* ipa-utils.h (method_class_type): Declare.
[official-gcc.git] / gcc / ChangeLog
blob1f99c1e73226d9c85a81aa686b8937d3b0ccee81
1 2013-08-23  Jan Hubicka  <jh@suse.cz>
3         * ipa-utils.h (method_class_type): Declare.
4         * ipa-devirt.c (method_class_type): Export.
6         * cgraphunit.c (analyze_functions): Do basic devirtualization;
7         do not walk base classes of anonymous types.
9 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
11         PR rtl-optimization/58220
12         PR regression/58221
13         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
14         handle SEQUENCE insns properly.
16 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
18         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
19         definition.
20         (pp_newline_and_indent): Likewise.
21         (pp_separate_with): Likewise.
22         * pretty-print.c (pp_newline_and_flush): Define.
23         (pp_newline_and_indent): Likewise.
24         (pp_separate_with): Likewise.
26 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
28         PR target/58218
29         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
30         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
32 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
34         * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
35         * gcc/config/i386/i386.md (*movti_internal): Use
36         predicate to determine if EVEX is needed.
37         (*movsi_internal): Ditto.
38         (*movdf_internal): Ditto.
39         (*movsf_internal): Ditto.
40         * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
42 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
44         PR tree-optimization/58209
45         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
46         (find_tail_calls): Give up for pointer result types if m is non-NULL.
47         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
48         emit POINTER_PLUS_EXPR.
49         (create_tailcall_accumulator): For pointer result type accumulate in
50         sizetype type.
52 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
54         * configure.ac: Add backslashes missing from the last change.
55         * configure: Regenerate.
57 2013-08-22  Jan Hubicka  <jh@susue.cz>
59         * ipa.c (function_and_variable_visibility): First remember function
60         was global and then make it local.
62 2013-08-22  Julian Brown  <julian@codesourcery.com>
64         * configure.ac: Add aarch64 to list of arches which use "nop" in
65         debug_line test.
66         * configure: Regenerate.
68 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
70         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
71         gnu_libc_has_function.
72         * config/s390/tpf.h: Likewise.
74 2013-08-22  Jan Hubicka  <jh@susue.cz>
76         * timevar.c (validate_phases): Add cast.
78 2013-08-22  Jan Hubicka  <jh@susue.cz>
80         * timevar.c (validate_phases): Use size_t for memory.
81         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
83 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
85         * pretty-print.h (output_buffer::output_buffer): Declare.
86         (pretty_printer::pretty_printer): Likewise.
87         (pp_construct): Remove.
88         * pretty-print.c (output_buffer::output_buffer): Define.
89         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
90         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
91         (print_gimple_expr): Likewise.
92         (print_gimple_seq): Likewise.
93         (gimple_dump_bb): Likewise.
94         * sched-vis.c (dump_value_slim): Likewise.
95         (dump_insn_slim): Likewise.
96         (dump_rtl_slim): Likewise.
97         (str_pattern_slim): Likewise.
98         * tree-mudflap.c (mf_varname_tree): Likewise.
99         * graph.c (print_graph_cfg): Likewise.
100         (start_graph_dump): Likewise.
101         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
102         placement-new.
103         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
104         pretty printer initialization.
105         * coretypes.h (diagnostic_context): Remove superflous type alias
106         declaration.
107         (pretty_printer): Likewise.  Declare directly as a class.
108         (pretty_print_info): Remove declaration as class.
109         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
110         and pp_clear_output_area.
111         (asan_add_global): Likewise.
113 2013-08-22  Jan Hubicka  <jh@suse.cz>
115         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
116         * ipa-utils.h (update_type_inheritance_graph): Declare.
117         (possible_polymorphic_call_target_p): Declare.
118         (possible_polymorphic_call_target_p): New.
119         * ipa-devirt.c: Update toplevel comments.
120         (cached_polymorphic_call_targets): Move up.
121         (odr_type_d): Move ID down.
122         (polymorphic_type_binfo_p): Update comment.
123         (odr_hasher::remove): Likewise;
124         (get_odr_type): Set anonymous_namespace.
125         (dump_odr_type): Dump it.
126         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
127         (maybe_record_node): Record node in cached_polymorphic_call_targets.
128         (record_binfo): Add comment.
129         (free_polymorphic_call_targets_hash): Do not ICE when cache is not built.
130         (devirt_node_removal_hook): Do not iCE when cache is freed.
131         (possible_polymorphic_call_target_p): New predicate.
132         (update_type_inheritance_graph): New function.
134 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
135             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
136             Sergey Lega  <sergey.s.lega@intel.com>
137             Anna Tikhonova  <anna.tikhonova@intel.com>
138             Ilya Tocar  <ilya.tocar@intel.com>
139             Andrey Turetskiy  <andrey.turetskiy@intel.com>
140             Ilya Verbin  <ilya.verbin@intel.com>
141             Kirill Yukhin  <kirill.yukhin@intel.com>
142             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
144         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
145         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
146         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
147         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
148         (OPTION_MASK_ISA_AVX2_UNSET): Update.
149         (OPTION_MASK_ISA_AVX512F_UNSET): New.
150         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
151         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
152         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
153         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
154         OPT_mavx512pf, OPT_mavx512er cases.
155         * config/i386/constraints.md (v): New constraint.
156         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
157         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
158         (bit_AVX512CD): New.
159         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
160         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
161         * config/i386/i386-c.c (ix86_target_macros_internal):
162         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
163         __AVX512PF__.
164         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
165         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
166         * config/i386/i386.c (regclass_map, dbx_register_map)
167         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
168         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
169         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
170         -mavx512pf options.
171         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
172         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
173         -mavx512pf options.  Fix formatting.
174         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
175         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
176         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
177         -mavx512cd, -mavx512pf options.
178         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
179         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
180         (ix86_preferred_output_reload_class): Replace SSE_REGS with
181         ALL_SSE_REGS.
182         (ix86_hard_regno_mode_ok): Support 512-bit registers.
183         (ix86_set_reg_reg_cost): Ditto.
184         (x86_order_regs_for_local_alloc): Ditto.
185         (MAX_VECT_LEN): Extend to 64-byte.
186         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
187         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
188         (TARGET_AVX512ER, TARGET_AVX512CD): New.
189         (BIGGEST_ALIGNMENT): Extend to 512-bits.
190         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
191         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
192         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
193         (SSE_REG_MODE_P): Support new modes.
194         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
195         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
196         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
197         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
198         (REG_CLASS_CONTENTS): Add new registers.
199         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
200         (EXT_REX_SSE_REGNO_P): New.
201         (HI_REGISTER_NAMES): Add new registers.
202         * config/i386/i386.md: Define constants for new registers.
203         (mode): Add new 512-bit modes.
204         (prefix): Support evex prefix.
205         (isa): Support avx512f, noavx512f, fma_avx512f.
206         (ssemodesuffix): Add new 512-bit modes.
207         (movxi): New.
208         (*movxi_internal_avx512f): Ditto.
209         (*movdi_internal): Replace constraint "x" with the new constraint "v".
210         Support MODE_XI.
211         (*movsi_internal): Likewise.
212         (*movdf_internal): Likewise.
213         (*movsf_internal): Likewise.
214         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
215         (<code><mode>3): Likewise.
216         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd): New.
217         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
218         with the new constraint "v".
219         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
220         modes.
221         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
222         with the new constraint "v".
223         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
224         (<sse2>_storedqu<avxsizesuffix>): Likewise.
225         (*<plusminus_insn><mode>3): Likewise.
226         (<sse>_vm<plusminus_insn><mode>3): Likewise.
227         (*mul<mode>3): Likewise.
228         (<sse>_vmmul<mode>3): Likewise.
229         (<sse>_div<mode>3): Likewise.
230         (<sse>_vmdiv<mode>3): Likewise.
231         (<sse>_sqrt<mode>2): Likewise.
232         (<sse>_vmsqrt<mode>2): Likewise.
233         (*<code><mode>3_finite): Likewise.
234         (*<code><mode>3) <smaxmin>: Likewise.
235         (<sse>_vm<code><mode>3): Likewise.
236         (*<code><mode>3) <any_logic>: Likewise.
237         (*fma_fmadd_<mode>): Likewise.
238         (*fma_fmsub_<mode>): Likewise.
239         (*fma_fnmadd_<mode>): Likewise.
240         (*fma_fnmsub_<mode>): Likewise.
241         (*fma_fmaddsub_<mode>): Likewise.
242         (*fma_fmsubadd_<mode>): Likewise.
243         (*fmai_fmadd_<mode>): Likewise.
244         (*fmai_fmsub_<mode>): Likewise.
245         (*fmai_fnmadd_<mode>): Likewise.
246         (*fmai_fnmsub_<mode>): Likewise.
247         (sse_cvtsi2ss): Likewise.
248         (sse_cvtsi2ssq): Likewise.
249         (sse_cvtss2si): Likewise.
250         (sse_cvtss2si_2): Likewise.
251         (sse_cvtss2siq): Likewise.
252         (sse_cvtss2siq_2): Likewise.
253         (sse_cvttss2si): Likewise.
254         (sse_cvtss2siq_2): Likewise.
255         (float<sseintvecmodelower><mode>2): Likewise.
256         (sse2_cvtsd2si_2): Likewise.
257         (sse2_cvtsd2siq_2): Likewise.
258         (*<plusminus_insn><mode>3): Likewise.
259         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
260         (*<sse4_1_avx2>_mul<mode>3): Likewise.
261         (ashr<mode>3): Likewise.
262         (<shift_insn><mode>3): Likewise.
263         (avx2_<code><mode>3): Likewise.
264         (*avx2_<code><mode>3): Likewise.
265         (*andnot<mode>3): Likewise.
266         (*<code><mode>3) <any_logic>: Likewise.
267         (abs<mode>2): Likewise.
268         (avx2_permvar<mode>): Likewise.
269         (avx2_perm<mode>_1): Likewise.
270         (*avx_vpermilp<mode>): Likewise.
271         (avx_vpermilvar<mode>3): Likewise.
272         (avx2_ashrv<mode>): Likewise.
273         (avx2_<shift_insn>v<mode>): Likewise.
274         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
275         -mavx512cd.
276         * doc/rtl.texi: Document XImode.
278 2013-08-21  Jeff Law  <law@redhat.com>
280         * tree-flow.h (register_jump_thread): Pass vector of edges
281         instead of each important edge.
282         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
283         thread path into a vector and pass that to register_jump_thread.
284         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
285         passed in edge vector to the current 3-edge form.
287         Revert:
288         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
290         * dce.c (fini_dce): Call df_analyze again just in case
291         delete_unmarked_insns removed anything.
293 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
295         * reload.h (struct reg_equivs): Rename to ..
296         (struct reg_equivs_s): .. this.
298 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
300         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
302 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
304         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest
305         comment.
307 2013-08-21  Jeff Law  <law@redhat.com>
309         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
310         simplify assignments too.  If the RHS collapses to a singleton
311         range, then return the value for the range.
313 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
315         * config/i386/sse.md (V16): Rename to...
316         (VMOVE): this.
317         (mov<mode>): Update iterator name.
318         (*mov<mode>_internal): Ditto.
319         (push<mode>1): Ditto.
320         (movmisalign<mode>): Ditto.
322 2013-08-20  Jan Hubicka  <jh@suse.cz>
324         PR bootstrap/58186
325         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
326         entry for direct edges.
327         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
329 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
331         Revert my last two changes, r201865 and r201864:
333         Revert r201865:
334         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
336         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
337         instances can own GC refs.
339         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
340         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
341         (gcc::context::gt_pch_nx): Likewise.
342         (gcc::context::gt_pch_nx):  Likewise.
343         * ggc.h (gt_ggc_mx <T>): New.
344         (gt_pch_nx_with_op <T>): New.
345         (gt_pch_nx <T>): New.
346         * passes.c (opt_pass::gt_ggc_mx): New.
347         (opt_pass::gt_pch_nx): New.
348         (opt_pass::gt_pch_nx_with_op): New.
349         (pass_manager::gt_ggc_mx): New.
350         (pass_manager::gt_pch_nx): New.
351         (pass_manager::gt_pch_nx_with_op): New.
352         (pass_manager::operator new): Use
353         ggc_internal_cleared_alloc_stat rather than xcalloc.
354         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
355         (pass_manager::gt_ggc_mx): New.
356         (pass_manager::gt_pch_nx): New.
357         (pass_manager::gt_pch_nx_with_op): New.
358         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
359         (opt_pass::operator new): New.
360         (opt_pass::gt_ggc_mx): New.
361         (opt_pass::gt_pch_nx): New.
362         (opt_pass::gt_pch_nx_with_op): New.
364         Revert r201864:
365         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
367         * Makefile.in (GTFILES): Add context.h.
368         * context.c (gcc::context::operator new): New.
369         (gcc::context::gt_ggc_mx): New.
370         (gcc::context::gt_pch_nx): New.
371         (gcc::context::gt_pch_nx): New.
372         * context.h (gcc::context): Add GTY((user)) marking.
373         (gcc::context::operator new): New.
374         (gcc::context::gt_ggc_mx): New.
375         (gcc::context::gt_pch_nx): New.
376         (gcc::context::gt_pch_nx): New.
377         (g): Add GTY marking.
378         (gt_ggc_mx (gcc::context *)): New.
379         (gt_pch_nx (gcc::context *)): New.
380         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
381         void *cookie)): New.
382         * gengtype.c (open_base_files) <ifiles>: Add context.h.
384 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
386         * dce.c (fini_dce): Call df_analyze again just in case
387         delete_unmarked_insns removed anything.
389 2013-08-20  Teresa Johnson  <tejohnson@google.com>
391         PR rtl-optimizations/57451
392         * final.c (reemit_insn_block_notes): Prevent lexical blocks
393         from crossing split section boundaries.
395 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
397         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
398         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
399         with MULTLIB_DEFAULTS.
401 2013-08-20  Nick Clifton  <nickc@redhat.com>
403         * target.def (narrow_volatile_bitfield): Note that the default
404         value is false, not !TARGET_STRICT_ALIGN.
405         * doc/tm.texi: Regenerate.
407 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
409         Fix LIB_SPEC for systems without libpthread.
411         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
412         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
413         for Android.
414         * config/i386/linux-common.h: Likewise.
415         * config/mips/linux-common.h: Likewise.
417 2013-08-20  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
419         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
420         checks.
422 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
424         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
425         instances can own GC refs.
427         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
428         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
429         (gcc::context::gt_pch_nx): Likewise.
430         (gcc::context::gt_pch_nx):  Likewise.
431         * ggc.h (gt_ggc_mx <T>): New.
432         (gt_pch_nx_with_op <T>): New.
433         (gt_pch_nx <T>): New.
434         * passes.c (opt_pass::gt_ggc_mx): New.
435         (opt_pass::gt_pch_nx): New.
436         (opt_pass::gt_pch_nx_with_op): New.
437         (pass_manager::gt_ggc_mx): New.
438         (pass_manager::gt_pch_nx): New.
439         (pass_manager::gt_pch_nx_with_op): New.
440         (pass_manager::operator new): Use
441         ggc_internal_cleared_alloc_stat rather than xcalloc.
442         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
443         (pass_manager::gt_ggc_mx): New.
444         (pass_manager::gt_pch_nx): New.
445         (pass_manager::gt_pch_nx_with_op): New.
446         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
447         (opt_pass::operator new): New.
448         (opt_pass::gt_ggc_mx): New.
449         (opt_pass::gt_pch_nx): New.
450         (opt_pass::gt_pch_nx_with_op): New.
452 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
454         * Makefile.in (GTFILES): Add context.h.
455         * context.c (gcc::context::operator new): New.
456         (gcc::context::gt_ggc_mx): New.
457         (gcc::context::gt_pch_nx): New.
458         (gcc::context::gt_pch_nx): New.
459         * context.h (gcc::context): Add GTY((user)) marking.
460         (gcc::context::operator new): New.
461         (gcc::context::gt_ggc_mx): New.
462         (gcc::context::gt_pch_nx): New.
463         (gcc::context::gt_pch_nx): New.
464         (g): Add GTY marking.
465         (gt_ggc_mx (gcc::context *)): New.
466         (gt_pch_nx (gcc::context *)): New.
467         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
468         void *cookie)): New.
469         * gengtype.c (open_base_files) <ifiles>: Add context.h.
471 2013-08-20  Alan Modra  <amodra@gmail.com>
473         PR target/57865
474         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
475         (rs6000_emit_epilogue): Likewise.
477 2013-08-19  Dehao Chen (dehao@google.com)
479         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
481 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
482             Jakub Jelinek  <jakub@redhat.com>
484         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
485         (BUILT_IN_FABSD64): Likewise.
486         (BUILT_IN_FABSD128): Likewise.
487         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
488         (fold_builtin_1): Likewise.
489         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping destination
490         and source operands.
491         (*abstd2_fpr): Likewise.
492         (*nabstd2_fpr): Likewise.
494 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
496         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
497         JUMP_P and INSN_P.
499 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
501         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
502         incomplete.
504 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
506         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
507         * builtins.c (default_libc_has_function): New.
508         (gnu_libc_has_function): Ditto.
509         (no_c99_libc_has_function): Ditto.
510         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
511         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
512         (fold_builtin_sincos): Likewise.
513         (fold_builtin_cexp): Likewise.
514         * builtins.def (DEF_C94_BUILTIN): Likewise.
515         (DEF_C99_BUILTIN): Likewise.
516         (DEF_C99_C90RES_BUILTIN): Likewise.
517         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
518         definitions to using this define.
519         * config/darwin-protos.h (darwin_libc_has_function): New.
520         * config/darwin.c: (darwin_libc_has_function: Ditto.
521         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
522         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
523         * config/darwin.h: Ditto.
524         * config/elfos.h: Ditto.
525         * config/freebsd.h: Ditto.
526         * config/i386/cygming.h: Ditto.
527         * config/i386/djgpp.h: Ditto.
528         * config/i386/i386-interix.h: Ditto.
529         * config/microblaze/microblaze.h: Ditto.
530         * config/mmix/mmix.h: Ditto.
531         * config/gnu-user.h: Ditto.
532         * config/ia64/hpux.h: Ditto.
533         * config/pa/pa-hpux.h: Ditto.
534         * config/pdp11/pdp11.h: Ditto.
535         * config/picochip/picochip.h: Ditto.
536         * config/linux.h: Ditto.
537         * config/netbsd.h: Ditto.
538         * config/openbsd.h: Ditto.
539         * config/rs6000/aix43.h: Ditto.
540         * config/rs6000/aix51.h: Ditto.
541         * config/rs6000/aix52.h: Ditto.
542         * config/rs6000/aix53.h: Ditto.
543         * config/rs6000/aix61.h: Ditto.
544         * config/rs6000/darwin.h: Ditto.
545         * config/rs6000/linux.h: Ditto.
546         * config/rs6000/linux64.h: Ditto.
547         * config/s390/tpf.h: Ditto.
548         * config/sol2-10.h: Ditto.
549         * config/sol2.h: Ditto.
550         * config/vms/vms.h: Ditto.
551         * config/vxworks.h: Ditto.
552         * config/linux-android.c (linux_android_libc_has_function):
553         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
554         * config/linux-protos.h (linux_android_libc_has_function):
555         New declaration.
556         * config/i386/i386.c (ix86_libc_has_function): New.
557         * config/i386/i386-protos.h
558         (ix86_libc_has_function): New declaration.
559         * config/i386/i386.md
560         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
561         ("isinf<mode>2): Likewise.
562         * convert.c (convert_to_integer): Using new target hook
563         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
564         TARGET_C99_FUNCTIONS.
565         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
566         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
567         * coretypes.h (function_class): New enum for different
568         classes of functions.
569         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
570         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
571         (TARGET_HAS_SINCOS): Likewise.
572         (TARGET_LIBC_HAS_FUNCTION): New.
573         * doc/tm.texi: Regenerated.
574         * targhooks.h (default_libc_has_function): New declaration.
575         (no_c99_libc_has_function): Ditto.
576         (gnu_libc_has_function): Ditto.
577         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
578         and TARGET_HAS_SINCOS.
580 2013-08-18  Jan Hubicka  <jh@suse.cz>
582         * Makeifle-in (ipa-devirt.o): New.
583         (GTFILES): Add ipa-utils.h and ipa-devirt.c
584         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
585         (analyze_functions): Look into possible targets of polymorphic call.
586         * dumpfile.c (dump_files): Add type-inheritance dump.
587         * dumpfile.h (TDI_inheritance): New.
588         * ipa-devirt.c: New file.
589         * ipa-utils.h (odr_type_d): Forward declare.
590         (odr_type): New type.
591         (build_type_inheritance_graph): Declare.
592         (possible_polymorphic_call_targets): Declare and introduce inline
593         variant when only edge is pased.
594         (dump_possible_polymorphic_call_targets): Likewise.
595         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
596         * tree.c (type_in_anonymous_namespace_p): Break out from ...
597         (types_same_for_odr): ... here.
598         * tree.h (type_in_anonymous_namespace_p): Declare.
600 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
602         PR tree-optimization/58006
603         * tree-parloops.c (take_address_of): Don't ICE if get_name
604         returns NULL.
605         (eliminate_local_variables_stmt): Remove clobber stmts.
607 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
609         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
610         error message is issued for an alias to undefined symbol.
612 2013-08-18  Jan Hubicka  <jh@suse.cz>
614         * cgraph.c (cgraph_create_indirect_edge): Discover
615         polymorphic calls and record basic info into indirect_info.
616         * gimple-fold.c (gimple_fold_call): When doing BINFO based
617         devirtualization, ignore objc function calls.
618         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
619         call with no parm index info.
620         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
621         * tree.c (virtual_method_call_p): New function.
622         * tree.h (virtual_method_call_p): Declare.
624 2013-08-16  Jan Hubicka  <jh@suse.cz>
626         PR middle-end/58179
627         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
629 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
631         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
632         attribute.
634 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
636         * gengtype.c (type_for_name): Add special-case support for
637         locating types within the "gcc::" namespace.
638         (open_base_files): Emit a "using namespace gcc" directive.
640 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
642         PR target/58160
643         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
644         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
646         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
647         array instead of each individual operand as a separate argument.
648         (emit_fusion_gpr_load): Likewise.
649         (expand_fusion_gpr_load): Add new function declaration.
651         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
652         signature to have the operands passed as an array, instead of as
653         separate arguments.  Allow ZERO_EXTEND to be in the memory
654         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
655         depend on the register live/dead flags when peepholes are run.
656         (expand_fusion_gpr_load): New function to be called from the
657         peephole2 pass, to change the register that addis sets to be the
658         target register.
659         (emit_fusion_gpr_load): Change the calling signature to have the
660         operands passed as an array, instead of as separate arguments.
661         Allow ZERO_EXTEND to be in the memory address, and also
662         SIGN_EXTEND if -mpower8-fusion-sign.
664         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
665         unspec enumeration.
666         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
667         adjust the register addis loads up in the peephole2 pass.  Do not
668         depend on the register live/dead state when the peephole pass is
669         done.
671 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
673         * gengtype.c (create_user_defined_type): Ensure that the kind
674         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
675         declaration is seen before the GTY((user)) marking.
677 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
679         PR target/58105
680         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
682 2013-08-16  Jan Hubicka  <jh@suse.cz>
684         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
685         arugment expected_type.
686         (gimple_fold_call): Use it.
687         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
688         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
689         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
690         (try_make_edge_direct_virtual_call): Likewise.
691         * tree.c (obj_type_ref_class): New.
692         * tree.h (obj_type_ref_class): Use it.
694 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
696         * sched-vis.c (rtl_slim_pp_initialized): Remove.
697         (rtl_slim_pp): Likewise.
698         (init_rtl_slim_pretty_print): Likewise.
699         (dump_value_slim):  Don't call it.  Use local pretty printer.
700         (dump_insn_slim): Likewise.
701         (dump_rtl_slim): Likewise.
702         (str_pattern_slim): Likewise.
703         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
704         Simplify.
706 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
708         PR tree-optimization/58164
709         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
710         walk gimple_goto_dest of GIMPLE_GOTO.
712         PR tree-optimization/58165
713         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
714         bi_call must be the last stmt in a bb, don't split_block, instead
715         use fallthru edge from it and give up if there is none.
716         Release conds vector when returning early.
718 2013-08-14  Xinliang David Li  <davidxl@google.com>
720         * config/i386/i386.c (ix86_option_override_internal):
721         Remove unused variable and field.
723 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
725         PR target/57949
726         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
727         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
728         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
729         and Linux, correct BLKmode alignment when 128-bit alignment is
730         required and compatibility flag is not set.
731         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
732         for zero-size arguments when compatibility flag is not set.
734 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
736         PR tree-optimization/58145
737         * tree-sra.c (build_ref_for_offset): If prev_base has
738         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
740 2013-08-14  Xinliang David Li  <davidxl@google.com>
742         * config/i386/i386.c (ix86_option_override_internal):
743         Fix uninitialized variable error.
745 2013-08-14  Xinliang David Li  <davidxl@google.com>
747         * config/i386/i386.opt: Define two new options.
748         * config/i386/x86-tune.def: Add arch selector field in macros.
749         * config/i386/i386.h: Adjust macro definition.
750         * config/i386/i386.c (ix86_option_override_internal):
751         Refactor the code.
752         (parse_mtune_ctrl_str): New function.
753         (set_ix86_tune_features): New function.
754         (ix86_function_specific_restore): Call the new helper function.
756 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
758         PR rtl-optimization/57662
759         * sel-sched.c (code_motion_process_successors): When the current insn
760         is removed after the recursive traversal, break from the loop.
761         Add comments and debug printouts.
763 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
764             Alexandre Oliva  <aoliva@redhat.com>
766         PR target/58067
767         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
768         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
769         there also UNSPEC_PLTOFF.
771 2013-08-14  Marek Polacek  <polacek@redhat.com>
773         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
774         NUM_CONDITIONS bit positions.
776 2013-08-13  Cary Coutant  <ccoutant@google.com>
778         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
779         (attr_checksum): Hash vector contents instead of pointer.
780         (attr_checksum_ordered): Likewise.
782 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
784         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
785         when Pmode != word_mode.  Add length_address attribute.
786         (sse3_monitor_<mode>): Merge from sse3_monitor and
787         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
788         Pmode != word_mode.  Update insn length attribute.
789         * config/i386/i386.c (ix86_option_override_internal): Update
790         ix86_gen_monitor selection for merged sse3_monitor insn.
792 2013-08-13  Julian Brown  <julian@codesourcery.com>
794         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
795         perform invalid legitimization on greater-than-word-size modes for
796         TARGET_E500_DOUBLE.
798 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
800         * ira.c (setup_class_translate_array): Use aclass instead of cl
801         for classes not fully covered by allocno classes.
803 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
805         PR tree-optimization/57661
806         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
807         * tree-inline.c (tree_function_versioning): Initialize it.
808         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
809         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
810         that is not being copied.
812         PR sanitizer/56417
813         * asan.c (instrument_strlen_call): Fix typo in comment.
814         Use char * type even for the lhs of POINTER_PLUS_EXPR.
816 2013-08-13  Steve Ellcey  <sellcey@mips.com>
818         * config/mips/mips.md (prefetch): Use lw instead of ld on
819         loongson in 32bit mode.
821 2013-08-13  Nick Clifton  <nickc@redhat.com>
823         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
825 2013-08-13  Jan Hubicka  <jh@suse.cz>
827         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
828         introduced edge; fix typo in sanity check.
829         (cgraph_resolve_speculation): Export; improve diagnostic.
830         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
831         speculation at type mismatch.
832         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
833         (cgraph_resolve_speculation): Declare.
834         (symtab_can_be_discarded): New function.
835         * value-prof.c (gimple_ic_transform): Remove actual transform code.
836         * ipa-inline-transform.c (speculation_removed): New global var.
837         (clone_inlined_nodes): See if speculation can be removed.
838         (inline_call): If speculations was removed, we growths may not match.
839         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
840         (speculation_useful_p): New function.
841         (resolve_noninline_speculation): New function.
842         (inline_small_functions): Resolve useless speculations.
843         * ipa-inline.h (speculation_useful_p): Declare
844         * ipa.c (can_replace_by_local_alias): Simplify.
845         (ipa_profile): Produce speculative calls in non-lto, too;
846         add simple cost model; produce local aliases.
848 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
850         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
851         PASS_MANAGER_H.
853 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
855         * config/i386/i386.c (ix86_function_versions): Use error + inform.
857 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
859         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
860         iterator instead of X87MODEF.
862 2013-08-12  Perez Read  <netfirewall@gmail.com>
864         PR target/58132
865         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
866         operand 0 for intel asm alternative.
867         (*movabs<mode>_2): Ditto for operand 1.
869 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
871         * config/aarch64/arm_none.h
872         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
874 2013-08-12  Nick Clifton  <nickc@redhat.com>
876         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
878 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
880         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
881         expand for QI/HImode operand to produce more effictive code for
882         unsigned char(short) --> float(double) conversion.
884 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
886         * doc/invoke.texi: Mention that -ftls-model does not force the final
887         model.
889 2013-08-12  Marek Polacek  <polacek@redhat.com>
890             Marc Glisse  <marc.glisse@inria.fr>
892         PR tree-optimization/57980
893         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
894         when creating -1 constant.
896 2013-08-10  Jan Hubicka  <jh@suse.cz>
898         Workaround binutils PR14342.
899         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
900         (gimple_init_edge_profiler): Likewise.
901         (gimple_gen_ic_func_profiler): Likewise.
903 2013-08-09  Jan Hubicka  <jh@suse.cz>
905         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
907 2013-08-09  Xinliang David Li  <davidxl@google.com>
909         * config/i386/stringop.def: New file.
910         * config/i386/stringop.opt: New file.
911         * config/i386/i386-opts.h: Include stringopt.def.
912         * config/i386/i386.opt: Include stringopt.opt.
913         * config/i386/i386.c (ix86_option_override_internal):
914         Override default size based stringop inline strategies with options.
915         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
916         New function.
918 2013-08-09  Jan Hubicka  <jh@suse.cz>
920         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
922 2013-08-09  Jan Hubicka  <jh@suse.cz>
924         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
925         CGRAPH_FREQ_MAX.
926         (dump_cgraph_node): Dump profile-id.
927         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
928         and common_target_probability.
929         * lto-cgraph.c (lto_output_edge): Stream common targets.
930         (lto_output_node): Stream profile ids.
931         (input_node): Stream profile ids.
932         (input_edge): Stream common targets.
933         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
934         * ipa.c: Include value-prof.h
935         (ipa_profile_generate_summary): Turn indirect call statement histograms
936         into common targets.
937         (ipa_profile): Turn common targets into speculative edges.
939 2013-08-09  Jan Hubicka  <jh@suse.cz>
941         * cgraph.h (cgraph_node): Add profile_id.
942         * value-prof.c (cgraph_node_map): Turn into pointer_map.
943         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
944         (del_node_map): Update.
945         (find_func_by_funcdef_no): Replace by ...
946         (find_func_by_profile_id): ... this one.
947         (gimple_ic_transform): Do not remove useful histograms when
948         speculation is not done; dump info when indirect call removal
949         can happen at LTO.
950         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
951         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
952         (__gcov_indirect_call_profiler_v2): .. this one.
953         * profile.h (init_node_map): Update.
954         * coverage.c (coverage_compute_profile_id): New function.
955         * coverage.h (coverage_compute_profile_id): Declare.
956         * tree-profile.c (init_ic_make_global_vars): Make
957         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
958         (gimple_init_edge_profiler): Update prototype of
959         __gcov_indirect_call_profiler.
960         (gimple_gen_ic_func_profiler): Simplify.
961         (tree_profiling): Use init_node_map
963 2013-08-09  Jan Hubicka  <jh@suse.cz>
965         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
966         non-speculative refs.
967         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
968         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
969         (cgraph_set_call_stmt): Likewise.
970         (cgraph_create_edge_1): Fix release checking compilatoin;
971         clear lto_stmt_uid.
972         (cgraph_free_edge): Free indirect info.
973         (cgraph_turn_edge_to_speculative): New function.
974         (cgraph_speculative_call_info): New function.
975         (cgraph_make_edge_direct): Return direct edge; handle speculation.
976         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
977         (dump_cgraph_node): Dump speculation.
978         (verify_edge_count_and_frequency): Accept speculative edges.
979         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
980         (verify_cgraph_node): Handle speculation.
981         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
982         (cgraph_set_call_stmt): Update prototype.
983         (cgraph_make_edge_direct): Update prototype.
984         (cgraph_speculative_call_info): Declare.
985         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
986         to change; update call of ipa_find_references.
987         * ipa-ref.c (ipa_record_reference): Fix return value; clear
988         lto_stmt_uid and speculative flags.
989         (ipa_dump_references): Dump speculation.
990         (ipa_clone_references): Clone speculative flag.
991         (ipa_clone_referring): Likewise.
992         (ipa_clone_ref): New function.
993         (ipa_find_reference): Look into lto_stmt_uids
994         (ipa_clear_stmts_in_references): Do not clear speculative calls.
995         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
996         (ipa_find_reference): Update declaration.
997         (ipa_clone_ref): Declare.
998         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
999         stream speculative flag.
1000         (lto_output_ref): Stream statements uids and speculation.
1001         (input_ref): Likewise.
1002         (input_edge): Stream speuclation.
1003         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1004         (cgraph_set_call_stmt_including_clones): Handle speculation.
1005         * ipa-inline.c (heap_edge_removal_hook): New function.
1006         (inline_small_functions): Register it.
1007         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1008         also initialize refs.
1009         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1010         edge to change.
1011         (try_make_edge_direct_simple_call): Likewise.
1012         (try_make_edge_direct_simple_call): Likewise.
1013         (update_indirect_edges_after_inlining): Likewise.
1014         (remove_described_reference): Look proper lto_stmt_uid.
1015         (propagate_controlled_uses): Likewise.
1016         (propagate_controlled_uses): Liekwise.
1017         * tree-inline.c (copy_bb): Copy speculative edges.
1018         (redirect_all_calls): New function.
1019         (copy_cfg_body): Do redirection after loop info is updated.
1020         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1022 2013-08-09  Jan Hubicka  <jh@suse.cz>
1024         * lto-streamer-out.c (output_function): Renumber PHIs.
1025         * lto-streamer-in.c (input_function): Likewise.
1027 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
1029         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1030         (get_lane_unsigned): Likewise.
1031         (dup_lane_scalar): Likewise.
1032         (get_lane): enable for VALL.
1033         * config/aarch64/aarch64-simd.md
1034         (aarch64_dup_lane_scalar<mode>): Remove.
1035         (aarch64_get_lane_signed<mode>): Likewise.
1036         (aarch64_get_lane_unsigned<mode>): Likewise.
1037         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1038         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1039         (aarch64_get_lane<mode>): Enable for all vector modes.
1040         (aarch64_get_lanedi): Remove misleading constraints.
1041         * config/aarch64/arm_neon.h
1042         (__aarch64_vget_lane_any): Define.
1043         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1044         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1045         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1046         * config/aarch64/iterators.md (VDQQH): New.
1047         (VDQQHS): Likewise.
1048         (vwcore): Likewise.
1050 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
1052         * configure.ac: Add GAS check for LEON instructions on SPARC.
1053         * configure: Regenerate.
1054         * config.in: Likewise.
1055         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1056         sparc*-*-* block.
1057         * config/sparc/sparc.opt (LEON, LEON3): New masks.
1058         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1059         for LEON or LEON3.
1060         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1061         (AS_LEON_FLAG): New macro.
1062         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1063         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1064         Deal with LEON and LEON3 for the memory model.
1065         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1066         (atomic_compare_and_swap<mode>_1): Likewise.
1067         (*atomic_compare_and_swap<mode>_1): Likewise.
1069 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1071         * config/arm/neon.md (vcond): Fix floating-point vector
1072         comparisons against 0.
1074 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
1076         * lra-constraints.c (emit_spill_move): Remove assert.
1077         (process_alt_operands): Add more debugging
1078         output.  Increase reject for spilling into memory.  Decrease
1079         reject for reloading scratch.
1080         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1082 2013-08-08  Steve Ellcey  <sellcey@mips.com>
1084         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1085         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1086         micromips incompatible.  Add nan2008.
1087         (MULTILIB_DIRNAMES): Add nan2008.
1088         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1089         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1090         and micromips incompatible.  Add nan2008.
1091         (MULTILIB_DIRNAMES): Add nan2008.
1092         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1094 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
1096         PR rtl-optimization/58079
1097         * combine.c (combine_simplify_rtx): Avoid using SUBST if
1098         simplify_comparison has widened a comparison with an integer.
1100 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1102         * config/arm/neon.md (movmisalign<mode>): Disable when we
1103         don't allow unaligned accesses.
1104         (*movmisalign<mode>_neon_store): Likewise.
1105         (*movmisalign<mode>_neon_load): Likewise.
1106         (*movmisalign<mode>_neon_store): Likewise.
1107         (*movmisalign<mode>_neon_load): Likewise.
1109 2013-08-08  Jan Hubicka  <jh@suse.cz>
1111         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1112         (make_pass_rebuild_cgraph_edges): Also clear references.
1113         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1114         * ipa-inline-transform.c (inline_transform): Remove all references
1115         after inlining.
1116         * cgraphunit.c (expand_function): Remove all references after
1117         expansion.
1118         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1119         (ipa_find_reference): Rewrite to iterator.
1120         (remove_stmt_references): Likewise.
1121         (ipa_clear_stmts_in_references): New function.
1122         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
1123         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
1124         clear references.
1125         * ipa-split.c (split_function): Remove references in split function.
1127 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1129         PR target/57431
1130         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
1131         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
1133 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1135         PR target/56979
1136         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
1137         suggested mode for the assignment isn't compatible with the
1138         registers required.
1140 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1142         PR target/58065
1143         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
1145 2013-08-07  Xinliang David Li  <davidxl@google.com>
1147         * config/i386/i386.opt: New option -mtune-ctrl=.
1148         * config/i386/x86-tune.def: New file.
1149         * config/i386/i386.h: include x86-tune.def.
1150         * config/i386/i386.c (ix86_option_override_internal):
1151         Parsing -mtune-ctrl= option and set tune features.
1153 2013-08-07  Oleg Endo <olegendo@gcc.gnu.org>
1155         PR other/12081
1156         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
1157         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
1158         to gen_2arg_fn_t.
1160 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1162         * rtl.h (update_alignments): Declare.
1163         * final.c (grow_label_align): New function extracted from...
1164         (shorten_branches): ...here.  Call it.
1165         (update_alignments): New function.
1166         * reorg.c (sibling_labels): New variable.
1167         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
1168         the new label along with it onto the sibling_labels vector.
1169         (fill_simple_delay_slots): Adjust call to get_label_before.
1170         (fill_slots_from_thread): Likewise.
1171         (relax_delay_slots): Likewise.
1172         (make_return_insns): Likewise.
1173         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
1175 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1177         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
1178         document its semantics.
1179         (diagnostic_report_diagnostic): Adjust accordingly.
1181 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1183         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
1184         (sparc_option_override): ...and port to new C++ pass API.
1185         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
1187 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
1189         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
1191 2013-08-06  Caroline Tice  <cmtice@google.com>
1193         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
1194         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
1195         * tree-pass.h: Add pass_vtable_verify.
1196         * varasm.c (assemble_variable): Add code to properly set the comdat
1197         section and name for the .vtable_map_vars section.
1198         (assemble_vtyv_preinit_initializer): New function.
1199         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
1200         LINK_ONCE flag.
1201         * output.h: Add function decl for assemble_vtv_preinit_initializer.
1202         * vtable-verify.c: New file.
1203         * vtable-verify.h: New file.
1204         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
1205         initialiation levels.
1206         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
1207         * passes.def: Insert pass_vtable_verify.
1208         * aclocal.m4: Reorder includes.
1209         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
1210         -fvtv-counts options.
1211         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
1212         as appropriate, if -fvtable-verify=... is used.
1213         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
1214         -fvtable-verify=... is used.
1215         * Makefile.in (OBJS):  Add vtable-verify.o to list.
1216         (vtable-verify.o): Add new build rule.
1217         (GTFILES): Add vtable-verify.c to list.
1218         * common.opt (fvtable-verify=): New flag.
1219         (vtv_priority): Values for fvtable-verify= flag.
1220         (fvtv-counts): New flag.
1221         (fvtv-debug): New flag.
1222         * tree.h (save_vtable_map_decl): New extern function decl.
1224 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1226         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
1227         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
1228         (pass_data_rl78_devirt): ...new pass_data instance and...
1229         (make_pass_rl78_devirt): ...new function.
1230         (rl78_asm_file_start): Port pass registration to new C++ API.
1232 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1234         * coretypes.h (rtl_opt_pass): Add.
1235         (gcc::context): Add.
1236         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
1237         (epiphany_init): Port to new C++ pass API.
1238         (epiphany_optimize_mode_switching): Likewise.
1239         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
1240         (pass_manager::get_pass_mode_switching): New.
1241         (pass_manager::get_pass_peephole2): New.
1242         * mode-switching.c (pass_mode_switching): Add clone method.
1243         * recog.c (pass_peephole2): Add clone method.
1244         (pass_split_all_insns): Add clone method.
1246 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
1248         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
1249         (mips_option_override): ...here, porting to new C++ API for passes.
1251 2013-08-06  Jan Hubicka  <jh@suse.cz>
1253         * cgraph.c (cgraph_get_body): New function based on lto.c
1254         implementation.
1255         * cgraph.h (cgraph_get_body): Declare.
1256         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
1257         LTO paths.
1258         * cgraphunit.c (expand_function): Get body prior expanding.
1259         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
1260         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
1261         really need.
1262         * passes.c (do_per_function_toporder): Get body.
1263         * tree-inline.c (expand_call_inline): Get body prior inlining it.
1264         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
1266 2013-08-06  Martin Jambor  <mjambor@suse.cz>
1268         PR fortran/57987
1269         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
1270         is not re-finalized.  Rename second parameter to no_collect.
1272 2013-08-06  Martin Jambor  <mjambor@suse.cz>
1274         PR middle-end/58041
1275         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
1276         MEM_REF has proper alignment information.
1278 2013-08-05  Oleg Endo <olegendo@gcc.gnu.org>
1280         PR other/12081
1281         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
1282         class insn_gen_fn.
1283         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
1284         rtx (*) (rtx, ...) with insn_gen_fn.
1285         * genoutput.c (output_insn_data): Cast gen_? function pointers to
1286         insn_gen_fn::stored_funcptr.  Add initializer braces.
1288 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1290         Rewrite how instances of passes are cloned to remove assumptions
1291         about their sizes (thus allowing pass subclasses to have
1292         additional data fields, albeit non-GC-managed ones at this point).
1294         * passes.c (make_pass_instance): Now that passes have clone
1295         methods, rewrite this function to eliminate XNEW and memcpy
1296         calls that used hardcoded sizes.  Since this function no longer
1297         creates pass instances, rename it to...
1298         (add_pass_instance): ...this.  Document the old way that passes were
1299         numbered and flagged, and rework this function to continue using it.
1300         (next_pass_1): Add an initial_pass argument for use by
1301         add_pass_instance.
1302         (position_pass): When adding multiple instances of a pass, use
1303         the pass's clone method, rather than relying on the XNEW/memcpy
1304         within the former make_pass_instance (now add_pass_instance).
1305         (pass_manager::pass_manager): When invoking next_pass_1, also supply
1306         the initial instance of the current pass within the pass manager.
1308 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1310         This is the automated part of the conversion of passes from C
1311         structs to C++ classes.
1313         Patch autogenerated by refactor_passes.py from
1314         https://github.com/davidmalcolm/gcc-refactoring-scripts
1315         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
1317         * asan.c (pass_asan): Convert from a global struct to a subclass of
1318         gimple_opt_pass along with...
1319         (pass_data_asan): ...new pass_data instance and...
1320         (make_pass_asan): ...new function.
1321         (pass_asan_O0): Convert from a global struct to a subclass of
1322         gimple_opt_pass along with...
1323         (pass_data_asan_O0): ...new pass_data instance and...
1324         (make_pass_asan_O0): ...new function.
1325         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
1326         subclass of rtl_opt_pass along with...
1327         (pass_data_inc_dec): ...new pass_data instance and...
1328         (make_pass_inc_dec): ...new function.
1329         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
1330         a subclass of rtl_opt_pass along with...
1331         (pass_data_reorder_blocks): ...new pass_data instance and...
1332         (make_pass_reorder_blocks): ...new function.
1333         (pass_duplicate_computed_gotos): Convert from a global struct to a
1334         subclass of rtl_opt_pass along with...
1335         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
1336         (make_pass_duplicate_computed_gotos): ...new function.
1337         (pass_partition_blocks): Convert from a global struct to a subclass of
1338         rtl_opt_pass along with...
1339         (pass_data_partition_blocks): ...new pass_data instance and...
1340         (make_pass_partition_blocks): ...new function.
1341         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
1342         struct to a subclass of rtl_opt_pass along with...
1343         (pass_data_branch_target_load_optimize1): ...new pass_data instance
1344         and...
1345         (make_pass_branch_target_load_optimize1): ...new function.
1346         (pass_branch_target_load_optimize2): Convert from a global struct to a
1347         subclass of rtl_opt_pass along with...
1348         (pass_data_branch_target_load_optimize2): ...new pass_data instance
1349         and...
1350         (make_pass_branch_target_load_optimize2): ...new function.
1351         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
1352         of rtl_opt_pass along with...
1353         (pass_data_jump): ...new pass_data instance and...
1354         (make_pass_jump): ...new function.
1355         (pass_jump2): Convert from a global struct to a subclass of
1356         rtl_opt_pass along with...
1357         (pass_data_jump2): ...new pass_data instance and...
1358         (make_pass_jump2): ...new function.
1359         * cfgexpand.c (pass_expand): Convert from a global struct to a
1360         subclass of rtl_opt_pass along with...
1361         (pass_data_expand): ...new pass_data instance and...
1362         (make_pass_expand): ...new function.
1363         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
1364         of rtl_opt_pass along with...
1365         (pass_data_free_cfg): ...new pass_data instance and...
1366         (make_pass_free_cfg): ...new function.
1367         (pass_into_cfg_layout_mode): Convert from a global struct to a
1368         subclass of rtl_opt_pass along with...
1369         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
1370         (make_pass_into_cfg_layout_mode): ...new function.
1371         (pass_outof_cfg_layout_mode): Convert from a global struct to a
1372         subclass of rtl_opt_pass along with...
1373         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
1374         (make_pass_outof_cfg_layout_mode): ...new function.
1375         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
1376         struct to a subclass of gimple_opt_pass along with...
1377         (pass_data_build_cgraph_edges): ...new pass_data instance and...
1378         (make_pass_build_cgraph_edges): ...new function.
1379         (pass_rebuild_cgraph_edges): Convert from a global struct to a
1380         subclass of gimple_opt_pass along with...
1381         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
1382         (make_pass_rebuild_cgraph_edges): ...new function.
1383         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
1384         subclass of gimple_opt_pass along with...
1385         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
1386         and...
1387         (make_pass_remove_cgraph_callee_edges): ...new function.
1388         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
1389         struct to a subclass of rtl_opt_pass along with...
1390         (pass_data_stack_adjustments): ...new pass_data instance and...
1391         (make_pass_stack_adjustments): ...new function.
1392         * combine.c (pass_combine): Convert from a global struct to a subclass
1393         of rtl_opt_pass along with...
1394         (pass_data_combine): ...new pass_data instance and...
1395         (make_pass_combine): ...new function.
1396         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
1397         global struct to a subclass of rtl_opt_pass along with...
1398         (pass_data_compare_elim_after_reload): ...new pass_data instance
1399         and...
1400         (make_pass_compare_elim_after_reload): ...new function.
1401         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
1402         of rtl_opt_pass along with...
1403         (pass_data_rtl_cprop): ...new pass_data instance and...
1404         (make_pass_rtl_cprop): ...new function.
1405         * cse.c (pass_cse): Convert from a global struct to a subclass of
1406         rtl_opt_pass along with...
1407         (pass_data_cse): ...new pass_data instance and...
1408         (make_pass_cse): ...new function.
1409         (pass_cse2): Convert from a global struct to a subclass of
1410         rtl_opt_pass along with...
1411         (pass_data_cse2): ...new pass_data instance and...
1412         (make_pass_cse2): ...new function.
1413         (pass_cse_after_global_opts): Convert from a global struct to a
1414         subclass of rtl_opt_pass along with...
1415         (pass_data_cse_after_global_opts): ...new pass_data instance and...
1416         (make_pass_cse_after_global_opts): ...new function.
1417         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
1418         of rtl_opt_pass along with...
1419         (pass_data_ud_rtl_dce): ...new pass_data instance and...
1420         (make_pass_ud_rtl_dce): ...new function.
1421         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
1422         rtl_opt_pass along with...
1423         (pass_data_fast_rtl_dce): ...new pass_data instance and...
1424         (make_pass_fast_rtl_dce): ...new function.
1425         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
1426         a subclass of rtl_opt_pass along with...
1427         (pass_data_df_initialize_opt): ...new pass_data instance and...
1428         (make_pass_df_initialize_opt): ...new function.
1429         (pass_df_initialize_no_opt): Convert from a global struct to a
1430         subclass of rtl_opt_pass along with...
1431         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
1432         (make_pass_df_initialize_no_opt): ...new function.
1433         (pass_df_finish): Convert from a global struct to a subclass of
1434         rtl_opt_pass along with...
1435         (pass_data_df_finish): ...new pass_data instance and...
1436         (make_pass_df_finish): ...new function.
1437         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
1438         rtl_opt_pass along with...
1439         (pass_data_rtl_dse1): ...new pass_data instance and...
1440         (make_pass_rtl_dse1): ...new function.
1441         (pass_rtl_dse2): Convert from a global struct to a subclass of
1442         rtl_opt_pass along with...
1443         (pass_data_rtl_dse2): ...new pass_data instance and...
1444         (make_pass_rtl_dse2): ...new function.
1445         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
1446         subclass of rtl_opt_pass along with...
1447         (pass_data_dwarf2_frame): ...new pass_data instance and...
1448         (make_pass_dwarf2_frame): ...new function.
1449         * except.c (pass_set_nothrow_function_flags): Convert from a global
1450         struct to a subclass of rtl_opt_pass along with...
1451         (pass_data_set_nothrow_function_flags): ...new pass_data instance
1452         and...
1453         (make_pass_set_nothrow_function_flags): ...new function.
1454         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
1455         subclass of rtl_opt_pass along with...
1456         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
1457         and...
1458         (make_pass_convert_to_eh_region_ranges): ...new function.
1459         * final.c (pass_compute_alignments): Convert from a global struct to a
1460         subclass of rtl_opt_pass along with...
1461         (pass_data_compute_alignments): ...new pass_data instance and...
1462         (make_pass_compute_alignments): ...new function.
1463         (pass_final): Convert from a global struct to a subclass of
1464         rtl_opt_pass along with...
1465         (pass_data_final): ...new pass_data instance and...
1466         (make_pass_final): ...new function.
1467         (pass_shorten_branches): Convert from a global struct to a subclass of
1468         rtl_opt_pass along with...
1469         (pass_data_shorten_branches): ...new pass_data instance and...
1470         (make_pass_shorten_branches): ...new function.
1471         (pass_clean_state): Convert from a global struct to a subclass of
1472         rtl_opt_pass along with...
1473         (pass_data_clean_state): ...new pass_data instance and...
1474         (make_pass_clean_state): ...new function.
1475         * function.c (pass_instantiate_virtual_regs): Convert from a global
1476         struct to a subclass of rtl_opt_pass along with...
1477         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
1478         (make_pass_instantiate_virtual_regs): ...new function.
1479         (pass_leaf_regs): Convert from a global struct to a subclass of
1480         rtl_opt_pass along with...
1481         (pass_data_leaf_regs): ...new pass_data instance and...
1482         (make_pass_leaf_regs): ...new function.
1483         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
1484         subclass of rtl_opt_pass along with...
1485         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
1486         and...
1487         (make_pass_thread_prologue_and_epilogue): ...new function.
1488         (pass_match_asm_constraints): Convert from a global struct to a
1489         subclass of rtl_opt_pass along with...
1490         (pass_data_match_asm_constraints): ...new pass_data instance and...
1491         (make_pass_match_asm_constraints): ...new function.
1492         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
1493         subclass of rtl_opt_pass along with...
1494         (pass_data_rtl_fwprop): ...new pass_data instance and...
1495         (make_pass_rtl_fwprop): ...new function.
1496         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
1497         rtl_opt_pass along with...
1498         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
1499         (make_pass_rtl_fwprop_addr): ...new function.
1500         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
1501         rtl_opt_pass along with...
1502         (pass_data_rtl_pre): ...new pass_data instance and...
1503         (make_pass_rtl_pre): ...new function.
1504         (pass_rtl_hoist): Convert from a global struct to a subclass of
1505         rtl_opt_pass along with...
1506         (pass_data_rtl_hoist): ...new pass_data instance and...
1507         (make_pass_rtl_hoist): ...new function.
1508         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
1509         subclass of gimple_opt_pass along with...
1510         (pass_data_lower_cf): ...new pass_data instance and...
1511         (make_pass_lower_cf): ...new function.
1512         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
1513         from a global struct to a subclass of gimple_opt_pass along with...
1514         (pass_data_strength_reduction): ...new pass_data instance and...
1515         (make_pass_strength_reduction): ...new function.
1516         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
1517         of rtl_opt_pass along with...
1518         (pass_data_rtl_ifcvt): ...new pass_data instance and...
1519         (make_pass_rtl_ifcvt): ...new function.
1520         (pass_if_after_combine): Convert from a global struct to a subclass of
1521         rtl_opt_pass along with...
1522         (pass_data_if_after_combine): ...new pass_data instance and...
1523         (make_pass_if_after_combine): ...new function.
1524         (pass_if_after_reload): Convert from a global struct to a subclass of
1525         rtl_opt_pass along with...
1526         (pass_data_if_after_reload): ...new pass_data instance and...
1527         (make_pass_if_after_reload): ...new function.
1528         * init-regs.c (pass_initialize_regs): Convert from a global struct to
1529         a subclass of rtl_opt_pass along with...
1530         (pass_data_initialize_regs): ...new pass_data instance and...
1531         (make_pass_initialize_regs): ...new function.
1532         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
1533         of ipa_opt_pass_d along with...
1534         (pass_data_ipa_cp): ...new pass_data instance and...
1535         (make_pass_ipa_cp): ...new function.
1536         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
1537         global struct to a subclass of gimple_opt_pass along with...
1538         (pass_data_inline_parameters): ...new pass_data instance and...
1539         (make_pass_inline_parameters): ...new function.
1540         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
1541         subclass of gimple_opt_pass along with...
1542         (pass_data_early_inline): ...new pass_data instance and...
1543         (make_pass_early_inline): ...new function.
1544         (pass_ipa_inline): Convert from a global struct to a subclass of
1545         ipa_opt_pass_d along with...
1546         (pass_data_ipa_inline): ...new pass_data instance and...
1547         (make_pass_ipa_inline): ...new function.
1548         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
1549         struct to a subclass of gimple_opt_pass along with...
1550         (pass_data_local_pure_const): ...new pass_data instance and...
1551         (make_pass_local_pure_const): ...new function.
1552         (pass_ipa_pure_const): Convert from a global struct to a subclass of
1553         ipa_opt_pass_d along with...
1554         (pass_data_ipa_pure_const): ...new pass_data instance and...
1555         (make_pass_ipa_pure_const): ...new function.
1556         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
1557         to a subclass of ipa_opt_pass_d along with...
1558         (pass_data_ipa_reference): ...new pass_data instance and...
1559         (make_pass_ipa_reference): ...new function.
1560         * ipa-split.c (pass_split_functions): Convert from a global struct to
1561         a subclass of gimple_opt_pass along with...
1562         (pass_data_split_functions): ...new pass_data instance and...
1563         (make_pass_split_functions): ...new function.
1564         (pass_feedback_split_functions): Convert from a global struct to a
1565         subclass of gimple_opt_pass along with...
1566         (pass_data_feedback_split_functions): ...new pass_data instance and...
1567         (make_pass_feedback_split_functions): ...new function.
1568         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
1569         global struct to a subclass of simple_ipa_opt_pass along with...
1570         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
1571         instance and...
1572         (make_pass_ipa_function_and_variable_visibility): ...new function.
1573         (pass_ipa_free_inline_summary): Convert from a global struct to a
1574         subclass of simple_ipa_opt_pass along with...
1575         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
1576         (make_pass_ipa_free_inline_summary): ...new function.
1577         (pass_ipa_whole_program_visibility): Convert from a global struct to a
1578         subclass of ipa_opt_pass_d along with...
1579         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
1580         and...
1581         (make_pass_ipa_whole_program_visibility): ...new function.
1582         (pass_ipa_profile): Convert from a global struct to a subclass of
1583         ipa_opt_pass_d along with...
1584         (pass_data_ipa_profile): ...new pass_data instance and...
1585         (make_pass_ipa_profile): ...new function.
1586         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
1587         ipa_opt_pass_d along with...
1588         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
1589         (make_pass_ipa_cdtor_merge): ...new function.
1590         * ira.c (pass_ira): Convert from a global struct to a subclass of
1591         rtl_opt_pass along with...
1592         (pass_data_ira): ...new pass_data instance and...
1593         (make_pass_ira): ...new function.
1594         (pass_reload): Convert from a global struct to a subclass of
1595         rtl_opt_pass along with...
1596         (pass_data_reload): ...new pass_data instance and...
1597         (make_pass_reload): ...new function.
1598         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
1599         subclass of rtl_opt_pass along with...
1600         (pass_data_cleanup_barriers): ...new pass_data instance and...
1601         (make_pass_cleanup_barriers): ...new function.
1602         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
1603         of rtl_opt_pass along with...
1604         (pass_data_loop2): ...new pass_data instance and...
1605         (make_pass_loop2): ...new function.
1606         (pass_rtl_loop_init): Convert from a global struct to a subclass of
1607         rtl_opt_pass along with...
1608         (pass_data_rtl_loop_init): ...new pass_data instance and...
1609         (make_pass_rtl_loop_init): ...new function.
1610         (pass_rtl_loop_done): Convert from a global struct to a subclass of
1611         rtl_opt_pass along with...
1612         (pass_data_rtl_loop_done): ...new pass_data instance and...
1613         (make_pass_rtl_loop_done): ...new function.
1614         (pass_rtl_move_loop_invariants): Convert from a global struct to a
1615         subclass of rtl_opt_pass along with...
1616         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
1617         (make_pass_rtl_move_loop_invariants): ...new function.
1618         (pass_rtl_unswitch): Convert from a global struct to a subclass of
1619         rtl_opt_pass along with...
1620         (pass_data_rtl_unswitch): ...new pass_data instance and...
1621         (make_pass_rtl_unswitch): ...new function.
1622         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
1623         subclass of rtl_opt_pass along with...
1624         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
1625         and...
1626         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1627         (pass_rtl_doloop): Convert from a global struct to a subclass of
1628         rtl_opt_pass along with...
1629         (pass_data_rtl_doloop): ...new pass_data instance and...
1630         (make_pass_rtl_doloop): ...new function.
1631         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
1632         a subclass of rtl_opt_pass along with...
1633         (pass_data_lower_subreg): ...new pass_data instance and...
1634         (make_pass_lower_subreg): ...new function.
1635         (pass_lower_subreg2): Convert from a global struct to a subclass of
1636         rtl_opt_pass along with...
1637         (pass_data_lower_subreg2): ...new pass_data instance and...
1638         (make_pass_lower_subreg2): ...new function.
1639         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
1640         struct to a subclass of ipa_opt_pass_d along with...
1641         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
1642         (make_pass_ipa_lto_gimple_out): ...new function.
1643         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
1644         of ipa_opt_pass_d along with...
1645         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
1646         (make_pass_ipa_lto_finish_out): ...new function.
1647         * mode-switching.c (pass_mode_switching): Convert from a global struct
1648         to a subclass of rtl_opt_pass along with...
1649         (pass_data_mode_switching): ...new pass_data instance and...
1650         (make_pass_mode_switching): ...new function.
1651         * modulo-sched.c (pass_sms): Convert from a global struct to a
1652         subclass of rtl_opt_pass along with...
1653         (pass_data_sms): ...new pass_data instance and...
1654         (make_pass_sms): ...new function.
1655         * omp-low.c (pass_expand_omp): Convert from a global struct to a
1656         subclass of gimple_opt_pass along with...
1657         (pass_data_expand_omp): ...new pass_data instance and...
1658         (make_pass_expand_omp): ...new function.
1659         (pass_lower_omp): Convert from a global struct to a subclass of
1660         gimple_opt_pass along with...
1661         (pass_data_lower_omp): ...new pass_data instance and...
1662         (make_pass_lower_omp): ...new function.
1663         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
1664         of gimple_opt_pass along with...
1665         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
1666         (make_pass_diagnose_omp_blocks): ...new function.
1667         * passes.c (pass_early_local_passes): Convert from a global struct to
1668         a subclass of simple_ipa_opt_pass along with...
1669         (pass_data_early_local_passes): ...new pass_data instance and...
1670         (make_pass_early_local_passes): ...new function.
1671         (pass_all_early_optimizations): Convert from a global struct to a
1672         subclass of gimple_opt_pass along with...
1673         (pass_data_all_early_optimizations): ...new pass_data instance and...
1674         (make_pass_all_early_optimizations): ...new function.
1675         (pass_all_optimizations): Convert from a global struct to a subclass
1676         of gimple_opt_pass along with...
1677         (pass_data_all_optimizations): ...new pass_data instance and...
1678         (make_pass_all_optimizations): ...new function.
1679         (pass_all_optimizations_g): Convert from a global struct to a subclass
1680         of gimple_opt_pass along with...
1681         (pass_data_all_optimizations_g): ...new pass_data instance and...
1682         (make_pass_all_optimizations_g): ...new function.
1683         (pass_rest_of_compilation): Convert from a global struct to a subclass
1684         of rtl_opt_pass along with...
1685         (pass_data_rest_of_compilation): ...new pass_data instance and...
1686         (make_pass_rest_of_compilation): ...new function.
1687         (pass_postreload): Convert from a global struct to a subclass of
1688         rtl_opt_pass along with...
1689         (pass_data_postreload): ...new pass_data instance and...
1690         (make_pass_postreload): ...new function.
1691         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
1692         subclass of rtl_opt_pass along with...
1693         (pass_data_gcse2): ...new pass_data instance and...
1694         (make_pass_gcse2): ...new function.
1695         * postreload.c (pass_postreload_cse): Convert from a global struct to
1696         a subclass of rtl_opt_pass along with...
1697         (pass_data_postreload_cse): ...new pass_data instance and...
1698         (make_pass_postreload_cse): ...new function.
1699         * predict.c (pass_profile): Convert from a global struct to a subclass
1700         of gimple_opt_pass along with...
1701         (pass_data_profile): ...new pass_data instance and...
1702         (make_pass_profile): ...new function.
1703         (pass_strip_predict_hints): Convert from a global struct to a subclass
1704         of gimple_opt_pass along with...
1705         (pass_data_strip_predict_hints): ...new pass_data instance and...
1706         (make_pass_strip_predict_hints): ...new function.
1707         * recog.c (pass_peephole2): Convert from a global struct to a subclass
1708         of rtl_opt_pass along with...
1709         (pass_data_peephole2): ...new pass_data instance and...
1710         (make_pass_peephole2): ...new function.
1711         (pass_split_all_insns): Convert from a global struct to a subclass of
1712         rtl_opt_pass along with...
1713         (pass_data_split_all_insns): ...new pass_data instance and...
1714         (make_pass_split_all_insns): ...new function.
1715         (pass_split_after_reload): Convert from a global struct to a subclass
1716         of rtl_opt_pass along with...
1717         (pass_data_split_after_reload): ...new pass_data instance and...
1718         (make_pass_split_after_reload): ...new function.
1719         (pass_split_before_regstack): Convert from a global struct to a
1720         subclass of rtl_opt_pass along with...
1721         (pass_data_split_before_regstack): ...new pass_data instance and...
1722         (make_pass_split_before_regstack): ...new function.
1723         (pass_split_before_sched2): Convert from a global struct to a subclass
1724         of rtl_opt_pass along with...
1725         (pass_data_split_before_sched2): ...new pass_data instance and...
1726         (make_pass_split_before_sched2): ...new function.
1727         (pass_split_for_shorten_branches): Convert from a global struct to a
1728         subclass of rtl_opt_pass along with...
1729         (pass_data_split_for_shorten_branches): ...new pass_data instance
1730         and...
1731         (make_pass_split_for_shorten_branches): ...new function.
1732         * ree.c (pass_ree): Convert from a global struct to a subclass of
1733         rtl_opt_pass along with...
1734         (pass_data_ree): ...new pass_data instance and...
1735         (make_pass_ree): ...new function.
1736         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
1737         subclass of rtl_opt_pass along with...
1738         (pass_data_stack_regs): ...new pass_data instance and...
1739         (make_pass_stack_regs): ...new function.
1740         (pass_stack_regs_run): Convert from a global struct to a subclass of
1741         rtl_opt_pass along with...
1742         (pass_data_stack_regs_run): ...new pass_data instance and...
1743         (make_pass_stack_regs_run): ...new function.
1744         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
1745         subclass of rtl_opt_pass along with...
1746         (pass_data_cprop_hardreg): ...new pass_data instance and...
1747         (make_pass_cprop_hardreg): ...new function.
1748         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
1749         subclass of rtl_opt_pass along with...
1750         (pass_data_reginfo_init): ...new pass_data instance and...
1751         (make_pass_reginfo_init): ...new function.
1752         * regmove.c (pass_regmove): Convert from a global struct to a subclass
1753         of rtl_opt_pass along with...
1754         (pass_data_regmove): ...new pass_data instance and...
1755         (make_pass_regmove): ...new function.
1756         * regrename.c (pass_regrename): Convert from a global struct to a
1757         subclass of rtl_opt_pass along with...
1758         (pass_data_regrename): ...new pass_data instance and...
1759         (make_pass_regrename): ...new function.
1760         * reorg.c (pass_delay_slots): Convert from a global struct to a
1761         subclass of rtl_opt_pass along with...
1762         (pass_data_delay_slots): ...new pass_data instance and...
1763         (make_pass_delay_slots): ...new function.
1764         (pass_machine_reorg): Convert from a global struct to a subclass of
1765         rtl_opt_pass along with...
1766         (pass_data_machine_reorg): ...new pass_data instance and...
1767         (make_pass_machine_reorg): ...new function.
1768         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
1769         of rtl_opt_pass along with...
1770         (pass_data_sched): ...new pass_data instance and...
1771         (make_pass_sched): ...new function.
1772         (pass_sched2): Convert from a global struct to a subclass of
1773         rtl_opt_pass along with...
1774         (pass_data_sched2): ...new pass_data instance and...
1775         (make_pass_sched2): ...new function.
1776         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
1777         to a subclass of rtl_opt_pass along with...
