi8): Define.
2663         (_mm512_set1_epi16): Define.
2664         (_mm512_set4_epi32): Define.
2665         (_mm512_set4_epi64): Define.
2666         (_mm512_set4_pd): Define.
2667         (_mm512_set4_ps): Define.
2668         (_mm512_setr4_epi64): Define.
2669         (_mm512_setr4_epi32): Define.
2670         (_mm512_setr4_pd): Define.
2671         (_mm512_setr4_ps): Define.
2672         (_mm512_setzero_epi32): Define.
2674 2014-03-31  Martin Jambor  <mjambor@suse.cz>
2676         PR middle-end/60647
2677         * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
2678         callsite_arguments_match_p.  Updated all callers.  Also check types of
2679         corresponding formal parameters and actual arguments.
2680         (not_all_callers_have_enough_arguments_p) Renamed to
2681         some_callers_have_mismatched_arguments_p.
2683 2014-03-31  Yuri Rumyantsev  <ysrumyan@gmail.com>
2685         * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
2687 2014-03-31  Kugan Vivekanandarajah  <kuganv@linaro.org>
2689         PR target/60034
2690         * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
2691         section anchor.
2693 2014-03-30  Uros Bizjak  <ubizjak@gmail.com>
2695         * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
2696         (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
2697         Split out
2698         <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
2699         Use FMAMODE_NOVF512 mode iterator.
2700         (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
2701         (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
2702         (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
2703         (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
2704         Split out
2705         <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
2706         Use VF_128_256 mode iterator.
2707         (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
2708         Ditto.
2710 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2712         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
2713         static chain if needed.
2715 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
2717         PR target/60697
2718         * lra-constraints.c (index_part_to_reg): New.
2719         (process_address): Use it.
2721 2014-03-27  Jeff Law  <law@redhat.com>
2722             Jakub Jelinek  <jakub@redhat.com>
2724         PR target/60648
2725         * expr.c (do_tablejump): Use simplify_gen_binary rather than
2726         gen_rtx_{PLUS,MULT} to build up the address expression.
2728         * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
2729         creating non-canonical RTL.
2731 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2733         PR ipa/60243
2734         * ipa-inline.c (want_inline_small_function_p): Short circuit large
2735         functions; reorganize to make cheap checks first.
2736         (inline_small_functions): Do not estimate growth when dumping;
2737         it is expensive.
2738         * ipa-inline.h (inline_summary): Add min_size.
2739         (growth_likely_positive): New function.
2740         * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
2741         (set_cond_stmt_execution_predicate): Cleanup.
2742         (estimate_edge_size_and_time): Compute min_size.
2743         (estimate_calls_size_and_time): Likewise.
2744         (estimate_node_size_and_time): Likewise.
2745         (inline_update_overall_summary): Update min_size.
2746         (do_estimate_edge_time): Likewise.
2747         (do_estimate_edge_size): Update.
2748         (do_estimate_edge_hints): Update.
2749         (growth_likely_positive): New function.
2751 2014-03-28  Jakub Jelinek  <jakub@redhat.com>
2753         PR target/60693
2754         * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
2755         also if addr has VOIDmode.
2757 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2759         * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
2760         * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
2761         Declare extern.
2762         * config/arm/cortex-a53.md: Add reservations and bypass for crypto
2763         instructions as well as AdvancedSIMD loads.
2765 2014-03-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2767         * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
2768         Use crypto_aese type.
2769         (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
2770         * config/arm/arm.md (is_neon_type): Replace crypto_aes with
2771         crypto_aese, crypto_aesmc.  Move to types.md.
2772         * config/arm/types.md (crypto_aes): Split into crypto_aese,
2773         crypto_aesmc.
2774         * config/arm/iterators.md (crypto_type): Likewise.
2776 2014-03-28  Jan Hubicka  <hubicka@ucw.cz>
2778         * cgraph.c: Include expr.h and tree-dfa.h.
2779         (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
2780         remove LHS.
2782 2014-03-28  Vladimir Makarov  <vmakarov@redhat.com>
2784         PR target/60675
2785         * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
2786         regs from checking multi-reg pseudos.
2788 2014-03-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
2790         * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
2792 2014-03-28  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
2794         * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
2795         if it would clobber the stack pointer, even temporarily.
2797 2014-03-28  Eric Botcazou  <ebotcazou@adacore.com>
2799         * mode-switching.c: Make small adjustments to the top comment.
2801 2014-03-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
2803         * config/rs6000/constraints.md (wD constraint): New constraint to
2804         match the constant integer to get the top DImode/DFmode out of a
2805         vector in a VSX register.
2807         * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
2808         match the constant integer to get the top DImode/DFmode out of a
2809         vector in a VSX register.
2811         * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
2812         for ISA 2.07.
2814         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2815         vbpermq builtins.
2817         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
2818         -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
2820         * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
2821         Optimize vec_extract of 64-bit values, where the value being
2822         extracted is in the top word, where we can use scalar
2823         instructions.  Add direct move and store support.  Combine the big
2824         endian/little endian vector select load support into a single insn.
2825         (vsx_extract_<mode>_internal1): Likewise.
2826         (vsx_extract_<mode>_internal2): Likewise.
2827         (vsx_extract_<mode>_load): Likewise.
2828         (vsx_extract_<mode>_store): Likewise.
2829         (vsx_extract_<mode>_zero): Delete, big and little endian insns are
2830         combined into vsx_extract_<mode>_load.
2831         (vsx_extract_<mode>_one_le): Likewise.
2833         * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
2834         define the top 64-bit vector element.
2836         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
2837         constraint.
2839         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2840         Document vec_vbpermq builtin.
2842         PR target/60672
2843         * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
2844         enable use of xxsldwi and xxpermdi builtin functions.
2845         (vec_xxpermdi): Likewise.
2847         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
2848         Document use of vec_xxsldwi and vec_xxpermdi builtins.
2850 2014-03-27  Vladimir Makarov  <vmakarov@redhat.com>
2852         PR rtl-optimization/60650
2853         * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
2854         first_p.  Use it.
2855         (find_spills_for): New.
2856         (assign_by_spills): Pass the new parameter to find_hard_regno_for.
2857         Spill all pseudos on the second iteration.
2859 2014-03-27  Marek Polacek  <polacek@redhat.com>
2861         PR c/50347
2862         * doc/extend.texi (ffs Builtins): Change unsigned types to signed
2863         types.
2865 2014-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2867         * config/s390/s390.c (s390_can_use_return_insn): Check for
2868         call-saved FPRs on 31 bit.
2870 2014-03-27  Jakub Jelinek  <jakub@redhat.com>
2872         PR middle-end/60682
2873         * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
2874         if they need regimplification, just drop them instead of
2875         calling gimple_regimplify_operands on them.
2877 2014-03-27  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2879         PR target/60580
2880         * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
2881         (aarch64_frame_pointer_required): Adjust logic.
2882         (aarch64_can_eliminate): Adjust logic.
2883         (aarch64_override_options_after_change): Adjust logic.
2885 2014-03-27  Dehao Chen  <dehao@google.com>
2887         * ipa-inline.c (early_inliner): Update node's inline info.
2889 2014-03-26  Dehao Chen  <dehao@google.com>
2891         * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
2892         compiler inserted conditional jumps for NAN float check.
2894 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
2896         * ubsan.h (ubsan_create_data): Change second argument's type
2897         to const location_t *.
2898         * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
2899         _("<unknown>").
2900         (ubsan_create_data): Change second argument to const location_t *PLOC.
2901         Create Loc field whenever PLOC is non-NULL.
2902         (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
2903         ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
2904         callers.
2906         PR other/59545
2907         * real.c (real_to_integer2): Change type of low to UHWI.
2909 2014-03-26  Tobias Burnus  <burnus@net-b.de>
2911         * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
2912         (CILK_SELF_SPECS): New define.
2913         (driver_self_specs): Use it.
2915 2014-03-26  Richard Biener  <rguenther@suse.de>
2917         * tree-pretty-print.c (percent_K_format): Implement special
2918         case for LTO and its stripped down BLOCK tree.
2920 2014-03-26  Jakub Jelinek  <jakub@redhat.com>
2922         PR sanitizer/60636
2923         * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
2925         * tree-vrp.c (simplify_internal_call_using_ranges): If only
2926         one range is range_int_cst_p, but not both, at least optimize
2927         addition/subtraction of 0 and multiplication by 0 or 1.
2928         * gimple-fold.c (gimple_fold_call): Fold
2929         IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
2930         (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
2931         INTEGER_CSTs, try to fold at least x * 0 and y - y.
2933 2014-03-26  Eric Botcazou  <ebotcazou@adacore.com>
2935         PR rtl-optimization/60452
2936         * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
2937         <case REG>: Return 1 for invalid offsets from the frame pointer.
2939 2014-03-26  Marek Polacek  <polacek@redhat.com>
2941         PR c/37428
2942         * doc/extend.texi (C Extensions): Mention variable-length arrays in
2943         a structure/union.
2945 2014-03-26  Marek Polacek  <polacek@redhat.com>
2947         PR c/39525
2948         * doc/extend.texi (Designated Inits): Describe what happens to omitted
2949         field members.
2951 2014-03-26  Marek Polacek  <polacek@redhat.com>
2953         PR other/59545
2954         * ira-color.c (update_conflict_hard_regno_costs): Perform the
2955         multiplication in unsigned type.
2957 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
2959         * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
2961 2014-03-26  Chung-Ju Wu  <jasonwucj@gmail.com>
2963         * doc/contrib.texi: Add myself as Andes nds32 port contributor.
2965 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
2967         PR ipa/60315
2968         * cif-code.def (UNREACHABLE) New code.
2969         * ipa-inline.c (inline_small_functions): Skip edges to
2970         __builtlin_unreachable.
2971         (estimate_edge_growth): Allow edges to __builtlin_unreachable.
2972         * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
2973         predicate to __bulitin_unreachable.
2974         (set_cond_stmt_execution_predicate): Fix issue when
2975         invert_tree_comparison returns ERROR_MARK.
2976         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
2977         propagate to inline clones.
2978         * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
2979         to unreachable.
2980         * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
2981         * cgraphclones.c (cgraph_clone_node): If call destination is already
2982         ureachable, do not redirect it back.
2983         * tree-inline.c (fold_marked_statements): Hanlde calls becoming
2984         unreachable.
2986 2014-03-25  Jan Hubicka  <hubicka@ucw.cz>
2988         * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
2989         Do not modify inline clones.
2991 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
2993         * config/i386/i386.md (general_sext_operand): New mode attr.
2994         (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
2995         don't generate (sign_extend (const_int)).
2996         (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
2997         operands[2].  Use We constraint instead of <i> and
2998         <general_sext_operand> predicate instead of <general_operand>.
2999         (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
3000         * config/i386/constraints.md (We): New constraint.
3001         * config/i386/predicates.md (x86_64_sext_operand,
3002         sext_operand): New predicates.
3004 2014-03-25  Martin Jambor  <mjambor@suse.cz>
3006         PR ipa/60600
3007         * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
3008         inconsistent devirtualizations to __builtin_unreachable.
3010 2014-03-25  Marek Polacek  <polacek@redhat.com>
3012         PR c/35449
3013         * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
3015 2014-03-25  Alan Lawrence  <alan.lawrence@arm.com>
3017         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
3018         order of elements for big-endian.
3020 2014-03-25  Richard Biener  <rguenther@suse.de>
3022         PR middle-end/60635
3023         * gimplify-me.c (gimple_regimplify_operands): Update the
3024         re-gimplifed stmt.
3026 2014-03-25  Martin Jambor  <mjambor@suse.cz>
3028         PR ipa/59176
3029         * lto-cgraph.c (lto_output_node): Stream body_removed flag.
3030         (lto_output_varpool_node): Likewise.
3031         (input_overwrite_node): Likewise.
3032         (input_varpool_node): Likewise.
3034 2014-03-25  Richard Biener  <rguenther@suse.de>
3036         * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
3037         (run_gcc): Likewise.
3039 2014-03-25  Jakub Jelinek  <jakub@redhat.com>
3041         * combine.c (simplify_compare_const): Add MODE argument.
3042         Handle mode_width 0 as very large mode_width.
3043         (try_combine, simplify_comparison): Adjust callers.
3045         * cselib.c (cselib_hash_rtx): Perform addition in unsigned
3046         type to avoid signed integer overflow.
3047         * explow.c (plus_constant): Likewise.
3049 2014-03-25  Dominik Vogt  <vogt@linux.vnet.ibm.com>
3051         * doc/generic.texi: Correct typos.
3053 2014-03-24  Tobias Burnus  <burnus@net-b.de>
3055         * doc/invoke.texi (-flto): Expand section about
3056         using static libraries with LTO.
3058 2014-03-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3060         PR rtl-optimization/60501
3061         * optabs.def (addptr3_optab): New optab.
3062         * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
3063         * doc/md.texi ("addptrm3"): Document new RTL standard expander.
3064         * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
3066         * lra.c (emit_add3_insn): Use the addptr pattern if available.
3068         * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
3070 2014-03-24  Ulrich Drepper  <drepper@gmail.com>
3072         * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
3073         _mm512_set1_pd.
3075         * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
3076         (_mm256_undefined_ps): Define.
3077         (_mm256_undefined_pd): Define.
3078         * config/i386/emmintrin.h (_mm_undefined_si128): Define.
3079         (_mm_undefined_pd): Define.
3080         * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
3081         * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
3082         (_mm512_undefined_ps): Define.
3083         (_mm512_undefined_pd): Define.
3084         Use _mm*_undefined_*.
3085         * config/i386/avx2intrin.h: Use _mm*_undefined_*.
3087 2014-03-24  Alex Velenko  <Alex.Velenko@arm.com>
3089         * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
3090         (lshr_simd): DI mode added.
3091         * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
3092         (aarch64_ushr_simddi): Likewise.
3093         * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
3094         * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
3095         (vshrd_n_u64): Likewise.
3097 2014-03-24  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3099         * Makefile.in (s-macro_list): Depend on cc1.
3101 2014-03-23  Teresa Johnson  <tejohnson@google.com>
3103         * ipa-utils.c (ipa_print_order): Use specified dump file.
3105 2014-03-23  Eric Botcazou  <ebotcazou@adacore.com>
3107         PR rtl-optimization/60601
3108         * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
3110         * gcc.c (eval_spec_function): Initialize save_growing_value.
3112 2014-03-22  Jakub Jelinek  <jakub@redhat.com>
3114         PR sanitizer/60613
3115         * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
3116         code == MINUS_EXPR, never swap op0 with op1.
3118         * toplev.c (init_local_tick): Avoid signed integer multiplication
3119         overflow.
3120         * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
3121         shift by first operand's bitsize.
3123 2014-03-21  Jakub Jelinek  <jakub@redhat.com>
3125         PR target/60610
3126         * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
3127         redefine to 1 or 0.
3128         * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
3129         TARGET_ISA_64BIT_P(x).
3131 2014-03-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3133         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
3134         pattern for vector nor instead of subtract from splat(-1).
3135         (altivec_expand_vec_perm_const_le): Likewise.
3137 2014-03-21  Richard Henderson  <rth@twiddle.net>
3139         PR target/60598
3140         * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
3141         related insns after epilogue_completed.
3143 2014-03-21  Martin Jambor  <mjambor@suse.cz>
3145         PR ipa/59176
3146         * cgraph.h (symtab_node): New flag body_removed.
3147         * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
3148         when removing bodies.
3149         * symtab.c (dump_symtab_base): Dump body_removed flag.
3150         * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
3151         had their bodies removed.
3153 2014-03-21  Martin Jambor  <mjambor@suse.cz>
3155         PR ipa/60419
3156         * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
3157         in the border.
3159 2014-03-21  Richard Biener  <rguenther@suse.de>
3161         PR tree-optimization/60577
3162         * tree-core.h (struct tree_base): Document nothrow_flag use
3163         in VAR_DECL_NONALIASED.
3164         * tree.h (VAR_DECL_NONALIASED): New.
3165         (may_be_aliased): Adjust.
3166         * coverage.c (build_var): Set VAR_DECL_NONALIASED.
3168 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3170         * expr.c (expand_expr_real_1): Remove outdated comment.
3172 2014-03-20  Jakub Jelinek  <jakub@redhat.com>
3174         PR middle-end/60597
3175         * ira.c (adjust_cleared_regs): Call copy_rtx on
3176         *reg_equiv[REGNO (loc)].src_p before passing it to
3177         simplify_replace_fn_rtx.
3179         PR target/60568
3180         * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
3181         into CONST, put pic register as first operand of PLUS.  Use
3182         gen_const_mem for both 32-bit and 64-bit PIC got loads.
3184 2014-03-20  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3186         * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
3188 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3190         * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
3191         around for store forwarding issue in the FPU on the UT699.
3192         * config/sparc/sparc.md (in_branch_delay): Return false for single FP
3193         loads and operations if -mfix-ut699 is specified.
3194         (divtf3_hq): Tweak attribute.
3195         (sqrttf2_hq): Likewise.
3197 2014-03-20  Eric Botcazou  <ebotcazou@adacore.com>
3199         * calls.c (store_one_arg): Remove incorrect const qualification on the
3200         type of the temporary.
3201         * cfgexpand.c (expand_return): Likewise.
3202         * expr.c (expand_constructor): Likewise.
3203         (expand_expr_real_1): Likewise.
3205 2014-03-20  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3207         * config/arm/arm.c (arm_dwarf_register_span): Update the element number
3208         of parts.
3210 2014-03-19  Kaz Kojima  <kkojima@gcc.gnu.org>
3212         PR target/60039
3213         * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
3215 2014-03-19  James Greenhalgh  <james.greenhalgh@arm.com>
3217         * config/arm/aarch-common-protos.h
3218         (alu_cost_table): Fix spelling of "extend".
3219         * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
3221 2014-03-19  Richard Biener  <rguenther@suse.de>
3223         PR middle-end/60553
3224         * tree-core.h (tree_type_common): Re-order pointer members
3225         to reduce recursion depth during GC walks.
3227 2014-03-19  Marek Polacek  <polacek@redhat.com>
3229         PR sanitizer/60569
3230         * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
3231         before accessing it.
3233 2014-03-19  Richard Biener  <rguenther@suse.de>
3235         PR lto/59543
3236         * lto-streamer-in.c (input_function): In WPA stage do not drop
3237         debug stmts.
3239 2014-03-19  Jakub Jelinek  <jakub@redhat.com>
3241         PR tree-optimization/60559
3242         * vectorizable_mask_load_store): Replace scalar MASK_LOAD
3243         with build_zero_cst assignment.
3245 2014-03-18  Kai Tietz  <ktietz@redhat.com>
3247         PR rtl-optimization/56356
3248         * sdbout.c (sdbout_parms): Verify that parms'
3249         incoming argument is valid.
3250         (sdbout_reg_parms): Likewise.
3252 2014-03-18  Richard Henderson  <rth@redhat.com>
3254         PR target/60562
3255         * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
3256         be shadowed by *float<SWI48><MODEF>2_sse.  Test X87_ENABLE_FLOAT.
3257         (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
3259 2014-03-18  Basile Starynkevitch  <basile@starynkevitch.net>
3261         * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
3262         * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
3263         Italicize plugin event names in description.  Explain that
3264         PLUGIN_PRAGMAS has no sense for lto1.  Explain PLUGIN_INCLUDE_FILE.
3265         Remind that no GCC functions should be called after PLUGIN_FINISH.
3266         Explain what pragmas with expansion are.
3268 2014-03-18  Martin Liska  <mliska@suse.cz>
3270         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
3271         gimple call statement is update.
3272         * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
3273         GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
3275 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
3277         PR sanitizer/60557
3278         * ubsan.c (ubsan_instrument_unreachable): Call
3279         initialize_sanitizer_builtins.
3280         (ubsan_pass): Likewise.
3282         PR sanitizer/60535
3283         * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
3284         varpool_finalize_decl instead of rest_of_decl_compilation.
3286 2014-03-18  Richard Biener  <rguenther@suse.de>
3288         * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
3289         by using bitmap_and_compl instead of bitmap_and_compl_into.
3290         (df_rd_transfer_function): Likewise.
3292 2014-03-18  Richard Biener  <rguenther@suse.de>
3294         * doc/lto.texi (fresolution): Fix typo.
3296 2014-03-18  Richard Biener  <rguenther@suse.de>
3298         * doc/invoke.texi (flto): Update for changes in 4.9.
3300 2014-03-18  Richard Biener  <rguenther@suse.de>
3302         * doc/loop.texi: Remove section on the removed lambda framework.
3303         Update loop docs with recent changes in preserving loop structure.
3305 2014-03-18  Richard Biener  <rguenther@suse.de>
3307         * doc/lto.texi (-fresolution): Document.
3309 2014-03-18  Richard Biener  <rguenther@suse.de>
3311         * doc/contrib.texi: Adjust my name.
3313 2014-03-18  Jakub Jelinek  <jakub@redhat.com>
3315         PR ipa/58721
3316         * internal-fn.c: Include diagnostic-core.h.
3317         (expand_BUILTIN_EXPECT): New function.
3318         * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
3319         (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
3320         IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
3321         * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
3322         IFN_BUILTIN_EXPECT.
3323         * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
3324         Revert 3 argument __builtin_expect code.
3325         (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
3326         * gimple-fold.c (gimple_fold_call): Likewise.
3327         * tree.h (fold_builtin_expect): New prototype.
3328         * builtins.c (build_builtin_expect_predicate): Add predictor
3329         argument, if non-NULL, create 3 argument __builtin_expect.
3330         (fold_builtin_expect): No longer static.  Add ARG2 argument,
3331         pass it through to build_builtin_expect_predicate.
3332         (fold_builtin_2): Adjust caller.
3333         (fold_builtin_3): Handle BUILT_IN_EXPECT.
3334         * internal-fn.def (BUILTIN_EXPECT): New.
3336 2014-03-18  Tobias Burnus  <burnus@net-b.de>
3338         PR ipa/58721
3339         * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
3340         PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
3341         PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
3343 2014-03-18  Jan Hubicka  <hubicka@ucw.cz>
3345         PR ipa/58721
3346         * predict.c (combine_predictions_for_bb): Fix up formatting.
3347         (expr_expected_value_1, expr_expected_value): Add predictor argument,
3348         fill what it points to if non-NULL.
3349         (tree_predict_by_opcode): Adjust caller, use the predictor.
3350         * predict.def (PRED_COMPARE_AND_SWAP): Add.
3352 2014-03-18  Eric Botcazou  <ebotcazou@adacore.com>
3354         * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
3355         proper constant for the store mode.
3357 2014-03-18  Ilya Enkovich  <ilya.enkovich@intel.com>
3359         * symtab.c (change_decl_assembler_name): Fix transparent alias
3360         chain construction.
3362 2014-03-16  Renlin Li  <Renlin.Li@arm.com>
3364         * config/aarch64/aarch64.c: Correct the comments about the
3365         aarch64 stack layout.
3367 2014-03-18  Thomas Schwinge  <thomas@codesourcery.com>
3369         * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
3370         check for GF_OMP_FOR_KIND_FOR.
3372 2013-03-18  Kirill Yukhin  <kirill.yukhin@intel.com>
3374         * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
3375         ymm and zmm register names.
3377 2014-03-17  Jakub Jelinek  <jakub@redhat.com>
3379         PR target/60516
3380         * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
3381         note creation for the 2010-08-31 changes.
