* config/msp430/msp430.h (LIB_SPEC): Add -lcrt
[official-gcc.git] / gcc / ChangeLog
blob158f2d6a12ee91e698c734ad782704f32d099ae2
1 2014-01-30  DJ Delorie  <dj@redhat.com>
3         * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4         * config/msp430/msp430.md (msp430_refsym_need_exit): New.
5         * config/msp430/msp430.c (msp430_expand_epilogue): Call it
6         whenever main() has an epilogue.
8 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
11         unused variable "field".
12         * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
13         (vsx_mergeh_<mode>): Likewise.
14         * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
15         (altivec_vmrghh): Likewise.
16         (altivec_vmrghw): Likewise.
17         (altivec_vmrglb): Likewise.
18         (altivec_vmrglh): Likewise.
19         (altivec_vmrglw): Likewise.
20         (altivec_vspltb): Add missing uses.
21         (altivec_vsplth): Likewise.
22         (altivec_vspltw): Likewise.
23         (altivec_vspltsf): Likewise.
25 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
27         PR target/59923
28         * ifcvt.c (cond_exec_process_insns): Don't conditionalize
29         frame related instructions.
31 2014-01-30  Vladimir Makarov  <vmakarov@redhat.com>
33         PR rtl-optimization/59959
34         * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
35         any reload of register whose subreg is invalid.
37 2014-01-30  Jakub Jelinek  <jakub@redhat.com>
39         * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing
40         warning.
41         * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
42         Add missing return type - void.
44 2014-01-30  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
46         * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
47         gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
48         remove element index adjustment for endian (now handled in vsx.md
49         and altivec.md).
50         (altivec_expand_vec_perm_const): Use
51         gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
52         * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
53         (vsx_xxspltw_<mode>): Adjust element index for little endian.
54         * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
55         define_expand and a new define_insn *altivec_vspltb_internal;
56         adjust for -maltivec=be on a little endian target.
57         (altivec_vspltb_direct): New.
58         (altivec_vsplth): Divide into a define_expand and a new
59         define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
60         little endian target.
61         (altivec_vsplth_direct): New.
62         (altivec_vspltw): Divide into a define_expand and a new
63         define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
64         little endian target.
65         (altivec_vspltw_direct): New.
66         (altivec_vspltsf): Divide into a define_expand and a new
67         define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
68         a little endian target.
70 2014-01-30  Richard Biener  <rguenther@suse.de>
72         PR tree-optimization/59993
73         * tree-ssa-forwprop.c (associate_pointerplus): Check we
74         can propagate form the earlier stmt and avoid the transform
75         when the intermediate result is needed.
77 2014-01-30  Alangi Derick  <alangiderick@gmail.com>
79         * README.Portability: Fix typo.
81 2014-01-30  David Holsgrove <david.holsgrove@xilinx.com>
83         * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
84         comparison_operator with ordered_comparison_operator.
86 2014-01-30  Nick Clifton  <nickc@redhat.com>
88         * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
89         Rename to mn10300_store_multiple_regs.
90         * config/mn10300/mn10300.c: Likewise.
91         * config/mn10300/mn10300.md (store_movm): Fix typo: call
92         store_multiple_regs.
93         * config/mn10300/predicates.md (mn10300_store_multiple_operation):
94         Call mn10300_store_multiple_regs.
96 2014-01-30  Nick Clifton  <nickc@redhat.com>
97             DJ Delorie  <dj@redhat.com>
99         * config/rl78/rl78.c (register_sizes): Make the "upper half" of
100         %fp 2 to keep registers after it properly word-aligned.
101         (rl78_alloc_physical_registers_umul): Handle the case where both
102         input operands are the same.
104 2014-01-30  Richard Biener  <rguenther@suse.de>
106         PR tree-optimization/59903
107         * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
108         check properly.
110 2014-01-30  Jason Merrill  <jason@redhat.com>
112         PR c++/59633
113         * tree.c (walk_type_fields): Handle VECTOR_TYPE.
115         PR c++/59645
116         * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
118 2014-01-30  Richard Biener  <rguenther@suse.de>
120         PR tree-optimization/59951
121         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug
122         insns.
124 2014-01-30  Savin Zlobec  <savin.zlobec@gmail.com>
126         PR target/59784
127         * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
128         SFmode to DFmode case.
130 2014-01-29  DJ Delorie  <dj@redhat.com>
132         * config/msp430/msp430.opt (-minrt): New.
133         * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
134         if -minrt given.
135         (ENDFILE_SPEC): Likewise.
137 2014-01-29  Jan Hubicka  <jh@suse.cz>
139         * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
140         (estimate_function_body_sizes): Use it.
142 2014-01-29  Paolo Carlini  <paolo.carlini@oracle.com>
144         PR c++/58561
145         * dwarf2out.c (is_cxx_auto): New.
146         (is_base_type): Use it.
147         (gen_type_die_with_usage): Likewise.
149 2014-01-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
151         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const):  Use
152         CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
153         * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
154         -maltivec=be with LE targets.
155         (vsx_mergeh_<mode>): Likewise.
156         * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New
157         unspecs.
158         (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
159         (altivec_vmrghb): Replace with define_expand and new
160         *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE
161         targets.
162         (altivec_vmrghb_direct): New define_insn.
163         (altivec_vmrghh): Replace with define_expand and new
164         *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE
165         targets.
166         (altivec_vmrghh_direct): New define_insn.
167         (altivec_vmrghw): Replace with define_expand and new
168         *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE
169         targets.
170         (altivec_vmrghw_direct): New define_insn.
171         (*altivec_vmrghsf): Adjust for endianness.
172         (altivec_vmrglb): Replace with define_expand and new
173         *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE
174         targets.
175         (altivec_vmrglb_direct): New define_insn.
176         (altivec_vmrglh): Replace with define_expand and new
177         *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE
178         targets.
179         (altivec_vmrglh_direct): New define_insn.
180         (altivec_vmrglw): Replace with define_expand and new
181         *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE
182         targets.
183         (altivec_vmrglw_direct): New define_insn.
184         (*altivec_vmrglsf): Adjust for endianness.
185         (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
186         (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
187         (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
188         (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
189         (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
190         (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
191         (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
192         (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
194 2014-01-29  Marcus Shawcroft  <marcus.shawcroft@arm.com>
196         * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
197         (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
198         whitespace.
200 2014-01-29  Richard Biener  <rguenther@suse.de>
202         PR tree-optimization/58742
203         * tree-ssa-forwprop.c (associate_pointerplus): Rename to
204         associate_pointerplus_align.
205         (associate_pointerplus_diff): New function.
206         (associate_pointerplus): Likewise.  Call associate_pointerplus_align
207         and associate_pointerplus_diff.
209 2014-01-29  Richard Biener  <rguenther@suse.de>
211         * lto-streamer.h (LTO_major_version): Bump to 3.
212         (LTO_minor_version): Reset to 0.
214 2014-01-29  Renlin Li  <Renlin.Li@arm.com>
216         * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
217         * config/arm/arm.c (FL_FOR_ARCH7VE): New.
218         (arm_file_start): Generate correct asm header for armv7ve.
219         * config/arm/bpabi.h: Add multilib support for armv7ve.
220         * config/arm/driver-arm.c: Change the architectures of cortex-a7
221         and cortex-a15 to armv7ve.
222         * config/arm/t-aprofile: Add multilib support for armv7ve.
223         * doc/invoke.texi: Document -march=armv7ve.
225 2014-01-29  Richard Biener  <rguenther@suse.de>
227         PR tree-optimization/58742
228         * tree-ssa-forwprop.c (associate_plusminus): Return true
229         if we changed sth, defer EH cleanup to ...
230         (ssa_forward_propagate_and_combine): ... here.  Call simplify_mult.
231         (simplify_mult): New function.
233 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
235         PR middle-end/59917
236         PR tree-optimization/59920
237         * tree.c (build_common_builtin_nodes): Remove
238         __builtin_setjmp_dispatcher initialization.
239         * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
240         * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
241         instead of gsi_after_labels + manually skipping debug stmts.
242         Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
243         ignore bbs with IFN_ABNORMAL_DISPATCHER.
244         * tree-inline.c (copy_edges_for_bb): Remove
245         can_make_abnormal_goto argument, instead add abnormal_goto_dest
246         argument.  Ignore computed_goto_p stmts.  Don't call
247         make_abnormal_goto_edges.  If a call might need abnormal edges
248         for non-local gotos, see if it already has an edge to
249         IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
250         with true argument, don't do anything then, otherwise add
251         EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
252         (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
253         caller.
254         * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
255         (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
256         (lower_stmt): Don't set data->calls_builtin_setjmp.
257         (lower_builtin_setjmp): Adjust comment.
258         * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
259         * tree-cfg.c (found_computed_goto): Remove.
260         (factor_computed_gotos): Remove.
261         (make_goto_expr_edges): Return bool, true for computed gotos.
262         Don't call make_abnormal_goto_edges.
263         (build_gimple_cfg): Don't set found_computed_goto, don't call
264         factor_computed_gotos.
265         (computed_goto_p): No longer static.
266         (make_blocks): Don't set found_computed_goto.
267         (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
268         (make_edges): If make_goto_expr_edges returns true, push bb
269         into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
270         instead of calling make_abnormal_goto_edges push bb into ab_edge_call
271         vector.  Record mapping between bbs and OpenMP regions if there
272         are any, adjust make_gimple_omp_edges caller.  Call
273         handle_abnormal_edges.
274         (make_abnormal_goto_edges): Remove.
275         * tree-cfg.h (make_abnormal_goto_edges): Remove.
276         (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
277         * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
278         * builtins.c (expand_builtin): Don't handle
279         BUILT_IN_SETJMP_DISPATCHER.
280         * internal-fn.def (ABNORMAL_DISPATCHER): New.
281         * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
282         filling *region also set *region_idx to (*region)->entry->index.