1778         (pass_data_stack_ptr_mod): ...new pass_data instance and...
1779         (make_pass_stack_ptr_mod): ...new function.
1780         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
1781         to a subclass of rtl_opt_pass along with...
1782         (pass_data_rtl_store_motion): ...new pass_data instance and...
1783         (make_pass_rtl_store_motion): ...new function.
1784         * tracer.c (pass_tracer): Convert from a global struct to a subclass
1785         of gimple_opt_pass along with...
1786         (pass_data_tracer): ...new pass_data instance and...
1787         (make_pass_tracer): ...new function.
1788         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
1789         to a subclass of gimple_opt_pass along with...
1790         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
1791         (make_pass_diagnose_tm_blocks): ...new function.
1792         (pass_lower_tm): Convert from a global struct to a subclass of
1793         gimple_opt_pass along with...
1794         (pass_data_lower_tm): ...new pass_data instance and...
1795         (make_pass_lower_tm): ...new function.
1796         (pass_tm_init): Convert from a global struct to a subclass of
1797         gimple_opt_pass along with...
1798         (pass_data_tm_init): ...new pass_data instance and...
1799         (make_pass_tm_init): ...new function.
1800         (pass_tm_mark): Convert from a global struct to a subclass of
1801         gimple_opt_pass along with...
1802         (pass_data_tm_mark): ...new pass_data instance and...
1803         (make_pass_tm_mark): ...new function.
1804         (pass_tm_edges): Convert from a global struct to a subclass of
1805         gimple_opt_pass along with...
1806         (pass_data_tm_edges): ...new pass_data instance and...
1807         (make_pass_tm_edges): ...new function.
1808         (pass_tm_memopt): Convert from a global struct to a subclass of
1809         gimple_opt_pass along with...
1810         (pass_data_tm_memopt): ...new pass_data instance and...
1811         (make_pass_tm_memopt): ...new function.
1812         (pass_ipa_tm): Convert from a global struct to a subclass of
1813         simple_ipa_opt_pass along with...
1814         (pass_data_ipa_tm): ...new pass_data instance and...
1815         (make_pass_ipa_tm): ...new function.
1816         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
1817         subclass of gimple_opt_pass along with...
1818         (pass_data_call_cdce): ...new pass_data instance and...
1819         (make_pass_call_cdce): ...new function.
1820         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
1821         subclass of gimple_opt_pass along with...
1822         (pass_data_build_cfg): ...new pass_data instance and...
1823         (make_pass_build_cfg): ...new function.
1824         (pass_split_crit_edges): Convert from a global struct to a subclass of
1825         gimple_opt_pass along with...
1826         (pass_data_split_crit_edges): ...new pass_data instance and...
1827         (make_pass_split_crit_edges): ...new function.
1828         (pass_warn_function_return): Convert from a global struct to a
1829         subclass of gimple_opt_pass along with...
1830         (pass_data_warn_function_return): ...new pass_data instance and...
1831         (make_pass_warn_function_return): ...new function.
1832         (pass_warn_function_noreturn): Convert from a global struct to a
1833         subclass of gimple_opt_pass along with...
1834         (pass_data_warn_function_noreturn): ...new pass_data instance and...
1835         (make_pass_warn_function_noreturn): ...new function.
1836         (pass_warn_unused_result): Convert from a global struct to a subclass
1837         of gimple_opt_pass along with...
1838         (pass_data_warn_unused_result): ...new pass_data instance and...
1839         (make_pass_warn_unused_result): ...new function.
1840         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
1841         a subclass of gimple_opt_pass along with...
1842         (pass_data_merge_phi): ...new pass_data instance and...
1843         (make_pass_merge_phi): ...new function.
1844         * tree-complex.c (pass_lower_complex): Convert from a global struct to
1845         a subclass of gimple_opt_pass along with...
1846         (pass_data_lower_complex): ...new pass_data instance and...
1847         (make_pass_lower_complex): ...new function.
1848         (pass_lower_complex_O0): Convert from a global struct to a subclass of
1849         gimple_opt_pass along with...
1850         (pass_data_lower_complex_O0): ...new pass_data instance and...
1851         (make_pass_lower_complex_O0): ...new function.
1852         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
1853         subclass of gimple_opt_pass along with...
1854         (pass_data_lower_eh): ...new pass_data instance and...
1855         (make_pass_lower_eh): ...new function.
1856         (pass_refactor_eh): Convert from a global struct to a subclass of
1857         gimple_opt_pass along with...
1858         (pass_data_refactor_eh): ...new pass_data instance and...
1859         (make_pass_refactor_eh): ...new function.
1860         (pass_lower_resx): Convert from a global struct to a subclass of
1861         gimple_opt_pass along with...
1862         (pass_data_lower_resx): ...new pass_data instance and...
1863         (make_pass_lower_resx): ...new function.
1864         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1865         of gimple_opt_pass along with...
1866         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1867         (make_pass_lower_eh_dispatch): ...new function.
1868         (pass_cleanup_eh): Convert from a global struct to a subclass of
1869         gimple_opt_pass along with...
1870         (pass_data_cleanup_eh): ...new pass_data instance and...
1871         (make_pass_cleanup_eh): ...new function.
1872         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1873         to a subclass of simple_ipa_opt_pass along with...
1874         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1875         (make_pass_ipa_lower_emutls): ...new function.
1876         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1877         a subclass of gimple_opt_pass along with...
1878         (pass_data_if_conversion): ...new pass_data instance and...
1879         (make_pass_if_conversion): ...new function.
1880         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1881         subclass of gimple_opt_pass along with...
1882         (pass_data_build_ssa): ...new pass_data instance and...
1883         (make_pass_build_ssa): ...new function.
1884         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1885         global struct to a subclass of gimple_opt_pass along with...
1886         (pass_data_loop_distribution): ...new pass_data instance and...
1887         (make_pass_loop_distribution): ...new function.
1888         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1889         subclass of gimple_opt_pass along with...
1890         (pass_data_mudflap_1): ...new pass_data instance and...
1891         (make_pass_mudflap_1): ...new function.
1892         (pass_mudflap_2): Convert from a global struct to a subclass of
1893         gimple_opt_pass along with...
1894         (pass_data_mudflap_2): ...new pass_data instance and...
1895         (make_pass_mudflap_2): ...new function.
1896         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1897         subclass of gimple_opt_pass along with...
1898         (pass_data_mudflap_1): ...new pass_data instance and...
1899         (make_pass_mudflap_1): ...new function.
1900         (pass_mudflap_2): Convert from a global struct to a subclass of
1901         gimple_opt_pass along with...
1902         (pass_data_mudflap_2): ...new pass_data instance and...
1903         (make_pass_mudflap_2): ...new function.
1904         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1905         gimple_opt_pass along with...
1906         (pass_data_nrv): ...new pass_data instance and...
1907         (make_pass_nrv): ...new function.
1908         (pass_return_slot): Convert from a global struct to a subclass of
1909         gimple_opt_pass along with...
1910         (pass_data_return_slot): ...new pass_data instance and...
1911         (make_pass_return_slot): ...new function.
1912         * tree-object-size.c (pass_object_sizes): Convert from a global struct
1913         to a subclass of gimple_opt_pass along with...
1914         (pass_data_object_sizes): ...new pass_data instance and...
1915         (make_pass_object_sizes): ...new function.
1916         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1917         global struct to a subclass of gimple_opt_pass along with...
1918         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1919         and...
1920         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1921         (pass_fixup_cfg): Convert from a global struct to a subclass of
1922         gimple_opt_pass along with...
1923         (pass_data_fixup_cfg): ...new pass_data instance and...
1924         (make_pass_fixup_cfg): ...new function.
1925         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1926         (make_pass_mudflap_1): ...new function.
1927         (pass_mudflap_2): Replace declaration with that of...
1928         (make_pass_mudflap_2): ...new function.
1929         (pass_asan): Replace declaration with that of...
1930         (make_pass_asan): ...new function.
1931         (pass_asan_O0): Replace declaration with that of...
1932         (make_pass_asan_O0): ...new function.
1933         (pass_tsan): Replace declaration with that of...
1934         (make_pass_tsan): ...new function.
1935         (pass_tsan_O0): Replace declaration with that of...
1936         (make_pass_tsan_O0): ...new function.
1937         (pass_lower_cf): Replace declaration with that of...
1938         (make_pass_lower_cf): ...new function.
1939         (pass_refactor_eh): Replace declaration with that of...
1940         (make_pass_refactor_eh): ...new function.
1941         (pass_lower_eh): Replace declaration with that of...
1942         (make_pass_lower_eh): ...new function.
1943         (pass_lower_eh_dispatch): Replace declaration with that of...
1944         (make_pass_lower_eh_dispatch): ...new function.
1945         (pass_lower_resx): Replace declaration with that of...
1946         (make_pass_lower_resx): ...new function.
1947         (pass_build_cfg): Replace declaration with that of...
1948         (make_pass_build_cfg): ...new function.
1949         (pass_early_tree_profile): Replace declaration with that of...
1950         (make_pass_early_tree_profile): ...new function.
1951         (pass_cleanup_eh): Replace declaration with that of...
1952         (make_pass_cleanup_eh): ...new function.
1953         (pass_sra): Replace declaration with that of...
1954         (make_pass_sra): ...new function.
1955         (pass_sra_early): Replace declaration with that of...
1956         (make_pass_sra_early): ...new function.
1957         (pass_early_ipa_sra): Replace declaration with that of...
1958         (make_pass_early_ipa_sra): ...new function.
1959         (pass_tail_recursion): Replace declaration with that of...
1960         (make_pass_tail_recursion): ...new function.
1961         (pass_tail_calls): Replace declaration with that of...
1962         (make_pass_tail_calls): ...new function.
1963         (pass_tree_loop): Replace declaration with that of...
1964         (make_pass_tree_loop): ...new function.
1965         (pass_tree_loop_init): Replace declaration with that of...
1966         (make_pass_tree_loop_init): ...new function.
1967         (pass_lim): Replace declaration with that of...
1968         (make_pass_lim): ...new function.
1969         (pass_tree_unswitch): Replace declaration with that of...
1970         (make_pass_tree_unswitch): ...new function.
1971         (pass_predcom): Replace declaration with that of...
1972         (make_pass_predcom): ...new function.
1973         (pass_iv_canon): Replace declaration with that of...
1974         (make_pass_iv_canon): ...new function.
1975         (pass_scev_cprop): Replace declaration with that of...
1976         (make_pass_scev_cprop): ...new function.
1977         (pass_empty_loop): Replace declaration with that of...
1978         (make_pass_empty_loop): ...new function.
1979         (pass_record_bounds): Replace declaration with that of...
1980         (make_pass_record_bounds): ...new function.
1981         (pass_graphite): Replace declaration with that of...
1982         (make_pass_graphite): ...new function.
1983         (pass_graphite_transforms): Replace declaration with that of...
1984         (make_pass_graphite_transforms): ...new function.
1985         (pass_if_conversion): Replace declaration with that of...
1986         (make_pass_if_conversion): ...new function.
1987         (pass_loop_distribution): Replace declaration with that of...
1988         (make_pass_loop_distribution): ...new function.
1989         (pass_vectorize): Replace declaration with that of...
1990         (make_pass_vectorize): ...new function.
1991         (pass_slp_vectorize): Replace declaration with that of...
1992         (make_pass_slp_vectorize): ...new function.
1993         (pass_complete_unroll): Replace declaration with that of...
1994         (make_pass_complete_unroll): ...new function.
1995         (pass_complete_unrolli): Replace declaration with that of...
1996         (make_pass_complete_unrolli): ...new function.
1997         (pass_parallelize_loops): Replace declaration with that of...
1998         (make_pass_parallelize_loops): ...new function.
1999         (pass_loop_prefetch): Replace declaration with that of...
2000         (make_pass_loop_prefetch): ...new function.
2001         (pass_iv_optimize): Replace declaration with that of...
2002         (make_pass_iv_optimize): ...new function.
2003         (pass_tree_loop_done): Replace declaration with that of...
2004         (make_pass_tree_loop_done): ...new function.
2005         (pass_ch): Replace declaration with that of...
2006         (make_pass_ch): ...new function.
2007         (pass_ccp): Replace declaration with that of...
2008         (make_pass_ccp): ...new function.
2009         (pass_phi_only_cprop): Replace declaration with that of...
2010         (make_pass_phi_only_cprop): ...new function.
2011         (pass_build_ssa): Replace declaration with that of...
2012         (make_pass_build_ssa): ...new function.
2013         (pass_build_alias): Replace declaration with that of...
2014         (make_pass_build_alias): ...new function.
2015         (pass_build_ealias): Replace declaration with that of...
2016         (make_pass_build_ealias): ...new function.
2017         (pass_dominator): Replace declaration with that of...
2018         (make_pass_dominator): ...new function.
2019         (pass_dce): Replace declaration with that of...
2020         (make_pass_dce): ...new function.
2021         (pass_dce_loop): Replace declaration with that of...
2022         (make_pass_dce_loop): ...new function.
2023         (pass_cd_dce): Replace declaration with that of...
2024         (make_pass_cd_dce): ...new function.
2025         (pass_call_cdce): Replace declaration with that of...
2026         (make_pass_call_cdce): ...new function.
2027         (pass_merge_phi): Replace declaration with that of...
2028         (make_pass_merge_phi): ...new function.
2029         (pass_split_crit_edges): Replace declaration with that of...
2030         (make_pass_split_crit_edges): ...new function.
2031         (pass_pre): Replace declaration with that of...
2032         (make_pass_pre): ...new function.
2033         (pass_profile): Replace declaration with that of...
2034         (make_pass_profile): ...new function.
2035         (pass_strip_predict_hints): Replace declaration with that of...
2036         (make_pass_strip_predict_hints): ...new function.
2037         (pass_lower_complex_O0): Replace declaration with that of...
2038         (make_pass_lower_complex_O0): ...new function.
2039         (pass_lower_complex): Replace declaration with that of...
2040         (make_pass_lower_complex): ...new function.
2041         (pass_lower_vector): Replace declaration with that of...
2042         (make_pass_lower_vector): ...new function.
2043         (pass_lower_vector_ssa): Replace declaration with that of...
2044         (make_pass_lower_vector_ssa): ...new function.
2045         (pass_lower_omp): Replace declaration with that of...
2046         (make_pass_lower_omp): ...new function.
2047         (pass_diagnose_omp_blocks): Replace declaration with that of...
2048         (make_pass_diagnose_omp_blocks): ...new function.
2049         (pass_expand_omp): Replace declaration with that of...
2050         (make_pass_expand_omp): ...new function.
2051         (pass_expand_omp_ssa): Replace declaration with that of...
2052         (make_pass_expand_omp_ssa): ...new function.
2053         (pass_object_sizes): Replace declaration with that of...
2054         (make_pass_object_sizes): ...new function.
2055         (pass_strlen): Replace declaration with that of...
2056         (make_pass_strlen): ...new function.
2057         (pass_fold_builtins): Replace declaration with that of...
2058         (make_pass_fold_builtins): ...new function.
2059         (pass_stdarg): Replace declaration with that of...
2060         (make_pass_stdarg): ...new function.
2061         (pass_early_warn_uninitialized): Replace declaration with that of...
2062         (make_pass_early_warn_uninitialized): ...new function.
2063         (pass_late_warn_uninitialized): Replace declaration with that of...
2064         (make_pass_late_warn_uninitialized): ...new function.
2065         (pass_cse_reciprocals): Replace declaration with that of...
2066         (make_pass_cse_reciprocals): ...new function.
2067         (pass_cse_sincos): Replace declaration with that of...
2068         (make_pass_cse_sincos): ...new function.
2069         (pass_optimize_bswap): Replace declaration with that of...
2070         (make_pass_optimize_bswap): ...new function.
2071         (pass_optimize_widening_mul): Replace declaration with that of...
2072         (make_pass_optimize_widening_mul): ...new function.
2073         (pass_warn_function_return): Replace declaration with that of...
2074         (make_pass_warn_function_return): ...new function.
2075         (pass_warn_function_noreturn): Replace declaration with that of...
2076         (make_pass_warn_function_noreturn): ...new function.
2077         (pass_cselim): Replace declaration with that of...
2078         (make_pass_cselim): ...new function.
2079         (pass_phiopt): Replace declaration with that of...
2080         (make_pass_phiopt): ...new function.
2081         (pass_forwprop): Replace declaration with that of...
2082         (make_pass_forwprop): ...new function.
2083         (pass_phiprop): Replace declaration with that of...
2084         (make_pass_phiprop): ...new function.
2085         (pass_tree_ifcombine): Replace declaration with that of...
2086         (make_pass_tree_ifcombine): ...new function.
2087         (pass_dse): Replace declaration with that of...
2088         (make_pass_dse): ...new function.
2089         (pass_nrv): Replace declaration with that of...
2090         (make_pass_nrv): ...new function.
2091         (pass_rename_ssa_copies): Replace declaration with that of...
2092         (make_pass_rename_ssa_copies): ...new function.
2093         (pass_sink_code): Replace declaration with that of...
2094         (make_pass_sink_code): ...new function.
2095         (pass_fre): Replace declaration with that of...
2096         (make_pass_fre): ...new function.
2097         (pass_check_data_deps): Replace declaration with that of...
2098         (make_pass_check_data_deps): ...new function.
2099         (pass_copy_prop): Replace declaration with that of...
2100         (make_pass_copy_prop): ...new function.
2101         (pass_vrp): Replace declaration with that of...
2102         (make_pass_vrp): ...new function.
2103         (pass_uncprop): Replace declaration with that of...
2104         (make_pass_uncprop): ...new function.
2105         (pass_return_slot): Replace declaration with that of...
2106         (make_pass_return_slot): ...new function.
2107         (pass_reassoc): Replace declaration with that of...
2108         (make_pass_reassoc): ...new function.
2109         (pass_rebuild_cgraph_edges): Replace declaration with that of...
2110         (make_pass_rebuild_cgraph_edges): ...new function.
2111         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2112         (make_pass_remove_cgraph_callee_edges): ...new function.
2113         (pass_build_cgraph_edges): Replace declaration with that of...
2114         (make_pass_build_cgraph_edges): ...new function.
2115         (pass_local_pure_const): Replace declaration with that of...
2116         (make_pass_local_pure_const): ...new function.
2117         (pass_tracer): Replace declaration with that of...
2118         (make_pass_tracer): ...new function.
2119         (pass_warn_unused_result): Replace declaration with that of...
2120         (make_pass_warn_unused_result): ...new function.
2121         (pass_diagnose_tm_blocks): Replace declaration with that of...
2122         (make_pass_diagnose_tm_blocks): ...new function.
2123         (pass_lower_tm): Replace declaration with that of...
2124         (make_pass_lower_tm): ...new function.
2125         (pass_tm_init): Replace declaration with that of...
2126         (make_pass_tm_init): ...new function.
2127         (pass_tm_mark): Replace declaration with that of...
2128         (make_pass_tm_mark): ...new function.
2129         (pass_tm_memopt): Replace declaration with that of...
2130         (make_pass_tm_memopt): ...new function.
2131         (pass_tm_edges): Replace declaration with that of...
2132         (make_pass_tm_edges): ...new function.
2133         (pass_split_functions): Replace declaration with that of...
2134         (make_pass_split_functions): ...new function.
2135         (pass_feedback_split_functions): Replace declaration with that of...
2136         (make_pass_feedback_split_functions): ...new function.
2137         (pass_strength_reduction): Replace declaration with that of...
2138         (make_pass_strength_reduction): ...new function.
2139         (pass_ipa_lower_emutls): Replace declaration with that of...
2140         (make_pass_ipa_lower_emutls): ...new function.
2141         (pass_ipa_function_and_variable_visibility): Replace declaration with
2142         that of...
2143         (make_pass_ipa_function_and_variable_visibility): ...new function.
2144         (pass_ipa_tree_profile): Replace declaration with that of...
2145         (make_pass_ipa_tree_profile): ...new function.
2146         (pass_early_local_passes): Replace declaration with that of...
2147         (make_pass_early_local_passes): ...new function.
2148         (pass_ipa_whole_program_visibility): Replace declaration with that
2149         of...
2150         (make_pass_ipa_whole_program_visibility): ...new function.
2151         (pass_ipa_lto_gimple_out): Replace declaration with that of...
2152         (make_pass_ipa_lto_gimple_out): ...new function.
2153         (pass_ipa_increase_alignment): Replace declaration with that of...
2154         (make_pass_ipa_increase_alignment): ...new function.
2155         (pass_ipa_inline): Replace declaration with that of...
2156         (make_pass_ipa_inline): ...new function.
2157         (pass_ipa_free_lang_data): Replace declaration with that of...
2158         (make_pass_ipa_free_lang_data): ...new function.
2159         (pass_ipa_free_inline_summary): Replace declaration with that of...
2160         (make_pass_ipa_free_inline_summary): ...new function.
2161         (pass_ipa_cp): Replace declaration with that of...
2162         (make_pass_ipa_cp): ...new function.
2163         (pass_ipa_reference): Replace declaration with that of...
2164         (make_pass_ipa_reference): ...new function.
2165         (pass_ipa_pure_const): Replace declaration with that of...
2166         (make_pass_ipa_pure_const): ...new function.
2167         (pass_ipa_pta): Replace declaration with that of...
2168         (make_pass_ipa_pta): ...new function.
2169         (pass_ipa_lto_finish_out): Replace declaration with that of...
2170         (make_pass_ipa_lto_finish_out): ...new function.
2171         (pass_ipa_tm): Replace declaration with that of...
2172         (make_pass_ipa_tm): ...new function.
2173         (pass_ipa_profile): Replace declaration with that of...
2174         (make_pass_ipa_profile): ...new function.
2175         (pass_ipa_cdtor_merge): Replace declaration with that of...
2176         (make_pass_ipa_cdtor_merge): ...new function.
2177         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
2178         of...
2179         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2180         (pass_init_datastructures): Replace declaration with that of...
2181         (make_pass_init_datastructures): ...new function.
2182         (pass_fixup_cfg): Replace declaration with that of...
2183         (make_pass_fixup_cfg): ...new function.
2184         (pass_expand): Replace declaration with that of...
2185         (make_pass_expand): ...new function.
2186         (pass_instantiate_virtual_regs): Replace declaration with that of...
2187         (make_pass_instantiate_virtual_regs): ...new function.
2188         (pass_rtl_fwprop): Replace declaration with that of...
2189         (make_pass_rtl_fwprop): ...new function.
2190         (pass_rtl_fwprop_addr): Replace declaration with that of...
2191         (make_pass_rtl_fwprop_addr): ...new function.
2192         (pass_jump): Replace declaration with that of...
2193         (make_pass_jump): ...new function.
2194         (pass_jump2): Replace declaration with that of...
2195         (make_pass_jump2): ...new function.
2196         (pass_lower_subreg): Replace declaration with that of...
2197         (make_pass_lower_subreg): ...new function.
2198         (pass_cse): Replace declaration with that of...
2199         (make_pass_cse): ...new function.
2200         (pass_fast_rtl_dce): Replace declaration with that of...
2201         (make_pass_fast_rtl_dce): ...new function.
2202         (pass_ud_rtl_dce): Replace declaration with that of...
2203         (make_pass_ud_rtl_dce): ...new function.
2204         (pass_rtl_dce): Replace declaration with that of...
2205         (make_pass_rtl_dce): ...new function.
2206         (pass_rtl_dse1): Replace declaration with that of...
2207         (make_pass_rtl_dse1): ...new function.
2208         (pass_rtl_dse2): Replace declaration with that of...
2209         (make_pass_rtl_dse2): ...new function.
2210         (pass_rtl_dse3): Replace declaration with that of...
2211         (make_pass_rtl_dse3): ...new function.
2212         (pass_rtl_cprop): Replace declaration with that of...
2213         (make_pass_rtl_cprop): ...new function.
2214         (pass_rtl_pre): Replace declaration with that of...
2215         (make_pass_rtl_pre): ...new function.
2216         (pass_rtl_hoist): Replace declaration with that of...
2217         (make_pass_rtl_hoist): ...new function.
2218         (pass_rtl_store_motion): Replace declaration with that of...
2219         (make_pass_rtl_store_motion): ...new function.
2220         (pass_cse_after_global_opts): Replace declaration with that of...
2221         (make_pass_cse_after_global_opts): ...new function.
2222         (pass_rtl_ifcvt): Replace declaration with that of...
2223         (make_pass_rtl_ifcvt): ...new function.
2224         (pass_into_cfg_layout_mode): Replace declaration with that of...
2225         (make_pass_into_cfg_layout_mode): ...new function.
2226         (pass_outof_cfg_layout_mode): Replace declaration with that of...
2227         (make_pass_outof_cfg_layout_mode): ...new function.
2228         (pass_loop2): Replace declaration with that of...
2229         (make_pass_loop2): ...new function.
2230         (pass_rtl_loop_init): Replace declaration with that of...
2231         (make_pass_rtl_loop_init): ...new function.
2232         (pass_rtl_move_loop_invariants): Replace declaration with that of...
2233         (make_pass_rtl_move_loop_invariants): ...new function.
2234         (pass_rtl_unswitch): Replace declaration with that of...
2235         (make_pass_rtl_unswitch): ...new function.
2236         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
2237         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2238         (pass_rtl_doloop): Replace declaration with that of...
2239         (make_pass_rtl_doloop): ...new function.
2240         (pass_rtl_loop_done): Replace declaration with that of...
2241         (make_pass_rtl_loop_done): ...new function.
2242         (pass_web): Replace declaration with that of...
2243         (make_pass_web): ...new function.
2244         (pass_cse2): Replace declaration with that of...
2245         (make_pass_cse2): ...new function.
2246         (pass_df_initialize_opt): Replace declaration with that of...
2247         (make_pass_df_initialize_opt): ...new function.
2248         (pass_df_initialize_no_opt): Replace declaration with that of...
2249         (make_pass_df_initialize_no_opt): ...new function.
2250         (pass_reginfo_init): Replace declaration with that of...
2251         (make_pass_reginfo_init): ...new function.
2252         (pass_inc_dec): Replace declaration with that of...
2253         (make_pass_inc_dec): ...new function.
2254         (pass_stack_ptr_mod): Replace declaration with that of...
2255         (make_pass_stack_ptr_mod): ...new function.
2256         (pass_initialize_regs): Replace declaration with that of...
2257         (make_pass_initialize_regs): ...new function.
2258         (pass_combine): Replace declaration with that of...
2259         (make_pass_combine): ...new function.
2260         (pass_if_after_combine): Replace declaration with that of...
2261         (make_pass_if_after_combine): ...new function.
2262         (pass_ree): Replace declaration with that of...
2263         (make_pass_ree): ...new function.
2264         (pass_partition_blocks): Replace declaration with that of...
2265         (make_pass_partition_blocks): ...new function.
2266         (pass_match_asm_constraints): Replace declaration with that of...
2267         (make_pass_match_asm_constraints): ...new function.
2268         (pass_regmove): Replace declaration with that of...
2269         (make_pass_regmove): ...new function.
2270         (pass_split_all_insns): Replace declaration with that of...
2271         (make_pass_split_all_insns): ...new function.
2272         (pass_fast_rtl_byte_dce): Replace declaration with that of...
2273         (make_pass_fast_rtl_byte_dce): ...new function.
2274         (pass_lower_subreg2): Replace declaration with that of...
2275         (make_pass_lower_subreg2): ...new function.
2276         (pass_mode_switching): Replace declaration with that of...
2277         (make_pass_mode_switching): ...new function.
2278         (pass_sms): Replace declaration with that of...
2279         (make_pass_sms): ...new function.
2280         (pass_sched): Replace declaration with that of...
2281         (make_pass_sched): ...new function.
2282         (pass_ira): Replace declaration with that of...
2283         (make_pass_ira): ...new function.
2284         (pass_reload): Replace declaration with that of...
2285         (make_pass_reload): ...new function.
2286         (pass_clean_state): Replace declaration with that of...
2287         (make_pass_clean_state): ...new function.
2288         (pass_branch_prob): Replace declaration with that of...
2289         (make_pass_branch_prob): ...new function.
2290         (pass_value_profile_transformations): Replace declaration with that
2291         of...
2292         (make_pass_value_profile_transformations): ...new function.
2293         (pass_postreload_cse): Replace declaration with that of...
2294         (make_pass_postreload_cse): ...new function.
2295         (pass_gcse2): Replace declaration with that of...
2296         (make_pass_gcse2): ...new function.
2297         (pass_split_after_reload): Replace declaration with that of...
2298         (make_pass_split_after_reload): ...new function.
2299         (pass_branch_target_load_optimize1): Replace declaration with that
2300         of...
2301         (make_pass_branch_target_load_optimize1): ...new function.
2302         (pass_thread_prologue_and_epilogue): Replace declaration with that
2303         of...
2304         (make_pass_thread_prologue_and_epilogue): ...new function.
2305         (pass_stack_adjustments): Replace declaration with that of...
2306         (make_pass_stack_adjustments): ...new function.
2307         (pass_peephole2): Replace declaration with that of...
2308         (make_pass_peephole2): ...new function.
2309         (pass_if_after_reload): Replace declaration with that of...
2310         (make_pass_if_after_reload): ...new function.
2311         (pass_regrename): Replace declaration with that of...
2312         (make_pass_regrename): ...new function.
2313         (pass_cprop_hardreg): Replace declaration with that of...
2314         (make_pass_cprop_hardreg): ...new function.
2315         (pass_reorder_blocks): Replace declaration with that of...
2316         (make_pass_reorder_blocks): ...new function.
2317         (pass_branch_target_load_optimize2): Replace declaration with that
2318         of...
2319         (make_pass_branch_target_load_optimize2): ...new function.
2320         (pass_leaf_regs): Replace declaration with that of...
2321         (make_pass_leaf_regs): ...new function.
2322         (pass_split_before_sched2): Replace declaration with that of...
2323         (make_pass_split_before_sched2): ...new function.
2324         (pass_compare_elim_after_reload): Replace declaration with that of...
2325         (make_pass_compare_elim_after_reload): ...new function.
2326         (pass_sched2): Replace declaration with that of...
2327         (make_pass_sched2): ...new function.
2328         (pass_stack_regs): Replace declaration with that of...
2329         (make_pass_stack_regs): ...new function.
2330         (pass_stack_regs_run): Replace declaration with that of...
2331         (make_pass_stack_regs_run): ...new function.
2332         (pass_df_finish): Replace declaration with that of...
2333         (make_pass_df_finish): ...new function.
2334         (pass_compute_alignments): Replace declaration with that of...
2335         (make_pass_compute_alignments): ...new function.
2336         (pass_duplicate_computed_gotos): Replace declaration with that of...
2337         (make_pass_duplicate_computed_gotos): ...new function.
2338         (pass_variable_tracking): Replace declaration with that of...
2339         (make_pass_variable_tracking): ...new function.
2340         (pass_free_cfg): Replace declaration with that of...
2341         (make_pass_free_cfg): ...new function.
2342         (pass_machine_reorg): Replace declaration with that of...
2343         (make_pass_machine_reorg): ...new function.
2344         (pass_cleanup_barriers): Replace declaration with that of...
2345         (make_pass_cleanup_barriers): ...new function.
2346         (pass_delay_slots): Replace declaration with that of...
2347         (make_pass_delay_slots): ...new function.
2348         (pass_split_for_shorten_branches): Replace declaration with that of...
2349         (make_pass_split_for_shorten_branches): ...new function.
2350         (pass_split_before_regstack): Replace declaration with that of...
2351         (make_pass_split_before_regstack): ...new function.
2352         (pass_convert_to_eh_region_ranges): Replace declaration with that
2353         of...
2354         (make_pass_convert_to_eh_region_ranges): ...new function.
2355         (pass_shorten_branches): Replace declaration with that of...
2356         (make_pass_shorten_branches): ...new function.
2357         (pass_set_nothrow_function_flags): Replace declaration with that of...
2358         (make_pass_set_nothrow_function_flags): ...new function.
2359         (pass_dwarf2_frame): Replace declaration with that of...
2360         (make_pass_dwarf2_frame): ...new function.
2361         (pass_final): Replace declaration with that of...
2362         (make_pass_final): ...new function.
2363         (pass_rtl_seqabstr): Replace declaration with that of...
2364         (make_pass_rtl_seqabstr): ...new function.
2365         (pass_release_ssa_names): Replace declaration with that of...
2366         (make_pass_release_ssa_names): ...new function.
2367         (pass_early_inline): Replace declaration with that of...
2368         (make_pass_early_inline): ...new function.
2369         (pass_inline_parameters): Replace declaration with that of...
2370         (make_pass_inline_parameters): ...new function.
2371         (pass_update_address_taken): Replace declaration with that of...
2372         (make_pass_update_address_taken): ...new function.
2373         (pass_convert_switch): Replace declaration with that of...
2374         (make_pass_convert_switch): ...new function.
2375         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
2376         to a subclass of simple_ipa_opt_pass along with...
2377         (pass_data_ipa_tree_profile): ...new pass_data instance and...
2378         (make_pass_ipa_tree_profile): ...new function.
2379         * tree-sra.c (pass_sra_early): Convert from a global struct to a
2380         subclass of gimple_opt_pass along with...
2381         (pass_data_sra_early): ...new pass_data instance and...
2382         (make_pass_sra_early): ...new function.
2383         (pass_sra): Convert from a global struct to a subclass of
2384         gimple_opt_pass along with...
2385         (pass_data_sra): ...new pass_data instance and...
2386         (make_pass_sra): ...new function.
2387         (pass_early_ipa_sra): Convert from a global struct to a subclass of
2388         gimple_opt_pass along with...
2389         (pass_data_early_ipa_sra): ...new pass_data instance and...
2390         (make_pass_early_ipa_sra): ...new function.
2391         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
2392         subclass of gimple_opt_pass along with...
2393         (pass_data_ccp): ...new pass_data instance and...
2394         (make_pass_ccp): ...new function.
2395         (pass_fold_builtins): Convert from a global struct to a subclass of
2396         gimple_opt_pass along with...
2397         (pass_data_fold_builtins): ...new pass_data instance and...
2398         (make_pass_fold_builtins): ...new function.
2399         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
2400         subclass of gimple_opt_pass along with...
2401         (pass_data_copy_prop): ...new pass_data instance and...
2402         (make_pass_copy_prop): ...new function.
2403         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
2404         global struct to a subclass of gimple_opt_pass along with...
2405         (pass_data_rename_ssa_copies): ...new pass_data instance and...
2406         (make_pass_rename_ssa_copies): ...new function.
2407         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
2408         subclass of gimple_opt_pass along with...
2409         (pass_data_dce): ...new pass_data instance and...
2410         (make_pass_dce): ...new function.
2411         (pass_dce_loop): Convert from a global struct to a subclass of
2412         gimple_opt_pass along with...
2413         (pass_data_dce_loop): ...new pass_data instance and...
2414         (make_pass_dce_loop): ...new function.
2415         (pass_cd_dce): Convert from a global struct to a subclass of
2416         gimple_opt_pass along with...
2417         (pass_data_cd_dce): ...new pass_data instance and...
2418         (make_pass_cd_dce): ...new function.
2419         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
2420         subclass of gimple_opt_pass along with...
2421         (pass_data_dominator): ...new pass_data instance and...
2422         (make_pass_dominator): ...new function.
2423         (pass_phi_only_cprop): Convert from a global struct to a subclass of
2424         gimple_opt_pass along with...
2425         (pass_data_phi_only_cprop): ...new pass_data instance and...
2426         (make_pass_phi_only_cprop): ...new function.
2427         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
2428         subclass of gimple_opt_pass along with...
2429         (pass_data_dse): ...new pass_data instance and...
2430         (make_pass_dse): ...new function.
2431         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
2432         a subclass of gimple_opt_pass along with...
2433         (pass_data_forwprop): ...new pass_data instance and...
2434         (make_pass_forwprop): ...new function.
2435         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
2436         struct to a subclass of gimple_opt_pass along with...
2437         (pass_data_tree_ifcombine): ...new pass_data instance and...
2438         (make_pass_tree_ifcombine): ...new function.
2439         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
2440         subclass of gimple_opt_pass along with...
2441         (pass_data_ch): ...new pass_data instance and...
2442         (make_pass_ch): ...new function.
2443         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
2444         subclass of gimple_opt_pass along with...
2445         (pass_data_tree_loop): ...new pass_data instance and...
2446         (make_pass_tree_loop): ...new function.
2447         (pass_tree_loop_init): Convert from a global struct to a subclass of
2448         gimple_opt_pass along with...
2449         (pass_data_tree_loop_init): ...new pass_data instance and...
2450         (make_pass_tree_loop_init): ...new function.
2451         (pass_lim): Convert from a global struct to a subclass of
2452         gimple_opt_pass along with...
2453         (pass_data_lim): ...new pass_data instance and...
2454         (make_pass_lim): ...new function.
2455         (pass_tree_unswitch): Convert from a global struct to a subclass of
2456         gimple_opt_pass along with...
2457         (pass_data_tree_unswitch): ...new pass_data instance and...
2458         (make_pass_tree_unswitch): ...new function.
2459         (pass_predcom): Convert from a global struct to a subclass of
2460         gimple_opt_pass along with...
2461         (pass_data_predcom): ...new pass_data instance and...
2462         (make_pass_predcom): ...new function.
2463         (pass_vectorize): Convert from a global struct to a subclass of
2464         gimple_opt_pass along with...
2465         (pass_data_vectorize): ...new pass_data instance and...
2466         (make_pass_vectorize): ...new function.
2467         (pass_graphite): Convert from a global struct to a subclass of
2468         gimple_opt_pass along with...
2469         (pass_data_graphite): ...new pass_data instance and...
2470         (make_pass_graphite): ...new function.
2471         (pass_graphite_transforms): Convert from a global struct to a subclass
2472         of gimple_opt_pass along with...
2473         (pass_data_graphite_transforms): ...new pass_data instance and...
2474         (make_pass_graphite_transforms): ...new function.
2475         (pass_check_data_deps): Convert from a global struct to a subclass of
2476         gimple_opt_pass along with...
2477         (pass_data_check_data_deps): ...new pass_data instance and...
2478         (make_pass_check_data_deps): ...new function.
2479         (pass_iv_canon): Convert from a global struct to a subclass of
2480         gimple_opt_pass along with...
2481         (pass_data_iv_canon): ...new pass_data instance and...
2482         (make_pass_iv_canon): ...new function.
2483         (pass_scev_cprop): Convert from a global struct to a subclass of
2484         gimple_opt_pass along with...
2485         (pass_data_scev_cprop): ...new pass_data instance and...
2486         (make_pass_scev_cprop): ...new function.
2487         (pass_record_bounds): Convert from a global struct to a subclass of
2488         gimple_opt_pass along with...
2489         (pass_data_record_bounds): ...new pass_data instance and...
2490         (make_pass_record_bounds): ...new function.
2491         (pass_complete_unroll): Convert from a global struct to a subclass of
2492         gimple_opt_pass along with...
2493         (pass_data_complete_unroll): ...new pass_data instance and...
2494         (make_pass_complete_unroll): ...new function.
2495         (pass_complete_unrolli): Convert from a global struct to a subclass of
2496         gimple_opt_pass along with...
2497         (pass_data_complete_unrolli): ...new pass_data instance and...
2498         (make_pass_complete_unrolli): ...new function.
2499         (pass_parallelize_loops): Convert from a global struct to a subclass
2500         of gimple_opt_pass along with...
2501         (pass_data_parallelize_loops): ...new pass_data instance and...
2502         (make_pass_parallelize_loops): ...new function.
2503         (pass_loop_prefetch): Convert from a global struct to a subclass of
2504         gimple_opt_pass along with...
2505         (pass_data_loop_prefetch): ...new pass_data instance and...
2506         (make_pass_loop_prefetch): ...new function.
2507         (pass_iv_optimize): Convert from a global struct to a subclass of
2508         gimple_opt_pass along with...
2509         (pass_data_iv_optimize): ...new pass_data instance and...
2510         (make_pass_iv_optimize): ...new function.
2511         (pass_tree_loop_done): Convert from a global struct to a subclass of
2512         gimple_opt_pass along with...
2513         (pass_data_tree_loop_done): ...new pass_data instance and...
2514         (make_pass_tree_loop_done): ...new function.
2515         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
2516         struct to a subclass of gimple_opt_pass along with...
2517         (pass_data_cse_reciprocals): ...new pass_data instance and...
2518         (make_pass_cse_reciprocals): ...new function.
2519         (pass_cse_sincos): Convert from a global struct to a subclass of
2520         gimple_opt_pass along with...
2521         (pass_data_cse_sincos): ...new pass_data instance and...
2522         (make_pass_cse_sincos): ...new function.
2523         (pass_optimize_bswap): Convert from a global struct to a subclass of
2524         gimple_opt_pass along with...
2525         (pass_data_optimize_bswap): ...new pass_data instance and...
2526         (make_pass_optimize_bswap): ...new function.
2527         (pass_optimize_widening_mul): Convert from a global struct to a
2528         subclass of gimple_opt_pass along with...
2529         (pass_data_optimize_widening_mul): ...new pass_data instance and...
2530         (make_pass_optimize_widening_mul): ...new function.
2531         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
2532         subclass of gimple_opt_pass along with...
2533         (pass_data_phiopt): ...new pass_data instance and...
2534         (make_pass_phiopt): ...new function.
2535         (pass_cselim): Convert from a global struct to a subclass of
2536         gimple_opt_pass along with...
2537         (pass_data_cselim): ...new pass_data instance and...
2538         (make_pass_cselim): ...new function.
2539         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
2540         subclass of gimple_opt_pass along with...
2541         (pass_data_phiprop): ...new pass_data instance and...
2542         (make_pass_phiprop): ...new function.
2543         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
2544         subclass of gimple_opt_pass along with...
2545         (pass_data_pre): ...new pass_data instance and...
2546         (make_pass_pre): ...new function.
2547         (pass_fre): Convert from a global struct to a subclass of
2548         gimple_opt_pass along with...
2549         (pass_data_fre): ...new pass_data instance and...
2550         (make_pass_fre): ...new function.
2551         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
2552         subclass of gimple_opt_pass along with...
2553         (pass_data_reassoc): ...new pass_data instance and...
2554         (make_pass_reassoc): ...new function.
2555         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
2556         subclass of gimple_opt_pass along with...
2557         (pass_data_sink_code): ...new pass_data instance and...
2558         (make_pass_sink_code): ...new function.
2559         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
2560         subclass of gimple_opt_pass along with...
2561         (pass_data_strlen): ...new pass_data instance and...
2562         (make_pass_strlen): ...new function.
2563         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
2564         struct to a subclass of gimple_opt_pass along with...
2565         (pass_data_build_alias): ...new pass_data instance and...
2566         (make_pass_build_alias): ...new function.
2567         (pass_build_ealias): Convert from a global struct to a subclass of
2568         gimple_opt_pass along with...
2569         (pass_data_build_ealias): ...new pass_data instance and...
2570         (make_pass_build_ealias): ...new function.
2571         (pass_ipa_pta): Convert from a global struct to a subclass of
2572         simple_ipa_opt_pass along with...
2573         (pass_data_ipa_pta): ...new pass_data instance and...
2574         (make_pass_ipa_pta): ...new function.
2575         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
2576         subclass of gimple_opt_pass along with...
2577         (pass_data_uncprop): ...new pass_data instance and...
2578         (make_pass_uncprop): ...new function.
2579         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
2580         global struct to a subclass of gimple_opt_pass along with...
2581         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
2582         (make_pass_late_warn_uninitialized): ...new function.
2583         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
2584         to a subclass of gimple_opt_pass along with...
2585         (pass_data_init_datastructures): ...new pass_data instance and...
2586         (make_pass_init_datastructures): ...new function.
2587         (pass_early_warn_uninitialized): Convert from a global struct to a
2588         subclass of gimple_opt_pass along with...
2589         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
2590         (make_pass_early_warn_uninitialized): ...new function.
2591         (pass_update_address_taken): Convert from a global struct to a
2592         subclass of gimple_opt_pass along with...
2593         (pass_data_update_address_taken): ...new pass_data instance and...
2594         (make_pass_update_address_taken): ...new function.
2595         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
2596         struct to a subclass of gimple_opt_pass along with...
2597         (pass_data_release_ssa_names): ...new pass_data instance and...
2598         (make_pass_release_ssa_names): ...new function.
2599         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
2600         subclass of gimple_opt_pass along with...
2601         (pass_data_stdarg): ...new pass_data instance and...
2602         (make_pass_stdarg): ...new function.
2603         * tree-switch-conversion.c (pass_convert_switch): Convert from a
2604         global struct to a subclass of gimple_opt_pass along with...
2605         (pass_data_convert_switch): ...new pass_data instance and...
2606         (make_pass_convert_switch): ...new function.
2607         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
2608         to a subclass of gimple_opt_pass along with...
2609         (pass_data_tail_recursion): ...new pass_data instance and...
2610         (make_pass_tail_recursion): ...new function.
2611         (pass_tail_calls): Convert from a global struct to a subclass of
2612         gimple_opt_pass along with...
2613         (pass_data_tail_calls): ...new pass_data instance and...
2614         (make_pass_tail_calls): ...new function.
2615         * tree-vect-generic.c (pass_lower_vector): Convert from a global
2616         struct to a subclass of gimple_opt_pass along with...
2617         (pass_data_lower_vector): ...new pass_data instance and...
2618         (make_pass_lower_vector): ...new function.
2619         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
2620         gimple_opt_pass along with...
2621         (pass_data_lower_vector_ssa): ...new pass_data instance and...
2622         (make_pass_lower_vector_ssa): ...new function.
2623         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
2624         to a subclass of gimple_opt_pass along with...
2625         (pass_data_slp_vectorize): ...new pass_data instance and...
2626         (make_pass_slp_vectorize): ...new function.
2627         (pass_ipa_increase_alignment): Convert from a global struct to a
2628         subclass of simple_ipa_opt_pass along with...
2629         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
2630         (make_pass_ipa_increase_alignment): ...new function.
2631         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
2632         gimple_opt_pass along with...
2633         (pass_data_vrp): ...new pass_data instance and...
2634         (make_pass_vrp): ...new function.
2635         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
2636         subclass of simple_ipa_opt_pass along with...
2637         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
2638         (make_pass_ipa_free_lang_data): ...new function.
2639         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
2640         gimple_opt_pass along with...
2641         (pass_data_tsan): ...new pass_data instance and...
2642         (make_pass_tsan): ...new function.
2643         (pass_tsan_O0): Convert from a global struct to a subclass of
2644         gimple_opt_pass along with...
2645         (pass_data_tsan_O0): ...new pass_data instance and...
2646         (make_pass_tsan_O0): ...new function.
2647         * var-tracking.c (pass_variable_tracking): Convert from a global
2648         struct to a subclass of rtl_opt_pass along with...
2649         (pass_data_variable_tracking): ...new pass_data instance and...
2650         (make_pass_variable_tracking): ...new function.
2651         * web.c (pass_web): Convert from a global struct to a subclass of
2652         rtl_opt_pass along with...
2653         (pass_data_web): ...new pass_data instance and...
2654         (make_pass_web): ...new function.
2655         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
2656         declaration with that of...
2657         (make_pass_mode_switch_use): ...new function.
2658         (pass_resolve_sw_modes): Replace declaration with that of...
2659         (make_pass_resolve_sw_modes): ...new function.
2660         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
2661         from a global struct to a subclass of rtl_opt_pass along with...
2662         (pass_data_mode_switch_use): ...new pass_data instance and...
2663         (make_pass_mode_switch_use): ...new function.
2664         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
2665         from a global struct to a subclass of rtl_opt_pass along with...
2666         (pass_data_resolve_sw_modes): ...new pass_data instance and...
2667         (make_pass_resolve_sw_modes): ...new function.
2668         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
2669         struct to a subclass of rtl_opt_pass along with...
2670         (pass_data_insert_vzeroupper): ...new pass_data instance and...
2671         (make_pass_insert_vzeroupper): ...new function.
2672         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
2673         global struct to a subclass of rtl_opt_pass along with...
2674         (pass_data_work_around_errata): ...new pass_data instance and...
2675         (make_pass_work_around_errata): ...new function.
2676         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
2677         struct to a subclass of rtl_opt_pass along with...
2678         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
2679         (make_pass_mips_machine_reorg2): ...new function.
2681 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2683         * passes.c (pass_manager::operator new): New.
2685 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2687         Handwritten part of conversion of passes to C++ classes.
2689         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
2690         (toplev.o): Add dep on PASS_MANAGER_H.
2691         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
2692         of early local pases to reflect this moving from a global to a
2693         member of gcc::pass_manager.
2694         (cgraph_add_new_function): Likewise.
2695         * lto-cgraph.c (lto_output_node): Update for conversion of
2696         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
2697         * passes.c (opt_pass::clone): New.
2698         (opt_pass::gate): New.
2699         (opt_pass::execute): New.
2700         (opt_pass::opt_pass): New.
2701         (pass_manager::execute_early_local_passes): New.
2702         (pass_manager::execute_pass_mode_switching): new.
2703         (finish_optimization_passes): Convert to...
2704         (pass_manager::finish_optimization_passes): ...this.
2705         (finish_optimization_passes): Update for conversion of passes to
2706         C++ classes.
2707         (register_dump_files_1): Use has_gate since we cannot portably
2708         check a vtable entry against NULL.
2709         (dump_one_pass): Likewise.
2710         (ipa_write_summaries_2): Likewise.
2711         (ipa_write_optimization_summaries_1): Likewise.
2712         (ipa_read_summaries_1): Likewise.
2713         (ipa_read_optimization_summaries_1): Likewise.
2714         (execute_ipa_stmt_fixups): Likewise.
2715         (pass_manager::pass_manager): Rewrite pass-creation, invoking
2716         pass-creation functions rather than wiring up globals, and
2717         storing the results in fields of pass_manager generated using
2718         pass-instances.def.
2719         (pass_manager::dump_profile_report): Update for conversion of
2720         passes to C++ classes.
2721         (pass_manager::execute_ipa_summary_passes): Likewise.
2722         (execute_one_ipa_transform_pass): Likewise.
2723         (execute_one_pass): Use has_gate and has_execute since we cannot
2724         portably check a vtable entry against NULL.
2725         * pass_manager.h (pass_manager::finish_optimization_passes): New.
2726         (pass_manager): Use pass-instances.def to add fields for the
2727         various pass instances.
2728         * toplev.c (finalize): Update for move of
2729         finish_optimization_passes to a method of gcc::pass_manager.
2730         * toplev.h (finish_optimization_passes): Move to method of class
2731         pass_manager.
2732         * tree-pass.h (struct pass_data): New.
2733         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
2734         (opt_pass::gate): Convert to virtual function.
2735         (opt_pass::~opt_pass): New.
2736         (opt_pass::clone): New.
2737         (opt_pass::execute): Convert to virtual function.
2738         (opt_pass::opt_pass): New.
2739         (opt_pass::ctxt_): new.
2740         (gimple_opt_pass): Convert to subclass of opt_pass.
2741         (gimple_opt_pass::gimple_opt_pass): New.
2742         (rtl_opt_pass): Convert to subclass of opt_pass.
2743         (rtl_opt_pass::rtl_opt_pass): New.
2744         (ipa_opt_pass_d): Convert to subclass of opt_pass.
2745         (ipa_opt_pass_d::ipa_opt_pass_d): New.
2746         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
2747         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
2748         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
2749         invocation of pass_mode_switching to reflect this moving from a
2750         global to a member of gcc::pass_manager.
2751         (ix86_option_override): Rework how pass_insert_vzeroupper is
2752         added to the pass_manager to reflect autogenerated changes.
2753         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
2755 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
2757         PR rtl-optimization/57708
2758         * recog.c (peep2_find_free_register): Validate all regs in a
2759         multi-reg mode.
2761 2013-08-05  Jan Hubicka  <jh@suse.cz>
2763         PR lto/57602
2764         * cgraph.c (verify_cgraph_node): Accept local flags from other
2765         partitions.
2766         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
2767         (function_and_variable_visibility): Likewise.
2768         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
2770 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2772         * graph.c (init_graph_slim_pretty_print): Remove.
2773         (print_graph_cfg): Do not call it.  Use local pretty printer.
2774         (start_graph_dump): Likewise.
2776 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2778         * gimple-pretty-print.c (buffer): Remove.
2779         (initialized): Likewise.
2780         (maybe_init_pretty_print): Likewise.
2781         (print_gimple_stmt): Do not call it.  Use non-static local
2782         pretty_printer variable.
2783         (print_gimple_expr): Likewise.
2784         (print_gimple_seq): Likewise.
2785         (gimple_dump_bb): Likewise.
2787 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2789         * asan.c (asan_pp): Remove.
2790         (asan_pp_initialized): Likewise.
2791         (asan_pp_initialize): Likewise.
2792         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
2793         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
2794         (asan_add_global): Likewise.
2796 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2798         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
2799         * diagnostic.h (diagnostic_flush_buffer): Adjust.
2800         * pretty-print.c (pp_formatted_text_data): Likewise.
2801         (pp_indent): Rename from pp_base_indent.
2802         (pp_format): Rename from pp_base_format.
2803         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
2804         (pp_format_verbatim): Rename from pp_base_format_verbatim.
2805         (pp_flush): Rename from pp_base_flush.
2806         (pp_set_line_maximum_length): Rename from
2807         pp_base_set_line_maximum_length.
2808         (pp_clear_output_area): Rename from pp_base_clear_output_area.
2809         (pp_set_prefix): Rename from pp_base_set_prefix.
2810         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
2811         (pp_emit_prefix): Rename from pp_base_emit_prefix.
2812         (pp_append_text): Rename from pp_base_append_text.
2813         (pp_formatted_text): Rename from pp_base_formatted_text.
2814         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
2815         (pp_remaining_character_count_for_line): Rename from
2816         pp_base_remaining_character_count_for_line.
2817         (pp_newline): Rename from pp_base_newline.
2818         (pp_character): Rename from pp_base_character.
2819         (pp_string): Rename from pp_base_string.
2820         (pp_maybe_space): Rename from pp_base_maybe_space.
2821         * asan.c (asan_pp_string): Adjust.
2822         (asan_emit_stack_protection): Likewise.
2823         (asan_add_global): Likewise.
2824         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
2825         * tree-mudflap.c (mf_varname_tree): Likewise.
2826         * tree-pretty-print.c (pp_tree_identifier): Rename from
2827         pp_base_tree_identifier.
2828         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
2829         Declare as function.
2831 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2833         * pretty-print.h (pp_bar_bar): New.
2834         (pp_ampersand_ampersand): Likewise.
2835         (pp_less_equal): Likewise.
2836         (pp_greater_equal): Likewise.
2837         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
2838         printer functions instead of pp_string or operators and punctuators.
2839         (dump_gimple_call): Likewise.
2840         (dump_gimple_omp_for): Likewise.
2841         (dump_gimple_transaction): Likewise.
2842         (dump_gimple_phi): Likewise.
2843         (pp_gimple_stmt_1): Likewise.
2844         * sched-vis.c (print_insn): Likewise.
2845         * tree-mudflap.c (mf_varname_tree): Likewise.
2846         * tree-pretty-print.c (dump_block_node): Likewise.
2847         (dump_generic_node): Likewise.
2849 2013-08-02  Jan Hubicka  <jh@suse.cz>
2851         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
2852         boundaries.
2853         * lto-streamer-out.c (tree_is_indexable): Results decls and
2854         parm decls are not indexable.
2855         (DFS_write_tree_body): Do not follow args and results.
2856         (hash_tree): Likewise.
2857         (output_functions): Rearrange so struct function is needed
2858         only when real body is output; be able to also ouptut abstract
2859         functions; output DECL_ARGUMENTS and DECL_RESULT.
2860         (lto_output): When not in WPA, ale store abstract functions.
2861         (write_symbol): Do not care about RESULT_DECL.
2862         (output_symbol_p): Handle correctly sbtract decls.
2863         * lto-streamer-in.c (input_function): Rearrange so struct
2864         function can be NULL at entry; allow streaming of
2865         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2866         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2867         sanity check during LTO.
2868         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2869         RESULT_DECl and DECL_ARGUMENTS.
2870         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2871         Likewise.
2873 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2875         * pretty-print.h (pp_underscore): New.
2876         (pp_comma): Tidy.
2877         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2878         printer functions instead of pp_character.
2879         (dump_binary_rhs): Likewise.
2880         (dump_ternary_rhs): Likewise.
2881         (dump_gimple_call_args): Likewise.
2882         (pp_points_to_solution): Likewise.
2883         (dump_gimple_call): Likewise.
2884         (dump_gimple_switch): Likewise.
2885         (dump_gimple_cond): Likewise.
2886         (dump_gimple_bind): Likewise.
2887         (dump_gimple_try): Likewise.
2888         (dump_gimple_omp_for): Likewise.
2889         (dump_gimple_omp_continue): Likewise.
2890         (dump_gimple_omp_single): Likewise.
2891         (dump_gimple_omp_sections): Likewise.
2892         (dump_gimple_omp_block): Likewise.
2893         (dump_gimple_omp_critical): Likewise.
2894         (dump_gimple_transaction): Likewise.
2895         (dump_gimple_asm): Likewise.
2896         (dump_gimple_phi): Likewise.
2897         (dump_gimple_omp_parallel): Likewise.
2898         (dump_gimple_omp_task): Likewise.
2899         (dump_gimple_omp_atomic_load): Likewise.
2900         (dump_gimple_omp_atomic_store): Likewise.
2901         (dump_gimple_mem_ops): Likewise.
2902         (pp_gimple_stmt_1): Likewise.
2903         (pp_cfg_jump): Likewise.
2904         (dump_implicit_edges): Likewise.
2905         (gimple_dump_bb_for_graph): Likewise.
2906         * graph.c (draw_cfg_node): Likewise.
2907         * langhooks.c (lhd_print_error_function): Likewise.
2908         * sched-vis.c (print_exp): Likewise.
2909         (print_value): Likewise.
2910         (print_pattern): Likewise.
2911         (print_insn): Likewise.
2912         (rtl_dump_bb_for_graph): Likewise.
2913         * tree-pretty-print.c (dump_function_declaration): Likewise.
2914         (dump_array_domain): Likewise.
2915         (dump_omp_clause): Likewise.
2916         (dump_location): Likewise.
2917         (dump_generic_node): Likewise.
2918         (print_struct_decl): Likewise.
2919         * diagnostic.c (diagnostic_show_locus): Use pp_space.
2921 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2923         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2924         candidate table when replacing a candidate statement.
2925         (replace_rhs_if_not_dup): Likewise.
2926         (replace_one_candidate): Likewise.
2928 2013-08-02  Jan Hubicka  <jh@suse.cz>
2929             Martin Liska  <marxin.liska@gmail.com>
2931         * cgraphunit.c (add_new_function): Fix logic when adding from
2932         late IPA pass.
2933         (assemble_thunk): Rename to ...
2934         (expand_thunk); .. this one; export; get it working with
2935         general functions; make produced gimple valid.
2936         * cgraph.h (expand_thunk): Declare.
2938 2013-08-02  Jan Hubicka  <jh@suse.cz>
2940         * ipa-cp.c (gather_context_independent_values): Use
2941         ipa_get_param_move_cost.
2942         (get_replacement_map): Remove PARAM; move parameter folding
2943         into tree-inline.c
2944         (create_specialized_node): Update.
2945         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
2946         assert that we have gimple body; update move_cost.
2947         (count_formal_params): Assert that we have gimple body.
2948         (ipa_dump_param): New function.
2949         (ipa_alloc_node_params): Break out from ...
2950         (ipa_initialize_node_params): ... here.
2951         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
2952         (ipa_write_node_info): Stream move costs.
2953         (ipa_read_node_info): Read move costs.
2954         (ipa_update_after_lto_read): Do not recompute node params.
2955         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
2956         (ipa_get_param): Check we are not in WPA.
2957         (ipa_get_param_move_cost): New.
2958         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
2959         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
2960         parm numbers to be present.
2962 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2964         PR rtl-optimization/58048
2965         * lra-constraints.c (process_alt_operands): Don't check asm
2966         operand on register.
2968 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
2970         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
2971         the implied StoreLoad barrier for atomic operations if before.
2973 2013-08-02  Jan Hubicka  <jh@suse.cz>
2974             Martin Liska  <marxin.liska@gmail.com>
2976         * cgraph.c (cgraph_function_body_availability): Do not check
2977         cgraph flags.
2978         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
2979         symtab_node_availability): Declare.
2980         * ipa.c (can_replace_by_local_alias): New.
2981         (function_and_variable_visibility): Use it.
2982         * symtab.c (symtab_for_node_and_aliases,
2983         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
2985 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
2987         PR rtl-optimization/57963
2988         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
2989         (lra_constraints): Use them.
2991 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
2993         * config/arm/types.md (define_attr "type"): Add "load_acq"
2994         and "store_rel".
2995         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
2996         changes.
2997         (cortex_a53_store1): Likewise.
2999 2013-08-01  Jan Hubicka  <jh@suse.cz>
3001         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3002         partitions are not needed.
3004 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
3006         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3007         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3008         MAYBE_NON_Q_CLASS_P where appropriate.
3010 2013-08-01  Jan Hubicka  <jh@suse.cz>
3012         * cgraph.h (release_function_body): Declare.
3013         * tree.c (free_lang_data_in_decl): Free, parameters and return values
3014         of unused delcarations.
3016 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3018         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3019         RTL form when subtracting a constant.
3021 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3023         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3024         Generate canonical plus rtx with negated immediate instead of minus
3025         where appropriate.
3026         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3028 2013-08-01  Jan Hubicka  <jh@suse.cz>
3030         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3031         (cgraph_release_function_body): Likewise.
3032         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3033         * cgraph.h (cgrpah_node): Rename abstract_and_needed
3034         to used_as_abstract_origin.
3035         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3036         symbols used as abstract origins.
3037         * cgraphunit.c (analyze_functions): Update.
3038         * ipa.c (symtab_remove_unreachable_nodes): Recompute
3039         used_as_abstract_origin.
3040         * tree-inline.c (tree_function_versioning): Update
3041         used_as_abstract_origin; be ready for DECL_RESULT and
3042         DECL_ARGUMENTS to be NULL.
3044         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3045         for abstract functions.
3046         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3047         real symbols.
3049 2013-08-01  Jan Hubicka  <jh@suse.cz>
3051         * profile.c (compute_value_histograms): Fix thinko.
3053 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3055         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
3056         aarch-common-protos.h to extra_headers.
3057         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3058         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3059         * config/aarch64/t-aarch64 (aarch-common.o): Define.
3061 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3063         * config/aarch64/aarch64.md (define_attr "type"): Delete.
3064         Include "../arm/types.md".  Define "type" attribute for all patterns.
3065         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3066         attribute changes.
3068 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
3070         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3071         to support power8 load fusion.
3072         (fusion_gpr_mem_load): Likewise.
3074         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3076         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3077         declarations for power8 load fusion.
3078         (emit_fusion_gpr_load): Likewise.
3080         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3081         tuning for power8, turn on fusion mode by default.  Turn on sign
3082         extending fusion mode if normal fusion mode is on, and we are at
3083         -O2 or -O3.
3084         (fusion_gpr_load_p): New function, return true if we can fuse an
3085         addis instruction with a dependent load to a GPR.
3086         (emit_fusion_gpr_load): Emit the instructions for power8 load
3087         fusion to GPRs.
3089         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3090         (VSX load fusion peepholes): New peepholes to fuse together an
3091         addi instruction with a VSX load instruction.
3093         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3094         peepholes to fuse an addis instruction with a load to a GPR base
3095         register.  If we are supporting sign extending fusions, convert
3096         sign extending loads to zero extending loads and add an explicit
3097         sign extension.
3099 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3101         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
3102         aarch-common-protos.h to extra_headers.
3103         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3104         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3105         (arm_early_store_addr_dep): Likewise.
3106         (arm_no_early_alu_shift_dep): Likewise.
3107         (arm_no_early_alu_shift_value_dep): Likewise.
3108         (arm_no_early_mul_dep): Likewise.
3109         (arm_no_early_store_addr_dep): Likewise.
3110         (arm_mac_accumulator_is_mul_result): Likewise.
3111         (arm_mac_accumulator_is_result): Likewise.
3112         * config/arm/aarch-common.c: ... here.  New file.
3113         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3114         here to ...
3115         (arm_early_store_addr_dep): Likewise.
3116         (arm_no_early_alu_shift_dep): Likewise.
3117         (arm_no_early_alu_shift_value_dep): Likewise.
3118         (arm_no_early_mul_dep): Likewise.
3119         (arm_no_early_store_addr_dep): Likewise.
3120         (arm_mac_accumulator_is_mul_result): Likewise.
3121         (arm_mac_accumulator_is_result): Likewise.
3122         * config/arm/aarch-common-protos.h: ... here.  New file.
3123         * config/arm/t-arm (aarch-common.o): Define.
3125 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3127         * config/arm/arm.md: Include new file "types.md".
3128         (define_attr "type"): Move from here to ...
3129         (define_attr "mul32"): Likewise.
3130         (define_attr "mul64"): Likewise.
3131         * config/arm/types.md: ... here.  New file.
3133 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
3135         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
3136         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
3138 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3140         * gen-pass-instances.awk: Fix offset of substr().
3142 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
3144         * Makefile.in (pass-instances.def): New.
3145         (passes.o): Replace dependency on passes.def with one on
3146         pass-instances.def
3148         * gen-pass-instances.awk: New.
3150         * passes.c (pass_manager::pass_manager): Use pass-instances.def
3151         rather than passes.def, updating local definition of NEXT_PASS
3152         macro to add an extra NUM parameter (currently unused).
3154 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
3156         * Makefile.in (PASS_MANAGER_H): New.
3157         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
3158         (passes.o): Likewise.
3159         (statistics.o): Likewise.
3160         (cgraphunit.o): Likewise.
3161         (context.o): Depend on PASS_MANAGER_H.
3163         * pass_manager.h: New.
3165         * cgraphunit.c (cgraph_add_new_function): Update for moves
3166         of globals to fields of pass_manager.
3167         (analyze_function): Likewise.
3168         (expand_function): Likewise.
3169         (ipa_passes): Likewise.
3170         (compile): Likewise.
3172         * context.c (context::context): New.
3173         * context.h  (context::context): New.
3174         (context::get_passes): New.
3175         (context::passes_): New.
3177         * lto-cgraph.c (input_node): Update for moves of globals to
3178         fields of pass_manager.
3180         * passes.c (all_passes): Remove, in favor of a field of the
3181         same name within the new class pass_manager.
3182         (all_small_ipa_passes): Likewise.
3183         (all_lowering_passes): Likewise.
3184         (all_regular_ipa_passes): Likewise.
3185         (all_late_ipa_passes): Likewise.
3186         (all_lto_gen_passes): Likewise.
3187         (passes_by_id): Likewise.
3188         (passes_by_id_size): Likewise.
3189         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3190         the new class pass_manager.
3191         (set_pass_for_id): Convert to...