3383 2014-03-17  Marek Polacek  <polacek@redhat.com>
3385         PR middle-end/60534
3386         * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
3387         as -fno-tree-loop-vectorize.
3388         (expand_omp_simd): Likewise.
3390 2014-03-15  Eric Botcazou  <ebotcazou@adacore.com>
3392         * config/sparc/sparc-protos.h (tls_call_delay): Delete.
3393         (eligible_for_call_delay): New prototype.
3394         * config/sparc/sparc.c (tls_call_delay): Rename into...
3395         (eligible_for_call_delay): ...this.  Return false if the instruction
3396         cannot be put in the delay slot of a branch.
3397         (eligible_for_restore_insn): Simplify.
3398         (eligible_for_return_delay): Return false if the instruction cannot be
3399         put in the delay slot of a branch and simplify.
3400         (eligible_for_sibcall_delay): Return false if the instruction cannot be
3401         put in the delay slot of a branch.
3402         * config/sparc/sparc.md (fix_ut699): New attribute.
3403         (tls_call_delay): Delete.
3404         (in_call_delay): Reimplement.
3405         (eligible_for_sibcall_delay): Rename into...
3406         (in_sibcall_delay): ...this.
3407         (eligible_for_return_delay): Rename into...
3408         (in_return_delay): ...this.
3409         (in_branch_delay): Reimplement.
3410         (in_uncond_branch_delay): Delete.
3411         (in_annul_branch_delay): Delete.
3413 2014-03-14  Richard Henderson  <rth@redhat.com>
3415         PR target/60525
3416         * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
3417         define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
3418         (*floathi<X87MODEF>2_i387_with_temp): Remove.
3419         (floathi splitters): Remove.
3420         (float<SWI48x>xf2): New pattern.
3421         (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2.  Drop
3422         code that tried to handle DImode for 32-bit, but which was excluded
3423         by the pattern's condition.  Drop allocation of stack temporary.
3424         (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
3425         (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
3426         (*float<SWI48><MODEF>2_mixed_interunit): Remove.
3427         (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
3428         (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
3429         (*float<SWI48><MODEF>2_sse_with_temp): Remove.
3430         (*float<SWI48><MODEF>2_sse_interunit): Remove.
3431         (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
3432         (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
3433         (*float<SWI48x><X87MODEF>2_i387): Remove.
3434         (all float _with_temp splitters): Remove.
3435         (*float<SWI48x><MODEF>2_i387): New pattern.
3436         (*float<SWI48><MODEF>2_sse): New pattern.
3437         (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
3438         (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
3440 2014-03-14  Jakub Jelinek  <jakub@redhat.com>
3441             Marek Polacek  <polacek@redhat.com>
3443         PR middle-end/60484
3444         * common.opt (dump_base_name_prefixed): New Variable.
3445         * opts.c (finish_options): Don't prepend directory to x_dump_base_name
3446         if x_dump_base_name_prefixed is already set, set it at the end.
3448 2014-03-14  Vladimir Makarov  <vmakarov@redhat.com>
3450         PR rtl-optimization/60508
3451         * lra-constraints.c (get_reload_reg): Add new parameter
3452         in_subreg_p.
3453         (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
3454         Pass the new parameter values.
3456 2014-03-14  Richard Biener  <rguenther@suse.de>
3458         * common.opt: Revert unintented changes from r205065.
3459         * opts.c: Likewise.
3461 2014-03-14  Richard Biener  <rguenther@suse.de>
3463         PR middle-end/60518
3464         * cfghooks.c (split_block): Properly adjust all loops the
3465         block was a latch of.
3467 2014-03-14  Martin Jambor  <mjambor@suse.cz>
3469         PR lto/60461
3470         * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
3471         and simplify it.
3473 2014-03-14  Georg-Johann Lay  <avr@gjlay.de>
3475         PR target/59396
3476         * config/avr/avr.c (avr_set_current_function): Pass function name
3477         through default_strip_name_encoding before sanity checking instead
3478         of skipping the first char of the assembler name.
3480 2014-03-13  Richard Henderson  <rth@redhat.com>
3482         PR debug/60438
3483         * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
3484         (ix86_force_to_memory, ix86_free_from_memory): Remove.
3485         * config/i386/i386-protos.h: Likewise.
3486         * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
3487         in the expander instead of a splitter.
3488         (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
3489         any possibility of requiring a memory.
3490         (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
3491         (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
3492         (fp branch splitters): Update for ix86_split_fp_branch.
3493         (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
3494         (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
3495         (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
3496         (*fop_<MODEF>_2_i387): Remove f/r alternative.
3497         (*fop_<MODEF>_3_i387): Likewise.
3498         (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
3499         (splitters for the fop_* register patterns): Remove.
3500         (fscalexf4_i387): Rename from *fscalexf4_i387.
3501         (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
3503 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
3505         PR tree-optimization/59779
3506         * tree-dfa.c (get_ref_base_and_extent): Use double_int
3507         type for bitsize and maxsize instead of HOST_WIDE_INT.
3509 2014-03-13  Steven Bosscher  <steven@gcc.gnu.org>
3511         PR rtl-optimization/57320
3512         * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
3513         the CFG after thread_prologue_and_epilogue_insns.
3515 2014-03-13  Vladimir Makarov  <vmakarov@redhat.com>
3517         PR rtl-optimization/57189
3518         * lra-constraints.c (process_alt_operands): Disfavor spilling
3519         vector pseudos.
3521 2014-03-13  Cesar Philippidis  <cesar@codesourcery.com>
3523         * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
3525 2014-03-13  Jakub Jelinek  <jakub@redhat.com>
3527         PR tree-optimization/59025
3528         PR middle-end/60418
3529         * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
3530         same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
3532 2014-03-13  Georg-Johann Lay  <avr@gjlay.de>
3534         PR target/60486
3535         * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
3536         calls of avr_out_plus_1.
3538 2014-03-13  Bin Cheng  <bin.cheng@arm.com>
3540         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
3541         BB's single pred and update the father loop's latch info later.
3543 2014-03-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
3545         * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
3546         (VEC_M): Likewise.
3547         (VEC_N): Likewise.
3548         (VEC_R): Likewise.
3549         (VEC_base): Likewise.
3550         (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
3551         registers, we need to swap double words in little endian mode.
3553         * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
3554         to be a container mode for 128-bit integer operations added in ISA
3555         2.07.  Unlike TImode and PTImode, the preferred register set is
3556         the Altivec/VMX registers for the 128-bit operations.
3558         * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
3559         declarations.
3560         (rs6000_split_128bit_ok_p): Likewise.
3562         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
3563         macros for creating ISA 2.07 normal and overloaded builtin
3564         functions with 3 arguments.
3565         (BU_P8V_OVERLOAD_3): Likewise.
3566         (VPERM_1T): Add support for V1TImode in 128-bit vector operations
3567         for use as overloaded functions.
3568         (VPERM_1TI_UNS): Likewise.
3569         (VSEL_1TI): Likewise.
3570         (VSEL_1TI_UNS): Likewise.
3571         (ST_INTERNAL_1ti): Likewise.
3572         (LD_INTERNAL_1ti): Likewise.
3573         (XXSEL_1TI): Likewise.
3574         (XXSEL_1TI_UNS): Likewise.
3575         (VPERM_1TI): Likewise.
3576         (VPERM_1TI_UNS): Likewise.
3577         (XXPERMDI_1TI): Likewise.
3578         (SET_1TI): Likewise.
3579         (LXVD2X_V1TI): Likewise.
3580         (STXVD2X_V1TI): Likewise.
3581         (VEC_INIT_V1TI): Likewise.
3582         (VEC_SET_V1TI): Likewise.
3583         (VEC_EXT_V1TI): Likewise.
3584         (EQV_V1TI): Likewise.
3585         (NAND_V1TI): Likewise.
3586         (ORC_V1TI): Likewise.
3587         (VADDCUQ): Add support for 128-bit integer arithmetic instructions
3588         added in ISA 2.07.  Add both normal 'altivec' builtins, and the
3589         overloaded builtin.
3590         (VADDUQM): Likewise.
3591         (VSUBCUQ): Likewise.
3592         (VADDEUQM): Likewise.
3593         (VADDECUQ): Likewise.
3594         (VSUBEUQM): Likewise.
3595         (VSUBECUQ): Likewise.
3597         * config/rs6000/rs6000-c.c (__int128_type): New static to hold
3598         __int128_t and __uint128_t types.
3599         (__uint128_type): Likewise.
3600         (altivec_categorize_keyword): Add support for vector __int128_t,
3601         vector __uint128_t, vector __int128, and vector unsigned __int128
3602         as a container type for TImode operations that need to be done in
3603         VSX/Altivec registers.
3604         (rs6000_macro_to_expand): Likewise.
3605         (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
3606         to support 128-bit integer instructions vaddcuq, vadduqm,
3607         vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
3608         (altivec_resolve_overloaded_builtin): Add support for V1TImode.
3610         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
3611         for V1TImode, and set up preferences to use VSX/Altivec registers.
3612         Setup VSX reload handlers.
3613         (rs6000_debug_reg_global): Likewise.
3614         (rs6000_init_hard_regno_mode_ok): Likewise.
3615         (rs6000_preferred_simd_mode): Likewise.
3616         (vspltis_constant): Do not allow V1TImode as easy altivec constants.
3617         (easy_altivec_constant): Likewise.
3618         (output_vec_const_move): Likewise.
3619         (rs6000_expand_vector_set): Convert V1TImode set and extract to
3620         simple move.
3621         (rs6000_expand_vector_extract): Likewise.
3622         (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
3623         addressing.
3624         (rs6000_const_vec): Add support for V1TImode.
3625         (rs6000_emit_le_vsx_load): Swap double words when loading or
3626         storing TImode/V1TImode.
3627         (rs6000_emit_le_vsx_store): Likewise.
3628         (rs6000_emit_le_vsx_move): Likewise.
3629         (rs6000_emit_move): Add support for V1TImode.
3630         (altivec_expand_ld_builtin): Likewise.
3631         (altivec_expand_st_builtin): Likewise.
3632         (altivec_expand_vec_init_builtin): Likewise.
3633         (altivec_expand_builtin): Likewise.
3634         (rs6000_init_builtins): Add support for V1TImode type.  Add
3635         support for ISA 2.07 128-bit integer builtins.  Define type names
3636         for the VSX/Altivec vector types.
3637         (altivec_init_builtins): Add support for overloaded vector
3638         functions with V1TImode type.
3639         (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
3640         (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
3641         external function.
3642         (rs6000_split_128bit_ok_p): Likewise.
3643         (rs6000_handle_altivec_attribute): Create V1TImode from vector
3644         __int128_t and vector __uint128_t.
3646         * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
3647         and mode attributes.
3648         (VSX_M): Likewise.
3649         (VSX_M2): Likewise.
3650         (VSm): Likewise.
3651         (VSs): Likewise.
3652         (VSr): Likewise.
3653         (VSv): Likewise.
3654         (VS_scalar): Likewise.
3655         (VS_double): Likewise.
3656         (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
3658         * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
3659         we support the ISA 2.07 128-bit integer arithmetic instructions.
3660         (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
3661         (enum rs6000_builtin_type_index): Add fields to hold V1TImode
3662         and TImode types for use with the builtin functions.
3663         (V1TI_type_node): Likewise.
3664         (unsigned_V1TI_type_node): Likewise.
3665         (intTI_type_internal_node): Likewise.
3666         (uintTI_type_internal_node): Likewise.
3668         * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
3669         128-bit builtin functions.
3670         (UNSPEC_VADDEUQM): Likewise.
3671         (UNSPEC_VADDECUQ): Likewise.
3672         (UNSPEC_VSUBCUQ): Likewise.
3673         (UNSPEC_VSUBEUQM): Likewise.
3674         (UNSPEC_VSUBECUQ): Likewise.
3675         (VM): Add V1TImode to vector mode iterators.
3676         (VM2): Likewise.
3677         (VI_unit): Likewise.
3678         (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
3679         (altivec_vaddcuq): Likewise.
3680         (altivec_vsubuqm): Likewise.
3681         (altivec_vsubcuq): Likewise.
3682         (altivec_vaddeuqm): Likewise.
3683         (altivec_vaddecuq): Likewise.
3684         (altivec_vsubeuqm): Likewise.
3685         (altivec_vsubecuq): Likewise.
3687         * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
3688         mode iterators.
3689         (BOOL_128): Likewise.
3690         (BOOL_REGS_OUTPUT): Likewise.
3691         (BOOL_REGS_OP1): Likewise.
3692         (BOOL_REGS_OP2): Likewise.
3693         (BOOL_REGS_UNARY): Likewise.
3694         (BOOL_REGS_AND_CR0): Likewise.
3696         * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
3697         128-bit integer builtin support.
3698         (vec_vadduqm): Likewise.
3699         (vec_vaddecuq): Likewise.
3700         (vec_vaddeuqm): Likewise.
3701         (vec_vsubecuq): Likewise.
3702         (vec_vsubeuqm): Likewise.
3703         (vec_vsubcuq): Likewise.
3704         (vec_vsubuqm): Likewise.
3706         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
3707         Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
3708         vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
3709         128-bit integer add/subtract to ISA 2.07.
3711 2014-03-12  Joern Rennecke  <joern.rennecke@embecosm.com>
3713         * config/arc/arc.c (arc_predicate_delay_insns):
3714         Fix third argument passed to conditionalize_nonjump.
3716 2014-03-12  Yufeng Zhang  <yufeng.zhang@arm.com>
3718         * config/aarch64/aarch64-builtins.c
3719         (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
3720         BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
3721         * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
3722         instead of __builtin_lfloor.
3723         (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
3725 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3727         * tree-ssa-ifcombine.c (forwarder_block_to): New function.
3728         (tree_ssa_ifcombine_bb_1): New function.
3729         (tree_ssa_ifcombine_bb): Use it.  Handle also cases where else_bb
3730         is an empty forwarder block to then_bb or vice versa and then_bb
3731         and else_bb are effectively swapped.
3733 2014-03-12  Christian Bruel  <christian.bruel@st.com>
3735         PR target/60264
3736         * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
3737         REG_CFA_DEF_CFA note.
3738         (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
3739         (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
3741 2014-03-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>
3743         PR tree-optimization/60454
3744         * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
3746 2014-03-12  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3748         * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
3749         Do not define target_cpu_default2 to generic.
3750         * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
3751         * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
3752         * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
3754 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3755             Marc Glisse  <marc.glisse@inria.fr>
3757         PR tree-optimization/60502
3758         * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
3759         instead of build_low_bits_mask.
3761 2014-03-12  Jakub Jelinek  <jakub@redhat.com>
3763         PR middle-end/60482
3764         * tree-vrp.c (register_edge_assert_for_1): Don't add assert
3765         if there are multiple uses, but op doesn't live on E edge.
3766         * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
3767         clobber stmts before __builtin_unreachable.
3769 2014-03-11  Richard Sandiford  <rdsandiford@googlemail.com>
3771         * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
3772         hard_frame_pointer_rtx.
3773         * cse.c (cse_insn): Remove volatile check.
3774         * cselib.c (cselib_process_insn): Likewise.
3775         * dse.c (scan_insn): Likewise.
3777 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
3779         * config/arc/arc.c (conditionalize_nonjump): New function,
3780         broken out of ...
3781         (arc_ifcvt): ... this.
3782         (arc_predicate_delay_insns): Use it.
3784 2014-03-11  Joern Rennecke  <joern.rennecke@embecosm.com>
3786         * config/arc/predicates.md (extend_operand): During/after reload,
3787         allow const_int_operand.
3788         * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
3789         (umulsidi3_700): Likewise.  Change operand 2 constraint back to "cL".
3790         (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
3791         to "i".
3792         (umulsi3_highpart_i): Likewise.
3794 2014-03-11  Richard Biener  <rguenther@suse.de>
3796         * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
3797         Add asserts to guard possible wrong-code bugs.
3799 2014-03-11  Richard Biener  <rguenther@suse.de>
3801         PR tree-optimization/60429
3802         PR tree-optimization/60485
3803         * tree-ssa-structalias.c (set_union_with_increment): Properly
3804         take into account all fields that overlap the shifted vars.
3805         (do_sd_constraint): Likewise.
3806         (do_ds_constraint): Likewise.
3807         (get_constraint_for_ptr_offset): Likewise.
3809 2014-03-11  Chung-Lin Tang  <cltang@codesourcery.com>
3811         * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
3812         (nios2_compute_frame_layout):
3813         Add calculation of cfun->machine->fp_save_offset.
3814         (nios2_expand_prologue): Correct setting of frame pointer register
3815         in prologue.
3816         (nios2_expand_epilogue): Update recovery of stack pointer from
3817         frame pointer accordingly.
3818         (nios2_initial_elimination_offset): Update calculation of offset
3819         for eliminating to HARD_FRAME_POINTER_REGNUM.
3821 2014-03-10  Jakub Jelinek  <jakub@redhat.com>
3823         PR ipa/60457
3824         * ipa.c (symtab_remove_unreachable_nodes): Don't call
3825         cgraph_get_create_node on VAR_DECLs.
3827 2014-03-10  Richard Biener  <rguenther@suse.de>
3829         PR middle-end/60474
3830         * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
3832 2014-03-08  Douglas B Rupp  <rupp@gnat.com>
3834         * config/vms/vms.opt (vms_float_format): New variable.
3836 2014-03-08  Tobias Burnus  <burnus@net-b.de>
3838         * doc/invoke.texi (-fcilkplus): Update implementation status.
3840 2014-03-08  Paulo Matos  <paulo@matos-sorge.com>
3841             Richard Biener  <rguenther@suse.de>
3843         * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
3844         consistently accross all TUs.
3845         (run_gcc): Enable -fshort-double automatically at link at link-time
3846         and disallow override.
3848 2014-03-08  Richard Sandiford  <rdsandiford@googlemail.com>
3850         PR target/58271
3851         * config/mips/mips.c (mips_option_override): Promote -mpaired-single
3852         warning to an error.  Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
3853         if they can't be used.
3855 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3857         * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
3858         for Solaris 11/x86 ld.
3859         * configure: Regenerate.
3861 2014-03-07  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3863         * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
3864         (LIB_TLS_SPEC): Save as ld_tls_libs.
3865         (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
3866         (HAVE_AS_IX86_TLSLDM): New test.
3867         * configure, config.in: Regenerate.
3868         * config/i386/i386.c (legitimize_tls_address): Fall back to
3869         TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
3870         cannot support TLS_MODEL_LOCAL_DYNAMIC.
3871         * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
3872         instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
3874 2014-03-07  Paulo Matos  <paulo@matos-sorge.com>
3876         * common.opt (fira-loop-pressure): Mark as optimization.
3878 2014-03-07  Thomas Schwinge  <thomas@codesourcery.com>
3880         * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
3881         an OpenMP mappable type.
3883 2014-03-06  Matthias Klose  <doko@ubuntu.com>
3885         * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
3886         MULTILIB_OSDIRNAMES is not defined.
3888 2014-03-06  Jakub Jelinek  <jakub@redhat.com>
3889             Meador Inge  <meadori@codesourcery.com>
3891         PR target/58595
3892         * config/arm/arm.c (arm_tls_symbol_p): Remove.
3893         (arm_legitimize_address): Call legitimize_tls_address for any
3894         arm_tls_referenced_p expression, handle constant addend.  Call it
3895         before testing for !TARGET_ARM.
3896         (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
3898 2014-03-06  Richard Biener  <rguenther@suse.de>
3900         PR middle-end/60445
3901         PR lto/60424
3902         PR lto/60427
3903         Revert
3904         2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
3906         * tree-streamer.c (record_common_node): Assert we don't record
3907         nodes with type double.
3908         (preload_common_node): Skip type double, complex double and double
3909         pointer since it is now frontend dependent due to fshort-double option.
3911 2014-03-06  Richard Biener  <rguenther@suse.de>
3913         * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
3914         or -fno-lto is specified and the linker has full plugin support.
3915         * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
3916         (main): Remove -flto processing, adjust lto_mode using use_plugin late.
3917         * lto-wrapper.c (merge_and_complain): Merge compile-time
3918         optimization levels.
3919         (run_gcc): And pass it through to the link options.
3921 2014-03-06  Alexandre Oliva  <aoliva@redhat.com>
3923         PR debug/60381
3924         Revert:
3925         2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
3926         PR debug/59992
3927         * cselib.c (remove_useless_values): Skip to avoid quadratic
3928         behavior if the condition moved from...
3929         (cselib_process_insn): ... here holds.
3931 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
3933         PR plugins/59335
3934         * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
3935         ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
3937         PR plugins/59335
3938         * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
3939         (TM_H): Add x86-tune.def.
3941 2014-03-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3943         * config/aarch64/aarch64.c (generic_tunings):
3944         Use cortexa57_extra_costs.
3946 2014-03-05  Jakub Jelinek  <jakub@redhat.com>
3948         PR lto/60404
3949         * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
3950         of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
3951         * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
3952         cost for in_lto_p.
3954 2014-03-04  Heiher  <r@hev.cc>
3956         * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
3957         * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
3959 2014-03-04  Uros Bizjak  <ubizjak@gmail.com>
3961         * config/i386/predicates.md (const2356_operand): Change to ...
3962         (const2367_operand): ... this.
3963         * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
3964         const2367_operand.
3965         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3966         (*avx512pf_scatterpf<mode>sf): Ditto.
3967         (avx512pf_scatterpf<mode>df): Ditto.
3968         (*avx512pf_scatterpf<mode>df_mask): Ditto.
3969         (*avx512pf_scatterpf<mode>df): Ditto.
3970         * config/i386/i386.c (ix86_expand_builtin): Update
3971         incorrect hint operand error message.
3973 2014-03-04  Richard Biener  <rguenther@suse.de>
3975         * lto-section-in.c (lto_get_section_data): Fix const cast.
3977 2014-03-04  Paulo Matos  <paulo@matos-sorge.com>
3979         * tree-streamer.c (record_common_node): Assert we don't record
3980         nodes with type double.
3981         (preload_common_node): Skip type double, complex double and double
3982         pointer since it is now frontend dependent due to fshort-double option.
3984 2014-03-04  Richard Biener  <rguenther@suse.de>
3986         PR lto/60405
3987         * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
3988         (lto_input_toplevel_asms): Likewise.
3989         * lto-section-in.c (lto_get_section_data): Instead do it here
3990         for every section.
3992 2014-03-04  Richard Biener  <rguenther@suse.de>
3994         PR tree-optimization/60382
3995         * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
3996         dead PHIs a reduction.
3998 2014-03-03  Uros Bizjak  <ubizjak@gmail.com>
4000         * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
4001         hint value.
4002         (_mm_prefetch): Move out of GCC target("sse") pragma.
4003         * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
4004         GCC target("prfchw") pragma.
4005         * config/i386/i386.md (prefetch): Emit prefetchwt1 only
4006         for locality <= 2.
4007         * config/i386/i386.c (ix86_option_override_internal): Enable
4008         -mprfchw with -mprefetchwt1.
4010 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
4012         * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
4013         Mark as varying.
4015 2014-03-03  Joern Rennecke  <joern.rennecke@embecosm.com>
4017         * opts.h (CL_PCH_IGNORE): Define.
4018         * targhooks.c (option_affects_pch_p):
4019         Return false for options that have CL_PCH_IGNORE set.
4020         * opt-functions.awk: Process PchIgnore.