284         PR other/58712
285         * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
286         For REGs set ORIGINAL_REGNO.
288 2014-01-29  Bingfeng Mei  <bmei@broadcom.com>
290         * doc/md.texi: Mention that a target shouldn't implement 
291         vec_widen_(s|u)mul_even/odd pair if it is less efficient
292         than hi/lo pair.
294 2014-01-29  Jakub Jelinek  <jakub@redhat.com>
296         PR tree-optimization/59594
297         * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
298         a copy of the datarefs vector rather than the vector itself.
300 2014-01-28  Jason Merrill  <jason@redhat.com>
302         PR c++/53756
303         * dwarf2out.c (auto_die): New static.
304         (gen_type_die_with_usage): Handle C++1y 'auto'.
305         (gen_subprogram_die): If in-class DIE had 'auto', emit type again
306         on definition.
308 2014-01-28  H.J. Lu  <hongjiu.lu@intel.com>
310         PR target/59672
311         * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
312         (SPEC_X32): Likewise.
313         (SPEC_64): Likewise.
314         * config/i386/i386.c (ix86_option_override_internal): Turn off
315         OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
316         for TARGET_16BIT.
317         (x86_file_start): Output .code16gcc for TARGET_16BIT.
318         * config/i386/i386.h (TARGET_16BIT): New macro.
319         (TARGET_16BIT_P): Likewise.
320         * config/i386/i386.opt: Add m16.
321         * doc/invoke.texi: Document -m16.
323 2014-01-28  Jakub Jelinek  <jakub@redhat.com>
325         PR preprocessor/59935
326         * input.c (location_get_source_line): Bail out on when line number
327         is zero, and test the return value of
328         lookup_or_add_file_to_cache_tab.
330 2014-01-28  Richard Biener  <rguenther@suse.de>
332         PR tree-optimization/58742
333         * tree-ssa-forwprop.c (associate_plusminus): Handle
334         pointer subtraction of the form (T)(P + A) - (T)P.
336 2014-01-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
338         * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
339         at const_int_cost.
341 2014-01-28  Richard Biener  <rguenther@suse.de>
343         Revert
344         2014-01-28  Richard Biener  <rguenther@suse.de>
346         PR rtl-optimization/45364
347         PR rtl-optimization/59890
348         * var-tracking.c (local_get_addr_clear_given_value): Handle
349         already cleared slot.
350         (val_reset): Handle not allocated local_get_addr_cache.
351         (vt_find_locations): Use post-order on the inverted CFG.
353 2014-01-28  Richard Biener  <rguenther@suse.de>
355         * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps):
356         Remove.
358 2014-01-28  Richard Biener  <rguenther@suse.de>
360         PR rtl-optimization/45364
361         PR rtl-optimization/59890
362         * var-tracking.c (local_get_addr_clear_given_value): Handle
363         already cleared slot.
364         (val_reset): Handle not allocated local_get_addr_cache.
365         (vt_find_locations): Use post-order on the inverted CFG.
367 2014-01-28  Alan Modra  <amodra@gmail.com>
369         * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
370         * configure.ac <recursive call for build != host>: Define
371         GENERATOR_FILE.  Comment.  Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
372         and LD_FOR_BUILD too.
373         * configure: Regenerate.
375 2014-01-27  Allan Sandfeld Jensen  <sandfeld@kde.org>
377         * config/i386/i386.c (get_builtin_code_for_version): Separate
378         Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
379         Broadwell from Haswell.
381 2014-01-27  Steve Ellcey  <sellcey@mips.com>
383         * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
384         Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
385         * config/mips/mips.c (mips_option_override): Change setting
386         of TARGET_DSP.
387         * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
388         * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
389         Change from Mask to Var.
391 2014-01-27  Jeff Law  <law@redhat.com>
393         * ipa-inline.c (inline_small_functions): Fix typo.
395 2014-01-27  Ilya Tocar  <ilya.tocar@intel.com>
397         * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
398         (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
399         (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
400         (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
401         (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
402         (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
403         (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
404         (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
405         (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
406         (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
407         (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
408         (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
409         (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
410         (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
411         (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
412         (_mm512_storeu_epi64): Ditto.
413         (_mm512_cmpge_epi32_mask): Ditto.
414         (_mm512_cmpge_epu32_mask): Ditto.
415         (_mm512_cmpge_epi64_mask): Ditto.
416         (_mm512_cmpge_epu64_mask): Ditto.
417         (_mm512_cmple_epi32_mask): Ditto.
418         (_mm512_cmple_epu32_mask): Ditto.
419         (_mm512_cmple_epi64_mask): Ditto.
420         (_mm512_cmple_epu64_mask): Ditto.
421         (_mm512_cmplt_epi32_mask): Ditto.
422         (_mm512_cmplt_epu32_mask): Ditto.
423         (_mm512_cmplt_epi64_mask): Ditto.
424         (_mm512_cmplt_epu64_mask): Ditto.
425         (_mm512_cmpneq_epi32_mask): Ditto.
426         (_mm512_cmpneq_epu32_mask): Ditto.
427         (_mm512_cmpneq_epi64_mask): Ditto.
428         (_mm512_cmpneq_epu64_mask): Ditto.
429         (_mm512_expand_pd): Ditto.
430         (_mm512_expand_ps): Ditto.
431         * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
432         VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
433         VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
434         * config/i386/i386.c (ix86_builtins): Add
435         IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
436         IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
437         IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
438         IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
439         IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
440         IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
441         IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
442         IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
443         IX86_BUILTIN_PMOVUSQW512_MEM.
444         (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
445         __builtin_ia32_pmovsqd512mem_mask,
446         __builtin_ia32_pmovqd512mem_mask,
447         __builtin_ia32_pmovusqw512mem_mask,
448         __builtin_ia32_pmovsqw512mem_mask,
449         __builtin_ia32_pmovqw512mem_mask,
450         __builtin_ia32_pmovusdw512mem_mask,
451         __builtin_ia32_pmovsdw512mem_mask,
452         __builtin_ia32_pmovdw512mem_mask,
453         __builtin_ia32_pmovqb512mem_mask,
454         __builtin_ia32_pmovusqb512mem_mask,
455         __builtin_ia32_pmovsqb512mem_mask,
456         __builtin_ia32_pmovusdb512mem_mask,
457         __builtin_ia32_pmovsdb512mem_mask,
458         __builtin_ia32_pmovdb512mem_mask.
459         (bdesc_args): Add __builtin_ia32_expanddf512,
460         __builtin_ia32_expandsf512.
461         (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
462         VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
463         VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
464         * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
465         (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
466         (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
467         (avx512f_<code>v8div16qi2_mask_store): This.
468         (avx512f_expand<mode>): New.
470 2014-01-27  Kirill Yukhin  <kirill.yukhin@intel.com>
472         * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
473         New.
474         (_mm512_mask_prefetch_i64gather_pd): Ditto.
475         (_mm512_prefetch_i32scatter_pd): Ditto.
476         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
477         (_mm512_prefetch_i64scatter_pd): Ditto.
478         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
479         (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
480         (_mm512_mask_prefetch_i64gather_ps): Ditto.
481         (_mm512_prefetch_i32scatter_ps): Ditto.
482         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
483         (_mm512_prefetch_i64scatter_ps): Ditto.
484         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
485         * config/i386/i386-builtin-types.def: Define
486         VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
487         and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
488         * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
489         IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
490         IX86_BUILTIN_SCATTERPFQPD.
491         (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
492         __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
493         __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
494         __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
495         __builtin_ia32_scatterpfqps.
496         (ix86_expand_builtin): Expand new built-ins.
497         * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
498         fix memory access data type.
499         (*avx512pf_gatherpf<mode>_mask): Ditto.
500         (*avx512pf_gatherpf<mode>): Ditto.
501         (avx512pf_scatterpf<mode>): Ditto.
502         (*avx512pf_scatterpf<mode>_mask): Ditto.
503         (*avx512pf_scatterpf<mode>): Ditto.
504         (GATHER_SCATTER_SF_MEM_MODE): New.
505         (avx512pf_gatherpf<mode>df): Ditto.
506         (*avx512pf_gatherpf<mode>df_mask): Ditto.
507         (*avx512pf_scatterpf<mode>df): Ditto.
509 2014-01-27  Jakub Jelinek  <jakub@redhat.com>
511         PR bootstrap/59934
512         * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
513         know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
514         reached.
516 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
518         * common/config/arm/arm-common.c
519         (arm_rewrite_mcpu): Handle multiple names.
520         * config/arm/arm.h
521         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
523 2014-01-27  James Greenhalgh  <james.greenhalgh@arm.com>
525         * gimple-builder.h (create_gimple_tmp): Delete.
527 2014-01-27  Christian Bruel  <christian.bruel@st.com>
529         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
530         words comparisons.
532 2014-01-26  John David Anglin  <danglin@gcc.gnu.org>
534         * config/pa/pa.md (call): Generate indirect long calls to non-local
535         functions when outputing 32-bit code.
536         (call_value): Likewise except for special call to buggy powf function.
538         * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
539         portable runtime and PIC indirect calls.
540         (pa_output_indirect_call): Remove unnecessary nop from portable runtime
541         and PIC call sequences.  Use ldo instead of blr to set return register
542         in PIC call sequence.
544 2014-01-25  Walter Lee  <walt@tilera.com>
546         * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
547         avoid clobbering a live register.
549 2014-01-25  Walter Lee  <walt@tilera.com>
551         * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
552         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
553         * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
554         Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
556 2014-01-25  Walter Lee  <walt@tilera.com>
558         * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
559         arguments on even registers.
560         (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
561         STACK_BOUNDARY.
562         * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
563         (BIGGEST_ALIGNMENT): Ditto.
564         (BIGGEST_FIELD_ALIGNMENT): Ditto.
566 2014-01-25  Walter Lee  <walt@tilera.com>
568         * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
569         insns before bundling.
570         * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
572 2014-01-25  Walter Lee  <walt@tilera.com>
574         * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
575         PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
576         * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
578 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
580         * config/mips/constraints.md (kl): Delete.
581         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
582         define expands, using...
583         (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
584         instructions for MIPS16.