3192         (pass_manager::set_pass_for_id): ...method.
3193         (get_pass_for_id): Convert to...
3194         (pass_manager::get_pass_for_id): ...method.
3195         (register_one_dump_file): Move body of implementation into...
3196         (pass_manager::register_one_dump_file): ...here.
3197         (register_dump_files_1): Convert to...
3198         (pass_manager::register_dump_files_1): ...method.
3199         (register_dump_files): Convert to...
3200         (pass_manager::register_dump_files): ...method.
3201         (create_pass_tab): Update for moves of globals to fields of
3202         pass_manager.
3203         (dump_passes): Move body of implementation into...
3204         (pass_manager::dump_passes): ...here.
3205         (register_pass): Move body of implementation into...
3206         (pass_manager::register_pass): ...here.
3207         (init_optimization_passes): Convert into...
3208         (pass_manager::pass_manager): ...constructor for new
3209         pass_manager class, and initialize the pass_lists array.
3210         (check_profile_consistency): Update for moves of globals to
3211         fields of pass_manager.
3212         (dump_profile_report): Move body of implementation into...
3213         (pass_manager::dump_profile_report): ...here.
3214         (ipa_write_summaries_1): Update for moves of pass lists from
3215         being globals to fields of pass_manager.
3216         (ipa_write_optimization_summaries): Likewise.
3217         (ipa_read_summaries):  Likewise.
3218         (ipa_read_optimization_summaries): Likewise.
3219         (execute_all_ipa_stmt_fixups): Likewise.
3221         * statistics.c (statistics_fini): Update for moves of globals to
3222         fields of pass_manager.
3224         * toplev.c (general_init): Replace call to
3225         init_optimization_passes with construction of the pass_manager
3226         instance.
3228         * tree-pass.h (all_passes): Remove, in favor of a field of the
3229         same name within the new class pass_manager.
3230         (all_small_ipa_passes): Likewise.
3231         (all_lowering_passes): Likewise.
3232         (all_regular_ipa_passes): Likewise.
3233         (all_lto_gen_passes): Likewise.
3234         (all_late_ipa_passes): Likewise.
3235         (passes_by_id): Likewise.
3236         (passes_by_id_size): Likewise.
3237         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3238         the new class pass_manager.
3239         (get_pass_for_id): Remove.
3241 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
3243         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
3244         configs.
3246 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
3248         * arm.md (mulhi3): New expand pattern.
3250 2013-07-30  Jan Hubicka  <jh@suse.cz>
3251             Martin Liska  <marxin.liska@gmail.com>
3253         * profile.c (compute_value_histograms): Do not ICE when
3254         there is mismatch only on some counters.
3256 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3258         PR rtl-optimization/57637
3259         * function.c (move_insn_for_shrink_wrap): Also check the
3260         GEN set of the LIVE problem for the liveness analysis
3261         if it exists, otherwise give up.
3263 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3265         PR tree-optimization/57993
3266         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
3267         replaced statement in the candidate table.
3268         (phi_add_costs): Return infinite cost when the hidden basis does
3269         not dominate all phis on which the candidate is dependent.
3270         (replace_one_candidate): Record replaced statement in the
3271         candidate table.
3273 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
3275         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
3276         (ashlv2si3): New expander.
3277         (*ashlv2si3_i): New define_insn_and_split.
3278         * predicates.md (float_operation): Allow patterns with three
3279         basic sub-patterns.
3281         PR rtl-optimization/58021
3282         * mode-switching.c (create_pre_exit): Always split off preceding
3283         insns if we are not at the basic block head.
3285 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
3287         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
3288         (UCLIBC_DYNAMIC_LINKER): New macro.
3289         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
3290         `-mnan=2008'.
3291         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
3292         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
3293         `-mnan=2008'.
3294         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
3295         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
3296         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
3297         for SF and DF modes.  Use ieee_quad_format for TF mode.
3298         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
3299         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
3300         (mips_option_override): Handle `-mnan=legacy'.
3301         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
3302         `-mabs=2008' and `-mnan=2008'.
3303         (OPTION_DEFAULT_SPECS): Add "nan" default.
3304         (ASM_SPEC): Handle `-mnan='.
3305         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
3306         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
3307         comment accordingly.
3308         (neg<mode>2): Likewise.
3309         * config/mips/mips.opt (mabs, mnan): New options.
3310         * doc/install.texi (Configuration): Document `--with-nan=' option.
3311         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
3312         `-mnan=' options.
3313         (MIPS Options): Document them.
3314         * config.gcc <mips*-*-*>: Handle `--with-nan='.
3315         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
3316         * configure: Regenerate.
3317         * config.in: Regenerate.
3319 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
3321         * config/i386/i386.md (float post-reload splitters): Do not check
3322         for subregs of SSE registers.
3324 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
3325             H.J. Lu  <hongjiu.lu@intel.com>
3327         PR target/57954
3328         PR target/57988
3329         * config/i386/i386.md (post-reload splitter
3330         to avoid partial SSE reg dependency stalls): New pattern.
3332 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
3334         * config/s390/s390.md ("movcc"): Swap load and store instructions.
3336 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
3338         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
3339         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
3341 2013-07-26  Cary Coutant  <ccoutant@google.com>
3343         * dwarf2out.c (die_checksum_ordered): Don't include template
3344         instantiations in signature.
3345         (is_template_parameter): New function.
3346         (is_template_instantiation): New function.
3347         (generate_skeleton_bottom_up): Don't include template instantiations
3348         in type unit DIE.
3349         (generate_skeleton): Likewise.
3350         (break_out_comdat_types): Move recursive call to break out nested
3351         types earlier.
3352         (prune_unused_types_mark_generic_parms_dies): Call
3353         is_template_parameter.
3355 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
3357         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
3358         uses vector registers.
3359         * config/aarch64/iterators.md: Add attributes rtn and vas.
3361 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3362             Richard Earnshaw  <richard.earnshaw@arm.com>
3364         * combine.c (simplify_comparison): Re-canonicalize operands
3365         where appropriate.
3366         * config/arm/arm.md (movcond_addsi): New splitter.
3368 2013-07-25  Sterling Augustine  <saugustine@google.com>
3370         * dwarf2out.c (size_of_pubnames): Move code to...
3371         (include_pubname_in_output): ...here.  New.
3372         (want_pubnames): Rearrange.
3373         (output_pubnames): Call include_pubname_in_output.  Move assertion.
3375 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
3377         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
3379 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
3381         PR target/38836
3382         * doc/extend.texi: Remove obsolete builtins. Fix
3383         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
3385 2013-07-25  Jan Hubicka  <jh@suse.cz>
3387         * cgraph.c (release_function_body): Break out from ...
3388         (cgraph_release_function_body): ... this one; also release DECL_RESULT
3389         and DECL_ARGUMENTS.
3390         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
3391         old_tree in the map.
3392         (create_specialized_node): Update.
3393         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
3394         into index.
3395         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
3396         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
3397         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
3398         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
3399         DECL_RESULT.
3401 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3403         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
3404         addsi3_carryin_alt2_<optab>): Correct output template.
3406 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3408         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
3409         Adjust for arm_restrict_it.
3410         Remove trailing whitespace.
3412 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
3414         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
3415         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
3417         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
3419 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
3421         PR rtl-optimization/57960
3422         * lra-constraints.c (process_alt_operands): Use the right mode
3423         when checking strict_low.
3425 2013-07-25  Jan Hubicka  <jh@suse.cz>
3427         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
3428         * cgraph.c (cgraph_remove_node): Do not release function body
3429         when in cgraph streaming.
3430         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
3431         in other partitions are not considered reachable; fix handling of
3432         clones.
3434 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3436         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
3438 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3440         PR target/19599
3441         PR target/57731
3442         PR target/57837
3443         * config/arm/arm.md ("*sibcall_insn): Replace use of
3444         Ss with US. Adjust output for v5 and v4t.
3445         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
3447         * config/arm/constraints.md ("Ss"): Rename to US.
3449 2013-07-25  Terry Guo  <terry.guo@arm.com>
3451         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
3452         shift_add/shift_sub0/shift_sub1 RTXs.
3454 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
3455             Anton Blanchard <anton@au1.ibm.com>
3457         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
3458         (altivec_vpks<VI_char>ss): Likewise.
3459         (altivec_vpks<VI_char>us): Likewise.
3460         (altivec_vpku<VI_char>us): Likewise.
3461         (altivec_vpku<VI_char>um): Likewise.
3463 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
3465         Introduce context class.
3467         * Makefile.in (CONTEXT_H): New.
3468         (OBJS): Add context.o.
3469         (toplev.o): Add CONTEXT_H to dependencies.
3470         (context.o): New.
3472         * toplev.c (general_init): Create the singleton gcc::context instance.
3474         * context.c: New.
3476         * context.h: New.
3478 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
3480         PR rtl-optimization/57968
3481         * mode-switching.c (create_pre_exit): Allow instructions that
3482         don't set a return register to need a non-exit mode.
3484 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3485             Anton Blanchard <anton@au1.ibm.com>
3487         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
3488         operands to vperm for little endian.
3489         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
3490         of lvsl to create the control mask for a vperm for little endian.
3492 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3493             Anton Blanchard <anton@au1.ibm.com>
3495         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
3496         two operands for little-endian.
3498 2013-07-23  Steve Ellcey  <sellcey@mips.com>
3500         * config/mips/mips.c (mips_case_values_threshold): New.
3501         (TARGET_CASE_VALUES_THRESHOLD): Define.
3503 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3504             Anton Blanchard <anton@au1.ibm.com>
3506         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
3507         selection of field for vector splat in little endian mode.
3509 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
3511         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
3512         expanders to rs6000.md.
3513         (ior<mode>3): Likewise.
3514         (and<mode>3): Likewise.
3515         (one_cmpl<mode>2): Likewise.
3516         (nor<mode>3): Likewise.
3517         (andc<mode>3): Likewise.
3518         (eqv<mode>3): Likewise.
3519         (nand<mode>3): Likewise.
3520         (orc<mode>3): Likewise.
3522         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
3523         declaration.
3525         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
3526         to split multi-word logical operations.
3527         (rs6000_split_logical_di): Likewise.
3528         (rs6000_split_logical): Likewise.
3530         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
3531         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
3532         and allow TImode operations in 32-bit.
3533         (vsx_and<mode>3_64bit): Likewise.
3534         (vsx_ior<mode>3_32bit): Likewise.
3535         (vsx_ior<mode>3_64bit): Likewise.
3536         (vsx_xor<mode>3_32bit): Likewise.
3537         (vsx_xor<mode>3_64bit): Likewise.
3538         (vsx_one_cmpl<mode>2_32bit): Likewise.
3539         (vsx_one_cmpl<mode>2_64bit): Likewise.
3540         (vsx_nor<mode>3_32bit): Likewise.
3541         (vsx_nor<mode>3_64bit): Likewise.
3542         (vsx_andc<mode>3_32bit): Likewise.
3543         (vsx_andc<mode>3_64bit): Likewise.
3544         (vsx_eqv<mode>3_32bit): Likewise.
3545         (vsx_eqv<mode>3_64bit): Likewise.
3546         (vsx_nand<mode>3_32bit): Likewise.
3547         (vsx_nand<mode>3_64bit): Likewise.
3548         (vsx_orc<mode>3_32bit): Likewise.
3549         (vsx_orc<mode>3_64bit): Likewise.
3551         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
3552         logical types in GPRs.
3554         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
3555         logical insns to rs6000.md, and allow TImode operations in
3556         32-bit.
3557         (altivec_ior<mode>3): Likewise.
3558         (altivec_xor<mode>3): Likewise.
3559         (altivec_one_cmpl<mode>2): Likewise.
3560         (altivec_nor<mode>3): Likewise.
3561         (altivec_andc<mode>3): Likewise.
3563         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
3564         attributes for moving the 128-bit logical operations into
3565         rs6000.md.
3566         (BOOL_REGS_OUTPUT): Likewise.
3567         (BOOL_REGS_OP1): Likewise.
3568         (BOOL_REGS_OP2): Likewise.
3569         (BOOL_REGS_UNARY): Likewise.
3570         (BOOL_REGS_AND_CR0): Likewise.
3571         (one_cmpl<mode>2): Add support for DI logical operations on
3572         32-bit, splitting the operations to 32-bit.
3573         (anddi3): Likewise.
3574         (iordi3): Likewise.
3575         (xordi3): Likewise.
3576         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
3577         changes to combine the 32/64-bit code, allow logical operations on
3578         TI mode in 32-bit, and to use similar match_operator patterns like
3579         scalar mode uses.  Combine the Altivec and VSX code for logical
3580         operations, and move it here.
3581         (ior<mode>3, 128-bit types): Likewise.
3582         (xor<mode>3, 128-bit types): Likewise.
3583         (one_cmpl<mode>3, 128-bit types): Likewise.
3584         (nor<mode>3, 128-bit types): Likewise.
3585         (andc<mode>3, 128-bit types): Likewise.
3586         (eqv<mode>3, 128-bit types): Likewise.
3587         (nand<mode>3, 128-bit types): Likewise.
3588         (orc<mode>3, 128-bit types): Likewise.
3589         (and<mode>3_internal): Likewise.
3590         (bool<mode>3_internal): Likewise.
3591         (boolc<mode>3_internal1): Likewise.
3592         (boolc<mode>3_internal2): Likewise.
3593         (boolcc<mode>3_internal1): Likewise.
3594         (boolcc<mode>3_internal2): Likewise.
3595         (eqv<mode>3_internal1): Likewise.
3596         (eqv<mode>3_internal2): Likewise.
3597         (one_cmpl1<mode>3_internal): Likewise.
3599 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3601         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3602         Rename flag_stack_usage to flag_stack_usage_info.
3604 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3606         * config/microblaze/sync.md: New file.
3607         * config/microblaze/microblaze.md: Include sync.md
3608         * config/microblaze/microblaze.c: Add print_operand 'y'.
3609         * config/microblaze/constraints.md: Add memory_contraint
3610         'Q' which is a single register.
3612 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
3614         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
3616 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3618         * reload.c (find_reloads): Exit loop once we find this operand
3619         cannot be reloaded somehow for this alternative.
3621         * reload.c (find_reloads): Exit loop once we find a hard register.
3623         * rtlanal.c (computed_jump_p): Exit loop once we find label
3624         reference is used.
3626         * i386.c (ix86_pad_returns): Exit loop after setting replace.
3628         * cfgloopmanip.c (remove_path): Exit loop after setting
3629         irred_invalidated.
3631         * gensupport.c (subst_dup): Avoid loop if code is not
3632         MATCH_DUP nor MATCH_OP_DUP.
3634 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
3636         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
3638 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3640         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
3641         true for SP_REGNUM if mode == ptr_mode.
3642         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
3643         with value R0_REGNUM + 31.
3645 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3647         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
3648         pad pointer-typed argument downward.
3650 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3652         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
3653         and __ILP32__ when the ILP32 model is in use.
3655 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3657         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
3658         (aarch64_load_symref_appropriately): In the case of
3659         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
3660         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
3661         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
3662         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
3663         if 'mode' doesn't equal to 'ptr_mode'.
3664         (aarch64_output_mi_thunk): Add an assertion on the alignment of
3665         'vcall_offset'; change to call aarch64_emit_move differently depending
3666         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
3667         to calculate the upper bound of 'vcall_offset'.
3668         (aarch64_cannot_force_const_mem): Change to also return true if
3669         mode != ptr_mode.
3670         (aarch64_legitimize_reload_address): In the case of large
3671         displacements, add new local variable 'xmode' and an assertion
3672         based on it; change to use 'xmode' to generate the new rtx and
3673         reload.
3674         (aarch64_asm_trampoline_template): Change to generate the template
3675         differently depending on TARGET_ILP32 or not; change to use
3676         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
3677         (aarch64_trampoline_size): Removed.
3678         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
3679         and replace immediate literals with it.  Change to use 'ptr_mode'
3680         instead of 'DImode' and call convert_memory_address if the mode
3681         of 'fnaddr' doesn't equal to 'ptr_mode'.
3682         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
3683         to output symbol.
3684         (aarch64_elf_asm_destructor): Likewise.
3685         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
3686         on TARGET_ILP32 instead of aarch64_trampoline_size.
3687         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
3688         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
3689         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
3690         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
3691         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
3692         (storewb_pair<GPI:mode>_<P:mode>): ... this.
3693         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
3694         depending on the value of 'mode'.
3695         (add_losym_<mode>): New.
3696         (ldr_got_small_<mode>): New, based on ldr_got_small.
3697         (ldr_got_small): Remove.
3698         (ldr_got_small_sidi): New.
3699         * config/aarch64/iterators.md (P): New.
3700         (PTR): Change to 'ptr_mode' in the condition.
3702 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3704         * config.gcc (aarch64*-*-*): Support --with-abi.
3705         (aarch64*-*-elf): Support --with-multilib-list.
3706         (aarch64*-*-linux*): Likewise.
3707         (supported_defaults): Add abi to aarch64*-*-*.
3708         * configure.ac: Mention AArch64 for --with-multilib-list.
3709         * configure: Re-generated.
3710         * config/aarch64/biarchilp32.h: New file.
3711         * config/aarch64/biarchlp64.h: New file.
3712         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
3713         (ABI_SPEC): Ditto.
3714         (MULTILIB_DEFAULTS): Ditto.
3715         (DRIVER_SELF_SPECS): Ditto.
3716         (ASM_SPEC): Update to also substitute -mabi.
3717         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
3718         file whose name depends on -mabi= and -mbig-endian.
3719         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
3720         TARGET_ILP32.
3721         (POINTER_SIZE): New define.
3722         (POINTERS_EXTEND_UNSIGNED): Ditto.
3723         (enum aarch64_abi_type): New enumeration tag.
3724         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
3725         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
3726         (TARGET_ILP32): New define.
3727         * config/aarch64/aarch64.opt (mabi): New.
3728         (aarch64_abi): New.
3729         (ilp32, lp64): New values for -mabi.
3730         * config/aarch64/t-aarch64 (comma): New define.
3731         (MULTILIB_OPTIONS): Ditto.
3732         (MULTILIB_DIRNAMES): Ditto.
3733         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
3734         * doc/invoke.texi: Document -mabi for AArch64.
3736 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
3738         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
3740 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3741             Anton Blanchard <anton@au1.ibm.com>
3743         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
3744         endianness when selecting field to splat.
3746 2013-07-22  Eric Christopher  <echristo@gmail.com>
3748         * dwarf2out.c (die_odr_checksum): New function to use
3749         CHECKSUM_ macros and ULEB128 for DIE tag.
3750         (generate_type_signature): Use.
3752 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
3754         * config.gcc (sparc*-*-*): Accept leon3 processor.
3755         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
3756         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
3757         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
3758         * config/sparc/sparc.opt (enum processor_type): Add leon3.
3759         (mfix-ut699): Adjust comment.
3760         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
3761         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
3762         (CPP_CPU_SPEC): Likewise.
3763         (ASM_CPU_SPEC): Likewise.
3764         * config/sparc/sparc.c (leon3_cost): New constant.
3765         (sparc_option_override): Add leon3 support.
3766         (mem_ref): New function.
3767         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
3768         (sparc_do_work_around_errata): Look into the instruction in the delay
3769         slot and adjust accordingly.  Add fix for the data cache nullify issues
3770         of the UT699.  Change insertion position for the NOP.
3771         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
3772         (leon3_load): New reservation.
3773         (leon_store): Bump latency to 2.
3774         (grfpu): New automaton.
3775         (grfpu_alu): New unit.
3776         (grfpu_ds): Likewise.
3777         (leon_fp_alu): Adjust.
3778         (leon_fp_mult): Delete.
3779         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
3780         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
3781         * config/sparc/sparc.md (cpu): Add leon3.
3782         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
3783         (swapsi): Likewise.
3784         (atomic_test_and_set): Likewise.
3785         (ldstub): Likewise.
3787 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
3789         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
3790         default for R5900 targets.
3791         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
3792         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
3793         * config/mips/mips.c (mips_option_override): Report an error for
3794         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
3795         for -march=r5900 -mhard-float.
3797 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3799         * df-problems.c (can_move_insns_across): Exit loop once we
3800         find a non-fixed, non-global register.
3802         * ipa-pure-const.c (propagate_nothrow): Exit loop after
3803         setting can_throw.
3805         * omega.c (omega_eliminate_red): Break after setting red_found.
3806         (omega_problem_has_red_equations): Similarly after setting found.
3807         (omega_query_variable): Similarly after setting coupled.
3809 2013-07-22  Marek Polacek  <polacek@redhat.com>
3811         * gimplify.c: Don't include gimple.h twice.
3813 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3815         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
3816         instead of TARGET_THUMB1.
3817         (Pz): New constraint.
3818         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
3819         encodings.
3820         (compare_negsi_si): Likewise.
3821         (compare_addsi2_op0): Likewise.
3822         (compare_addsi2_op1): Likewise.
3823         (addsi3_carryin_<optab>): Likewise.
3824         (addsi3_carryin_alt2_<optab>): Likewise.
3825         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
3826         for arm_restrict_it.
3827         (subsi3_carryin): Likewise.
3828         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
3829         (minmax_arithsi): Disable for arm_restrict_it.
3830         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
3831         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
3832         (satsi_<SAT:code>_shift): Likewise.
3833         (arm_shiftsi3): Add alternative for 16-bit encoding.
3834         (arm32_movhf): Disable for arm_restrict_it.
3835         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
3836         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
3838 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
3840         * config/arm/arm.md (attribute "insn"): Delete.
3841         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
3842         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
3843         (not_shiftsi): Update for attribute change.
3844         (not_shiftsi_compare0): Likewise.
3845         (not_shiftsi_compare0_scratch): Likewise.
3846         (arm_one_cmplsi2): Likewise.
3847         (thumb1_one_cmplsi2): Likewise.
3848         (notsi_compare0): Likewise.
3849         (notsi_compare0_scratch): Likewise.
3850         (thumb1_movdi_insn): Likewise.
3851         (arm_movsi_insn): Likewise.
3852         (movhi_insn_arch4): Likewise.
3853         (movhi_bytes): Likewise.
3854         (arm_movqi_insn): Likewise.
3855         (thumb1_movqi_insn): Likewise.
3856         (arm32_movhf): Likewise.
3857         (thumb1_movhf): Likewise.
3858         (arm_movsf_soft_insn): Likewise.
3859         (thumb1_movsf_insn): Likewise.
3860         (thumb_movdf_insn): Likewise.
3861         (movsicc_insn): Likewise.
3862         (movsfcc_soft_insn): Likewise.
3863         (and_scc): Likewise.
3864         (cond_move): Likewise.
3865         (if_move_not): Likewise.
3866         (if_not_move): Likewise.
3867         (if_shift_move): Likewise.
3868         (if_move_shift): Likewise.
3869         (if_shift_shift): Likewise.
3870         (if_not_arith): Likewise.
3871         (if_arith_not): Likewise.
3872         (cond_move_not): Likewise.
3873         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3874         (neon_mov<mode>): Likewise.
3875         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3876         (thumb2_movsi_vfp): Likewise.
3877         (movsf_vfp): Likewise.
3878         (thumb2_movsf_vfp): Likewise.
3879         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3880         change.
3881         (cortexa7_older_only): Likewise.
3882         (cortexa7_younger): Likewise.
3883         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3884         (1020alu_shift_op): Likewise.
3885         (1020alu_shift_reg_op): Likewise.
3886         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3887         (alu_shift_op): Likewise.
3888         (alu_shift_reg_op): Likewise.
3889         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3890         (11_alu_shift_op): Likewise.
3891         (11_alu_shift_reg_op): Likewise.
3892         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3893         (9_alu_shift_reg_op): Likewise.
3894         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3895         change.
3896         (cortex_a15_alu_shift): Likewise.
3897         (cortex_a15_alu_shift_reg): Likewise.
3898         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3899         (cortex_a5_alu_shift): Likewise.
3900         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3901         change.
3902         (cortex_a53_alu_shift): Likewise.
3903         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3904         change.
3905         (cortex_a7_alu_reg): Likewise.
3906         (cortex_a7_alu_shift): Likewise.
3907         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3908         (cortex_a8_alu_shift): Likewise.
3909         (cortex_a8_alu_shift_reg): Likewise.
3910         (cortex_a8_mov): Likewise.
3911         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3912         (cortex_a9_dp_shift): Likewise.
3913         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3914         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3915         (cortex_r4_mov): Likewise.
3916         (cortex_r4_alu_shift): Likewise.
3917         (cortex_r4_alu_shift_reg): Likewise.
3918         * config/arm/fa526.md (526_alu_op): Update for attribute change.
3919         (526_alu_shift_op): Likewise.
3920         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3921         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3922         (626te_alu_shift_op): Likewise.
3923         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3924         (726te_alu_op): Likewise.
3925         (726te_alu_shift_op): Likewise.
3926         (726te_alu_shift_reg_op): Likewise.
3927         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3928         (mp626_alu_shift_op): Likewise.
3929         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3930         (pj4_alu_e1_conds): Likewise.
3931         (pj4_alu): Likewise.
3932         (pj4_alu_conds): Likewise.
3933         (pj4_shift): Likewise.
3934         (pj4_shift_conds): Likewise.
3935         (pj4_alu_shift): Likewise.
3936         (pj4_alu_shift_conds): Likewise.
3938 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3940         * config/arm/predicates.md (shiftable_operator_strict_it):
3941         New predicate.
3942         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
3943         Disable cond_exec version for arm_restrict_it.
3944         (thumb2_smaxsi3): Convert to generate cond_exec.
3945         (thumb2_sminsi3): Likewise.
3946         (thumb32_umaxsi3): Likewise.
3947         (thumb2_uminsi3): Likewise.
3948         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
3949         (thumb2_neg_abssi2): Likewise.
3950         (thumb2_mov_scc): Add alternative for 16-bit encoding.
3951         (thumb2_movsicc_insn): Adjust alternatives.
3952         (thumb2_mov_negscc): Disable for arm_restrict_it.
3953         (thumb2_mov_negscc_strict_it): New pattern.
3954         (thumb2_mov_notscc_strict_it): New pattern.
3955         (thumb2_mov_notscc): Disable for arm_restrict_it.
3956         (thumb2_ior_scc): Likewise.
3957         (thumb2_ior_scc_strict_it): New pattern.
3958         (thumb2_cond_move): Adjust for arm_restrict_it.
3959         (thumb2_cond_arith): Disable for arm_restrict_it.
3960         (thumb2_cond_arith_strict_it): New pattern.
3961         (thumb2_cond_sub): Adjust for arm_restrict_it.
3962         (thumb2_movcond): Likewise.
3963         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
3964         (thumb2_zero_extendhisi2_v6): Likewise.
3965         (thumb2_zero_extendqisi2_v6): Likewise.
3966         (orsi_notsi_si): Likewise.
3967         (orsi_not_shiftsi_si): Likewise.
3969 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
3971         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
3972         instruction sequence is 1 byte shorter.
3974 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
3976         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
3977         it is not needed after split.
3979 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
3981         PR target/51784
3982         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
3983         second label for nonlocal goto receivers. Don't output pic base labels
3984         unless we're producing PIC; mark that action unreachable().
3985         (ix86_save_reg): If the function contains a nonlocal label, save the
3986         PIC base reg.
3987         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
3988         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
3989         (update_pic_label_number_if_needed): New.
3990         (machopic_output_function_base_name): Adjust for nonlocal receiver
3991         case.
3992         (machopic_should_output_picbase_label): New.
3993         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
3994         (nonlocal_goto_receiver): New insn and split.
3996 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
3998         * config/aarch64/aarch64-builtins.c
3999         (aarch64_fold_builtin): Fold abs in all modes.
4000         * config/aarch64/aarch64-simd-builtins.def
4001         (abs): Enable for all modes.
4002         * config/aarch64/arm_neon.h
4003         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4004         (vabs_f64): Add missing intrinsic.
4006 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
4008         * config/aarch64/arm_neon.h (vabs_s64): New function
4010 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
4012         PR target/57516
4013         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4014         * config/avr/avr.md (adjust_len): Add `round'.
4015         * config/avr/avr-protos.h (avr_out_round): New prototype.
4016         (avr_out_plus): Add `out_label' argument.
4017         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4018         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4019         Handle the case where `insn' is just a pattern.
4020         (avr_out_bitop): Handle the case where `insn' is just a pattern.
4021         (avr_out_round): New function.
4022         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4024 2013-07-18  David Holsgrove <david.holsgrove@xilinx.com>
4026         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4027         Add check for flag_stack_usage to handle -fstack-usage support
4029 2013-07-18  Pat Haugen <pthaugen@us.ibm.com>
4031         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4032         interaction for new Power8 flags and VSX.
4034 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
4036         PR middle-end/57698
4037         * tree-inline.c (expand_call_inline): Emit errors during
4038         early_inlining only if optimization is not turned on.
4040 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4042         * passes.def: New.
4044         * passes.c (init_optimization_passes): Move the construction of
4045         the pass hierarchy into a new passes.def file.
4047         * Makefile.in (passes.o): Add dependency on passes.def.
4049 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4051         * passes.c (init_optimization_passes): Introduce macros for
4052         constructing the tree of passes (INSERT_PASSES_AFTER,
4053         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4054         TERMINATE_PASS_LIST).
4056 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
4057             Wei Mi  <wmi@google.com>
4059         PR rtl-optimization/57878
4060         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4061         top.
4062         (reload_pseudo_compare_func): Check nregs first for reload
4063         pseudos.
4065 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4067         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4069 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
4071         * read-rtl.c (validate_const_int): Once an invalid character is
4072         seen, quit the loop.
4074         * gengtype.c (write_roots): Similarly once we find the "deletable"
4075         or "if_marked" option.
4077 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4079         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4080         "xtab" and "sat".  Move value "clz" from here to ...
4081         (attriubte "type"): ... here.
4082         (satsi_<SAT:code>): Delete "insn" attribute.
4083         (satsi_<SAT:code>_shift): Likewise.
4084         (arm_zero_extendqisi2addsi): Likewise.
4085         (arm_extendqisi2addsi): Likewise.
4086         (clzsi2): Update for attribute changes.
4087         (rbitsi2): Likewise.
4088         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4089         attribute.
4090         (arm_usatsihi): Likewise.
4091         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4093 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4095         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4096         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
4097         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
4098         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
4099         in alphabetical order.
4100         (attribute "core_cycles"): Update for attribute changes.
4101         (arm_addsi3): Likewise.
4102         (addsi3_compare0): Likewise.
4103         (addsi3_compare0_scratch): Likewise.
4104         (addsi3_compare_op1): Likewise.
4105         (addsi3_compare_op2): Likewise.
4106         (compare_addsi2_op0): Likewise.
4107         (compare_addsi2_op1): Likewise.
4108         (addsi3_carryin_shift_<optab>): Likewise.
4109         (subsi3_carryin_shift): Likewise.
4110         (rsbsi3_carryin_shift): Likewise.
4111         (arm_subsi3_insn): Likewise.
4112         (subsi3_compare0): Likewise.
4113         (subsi3_compare): Likewise.
4114         (arm_andsi3_insn): Likewise.
4115         (thumb1_andsi3_insn): Likewise.
4116         (andsi3_compare0): Likewise.
4117         (andsi3_compare0_scratch): Likewise.
4118         (zeroextractsi_compare0_scratch
4119         (andsi_not_shiftsi_si): Likewise.
4120         (iorsi3_insn): Likewise.
4121         (iorsi3_compare0): Likewise.
4122         (iorsi3_compare0_scratch): Likewise.
4123         (arm_xorsi3): Likewise.
4124         (thumb1_xorsi3_insn): Likewise.
4125         (xorsi3_compare0): Likewise.
4126         (xorsi3_compare0_scratch): Likewise.
4127         (satsi_<SAT:code>_shift): Likewise.
4128         (rrx): Likewise.
4129         (arm_shiftsi3): Likewise.
4130         (shiftsi3_compare0): Likewise.
4131         (not_shiftsi): Likewise.
4132         (not_shiftsi_compare0): Likewise.
4133         (not_shiftsi_compare0_scratch): Likewise.
4134         (arm_one_cmplsi2): Likewise.
4135         (thumb_one_complsi2): Likewise.
4136         (notsi_compare0): Likewise.
4137         (notsi_compare0_scratch): Likewise.
4138         (thumb1_zero_extendhisi2): Likewise.
4139         (arm_zero_extendhisi2): Likewise.
4140         (arm_zero_extendhisi2_v6): Likewise.
4141         (arm_zero_extendhisi2addsi): Likewise.
4142         (thumb1_zero_extendqisi2): Likewise.
4143         (thumb1_zero_extendqisi2_v6): Likewise.
4144         (arm_zero_extendqisi2): Likewise.
4145         (arm_zero_extendqisi2_v6): Likewise.
4146         (arm_zero_extendqisi2addsi): Likewise.
4147         (thumb1_extendhisi2): Likewise.
4148         (arm_extendhisi2): Likewise.
4149         (arm_extendhisi2_v6): Likewise.
4150         (arm_extendqisi): Likewise.
4151         (arm_extendqisi_v6): Likewise.
4152         (arm_extendqisi2addsi): Likewise.
4153         (thumb1_extendqisi2): Likewise.
4154         (thumb1_movdi_insn): Likewise.
4155         (arm_movsi_insn): Likewise.
4156         (movsi_compare0): Likewise.
4157         (movhi_insn_arch4): Likewise.
4158         (movhi_bytes): Likewise.
4159         (arm_movqi_insn): Likewise.
4160         (thumb1_movqi_insn): Likewise.
4161         (arm32_movhf): Likewise.
4162         (thumb1_movhf): Likewise.
4163         (arm_movsf_soft_insn): Likewise.
4164         (thumb1_movsf_insn): Likewise.
4165         (movdf_soft_insn): Likewise.
4166         (thumb_movdf_insn): Likewise.
4167         (arm_cmpsi_insn): Likewise.
4168         (cmpsi_shiftsi): Likewise.
4169         (cmpsi_shiftsi_swp): Likewise.
4170         (arm_cmpsi_negshiftsi_si): Likewise.
4171         (movsicc_insn): Likewise.
4172         (movsfcc_soft_insn): Likewise.
4173         (arith_shiftsi): Likewise.
4174         (arith_shiftsi_compare0
4175         (arith_shiftsi_compare0_scratch
4176         (sub_shiftsi): Likewise.
4177         (sub_shiftsi_compare0
4178         (sub_shiftsi_compare0_scratch
4179         (and_scc): Likewise.
4180         (cond_move): Likewise.
4181         (if_plus_move): Likewise.
4182         (if_move_plus): Likewise.
4183         (if_move_not): Likewise.
4184         (if_not_move): Likewise.
4185         (if_shift_move): Likewise.
4186         (if_move_shift): Likewise.
4187         (if_shift_shift): Likewise.
4188         (if_not_arith): Likewise.
4189         (if_arith_not): Likewise.
4190         (cond_move_not): Likewise.
4191         (thumb1_ashlsi3): Set type attribute.
4192         (thumb1_ashrsi3): Likewise.
4193         (thumb1_lshrsi3): Likewise.
4194         (thumb1_rotrsi3): Likewise.
4195         (shiftsi3_compare0_scratch): Likewise.
4196         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
4197         (neon_mov<mode>): Likewise.
4198         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
4199         attribute changes.
4200         (thumb2_movsi_insn): Likewise.
4201         (thumb2_cmpsi_neg_shiftsi): Likewise.
4202         (thumb2_extendqisi_v6): Likewise.
4203         (thumb2_zero_extendhisi2_v6): Likewise.
4204         (thumb2_zero_extendqisi2_v6): Likewise.
4205         (thumb2_shiftsi3_short): Likewise.
4206         (thumb2_addsi3_compare0_scratch): Likewise.
4207         (orsi_not_shiftsi_si): Likewise.
4208         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
4209         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
4210         changes.
4211         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
4212         (1020alu_shift_op): Likewise.
4213         (1020alu_shift_reg_op): Likewise.
4214         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
4215         (alu_shift_op): Likewise.
4216         (alu_shift_reg_op): Likewise.
4217         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
4218         (11_alu_shift_op): Likewise.
4219         (11_alu_shift_reg_op): Likewise.
4220         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
4221         (9_alu_shift_reg_op): Likewise.
4222         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4223         changes.
4224         (cortex_a15_alu_shift): Likewise.
4225         (cortex_a15_alu_shift_reg): Likewise.
4226         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
4227         changes.
4228         (cortex_a5_alu_shift): Likewise.
4229         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
4230         changes.
4231         (cortex_a53_alu_shift): Likewise.
4232         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4233         changes.
4234         (cortex_a7_alu_reg): Likewise.
4235         (cortex_a7_alu_shift): Likewise.
4236         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
4237         changes.
4238         (cortex_a8_alu_shift): Likewise.
4239         (cortex_a8_alu_shift_reg): Likewise.
4240         (cortex_a8_mov): Likewise.
4241         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
4242         (cortex_a9_dp_shift): Likewise.
4243         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
4244         changes.
4245         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
4246         changes.
4247         (cortex_r4_mov): Likewise.
4248         (cortex_r4_alu_shift): Likewise.
4249         (cortex_r4_alu_shift_reg): Likewise.
4250         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
4251         (526_alu_shift_op): Likewise.
4252         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
4253         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
4254         (626te_alu_shift_op): Likewise.
4255         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
4256         (726te_alu_op): Likewise.
4257         (726te_alu_shift_op): Likewise.
4258         (726te_alu_shift_reg_op): Likewise.
4259         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
4260         (mp626_alu_shift_op): Likewise.
4261         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
4262         (pj4_alu_e1_conds): Likewise.
4263         (pj4_alu): Likewise.
4264         (pj4_alu_conds): Likewise.
4265         (pj4_shift): Likewise.
4266         (pj4_shift_conds): Likewise.
4267         (pj4_alu_shift): Likewise.
4268         (pj4_alu_shift_conds): Likewise.
4269         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4270         changes.
4271         (cortexa7_older_only): Likewise.
4272         (cortexa7_younger): Likewise.
4274 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4276         * ipa-pure-const.c (generate_summary): Rename to...
4277         (pure_const_generate_summary): ... this.
4279 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
4281         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
4283 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
4285         PR target/57909
4286         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
4287         usage in HI mode.
4289 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4291         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
4292         enabled without -march=zEC12.
4293         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
4294         flags to be set.
4296 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4298         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
4299         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
4300         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
4301         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
4302         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
4303         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
4304         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
4305         ISA_HAS_FP4.
4306         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
4307         and ISA_HAS_NMADD3_NMSUB3.
4308         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
4309         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
4310         (nmsub4<mode>, nmsub3<mode>): Likewise.
4311         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
4313 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4315         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
4316         TARGET_MIPS5400 checking.
4318 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
4319             Peter Bergner  <bergner@vnet.ibm.com>
4321         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
4322         registers in the comment.
4323         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
4324         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
4325         rather than FIRST_PSEUDO_REGISTERS.
4327 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
4329         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
4330         enable extra ISA flags with TARGET_HTM.
4332 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4334         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
4335         Fix comment typos.
4337 2013-07-15  Cong Hou  <congh@google.com>
4339         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
4340         in compare function for sorting.
4342 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
4344         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
4345         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
4346         * config/rs6000/rs6000.opt: Add -mhtm option.
4347         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
4348         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
4349         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
4350         __HTM__ if the HTM instructions are available.
4351         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
4352         htm_spr_reg_operand): New define_predicates.
4353         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
4354         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
4355         Include htm.md.
4356         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
4357         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
4358         HTM builtin functions.
4359         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
4360         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
4361         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
4362         (rs6000_builtin_mask_calculate): Likewise.
4363         (rs6000_option_override_internal): Likewise.
4364         (bdesc_htm): Add new HTM builtin support.
4365         (htm_spr_num): New function.
4366         (htm_spr_regno): Likewise.
4367         (rs6000_htm_spr_icode): Likewise.
4368         (htm_expand_builtin): Likewise.
4369         (htm_init_builtins): Likewise.
4370         (rs6000_expand_builtin): Add support for HTM builtin functions.
4371         (rs6000_init_builtins): Likewise.
4372         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
4373         option.
4374         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
4375         (TARGET_HTM, MASK_HTM): Define macros.
4376         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
4377         (FIXED_REGISTERS): Likewise.
4378         (CALL_USED_REGISTERS): Likewise.
4379         (CALL_REALLY_USED_REGISTERS): Likewise.
4380         (REG_ALLOC_ORDER): Likewise.
4381         (enum reg_class): Likewise.
4382         (REG_CLASS_NAMES): Likewise.
4383         (REG_CLASS_CONTENTS): Likewise.
4384         (REGISTER_NAMES): Likewise.
4385         (ADDITIONAL_REGISTER_NAMES): Likewise.
4386         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
4387         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
4388         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
4389         * config/rs6000/htm.md: New file.
4390         * config/rs6000/htmintrin.h: New file.
4391         * config/rs6000/htmxlintrin.h: New file.
4393 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4395         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4396         Define SYMBOL_TINY_GOT, update comment.
4397         * config/aarch64/aarch64.c
4398         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
4399         (aarch64_expand_mov_immediate): Likewise.
4400         (aarch64_print_operand): Likewise.
4401         (aarch64_classify_symbol): Likewise.
4402         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
4403         (ldr_got_tiny): Define.
4405 2013-07-13  Tobias Grosser  <tobias@grosser.es>
4407         PR tree-optimization/54094
4408         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
4409           scheduling dimension for the parallelism check from the polyhedral
4410           information in the AST.
4411         * graphite-dependences.c (carries_deps): Do not assume the schedule is
4412           in 2D + 1 form.
4414 2013-07-13  Jason Merrill  <jason@redhat.com>
4416         * print-tree.c (debug_vec_tree): Use debug_raw.
4417         (debug_raw (vec<tree, va_gc> &)): New.
4418         (debug_raw (vec<tree, va_gc> *)): New.
4419         * tree.h: Declare them.
4421 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
4423         * ifcvt.c (ifcvt_after_combine): New static variable.
4424         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
4425         for size.
4426         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
4427         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
4428         rest_of_handle_if_after_reload): Pass new argument for if_convert.
4430 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
4432         * config/mips/mips.c (mips_expand_call): Remove empty statement.
4434 2013-07-12  Michael Matz  <matz@suse.de>
4436         PR middle-end/55771
4437         * convert.c (convert_to_real): Reject non-float inner types.
4439 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
4441         * config/aarch64/aarch64-protos.h
4442         (aarch64_simd_immediate_valid_for_move): Remove.
4443         * config/aarch64/aarch64.c (simd_immediate_info): New member.
4444         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
4445         cases.
4446         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
4448 2013-07-11  Steve Ellcey  <sellcey@mips.com>
4450         * config/mips/mips.c (mips_conditional_register_usage): Do not
4451         use t[0-7] registers in MIPS16 mode when optimizing for size.
4453 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
4455         * config/i386/i386.c (dispatch_function_versions): Fix array
4456         indexing of function_version_info to match actual_versions.
4458 2013-07-11  Teresa Johnson  <tejohnson@google.com>
4460         * vec.h (struct va_gc): Move release out-of-line.
4461         (va_gc::release): Call ggc_free on released vec.
4463 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4465         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
4466         Require GOT register as additional operand in UNSPEC.
4467         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
4468         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4469         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4470         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
4471         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
4472         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
4473         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
4475 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
4477         PR target/57631
4478         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
4479         name seen by assembler/linker rather if available.
4481 2013-07-11  Andreas Schwab  <schwab@suse.de>
4483         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
4485 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
4487         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
4489 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
4491         * doc/tm.texi.in: Move hook documentation to ....
4492         * target.def: ... here.
4494         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
4495         text on @hook line.
4496         * doc/tm.texi: Regenerate.
4498 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
4500         PR c++/57869
4501         * doc/invoke.texi: Document -Wconditionally-supported.
4503 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4505         PR target/57844
4506         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
4507         of my_fp.
4509 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4511         PR target/57506
4512         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
4513         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
4514         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
4515         Remove duplicate devices.
4516         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
4517         * config/avr/t-multilib: Regenerate.
4518         * config/avr/avr-tables.opt: Regenerate.
4519         * doc/avr-mmcu.texi: Regenerate.
4521 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4523         PR target/56987
4524         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
4526 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
4528         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
4529         the cost of MULT when optimizing for size.
4531 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4533         * config/cr16/cr16-protos.h: Don't include target.h.
4535 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
4537         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
4538         adjust register size for TDmode and TFmode for VSX registers.
4540 2013-07-08  Kai Tietz  <ktietz@redhat.com>
4542         PR target/56892
4543         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
4544         hook_bool_const_tree_true.
4546 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4548         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
4549         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
4550         * config/s390/s390.md: Define FPR*_REGNUM constants.
4551         Fix FPR2_REGNUM constant (18 -> 17).
4552         ("*trunc<BFP:mode><DFP_ALL:mode>2")
4553         ("*trunc<DFP_ALL:mode><BFP:mode>2")
4554         ("trunc<BFP:mode><DFP_ALL:mode>2")
4555         ("trunc<DFP_ALL:mode><BFP:mode>2")
4556         ("*extend<BFP:mode><DFP_ALL:mode>2")
4557         ("*extend<DFP_ALL:mode><BFP:mode>2")
4558         ("extend<BFP:mode><DFP_ALL:mode>2")
4559         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
4560         FPR4_REGNUM.
4562 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
4564         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
4566 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4568         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
4569         and cfun_fpr_bit_p to cfun_fpr_save_p.
4570         (s390_frame_area, s390_register_info, s390_frame_info)
4571         (s390_emit_prologue, s390_emit_epilogue)
4572         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
4573         register numbers.
4574         * config/s390/s390.h: Define *_REGNUM macros for floating point
4575         register numbers.
4577 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
4579         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
4581 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
4583         PR rtl-optimization/57786
4584         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
4585         and break out of the loop when it is set to false.
4587 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
4589         PR target/57819
4590         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
4591         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
4592         (const_int 63)) 0)).
4593         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
4594         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
4595         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
4597         PR rtl-optimization/57829
4598         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
4599         mask bits outside of mode are just sign-extension from mode to HWI.
4601 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
4603         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
4604         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
4605         adjust_address instead of change_address to keep info about alignment.
4606         (emit_strmov): Remove.
4607         (emit_memmov): New function.
4608         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
4609         (expand_movmem_epilogue): Likewise and return updated rtx for
4610         destination.
4611         (expand_constant_movmem_prologue): Likewise and return updated rtx for
4612         destination and source.
4613         (decide_alignment): Refactor, handle vector_loop.
4614         (ix86_expand_movmem): Likewise.
4615         (ix86_expand_setmem): Likewise.
4616         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
4618 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
4620         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
4621         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
4623 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4625         * config/i386/sse.md (sse_movlhps): Change alternative 3
4626         of operand 2 to "m".
4628 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4630         PR target/57807
4631         * config/i386/sse.md (iptr): New mode attribute.
4632         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
4633         (<sse>_vm<plusminus_insn><mode>3): Ditto.
4634         (<sse>_vmmul<mode>3): Ditto.
4635         (<sse>_vmdiv<mode>3): Ditto.
4636         (sse_vmrcpv4sf2): Ditto.
4637         (<sse>_vmsqrt<mode>2): Ditto.
4638         (sse_vmrsqrtv4sf2): Ditto.
4639         (<sse>_vm<code><mode>3): Ditto.
4640         (avx_vmcmp<mode>3): Ditto.
4641         (<sse>_vmmaskcmp<mode>3): Ditto.
4642         (<sse>_comi): Ditto.
4643         (<sse>_ucomi): Ditto.
4644         (*xop_vmfrcz_<mode>): Ditto.
4645         (*fmai_fmadd_<mode>): Ditto.
4646         (*fmai_fmsub_<mode>): Ditto.
4647         (*fmai_fnmadd_<mode>): Ditto.
4648         (*fmai_fnmsub_<mode>): Ditto.
4649         (*fma4i_vmfmadd_<mode>): Ditto.
4650         (*fma4i_vmfmsub_<mode>): Ditto.
4651         (*fma4i_vmfnmadd_<mode>): Ditto.
4652         (*fma4i_vmfnmsub_<mode>): Ditto.
4653         (*xop_vmfrcz_<mode>): Ditto.
4654         (sse_cvtps2pi): Ditto.
4655         (sse_cvttps2pi): Ditto.
4656         (sse_cvtss2si): Ditto.
4657         (sse_cvtss2si_2): Ditto.
4658         (sse_cvtss2siq_2): Ditto.
4659         (sse_cvttss2si): Ditto.
4660         (sse_cvttss2siq): Ditto.
4661         (sse_cvtsd2si): Ditto.
4662         (sse_cvtsd2si_2): Ditto.
4663         (sse_cvtsd2siq_2): Ditto.
4664         (sse_cvttsd2si): Ditto.
4665         (sse_cvttsd2siq): Ditto.
4666         (sse_cvtsd2ss): Ditto.
4667         (sse_cvtss2sd): Ditto.
4668         (avx2_pbroadcast<mode>): Ditto.
4669         (avx2_pbroadcast<mode>_1): Ditto.
4670         (*avx_vperm_broadcast_v4sf): Ditto.
4672         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
4673         (sse_movlhps): Ditto.
4674         (sse_storehps): Ditto.
4675         (sse_loadhps): Ditto.
4676         (sse_storelps): Ditto.
4677         (sse_loadlps): Ditto.
4678         (*vec_concatv4sf): Ditto.
4679         (*vec_interleave_highv2df): Ditto.
4680         (*vec_interleave_lowv2df): Ditto.
4681         (*vec_extractv2df_1_sse): Ditto.
4682         (*vec_extractv2df_0_sse): Ditto.
4683         (sse2_storelpd): Ditto.
4684         (sse2_loadlpd): Ditto.
4685         (sse2_movsd): Ditto.
4686         (*vec_concatv4si): Ditto.
4687         (vec_concatv2di): Ditto.
4689         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
4690         for Intel asm dialect.
4691         (mmx_punpcklwd): Ditto.
4692         (mmx_punpckldq): Ditto.
4694         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
4695         for intel assembler dialect.
4697 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
4699         PR target/29776
4700         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
4701         for BUILT_IN_C{LZ,LRSB}*.
4702         * tree.h (CASE_INT_FN): Add FN##IMAX case.
4703         * tree-vrp.c (extract_range_basic): Handle
4704         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
4705         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
4706         fall thru to code calling set_value*.
4707         * builtins.c (expand_builtin): Remove *IMAX cases.
4708         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
4709         if width is bigger than 2*HWI.
4711 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
4713         PR rtl-optimization/55342
4714         * lra-int.h (lra_subreg_reload_pseudos): New.
4715         * lra.c: Add undoing optional reloads to the block diagram.
4716         (lra_subreg_reload_pseudos): New.
4717         (lra_optional_reload_pseudos): Change comments.
4718         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
4719         lra_optional_reload_pseudos after undo transformations.
4720         * lra-assigns.c (pseudo_prefix_title): New.
4721         (lra_setup_reg_renumber): Use it.
4722         (spill_for): Ditto.  Check subreg reload pseudos too.
4723         (assign_by_spills): Consider subreg reload pseudos too.
4724         * lra-constraints.c (simplify_operand_subreg): Use
4725         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
4726         (curr_insn_transform): Recognize and do optional reloads.
4727         (undo_optional_reloads): New.
4728         (lra_undo_inheritance): Call undo_optional_reloads.
4730 2013-07-05  Thomas Quinot  <quinot@adacore.com>
4732         * tree-complex.c (expand_complex_operations_1): Fix typo.
4734 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
4736         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
4737         (tune_params): New member 'const vec_costs'.
4738         * config/aarch64/aarch64.c (generic_vector_cost): New.
4739         (generic_tunings): New member 'generic_vector_cost'.
4740         (aarch64_builtin_vectorization_cost): New.
4741         (aarch64_add_stmt_cost): New.
4742         (TARGET_VECTORIZE_ADD_STMT_COST): New.
4743         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
4745 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
4747         PR target/57777
4748         * config/i386/predicates.md (vsib_address_operand): Disallow
4749         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
4751 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
4753         PR middle-end/55030
4754         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
4755         expand_builtin_setjmp_receiver.
4756         (expand_label): Adjust, call expand_builtin_setjmp_receiver
4757         with NULL for the label parameter.
4758         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
4759         the frame-pointer.  Adjust comments.
4760         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
4761         only if LABEL is non-NULL.
4763 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
4765         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
4766         (ARM_ABI_AAPCS64): Ditto.
4767         (arm_abi): Ditto.
4768         (ARM_DEFAULT_ABI): Ditto.
4770 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
4772         * config/aarch64/aarch64-builtins.c
4773         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
4774         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
4775         (st1): Likewise.
4776         * config/aarch64/aarch64-simd.md
4777         (aarch64_ld1<VALL:mode>): New.
4778         (aarch64_st1<VALL:mode>): Likewise.
4779         * config/aarch64/arm_neon.h
4780         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
4782 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
4784         * config/i386/i386.c (gate_insert_vzeroupper): Check if
4785         target ISA is AVX.
4786         (ix86_option_override_internal):Turn on all -mavx target flags by
4787         default as they are dependent on AVX anyway.
4789 2013-07-02  Cary Coutant  <ccoutant@google.com>
4791         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
4792         deterministic hash.
4793         (loc_checksum_ordered): Likewise.
4794         (hash_loc_operands): Remove inline keyword.
4796 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
4798         PR tree-optimization/57741
4799         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
4800         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
4801         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
4802         Allow REAL_CST step_exprs if flag_associative_math.
4803         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
4805 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4807         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
4809 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4811         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
4813 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4815         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
4816         encoding.
4817         (iorsi3_insn): Likewise.
4818         (arm_xorsi3): Likewise.
4820 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
4822         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
4823         to ...
4824         (attribute "type"): ... here, and prefix with "wmmx_".
4825         (attribute "core_cycles"): Update for attribute changes.
4826         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
4827         (tbcstv4hi): Likewise.
4828         (tbcstv2si): Likewise.
4829         (iwmmxt_iordi3): Likewise.
4830         (iwmmxt_xordi3): Likewise.
4831         (iwmmxt_anddi3): Likewise.
4832         (iwmmxt_nanddi3): Likewise.
4833         (iwmmxt_arm_movdi): Likewise.
4834         (iwmmxt_movsi_insn): Likewise.
4835         (mov<mode>_internal): Likewise.
4836         (and<mode>3_iwmmxt): Likewise.
4837         (ior<mode>3_iwmmxt): Likewise.
4838         (xor<mode>3_iwmmxt): Likewise.
4839         (add<mode>3_iwmmxt): Likewise.
4840         (ssaddv8qi3): Likewise.
4841         (ssaddv4hi3): Likewise.
4842         (ssaddv2si3): Likewise.
4843         (usaddv8qi3): Likewise.
4844         (usaddv4hi3): Likewise.
4845         (usaddv2si3): Likewise.
4846         (sub<mode>3_iwmmxt): Likewise.
4847         (sssubv8qi3): Likewise.
4848         (sssubv4hi3): Likewise.
4849         (sssubv2si3): Likewise.
4850         (ussubv8qi3): Likewise.
4851         (ussubv4hi3): Likewise.
4852         (ussubv2si3): Likewise.
4853         (mulv4hi3_iwmmxt): Likewise.
4854         (smulv4hi3_highpart): Likewise.
4855         (umulv4hi3_highpart): Likewise.
4856         (iwmmxt_wmacs): Likewise.
4857         (iwmmxt_wmacsz): Likewise.
4858         (iwmmxt_wmacu): Likewise.
4859         (iwmmxt_wmacuz): Likewise.
4860         (iwmmxt_clrdi): Likewise.
4861         (iwmmxt_clrv8qi): Likewise.
4862         (iwmmxt_clr4hi): Likewise.
4863         (iwmmxt_clr2si): Likewise.
4864         (iwmmxt_uavgrndv8qi3): Likewise.
4865         (iwmmxt_uavgrndv4hi3): Likewise.
4866         (iwmmxt_uavgv8qi3): Likewise.
4867         (iwmmxt_uavgv4hi3): Likewise.
4868         (iwmmxt_tinsrb): Likewise.
4869         (iwmmxt_tinsrh): Likewise.
4870         (iwmmxt_tinsrw): Likewise.
4871         (iwmmxt_textrmub): Likewise.
4872         (iwmmxt_textrmsb): Likewise.
4873         (iwmmxt_textrmuh): Likewise.
4874         (iwmmxt_textrmsh): Likewise.
4875         (iwmmxt_textrmw): Likewise.
4876         (iwmxxt_wshufh): Likewise.
4877         (eqv8qi3): Likewise.
4878         (eqv4hi3): Likewise.
4879         (eqv2si3): Likewise.
4880         (gtuv8qi3): Likewise.
4881         (gtuv4hi3): Likewise.
4882         (gtuv2si3): Likewise.
4883         (gtv8qi3): Likewise.
4884         (gtv4hi3): Likewise.
4885         (gtv2si3): Likewise.
4886         (smax<mode>3_iwmmxt): Likewise.
4887         (umax<mode>3_iwmmxt): Likewise.
4888         (smin<mode>3_iwmmxt): Likewise.
4889         (umin<mode>3_iwmmxt): Likewise.
4890         (iwmmxt_wpackhss): Likewise.
4891         (iwmmxt_wpackwss): Likewise.
4892         (iwmmxt_wpackdss): Likewise.
4893         (iwmmxt_wpackhus): Likewise.
4894         (iwmmxt_wpackwus): Likewise.
4895         (iwmmxt_wpackdus): Likewise.
4896         (iwmmxt_wunpckihb): Likewise.
4897         (iwmmxt_wunpckihh): Likewise.
4898         (iwmmxt_wunpckihw): Likewise.
4899         (iwmmxt_wunpckilb): Likewise.
4900         (iwmmxt_wunpckilh): Likewise.
4901         (iwmmxt_wunpckilw): Likewise.
4902         (iwmmxt_wunpckehub): Likewise.
4903         (iwmmxt_wunpckehuh): Likewise.
4904         (iwmmxt_wunpckehuw): Likewise.
4905         (iwmmxt_wunpckehsb): Likewise.
4906         (iwmmxt_wunpckehsh): Likewise.
4907         (iwmmxt_wunpckehsw): Likewise.
4908         (iwmmxt_wunpckelub): Likewise.
4909         (iwmmxt_wunpckeluh): Likewise.
4910         (iwmmxt_wunpckeluw): Likewise.
4911         (iwmmxt_wunpckelsb): Likewise.
4912         (iwmmxt_wunpckelsh): Likewise.
4913         (iwmmxt_wunpckelsw): Likewise.
4914         (ror<mode>3): Likewise.
4915         (ashr<mode>3_iwmmxt): Likewise.
4916         (lshr<mode>3_iwmmxt): Likewise.
4917         (ashl<mode>3_iwmmxt): Likewise.
4918         (ror<mode>3_di): Likewise.
4919         (ashr<mode>3_di): Likewise.
4920         (lshr<mode>3_di): Likewise.
4921         (ashl<mode>3_di): Likewise.
4922         (iwmmxt_wmadds): Likewise.
4923         (iwmmxt_wmaddu): Likewise.
4924         (iwmmxt_tmia): Likewise.
4925         (iwmmxt_tmiaph): Likewise.
4926         (iwmmxt_tmiabb): Likewise.
4927         (iwmmxt_tmiatb): Likewise.
4928         (iwmmxt_tmiabt): Likewise.
4929         (iwmmxt_tmiatt): Likewise.
4930         (iwmmxt_tmovmskb): Likewise.
4931         (iwmmxt_tmovmskh): Likewise.
4932         (iwmmxt_tmovmskw): Likewise.
4933         (iwmmxt_waccb): Likewise.
4934         (iwmmxt_wacch): Likewise.
4935         (iwmmxt_waccw): Likewise.
4936         (iwmmxt_waligni): Likewise.
4937         (iwmmxt_walignr): Likewise.
4938         (iwmmxt_walignr0): Likewise.
4939         (iwmmxt_walignr1): Likewise.
4940         (iwmmxt_walignr2): Likewise.
4941         (iwmmxt_walignr3): Likewise.
4942         (iwmmxt_wsadb): Likewise.
4943         (iwmmxt_wsadh): Likewise.
4944         (iwmmxt_wsadbz): Likewise.
4945         (iwmmxt_wsadhz): Likewise.
4946         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
4947         (iwmmxt_wabsdiffb): Likewise.
4948         (iwmmxt_wabsdiffh): Likewise.
4949         (iwmmxt_wabsdiffw): Likewise.
4950         (iwmmxt_waddsubhx): Likewise
4951         (iwmmxt_wsubaddhx): Likewise.
4952         (addc<mode>3): Likewise.
4953         (iwmmxt_avg4): Likewise.
4954         (iwmmxt_avg4r): Likewise.
4955         (iwmmxt_wmaddsx): Likewise.
4956         (iwmmxt_wmaddux): Likewise.
4957         (iwmmxt_wmaddsn): Likewise.
4958         (iwmmxt_wmaddun): Likewise.
4959         (iwmmxt_wmulwsm): Likewise.
4960         (iwmmxt_wmulwum): Likewise.
4961         (iwmmxt_wmulsmr): Likewise.
4962         (iwmmxt_wmulumr): Likewise.
4963         (iwmmxt_wmulwsmr): Likewise.
4964         (iwmmxt_wmulwumr): Likewise.
4965         (iwmmxt_wmulwl): Likewise.
4966         (iwmmxt_wqmulm): Likewise.
4967         (iwmmxt_wqmulwm): Likewise.
4968         (iwmmxt_wqmulmr): Likewise.
4969         (iwmmxt_wqmulwmr): Likewise.
4970         (iwmmxt_waddbhusm): Likewise.
4971         (iwmmxt_waddbhusl): Likewise.
4972         (iwmmxt_wqmiabb): Likewise.
4973         (iwmmxt_wqmiabt): Likewise.
4974         (iwmmxt_wqmiatb): Likewise.
4975         (iwmmxt_wqmiatt): Likewise.
4976         (iwmmxt_wqmiabbn): Likewise.
4977         (iwmmxt_wqmiabtn): Likewise.
4978         (iwmmxt_wqmiatbn): Likewise.
4979         (iwmmxt_wqmiattn): Likewise.
4980         (iwmmxt_wmiabb): Likewise.
4981         (iwmmxt_wmiabt): Likewise.
4982         (iwmmxt_wmiatb): Likewise.
4983         (iwmmxt_wmiatt): Likewise.
4984         (iwmmxt_wmiabbn): Likewise.
4985         (iwmmxt_wmiabtn): Likewise.
4986         (iwmmxt_wmiatbn): Likewise.
4987         (iwmmxt_wmiattn): Likewise.
4988         (iwmmxt_wmiawbb): Likewise.
4989         (iwmmxt_wmiawbt): Likewise.
4990         (iwmmxt_wmiawtb): Likewise.
4991         (iwmmxt_wmiawtt): Likewise.
4992         (iwmmxt_wmiawbbn): Likewise.
4993         (iwmmxt_wmiawbtn): Likewise.
4994         (iwmmxt_wmiawtbn): Likewise.
4995         (iwmmxt_wmiawttn): Likewise.
4996         (iwmmxt_wmerge): Likewise.
4997         (iwmmxt_tandc<mode>3): Likewise.
4998         (iwmmxt_torc<mode>3): Likewise.
4999         (iwmmxt_torvsc<mode>3): Likewise.
5000         (iwmmxt_textrc<mode>3): Likewise.
5001         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5002         (wmmxt_pack): Likewise.
5003         (wmmxt_mult_c1): Likewise.
5004         (wmmxt_mult_c2): Likewise.
5005         (wmmxt_alu_c1): Likewise.
5006         (wmmxt_alu_c2): Likewise.
5007         (wmmxt_alu_c3): Likewise.
5008         (wmmxt_transfer_c1): Likewise.
5009         (wmmxt_transfer_c2): Likewise.
5010         (wmmxt_transfer_c3): Likewise.
5011         (marvell_f_iwmmxt_wstr): Likewise.
5012         (marvell_f_iwmmxt_wldr): Likewise.
5014 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
5016         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5018 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5020         Revert:
5021         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5022         * lra-constraints.c (need_for_split_p): Check call used hard regs
5023         living through calls.
5025         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5026         call used regs for call insn.
5028 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
5030         PR target/57736
5031         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5032         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5034 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
5036         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5038 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5040         * lra-constraints.c (need_for_split_p): Check call used hard regs
5041         living through calls.
5043 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
5045         PR target/57744
5046         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5047         to tie with any other modes.  Eliminate Altivec vector mode tests,
5048         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
5049         code, to return 0 if testing MODE2 for a condition, if we've
5050         already tested MODE1 for the same condition.
5052 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5054         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5055         layout.
5057 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5059         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5060         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5062 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5064         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5065         Define.
5066         (aarch64_symbolic_constant_p): Remove.
5067         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5068         static.  Fix line length and white space.
5069         (aarch64_symbolic_constant_p): Remove.
5070         * config/aarch64/predicates.md (aarch64_valid_symref):
5071         Use aarch64_classify_symbol_expression.
5073 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5075         * config/arm/constraints.md (Ts): New constraint.
5076         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5077         16-bit encodings.
5078         (compare_scc): Use "Ts" constraint for operand 0.
5079         (ior_scc_scc): Likewise.
5080         (and_scc_scc): Likewise.
5081         (and_scc_scc_nodom): Likewise.
5082         (ior_scc_scc_cmp): Likewise for operand 7.
5083         (and_scc_scc_cmp): Likewise.
5084         * config/arm/thumb2.md (thumb2_movsi_insn):
5085         Add alternatives for 16-bit encodings.
5086         (thumb2_movhi_insn): Likewise.
5087         (thumb2_movsicc_insn): Likewise.
5088         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
5089         (thumb2_negscc): Use "Ts" constraint.
5090         Move mvn instruction outside cond_exec block.
5091         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5092         for 16-bit encodings.
5094 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5096         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5097         encoding.
5098         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5099         (mulsi3subsi): Likewise.
5100         (mulsidi3adddi): Likewise.
5101         (mulsidi3_v6): Likewise.
5102         (umulsidi3_v6): Likewise.
5103         (umulsidi3adddi_v6): Likewise.
5104         (smulsi3_highpart_v6): Likewise.
5105         (umulsi3_highpart_v6): Likewise.
5106         (mulhisi3tb): Likewise.
5107         (mulhisi3bt): Likewise.
5108         (mulhisi3tt): Likewise.
5109         (maddhisi4): Likewise.
5110         (maddhisi4tb): Likewise.
5111         (maddhisi4tt): Likewise.
5112         (maddhidi4): Likewise.
5113         (maddhidi4tb): Likewise.
5114         (maddhidi4tt): Likewise.
5115         (zeroextractsi_compare0_scratch): Likewise.
5116         (insv_zero): Likewise.
5117         (insv_t2): Likewise.
5118         (anddi_notzesidi_di): Likewise.
5119         (anddi_notsesidi_di): Likewise.
5120         (andsi_notsi_si): Likewise.