4021         * doc/options.texi: Document PchIgnore.
4023         * config/arc/arc.opt (misize): Add PchIgnore property.
4025 2014-03-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4027         * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
4028         reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
4029         constraint on constants to permit them being loaded into
4030         GENERAL_REGS or BASE_REGS.
4032 2014-03-03  Nick Clifton  <nickc@redhat.com>
4034         * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
4035         anti-cacnonical alternatives.
4036         (negandhi3_real): New pattern.
4037         * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
4039 2014-03-03  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
4041         * config/avr/avr-mcus.def: Remove atxmega16x1.
4042         * config/avr/avr-tables.opt: Regenerate.
4043         * config/avr/t-multilib: Regenerate.
4044         * doc/avr-mmcu.texi: Regenerate.
4046 2014-03-03  Tobias Grosser  <tobias@grosser.es>
4047             Mircea Namolaru  <mircea.namolaru@inria.fr>
4049         PR tree-optimization/58028
4050         * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
4051         scalar dimensions.
4053 2014-03-03  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4055         * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
4056         not handled by recognizers.
4058 2014-03-03  Jakub Jelinek  <jakub@redhat.com>
4060         PR middle-end/60175
4061         * function.c (expand_function_end): Don't emit
4062         clobber_return_register sequence if clobber_after is a BARRIER.
4063         * cfgexpand.c (construct_exit_block): Append instructions before
4064         return_label to prev_bb.
4066 2014-03-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4068         * config/rs6000/constraints.md: Document reserved use of "wc".
4070 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
4072         PR ipa/60150
4073         * ipa.c (function_and_variable_visibility): When dissolving comdat
4074         group, also set all symbols to local.
4076 2014-03-02  Jan Hubicka  <hubicka@ucw.cz>
4078         PR ipa/60306
4080         Revert:
4081         2013-12-14   Jan Hubicka  <jh@suse.cz>
4082         PR middle-end/58477
4083         * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
4085 2014-03-02  Jon Beniston  <jon@beniston.com>
4087         PR bootstrap/48230
4088         PR bootstrap/50927
4089         PR bootstrap/52466
4090         PR target/46898
4091         * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
4092           (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
4093         * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
4094         (simple_return, *simple_return): New patterns
4095         * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
4096         * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
4098 2014-03-01  Paolo Carlini  <paolo.carlini@oracle.com>
4100         * dwarf2out.c (gen_subprogram_die): Tidy.
4102 2014-03-01  Oleg Endo  <olegendo@gcc.gnu.org>
4104         PR target/60071
4105         * config/sh/sh.md (*mov_t_msb_neg): Split into ...
4106         (*mov_t_msb_neg_negc): ... this new insn.
4108 2014-02-28  Jason Merrill  <jason@redhat.com>
4110         PR c++/58678
4111         * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
4112         function.
4114 2014-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
4116         PR c++/60314
4117         * dwarf2out.c (decltype_auto_die): New static.
4118         (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
4119         (gen_type_die_with_usage): Handle 'decltype(auto)'.
4120         (is_cxx_auto): Likewise.
4122 2014-02-28  Ian Bolton  <ian.bolton@arm.com>
4124         * config/aarch64/aarch64.h: Define __ARM_NEON by default if
4125         we are not using general regs only.
4127 2014-02-28  Richard Biener  <rguenther@suse.de>
4129         PR target/60280
4130         * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
4131         previous fix and only allow to remove trivial pre-headers
4132         and latches.  Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
4133         (remove_forwarder_block): Properly update the latch of a loop.
4135 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4137         PR debug/59992
4138         * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
4139         (cselib_preserved_hash_table): New.
4140         (preserve_constants_and_equivs): Move preserved vals to it.
4141         (cselib_find_slot): Look it up first.
4142         (cselib_init): Initialize it.
4143         (cselib_finish): Release it.
4144         (dump_cselib_table): Dump it.
4146 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4148         PR debug/59992
4149         * cselib.c (remove_useless_values): Skip to avoid quadratic
4150         behavior if the condition moved from...
4151         (cselib_process_insn): ... here holds.
4153 2014-02-28  Alexandre Oliva  <aoliva@redhat.com>
4155         PR debug/57232
4156         * var-tracking.c (vt_initialize): Apply the same condition to
4157         preserve the CFA base value.
4159 2014-02-28  Joey Ye  <joey.ye@arm.com>
4161         PR target/PR60169
4162         * config/arm/arm.c (thumb_far_jump_used_p): Don't change
4163         if reload in progress or completed.
4165 2014-02-28  Tobias Burnus  <burnus@net-b.de>
4167         PR middle-end/60147
4168         * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
4169         NAMELIST_DECL.
4171 2014-02-27  H.J. Lu  <hongjiu.lu@intel.com>
4173         * doc/tm.texi.in (Condition Code Status): Update documention for
4174         relative locations of cc0-setter and cc0-user.
4176 2014-02-27  Jeff Law  <law@redhat.com>
4178         PR rtl-optimization/52714
4179         * combine.c (try_combine): When splitting an unrecognized PARALLEL
4180         into two independent simple sets, if I3 is a jump, ensure the
4181         pattern we place into I3 is a (set (pc) ...).
4183 2014-02-27  Mikael Pettersson  <mikpe@it.uu.se>
4184             Jeff Law  <law@redhat.com>
4186         PR rtl-optimization/49847
4187         * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
4188         are in different blocks.
4189         * doc/tm.texi (Condition Code Status): Update documention for
4190         relative locations of cc0-setter and cc0-user.
4192 2014-02-27  Vladimir Makarov  <vmakarov@redhat.com>
4194         PR target/59222
4195         * lra.c (lra_emit_add): Check SUBREG too.
4197 2014-02-27  Andreas Schwab  <schwab@suse.de>
4199         * config/m68k/m68k.c (m68k_option_override): Disable
4200         -flive-range-shrinkage for classic m68k.
4201         (m68k_override_options_after_change): Likewise.
4203 2014-02-27  Marek Polacek  <polacek@redhat.com>
4205         PR middle-end/59223
4206         * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
4207         -Wmaybe-uninitialized.
4209 2014-02-27  Alan Modra  <amodra@gmail.com>
4211         PR target/57936
4212         * reload1.c (emit_input_reload_insns): When reload_override_in,
4213         set old to rl->in_reg when rl->in_reg is a subreg.
4215 2014-02-26  Richard Biener  <rguenther@suse.de>
4217         PR bootstrap/60343
4218         * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
4220 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
4222         * common/config/i386/predicates.md (const1256_operand): Remove.
4223         (const2356_operand): New.
4224         (const_1_to_2_operand): Remove.
4225         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
4226         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4227         (*avx512pf_gatherpf<mode>sf): Ditto.
4228         (avx512pf_gatherpf<mode>df): Ditto.
4229         (*avx512pf_gatherpf<mode>df_mask): Ditto.
4230         (*avx512pf_gatherpf<mode>df): Ditto.
4231         (avx512pf_scatterpf<mode>sf): Ditto.
4232         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4233         (*avx512pf_scatterpf<mode>sf): Ditto.
4234         (avx512pf_scatterpf<mode>df): Ditto.
4235         (*avx512pf_scatterpf<mode>df_mask): Ditto.
4236         (*avx512pf_scatterpf<mode>df): Ditto.
4237         * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
4239 2014-02-26  Ilya Tocar  <ilya.tocar@intel.com>
4241         * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
4242         (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
4243         (_mm512_mask_testn_epi64_mask): Move to ...
4244         * config/i386/avx512cdintrin.h: Here.
4245         * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
4246         * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
4247         (avx512f_scalef<mode><mask_name><round_name>): Ditto.
4248         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
4249         TARGET_AVX512F from TARGET_AVX512CD.
4251 2014-02-26  Richard Biener  <rguenther@suse.de>
4253         PR ipa/60327
4254         * ipa.c (walk_polymorphic_call_targets): Properly guard
4255         call to inline_update_overall_summary.
4257 2014-02-26  Bin Cheng  <bin.cheng@arm.com>
4259         PR target/60280
4260         * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
4261         and latches only if requested.  Fix latch if it is removed.
4262         * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
4263         LOOPS_HAVE_PREHEADERS.
4265 2014-02-25  Andrew Pinski  <apinski@cavium.com>
4267         * builtins.c (expand_builtin_thread_pointer): Create a new target
4268         when the target is NULL.
4270 2014-02-25  Vladimir Makarov  <vmakarov@redhat.com>
4272         PR rtl-optimization/60317
4273         * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
4274         * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
4275         * lra-assigns.c: Include params.h.
4276         (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
4277         other reload pseudos considerations.
4279 2014-02-25  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4281         * config/rs6000/vector.md (*vector_unordered<mode>): Change split
4282         to use canonical form for nor<mode>3.
4284 2014-02-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4286         PR target/55426
4287         * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
4288         conversions.
4290 2014-02-25  Ilya Tocar  <ilya.tocar@intel.com>
4292         * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
4293         (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
4294         (ix86_handle_option): Handle OPT_mprefetchwt1.
4295         * config/i386/cpuid.h (bit_PREFETCHWT1): New.
4296         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
4297         PREFETCHWT1 CPUID.
4298         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4299         OPTION_MASK_ISA_PREFETCHWT1.
4300         * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
4301         (PTA_PREFETCHWT1): New.
4302         (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
4303         (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
4304         * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
4305         * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
4306         (*prefetch_avx512pf_<mode>_: Change into ...
4307         (*prefetch_prefetchwt1_<mode>: This.
4308         * config/i386/i386.opt (mprefetchwt1): New.
4309         * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
4310         (_mm_prefetch): Handle intent to write.
4311         * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
4313 2014-02-25  Richard Biener  <rguenther@suse.de>
4315         PR middle-end/60291
4316         * emit-rtl.c (mem_attrs_htab): Remove.
4317         (mem_attrs_htab_hash): Likewise.
4318         (mem_attrs_htab_eq): Likewise.
4319         (set_mem_attrs): Always allocate new mem-attrs when something changed.
4320         (init_emit_once): Do not allocate mem_attrs_htab.
4322 2014-02-25  Richard Biener  <rguenther@suse.de>
4324         PR lto/60319
4325         * lto-opts.c (lto_write_options): Output non-explicit conservative
4326         -fwrapv, -fno-trapv and -fno-strict-overflow.
4327         * lto-wrapper.c (merge_and_complain): Handle merging those options.
4328         (run_gcc): And pass them through.
4330 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
4332         * sel-sched.c (calculate_new_fences): New parameter ptime.
4333         Calculate it as a maximum over all fence cycles.
4334         (sel_sched_region_2): Adjust the call to calculate_new_fences.
4335         Print the final schedule timing when sched_verbose.
4337 2014-02-25  Andrey Belevantsev  <abel@ispras.ru>
4339         PR rtl-optimization/60292
4340         * sel-sched.c (fill_vec_av_set): Do not reset target availability
4341         bit fot the fence instruction.
4343 2014-02-24  Alangi Derick  <alangiderick@gmail.com>
4345         * calls.h: Fix typo in comment.
4347 2014-02-24  John David Anglin  <danglin@gcc.gnu.org>
4349         * config/pa/pa.c (pa_output_move_double): Don't valididate when
4350         adjusting offsetable addresses.
4352 2014-02-24  Guozhi Wei  <carrot@google.com>
4354         * sparseset.h (sparseset_pop): Fix the wrong index.
4356 2014-02-24  Walter Lee  <walt@tilera.com>
4358         * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
4359         (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
4360         triplet.
4361         * common/config/tilegx/tilegx-common.c
4362         (TARGET_DEFAULT_TARGET_FLAGS): Define.
4363         * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
4364         (LINK_SPEC): Ditto.
4365         * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
4366         * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
4367         (tilegx_gimplify_va_arg_expr): Handle big endian.
4368         (tilegx_expand_unaligned_load): Ditto.
4369         (tilegx_expand_unaligned_store): Ditto.
4370         (TARGET_RETURN_IN_MSB): New.
4371         * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
4372         (TARGET_ENDIAN_DEFAULT): New.
4373         (TARGET_BIG_ENDIAN): Handle big endian.
4374         (BYTES_BIG_ENDIAN): Ditto.
4375         (WORDS_BIG_ENDIAN): Ditto.
4376         (FLOAT_WORDS_BIG_ENDIAN): Ditto.
4377         (ENDIAN_SPEC): New.
4378         (EXTRA_SPECS): New.
4379         * config/tilegx/tilegx.md (extv): Handle big endian.
4380         (extzv): Ditto.
4381         (insn_st<n>): Ditto.
4382         (insn_st<n>_add<bitsuffix>): Ditto.
4383         (insn_stnt<n>): Ditto.
4384         (insn_stnt<n>_add<bitsuffix>):Ditto.
4385         (vec_interleave_highv8qi): Handle big endian.
4386         (vec_interleave_highv8qi_be): New.
4387         (vec_interleave_highv8qi_le): New.
4388         (insn_v1int_h): Handle big endian.
4389         (vec_interleave_lowv8qi): Handle big endian.
4390         (vec_interleave_lowv8qi_be): New.
4391         (vec_interleave_lowv8qi_le): New.
4392         (insn_v1int_l): Handle big endian.
4393         (vec_interleave_highv4hi): Handle big endian.
4394         (vec_interleave_highv4hi_be): New.
4395         (vec_interleave_highv4hi_le): New.
4396         (insn_v2int_h): Handle big endian.
4397         (vec_interleave_lowv4hi): Handle big endian.
4398         (vec_interleave_lowv4hi_be): New.
4399         (vec_interleave_lowv4hi_le): New.
4400         (insn_v2int_l): Handle big endian.
4401         (vec_interleave_highv2si): Handle big endian.
4402         (vec_interleave_highv2si_be): New.
4403         (vec_interleave_highv2si_le): New.
4404         (insn_v4int_h): Handle big endian.
4405         (vec_interleave_lowv2si): Handle big endian.
4406         (vec_interleave_lowv2si_be): New.
4407         (vec_interleave_lowv2si_le): New.
4408         (insn_v4int_l): Handle big endian.
4409         * config/tilegx/tilegx.opt (mbig-endian): New option.
4410         (mlittle-endian): New option.
4411         * doc/install.texi: Document tilegxbe-linux.
4412         * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
4414 2014-02-24  Martin Jambor  <mjambor@suse.cz>
4416         PR ipa/60266
4417         * ipa-cp.c (propagate_constants_accross_call): Bail out early if
4418         there are no parameter descriptors.
4420 2014-02-24  Andrey Belevantsev  <abel@ispras.ru>
4422         PR rtl-optimization/60268
4423         * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
4424         initialization to ...
4425         (sched_rgn_init): ... here.
4426         (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
4428 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4430         * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
4431         names.
4433 2014-02-23  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
4435         * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
4436         definition.
4438 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4440         * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
4441         define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
4443 2014-02-23  David Holsgrove  <david.holsgrove@xilinx.com>
4445         * config/microblaze/predicates.md: Add cmp_op predicate.
4446         * config/microblaze/microblaze.md: Add branch_compare instruction
4447         which uses cmp_op predicate and emits cmp insn before branch.
4448         * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
4449         to microblaze_expand_conditional_branch and consolidate logic.
4450         (microblaze_expand_conditional_branch): emit branch_compare
4451         insn instead of handling cmp op separate from branch insn.
4453 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4455         * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
4456         to permit subregs.
4458 2014-02-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4460         * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
4461         define_insn with define_expand and new define_insn
4462         *altivec_lve<VI_char>x_internal.
4463         (altivec_stve<VI_char>x): Replace define_insn with define_expand
4464         and new define_insn *altivec_stve<VI_char>x_internal.
4465         * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
4466         prototype.
4467         * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
4468         lve*x built-ins.
4469         (altivec_expand_stvex_be): New function.
4471 2014-02-22  Joern Rennecke  <joern.rennecke@embecosm.com>
4473         * config/avr/avr.c (avr_can_eliminate): Allow elimination from
4474         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
4475         * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
4476         ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
4478 2014-02-21  Vladimir Makarov  <vmakarov@redhat.com>
4480         PR target/60298
4481         * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
4482         instead of emit_move_insn.
4484 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4486         * config/rs6000/altivec.md (altivec_vsumsws): Replace second
4487         vspltw with vsldoi.
4488         (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
4489         gen_altivec_vsumsws.
4491 2014-02-21  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4493         * config/rs6000/altivec.md (altivec_lvxl): Rename as
4494         *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
4495         (altivec_lvxl_<mode>): New define_expand incorporating
4496         -maltivec=be semantics where needed.
4497         (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
4498         (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
4499         semantics where needed.
4500         (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
4501         (altivec_stvx_<mode>): New define_expand incorporating
4502         -maltivec=be semantics where needed.
4503         (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
4504         VM2 iterator instead of V4SI.
4505         (altivec_stvxl_<mode>): New define_expand incorporating
4506         -maltivec=be semantics where needed.
4507         * config/rs6000/rs6000-builtin.def: Add new built-in definitions
4508         LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
4509         LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
4510         STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
4511         STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
4512         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
4513         ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
4514         similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
4515         ALTIVEC_BUILTIN_STVXL.
4516         * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
4517         (altivec_expand_stvx_be): Likewise.
4518         * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
4519         (altivec_expand_lvx_be): Likewise.
4520         (altivec_expand_stvx_be): Likewise.
4521         (altivec_expand_builtin): Add cases for
4522         ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
4523         ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
4524         (altivec_init_builtins): Add definitions for
4525         __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
4526         __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
4528 2014-02-21  Catherine Moore  <clm@codesourcery.com>
4530         * doc/invoke.texi (mvirt, mno-virt): Document.
4531         * config/mips/mips.opt (mvirt): New option.
4532         * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
4534 2014-02-21  Richard Biener  <rguenther@suse.de>
4536         PR tree-optimization/60276
4537         * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
4538         (STMT_VINFO_MIN_NEG_DIST): New macro.
4539         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
4540         STMT_VINFO_MIN_NEG_DIST.
4541         * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
4542         made for negative dependence distances still hold.
4544 2014-02-21  Richard Biener  <rguenther@suse.de>
4546         PR middle-end/60291
4547         * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
4548         DECL_INITIAL for globals not in the current function context.
4550 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
4552         PR tree-optimization/56490
4553         * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
4554         * tree-ssa-uninit.c: Include params.h.
4555         (compute_control_dep_chain): Add num_calls argument, return false
4556         if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
4557         num_calls to recursive call.
4558         (find_predicates): Change dep_chain into normal array,
4559         cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
4560         variable and adjust compute_control_dep_chain caller.
4561         (find_def_preds): Likewise.
4563 2014-02-21  Thomas Schwinge  <thomas@codesourcery.com>
4565         * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
4566         <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
4568 2014-02-21  Nick Clifton  <nickc@redhat.com>
4570         * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
4571         (pushhi1): Likewise.
4572         (popqi1): Add mode to pre_dec.
4573         (pophi1): Likewise.
4575 2014-02-21  Jakub Jelinek  <jakub@redhat.com>
4577         * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
4578         mode for mask of V8SFmode permutation.
4580 2014-02-20  Richard Henderson  <rth@redhat.com>
4582         PR c++/60272
4583         * builtins.c (expand_builtin_atomic_compare_exchange): Always make
4584         a new pseudo for OLDVAL.
4586 2014-02-20  Jakub Jelinek  <jakub@redhat.com>
4588         PR target/57896
4589         * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
4590         gen_reg_rtx if d->testing_p.
4591         (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
4592         if d->testing_p and we will certainly return true.
4593         (expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
4594         if d->testing_p.
4596 2014-02-20  Uros Bizjak  <ubizjak@gmail.com>
4598         * emit-rtl.c (gen_reg_rtx): Assert that
4599         crtl->emit.regno_pointer_align_length is non-zero.
4601 2014-02-20  Richard Henderson  <rth@redhat.com>
4603         PR c++/60272
4604         * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
4605         on failure the store back into EXPECT.
4607 2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
4608             Sandra Loosemore  <sandra@codesourcery.com>
4610         * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
4611         * config/nios2/nios2.c (nios2_function_profiler): Add
4612         -fPIC (flag_pic == 2) support.
4613         (nios2_handle_custom_fpu_cfg): Fix warning parameter.
4614         (nios2_large_offset_p): New function.
4615         (nios2_unspec_reloc_p): Move up position, update to use
4616         nios2_large_offset_p.
4617         (nios2_unspec_address): Remove function.
4618         (nios2_unspec_offset): New function.
4619         (nios2_large_got_address): New function.
4620         (nios2_got_address): Add large offset support.
4621         (nios2_legitimize_tls_address): Update usage of removed and new
4622         functions.
4623         (nios2_symbol_binds_local_p): New function.
4624         (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
4625         (nios2_legitimize_address): Update to use nios2_large_offset_p.
4626         (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
4627         (nios2_print_operand): Merge H/L processing, add hiadj/lo
4628         processing for (const (unspec ...)).
4629         (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
4631 2014-02-20  Richard Biener  <rguenther@suse.de>
4633         * tree-cfg.c (replace_uses_by): Mark altered BBs before
4634         doing the substitution.
4635         (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
4637 2014-02-20  Martin Jambor  <mjambor@suse.cz>
4639         PR ipa/55260
4640         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
4641         info when checking whether lattices are bottom.
4643 2014-02-20  Richard Biener  <rguenther@suse.de>
4645         PR middle-end/60221
4646         * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
4647         regions at -O0.
4649 2014-02-20  Jan Hubicka  <hubicka@ucw.cz>
4651         PR ipa/58555
4652         * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
4653         parameter specifying the scaling.
4654         (inline_call): Update.
4655         (want_inline_recursively): Guard division by zero.
4656         (recursive_inlining): Update.
4657         * ipa-inline.h (clone_inlined_nodes): Update.
4659 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
4661         PR target/60204
4662         * config/i386/i386.c (classify_argument): Pass structures of size
4663         64 bytes or less in register.
4665 2014-02-20  Ilya Tocar  <ilya.tocar@intel.com>
4666             Kirill Yukhin  <kirill.yukhin@intel.com>
4668         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
4669         (_mm_rcp28_round_ss): Ditto.
4670         (_mm_rsqrt28_round_sd): Ditto.
4671         (_mm_rsqrt28_round_ss): Ditto.
4672         * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
4673         (_mm_rcp14_round_ss): Ditto.
4674         (_mm_rsqrt14_round_sd): Ditto.
4675         (_mm_rsqrt14_round_ss): Ditto.
4676         * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
4677         the first input operand, get rid of match_dup.
4678         (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
4679         attribute to sse.
4680         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4681         Ditto.
4682         (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
4683         operand as the first input operand, set type attribute.
4684         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4685         Set type attribute.
4686         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
4687         operand as the first input operand, set type attribute.
4689 2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4691         * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
4692         bit of zero.
4694 2014-02-19  H.J. Lu  <hongjiu.lu@intel.com>
4696         PR target/60207
4697         * config/i386/i386.c (construct_container): Remove TFmode check
4698         for X86_64_INTEGER_CLASS.
4700 2014-02-19  Uros Bizjak  <ubizjak@gmail.com>
4702         PR target/59794
4703         * config/i386/i386.c (type_natural_mode): Warn for ABI changes
4704         only when -Wpsabi is enabled.
4706 2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>
4708          PR target/59799
4709         * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
4710         passing arrays in registers are the same as for structs, so remove the
4711         special case for them.