585         (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
586         non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
588 2014-01-25  Walter Lee  <walt@tilera.com>
590         * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
591         (clzdi2): Ditto.
592         (ffsdi2): Ditto.
594 2014-01-25  Walter Lee  <walt@tilera.com>
596         * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
597         (TARGET_EXPAND_TO_RTL_HOOK): Define.
599 2014-01-25  Richard Sandiford  <rdsandiford@googlemail.com>
601         * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
602         Handle XOR.
604 2014-01-25  Jakub Jelinek  <jakub@redhat.com>
606         * print-rtl.c (in_call_function_usage): New var.
607         (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
608         EXPR_LIST mode as mode and not as reg note name.
610         PR middle-end/59561
611         * cfgloopmanip.c (copy_loop_info): If
612         loop->warned_aggressive_loop_optimizations, make sure
613         the flag is set in target loop too.
615 2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
617         * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
618         flag_cilkplus.
619         * builtins.def: Likewise.
620         * cilk.h (fn_contains_cilk_spawn_p): Likewise.
621         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
622         * ira.c (ira_setup_eliminable_regset): Likewise.
623         * omp-low.c (gate_expand_omp): Likewise.
624         (execute_lower_omp): Likewise.
625         (diagnose_sb_0): Likewise.
626         (gate_diagnose_omp_blocks): Likewise.
627         (simd_clone_clauses_extract): Likewise.
628         (gate): Likewise.
630 2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
632         * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
633         correction for little endian...
634         * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
635         here.
637 2014-01-24  Jeff Law  <law@redhat.com>
639         PR tree-optimization/59919
640         * tree-vrp.c (find_assert_locations_1): Do not register asserts
641         for non-returning calls.
643 2014-01-24  James Greenhalgh  <james.greenhalgh@arm.com>
645         * common/config/aarch64/aarch64-common.c
646         (aarch64_rewrite_mcpu): Handle multiple names.
647         * config/aarch64/aarch64.h
648         (BIG_LITTLE_SPEC): Do not discard mcpu switches.
650 2014-01-24  Dodji Seketeli  <dodji@redhat.com>
652         * input.c (add_file_to_cache_tab): Handle the case where fopen
653         returns NULL.
655 2014-01-23  H.J. Lu  <hongjiu.lu@intel.com>
657         PR target/59929
658         * config/i386/i386.md (pushsf splitter): Get stack adjustment
659         from push operand if code of push isn't PRE_DEC.
661 2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
663         PR target/59909
664         * doc/invoke.texi (RS/6000 and PowerPC Options): Document
665         -mquad-memory-atomic.  Update -mquad-memory documentation to say
666         it is only used for non-atomic loads/stores.
668         * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
669         -mquad-memory or -mquad-memory-atomic switches.
671         * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
672         -mquad-memory-atomic to ISA 2.07 support.
674         * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
675         to separate support of normal quad word memory operations (ldq, stq)
676         from the atomic quad word memory operations.
678         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
679         support to separate non-atomic quad word operations from atomic
680         quad word operations.  Disable non-atomic quad word operations in
681         little endian mode so that we don't have to swap words after the
682         load and before the store.
683         (quad_load_store_p): Add comment about atomic quad word support.
684         (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
685         options printed with -mdebug=reg.
687         * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
688         -mquad-memory-atomic as the test for whether we have quad word
689         atomic instructions.
690         (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
691         or -mp8-vector are used, allow byte/half-word atomic operations.
693         * config/rs6000/sync.md (load_lockedti): Insure that the address
694         is a proper indexed or indirect address for the lqarx instruction.
695         On little endian systems, swap the hi/lo registers after the lqarx
696         instruction.
697         (load_lockedpti): Use indexed_or_indirect_operand predicate to
698         insure the address is valid for the lqarx instruction.
699         (store_conditionalti): Insure that the address is a proper indexed
700         or indirect address for the stqcrx. instruction.  On little endian
701         systems, swap the hi/lo registers before doing the stqcrx.
702         instruction.
703         (store_conditionalpti): Use indexed_or_indirect_operand predicate to
704         insure the address is valid for the stqcrx. instruction.
706         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
707         Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
708         type of quad memory support is available.
710 2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>
712         PR regression/59915
713         * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
714         there is a danger of looping.
716 2014-01-23  Pat Haugen  <pthaugen@us.ibm.com>
718         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
719         force flag_ira_loop_pressure if set via command line.
721 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
723         * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
724         (ashr_simd): New builtin handling DI mode.
725         * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
726         (aarch64_sshr_simddi): New match pattern.
727         * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
728         (vshrd_n_s64): Likewise.
729         * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
731 2014-01-23  Nick Clifton  <nickc@redhat.com>
733         * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
734         (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
735         favour of mcu specific scripts.
736         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
737         430x multilibs.
739 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
740             Alex Velenko  <Alex.Velenko@arm.com>
742         * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
743         (vaddv_s16): Likewise.
744         (vaddv_s32): Likewise.
745         (vaddv_u8): Likewise.
746         (vaddv_u16): Likewise.
747         (vaddv_u32): Likewise.
748         (vaddvq_s8): Likewise.
749         (vaddvq_s16): Likewise.
750         (vaddvq_s32): Likewise.
751         (vaddvq_s64): Likewise.
752         (vaddvq_u8): Likewise.
753         (vaddvq_u16): Likewise.
754         (vaddvq_u32): Likewise.
755         (vaddvq_u64): Likewise.
756         (vaddv_f32): Likewise.
757         (vaddvq_f32): Likewise.
758         (vaddvq_f64): Likewise.
759         (vmaxv_f32): Likewise.
760         (vmaxv_s8): Likewise.
761         (vmaxv_s16): Likewise.
762         (vmaxv_s32): Likewise.
763         (vmaxv_u8): Likewise.
764         (vmaxv_u16): Likewise.
765         (vmaxv_u32): Likewise.
766         (vmaxvq_f32): Likewise.
767         (vmaxvq_f64): Likewise.
768         (vmaxvq_s8): Likewise.
769         (vmaxvq_s16): Likewise.
770         (vmaxvq_s32): Likewise.
771         (vmaxvq_u8): Likewise.
772         (vmaxvq_u16): Likewise.
773         (vmaxvq_u32): Likewise.
774         (vmaxnmv_f32): Likewise.
775         (vmaxnmvq_f32): Likewise.
776         (vmaxnmvq_f64): Likewise.
777         (vminv_f32): Likewise.
778         (vminv_s8): Likewise.
779         (vminv_s16): Likewise.
780         (vminv_s32): Likewise.
781         (vminv_u8): Likewise.
782         (vminv_u16): Likewise.
783         (vminv_u32): Likewise.
784         (vminvq_f32): Likewise.
785         (vminvq_f64): Likewise.
786         (vminvq_s8): Likewise.
787         (vminvq_s16): Likewise.
788         (vminvq_s32): Likewise.
789         (vminvq_u8): Likewise.
790         (vminvq_u16): Likewise.
791         (vminvq_u32): Likewise.
792         (vminnmv_f32): Likewise.
793         (vminnmvq_f32): Likewise.
794         (vminnmvq_f64): Likewise.
796 2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
798         * config/aarch64/aarch64-simd.md
799         (aarch64_dup_lane<mode>): Correct lane number on big-endian.
800         (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
801         (*aarch64_mul3_elt<mode>): Likewise.
802         (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
803         (*aarch64_mul3_elt_to_64v2df): Likewise.
804         (*aarch64_mla_elt<mode>): Likewise.
805         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
806         (*aarch64_mls_elt<mode>): Likewise.
807         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
808         (*aarch64_fma4_elt<mode>): Likewise.
809         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
810         (*aarch64_fma4_elt_to_64v2df): Likewise.
811         (*aarch64_fnma4_elt<mode>): Likewise.
812         (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
813         (*aarch64_fnma4_elt_to_64v2df): Likewise.
814         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
815         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
816         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
817         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
818         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
819         (aarch64_sqdmull_lane<mode>_internal): Likewise.
820         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
822 2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>
824         * config/aarch64/aarch64-simd.md
825         (aarch64_be_checked_get_lane<mode>): New define_expand.
826         * config/aarch64/aarch64-simd-builtins.def
827         (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
828         New builtin definition.
829         * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
830         Use new safe be builtin.
832 2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>
834         * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
835         New define_insn.
836         (aarch64_be_st1<mode>): Likewise.
837         (aarch_ld1<VALL:mode>): Define_expand modified.
838         (aarch_st1<VALL:mode>): Likewise.
839         * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
840         (UNSPEC_ST1): Likewise.
842 2014-01-23  David Holsgrove <david.holsgrove@xilinx.com>
844         * config/microblaze/microblaze.md: Add trap insn and attribute
846 2014-01-23  Dodji Seketeli  <dodji@redhat.com>
848         PR preprocessor/58580
849         * input.h (location_get_source_line): Take an additional line_size
850         parameter.
851         (void diagnostics_file_cache_fini): Declare new function.
852         * input.c (struct fcache): New type.
853         (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
854         New static constants.
855         (diagnostic_file_cache_init, total_lines_num)
856         (lookup_file_in_cache_tab, evicted_cache_tab_entry)
857         (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
858         (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
859         (get_next_line, read_next_line, goto_next_line, read_line_num):
860         New static function definitions.
861         (diagnostic_file_cache_fini): New function.
862         (location_get_source_line): Take an additional output line_len
863         parameter.  Re-write using lookup_or_add_file_to_cache_tab and
864         read_line_num.
865         * diagnostic.c (diagnostic_finish): Call
866         diagnostic_file_cache_fini.
867         (adjust_line): Take an additional input parameter for the length
868         of the line, rather than calculating it with strlen.
869         (diagnostic_show_locus): Adjust the use of
870         location_get_source_line and adjust_line with respect to their new
871         signature.  While displaying a line now, do not stop at the first
872         null byte.  Rather, display the zero byte as a space and keep
873         going until we reach the size of the line.
874         * Makefile.in: Add vec.o to OBJS-libcommon
876 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
877             Ilya Tocar     <ilya.tocar@intel.com>
879         * config/i386/avx512fintrin.h (_mm512_kmov): New.