5121         (iordi_zesidi_di): Likewise.
5122         (xordi_zesidi_di): Likewise.
5123         (andsi_iorsi3_notsi): Likewise.
5124         (smax_0): Likewise.
5125         (smax_m1): Likewise.
5126         (smin_0): Likewise.
5127         (not_shiftsi): Likewise.
5128         (unaligned_loadsi): Likewise.
5129         (unaligned_loadhis): Likewise.
5130         (unaligned_loadhiu): Likewise.
5131         (unaligned_storesi): Likewise.
5132         (unaligned_storehi): Likewise.
5133         (extv_reg): Likewise.
5134         (extzv_t2): Likewise.
5135         (divsi3): Likewise.
5136         (udivsi3): Likewise.
5137         (arm_zero_extendhisi2addsi): Likewise.
5138         (arm_zero_extendqisi2addsi): Likewise.
5139         (compareqi_eq0): Likewise.
5140         (arm_extendhisi2_v6): Likewise.
5141         (arm_extendqisi2addsi): Likewise.
5142         (arm_movt): Likewise.
5143         (thumb2_ldrd): Likewise.
5144         (thumb2_ldrd_base): Likewise.
5145         (thumb2_ldrd_base_neg): Likewise.
5146         (thumb2_strd): Likewise.
5147         (thumb2_strd_base): Likewise.
5148         (thumb2_strd_base_neg): Likewise.
5149         (arm_negsi2): Add alternative for 16-bit encoding.
5150         (arm_one_cmplsi2): Likewise.
5152 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5154         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
5155         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
5156         (movdfcc): Likewise.
5157         * config/arm/vfp.md (*thumb2_movsf_vfp):
5158         Disable predication for arm_restrict_it.
5159         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
5160         (*thumb2_movdfcc_vfp): Likewise.
5161         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
5162         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
5163         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
5164         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
5165         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
5166         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
5167         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
5168         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
5169         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
5170         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
5171         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
5172         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
5173         Disable predication for arm_restrict_it.
5175 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
5177         * config/i386/bmiintrin.h (_bextr_u32): New.
5178         (_bextr_u64): Ditto.
5180 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
5182         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
5183         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
5184         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
5185         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
5186         * config/mips/n32-elf.h: ...this new file.
5188 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
5190         PR target/57224
5191         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
5192         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
5194 2013-06-27  Catherine Moore <clm@codesourcery.com>
5196         * config/mips/mips-tables.opt: Regenerate.
5197         * config/mips/mips-cpus.def: Add m14ke and m14kec.
5198         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
5199         * doc/invoke.texi: Add -m14kc.
5201 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5203         PR target/57623
5204         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
5205         constraints of operand 1 and 2.
5207         PR target/57623
5208         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
5209         to match RTL canonicalization.  Swap predicates and
5210         constraints of operand 1 and 2.
5212 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
5214         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
5215         Process OP_INOUT regs for splitting too.
5217 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5219         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
5220         decl before the loop, initialize to NULL.
5221         (vectorizable_load): Initialize ptr_incr to NULL.
5223 2013-06-27  Martin Jambor  <mjambor@suse.cz>
5225         PR lto/57208
5226         * ipa-ref.h (ipa_maybe_record_reference): Declare.
5227         * ipa-ref.c (ipa_maybe_record_reference): New function.
5228         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
5229         * ipa-cp.c (create_specialized_node): Record potential references from
5230         aggvals.
5231         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
5233 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5235         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
5236         parameter 'mode' of type 'enum machine_mode mode'; change to pass
5237         'mode' to force_reg.
5238         (aarch64_add_offset): Update calls to aarch64_force_temporary.
5239         (aarch64_expand_mov_immediate): Likewise.
5241 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5243         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
5244         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
5246 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5248         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
5249         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5250         (struct machine_function): Add tbegin_p.
5251         (s390_canonicalize_comparison): Fold CC mode compares to
5252         conditional jump if possible.
5253         (s390_emit_jump): Return the emitted jump.
5254         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
5255         Handle CCRAWmode compares.
5256         (s390_option_override): Default to -mhtm if available.
5257         (s390_reg_clobbered_rtx): Handle floating point regs as well.
5258         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
5259         FPRs instead of df_regs_ever_live_p.
5260         (s390_optimize_nonescaping_tx): New function.
5261         (s390_init_frame_layout): Extend clobbered_regs array to cover
5262         FPRs as well.
5263         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
5264         (s390_expand_tbegin): New function.
5265         (enum s390_builtin): New enum definition.
5266         (code_for_builtin): New array definition.
5267         (s390_init_builtins): New function.
5268         (s390_expand_builtin): New function.
5269         (TARGET_INIT_BUILTINS): Define.
5270         (TARGET_EXPAND_BUILTIN): Define.
5271         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
5272         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
5273         (s390_alc_comparison): Likewise.
5274         * config/s390/s390-modes.def: Add CCRAWmode.
5275         * config/s390/s390.h (processor_flags): Add PF_TX.
5276         (TARGET_CPU_HTM): Define macro.
5277         (TARGET_HTM): Define macro.
5278         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
5279         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
5280         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5281         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
5282         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
5283         (TBEGIN_MASK, TBEGINC_MASK): New constants.
5284         ("*cc_to_int"): Move up.
5285         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
5286         constants other than 0.
5287         ("*ccraw_to_int"): New insn and splitter definition.
5288         ("tbegin", "tbegin_nofloat", "tbegin_retry")
5289         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
5290         ("tx_assist"): New expander.
5291         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
5292         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
5293         * config/s390/s390.opt: Add -mhtm option.
5294         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
5295         * config/s390/htmxlintrin.h: New file.
5296         * config/s390/htmintrin.h: New file.
5297         * config/s390/s390intrin.h: New file.
5298         * doc/extend.texi: Document htm builtins.
5299         * config.gcc: Add the new header files to extra_headers.
5301 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
5303         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
5304         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
5306 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
5307             Pat Haugen <pthaugen@us.ibm.com>
5308             Peter Bergner <bergner@vnet.ibm.com>
5310         * config/rs6000/power8.md: New.
5311         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
5312         setting for power8 entry.
5313         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
5314         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
5315         test for Power4/Power5 only.
5316         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
5317         support.
5318         (force_new_group): Adjust comment.
5319         * config/rs6000/rs6000.md: Include power8.md.
5321 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
5323         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
5324         * config/arm/arm-protos.h (arm_max_conditional_execute): New
5325         declaration.
5326         (tune_params): Update comment.
5327         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
5328         (arm_max_conditional_execute): New function.
5329         (thumb2_final_prescan_insn): Use max_insn_skipped and
5330         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
5332 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
5334         PR tree-optimization/57705
5335         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
5336         SSA_NAME step, provided that it is not defined inside the loop.
5337         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
5338         (get_initial_def_for_induction): Handle SSA_NAME IV step.
5340 2013-06-25  Martin Jambor  <mjambor@suse.cz>
5342         PR middle-end/57670
5343         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
5344         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
5345         calls in the dump.
5346         (ipa_note_param_call): Initialize member_ptr flag.
5347         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
5348         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
5349         (ipa_write_indirect_edge_info): Stream member_ptr flag.
5350         (ipa_read_indirect_edge_info): Likewise.
5352 2013-06-25  Richard Biener  <rguenther@suse.de>
5354         PR middle-end/56977
5355         * passes.c (init_optimization_passes): Move pass_fold_builtins
5356         and pass_dce earlier with -Og.
5358 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
5360         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
5361         <BIT_FIELD_REF>: Remove trailing TAB.
5362         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
5363         remove blank line.
5365 2013-06-24  Martin Jambor  <mjambor@suse.cz>
5367         PR tree-optimization/57358
5368         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
5369         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
5370         (ipa_analyze_params_uses): Generate pessimistic info when true.
5372 2013-06-24  Martin Jambor  <mjambor@suse.cz>
5374         PR tree-optimization/57539
5375         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
5376         global.inlined_to of the new node to it.  All callers changed.
5377         * ipa-inline-transform.c (clone_inlined_nodes): New variable
5378         inlining_into, pass it to cgraph_clone_node.
5379         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
5380         ipa_free_edge_args_substructures.
5381         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
5382         rdesc linked list.  Do not assert rdesc edges have inlined caller.
5383         Assert we have found an rdesc in the rdesc list.
5385 2013-06-24  Richard Biener  <rguenther@suse.de>
5387         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
5388         (pointer_set_lookup): Declare.
5389         (class pointer_map): New template class implementing a
5390         generic pointer to T map.
5391         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
5392         pointer_map<T>::contains, pointer_map<T>::insert,
5393         pointer_map<T>::traverse): New functions.
5394         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
5395         (pointer_set_lookup): New function.
5396         (pointer_set_contains): Use pointer_set_lookup.
5397         (pointer_set_insert): Likewise.
5398         (insert_aux): Remove.
5399         (struct pointer_map_t): Embed a pointer_set_t.
5400         (pointer_map_create): Adjust.
5401         (pointer_map_destroy): Likewise.
5402         (pointer_map_contains): Likewise.
5403         (pointer_map_insert): Likewise.
5404         (pointer_map_traverse): Likewise.
5405         * tree-streamer.h (struct streamer_tree_cache_d): Use a
5406         pointer_map<unsigned> instead of a pointer_map_t.
5407         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
5408         (streamer_tree_cache_lookup): Likewise.
5409         (streamer_tree_cache_create): Likewise.
5410         (streamer_tree_cache_delete): Likewise.
5411         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
5412         pointer_map<unsigned> instead of a pointer_map_t.
5413         (lto_init_tree_ref_encoder): Adjust.
5414         (lto_destroy_tree_ref_encoder): Likewise.
5415         * lto-section-out.c (lto_output_decl_index): Likewise.
5416         (lto_record_function_out_decl_state): Likewise.
5417         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
5419 2013-06-24  Richard Biener  <rguenther@suse.de>
5421         PR tree-optimization/57488
5422         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
5424 2013-06-24  Alan Modra  <amodra@gmail.com>
5426         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
5427         (gen_easy_altivec_constant): Likewise.
5428         * config/rs6000/predicates.md (easy_vector_constant_add_self,
5429         easy_vector_constant_msb): Likewise.
5431 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
5433         PR target/57688
5434         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
5435         add missing return true.
5437 2013-06-23  Oleg Endo <olegendo@gcc.gnu.org>
5439         PR target/52483
5440         * config/sh/predicates.md (general_extend_operand): Invoke
5441         general_movsrc_operand for memory operands.
5442         (general_movsrc_operand): Allow reg+reg addressing, do not use
5443         general_operand for memory operands.
5445 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
5447         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
5448         when current target options does not apply.
5449         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
5450         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
5451         * config/i386/bmiintrin.h: Pass appropriate target
5452         attributes to header.
5453         * config/i386/mmintrin.h: Ditto.
5454         * config/i386/nmmintrin.h: Ditto.
5455         * config/i386/avx2intrin.h: Ditto.
5456         * config/i386/fxsrintrin.h: Ditto.
5457         * config/i386/tbmintrin.h: Ditto.
5458         * config/i386/xsaveintrin.h: Ditto.
5459         * config/i386/f16cintrin.h: Ditto.
5460         * config/i386/xtestintrin.h: Ditto.
5461         * config/i386/xsaveoptintrin.h: Ditto.
5462         * config/i386/bmi2intrin.h: Ditto.
5463         * config/i386/lzcntintrin.h: Ditto.
5464         * config/i386/smmintrin.h: Ditto.
5465         * config/i386/wmmintrin.h: Ditto.
5466         * config/i386/x86intrin.h: Remove all header include guards.
5467         * config/i386/prfchwintrin.h: Ditto.
5468         * config/i386/pmmintrin.h: Ditto.
5469         * config/i386/tmmintrin.h: Ditto.
5470         * config/i386/xmmintrin.h: Ditto.
5471         * config/i386/popcntintrin.h: Ditto.
5472         * config/i386/rdseedintrin.h: Ditto.
5473         * config/i386/ammintrin.h: Ditto.
5474         * config/i386/emmintrin.h: Ditto.
5475         * config/i386/immintrin.h: Remove all header include guards.
5476         * config/i386/fma4intrin.h: Ditto.
5477         * config/i386/lwpintrin.h: Ditto.
5478         * config/i386/xopintrin.h: Ditto.
5479         * config/i386/ia32intrin.h: Ditto.
5480         * config/i386/avxintrin.h: Ditto.
5481         * config/i386/rtmintrin.h: Ditto.
5482         * config/i386/fmaintrin.h: Ditto.
5483         * config/i386/mm3dnow.h: Ditto.
5485 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
5487         * common/config/i386/i386-common.c: Handle LZCNT.
5489 2013-06-22  Andi Kleen  <ak@linux.intel.com>
5491         * doc/extend.texi: Use __atomic_store_n instead of
5492         __atomic_store in HLE example.
5494 2013-06-22  Oleg Endo <olegendo@gcc.gnu.org>
5496         * config/sh/sh.c: Remove <cstdlib> workaround.
5498 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5500         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
5502 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5504         * doc/extend.texi: Document that __atomic_clear and
5505           __atomic_test_and_set should only be used with bool.
5507 2013-06-20  Jan Hubicka  <jh@suse.cz>
5509         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
5510         types_same_for_odr.
5511         * tree.c (decls_same_for_odr): New function.
5512         (same_for_edr): New function.
5513         (types_same_for_odr): New function.
5514         (get_binfo_at_offset): Use it.
5515         * tree.h (types_same_for_odr): Declare.
5517 2013-06-20  Oleg Endo <olegendo@gcc.gnu.org>
5518             Jason Merrill  <jason@redhat.com>
5520         * system.h: Include <cstdlib> as well as <stdlib.h>.
5522 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
5524         PR target/57655
5525         * config/i386/i386.c (construct_container): Report error if
5526         long double is used with disabled x87 float returns.
5528 2013-06-20  Jan Hubicka  <jh@suse.cz>
5530         * lto-cgraph.c (input_symtab): Do not set cgraph state.
5532 2013-06-20  Joern Rennecke <joern.rennecke@embecosm.com>
5534         PR rtl-optimization/57425
5535         PR rtl-optimization/57569
5536         * alias.c (write_dependence_p): Remove parameters mem_mode and
5537         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
5538         Changed all callers.
5539         (canon_anti_dependence): Get comments and semantics in sync.
5540         Add parameter mem_canonicalized.  Changed all callers.
5541         * rtl.h (canon_anti_dependence): Update prototype.
5543 2013-06-20  Richard Biener  <rguenther@suse.de>
5545         * data-streamer-in.c (streamer_read_uhwi): Optimize single
5546         byte case, inline streamer_read_uchar and defer section
5547         overrun check.
5549 2013-06-20  Richard Biener  <rguenther@suse.de>
5551         PR tree-optimization/57584
5552         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
5553         SSA names into the expanded expression that take part in
5554         abnormal coalescing.
5556 2013-06-19  Sharad Singhai  <singhai@google.com>
5558         * gcov.c (print_usage): Handle new option.
5559         (process_args): Ditto.
5560         (get_gcov_intermediate_filename): New function.
5561         (output_intermediate_file): New function.
5562         (output_gcov_file): New function
5563         (generate_results): Handle new option.
5564         (release_function): Relase demangled name.
5565         (read_graph_file): Handle demangled name.
5566         (output_lines): Ditto.
5567         * doc/gcov.texi: Document gcov intermediate format.
5569 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
5571         PR bootstrap/57604
5572         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
5573         (lra_emit_add): Use the functions.  Add comment about Y as an
5574         address segment.
5576 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
5578         PR driver/57652
5579         * collect2.c (collect_atexit): New.
5580         (collect_exit): Delete.
5581         (main): Register collect_atexit with atexit.
5582         (collect_wait): Change collect_exit to exit.
5583         (do_wait): Same.
5584         * collect2.h (collect_exit): Delete.
5585         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
5587 2013-06-19  Wei Mi  <wmi@google.com>
5589         PR rtl-optimization/57518
5590         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
5591         if regno is used in paradoxical subreg.
5592         (update_equiv_regs): Check pdx_subregs[regno] before
5593         set a reg to be equivalent with a mem.
5595 2013-06-19  Matthias Klose  <doko@ubuntu.com>
5597         PR driver/57651
5598         * file-find.h (find_a_file): Add a mode parameter.
5599         * file-find.c (find_a_file): Likewise.
5600         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
5601         with X_OK for the executables.
5602         * collect2.c (main): Call find_a_file with X_OK.
5604 2013-06-19  Steve Ellcey  <sellcey@mips.com>
5606         PR target/56942
5607         * config/mips/mips.md (casesi_internal_mips16_<mode>):
5608         Use NEXT_INSN instead of next_real_insn.
5610 2013-06-19  Jan Hubicka  <jh@suse.cz>
5612         * cgraph.h (const_value_known_p): Replace by ...
5613         (ctor_for_folding): .. this one.
5614         * cgraphunit.c (process_function_and_variable_attributes): Use it.
5615         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
5616         * expr.c (expand_expr_real_1): Likewise.
5617         (string_constant): Likewise.
5618         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
5619         * ipa.c (process_references): Likewise.
5620         (symtab_remove_unreachable_nodes): Likewise.
5621         * ipa-inline-analysis.c (param_change_prob): Likewise.
5622         * gimple-fold.c (canonicalize_constructor_val): Likewise.
5623         (get_base_constructor): Likwise.
5624         * varpool.c (varpool_remove_node): Likewise.
5625         (varpool_remove_initializer): LIkewise.
5626         (dump_varpool_node): LIkwise.
5627         (const_value_known_p): Rewrite to ...
5628         (ctor_for_folding): ... this one.
5630 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
5632         PR driver/57651
5633         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
5634         PERSONALITY in $PATH derived prefixes.
5636 2013-06-19  Jeff Law  <law@redhat.com>
5638         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
5639         in comment.
5641         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
5642         (simplify_bitwise_binary): Use it to simpify certain binary ops on
5643         booleans.
5645 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
5647         * config/arm/vfp.md: Move VFP instruction classification documentation
5648         to ...
5649         * config/arm/arm.md: ... here.  Update instruction classification
5650         documentation.
5652 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
5654         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
5655         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
5656         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
5658 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
5660         PR target/57609
5661         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5662         with NEXT_INSN.  Use tablejump_p to check for jump table data
5663         insns.
5665 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
5667         PR c++/56544
5668         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
5669         that now in C++ the value is correct per the C++ standards.
5671 2013-06-19  Richard Biener  <rguenther@suse.de>
5673         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
5674         for global context.
5676 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5678         Revert:
5679         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5681         PR target/57609
5682         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5683         with next_active_insn.
5685 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
5687         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
5688         functions are inlined during failures to flag an error.
5689         * tree-inline.c (expand_call_inline): Allow the error to be flagged
5690         in early inline pass.
5692 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
5694         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
5695         in comments.
5697 2013-06-18  Julian Brown  <julian@codesourcery.com>
5699         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
5700         Permit virtual register pre-reload if !strict.
5701         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
5702         change.
5703         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
5704         prototype.
5705         * config/arm/neon.md (movmisalign<mode>): Use
5706         neon_perm_struct_or_reg_operand instead of
5707         neon_struct_or_register_operand.
5708         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
5709         neon_permissive_struct_operand instead of neon_struct_operand.
5710         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
5711         neon_vector_mem_operand.
5712         * config/arm/predicates.md (neon_struct_operand): Adjust call to
5713         neon_vector_mem_operand.
5714         (neon_permissive_struct_operand): New.
5715         (neon_struct_or_register_operand): Rename to...
5716         (neon_perm_struct_or_reg_operand): This. Adjust call to
5717         neon_vector_mem_operand.
5719 2013-06-18  Richard Biener  <rguenther@suse.de>
5721         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
5722         * lto-streamer.h: Include pointer-set.h.
5723         (struct lto_decl_slot): Remove.
5724         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
5725         Remove next_index entry.
5726         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5727         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5728         (lto_init_tree_ref_encoder): Adjust.
5729         (lto_destroy_tree_ref_encoder): Likewise.
5730         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5731         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5732         (lto_output_decl_index): Adjust.
5733         (lto_new_out_decl_state): Likewise.
5734         (lto_record_function_out_decl_state): Likewise.
5735         * lto-streamer-out.c (copy_function): Likewise.
5737 2013-06-18  Richard Biener  <rguenther@suse.de>
5739         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
5740         * cgraphunit.c: Include cfgloop.h.
5741         (init_lowered_empty_function): Initialize the loop tree.
5742         (assemble_thunk): Insert new BBs into loops.
5744 2013-06-18  Richard Biener  <rguenther@suse.de>
5746         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
5747         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
5748         the map from cache entry to cache index optional.
5749         (streamer_tree_cache_replace_tree): Adjust accordingly.
5750         (streamer_tree_cache_append): Likewise.
5751         (streamer_tree_cache_delete): Likewise.
5752         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
5753         streamer cache map from cache entry to cache index.
5754         * lto-streamer-out.c (create_output_block): Adjust.
5756 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
5758         * config/arm/arm.md (attribute "insn"): Move multiplication and
5759         division attributes to...
5760         (attribute "type"): ... here.  Remove mult.
5761         (attribute "mul32"): New attribute.
5762         (attribute "mul64"): Add umaal.
5763         (*arm_mulsi3): Update attributes.
5764         (*arm_mulsi3_v6): Likewise.
5765         (*thumb_mulsi3): Likewise.
5766         (*thumb_mulsi3_v6): Likewise.
5767         (*mulsi3_compare0): Likewise.
5768         (*mulsi3_compare0_v6): Likewise.
5769         (*mulsi_compare0_scratch): Likewise.
5770         (*mulsi_compare0_scratch_v6): Likewise.
5771         (*mulsi3addsi): Likewise.
5772         (*mulsi3addsi_v6): Likewise.
5773         (*mulsi3addsi_compare0): Likewise.
5774         (*mulsi3addsi_compare0_v6): Likewise.
5775         (*mulsi3addsi_compare0_scratch): Likewise.
5776         (*mulsi3addsi_compare0_scratch_v6): Likewise.
5777         (*mulsi3subsi): Likewise.
5778         (*mulsidi3adddi): Likewise.
5779         (*mulsi3addsi_v6): Likewise.
5780         (*mulsidi3adddi_v6): Likewise.
5781         (*mulsidi3_nov6): Likewise.
5782         (*mulsidi3_v6): Likewise.
5783         (*umulsidi3_nov6): Likewise.
5784         (*umulsidi3_v6): Likewise.
5785         (*umulsidi3adddi): Likewise.
5786         (*umulsidi3adddi_v6): Likewise.
5787         (*smulsi3_highpart_nov6): Likewise.
5788         (*smulsi3_highpart_v6): Likewise.
5789         (*umulsi3_highpart_nov6): Likewise.
5790         (*umulsi3_highpart_v6): Likewise.
5791         (mulhisi3): Likewise.
5792         (*mulhisi3tb): Likewise.
5793         (*mulhisi3bt): Likewise.
5794         (*mulhisi3tt): Likewise.
5795         (maddhisi4): Likewise.
5796         (*maddhisi4tb): Likewise.
5797         (*maddhisi4tt): Likewise.
5798         (maddhidi4): Likewise.
5799         (*maddhidi4tb): Likewise.
5800         (*maddhidi4tt): Likewise.
5801         (divsi3): Likewise.
5802         (udivsi3): Likewise.
5803         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
5804         (thumb2_mulsi_short_compare0): Likewise.
5805         (thumb2_mulsi_short_compare0_scratch): Likewise.
5806         * config/arm/arm1020e.md (1020mult1): Update attribute change.
5807         (1020mult2): Likewise.
5808         (1020mult3): Likewise.
5809         (1020mult4): Likewise.
5810         (1020mult5): Likewise.
5811         (1020mult6): Likewise.
5812         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
5813         change.
5814         (cortex_a15_mult64): Likewise.
5815         (cortex_a15_sdiv): Likewise.
5816         (cortex_a15_udiv): Likewise.
5817         * config/arm/arm1026ejs.md (mult1): Update attribute change.
5818         (mult2): Likewise.
5819         (mult3): Likewise.
5820         (mult4): Likewise.
5821         (mult5): Likewise.
5822         (mult6): Likewise.
5823         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
5824         (pj4_ir_div): Likewise.
5825         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
5826         (11_mult2): Likewise.
5827         (11_mult3): Likewise.
5828         (11_mult4): Likewise.
5829         (11_mult5): Likewise.
5830         (11_mult6): Likewise.
5831         (11_mult7): Likewise.
5832         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
5833         (cortex_a8_mla): Likewise.
5834         (cortex_a8_mull): Likewise.
5835         (cortex_a8_smulwy): Likewise.
5836         (cortex_a8_smlald): Likewise.
5837         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
5838         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
5839         (cortex_r4_mul_3): Likewise.
5840         (cortex_r4_mla_4): Likewise.
5841         (cortex_r4_mla_3): Likewise.
5842         (cortex_r4_smlald): Likewise.
5843         (cortex_r4_mull): Likewise.
5844         (cortex_r4_sdiv): Likewise.
5845         (cortex_r4_udiv): Likewise.
5846         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
5847         (cortex_a7_idiv): Likewise.
5848         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
5849         (9_mult2): Likewise.
5850         (9_mult3): Likewise.
5851         (9_mult4): Likewise.
5852         (9_mult5): Likewise.
5853         (9_mult6): Likewise.
5854         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
5855         (cortex_a53_sdiv): Likewise.
5856         (cortex_a53_udiv): Likewise.
5857         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5858         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5859         (mp626_mult2): Likewise.
5860         (mp626_mult3): Likewise.
5861         (mp626_mult4): Likewise.
5862         * config/arm/fa526.md (526_mult1): Update attribute change.
5863         (526_mult2): Likewise.
5864         * config/arm/arm-generic.md (mult): Update attribute change.
5865         (mult_ldsched_strongarm): Likewise.
5866         (mult_ldsched): Likewise.
5867         (multi_cycle): Likewise.
5868         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5869         * config/arm/fa606te.md (606te_mult1): Update attribute change.
5870         (606te_mult2): Likewise.
5871         (606te_mult3): Likewise.
5872         (606te_mult4): Likewise.
5873         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5874         (cortex_a9_mac16): Likewise.
5875         (cortex_a9_multiply): Likewise.
5876         (cortex_a9_mac): Likewise.
5877         (cortex_a9_multiply_long): Likewise.
5878         * config/arm/fa626te.md (626te_mult1): Update attribute change.
5879         (626te_mult2): Likewise.
5880         (626te_mult3): Likewise.
5881         (626te_mult4): Likewise.
5883 2013-06-18  Richard Biener  <rguenther@suse.de>
5885         PR lto/57334
5886         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5888 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5890         PR target/57609
5891         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5892         with next_active_insn.
5894 2013-06-18  Alan Modra  <amodra@gmail.com>
5896         * config/rs6000/rs6000.h (enum data_align): New.
5897         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5898         (DATA_ABI_ALIGNMENT): Define.
5899         (CONSTANT_ALIGNMENT): Correct comment.
5900         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5901         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5903 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
5905         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5906         ATTRIBUTE_UNUSED marking.
5908 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
5910         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5911         alternative and update.
5912         (aarch64_dup_lanedi): Delete.
5913         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5914         * config/aarch64/aarch64-simd-builtins.def: Update.
5916 2013-06-17  Richard Biener  <rguenther@suse.de>
5918         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5919         (lto_input_scc): Declare.
5920         (lto_input_tree_1): Likewise.
5921         (struct lto_stats_d): Add num_tree_bodies_output and
5922         num_pickle_refs_output.
5923         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5924         (lto_read_tree_1): Split out from ...
5925         (lto_read_tree): ... this.
5926         (lto_input_scc): New function.
5927         (lto_input_tree_1): Split out from ...
5928         (lto_input_tree): ... this.  Handle LTO_tree_scc.
5929         (lto_data_in_create): Create the streamer cache without hashes.
5930         * lto-streamer-out.c (create_output_block): Create the streamer
5931         cache with hashes when not doing WPA.
5932         (lto_write_tree_1): Split out from ...
5933         (lto_write_tree): ... this.
5934         (get_symbol_initial_value): New function.
5935         (lto_output_tree_1): Split out from ...
5936         (lto_output_tree): ... this.  Write trees as series of SCCs
5937         using a DFS walk via DFS_write_tree.
5938         (struct sccs, struct scc_entry): New types.
5939         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
5940         (DFS_write_tree_body): New function.
5941         (DFS_write_tree): Likewise.
5942         (hash_tree): Likewise.
5943         (scc_entry_compare): Likewise.
5944         (hash_scc): Likewise.
5945         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
5946         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
5947         TREE_CHAIN as regular reference.
5948         (streamer_read_integer_cst): Remove.
5949         (streamer_get_pickled_tree): Adjust.
5950         * tree-streamer-out.c (streamer_write_chain): Disable streaming
5951         of DECL_EXTERNALs in BLOCK_VARS for now.
5952         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
5953         reference.
5954         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
5955         Add hash value argument and record that if hashes are recorded
5956         in the cache.
5957         (streamer_tree_cache_insert_1): Adjust.
5958         (streamer_tree_cache_insert): Likewise.
5959         (streamer_tree_cache_insert_at): Rename to ...
5960         (streamer_tree_cache_replace_tree): ... this and adjust.
5961         (streamer_tree_cache_append): Adjust.
5962         (record_common_node): Likewise.
5963         (streamer_tree_cache_create): Add argument whether to
5964         record hash values together with trees.
5965         (streamer_tree_cache_delete): Adjust.
5966         * tree-streamer.h (struct streamer_tree_cache_d): Add
5967         vector of hashes.
5968         (streamer_read_integer_cst): Remove.
5969         (streamer_tree_cache_insert): Adjust.
5970         (streamer_tree_cache_append): Likewise.
5971         (streamer_tree_cache_insert_at): Rename to ...
5972         (streamer_tree_cache_replace_tree): ... this and adjust.
5973         (streamer_tree_cache_create): Add argument whether to record hashes.
5974         (streamer_tree_cache_get): Rename to ...
5975         (streamer_tree_cache_get_tree): ... this.
5976         (streamer_tree_cache_get_hash): New function.
5977         * tree.c (cache_integer_cst): New function.
5978         * tree.h (cache_integer_cst): Declare.
5979         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
5980         * lto-symtab.c (lto_varpool_replace_node): Only release
5981         DECL_INITIAL of non-prevailing decls.
5982         * varpool.c (varpool_remove_initializer): Do not release
5983         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
5985 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
5987         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
5988         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
5989         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
5990         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
5991         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
5992         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
5993         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
5994         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
5995         instead of TARGET_64BIT.
5996         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
5997         Require ISA_HAS_<D>DIV.
5999 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
6001         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6002         (mips*-*-linux*): Move default with_llsc setting to where other
6003         defaults are set.
6004         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6005         with_arch block.
6006         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6007         Likewise.  Remove default with_tune setting.  Move default float
6008         setting to its own block.  Handle with_llsc in the same block as above.
6010 2013-06-16  Joern Rennecke <joern.rennecke@embecosm.com>
6012         PR rtl-optimization/57425
6013         PR rtl-optimization/57569
6014         * alias.c (write_dependence_p): Add new parameters mem_mode,
6015         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
6016         Changed all callers.
6017         (canon_anti_dependence): New function.
6018         * cse.c (check_dependence): Use canon_anti_dependence.
6019         * cselib.c (cselib_invalidate_mem): Likewise.
6020         * rtl.h (canon_anti_dependence): Declare.
6022 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6024         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6025         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6026         ".set mips3" for 64-bit targets.
6028 2013-06-15  Dehao Chen  <dehao@google.com>
6030         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6031         * gimple-low.c (gimple_check_call_matching_types): Likewise.
6032         (gimple_check_call_args): Likewise.
6033         * value-prof.c (check_ic_target): Likewise.
6034         * ipa-inline.c (early_inliner): Likewise.
6035         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6036         * cgraph.c (cgraph_create_edge_1): Likewise.
6037         (cgraph_make_edge_direct): Likewise.
6039 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
6041         PR target/57615
6042         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6043         rs6000_output_move_128bit to handle emitting quad memory
6044         operations.  Set attribute length to 8 bytes.
6046 2013-06-14  Vidya Praveen <vidyapraveen@arm.com>
6048         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6049         New pattern.
6050         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6051         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6052         (aarch64_<su>mlsl<mode>): Likewise.
6054 2013-06-14  Mike Stump  <mikestump@comcast.net>
6056         * Makefile.in (TARGET_H): Add insn-codes.h.
6058 2013-06-14  Alan Modra  <amodra@gmail.com>
6060         PR middle-end/57134
6061         PR middle-end/57586
6062         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6063         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
6064         bitfield expansion when EXPAND_MEMORY.
6065         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6067 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
6069         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6070         test for clearing quad memory on 32-bit later.
6072 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6074         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6075         (fold_negate_expr): Likewise.
6076         (fold_real_zero_addition_p): Handle vectors.
6077         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6079 2013-06-14  Alan Modra  <amodra@gmail.com>
6081         * varasm.c (force_const_mem): Revert 2013-06-07 change.
6083 2013-06-13  Jan Hubicka  <jh@suse.cz>
6085         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6086         Local comdats are not externally visible.
6087         * symtab.c (dump_symtab_base): Dump externally visible.
6088         (verify_symtab_base): Verify back links in the symtab hash.
6090 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6092         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6093         CONVERT_EXPR as equal nodes.
6095 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6097         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6099 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6101         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6102         Generalize to complex and vector.
6103         * tree.c (build_all_ones_cst): New function.
6104         * tree.h (build_all_ones_cst): Declare it.
6106 2013-06-13  Alan Modra  <amodra@gmail.com>
6108         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6109         * config/rs6000/rs6000.md (signbittf2): New insn.
6110         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6111         (abstf2_internal, cmptf_internal2): Likewise.
6112         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6114 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
6115             Pat Haugen <pthaugen@us.ibm.com>
6116             Peter Bergner <bergner@vnet.ibm.com>
6118         * config/rs6000/rs6000.c (emit_load_locked): Add support for
6119         power8 byte, half-word, and quad-word atomic instructions.
6120         (emit_store_conditional): Likewise.
6121         (rs6000_expand_atomic_compare_and_swap): Likewise.
6122         (rs6000_expand_atomic_op): Likewise.
6124         * config/rs6000/sync.md (larx): Add new modes for power8.
6125         (stcx): Likewise.
6126         (AINT): New mode iterator to include TImode as well as normal
6127         integer modes on power8.
6128         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
6129         that VSX registers are not considered.  Use AINT mode iterator
6130         instead of INT1 to allow inclusion of quad word atomic operations
6131         on power8.
6132         (load_locked<mode>): Likewise.
6133         (store_conditional<mode>): Likewise.
6134         (atomic_compare_and_swap<mode>): Likewise.
6135         (atomic_exchange<mode>): Likewise.
6136         (atomic_nand<mode>): Likewise.
6137         (atomic_fetch_<fetchop_name><mode>): Likewise.
6138         (atomic_nand_fetch<mode>): Likewise.
6139         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
6140         each type.
6141         (ATOMIC): On power8, add QImode, HImode modes.
6142         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
6143         modes that promote to SImode.
6144         (load_lockedti): Convert TImode arguments to PTImode, so that we
6145         get a guaranteed even/odd register pair.
6146         (load_lockedpti): Likewise.
6147         (store_conditionalti): Likewise.
6148         (store_conditionalpti): Likewise.
6150         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
6151         atomic load/store instructions.
6152         (HSI): Likewise.
6154 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
6156         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
6157         loads.
6158         (insn_count): New attribute, with most cases extracted from...
6159         (length): ...here.  Redefine most cases in terms of insn_count.
6160         (single_insn): Delete.
6161         (can_delay): Use insn_count to check for single instructions.
6162         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
6163         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
6164         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
6165         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
6166         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
6167         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
6168         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
6169         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
6170         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
6171         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
6172         rather than "length".
6173         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
6174         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
6175         Use "insn_count" rather than "length".
6176         * config/mips/mips-dsp.md
6177         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
6178         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
6179         length attributes.
6181 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
6183         PR tree-optimization/57361
6184         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
6186 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
6188         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
6189         to split.
6190         (aarch64_simd_combine<mode>): New instruction expansion.
6191         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
6192         function prototype.
6193         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
6194         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
6196 2013-06-12  Jan Hubicka  <jh@suse.cz>
6198         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
6199         decl has when in streaming stage.
6200         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
6201         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
6203 2013-06-12  Roland Stigge <stigge@antcom.de>
6205         PR target/57578
6206         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
6208 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6210         PR tree-optimization/57537
6211         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
6212         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
6214 2013-06-12  Richard Biener  <rguenther@suse.de>
6216         * data-streamer.h (streamer_write_char_stream): CSE
6217         obs->current_pointer.
6218         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
6219         streamer_write_char_stream manually and optimize the resulting loop.
6220         (streamer_write_hwi_stream): Likewise.
6222 2013-06-12  Jan Hubicka  <jh@suse.cz>
6224         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
6225         * cgraph.h (varpool_create_empty_node): Declare.
6226         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
6227         duplicated nodes.
6228         * symtab.c (symtab_unregister_node): Be lax about missin entries
6229         in node hash.
6230         (symtab_get_node): Update comment.
6231         * varpool.c (varpool_create_empty_node): Break out from ...
6232         (varpool_node_for_decl): ... here.
6233         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
6235 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
6237         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
6238         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
6239         part.  Use straight-line flow at the end.
6240         <COMPONENT_REF>: Remove superfluous else.
6241         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
6243 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6245         PR target/56564
6246         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
6247         target hook even for !TREE_PUBLIC decls.  If no resolution info
6248         is available, return false for common and external decls.
6250 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6252         * config/rl78/constraints.md (U): New constraint.
6253         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
6254         valloc attribute.
6256 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
6258         PR target/57589
6259         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
6260         to allow returning address to AT_PLATFORM name.
6262 2013-06-11  Jan Hubicka  <jh@suse.cz>
6264         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
6265         * cgraph.h (symtab_node_base): Add weakref flag.
6266         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
6267         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
6268         (output_weakrefs): Use weakref flag.
6269         * fold-const.c (simple_operand_p): Handle WEAK.
6270         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
6271         * ipa.c (varpool_externally_visible_p): Drop weakref.
6272         (function_and_variable_visibility): Update comment; fix weakref
6273         sanity checks; do not clear DECL_WEAK on them.
6274         * lto-cgraph.c (lto_output_node): update.
6275         (lto_output_varpool_node): Update.
6276         (input_overwrite_node): Update.
6277         (input_node): Update.
6278         (input_varpool_node): Update.
6279         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
6280         (lto_symtab_merge_symbols): Add sanity check.
6281         (lto_symtab_prevailing_decl): Do not special case weakrefs.
6282         * passes.c (rest_of_decl_compilation): Set static flag, too.
6283         * symtab.c (dump_symtab_base): Dump weakref.
6284         (verify_symtab_base): Sanity check weakrefs.
6285         (symtab_make_decl_local): Remove duplicated code.
6286         (symtab_alias_ultimate_target): Simplify.
6287         * varpool.c (varpool_create_variable_alias): Set weakref flag.
6289 2013-06-11  DJ Delorie  <dj@redhat.com>
6291         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
6292         (rl78_unwind_word_mode): New.
6294 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
6296         * final.c (debug_prefix_maps): Make static.
6298 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
6300         * function.c (initial_trampoline): Remove stray copy.
6302 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
6304         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
6306 2013-06-11  Martin Jambor  <mjambor@suse.cz>
6308         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
6309         within bounds at the beginning of the function.
6311 2013-06-11  Alan Modra  <amodra@gmail.com>
6313         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
6314         reporting.
6315         (get_named_section): Don't NULL !DECL_P decl.
6317 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
6319         * doc/invoke.texi (core-avx2): Document.
6320         (slm): Likewise.
6321         (atom): Updated with MOVBE.
6323 2013-06-11  Richard Biener  <rguenther@suse.de>
6325         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
6327 2013-06-11  Anton Blanchard  <anton@samba.org>
6329         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
6330         correct shift value in little-endian mode.
6332 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
6334         PR target/56564
6335         * varasm.c (get_variable_align): Move #endif to the right place.
6337 2013-06-10  Cary Coutant  <ccoutant@google.com>
6339         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
6340         for hash so that hash table traversal order is deterministic.
6342 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
6343             Pat Haugen <pthaugen@us.ibm.com>
6344             Peter Bergner <bergner@vnet.ibm.com>
6346         * config/rs6000/vector.md (GPR move splitter): Do not split moves
6347         of vectors in GPRS if they are direct moves or quad word load or
6348         store moves.
6350         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
6351         declaration.
6352         (direct_move_p): Likewise.
6353         (quad_load_store_p): Likewise.
6355         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
6356         classes into bins based on the physical register type.
6357         (reg_class_to_reg_type): Likewise.
6358         (IS_STD_REG_TYPE): Likewise.
6359         (IS_FP_VECT_REG_TYPE): Likewise.
6360         (reload_fpr_gpr): Arrays to determine what insn to use if we can
6361         use direct move instructions.
6362         (reload_gpr_vsx): Likewise.
6363         (reload_vsx_gpr): Likewise.
6364         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
6365         information that is a simplification of register classes.  Also
6366         precalculate direct move reload helpers.
6367         (direct_move_p): New function to return true if the operation can
6368         be done as a direct move instruciton.
6369         (quad_load_store_p): New function to return true if the operation
6370         is a quad memory operation.
6371         (rs6000_legitimize_address): If quad memory, only allow register
6372         indirect for TImode addresses.
6373         (rs6000_legitimate_address_p): Likewise.
6374         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
6375         (rs6000_reload_register_type): Likewise.
6376         (register_to_reg_type): Return register type.
6377         (rs6000_secondary_reload_simple_move): New helper function for
6378         secondary reload and secondary memory needed to identify anything
6379         that is a simple move, and does not need reloading.
6380         (rs6000_secondary_reload_direct_move): New helper function for
6381         secondary reload to identify cases that can be done with several
6382         instructions via the direct move instructions.
6383         (rs6000_secondary_reload_move): New helper function for secondary
6384         reload to identify moves between register types that can be done.
6385         (rs6000_secondary_reload): Add support for quad memory operations
6386         and for direct move.
6387         (rs6000_secondary_memory_needed): Likewise.
6388         (rs6000_debug_secondary_memory_needed): Change argument names.
6389         (rs6000_output_move_128bit): New function to return the move to
6390         use for 128-bit moves, including knowing about the various
6391         limitations of quad memory operations.
6393         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
6394         memory operations.  call rs6000_output_move_128bit for the actual
6395         instruciton(s) to generate.
6396         (vsx_movti_64bit): Likewise.
6398         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
6399         (UNSPEC_P8V_MTVSRWZ): Likewise.
6400         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
6401         (UNSPEC_P8V_MTVSRD): Likewise.
6402         (UNSPEC_P8V_XXPERMDI): Likewise.
6403         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
6404         (UNSPEC_FUSION_GPR): Likewise.
6405         (FMOVE128_GPR): New iterator for direct move.
6406         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
6407         (f32_sv): Likewise.
6408         (f32_dm): Likewise.
6409         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
6410         loads and direct move instructions.
6411         (zero_extendsidi2_lfiwzx): Likewise.
6412         (extendsidi2_lfiwax): Likewise.
6413         (extendsidi2_nocell): Likewise.
6414         (floatsi<mode>2_lfiwax): Likewise.
6415         (lfiwax): Likewise.
6416         (floatunssi<mode>2_lfiwzx): Likewise.
6417         (lfiwzx): Likewise.
6418         (fix_trunc<mode>_stfiwx): Likewise.
6419         (fixuns_trunc<mode>_stfiwx): Likewise.
6420         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
6421         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
6422         (parity<mode>2_cmpb): Set length/type attr.
6423         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
6424         for 'mr.' to fast_compare.
6425         (bpermd_<mode>): Change type attr to popcnt.
6426         (p8_fmrgow_<mode>): New insns for power8 direct move support.
6427         (p8_mtvsrwz_1): Likewise.
6428         (p8_mtvsrwz_2): Likewise.
6429         (reload_fpr_from_gpr<mode>): Likewise.
6430         (p8_mtvsrd_1): Likewise.
6431         (p8_mtvsrd_2): Likewise.
6432         (p8_xxpermdi_<mode>): Likewise.
6433         (reload_vsx_from_gpr<mode>): Likewise.
6434         (reload_vsx_from_gprsf): Likewise.
6435         (p8_mfvsrd_3_<mode>): LIkewise.
6436         (reload_gpr_from_vsx<mode>): Likewise.
6437         (reload_gpr_from_vsxsf): Likewise.
6438         (p8_mfvsrd_4_disf): Likewise.
6439         (multi-word GPR splits): Do not split direct moves or quad memory
6440         operations.
6442 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
6444         * tree-into-ssa.c (interesting_blocks): Make static.
6446 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
6448         PR target/56564
6449         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
6450         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
6451         Use DATA_ABI_ALIGNMENT for that case instead if defined.
6452         (get_variable_align): New function.
6453         (get_variable_section, emit_bss, emit_common,
6454         assemble_variable_contents, place_block_symbol): Use
6455         get_variable_align instead of DECL_ALIGN.
6456         (assemble_noswitch_variable): Add align argument, use it
6457         instead of DECL_ALIGN.
6458         (assemble_variable): Adjust caller.  Use get_variable_align
6459         instead of DECL_ALIGN.
6460         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
6461         caller.
6462         (DATA_ABI_ALIGNMENT): Define.
6463         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
6464         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
6465         opt is false, only return the psABI mandated alignment increase.
6466         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
6467         (DATA_ABI_ALIGNMENT): ... this.
6468         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
6469         (DATA_ABI_ALIGNMENT): ... this.
6470         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
6471         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
6472         (DATA_ABI_ALIGNMENT): ... this.
6473         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
6474         * doc/tm.texi: Regenerated.
6476 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
6478         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
6479         cmp_code to construct REG_EQUAL note.
6481 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
6483         PR target/57568
6484         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
6485         that operands[2] doesn't overlap with operands[0].
6487 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
6488             Jan Hubicka  <jh@suse.cz>
6490         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
6491         hack to mark symbols as used.
6493 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
6495         PR rtl-optimization/57559
6496         * lra-constraints.c (process_alt_operands): Don't discourage
6497         memory with known offset for offsetable memory constraint.
6498         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
6500 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
6502         * varasm.c (struct oc_local_state): Reorder fields.
6503         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
6504         and adjust accordingly.
6505         (output_constructor): Reorder initialization code and adjust call to
6506         output_constructor_bitfield.
6508 2013-06-07  Jan Hubicka  <jh@suse.cz>
6510         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
6512 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
6514         * tree-object-size.c (unknown): Make const.
6516 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6518         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
6519         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
6520         for last alternative in the cpu_facility attribute.
6522 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6524         PR target/56315
6525         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
6526         (xordi3): Change operand 2 constraint to arm_xordi_operand.
6527         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
6528         * config/arm/constraints.md (Dg): New constraint.
6529         * config/arm/neon.md (xordi3_neon): Remove.
6530         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
6531         * config/arm/predicates.md (arm_xordi_operand): New predicate.
6533 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6535         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
6536         Clean up alternatives.
6538 2013-06-07  Alan Modra  <amodra@gmail.com>
6540         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
6541         va_list_gpr_size.
6543 2013-06-07  Alan Modra  <amodra@gmail.com>
6545         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
6547 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6549         * config/arm/constraints.md (Df): New constraint.
6550         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
6551         Correct length attribute for last two alternatives.
6553 2013-06-07  Alan Modra  <amodra@gmail.com>
6555         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6556         override user -mfp-in-toc.
6557         (offsettable_ok_by_alignment): Consider just the current access
6558         rather than the whole object, unless BLKmode.  Handle
6559         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
6560         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
6561         for -mcmodel=medium.
6562         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
6563         override user -mfp-in-toc or -msum-in-toc.  Default to
6564         -mno-fp-in-toc for -mcmodel=medium.
6566 2013-06-06  DJ Delorie  <dj@redhat.com>
6568         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
6569         TARGET_VALID_POINTER_MODE.
6571 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
6572             Pat Haugen <pthaugen@us.ibm.com>
6573             Peter Bergner <bergner@vnet.ibm.com>
6575         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6576         Document new power8 builtins.
6578         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
6579         condition code register, to allow 128-bit logical operations to be
6580         done in the VSX or GPR registers.
6581         (nor<mode>3): Use the canonical form for nor.
6582         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
6583         vclz*, and vpopcnt* vector instructions.
6584         (nand<mode>3): Likewise.
6585         (orc<mode>3): Likewise.
6586         (clz<mode>2): LIkewise.
6587         (popcount<mode>2): Likewise.
6589         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
6590         that only the GPRs are recognized.
6592         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6593         support for new power8 builtins.
6595         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
6596         builtin functions.
6597         (xscvdpspn): Likewise.
6598         (vclz): Likewise.
6599         (vclzb): Likewise.
6600         (vclzh): Likewise.
6601         (vclzw): Likewise.
6602         (vclzd): Likewise.
6603         (vpopcnt): Likewise.
6604         (vpopcntb): Likewise.
6605         (vpopcnth): Likewise.
6606         (vpopcntw): Likewise.
6607         (vpopcntd): Likewise.
6608         (vgbbd): Likewise.
6609         (vmrgew): Likewise.
6610         (vmrgow): Likewise.
6611         (eqv): Likewise.
6612         (eqv_v16qi3): Likewise.
6613         (eqv_v8hi3): Likewise.
6614         (eqv_v4si3): Likewise.
6615         (eqv_v2di3): Likewise.
6616         (eqv_v4sf3): Likewise.
6617         (eqv_v2df3): Likewise.
6618         (nand): Likewise.
6619         (nand_v16qi3): Likewise.
6620         (nand_v8hi3): Likewise.
6621         (nand_v4si3): Likewise.
6622         (nand_v2di3): Likewise.
6623         (nand_v4sf3): Likewise.
6624         (nand_v2df3): Likewise.
6625         (orc): Likewise.
6626         (orc_v16qi3): Likewise.
6627         (orc_v8hi3): Likewise.
6628         (orc_v4si3): Likewise.
6629         (orc_v2di3): Likewise.
6630         (orc_v4sf3): Likewise.
6631         (orc_v2df3): Likewise.
6633         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
6634         allow power8 quad mode in 64-bit.
6635         (rs6000_builtin_vectorized_function): Add support to vectorize
6636         ISA 2.07 count leading zeros, population count builtins.
6637         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
6638         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
6639         (builtin_function_type): Add vgbbd builtin function which takes an
6640         unsigned argument.
6641         (altivec_expand_vec_perm_const): Add support for new power8 merge
6642         instructions.
6644         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
6645         that does not include TImdoe for use with 32-bit.
6646         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
6647         instructions.
6648         (UNSPEC_VSX_CVDPSPN): Likewise.
6649         (vsx_xscvdpspn): Likewise.
6650         (vsx_xscvspdpn): Likewise.
6651         (vsx_xscvdpspn_scalar): Likewise.
6652         (vsx_xscvspdpn_directmove): Likewise.
6653         (vsx_and<mode>3): Split logical operations into 32-bit and
6654         64-bit. Add support to do logical operations on TImode as well as
6655         VSX vector types.  Allow logical operations to be done in either
6656         VSX registers or in general purpose registers in 64-bit mode.  Add
6657         splitters if GPRs were used. For AND, add clobber of CCmode to
6658         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
6659         encoding.
6660         (vsx_and<mode>3_32bit): Likewise.
6661         (vsx_and<mode>3_64bit): Likewise.
6662         (vsx_ior<mode>3): Likewise.
6663         (vsx_ior<mode>3_32bit): Likewise.
6664         (vsx_ior<mode>3_64bit): Likewise.
6665         (vsx_xor<mode>3): Likewise.
6666         (vsx_xor<mode>3_32bit): Likewise.
6667         (vsx_xor<mode>3_64bit): Likewise.
6668         (vsx_one_cmpl<mode>2): Likewise.
6669         (vsx_one_cmpl<mode>2_32bit): Likewise.
6670         (vsx_one_cmpl<mode>2_64bit): Likewise.
6671         (vsx_nor<mode>3): Likewise.
6672         (vsx_nor<mode>3_32bit): Likewise.
6673         (vsx_nor<mode>3_64bit): Likewise.
6674         (vsx_andc<mode>3): Likewise.
6675         (vsx_andc<mode>3_32bit): Likewise.
6676         (vsx_andc<mode>3_64bit): Likewise.
6677         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
6678         and xxlorc instructions.
6679         (vsx_eqv<mode>3_64bit): Likewise.
6680         (vsx_nand<mode>3_32bit): Likewise.
6681         (vsx_nand<mode>3_64bit): Likewise.
6682         (vsx_orc<mode>3_32bit): Likewise.
6683         (vsx_orc<mode>3_64bit): Likewise.
6685         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
6687         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
6688         instruction.
6689         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
6690         (p8_vmrgow): Likewise.
6691         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
6692         GPRs to be split under VSX.
6693         (p8v_clz<mode>2): Add power8 count leading zero support.
6694         (p8v_popcount<mode>2): Add power8 population count support.
6695         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
6696         support.
6698         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
6699         instruction.
6701         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
6702         builtin functions.
6703         (vec_nand): Likewise.
6704         (vec_vclz): Likewise.
6705         (vec_vclzb): Likewise.
6706         (vec_vclzd): Likewise.
6707         (vec_vclzh): Likewise.
6708         (vec_vclzw): Likewise.
6709         (vec_vgbbd): Likewise.
6710         (vec_vmrgew): Likewise.
6711         (vec_vmrgow): Likewise.
6712         (vec_vpopcnt): Likewise.
6713         (vec_vpopcntb): Likewise.
6714         (vec_vpopcntd): Likewise.
6715         (vec_vpopcnth): Likewise.
6716         (vec_vpopcntw): Likewise.
6718 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6720         PR rtl-optimization/57468
6721         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
6722         spilled pseudos.
6724 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6726         PR rtl-optimization/57459
6727         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
6728         type when setting live regs.
6730 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6732         * config/s390/s390.opt (mlra): New option.
6733         * config/s390/s390.c (s390_decompose_address): Check displacement
6734         for all registers for LRA.
6735         (s390_secondary_reload): Don't used secondary reloads for LRA.
6736         (s390_lra_p): New function.
6737         (TARGET_LRA_P): Define.
6738         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
6739         of attribute cpu_facility to zarch for the last alternative.
6740         (*cmpmem_short): Ditto.
6742 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
6744         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
6745         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
6746         (arm_expand_prologue): Likewise.
6748 2013-06-06  Teresa Johnson  <tejohnson@google.com>
6750         PR c++/53743
6751         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
6752         as this is now done by redirect_edge_and_branch_force.
6753         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
6754         barriers, and fix interaction with splitting.
6755         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
6756         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
6757         reflect changes made in the routine.
6758         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
6759         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
6760         since this is called in cfglayout mode, and replace partition fixup
6761         with assert as that is now done by force_nonfallthru_and_redirect.
6762         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
6763         already be marked with region crossing note.
6764         (insert_section_boundary_note): Make non-static, gate on flag
6765         has_bb_partition, rewrite to also check for multiple partitions.
6766         (rest_of_handle_reorder_blocks): Remove call to
6767         insert_section_boundary_note, now done later during free_cfg.
6768         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
6769         * bb-reorder.h (insert_section_boundary_note): Declare.
6770         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
6771         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
6772         invoke insert_section_boundary_note.
6773         (try_redirect_by_replacing_jump): Remove unnecessary
6774         check for region crossing note.
6775         (fixup_partition_crossing): New function.
6776         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
6777         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
6778         in non-cfglayout mode.
6779         (force_nonfallthru_and_redirect): Fixup partition boundaries,
6780         remove old code that tried to do this. Emit barrier correctly
6781         when we are in cfglayout mode.
6782         (last_bb_in_partition): New function.
6783         (rtl_split_edge): Correctly fixup partition boundaries.
6784         (commit_one_edge_insertion): Remove old code that tried to
6785         fixup region crossing edge since this is now handled in
6786         split_block, and set up insertion point correctly since
6787         block may now end in a jump.
6788         (verify_hot_cold_block_grouping): Guard against checking when not in
6789         linearized RTL mode.
6790         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
6791         notes.
6792         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
6793         rtl_verify_flow_info, so not called in cfglayout mode.
6794         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
6795         (fixup_reorder_chain): Remove old code that attempted to fixup region
6796         crossing note as this is now handled in force_nonfallthru_and_redirect.
6797         (duplicate_insn_chain): Don't duplicate switch section notes.
6798         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
6799         note.
6800         * basic-block.h (emit_barrier_after_bb): Declare.
6802 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6804         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
6805         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
6806         arm_usatsihi): Adjust alternatives for arm_restrict_it.
6808 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6810         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
6811         where appropriate.
6812         * config/arm/ldmstm.md: Regenerate.
6814 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6816         * config/arm/sync.md (atomic_loaddi_1):
6817         Disable predication for arm_restrict_it.
6818         (arm_load_exclusive<mode>): Likewise.
6819         (arm_load_exclusivesi): Likewise.
6820         (arm_load_exclusivedi): Likewise.
6821         (arm_load_acquire_exclusive<mode>): Likewise.
6822         (arm_load_acquire_exclusivesi): Likewise.
6823         (arm_load_acquire_exclusivedi): Likewise.
6824         (arm_store_exclusive<mode>): Likewise.
6825         (arm_store_exclusive<mode>): Likewise.
6826         (arm_store_release_exclusivedi): Likewise.
6827         (arm_store_release_exclusive<mode>): Likewise.
6829 2013-06-06  Richard Biener  <rguenther@suse.de>
6831         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
6832         after LTO_null.
6833         (lto_tag_is_tree_code_p): Adjust.
6834         (lto_tag_is_gimple_code_p): Likewise.
6835         (lto_gimple_code_to_tag): Likewise.
6836         (lto_tag_to_gimple_code): Likewise.
6837         (lto_tree_code_to_tag): Likewise.
6838         (lto_tag_to_tree_code): Likewise.
6839         * data-streamer.h (streamer_write_hwi_in_range): Use
6840         uhwi streaming to stream the normalized range.
6841         (streamer_read_hwi_in_range): Likewise.
6843 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6845         * config/arm/arm.md (enabled_for_depr_it): New attribute.
6846         (predicable_short_it): Likewise.
6847         (predicated): Likewise.
6848         (enabled): Handle above.
6849         (define_cond_exec): Set predicated attribute to yes.
6851 2013-06-05  Mike Stump  <mikestump@comcast.net>
6853         * gdbinit.in (__FUNCTION__): Add.
6855 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
6857         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6858         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6860 2013-06-05  Jan Hubicka  <jh@suse.cz>
6862         * varasm.c (mark_decl_referenced): Revert the removal until targets
6863         are fixed.
6865 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
6867         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6868         instead of mark_decl_referenced.
6870 2013-06-05  Jan Hubicka  <jh@suse.cz>
6872         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6873         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6874         and symtab_used_from_object_file_p.
6875         (cgraph_make_node_local_1): Clear forced_by_abi.
6876         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6877         * cgraph.h (symtab_node_base): Add forced_by_abi.
6878         (decide_is_variable_needed): Remove.
6879         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6880         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6881         (decide_is_symbol_needed): ... this one; handle symbols in general;
6882         always analyze virtuals; honnor forced_by_abi.
6883         (cgraph_finalize_function): Update.
6884         (varpool_finalize_decl): Update.
6885         (symbol_defined_and_needed): Remove.
6886         (analyze_functions): Update.
6887         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6888         output_refs, input_overwrite_node): Handle forced_by_abi.
6889         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6890         (address_taken_from_non_vtable_p): ... this one.
6891         (comdat_can_be_unshared_p_1): New function.
6892         (cgraph_comdat_can_be_unshared_p): Rename to ...
6893         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6894         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6895         (function_and_variable_visibility): Clear forced_by_abi as needed.
6896         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6897         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6898         * symtab.c (dump_symtab_base): Dump forced_by_abi.
6899         * varpool.c (decide_is_variable_needed): Remove.
6901 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6903         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6904         (arm_option_override): Override arm_restrict_it where appropriate.
6905         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6906         * config/arm/arm.opt (mrestrict-it): New command-line option.
6907         * doc/invoke.texi: Document -mrestrict-it.
6909 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6911         * tsan.c (tsan_atomic_table): Make const.
6913 2013-06-05  Richard Biener  <rguenther@suse.de>
6915         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6916         index associated with the tree we are supposed to replace.
6917         * tree-streamer-out.c (pack_ts_base_value_fields): Output
6918         TREE_ASM_WRITTEN as zero for everything but SSA names.
6920 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6922         * tree-ssa-structalias.c (call_stmt_vars): Make static.
6924 2013-06-04  Jan Hubicka  <jh@suse.cz>
6926         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6927         (input_node, input_varpool_node): Handle correctly external same
6928         body aliases.
6929         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6930         nodes at ltrans stage.
6932 2013-06-04  Jan Hubicka  <jh@suse.cz>
6934         * ipa-inline.c (update_caller_keys): Fix availability test.
6935         (update_callee_keys): Likewise.
6936         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6937         to follow ELF standard.
6939 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
6941         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
6942         (mips64r5900el-*-elf*): New configurations.
6943         * config/mips/mips-cpus.def (r5900): New processor.
6944         * config/mips/mips-tables.opt: Regenerate.
6945         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
6946         (mips_issue_rate): Handle PROCESSOR_R5900.
6947         (mips_reorg_process_insns): Force reorder mode for the R5900.
6948         * config/mips/mips.h (TARGET_MIPS5900): Define.
6949         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
6950         TARGET_MIPS5900.
6951         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
6952         TARGET_MIPS5900.
6953         * config/mips/mips.md (processor): Add r5900.
6954         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
6956 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6958         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
6959         into function to generate MOVI instruction.
6960         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
6961         (aarch64_preferred_simd_mode): Turn into wrapper.
6962         (aarch64_output_scalar_simd_mov_immediate): New function.
6963         * config/aarch64/aarch64-protos.h: Add prototype for above.
6965 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6967         * config/aarch64/aarch64.c (simd_immediate_info): Remove
6968         element_char member.
6969         (sizetochar): Return signed char.
6970         (aarch64_simd_valid_immediate): Remove elchar and other
6971         unnecessary variables.
6972         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
6973         Calculate element_char as required.
6974         * config/aarch64/aarch64-protos.h: Update and move prototype
6975         for aarch64_output_simd_mov_immediate.
6976         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6977         Update arguments.
6979 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6981         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
6982         information completed by aarch64_simd_valid_immediate.
6983         (aarch64_legitimate_constant_p): Update arguments.
6984         (aarch64_simd_valid_immediate): Work with struct rather than many
6985         pointers.
6986         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
6987         (aarch64_simd_make_constant): Update arguments.
6988         (aarch64_output_simd_mov_immediate): Work with struct rather than
6989         many pointers.  Output immediate directly rather than as operand.
6990         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6991         Update prototype.
6992         * config/aarch64/constraints.md (Dn): Update arguments.
6994 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
6996         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
6997         longer static.
6998         (aarch64_simd_immediate_valid_for_move): Remove.
6999         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7000         (aarch64_simd_make_constant): Update call.
7001         (aarch64_output_simd_mov_immediate): Update call.
7002         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7003         Add prototype.
7004         * config/aarch64/constraints.md (Dn): Update call.
7006 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7008         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7009         return type to bool for prototype.
7010         (aarch64_legitimate_constant_p): Check for true instead of not -1.
7011         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7012         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7014 2013-06-04  Catherine Moore  <clm@codesourcery.com>
7016         * config/mips/mips.opt (meva): New.
7017         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7018         (ASM_SPEC): Handle -meva.
7019         * doc/invoke.texi (meva):  Document.
7021 2013-06-04  Alan Modra  <amodra@gmail.com>
7023         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7024         constant output.
7026 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7028         * rtl.def: Add extra fourth optional field to define_cond_exec.
7029         * gensupport.c (process_one_cond_exec): Process attributes from
7030         define_cond_exec.
7031         * doc/md.texi: Document fourth field in define_cond_exec.
7033 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
7035         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7036         out the processing order as in store_bit_field_1.
7038 2013-06-04  Jan Hubicka  <jh@suse.cz>
7040         PR middle-end/57500
7041         * cgraphunit.c (cgraph_process_same_body_aliases): Create
7042         non-VAR_DECL node if it does not exist yet.
7044 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
7046         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7047         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7048         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7049         target_cpu_default setting.
7051 2013-06-03  Teresa Johnson  <tejohnson@google.com>
7053         * dumpfile.c (opt_info_switch_p): Change -fopt-info
7054         default to -fopt-info=optimized instead of all.
7055         * doc/invoke.texi: Ditto.
7056         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7057         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7058         (execute_vect_slp): Emit BB vectorization success under
7059         MSG_OPTIMIZED_LOCATIONS.
7060         * tree-vect-slp.c (vect_slp_transform_bb): Change
7061         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7062         * tree-vect-loop.c (vect_transform_loop): Ditto.
7064 2013-06-03  Jason Merrill  <jason@redhat.com>
7066         PR c++/57415
7067         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7068         Use TARGET_EXPR for C++.
7070 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
7072         PR rtl-optimization/57268
7073         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7074         if DEBUG_INSN_P (insn).
7076         Reapply
7077         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7079         PR rtl-optimization/57268
7080         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7081         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7083 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7085         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7086         (ix86_avoid_lea_for_addr): Likewise.
7087         (exact_dependency_1): Likewise.
7088         (ix86_adjust_cost): Likewise.
7089         (swap_top_of_ready_list): Fix formatting and !reload_completed check
7090         removed.
7091         (do_reorder_for_imul): Fix typo, formatting and
7092         !reload_completed check removed.
7093         (ix86_sched_reorder): Fix typo and formatting.
7094         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7095         list.
7097 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
7099         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7101 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
7103         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7104         <STRING_CST>: Likewise.
7105         <VECTOR_CST>: Likewise.
7107 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
7108             Mikael Morin  <mikael@gcc.gnu.org>
7110         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7111         * config.in: Regenerated.
7112         * configure: Regenerated.
7114 2013-06-01  Jan Hubicka  <jh@suse.cz>
7116         PR middle-end/57366
7117         * cgraphunit.c (compile): When weakref is not supported,
7118         set up transparent aliases before final output pass.
7119         * varasm.c (assemble_alias): Do not try to do it here.
7121 2013-06-01  Jan Hubicka  <jh@suse.cz>
7123         PR middle-end/57467
7124         * passes.c (for_per_function): Skip unanalyzed functions.
7126 2013-06-01  Jan Hubicka  <jh@suse.cz>
7128         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
7129         (lto_symtab_merge_symbols_1): ... this one.
7130         (lto_symtab_merge_cgraph_nodes): Rename to ...
7131         (lto_symtab_merge_symbols): ... this one; simplify.
7132         * cgraph.c (same_body_aliases_done): Rename to ...
7133         (cpp_implicit_aliases_done): ... this one.
7134         (cgraph_create_function_alias): Update.
7135         (cgraph_same_body_alias): Update.