4713 2014-02-19  Eric Botcazou  <ebotcazou@adacore.com>
4715         * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
4716         destination type, extract only the valid bits if the source type is not
4717         integral and has a different mode.
4719 2014-02-19  Richard Biener  <rguenther@suse.de>
4721         PR ipa/60243
4722         * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
4723         for all calls.
4725 2014-02-19  Richard Biener  <rguenther@suse.de>
4727         PR ipa/60243
4728         * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
4729         (ipa_modify_call_arguments): Emit an argument load explicitely and
4730         preserve virtual SSA form there and for the replacement call.
4731         Do not update SSA form nor free dominance info.
4733 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4735         * ipa.c (function_and_variable_visibility): Also clear WEAK
4736         flag when disolving COMDAT_GROUP.
4738 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4740         * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
4741         * ipa-prop.c (ipa_set_jf_known_type): Return early when
4742         not devirtualizing.
4743         (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
4744         do more sanity checks.
4745         (detect_type_change): Return true when giving up early.
4746         (compute_complex_assign_jump_func): Fix type parameter of
4747         ipa_set_ancestor_jf.
4748         (compute_complex_ancestor_jump_func): Likewise.
4749         (update_jump_functions_after_inlining): Fix updating of
4750         ancestor function.
4751         * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
4753 2014-02-18  Jan Hubicka  <hubicka@ucw.cz>
4755         * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
4756         inline clones when edge disappears.
4758 2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4760         PR target/60203
4761         * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
4762         Split 64-bit moves into 2 patterns.  Do not allow the use of
4763         direct move for TDmode in little endian, since the decimal value
4764         has little endian bytes within a word, but the 64-bit pieces are
4765         ordered in a big endian fashion, and normal subreg's of TDmode are
4766         not allowed.
4767         (mov<mode>_64bit_dm): Likewise.
4768         (movtd_64bit_nodm): Likewise.
4770 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
4772         PR tree-optimization/60174
4773         * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
4774         statement of an SSA_NAME that occurs in an abnormal PHI node.
4776 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
4778         PR sanitizer/60142
4779         * final.c (SEEN_BB): Remove.
4780         (SEEN_NOTE, SEEN_EMITTED): Renumber.
4781         (final_scan_insn): Don't force_source_line on second
4782         NOTE_INSN_BASIC_BLOCK.
4784 2014-02-18  Uros Bizjak  <ubizjak@gmail.com>
4786         PR target/60205
4787         * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
4788         * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
4789         (type_natural_mode): Warn ABI change when %zmm register is not
4790         available for AVX512F vector value passing.
4792 2014-02-18  Kai Tietz  <ktietz@redhat.com>
4794         PR target/60193
4795         * config/i386/i386.c (ix86_expand_prologue): Use value in
4796         rax register as displacement when restoring %r10 or %rax.
4797         Fix wrong offset when restoring both registers.
4799 2014-02-18  Eric Botcazou  <ebotcazou@adacore.com>
4801         * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
4802         assertion with conditional return.
4804 2014-02-18  Jakub Jelinek  <jakub@redhat.com>
4805             Uros Bizjak  <ubizjak@gmail.com>
4807         PR driver/60233
4808         * config/i386/driver-i386.c (host_detect_local_cpu): If
4809         YMM state is not saved by the OS, also clear has_f16c.  Move
4810         CPUID 0x80000001 handling before YMM state saving checking.
4812 2014-02-18  Andrey Belevantsev  <abel@ispras.ru>
4814         PR rtl-optimization/58960
4815         * haifa-sched.c (alloc_global_sched_pressure_data): New,
4816         factored out from ...
4817         (sched_init): ... here.
4818         (free_global_sched_pressure_data): New, factored out from ...
4819         (sched_finish): ... here.
4820         * sched-int.h (free_global_sched_pressure_data): Declare.
4821         * sched-rgn.c (nr_regions_initial): New static global.
4822         (haifa_find_rgns): Initialize it.
4823         (schedule_region): Disable sched-pressure for the newly
4824         generated regions.
4826 2014-02-17  Richard Biener  <rguenther@suse.de>
4828         * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
4829         release SSA defs of pattern stmts.
4831 2014-02-17  Richard Biener  <rguenther@suse.de>
4833         * tree-inline.c (expand_call_inline): Release the virtual
4834         operand defined by the call we are about to inline.
4836 2014-02-17  Richard Biener  <rguenther@suse.de>
4838         * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
4840 2014-02-17  Kirill Yukhin  <kirill.yukhin@intel.com>
4841             Ilya Tocar  <ilya.tocar@intel.com>
4843         * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
4844         arguments order in builtin.
4845         (_mm512_permutexvar_epi64): Ditto.
4846         (_mm512_mask_permutexvar_epi64): Ditto
4847         (_mm512_maskz_permutexvar_epi32): Ditto
4848         (_mm512_permutexvar_epi32): Ditto
4849         (_mm512_mask_permutexvar_epi32): Ditto
4851 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4853         * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
4854         (p8_vmrgow): Likewise.
4856 2014-02-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4858         * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
4859         endian targets.
4861 2014-02-15  Michael Meissner  <meissner@linux.vnet.ibm.com>
4863         PR target/60203
4864         * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
4865         (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
4866         into 64-bit and 32-bit moves.  On 64-bit moves, add support for
4867         using direct move instructions on ISA 2.07.  Also adjust
4868         instruction length for 64-bit.
4869         (mov<mode>_64bit, TFmode/TDmode): Likewise.
4870         (mov<mode>_32bit, TFmode/TDmode): Likewise.
4872 2014-02-15  Alan Modra  <amodra@gmail.com>
4874         PR target/58675
4875         PR target/57935
4876         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
4877         find_replacement on parts of insn rtl that might be reloaded.
4879 2014-02-15  Richard Biener  <rguenther@suse.de>
4881         PR tree-optimization/60183
4882         * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
4883         (tree_ssa_phiprop): Calculate and free post-dominators.
4885 2014-02-14  Jeff Law  <law@redhat.com>
4887         PR rtl-optimization/60131
4888         * ree.c (get_extended_src_reg): New function.
4889         (combine_reaching_defs): Use it rather than assuming location of REG.
4890         (find_and_remove_re): Verify first operand of extension is
4891         a REG before adding the insns to the copy list.
4893 2014-02-14  Roland McGrath  <mcgrathr@google.com>
4895         * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
4896         * configure: Regenerated.
4897         * config.in: Regenerated.
4898         * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
4899         instead of ASM_SHORT.
4901 2014-02-14  Vladimir Makarov  <vmakarov@redhat.com>
4902             Richard Earnshaw  <rearnsha@arm.com>
4904         PR rtl-optimization/59535
4905         * lra-constraints.c (process_alt_operands): Encourage alternative
4906         when unassigned pseudo class is superset of the alternative class.
4907         (inherit_reload_reg): Don't inherit when optimizing for code size.
4908         * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
4909         returning CORE_REGS for anything but Thumb1 and BASE_REGS for
4910         modes not less than 4 for Thumb1.
4912 2014-02-14  Kyle McMartin  <kyle@redhat.com>
4914         PR pch/60010
4915         * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
4917 2014-02-14  Richard Biener  <rguenther@suse.de>
4919         * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
4920         (get_frame_arg): Drop the assert with langhook types_compatible_p.
4921         Do not strip INDIRECT_REFs.
4923 2014-02-14  Richard Biener  <rguenther@suse.de>
4925         PR lto/60179
4926         * lto-streamer-out.c (DFS_write_tree_body): Do not follow
4927         DECL_FUNCTION_SPECIFIC_TARGET.
4928         (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
4929         * tree-streamer-out.c (pack_ts_target_option): Remove.
4930         (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
4931         (write_ts_function_decl_tree_pointers): Do not stream
4932         DECL_FUNCTION_SPECIFIC_TARGET.
4933         * tree-streamer-in.c (unpack_ts_target_option): Remove.
4934         (unpack_value_fields): Do not stream TS_TARGET_OPTION.
4935         (lto_input_ts_function_decl_tree_pointers): Do not stream
4936         DECL_FUNCTION_SPECIFIC_TARGET.
4938 2014-02-14  Jakub Jelinek  <jakub@redhat.com>
4940         * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
4941         (get_initial_def_for_induction, vectorizable_induction): Ignore
4942         debug stmts when looking for exit_phi.
4943         (vectorizable_live_operation): Fix up condition.
4945 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
4947         * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
4948         nreverse() because it changes the content of original tree list.
4950 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
4952         * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
4953         * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
4955 2014-02-14  Chung-Ju Wu  <jasonwucj@gmail.com>
4957         * config/nds32/nds32.c (nds32_naked_function_p): Follow the
4958         GNU coding standards.
4960 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
4962         PR debug/60152
4963         * dwarf2out.c (gen_subprogram_die): Don't call
4964         add_calling_convention_attribute if subr_die is old_die.
4966 2014-02-13  Sharad Singhai  <singhai@google.com>
4968         * doc/optinfo.texi: Fix order of nodes.
4970 2014-02-13  Uros Bizjak  <ubizjak@gmail.com>
4972         * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
4973         operands[2], not operands[3].
4975 2014-02-13  Richard Biener  <rguenther@suse.de>
4977         PR bootstrap/59878
4978         * doc/install.texi (ISL): Update recommended version to 0.12.2,
4979         mention the possibility of an in-tree build.
4980         (CLooG): Update recommended version to 0.18.1, mention the
4981         possibility of an in-tree build and clarify that the ISL
4982         bundled with CLooG does not work.
4984 2014-02-13  Jakub Jelinek  <jakub@redhat.com>
4986         PR target/43546
4987         * expr.c (compress_float_constant): If x is a hard register,
4988         extend into a pseudo and then move to x.
4990 2014-02-13  Dominik Vogt  <vogt@linux.vnet.ibm.com>
4992         * config/s390/s390.c (s390_asm_output_function_label): Fix crash
4993         caused by bad second argument to warning_at() with -mhotpatch and
4994         nested functions (e.g. with gfortran).
4996 2014-02-13  Richard Sandiford  <rdsandiford@googlemail.com>
4998         * opts.c (option_name): Remove "enabled by default" rider.
5000 2014-02-12  John David Anglin  <danglin@gcc.gnu.org>
5002         * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
5004 2014-02-12  H.J. Lu  <hongjiu.lu@intel.com>
5005             Uros Bizjak  <ubizjak@gmail.com>
5007         PR target/60151
5008         * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
5009         * configure: Regenerated.
5011 2014-02-12  Richard Biener  <rguenther@suse.de>
5013         * vec.c (vec_prefix::calculate_allocation): Move as
5014         inline variant to vec.h.
5015         (vec_prefix::calculate_allocation_1): New out-of-line version.
5016         * vec.h (vec_prefix::calculate_allocation_1): Declare.
5017         (vec_prefix::m_has_auto_buf): Rename to ...
5018         (vec_prefix::m_using_auto_storage): ... this.
5019         (vec_prefix::calculate_allocation): Inline the easy cases
5020         and dispatch to calculate_allocation_1 which doesn't need the
5021         prefix address.
5022         (va_heap::reserve): Use gcc_checking_assert.
5023         (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
5024         m_using_auto_storage.
5025         (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
5026         member and adjust.
5027         (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
5028         (vec<T, va_heap, vl_ptr>::release): Avoid casting.
5029         (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
5031 2014-02-12  Richard Biener  <rguenther@suse.de>
5033         * gcse.c (compute_transp): break from loop over canon_modify_mem_list
5034         when we found a dependence.
5036 2014-02-12  Thomas Schwinge  <thomas@codesourcery.com>
5038         * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
5039         common code...
5040         (maybe_fold_stmt): ... into this new function.
5041         * omp-low.c (lower_omp): Update comment.
5043         * omp-low.c (lower_omp_target): Add clobber for sizes array, after
5044         last use.
5046         * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
5047         dereference.
5049 2014-02-12  James Greenhalgh  <james.greenhalgh@arm.com>
5051         * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
5052         identifiers in comments.
5053         (cortexa53_extra_costs): Likewise.
5054         * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
5055         (cortexa7_extra_costs): Likewise.
5056         (cortexa12_extra_costs): Likewise.
5057         (cortexa15_extra_costs): Likewise.
5058         (v7m_extra_costs): Likewise.
5060 2014-02-12  Richard Biener  <rguenther@suse.de>
5062         PR middle-end/60092
5063         * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
5064         of posix_memalign being successful.
5065         (lower_stmt): Restrict lowering of posix_memalign to when
5066         -ftree-bit-ccp is enabled.
5068 2014-02-12  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
5070         * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
5071         arg_loc.
5072         * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
5074 2014-02-12  Eric Botcazou  <ebotcazou@adacore.com>
5076         PR rtl-optimization/60116
5077         * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
5078         other_insn once the combination has been validated.
5080 2014-02-11  Jan Hubicka  <hubicka@ucw.cz>
5082         PR lto/59468
5083         * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
5084         and wrapper.
5085         * ipa-devirt.c: Include demangle.h
5086         (odr_violation_reported): New static variable.
5087         (add_type_duplicate): Update odr_violations.
5088         (maybe_record_node): Add completep parameter; update it.
5089         (record_target_from_binfo): Add COMPLETEP parameter;
5090         update it as needed.
5091         (possible_polymorphic_call_targets_1): Likewise.
5092         (struct polymorphic_call_target_d): Add nonconstruction_targets;
5093         rename FINAL to COMPLETE.
5094         (record_targets_from_bases): Sanity check we found the binfo;
5095         fix COMPLETEP updating.
5096         (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
5097         parameter, fix computing of COMPLETEP.
5098         (dump_possible_polymorphic_call_targets): Imrove readability of dump;
5099         at LTO time do demangling.
5100         (ipa_devirt): Use nonconstruction_targets; Improve dumps.
5101         * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
5102         parameter.
5103         (gimple_get_virt_method_for_binfo): Likewise.
5104         * gimple-fold.h (gimple_get_virt_method_for_binfo,
5105         gimple_get_virt_method_for_vtable): Update prototypes.
5107 2014-02-11  Vladimir Makarov  <vmakarov@redhat.com>
5109         PR target/49008
5110         * genautomata.c (add_presence_absence): Fix typo with
5111         {final_}presence_list.
5113 2014-02-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
5115         PR target/60137
5116         * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
5117         for VSX/Altivec vectors that land in GPR registers.
5119 2014-02-11  Richard Henderson  <rth@redhat.com>
5120             Jakub Jelinek  <jakub@redhat.com>
5122         PR debug/59776
5123         * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
5124         around drhs if type conversion to lacc->type is not useless.
5126 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5128         * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
5129         tuning struct.
5130         (cortex-a57.cortex-a53): Likewise.
5131         * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
5133 2014-02-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5135         * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
5136         arm_restrict_it.
5138 2014-02-11  Renlin Li  <Renlin.Li@arm.com>
5140         * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
5141         add_options_for_arm_vfp3.
5143 2014-02-11  Jeff Law  <law@redhat.com>
5145         PR middle-end/54041
5146         * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
5147         object with an undesirable mode.
5149 2014-02-11  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5151         PR libgomp/60107
5152         * config/i386/sol2-9.h: New file.
5153         * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
5154         *-*-solaris2.9*): Use it.
5156 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
5158         * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
5159         * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
5161 2014-02-10  Nagaraju Mekala  <nagaraju.mekala@xilinx.com>
5163         * config/microblaze/microblaze.c: Extend mcpu version format
5165 2014-02-10  David Holsgrove  <david.holsgrove@xilinx.com>
5167         * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
5169 2014-02-10  Richard Henderson  <rth@redhat.com>
5171         PR target/59927
5172         * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
5173         * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
5174         ms-abi vs -mno-accumulate-outgoing-args.
5175         (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
5176         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
5177         respect to ms-abi.
5179 2014-02-10  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5181         PR middle-end/60080
5182         * cfgexpand.c (expand_asm_operands): Attach source location to
5183         ASM_INPUT rtx objects.
5184         * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
5186 2014-02-10  Nick Clifton  <nickc@redhat.com>
5188         * config/mn10300/mn10300.c (popcount): New function.
5189         (mn10300_expand_prologue): Include saved registers in stack usage
5190         count.
5192 2014-02-10  Jeff Law  <law@redhat.com>
5194         PR middle-end/52306
5195         * reload1.c (emit_input_reload_insns): Do not create invalid RTL
5196         when changing the SET_DEST of a prior insn to avoid an input reload.
5198 2014-02-10  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
5200         * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
5201         big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
5202         -mcall-openbsd, or -mcall-linux.
5203         (CC1_ENDIAN_BIG_SPEC): Remove.
5204         (CC1_ENDIAN_LITTLE_SPEC): Remove.
5205         (CC1_ENDIAN_DEFAULT_SPEC): Remove.
5206         (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
5207         (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
5208         and %cc1_endian_default.
5209         * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
5211 2014-02-10  Richard Biener  <rguenther@suse.de>
5213         PR tree-optimization/60115
5214         * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
5215         MEM_REF handling.  Properly verify that the accesses are not
5216         out of the objects bound.
5218 2014-02-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5220         * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
5221         coretex to cortex.
5223 2014-02-10  Eric Botcazou  <ebotcazou@adacore.com>
5225         * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
5226         proper constants and fix formatting.
5227         (possible_polymorphic_call_targets): Fix formatting.
5229 2014-02-10  Kirill Yukhin  <kirill.yukhin@intel.com>
5230             Ilya Tocar  <ilya.tocar@intel.com>
5232         * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
5233         (_mm512_loadu_epi32): Renamed into...
5234         (_mm512_loadu_si512): This.
5235         (_mm512_storeu_epi32): Renamed into...
5236         (_mm512_storeu_si512): This.
5237         (_mm512_maskz_ceil_ps): Removed.
5238         (_mm512_maskz_ceil_pd): Ditto.
5239         (_mm512_maskz_floor_ps): Ditto.
5240         (_mm512_maskz_floor_pd): Ditto.
5241         (_mm512_floor_round_ps): Ditto.
5242         (_mm512_floor_round_pd): Ditto.
5243         (_mm512_ceil_round_ps): Ditto.
5244         (_mm512_ceil_round_pd): Ditto.
5245         (_mm512_mask_floor_round_ps): Ditto.
5246         (_mm512_mask_floor_round_pd): Ditto.
5247         (_mm512_mask_ceil_round_ps): Ditto.
5248         (_mm512_mask_ceil_round_pd): Ditto.
5249         (_mm512_maskz_floor_round_ps): Ditto.
5250         (_mm512_maskz_floor_round_pd): Ditto.
5251         (_mm512_maskz_ceil_round_ps): Ditto.
5252         (_mm512_maskz_ceil_round_pd): Ditto.
5253         (_mm512_expand_pd): Ditto.
5254         (_mm512_expand_ps): Ditto.
5255         * config/i386/i386.c (ix86_builtins): Remove
5256         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
5257         (bdesc_args): Ditto.
5258         * config/i386/predicates.md (const1256_operand): New.
5259         (const_1_to_2_operand): Ditto.
5260         * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
5261         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
5262         (*avx512pf_gatherpf<mode>sf): Ditto.
5263         (avx512pf_gatherpf<mode>df): Ditto.
5264         (*avx512pf_gatherpf<mode>df_mask): Ditto.
5265         (*avx512pf_gatherpf<mode>df): Ditto.
5266         (avx512pf_scatterpf<mode>sf): Ditto.
5267         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
5268         (*avx512pf_scatterpf<mode>sf): Ditto.
5269         (avx512pf_scatterpf<mode>df): Ditto.
5270         (*avx512pf_scatterpf<mode>df_mask): Ditto.
5271         (*avx512pf_scatterpf<mode>df): Ditto.
5272         (avx512f_expand<mode>): Removed.
5273         (<shift_insn><mode>3<mask_name>): Change predicate type.
5275 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
5277         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
5278         not at the end of datarefs vector use ordered_remove to avoid
5279         reordering datarefs vector.
5281         PR c/59984
5282         * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
5283         mark local addressable non-static vars as GOVD_PRIVATE
5284         instead of GOVD_LOCAL.
5285         * omp-low.c (lower_omp_for): Move gimple_bind_vars
5286         and BLOCK_VARS of gimple_bind_block to new_stmt rather
5287         than copying them.
5289         PR middle-end/60092
5290         * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
5291         if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
5292         assume_aligned or alloc_align attributes.
5293         (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
5294         arguments.  Handle also assume_aligned and alloc_align attributes.
5295         (evaluate_stmt): Adjust bit_value_assume_aligned caller.  Handle
5296         calls to functions with assume_aligned or alloc_align attributes.
5297         * doc/extend.texi: Document assume_aligned and alloc_align attributes.
5299 2014-02-08  Terry Guo  <terry.guo@arm.com>
5301         * doc/invoke.texi: Document ARM -march=armv7e-m.
5303 2014-02-08  Jakub Jelinek  <jakub@redhat.com>
5305         * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
5306         flag on __cilkrts_rethrow builtin.
5308         PR ipa/60026
5309         * ipa-cp.c (determine_versionability): Fail at -O0
5310         or __attribute__((optimize (0))) or -fno-ipa-cp functions.
5311         * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
5313         Revert:
5314         2014-02-04  Jakub Jelinek  <jakub@redhat.com>
5316         PR ipa/60026
5317         * tree-inline.c (copy_forbidden): Fail for
5318         __attribute__((optimize (0))) functions.
5320 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
5322         * varpool.c: Include pointer-set.h.
5323         (varpool_remove_unreferenced_decls): Variables in other partitions
5324         will not be output; be however careful to not lose information
5325         about partitioning.
5327 2014-02-07  Jan Hubicka  <hubicka@ucw.cz>
5329         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
5330         lookup in the vtable constructor.
5332 2014-02-07  Jeff Law  <law@redhat.com>
5334         PR target/40977
5335         * config/m68k/m68k.md (ashldi_extsi): Turn into a
5336         define_insn_and_split.
5338         * ipa-inline.c (inline_small_functions): Fix typos.
5340 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5342         * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
5343         (s390_can_use_return_insn): Declare.
5344         * config/s390/s390.h (EPILOGUE_USES): Define.
5345         * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
5346         instructions.
5347         (s390_chunkify_start): Handle return JUMP_LABELs.
5348         (s390_early_mach): Emit a main_pool instruction on the entry edge.
5349         (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
5350         (s390_can_use_return_insn): New functions.
5351         (s390_fix_long_loop_prediction): Handle conditional returns.
5352         (TARGET_SET_UP_BY_PROLOGUE): Define.
5353         * config/s390/s390.md (ANY_RETURN): New code iterator.
5354         (*creturn, *csimple_return, return, simple_return): New patterns.
5356 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5358         * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
5359         notes to each restore.  Also add REG_CFA_DEF_CFA when restoring %r15.
5360         (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P.  Update the
5361         REG_CFA_RESTORE list when deciding not to restore a register.
5363 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5365         * config/s390/s390.c: Include tree-pass.h and context.h.
5366         (s390_early_mach): New function, split out from...
5367         (s390_emit_prologue): ...here.
5368         (pass_data_s390_early_mach): New pass structure.
5369         (pass_s390_early_mach): New class.
5370         (s390_option_override): Create and register early_mach pass.
5371         Move to end of file.
5373 2014-02-07  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
5375         * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
5376         to match for the exit block.
5378 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5380         * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
5381         ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
5382         Reject misaligned operands.
5384 2014-02-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5386         * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
5388 2014-02-07  Richard Biener  <rguenther@suse.de>
5390         PR middle-end/60092
5391         * gimple-low.c (lower_builtin_posix_memalign): New function.