880         * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
881         (__builtin_ia32_kmov16): Ditto.
882         * config/i386/i386.md (UNSPEC_KMOV): New.
883         (kmovw): Ditto.
885 2014-01-23  Kirill Yukhin  <kirill.yukhin@intel.com>
887         * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
888         (_mm512_storeu_si512): Ditto.
890 2014-01-23  Richard Sandiford  <rdsandiford@googlemail.com>
892         PR target/52125
893         * rtl.h (get_referenced_operands): Declare.
894         * recog.c (get_referenced_operands): New function.
895         * config/mips/mips.c (mips_reorg_process_insns): Check which asm
896         operands have been referenced when recording LO_SUM references.
898 2014-01-22  David Holsgrove <david.holsgrove@xilinx.com>
900         * config/microblaze/microblaze.md: Correct bswaphi2 insn.
902 2014-01-22  Jan Hubicka  <jh@suse.cz>
904         * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
905         Enable for generic and recent AMD targets.
907 2014-01-22  Jan Hubicka  <jh@suse.cz>
909         * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
910         ARG_SIZE note when adjustment was eliminated.
912 2014-01-22  Jeff Law  <law@redhat.com>
914         PR tree-optimization/59597
915         * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
916         in file.  Accept new argument REGISTERING and use it to modify
917         dump output appropriately.
918         (register_jump_thread): Corresponding changes.
919         (mark_threaded_blocks): Reinstate code to cancel unprofitable
920         thread paths involving joiner blocks.  Add code to dump cancelled
921         jump threading paths.
923 2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>
925         PR rtl-optimization/59477
926         * lra-constraints.c (inherit_in_ebb): Process call for living hard
927         regs.  Update reloads_num and potential_reload_hard_regs for all insns.
929 2014-01-22  Tom Tromey  <tromey@redhat.com>
931         * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
932         PARAMS.
933         * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
935 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
937         PR rtl-optimization/59896
938         * lra-constraints.c (process_alt_operands): Check unused note for
939         matched operands of insn with no output reloads.
941 2014-01-21  Richard Sandiford  <rdsandiford@googlemail.com>
943         * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
944         (mips_move_from_gpr_cost): Likewise.
946 2014-01-21  Vladimir Makarov  <vmakarov@redhat.com>
948         PR rtl-optimization/59858
949         * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
950         ira_class_hard_regs_num.
951         (process_alt_operands): Increase reject for dying matched operand.
953 2014-01-21  Jakub Jelinek  <jakub@redhat.com>
955         PR target/59003
956         * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
957         smaller than size, perform several stores or loads and stores
958         at dst + count - size to store or copy all of size bytes, rather
959         than just last modesize bytes.
961 2014-01-20  DJ Delorie  <dj@redhat.com>
963         * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
964         that CLOBBERs are REGs before propogating their values.
966 2014-01-20  H.J. Lu  <hongjiu.lu@intel.com>
968         PR middle-end/59789
969         * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
970         (cgraph_inline_failed_type): New function.
971         * cgraph.h (DEFCIFCODE): Add type.
972         (cgraph_inline_failed_type_t): New enum.
973         (cgraph_inline_failed_type): New prototype.
974         * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
975         FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
976         FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
977         LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
978         MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
979         RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
980         OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
981         INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
982         Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
983         FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
984         EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
985         OPTIMIZATION_MISMATCH.
986         * tree-inline.c (expand_call_inline): Emit errors during
987         early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
989 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
991         PR target/59685
992         * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
993         mode attribute in insn output.
995 2014-01-20  Eric Botcazou  <ebotcazou@adacore.com>
997         * output.h (output_constant): Delete.
998         * varasm.c (output_constant): Make private.
1000 2014-01-20  Alex Velenko  <Alex.Velenko@arm.com>
1002         * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1004 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
1006         PR middle-end/59860
1007         * tree.h (fold_builtin_strcat): New prototype.
1008         * builtins.c (fold_builtin_strcat): No longer static.  Add len
1009         argument, if non-NULL, don't call c_strlen.  Optimize
1010         directly into __builtin_memcpy instead of __builtin_strcpy.
1011         (fold_builtin_2): Adjust fold_builtin_strcat caller.
1012         * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1014 2014-01-20  Uros Bizjak  <ubizjak@gmail.com>
1016         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1017         for SImode_address_operand operands, having only a REG argument.
1019 2014-01-20  Marcus Shawcroft  <marcus.shawcroft@arm.com>
1021         * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1022         loader name using mbig-endian.
1023         (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1025 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
1027         * doc/invoke.texi (-march): Clarify documentation for AArch64.
1028         (-mtune): Likewise.
1029         (-mcpu): Likewise.
1031 2014-01-20  Tejas Belagod  <tejas.belagod@arm.com>
1033         * config/aarch64/aarch64-protos.h
1034         (aarch64_cannot_change_mode_class_ptr): Declare.
1035         * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1036         aarch64_cannot_change_mode_class_ptr): New.
1037         * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1038         backend hook aarch64_cannot_change_mode_class.
1040 2014-01-20  James Greenhalgh  <james.greenhalgh@arm.com>
1042         * common/config/aarch64/aarch64-common.c
1043         (aarch64_handle_option): Don't handle any option order logic here.
1044         * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1045         selected_cpu, warn on architecture version mismatch.
1046         (aarch64_override_options): Fix parsing order for option strings.
1048 2014-01-20  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1049             Iain Sandoe  <iain@codesourcery.com>
1051         PR bootstrap/59496
1052         * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1053         warning.  Amend comment to reflect current functionality.
1055 2014-01-20  Richard Biener  <rguenther@suse.de>
1057         PR middle-end/59860
1058         * builtins.c (fold_builtin_strcat): Remove case better handled
1059         by tree-ssa-strlen.c.
1061 2014-01-20  Alan Lawrence  <alan.lawrence@arm.com>
1063         * config/aarch64/aarch64.opt
1064         (mcpu, march, mtune): Make case-insensitive.
1066 2014-01-20  Jakub Jelinek  <jakub@redhat.com>
1068         PR target/59880
1069         * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1070         if operands[1] is a REG or ZERO_EXTEND of a REG.
1072 2014-01-19  Jan Hubicka  <jh@suse.cz>
1074         * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1076 2014-01-19  John David Anglin  <danglin@gcc.gnu.org>
1078         * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1079         long non-pic millicode calls.
1081 2014-01-19  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
1083         * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1085 2014-01-19  Kito Cheng  <kito@0xlab.org>
1087         * builtins.c (expand_movstr): Check movstr expand done or fail.
1089 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
1090             H.J. Lu  <hongjiu.lu@intel.com>
1092         PR target/59379
1093         * config/i386/i386.md (*lea<mode>): Zero-extend return register
1094         to DImode for zero-extended addresses.
1096 2014-01-19  Jakub Jelinek  <jakub@redhat.com>
1098         PR rtl-optimization/57763
1099         * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1100         on the new indirect jump_insn and increment LABEL_NUSES (label).
1102 2014-01-18  H.J. Lu  <hongjiu.lu@intel.com>
1104         PR bootstrap/59580
1105         PR bootstrap/59583
1106         * config.gcc (x86_archs): New variable.
1107         (x86_64_archs): Likewise.
1108         (x86_cpus): Likewise.
1109         Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1110         --with-arch/--with-cpu= options.
1111         Support --with-arch=/--with-cpu={nehalem,westmere,
1112         sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1114 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
1116         * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1117         and PROCESSOR_ATHLON to simplify code.  Move "memory" calculation.
1119 2014-01-18  Uros Bizjak  <ubizjak@gmail.com>
1121         * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1123 2014-01-18  Jakub Jelinek  <jakub@redhat.com>
1125         PR target/58944
1126         * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1127         clear cpp_get_options (parse_in)->warn_unused_macros for
1128         ix86_target_macros_internal with cpp_define.
1130 2014-01-18  Richard Sandiford  <rdsandiford@googlemail.com>
1132         * jump.c (delete_related_insns): Keep (use (insn))s.
1133         * reorg.c (redundant_insn): Check for barriers too.
1135 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
1137         * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1139 2014-01-17  John David Anglin  <danglin@gcc.gnu.org>
1141         * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1142         call to $$dyncall when TARGET_LONG_CALLS is true.
1144 2014-01-17  Jeff Law  <law@redhat.com>
1146         * ree.c (combine_set_extension): Temporarily disable test for
1147         changing number of hard registers.
1149 2014-01-17  Jan Hubicka  <jh@suse.cz>
1151         PR middle-end/58125
1152         * ipa-inline-analysis.c (inline_free_summary):
1153         Do not free summary of aliases.
1155 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
1157         PR middle-end/59706
1158         * gimplify.c (gimplify_expr): Use create_tmp_var
1159         instead of create_tmp_var_raw.  If cond doesn't have
1160         integral type, don't add the IFN_ANNOTATE builtin at all.
1162 2014-01-17  Martin Jambor  <mjambor@suse.cz>
1164         PR ipa/59736
1165         * ipa-cp.c (prev_edge_clone): New variable.
1166         (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1167         Also resize prev_edge_clone vector.
1168         (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1169         (ipcp_edge_removal_hook): New function.
1170         (ipcp_driver): Register ipcp_edge_removal_hook.
1172 2014-01-17  Andrew Pinski <apinski@cavium.com>
1173             Steve Ellcey  <sellcey@mips.com>
1175         PR target/59462
1176         * config/mips/mips.c (mips_print_operand): Check operand mode instead
1177         of operator mode.
1179 2014-01-17  Jeff Law  <law@redhat.com>
1181         PR middle-end/57904
1182         * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1183         so that pass_ccp runs first.
1185 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
1187         * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1188         (ix86_adjust_cost): Use !TARGET_XXX.
1189         (do_reorder_for_imul): Likewise.
1190         (swap_top_of_ready_list): Likewise.
1191         (ix86_sched_reorder): Likewise.