7136         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
7137         (verify_edge_corresponds_to_fndecl): Simplify.
7138         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
7139         (cgraph_node): Remove same_body_alias.
7140         (varpool_node): Remove alias_of and extra_name_alias.
7141         (same_body_aliases_done): Rename to ..
7142         (cpp_implicit_aliases_done): ... this one.
7143         (symtab_alias_ultimate_target): Add default parameter.
7144         (symtab_resolve_alias): New function.
7145         (fixup_same_cpp_alias_visibility): Declare.
7146         (cgraph_function_node): Add default parameter.
7147         (cgraph_node_asm_name): Likewise.
7148         (cgraph_function_or_thunk_node): Add default parameter; do
7149         not ICE when it is NULL.
7150         (varpool_variable_node): Likewise.
7151         * tree-emutls.c (create_emultls_var): Update.
7152         (ipa_lower_emutls): Update.
7153         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
7154         (cgraph_reset_node): Reset alias info.
7155         (cgraph_finalize_function): Update.
7156         (fixup_same_cpp_alias_visibility): Move to symtab.c.
7157         (analyze_function): Simplify.
7158         (cgraph_process_same_body_aliases): Simplify.
7159         (analyze_functions): Fixup same body aliases.
7160         (handle_alias_pairs): Simplify.
7161         (assemble_thunk): Update.
7162         (assemble_thunks_and_aliases): Update.
7163         (output_weakrefs): Rewrite.
7164         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
7165         (lto_output_varpool_node): Likewise.
7166         (compute_ltrans_boundary): Remve assert.
7167         (get_alias_symbol): New functoin.
7168         (input_node): Rewrite alias handling.
7169         (input_varpool_node): Likewise.
7170         * ipa-pure-const.c (propagate_pure_const): Fix formating.
7171         * ipa.c (process_references): Handle weakrefs correctly.
7172         (symtab_remove_unreachable_nodes): Likewise.
7173         * trans-mem.c (get_cg_data): Update.
7174         (ipa_tm_create_version_alias): Update.
7175         (ipa_tm_execute): Update.
7176         * symtab.c (dump_symtab_base): Dump aliases.
7177         (verify_symtab_base): Verify aliases.
7178         (symtab_node_availability): New function.
7179         (symtab_alias_ultimate_target): Simplify.
7180         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
7181         handle all the fixup cases.
7182         (symtab_resolve_alias): New function.
7183         * passes.c (ipa_write_summaries): Handle weakrefs.
7184         * varpool.c (varpool_analyze_node): Simplify.
7185         (assemble_aliases): Update.
7186         (varpool_create_variable_alias): Simplify.
7187         (varpool_extra_name_alias): Simplify.
7188         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
7189         (lto_symtab_merge_symbols): ... this one.
7191 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
7193         Revert
7194         PR rtl-optimization/57268
7195         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7196         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7198 2013-06-01  Tobias Burnus  <burnus@net-b.de>
7200         Partially reverted:
7201         2013-05-31  Tobias Burnus  <burnus@net-b.de>
7203         PR middle-end/57073
7204         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
7205         further up.
7207 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7209         PR rtl-optimization/57268
7210         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7211         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7213 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7215         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
7216         unordered comparison operators when -fno-trapping-math is in effect
7217         on the e500.
7218         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
7219         and implement unordered comparison operators properly on the e500.
7221 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7223         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
7224         for constant scalar integers.
7225         (simplify_relational_operation_1): Likewise.
7227 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
7229         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
7230         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
7231         Fix comment.
7233 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7234             Igor Zamyatin  <igor.zamyatin@intel.com>
7236         Silvermont (SLM) architecture performance tuning.
7237         * config/i386/i386.h (enum ix86_tune_indices): Add
7238         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
7239         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
7241         * config/i386/i386.c (initial_ix86_tune_features)
7242         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
7243         (ix86_lea_outperforms): Handle Silvermont tuning.
7244         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
7245         call.
7246         (ix86_use_lea_for_mov): Likewise.
7247         (ix86_avoid_lea_for_addr): Likewise.
7248         (ix86_lea_for_add_ok): Likewise.
7249         (exact_dependency_1): New function.
7250         (exact_store_load_dependency): Likewise.
7251         (ix86_adjust_cost): Handle Silvermont tuning.
7252         (do_reoder_for_imul): Likewise.
7253         (swap_top_of_ready_list): New function.
7254         (ix86_sched_reorder): Changed to handle Silvermont tuning.
7256         * config/i386/i386.md (peepholes that split memory operand in fp
7257         converts): New.
7259 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7261         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7262         Remove un-necessary braces.
7264 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7266         * config/aarch64/aarch64.c (aarch64_classify_symbol):
7267         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
7269 2013-05-31  Tobias Burnus  <burnus@net-b.de>
7271         PR middle-end/57073
7272         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
7274 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7276         PR target/56315
7277         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
7278         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
7279         * config/arm/neon.md (iordi3_neon): Remove.
7280         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
7281         * config/arm/predicates.md (imm_for_neon_logic_operand):
7282         Move to earlier in the file.
7283         (neon_logic_op2): Likewise.
7284         (arm_iordi_operand_neon): New predicate.
7286 2013-05-31  Richard Biener  <rguenther@suse.de>
7288         PR tree-optimization/57478
7289         PR tree-optimization/57453
7290         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
7291         are life as well.
7293 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
7295         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
7296         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
7298 2013-05-30  Tobias Burnus  <burnus@net-b.de>
7299             Thomas Koenig  <tkoenig@gcc.gnu.org>
7301         PR middle-end/57073
7302         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
7303         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
7305 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
7307         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
7309 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
7311         * target.def (register_usage_leveling_p): New hook.
7312         * targhooks.c (default_register_usage_leveling_p): New.
7313         * targhooks.h (default_register_usage_leveling_p): New prototype.
7314         * lra-assigns.c (register_usage_leveling_p): Use the hook.
7315         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
7316         * doc/tm.texi: Update.
7317         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
7319 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
7321         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
7322         (*insv_reg<mode>): New define_insn.
7324 2013-05-30  Joern Rennecke <joern.rennecke@embecosm.com>
7326         PR rtl-optimization/57439
7327         * postreload.c (move2add_valid_value_p): Check that we have
7328         a zero subreg_regno_offset when accessing the register in
7329         the requested mode.
7331 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7332             Igor Zamyatin  <igor.zamyatin@intel.com>
7334         Silvermont (SLM) architecture pipeline model, tuning and
7335         insn selection.
7336         * config.gcc: Add slm config options and target.
7338         * config/i386/slm.md: New.
7340         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
7342         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
7343         PROCESSOR_SLM.
7344         (ix86_target_macros_internal): Likewise.
7346         * gcc/config/i386/i386.c (slm_cost): New cost.
7347         (m_SLM): New macro flag.
7348         (initial_ix86_tune_features): Set m_SLM.
7349         (x86_accumulate_outgoing_args): Likewise.
7350         (x86_arch_always_fancy_math_387): Likewise.
7351         (processor_target_table): Add slm cost.
7352         (cpu_names): Add slm cpu name.
7353         (x86_option_override_internal): Set SLM ISA.
7354         (ix86_issue_rate): New case PROCESSOR_SLM.
7355         (ia32_multipass_dfa_lookahead): Likewise.
7356         (fold_builtin_cpu): Add slm.
7358         * config/i386/i386.h (TARGET_SLM): New target macro.
7359         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
7360         (processor_type): Add PROCESSOR_SLM.
7362         * config/i386/i386.md (cpu): Add new value "slm".
7363         (slm.md): Include slm.md.
7365 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
7366             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7368         * config/arm/arm-protos.h: Add and update function protos.
7369         * config/arm/arm.c (use_simple_return_p): New added.
7370         (thumb2_expand_return): Check simple_return flag.
7371         * config/arm/arm.md: Add simple_return and conditional simple_return.
7372         * config/arm/iterators.md: Add iterator for return and simple_return.
7374 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7376         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
7377         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
7378         (arm_emit_vfp_multi_reg_pop): Likewise.
7379         (thumb2_emit_ldrd_pop): Likewise.
7380         (arm_expand_epilogue): Add misc REG_CFA notes.
7381         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
7383 2013-05-29  Lawrence Crowl  <crowl@google.com>
7385         * config/arm/t-arm: Update for below.
7387         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
7388         Change type to hash_table.  Update dependent calls and types.
7390         * config/i386/t-cygming: Update for below.
7392         * config/i386/t-interix: Update for below.
7394         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
7395         Change type to hash_table.  Update dependent calls and types.
7396         (i386_find_on_wrapper_list::wrappers): Likewise.
7398         * config/ia64/t-ia64: Update for below.
7400         * config/ia64/ia64.c (bundle_state_table):
7401         Change type to hash_table.  Update dependent calls and types.
7403         * config/mips/mips.c (mips_reorg_process_insns::htab):
7404         Change type to hash_table.  Update dependent calls and types.
7406         * config/sol2.c (solaris_comdat_htab):
7407         Change type to hash_table.  Update dependent calls and types.
7409         * config/t-sol2: Update for above.
7411 2013-05-29  Teresa Johnson  <tejohnson@google.com>
7413         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
7414         functions are not yet marked as defined.
7416 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
7417             Pat Haugen <pthaugen@us.ibm.com>
7418             Peter Bergner <bergner@vnet.ibm.com>
7420         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
7421         instructions.
7422         (VEC_A): Likewise.
7423         (VEC_C): Likewise.
7424         (vrotl<mode>3): Likewise.
7425         (vashl<mode>3): Likewise.
7426         (vlshr<mode>3): Likewise.
7427         (vashr<mode>3): Likewise.
7429         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7430         support for power8 V2DI builtins.
7432         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
7433         power8 V2DI builtins.
7434         (vupkhsw): Likewise.
7435         (vupklsw): Likewise.
7436         (vaddudm): Likewise.
7437         (vminsd): Likewise.
7438         (vmaxsd): Likewise.
7439         (vminud): Likewise.
7440         (vmaxud): Likewise.
7441         (vpkudum): Likewise.
7442         (vpksdss): Likewise.
7443         (vpkudus): Likewise.
7444         (vpksdus): Likewise.
7445         (vrld): Likewise.
7446         (vsld): Likewise.
7447         (vsrd): Likewise.
7448         (vsrad): Likewise.
7449         (vsubudm): Likewise.
7450         (vcmpequd): Likewise.
7451         (vcmpgtsd): Likewise.
7452         (vcmpgtud): Likewise.
7453         (vcmpequd_p): Likewise.
7454         (vcmpgtsd_p): Likewise.
7455         (vcmpgtud_p): Likewise.
7456         (vupkhsw): Likewise.
7457         (vupklsw): Likewise.
7458         (vaddudm): Likewise.
7459         (vmaxsd): Likewise.
7460         (vmaxud): Likewise.
7461         (vminsd): Likewise.
7462         (vminud): Likewise.
7463         (vpksdss): Likewise.
7464         (vpksdus): Likewise.
7465         (vpkudum): Likewise.
7466         (vpkudus): Likewise.
7467         (vrld): Likewise.
7468         (vsld): Likewise.
7469         (vsrad): Likewise.
7470         (vsrd): Likewise.
7471         (vsubudm): Likewise.
7473         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
7474         support for power8 V2DI instructions.
7476         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
7477         power8 V2DI instructions.  Combine pack and unpack insns to use an
7478         iterator for each mode.  Check whether a particular mode supports
7479         Altivec instructions instead of just checking TARGET_ALTIVEC.
7480         (UNSPEC_VPKUWUM): Likewise.
7481         (UNSPEC_VPKSHSS): Likewise.
7482         (UNSPEC_VPKSWSS): Likewise.
7483         (UNSPEC_VPKUHUS): Likewise.
7484         (UNSPEC_VPKSHUS): Likewise.
7485         (UNSPEC_VPKUWUS): Likewise.
7486         (UNSPEC_VPKSWUS): Likewise.
7487         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
7488         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
7489         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
7490         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
7491         (UNSPEC_VUPKHSB): Likewise.
7492         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
7493         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
7494         (UNSPEC_VUPKHSH): Likewise.
7495         (UNSPEC_VUPKLSB): Likewise.
7496         (UNSPEC_VUPKLSH): Likewise.
7497         (VI2): Likewise.
7498         (VI_char): Likewise.
7499         (VI_scalar): Likewise.
7500         (VI_unit): Likewise.
7501         (VP): Likewise.
7502         (VP_small): Likewise.
7503         (VP_small_lc): Likewise.
7504         (VU_char): Likewise.
7505         (add<mode>3): Likewise.
7506         (altivec_vaddcuw): Likewise.
7507         (altivec_vaddu<VI_char>s): Likewise.
7508         (altivec_vadds<VI_char>s): Likewise.
7509         (sub<mode>3): Likewise.
7510         (altivec_vsubcuw): Likewise.
7511         (altivec_vsubu<VI_char>s): Likewise.
7512         (altivec_vsubs<VI_char>s): Likewise.
7513         (altivec_vavgs<VI_char>): Likewise.
7514         (altivec_vcmpbfp): Likewise.
7515         (altivec_eq<mode>): Likewise.
7516         (altivec_gt<mode>): Likewise.
7517         (altivec_gtu<mode>): Likewise.
7518         (umax<mode>3): Likewise.
7519         (smax<mode>3): Likewise.
7520         (umin<mode>3): Likewise.
7521         (smin<mode>3): Likewise.
7522         (altivec_vpkuhum): Likewise.
7523         (altivec_vpkuwum): Likewise.
7524         (altivec_vpkshss): Likewise.
7525         (altivec_vpkswss): Likewise.
7526         (altivec_vpkuhus): Likewise.
7527         (altivec_vpkshus): Likewise.
7528         (altivec_vpkuwus): Likewise.
7529         (altivec_vpkswus): Likewise.
7530         (altivec_vpks<VI_char>ss): Likewise.
7531         (altivec_vpks<VI_char>us): Likewise.
7532         (altivec_vpku<VI_char>us): Likewise.
7533         (altivec_vpku<VI_char>um): Likewise.
7534         (altivec_vrl<VI_char>): Likewise.
7535         (altivec_vsl<VI_char>): Likewise.
7536         (altivec_vsr<VI_char>): Likewise.
7537         (altivec_vsra<VI_char>): Likewise.
7538         (altivec_vsldoi_<mode>): Likewise.
7539         (altivec_vupkhsb): Likewise.
7540         (altivec_vupkhs<VU_char>): Likewise.
7541         (altivec_vupkls<VU_char>): Likewise.
7542         (altivec_vupkhsh): Likewise.
7543         (altivec_vupklsb): Likewise.
7544         (altivec_vupklsh): Likewise.
7545         (altivec_vcmpequ<VI_char>_p): Likewise.
7546         (altivec_vcmpgts<VI_char>_p): Likewise.
7547         (altivec_vcmpgtu<VI_char>_p): Likewise.
7548         (abs<mode>2): Likewise.
7549         (vec_unpacks_hi_v16qi): Likewise.
7550         (vec_unpacks_hi_v8hi): Likewise.
7551         (vec_unpacks_lo_v16qi): Likewise.
7552         (vec_unpacks_hi_<VP_small_lc>): Likewise.
7553         (vec_unpacks_lo_v8hi): Likewise.
7554         (vec_unpacks_lo_<VP_small_lc>): Likewise.
7555         (vec_pack_trunc_v8h): Likewise.
7556         (vec_pack_trunc_v4si): Likewise.
7557         (vec_pack_trunc_<mode>): Likewise.
7559         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
7560         V2DI builtins.
7561         (vec_vmaxsd): Likewise.
7562         (vec_vmaxud): Likewise.
7563         (vec_vminsd): Likewise.
7564         (vec_vminud): Likewise.
7565         (vec_vpksdss): Likewise.
7566         (vec_vpksdus): Likewise.
7567         (vec_vpkudum): Likewise.
7568         (vec_vpkudus): Likewise.
7569         (vec_vrld): Likewise.
7570         (vec_vsld): Likewise.
7571         (vec_vsrad): Likewise.
7572         (vec_vsrd): Likewise.
7573         (vec_vsubudm): Likewise.
7574         (vec_vupkhsw): Likewise.
7575         (vec_vupklsw): Likewise.
7577 2013-05-29  Jan Hubicka  <jh@suse.cz>
7579         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
7580         flags; reorder rest of fields in more consistent way.
7581         (varpool_node): Remove analyzed, finalized and alias.
7582         (cgraph_ndoe): Likewise.
7583         (symtab_alias_ultimate_target): New function.
7584         (cgraph_function_node): Move offline.
7585         (cgraph_reset_node): Declare.
7586         (cgraph_comdat_can_be_unshared_p): Remove.
7587         (varpool_remove_initializer): Declare.
7588         (varpool_first_defined_variable, varpool_next_defined_variable
7589         cgraph_first_defined_function, cgraph_next_defined_function): Update.
7590         (cgraph_function_with_gimple_body_p): Update.
7591         (varpool_all_refs_explicit_p): Update.
7592         (symtab_alias_target): New function.
7593         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
7594         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
7595         (cgraph_function_or_thunk_node): Simplify using
7596         symtab_alias_ultimate_target.
7597         (varpool_variable_node): Likewise.
7598         * cgraph.c (cgraph_create_function_alias): Update.
7599         (cgraph_add_thunk): Update.
7600         (cgraph_remove_node): Update.
7601         (dump_cgraph_node): Do not dump removed flags.
7602         (cgraph_function_body_availability): Update.
7603         (cgraph_propagate_frequency): Update.
7604         (verify_cgraph_node): Check sanity of local flag.
7605         (cgraph_function_node): Move here from cgraph.h; revamp for
7606         cgraph_function_or_thunk_node.
7607         * lto-symtab.c (lto_varpool_replace_node): Update.
7608         (lto_symtab_resolve_can_prevail_p): Update.
7609         (lto_symtab_merge_cgraph_nodes): Update.
7610         * ipa-cp.c (determine_versionability, initialize_node_lattices,
7611         propagate_constants_accross_call, devirtualization_time_bonus,
7612         ipcp_propagate_stage): Update.
7613         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
7614         * ipa-inline-transform.c (clone_inlined_nodes,
7615         preserve_function_body_p): Update.
7616         * ipa-reference.c (propagate): Update.
7617         (write_node_summary_p): Update.
7618         * toplev.c (wrapup_global_declaration_2): Update.
7619         * cgraphunit.c (cgraph_analyze_function): Rename to ...
7620         (analyze_function) ... this one.
7621         (cgraph_process_new_functions): Update.
7622         (cgraph_reset_node): Export.
7623         (cgraph_finalize_function): Update.
7624         (cgraph_add_new_function): Update.
7625         (process_function_and_variable_attributes): Update.
7626         (varpool_finalize_decl): Update.
7627         (symbol_finalized): Remove.
7628         (symbol_finalized_and_needed): Rename to ...
7629         (symbol_defined_and_needed): ... update.
7630         (cgraph_analyze_functions): Update.
7631         (handle_alias_pairs): Update.
7632         (mark_functions_to_output): Update.
7633         (assemble_thunk): Update.
7634         (output_in_order): Update.
7635         (output_weakrefs): Update.
7636         (finalize_compilation_unit): Update.
7637         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
7638         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
7639         input_node, input_varpool_node): Update.
7640         * dbxout.c (dbxout_expand_expr): Update.
7641         * cgraphclones.c (cgraph_clone_node): Update.
7642         (cgraph_copy_node_for_versioning): Update.
7643         (cgraph_materialize_clone): Update.
7644         (cgraph_materialize_all_clones): Update.
7645         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
7646         propagate_pure_const, propagate_nothrow): Update.
7647         * lto-streamer-out.c (lto_output, write_symbol): Update.
7648         * ipa-utils.c (ipa_reverse_postorder): Update.
7649         * ipa-inline.c (can_inline_edge_p): Update.
7650         (update_caller_keys, ipa_inline): Update.
7651         * dwarf2out.c (reference_to_unused,
7652         premark_types_used_by_global_vars_helper): Update.
7653         * tree-eh.c (tree_could_trap_p): Update.
7654         * ipa-split.c (consider_split, execute_split_functions): Update.
7655         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
7656         has_addr_references_p): Update; move ahead in file for better
7657         readability.
7658         (process_references): Simplify.
7659         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
7660         bodies are removed.
7661         (cgraph_comdat_can_be_unshared_p): Make static.
7662         (cgraph_externally_visible_p): Update.
7663         (varpool_externally_visible_p): Update.
7664         (function_and_variable_visibility): Update.
7665         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
7666         ipa_tm_mark_force_output_node): Update.
7667         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
7668         estimate_edge_devirt_benefit, inline_generate_summary,
7669         inline_write_summary): Update.
7670         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
7671         * ipa-prop.c (ipa_compute_jump_functions): Update.
7672         (ipa_print_node_params, ipa_prop_read_section,
7673         ipa_update_after_lto_read, read_replacements_section): Update.
7674         * varasm.c (mark_decl_referenced): Update.
7675         (assemble_alias, dump_tm_clone_pairs): Update.
7676         * tree-inline.c (copy_bb): Update.
7677         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
7678         Update.
7679         * symtab.c (dump_symtab_base): Print new flags.
7680         (verify_symtab_base): Verify new flags.
7681         (symtab_alias_ultimate_target): New function.
7682         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
7683         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
7684         Update.
7685         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
7686         Update.
7687         * i386.c (ix86_get_function_versions_dispatcher,
7688         ix86_generate_version_dispatcher_body): Update.
7689         (fold_builtin_cpu): Use varpool_add_new_variable.
7690         * varpool.c (varpool_remove_initializer): Break out from ...
7691         (varpool_remove_node): ... this one.
7692         (dump_varpool_node, varpool_node_for_asm,
7693         cgraph_variable_initializer_availability, varpool_analyze_node,
7694         varpool_assemble_decl, varpool_remove_unreferenced_decls,
7695         varpool_finalize_named_section_flags, varpool_create_variable_alias):
7696         Update.
7698 2013-05-29  Jan Hubicka  <jh@suse.cz>
7700         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
7702 2013-05-29  Easwaran Raman  <eraman@google.com>
7704         PR tree-optimization/57442
7705         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
7706         when control exits the main loop.
7708 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
7710         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
7711         and RX600.
7712         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
7713         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
7714         * rx/t-rx: Add rx100 under multi library matches option for nofpu
7715         option.
7717 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7719         PR tree-optimization/57441
7720         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
7721         Don't limit size of incr_vec to number of candidates.
7723 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
7725         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
7726         and mips16 directories.
7727         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
7728         (MULTILIB_DIRNAMES): Ditto.
7729         (MULTILIB_EXCEPTIONS): Add new exceptions.
7730         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
7731         (MULTILIB_DIRNAMES): Ditto.
7732         (MULTILIB_EXCEPTIONS): Add new exceptions.
7734 2012-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7735             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7737         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
7738         SYMBOL_TINY_ABSOLUTE.
7739         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
7740         SYMBOL_TINY_ABSOLUTE.
7741         (aarch64_expand_mov_immediate): Likewise.
7742         (aarch64_classify_symbol): Likewise.
7743         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
7744         Permit SYMBOL_TINY_ABSOLUTE.
7745         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
7747 2013-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7748             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7750         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
7751         Refactor if/switch.  Replace gcc_assert with if.
7753 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
7755         * config/i386/i386.c (initial_ix86_tune_features): Enable
7756         FP Reassociation for AMD bdver1 and bdver2.
7758 2013-05-29  Martin Jambor  <mjambor@suse.cz>
7760         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
7761         and IMAGPART_EXPR do not occur within other handled_components.
7763 2013-05-29  Richard Biener  <rguenther@suse.de>
7765         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
7766         access on whether the use is in the BB we currently try to
7767         vectorize.
7768         (vect_bb_vectorization_profitable_p): Pass the BB we currently
7769         vectorize to vect_bb_slp_scalar_cost.
7771 2013-05-29  Richard Biener  <rguenther@suse.de>
7773         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
7774         computing scalar cost offsetted by stmts that are kept live
7775         by scalar uses.
7776         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
7777         for computation of scalar cost.
7779 2013-05-28  Steve Ellcey  <sellcey@mips.com>
7781         * config/mips/mips-cpus.def (mips32r2): Change processor type.
7783 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7785         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
7786         array notation built-in reduction functions.
7787         * doc/passes.texi (Passes): Added documentation about changes done
7788         for Cilk Plus.
7789         * doc/invoke.texi (C Dialect Options): Added documentation about
7790         the -fcilkplus flag.
7791         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
7792         (BUILTINS_DEF): Depend on cilkplus.def.
7793         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
7794         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
7795         * cilkplus.def: New file.
7797 2013-05-28  Joern Rennecke <joern.rennecke@embecosm.com>
7799         PR rtl-optimization/57439
7800         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
7802 2013-05-28  Easwaran Raman  <eraman@google.com>
7804         PR tree-optimization/57337
7805         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
7806         (find_insert_point): Correctly identify the insertion point
7807         when two statements with the same UID is compared.
7809 2013-05-28  Richard Biener  <rguenther@suse.de>
7811         PR tree-optimization/56787
7812         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
7813         from the list of data references.
7814         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
7815         clobbers.
7816         (vect_analyze_loop_operations): Likewise.
7817         (vect_transform_loop): Remove clobbers.
7819 2013-05-28  Martin Jambor  <mjambor@suse.cz>
7821         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
7822         and REALPART_EXPRs have scalar type.
7824 2013-05-28  Richard Biener  <rguenther@suse.de>
7826         PR tree-optimization/57411
7827         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
7828         virtual operands.
7829         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
7830         virtual operand propagation.
7832 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7834         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
7835         destination register for bmasksi_vis.
7836         (vector_init_bshuffle): Likewise.
7837         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
7839 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7841         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
7842         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
7843         mode if the instruction isn't available in the original mode.
7844         * config/sparc/sparc.opt (mfix-ut699): New option.
7845         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
7846         (divdf3): Turn into expander.
7847         (divdf3_nofix): New insn.
7848         (divdf3_fix): Likewise.
7849         (divsf3): Disable if -mfix-ut699.
7850         (sqrtdf2): Turn into expander.
7851         (sqrtdf2_nofix): New insn.
7852         (sqrtdf2_fix): Likewise.
7853         (sqrtsf2): Disable if -mfix-ut699.
7855 2013-05-27  Richard Biener  <rguenther@suse.de>
7857         PR middle-end/57412
7858         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7859         block for the new loop.
7861 2013-05-27  Richard Biener  <rguenther@suse.de>
7863         PR tree-optimization/57343
7864         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7865         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7866         (number_of_iterations_cond): Do not build the folded tree.
7868 2013-05-27  Richard Biener  <rguenther@suse.de>
7870         Revert
7871         PR middle-end/57381
7872         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7873         OEP_CONSTANT_ADDRESS_OF retained.
7875         PR tree-optimization/57417
7876         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7877         for unchanged base.
7878         (set_ssa_val_to): Compare addresses using
7879         get_addr_base_and_unit_offset.
7881 2013-05-27  Joern Rennecke <joern.rennecke@embecosm.com>
7883         PR rtl-optimization/56833
7884         * postreload.c (move2add_record_mode): New function.
7885         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7886         (move2add_use_add2_insn): Use move2add_record_sym_value.
7887         (move2add_use_add3_insn): Likewise.
7888         (reload_cse_move2add): Use move2add_valid_value_p and
7889         move2add_record_mode.  Invalidate call-clobbered and REG_INC
7890         affected regs by setting reg_mode to VOIDmode.
7891         (move2add_note_store): Don't pretend the inside of a SUBREG is
7892         the actual destination.  Invalidate single/leading registers by
7893         setting reg_mode to VOIDmode.
7894         Use move2add_record_sym_value, move2add_valid_value_p and
7895         move2add_record_mode.
7897 2013-05-27  Richard Biener  <rguenther@suse.de>
7899         PR tree-optimization/57396
7900         * tree-affine.c (double_int_constant_multiple_p): Properly
7901         return false for val == 0 and div != 0.
7903 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7905         * config/mips/mips.h: Use #elif in preprocessor conditions.
7907 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7909         PR target/53916
7910         * config/mips/constraints.md (kl): New constraint.
7911         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7912         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
7913         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
7914         from LO for MIPS16.
7915         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7917 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7919         PR target/55777
7920         * config/mips/mips.c (mips_can_inline_p): New function.
7921         (TARGET_CAN_INLINE_P): Define.
7923 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
7925         * sched-int.h (ds_t, dw_t): Make unsigned int.
7926         Fix documentation that describes how all the ds_t bits are used.
7927         Reserve the last bit for delayed-branch scheduling.
7928         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7929         (BITS_PER_DEP_WEAK): Fix definition and documentation.
7930         (gen_dep_weak_1): Remove prototype.
7931         * sched-deps.c (get_dep_weak_1): Make static.
7932         * target.def (speculate_insn, needs_block_p, gen_spec_check,
7933         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7934         * doc/tm.texi: Regenerate.
7935         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7937 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
7939         PR debug/56950
7940         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
7942 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
7943             Sandra Loosemore  <sandra@codesourcery.com>
7945         * config.gcc (powerpc-*): Allow native for with-cpu.
7947 2013-05-24  Jeff Law  <law@redhat.com>
7949         PR tree-optimization/57124
7950         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
7951         conversion feeding a condition if the range has an overflow
7952         if -fstrict-overflow.  Add warnings for when we do make the
7953         transformation.
7955 2013-05-24  Dehao Chen  <dehao@google.com>
7957         * tree-cfg.c (locus_discrim_map): Fix the typo.
7958         (locus_discrim_hasher): Likewise.
7959         (locus_discrim_hasher::hash): Likewise.
7960         (locus_discrim_hasher::equal): Likewise.
7962 2013-05-24  Martin Jambor  <mjambor@suse.cz>
7964         PR tree-optimization/57294
7965         * cgraph.h (ipa_record_stmt_references): Declare.
7966         * cgraphbuild.c (ipa_record_stmt_references): New function.
7967         (build_cgraph_edges): Use ipa_record_stmt_references.
7968         (rebuild_cgraph_edges): Likewise.
7969         (cgraph_rebuild_references): Likewise.
7970         * ipa-prop.c (ipa_modify_call_arguments): Discard references
7971         associated with the old statement and build references from the
7972         newly built statements.
7973         * ipa-ref.c (ipa_remove_stmt_references): New function.
7974         * ipa-ref.h (ipa_remove_stmt_references): Declare.
7976 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
7978         * lra-constraints.c (emit_spill_move): Use smaller mode for
7979         mem-mem moves.
7980         (check_and_process_move): Consider mem-reg moves for secondary
7981         too.
7982         (curr_insn_transform): Don't lose insns emitted before for
7983         secondary memory moves.
7984         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
7985         reg set up in the current insn.
7987 2013-05-24  Dehao Chen  <dehao@google.com>
7989         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
7990         hash function.
7991         (locus_descrim_hasher::equal): Likewise.
7992         (build_gimple_cfg): New discriminator assignment algorithm.
7993         (make_edges): Likewise.
7994         (next_discriminator_for_locus): Likewise.
7995         (same_line_p): Likewise.
7996         (assign_discriminators): Likewise.
7997         (make_cond_expr_edges): Likewise.
7998         (make_gimple_switch_edges): Likewise.
7999         (make_goto_expr_edges): Likewise.
8000         (make_gimple_asm_edges): Likewise.
8002 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
8004         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8005         X format specifier to only display bottom 16 bits.
8006         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8007         immediate to match for operand 2, since it will be masked.
8009 2013-05-24  Richard Biener  <rguenther@suse.de>
8011         PR tree-optimization/57287
8012         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8013         all SSA names that occur in abnormal PHIs.
8015 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8017         PR tree-ssa/57385
8018         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8019         that index is not negative.
8021 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
8023         PR rtl-optimization/55177
8024         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8025         (simplify_byte_swapping_operation): New.
8026         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8027         (simplify_relational_operation_1): Deal with BSWAP.
8029 2013-05-23  Richard Henderson  <rth@redhat.com>
8031         PR target/56742
8032         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8033         (ix86_reorg): Call it.
8035 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
8037         PR target/57379
8038         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8039         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8040         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8042 2013-05-23  Christian Bruel  <christian.bruel@st.com>
8044         PR debug/57351
8045         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8047 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8048             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8050         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8051         * config/aarch64/constraints.md (Usa): Remove.
8052         * doc/md.texi (AArch64 Usa): Remove.
8054 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8055             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8057         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8058         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8059         * config/aarch64/predicates.md (aarch64_const_address): Remove.
8060         (aarch64_mov_operand): Use aarch64_mov_operand_p.
8062 2013-05-23  Vidya Praveen <vidyapraveen@arm.com>
8064         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8065         instruction (AdvSIMD).
8066         * config/aarch64/aarch64-builtins.c
8067         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8068         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8070 2013-05-23  Martin Jambor  <mjambor@suse.cz>
8072         PR middle-end/57347
8073         * tree.h (contains_bitfld_component_ref_p): Declare.
8074         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8075         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
8076         caller.
8077         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8078         not access a bit-field.  Assert all final offsets are byte-aligned.
8080 2013-05-23  Richard Biener  <rguenther@suse.de>
8082         PR tree-optimization/57380
8083         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8084         least one invariant or re-used load.
8085         * passes.c (init_optimization_passes): Move pass_phiprop before
8086         pass_forwprop.
8088 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
8090         * config/aarch64/aarch64-simd.md
8091         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8093 2013-05-23  Richard Biener  <rguenther@suse.de>
8095         PR middle-end/57381
8096         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8097         OEP_CONSTANT_ADDRESS_OF retained.
8099 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
8101         PR middle-end/57344
8102         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8103         don't lower unit.  Handle unit not being always BITS_PER_WORD.
8105 2013-05-23  Richard Biener  <rguenther@suse.de>
8107         PR rtl-optimization/57341
8108         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8109         instead of true_dependence.
8111 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
8113         * bb-reorder.c (branch_threshold): Make const.
8114         (exec_threshold): Ditto.
8116 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8117             Pat Haugen <pthaugen@us.ibm.com>
8118             Peter Bergner <bergner@vnet.ibm.com>
8120         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8121         documentation for the power8 crypto builtins.
8123         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
8125         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
8126         macros for defining power8 builtin functions.
8127         (BU_P8V_AV_2): Likewise.
8128         (BU_P8V_AV_P): Likewise.
8129         (BU_P8V_VSX_1): Likewise.
8130         (BU_P8V_OVERLOAD_1): Likewise.
8131         (BU_P8V_OVERLOAD_2): Likewise.
8132         (BU_CRYPTO_1): Likewise.
8133         (BU_CRYPTO_2): Likewise.
8134         (BU_CRYPTO_3): Likewise.
8135         (BU_CRYPTO_OVERLOAD_1): Likewise.
8136         (BU_CRYPTO_OVERLOAD_2): Likewise.
8137         (XSCVSPDP): Fix typo, point to the correct instruction.
8138         (VCIPHER): Add power8 crypto builtins.
8139         (VCIPHERLAST): Likewise.
8140         (VNCIPHER): Likewise.
8141         (VNCIPHERLAST): Likewise.
8142         (VPMSUMB): Likewise.
8143         (VPMSUMH): Likewise.
8144         (VPMSUMW): Likewise.
8145         (VPERMXOR_V2DI): Likewise.
8146         (VPERMXOR_V4SI: Likewise.
8147         (VPERMXOR_V8HI: Likewise.
8148         (VPERMXOR_V16QI: Likewise.
8149         (VSHASIGMAW): Likewise.
8150         (VSHASIGMAD): Likewise.
8151         (VPMSUM): Likewise.
8152         (VPERMXOR): Likewise.
8153         (VSHASIGMA): Likewise.
8155         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8156         __CRYPTO__ if the crypto instructions are available.
8157         (altivec_overloaded_builtins): Add support for overloaded power8
8158         builtins.
8160         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
8161         support for power8 crypto builtins.
8162         (builtin_function_type): Likewise.
8163         (altivec_init_builtins): Add support for builtins that take vector
8164         long long (V2DI) arguments.
8166         * config/rs6000/crypto.md: New file, define power8 crypto
8167         instructions.
8169 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8170             Pat Haugen <pthaugen@us.ibm.com>
8171             Peter Bergner <bergner@vnet.ibm.com>
8173         * doc/invoke.texi (Option Summary): Add power8 options.
8174         (RS/6000 and PowerPC Options): Likewise.
8176         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
8177         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
8178         wm, wn, wr documentation.
8180         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
8181         registers if direct move instructions are enabled.
8182         (wn): New constraint for no registers.
8183         (wq): New constraint for quad word even GPR registers.
8184         (wr): New constraint if 64-bit instructions are enabled.
8185         (wv): New constraint if power8 vector instructions are enabled.
8186         (wQ): New constraint for quad word memory locations.
8188         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
8189         constraint for 0..15 for crypto instructions.
8190         (gpc_reg_operand): If VSX allow registers in VSX registers as well
8191         as GPR and floating point registers.
8192         (int_reg_operand): New predicate to match only GPR registers.
8193         (base_reg_operand): New predicate to match base registers.
8194         (quad_int_reg_operand): New predicate to match even GPR registers
8195         for quad memory operations.
8196         (vsx_reg_or_cint_operand): New predicate to allow vector logical
8197         operations in both GPR and VSX registers.
8198         (quad_memory_operand): New predicate for quad memory operations.
8199         (reg_or_indexed_operand): New predicate for direct move support.
8201         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
8202         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
8203         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
8204         (POWERPC_MASKS): Add power8 options.
8205         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
8206         various options.
8208         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8209         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
8211         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
8212         (-mpower8-fusion): New power8 options.
8213         (-mpower8-fusion-sign): Likewise.
8214         (-mpower8-vector): Likewise.
8215         (-mcrypto): Likewise.
8216         (-mdirect-move): Likewise.
8217         (-mquad-memory): Likewise.
8219         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
8220         power8.
8221         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
8222         registers.
8223         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
8224         (rs6000_debug_vector_unit): Add p8_vector.
8225         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
8226         definitions.  Also print fusion state.
8227         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
8228         (rs6000_builtin_mask_calculate): Add power8 builtin support.
8229         (rs6000_option_override_internal): Add support for power8.
8230         (rs6000_common_init_builtins): Add debugging for skipped builtins
8231         if -mdebug=builtin.
8232         (rs6000_adjust_cost): Add power8 support.
8233         (rs6000_issue_rate): Likewise.
8234         (insn_must_be_first_in_group): Likewise.
8235         (insn_must_be_last_in_group): Likewise.
8236         (force_new_group): Likewise.
8237         (rs6000_register_move_cost): Likewise.
8238         (rs6000_opt_masks): Likewise.
8240         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
8241         power8 capable assembler, default to power7 options.
8242         (TARGET_DIRECT_MOVE): Likewise.
8243         (TARGET_CRYPTO): Likewise.
8244         (TARGET_P8_VECTOR): Likewise.
8245         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
8246         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
8247         (VECTOR_MEM_P8_VECTOR_P): Likewise.
8248         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
8249         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
8250         (TARGET_XSCVDPSPN): Likewise.
8251         (TARGET_XSCVSPDPN): Likewsie.
8252         (TARGET_SYNC_HI_QI): Likewise.
8253         (TARGET_SYNC_TI): Likewise.
8254         (MASK_CRYPTO): Likewise.
8255         (MASK_DIRECT_MOVE): Likewise.
8256         (MASK_P8_FUSION): Likewise.
8257         (MASK_P8_VECTOR): Likewise.
8258         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
8259         temporary used by some of the direct move instructions to get two FP
8260         temporary registers does not force creation of a stack frame.
8261         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
8262         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
8263         that any VSX registers are tieable, even if they are also an
8264         Altivec vector mode.
8265         (r6000_reg_class_enum): Add wm, wr, wv constraints.
8266         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
8267         (RS6000_BTM_CRYPTO): Likewise.
8268         (RS6000_BTM_COMMON): Likewise.
8270         * config/rs6000/rs6000.md (cpu attribute): Add power8.
8271         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
8272         (enum rs6000_vector): Add power8 vector support.
8274 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8276         PR target/19599
8277         PR target/57340
8278         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
8279         (any_sibcall_could_use_r3): this and handle indirect calls.
8280         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
8282 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8284         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
8286 2013-05-22  Richard Biener  <rguenther@suse.de>
8288         PR middle-end/57349
8289         * profile.c (branch_prob): Do not split blocks that are
8290         abnormally receiving from ECF_RETURNS_TWICE functions.
8292 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
8294         * recog.c (offsettable_address_addr_space_p): Fix calculation of
8295         address mode.  Move pointer mode initialization to the same place.
8297 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8299         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
8300         while it has any effect.
8302 2013-05-21  Easwaran Raman  <eraman@google.com>
8304         PR tree-optimization/57322
8305         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
8306         UID of the statement added to the BB to be 1.
8308 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
8310         PR tree-optimization/57331
8311         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
8312         of conversion from pointer type to integral type with integer.
8314 2013-05-21  Martin Jambor  <mjambor@suse.cz>
8316         PR lto/57289
8317         * ipa-prop.c (ipa_read_node_info): Process param_used and
8318         controlled_uses in the same order as when writing.
8320 2013-05-21  Magnus Granberg  <baldrick@free.fr>
8322         PR plugins/56754
8323         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
8325 2013-05-21  Richard Biener  <rguenther@suse.de>
8327         PR tree-optimization/57318
8328         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
8329         estimate stmts with side-effects as likely eliminated.
8331 2013-05-21  Richard Biener  <rguenther@suse.de>
8333         PR tree-optimization/57330
8334         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
8335         preserve the call stmts fntype.
8337 2013-05-21  Richard Biener  <rguenther@suse.de>
8339         PR tree-optimization/57303
8340         * tree-ssa-sink.c (statement_sink_location): Improve killing
8341         stmt detection and properly handle self-assignments.
8343 2013-05-21  Christian Bruel  <christian.bruel@st.com>
8345         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
8346         spanning registers. LEAF_REG_REMAP is supported only for contiguous
8347         registers. Set register size out of the PARALLEL loop.
8349 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
8351         PR target/56547
8352         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
8353         (*fmasf4, *fmasf4_media): New insns.
8355 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8357         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
8358         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
8359         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
8360         (mips_idiv_insns): Update the comments to say that the returned
8361         instruction counts are in units of BASE_INSN_LENGTH.
8362         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
8363         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
8364         using 2 rather than 4 as the length of indirect MIPS16 and
8365         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
8366         length of a NOP.  Don't divide MIPS16 lengths by 2.
8367         (mips16_split_long_branches): Assume a branch is long if the
8368         length is greater than 4 rather than 8.
8369         * config/mips/mips.md (length): Give MIPS16 lengths directly,
8370         rather than multiplying them by 2.  Multiply instruction counts
8371         by BASE_INSN_LENGTH rather than 4.
8372         (*jump_mips16, tls_get_tp_mips16_<mode>)
8373         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
8375 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8377         * config/mips/mips.md (extended_mips16): Remove branch case.
8378         (length): Remove duplicated extended_mips16 test.
8380 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8382         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
8384 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
8386         * recog.h (Recog_data): Rename to...
8387         (recog_data_d): ...this.
8388         (recog_data): Update accordingly.
8389         * recog.c (recog_data): Likewise.
8390         * reload.c (save_recog_data): Likewise.
8391         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
8392         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
8394 2013-05-17  Julian Brown  <julian@codesourcery.com>
8396         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
8397         found in a REG_EQUAL note, invalidate it.
8399 2013-05-17  Easwaran Raman  <eraman@google.com>
8401         * tree-ssa-reassoc.c (find_insert_point): New function.
8402         (insert_stmt_after): Likewise.
8403         (get_def_stmt): Likewise.
8404         (ensure_ops_are_available): Likewise.
8405         (not_dominated_by): Likewise.
8406         (rewrite_expr_tree): Do not move statements beyond what is
8407         necessary. Remove call to swap_ops_for_binary_stmt...
8408         (reassociate_bb): ... and move it here.
8409         (build_and_add_sum): Assign UIDs for new statements.
8410         (linearize_expr): Likewise.
8411         (do_reassoc): Renumber gimple statement UIDs.
8413 2013-05-17  Jan Hubicka  <jh@suse.cz>
8415         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
8416         weakrefs.
8417         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
8418         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
8419         weakrefs.
8420         (output_weakrefs): Update.
8422 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
8423             Martin Jambor  <mjambor@suse.cz>
8425         PR middle-end/57276
8426         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
8427         value that corresponds to the given aggval is found in values vector.
8429 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
8431         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
8432         sse, sse2, sse3, ssse3 and sse4a flags to options.
8434 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
8436         * gengtype-state.c: (s_expr_writer): New class, to handle
8437         prettifying of output layout of s-expressions.
8438         (state_writer): New class, to write out gtype.state.
8439         (state_written_type_count): Move this variable into member data of
8440         state_writer.
8441         (s_expr_writer::s_expr_writer): New code: constructor for new class
8442         (state_writer::state_writer(): ditto
8443         (s_expr_writer::write_new_line): New function
8444         (s_expr_writer::write_any_indent): ditto
8445         (s_expr_writer::begin_s_expr): ditto
8446         (s_expr_writer::end_s_expr): ditto
8447         (write_state_fileloc): convert to method of state_writer...
8448         (state_writer:: write_state_fileloc): ...and use methods of
8449         s_expr_writer to write indentation into the gtype.state output file
8450         to visually represent the hierarchical structure of the list
8451         structures
8452         (write_state_fields): ditto, renaming to...
8453         (state_writer::write_state_fields)
8454         (write_state_a_string): ditto, renaming to...
8455         (state_writer::write_state_a_string)
8456         (write_state_string_option): ditto, renaming to...
8457         (state_writer::write_state_string_option)
8458         (write_state_type_option): ditto, renaming to...
8459         (state_writer::write_state_type_option)
8460         (write_state_nested_option): ditto, renaming to...
8461         (state_writer::write_state_nested_option)
8462         (write_state_option): ditto, renaming to...
8463         (state_writer::write_state_option)
8464         (write_state_options): ditto, renaming to...
8465         (state_writer::write_state_options)
8466         (write_state_lang_bitmap): ditto, renaming to...
8467         (state_writer::write_state_lang_bitmap)
8468         (write_state_version): ditto, renaming to...
8469         (state_writer::write_state_version)
8470         (write_state_scalar_type): ditto, renaming to...
8471         (state_writer::write_state_scalar_type)
8472         (write_state_string_type): ditto, renaming to...
8473         (state_writer::write_state_string_type)
8474         (write_state_undefined_type): ditto, renaming to...
8475         (state_writer::write_state_undefined_type)
8476         (write_state_struct_union_type): ditto, renaming to...
8477         (state_writer::write_state_struct_union_type)
8478         (write_state_struct_type): ditto, renaming to...
8479         (state_writer::write_state_struct_type)
8480         (write_state_user_struct_type): ditto, renaming to...
8481         (state_writer::write_state_user_struct_type)
8482         (write_state_lang_struct_type): ditto, renaming to...
8483         (state_writer::write_state_lang_struct_type)
8484         (write_state_param_struct_type): ditto, renaming to...
8485         (state_writer::write_state_param_struct_type)
8486         (write_state_pointer_type): ditto, renaming to...
8487         (state_writer::write_state_pointer_type)
8488         (write_state_array_type): ditto, renaming to...
8489         (state_writer::write_state_array_type)
8490         (write_state_gc_used): ditto, renaming to...
8491         (state_writer::write_state_gc_used)
8492         (write_state_common_type_content): ditto, renaming to...
8493         (state_writer::write_state_common_type_content)
8494         (write_state_type): ditto, renaming to...
8495         (state_writer::write_state_type)
8496         (write_state_pair_list): ditto, renaming to...
8497         (state_writer::write_state_pair_list)
8498         (write_state_pair): ditto, renaming to...
8499         (state_writer::write_state_pair)
8500         (write_state_typedefs): ditto, renaming to...
8501         (state_writer::write_state_typedefs)
8502         (write_state_structures): ditto, renaming to...
8503         (state_writer::write_state_structures)
8504         (write_state_param_structs): ditto, renaming to...
8505         (state_writer::write_state_param_structs)
8506         (write_state_variables): ditto, renaming to...
8507         (state_writer::write_state_variables)
8508         (write_state_srcdir): ditto, renaming to...
8509         (state_writer::write_state_srcdir)
8510         (write_state_files_list): ditto, renaming to...
8511         (state_writer::write_state_files_list)
8512         (write_state_languages): ditto, renaming to...
8513         (state_writer::write_state_languages)
8514         (write_state): create a state_writer instance and use it when
8515         writing out the state file
8517 2013-05-17  Mike Stump  <mikestump@comcast.net>
8519         PR rtl-optimization/57304
8520         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
8521         accessing DF_REF_REAL_LOC.
8523 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8525         PR rtl-optimization/57281
8526         PR rtl-optimization/57300
8527         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
8528         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
8529         what the other splitter did if the registers are dead.
8531 2013-05-17  Richard Biener  <rguenther@suse.de>
8533         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
8534         MEM_REF offsets.
8536 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8538         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
8539         linking.
8541 2013-05-17  Marek Polacek  <polacek@redhat.com>
8543         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
8544         length when doing non-zero store of storing '\0' to '\0'.
8546 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8548         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
8549         vect_external_def oprnd1 with loop_vinfo, try to emit
8550         optional cast, negation and and stmts on the loop preheader
8551         edge instead of into the pattern def seq.
8553         PR tree-optimization/57051
8554         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
8555         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
8557 2013-05-16  Nick Clifton  <nickc@redhat.com>
8559         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
8560         (rl78_is_naked_func): New function.
8561         (rl78_expand_prologue): Skip prologue generation for naked functions.
8562         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
8563         * doc/extend.texi (naked): Add RL78 to the list of processors
8564         that supports this attribute.
8566 2013-05-16  Jeff Law  <law@redhat.com>
8568         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
8570 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8572         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
8573         cache parameters using detect_caches_amd also for CYRIX,
8574         NSC and TM2 signatures.
8576 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8577             Dzianis Kahanovich  <mahatma@eu.by>
8579         PR target/45359
8580         PR target/46396
8581         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8582         VIA/Centaur processors and determine their cache parameters
8583         using detect_caches_amd.
8585 2013-05-16  Teresa Johnson  <tejohnson@google.com>
8587         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
8588         (rtl_verify_edges): New function.
8589         (rtl_verify_bb_insns): Ditto.
8590         (rtl_verify_bb_pointers): Ditto.
8591         (rtl_verify_bb_insn_chain): Ditto.
8592         (rtl_verify_fallthru): Ditto.
8593         (rtl_verify_bb_layout): Ditto.
8594         (rtl_verify_flow_info_1): Outline checks into new functions.
8595         (rtl_verify_flow_info): Ditto.
8597 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
8599         * cfghooks.c (copy_bbs): Add update_dominance argument.
8600         * cfghooks.h (copy_bbs): Update prototype.
8601         * tree-cfg.c (gimple_duplicate_sese_region):
8602         Add update_dominance argument.
8603         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
8604         * tree-ssa-loop-ch.c (copy_loop_headers): Update
8605         gimple_duplicate_sese_region call.
8606         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
8607         Update copy_bbs call.
8608         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
8609         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
8611 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8613         * tree-vectorizer.h (NUM_PATTERNS): Increment.
8614         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
8615         vect_recog_rotate_pattern.
8616         (vect_recog_rotate_pattern): New function.
8618 2013-05-16  Jason Merrill  <jason@redhat.com>
8620         * Makefile.in (LLINKER): New variable.
8621         (mostlyclean): Remove link mutex.
8622         * configure.ac: Handle --enable-link-mutex.
8623         * lock-and-run.sh: New script.
8625 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8627         PR target/19599
8628         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
8629         for NULL decl.
8631 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8633         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
8635 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
8637         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
8638         * config/arm/arm.c (next_consecutive_mem): New function.
8639         (gen_movmem_ldrd_strd): Likewise.
8640         * config/arm/arm.md (movmemqi): Update condition and code.
8641         (unaligned_loaddi, unaligned_storedi): New patterns.
8643 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8645         * config.gcc: Obsolete *-*-solaris2.9*.
8646         * doc/install.texi (Specific, *-*-solaris2*): Document it.
8648 2013-05-16  Richard Biener  <rguenther@suse.de>
8650         * passes.c (init_optimization_passes): Move pass_parallelize_loops
8651         earlier, after GRAPHITE transforms and IV canonicalization.
8653 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8655         * omp-low.c (extract_omp_for_data): For collapsed loops,
8656         if at least one of the loops is known at compile time to
8657         iterate zero times, set count to 0.
8658         (expand_omp_regimplify_p): New function.
8659         (expand_omp_for_generic): For collapsed loops, if at least
8660         one of the loops isn't known to iterate at least once,
8661         add runtime check with setting count to 0.
8662         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
8663         For unsigned types if it isn't known at compile time that
8664         the loop will iterate at least once, add runtime check to bypass
8665         the whole loop if initial condition isn't true.
8667 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
8669         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
8671 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
8673         PR middle-end/57286
8674         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
8675         transformations to avoid an infinite loop.
8677 2013-05-16  Marek Polacek  <polacek@redhat.com>
8679         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
8681 2013-05-15  Leif Ekblad  <leif@rdos.net>
8683         * config/i386/i386.c (ix86_decompose_address): Use
8684         DEFAULT_TLS_SEG_REG to access TLS segment register.
8685         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
8686         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
8687         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
8689 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
8691         PR target/57260
8692         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
8693         sibling calls to functions that would normally be lazily bound,
8694         unless $gp is call-clobbered.
8696 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
8698         * config/i386/i386.c (ix86_option_override_internal): Update
8699         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
8700         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
8701         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
8702         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
8703         of TARGET_3DNOW.
8704         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
8706 2013-05-15  Andreas Schwab  <schwab@suse.de>
8708         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
8709         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
8710         third operand.
8712 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8714         * loop-unroll.c (report_unroll_peel): Check decision before
8715         emitting unroll/peel message.
8717 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8719         * function.h (has_bb_partition): New rtl_data flag.
8720         (bb_reorder_complete): Ditto.
8721         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
8722         instead of flag_reorder_blocks_and_partition.
8723         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
8724         with some enhancements.
8725         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
8726         * bb-reorder.c (connect_traces): Check for has_bb_partition
8727         instead of flag_reorder_blocks_and_partition.
8728         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
8729         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
8730         verify_hot_cold_block_grouping.
8731         (partition_hot_cold_basic_blocks): Set has_bb_partition.
8733 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8735         PR target/19599
8736         * config/arm/predicates.md (call_insn_operand): New predicate.
8737         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
8738         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
8739         if insn is not a tail call.
8740         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
8741         registers.
8742         * config/arm/arm.h (enum reg_class): New caller save register class.
8743         (REG_CLASS_NAMES): Likewise.
8744         (REG_CLASS_CONTENTS): Likewise.
8745         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
8746         without decls.
8748 2013-05-15  Richard Biener  <rguenther@suse.de>
8750         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
8751         of MSG_OPTIMIZED_LOCATIONS.
8752         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
8753         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
8754         message.
8755         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
8756         of MSG_OPTIMIZED_LOCATIONS.
8757         (execute_vect_slp): Likewise.
8758         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
8759         (vect_create_cond_for_alias_checks): Likewise.
8760         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8761         (vect_recog_widen_mult_pattern): Likewise.
8762         (vect_recog_widen_sum_pattern): Likewise.
8763         (vect_recog_over_widening_pattern): Likewise.
8764         (vect_recog_widen_shift_pattern): Likewise.
8765         (vect_recog_vector_vector_shift_pattern): Likewise.
8766         (vect_recog_divmod_pattern): Likewise.
8767         (vect_recog_mixed_size_cond_pattern): Likewise.
8768         (vect_recog_bool_pattern): Likewise.
8769         (vect_pattern_recog_1): Likewise.
8771 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8773         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
8774         non-functions to builtin_unreachable.
8775         * ipa-inline-transform.c (inline_call): Do not assert estimates were
8776         correct when new direct edges were discovered.
8778 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8780         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
8781         header, print symbol order instead of node uid, print more information
8782         about indirect edge targets.
8783         (ipa_make_edge_direct_to_target): Print symbol order instead of node
8784         uids.
8785         (ipa_make_edge_direct_to_target): Likewise.
8786         (remove_described_reference): Likewise.
8787         (propagate_controlled_uses): Likewise.
8788         (ipa_print_node_params): Also print symbol order.
8789         (ipcp_transform_function): Print symbol order instead of node uids.
8790         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
8791         (cgraph_get_create_real_symbol_node): Likewise.
8792         * ipa-cp.c (print_lattice): Likewise.
8793         (print_all_lattices): Likewise.
8794         (determine_versionability): Likewise.
8795         (initialize_node_lattices): Likewise.
8796         (estimate_local_effects): Likewise.
8797         (update_profiling_info): Likewise.
8798         (create_specialized_node): Likewise.
8799         (perhaps_add_new_callers): Likewise.
8800         (decide_about_value): Likewise.
8801         (decide_whether_version_node): Likewise.
8802         (identify_dead_nodes): Likewise.
8803         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
8804         (dump_inline_summary): Likewise.
8805         (estimate_node_size_and_time): Likewise.
8806         (inline_analyze_function): Likewise.
8807         * ipa-inline.c (report_inline_failed_reason): Likewise.
8808         (want_early_inline_function_p): Likewise.
8809         (edge_badness): Likewise.
8810         (update_edge_key): Likewise.
8811         (inline_small_functions): Likewise.  Add dumping of order to two other
8812         dumps.
8813         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
8814         instead of node uids.
8815         (propagate_pure_const): Likewise.
8816         (propagate_pure_const): Likewise.
8817         * ipa-utils.c (dump_cgraph_node_set): Likewise.
8818         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
8819         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
8820         of node uids.
8821         * tree-pretty-print.c (dump_function_header): Likewise.
8822         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
8823         Print symbol order instead of node uids.
8825 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8827         * config/s390/s390.c (s390_register_move_cost): Don't impose the
8828         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
8830 2013-05-15  Richard Biener  <rguenther@suse.de>
8832         PR tree-optimization/57275
8833         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
8834         return value for fail to do runtime alias checks for gather loads.
8836 2013-05-15  Jan Hubicka  <jh@suse.cz>
8838         PR lto/57038
8839         PR lto/47375
8840         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
8841         weakrefs are not external.
8842         (lto_symtab_merge_decls): Fix thinko when dealing with
8843         non-lto_symtab decls.
8844         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
8845         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
8846         * varpool.c (dump_varpool_node): Dump more flags.
8848 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8850         * config/i386/i386.c (processor_alias_table): Add instruction
8851         FSGSBASE for AMD bdver3 architecture.
8853 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8855         * tree.c (warn_deprecated_use): Print file:line using locus color.
8856         * diagnostic.c (diagnostic_report_current_module): Print file:line
8857         and file:line:column using locus color.
8859 2013-05-14  Mike Stump  <mikestump@comcast.net>
8861         * gdbinit.in: Add __null.
8863 2013-05-14  Mike Stump  <mikestump@comcast.net>
8865         * recog.h: Rename struct recog_data to Recog_data.
8866         * recog.c: Likewise.
8867         * reload.c (can_reload_into): Likewise.
8868         * config/picochip/picochip.c: Likewise.
8870 2013-05-14  Mike Stump  <mikestump@comcast.net>
8872         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8874 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
8876         * resource.h (struct resources): Remove unch_memory member.
8877         (CLEAR_RESOURCE): Don't clear unch_memory.
8878         * resource.c (mark_referenced_resources): Don't set it.
8879         (mark_set_resources): Likewise.
8880         (mark_target_live_regs): Don't clear it.
8881         (init_resource_info): Likewise.
8882         * reorg.c (resource_conflicts_p): Don't compare it.
8883         (redundant_insn): Don't set it.
8885         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8886         Remove prototypes.
8887         * emit-rtl.c (next_label): Remove unused function.
8888         (skip_consecutive_labels, link_cc0_insns): Move to ...
8889         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8890         only place where these functions are used, and make them static.
8892 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8894         * fold-const.c (fold_negate_expr): Handle vectors.
8895         (fold_truth_not_expr): Make it static.
8896         (fold_invert_truthvalue): New static function.
8897         (invert_truthvalue_loc): Handle vectors. Do not call
8898         fold_truth_not_expr directly.
8899         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8900         <TRUTH_NOT_EXPR>: Do not cast to boolean.
8901         (fold_comparison): Handle vector constants.
8902         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8903         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8904         * tree.h (fold_truth_not_expr): Remove declaration.
8906 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
8908         * config/aarch64/aarch64-simd.md
8909         (aarch64_vcond_internal<mode>): Rename to...
8910         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8911         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8912         float modes. Clarify all iterator modes.
8913         (vcond<mode><mode>): Use new name for vcond expanders.
8914         (vcond<v_cmp_result><mode>): Likewise.
8915         (vcondu<mode><mode>: Likewise.
8916         * config/aarch64/iterators.md (VDQF_COND): New.
8918 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8920         PR bootstrap/57266
8921         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8922         variable for the shift amount. Check that we shift by non-negative
8923         amounts.
8925 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
8927         PR target/42017
8928         * config/arm/arm.h (EPILOGUE_USES): Only return true
8929         for LR_REGNUM after epilogue_completed.
8931 2013-05-14  Joern Rennecke <joern.rennecke@embecosm.com>
8933         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8934         is error_mark_node.
8936 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8938         PR target/57261
8939         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
8940         and Solaris 11+/x86 with gld.
8941         * configure: Regenerate.
8943 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8945         * expmed.c (expand_shift_1): Canonicalize rotates by
8946         constant bitsize / 2 to bitsize - 1.
8947         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
8948         case ROTATERT>: Likewise.
8950         Revert:
8951         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
8953         * config/i386/i386.md (rotateinv): New code attr.
8954         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8955         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8956         roll $31, %eax, etc.
8958 2013-05-14  Richard Biener  <rguenther@suse.de>
8960         PR middle-end/57235
8961         * tree-eh.c (sink_clobbers): Give up for successors with
8962         multiple predecessors and no virtual uses.
8964 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
8966         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
8967         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
8969 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8971         PR middle-end/57251
8972         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
8973         the case when both op0 and op1 have VOIDmode.
8975 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
8977         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
8978         in multiply-accumulate mode.
8980 2013-05-13  Guozhi Wei  <carrot@google.com>
8982         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
8984 2013-05-13  Kai Tietz  <ktietz@redhat.com>
8986         PR target/56975
8987         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
8988         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
8989         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
8990         * config/i386/i386.c (ix86_option_override_internal): Likewise.
8991         (ix86_expand_prologue): Likewise.
8992         (ix86_expand_split_stack_prologue): Likewise.
8993         (legitimate_pic_address_disp_p): Likewise.
8994         (legitimize_pic_address): Likewise.
8995         (legitimize_tls_address): Likewise.
8996         (legitimize_pe_coff_symbol): Likewise.
8997         (output_pic_addr_const): Likewise.
8998         (construct_plt_address): Likewise.
8999         (ix86_expand_call): Likewise.
9000         (x86_output_mi_thunk): Likewise.
9001         (x86_function_profiler): Likewise.
9003 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
9005         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9006         similar switch cases.
9007         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9008         (aarch64_simd_mov_to_<mode>low): Delete.
9009         (aarch64_simd_mov_to_<mode>high): Delete.
9010         (move_lo_quad_<mode>): Add w<-r alternative.
9011         (aarch64_simd_move_hi_quad_<mode>): Likewise.
9012         (aarch64_simd_mov_from_*): Update type attribute.
9013         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9014         statement.
9016 2013-05-13  Jan Hubicka  <jh@suse.cz>
9018         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9019         * config/i386/i386.c (ix86_compute_frame_layout,
9020         ix86_expand_epilogue, emit_i387_cw_initialization,
9021         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9022         ix86_local_alignment): Fix use of size/speed predicates.
9024 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9026         PR tree-optimization/45216
9027         PR tree-optimization/57157
9028         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9029         the (-Y) & (B - 1) variant if OP is |.
9030         * expmed.c (expand_shift_1): For rotations by const0_rtx just
9031         return shifted.  Use (-op1) & (prec - 1) as other_amount
9032         instead of prec - op1.
9034 2013-05-13  Martin Jambor  <mjambor@suse.cz>
9036         PR middle-end/42371
9037         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9038         (ipa_constant_data): New type.
9039         (ipa_jump_func): Use ipa_constant_data to hold information about
9040         constant jump functions.
9041         (ipa_get_jf_constant): Adjust to jump function type changes.
9042         (ipa_get_jf_constant_rdesc): New function.
9043         (ipa_param_descriptor): New field controlled_uses.
9044         (ipa_get_controlled_uses): New function.
9045         (ipa_set_controlled_uses): Likewise.
9046         * ipa-ref.h (ipa_find_reference): Declare.
9047         * ipa-prop.c (ipa_cst_ref_desc): New type.
9048         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9049         changes.
9050         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
9051         New parameter cs.  Adjust all callers.
9052         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9053         (remove_described_reference): New function.
9054         (jfunc_rdesc_usable): Likewise.
9055         (try_make_edge_direct_simple_call): Decrement controlled use count,
9056         attempt to remove reference if it hits zero.
9057         (combine_controlled_uses_counters): New function.
9058         (propagate_controlled_uses): Likewise.
9059         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9060         (ipa_edge_duplication_hook): Duplicate reference descriptions.
9061         (ipa_print_node_params): Print described use counter.
9062         (ipa_write_jump_function): Adjust to jump function type changes.
9063         (ipa_read_jump_function): New parameter CS, pass it to
9064         ipa_set_jf_constant.  Adjust caller.
9065         (ipa_write_node_info): Stream controlled use count
9066         (ipa_read_node_info): Likewise.
9067         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9068         asserting.
9069         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9070         count.  Remove cloning-added reference if it reaches zero.
9071         * ipa-ref.c (ipa_find_reference): New function.
9073 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9075         * config/i386/i386.c (processor_target_table): Modified default
9076         alignment values for AMD BD and BT architectures.
9078 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
9080         * tree-vect-generic.c (uniform_vector_p): Move ...
9081         * tree.c (uniform_vector_p): ... here.
9082         * tree.h (uniform_vector_p): Declare it.
9083         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9084         into a scalar.
9086 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9088         PR tree-optimization/57230
9089         * tree-ssa-strlen.c (handle_char_store): Record length for
9090         array store from STRING_CST.
9092         PR tree-optimization/57230
9093         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9094         check.
9096 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
9098         * config/epiphany/epiphany.c (epiphany_init): Check size of
9099         NUM_MODES_FOR_MODE_SWITCHING.
9100         (epiphany_expand_prologue):
9101         Remove CONFIG_REGNUM initial value handling code.
9102         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9103         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9104         (emit_set_fp_mode, epiphany_mode_after): Likewise.
9105         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9106         Don't return 1 for FP_MODE_NONE.
9107         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9108         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9109         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9110         * config/epiphany/epiphany.md (save_config): New pattern.
9112 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
9114         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9116 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
9118         * config/i386/i386.md (memory): Handle sseishft1.
9119         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9120         (*vec_extractv2di_1): Ditto.