5392         (lower_stmt): Call it to lower posix_memalign in a way
5393         to make alignment info accessible.
5395 2014-02-07  Jakub Jelinek  <jakub@redhat.com>
5397         PR c++/60082
5398         * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
5399         __builtin_setjmp_receiver.
5401 2014-02-07  Richard Biener  <rguenther@suse.de>
5403         PR middle-end/60092
5404         * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
5405         * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
5406         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
5407         Handle BUILT_IN_POSIX_MEMALIGN.
5408         (find_func_clobbers): Likewise.
5409         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5410         (call_may_clobber_ref_p_1): Likewise.
5412 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5414         PR ipa/59918
5415         * ipa-devirt.c (record_target_from_binfo): Remove overactive
5416         sanity check.
5418 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5420         PR ipa/59469
5421         * lto-cgraph.c (lto_output_node): Use
5422         symtab_get_symbol_partitioning_class.
5423         (lto_output_varpool_node): likewise.
5424         (symtab_get_symbol_partitioning_class): Move here from
5425         lto/lto-partition.c
5426         * cgraph.h (symbol_partitioning_class): Likewise.
5427         (symtab_get_symbol_partitioning_class): Declare.
5429 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5431         * ggc.h (ggc_internal_cleared_alloc): New macro.
5432         * vec.h (vec_safe_copy): Handle memory stats.
5433         * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
5434         * target-globals.c (save_target_globals): Likewise.
5436 2014-02-06  Jan Hubicka  <hubicka@ucw.cz>
5438         PR target/60077
5439         * expr.c (emit_move_resolve_push): Export; be bit more selective
5440         on when to clear alias set.
5441         * expr.h (emit_move_resolve_push): Declare.
5442         * function.h (struct function): Add tail_call_marked.
5443         * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
5444         * config/i386/i386-protos.h (ix86_expand_push): Remove.
5445         * config/i386/i386.md (TImode move expander): De not call
5446         ix86_expand_push.
5447         (FP push expanders): Preserve memory attributes.
5448         * config/i386/sse.md (push<mode>1): Remove.
5449         * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
5450         (ix86_expand_push): Remove.
5451         * config/i386/mmx.md (push<mode>1): Remove.
5453 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5455         PR rtl-optimization/60030
5456         * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
5457         lopart with paradoxical subreg before shifting it up by hprec.
5459 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5461         * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
5462         Remove extra newline at end of file.
5463         * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
5464         (arm_issue_rate): Handle cortexa57.
5465         * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
5466         (cortex-a57.cortex-a53): Likewise.
5468 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5470         PR target/59575
5471         * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
5472         don't record in REG_FRAME_RELATED_EXPR registers not set in that
5473         bitmask.
5474         (arm_expand_prologue): Adjust all callers.
5475         (arm_unwind_emit_sequence): Allow saved, but not important for unwind
5476         info, registers also at the lowest numbered registers side.  Use
5477         gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
5478         XEXP.
5480         PR debug/59992
5481         * var-tracking.c (adjust_mems): Before adding a SET to
5482         amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
5484 2014-02-06  Alan Modra  <amodra@gmail.com>
5486         PR target/60032
5487         * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
5488         change SDmode to DDmode when lra_in_progress.
5490 2014-02-06  Jakub Jelinek  <jakub@redhat.com>
5492         PR middle-end/59150
5493         * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
5494         free_data_ref on the dr first, and before goto again also set dr
5495         to the next dr.  For simd_lane_access, free old datarefs[i] before
5496         overwriting it.  For get_vectype_for_scalar_type failure, don't
5497         free_data_ref if simd_lane_access.
5499         * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
5501         PR target/60062
5502         * tree.h (opts_for_fn): New inline function.
5503         (opt_for_fn): Define.
5504         * config/i386/i386.c (ix86_function_regparm): Use
5505         opt_for_fn (decl, optimize) instead of optimize.
5507 2014-02-06  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5509         * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
5510         for SYMBOL_REF in large memory model.
5512 2014-02-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5514         * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
5515         and crypto support.
5516         (cortex-a57): Likewise.
5517         (cortex-a57.cortex-a53): Likewise.
5519 2014-02-06  Yury Gribov  <y.gribov@samsung.com>
5520             Kugan Vivekanandarajah  <kuganv@linaro.org>
5522         * config/arm/arm.c (arm_vector_alignment_reachable): Check
5523         unaligned_access.
5524         * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
5526 2014-02-06  Richard Biener  <rguenther@suse.de>
5528         * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
5529         set_loop_copy and initialize_original_copy_tables.
5531 2014-02-06  Alex Velenko  <Alex.Velenko@arm.com>
5533         * config/aarch64/aarch64-simd.md
5534         (aarch64_ashr_simddi): Change QI to SI.
5536 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
5537             Jakub Jelinek  <jakub@redhat.com>
5539         PR middle-end/60013
5540         * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
5541         of the dataflow.
5543 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5545         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
5546         CODE_FOR_altivec_vpku[hw]um to
5547         CODE_FOR_altivec_vpku[hw]um_direct.
5548         * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
5549         UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
5550         (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
5551         UNSPEC_VUNPACK_LO_SIGN_DIRECT.
5553 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5555         * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
5556         generation for -maltivec=be.
5557         (altivec_vsumsws): Simplify redundant test.
5559 2014-02-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5561         * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
5562         (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
5563         (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
5564         (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
5565         gen_altivec_vpkuwum.
5566         (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
5567         BYTES_BIG_ENDIAN.
5568         (altivec_vpks<VI_char>ss): Likewise.
5569         (altivec_vpks<VI_char>us): Likewise.
5570         (altivec_vpku<VI_char>us): Likewise.
5571         (altivec_vpku<VI_char>um): Likewise.
5572         (altivec_vpku<VI_char>um_direct): New (copy of
5573         altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
5574         internal use).
5575         (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
5576         target is little endian and -maltivec=be is not specified.
5577         (*altivec_vupkhs<VU_char>_direct): New (copy of
5578         altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
5579         (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
5580         target is little endian and -maltivec=be is not specified.
5581         (*altivec_vupkls<VU_char>_direct): New (copy of
5582         altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
5583         (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
5584         little endian and -maltivec=be is not specified.
5585         (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
5586         little endian and -maltivec=be is not specified.
5588 2014-02-05  Richard Henderson  <rth@redhat.com>
5590         PR debug/52727
5591         * combine-stack-adj.c: Revert r206943.
5592         * sched-int.h (struct deps_desc): Add last_args_size.
5593         * sched-deps.c (init_deps): Initialize it.
5594         (sched_analyze_insn): Add OUTPUT dependencies between insns that
5595         contain REG_ARGS_SIZE notes.
5597 2014-02-05  Jan Hubicka  <hubicka@ucw.cz>
5599         * lto-cgraph.c (asm_nodes_output): Make global.
5600         * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
5601         * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
5602         (driver_handle_option): Handle OPT_fwpa.
5604 2014-02-05  Jakub Jelinek  <jakub@redhat.com>
5606         PR ipa/59947
5607         * ipa-devirt.c (possible_polymorphic_call_targets): Fix
5608         a comment typo and formatting issue.  If odr_hash hasn't been
5609         created, return vNULL and set *completep to false.
5611         PR middle-end/57499
5612         * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
5613         bb with no successors.
5615 2014-02-05  James Greenhalgh  <james.greenhalgh@arm.com>
5617         PR target/59718
5618         * doc/invoke.texi (-march): Clarify documentation for ARM.
5619         (-mtune): Likewise.
5620         (-mcpu): Likewise.
5622 2014-02-05  Richard Biener  <rguenther@suse.de>
5624         * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
5625         when not vectorizing because of too many alias checks.
5626         * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
5627         Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
5629 2014-02-05  Nick Clifton  <nickc@redhat.com>
5631         * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
5632         accept extended registers in any mode when compiling for the MN10300.
5634 2014-02-05  Yury Gribov  <y.gribov@samsung.com>
5636         * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
5637         * ipa-inline.c (report_inline_failed_reason): Handle mismatched
5638         sanitization attributes.
5639         (can_inline_edge_p): Likewise.
5640         (sanitize_attrs_match_for_inline_p): New function.
5642 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5644         * ipa-prop.c (detect_type_change): Shor circuit testing of
5645         type changes on THIS pointer.
5647 2014-02-04  John David Anglin  <danglin@gcc.gnu.org>
5649         PR target/59777
5650         * config/pa/pa.c (legitimize_tls_address): Return original address
5651         if not passed a SYMBOL_REF rtx.
5652         (hppa_legitimize_address): Call legitimize_tls_address for all TLS
5653         addresses.
5654         (pa_emit_move_sequence): Simplify TLS source operands.
5655         (pa_legitimate_constant_p): Reject all TLS constants.
5656         * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
5657         (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
5659 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5661         * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
5662         groups when we know they are controlled by LTO.
5663         * varasm.c (default_binds_local_p_1): If object is in other partition,
5664         it will be resolved locally.
5666 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5668         * config/host-linux.c (linux_gt_pch_use_address): Don't
5669         use SSIZE_MAX because it is not always defined.
5671 2014-02-04  Vladimir Makarov  <vmakarov@redhat.com>
5673         PR bootstrap/59913
5674         * lra-constraints.c (need_for_split_p): Use more 3 reloads as
5675         threshold for pseudo splitting.
5676         (update_ebb_live_info): Process call argument hard registers and
5677         hard registers from insn definition too.
5678         (max_small_class_regs_num): New constant.
5679         (inherit_in_ebb): Update live hard regs through EBBs.  Update
5680         reloads_num only for small register classes.  Don't split for
5681         outputs of jumps.
5683 2014-02-04  Markus Trippelsdorf  <markus@trippelsdorf.de>
5685         PR ipa/60058
5686         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
5687         is non-null.
5689 2014-02-04  Jan Hubicka  <hubicka@ucw.cz>
5691         * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
5692         visibility is safe.
5694 2014-02-04  Marek Polacek  <polacek@redhat.com>
5696         * gdbinit.in (pel): Define.
5698 2014-02-04  Bernd Edlinger  <bernd.edlinger@hotmail.de>
5700         * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
5701         behavior.
5703 2014-02-04  Richard Biener  <rguenther@suse.de>
5705         PR lto/59723
5706         * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
5707         in function context local.
5708         (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
5709         * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
5710         similar to LTO_imported_decl_ref.
5712 2014-02-04  Jakub Jelinek  <jakub@redhat.com>
5714         PR tree-optimization/60002
5715         * cgraphclones.c (build_function_decl_skip_args): Clear
5716         DECL_LANG_SPECIFIC.
5718         PR tree-optimization/60023
5719         * tree-if-conv.c (predicate_mem_writes): Pass true instead of
5720         false to gsi_replace.
5721         * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
5722         has been in some EH region and vec_stmt could throw, add
5723         vec_stmt into the same EH region.
5724         * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
5725         has no lhs, ignore it.
5726         * internal-fn.c (expand_MASK_LOAD): Likewise.
5728         PR ipa/60026
5729         * tree-inline.c (copy_forbidden): Fail for
5730         __attribute__((optimize (0))) functions.
5732         PR other/58712
5733         * omp-low.c (simd_clone_struct_copy): If from->inbranch
5734         is set, copy one less argument.
5735         (expand_simd_clones): Don't subtract clone_info->inbranch
5736         from simd_clone_struct_alloc argument.
5738         PR rtl-optimization/57915
5739         * recog.c (simplify_while_replacing): If all unary/binary/relational
5740         operation arguments are constant, attempt to simplify those.
5742         PR middle-end/59261
5743         * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
5744         if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
5746 2014-02-04  Richard Biener  <rguenther@suse.de>
5748         PR tree-optimization/60012
5749         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
5750         TBAA disambiguation to all DDRs.
5752 2014-02-04  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5754         PR target/59788
5755         * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
5756         (LINK_SPEC): Use it for -shared, -shared-libgcc.
5758 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5760         PR ipa/59882
5761         * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
5763 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5765         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
5766         * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
5768 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5770         PR ipa/59831
5771         * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
5772         to figure out targets of polymorphic calls with known decl.
5773         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
5774         * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
5775         * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
5776         (get_polymorphic_call_info): ... here.
5777         (get_polymorphic_call_info_from_invariant): New function.
5779 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5781         * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
5782         lookup via vtable pointer; check for type consistency
5783         and turn inconsitent facts into UNREACHABLE.
5784         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
5785         * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
5786         type inconsistent querries; return UNREACHABLE instead.
5788 2014-02-03  Richard Henderson  <rth@twiddle.net>
5790         PR tree-opt/59924
5791         * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
5792         already processed this node.
5793         (normalize_one_pred_1): Pass along mark_set.
5794         (normalize_one_pred): Create and destroy a pointer_set_t.
5795         (normalize_one_pred_chain): Likewise.
5797 2014-02-03  Laurent Aflonsi  <laurent.alfonsi@st.com>
5799         PR gcov-profile/58602
5800         * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
5802 2014-02-03  Jan Hubicka  <hubicka@ucw.cz>
5804         PR ipa/59831
5805         * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
5806         -fno-devirtualize; try to devirtualize by the knowledge of
5807         virtual table pointer given by aggregate propagation.
5808         * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
5809         (ipa_print_node_jump_functions): Dump also offset that
5810         is relevant for polymorphic calls.
5811         (determine_known_aggregate_parts): Add arg_type parameter; use it
5812         instead of determining the type from pointer type.
5813         (ipa_compute_jump_functions_for_edge): Update call of
5814         determine_known_aggregate_parts.
5815         * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
5816         (gimple_get_virt_method_for_binfo): ... here; simplify using
5817         vtable_pointer_value_to_vtable.
5818         * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
5819         * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
5820         to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
5821         (vtable_pointer_value_to_vtable): Break out from ...; handle also
5822         POINTER_PLUS_EXPR.
5823         (vtable_pointer_value_to_binfo): ... here.
5824         * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
5826 2014-02-03  Teresa Johnson  <tejohnson@google.com>
5828         * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
5829         redef of outer loop index variable.
5831 2014-02-03  Marc Glisse  <marc.glisse@inria.fr>
5833         PR c++/53017
5834         PR c++/59211
5835         * doc/extend.texi (Function Attributes): Typo.
5837 2014-02-03  Cong Hou  <congh@google.com>
5839         PR tree-optimization/60000
5840         * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
5841         if the vectorized statement is a store.  A store statement can only
5842         appear at the end of pattern statements.
5844 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
5846         * config/i386/i386.c (flag_opts): Add -mlong-double-128.
5847         (ix86_option_override_internal): Default long double to 64-bit for
5848         32-bit Bionic and to 128-bit for 64-bit Bionic.
5850         * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
5851         TARGET_LONG_DOUBLE_128 is true.
5852         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
5854         * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
5855         (mlong-double-64): Negate -mlong-double-128.
5856         (mlong-double-128): New option.
5858         * config/i386/i386-c.c (ix86_target_macros): Define
5859         __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
5861         * doc/invoke.texi: Document -mlong-double-128.
5863 2014-02-03  H.J. Lu  <hongjiu.lu@intel.com>
5865         PR rtl-optimization/60024
5866         * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
5868 2014-02-03  Markus Trippelsdorf  <markus@trippelsdorf.de>
5870         * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
5872 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
5874         PR rtl-optimization/57662
5875         * sel-sched.c (code_motion_path_driver): Do not mark already not
5876         existing blocks in the visiting bitmap.
5878 2014-02-03  Andrey Belevantsev  <abel@ispras.ru>
5880         * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
5881         on the insn being emitted.
5883 2014-02-03  James Greenhalgh  <james.greenhalgh@arm.com>
5884             Will Deacon  <will.deacon@arm.com>
5886         * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
5888 2014-02-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5890         * config/arm/arm-tables.opt: Regenerate.
5892 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5894         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
5895         for vector types other than V16QImode.
5896         * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
5897         define_expand, and call altivec_expand_vec_perm_le when producing
5898         code with little endian element order.
5899         (*altivec_vperm_<mode>_internal): New insn having previous
5900         behavior of altivec_vperm_<mode>.
5901         (altivec_vperm_<mode>_uns): Change to a define_expand, and call
5902         altivec_expand_vec_perm_le when producing code with little endian
5903         element order.
5904         (*altivec_vperm_<mode>_uns_internal): New insn having previous
5905         behavior of altivec_vperm_<mode>_uns.
5907 2014-02-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5909         * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
5910         (altivec_vsumsws): Add handling for -maltivec=be with a little
5911         endian target.
5912         (altivec_vsumsws_direct): New.
5913         (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
5914         gen_altivec_vsumsws.
5916 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
5918         * ipa-devirt.c (subbinfo_with_vtable_at_offset,
5919         vtable_pointer_value_to_binfo): New functions.
5920         * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
5921         * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
5923 2014-02-02  Sandra Loosemore  <sandra@codesourcery.com>
5925         * config/nios2/nios2.md (load_got_register): Initialize GOT
5926         pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
5927         * config/nios2/nios2.c (nios2_function_profiler): Likewise.
5929 2014-02-02  Jan Hubicka  <hubicka@ucw.cz>
5931         * ipa-prop.c (update_jump_functions_after_inlining): When type is not
5932         preserverd by passthrough, do not propagate the type.
5934 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
5936         * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
5937         (mips_atomic_assign_expand_fenv): New function.
5938         (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
5940 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
5942         * doc/extend.texi (__builtin_mips_get_fcsr): Document.
5943         (__builtin_mips_set_fcsr): Likewise.
5944         * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
5945         MIPS_USI_FTYPE_VOID.
5946         * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
5947         (mips16_expand_set_fcsr): Likewise.
5948         * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
5949         (mips16_set_fcsr_stub): Likewise.
5950         (mips16_get_fcsr_one_only_stub): New class.
5951         (mips16_set_fcsr_one_only_stub): Likewise.
5952         (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
5953         (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
5954         (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
5955         (hard_float): New availability predicate.
5956         (mips_builtins): Add get_fcsr and set_fcsr.
5957         (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
5958         * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
5959         (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
5960         (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
5961         (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
5962         patterns.
5964 2014-02-02  Richard Sandiford  <rdsandiford@googlemail.com>
5966         * config/mips/mips.c (mips_one_only_stub): New class.
5967         (mips_need_mips16_rdhwr_p): Replace with...
5968         (mips16_rdhwr_stub): ...this new variable.
5969         (mips16_stub_call_address): New function.
5970         (mips16_rdhwr_one_only_stub): New class.
5971         (mips_expand_thread_pointer): Use mips16_stub_call_address.
5972         (mips_output_mips16_rdhwr): Delete.
5973         (mips_finish_stub): New function.
5974         (mips_code_end): Use it to handle rdhwr stubs.
5976 2014-02-02  Uros Bizjak  <ubizjak@gmail.com>
5978         PR target/60017
5979         * config/i386/i386.c (classify_argument): Fix handling of bit_offset
5980         when calculating size of integer atomic types.
5982 2014-02-02  H.J. Lu  <hongjiu.lu@intel.com>
5984         * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
5986 2014-02-01  Jakub Jelinek  <jakub@redhat.com>
5988         PR tree-optimization/60003
5989         * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
5990         * profile.c (branch_prob): Use gimple_call_builtin_p
5991         to check for BUILT_IN_SETJMP_RECEIVER.
5992         * tree-inline.c (copy_bb): Call notice_special_calls.
5994 2014-01-31  Vladimir Makarov  <vmakarov@redhat.com>
5996         PR bootstrap/59985
5997         * lra-constraints.c (process_alt_operands): Update reload_sum only
5998         on the first pass.
6000 2014-01-31  Richard Henderson  <rth@redhat.com>
6002         PR middle-end/60004
6003         * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
6004         until after else_eh is processed.
6006 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
6008         * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
6009         (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
6010         (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
6011         in smmintrin.h, remove them.
6012         (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
6013         * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
6014         * config/i386/i386.md (ROUND_SAE): Fix value.
6015         * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
6016         (const48_operand): New.
6017         * config/i386/subst.md (round), (round_expand): Use
6018         const_4_or_8_to_11_operand.
6019         (round_saeonly), (round_saeonly_expand): Use const48_operand.
6021 2014-01-31  Ilya Tocar  <ilya.tocar@intel.com>
6023         * config/i386/constraints.md (Yk): Swap meaning with k.
6024         * config/i386/i386.md (movhi_internal): Change Yk to k.
6025         (movqi_internal): Ditto.
6026         (*k<logic><mode>): Ditto.
6027         (*andhi_1): Ditto.
6028         (*andqi_1): Ditto.
6029         (kandn<mode>): Ditto.
6030         (*<code>hi_1): Ditto.
6031         (*<code>qi_1): Ditto.
6032         (kxnor<mode>): Ditto.
6033         (kortestzhi): Ditto.
6034         (kortestchi): Ditto.
6035         (kunpckhi): Ditto.
6036         (*one_cmplhi2_1): Ditto.
6037         (*one_cmplqi2_1): Ditto.
6038         * config/i386/sse.md (): Change k to Yk.
6039         (avx512f_load<mode>_mask): Ditto.
6040         (avx512f_blendm<mode>): Ditto.
6041         (avx512f_store<mode>_mask): Ditto.
6042         (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
6043         (avx512f_storedqu<mode>_mask): Ditto.
6044         (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
6045         Ditto.
6046         (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
6047         (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
6048         (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
6049         (avx512f_maskcmp<mode>3): Ditto.
6050         (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
6051         (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
6052         (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
6053         (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
6054         (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
6055         (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
6056         (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
6057         (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
6058         (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
6059         (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
6060         (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
6061         (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
6062         (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
6063         (vec_extract_lo_<mode>_maskm): Ditto.
6064         (vec_extract_hi_<mode>_maskm): Ditto.
6065         (avx512f_vternlog<mode>_mask): Ditto.
6066         (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
6067         (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
6068         (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
6069         (avx512f_<code>v8div16qi2_mask): Ditto.
6070         (avx512f_<code>v8div16qi2_mask_store): Ditto.
6071         (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
6072         (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
6073         (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
6074         (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
6075         (*avx512pf_gatherpf<mode>sf_mask): Ditto.
6076         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6077         (*avx512pf_scatterpf<mode>sf_mask): Ditto.
6078         (*avx512pf_scatterpf<mode>df_mask): Ditto.
6079         (avx512cd_maskb_vec_dupv8di): Ditto.
6080         (avx512cd_maskw_vec_dupv16si): Ditto.
6081         (avx512f_vpermi2var<mode>3_maskz): Ditto.
6082         (avx512f_vpermi2var<mode>3_mask): Ditto.
6083         (avx512f_vpermi2var<mode>3_mask): Ditto.
6084         (avx512f_vpermt2var<mode>3_maskz): Ditto.
6085         (*avx512f_gathersi<mode>): Ditto.
6086         (*avx512f_gathersi<mode>_2): Ditto.
6087         (*avx512f_gatherdi<mode>): Ditto.
6088         (*avx512f_gatherdi<mode>_2): Ditto.
6089         (*avx512f_scattersi<mode>): Ditto.
6090         (*avx512f_scatterdi<mode>): Ditto.
6091         (avx512f_compress<mode>_mask): Ditto.
6092         (avx512f_compressstore<mode>_mask): Ditto.
6093         (avx512f_expand<mode>_mask): Ditto.