1193 2014-01-17  H.J. Lu  <hongjiu.lu@intel.com>
1195         * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1196         PROCESSOR_INTEL.  Treat like PROCESSOR_GENERIC.
1197         * config/i386/i386.c (intel_memcpy): New.  Duplicate slm_memcpy.
1198         (intel_memset): New.  Duplicate slm_memset.
1199         (intel_cost): New.  Duplicate slm_cost.
1200         (m_INTEL): New macro.
1201         (processor_target_table): Add "intel".
1202         (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1203         with PROCESSOR_INTEL for "intel".
1204         (ix86_lea_outperforms): Support PROCESSOR_INTEL.  Duplicate
1205         PROCESSOR_SILVERMONT.
1206         (ix86_issue_rate): Likewise.
1207         (ix86_adjust_cost): Likewise.
1208         (ia32_multipass_dfa_lookahead): Likewise.
1209         (swap_top_of_ready_list): Likewise.
1210         (ix86_sched_reorder): Likewise.
1211         (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1212         instead of TARGET_OPT_AGU.
1213         * config/i386/i386.h (TARGET_INTEL): New.
1214         (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1215         (processor_type): Add PROCESSOR_INTEL.
1216         * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1217         Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1219 2014-01-17  Marek Polacek  <polacek@redhat.com>
1221         PR c/58346
1222         * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1223         size is zero.
1225 2014-01-17  Richard Biener  <rguenther@suse.de>
1227         PR tree-optimization/46590
1228         * opts.c (default_options_table): Add entries for
1229         OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1230         all enabled at -O1 but not for -Og.
1231         * common.opt (fbranch-count-reg): Remove Init(1).
1232         (fmove-loop-invariants): Likewise.
1233         (ftree-pta): Likewise.
1235 2014-01-17  Jakub Jelinek  <jakub@redhat.com>
1237         * config/i386/i386.c (ix86_data_alignment): For compatibility with
1238         (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1239         decls to at least the GCC 4.8 used alignments.
1241         PR fortran/59440
1242         * tree-nested.c (convert_nonlocal_reference_stmt,
1243         convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1244         of GIMPLE_BIND stmts, adjust associated decls.
1246 2014-01-17  Richard Biener  <rguenther@suse.de>
1248         PR tree-optimization/46590
1249         * vec.h (vec<>::bseach): New member function implementing
1250         binary search according to C89 bsearch.
1251         (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1252         * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1253         bitmap pointer again.  Make accesses_in_loop a flat array.
1254         (mem_ref_obstack): New global.
1255         (outermost_indep_loop): Adjust for mem_ref->stored changes.
1256         (mark_ref_stored): Likewise.
1257         (ref_indep_loop_p_2): Likewise.
1258         (set_ref_stored_in_loop): New helper function.
1259         (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1260         (memref_free): Adjust.
1261         (record_mem_ref_loc): Simplify.
1262         (gather_mem_refs_stmt): Adjust.
1263         (sort_locs_in_loop_postorder_cmp): New function.
1264         (analyze_memory_references): Sort accesses_in_loop after
1265         loop postorder number.
1266         (find_ref_loc_in_loop_cmp): New function.
1267         (for_all_locs_in_loop): Find relevant cluster of locs in
1268         accesses_in_loop and iterate without recursion.
1269         (execute_sm): Avoid uninit warning.
1270         (struct ref_always_accessed): Simplify.
1271         (ref_always_accessed::operator ()): Likewise.
1272         (ref_always_accessed_p): Likewise.
1273         (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1274         loop postorder numbers here.
1275         (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1276         numbers.
1278 2014-01-17  Jan Hubicka  <hubicka@ucw.cz>
1280         PR c++/57945
1281         * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1282         on decls for which assemble_alias has been called.
1284 2014-01-17  Nick Clifton  <nickc@redhat.com>
1286         * config/msp430/msp430.opt: (mcpu): New option.
1287         * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1288         (msp430_option_override): Parse target_cpu.  If the MCU name
1289         matches a generic string, clear target_mcu.
1290         (msp430_attr): Allow numeric interrupt values up to 63.
1291         (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1292         * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1293         option.
1294         * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1295         Add mcpu matches.
1296         * config/msp430/msp430.md (popm): Use %J rather than %I.
1297         (addsi3): Use msp430_nonimmediate_operand for operand 2.
1298         (addhi_cy_i): Use immediate_operand for operand 2.
1299         * doc/invoke.texi: Document -mcpu option.
1301 2014-01-17  Richard Biener  <rguenther@suse.de>
1303         PR rtl-optimization/38518
1304         * df.h (df_analyze_loop): Declare.
1305         * df-core.c: Include cfgloop.h.
1306         (df_analyze_1): Split out main part of df_analyze.
1307         (df_analyze): Adjust.
1308         (loop_inverted_post_order_compute): New function.
1309         (loop_post_order_compute): Likewise.
1310         (df_analyze_loop): New function avoiding whole-function
1311         postorder computes.
1312         * loop-invariant.c (find_defs): Use df_analyze_loop.
1313         (find_invariants): Adjust.
1314         * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1316 2014-01-17  Zhenqiang Chen  <zhenqiang.chen@arm.com>
1318         * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1319         (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1321 2014-01-16  Ilya Enkovich  <ilya.enkovich@intel.com>
1323         * ipa-ref.c (ipa_remove_stmt_references): Fix references
1324         traversal when removing references.
1326 2014-01-16  Jan Hubicka  <jh@suse.cz>
1328         PR ipa/59775
1329         * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1331 2014-01-16  Bernd Schmidt  <bernds@codesourcery.com>
1333         PR middle-end/56791
1334         * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1335         pushing a reload for an autoinc when we had previously reloaded an
1336         inner part of the address.
1338 2014-01-16  Jakub Jelinek  <jakub@redhat.com>
1340         * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1341         field.
1342         (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1343         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1344         when not giving up or versioning for alias only because of
1345         loop->safelen.
1346         (vect_analyze_data_ref_dependences): Set to true.
1347         * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1348         is a GIMPLE_PHI.
1349         (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1350         LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1351         to the condition.
1353         PR middle-end/58344
1354         * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1356         PR target/59839
1357         * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1358         operand 0 predicate for gathers, use a new pseudo as subtarget.
1360 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
1362         PR middle-end/59609
1363         * lra-constraints.c (process_alt_operands): Add printing debug info.
1364         Check absence of input/output reloads for matched operands too.
1366 2014-01-16  Vladimir Makarov  <vmakarov@redhat.com>
1368         PR rtl-optimization/59835
1369         * ira.c (ira_init_register_move_cost): Increase cost for
1370         impossible modes.
1372 2014-01-16  Alan Lawrence  <alan.lawrence@arm.com>
1374         * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1376 2014-01-16  Richard Earnshaw  <rearnsha@arm.com>
1378         PR target/59780
1379         * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1380         non-register objects.  Use gen_(high/low)part more consistently.
1381         Fix assertions.
1383 2014-01-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
1385         PR target/59844
1386         * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1387         endian support, remove tests for WORDS_BIG_ENDIAN.
1388         (p8_mfvsrd_3_<mode>): Likewise.
1389         (reload_gpr_from_vsx<mode>): Likewise.
1390         (reload_gpr_from_vsxsf): Likewise.
1391         (p8_mfvsrd_4_disf): Likewise.
1393 2014-01-16  Richard Biener  <rguenther@suse.de>
1395         PR rtl-optimization/46590
1396         * lcm.c (compute_antinout_edge): Use postorder iteration.
1397         (compute_laterin): Use inverted postorder iteration.
1399 2014-01-16  Nick Clifton  <nickc@redhat.com>
1401         PR middle-end/28865
1402         * varasm.c (output_constant): Return the number of bytes actually
1403         emitted.
1404         (output_constructor_array_range): Update the field size with the
1405         number of bytes emitted by output_constant.
1406         (output_constructor_regular_field): Likewise.  Also do not
1407         complain if the total number of bytes emitted is now greater
1408         than the expected fieldpos.
1409         * output.h (output_constant): Update prototype and descriptive comment.
1411 2014-01-16  Marek Polacek  <polacek@redhat.com>
1413         PR middle-end/59827
1414         * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1415         it is error_mark_node.
1417 2014-01-15  Uros Bizjak  <ubizjak@gmail.com>
1419         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1420         VALID_AVX256_REG_OR_OI_MODE.
1422 2014-01-15  Pat Haugen  <pthaugen@us.ibm.com>
1424         * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1425         current procedure should be profiled.
1427 2014-01-15  Andrew Pinski  <apinski@cavium.com>
1429         * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1430         of moving from/to the STACK_REG register class.
1432 2014-01-15  Richard Henderson <rth@redhat.com>
1434         PR debug/54694
1435         * reginfo.c (global_regs_decl): Globalize.
1436         * rtl.h (global_regs_decl): Declare.
1437         * ira.c (do_reload): Diagnose frame_pointer_needed and it
1438         reserved via global_regs.
1440 2014-01-15  Teresa Johnson  <tejohnson@google.com>
1442         * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
1444 2014-01-15  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
1446         * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
1447         and vmulosh rather than call gen_vec_widen_smult_*.
1448         (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
1449         than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
1450         (vec_widen_smult_even_v16qi): Likewise.
1451         (vec_widen_umult_even_v8hi): Likewise.
1452         (vec_widen_smult_even_v8hi): Likewise.
1453         (vec_widen_umult_odd_v16qi): Likewise.
1454         (vec_widen_smult_odd_v16qi): Likewise.
1455         (vec_widen_umult_odd_v8hi): Likewise.
1456         (vec_widen_smult_odd_v8hi): Likewise.
1457         (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
1458         vmuloub rather than call gen_vec_widen_umult_*.
1459         (vec_widen_umult_lo_v16qi): Likewise.
1460         (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
1461         vmulosb rather than call gen_vec_widen_smult_*.
1462         (vec_widen_smult_lo_v16qi): Likewise.
1463         (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
1464         rather than call gen_vec_widen_umult_*.
1465         (vec_widen_umult_lo_v8hi): Likewise.
1466         (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
1467         rather than call gen_vec_widen_smult_*.