9122 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
9124         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
9125         saved registers.
9127 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
9129         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
9130         Add mthumb/march=armv7-a multilib.
9131         Add mthumb/march=armv7-r multilib.
9132         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
9134 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
9136         * config/v850/t-rtems: Add more multilibs.
9138 2013-05-10  Richard Biener  <rguenther@suse.de>
9140         PR tree-optimization/57214
9141         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
9142         not propagate from SSA names that occur in abnormal PHI nodes.
9144 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
9146         * stor-layout.c (element_precision): New function.
9147         * machmode.h (element_precision): Declare it.
9148         * tree.c (build_minus_one_cst): New function.
9149         (element_precision): Likewise.
9150         * tree.h (build_minus_one_cst): Declare new function.
9151         (element_precision): Likewise.
9152         * fold-const.c (operand_equal_p): Use element_precision.
9153         (fold_binary_loc): Handle vector types.
9154         * convert.c (convert_to_integer): Use element_precision.
9155         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
9156         separately.
9158 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
9160         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
9161         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
9162         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
9163         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9164         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
9165         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
9166         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
9167         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9168         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
9169         (Uuw8): New constraints.
9170         (Usb4): Move into alphabetical order.
9171         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
9172         (sd8_operand, ub8_operand, uw8_operand): New predicates.
9173         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
9174         previously unnamed patterns.
9175         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
9176         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
9177         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
9178         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
9179         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
9180         of specific lengths.
9182 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9184         * config/i386/i386.md (rotateinv): New code attr.
9185         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9186         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9187         roll $31, %eax, etc.
9189         PR tree-optimization/45216
9190         PR tree-optimization/57157
9191         * tree-ssa-forwprop.c (simplify_rotate): New function.
9192         (ssa_forward_propagate_and_combine): Call it.
9194 2013-05-10  Richard Biener  <rguenther@suse.de>
9196         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
9197         disable peeling when we version for aliasing.
9198         (vector_alignment_reachable_p): Honor explicit user alignment.
9199         (vect_supportable_dr_alignment): Likewise.
9200         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
9201         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
9202         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
9203         then peeling to arrange for the cost-model check to come first.
9205 2013-05-10  Alan Modra  <amodra@gmail.com>
9207         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
9208         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
9209         * configure: Regenerate.
9211 2013-05-10  Alan Modra  <amodra@gmail.com>
9213         PR target/55033
9214         * varasm.c (default_elf_select_section): Move !DECL_P check..
9215         (get_named_section): ..to here before calling get_section_name.
9216         Adjust assertion.
9217         (default_section_type_flags): Add DECL_P check.
9218         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
9219         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
9221 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
9223         * config/epiphany/epiphany.c (epiphany_expand_prologue):
9224         When using gen_stack_adjust_str with a register offset, add a
9225         REG_FRAME_RELATED_EXPR note.
9227 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
9229         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
9230         (*vec_extractv4si_zext_mem): Ditto.
9231         (*vec_extractv2di): Add 0->x and x->x alternatives.
9232         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
9233         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
9235 2013-05-09  Jason Merrill  <jason@redhat.com>
9237         N3639 C++1y VLA support
9238         * gimplify.c (gimplify_vla_decl): Don't touch an existing
9239         DECL_VALUE_EXPR.
9241         * tree.c (build_constructor_va): New.
9242         * tree.h: Declare it.
9244 2013-05-09  Martin Jambor  <mjambor@suse.cz>
9246         PR lto/57084
9247         * gimple-fold.c (canonicalize_constructor_val): Call
9248         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
9250 2013-05-09  Jan Hubicka  <jh@suse.cz>
9251             Richard Biener  <rguenther@suse.de>
9253         PR lto/54095
9254         * symtab.c (symtab_make_decl_local): Do not add private names.
9256 2013-05-09  Jan Hubicka  <jh@suse.cz>
9258         PR lto/54095
9259         * symtab.c (insert_to_assembler_name_hash): Handle clones.
9260         (unlink_from_assembler_name_hash): Likewise.
9261         (symtab_prevail_in_asm_name_hash, symtab_register_node,
9262         symtab_unregister_node, symtab_initialize_asm_name_hash,
9263         change_decl_assembler_name): Update.
9265 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
9267         * config/aarch64/aarch64.md: New movtf split.
9268         (*movtf_aarch64): Update.
9269         (aarch64_movdi_tilow): Handle TF modes and rename to
9270         aarch64_movdi_<mode>low.
9271         (aarch64_movdi_tihigh): Handle TF modes and rename to
9272         aarch64_movdi_<mode>high
9273         (aarch64_movtihigh_di): Handle TF modes and rename to
9274         aarch64_mov<mode>high_di
9275         (aarch64_movtilow_di): Handle TF modes and rename to
9276         aarch64_mov<mode>low_di
9277         (aarch64_movtilow_tilow): Remove spurious whitespace.
9278         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
9279         splits.
9280         (aarch64_print_operand): Update.
9282 2013-05-09  Alan Modra  <amodra@gmail.com>
9284         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
9285         powerpc64le.
9286         * configure: Regenerate.
9288 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
9290         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
9291         splitter preparation statements.
9292         * config/i386/sse.md (*vec_extract* splitters): Ditto.
9293         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
9294         adjust_address_nv.
9296 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9298         * gimple-ssa-strength-reduction.c (count_candidates): Change
9299         return value to int.
9300         (analyze_candidates_and_replace): Change type of length to int.
9302 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
9304         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
9305         (*vec_extract<mode>): Use VI12_128 mode iterator.
9306         (*vec_extract<mode>_mem): Ditto.
9307         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
9308         attribute.
9310 2013-05-08  Diego Novillo  <dnovillo@google.com>
9312         PR bootstrap/54659
9314         Revert:
9315         2012-08-17  Diego Novillo  <dnovillo@google.com>
9317         PR bootstrap/54281
9318         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
9319         * config.in: Regenerate.
9320         * configure: Regenerate.
9321         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
9323 2013-05-08  Jan Hubicka  <jh@suse.cz>
9325         PR lto/54095
9326         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
9327         * cgraph.h (symtab_node_base): Add unique_name.
9328         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9329         input_overwrite_node, input_varpool_node): Stream unique_name.
9330         * cgraphclones.c (cgraph_create_virtual_clone,
9331         cgraph_function_versioning): Set unique_name.
9332         * ipa.c (function_and_variable_visibility): Set unique_name.
9334 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9336         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
9337         (alloc_cand_and_find_basis): Restrict conditional candidate
9338         processing to CAND_MULTs.
9340 2013-05-08  Jan Hubicka  <jh@suse.cz>
9342         PR lto/54095
9343         lto-symtab.c (lto_symtab_symbol_p): New function.
9344         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
9345         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
9346         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
9347         Skip static symbols.
9349 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
9351         PR tree-optimization/57200
9352         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
9353         Only call inform if the preceding warning_at returns true.
9355 2013-05-07  Han Shen  <shenhan@google.com>
9357         * cfgexpand.c (record_or_union_type_has_array_p): New function.
9358         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
9359         * common.opt (fstack-protector-strong): New option.
9360         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
9361         * doc/invoke.texi (Optimization Options): Document
9362         "-fstack-protector-strong".
9363         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
9365 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9367         * config/mips/mips.c (mips_machine_reorg2): Return 0.
9369 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
9371         * ira.c (update_equiv_regs): Add insn having equiv memory even if
9372         it is not lhs of the insn.
9373         (setup_reg_equiv): Remove insn having equiv memory which it is not
9374         lhs of the insn.
9375         * lra-constraints.c (process_address): Try to improve generation
9376         code for address base + disp.
9377         (lra_constraints): Make correct the code for checking insn setting
9378         up backward equivalence.  Remove insn only if it is in the init
9379         insn list.
9380         * lra-eliminations.c (update_reg_eliminate): Change return value.
9381         (lra_eliminate): Use the result.
9383 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
9385         * config/i386/sse.md (ssescalarnummask): New mode attribute.
9386         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
9387         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
9388         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
9389         register target operands.
9390         (*vec_extractv8hi_sse2): New pattern.
9391         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
9392         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
9393         (*vec_extract<mode>_mem): New insn and split pattern.
9395 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
9397         * config/arm/arm.c (arm_asan_shadow_offset): New function.
9398         (TARGET_ASAN_SHADOW_OFFSET): Define.
9399         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
9400         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
9402 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9404         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
9405         (incr_vec_index): Return -1 if increment not found.
9406         (create_add_on_incoming_edge): Assert if increment not found.
9407         (record_increment): Limit number of increments recorded.
9408         (all_phi_incrs_profitable): Return false if an increment not found.
9409         (replace_profitable_candidates): Don't process increments that were
9410         not recorded.
9411         (analyze_candidates_and_replace): Limit size of incr_vec.
9413 2013-05-07  Richard Biener  <rguenther@suse.de>
9415         * calls.c (special_function_p): setjmp-like functions are leaf.
9416         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
9417         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
9419 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
9421         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
9422         (aarch64_simd_mov<mode>): New expander.
9423         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
9424         (aarch64_simd_mov_to_<mode>high): Likewise.
9425         (aarch64_simd_mov_from_<mode>low): Likewise.
9426         (aarch64_simd_mov_from_<mode>high): Likewise.
9427         (aarch64_dup_lane<mode>): Update.
9428         (aarch64_dup_lanedi): New instruction pattern.
9429         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
9430         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
9432 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9434         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
9435         (replace_mult_candidate): Remove unnecessary argument; remove
9436         unnecessary parameter from call to introduce_cast_before_cand.
9437         (replace_unconditional_candidate): Remove unnecessary parameter
9438         from call to replace_mult_candidate.
9439         (replace_conditional_candidate): Likewise.
9440         (insert_initializers): Use make_temp_ssa_name.
9441         (introduce_cast_before_cand): Remove unnecessary argument; use
9442         make_temp_ssa_name.
9443         (replace_one_candidate): Remove unnecessary argument; remove
9444         unnecessary parameter from calls to introduce_cast_before_cand.
9445         (replace_profitable_candidates): Remove unnecessary parameters
9446         from calls to replace_one_candidate.
9448 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9450         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9451         phi def as possibly hiding a basis for a CAND_ADD whose operands
9452         have been commuted in the analysis.
9453         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
9455 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
9457         * config/aarch64/aarch64.md
9458         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
9459         shift value between 0-4.
9461 2013-05-07  Richard Biener  <rguenther@suse.de>
9463         * double-int.h (rshift): New overload.
9464         * double-int.c (rshift): New function.
9465         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
9466         (create_reference_ops_from_ref): Remove.
9467         (vn_reference_insert): Use shared ops for constructing the
9468         reference and copy it.
9470 2013-05-07  Richard Biener  <rguenther@suse.de>
9472         PR middle-end/57190
9473         * tree-eh.c (sink_clobbers): Properly propagate
9474         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
9476 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
9478         PR tree-optimization/57149
9479         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
9480         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
9481         collect_phi_def_edges, execute_late_warn_uninitialized): Use
9482         uninit_undefined_value_p instead of ssa_undefined_value_p.
9484         PR debug/57184
9485         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
9486         for modifier == EXPAND_INITIALIZER.
9488 2013-05-07  Anton Blanchard  <anton@samba.org>
9490         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
9491         for powerpc64 little endian.
9492         * configure: Regenerate.
9494 2013-05-06  Graham Stott  <grahams@btinternet.com>
9496         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
9497         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
9498         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
9499         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
9501 2013-05-06  Graham Stott  <grahams@btinternet.com>
9503         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
9504         codes which allow non-lvalues.
9506 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
9508         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
9509         components are all 1s.
9510         (integer_minus_onep): New function.
9511         * tree.h (integer_minus_onep): Declare it.
9512         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
9513         integer_minus_onep instead of integer_all_onesp.
9515 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9517         PR target/52933
9518         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
9519         variations of these patterns.
9521 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9523         * config/i386/i386.md (isa): Add x64_sse4 member.
9524         (enabled): Handle x64_sse4.
9525         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
9526         instruction for 64bit SSE4_1 targets.  Update insn attributes.
9527         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
9528         instruction for SSE4_1 targets.  Update insn attributes.
9529         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9530         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
9531         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
9532         const_1 selector.
9533         (*vec_extractv4si): Rename from *sse4_1_pextrd.
9534         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
9535         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
9537 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9539         PR target/57108
9540         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
9542 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
9544         * final.c (do_assembler_dialects): Don't handle curly braces and
9545         vertical bar escaped by % as dialect delimiters.
9546         (output_asm_insn): Print curly braces and vertical bar if escaped
9547         by % and ASSEMBLER_DIALECT defined.
9548         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
9549         * doc/tm.texi: Regenerated.
9551 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9553         * config/mips/mips.c: Include tree-pass.h.
9554         (mips_reorg): Split in pre- and post-dbr_schedule parts.
9555         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
9556         (pass_mips_machine_reorg2): New machine specific pass.
9557         (insert_pass_mips_machine_reorg2): New pass plugin definition.
9558         (mips_option_override): Register the new pass.
9559         * rtl.h (cleanup_barriers): Remove prototype.
9560         (dbr_schedule): Likewise.
9561         * jump.c (cleanup_barriers): Make static.
9562         * reorg.c (dbr_schedule): Likewise.
9564 2013-05-06  Richard Biener  <rguenther@suse.de>
9566         PR tree-optimization/57185
9567         * tree-parloops.c (add_field_for_reduction): Handle anonymous
9568         SSA names properly.
9570 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9572         PR target/57106
9573         * config/i386/i386.c (add_parameter_dependencies): Add dependence
9574         between "first_arg" and "insn", not "last" and "insn".
9576 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9578         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
9579         (find_candidates_in_block): Re-enable slsr_process_phi.
9580         (create_phi_basis): Fix double counting of candidate adjustment.
9582 2013-05-06  Richard Biener  <rguenther@suse.de>
9584         PR middle-end/57147
9585         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
9586         the edge is also fallthru, preserve it and just clear the
9587         abnormal flag.
9588         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
9589         also complex, preserve that and just clear the fallthru flag.
9590         * tree-inline.c (update_ssa_across_abnormal_edges): Also
9591         update virtual operands.
9593 2013-05-06  Alan Modra  <amodra@gmail.com>
9595         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
9596         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
9597         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
9598         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
9599         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9600         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
9602 2013-05-06  Alan Modra  <amodra@gmail.com>
9604         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
9605         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
9606         (DEFAULT_ASM_ENDIAN): Define.
9607         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9608         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
9609         Update -K PIC clause from sysv4.h.
9610         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
9611         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
9613 2013-05-06  Alan Modra  <amodra@gmail.com>
9615         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
9616         twice for little-endian.
9617         (ashrdi3_no_power, ashrdi3): Support little-endian.
9619 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9621         PR target/55303
9622         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
9623         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
9624         related expanders.
9625         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
9626         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
9627         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
9628         New predicates.
9630 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
9631             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
9633         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
9634         * config/pa/pa.opt: Make mbig-switch a no-op.
9635         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
9636         (CASE_VECTOR_MODE): Always return SImode.
9637         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
9638         for the !TARGET_BIG_SWITCH case.
9639         * config/pa/pa-linux.h: Likewise.
9640         * config/pa/pa-openbsd.h: Likewise.
9641         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
9642         * config/pa/pa.md (short_jump): Remove define_insn.
9643         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
9644         (casesi0): Remove define_insn.
9645         (type): Remove btable_branch.
9646         (pa_combine_type): Likewise.
9647         (in_nullified_branch_delay): Likewise.
9648         (in_call_delay): Likewise.
9649         (define_delay): Likewise.
9650         (define_insn_reservation "Z3"): Likewise.
9651         (define_insn_reservation "Z4"): Likewise.
9652         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
9653         (pa_adjust_insn_length): Remove adjustment for btable branches.
9654         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
9655         and mno-big-switch
9657 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
9659         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9660         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
9661         Add m->r,x alternatives.
9662         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
9663         splitters using SWI48x mode iterator.
9664         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
9665         TARGET_64BIT.  Add m->x alternative.
9666         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
9667         Add o->x alternative.  Enable for TARGET_SSE.
9668         (sse_storeq): Remove expander.
9669         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
9670         with memory input operand.
9671         (*vec_extractv2di_1 splitter): New.
9672         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
9673         * config/i386/i386.md (ssevecmodelower): New mode attribute.
9675 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9677         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
9678         (INT_LOWPART): Delete.
9679         (extract_MB): Adjust.
9680         (extract_ME): Adjust.
9681         (print_operand): Adjust.
9683 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9685         * config/rs6000/predicates.md (reg_or_add_cint_operand,
9686         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9687         (reg_or_logical_cint_operand, easy_fp_constant,
9688         logical_const_operand): Delete "CONST_DOUBLE" case.
9689         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
9690         "HOST_BITS_PER_WIDE_INT == 64" test.
9691         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
9692         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
9693         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
9694         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
9695         test.
9696         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
9697         CONST_DOUBLE DImode/VOIDmode case.
9698         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
9699         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9700         CONST_DOUBLE VOIDmode case.
9701         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9702         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
9703         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
9704         Delete CONST_DOUBLE case.
9705         (splitters for mov FMOVE64 const_double): Delete
9706         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9707         "HOST_BITS_PER_WIDE_INT >= 64" test.
9708         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
9709         case.
9710         (mov DI const_double): Delete.
9712 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
9714         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
9715         on op shows all bits zero in mode of a lowpart subreg, return zero.
9717 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
9719         PR target/57150
9720         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
9721         to save TFmode registers and DImode to save TImode registers for
9722         caller save operations.
9723         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
9724         mark being partially clobbered since they only use the first
9725         double word.
9727         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
9728         and TDmode only use the upper 64-bits of each VSX register.
9730 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9732         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
9733         (find_candidates_in_block): Disable slsr_process_phi.
9735 2013-05-03  Guozhi Wei  <carrot@google.com>
9737         * coverage.c (coverage_obj_init): Move the construction of gcov
9738         constructor to ...
9739         (build_init_ctor): ... here.
9741 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9743         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
9744         (slsr_cand_d): Redefine def_phi.
9745         (stride_status, phi_adjust_status, count_phis_status): New enums.
9746         (find_phi_def): New.
9747         (find_basis_for_base_expr): New.
9748         (find_basis_for_candidate): Handle hidden bases.
9749         (alloc_cand_and_find_basis): Handle phi candidates.
9750         (slsr_process_phi): New.
9751         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
9752         (create_mul_imm_cand): Likewise.
9753         (create_add_ssa_cand): Exclude phi base candidates.
9754         (create_add_imm_cand): Likewise.
9755         (slsr_process_cast): Likewise.
9756         (slsr_process_copy): Likewise.
9757         (find_candidates_in_block): Handle phi candidates.
9758         (dump_candidate): Likewise.
9759         (unconditional_cands): Delete.
9760         (unconditional_cands_with_known_stride_p): Delete.
9761         (phi_dependent_cand_p): New.
9762         (cand_increment): Handle phi-dependent candidates.
9763         (replace_dependent): Delete.
9764         (replace_mult_candidate): New.
9765         (replace_unconditional_candidate): New.
9766         (incr_vec_index): Move to avoid forward reference.
9767         (create_add_on_incoming_edge): New.
9768         (create_phi_basis): New.
9769         (replace_dependents): Delete.
9770         (replace_conditional_candidate): New.
9771         (phi_add_costs): New.
9772         (replace_uncond_cands_and_profitable_phis): New.
9773         (record_increment): Handle phi adjustments.
9774         (record_phi_increments): New.
9775         (record_increments): Handle phi adjustments.
9776         (phi_incr_cost): New.
9777         (lowest_cost_path): Handle phis.
9778         (total_savings): Likewise.
9779         (analyze_increments): Likewise.
9780         (ncd_with_phi): New.
9781         (ncd_of_cand_and_phis): New.
9782         (nearest_common_dominator_for_cands): Handle phi increments.
9783         (all_phi_incrs_profitable): New.
9784         (replace_profitable_candidates): Handle phi-dependent candidates.
9785         (analyze_candidates_and_replace): Likewise.
9787 2013-05-03  Teresa Johnson  <tejohnson@google.com>
9789         PR bootstrap/57154
9790         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
9791         do not exceed REG_BR_PROB_BASE.
9793 2013-05-03  Jeff Law  <law@redhat.com>
9795         PR tree-optimization/57144
9796         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
9797         operand of the condition will bit into the new type when eliminating
9798         a cast feeding a condition.
9800 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
9802         PR rtl-optimization/57130
9803         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
9804         of COMPARE as in_code to the recursive call if needed.
9806 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
9808         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
9809         (enabled): Handle new members.
9810         * config/i386/sse.md (*vec_concatv2si): Merge from
9811         *vec_concatv2si_sse2 and vec_concatv2si_sse.
9812         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
9814 2013-05-03  Joern Rennecke <joern.rennecke@embecosm.com>
9816         PR tree-optimization/57027
9817         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
9818         for fnms opportunity, check we got the prerequisite kind
9819         of tree / gimple before using accessor functions.
9821 2013-05-03  Richard Biener  <rguenther@suse.de>
9823         * double-int.h (lshift): New overload without precision
9824         and arith argument.
9825         (operator *=, operator +=, operator -=): Move ...
9826         * double-int.c (operator *=, operator +=, operator -=): ... here
9827         and implement more efficiently.
9828         (mul_double_with_sign): Remove.
9829         (lshift_double): Adjust to take unsinged shift argument, push
9830         dispatching code to callers.
9831         (mul_double_wide_with_sign): Add early out for callers that
9832         are not interested in high parts or overflow.
9833         (lshift): New function.
9834         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
9835         dispatch code here.
9836         (lrotate, rrotate): Use logical shifts.
9837         * expr.c (get_inner_reference): Use lshift.
9838         * fixed-value.c (do_fixed_divide): Likewise.
9839         * tree-dfa.c (get_ref_base_and_extent): Likewise.
9840         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
9841         (indirect_refs_may_alias_p): Likewise.
9842         (stmt_kills_ref_p_1): Likewise.
9844 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9846         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
9848 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9850         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
9851         scalar form of FABD instruction.
9853 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
9855         * lra-constraints.c (process_alt_operands): Add checking alt
9856         number to choose the best alternative.
9858 2013-05-02  Richard Biener  <rguenther@suse.de>
9860         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9861         bitmap and its handling.
9862         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9864 2013-05-02  Richard Biener  <rguenther@suse.de>
9866         PR middle-end/57140
9867         * tree-inline.c (copy_loops): Properly handle removed loops.
9868         (copy_cfg_body): Mark destination loops for fixup if source
9869         loops needed fixup.
9871 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
9873         PR target/56732
9874         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9875         generating simple_return for naked functions.
9877 2013-05-02  Martin Jambor  <mjambor@suse.cz>
9879         PR middle-end/56988
9880         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9881         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9882         flags match.
9883         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9884         ipa_agg_replacement_value structures.
9885         (known_aggs_to_agg_replacement_list): Likewise.
9886         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9887         (read_agg_replacement_chain): Likewise.
9888         (ipcp_transform_function): Also check that by_ref flags match.
9890 2013-05-02  Richard Biener  <rguenther@suse.de>
9892         * graphds.h (struct graph): Add obstack member.
9893         * graphds.c (new_graph): Initialize obstack and allocate
9894         vertices from it.
9895         (add_edge): Allocate edge from the obstack.
9896         (free_graph): Free the obstack instead of all edges and vertices.
9898 2013-05-02  Teresa Johnson  <tejohnson@google.com>
9900         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9901         divides.
9902         * cfg.c (update_bb_profile_for_threading): Ditto.
9903         * tree-inline.c (copy_bb): Ditto.
9904         (copy_edges_for_bb): Ditto.
9905         (initialize_cfun): Ditto.
9906         (copy_cfg_body): Ditto.
9907         (expand_call_inline): Ditto.
9908         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9909         (estimate_node_size_and_time): Ditto.
9910         (inline_merge_summary): Ditto.
9911         * cgraphclones.c (cgraph_clone_edge): Ditto.
9912         (cgraph_clone_node): Ditto.
9913         * sched-rgn.c (compute_dom_prob_ps): Ditto.
9914         (compute_trg_info): Ditto.
9916 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9918         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9919         S reg when fp attribute set.
9920         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9922 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9924         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9925         New pattern.
9926         (*and_one_cmplsi3_compare0_uxtw): Likewise.
9927         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9928         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9930 2013-05-02  Richard Biener  <rguenther@suse.de>
9932         * tree-scalar-evolution.c (scev_info_hasher): Remove.
9933         (struct instantiate_cache_entry): New type.
9934         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9935         (struct instantiate_cache_type): New type.
9936         (set_instantiated_value, get_instantiated_value): Remove.
9937         (get_instantiated_value_entry): New function.
9938         (instantiate_scev_name): Use the new cache and adjust.
9939         (instantiate_scev_poly): Adjust.
9940         (instantiate_scev_binary): Likewise.
9941         (instantiate_array_ref): Likewise.
9942         (instantiate_scev_convert): Likewise.
9943         (instantiate_scev_not): Likewise.
9944         (instantiate_scev_3): Likewise.
9945         (instantiate_scev_2): Likewise.
9946         (instantiate_scev_r): Likewise.
9947         (instantiate_scev): Likewise.
9948         (resolve_mixers): Likewise.
9950 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
9952         PR target/57091
9953         * lra-constraints.c (best_small_class_operands_num): Remove.
9954         (process_alt_operands): Remove small_class_operands_num.  Take
9955         small classes operands into losers and only if the operand is not
9956         matched.  Modify debugging output.
9957         (curr_insn_transform): Remove best_small_class_operands_num.
9958         Print insn name.
9960 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9962         * config/aarch64/aarch64-builtins.c
9963         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
9964         * config/aarch64/aarch64-simd-builtins.def
9965         (reduc_splus_): Add new modes.
9966         (reduc_uplus_): New.
9967         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
9968         (reduc_uplus_v4sf): Likewise.
9969         (reduc_splus_v4sf): Likewise.
9970         (aarch64_addv<mode>): Likewise.
9971         (reduc_uplus_<mode>): Likewise.
9972         (reduc_splus_<mode>): Likewise.
9973         (aarch64_addvv2di): Likewise.
9974         (reduc_uplus_v2di): Likewise.
9975         (reduc_splus_v2di): Likewise.
9976         (aarch64_addvv2si): Likewise.
9977         (reduc_uplus_v2si): Likewise.
9978         (reduc_splus_v2si): Likewise.
9979         (reduc_<sur>plus_<mode>): New.
9980         (reduc_<sur>plus_v2di): Likewise.
9981         (reduc_<sur>plus_v2si): Likewise.
9982         (reduc_<sur>plus_v4sf): Likewise.
9983         (aarch64_addpv4sf): Likewise.
9984         * config/aarch64/arm_neon.h
9985         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
9986         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
9987         add UNSPEC_SADDV, UNSPEC_UADDV.
9988         (SUADDV): New.
9989         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
9991 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9993         * config/aarch64/arm_neon.h
9994         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
9996 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
9998         * config/aarch64/aarch64-builtins
9999         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10001 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10003         * config/aarch64/aarch64-simd-builtins.def
10004         (reduc_smax_): New.
10005         (reduc_smin_): Likewise.
10006         (reduc_umax_): Likewise.
10007         (reduc_umin_): Likewise.
10008         (reduc_smax_nan_): Likewise.
10009         (reduc_smin_nan_): Likewise.
10010         (fmax): Remove.
10011         (fmin): Likewise.
10012         (smax): Update for V2SF, V4SF and V2DF modes.
10013         (smin): Likewise.
10014         (smax_nan): New.
10015         (smin_nan): Likewise.
10016         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10017         (<su><maxmin><mode>3): ...This, refactor.
10018         (s<maxmin><mode>3): New.
10019         (<maxmin_uns><mode>3): Likewise.
10020         (reduc_<maxmin_uns>_<mode>): Refactor.
10021         (reduc_<maxmin_uns>_v4sf): Likewise.
10022         (reduc_<maxmin_uns>_v2si): Likewise.
10023         (aarch64_<fmaxmin><mode>: Remove.
10024         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10025         new builtin names.
10026         (vmin<q>_f<32,64>): Likewise.
10027         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10028         (FMAXMIN): New.
10029         (su): Add mappings for smax, smin, umax, umin.
10030         (maxmin): New.
10031         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10032         (FMAXMIN): Rename as...
10033         (FMAXMIN_UNS): ...This.
10034         (maxminv): Remove.
10035         (fmaxminv): Likewise.
10036         (fmaxmin): Likewise.
10037         (maxmin_uns): New.
10038         (maxmin_uns_op): Likewise.
10040 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10042         * config/aarch64/arm_neon.h
10043         (vac<ge, gt><sd>_f<32, 64>): Rename to...
10044         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10045         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10047 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10049         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10050         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10052 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10054         * config/aarch64/aarch64-simd.md
10055         (vcond<mode>_internal): Handle special cases for constant masks.
10056         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10057         (vcondu<mode><mode>): Likewise.
10058         (vcond<v_cmp_result><mode>): New.
10060 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10062         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10063         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10064         * config/aarch64/aarch64-simd-builtins.def
10065         (cmeq): Update to BUILTIN_VALLDI.
10066         (cmgt): Likewise.
10067         (cmge): Likewise.
10068         (cmle): Likewise.
10069         (cmlt): Likewise.
10070         * config/aarch64/arm_neon.h
10071         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10072         to builtins or C as appropriate.
10074 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10076         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10077         (cmgeu): ...This.
10078         (cmhi): Rename to...
10079         (cmgtu): ...This.
10080         * config/aarch64/aarch64-simd.md
10081         (simd_mode): Add SF.
10082         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10083         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10084         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10085         (cstore<mode>_neg): ...This.
10086         * config/aarch64/iterators.md
10087         (VALLF): new.
10088         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10089         (COMPARISONS): New.
10090         (UCOMPARISONS): Likewise.
10091         (optab): Add missing comparisons.
10092         (n_optab): New.
10093         (cmp_1): Likewise.
10094         (cmp_2): Likewise.
10095         (CMP): Likewise.
10096         (cmp): Remove.
10097         (VCMP_S): Likewise.
10098         (VCMP_U): Likewise.
10099         (V_cmp_result): Add DF, SF modes.
10100         (v_cmp_result): Likewise.
10101         (v): Likewise.
10102         (vmtype): Likewise.
10103         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10105 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
10107         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10108         define_insn to define_insn_and_split.
10109         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10110         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10111         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10112         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10113         (thumb2_negscc): Likewise.
10115 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10117         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10119 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10121         * config/arm/thumb2.md: Remove trailing whitespaces.
10123 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
10125         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
10126         Use gen_int_mode rather than GEN_INT.
10128 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
10130         * value-prof.c (stream_in_histogram_value): Remove the strayed
10131         debug_gimple_stmt.
10133 2013-04-30  Richard Biener  <rguenther@suse.de>
10135         PR middle-end/57122
10136         * cfghooks.c (split_edge): Properly check for the loop latch edge.
10138 2013-04-30  Richard Biener  <rguenther@suse.de>
10140         PR middle-end/57107
10141         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
10143 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
10145         PR rtl-optimization/56957
10146         PR rtl-optimization/57105
10147         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
10148         variable.  Use just INSN_UID for determining whether an insn
10149         should be only disconnected from the insn stream.
10150         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
10152 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
10154         PR tree-optimization/57104
10155         * tsan.c (instrument_expr): Don't instrument accesses to
10156         DECL_HARD_REGISTER VAR_DECLs.
10158 2013-04-30  Richard Biener  <rguenther@suse.de>
10160         * function.h (loops_for_fn): New inline function.
10161         (set_loops_for_fn): Likewise.
10162         * cfgloop.h (place_new_loop): Add struct function parameter.
10163         (get_loop): Likewise.
10164         (get_loops): Likewise.
10165         (number_of_loops): Likewise.
10166         (fel_next): Adjust.
10167         (fel_init): Likewise.
10168         * cfg.c (get_loop_copy): Adjust.
10169         * cfgloop.c (flow_loops_dump): Likewise.
10170         (record_loop_exits): Likewise.
10171         (verify_loop_structure): Likewise.
10172         * cfgloopanal.c (mark_irreducible_loops): Likewise.
10173         (estimate_reg_pressure_cost): Likewise.
10174         (mark_loop_exit_edges): Likewise.
10175         * cfgloopmanip.c (place_new_loop): Likewise.
10176         (add_loop): Likewise.
10177         (duplicate_loop): Likewise.
10178         * graph.c (draw_cfg_nodes): Likewise.
10179         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
10180         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
10181         (extract_affine_chrec): Likewise.
10182         (build_scop_iteration_domain): Likewise.
10183         * graphite.c (graphite_initialize): Likewise.
10184         * ira-build.c (create_loop_tree_nodes): Likewise.
10185         (more_one_region_p): Likewise.
10186         (rebuild_regno_allocno_maps): Likewise.
10187         (mark_loops_for_removal): Likewise.
10188         (mark_all_loops_for_removal): Likewise.
10189         (remove_unnecessary_regions): Likewise.
10190         (ira_build): Likewise.
10191         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
10192         * loop-init.c (fix_loop_structure): Likewise.
10193         (gate_rtl_move_loop_invariants): Likewise.
10194         (gate_rtl_unswitch): Likewise.
10195         (gate_rtl_unroll_and_peel_loops): Likewise.
10196         (rtl_doloop): Likewise.
10197         * lto-streamer-in.c (input_cfg): Likewise.
10198         * lto-streamer-out.c (output_cfg): Likewise.
10199         * modulo-sched.c (sms_schedule): Likewise.
10200         * predict.c (tree_estimate_probability): Likewise.
10201         (tree_estimate_probability_driver): Likewise.
10202         (estimate_loops): Likewise.
10203         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
10204         (move_sese_region_to_fn): Likewise.
10205         (debug_loop_num): Likewise.
10206         * tree-chrec.c (chrec_evaluate): Likewise.
10207         (hide_evolution_in_other_loops_than_loop): Likewise.
10208         (chrec_component_in_loop_num): Likewise.
10209         (reset_evolution_in_loop): Likewise.
10210         (evolution_function_is_invariant_rec_p): Likewise.
10211         * tree-if-conv.c (main_tree_if_conversion): Likewise.
10212         * tree-inline.c (copy_loops): Likewise.
10213         (copy_cfg_body): Likewise.
10214         (tree_function_versioning): Likewise.
10215         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
10216         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
10217         Likewise.
10218         (add_to_evolution_1): Likewise.
10219         (scev_const_prop): Likewise.
10220         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
10221         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
10222         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
10223         (tree_ssa_lim_initialize): Likewise.
10224         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
10225         (verify_loop_closed_ssa): Likewise.
10226         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
10227         (tree_ssa_loop_im): Likewise.
10228         (tree_ssa_loop_unswitch): Likewise.
10229         (tree_vectorize): Likewise.
10230         (check_data_deps): Likewise.
10231         (tree_ssa_loop_ivcanon): Likewise.
10232         (tree_ssa_loop_bounds): Likewise.
10233         (tree_complete_unroll): Likewise.
10234         (tree_complete_unroll_inner): Likewise.
10235         (tree_parallelize_loops): Likewise.
10236         (tree_ssa_loop_prefetch): Likewise.
10237         (tree_ssa_loop_ivopts): Likewise.
10238         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
10239         * tree-vectorizer.c (vectorize_loops): Likewise.
10241 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
10243         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
10244         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
10245         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
10246         with EABI_LINK_SPEC.
10248 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10250         PR target/44578
10251         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
10252         alternative.
10254 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
10256         PR target/57097
10257         * lra-constraints.c (process_alt_operands): Discourage a bit more
10258         using memory for pseudos.  Print cost dump for alternatives.
10259         Modify cost values for conflicts with early clobbers.
10260         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
10262 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10264         PR target/57098
10265         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
10267 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
10269         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
10270         from/to S register.
10271         (movdi_aarch64): Support LDR/STR from/to D register.
10273 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
10275         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
10276         or higher by default.
10278 2013-04-29  Richard Biener  <rguenther@suse.de>
10280         PR middle-end/57075
10281         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
10282         even if not adding abnormal edges for calls that can make
10283         abnormal gotos.
10285 2013-04-29  Richard Biener  <rguenther@suse.de>
10287         PR middle-end/57103
10288         * tree-cfg.c (move_stmt_op): Fix condition under which to update
10289         TREE_BLOCK.
10290         (move_stmt_r): Remove redundant checking.
10292 2013-04-29  Teresa Johnson  <tejohnson@google.com>
10294         PR bootstrap/57077
10295         * basic-block.h (apply_scale): New function.
10296         (apply_probability): Use apply_scale.
10297         * gimple-streamer-in.c (input_bb): Ditto.
10298         * lto-streamer-in.c (input_cfg): Ditto.
10299         * lto-cgraph.c (merge_profile_summaries): Ditto.
10300         * tree-optimize.c (execute_fixup_cfg): Ditto.
10301         * tree-inline.c (copy_bb): Update comment to use apply_scale.
10302         (copy_edges_for_bb): Ditto.
10303         (copy_cfg_body): Ditto.
10305 2013-04-29  Tom de Vries  <tom@codesourcery.com>
10307         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
10308         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
10309         (tail_merge_optimize): Handle current_loops == NULL.
10311 2013-04-26  Jeff Law  <law@redhat.com>
10313         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
10314         (simplify_cond_using_ranges): Generalize code to simplify
10315         COND_EXPRs where one argument is a constant and the other
10316         is an SSA_NAME created by an integral type conversion.
10318 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10320         * config/arm/arm.md (store_minmaxsi): Use only when
10321         optimize_insn_for_size_p.
10323 2013-04-29  Christian Bruel  <christian.bruel@st.com>
10325         PR target/57108
10326         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
10328 2013-04-29  Richard Biener  <rguenther@suse.de>
10330         PR middle-end/57089
10331         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
10332         loop tree make sure to schedule a fixup for the child as well.
10333         (expand_omp_for_generic): Properly add loops.
10334         (expand_omp_for_static_nochunk): Likewise.
10335         (expand_omp_for_static_chunk): Likewise.
10336         (expand_omp_for): For the degenerate case fixup loops.
10337         (expand_omp_sections): Fix default bb placement in loops.
10338         (expand_omp_atomic_pipeline): Properly add loops.
10340 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10342         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
10344 2013-04-29  Tom de Vries  <tom@codesourcery.com>
10346         * tree-ssa-tail-merge.c: Update header comment.
10348 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10350         * config/aarch64/arm_neon.h
10351         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
10352         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
10353         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
10354         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
10355         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
10356         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
10357         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
10358         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
10360 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10362         * config/aarch64/aarch64-simd.md
10363         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
10364         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
10365         fix_trunc, fixuns_trunc.
10366         (ftrunc<VDQF:mode>2): New.
10367         * config/aarch64/iterators.md (optab): Add fix, fixuns.
10368         (fix_trunc_optab): New.
10370 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10372         * config/aarch64/aarch64-builtins.c
10373         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
10374         iceilf, lround, iroundf.
10376 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10378         PR target/54349
10379         * config/i386/i386.h (enum ix86_tune_indices)
10380         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
10381         New, split from X86_TUNE_INTER_UNIT_MOVES.
10382         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
10383         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
10384         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
10385         (TARGET_INTER_UNIT_MOVES): Remove.
10386         * config/i386/i386.c (initial_ix86_tune_features): Update.
10387         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
10388         (ix86_expand_convert_uns_didf_sse): Use
10389         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10390         (ix86_expand_vector_init_one_nonzero): Ditto.
10391         (ix86_expand_vector_init_interleave): Ditto.
10392         (inline_secondary_memory_needed): Return true for moves from SSE class
10393         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
10394         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
10395         * config/i386/constraints.md (Yi, Ym): Depend on
10396         TARGET_INTER_UNIT_MOVES_TO_VEC.
10397         (Yj, Yn): New constraints.
10398         * config/i386/i386.md (*movdi_internal): Change constraints of
10399         operand 1 from Yi to Yj and from Ym to Yn.
10400         (*movsi_internal): Ditto.
10401         (*movdf_internal): Ditto.
10402         (*movsf_internal): Ditto.
10403         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
10404         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10405         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
10406         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
10407         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
10408         * config/i386/sse.md (movdi_to_sse): Ditto.
10409         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
10410         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
10411         TARGET_INTER_UNIT_MOVES.
10412         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
10413         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
10414         instead of TARGET_INTER_UNIT_MOVES.
10415         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
10416         operand 1 from Yi to Yj and from Ym to Yn.
10418 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10420         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
10421         (float_truncate_hi_): Likewise.
10422         (float_extend_lo_): Likewise.
10423         (float_truncate_lo_): Likewise.
10424         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
10425         (aarch64_float_extend_lo_v2df): Likewise.
10426         (vec_unpacks_hi_v4sf): Likewise.
10427         (aarch64_float_truncate_lo_v2sf): Likewise.
10428         (aarch64_float_truncate_hi_v4sf): Likewise.
10429         (vec_pack_trunc_v2df): Likewise.
10430         (vec_pack_trunc_df): Likewise.
10432 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10434         * config/aarch64/aarch64-builtins.c
10435         (aarch64_fold_builtin): Fold float conversions.
10436         * config/aarch64/aarch64-simd-builtins.def
10437         (floatv2si, floatv4si, floatv2di): New.
10438         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
10439         * config/aarch64/aarch64-simd.md
10440         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
10441         * config/aarch64/iterators.md (FLOATUORS): New.
10442         (optab): Add float, floatuns.
10443         (su_optab): Likewise.
10445 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10447         * config/aarch64/aarch64-builtins.c
10448         (aarch64_builtin_vectorized_function): Use new names for
10449         fcvt builtins.
10450         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
10451         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
10452         (fcvtzu): Split as...
10453         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
10454         (fcvtas): Split as...
10455         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
10456         (fcvtau): Split as...
10457         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
10458         (fcvtps): Split as...
10459         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
10460         (fcvtpu): Split as...
10461         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
10462         (fcvtms): Split as...
10463         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
10464         (fcvtmu): Split as...
10465         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
10466         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
10467         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
10468         (lfrintnusf, lfrintnudf): Likewise.
10469         * config/aarch64/aarch64-simd.md
10470         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
10471         define_insn.
10472         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
10473         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
10474         (fcvt_pattern): Likewise.
10476 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10478         * config/aarch64/aarch64-simd.md
10479         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
10480         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
10482 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10484         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
10485         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
10486         (vrnd<a,m,n,p>_f32): Implement using builtins.
10487         (vrnd<i,x><q>_f<32, 64>): New.
10489 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10491         * config/aarch64/aarch64-builtins.c
10492         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
10493         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
10494         (frintz): Rename to...
10495         (btrunc): ...this.
10496         (frintp): Rename to...
10497         (ceil): ...this.
10498         (frintm): Rename to...
10499         (floor): ...this.
10500         (frinti): Rename to...
10501         (nearbyint): ...this.
10502         (frintx): Rename to...
10503         (rint): ...this.
10504         (frinta): Rename to...
10505         (round): ...this.
10506         * config/aarch64/aarch64-simd.md
10507         (aarch64_frint<frint_suffix><mode>): Delete.
10508         (<frint_pattern><mode>2): Convert to insn.
10509         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
10510         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
10511         (frint_pattern): Likewise.
10512         (frint_suffix): Likewise.
10514 2013-04-29  Richard Biener  <rguenther@suse.de>
10516         PR tree-optimization/57081
10517         * loop-init.c: Include tree-flow.h.
10518         (loop_optimizer_finalize): Free number of iteration estimates.
10519         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
10521 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
10523         PR tree-optimization/57083
10524         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
10525         non-singleton shift count range, zero extend low_bound for uns case.
10527         * config/i386/predicates.md (general_vector_operand): New predicate.
10528         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
10529         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
10530         if they aren't nonimmediate operands.  If their original values
10531         satisfy const_vector_equal_evenodd_p, don't shift them.
10532         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
10533         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
10534         if not nonimmediate_operand.
10535         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
10536         instead of register_operand.
10537         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
10539 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
10541         * stor-layout.c (finalize_size_functions): Allocate a structure and
10542         reset cfun before dumping the functions.
10544 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
10546         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
10548         PR target/56866
10549         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
10550         use xop_pmacsdqh if uns_p.
10551         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
10552         the immediate rotate count.
10554 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
10556         * rtl.h (struct rtx_def): Add comment for field jump.
10557         (LRA_SUBREG_P): New macro.
10558         * recog.c (register_operand): Check LRA_SUBREG_P.
10559         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10560         stack frame.
10561         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10562         (lra_final_code_change): Skip subreg change for operators.
10563         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10564         if there are no operand changes.
10565         * lra-constraints.c (curr_insn_set): New.
10566         (match_reload): Set LRA_SUBREG_P.
10567         (emit_spill_move): Ditto.
10568         (check_and_process_move): Use curr_insn_set. Process only single
10569         set insns.  Don't initialize sec_mem_p and change_p.
10570         (simplify_operand_subreg): Use LRA_SUBREG_P.
10571         (reg_in_class_p): New function.
10572         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10573         of #ifdef.  Add code to remove cycling.
10574         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10575         non-null disp.  Reload inner instead of disp when base and index
10576         are null.  Try to put lo_sum into register.
10577         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10578         (check_and_process_move): Move code for move cost check to
10579         simple_move_p.  Remove equiv_substitution.
10580         (simple_move_p): New function.
10581         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10582         curr_insn_set.  Call check_and_process_move only for single set
10583         insns.  Use the new function.  Move call of check_and_process_move
10584         after operand equiv substitution and address process.
10586 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10588         PR go/57045
10589         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
10590         with nonlocal goto receivers or returns twice calls, ignore
10591         unininitialized values from abnormal edges to nl goto receiver
10592         or returns twice call.
10594 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10596         PR tree-optimization/57051
10597         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
10598         and VEC_RSHIFT_EXPR if shift count is a multiple of element
10599         bitsize.
10601 2013-04-26  Richard Biener  <rguenther@suse.de>
10603         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
10604         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
10605         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
10606         (fixup_loop_arrays_after_move): New function.
10607         (move_sese_region_to_fn): Properly outline the loop tree parts
10608         of the SESE region.
10610 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
10612         * config/i386/i386.md (type, unit): Fix long lines.
10614 2013-04-26  Richard Biener  <rguenther@suse.de>
10616         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
10617         (lto-streamer-out.o): Likewise.
10618         * cfgloop.c (init_loops_structure): Export, add struct function
10619         argument and adjust.
10620         (flow_loops_find): Adjust.
10621         * cfgloop.h (enum loop_estimation): Add EST_LAST.
10622         (init_loops_structure): Declare.
10623         * lto-streamer-in.c: Include cfgloop.h.
10624         (input_cfg): Input the loop tree.
10625         * lto-streamer-out.c: Include cfgloop.h.
10626         (output_cfg): Output the loop tree.
10627         (output_struct_function_base): Do not drop PROP_loops.
10629 2013-03-26  Richard Biener  <rguenther@suse.de>
10631         * tree-cfg.c (execute_build_cfg): Build the loop tree.
10632         (pass_build_cfg): Provide PROP_loops.
10633         (move_sese_region_to_fn): Remove loops that are outlined into fn
10634         for now.
10635         * tree-inline.c: Include cfgloop.h.
10636         (initialize_cfun): Do not drop PROP_loops.
10637         (copy_loops): New function.
10638         (copy_cfg_body): Copy loop structure.
10639         (tree_function_versioning): Initialize destination loop tree.
10640         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
10641         (pass_parallelize_loops): Do IL verification.
10642         * loop-init.c (loop_optimizer_init): Fixup loops if required.
10643         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
10644         the CFG make sure we fixup loops as well.
10645         * tree-ssa-tail-merge.c: Include cfgloop.h.
10646         (replace_block_by): When merging loop latches mark loops for fixup.
10647         * lto-streamer-out.c (output_struct_function_base): Drop
10648         PROP_loops for now.
10649         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
10650         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
10651         * ipa-split.c: Include cfgloop.h.
10652         (split_function): Add the new return block to the loop tree root.
10653         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
10654         whether we have removed the forwarder block.
10655         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
10656         * cfgloop.h (place_new_loop): Declare.
10657         * cfgloopmanip.c (place_new_loop): Export.
10658         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
10659         (tree-switch-conversion.o): Likewise.
10660         (tree-complex.o): Likewise.
10661         (tree-inline.o): Likewise.
10662         (tree-ssa-tailmerge.o): Likewise.
10663         (ipa-split.o): Likewise.
10664         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
10665         (tree-ssa-copy.o): Likewise.
10666         * tree-switch-conversion.c: Include cfgloop.h
10667         (process_switch): If we emit a bit-test cascade, schedule loops
10668         for fixup.
10669         * tree-complex.c: Include cfgloop.h.
10670         (expand_complex_div_wide): Properly add new basic-blocks to loops.
10671         * asan.c: Include cfgloop.h.
10672         (create_cond_insert_point): Properly add new basic-blocks to
10673         loops, schedule loop fixup.
10674         * cfgloop.c (verify_loop_structure): Check that looks are not
10675         marked for fixup.
10676         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
10677         to loops.
10678         (expand_omp_for_generic): Likewise.
10679         (expand_omp_sections): Likewise.
10680         (expand_omp_atomic_pipeline): Schedule loops for fixup.
10681         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
10682         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
10683         is initialized, not when loops are present.
10684         * tree-parloops.c (parallelize_loops): Remove checking here.
10685         * passes.c (init_optimization_passes): Schedule a copy-propagation
10686         pass before complete unrolling of inner loops.
10688 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10690         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
10691         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
10692         (colorize_init): Add argument to _WIN32 version.
10693         * toplev.c: Include diagnostic-color.h.
10694         (process_options): Default to -fdiagnostics-color=auto if
10695         GCC_COLORS env var is in the environment.
10696         * common.opt (fdiagnostics-color=): Add Var and Init.
10697         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
10698         env var is in the environment, the default is auto rather than never.
10700         * diagnostic.h (file_name_as_prefix): Add context argument.
10701         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
10702         the string as locus.
10703         * langhooks.c (lhd_print_error_function): Adjust caller.
10705 2013-04-25  Lawrence Crowl  <crowl@google.com>
10707         * var-tracking.c (shared_hash_def::htab):
10708         Change type to hash_table.  Update dependent calls and types.
10710 2013-04-25  Lawrence Crowl  <crowl@google.com>
10712         * Makefile.in: Update as needed below.
10714         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
10715         Move declaration to after the type's method definitons.
10717         * attribs.c (htab_t scoped_attributes::attribute_hash):
10718         Change type to hash_table.  Update dependent calls and types.
10720         * bitmap.c (htab_t bitmap_desc_hash):
10721         Change type to hash_table.  Update dependent calls and types.
10723         * cselib.c (htab_t cselib_hash_table):
10724         Change type to hash_table.  Update dependent calls and types.
10726         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
10727         (hash_string_slot_node): Move implementation into lto-streamer.h
10728         struct string_slot_hasher.
10729         (eq_string_slot_node): Likewise.
10731         * data-streamer-out.c: Update output_block::string_hash_table
10732         dependent calls and types.
10734         * dwarf2cfi.c (htab_t trace_index):
10735         Change type to hash_table.  Update dependent calls and types.
10737         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
10738         Change type to hash_table.  Update dependent calls and types.
10739         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
10740         (htab_t optimize_external_refs::map): Likewise.
10741         (htab_t output_comp_unit::extern_map): Likewise.
10742         (htab_t output_comdat_type_unit::extern_map): Likewise.
10743         (htab_t output_macinfo::macinfo_htab): Likewise.
10744         (htab_t optimize_location_lists::htab): Likewise.
10745         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
10747         * except.c (htab_t ehspec_hash_type):
10748         Change type to hash_table.  Update dependent calls and types.
10749         (assign_filter_values::ttypes): Likewise.
10750         (assign_filter_values::ehspec): Likewise.
10751         (sjlj_assign_call_site_values::ar_hash): Likewise.
10752         (convert_to_eh_region_ranges::ar_hash): Likewise.
10754         * gcse.c (htab_t pre_ldst_table):
10755         Change type to hash_table.  Update dependent calls and types.
10757         * ggc-common.c (htab_t saving_htab):
10758         Change type to hash_table.  Update dependent calls and types.
10759         (htab_t loc_hash): Likewise.
10760         (htab_t ptr_hash): Likewise.
10761         (call_count): Rename ggc_call_count.
10762         (call_alloc): Rename ggc_call_alloc.
10763         (loc_descriptor): Rename make_loc_descriptor.
10764         (add_statistics): Rename ggc_add_statistics.
10766         * ggc-common.c (saving_htab):
10767         Change type to hash_table.  Update dependent calls and types.
10769         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
10770         (push_gimplify_context): Likewise.
10771         (pop_gimplify_context): Likewise.
10772         (struct gimple_temp_hash_elt): Added.
10773         (struct gimplify_hasher): Likewise.
10774         (struct gimplify_ctx.temp_htab):
10775         Change type to hash_table.  Update dependent calls and types.
10777         * gimple-fold.c: Include gimplify-ctx.h.
10779         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
10780         Change type to hash_table.  Update dependent calls and types.
10781         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
10782         avoid potential global name collision.
10784         * gimplify.c: Include gimplify-ctx.h.
10785         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
10786         (htab_t gimplify_ctx::temp_htab):
10787         Update dependent calls and types for new type hash_table.
10788         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
10789         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
10791         * gimplify-ctx.h: New.
10792         (struct gimple_temp_hash_elt): Move from gimplify.c.
10793         (class gimplify_hasher): New.
10794         (struct gimplify_ctx): Move from gimple.h.
10795         (htab_t gimplify_ctx::temp_htab):
10796         Change type to hash_table.  Update dependent calls and types.
10798         * graphite-clast-to-gimple.c: Include graphite-htab.h.
10799         (htab_t ivs_params::newivs_index):
10800         Change type to hash_table.  Update dependent calls and types.
10801         (htab_t ivs_params::params_index): Likewise.
10802         (htab_t print_generated_program::params_index): Likewise.
10803         (htab_t gloog::newivs_index): Likewise.
10804         (htab_t gloog::params_index): Likewise.
10806         * graphite.c: Include graphite-htab.h.
10807         4htab_t graphite_transform_loops::bb_pbb_mapping):
10808         Change type to hash_table.  Update dependent calls and types.
10810         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
10811         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
10812         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
10814         * graphite-dependences.c: Include graphite-htab.h.
10815         (loop_is_parallel_p): Change hash table type of parameter.
10817         * graphite-htab.h: New.
10818         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
10819         (extern find_pbb_via_hash): Move from graphite-poly.h.
10820         (extern loop_is_parallel_p): Move from graphite-poly.h.
10821         (extern get_loop_body_pbbs): Move from graphite-poly.h.
10823         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
10824         (extern loop_is_parallel_p): Move to graphite-htab.h.
10825         (extern get_loop_body_pbbs): Move to graphite-htab.h.
10827         * haifa-sched.c (htab_t delay_htab):
10828         Change type to hash_table.  Update dependent calls and types.
10829         (htab_t delay_htab_i2): Likewise.
10831         * ira-color.c (htab_t allocno_hard_regs_htab):
10832         Change type to hash_table.  Update dependent calls and types.
10834         * ira-costs.c (htab_t cost_classes_htab):
10835         Change type to hash_table.  Update dependent calls and types.
10837         * loop-invariant.c (htab_t merge_identical_invariants::eq):
10838         Change type to hash_table.  Update dependent calls and types.
10840         * loop-iv.c (htab_t bivs):
10841         Change type to hash_table.  Update dependent calls and types.
10843         * loop-unroll.c (htab_t opt_info::insns_to_split):
10844         Change type to hash_table.  Update dependent calls and types.
10845         (htab_t opt_info::insns_with_var_to_expand): Likewise.
10847         * lto-streamer.h (struct string_slot): Move from data-streamer.h
10848         (struct string_slot_hasher): New.
10849         (htab_t output_block::string_hash_table):
10850         Change type to hash_table.  Update dependent calls and types.
10852         * lto-streamer-in.c (freeing_string_slot_hasher): New.
10853         (htab_t file_name_hash_table):
10854         Change type to hash_table.  Update dependent calls and types.
10856         * lto-streamer-out.c: Update output_block::string_hash_table dependent
10857         calls and types.
10859         * lto-streamer.c (htab_t tree_htab):
10860         Change type to hash_table.  Update dependent calls and types.
10862         * omp-low.c: Include gimplify-ctx.h.
10864         * passes.c (htab_t name_to_pass_map):
10865         Change type to hash_table.  Update dependent calls and types.
10866         (pass_traverse): Rename to passes_pass_traverse.
10868         * plugin.c (htab_t event_tab):
10869         Change type to hash_table.  Update dependent calls and types.
10871         * postreload-gcse.c (htab_t expr_table):
10872         Change type to hash_table.  Update dependent calls and types.
10873         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10875         * sese.c (debug_rename_map_1): Make extern.
10876         (htab_t copy_bb_and_scalar_dependences::rename_map):
10877         Change type to hash_table.  Update dependent calls and types.
10879         * sese.h (extern debug_rename_map): Move to .c file.
10881         * store-motion.c (htab_t store_motion_mems_table):
10882         Change type to hash_table.  Update dependent calls and types.
10884         * trans-mem.c (htab_t tm_new_mem_hash):
10885         Change type to hash_table.  Update dependent calls and types.
10887         * tree-browser.c (htab_t TB_up_ht):
10888         Change type to hash_table.  Update dependent calls and types.
10890         * tree-cfg.c (htab_t discriminator_per_locus):
10891         Change type to hash_table.  Update dependent calls and types.
10893         * tree-complex.c: Include tree-hasher.h
10894         (htab_t complex_variable_components):
10895         Change type to hash_table.  Update dependent calls and types.
10897         * tree-eh.c (htab_t finally_tree):
10898         Change type to hash_table.  Update dependent calls and types.
10900         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10901         struct int_tree_hasher.
10902         (extern int_tree_map_eq): Likewise.
10903         (uid_decl_map_hash): Removed.
10904         (extern decl_tree_map_eq): Likewise.
10906         * tree-hasher.h: New.
10907         (struct int_tree_hasher): New.
10908         (typedef int_tree_htab_type): New.
10910         * tree-inline.c: Include gimplify-ctx.h.
10912         * tree-mudflap.c: Include gimplify-ctx.h.
10914         * tree-parloops.c: Include tree-hasher.h.
10915         (htab_t eliminate_local_variables_stmt::decl_address):
10916         Change type to hash_table.  Update dependent calls and types.
10917         (htab_t separate_decls_in_region::decl_copies): Likewise.
10919         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10920         Change type to hash_table.  Update dependent calls and types.
10922         * tree-sra.c (candidates):
10923         Change type to hash_table.  Update dependent calls and types.
10925         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10926         in tree-flow.h.
10927         (int_tree_map_hash): Likewise.
10929         * tree-ssa-dom.c (htab_t avail_exprs):
10930         Change type to hash_table.  Update dependent calls and types.
10932         * tree-ssa-live.c (var_map_base_init::tree_to_index):
10933         Change type to hash_table.  Update dependent calls and types.
10935         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10936         Change type to hash_table.  Update dependent calls and types.
10938         * tree-ssa-phiopt.c (seen_ssa_names):
10939         Change type to hash_table.  Update dependent calls and types.
10941         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
10942         Change type to hash_table.  Update dependent calls and types.
10944         * tree-ssa-uncprop.c (equiv):
10945         Change type to hash_table.  Update dependent calls and types.
10947 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
10949         PR rtl-optimization/57003
10950         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
10951         call note_stores with kill_clobbered_value callback again after
10952         killing regs_invalidated_by_call.
10954 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10956         * config/aarch64/aarch64-simd.md
10957         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
10958         (aarch64_simd_bsl<mode>): Likewise.
10959         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
10961 2013-04-25  Marek Polacek  <polacek@redhat.com>
10963         PR tree-optimization/57066
10964         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
10966 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
10968         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
10970 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10972         * config/aarch64/aarch64-builtins.c
10973         (aarch64_fold_builtin): New.
10974         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
10975         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
10976         * config/aarch64/aarch64-simd-builtins.def (abs): New.
10977         * config/aarch64/arm_neon.h
10978         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
10980 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
10981             Tejas Belagod  <tejas.belagod@arm.com>
10983         * config/aarch64/aarch64-builtins.c
10984         (aarch64_gimple_fold_builtin): New.
10985         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
10986         * config/aarch64/aarch64-simd-builtins.def (addv): New.
10987         * config/aarch64/aarch64-simd.md (addpv4sf): New.
10988         (addvv4sf): Update.
10989         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
10991 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10993         * config/aarch64/aarch64.md
10994         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
10996 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
10998         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
10999         (*ngcsi_uxtw): New pattern.
11001 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11002             Julian Brown  <julian@codesourcery.com>
11004         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11005         (TB_DREG): Add T_V4HF.
11006         (v4hf_UP): New macro.
11007         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11008         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11009         Handle initialisation of V4HF. Adjust initialisation of reinterpret
11010         built-ins.
11011         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11012         (arm_vector_mode_supported_p): Handle V4HF.
11013         (arm_mangle_map): Handle V4HFmode.
11014         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11015         * config/arm/arm_neon_builtins.def: Add entries for
11016         vcvtv4hfv4sf, vcvtv4sfv4hf.
11017         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11018         (neon_vcvtv4hfv4sf): Likewise.
11019         * config/arm/neon-gen.ml: Handle half-precision floating point
11020         features.
11021         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11022         * config/arm/arm_neon.h: Regenerate.
11023         * config/arm/neon.ml (type elts): Add F16.
11024         (type vectype): Add T_float16x4, T_floatHF.
11025         (type vecmode): Add V4HF.
11026         (type features): Add Requires_FP_bit feature.
11027         (elt_width): Handle F16.
11028         (elt_class): Likewise.
11029         (elt_of_class_width): Likewise.
11030         (mode_of_elt): Refactor.
11031         (type_for_elt): Handle F16, fix error messages.
11032         (vectype_size): Handle T_float16x4.
11033         (vcvt_sh): New function.
11034         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11035         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11036         (string_of_mode): Handle V4HF.
11037         * doc/arm-neon-intrinsics.texi: Regenerate.
11039 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11041         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11042         format specifier in 'X' case.
11044 2013-04-25  Alan Modra  <amodra@gmail.com>
11046         PR target/57052
11047         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11048         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11049         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11050         Repeat for many other rotate/shift and mask patterns using subregs.
11051         Name lshiftrt insns.
11052         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11053         on WORDS_BIG_ENDIAN.
11055 2013-04-25  Alan Modra  <amodra@gmail.com>
11057         * config.gcc: Support little-endian powerpc-linux targets.
11058         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11059         (LINK_OS_LINUX_SPEC): Define.
11060         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11061         Preserve MASK_LITTLE_ENDIAN.
11062         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11063         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11064         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11065         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11066         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11067         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11068         Correct fp word order for little-endian.  Don't shift toc entries
11069         smaller than a word for little-endian.
11070         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11071         (bswapdi2 splits): Correct low-part subreg for little-endian.
11072         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11073         low/high where such is correct only for be.
11074         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11075         little-endian for -mcall-aixdesc.
11077 2013-04-25  Alan Modra  <amodra@gmail.com>
11079         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11080         replace_equiv_address_nv.
11082 2013-04-25  Alan Modra  <amodra@gmail.com>
11084         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11086 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11088         Revert:
11089         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11090         * rtl.h (struct rtx_def): ...
11092 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11094         PR rtl-optimizations/57046
11095         * lra-constraints (split_reg): Set up lra_risky_transformations_p
11096         for multi-reg splits.
11098 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
11100         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11102 2013-04-24  Sterling Augustine  <saugustine@google.com>
11104         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11105         (comp_dir_string, debug_str_dwo_section): New.
11106         (DEBUG_STR_DWO_SECTION): Rename to ...
11107         (DEBUG_DWO_STR_SECTION): ... this.
11108         (DEBUG_NORM_STR_SECTION): Delete.
11109         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11110         (DEBUG_STR_DWO_SECTION_FLAGS): New.
11111         (find_AT_string): Move most logic to ...
11112         (find_AT_string_in_table): ... here.  New.
11113         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11114         add_skeleton_AT_string.  Delete logic.
11115         (output_skeleton_debug_sections): Remove call to
11116         add_top_level_skeleton_die_attrs.
11117         (add_comp_dir_attribute): Move logic to comp_dir_string.
11118         (dwarf2out_init): Initialize debug_str_dwo_section.
11119         (output_indirect_string): Call find_string_form.
11120         (output_indirect_strings): Rewrite.
11121         (prune_unused_types): Empty skeleton_debug_str_hash.
11122         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
11123         (dwarf2out_finish):  Call output_indirect_strings.
11125 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11127         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
11129 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11131         * rtl.h (struct rtx_def): Add comment for field jump.
11132         (LRA_SUBREG_P): New macro.
11133         * recog.c (register_operand): Check LRA_SUBREG_P.
11134         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11135         stack frame.
11136         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11137         (lra_final_code_change): Skip subreg change for operators.
11138         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11139         if there are no operand changes.
11140         * lra-constraints.c (curr_insn_set): New.
11141         (match_reload): Set LRA_SUBREG_P.
11142         (emit_spill_move): Ditto.
11143         (check_and_process_move): Use curr_insn_set. Process only single
11144         set insns.  Don't initialize sec_mem_p and change_p.
11145         (simplify_operand_subreg): Use LRA_SUBREG_P.
11146         (reg_in_class_p): New function.
11147         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11148         of #ifdef.  Add code to remove cycling.
11149         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11150         non-null disp.  Reload inner instead of disp when base and index
11151         are null.  Try to put lo_sum into register.
11152         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11153         (check_and_process_move): Move code for move cost check to
11154         simple_move_p.  Remove equiv_substitution.
11155         (simple_move_p): New function.
11156         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11157         curr_insn_set.  Call check_and_process_move only for single set
11158         insns.  Use the new function.  Move call of check_and_process_move
11159         after operand equiv substitution and address process.
11161 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
11163         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
11164         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
11165         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
11167 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11169         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
11171 2013-04-24  Marek Polacek  <polacek@redhat.com>
11173         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
11174         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
11175         (select_loops_exit_conditions): Likewise.
11176         (number_of_iterations_for_all_loops): Likewise.
11177         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
11178         (scev_analysis): Likewise.
11180 2013-04-02  Catherine Moore  <clm@codesourcery.com>
11181             Chao-ying Fu <fu@mips.com>
11183         * config/mips/micromips.md (jraddiusp): New pattern.
11184         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
11185         instruction if possible.
11187 2013-04-24  Alan Modra  <amodra@gmail.com>
11189         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
11191 2013-04-24  Julian Brown  <julian@codesourcery.com>
11192             Chung-Lin Tang  <cltang@codesourcery.com>
11194         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
11195         dependency behavior in enumeration type DIE generation. Add TODO note
11196         to comments about future DW_FORM_sdata/udata re-work of related code.
11198 2013-04-23  Lawrence Crowl  <crowl@google.com>
11200         * Makefile.in: Update as needed below.