6094         * config/i386/subst.md (mask): Change k to Yk.
6095         (mask_scalar_merge): Ditto.
6096         (sd): Ditto.
6098 2014-01-31  Marc Glisse  <marc.glisse@inria.fr>
6100         * doc/extend.texi (Vector Extensions): Document ?: in C++.
6102 2014-01-31  Richard Biener  <rguenther@suse.de>
6104         PR middle-end/59990
6105         * builtins.c (fold_builtin_memory_op): Make sure to not
6106         use a floating-point mode or a boolean or enumeral type for
6107         the copy operation.
6109 2014-01-30  DJ Delorie  <dj@redhat.com>
6111         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
6112         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
6113         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
6114         whenever main() has an epilogue.
6116 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6118         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
6119         unused variable "field".
6120         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
6121         (vsx_mergeh_<mode>): Likewise.
6122         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
6123         (altivec_vmrghh): Likewise.
6124         (altivec_vmrghw): Likewise.
6125         (altivec_vmrglb): Likewise.
6126         (altivec_vmrglh): Likewise.
6127         (altivec_vmrglw): Likewise.
6128         (altivec_vspltb): Add missing uses.
6129         (altivec_vsplth): Likewise.
6130         (altivec_vspltw): Likewise.
6131         (altivec_vspltsf): Likewise.
6133 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
6135         PR target/59923
6136         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
6137         frame related instructions.
6139 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
6141         PR rtl-optimization/59959
6142         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
6143         any reload of register whose subreg is invalid.
6145 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
6147         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
6148         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
6149         Add missing return type - void.
6151 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6153         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
6154         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
6155         remove element index adjustment for endian (now handled in vsx.md
6156         and altivec.md).
6157         (altivec_expand_vec_perm_const): Use
6158         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
6159         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
6160         (vsx_xxspltw_<mode>): Adjust element index for little endian.
6161         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
6162         define_expand and a new define_insn *altivec_vspltb_internal;
6163         adjust for -maltivec=be on a little endian target.
6164         (altivec_vspltb_direct): New.
6165         (altivec_vsplth): Divide into a define_expand and a new
6166         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
6167         little endian target.
6168         (altivec_vsplth_direct): New.
6169         (altivec_vspltw): Divide into a define_expand and a new
6170         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
6171         little endian target.
6172         (altivec_vspltw_direct): New.
6173         (altivec_vspltsf): Divide into a define_expand and a new
6174         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
6175         a little endian target.
6177 2014-01-30  Richard Biener  <rguenther@suse.de>
6179         PR tree-optimization/59993
6180         * tree-ssa-forwprop.c (associate_pointerplus): Check we
6181         can propagate form the earlier stmt and avoid the transform
6182         when the intermediate result is needed.
6184 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
6186         * README.Portability: Fix typo.
6188 2014-01-30  David Holsgrove  <david.holsgrove@xilinx.com>
6190         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
6191         comparison_operator with ordered_comparison_operator.
6193 2014-01-30  Nick Clifton  <nickc@redhat.com>
6195         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
6196         Rename to mn10300_store_multiple_regs.
6197         * config/mn10300/mn10300.c: Likewise.
6198         * config/mn10300/mn10300.md (store_movm): Fix typo: call
6199         store_multiple_regs.
6200         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
6201         Call mn10300_store_multiple_regs.
6203 2014-01-30  Nick Clifton  <nickc@redhat.com>
6204             DJ Delorie  <dj@redhat.com>
6206         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
6207         %fp 2 to keep registers after it properly word-aligned.
6208         (rl78_alloc_physical_registers_umul): Handle the case where both
6209         input operands are the same.
6211 2014-01-30  Richard Biener  <rguenther@suse.de>
6213         PR tree-optimization/59903
6214         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
6215         check properly.
6217 2014-01-30  Jason Merrill  <jason@redhat.com>
6219         PR c++/59633
6220         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
6222         PR c++/59645
6223         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
6225 2014-01-30  Richard Biener  <rguenther@suse.de>
6227         PR tree-optimization/59951
6228         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
6230 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
6232         PR target/59784
6233         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
6234         SFmode to DFmode case.
6236 2014-01-29  DJ Delorie  <dj@redhat.com>
6238         * config/msp430/msp430.opt (-minrt): New.
6239         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
6240         if -minrt given.
6241         (ENDFILE_SPEC): Likewise.
6243 2014-01-29  Jan Hubicka  <hubicka@ucw.cz>
6245         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
6246         (estimate_function_body_sizes): Use it.
6248 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
6250         PR c++/58561
6251         * dwarf2out.c (is_cxx_auto): New.
6252         (is_base_type): Use it.
6253         (gen_type_die_with_usage): Likewise.
6255 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6257         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
6258         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
6259         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
6260         -maltivec=be with LE targets.
6261         (vsx_mergeh_<mode>): Likewise.
6262         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
6263         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
6264         (altivec_vmrghb): Replace with define_expand and new
6265         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
6266         (altivec_vmrghb_direct): New define_insn.
6267         (altivec_vmrghh): Replace with define_expand and new
6268         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
6269         (altivec_vmrghh_direct): New define_insn.
6270         (altivec_vmrghw): Replace with define_expand and new
6271         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
6272         (altivec_vmrghw_direct): New define_insn.
6273         (*altivec_vmrghsf): Adjust for endianness.
6274         (altivec_vmrglb): Replace with define_expand and new
6275         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
6276         (altivec_vmrglb_direct): New define_insn.
6277         (altivec_vmrglh): Replace with define_expand and new
6278         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
6279         (altivec_vmrglh_direct): New define_insn.
6280         (altivec_vmrglw): Replace with define_expand and new
6281         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
6282         (altivec_vmrglw_direct): New define_insn.
6283         (*altivec_vmrglsf): Adjust for endianness.
6284         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
6285         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
6286         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
6287         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
6288         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
6289         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
6290         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
6291         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
6293 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
6295         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
6296         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
6297         whitespace.
6299 2014-01-29  Richard Biener  <rguenther@suse.de>
6301         PR tree-optimization/58742
6302         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
6303         associate_pointerplus_align.
6304         (associate_pointerplus_diff): New function.
6305         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
6306         and associate_pointerplus_diff.
6308 2014-01-29  Richard Biener  <rguenther@suse.de>
6310         * lto-streamer.h (LTO_major_version): Bump to 3.
6311         (LTO_minor_version): Reset to 0.
6313 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
6315         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
6316         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
6317         (arm_file_start): Generate correct asm header for armv7ve.
6318         * config/arm/bpabi.h: Add multilib support for armv7ve.
6319         * config/arm/driver-arm.c: Change the architectures of cortex-a7
6320         and cortex-a15 to armv7ve.
6321         * config/arm/t-aprofile: Add multilib support for armv7ve.
6322         * doc/invoke.texi: Document -march=armv7ve.
6324 2014-01-29  Richard Biener  <rguenther@suse.de>
6326         PR tree-optimization/58742
6327         * tree-ssa-forwprop.c (associate_plusminus): Return true
6328         if we changed sth, defer EH cleanup to ...
6329         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
6330         (simplify_mult): New function.
6332 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
6334         PR middle-end/59917
6335         PR tree-optimization/59920
6336         * tree.c (build_common_builtin_nodes): Remove
6337         __builtin_setjmp_dispatcher initialization.
6338         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
6339         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
6340         instead of gsi_after_labels + manually skipping debug stmts.
6341         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
6342         ignore bbs with IFN_ABNORMAL_DISPATCHER.
6343         * tree-inline.c (copy_edges_for_bb): Remove
6344         can_make_abnormal_goto argument, instead add abnormal_goto_dest
6345         argument.  Ignore computed_goto_p stmts.  Don't call
6346         make_abnormal_goto_edges.  If a call might need abnormal edges
6347         for non-local gotos, see if it already has an edge to
6348         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
6349         with true argument, don't do anything then, otherwise add
6350         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
6351         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
6352         caller.
6353         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
6354         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
6355         (lower_stmt): Don't set data->calls_builtin_setjmp.
6356         (lower_builtin_setjmp): Adjust comment.
6357         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
6358         * tree-cfg.c (found_computed_goto): Remove.
6359         (factor_computed_gotos): Remove.
6360         (make_goto_expr_edges): Return bool, true for computed gotos.
6361         Don't call make_abnormal_goto_edges.
6362         (build_gimple_cfg): Don't set found_computed_goto, don't call
6363         factor_computed_gotos.
6364         (computed_goto_p): No longer static.
6365         (make_blocks): Don't set found_computed_goto.
6366         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
6367         (make_edges): If make_goto_expr_edges returns true, push bb
6368         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
6369         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
6370         vector.  Record mapping between bbs and OpenMP regions if there
6371         are any, adjust make_gimple_omp_edges caller.  Call
6372         handle_abnormal_edges.
6373         (make_abnormal_goto_edges): Remove.
6374         * tree-cfg.h (make_abnormal_goto_edges): Remove.
6375         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
6376         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
6377         * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
6378         * internal-fn.def (ABNORMAL_DISPATCHER): New.
6379         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
6380         filling *region also set *region_idx to (*region)->entry->index.
6382         PR other/58712
6383         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
6384         For REGs set ORIGINAL_REGNO.
6386 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
6388         * doc/md.texi: Mention that a target shouldn't implement
6389         vec_widen_(s|u)mul_even/odd pair if it is less efficient
6390         than hi/lo pair.
6392 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
6394         PR tree-optimization/59594
6395         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
6396         a copy of the datarefs vector rather than the vector itself.
6398 2014-01-28  Jason Merrill  <jason@redhat.com>
6400         PR c++/53756
6401         * dwarf2out.c (auto_die): New static.
6402         (gen_type_die_with_usage): Handle C++1y 'auto'.
6403         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
6404         on definition.
6406 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
6408         PR target/59672
6409         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
6410         (SPEC_X32): Likewise.
6411         (SPEC_64): Likewise.
6412         * config/i386/i386.c (ix86_option_override_internal): Turn off
6413         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
6414         for TARGET_16BIT.
6415         (x86_file_start): Output .code16gcc for TARGET_16BIT.
6416         * config/i386/i386.h (TARGET_16BIT): New macro.
6417         (TARGET_16BIT_P): Likewise.
6418         * config/i386/i386.opt: Add m16.
6419         * doc/invoke.texi: Document -m16.
6421 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
6423         PR preprocessor/59935
6424         * input.c (location_get_source_line): Bail out on when line number
6425         is zero, and test the return value of lookup_or_add_file_to_cache_tab.
6427 2014-01-28  Richard Biener  <rguenther@suse.de>
6429         PR tree-optimization/58742
6430         * tree-ssa-forwprop.c (associate_plusminus): Handle
6431         pointer subtraction of the form (T)(P + A) - (T)P.
6433 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6435         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
6436         at const_int_cost.
6438 2014-01-28  Richard Biener  <rguenther@suse.de>
6440         Revert
6441         2014-01-28  Richard Biener  <rguenther@suse.de>
6443         PR rtl-optimization/45364
6444         PR rtl-optimization/59890
6445         * var-tracking.c (local_get_addr_clear_given_value): Handle
6446         already cleared slot.
6447         (val_reset): Handle not allocated local_get_addr_cache.
6448         (vt_find_locations): Use post-order on the inverted CFG.
6450 2014-01-28  Richard Biener  <rguenther@suse.de>
6452         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
6454 2014-01-28  Richard Biener  <rguenther@suse.de>
6456         PR rtl-optimization/45364
6457         PR rtl-optimization/59890
6458         * var-tracking.c (local_get_addr_clear_given_value): Handle
6459         already cleared slot.
6460         (val_reset): Handle not allocated local_get_addr_cache.
6461         (vt_find_locations): Use post-order on the inverted CFG.
6463 2014-01-28  Alan Modra  <amodra@gmail.com>
6465         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
6466         * configure.ac <recursive call for build != host>: Define
6467         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
6468         and LD_FOR_BUILD too.
6469         * configure: Regenerate.
6471 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
6473         * config/i386/i386.c (get_builtin_code_for_version): Separate
6474         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
6475         Broadwell from Haswell.
6477 2014-01-27  Steve Ellcey  <sellcey@mips.com>
6479         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
6480         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
6481         * config/mips/mips.c (mips_option_override): Change setting
6482         of TARGET_DSP.
6483         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
6484         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
6485         Change from Mask to Var.
6487 2014-01-27  Jeff Law  <law@redhat.com>
6489         * ipa-inline.c (inline_small_functions): Fix typo.
6491 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
6493         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
6494         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
6495         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
6496         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
6497         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
6498         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
6499         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
6500         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
6501         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
6502         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
6503         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
6504         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
6505         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
6506         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
6507         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
6508         (_mm512_storeu_epi64): Ditto.
6509         (_mm512_cmpge_epi32_mask): Ditto.
6510         (_mm512_cmpge_epu32_mask): Ditto.
6511         (_mm512_cmpge_epi64_mask): Ditto.
6512         (_mm512_cmpge_epu64_mask): Ditto.
6513         (_mm512_cmple_epi32_mask): Ditto.
6514         (_mm512_cmple_epu32_mask): Ditto.
6515         (_mm512_cmple_epi64_mask): Ditto.
6516         (_mm512_cmple_epu64_mask): Ditto.
6517         (_mm512_cmplt_epi32_mask): Ditto.
6518         (_mm512_cmplt_epu32_mask): Ditto.
6519         (_mm512_cmplt_epi64_mask): Ditto.
6520         (_mm512_cmplt_epu64_mask): Ditto.
6521         (_mm512_cmpneq_epi32_mask): Ditto.
6522         (_mm512_cmpneq_epu32_mask): Ditto.
6523         (_mm512_cmpneq_epi64_mask): Ditto.
6524         (_mm512_cmpneq_epu64_mask): Ditto.
6525         (_mm512_expand_pd): Ditto.
6526         (_mm512_expand_ps): Ditto.
6527         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
6528         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
6529         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
6530         * config/i386/i386.c (ix86_builtins): Add
6531         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
6532         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
6533         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
6534         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
6535         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
6536         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
6537         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
6538         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
6539         IX86_BUILTIN_PMOVUSQW512_MEM.
6540         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
6541         __builtin_ia32_pmovsqd512mem_mask,
6542         __builtin_ia32_pmovqd512mem_mask,
6543         __builtin_ia32_pmovusqw512mem_mask,
6544         __builtin_ia32_pmovsqw512mem_mask,
6545         __builtin_ia32_pmovqw512mem_mask,
6546         __builtin_ia32_pmovusdw512mem_mask,
6547         __builtin_ia32_pmovsdw512mem_mask,
6548         __builtin_ia32_pmovdw512mem_mask,
6549         __builtin_ia32_pmovqb512mem_mask,
6550         __builtin_ia32_pmovusqb512mem_mask,
6551         __builtin_ia32_pmovsqb512mem_mask,
6552         __builtin_ia32_pmovusdb512mem_mask,
6553         __builtin_ia32_pmovsdb512mem_mask,
6554         __builtin_ia32_pmovdb512mem_mask.
6555         (bdesc_args): Add __builtin_ia32_expanddf512,
6556         __builtin_ia32_expandsf512.
6557         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
6558         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
6559         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
6560         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
6561         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
6562         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
6563         (avx512f_<code>v8div16qi2_mask_store): This.
6564         (avx512f_expand<mode>): New.
6566 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
6568         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
6569         New.
6570         (_mm512_mask_prefetch_i64gather_pd): Ditto.
6571         (_mm512_prefetch_i32scatter_pd): Ditto.
6572         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
6573         (_mm512_prefetch_i64scatter_pd): Ditto.
6574         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
6575         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
6576         (_mm512_mask_prefetch_i64gather_ps): Ditto.
6577         (_mm512_prefetch_i32scatter_ps): Ditto.
6578         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
6579         (_mm512_prefetch_i64scatter_ps): Ditto.
6580         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
6581         * config/i386/i386-builtin-types.def: Define
6582         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
6583         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
6584         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
6585         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
6586         IX86_BUILTIN_SCATTERPFQPD.
6587         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
6588         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
6589         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
6590         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
6591         __builtin_ia32_scatterpfqps.
6592         (ix86_expand_builtin): Expand new built-ins.
6593         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
6594         fix memory access data type.
6595         (*avx512pf_gatherpf<mode>_mask): Ditto.
6596         (*avx512pf_gatherpf<mode>): Ditto.
6597         (avx512pf_scatterpf<mode>): Ditto.
6598         (*avx512pf_scatterpf<mode>_mask): Ditto.
6599         (*avx512pf_scatterpf<mode>): Ditto.
6600         (GATHER_SCATTER_SF_MEM_MODE): New.
6601         (avx512pf_gatherpf<mode>df): Ditto.
6602         (*avx512pf_gatherpf<mode>df_mask): Ditto.
6603         (*avx512pf_scatterpf<mode>df): Ditto.
6605 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
6607         PR bootstrap/59934
6608         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
6609         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
6610         reached.
6612 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
6614         * common/config/arm/arm-common.c
6615         (arm_rewrite_mcpu): Handle multiple names.
6616         * config/arm/arm.h
6617         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
6619 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
6621         * gimple-builder.h (create_gimple_tmp): Delete.
6623 2014-01-27  Christian Bruel  <christian.bruel@st.com>
6625         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
6626         words comparisons.
6628 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
6630         * config/pa/pa.md (call): Generate indirect long calls to non-local
6631         functions when outputing 32-bit code.
6632         (call_value): Likewise except for special call to buggy powf function.
6634         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
6635         portable runtime and PIC indirect calls.
6636         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
6637         and PIC call sequences.  Use ldo instead of blr to set return register
6638         in PIC call sequence.
6640 2014-01-25  Walter Lee  <walt@tilera.com>
6642         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
6643         avoid clobbering a live register.
6645 2014-01-25  Walter Lee  <walt@tilera.com>
6647         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
6648         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
6649         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
6650         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
6652 2014-01-25  Walter Lee  <walt@tilera.com>
6654         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
6655         arguments on even registers.
6656         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
6657         STACK_BOUNDARY.
6658         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
6659         (BIGGEST_ALIGNMENT): Ditto.
6660         (BIGGEST_FIELD_ALIGNMENT): Ditto.
6662 2014-01-25  Walter Lee  <walt@tilera.com>
6664         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
6665         insns before bundling.
6666         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
6668 2014-01-25  Walter Lee  <walt@tilera.com>
6670         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
6671         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
6672         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
6674 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
6676         * config/mips/constraints.md (kl): Delete.
6677         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
6678         define expands, using...
6679         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
6680         instructions for MIPS16.
6681         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
6682         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
6684 2014-01-25  Walter Lee  <walt@tilera.com>
6686         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
6687         (clzdi2): Ditto.
6688         (ffsdi2): Ditto.
6690 2014-01-25  Walter Lee  <walt@tilera.com>
6692         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
6693         (TARGET_EXPAND_TO_RTL_HOOK): Define.
6695 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
6697         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
6698         Handle XOR.
6700 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
6702         * print-rtl.c (in_call_function_usage): New var.
6703         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
6704         EXPR_LIST mode as mode and not as reg note name.
6706         PR middle-end/59561
6707         * cfgloopmanip.c (copy_loop_info): If
6708         loop->warned_aggressive_loop_optimizations, make sure
6709         the flag is set in target loop too.
6711 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
6713         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
6714         flag_cilkplus.
6715         * builtins.def: Likewise.
6716         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
6717         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
6718         * ira.c (ira_setup_eliminable_regset): Likewise.
6719         * omp-low.c (gate_expand_omp): Likewise.
6720         (execute_lower_omp): Likewise.
6721         (diagnose_sb_0): Likewise.
6722         (gate_diagnose_omp_blocks): Likewise.
6723         (simd_clone_clauses_extract): Likewise.
6724         (gate): Likewise.
6726 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6728         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
6729         correction for little endian...
6730         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
6731         here.
6733 2014-01-24  Jeff Law  <law@redhat.com>
6735         PR tree-optimization/59919
6736         * tree-vrp.c (find_assert_locations_1): Do not register asserts
6737         for non-returning calls.
6739 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
6741         * common/config/aarch64/aarch64-common.c
6742         (aarch64_rewrite_mcpu): Handle multiple names.
6743         * config/aarch64/aarch64.h
6744         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
6746 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
6748         * input.c (add_file_to_cache_tab): Handle the case where fopen
6749         returns NULL.
6751 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
6753         PR target/59929
6754         * config/i386/i386.md (pushsf splitter): Get stack adjustment
6755         from push operand if code of push isn't PRE_DEC.
6757 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
6759         PR target/59909
6760         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
6761         -mquad-memory-atomic.  Update -mquad-memory documentation to say
6762         it is only used for non-atomic loads/stores.
6764         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
6765         -mquad-memory or -mquad-memory-atomic switches.
6767         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
6768         -mquad-memory-atomic to ISA 2.07 support.
6770         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
6771         to separate support of normal quad word memory operations (ldq, stq)
6772         from the atomic quad word memory operations.
6774         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
6775         support to separate non-atomic quad word operations from atomic
6776         quad word operations.  Disable non-atomic quad word operations in
6777         little endian mode so that we don't have to swap words after the
6778         load and before the store.
6779         (quad_load_store_p): Add comment about atomic quad word support.
6780         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
6781         options printed with -mdebug=reg.
6783         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
6784         -mquad-memory-atomic as the test for whether we have quad word
6785         atomic instructions.
6786         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
6787         or -mp8-vector are used, allow byte/half-word atomic operations.
6789         * config/rs6000/sync.md (load_lockedti): Insure that the address
6790         is a proper indexed or indirect address for the lqarx instruction.
6791         On little endian systems, swap the hi/lo registers after the lqarx
6792         instruction.
6793         (load_lockedpti): Use indexed_or_indirect_operand predicate to
6794         insure the address is valid for the lqarx instruction.
6795         (store_conditionalti): Insure that the address is a proper indexed
6796         or indirect address for the stqcrx. instruction.  On little endian
6797         systems, swap the hi/lo registers before doing the stqcrx.
6798         instruction.
6799         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
6800         insure the address is valid for the stqcrx. instruction.
6802         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
6803         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
6804         type of quad memory support is available.
6806 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
6808         PR regression/59915
6809         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
6810         there is a danger of looping.
6812 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
6814         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6815         force flag_ira_loop_pressure if set via command line.
6817 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
6819         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
6820         (ashr_simd): New builtin handling DI mode.
6821         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
6822         (aarch64_sshr_simddi): New match pattern.
6823         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
6824         (vshrd_n_s64): Likewise.
6825         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
6827 2014-01-23  Nick Clifton  <nickc@redhat.com>
6829         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
6830         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
6831         favour of mcu specific scripts.
6832         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
6833         430x multilibs.
6835 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
6836             Alex Velenko  <Alex.Velenko@arm.com>
6838         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
6839         (vaddv_s16): Likewise.
6840         (vaddv_s32): Likewise.
6841         (vaddv_u8): Likewise.
6842         (vaddv_u16): Likewise.
6843         (vaddv_u32): Likewise.
6844         (vaddvq_s8): Likewise.
6845         (vaddvq_s16): Likewise.
6846         (vaddvq_s32): Likewise.
6847         (vaddvq_s64): Likewise.
6848         (vaddvq_u8): Likewise.
6849         (vaddvq_u16): Likewise.
6850         (vaddvq_u32): Likewise.
6851         (vaddvq_u64): Likewise.
6852         (vaddv_f32): Likewise.