1468         (vec_widen_smult_lo_v8hi): Likewise.
1470 2014-01-15  Jeff Law  <law@redhat.com>
1472         PR tree-optimization/59747
1473         * ree.c (find_and_remove_re): Properly handle case where a second
1474         eliminated extension requires widening a copy created for elimination
1475         of a prior extension.
1476         (combine_set_extension): Ensure that the number of hard regs needed
1477         for a destination register does not change when we widen it.
1479 2014-01-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
1481         * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
1482         (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
1483         (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
1484         (arm*-*-rtems*): Use t-rtems from existing tmake_file.
1485         (avr-*-rtems*): Likewise.
1486         (bfin*-rtems*): Likewise.
1487         (moxie-*-rtems*): Likewise.
1488         (h8300-*-rtems*): Likewise.
1489         (i[34567]86-*-rtems*): Likewise.
1490         (lm32-*-rtems*): Likewise.
1491         (m32r-*-rtems*): Likewise.
1492         (m68k-*-rtems*): Likewise.
1493         (microblaze*-*-rtems*): Likewise.
1494         (mips*-*-rtems*): Likewise.
1495         (powerpc-*-rtems*): Likewise.
1496         (sh-*-rtems*): Likewise.
1497         (sparc-*-rtems*): Likewise.
1498         (sparc64-*-rtems*): Likewise.
1499         (v850-*-rtems*): Likewise.
1500         (m32c-*-rtems*): Likewise.
1502 2014-01-15  Vladimir Makarov  <vmakarov@redhat.com>
1504         PR rtl-optimization/59511
1505         * ira.c (ira_init_register_move_cost): Use memory costs for some
1506         cases of register move cost calculations.
1507         * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
1508         instead of BB frequency.
1509         * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
1510         * lra-assigns.c (find_hard_regno_for): Ditto.
1512 2014-01-15  Richard Biener  <rguenther@suse.de>
1514         PR tree-optimization/59822
1515         * tree-vect-stmts.c (hoist_defs_of_uses): New function.
1516         (vectorizable_load): Use it to hoist defs of uses of invariant
1517         loads out of the loop.
1519 2014-01-15  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
1520             Kugan Vivekanandarajah  <kuganv@linaro.org>
1522         PR target/59695
1523         * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
1524         truncation.
1526 2014-01-15  Richard Biener  <rguenther@suse.de>
1528         PR rtl-optimization/59802
1529         * lcm.c (compute_available): Use inverted postorder to seed
1530         the initial worklist.
1532 2014-01-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1534         PR target/59803
1535         * config/s390/s390.c (s390_preferred_reload_class): Don't return
1536         ADDR_REGS for invalid symrefs in non-PIC code.
1538 2014-01-15  Jakub Jelinek  <jakub@redhat.com>
1540         PR other/58712
1541         * builtins.c (determine_block_size): Initialize *probable_max_size
1542         even if len_rtx is CONST_INT.
1544 2014-01-14  Andrew Pinski  <apinski@cavium.com>
1546         * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
1547         * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
1548         (cortexa53_tunings): Likewise.
1549         (aarch64_sched_issue_rate): New function.
1550         (TARGET_SCHED_ISSUE_RATE): Define.
1552 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
1554         * ira-costs.c (find_costs_and_classes): Add missed
1555         ira_init_register_move_cost_if_necessary.
1557 2014-01-14  Vladimir Makarov  <vmakarov@redhat.com>
1559         PR target/59787
1560         * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
1562 2014-01-14  H.J. Lu  <hongjiu.lu@intel.com>
1564         PR target/59794
1565         * config/i386/i386.c (type_natural_mode): Add a bool parameter
1566         to indicate if type is used for function return value.  Warn ABI
1567         change if the vector mode isn't available for function return value.
1568         (ix86_function_arg_advance): Pass false to type_natural_mode.
1569         (ix86_function_arg): Likewise.
1570         (ix86_gimplify_va_arg): Likewise.
1571         (function_arg_32): Don't warn ABI change.
1572         (ix86_function_value): Pass true to type_natural_mode.
1573         (ix86_return_in_memory): Likewise.
1574         (ix86_struct_value_rtx): Removed.
1575         (TARGET_STRUCT_VALUE_RTX): Likewise.
1577 2014-01-14  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
1579         * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
1580         converting a conditional jump into a conditional return.
1582 2014-01-14  Richard Biener  <rguenther@suse.de>
1584         PR tree-optimization/58921
1585         PR tree-optimization/59006
1586         * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
1587         hoisting invariant stmts.
1588         * tree-vect-stmts.c (vectorizable_load): Insert the splat of
1589         invariant loads on the preheader edge if possible.
1591 2014-01-14  Joey Ye  <joey.ye@arm.com>
1593         * doc/plugin.texi (Building GCC plugins): Update to C++.
1595 2014-01-14  Kirill Yukhin  <kirill.yukhin@intel.com>
1597         * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
1598         (_mm_rcp28_round_ss): Ditto.
1599         (_mm_rsqrt28_round_sd): Ditto.
1600         (_mm_rsqrt28_round_ss): Ditto.
1601         (_mm_rcp28_sd): Ditto.
1602         (_mm_rcp28_ss): Ditto.
1603         (_mm_rsqrt28_sd): Ditto.
1604         (_mm_rsqrt28_ss): Ditto.
1605         * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
1606         * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
1607         * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
1608         (IX86_BUILTIN_RCP28SD): Ditto.
1609         (IX86_BUILTIN_RCP28SS): Ditto.
1610         (IX86_BUILTIN_RSQRT28SD): Ditto.
1611         (IX86_BUILTIN_RSQRT28SS): Ditto.
1612         (bdesc_special_args): Define __builtin_ia32_movntdqa512,
1613         __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
1614         __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
1615         (ix86_expand_special_args_builtin): Expand new FTYPE.
1616         * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
1617         (srcp14<mode>): Make insn unary.
1618         (avx512f_vmscalef<mode><round_name>): Use substed predicate.
1619         (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
1620         (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
1621         (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
1622         (avx512er_exp2<mode><mask_name><round_saeonly_name>):
1623         Fix rounding: make it SAE only.
1624         (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1625         Ditto.
1626         (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1627         Ditto.
1628         (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
1629         (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
1630         (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
1631         * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
1632         (round_saeonly_mask_scalar_operand4): Ditto.
1633         (round_saeonly_mask_scalar_op3): Ditto.
1634         (round_saeonly_mask_scalar_op4): Ditto.
1636 2014-01-13  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1638         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1639         Implement -maltivec=be for vec_insert and vec_extract.
1641 2014-01-10  DJ Delorie  <dj@redhat.com>
1643         * config/msp430/msp430.md (call_internal): Don't allow memory
1644         references with SP as the base register.
1645         (call_value_internal): Likewise.
1646         * config/msp430/constraints.md (Yc): New.  For memory references
1647         that don't use SP as a base register.
1649         * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
1650         "an integer without a # prefix"
1651         * config/msp430/msp430.md (epilogue_helper): Use it.
1653 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
1655         PR target/59617
1656         * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
1657         AVX512F gather builtins.
1658         * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
1659         on gather decls with INTEGER_TYPE masktype.
1660         (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
1661         directly into the builtin rather than hoisting it before loop.
1663         PR tree-optimization/59387
1664         * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
1665         (scev_const_prop): If folded_casts and type has undefined overflow,
1666         use force_gimple_operand instead of force_gimple_operand_gsi and
1667         for each added stmt if it is assign with
1668         arith_code_with_undefined_signed_overflow, call
1669         rewrite_to_defined_overflow.
1670         * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
1671         gimple-fold.h instead.
1672         (arith_code_with_undefined_signed_overflow,
1673         rewrite_to_defined_overflow): Moved to ...
1674         * gimple-fold.c (arith_code_with_undefined_signed_overflow,
1675         rewrite_to_defined_overflow): ... here.  No longer static.
1676         Include gimplify-me.h.
1677         * gimple-fold.h (arith_code_with_undefined_signed_overflow,
1678         rewrite_to_defined_overflow): New prototypes.
1680 2014-01-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1682         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
1684 2014-01-13  Eric Botcazou  <ebotcazou@adacore.com>
1686         * builtins.c (get_object_alignment_2): Minor tweak.
1687         * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
1689 2014-01-13  Christian Bruel  <christian.bruel@st.com>
1691         * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
1692         optimized non constant lengths.
1694 2014-01-13  Jakub Jelinek  <jakub@redhat.com>
1696         PR libgomp/59194
1697         * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
1698         load as __atomic_load_N if possible.
1700 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
1702         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
1703         target parameter.
1704         (rs6000_expand_builtin): Adjust call.
1706 2014-01-11  David Edelsohn  <dje.gcc@gmail.com>
1708         PR target/58115
1709         * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
1710         * config/rs6000/rs6000.c: Include target-globals.h.
1711         (rs6000_set_current_function): Instead of doing target_reinit
1712         unconditionally, use save_target_globals_default_opts and
1713         restore_target_globals.
1715         * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
1716         FPSCR.
1717         * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
1718         (rs6000_expand_builtin): Handle mffs and mtfsf.
1719         (rs6000_init_builtins): Define mffs and mtfsf.
1720         * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
1721         (rs6000_mffs): New pattern.
1722         (rs6000_mtfsf): New pattern.
1724 2014-01-11  Bin Cheng  <bin.cheng@arm.com>
1726         * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
1727         Start narrowing with START.  Apply candidate-use pair
1728         and check overall cost in narrowing.
1729         (iv_ca_prune): Pass new argument.
1731 2014-01-10  Jeff Law  <law@redhat.com>
1733         PR middle-end/59743
1734         * ree.c (combine_reaching_defs): Ensure the defining statement
1735         occurs before the extension when optimizing extensions with
1736         different source and destination hard registers.
1738 2014-01-10  Jan Hubicka  <jh@suse.cz>
1740         PR ipa/58585
1741         * ipa-devirt.c (build_type_inheritance_graph): Also add types of
1742         vtables into the type inheritance graph.