11202         * hash-table.h (class hash_table):
11203         Correct many methods with parameter types compare_type to the correct
11204         value_type.  (Correct code was unlikely to notice the change.)
11205         (hash_table::elements_with_deleted) New.
11206         (class hashtable::iterator): New.
11207         (hashtable::begin()): New.
11208         (hashtable::end()): New.
11209         (FOR_EACH_HASH_TABLE_ELEMENT): New.
11211         * statistics.c (statistics_hashes):
11212         Change type to hash_table.  Update dependent calls and types.
11214         * tree-into-ssa.c (var_infos):
11215         Change type to hash_table.  Update dependent calls and types.
11217         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
11218         Change type to hash_table.  Update dependent calls and types.
11220         * tree-ssa-loop-im.c (struct mem_ref.refs):
11221         Change type to hash_table.  Update dependent calls and types.
11223         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
11224         Change type to hash_table.  Update dependent calls and types.
11226         * tree-ssa-sccvn.c (vn_tables_s::nary):
11227         Change type to hash_table.  Update dependent calls and types.
11228         (vn_tables_s::phis): Likewise.
11229         (vn_tables_s::references): Likewise.
11231         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
11232         (vn_reference_eq): Update parameter and return types.
11234         * tree-ssa-structalias.c (pointer_equiv_class_table):
11235         Change type to hash_table.  Update dependent calls and types.
11236         (location_equiv_class_table): Likewise.
11238         * tree-vect-data-refs.c: Consequential changes for making
11239         peeling a hash_table.
11241         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
11242         (destroy_loop_vec_info): Dependent hash_table update.
11244         * tree-vectorizer.h (peeling_htab):
11245         Change type to hash_table.  Update dependent calls and types.
11247 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
11249         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
11250         to check the register content is equal or not.
11251         * lra-constraints.c (match_reload): Use lra_assign_reg_val
11252         to assign register content record.
11253         * lra-eliminations.c (update_reg_eliminate): Use
11254         lra_update_reg_val_offset to update register content offset.
11255         * lra-int.h (struct lra_reg): Add offset member.
11256         (lra_reg_val_equal_p): New static inline function.
11257         (lra_update_reg_val_offset): New static inline function.
11258         (lra_assign_reg_val): New static inline function.
11259         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
11260         to assign register content record.
11261         (initialize_lra_reg_info_element): Initial offset to zero.
11263 2013-04-23  Catherine Moore  <clm@codesourcery.com>
11265         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
11266         operands.  Record compression.
11268 2013-04-23  Xinliang David Li  <davidxl@google.com>
11270         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
11272 2013-04-23  Richard Biener  <rguenther@suse.de>
11274         PR middle-end/57036
11275         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
11276         parameter, only add abnormal goto edges from the copied body
11277         if the call could perform abnormal gotos.
11278         (copy_cfg_body): Adjust.
11280 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
11282         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
11284 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
11286         * coretypes.h (gimple_stmt_iterator): Add struct to make
11287         compatible with C.
11289 2013-04-23  Richard Biener  <rguenther@suse.de>
11291         PR tree-optimization/57026
11292         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
11293         from SSA names occuring in abnormal PHI nodes.
11295 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11297         * lto/lto.c (print_lto_report_1): Fix LTO report names.
11299 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11301         * lto/lto.c (print_lto_report_1): Declare early.
11302         (read_cgraph_and_symbols): Call print_lto_report_1 early.
11304 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11306         * common.opt (-flto-report-wpa): Add.
11307         * doc/invoke.texi (-flto-report-wpa): Add.
11308         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
11309         (lto_main): dito.
11311 2013-04-22  Xinliang David Li  <davidxl@google.com>
11313         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
11314         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
11315         * Makefile.in: New dependency
11317             David Daney <ddaney.cavm@gmail.com>
11319         * configure.ac (gcc_cv_as_micromips_support):  Use the
11320         --fatal-warnings option.
11321         * configure: Regenerate.
11323 2013-04-22  Marek Polacek  <polacek@redhat.com>
11325         PR sanitizer/56990
11326         * tsan.c (instrument_expr): Don't instrument expression
11327         in case its size is zero.
11329 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
11331         PR target/57032
11332         Revert:
11333         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
11335         * config/alpha/alpha.c (TARGET_LRA_P): New define.
11337 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11339         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
11340         (gimple_stmt_iterator): New typedef.
11341         * gimple.h (gimple_stmt_iterator): Rename to...
11342         (gimple_stmt_iterator_d): ... This.
11343         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
11344         trees be valid for GIMPLE and GENERIC.
11345         (TARGET_GIMPLE_FOLD_BUILTIN): New.
11346         * gimple-fold.c (gimple_fold_call): Call target hook
11347         gimple_fold_builtin.
11348         * hooks.c (hook_bool_gsiptr_false): New.
11349         * hooks.h (hook_bool_gsiptr_false): New.
11350         * target.def (fold_stmt): New.
11351         * doc/tm.texi: Regenerate.
11353 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
11355         PR target/57018
11356         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
11357         a set sp if no stack realignment.
11359 2013-04-22  Nick Clifton  <nickc@redhat.com>
11361         * config.gcc (tilegx-linux): Extend extra_objs rather than
11362         overwriting it.
11363         (tilepro-linux): Likewise.
11365 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11367         * config/aarch64/aarch64-builtins.c
11368         (CF): Remove.
11369         (CF0, CF1, CF2, CF3, CF4, CF10): New.
11370         (VAR<1-12>): Add MAP parameter.
11371         (BUILTIN_*): Likewise.
11372         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
11373         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
11374         (aarch64_ushl_n<mode>): Likewise.
11375         (aarch64_sshr_n<mode>): Likewise.
11376         (aarch64_ushr_n<mode>): Likewise.
11377         (aarch64_<maxmin><mode>): Likewise.
11378         (aarch64_sqrt<mode>): Likewise.
11379         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
11380         (vshr<q>_n_*): Likewise.
11382 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11384         * config/aarch64/aarch64-builtins.c
11385         (aarch64_simd_builtin_type_mode): Handle SF types.
11386         (sf_UP): Define.
11387         (BUILTIN_GPF): Define.
11388         (aarch64_init_simd_builtins): Handle SF types.
11389         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
11390         (frecps): Likewise.
11391         (frecpx): Likewise.
11392         * config/aarch64/aarch64-simd.md
11393         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
11394         (aarch64_frecpe<mode>): New.
11395         (aarch64_frecps<mode>): Likewise.
11396         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
11397         (v8type): Add frecp<esx>.
11398         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
11399         (aarch64_frecps<mode>): Likewise.
11400         * config/aarch64/iterators.md (FRECP): New.
11401         (frecp_suffix): Likewise.
11402         * config/aarch64/arm_neon.h
11403         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
11405 2013-04-22  Christian Bruel  <christian.bruel@st.com>
11407         PR target/56995
11408         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
11409         (REG_CLASS_NAMES): Idem.
11410         (REG_CLASS_CONTENTS): Idem.
11411         (REGCLASS_HAS_FP_REG): Idem.
11412         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
11413         (sh_conditional_register_usage): Idem.
11415 2013-04-21  Jeff Law  <law@redhat.com>
11417         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
11418         (ssa_forward_propagate_and_combine): Use it.
11420 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11422         * lra.c: Update the flow chart diagram.
11424 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11426         PR rtl-optimization/56847
11427         * lra-constraints.c (process_alt_operands): Discourage alternative
11428         with non-matche doffsettable memory constraint fro memory with
11429         known offset.
11431 2013-04-19  Richard Biener  <rguenther@suse.de>
11433         PR tree-optimization/56982
11434         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
11435         function.
11436         * gimplify.c (gimplify_call_expr): Notice special calls.
11437         (gimplify_modify_expr): Likewise.
11438         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
11439         abnormal control flow receivers.
11440         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
11441         in the same way as cfun->has_nonlocal_labels.
11442         (gimple_purge_dead_abnormal_call_edges): Likewise.
11443         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
11444         receivers start a basic-block.
11446 2013-04-19  Richard Biener  <rguenther@suse.de>
11448         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
11449         member ...
11450         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
11451         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
11452         (SLP_TREE_LOAD_PERMUTATION): Add.
11453         (vect_transform_slp_perm_load): Adjust prototype.
11454         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
11455         (vect_free_slp_instance): Likewise.
11456         (vect_create_new_slp_node): Likewise.
11457         (vect_supported_slp_permutation_p): Remove.
11458         (vect_slp_rearrange_stmts): Adjust.
11459         (vect_supported_load_permutation_p): Likewise.  Inline
11460         vect_supported_slp_permutation_p here.
11461         (vect_analyze_slp_instance): Compute load permutations per
11462         slp node instead of per instance.
11463         (vect_get_slp_defs): Adjust.
11464         (vect_transform_slp_perm_load): Likewise.
11465         (vect_schedule_slp_instance): Remove redundant code.
11466         (vect_schedule_slp): Remove hack for PR56270, add it ...
11467         * tree-vect-stmts.c (vectorizable_load): ... here, do not
11468         CSE loads for SLP.  Adjust.
11470 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11472         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
11473         spelling in two comments.
11475 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11477         PR target/56797
11478         * config/arm/arm.c (load_multiple_sequence): Require SP
11479         as base register for loads if SP is in the register list.
11481 2013-04-19  Martin Jambor  <mjambor@suse.cz>
11483         PR tree-optimization/56718
11484         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
11485         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
11486         and made public.  Adjusted all callers.
11487         (ipa_intraprocedural_devirtualization): New function.
11488         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
11489         (ipa_intraprocedural_devirtualization): Likewise.
11490         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
11492 2013-04-19  Richard Biener  <rguenther@suse.de>
11494         PR tree-optimization/57000
11495         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
11497 2013-04-19  Terry Guo  <terry.guo@arm.com>
11499         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
11500         Replace with ...
11501         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
11502         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
11503         (cortex_m4_fmacs): Use new reservations.
11504         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
11506 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
11508         PR rtl-optimization/56999
11509         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
11510         related code.
11511         (lra_coalesce): Remove split_origin_bitmap and related code.
11512         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
11513         ranges if necessary.
11515 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
11517         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
11518         New array.
11519         (ix86_expand_call): Remove clobbered_registers array and use
11520         x86_64_ms_sysv_extra_clobbered_registers instead.
11521         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
11522         Declare here.
11523         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
11524         predicate.
11525         * config/i386/i386.md (*call_rex64_ms_sysv): Use
11526         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
11527         (*call_value_rex64_ms_sysv): Ditto.
11529 2013-04-18  Cary Coutant  <ccoutant@google.com>
11531         * dwarf2out.c (output_pubnames): Check die_perennial_p of
11532         parent instead of die_mark.
11534 2013-04-18  Diego Novillo  <dnovillo@google.com>
11536         * gimple.c (create_gimple_tmp): New.
11537         (get_expr_type): New.
11538         (build_assign): New.
11539         (build_type_cast): New.
11540         * gimple.h (enum ssa_mode): Define.
11541         (gimple_seq_set_location): New.
11542         * asan.c (build_check_stmt): Change some gimple_build_* calls
11543         to use build_assign and build_type_cast.
11545 2013-04-18  Richard Biener  <rguenther@suse.de>
11547         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
11548         handle negative step.  Remove redundant checks.
11549         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
11550         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
11551         for negative step and grouped loads fail to vectorize.
11553 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11555         * emit-rtl.c (reset_insn_used_flags): New function.
11556         (reset_all_used_flags): Use it.
11557         (verify_insn_sharing): New function.
11558         (verify_rtl_sharing): Fix verification for SEQUENCEs.
11560 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
11562         PR tree-optimization/56984
11563         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
11564         and (x >> M) >= N don't register any assertion if N << M is the
11565         minimum value.
11567 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11569         * lower-subreg.c (resolve_simple_move): If called self-recursive,
11570         do not delete_insn insns that have not yet been emitted, only
11571         unlink them with remove_insn.
11572         * df-scan.c (df_insn_delete): Revert r197492.
11574 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
11576         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
11577         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
11579 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11581         * config/arm/arm.md (movsicc_insn): Convert define_insn into
11582         define_insn_and_split.
11583         (and_scc,ior_scc,negscc): Likewise.
11584         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
11586 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11588         * config/arm/arm.c (use_return_insn): Return 0 for targets that
11589         can benefit from using a sequence of LDRD instructions in epilogue
11590         instead of a single LDM instruction.
11592 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11594         PR 45688
11595         * doc/extend.texi: Fix typo.
11597 2013-04-17  Richard Biener  <rguenther@suse.de>
11599         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
11600         (vect_build_slp_tree): ... here.
11601         (vect_build_slp_tree_1): Compute which stmts of the SLP group
11602         match.  Remove special-casing of mismatched complex loads.
11603         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
11604         re-try the match with swapped commutative operands.
11605         (vect_supported_load_permutation_p): Remove special-casing of
11606         mismatched complex loads.
11607         (vect_analyze_slp_instance): Adjust.
11609 2013-04-17  Richard Biener  <rguenther@suse.de>
11611         PR rtl-optimization/56921
11612         * cfgloop.h (struct loop): Add simple_loop_desc member.
11613         (struct niter_desc): Mark with GTY(()).
11614         (simple_loop_desc): Do not use aux field but simple_loop_desc.
11615         * loop-iv.c (get_simple_loop_desc): Likewise.
11616         (free_simple_loop_desc): Likewise.
11618         Revert
11619         2013-04-16  Richard Biener  <rguenther@suse.de>
11621         PR rtl-optimization/56921
11622         * loop-init.c (pass_rtl_move_loop_invariants): Add
11623         TODO_do_not_ggc_collect to todo_flags_finish.
11624         (pass_rtl_unswitch): Same.
11625         (pass_rtl_unroll_and_peel_loops): Same.
11626         (pass_rtl_doloop): Same.
11628 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
11630         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
11631         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
11632         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
11633         references.
11634         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
11635         * tree-streamer.c (record_common_node): Adjust reference in comment.
11637 2013-04-17  Terry Guo  <terry.guo@arm.com>
11639         * config/arm/cortex-m4.md: Add a new bypass.
11641 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11643         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
11644         New pattern.
11645         (*subs_<optab><mode>_multp2): New pattern.
11646         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11647         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11649 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11651         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
11652         (*subs_mul_imm_<mode>): New pattern.
11654 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
11656         PR target/56948
11657         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
11658         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
11659         (vsx_movti_32bit): Change j->wa to O->wa.
11661 2013-04-16  Richard Biener  <rguenther@suse.de>
11663         PR rtl-optimization/56921
11664         * loop-init.c (pass_rtl_move_loop_invariants): Add
11665         TODO_do_not_ggc_collect to todo_flags_finish.
11666         (pass_rtl_unswitch): Same.
11667         (pass_rtl_unroll_and_peel_loops): Same.
11668         (pass_rtl_doloop): Same.
11670 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11672         * config/arm/arm.c (emit_multi_reg_push): New declaration
11673         for an existing function.
11674         (arm_emit_strd_push): New function.
11675         (arm_expand_prologue): Used here.
11676         (arm_emit_ldrd_pop): New function.
11677         (arm_expand_epilogue): Used here.
11678         (arm_get_frame_offsets): Update condition.
11679         (arm_emit_multi_reg_pop): Add a special case for load of a single
11680         register with writeback.
11682 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
11684         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
11685         description.
11687 2013-04-16  Richard Biener  <rguenther@suse.de>
11689         PR tree-optimization/56756
11690         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
11691         (first_mem_ref_loc): New.
11692         (execute_sm): Place the load temporarily before a previous
11693         access instead of in the latch edge to ensure its SSA dependencies
11694         are defined at points dominating the load.
11696 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
11698         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
11699         correct fix by moving header and footer insn to the footer of
11700         the merged basic block.  Clear BB_END of the merged-away block.
11702         PR middle-end/43631
11703         * emit-rtl.c (make_note_raw): New function.
11704         (link_insn_into_chain): New static inline function.
11705         (add_insn): Use it.
11706         (add_insn_before, add_insn_after): Factor insn chain linking code...
11707         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
11708         using link_insn_into_chain.
11709         (note_outside_basic_block_p): New helper function for emit_note_after
11710         and emit_note_before.
11711         (emit_note_after): Use nobb variant of add_insn_after if the note
11712         should not be contained in a basic block.
11713         (emit_note_before): Use nobb variant of add_insn_before if the note
11714         should not be contained in a basic block.
11715         (emit_note_copy): Use make_note_raw.
11716         (emit_note): Likewise.
11717         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
11718         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
11719         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
11720         the moved barrier the tail of the basic block it follows.
11721         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
11723 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
11725         PR tree-optimization/56962
11726         * gimple-ssa-strength-reduction.c (record_increment): Only set
11727         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
11728         either rhs1 or rhs2 is equal to c->base_expr.
11730 2013-04-15  Richard Biener  <rguenther@suse.de>
11732         PR tree-optimization/56933
11733         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
11734         member.
11735         (GROUP_READ_WRITE_DEPENDENCE): Remove.
11736         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
11737         * tree-vect-data-refs.c (vect_analyze_group_access): Move
11738         dependence check ...
11739         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
11740         ... here.
11741         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
11742         GROUP_READ_WRITE_DEPENDENCE.
11744 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11746         * emit-rtl.c (reset_all_used_flags): New function.
11747         (verify_rtl_sharing): Call reset_all_used_flags before and after
11748         performing the checks.
11750 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11752         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
11753         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
11754         * config/arm/constraints.md (De): New constraint.
11755         * config/arm/neon.md (anddi3_neon): Delete.
11756         (neon_vand<mode>): Expand to standard anddi3 pattern.
11757         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
11758         Move earlier in the file.
11759         (neon_inv_logic_op2): Likewise.
11760         (arm_anddi_operand_neon): New predicate.
11762 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11764         * configure.ac (gcc_cv_ld_as_needed): Set
11765         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
11766         Use -z ignore, -z record on *-*-solaris2*.
11767         (HAVE_LD_AS_NEEDED): Update comment.
11768         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
11769         * configure: Regenerate.
11770         * config.in: Regenerate.
11771         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
11772         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
11773         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
11774         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
11775         equivalents.  Fix markup.
11776         * doc/tm.texi: Regenerate.
11778 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
11780         * config/i386/i386.opt: New option mstack-protector-guard=.
11781         * config/i386/i386-opts.h: Add enum stack_protector_guard.
11782         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
11783         TARGET_SSP_TLS_GUARD.
11784         * config/i386/i386.c (ix86_option_override_internal): Set
11785         ix86_stack_protector_guard.
11786         * config/i386/i386.md (stack_protect_set): Enable for
11787         TARGET_SSP_TLS_GUARD only.
11788         (stack_protect_set_<mode>): Ditto.
11789         (stack_protect_test): Ditto.
11790         (stack_protect_test_<mode>): Ditto.
11791         * doc/invoke.texi (i386 Option): Document.
11793 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
11795         PR target/56890
11796         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
11797         (S_MODES): Set H_MODE bit.
11798         (SF_MODES): Set only S_MODE and SF_MODE bits.
11799         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
11800         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
11801         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
11802         <MODE_FLOAT>: Likewise.
11804 2013-04-15  Joey Ye  <joey.ye@arm.com>
11806         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
11808 2013-04-15  Joey Ye  <joey.ye@arm.com>
11810         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
11811         for real far jump.
11812         (thumb_far_jump_used_p): Count instruction size and set
11813         far_jump_used.
11815 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
11817         * reorg.c (fill_simple_delay_slots): Reindent block of code.
11818         * resource.c (mark_target_live_regs): Reformat conditional block.
11820 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11822         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
11823         notes, they are emitted only just before final.
11824         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
11826 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11828         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
11829         * cfgrtl.c (delete_insn): Call it here instead.
11830         * lra-spills.c (lra_final_code_change): Use delete_insn.
11831         * haifa-sched.c (sched_remove_insn): Likewise.
11832         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
11833         returning to the nop pool.
11834         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
11835         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
11837 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11839         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
11840         * doc/tm.texi: Regenerated.
11842 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
11844         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
11845         QImode checks.
11847 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11849         * df-core.c (df_find_def): Compare register numbers.
11850         (df_find_use): Likewise.
11852 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
11854         PR target/56903
11855         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11856         lra_in_progress for return.
11858 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11860         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11861         define_insn into define_insn_and_split and emit movsicc patterns.
11863 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11865         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11867 2013-04-12  Richard Biener  <rguenther@suse.de>
11869         * tree-pass.h (TODO_do_not_ggc_collect): New.
11870         * passes.c (execute_one_ipa_transform_pass): Honor
11871         TODO_do_not_ggc_collect.
11872         (execute_one_pass): Likewise.
11874         Revert
11875         2013-04-10  Richard Biener  <rguenther@suse.de>
11877         * passes.c (init_optimization_passes): Remove reload pass.
11878         * ira.c (do_reload): Merge into ...
11879         (ira): ... this.
11880         (rest_of_handle_reload): Remove.
11881         (pass_reload): Likewise.
11882         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11883         of reload for vzeroupper pass placement.
11885 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
11887         PR tree-optimization/56918
11888         PR tree-optimization/56920
11889         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11890         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
11891         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
11892         use wide_mul_with_sign method.
11894 2013-04-12  Richard Biener  <rguenther@suse.de>
11896         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11897         not be considered a gimple constant.
11899 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
11901         * fold-const.c (const_binop): Handle vector shifts by a scalar.
11902         (fold_binary_loc): Call const_binop also for mixed vector-scalar
11903         operations.
11905 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11906             Jakub Jelinek  <jakub@redhat.com>
11908         * opts.c: Include diagnostic-color.h.
11909         (common_handle_option): Handle OPT_fdiagnostics_color_.
11910         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11911         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11912         (diagnostic-color.o): New.
11913         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11914         (diagnostic_color_rule): New enum.
11915         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11916         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11917         the location string.
11918         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11919         either NULL, or color kind.
11920         * diagnostic-color.c: New file.
11921         * diagnostic-color.h: New file.
11922         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11923         arguments.
11924         * doc/invoke.texi (-fdiagnostics-color): Document.
11925         * pretty-print.h (pp_show_color): Define.
11926         (struct pretty_print_info): Add show_color field.
11927         * diagnostic.c: Include diagnostic-color.h.
11928         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11929         macros.  Colorize error:, warning: etc. strings and also the location
11930         string.
11931         (diagnostic_show_locus): Colorize the caret line.
11932         * pretty-print.c: Include diagnostic-color.h.
11933         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
11934         inside of %< %> quotes or quoted through q format modifier.
11936 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11938         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
11940 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11942         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
11943         code in CC_NZ mode.
11944         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
11945         pattern.
11947 2013-04-11  Marek Polacek  <polacek@redhat.com>
11949         PR tree-optimization/48184
11950         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
11952 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
11954         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
11955         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
11956         (skip_simple_arithmetic): Tidy up.
11957         * tree.h (skip_simple_constant_arithmetic): Declare.
11959 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11961         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
11963 2013-04-11  Richard Biener  <rguenther@suse.de>
11965         * tree-vect-loop.c (get_initial_def_for_induction): Properly
11966         generate vector constants.
11968 2013-04-11  Richard Biener  <rguenther@suse.de>
11970         PR tree-optimization/56878
11971         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
11972         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
11973         New function.
11974         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
11975         Prefer to align the DR with the most invariant base address.
11977 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
11979         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
11980         comment.
11982 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
11984         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
11985         floating-point vector comparisons against 0.
11987 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
11989         PR tree-optimization/56899
11990         * fold-const.c (extract_muldiv_1): Apply distributive law
11991         only if TYPE_OVERFLOW_WRAPS (ctype).
11993 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
11995         PR target/56124
11996         * ira-costs.c (scan_one_insn): Check whether the source rtx of
11997         loading has side effect.
11999 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
12001         * config/sparc/sparc.c: Include tree-pass.h.
12002         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12003         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
12004         head of file.  Change return type.  Split off gate function.
12005         (sparc_gate_work_around_errata): New function.
12006         (pass_work_around_errata): New pass definition.
12007         (insert_pass_work_around_errata) New pass insert definition to
12008         insert pass_work_around_errata just after delayed-branch scheduling.
12009         (sparc_option_override): Insert the pass.
12010         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12012 2013-04-10  David S. Miller  <davem@davemloft.net>
12014         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12015         or -mcpu=hypersparc.
12017         * target.def (cstore_mode): New hook.
12018         * target.h: Include insn-codes.h
12019         * targhooks.c: Likewise.
12020         (default_cstore_mode): New function.
12021         * targhooks.h: Declare it.
12022         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12023         * doc/tm.texi: Rebuild.
12024         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12025         target hook, rather than inspecting the insn_data.
12026         * config/sparc/sparc.c (sparc_cstore_mode): New function.
12027         (TARGET_CSTORE_MODE): Redefine.
12028         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12029         result patterns.
12030         * config/sparc/predicates.md (cstore_result_operand): New special
12031         predicate.
12032         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12033         Use it for operand 0.
12034         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12035         (*snesi_special): Likewise.
12036         (*snesi_zero): Likewise.
12037         (*seqsi_zero): Likewise.
12038         (*sltu_insn): Likewise.
12039         (*sgeu_insn): Likewise.
12040         (*seqdi_special): Make operand 0 and comparison operation be of
12041         DImode.
12042         (*snedi_special): Likewise.
12043         (*snedi_special_vis3): Likewise.
12044         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12045         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12046         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12047         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12048         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12049         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12050         (*sltu_extend_sp64): Likewise.
12051         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12052         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12053         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12054         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12055         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12057 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
12059         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12060         (aarch64_start_file): Use the new function.
12062 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12063             Jason Merrill  <jason@redhat.com>
12065         * common.opt: Add -gdwarf.
12066         * opts.c (common_handle_option): Handle it.
12067         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12069 2013-04-10  Richard Biener  <rguenther@suse.de>
12071         * passes.c (execute_todo): Do not call ggc_collect conditional here.
12072         (execute_one_ipa_transform_pass): But unconditionally here.
12073         (execute_one_pass): And here.
12074         (init_optimization_passes): Remove reload pass.
12075         * tree-pass.h (TODO_ggc_collect): Remove.
12076         (pass_reload): Likewise.
12077         * ira.c (do_reload): Merge into ...
12078         (ira): ... this.
12079         (rest_of_handle_reload): Remove.
12080         (pass_reload): Likewise.
12081         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12082         of reload for vzeroupper pass placement.
12083         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12084         and todo_flags_finish of all passes.
12086 2013-04-10  Richard Biener  <rguenther@suse.de>
12088         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12089         first_const_oprnd field, rename first_def_type to first_op_type.
12090         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12091         (vect_get_and_check_slp_defs): Always use the type of the
12092         operand.  Allow mixed vect_external_def, vect_constant_def types.
12093         (vect_get_constant_vectors): Handle mixed vect_external_def,
12094         vect_constant_def types.
12096 2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
12098         PR tree-optimization/55524
12099         * tree-ssa-math-opts.c
12100         (convert_mult_to_fma): Don't use an fms construct
12101         when we don't have an fms operation, but fnma, and it looks
12102         likely that we'll be able to use the latter.
12104 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
12106         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12107         function.
12108         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12109         inline fail caused by overwritable functions.
12111 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
12113         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12114         unnecessary bits in the constant power of two case.
12116 2013-04-10  Richard Biener  <rguenther@suse.de>
12118         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12119         broken code swapping operands.
12120         (vect_build_slp_tree): Do not compute load permutations here.
12121         (vect_analyze_slp_instance): Compute load permutations here,
12122         after building the SLP tree.
12124 2013-04-09  Christian Bruel  <christian.bruel@st.com>
12126         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
12127         of next/prev_real_insn.
12129 2013-04-09  Jan Hubicka  <jh@suse.cz>
12131         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
12132         Drop aliased parameter.
12133         (function_and_variable_visibility): Do not handle alias pairs.
12134         * cgraph.c (varpool_externally_visible_p): Update prototype.
12135         * varpool.c (varpool_add_new_variable): Update.
12137 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12139         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
12141 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
12143         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
12145         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
12147 2013-04-09  Marek Polacek  <polacek@redhat.com>
12149         PR tree-optimization/48762
12150         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
12152 2013-04-09  Richard Biener  <rguenther@suse.de>
12154         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
12155         dealing with cost.
12156         (vect_build_slp_tree): Likewise.
12157         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
12158         calculating the cost of a SLP instance.
12159         (vect_analyze_slp_instance): Use it from here, after building
12160         the SLP tree.
12162 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12164         PR middle-end/56883
12165         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
12166         expand_omp_for_static_chunk): Use simple_p = true in
12167         force_gimple_operand_gsi calls when assigning to addressable decls.
12169 2013-04-09  Jeff Law  <law@redhat.com>
12171         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
12172         when the boolean was created by converting a wider object which
12173         had a boolean range.
12175 2013-04-09  Richard Biener  <rguenther@suse.de>
12177         * tree-vectorizer.h (slp_void_p): Remove.
12178         (slp_tree): Typedef before _slp_tree declaration.
12179         (struct _slp_tree): Use a vector of slp_tree as children.
12180         (vect_get_place_in_interleaving_chain): Remove.
12181         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
12182         Move ...
12183         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
12184         and make static.
12185         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
12186         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
12187         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
12188         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
12189         Use slp_node instead of slp_void_p and adjust.
12191 2013-04-09  Richard Biener  <rguenther@suse.de>
12193         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
12194         work that is not necessary.
12196 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12198         PR tree-optimization/56854
12199         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
12200         forward into clobber stmts if it would change MEM_REF lhs into
12201         non-MEM_REF.
12203 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
12205         * tree.c (type_hash_lookup, type_hash_add): Make static.
12206         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
12208 2013-04-09  Richard Biener  <rguenther@suse.de>
12210         * tree.h (unsave_expr_now): Remove.
12211         * tree-inline.c (mark_local_for_remap_r): Remove.
12212         (unsave_expr_1): Likewise.
12213         (unsave_r): Likewise.
12214         (unsave_expr_now): Likewise.
12215         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
12216         (propagate_tree_value): Likewise.
12218 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
12220         * doc/rtl.texi (sequence): Rewrite documentation to match the
12221         current use of SEQUENCE rtl objects.
12222         * rtl.def (SEQUENCE): Likewise.
12224         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
12225         Update documentation.
12226         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
12227         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
12229         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
12231 2013-04-08  Teresa Johnson  <tejohnson@google.com>
12233         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
12234         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
12235         methods.
12236         (estimate_edge_size_and_time): Add comment to suggest using rounding
12237         methods.
12238         (estimate_node_size_and_time): Ditto.
12239         (remap_edge_change_prob): Use helper rounding divide methods.
12240         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
12241         (gimple_mod_pow2_value_transform): Ditto.
12242         (gimple_mod_subtract_transform): Ditto.
12243         (gimple_ic_transform): Ditto.
12244         (gimple_stringops_transform): Ditto.
12245         * stmt.c (conditional_probability): Ditto.
12246         (emit_case_dispatch_table): Ditto.
12247         * lto-cgraph.c (merge_profile_summaries): Ditto.
12248         * tree-optimize.c (execute_fixup_cfg): Ditto.
12249         * cfgcleanup.c (try_forward_edges): Ditto.
12250         * cfgloopmanip.c (scale_loop_profile): Ditto.
12251         (loopify): Ditto.
12252         (duplicate_loop_to_header_edge): Ditto.
12253         (lv_adjust_loop_entry_edge): Ditto.
12254         * tree-vect-loop.c (vect_transform_loop): Ditto.
12255         * profile.c (compute_branch_probabilities): Ditto.
12256         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
12257         * lto-streamer-in.c (input_cfg): Ditto.
12258         * gimple-streamer-in.c (input_bb): Ditto.
12259         * ipa-cp.c (update_profiling_info): Ditto.
12260         (update_specialized_profile): Ditto.
12261         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
12262         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
12263         rounding methods.
12264         * sched-rgn.c (compute_dom_prob_ps): Ditto.
12265         (compute_trg_info): Ditto.
12266         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
12267         (purge_dead_edges): Ditto.
12268         * loop-unswitch.c (unswitch_loop): Ditto.
12269         * cgraphclones.c (cgraph_clone_edge): Ditto.
12270         (cgraph_clone_node): Ditto.
12271         * tree-inline.c (copy_bb): Ditto.
12272         (copy_edges_for_bb): Ditto.
12273         (initialize_cfun): Ditto.
12274         (copy_cfg_body): Ditto.
12275         (expand_call_inline): Ditto.
12277 2013-04-08  Kai Tietz  <ktietz@redhat.com>
12279         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
12280         TARGET_CYGWIN64 by TARGET_64BIT.
12282 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
12284         * config/epiphany/epiphany.md (GPR_1): New constant.
12285         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
12286         * config/epiphany/epiphany.c (gen_compare_reg):
12287         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
12288         is already in place.
12289         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
12290         Don't require being called during rtl expansion; If y operlaps r0,
12291         return 0.
12292         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
12293         (epiphany_expand_epilogue): Likewise.
12295         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
12296         Don't use CC_FPmode for ORDERED / UNORDERED.
12297         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
12299         * config/epiphany/constraints.md (CnL): New constraint.
12300         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
12301         * config/epiphany/predicates.md (add_operand): Allow 1024.
12303         * config/epiphany/epiphany.md (logical_op): New code iterator.
12304         (op_mnc): New code attribute.
12305         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
12306         (mov_f+1, mov_f+2): New peephole2 patterns.
12308         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
12309         (cstoresi4): Also allow re-use of zero result when doing a NE
12310         comparison to a non-zero operand.
12311         Use (clobber (scratch)) for first insn if the gpr output is not needed.
12313         * config/epiphany/epiphany.md (<insn_opname>v2si3):
12314         Use gen_addsi3_i / gen_subsi3_i.
12316 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
12318         PR c++/34949
12319         PR c++/50243
12320         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
12321         contain anything but clobbers, at most one __builtin_stack_restore,
12322         optionally debug stmts and final resx, and if it has at least one
12323         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
12324         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
12325         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
12326         which isn't defaut definition, remove them.
12327         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
12328         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
12329         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
12330         with MEM_REF LHS with SSA_NAME address.
12332 2013-04-08  Jeff Law  <law@redhat.com>
12334         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
12336 2013-04-08  Richard Biener  <rguenther@suse.de>
12338         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
12339         extra newline.
12340         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
12341         determined vector type.
12342         (vect_analyze_data_refs): Likewise.
12343         (vect_get_new_vect_var): Adjust.
12344         (vect_create_destination_var): Preserve SSA name versions.
12345         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
12346         not dump anything here.
12348 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
12350         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
12351         Add member lr_slot_known.
12352         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
12353         if necessary.
12354         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
12355         Remove code that sets lr_slot_offset according to what a previous
12356         version of epiphany_emit_save_restore used to do.
12357         (epiphany_emit_save_restore): When doing an lr save or restore,
12358         set/verify lr_slot_known and lr_slot_offset.
12360 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
12362         PR target/54338
12363         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
12364         in ALL_REGS.
12366 2013-04-08  Richard Biener  <rguenther@suse.de>
12368         * alias.c (find_base_term): Fix thinko in previous change.
12370 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
12372         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
12373         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
12374         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
12375         if possible to compute val.
12376         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
12377         For QImode integers don't require anything about precision.  Use
12378         const_with_all_bytes_same to find out if the constant doesn't have
12379         repeated bytes in it.
12381 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12383         * config/s390/s390.c (s390_expand_insv): Only accept insertions
12384         within mode size.
12386 2013-04-08  Marek Polacek  <polacek@redhat.com>
12388         PR rtl-optimization/48182
12389         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
12390         value to 1.
12392 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
12394         PR target/55487
12395         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
12396         nuses, make sure we have a label.
12398 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12400         PR target/56843
12401         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
12402         (rs6000_emit_swdiv_low_precision): Remove.
12403         (rs6000_emit_swdiv): Rewrite to handle between one and four
12404         iterations of Newton-Raphson generally; modify required number of
12405         iterations for some cases.
12406         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
12408 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
12410         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
12411         set-but-unused variable.
12413         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
12414         basic blocks of released function bodies garbage-collectable.
12416         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
12417         (struct rtl_opt_pass): Add TODO_df_finish.
12419         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
12421 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12423         * config/arm/constraints.md (q): New constraint.
12424         * config/arm/ldrdstrd.md: New file.
12425         * config/arm/arm.md (ldrdstrd.md) New include.
12426         (arm_movdi): Use "q" instead of "r" constraint
12427         for double-word memory access.
12428         (movdf_soft_insn): Likewise.
12429         * config/arm/vfp.md (movdi_vfp): Likewise.
12430         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
12431         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
12432         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
12433         (mem_ok_for_ldrd_strd): Likewise.
12434         (output_move_double): Update assertion.
12436 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12438         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
12440 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12442         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
12443         define_insn_and_split.
12444         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
12446 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12448         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
12449         define_insn_and_split.
12450         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
12451         (shiftsi3_compare): New pattern.
12452         (rrx): New pattern.
12453         * config/arm/unspecs.md (UNSPEC_RRX): New.
12455 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12457         * config/arm/arm.md (negdi_extendsidi): New pattern.
12458         (negdi_zero_extendsidi): Likewise.
12460 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12462         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
12463         define_insn_and_split.
12464         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
12465         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
12467 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12469         * config/arm/arm.md (arm_subdi3): Convert define_insn into
12470         define_insn_and_split.
12471         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
12472         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
12474 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12476         * config/arm/arm.md (subsi3_carryin): New pattern.
12477         (subsi3_carryin_const): Likewise.
12478         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
12479         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
12481 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12483         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
12485 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12487         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
12488         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
12490 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12492         * config/arm/arm.c (arm_expand_builtin): Change fcode
12493         type to unsigned int.
12495 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12497         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
12499 2013-04-04  Ian Lance Taylor  <iant@google.com>
12501         * doc/standards.texi (Standards): The Go frontend supports the Go 1
12502         language standard.
12504 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
12506         PR middle-end/56729
12507         * df-scan.c (df_insn_delete): Disable failing assert.
12509 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12511         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
12512         New function prototype.
12513         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
12514         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
12515         (arm_builtin_vectorized_function): New function.
12517 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12519         * config/arm/arm_neon_builtins.def: New file.
12520         * config/arm/arm.c (neon_builtin_data): Move contents to
12521         arm_neon_builtins.def.
12522         (enum arm_builtins): Include neon builtin definitions.
12523         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
12524         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
12526 2013-04-04  Marek Polacek  <polacek@redhat.com>
12528         PR tree-optimization/48186
12529         * predict.c (maybe_hot_frequency_p): Return false if
12530         HOT_BB_FREQUENCY_FRACTION is 0.
12531         (cgraph_maybe_hot_edge_p): Likewise.
12533 2013-04-04  Richard Biener  <rguenther@suse.de>
12535         PR tree-optimization/56826
12536         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
12537         more accurately.
12539 2013-04-04  Richard Biener  <rguenther@suse.de>
12541         PR tree-optimization/56213
12542         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
12543         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
12545 2013-04-04  Richard Biener  <rguenther@suse.de>
12547         PR tree-optimization/56837
12548         * tree-loop-distribution.c (classify_partition): For non-zero
12549         values require that the value has the same precision as its
12550         mode to be useful as memset value.
12552 2013-04-03  Nick Clifton  <nickc@redhat.com>
12554         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
12555         (fmssf4): Use fmsf.s on E3V5 architectures.
12556         (fnmasf4): Use fnmaf.s on E3V5 architectures.
12557         (fnmssf4): Use fnmsf.s on E3V5 architectures.
12559 2013-04-03  Jeff Law  <law@redhat.com>
12561         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
12562         (lra-eliminations.o): Likewise.
12564 2013-04-03  Teresa Johnson  <tejohnson@google.com>
12566         * gcov-io.c (compute_working_sets): Moved most of body of old
12567         compute_working_sets here from profile.c.
12568         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
12569         (gcov_working_set_t): Moved typedef here from basic-block.h
12570         (compute_working_set): Declare.
12571         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
12572         (get_working_sets): Renamed from compute_working_set,
12573         replace most of body with call to new compute_working_sets.
12574         (get_exec_counts): Replace call to compute_working_sets
12575         to get_working_sets.
12576         * profile.h (get_working_sets): Renamed from compute_working_set.
12577         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
12578         to get_working_sets.
12579         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
12580         * gcov-dump.c (dump_working_sets): New function.
12582 2013-04-03  Kenneth Zadeck <zadeck@naturalbridge.com>
12584         * hwint.c (sext_hwi, zext_hwi): New functions.
12585         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
12586         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
12587         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
12588         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
12589         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
12590         (sext_hwi, zext_hwi): New functions.
12592 2013-04-03  Jeff Law  <law@redhat.com>
12594         PR tree-optimization/56799
12595         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
12596         back test for widening conversion erroneously dropped in prior change.
12598 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12600         PR target/56809
12601         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
12602         instead of next_real_insn.
12604 2013-04-03  Marek Polacek  <polacek@redhat.com>
12606         PR sanitizer/55702
12607         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
12609 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12611         PR target/56809
12612         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
12613         next_real_insn.
12614         (thumb1_output_casesi): Likewise.
12615         (thumb2_output_casesi): Likewise.
12617 2013-04-03  Richard Biener  <rguenther@suse.de>
12619         PR tree-optimization/56817
12620         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
12621         Split out ...
12622         (tree_unroll_loops_completely_1): ... new function to manually
12623         walk the loop tree, properly defering outer loops of unrolled
12624         loops to later iterations.
12626 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12628         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
12629         (vectorizable_load): Likewise.
12630         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
12631         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
12633 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12635         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
12636         BIT_FIELD_REF.
12638 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12640         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
12642 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
12644         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
12646 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12648         PR tree-optimization/56790
12649         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
12650         folding.
12652 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12654         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
12655         Handle VEC_MERGE.
12656         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
12657         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
12658         equal arguments.
12660 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
12662         PR c/19449
12663         * tree.h (force_folding_builtin_constant_p): New decl.
12664         * builtins.c (force_folding_builtin_constant_p): New variable.
12665         (fold_builtin_constant_p): Fold immediately also if
12666         force_folding_builtin_constant_p.
12668 2013-04-03  Richard Biener  <rguenther@suse.de>
12670         PR tree-optimization/56812
12671         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
12672         DRs of the same interleaving chain are independent.
12674 2013-04-02  Jason Merrill  <jason@redhat.com>
12676         * gdbinit.in (pbb): Use debug fn.
12678 2013-04-02  Lawrence Crowl  <crowl@google.com>
12680         * sese.h (struct ivtype_map_elt_s): Remove unused.
12681         (extern debug_ivtype_map): Remove unused.
12682         (extern eq_ivtype_map_elts): Remove unused.
12683         * sese.c (debug_ivtype_map): Removed unused.
12684         (debug_ivtype_map_1): Removed unused.
12685         (debug_ivtype_elt): Remove unused.
12686         (eq_ivtype_map_elts): Remove unused.
12689 2013-04-02  Kai Tietz  <ktietz@redhat.com>
12691         PR target/52790
12692         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
12693         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
12694         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
12695         function.
12696         (legitimize_pe_coff_symbol): Likewise.
12697         (is_imported_p): New helper-function.
12698         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
12699         for Windows x64 targets.
12700         (ix86_expand_prologue): Optimize for pe-coff targets.
12701         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
12702         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
12703         medium/large code-model.
12704         (legitimize_pic_address): Likewise.
12705         (legitimize_tls_address): Likewise.
12706         (ix86_expand_call): Likewise.
12707         (x86_output_mi_thunk): Likewise.
12708         (get_dllimport_decl): Add new beimport argument.
12709         (construct_plt_address): Don't assert for x64 pe-coff targets.
12710         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
12711         targets.
12712         (SYMBOL_FLAG_STUBVAR): New macro.
12713         (SYMBOL_REF_STUBVAR_P): Likewise.
12714         * config/i386/winnt.c (stub_list): New structure.
12715         (stub_head): New local variable.
12716         (i386_pe_record_stub): New function.
12717         (i386_pe_file_end): Emit refptr-stubs.
12719 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
12721         PR rtl-optimization/56745
12722         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
12723         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
12725         PR c++/34949
12726         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
12727         and both of them are MEM_REFs, just compare first argument for
12728         equality and attempt to deal even with differing offsets.
12730         PR c++/34949
12731         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
12732         of gimple_clobber_p to be MEM_REF.
12733         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
12734         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
12735         after gimplification.
12736         * asan.c (get_mem_ref_of_assignment): Don't instrument
12737         gimple_clobber_p stmts.
12738         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
12739         gimple_clobber_p stmt if they have MEM_REF lhs and
12740         are dead because of another gimple_clobber_p stmt.
12741         * tree-ssa-live.c (clear_unused_block_pointer): Treat
12742         gimple_clobber_p stmts like debug stmts.
12743         (remove_unused_locals): Remove clobbers with MEM_REF lhs
12744         that refer to unused VAR_DECLs or uninitialized values.
12745         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
12746         gimple_clobber_p stmts if they refer to removed parameters.
12747         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
12748         formatting.
12750 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
12752         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
12753         using SWI48 mode attribute.
12755 2013-04-02  Wei Mi  <wmi@google.com>
12757         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
12758         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
12759         *<rotate_insn><mode>3_mask in i386.md.
12761 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12763         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
12765 2013-04-02  Richard Biener  <rguenther@suse.de>
12767         PR tree-optimization/56778
12768         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
12769         Runtime alias tests are not supported for gather loads.
12770         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
12771         stmts referenced from SSA operands before updating SSA form.
12773 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
12774             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12776         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
12777         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
12778         * config/arm/cortex-a53.md: New file.
12779         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
12780         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
12781         * config/arm/arm.c (arm_issue_rate): Likewise.
12782         * config/arm/arm-tune.md: Regenerate
12783         * config/arm/arm-tables.opt: Regenerate.
12784         * config/arm/arm-cores.def: Add cortex-a53.
12786 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
12788         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
12789         non-static link.
12791 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12793         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
12794         scalar load/store operations using B/H registers.
12795         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
12797 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12799         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
12800         scalar move.
12801         * config/aarch64/aarch64.c
12802         (aarch64_simd_scalar_immediate_valid_for_move): New.
12803         * config/aarch64/aarch64-protos.h
12804         (aarch64_simd_scalar_immediate_valid_for_move): New.
12805         * config/aarch64/constraints.md (Dh, Dq): New.
12806         * config/aarch64/iterators.md (hq): New.
12808 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
12810         * reorg.c (get_branch_condition): Deal with conditional returns.
12811         (fill_simple_delay_slots): Remove dead code dealing with jumps.
12813 2013-04-01  Wei Mi  <wmi@google.com>
12815         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
12816         Truncate operand 2 using %b asm operand modifier.
12817         (*<shift_insn><mode>3_mask): Ditto.
12818         (*<rotate_insn><mode>3_mask): Ditto.
12820 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
12822         PR middle-end/56798
12823         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
12825 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
12827         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
12828         of next_real_insn.
12829         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
12831 2013-03-30  Lawrence Crowl  <crowl@google.com>
12833         * dse.c (clear_alias_sets): Remove never set.
12834         (disqualified_clear_alias_sets): Remove never set.
12835         (clear_alias_mode_pool): Remove never set.
12836         (dse_step0): Remove condition that is never true.
12837         (canon_address): Remove condition that is never true.
12838         (dse_step7): Remove condition that is never true.
12839         (rest_of_handle_dse): Remove condition that is never true.
12840         (rest_of_handle_dse::did_global): Remove never read from above.
12841         (dse_step2_spill): Remove never called from above.
12842         (dse_step5_spill): Remove never called from above.
12844 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
12846         * doc/md.texi (Standard Names) <casesi>: Update documentation for
12847         JUMP_TABLE_DATA changes.
12848         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
12849         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
12850         (Insns) <jump_table_data>: New entry.
12851         * doc/tm.texi: Regenerate.
12853         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
12855         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12856         for table jump at the end of a basic block using tablejump_p.
12857         * targhooks.c (default_invalid_within_doloop): Likewise.
12858         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12859         target hook implementation that is identical to the default hook.
12860         (rs6000_invalid_within_doloop): Remove.
12862         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12863         unused variable from tablejump_p call.
12865         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12866         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12867         (INSN_DELETED_P): Likewise.
12868         (emit_jump_table_data): New prototype.
12869         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12870         after 4th as unused.
12871         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12872         * sched-vis.c (print_insn): Likewise.
12873         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12874         insn for compatibility with back ends that use next_active_insn to
12875         identify jump table data.
12876         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12877         (remove_insn): Likewise.
12878         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12879         to be emitted.
12880         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12881         (emit_jump_table_data): New function.
12883         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12884         basic block, a JUMP_TABLE_DATA never is.
12885         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12886         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12887         off from code handling real insns.
12888         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12889         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12890         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12891         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12892         is not a NONDEBUG_INSN_P.
12893         * ira-costs.c (scan_one_insn): Likewise.
12894         * jump.c (mark_all_labels): Likewise.
12895         (mark_jump_label_1): Likewise.
12896         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12897         * lra.c (get_insn_freq): Expect all insns reaching here to be in
12898         a basic block.
12899         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12900         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12901         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12902         JUMP_TABLE_DATA_P insns.
12903         (calculate_elim_costs_all_insns): Likewise.
12904         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12905         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12906         (delete_output_reload): Code style fixups.
12907         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12908         insn flags on this non-insn.
12909         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12910         as scheduling barriers, for pre-change compatibility.
12911         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12912         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12914         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12915         redundant JUMP_TABLE_DATA_P test.
12916         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12917         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12918         (frv_for_each_packet): Likewise.
12919         * config/i386/i386.c (min_insn_size): Likewise.
12920         (ix86_avoid_jump_mispredicts): Likewise.
12921         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12922         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12923         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12924         (mips16_insn_length): Robustify.
12925         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12926         (mips16_split_long_branches): Likewise.
12927         * config/pa/pa.c (pa_combine_instructions): Likewise.
12928         * config/rs6000/rs6000.c (get_next_active_insn): Treat
12929         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12930         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12931         as contributing to pool range lengths.
12932         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12933         Remove redundant JUMP_TABLE_DATA_P test.
12934         (sh_loop_align): Likewise.
12935         (split_branches): Likewise.
12936         (sh_insn_length_adjustment): Likewise.
12937         * config/spu/spu.c (get_branch_target): Likewise.
12939 2013-03-29  Jan Hubicka  <jh@suse.cz>
12941         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
12942         gcov streaming; stream hot bb threshold to ltrans.
12943         * predict.c (get_hot_bb_threshold): Break out from ....
12944         (maybe_hot_count_p): ... here.
12945         (set_hot_bb_threshold): New function.
12946         * lto-section-in.c (lto_section_name): Add profile.
12947         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
12948         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
12949         and data-streamer.h
12950         (histogram_entry): New structure.
12951         (histogram, histogram_pool): New global vars.
12952         (histogram_hash): New structure.
12953         (histogram_hash::hash): New method.
12954         (histogram_hash::equal): Likewise.
12955         (account_time_size): New function.
12956         (cmp_counts): New function.
12957         (dump_histogram): New function.
12958         (ipa_profile_generate_summary): New function.
12959         (ipa_profile_write_summary): New function.
12960         (ipa_profile_read_summary): New function.
12961         (ipa_profile): Decide on threshold.
12962         (pass_ipa_profile): Add ipa_profile_write_summary and
12963         ipa_profile_read_summary.
12964         * Makefile.in (ipa.o): Update dependencies.
12965         * lto-streamer.h (LTO_section_ipa_profile): New section.
12967 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
12969         * tree.h (VAR_P): New.
12971 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
12973         PR lto/56777
12974         * doc/invoke.texi ([-fwhole-program]): Fix typo.
12976 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
12978         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
12979         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12980         (control_flow_insn_p): Likewise.
12981         * cfgrtl.c (duplicate_insn_chain): Likewise.
12982         * final.c (get_attr_length_1): Likewise.
12983         (shorten_branches): Likewise.
12984         (final_scan_insn): Likewise.
12985         * function.c (instantiate_virtual_regs): Likewise.
12986         * gcse.c (insert_insn_end_basic_block): Likewise.
12987         * ira-costs.c (scan_one_insn): Likewise.
12988         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12989         * lra.c (check_rtl): Likewise.
12990         * reload1.c (elimination_costs_in_insn): Likewise.
12991         * reorg.c (follow_jumps): Likewise.
12993         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
12994         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12995         (thumb_far_jump_used_p): Likewise.
12996         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
12997         (workaround_speculation): Likewise.
12998         (add_sched_insns_for_speculation): Likewise.
12999         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13000         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13001         (frv_for_each_packet): Likewise.
13002         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13003         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13004         (final_emit_insn_group_barriers): Likewise.
13005         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13006         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13007         (mips16_insn_length): Likewise.
13008         * config/pa/pa.c (pa_reorg): Likewise.
13009         (pa_combine_instructions): Likewise.
13010         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13011         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13012         (sh_reorg): Likewise.
13013         (split_branches): Likewise.
13014         * config/spu/spu.c (get_branch_target): Likewise.
13016         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13017         JUMP_TABLE_DATA_P.
13019 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
13021         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13022         Fix declaration name.
13024 2013-03-28  Lawrence Crowl  <crowl@google.com>
13026         * graphds.h (struct graph.indicies): Remove unused.
13027         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13028         (SCOP_ORIGINAL_PDDRS): Remove unused.
13029         * sese.h (extern insert_loop_close_phis): Removed unused.
13030         (extern insert_guard_phis): Removed unused.
13031         (extern ivtype_map_elt_info): Removed unused.
13032         (new_ivtype_map_elt): Removed unused.
13033         * sese.c (ivtype_map_elt_info): Removed unused.
13035 2013-03-28  Lawrence Crowl  <crowl@google.com>
13037         * Makefile.in: Add several missing include dependences.
13038         (DUMPFILE_H): New.
13039         (test-dump.o): New.  This object is not added to any executable,
13040         but is present for ad-hoc testing.
13041         * bitmap.c
13042         (debug (const bitmap_head_def &)): New.
13043         (debug (const bitmap_head_def *)): New.
13044         * bitmap.h
13045         (extern debug (const bitmap_head_def &)): New.
13046         (extern debug (const bitmap_head_def *)): New.
13047         * cfg.c
13048         (debug (edge_def &)): New.
13049         (debug (edge_def *)): New.
13050         * cfghooks.c
13051         (debug (basic_block_def &)): New.
13052         (debug (basic_block_def *)): New.
13053         * dumpfile.h
13054         (dump_node (const_tree, int, FILE *)): Correct source file.
13055         * dwarf2out.c
13056         (debug (die_struct &)): New.
13057         (debug (die_struct *)): New.
13058         * dwarf2out.h
13059         (extern debug (die_struct &)): New.
13060         (extern debug (die_struct *)): New.
13061         * gimple-pretty-print.c
13062         (debug (gimple_statement_d &)): New.
13063         (debug (gimple_statement_d *)): New.
13064         * gimple-pretty-print.h
13065         (extern debug (gimple_statement_d &)): New.
13066         (extern debug (gimple_statement_d *)): New.
13067         * ira-build.c
13068         (debug (ira_allocno_copy &)): New.
13069         (debug (ira_allocno_copy *)): New.
13070         (debug (ira_allocno &)): New.
13071         (debug (ira_allocno *)): New.
13072         * ira-int.h
13073         (extern debug (ira_allocno_copy &)): New.
13074         (extern debug (ira_allocno_copy *)): New.
13075         (extern debug (ira_allocno &)): New.
13076         (extern debug (ira_allocno *)): New.
13077         * ira-lives.c
13078         (debug (live_range &)): New.
13079         (debug (live_range *)): New.
13080         * lra-int.h
13081         (debug (lra_live_range &)): New.
13082         (debug (lra_live_range *)): New.
13083         * lra-lives.c
13084         (debug (lra_live_range &)): New.
13085         (debug (lra_live_range *)): New.
13086         * omega.c
13087         (debug (omega_pb_d &)): New.
13088         (debug (omega_pb_d *)): New.
13089         * omega.h
13090         (extern debug (omega_pb_d &)): New.
13091         (extern debug (omega_pb_d *)): New.
13092         * print-rtl.c
13093         (debug (const rtx_def &)): New.
13094         (debug (const rtx_def *)): New.
13095         * print-tree.c
13096         (debug_tree (tree): Move within file.
13097         (debug_raw (const tree_node &)): New.
13098         (debug_raw (const tree_node *)): New.
13099         (dump_tree_via_hooks (const tree_node *, int)): New.
13100         (debug (const tree_node &)): New.
13101         (debug (const tree_node *)): New.
13102         (debug_verbose (const tree_node &)): New.
13103         (debug_verbose (const tree_node *)): New.
13104         (debug_head (const tree_node &)): New.
13105         (debug_head (const tree_node *)): New.
13106         (debug_body (const tree_node &)): New.
13107         (debug_body (const tree_node *)): New.
13108         (debug_vec_tree (tree): Move and reimplement in terms of dump.
13109         (debug (vec<tree, va_gc> &)): New.
13110         (debug (vec<tree, va_gc> *)): New.
13111         * rtl.h
13112         (extern debug (const rtx_def &)): New.
13113         (extern debug (const rtx_def *)): New.
13114         * sbitmap.c
13115         (debug_raw (simple_bitmap_def &)): New.
13116         (debug_raw (simple_bitmap_def *)): New.
13117         (debug (simple_bitmap_def &)): New.
13118         (debug (simple_bitmap_def *)): New.
13119         * sbitmap.h
13120         (extern debug (simple_bitmap_def &)): New.
13121         (extern debug (simple_bitmap_def *)): New.
13122         (extern debug_raw (simple_bitmap_def &)): New.
13123         (extern debug_raw (simple_bitmap_def *)): New.
13124         * sel-sched-dump.c
13125         (debug (vinsn_def &)): New.
13126         (debug (vinsn_def *)): New.
13127         (debug_verbose (vinsn_def &)): New.
13128         (debug_verbose (vinsn_def *)): New.
13129         (debug (expr_def &)): New.
13130         (debug (expr_def *)): New.
13131         (debug_verbose (expr_def &)): New.
13132         (debug_verbose (expr_def *)): New.
13133         (debug (vec<rtx> &)): New.
13134         (debug (vec<rtx> *)): New.
13135         * sel-sched-dump.h
13136         (extern debug (vinsn_def &)): New.
13137         (extern debug (vinsn_def *)): New.
13138         (extern debug_verbose (vinsn_def &)): New.
13139         (extern debug_verbose (vinsn_def *)): New.
13140         (extern debug (expr_def &)): New.
13141         (extern debug (expr_def *)): New.
13142         (extern debug_verbose (expr_def &)): New.
13143         (extern debug_verbose (expr_def *)): New.
13144         (extern debug (vec<rtx> &)): New.
13145         (extern debug (vec<rtx> *)): New.
13146         * sel-sched-ir.h
13147         (_list_iter_cond_expr): Make inline instead of static.
13148         * sreal.c
13149         (debug (sreal &)): New.
13150         (debug (sreal *)): New.
13151         * sreal.h
13152         (extern debug (sreal &)): New.
13153         (extern debug (sreal *)): New.
13154         * tree.h
13155         (extern debug_raw (const tree_node &)): New.
13156         (extern debug_raw (const tree_node *)): New.
13157         (extern debug (const tree_node &)): New.
13158         (extern debug (const tree_node *)): New.
13159         (extern debug_verbose (const tree_node &)): New.
13160         (extern debug_verbose (const tree_node *)): New.
13161         (extern debug_head (const tree_node &)): New.
13162         (extern debug_head (const tree_node *)): New.
13163         (extern debug_body (const tree_node &)): New.
13164         (extern debug_body (const tree_node *)): New.
13165         (extern debug (vec<tree, va_gc> &)): New.
13166         (extern debug (vec<tree, va_gc> *)): New.
13167         * tree-cfg.c
13168         (debug (struct loop &)): New.
13169         (debug (struct loop *)): New.
13170         (debug_verbose (struct loop &)): New.
13171         (debug_verbose (struct loop *)): New.
13172         * tree-dump.c: Add header dependence.
13173         * tree-flow.h
13174         (extern debug (struct loop &)): New.
13175         (extern debug (struct loop *)): New.
13176         (extern debug_verbose (struct loop &)): New.
13177         (extern debug_verbose (struct loop *)): New.
13178         * tree-data-ref.c
13179         (debug (data_reference &)): New.
13180         (debug (data_reference *)): New.
13181         (debug (vec<data_reference_p> &)): New.
13182         (debug (vec<data_reference_p> *)): New.
13183         (debug (vec<ddr_p> &)): New.
13184         (debug (vec<ddr_p> *)): New.
13185         * tree-data-ref.h
13186         (extern debug (data_reference &)): New.
13187         (extern debug (data_reference *)): New.
13188         (extern debug (vec<data_reference_p> &)): New.
13189         (extern debug (vec<data_reference_p> *)): New.
13190         (extern debug (vec<ddr_p> &)): New.
13191         (extern debug (vec<ddr_p> *)): New.
13192         * tree-ssa-alias.c
13193         (debug (pt_solution &)): New.
13194         (debug (pt_solution *)): New.
13195         * tree-ssa-alias.h
13196         (extern debug (pt_solution &)): New.
13197         (extern debug (pt_solution *)): New.
13198         * tree-ssa-alias.c
13199         (debug (_var_map &)): New.
13200         (debug (_var_map *)): New.
13201         (debug (tree_live_info_d &)): New.
13202         (debug (tree_live_info_d *)): New.
13203         * tree-ssa-alias.h
13204         (extern debug (_var_map &)): New.
13205         (extern debug (_var_map *)): New.
13206         (extern debug (tree_live_info_d &)): New.
13207         (extern debug (tree_live_info_d *)): New.
13209 2013-03-28  Jan Hubicka  <jh@suse.cz>
13211         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
13213 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
13215         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
13216         record only when desired or required.
13218 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
13220         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
13221         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
13223 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13225         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
13226         (*andsi3_compare0_uxtw): New pattern.
13227         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
13228         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
13230 2013-03-28  Jan Hubicka  <jh@suse.cz>
13232         * data-streamer-in.c (streamer_read_gcov_count): New function.
13233         * gimple-streamer-out.c: Include value-prof.h.
13234         (output_gimple_stmt): Output histogram.
13235         (output_bb): Use streamer_write_gcov_count.
13236         * value-prof.c: Include data-streamer.h
13237         (dump_histogram_value): Add HIST_TYPE_MAX.
13238         (stream_out_histogram_value): New function.
13239         (stream_in_histogram_value): New function.
13240         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
13241         (stream_out_histogram_value, stream_in_histogram_value): Declare.
13242         * data-streamer-out.c (streamer_write_gcov_count): New function.
13243         (streamer_write_gcov_count_stream): New function.
13244         * lto-cgraph.c (lto_output_edge): Update counter streaming.
13245         (lto_output_node): Likewise.
13246         (input_node, input_edge): Likewise.
13247         * lto-streamer-out.c (output_cfg): Update streaming.
13248         * lto-streamer-in.c (input_cfg): Likewise.
13249         * data-streamer.h (streamer_write_gcov_count,
13250         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
13251         * gimple-streamer-in.c: Include value-prof.h
13252         (input_gimple_stmt): Input histograms.
13253         (input_bb): Update profile streaming.
13255 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
13257         * genmodes.c (emit_max_int): New function.
13258         (emit_insn_modes_h): Added call to emit_max_function.
13259         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
13260         Added doc.
13261         * machmode.def: Fixed comment.
13263 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
13265         * combine.c (try_combine): Removed useless assert.
13266         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
13268 2013-03-28  Marek Polacek  <polacek@redhat.com>
13269             Richard Biener  <rguenther@suse.de>
13271         PR tree-optimization/56695
13272         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
13273         build signed result of a vector comparison.
13274         * tree-cfg.c (verify_gimple_comparison): Check that a result
13275         of a vector comparison has signed type.
13277 2013-03-28  Richard Biener  <rguenther@suse.de>
13279         PR tree-optimization/37021
13280         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
13281         do not restrict gaps between groups.
13282         * tree-vect-stmts.c (vectorizable_load): Properly account for
13283         a gap between groups.
13285 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
13287         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
13288         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
13289         is not enabled.
13291 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
13293         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
13294         * doc/extend.texi (Named Address Spaces): Ditto.
13295         (Variable Attributes): Ditto.
13297 2013-03-27  Kai Tietz  <ktietz@redhat.com>
13299         * config.build: Add support for cygwin x64 target.
13300         * config.gcc: Likewise.
13301         * config.host: Likewise.
13302         * configure.ac: Likewise
13303         * configure: Regenerated.
13305 2013-03-27  Kai Tietz  <ktietz@redhat.com>
13307         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
13308         * config/i386/t-cygwin-w64: New file.
13309         * config/i386/cygwin-w64.h: New file.
13310         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
13311         and add support for x64-cygwin target.
13312         (CPP_SPEC): Likewise.
13313         (CXX_WRAP_SPEC_LIST): Undefine before define.
13314         (LIBGCJ_SONAME): Use 15 as version.
13316 2013-03-27  Richard Biener  <rguenther@suse.de>
13318         PR tree-optimization/56716
13319         * tree-ssa-structalias.c (perform_var_substitution): Adjust
13320         dumping for ref nodes.
13322 2013-03-27  Martin Jambor  <mjambor@suse.cz>
13324         PR tree-optimization/55334
13325         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
13326         restricted pointers to arrays.
13328 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13330         * Makefile.in (.SUFFIXES): Add .cc.
13331         (.c.o): Apply same recipe for implicit rule .cc.o.
13333 2013-03-27  Richard Biener  <rguenther@suse.de>
13335         PR tree-optimization/37021
13336         * tree-vect-data-refs.c (vect_check_strided_load): Allow
13337         REALPART/IMAGPART_EXPRs around the supported refs.
13338         * tree-ssa-structalias.c (find_func_aliases): Assume that
13339         floating-point values are not used to transfer pointers.
13341 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
13343         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
13344         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
13345         * gcc/doc/tm.texi: Regenerate.
13346         * gcc/targhooks.h (default_has_ifunc_p): New.
13347         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
13348         * gcc/config/linux-protos.h: New file.
13349         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
13350         this hook for linux which disables support of indirect functions in
13351         android.
13352         * gcc/config/linux-android.c: New file.
13353         * gcc/config/t-linux-android.c: Ditto.
13354         * gcc/config.gcc: Added new object file linux-android.o.
13355         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
13356         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
13357         * gcc/varasm.c (do_assemble_alias): Likewise.
13358         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
13359         doesn't support indirect functions.
13360         * configure: Regenerate.
13362 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
13364         PR target/56102
13365         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
13366         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
13367         mult-word mode.
13369 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13371         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
13373 2013-03-27  Terry Guo  <terry.guo@arm.com>
13375         * config/arm/arm-cores.def: Added core cortex-r7.
13376         * config/arm/arm-tune.md: Regenerated.
13377         * config/arm/arm-tables.opt: Regenerated.
13378         * doc/invoke.texi: Added entry for core cortex-r7.
13380 2013-03-27  Walter Lee  <walt@tilera.com>
13382         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
13383         double-decrement of next_scratch_regno.