6853         (vaddvq_f32): Likewise.
6854         (vaddvq_f64): Likewise.
6855         (vmaxv_f32): Likewise.
6856         (vmaxv_s8): Likewise.
6857         (vmaxv_s16): Likewise.
6858         (vmaxv_s32): Likewise.
6859         (vmaxv_u8): Likewise.
6860         (vmaxv_u16): Likewise.
6861         (vmaxv_u32): Likewise.
6862         (vmaxvq_f32): Likewise.
6863         (vmaxvq_f64): Likewise.
6864         (vmaxvq_s8): Likewise.
6865         (vmaxvq_s16): Likewise.
6866         (vmaxvq_s32): Likewise.
6867         (vmaxvq_u8): Likewise.
6868         (vmaxvq_u16): Likewise.
6869         (vmaxvq_u32): Likewise.
6870         (vmaxnmv_f32): Likewise.
6871         (vmaxnmvq_f32): Likewise.
6872         (vmaxnmvq_f64): Likewise.
6873         (vminv_f32): Likewise.
6874         (vminv_s8): Likewise.
6875         (vminv_s16): Likewise.
6876         (vminv_s32): Likewise.
6877         (vminv_u8): Likewise.
6878         (vminv_u16): Likewise.
6879         (vminv_u32): Likewise.
6880         (vminvq_f32): Likewise.
6881         (vminvq_f64): Likewise.
6882         (vminvq_s8): Likewise.
6883         (vminvq_s16): Likewise.
6884         (vminvq_s32): Likewise.
6885         (vminvq_u8): Likewise.
6886         (vminvq_u16): Likewise.
6887         (vminvq_u32): Likewise.
6888         (vminnmv_f32): Likewise.
6889         (vminnmvq_f32): Likewise.
6890         (vminnmvq_f64): Likewise.
6892 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
6894         * config/aarch64/aarch64-simd.md
6895         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
6896         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
6897         (*aarch64_mul3_elt<mode>): Likewise.
6898         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
6899         (*aarch64_mul3_elt_to_64v2df): Likewise.
6900         (*aarch64_mla_elt<mode>): Likewise.
6901         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
6902         (*aarch64_mls_elt<mode>): Likewise.
6903         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
6904         (*aarch64_fma4_elt<mode>): Likewise.
6905         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
6906         (*aarch64_fma4_elt_to_64v2df): Likewise.
6907         (*aarch64_fnma4_elt<mode>): Likewise.
6908         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6909         (*aarch64_fnma4_elt_to_64v2df): Likewise.
6910         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
6911         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
6912         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
6913         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
6914         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6915         (aarch64_sqdmull_lane<mode>_internal): Likewise.
6916         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6918 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
6920         * config/aarch64/aarch64-simd.md
6921         (aarch64_be_checked_get_lane<mode>): New define_expand.
6922         * config/aarch64/aarch64-simd-builtins.def
6923         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
6924         New builtin definition.
6925         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
6926         Use new safe be builtin.
6928 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
6930         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
6931         New define_insn.
6932         (aarch64_be_st1<mode>): Likewise.
6933         (aarch_ld1<VALL:mode>): Define_expand modified.
6934         (aarch_st1<VALL:mode>): Likewise.
6935         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
6936         (UNSPEC_ST1): Likewise.
6938 2014-01-23  David Holsgrove  <david.holsgrove@xilinx.com>
6940         * config/microblaze/microblaze.md: Add trap insn and attribute
6942 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
6944         PR preprocessor/58580
6945         * input.h (location_get_source_line): Take an additional line_size
6946         parameter.
6947         (void diagnostics_file_cache_fini): Declare new function.
6948         * input.c (struct fcache): New type.
6949         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
6950         New static constants.
6951         (diagnostic_file_cache_init, total_lines_num)
6952         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
6953         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
6954         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
6955         (get_next_line, read_next_line, goto_next_line, read_line_num):
6956         New static function definitions.
6957         (diagnostic_file_cache_fini): New function.
6958         (location_get_source_line): Take an additional output line_len
6959         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
6960         read_line_num.
6961         * diagnostic.c (diagnostic_finish): Call
6962         diagnostic_file_cache_fini.
6963         (adjust_line): Take an additional input parameter for the length
6964         of the line, rather than calculating it with strlen.
6965         (diagnostic_show_locus): Adjust the use of
6966         location_get_source_line and adjust_line with respect to their new
6967         signature.  While displaying a line now, do not stop at the first
6968         null byte.  Rather, display the zero byte as a space and keep
6969         going until we reach the size of the line.
6970         * Makefile.in: Add vec.o to OBJS-libcommon
6972 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
6973             Ilya Tocar     <ilya.tocar@intel.com>
6975         * config/i386/avx512fintrin.h (_mm512_kmov): New.
6976         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
6977         (__builtin_ia32_kmov16): Ditto.
6978         * config/i386/i386.md (UNSPEC_KMOV): New.
6979         (kmovw): Ditto.
6981 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
6983         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
6984         (_mm512_storeu_si512): Ditto.
6986 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
6988         PR target/52125
6989         * rtl.h (get_referenced_operands): Declare.
6990         * recog.c (get_referenced_operands): New function.
6991         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
6992         operands have been referenced when recording LO_SUM references.
6994 2014-01-22  David Holsgrove  <david.holsgrove@xilinx.com>
6996         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
6998 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
7000         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
7001         Enable for generic and recent AMD targets.
7003 2014-01-22  Jan Hubicka  <hubicka@ucw.cz>
7005         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
7006         ARG_SIZE note when adjustment was eliminated.
7008 2014-01-22  Jeff Law  <law@redhat.com>
7010         PR tree-optimization/59597
7011         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
7012         in file.  Accept new argument REGISTERING and use it to modify
7013         dump output appropriately.
7014         (register_jump_thread): Corresponding changes.
7015         (mark_threaded_blocks): Reinstate code to cancel unprofitable
7016         thread paths involving joiner blocks.  Add code to dump cancelled
7017         jump threading paths.
7019 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
7021         PR rtl-optimization/59477
7022         * lra-constraints.c (inherit_in_ebb): Process call for living hard
7023         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
7025 2014-01-22  Tom Tromey  <tromey@redhat.com>
7027         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
7028         PARAMS.
7029         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
7031 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
7033         PR rtl-optimization/59896
7034         * lra-constraints.c (process_alt_operands): Check unused note for
7035         matched operands of insn with no output reloads.
7037 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
7039         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
7040         (mips_move_from_gpr_cost): Likewise.
7042 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
7044         PR rtl-optimization/59858
7045         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
7046         ira_class_hard_regs_num.
7047         (process_alt_operands): Increase reject for dying matched operand.
7049 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
7051         PR target/59003
7052         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
7053         smaller than size, perform several stores or loads and stores
7054         at dst + count - size to store or copy all of size bytes, rather
7055         than just last modesize bytes.
7057 2014-01-20  DJ Delorie  <dj@redhat.com>
7059         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
7060         that CLOBBERs are REGs before propogating their values.
7062 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
7064         PR middle-end/59789
7065         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
7066         (cgraph_inline_failed_type): New function.
7067         * cgraph.h (DEFCIFCODE): Add type.
7068         (cgraph_inline_failed_type_t): New enum.
7069         (cgraph_inline_failed_type): New prototype.
7070         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
7071         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
7072         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
7073         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
7074         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
7075         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
7076         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
7077         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
7078         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
7079         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
7080         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
7081         OPTIMIZATION_MISMATCH.
7082         * tree-inline.c (expand_call_inline): Emit errors during
7083         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
7085 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
7087         PR target/59685
7088         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
7089         mode attribute in insn output.
7091 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
7093         * output.h (output_constant): Delete.
7094         * varasm.c (output_constant): Make private.
7096 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
7098         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
7100 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
7102         PR middle-end/59860
7103         * tree.h (fold_builtin_strcat): New prototype.
7104         * builtins.c (fold_builtin_strcat): No longer static.  Add len
7105         argument, if non-NULL, don't call c_strlen.  Optimize
7106         directly into __builtin_memcpy instead of __builtin_strcpy.
7107         (fold_builtin_2): Adjust fold_builtin_strcat caller.
7108         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
7110 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
7112         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
7113         for SImode_address_operand operands, having only a REG argument.
7115 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7117         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
7118         loader name using mbig-endian.
7119         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
7121 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
7123         * doc/invoke.texi (-march): Clarify documentation for AArch64.
7124         (-mtune): Likewise.
7125         (-mcpu): Likewise.
7127 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
7129         * config/aarch64/aarch64-protos.h
7130         (aarch64_cannot_change_mode_class_ptr): Declare.
7131         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
7132         aarch64_cannot_change_mode_class_ptr): New.
7133         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
7134         backend hook aarch64_cannot_change_mode_class.
7136 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
7138         * common/config/aarch64/aarch64-common.c
7139         (aarch64_handle_option): Don't handle any option order logic here.
7140         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
7141         selected_cpu, warn on architecture version mismatch.
7142         (aarch64_override_options): Fix parsing order for option strings.
7144 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7145             Iain Sandoe  <iain@codesourcery.com>
7147         PR bootstrap/59496
7148         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
7149         warning.  Amend comment to reflect current functionality.
7151 2014-01-20  Richard Biener  <rguenther@suse.de>
7153         PR middle-end/59860
7154         * builtins.c (fold_builtin_strcat): Remove case better handled
7155         by tree-ssa-strlen.c.
7157 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
7159         * config/aarch64/aarch64.opt
7160         (mcpu, march, mtune): Make case-insensitive.
7162 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
7164         PR target/59880
7165         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
7166         if operands[1] is a REG or ZERO_EXTEND of a REG.
7168 2014-01-19  Jan Hubicka  <hubicka@ucw.cz>
7170         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
7172 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
7174         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
7175         long non-pic millicode calls.
7177 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
7179         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
7181 2014-01-19  Kito Cheng  <kito@0xlab.org>
7183         * builtins.c (expand_movstr): Check movstr expand done or fail.
7185 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7186             H.J. Lu  <hongjiu.lu@intel.com>
7188         PR target/59379
7189         * config/i386/i386.md (*lea<mode>): Zero-extend return register
7190         to DImode for zero-extended addresses.
7192 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
7194         PR rtl-optimization/57763
7195         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
7196         on the new indirect jump_insn and increment LABEL_NUSES (label).
7198 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
7200         PR bootstrap/59580
7201         PR bootstrap/59583
7202         * config.gcc (x86_archs): New variable.
7203         (x86_64_archs): Likewise.
7204         (x86_cpus): Likewise.
7205         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
7206         --with-arch/--with-cpu= options.
7207         Support --with-arch=/--with-cpu={nehalem,westmere,
7208         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
7210 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7212         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
7213         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
7215 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
7217         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
7219 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
7221         PR target/58944
7222         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
7223         clear cpp_get_options (parse_in)->warn_unused_macros for
7224         ix86_target_macros_internal with cpp_define.
7226 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
7228         * jump.c (delete_related_insns): Keep (use (insn))s.
7229         * reorg.c (redundant_insn): Check for barriers too.
7231 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7233         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
7235 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
7237         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
7238         call to $$dyncall when TARGET_LONG_CALLS is true.
7240 2014-01-17  Jeff Law  <law@redhat.com>
7242         * ree.c (combine_set_extension): Temporarily disable test for
7243         changing number of hard registers.
7245 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
7247         PR middle-end/58125
7248         * ipa-inline-analysis.c (inline_free_summary):
7249         Do not free summary of aliases.
7251 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
7253         PR middle-end/59706
7254         * gimplify.c (gimplify_expr): Use create_tmp_var
7255         instead of create_tmp_var_raw.  If cond doesn't have
7256         integral type, don't add the IFN_ANNOTATE builtin at all.
7258 2014-01-17  Martin Jambor  <mjambor@suse.cz>
7260         PR ipa/59736
7261         * ipa-cp.c (prev_edge_clone): New variable.
7262         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
7263         Also resize prev_edge_clone vector.
7264         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
7265         (ipcp_edge_removal_hook): New function.
7266         (ipcp_driver): Register ipcp_edge_removal_hook.
7268 2014-01-17  Andrew Pinski  <apinski@cavium.com>
7269             Steve Ellcey  <sellcey@mips.com>
7271         PR target/59462
7272         * config/mips/mips.c (mips_print_operand): Check operand mode instead
7273         of operator mode.
7275 2014-01-17  Jeff Law  <law@redhat.com>
7277         PR middle-end/57904
7278         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
7279         so that pass_ccp runs first.
7281 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7283         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
7284         (ix86_adjust_cost): Use !TARGET_XXX.
7285         (do_reorder_for_imul): Likewise.
7286         (swap_top_of_ready_list): Likewise.
7287         (ix86_sched_reorder): Likewise.
7289 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
7291         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7292         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
7293         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
7294         (intel_memset): New.  Duplicate slm_memset.
7295         (intel_cost): New.  Duplicate slm_cost.
7296         (m_INTEL): New macro.
7297         (processor_target_table): Add "intel".
7298         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
7299         with PROCESSOR_INTEL for "intel".
7300         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
7301         PROCESSOR_SILVERMONT.
7302         (ix86_issue_rate): Likewise.
7303         (ix86_adjust_cost): Likewise.
7304         (ia32_multipass_dfa_lookahead): Likewise.
7305         (swap_top_of_ready_list): Likewise.
7306         (ix86_sched_reorder): Likewise.
7307         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
7308         instead of TARGET_OPT_AGU.
7309         * config/i386/i386.h (TARGET_INTEL): New.
7310         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
7311         (processor_type): Add PROCESSOR_INTEL.
7312         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
7313         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
7315 2014-01-17  Marek Polacek  <polacek@redhat.com>
7317         PR c/58346
7318         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
7319         size is zero.
7321 2014-01-17  Richard Biener  <rguenther@suse.de>
7323         PR tree-optimization/46590
7324         * opts.c (default_options_table): Add entries for
7325         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
7326         all enabled at -O1 but not for -Og.
7327         * common.opt (fbranch-count-reg): Remove Init(1).
7328         (fmove-loop-invariants): Likewise.
7329         (ftree-pta): Likewise.
7331 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
7333         * config/i386/i386.c (ix86_data_alignment): For compatibility with
7334         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
7335         decls to at least the GCC 4.8 used alignments.
7337         PR fortran/59440
7338         * tree-nested.c (convert_nonlocal_reference_stmt,
7339         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
7340         of GIMPLE_BIND stmts, adjust associated decls.
7342 2014-01-17  Richard Biener  <rguenther@suse.de>
7344         PR tree-optimization/46590
7345         * vec.h (vec<>::bseach): New member function implementing
7346         binary search according to C89 bsearch.
7347         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
7348         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
7349         bitmap pointer again.  Make accesses_in_loop a flat array.
7350         (mem_ref_obstack): New global.
7351         (outermost_indep_loop): Adjust for mem_ref->stored changes.
7352         (mark_ref_stored): Likewise.
7353         (ref_indep_loop_p_2): Likewise.
7354         (set_ref_stored_in_loop): New helper function.
7355         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
7356         (memref_free): Adjust.
7357         (record_mem_ref_loc): Simplify.
7358         (gather_mem_refs_stmt): Adjust.
7359         (sort_locs_in_loop_postorder_cmp): New function.
7360         (analyze_memory_references): Sort accesses_in_loop after
7361         loop postorder number.
7362         (find_ref_loc_in_loop_cmp): New function.
7363         (for_all_locs_in_loop): Find relevant cluster of locs in
7364         accesses_in_loop and iterate without recursion.
7365         (execute_sm): Avoid uninit warning.
7366         (struct ref_always_accessed): Simplify.
7367         (ref_always_accessed::operator ()): Likewise.
7368         (ref_always_accessed_p): Likewise.
7369         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
7370         loop postorder numbers here.
7371         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
7372         numbers.
7374 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
7376         PR c++/57945
7377         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
7378         on decls for which assemble_alias has been called.
7380 2014-01-17  Nick Clifton  <nickc@redhat.com>
7382         * config/msp430/msp430.opt: (mcpu): New option.
7383         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
7384         (msp430_option_override): Parse target_cpu.  If the MCU name
7385         matches a generic string, clear target_mcu.
7386         (msp430_attr): Allow numeric interrupt values up to 63.
7387         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
7388         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
7389         option.
7390         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
7391         Add mcpu matches.
7392         * config/msp430/msp430.md (popm): Use %J rather than %I.
7393         (addsi3): Use msp430_nonimmediate_operand for operand 2.
7394         (addhi_cy_i): Use immediate_operand for operand 2.
7395         * doc/invoke.texi: Document -mcpu option.
7397 2014-01-17  Richard Biener  <rguenther@suse.de>
7399         PR rtl-optimization/38518
7400         * df.h (df_analyze_loop): Declare.
7401         * df-core.c: Include cfgloop.h.
7402         (df_analyze_1): Split out main part of df_analyze.
7403         (df_analyze): Adjust.
7404         (loop_inverted_post_order_compute): New function.
7405         (loop_post_order_compute): Likewise.
7406         (df_analyze_loop): New function avoiding whole-function
7407         postorder computes.
7408         * loop-invariant.c (find_defs): Use df_analyze_loop.
7409         (find_invariants): Adjust.
7410         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
7412 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
7414         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
7415         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
7417 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
7419         * ipa-ref.c (ipa_remove_stmt_references): Fix references
7420         traversal when removing references.
7422 2014-01-16  Jan Hubicka  <hubicka@ucw.cz>
7424         PR ipa/59775
7425         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
7427 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
7429         PR middle-end/56791
7430         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
7431         pushing a reload for an autoinc when we had previously reloaded an
7432         inner part of the address.
7434 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
7436         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
7437         field.
7438         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
7439         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
7440         when not giving up or versioning for alias only because of
7441         loop->safelen.
7442         (vect_analyze_data_ref_dependences): Set to true.
7443         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
7444         is a GIMPLE_PHI.
7445         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
7446         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
7447         to the condition.
7449         PR middle-end/58344
7450         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
7452         PR target/59839
7453         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
7454         operand 0 predicate for gathers, use a new pseudo as subtarget.
7456 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
7458         PR middle-end/59609
7459         * lra-constraints.c (process_alt_operands): Add printing debug info.
7460         Check absence of input/output reloads for matched operands too.
7462 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
7464         PR rtl-optimization/59835
7465         * ira.c (ira_init_register_move_cost): Increase cost for
7466         impossible modes.
7468 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
7470         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
7472 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
7474         PR target/59780
7475         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
7476         non-register objects.  Use gen_(high/low)part more consistently.
7477         Fix assertions.
7479 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
7481         PR target/59844
7482         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
7483         endian support, remove tests for WORDS_BIG_ENDIAN.
7484         (p8_mfvsrd_3_<mode>): Likewise.
7485         (reload_gpr_from_vsx<mode>): Likewise.
7486         (reload_gpr_from_vsxsf): Likewise.
7487         (p8_mfvsrd_4_disf): Likewise.
7489 2014-01-16  Richard Biener  <rguenther@suse.de>
7491         PR rtl-optimization/46590
7492         * lcm.c (compute_antinout_edge): Use postorder iteration.
7493         (compute_laterin): Use inverted postorder iteration.
7495 2014-01-16  Nick Clifton  <nickc@redhat.com>
7497         PR middle-end/28865
7498         * varasm.c (output_constant): Return the number of bytes actually
7499         emitted.
7500         (output_constructor_array_range): Update the field size with the
7501         number of bytes emitted by output_constant.
7502         (output_constructor_regular_field): Likewise.  Also do not
7503         complain if the total number of bytes emitted is now greater
7504         than the expected fieldpos.
7505         * output.h (output_constant): Update prototype and descriptive comment.
7507 2014-01-16  Marek Polacek  <polacek@redhat.com>
7509         PR middle-end/59827
7510         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
7511         it is error_mark_node.
7513 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
7515         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
7516         VALID_AVX256_REG_OR_OI_MODE.
7518 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
7520         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
7521         current procedure should be profiled.
7523 2014-01-15  Andrew Pinski  <apinski@cavium.com>
7525         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
7526         of moving from/to the STACK_REG register class.
7528 2014-01-15  Richard Henderson  <rth@redhat.com>
7530         PR debug/54694
7531         * reginfo.c (global_regs_decl): Globalize.
7532         * rtl.h (global_regs_decl): Declare.
7533         * ira.c (do_reload): Diagnose frame_pointer_needed and it
7534         reserved via global_regs.
7536 2014-01-15  Teresa Johnson  <tejohnson@google.com>
7538         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
7540 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
7542         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
7543         and vmulosh rather than call gen_vec_widen_smult_*.
7544         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
7545         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
7546         (vec_widen_smult_even_v16qi): Likewise.
7547         (vec_widen_umult_even_v8hi): Likewise.
7548         (vec_widen_smult_even_v8hi): Likewise.
7549         (vec_widen_umult_odd_v16qi): Likewise.
7550         (vec_widen_smult_odd_v16qi): Likewise.
7551         (vec_widen_umult_odd_v8hi): Likewise.
7552         (vec_widen_smult_odd_v8hi): Likewise.
7553         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
7554         vmuloub rather than call gen_vec_widen_umult_*.
7555         (vec_widen_umult_lo_v16qi): Likewise.
7556         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
7557         vmulosb rather than call gen_vec_widen_smult_*.
7558         (vec_widen_smult_lo_v16qi): Likewise.
7559         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
7560         rather than call gen_vec_widen_umult_*.
7561         (vec_widen_umult_lo_v8hi): Likewise.
7562         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
7563         rather than call gen_vec_widen_smult_*.
7564         (vec_widen_smult_lo_v8hi): Likewise.
7566 2014-01-15  Jeff Law  <law@redhat.com>
7568         PR tree-optimization/59747
7569         * ree.c (find_and_remove_re): Properly handle case where a second
7570         eliminated extension requires widening a copy created for elimination
7571         of a prior extension.
7572         (combine_set_extension): Ensure that the number of hard regs needed
7573         for a destination register does not change when we widen it.
7575 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
7577         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
7578         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
7579         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
7580         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
7581         (avr-*-rtems*): Likewise.
7582         (bfin*-rtems*): Likewise.
7583         (moxie-*-rtems*): Likewise.
7584         (h8300-*-rtems*): Likewise.
7585         (i[34567]86-*-rtems*): Likewise.
7586         (lm32-*-rtems*): Likewise.
7587         (m32r-*-rtems*): Likewise.
7588         (m68k-*-rtems*): Likewise.
7589         (microblaze*-*-rtems*): Likewise.
7590         (mips*-*-rtems*): Likewise.
7591         (powerpc-*-rtems*): Likewise.
7592         (sh-*-rtems*): Likewise.
7593         (sparc-*-rtems*): Likewise.
7594         (sparc64-*-rtems*): Likewise.
7595         (v850-*-rtems*): Likewise.
7596         (m32c-*-rtems*): Likewise.
7598 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
7600         PR rtl-optimization/59511
7601         * ira.c (ira_init_register_move_cost): Use memory costs for some
7602         cases of register move cost calculations.
7603         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
7604         instead of BB frequency.
7605         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
7606         * lra-assigns.c (find_hard_regno_for): Ditto.
7608 2014-01-15  Richard Biener  <rguenther@suse.de>
7610         PR tree-optimization/59822
7611         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
7612         (vectorizable_load): Use it to hoist defs of uses of invariant
7613         loads out of the loop.