1744 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
1746         PR rtl-optimization/59754
1747         * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
1748         modes in the REGNO != REGNO case.
1750 2014-01-10  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1752         * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
1754 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
1756         PR tree-optimization/59745
1757         * tree-predcom.c (tree_predictive_commoning_loop): Call
1758         free_affine_expand_cache if giving up because components is NULL.
1760         * target-globals.c (save_target_globals): Allocate < 4KB structs using
1761         GC in payload of target_globals struct instead of allocating them on
1762         the heap and the larger structs separately using GC.
1763         * target-globals.h (struct target_globals): Make regs, hard_regs,
1764         reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
1765         of GTY((skip)) and change type to void *.
1766         (reset_target_globals): Cast loads from those fields to corresponding
1767         types.
1769 2014-01-10  Steve Ellcey  <sellcey@mips.com>
1771         PR plugins/59335
1772         * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
1773         gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
1774         tree-ssanames.h, print-tree.h, varasm.h, and context.h.
1776 2014-01-10  Richard Earnshaw  <rearnsha@arm.com>
1778         PR target/59744
1779         * aarch64-modes.def (CC_Zmode): New flags mode.
1780         * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
1781         represents an equality.
1782         (aarch64_get_condition_code): Handle CC_Zmode.
1783         * aarch64.md (compare_neg<mode>): Restrict to equality operations.
1785 2014-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
1787         * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
1788         extraction in good case.
1790 2014-01-10  Richard Biener  <rguenther@suse.de>
1792         PR tree-optimization/59374
1793         * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
1794         checking after SLP discovery.  Mark stmts not participating
1795         in any SLP instance properly.
1797 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1799         * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
1800         when handling a SET rtx.
1802 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1804         * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
1805         (cortex-a57): Likewise.
1806         (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
1808 2014-01-10  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1810         * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
1811         non-iwmmxt builtins.
1813 2014-01-10  Jan Hubicka  <hubicka@ucw.cz>
1815         PR ipa/58252
1816         PR ipa/59226
1817         * ipa-devirt.c record_target_from_binfo): Take as argument
1818         stack of binfos and lookup matching one for virtual inheritance.
1819         (possible_polymorphic_call_targets_1): Update.
1821 2014-01-10  Huacai Chen  <chenhc@lemote.com>
1823         * config/mips/driver-native.c (host_detect_local_cpu): Handle new
1824         kernel strings for Loongson-2E/2F/3A.
1826 2014-01-10  Jakub Jelinek  <jakub@redhat.com>
1828         PR middle-end/59670
1829         * tree-vect-data-refs.c (vect_analyze_data_refs): Check
1830         is_gimple_call before calling gimple_call_internal_p.
1832 2014-01-09  Steve Ellcey  <sellcey@mips.com>
1834         * Makefile.in (TREE_FLOW_H): Remove.
1835         (TREE_SSA_H): Add file names from tree-flow.h.
1836         * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
1837         * tree.h: Remove tree-flow.h reference.
1838         * hash-table.h: Remove tree-flow.h reference.
1839         * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
1840         reference with tree-ssa-loop.h.
1842 2014-01-09  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1844         * doc/invoke.texi: Add -maltivec={be,le} options, and document
1845         default element-order behavior for -maltivec.
1846         * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
1847         * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
1848         that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
1849         when targeting big endian, at least for now.
1850         * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
1852 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
1854         PR middle-end/47735
1855         * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
1856         var satisfies use_register_for_decl, just take into account type
1857         alignment, rather than decl alignment.
1859         PR tree-optimization/59622
1860         * gimple-fold.c (gimple_fold_call): Fix a typo in message.  For
1861         __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
1862         __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
1863         Don't devirtualize for inplace at all.  For targets.length () == 1,
1864         if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
1866 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
1868         * config/i386/i386.md (cpu): Remove the unused btver1.
1870 2014-01-09  H.J. Lu  <hongjiu.lu@intel.com>
1872         * gdbasan.in: Put a breakpoint on __sanitizer::Report.
1874 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
1876         PR target/58115
1877         * tree-core.h (struct target_globals): New forward declaration.
1878         (struct tree_target_option): Add globals field.
1879         * tree.h (TREE_TARGET_GLOBALS): Define.
1880         (prepare_target_option_nodes_for_pch): New prototype.
1881         * target-globals.h (struct target_globals): Define even if
1882         !SWITCHABLE_TARGET.
1883         * tree.c (prepare_target_option_node_for_pch,
1884         prepare_target_option_nodes_for_pch): New functions.
1885         * config/i386/i386.h (SWITCHABLE_TARGET): Define.
1886         * config/i386/i386.c: Include target-globals.h.
1887         (ix86_set_current_function): Instead of doing target_reinit
1888         unconditionally, use save_target_globals_default_opts and
1889         restore_target_globals.
1891 2014-01-09  Richard Biener  <rguenther@suse.de>
1893         PR tree-optimization/59715
1894         * tree-cfg.h (split_critical_edges): Declare.
1895         * tree-cfg.c (split_critical_edges): Export.
1896         * tree-ssa-sink.c (execute_sink_code): Split critical edges.
1898 2014-01-09  Max Ostapenko  <m.ostapenko@partner.samsung.com>
1900         * cfgexpand.c (expand_stack_vars): Optionally disable
1901         asan stack protection.
1902         (expand_used_vars): Likewise.
1903         (partition_stack_vars): Likewise.
1904         * asan.c (asan_emit_stack_protection): Optionally disable
1905         after return stack usage.
1906         (instrument_derefs): Optionally disable memory access instrumentation.
1907         (instrument_builtin_call): Likewise.
1908         (instrument_strlen_call): Likewise.
1909         (asan_protect_global): Optionally disable global variables protection.
1910         * doc/invoke.texi: Added doc for new options.
1911         * params.def: Added new options.
1912         * params.h: Likewise.
1914 2014-01-09  Jakub Jelinek  <jakub@redhat.com>
1916         PR rtl-optimization/59724
1917         * ifcvt.c (cond_exec_process_if_block): Don't call
1918         flow_find_head_matching_sequence with 0 longest_match.
1919         * cfgcleanup.c (flow_find_head_matching_sequence): Count even
1920         non-active insns if !stop_after.
1921         (try_head_merge_bb): Revert 2014-01-07 changes.
1923 2014-01-08  Jeff Law  <law@redhat.com>
1925         * ree.c (get_sub_rtx): New function, extracted from...
1926         (merge_def_and_ext): Here.
1927         (combine_reaching_defs): Use get_sub_rtx.
1929 2014-01-08  Eric Botcazou  <ebotcazou@adacore.com>
1931         * cgraph.h (varpool_variable_node): Do not choke on null node.
1933 2014-01-08  Catherine Moore  <clm@codesourcery.com>
1935         * config/mips/mips.md (simple_return): Attempt to use JRC
1936         for microMIPS.
1937         * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
1939 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
1941         PR rtl-optimization/59137
1942         * reorg.c (steal_delay_list_from_target): Call update_block for
1943         elided insns.
1944         (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
1946 2014-01-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1948         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
1949         two duplicate entries.
1951 2014-01-08  Richard Sandiford  <rdsandiford@googlemail.com>
1953         Revert:
1954         2012-10-07  Richard Sandiford  <rdsandiford@googlemail.com>
1956         * config/mips/mips.c (mips_truncated_op_cost): New function.
1957         (mips_rtx_costs): Adjust test for BADDU.
1958         * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
1960         2012-10-02  Richard Sandiford  <rdsandiford@googlemail.com>
1962         * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
1963         (*baddu_si): ...this new pattern.
1965 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
1967         PR ipa/59722
1968         * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
1970 2014-01-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1972         PR middle-end/57748
1973         * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
1974         inner_reference_p.
1975         (expand_expr, expand_normal): Adjust.
1976         * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
1977         inner_reference_p. Use inner_reference_p to expand inner references.
1978         (store_expr): Adjust.
1979         * cfgexpand.c (expand_call_stmt): Adjust.
1981 2014-01-08  Rong Xu  <xur@google.com>
1983         * gcov-io.c (gcov_var): Move from gcov-io.h.
1984         (gcov_position): Ditto.
1985         (gcov_is_error): Ditto.
1986         (gcov_rewrite): Ditto.
1987         * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
1988         only part to libgcc/libgcov.h.
1990 2014-01-08  Marek Polacek  <polacek@redhat.com>
1992         PR middle-end/59669
1993         * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
1995 2014-01-08  Marek Polacek  <polacek@redhat.com>
1997         PR sanitizer/59667
1998         * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2000 2014-01-08  Jakub Jelinek  <jakub@redhat.com>
2002         PR rtl-optimization/59649
2003         * stor-layout.c (get_mode_bounds): For BImode return
2004         0 and STORE_FLAG_VALUE.
2006 2014-01-08  Richard Biener  <rguenther@suse.de>
2008         PR middle-end/59630
2009         * gimple.h (is_gimple_builtin_call): Remove.
2010         (gimple_builtin_call_types_compatible_p): New.
2011         (gimple_call_builtin_p): New overload.
2012         * gimple.c (is_gimple_builtin_call): Remove.
2013         (validate_call): Rename to ...
2014         (gimple_builtin_call_types_compatible_p): ... this and export.  Also
2015         check return types.
2016         (validate_type): New static function.
2017         (gimple_call_builtin_p): New overload and adjust.
2018         * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2019         (gimple_fold_call): Likewise.  Use gimple_call_builtin_p.
2020         (gimple_fold_stmt_to_constant_1): Likewise.
2021         * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2023 2014-01-08  Richard Biener  <rguenther@suse.de>
2025         PR middle-end/59471
2026         * gimplify.c (gimplify_expr): Gimplify register-register type
2027         VIEW_CONVERT_EXPRs to separate stmts.
2029 2014-01-07  Jeff Law  <law@redhat.com>
2031         PR middle-end/53623
2032         * ree.c (combine_set_extension): Handle case where source
2033         and destination registers in an extension insn are different.
2034         (combine_reaching_defs): Allow source and destination registers
2035         in extension to be different under limited circumstances.