7615 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
7616             Kugan Vivekanandarajah  <kuganv@linaro.org>
7618         PR target/59695
7619         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
7620         truncation.
7622 2014-01-15  Richard Biener  <rguenther@suse.de>
7624         PR rtl-optimization/59802
7625         * lcm.c (compute_available): Use inverted postorder to seed
7626         the initial worklist.
7628 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7630         PR target/59803
7631         * config/s390/s390.c (s390_preferred_reload_class): Don't return
7632         ADDR_REGS for invalid symrefs in non-PIC code.
7634 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
7636         PR other/58712
7637         * builtins.c (determine_block_size): Initialize *probable_max_size
7638         even if len_rtx is CONST_INT.
7640 2014-01-14  Andrew Pinski  <apinski@cavium.com>
7642         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
7643         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
7644         (cortexa53_tunings): Likewise.
7645         (aarch64_sched_issue_rate): New function.
7646         (TARGET_SCHED_ISSUE_RATE): Define.
7648 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
7650         * ira-costs.c (find_costs_and_classes): Add missed
7651         ira_init_register_move_cost_if_necessary.
7653 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
7655         PR target/59787
7656         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
7658 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
7660         PR target/59794
7661         * config/i386/i386.c (type_natural_mode): Add a bool parameter
7662         to indicate if type is used for function return value.  Warn ABI
7663         change if the vector mode isn't available for function return value.
7664         (ix86_function_arg_advance): Pass false to type_natural_mode.
7665         (ix86_function_arg): Likewise.
7666         (ix86_gimplify_va_arg): Likewise.
7667         (function_arg_32): Don't warn ABI change.
7668         (ix86_function_value): Pass true to type_natural_mode.
7669         (ix86_return_in_memory): Likewise.
7670         (ix86_struct_value_rtx): Removed.
7671         (TARGET_STRUCT_VALUE_RTX): Likewise.
7673 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
7675         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
7676         converting a conditional jump into a conditional return.
7678 2014-01-14  Richard Biener  <rguenther@suse.de>
7680         PR tree-optimization/58921
7681         PR tree-optimization/59006
7682         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
7683         hoisting invariant stmts.
7684         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
7685         invariant loads on the preheader edge if possible.
7687 2014-01-14  Joey Ye  <joey.ye@arm.com>
7689         * doc/plugin.texi (Building GCC plugins): Update to C++.
7691 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
7693         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
7694         (_mm_rcp28_round_ss): Ditto.
7695         (_mm_rsqrt28_round_sd): Ditto.
7696         (_mm_rsqrt28_round_ss): Ditto.
7697         (_mm_rcp28_sd): Ditto.
7698         (_mm_rcp28_ss): Ditto.
7699         (_mm_rsqrt28_sd): Ditto.
7700         (_mm_rsqrt28_ss): Ditto.
7701         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
7702         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
7703         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
7704         (IX86_BUILTIN_RCP28SD): Ditto.
7705         (IX86_BUILTIN_RCP28SS): Ditto.
7706         (IX86_BUILTIN_RSQRT28SD): Ditto.
7707         (IX86_BUILTIN_RSQRT28SS): Ditto.
7708         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
7709         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
7710         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
7711         (ix86_expand_special_args_builtin): Expand new FTYPE.
7712         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
7713         (srcp14<mode>): Make insn unary.
7714         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
7715         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
7716         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
7717         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
7718         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
7719         Fix rounding: make it SAE only.
7720         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
7721         Ditto.
7722         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
7723         Ditto.
7724         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
7725         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
7726         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
7727         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
7728         (round_saeonly_mask_scalar_operand4): Ditto.
7729         (round_saeonly_mask_scalar_op3): Ditto.
7730         (round_saeonly_mask_scalar_op4): Ditto.
7732 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7734         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7735         Implement -maltivec=be for vec_insert and vec_extract.
7737 2014-01-10  DJ Delorie  <dj@redhat.com>
7739         * config/msp430/msp430.md (call_internal): Don't allow memory
7740         references with SP as the base register.
7741         (call_value_internal): Likewise.
7742         * config/msp430/constraints.md (Yc): New.  For memory references
7743         that don't use SP as a base register.
7745         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
7746         "an integer without a # prefix"
7747         * config/msp430/msp430.md (epilogue_helper): Use it.
7749 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
7751         PR target/59617
7752         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
7753         AVX512F gather builtins.
7754         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
7755         on gather decls with INTEGER_TYPE masktype.
7756         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
7757         directly into the builtin rather than hoisting it before loop.
7759         PR tree-optimization/59387
7760         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
7761         (scev_const_prop): If folded_casts and type has undefined overflow,
7762         use force_gimple_operand instead of force_gimple_operand_gsi and
7763         for each added stmt if it is assign with
7764         arith_code_with_undefined_signed_overflow, call
7765         rewrite_to_defined_overflow.
7766         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
7767         gimple-fold.h instead.
7768         (arith_code_with_undefined_signed_overflow,
7769         rewrite_to_defined_overflow): Moved to ...
7770         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
7771         rewrite_to_defined_overflow): ... here.  No longer static.
7772         Include gimplify-me.h.
7773         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
7774         rewrite_to_defined_overflow): New prototypes.
7776 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7778         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
7780 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
7782         * builtins.c (get_object_alignment_2): Minor tweak.
7783         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
7785 2014-01-13  Christian Bruel  <christian.bruel@st.com>
7787         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
7788         optimized non constant lengths.
7790 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
7792         PR libgomp/59194
7793         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
7794         load as __atomic_load_N if possible.
7796 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
7798         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
7799         target parameter.
7800         (rs6000_expand_builtin): Adjust call.
7802 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
7804         PR target/58115
7805         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
7806         * config/rs6000/rs6000.c: Include target-globals.h.
7807         (rs6000_set_current_function): Instead of doing target_reinit
7808         unconditionally, use save_target_globals_default_opts and
7809         restore_target_globals.
7811         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
7812         FPSCR.
7813         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
7814         (rs6000_expand_builtin): Handle mffs and mtfsf.
7815         (rs6000_init_builtins): Define mffs and mtfsf.
7816         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
7817         (rs6000_mffs): New pattern.
7818         (rs6000_mtfsf): New pattern.
7820 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
7822         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
7823         Start narrowing with START.  Apply candidate-use pair
7824         and check overall cost in narrowing.
7825         (iv_ca_prune): Pass new argument.
7827 2014-01-10  Jeff Law  <law@redhat.com>
7829         PR middle-end/59743
7830         * ree.c (combine_reaching_defs): Ensure the defining statement
7831         occurs before the extension when optimizing extensions with
7832         different source and destination hard registers.
7834 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
7836         PR ipa/58585
7837         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
7838         vtables into the type inheritance graph.
7840 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
7842         PR rtl-optimization/59754
7843         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
7844         modes in the REGNO != REGNO case.
7846 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7848         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
7850 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
7852         PR tree-optimization/59745
7853         * tree-predcom.c (tree_predictive_commoning_loop): Call
7854         free_affine_expand_cache if giving up because components is NULL.
7856         * target-globals.c (save_target_globals): Allocate < 4KB structs using
7857         GC in payload of target_globals struct instead of allocating them on
7858         the heap and the larger structs separately using GC.
7859         * target-globals.h (struct target_globals): Make regs, hard_regs,
7860         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
7861         of GTY((skip)) and change type to void *.
7862         (reset_target_globals): Cast loads from those fields to corresponding
7863         types.
7865 2014-01-10  Steve Ellcey  <sellcey@mips.com>
7867         PR plugins/59335
7868         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
7869         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
7870         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
7872 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
7874         PR target/59744
7875         * aarch64-modes.def (CC_Zmode): New flags mode.
7876         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
7877         represents an equality.
7878         (aarch64_get_condition_code): Handle CC_Zmode.
7879         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
7881 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7883         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
7884         extraction in good case.
7886 2014-01-10  Richard Biener  <rguenther@suse.de>
7888         PR tree-optimization/59374
7889         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
7890         checking after SLP discovery.  Mark stmts not participating
7891         in any SLP instance properly.
7893 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7895         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
7896         when handling a SET rtx.
7898 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7900         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
7901         (cortex-a57): Likewise.
7902         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
7904 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7906         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
7907         non-iwmmxt builtins.
7909 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
7911         PR ipa/58252
7912         PR ipa/59226
7913         * ipa-devirt.c record_target_from_binfo): Take as argument
7914         stack of binfos and lookup matching one for virtual inheritance.
7915         (possible_polymorphic_call_targets_1): Update.
7917 2014-01-10  Huacai Chen  <chenhc@lemote.com>
7919         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
7920         kernel strings for Loongson-2E/2F/3A.
7922 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
7924         PR middle-end/59670
7925         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
7926         is_gimple_call before calling gimple_call_internal_p.
7928 2014-01-09  Steve Ellcey  <sellcey@mips.com>
7930         * Makefile.in (TREE_FLOW_H): Remove.
7931         (TREE_SSA_H): Add file names from tree-flow.h.
7932         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
7933         * tree.h: Remove tree-flow.h reference.
7934         * hash-table.h: Remove tree-flow.h reference.
7935         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
7936         reference with tree-ssa-loop.h.
7938 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7940         * doc/invoke.texi: Add -maltivec={be,le} options, and document
7941         default element-order behavior for -maltivec.
7942         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
7943         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
7944         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
7945         when targeting big endian, at least for now.
7946         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
7948 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
7950         PR middle-end/47735
7951         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
7952         var satisfies use_register_for_decl, just take into account type
7953         alignment, rather than decl alignment.
7955         PR tree-optimization/59622
7956         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
7957         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
7958         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
7959         Don't devirtualize for inplace at all.  For targets.length () == 1,
7960         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
7962 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
7964         * config/i386/i386.md (cpu): Remove the unused btver1.
7966 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
7968         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
7970 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
7972         PR target/58115
7973         * tree-core.h (struct target_globals): New forward declaration.
7974         (struct tree_target_option): Add globals field.
7975         * tree.h (TREE_TARGET_GLOBALS): Define.
7976         (prepare_target_option_nodes_for_pch): New prototype.
7977         * target-globals.h (struct target_globals): Define even if
7978         !SWITCHABLE_TARGET.
7979         * tree.c (prepare_target_option_node_for_pch,
7980         prepare_target_option_nodes_for_pch): New functions.
7981         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
7982         * config/i386/i386.c: Include target-globals.h.
7983         (ix86_set_current_function): Instead of doing target_reinit
7984         unconditionally, use save_target_globals_default_opts and
7985         restore_target_globals.
7987 2014-01-09  Richard Biener  <rguenther@suse.de>
7989         PR tree-optimization/59715
7990         * tree-cfg.h (split_critical_edges): Declare.
7991         * tree-cfg.c (split_critical_edges): Export.
7992         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
7994 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
7996         * cfgexpand.c (expand_stack_vars): Optionally disable
7997         asan stack protection.
7998         (expand_used_vars): Likewise.
7999         (partition_stack_vars): Likewise.
8000         * asan.c (asan_emit_stack_protection): Optionally disable
8001         after return stack usage.
8002         (instrument_derefs): Optionally disable memory access instrumentation.
8003         (instrument_builtin_call): Likewise.
8004         (instrument_strlen_call): Likewise.
8005         (asan_protect_global): Optionally disable global variables protection.
8006         * doc/invoke.texi: Added doc for new options.
8007         * params.def: Added new options.
8008         * params.h: Likewise.
8010 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
8012         PR rtl-optimization/59724
8013         * ifcvt.c (cond_exec_process_if_block): Don't call
8014         flow_find_head_matching_sequence with 0 longest_match.
8015         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
8016         non-active insns if !stop_after.
8017         (try_head_merge_bb): Revert 2014-01-07 changes.
8019 2014-01-08  Jeff Law  <law@redhat.com>
8021         * ree.c (get_sub_rtx): New function, extracted from...
8022         (merge_def_and_ext): Here.
8023         (combine_reaching_defs): Use get_sub_rtx.
8025 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
8027         * cgraph.h (varpool_variable_node): Do not choke on null node.
8029 2014-01-08  Catherine Moore  <clm@codesourcery.com>
8031         * config/mips/mips.md (simple_return): Attempt to use JRC
8032         for microMIPS.
8033         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
8035 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
8037         PR rtl-optimization/59137
8038         * reorg.c (steal_delay_list_from_target): Call update_block for
8039         elided insns.
8040         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
8042 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8044         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
8045         two duplicate entries.
8047 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
8049         Revert:
8050         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
8052         * config/mips/mips.c (mips_truncated_op_cost): New function.
8053         (mips_rtx_costs): Adjust test for BADDU.
8054         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
8056         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
8058         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
8059         (*baddu_si): ...this new pattern.
8061 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
8063         PR ipa/59722
8064         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
8066 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
8068         PR middle-end/57748
8069         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
8070         inner_reference_p.
8071         (expand_expr, expand_normal): Adjust.
8072         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
8073         inner_reference_p. Use inner_reference_p to expand inner references.
8074         (store_expr): Adjust.
8075         * cfgexpand.c (expand_call_stmt): Adjust.
8077 2014-01-08  Rong Xu  <xur@google.com>
8079         * gcov-io.c (gcov_var): Move from gcov-io.h.
8080         (gcov_position): Ditto.
8081         (gcov_is_error): Ditto.
8082         (gcov_rewrite): Ditto.
8083         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
8084         only part to libgcc/libgcov.h.
8086 2014-01-08  Marek Polacek  <polacek@redhat.com>
8088         PR middle-end/59669
8089         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
8091 2014-01-08  Marek Polacek  <polacek@redhat.com>
8093         PR sanitizer/59667
8094         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
8096 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
8098         PR rtl-optimization/59649
8099         * stor-layout.c (get_mode_bounds): For BImode return
8100         0 and STORE_FLAG_VALUE.
8102 2014-01-08  Richard Biener  <rguenther@suse.de>
8104         PR middle-end/59630
8105         * gimple.h (is_gimple_builtin_call): Remove.
8106         (gimple_builtin_call_types_compatible_p): New.
8107         (gimple_call_builtin_p): New overload.
8108         * gimple.c (is_gimple_builtin_call): Remove.
8109         (validate_call): Rename to ...
8110         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
8111         check return types.
8112         (validate_type): New static function.
8113         (gimple_call_builtin_p): New overload and adjust.
8114         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
8115         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
8116         (gimple_fold_stmt_to_constant_1): Likewise.
8117         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
8119 2014-01-08  Richard Biener  <rguenther@suse.de>
8121         PR middle-end/59471
8122         * gimplify.c (gimplify_expr): Gimplify register-register type
8123         VIEW_CONVERT_EXPRs to separate stmts.
8125 2014-01-07  Jeff Law  <law@redhat.com>
8127         PR middle-end/53623
8128         * ree.c (combine_set_extension): Handle case where source
8129         and destination registers in an extension insn are different.
8130         (combine_reaching_defs): Allow source and destination registers
8131         in extension to be different under limited circumstances.
8132         (add_removable_extension): Remove restriction that the
8133         source and destination registers in the extension are the same.
8134         (find_and_remove_re): Emit a copy from the extension's
8135         destination to its source after the defining insn if
8136         the source and destination registers are different.
8138         PR middle-end/59285
8139         * ifcvt.c (merge_if_block): If we are merging a block with more than
8140         one successor with a block with no successors, remove any BARRIER
8141         after the second block.
8143 2014-01-07  Dan Xio Qiang  <ziyan01@163.com>
8145         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
8147 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
8149         PR target/59652
8150         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
8151         for 14-bit register offsets when INT14_OK_STRICT is false.
8153 2014-01-07  Roland Stigge  <stigge@antcom.de>
8154             Michael Meissner  <meissner@linux.vnet.ibm.com>
8156         PR 57386/target
8157         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
8158         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
8160 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
8162         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
8163         -mcpu.
8165 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
8167         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
8168         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
8169         rtx is const0_rtx or not.
8171 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
8173         PR target/58115
8174         * target-globals.c (save_target_globals): Remove this_fn_optab
8175         handling.
8176         * toplev.c: Include optabs.h.
8177         (target_reinit): Temporarily restore the global options if another
8178         set of options are in force.
8180 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
8182         PR rtl-optimization/58668
8183         * cfgcleanup.c (flow_find_cross_jump): Don't count
8184         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
8185         to determine what is counted.
8186         (flow_find_head_matching_sequence): Use active_insn_p to determine
8187         what is counted.
8188         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
8189         counting change.
8190         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
8191         determine what is counted.
8193         PR tree-optimization/59643
8194         * tree-predcom.c (split_data_refs_to_components): If one dr is
8195         read and one write, determine_offset fails and the write isn't
8196         in the bad component, just put the read into the bad component.
8198 2014-01-07  Mike Stump  <mikestump@comcast.net>
8199             Jakub Jelinek  <jakub@redhat.com>
8201         PR pch/59436
8202         * tree-core.h (struct tree_optimization_option): Change optabs
8203         type from unsigned char * to void *.
8204         * optabs.c (init_tree_optimization_optabs): Adjust
8205         TREE_OPTIMIZATION_OPTABS initialization.
8207 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
8209         PR target/59644
8210         * config/i386/i386.h (struct machine_function): Add
8211         no_drap_save_restore field.
8212         * config/i386/i386.c (ix86_save_reg): Use
8213         !cfun->machine->no_drap_save_restore instead of
8214         crtl->stack_realign_needed.
8215         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
8216         this function clears frame_pointer_needed.  Set
8217         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
8218         and DRAP reg is needed.
8220 2014-01-06  Marek Polacek  <polacek@redhat.com>
8222         PR c/57773
8223         * doc/implement-c.texi: Mention that other integer types are
8224         permitted as bit-field types in strictly conforming mode.
8226 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
8228         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
8229         is newly allocated.
8231 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
8233         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
8235 2014-01-06  Martin Jambor  <mjambor@suse.cz>
8237         PR ipa/59008
8238         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
8239         to int.
8240         * ipa-prop.c (ipa_print_node_params): Fix indentation.
8242 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
8244         PR debug/59350
8245         PR debug/59510
8246         * var-tracking.c (add_stores): Preserve the value of the source even if
8247         we don't record the store.
8249 2014-01-06  Terry Guo  <terry.guo@arm.com>
8251         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
8253 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
8255         PR bootstrap/59541
8256         * config/darwin.c (darwin_function_section): Adjust return values to
8257         correspond to optimisation changes made in r206070.
8259 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
8261         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
8262         from prefetch_block tune setting.
8263         (nocona_cost): Correct size of prefetch block to 64.
8265 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
8267         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
8268         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
8269         used to save the static chain register in the computation of the offset
8270         from which the FP registers need to be restored.
8272 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
8274         PR tree-optimization/59519
8275         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
8276         ICE if get_current_def (current_new_name) is already non-NULL, as long
8277         as it is a phi result of some other phi in *new_exit_bb that has
8278         the same argument.
8280         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
8281         or vmovdqu* for misaligned_operand.
8282         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
8283         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
8284         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
8285         aligned_mem for AVX512F masked aligned load and store builtins and for
8286         non-temporal moves.
8288 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
8290         PR tree-optimization/59651
8291         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
8292         Address range for negative step should be added by TYPE_SIZE_UNIT.
8294 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
8296         * config/m68k/m68k.c (handle_move_double): Handle pushes with
8297         overlapping registers also for registers other than the stack pointer.
8299 2014-01-03  Marek Polacek  <polacek@redhat.com>
8301         PR other/59661
8302         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
8303         __builtin_FILE.
8305 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
8307         PR target/59625
8308         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
8309         asm goto as jump.
8311         * config/i386/i386.md (MODE_SIZE): New mode attribute.
8312         (push splitter): Use <P:MODE_SIZE> instead of
8313         GET_MODE_SIZE (<P:MODE>mode).
8314         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
8315         (mov -1, reg peephole2): Likewise.
8316         * config/i386/sse.md (*mov<mode>_internal,
8317         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
8318         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
8319         *<code><mode>3, *andnot<mode>3<mask_name>,
8320         <mask_codefor><code><mode>3<mask_name>): Likewise.
8321         * config/i386/subst.md (mask_mode512bit_condition,
8322         sd_mask_mode512bit_condition): Likewise.
8324 2014-01-02  Xinliang David Li  <davidxl@google.com>
8326         PR tree-optimization/59303
8327         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
8328         (dump_predicates): Better output format.
8329         (pred_equal_p): New function.
8330         (is_neq_relop_p): Ditto.
8331         (is_neq_zero_form_p): Ditto.
8332         (pred_expr_equal_p): Ditto.
8333         (pred_neg_p): Ditto.
8334         (simplify_pred): Ditto.
8335         (simplify_preds_2): Ditto.
8336         (simplify_preds_3): Ditto.
8337         (simplify_preds_4): Ditto.
8338         (simplify_preds): Ditto.
8339         (push_pred): Ditto.
8340         (push_to_worklist): Ditto.
8341         (get_pred_info_from_cmp): Ditto.
8342         (is_degenerated_phi): Ditto.
8343         (normalize_one_pred_1): Ditto.
8344         (normalize_one_pred): Ditto.
8345         (normalize_one_pred_chain): Ditto.
8346         (normalize_preds): Ditto.
8347         (normalize_cond_1): Remove function.
8348         (normalize_cond): Ditto.
8349         (is_gcond_subset_of): Ditto.
8350         (is_subset_of_any): Ditto.
8351         (is_or_set_subset_of): Ditto.
8352         (is_and_set_subset_of): Ditto.
8353         (is_norm_cond_subset_of): Ditto.
8354         (pred_chain_length_cmp): Ditto.
8355         (convert_control_dep_chain_into_preds): Type change.
8356         (find_predicates): Ditto.
8357         (find_def_preds): Ditto.
8358         (destroy_predicates_vecs): Ditto.
8359         (find_matching_predicates_in_rest_chains): Ditto.
8360         (use_pred_not_overlap_with_undef_path_pred): Ditto.
8361         (is_pred_expr_subset): Ditto.
8362         (is_pred_chain_subset_of): Ditto.
8363         (is_included_in): Ditto.
8364         (is_superset_of): Ditto.
8366 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
8368         Update copyright years.
8370 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
8372         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
8373         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
8374         config/arc/arc.md, config/arc/arc.opt,
8375         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
8376         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
8377         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
8378         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
8379         config/linux-protos.h, config/linux.c, config/winnt-c.c,
8380         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
8381         vtable-verify.c, vtable-verify.h: Use the standard form for the
8382         copyright notice.
8384 2014-01-02  Tobias Burnus  <burnus@net-b.de>
8386         * gcc.c (process_command): Update copyright notice dates.
8387         * gcov-dump.c: Ditto.
8388         * gcov.c: Ditto.
8389         * doc/cpp.texi: Bump @copying's copyright year.
8390         * doc/cppinternals.texi: Ditto.
8391         * doc/gcc.texi: Ditto.
8392         * doc/gccint.texi: Ditto.
8393         * doc/gcov.texi: Ditto.
8394         * doc/install.texi: Ditto.
8395         * doc/invoke.texi: Ditto.
8397 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8399         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
8401 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
8403         * config/i386/sse.md (*mov<mode>_internal): Guard
8404         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
8406         PR rtl-optimization/59647
8407         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
8408         new_rtx into UNSIGNED_FLOAT rtxes.
8410 Copyright (C) 2014 Free Software Foundation, Inc.
8412 Copying and distribution of this file, with or without modification,
8413 are permitted in any medium without royalty provided the copyright
8414 notice and this notice are preserved.