2036         (add_removable_extension): Remove restriction that the
2037         source and destination registers in the extension are the same.
2038         (find_and_remove_re): Emit a copy from the extension's
2039         destination to its source after the defining insn if
2040         the source and destination registers are different.
2042         PR middle-end/59285
2043         * ifcvt.c (merge_if_block): If we are merging a block with more than
2044         one successor with a block with no successors, remove any BARRIER
2045         after the second block.
2047 2014-01-07  Dan Xio Qiang <ziyan01@163.com>
2049         * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2051 2014-01-07  John David Anglin  <danglin@gcc.gnu.org>
2053         PR target/59652
2054         * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2055         for 14-bit register offsets when INT14_OK_STRICT is false.
2057 2014-01-07  Roland Stigge  <stigge@antcom.de>
2058             Michael Meissner  <meissner@linux.vnet.ibm.com>
2060         PR 57386/target
2061         * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2062         Only check TFmode for SPE constants.  Don't check TImode or TDmode.
2064 2014-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
2066         * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2067         -mcpu.
2069 2014-01-07  Yufeng Zhang  <yufeng.zhang@arm.com>
2071         * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2072         with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2073         rtx is const0_rtx or not.
2075 2014-01-07  Richard Sandiford  <rdsandiford@googlemail.com>
2077         PR target/58115
2078         * target-globals.c (save_target_globals): Remove this_fn_optab
2079         handling.
2080         * toplev.c: Include optabs.h.
2081         (target_reinit): Temporarily restore the global options if another
2082         set of options are in force.
2084 2014-01-07  Jakub Jelinek  <jakub@redhat.com>
2086         PR rtl-optimization/58668
2087         * cfgcleanup.c (flow_find_cross_jump): Don't count
2088         any jumps if dir_p is NULL.  Remove p1 variable, use active_insn_p
2089         to determine what is counted.
2090         (flow_find_head_matching_sequence): Use active_insn_p to determine
2091         what is counted.
2092         (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2093         counting change.
2094         * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2095         determine what is counted.
2097         PR tree-optimization/59643
2098         * tree-predcom.c (split_data_refs_to_components): If one dr is
2099         read and one write, determine_offset fails and the write isn't
2100         in the bad component, just put the read into the bad component.
2102 2014-01-07  Mike Stump  <mikestump@comcast.net>
2103             Jakub Jelinek  <jakub@redhat.com>
2105         PR pch/59436
2106         * tree-core.h (struct tree_optimization_option): Change optabs
2107         type from unsigned char * to void *.
2108         * optabs.c (init_tree_optimization_optabs): Adjust
2109         TREE_OPTIMIZATION_OPTABS initialization.
2111 2014-01-06  Jakub Jelinek  <jakub@redhat.com>
2113         PR target/59644
2114         * config/i386/i386.h (struct machine_function): Add
2115         no_drap_save_restore field.
2116         * config/i386/i386.c (ix86_save_reg): Use
2117         !cfun->machine->no_drap_save_restore instead of
2118         crtl->stack_realign_needed.
2119         (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2120         this function clears frame_pointer_needed.  Set
2121         cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2122         and DRAP reg is needed.
2124 2014-01-06  Marek Polacek  <polacek@redhat.com>
2126         PR c/57773
2127         * doc/implement-c.texi: Mention that other integer types are
2128         permitted as bit-field types in strictly conforming mode.
2130 2014-01-06  Felix Yang  <fei.yang0953@gmail.com>
2132         * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2133         is newly allocated.
2135 2014-01-06  Richard Earnshaw  <rearnsha@arm.com>
2137         * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2139 2014-01-06  Martin Jambor  <mjambor@suse.cz>
2141         PR ipa/59008
2142         * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2143         to int.
2144         * ipa-prop.c (ipa_print_node_params): Fix indentation.
2146 2014-01-06  Eric Botcazou  <ebotcazou@adacore.com>
2148         PR debug/59350
2149         PR debug/59510
2150         * var-tracking.c (add_stores): Preserve the value of the source even if
2151         we don't record the store.
2153 2014-01-06  Terry Guo  <terry.guo@arm.com>
2155         * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2157 2014-01-05  Iain Sandoe  <iain@codesourcery.com>
2159         PR bootstrap/59541
2160         * config/darwin.c (darwin_function_section): Adjust return values to
2161         correspond to optimisation changes made in r206070.
2163 2014-01-05  Uros Bizjak  <ubizjak@gmail.com>
2165         * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2166         from prefetch_block tune setting.
2167         (nocona_cost): Correct size of prefetch block to 64.
2169 2014-01-04  Eric Botcazou  <ebotcazou@adacore.com>
2171         * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2172         (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2173         used to save the static chain register in the computation of the offset
2174         from which the FP registers need to be restored.
2176 2014-01-04  Jakub Jelinek  <jakub@redhat.com>
2178         PR tree-optimization/59519
2179         * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2180         ICE if get_current_def (current_new_name) is already non-NULL, as long
2181         as it is a phi result of some other phi in *new_exit_bb that has
2182         the same argument.
2184         * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2185         or vmovdqu* for misaligned_operand.
2186         (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2187         <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2188         * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2189         aligned_mem for AVX512F masked aligned load and store builtins and for
2190         non-temporal moves.
2192 2014-01-03  Bingfeng Mei  <bmei@broadcom.com>
2194         PR tree-optimization/59651
2195         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2196         Address range for negative step should be added by TYPE_SIZE_UNIT.
2198 2014-01-03  Andreas Schwab  <schwab@linux-m68k.org>
2200         * config/m68k/m68k.c (handle_move_double): Handle pushes with
2201         overlapping registers also for registers other than the stack pointer.
2203 2014-01-03  Marek Polacek  <polacek@redhat.com>
2205         PR other/59661
2206         * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2207         __builtin_FILE.
2209 2014-01-03  Jakub Jelinek  <jakub@redhat.com>
2211         PR target/59625
2212         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2213         asm goto as jump.
2215         * config/i386/i386.md (MODE_SIZE): New mode attribute.
2216         (push splitter): Use <P:MODE_SIZE> instead of
2217         GET_MODE_SIZE (<P:MODE>mode).
2218         (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2219         (mov -1, reg peephole2): Likewise.
2220         * config/i386/sse.md (*mov<mode>_internal,
2221         <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2222         <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2223         *<code><mode>3, *andnot<mode>3<mask_name>,
2224         <mask_codefor><code><mode>3<mask_name>): Likewise.
2225         * config/i386/subst.md (mask_mode512bit_condition,
2226         sd_mask_mode512bit_condition): Likewise.
2228 2014-01-02  Xinliang David Li  <davidxl@google.com>
2230         PR tree-optimization/59303
2231         * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2232         (dump_predicates): Better output format.
2233         (pred_equal_p): New function.
2234         (is_neq_relop_p): Ditto.
2235         (is_neq_zero_form_p): Ditto.
2236         (pred_expr_equal_p): Ditto.
2237         (pred_neg_p): Ditto.
2238         (simplify_pred): Ditto.
2239         (simplify_preds_2): Ditto.
2240         (simplify_preds_3): Ditto.
2241         (simplify_preds_4): Ditto.
2242         (simplify_preds): Ditto.
2243         (push_pred): Ditto.
2244         (push_to_worklist): Ditto.
2245         (get_pred_info_from_cmp): Ditto.
2246         (is_degenerated_phi): Ditto.
2247         (normalize_one_pred_1): Ditto.
2248         (normalize_one_pred): Ditto.
2249         (normalize_one_pred_chain): Ditto.
2250         (normalize_preds): Ditto.
2251         (normalize_cond_1): Remove function.
2252         (normalize_cond): Ditto.
2253         (is_gcond_subset_of): Ditto.
2254         (is_subset_of_any): Ditto.
2255         (is_or_set_subset_of): Ditto.
2256         (is_and_set_subset_of): Ditto.
2257         (is_norm_cond_subset_of): Ditto.
2258         (pred_chain_length_cmp): Ditto.
2259         (convert_control_dep_chain_into_preds): Type change.
2260         (find_predicates): Ditto.
2261         (find_def_preds): Ditto.
2262         (destroy_predicates_vecs): Ditto.
2263         (find_matching_predicates_in_rest_chains): Ditto.
2264         (use_pred_not_overlap_with_undef_path_pred): Ditto.
2265         (is_pred_expr_subset): Ditto.
2266         (is_pred_chain_subset_of): Ditto.
2267         (is_included_in): Ditto.
2268         (is_superset_of): Ditto.
2270 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
2272         Update copyright years.
2274 2014-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
2276         * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2277         config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2278         config/arc/arc.md, config/arc/arc.opt,
2279         config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2280         config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2281         config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2282         config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2283         config/linux-protos.h, config/linux.c, config/winnt-c.c,
2284         diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2285         vtable-verify.c, vtable-verify.h: Use the standard form for the
2286         copyright notice.
2288 2014-01-02  Tobias Burnus  <burnus@net-b.de>
2290         * gcc.c (process_command): Update copyright notice dates.
2291         * gcov-dump.c: Ditto.
2292         * gcov.c: Ditto.
2293         * doc/cpp.texi: Bump @copying's copyright year.
2294         * doc/cppinternals.texi: Ditto.
2295         * doc/gcc.texi: Ditto.
2296         * doc/gccint.texi: Ditto.
2297         * doc/gcov.texi: Ditto.
2298         * doc/install.texi: Ditto.
2299         * doc/invoke.texi: Ditto.
2301 2014-01-01  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2303         * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2305 2014-01-01  Jakub Jelinek  <jakub@redhat.com>
2307         * config/i386/sse.md (*mov<mode>_internal): Guard
2308         EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2310         PR rtl-optimization/59647
2311         * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2312         new_rtx into UNSIGNED_FLOAT rtxes.
2314 Copyright (C) 2014 Free Software Foundation, Inc.
2316 Copying and distribution of this file, with or without modification,
2317 are permitted in any medium without royalty provided the copyright
2318 notice and this notice are preserved.