13385 2013-03-27  Walter Lee  <walt@tilera.com>
13387         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
13388         input operands.
13389         (insn_v1mulus): Ditto.
13390         (insn_v2muls): Ditto.
13392 2013-03-27  Walter Lee  <walt@tilera.com>
13394         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
13395         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
13397 2013-03-27  Walter Lee  <walt@tilera.com>
13399         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
13400         (*sibcall_value): Ditto.
13402 2013-03-27  Walter Lee  <walt@tilera.com>
13404         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
13405         (insn_mnz_v8qi): ... this ...
13406         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
13407         vector equivalent.
13408         (insn_v<n>mnz): Replaced by ...
13409         (insn_v1mnz): ... this ...
13410         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
13411         equivalent.
13412         (insn_mz_<mode>): Replaced by ...
13413         (insn_mz_v8qi): ... this ...
13414         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
13415         vector equivalent.
13416         (insn_v<n>mz): Replaced by ...
13417         (insn_v1mz): ... this ...
13418         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
13419         equivalent.
13421 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
13423         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
13425 2013-03-26  Roland McGrath  <mcgrathr@google.com>
13427         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
13428         than fprintf with a non-constant, non-format string.
13430 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13432         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
13433         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
13434         operand 0 predicate.
13435         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
13436         attribute.  Use general_x64nomem_operand as operand 1 predicate.
13437         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
13438         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13439         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
13440         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13441         (mov<mode>_insv_1): Remove expander.  Merge insn with
13442         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
13443         Use general_x64nomem_operand as operand 1 predicate.
13444         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
13445         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
13446         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13447         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
13448         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
13449         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
13450         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13451         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
13452         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
13453         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
13454         (general_x64nomem_operand): Ditto.
13456 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
13458         * config/rtems.opt: Add -pthread option.
13460 2013-03-26  Richard Biener  <rguenther@suse.de>
13462         * alias.c (find_base_term): Avoid redundant and not used recursion.
13463         (base_alias_check): Get the initial base term from the caller.
13464         (true_dependence_1): Compute and pass base terms to base_alias_check.
13465         (write_dependence_p): Likewise.
13466         (may_alias_p): Likewise.
13468 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
13470         * config/aarch64/aarch64.c (aarch64_classify_address): Support
13471         PC-relative load in SI modes and above only.
13473 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
13475         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
13476         * config/arm/iwmmxt.md (WCGR0): Update.
13477         (WCGR1, WCGR2, WCGR3): Likewise.
13479 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13481         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
13482         Use x64 and nox64 isa attributes.
13484 2013-03-26  Richard Biener  <rguenther@suse.de>
13486         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
13487         alignment computations and rely on get_object_alignment_1
13488         for the !TYPE_P case.
13489         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
13491 2013-03-26  Walter Lee  <walt@tilera.com>
13493         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
13494         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
13496 2013-03-25  Jeff Law  <law@redhat.com>
13498         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
13499         check for INTEGRAL_TYPE_P that was missing due to checking in
13500         wrong version of prior patch.
13502 2013-03-25  Walter Lee  <walt@tilera.com>
13504         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
13505         TILEGX_INSN_SHUFFLEBYTES1.
13506         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
13507         shufflebytes1.
13508         (tilegx_builtins): Ditto.
13509         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
13511 2013-03-25  Walter Lee  <walt@tilera.com>
13513         * config/tilegx/tilegx.md (floatsisf2): New pattern.
13514         (floatunssisf2): New pattern.
13515         (floatsidf2): New pattern.
13516         (floatunssidf2): New pattern.
13518 2013-03-25  Walter Lee  <walt@tilera.com>
13520         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
13521         tests for constraint J, K, N, P.
13523 2013-03-25  Walter Lee  <walt@tilera.com>
13525         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
13526         Use indirect/pcrel encoding.
13527         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
13528         Ditto.
13530 2013-03-25  Steve Ellcey  <sellcey@mips.com>
13532         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
13533         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
13534         * config/mips/mips.c (mips_option_override): Set IMADD default.
13535         * config/mips/mips.h (PTF_AVOID_IMADD): New.
13536         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
13537         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
13538         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
13539         * doc/invoke.texi (-mimadd/-mno-imadd): New.
13541 2013-03-25  Jeff Law  <law@redhat.com>
13543         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
13544         slightly to avoid creating and folding useless trees.  Simplify
13545         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
13547 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
13549         * config/i386/i386.md (*zero_extendsidi2): Merge with
13550         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
13551         * config/i386/predicates.md (x86_64_zext_operand): Rename from
13552         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
13553         targets.  Clarify comment.
13555 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13557         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
13558         pass-through jump functions differently.
13559         (ipa_read_jump_function): Likewise.  Also use setter functions to set
13560         up jump functions.
13562 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13564         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
13565         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
13566         process it.
13567         (ipa_get_indirect_edge_target): New function.
13568         (devirtualization_time_bonus): New parameter known_aggs, pass it to
13569         ipa_get_indirect_edge_target.  Update all callers.
13570         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
13571         ipa_get_indirect_edge_target_1 instead of calling
13572         ipa_get_indirect_edge_target.
13573         (create_specialized_node): Pass aggvlas to
13574         ipcp_discover_new_direct_edges.
13576 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13578         * config/arm/arm.md (f_sels, f_seld): New types.
13579         (*cmov<mode>): New pattern.
13580         * config/arm/predicates.md (arm_vsel_comparison_operator): New
13581         predicate.
13583 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13585         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
13586         POSIX-printf for mingw-hosted  builds.
13588 2013-03-25  Richard Biener  <rguenther@suse.de>
13590         PR middle-end/56694
13591         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
13592         must-not-throw stmt location.
13594 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13596         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
13597         Emit load-acquire versions when acq is true.
13598         (arm_emit_store_exclusive): Add rel parameter.
13599         Emit store-release versions when rel is true.
13600         (arm_split_compare_and_swap): Use acquire-release instructions
13601         instead.
13602         of barriers when appropriate.
13603         (arm_split_atomic_op): Likewise.
13604         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
13605         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
13606         (VUNSPEC_SLX): Likewise.
13607         (VUNSPEC_LDA): Likewise.
13608         (VUNSPEC_STL): Likewise.
13609         * config/arm/sync.md (atomic_load<mode>): New pattern.
13610         (atomic_store<mode>): Likewise.
13611         (arm_load_acquire_exclusive<mode>): Likewise.
13612         (arm_load_acquire_exclusivesi): Likewise.
13613         (arm_load_acquire_exclusivedi): Likewise.
13614         (arm_store_release_exclusive<mode>): Likewise.
13616 2013-03-25  Catherine Moore  <clm@codesourcery.com>
13618         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
13619         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
13620         * config/mip/predicates.md (lwsp_swsp_operand,
13621         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
13622         sb16_operand, db4_operand, db7_operand, ib3_operand,
13623         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
13624         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
13625         andi16_operand): New predicates.
13626         * config/mips/mips.md (compression): New attribute.
13627         (enabled): New attribute.
13628         (length): Consider compression in computing length.
13629         (shift_compression): New code attribute.
13630         (*add<mode>3): New operands. Record compression.
13631         (sub<mode>3): Likewise.
13632         (one_cmpl<mode>2): Likewise.
13633         (*and<mode>3): Likewise.
13634         (*ior<mode>3): Likewise.
13635         (unnamed pattern for xor): Likewise.
13636         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
13637         (*<optab><mode>3): Likewise.
13638         (*mov<mode>_internal: Likewise.
13639         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
13640         (mips_unsigned_immediate_p): New.
13641         (umips_lwsp_swsp_address_p): New.
13642         (m16_based_address_p): New.
13643         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
13644         (mips_unsigned_immediate_p): New prototype.
13645         (lwsp_swsp_address_p): New prototype.
13646         (m16_based_address_p): New prototype.
13647         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
13648         (mips_signed_immediate_p): New function.
13649         (m16_based_address_p): New function.
13650         (lwsp_swsp_address_p): New function.
13651         (mips_print_operand_punctuation): Recognize short delay slot insns
13652         for microMIPS.add<mode>3"
13654 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13656         PR target/56720
13657         * config/arm/iterators.md (v_cmp_result): New mode attribute.
13658         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
13660 2013-03-25  Richard Biener  <rguenther@suse.de>
13662         PR tree-optimization/56689
13663         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
13664         any edge.
13666 2013-03-25  Richard Biener  <rguenther@suse.de>
13668         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
13669         of bitmap.
13670         (memory_references): Likewise.
13671         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
13672         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
13673         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
13674         (gather_mem_refs_in_loops): Fold into ...
13675         (analyze_memory_references): ... this.  Move initialization
13676         to tree_ssa_lim_initialize.
13677         (fill_always_executed_in): Rename to ...
13678         (fill_always_executed_in_1): ... this.
13679         (fill_always_executed_in): Move contains_call computation to
13680         this new function from ...
13681         (tree_ssa_lim_initialize): ... here.
13682         (tree_ssa_lim): Call fill_always_executed_in.
13684 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13686         * postreload.c (reload_combine): Fix code detecting returns.
13688 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13690         * function.c (emit_use_return_register_into_block): On cc0 targets,
13691         do not emit the sequence between cc0 setter and user.
13693 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13695         * config/i386/predicates.md (local_symbolic_operand): Interpret
13696         dll-imported symbols as none-local.
13698 2013-03-25  Richard Biener  <rguenther@suse.de>
13700         * tree-ssa-loop-im.c (struct depend): Remove.
13701         (struct lim_aux_data): Make depends a vec of gimples.
13702         (free_lim_aux_data): Adjust.
13703         (add_dependency): Likewise.
13704         (set_level): Likewise.
13706 2013-03-25  Richard Biener  <rguenther@suse.de>
13708         PR middle-end/56434
13709         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
13710         the pointer returned by calls with ECF_MALLOC set.
13712 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13714         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
13716 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13718         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
13719         using MMXMODE mode iterator.
13720         (*move<mode>_internal): Merge with *movv2sf_internal and
13721         *movv2sf_internal_rex64 using MMXMODE mode iterator.
13723 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13725         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
13726         (record_last_mem_set_info): Likewise.
13728         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
13729         of XNEWVEC followed by memset.
13730         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
13732 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13734         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
13735         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
13736         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
13737         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
13738         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
13739         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
13740         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
13741         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
13742         BARRIER_P instead of GET_CODE.
13744 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13746         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
13747         inaccuracy in the probing code.
13749         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
13750         (ctrapdi4): Likewise.
13752 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13754         * calls.c (expand_call): Add missing guard to code handling return
13755         of non-BLKmode structures in MSB.
13756         * function.c (expand_function_end): Likewise.
13758 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13760         * combine.c (try_combine): Adjust comment.  Do not add the set of
13761         insn #0 if the destination indirectly is set or dies in insn #2.
13762         Tidy up code to distribute a new note.
13764 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13766         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
13767         also for alternatives 16 and 17.
13769 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13771         * config/i386/sse.md (*mov<mode>_internal): Merge with
13772         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
13773         Emit insn template depending on type attribute.  Use
13774         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13775         movd instead of movq mnemonic for interunit moves.  Rewrite mode
13776         attribute calculation.  Remove unit attribute calculation.
13777         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
13778         Set prefix_data16 attribute for DImode ssemov types.
13779         Use Ym instead of y for SSE-MMX conversion alternatives.
13780         Reorder operand constraints.
13782 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
13784         * df.h (df_insn_delete): Adjust prototype.
13785         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
13786         and let it decide whether mark the basic block dirty.
13787         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
13788         * df-scan.c (df_insn_info_delete): New helper function, split
13789         off from df_insn_delete.
13790         (df_scan_free_bb_info): Use it.
13791         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
13792         Likewise.
13793         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
13794         that the insn is actually an insn and it has a non-NULL basic block.
13795         Do not mark basic block dirty if only deleting a DEBUG_INSN.
13797 2013-03-22  Richard Biener  <rguenther@suse.de>
13799         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
13800         dep_ref members.
13801         (mem_ref_alloc): Do not allocate them.
13802         (refs_independent_p): Do not query or maintain a cache.
13804 2013-03-22  Richard Biener  <rguenther@suse.de>
13806         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
13807         (gather_mem_refs_in_loops): Do not compute it.
13808         (analyze_memory_references): Do not allocate it.
13809         (tree_ssa_lim_finalize): Do not free it.
13810         (for_all_locs_in_loop): Do not query all_refs_in_loop.
13812 2013-03-22  Richard Biener  <rguenther@suse.de>
13814         * is-a.h (as_a): Use gcc_checking_assert.
13816 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
13818         * config/aarch64/aarch64.c (aarch64_print_operand): New
13819         format specifier for printing a constant in hex.
13820         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
13821         format specifier for printing second operand.
13823 2013-03-22  Richard Biener  <rguenther@suse.de>
13825         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
13826         bitmaps.
13827         (gather_mem_refs_in_loops): Perform store accumulation here.
13828         (create_vop_ref_mapping_loop): Remove.
13829         (create_vop_ref_mapping): Likewise.
13830         (analyze_memory_references): Initialize refs_stored_in_loop.
13831         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
13832         (record_indep_loop): Remove.
13833         (record_dep_loop): New function.
13834         (ref_indep_loop_p_1): Adjust to only walk over references
13835         in the loop, not its subloops.
13836         (ref_indep_loop_p): Rename to ...
13837         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
13838         maintaining a more fine-grained cache.
13839         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
13840         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
13842 2013-03-22  Richard Biener  <rguenther@suse.de>
13844         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
13845         (struct mem_ref): Make accesses_in_loop a vec of a vec of
13846         aggregate mem_ref_loc.
13847         (free_mem_ref_locs): Inline into ...
13848         (memref_free): ... this and adjust.
13849         (mem_ref_alloc): Adjust.
13850         (mem_ref_locs_alloc): Remove.
13851         (record_mem_ref_loc): Adjust.
13852         (get_all_locs_in_loop): Rewrite into ...
13853         (for_all_locs_in_loop): ... this iterator.
13854         (rewrite_mem_ref_loc): New functor.
13855         (rewrite_mem_refs): Use for_all_locs_in_loop.
13856         (sm_set_flag_if_changed): New functor.
13857         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13858         (ref_always_accessed): New functor.
13859         (ref_always_accessed_p): Use for_all_locs_in_loop.
13861 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13863         * tree-pass.h (PROP_gimple_lvec): New.
13864         * passes.c (dump_properties): Handle PROP_gimple_lvec.
13865         (init_optimization_passes): Move pass_lower_vector.
13866         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13867         PROP_gimple_lvec.
13868         (pass_lower_vector): Provide PROP_gimple_lvec.
13869         (pass_lower_vector_ssa): Likewise.
13870         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13872 2013-03-21  Mark Wielaard  <mjw@redhat.com>
13874         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13876 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13878         * config/i386/i386.md (*movdi_internal): Disparage slightly
13879         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
13880         conversion alternatives.
13882 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
13884         PR middle-end/48087
13885         * diagnostic.def (DK_WERROR): New kind.
13886         * diagnostic.h (werrorcount): Define.
13887         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13888         promoted to DK_ERROR, increment DK_WERROR counter instead of
13889         DK_ERROR counter.
13890         * toplev.c (toplev_main): Call print_ignored_options even if
13891         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
13892         even if just werrorcount is non-zero.
13894         PR debug/55608
13895         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13896         on failure.
13897         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13898         (string_cst_pool_decl): New function.
13899         (optimize_one_addr_into_implicit_ptr): New function.
13900         (resolve_addr_in_expr): Optimize DWARF location expression
13901         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13902         which doesn't live in memory, but has DW_AT_location or
13903         DW_AT_const_value, or refers to a string literal, into
13904         DW_OP_GNU_implicit_pointer.
13905         (optimize_location_into_implicit_ptr): New function.
13906         (resolve_addr): If removing DW_AT_location of a variable because
13907         it was DW_OP_addr of address of the variable, but the variable doesn't
13908         live in memory, try to emit const value attribute for the initializer.
13910 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13912         * tree.h (VECTOR_TYPE_P): New macro.
13913         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13914         TYPE_MODE): Use it.
13915         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13916         VEC_COND_EXPR cannot be lvalues.
13917         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13919 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13921         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13922         Restrict the transformation to equal modes.
13924 2013-03-21  Richard Biener  <rguenther@suse.de>
13926         PR tree-optimization/39326
13927         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13928         (MEM_ANALYZABLE): Adjust.
13929         (record_mem_ref_loc): Move bitmap ops ...
13930         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
13931         unanalyzable refs, do not record locations for it.
13932         (analyze_memory_references): Allocate ref zero as shared
13933         unanalyzable ref.
13934         (refs_independent_p): Do not test for unanalyzed mems here.
13935         (ref_indep_loop_p_1): Special-case disambiguation against
13936         the unanalyzed ref.
13937         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
13939 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
13941         * config/arm/arm-protos.h (tune_params): Add
13942         prefer_neon_for_64bits field.
13943         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
13944         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
13945         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
13946         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
13947         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
13948         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
13949         (arm_option_override): Handle -mneon-for-64bits new option.
13950         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13951         (prefer_neon_for_64bits): Declare new variable.
13952         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
13953         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
13954         (arch_enabled): Handle new arch types. Remove support for onlya8
13955         and nota8.
13956         (one_cmpldi2): Use new arch names.
13957         (zero_extend<mode>di2, extend<mode>di2): Ditto.
13958         * config/arm/arm.opt (mneon-for-64bits): Add option.
13959         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
13960         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
13961         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
13962         of onlya8.
13963         * doc/invoke.texi (-mneon-for-64bits): Document.
13965 2013-03-21  Richard Biener  <rguenther@suse.de>
13967         PR tree-optimization/39326
13968         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
13969         (sort_bbs_in_loop_postorder_cmp): New function.
13970         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
13972 2013-03-21  Richard Biener  <rguenther@suse.de>
13974         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
13975         (vect_insert_into_interleaving_chain): Likewise.
13976         (vect_drs_dependent_in_basic_block): Inline ...
13977         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
13978         split out from ...
13979         (vect_analyze_data_ref_dependence): ... here.  Simplify.
13980         (vect_check_interleaving): Simplify.
13981         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
13982         (vect_slp_analyze_data_ref_dependences): ... this new function.
13983         (dr_group_sort_cmp): New function.
13984         (vect_analyze_data_ref_accesses): Compute data-reference groups
13985         here instead of in vect_analyze_data_ref_dependence.  Use
13986         a more efficient algorithm.
13987         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
13988         vect_slp_analyze_data_ref_dependences.  Call
13989         vect_analyze_data_ref_accesses earlier.
13990         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13991         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
13992         (vect_slp_analyze_data_ref_dependences): New prototype.
13994 2013-03-21  Richard Biener  <rguenther@suse.de>
13996         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
13997         ref is stored in the loop.
13998         (find_refs_for_sm): Walk only over all stores.
13999         (store_motion_loop): Allocate from lim_bitmap_obstack.
14000         (store_motion): Likewise.
14002 2013-03-21  Richard Biener  <rguenther@suse.de>
14004         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14005         Update virtual SSA form.
14007 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14009         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14010         * configure: Regenerate.
14011         * config.in: Regenerate.
14012         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14013         if !HAVE_LD_EH_FRAME_CIEV3.
14015 2013-03-21  Richard Biener  <rguenther@suse.de>
14017         * tree-cfg.c (verify_expr_no_block): New function.
14018         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14019         nor DECL_VALUE_EXPR have locations with associated blocks.
14020         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14021         (clear_unused_block_pointer): Remove code dealing with
14022         blocks in DECL_DEBUG_EXPR locations.
14024 2013-03-21  Richard Biener  <rguenther@suse.de>
14026         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14027         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
14028         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14029         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14030         instead of DECL_DEBUG_EXPR_IS_FROM.
14031         * gimplify.c (gimplify_modify_expr): Likewise.
14032         * tree-cfg.c (verify_expr_location_1): Likewise.
14033         * tree-complex.c (create_one_component_var): Likewise.
14034         * tree-sra.c (create_access_replacement): Likewise.
14035         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14036         (clear_unused_block_pointer): Likewise.
14037         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14038         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14039         * var-tracking.c (var_debug_decl): Likewise.
14040         (track_expr_p): Likewise.
14041         * tree-inline.c (add_local_variables): Likewise.  Set
14042         DECL_HAS_DEBUG_EXPR_P after copying it.
14043         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14044         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
14046 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14048         PR bootstrap/56656
14049         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14050         * configure: Regenerate.
14051         * config.in: Regenerate.
14052         * config/i386/i386.md (*movdf_internal): Use
14053         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14054         movd instead of movq mnemonic for interunit moves.
14055         (*movdi_internal): Ditto.
14057 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14059         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14060         (abd<mode>_3): New pattern.
14061         (aba<mode>_3): New pattern.
14062         (fabd<mode>_3): New pattern.
14064 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14066         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14067         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14068         occurrence of REGISTER_PREFIX as its empty string.
14070 2013-03-20  Jeff Law  <law@redhat.com>
14072         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14073         addititional equivalences for equality comparisons between an SSA_NAME
14074         and a constant where the SSA_NAME was set from a widening conversion.
14076 2013-03-20  Walter Lee  <walt@tilera.com>
14078         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14080 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14082         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14083         depending on type attribute.
14084         (*movti_internal): Ditto.
14085         (*movtf_internal): Ditto.
14086         (*movxf_internal): Ditto.
14087         (*movdf_internal): Ditto.
14088         (*movsf_internal): Ditto.
14090 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14092         * config/i386/i386.md (*movti_internal): Set prefix attribute to
14093         maybe_vex for sselog1 and ssemov types.
14094         (*movdi_internal): Reorder operand constraints.
14095         (*movsi_internal): Ditto.  Set prefix attribute to
14096         maybe_vex for sselog1 and ssemov types.
14097         (*movtf_internal): Set prefix attribute to maybe_vex
14098         for sselog1 and ssemov types.
14099         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
14100         DImode ssemov types.  Reorder operand constraints.
14101         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
14102         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
14103         attribute for SImode ssemov types.  Reorder operand constraints.
14105 2013-03-20  Martin Jambor  <mjambor@suse.cz>
14107         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14108         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14110 2013-03-20  Pat Haugen <pthaugen@us.ibm.com>
14112         * config/rs6000/predicates.md (indexed_address, update_address_mem
14113         update_indexed_address_mem): New predicates.
14114         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14115         attribute for load/store instructions.
14116         * config/rs6000/dfp.md (movsd_store): Likewise.
14117         (movsd_load): Likewise.
14118         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14119         (unnamed HI->DI extend define_insn): Likewise.
14120         (unnamed SI->DI extend define_insn): Likewise.
14121         (unnamed QI->SI extend define_insn): Likewise.
14122         (unnamed QI->HI extend define_insn): Likewise.
14123         (unnamed HI->SI extend define_insn): Likewise.
14124         (unnamed HI->SI extend define_insn): Likewise.
14125         (extendsfdf2_fpr): Likewise.
14126         (movsi_internal1): Likewise.
14127         (movsi_internal1_single): Likewise.
14128         (movhi_internal): Likewise.
14129         (movqi_internal): Likewise.
14130         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
14131         attribute for load/store instructions.
14132         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
14133         instructions.
14134         (mov<mode>_softfloat): Likewise.
14135         (mov<mode>_hardfloat32): Likewise.
14136         (mov<mode>_hardfloat64): Likewise.
14137         (mov<mode>_softfloat64): Likewise.
14138         (movdi_internal32): Likewise.
14139         (movdi_internal64): Likewise.
14140         (probe_stack_<mode>): Likewise.
14142 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
14144         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
14145         floating point, and decimal floating point to reload iterator.
14147         * config/rs6000/constraints.md (wl constraint): New constraints to
14148         return FLOAT_REGS if certain options are used to reduce the number
14149         of separate patterns that exist in the file.
14150         (wx constraint): Likewise.
14151         (wz constraint): Likewise.
14153         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14154         -mdebug=reg, print wg, wl, wx, and wz constraints.
14155         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14156         Initialize the reload functions for 64-bit binary/decimal floating
14157         point types.
14158         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14159         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14160         create the buffer on the stack to overcome not having a 32-bit
14161         load and store.
14162         (rs6000_emit_move): Likewise.
14163         (rs6000_secondary_memory_needed_rtx): Likewise.
14164         (rs6000_alloc_sdmode_stack_slot): Likewise.
14165         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14166         via xxlxor, just like DFmode 0.0.
14168         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
14169         define as 1 if we are running on a power7 or newer.
14170         (enum r6000_reg_class_enum): Add new constraints.
14172         * config/rs6000/dfp.md (movsd): Delete, combine with binary
14173         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14174         with other moves by using conditional constraits (wg).  Use LFIWZX
14175         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
14176         (movsd splitter): Likewise.
14177         (movsd_hardfloat): Likewise.
14178         (movsd_softfloat): Likewise.
14180         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
14181         binary and decimal floating point moves.
14182         (fmove_ok): New attributes to combine binary and decimal floating
14183         point moves, and to combine power6x (mfpgpr) moves along normal
14184         floating moves.
14185         (real_value_to_target): Likewise.
14186         (f32_lr): Likewise.
14187         (f32_lm): Likewise.
14188         (f32_li): Likewise.
14189         (f32_sr): Likewise.
14190         (f32_sm): Likewise.
14191         (f32_si): Likewise.
14192         (movsf): Combine binary and decimal floating point moves.  Combine
14193         power6x (mfpgpr) moves with other moves by using conditional
14194         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
14195         (mov<mode> for SFmode/SDmode); Likewise.
14196         (SFmode/SDmode splitters): Likewise.
14197         (movsf_hardfloat): Likewise.
14198         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
14199         (movsf_softfloat): Likewise.
14200         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
14202         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
14203         wx and wz constraints.
14205         * config/rs6000/constraints.md (wg constraint): New constraint to
14206         return FLOAT_REGS if -mmfpgpr (power6x) was used.
14208         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
14209         constraint.
14211         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14212         -mdebug=reg, print wg, wl, wx, and wz constraints.
14213         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14214         Initialize the reload functions for 64-bit binary/decimal floating
14215         point types.
14216         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14217         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14218         create the buffer on the stack to overcome not having a 32-bit
14219         load and store.
14220         (rs6000_emit_move): Likewise.
14221         (rs6000_secondary_memory_needed_rtx): Likewise.
14222         (rs6000_alloc_sdmode_stack_slot): Likewise.
14223         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14224         via xxlxor, just like DFmode 0.0.
14226         * config/rs6000/dfp.md (movdd): Delete, combine with binary
14227         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14228         with other moves by using conditional constraits (wg).  Use LFIWZX
14229         and STFIWX for loading SDmode on power7.
14230         (movdd splitters): Likewise.
14231         (movdd_hardfloat32): Likewise.
14232         (movdd_softfloat32): Likewise.
14233         (movdd_hardfloat64_mfpgpr): Likewise.
14234         (movdd_hardfloat64): Likewise.
14235         (movdd_softfloat64): Likewise.
14237         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
14238         64-bit binary and decimal floating point moves.
14239         (FMOVE64X): Likewise.
14240         (movdf): Combine 64-bit binary and decimal floating point moves.
14241         Combine power6x (mfpgpr) moves with other moves by using
14242         conditional constraits (wg).
14243         (mov<mode> for DFmode/DDmode): Likewise.
14244         (DFmode/DDmode splitters): Likewise.
14245         (movdf_hardfloat32): Likewise.
14246         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
14247         (movdf_softfloat32): Likewise.
14248         (movdf_hardfloat64_mfpgpr): Likewise.
14249         (movdf_hardfloat64): Likewise.
14250         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
14251         (movdf_softfloat64): Likewise.
14252         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
14253         (reload_<mode>_load): Move to later in the file so they aren't in
14254         the middle of the floating point move insns.
14255         (reload_<mode>_store): Likewise.
14257         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
14258         constraint.
14260         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
14261         constraint if -mdebug=reg.
14262         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
14263         Enable using dd reload support if needed.
14265         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
14266         binary and decimal floating point moves in rs6000.md.
14267         (movtd_internal): Likewise.
14269         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
14270         decimal floating point moves.
14271         (movtf): Likewise.
14272         (movtf_internal): Likewise.
14273         (mov<mode>_internal, TDmode/TFmode): Likewise.
14274         (movtf_softfloat): Likewise.
14275         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
14277         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
14278         movdi_internal64, using wg constraint for move direct operations.
14279         (movdi_internal64): Likewise.
14281         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
14282         MODES_TIEABLE_P for selected modes.  Print the numerical value of
14283         the various virtual registers. Use GPR/FPR first/last values,
14284         instead of hard coding the register numbers.  Print which modes
14285         have reload functions registered.
14286         (rs6000_option_override_internal): If -mdebug=reg, trace the options
14287         settings before/after setting cpu, target and subtarget settings.
14288         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
14289         and for secondary reload failures in rs6000_secondary_reload_inner.
14290         (rs6000_secondary_reload_fail): Likewise.
14291         (rs6000_secondary_reload_inner): Likewise.
14293         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
14294         macros for first/last GPR and FPR registers.
14295         (LAST_GPR_REGNO): Likewise.
14296         (FIRST_FPR_REGNO): Likewise.
14297         (LAST_FPR_REGNO): Likewise.
14299         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
14300         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
14301         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
14302         (vcond<mode><mode>): Likewise.
14303         (vcondu<mode><mode>): Likewise.
14304         (vector_gtu<mode>): Likewise.
14305         (vector_gte<mode>): Likewise.
14306         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
14307         to prevent the compiler from converting DImode operations to TImode.
14308         (ior<mode>3): Likewise.
14309         (and<mode>3): Likewise.
14310         (one_cmpl<mode>2): Likewise.
14311         (nor<mode>3): Likewise.
14312         (andc<mode>3): Likewise.
14314         * config/rs6000/constraints.md (wt constraint): New constraint
14315         that returns VSX_REGS if TImode is allowed in VSX registers.
14317         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
14318         constant under VSX.
14320         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
14321         similar to TImode, but it is restricted to being in the GPRs.
14323         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
14324         TImode to occupy a single VSX register.
14326         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
14327         -mvsx-timode for power7/power8.
14328         (power7 cpu): Likewise.
14329         (power8 cpu): Likewise.
14331         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
14332         sure that TFmode/TDmode take up two registers if they are ever
14333         allowed in the upper VSX registers.
14334         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
14335         registers.
14336         (rs6000_init_hard_regno_mode_ok): Likewise.
14337         (rs6000_debug_reg_global): Add debugging for PTImode and wt
14338         constraint.  Print if LRA is turned on.
14339         (rs6000_option_override_internal): Give an error if -mvsx-timode
14340         and VSX is not enabled.
14341         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
14342         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
14343         to reg+offset addressing.  Use PTImode when checking offset
14344         addresses for validity.
14345         (reg_offset_addressing_ok_p): Likewise.
14346         (rs6000_legitimate_offset_address_p): Likewise.
14347         (rs6000_legitimize_address): Likewise.
14348         (rs6000_legitimize_reload_address): Likewise.
14349         (rs6000_legitimate_address_p): Likewise.
14350         (rs6000_eliminate_indexed_memrefs): Likewise.
14351         (rs6000_emit_move): Likewise.
14352         (rs6000_secondary_reload): Likewise.
14353         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
14354         reloads to fpr registers to continue to use reg+offset addressing,
14355         but 64-bit reloads to altivec registers need reg+reg addressing.
14356         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
14357         it.  Treat LO_SUM like a PLUS operation.
14358         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
14359         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
14360         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
14361         registers to share a register with a smaller sized type, since VSX
14362         puts scalars in the upper 64-bits.
14363         (print_operand): Add support for PTImode.
14364         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
14365         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
14366         registers, but don't have arithmetic support.
14367         (rs6000_memory_move_cost): Add test for VSX.
14368         (rs6000_opt_masks): Add -mvsx-timode.
14370         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
14371         for TImode.
14372         (VSs): Likewise.
14373         (VSr): Use wt constraint for TImode.
14374         (VSv): Drop TImode support.
14375         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
14376         (vsx_movti_64bit): Likewise.
14377         (vsx_movti_32bit): Likewise.
14378         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
14379         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
14380         one '?' on the appropriate output constraint.  Do not allow TImode
14381         logical operations on 32-bit systems.
14382         (vsx_ior<mode>3): Likewise.
14383         (vsx_xor<mode>3): Likewise.
14384         (vsx_one_cmpl<mode>2): Likewise.
14385         (vsx_nor<mode>3): Likewise.
14386         (vsx_andc<mode>3): Likewise.
14387         (vsx_concat_<mode>): Likewise.
14388         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
14390         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
14391         OPTION_MASK_VSX_TIMODE.
14392         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
14393         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
14395         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
14396         (TI2 iterator): New iterator for TImode, PTImode.
14397         (wd mode attribute): Add values for vector types.
14398         (movti_string): Replace TI move operations with operations for TImode
14399         and PTImode.  Add support for TImode being allowed in VSX registers.
14400         (mov<mode>_string, TImode/PTImode): Likewise.
14401         (movti_ppc64): Likewise.
14402         (mov<mode>_ppc64, TImode/PTImode): Likewise.
14403         (TI mode splitters): Likewise.
14405         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
14406         constraint.
14408 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
14410         PR tree-optimization/56355
14411         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
14412         Also handle integers with undefined overflow.
14414 2013-03-20  Catherine Moore  <clm@codesourcery.com>
14415             Maciej W. Rozycki  <macro@codesourcery.com>
14416             Tom de Vries  <tom@codesourcery.com>
14417             Nathan Sidwell <nathan@codesourcery.com>
14418             Iain Sandoe  <iain@codesourcery.com>
14419             Nathan Froyd  <froydnj@codesourcery.com>
14420             Chao-ying Fu <fu@mips.com>
14422         * doc/extend.texi: (micromips, nomicromips, nocompression):
14423         Document new function attributes.
14424         * doc/invoke.texi (minterlink-compressed, mmicromips,
14425         m14k, m14ke, m14kec): Document new options.
14426         (minterlink-mips16): Update documentation.
14427         * doc/md.texi (ZC, ZD): Document new constraints.
14428         * configure.ac (gcc_cv_as_micromips): Check if linker
14429         supports the .set micromips directive.
14430         * configure: Regenerate.
14431         * config.in: Regenerate.
14432         * config/mips/mips-tables.opt: Regenerate.
14433         * config/mips/micromips.md: New file.
14434         * constraints.md (ZC, ZD): New constraints.
14435         * config/mips/predicates.md (movep_src_register): New predicate.
14436         (movep_src_operand): New predicate.
14437         (non_volatile_mem_operand): New predicate.
14438         * config/mips/mips.md (multimem): New type.
14439         (length): Differentiate between 17-bit and 18-bit branch offsets.
14440         (MOVEP1, MOVEP2): New mode iterator.
14441         (mov_<load>l): Use ZC constraint.
14442         (mov_<load>r): Likewise.
14443         (mov_<store>l): Likewise.
14444         (mov_<store>r): Likewise.
14445         (*branch_equality<mode>_inverted): Add microMIPS support.
14446         (*branch_equality<mode>): Likewise.
14447         (*jump_absolute): Likewise.
14448         (indirect_jump_<mode>): Likewise.
14449         (tablejump_<mode>): Likewise.
14450         (<optab>_internal): Likewise.
14451         (sibcall_internal): Likewise.
14452         (sibcall_value_internal): Likewise.
14453         (prefetch): Use constraint ZD.
14454         * config/mips/mips.opt (minterlink-compressed): New option.
14455         (minterlink-mips16): Now an alias for minterlink-compressed.
14456         (mmicromips): New option.
14457         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14458         (compare_and_swap_12): Likewise.
14459         (sync_add<mode>): Likewise.
14460         (sync_<optab>_12): Likewise.
14461         (sync_old_<optab>_12): Likewise.
14462         (sync_new_<optab>_12): Likewise.
14463         (sync_nand_12): Likewise.
14464         (sync_old_nand_12): Likewise.
14465         (sync_new_nand_12): Likewise.
14466         (sync_sub<mode>): Likewise.
14467         (sync_old_add<mode>): Likewise.
14468         (sync_old_sub<mode>): Likewise.
14469         (sync_new_add<mode>): Likewise.
14470         (sync_new_sub<mode>): Likewise.
14471         (sync_<optab><mode>): Likewise.
14472         (sync_old_<optab><mode>): Likewise.
14473         (sync_new_<optab><mode>): Likewise.
14474         (sync_nand<mode>): Likewise.
14475         (sync_old_nand<mode>): Likewise.
14476         (sync_new_nand<mode>): Likewise.
14477         (sync_lock_test_and_set<mode>): Likewise.
14478         (test_and_set_12): Likewise.
14479         (atomic_compare_and_swap<mode>): Likewise.
14480         (atomic_exchange<mode>_llsc): Likewise.
14481         (atomic_fetch_add<mode>_llsc): Likewise.
14482         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14483         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14484         (umips_save_restore_pattern_p): Likewise.
14485         (umips_load_store_pair_p): Likewise.
14486         (umips_output_load_store_pair): Likewise.
14487         (umips_movep_target_p): Likewise.
14488         (umips_12bit_offset_address_p): Likewise.
14489         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14490         (mips_base_mips16): Rename this...
14491         (mips_base_compression_flags): ...to this. Update all uses.
14492         (mips_attribute_table): Add micromips, nomicromips and nocompression.
14493         (mips_mips16_decl_p): Delete.
14494         (mips_nomips16_decl_p): Delete.
14495         (mips_get_compress_on_flags): New function.
14496         (mips_get_compress_off_flags): New function.
14497         (mips_get_compress_mode): New function.
14498         (mips_get_compress_on_name): New function.
14499         (mips_get_compress_off_name): New function.
14500         (mips_insert_attributes): Support multiple compression types.
14501         (mips_merge_decl_attributes): Likewise.
14502         (umips_12bit_offset_address_p): New function.
14503         (mips_start_function_definition): Emit .set micromips directive.
14504         (mips_call_may_need_jalx_p): New function.
14505         (mips_function_ok_for_sibcall): Add microMIPS support.
14506         (mips_print_operand_punctuation): Support short delay slots and
14507         compact jumps.
14508         (umips_swm_mask, umips_swm_encoding): New.
14509         (umips_build_save_restore): New function.
14510         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14511         (was_mips16_p): Remove.
14512         (old_compression_mode): New.
14513         (mips_set_compression_mode): New function.
14514         (mips_set_current_function): Add microMIPS support.
14515         (mips_option_override): Likewise.
14516         (umips_save_restore_pattern_p): New function.
14517         (umips_output_save_restore): New function.
14518         (umips_load_store_pair_p_1): New function.
14519         (umips_load_store_pair_p): New function.
14520         (umips_output_load_store_pair_1): New function.
14521         (umips_output_load_store_pair): New function.
14522         (umips_movep_target_p) New function.
14523         (mips_prepare_pch_save): Add microMIPS support.
14524         * config/mips/mips.h (TARGET_COMPRESSION): New.
14525         (TARGET_CPU_CPP_BUILTINS): Update macro
14526         to use new compression flags and to support microMIPS.
14527         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14528         (MIPS_ARCH_FLOAT_SPEC): Likewise.
14529         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14530         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14531         (ASM_SPEC): Support mmicromips and mno-micromips.
14532         (M16STORE_REG_P): New macro.
14533         (MIPS_CALL): Support TARGET_MICROMIPS.
14534         (MICROMIPS_J): New macro.
14535         (mips_base_mips16): Rename this...
14536         (mips_base_compression_flags): ...to this.
14537         (UMIPS_12BIT_OFFSET_P): New macro.
14538         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14539         (MULTILIB_DIRNAMES): Likewise.
14540 2013-03-20  Richard Biener  <rguenther@suse.de>
14542         PR tree-optimization/56661
14543         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
14544         the result does not have to be distinct.
14546 2013-03-20  Richard Biener  <rguenther@suse.de>
14548         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
14549         remap_gimple_op_r.
14551 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14552             Steven Bosscher <steven@gcc.gnu.org>
14554         PR rtl-optimization/56605
14555         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
14557 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14559         PR bootstrap/56656
14560         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
14561         that require movd instead of movq.
14563 2013-03-20  Richard Biener  <rguenther@suse.de>
14565         * tree-ssa-structalias.c (struct variable_info): Add pointer
14566         to the first field of an aggregate with sub-vars.  Make
14567         this and the pointer to the next subfield its ID.
14568         (vi_next): New function.
14569         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
14570         storedanything_id, integer_id): Increment by one.
14571         (new_var_info, get_call_vi, lookup_call_clobber_vi,
14572         get_call_clobber_vi): Adjust.
14573         (solution_set_expand): Simplify and speedup.
14574         (solution_set_add): Inline into ...
14575         (set_union_with_increment): ... this.  Adjust accordingly.
14576         (do_sd_constraint): Likewise.
14577         (do_ds_constraint): Likewise.
14578         (do_complex_constraint): Simplify.
14579         (build_pred_graph): Adjust.
14580         (solve_graph): Likewise.  Simplify and speedup.
14581         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
14582         get_constraint_for_component_ref, get_constraint_for_1,
14583         first_vi_for_offset, first_or_preceding_vi_for_offset,
14584         create_function_info_for, create_variable_info_for_1,
14585         create_variable_info_for, intra_create_variable_infos): Adjust.
14586         (init_base_vars): Push NULL for ID zero.
14587         (compute_points_to_sets): Adjust.
14589 2013-03-20  Richard Biener  <rguenther@suse.de>
14591         * cfgloop.c (verify_loop_structure): Streamline and avoid
14592         ICEing on corrupt loop tree.
14593         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
14594         loop tree.
14596 2013-03-20  Richard Biener  <rguenther@suse.de>
14598         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
14599         check whether an SSA update is needed.
14601 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
14603         * config/mips/constraints.md (T): Rename to...
14604         (Yf): ...this.
14605         (U): Rename to...
14606         (Yd): ...this.
14607         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
14608         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
14610 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14612         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
14613         (*subsi3_carryin_uxtw): Likewise.
14615 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14617         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
14618         (*rorsi3_insn_uxtw): Likewise.
14620 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14622         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
14623         (*extrsi5_insn_uxtw): Likewise.
14625 2013-03-19  Richard Biener  <rguenther@suse.de>
14627         PR tree-optimization/56273
14628         * passes.c (init_optimization_passes): Move second VRP after DOM.
14630 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
14632         * config/i386/i386.md (*movti_internal): Merge from
14633         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
14634         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
14635         nox64 isa attributes.
14637 2013-03-18  Richard Biener  <rguenther@suse.de>
14639         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
14640         (unite): Likewise.
14641         (merge_node_constraints): Likewise.
14642         (build_succ_graph): Likewise.
14643         (valid_graph_edge): Inline into single caller.
14644         (unify_nodes): Likewise.  Use bitmap_set_bit return value
14645         and cache varinfo.
14646         (scc_visit): Fix formatting and variable use.
14647         (do_sd_constraint): Use gcc_checking_assert.
14648         (do_ds_constraint): Likewise.
14649         (do_complex_constraint): Likewise.
14650         (condense_visit): Likewise.  Cleanup.
14651         (dump_pred_graph): New function.
14652         (perform_var_substitution): Dump the pred-graph before
14653         variable substitution.
14654         (find_equivalent_node): Use gcc_checking_assert.
14655         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
14657 2013-03-18  Richard Biener  <rguenther@suse.de>
14659         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
14660         Remove cond_expr_stmt_list argument and do not gimplify the
14661         built expression.
14662         (vect_loop_versioning): Adjust.
14663         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
14664         Cleanup to use less temporaries.
14665         (vect_create_data_ref_ptr): Cleanup.
14667 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14669         PR tree-optimization/56635
14670         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
14671         require types_compatible_p types.
14673 2013-03-18  Nick Clifton  <nickc@redhat.com>
14675         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
14676         spurious backslash.
14678         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
14679         Add missing line to comment describing function.
14681 2013-03-18  Richard Biener  <rguenther@suse.de>
14683         PR tree-optimization/56210
14684         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14685         Handle string / character search functions.
14686         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
14688 2013-03-18  Richard Biener  <rguenther@suse.de>
14690         PR middle-end/56483
14691         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
14692         and implement properly.
14693         * gimple.h (gimple_cond_single_var_p): Remove.
14695 2013-03-18  Richard Biener  <rguenther@suse.de>
14697         * tree-data-ref.h (find_data_references_in_loop): Declare.
14698         * tree-data-ref.c (get_references_in_stmt): Use a stack
14699         vector pre-allocated in the callers.
14700         (find_data_references_in_stmt): Adjust.
14701         (graphite_find_data_references_in_stmt): Likewise.
14702         (create_rdg_vertices): Likewise.
14703         (find_data_references_in_loop): Export.
14704         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14705         Compute dependences here...
14706         (vect_analyze_data_refs): ...not here.  When we encounter
14707         a non-vectorizable data reference in basic-block vectorization
14708         truncate the data reference vector.  Do not bother to
14709         fixup data-dependence information for gather loads.
14710         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
14711         of data references, as reported.
14713 2013-03-18  Richard Biener  <rguenther@suse.de>
14715         PR tree-optimization/3713
14716         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
14717         has_constants and expr.
14718         (stmt_has_constants): Properly valueize SSA names when deciding
14719         whether the stmt has constants.
14721 2013-03-18  Richard Biener  <rguenther@suse.de>
14723         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
14724         whole function when there is nothing to do.
14725         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
14726         * tree-vectorizer.c (vectorize_loops): Update virtual and
14727         loop-closed SSA once.
14728         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
14730 2013-03-18  Richard Biener  <rguenther@suse.de>
14732         PR middle-end/56113
14733         * domwalk.c (bb_postorder): New global static.
14734         (cmp_bb_postorder): New function.
14735         (walk_dominator_tree): Replace scheme imposing an order for
14736         visiting dominator sons by one sorting them at the time they
14737         are pushed on the stack.
14739 2013-03-18  Richard Biener  <rguenther@suse.de>
14741         PR tree-optimization/39326
14742         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
14743         (struct mem_ref): Replace mem member with ao_ref typed member.
14744         (MEM_ANALYZABLE): Adjust.
14745         (memref_eq): Likewise.
14746         (mem_ref_alloc): Likewise.
14747         (gather_mem_refs_stmt): Likewise.
14748         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
14749         (execute_sm_if_changed_flag_set): Adjust.
14750         (execute_sm): Likewise.
14751         (ref_always_accessed_p): Likewise.
14752         (refs_independent_p): Likewise.
14753         (can_sm_ref_p): Likewise.
14755 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14757         PR c/56566
14758         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
14759         return 1 even for !unsignedp.
14761 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14763         * config/i386/i386.md (isa): Add x64 and nox64.
14764         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
14765         (*pushtf): Enable *roF alternative for x64 isa only.
14766         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
14767         mode attribute of integer alternatives to DImode for TARGET_64BIT.
14768         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
14769         (*movtf_internal): Merge from *movtf_internal_rex64 and
14770         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
14771         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
14772         nox64 isa attributes.
14773         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
14774         nox64 isa attributes.
14775         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
14777 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14779         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14781 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
14783         PR target/56640
14784         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
14785         class names.  Remove trailing comma after "ALL_REGS".
14787 2013-03-16  Jan Hubicka  <jh@suse.cz>
14789         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
14790         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
14791         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
14792         of cgraph_get_create_node.
14793         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14795 2013-03-16  Jason Merrill  <jason@redhat.com>
14797         PR debug/49090
14798         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
14799         with DW_AT_default_value.
14801 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
14803         * BASE-VER: Set to 4.9.0.
14805 2013-03-14  Andi Kleen  <ak@linux.intel.com>
14807         PR target/56619
14808         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
14809         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
14810         Document _x* TSX intrinsics.
14812 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
14813             David Holsgrove  <david.holsgrove@xilinx.com>
14815         * configure.ac: Add MicroBlaze TLS support detection.
14816         * configure: Regenerate.
14817         * config/microblaze/microblaze-protos.h
14818         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
14819         symbol_mentioned_p, label_mentioned_p): Add prototypes.
14820         * config/microblaze/microblaze.c (microblaze_address_type): Add
14821         ADDRESS_TLS and tls_reloc address types.
14822         (microblaze_address_info): Add tls_reloc.
14823         (TARGET_HAVE_TLS): Define.
14824         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
14825          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
14826          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
14827         load_tls_operand,  microblaze_call_tls_get_addr,
14828         microblaze_legitimize_tls_address): New functions.
14829         (microblaze_classify_unspec): Handle UNSPEC_TLS.
14830         (get_base_reg): Use microblaze_tls_symbol_p.
14831         (microblaze_classify_address): Handle TLS.
14832         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
14833         label_mentioned_p and microblaze_tls_referenced_p.
14834         (microblaze_legitimize_address): Handle TLS.
14835         (microblaze_address_insns): Handle ADDRESS_TLS.
14836         (pic_address_needs_scratch): Handle TLS.
14837         (print_operand_address): Handle TLS.
14838         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
14839         (microblaze_expand_move): Handle TLS.
14840         (microblaze_legitimate_constant_p): Check
14841         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
14842         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14843         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
14844         (PIC_OFFSET_TABLE_REGNUM): Set.
14845         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
14846         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
14847         (addsi3, movsi_internal2, movdf_internal): Update constraints
14848         * config/microblaze/predicates.md (arith_plus_operand): Define
14849         (move_operand): Redefine as move_src_operand,
14850         check microblaze_tls_referenced_p.
14852 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14854         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
14855         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14857 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14859         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14860         CC mode for AND.
14862 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
14864         PR tree-optimization/53265
14865         * common.opt (Waggressive-loop-optimizations): New option.
14866         * tree-ssa-loop-niter.c: Include tree-pass.h.
14867         (do_warn_aggressive_loop_optimizations): New function.
14868         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
14869         if number_of_latch_executions returned constant.
14870         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14871         early.  If number_of_latch_executions returned constant, set
14872         nb_iterations_upper_bound back to it.
14873         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14874         field.
14875         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14876         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14878         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14879         (MULTILIB_OSDIRNAMES): Set.
14880         * genmultilib: If defaultosdirname doesn't start with :: , set
14881         defaultosdirname2 instead, clear it and emit two . multilib_raw
14882         entries instead of just one.
14884 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
14886         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14887         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14888         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14889         (SUBTARGET_OVERRIDE_OPTIONS): New.
14891 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
14893         PR target/49880
14894         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14895         (musermode): Convert to Var(TARGET_USERMODE).
14896         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14897         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14898         * config/sh/sh.c (sh_option_override): Use
14899         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14900         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14901         condition.
14902         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14903         TARGET_SH4.
14904         (udivsi3_i4_single, divsi3_i4_single): Use
14905         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14907 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
14909         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14910         default setting.
14912 2013-03-13  Richard Biener  <rguenther@suse.de>
14914         PR tree-optimization/56608
14915         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14916         calls when vectorizing basic-blocks.
14918 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
14920         PR plugins/45078
14921         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14922         tm_file.
14924 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
14926         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14928 2013-03-11  Jan Hubicka  <jh@suse.cz>
14930         PR lto/56557
14931         * lto-streamer-out.c (output_symbol_p): Skip references from
14932         constructors of external variables.
14934 2013-03-11  Jan Hubicka  <jh@suse.cz>
14936         PR middle-end/56571
14937         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
14938         from pseudos.
14939         * emit-rtl.c (verify_rtx_sharing): Likewise.
14940         (copy_insn_1): Likewise.
14941         * rtl.c (copy_rtx): Likewise.
14943 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
14945         PR target/56591
14946         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
14947         output_operand_lossage message.
14949 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
14951         PR target/56470
14952         * arm.c (shift_op): Validate RTL pattern on the fly.
14953         (arm_print_operand, case 'S'): Don't use shift_operator to validate
14954         the RTL.
14956 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
14958         PR target/56347
14959         * config/pa/pa.md (call_value): Check for calls to powf and direct to
14960         new call patterns that clobber %fr12.
14961         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
14962         split and postreload patterns.
14963         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
14964         registers %fr12 and %fr12R as call used.
14966 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
14968         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
14969         (canon_address, record_store, replace_read, check_mem_read_rtx,
14970         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
14971         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
14972         rest_of_handle_dse): Likewise.
14974 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
14976         PR middle-end/56524
14977         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
14978         Add base_optabs.
14979         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
14980         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
14981         (save_optabs_if_changed): Replace with...
14982         (init_tree_optimization_optabs): ...this.
14983         * optabs.c (save_optabs_if_changed): Rename to...
14984         (init_tree_optimization_optabs): ...this.  Take the optimization node
14985         as argument.  Do nothing if the base optabs are already correct.
14986         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
14987         to recompute optabs.
14988         * function.h (function): Remove optabs field.
14989         * function.c (invoke_set_current_function_hook): Call
14990         init_tree_optimization_optabs.  Use the result to initialize
14991         this_fn_optabs.
14993 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
14995         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
14996         if GTMA_HAS_NO_INSTRUMENTATION.
14997         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
14998         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
14999         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15000         * gimple-pretty-print.c (dump_gimple_transaction): Handle
15001         GTMA_HAS_NO_INSTRUMENTATION.
15003 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
15005         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15006         libasan_preinit.o.
15008 2013-03-08  Marek Polacek  <polacek@redhat.com>
15009             Jakub Jelinek  <jakub@redhat.com>
15011         PR tree-optimization/56478
15012         * predict.c (is_comparison_with_loop_invariant_p): Change the
15013         type of loop_step to tree.
15014         (predict_loops): Adjust.
15015         (predict_iv_comparison): Perform the computations on double_ints.
15017 2013-03-08  Richard Biener  <rguenther@suse.de>
15019         PR tree-optimization/56570
15020         * tree-cfg.c (verify_expr_location_1): Verify locations for
15021         DECL_DEBUG_EXPR.
15022         * tree-sra.c (create_access_replacement): Strip locations
15023         from DECL_DEBUG_EXPRs.
15025 2013-03-08  Richard Biener  <rguenther@suse.de>
15027         * tree-inline.c (expand_call_inline): Do not associate
15028         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15029         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15031 2013-03-08  Richard Biener  <rguenther@suse.de>
15033         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15034         or block changes with -Og.  Fix for location / block encoding
15035         changes and PHI arguments with locations.
15037 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
15039         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15040         for all counters.
15041         (struct output_info): Likewise.
15042         (register_overhead): Remove bad gcc_assert.
15043         (bitmap_find_bit): If there is only a single bitmap element, do not
15044         count a miss as a search.
15045         (print_statistics): Update for counter type changes.
15046         (dump_bitmap_statistics): Likewise.  Print headers such that they
15047         are properly lined up with the printed counters.
15049 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
15051         PR tree-optimization/56559
15052         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15053         check that it has only a single use.
15055 2013-03-07  Richard Biener  <rguenther@suse.de>
15057         * doc/invoke.texi (fwhole-program): Discourage use in combination
15058         with -flto.
15060 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
15062         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15064         PR tree-optimization/56539
15065         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15066         instead of GSI_CONTINUE_LINKING as last argument to
15067         force_gimple_operand_gsi.  Adjust function comment.
15069         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15070         aarch64-cores.def.
15072         PR middle-end/56548
15073         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15074         promoted mode, convert the result back to the original mode.
15076 2013-03-06  Richard Biener  <rguenther@suse.de>
15078         PR middle-end/56294
15079         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15080         (insert_updated_phi_nodes_compare_uids): New function.
15081         (update_ssa): Sort symbols_to_rename after UID before
15082         traversing it to insert PHI nodes.
15084 2013-03-06  Richard Biener  <rguenther@suse.de>
15086         PR middle-end/50494
15087         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15088         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15090         Revert
15091         2013-02-13  Richard Biener  <rguenther@suse.de>
15093         PR lto/50494
15094         * varasm.c (output_constant_def_1): Get the decl representing
15095         the constant as argument.
15096         (output_constant_def): Wrap output_constant_def_1.
15097         (make_decl_rtl): Use output_constant_def_1 with the decl
15098         representing the constant.
15099         (build_constant_desc): Optionally re-use a decl already
15100         representing the constant.
15101         (tree_output_constant_def): Adjust.
15103 2013-03-06  Joey Ye  <joey.ye@arm.com>
15105         PR lto/50293
15106         * gcc.c (convert_white_space): New function.
15107         (main): Handles white space in function name.
15109 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
15111         PR target/56529
15112         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15113         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
15114         to SH_DIV_CALL_TABLE for TARGET_SH2.
15115         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15116         list.
15117         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15118         call-table options.
15120 2013-03-05  Sterling Augustine  <saugustine@google.com>
15121             Cary Coutant  <ccoutant@google.com>
15123         PR debug/55364
15124         * dwarf2out.c (resolve_addr): Don't call
15125         remove_loc_list_addr_table_entries a second time for the same
15126         expression.
15128 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15130         PR debug/56510
15131         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
15132         (avoid_complex_debug_insns): New function.
15133         (expand_debug_locations): Call it.
15135         PR rtl-optimization/56484
15136         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
15137         lifetimes of hard registers on small register class machines.
15139 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15141         * config/microblaze/microblaze-protos.h: Rename
15142         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
15143         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
15144         fast_interrupt.
15145         (microblaze_fast_interrupt_function_p): New function.
15146         (microblaze_is_interrupt_handler): Rename to
15147         microblaze_is_interrupt_variant and add fast_interrupt check.
15148         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
15149         (save_restore_insns): Likewise.
15150         (compute_frame_size): Likewise.
15151         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
15152         (microblaze_globalize_label): Likewise.
15153         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
15154         * config/microblaze/microblaze.md: Use wrapper
15155         microblaze_is_interrupt_variant.
15157 2013-03-05  Kai Tietz  <ktietz@redhat.com>
15159         * sdbout.c (sdbout_one_type): Switch to current function's section
15160         supporting cold/hot.
15162 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15164         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
15165         -mxl-reorder.
15167 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15169         PR middle-end/56461
15170         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
15171         if VALGRIND_GET_VBITS is defined, temporarily make object
15172         memory all defined, and restore previous valgrind addressability
15173         and definability afterwards.  Free this_object at the end.
15175         PR middle-end/56461
15176         * lra.c (lra): Call lra_clear_live_ranges if live_p,
15177         right before calling lra_create_live_ranges, also call it
15178         when clearing live_p.  Only call lra_clear_live_ranges
15179         at the end if live_p.
15181         PR middle-end/56461
15182         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
15184 2013-03-05  Richard Biener  <rguenther@suse.de>
15186         PR tree-optimization/56521
15187         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
15188         value-id.
15190 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
15192         PR c++/55135
15193         * except.h (remove_unreachable_eh_regions): New prototype.
15194         * except.c (remove_eh_handler_splicer): New function, split out
15195         of remove_eh_handler.
15196         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
15197         warning about running it on many EH regions one at a time.
15198         (remove_unreachable_eh_regions_worker): New function, walk the
15199         EH tree in depth-first order and remove non-marked regions.
15200         (remove_unreachable_eh_regions): New function.
15201         * tree-eh.c (mark_reachable_handlers): New function, split out
15202         from remove_unreachable_handlers.
15203         (remove_unreachable_handlers): Use mark_reachable_handlers and
15204         remove_unreachable_eh_regions.
15205         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
15206         and remove_unreachable_eh_regions.
15208 2013-03-05  Richard Biener  <rguenther@suse.de>
15210         PR middle-end/56525
15211         * loop-init.c (fix_loop_structure): Remove loops in two stages,
15212         not freeing them until the end.
15214 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15216         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
15218 2013-03-05  Richard Biener  <rguenther@suse.de>
15220         PR tree-optimization/56270
15221         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
15222         of loads after scheduling an SLP instance.
15224 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15226         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
15227         tic6x.exp.
15228         (check_gcc_parallelize): Run guality.exp as a separate job from
15229         vect.exp with unsorted.exp and $(dg_target_exps) separately from
15230         struct-layout-1.exp with stackalign.exp.
15232         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
15234         PR middle-end/56461
15235         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
15236         load_index sbitmap even if some bit in it isn't set.
15238         PR middle-end/56461
15239         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
15240         (discover_iteration_bound_by_body_walk): Change queues to
15241         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
15242         spelling in comment.  Call safe_push on queues[bound_index] directly.
15243         Release queues[queue_index] in every iteration unconditionally.
15244         Release bounds vector.
15246         PR middle-end/56461
15247         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
15248         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
15249         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
15250         inner_phis vector.
15252 2013-03-05  Richard Biener  <rguenther@suse.de>
15254         PR lto/56515
15255         * tree-inline.c (remap_blocks_to_null): New function.
15256         (expand_call_inline): When expanding a call stmt without
15257         an associated block inline remap all callee blocks to NULL.
15259 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15261         PR rtl-optimization/56494
15262         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
15263         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
15264         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
15266         PR middle-end/56461
15267         * sel-sched-ir.c (free_sched_pools): Release
15268         succs_info_pool.stack[succs_info_pool.max_top] vectors too
15269         if succs_info_pool.max_top isn't -1.
15271         PR bootstrap/56509
15272         * opts.c (opts_obstack, opts_concat): Moved to...
15273         * opts-common.c (opts_obstack, opts_concat): ... here.
15275 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
15277         PR middle-end/56461
15278         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
15280 2013-03-04  Martin Jambor  <mjambor@suse.cz>
15282         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
15283         all appropriate places.
15285 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
15287         PR tree-optimization/56424
15288         * ipa-split.c (split_function): Do not set the RSO flag if result is
15289         not by reference and its type is a register type.
15291 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
15293         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
15294         (microblaze_legitimate_pic_operand): Likewise
15295         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
15296         new function microblaze_legitimate_pic_operand
15297         * config/microblaze/microblaze-protos.h
15298         (microblaze_legitimate_pic_operand): Declare.
15300 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15302         * config/microblaze/predicates.md (call_insn_simple_operand):
15303         New predicate for supported rtx code types.
15304         * config/microblaze/microblaze.md (call_internal1): Use
15305         call_insn_simple_operand predicate.
15307 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
15309         PR middle-end/56461
15310         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
15311         partitions.ordered_remove.
15313         PR middle-end/56461
15314         * tree-vect-stmts.c (vectorizable_conversion): Don't call
15315         vec_oprnds0.create (1) for modifier == NONE.
15317         PR middle-end/56461
15318         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
15319         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
15320         vec_oprnds1 right before pushing anything to it for
15321         scalar_shift_arg.
15323         PR middle-end/56461
15324         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
15325         set nbbs to 0 instead of having separate code path.
15326         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
15327         instead of false as last argument if returning NULL.
15329 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
15331         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
15332         the attribute is now called "target" instead of "option".
15333         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
15334         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
15335         attribute/pragma name for TARGET_OPTION_VALID_P and
15336         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
15337         * doc/tm.texi: Regenerated.
15339 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
15341         * config/microblaze/microblaze.c:
15342         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
15343         * config/microblaze/microblaze.h: Add -mxl-reorder to
15344         DRIVER_SELF_SPECS.
15345         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
15346         instructions emitted if TARGET_REORDER.
15347         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
15348         or 0 for -m/-mno case, but initialises as 2 to detect default use case
15349         separately.
15351 2013-03-01  Xinliang David Li  <davidxl@google.com>
15353         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
15354         walk length.
15356 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
15358         PR middle-end/56461
15359         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
15360         vector even when returning true.  Fix up function comment formatting.
15362         PR middle-end/56461
15363         * ira-build.c (ira_loop_nodes_count): New variable.
15364         (create_loop_tree_nodes): Initialize it.
15365         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
15367         PR middle-end/56461
15368         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
15369         method on dr_chain and result_chain.
15370         * tree-vect-stmts.c (vectorizable_store): Only call
15371         result_chain.create if j == 0.
15373         PR middle-end/56461
15374         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
15375         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
15376         before overwriting it.
15378 2013-03-01  Tobias Burnus  <burnus@net-b.de>
15380         * doc/extended.texi (C Extensions): Change order in @menu
15381         to match @node.
15382         (Other MIPS Built-in Functions): Move last MIPS entry before
15383         "picoChip Built-in Functions".
15384         (SH Built-in Functions): Move after RX Built-in Functions.
15385         * doc/gcc.texi (Introduction): Change order in @menu
15386         to match @node.
15387         * doc/md.texi (Constraints): Ditto.
15388         * gty.texi (Type Information): Ditto.
15389         (User-provided marking routines for template types): Make
15390         subsection.
15391         * doc/invoke.texi (AArch64 Options): Move before
15392         "Adapteva Epiphany Options".
15394 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
15395             Jakub Jelinek  <jakub@redhat.com>
15397         PR sanitizer/56454
15398         * asan.c (gate_asan): Lookup no_sanitize_address instead of
15399         no_address_safety_analysis attribute.
15400         * doc/extend.texi (no_address_safety_attribute): Rename to
15401         no_sanitize_address attribute, mention no_address_safety_analysis
15402         attribute as deprecated alias.
15404 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
15406         PR middle-end/56461
15407         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
15408         type to vec<vec<tree> > *.
15409         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
15410         to be vec<tree> instead of vec<tree> *, set vec_defs
15411         to vNULL and call vec_defs.create (number_of_vects), adjust other
15412         uses of vec_defs.
15413         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
15414         vectorizable_condition): Adjust vect_get_slp_defs callers.
15416 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15418         * config/aarch64/aarch64.c
15419         (aarch64_float_const_representable): Remove unused variable.
15421 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15423         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
15425 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15427         * config/aarch64/aarch64-builtins.c
15428         (aarch64_init_simd_builtins): Make static.
15430 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15432         * config/aarch64/aarch64.c
15433         (aarch64_simd_make_constant): Make static.
15435 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15437         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
15438         with no initialization to the RHS of debug statements.
15440 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15442         PR tree-optimization/56294
15443         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
15444         Adjust dumping.
15445         (get_access_replacement): Do not call create_access_replacement.
15446         Assert a replacement exists.
15447         (get_repl_default_def_ssa_name): Create the replacement declaration
15448         itself.
15450 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15452         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
15453         final_end_function.
15455 2013-02-28  Marek Polacek  <polacek@redhat.com>
15457         PR rtl-optimization/56466
15458         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
15459         if we're changing a loop.
15460         (peel_loops_completely): Likewise.
15462 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
15464         PR c++/55813
15465         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
15467 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15469         PR target/56445
15470         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
15471         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
15472         INTX_FTYPE_FX, FX_FTYPE_INTX.
15473         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
15475 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15477         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
15478         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
15479         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
15480         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
15481         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
15482         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
15483         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
15484         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
15485         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
15486         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
15487         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
15488         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15489         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
15490         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
15491         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
15492         (avrxmega6): Increase max flash segments from 5 to 6.
15493         * config/avr/t-multilib: Regenerate.
15494         * config/avr/avr-tables.opt: Regenerate.
15495         * doc/avr-mmcu.texi: Regenerate.
15497 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15499         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
15500         (avr_device_to_arch): Rename to avr_device_to_ld.
15501         (avr_device_to_as): New prototype.
15502         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
15503         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
15504         * config/avr/driver-avr.c (avr_device_to_as): New.
15505         (avr_device_to_arch): Rename to avr_device_to_ld.
15507 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15509         PR middle-end/56461
15510         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
15511         method on dr_chain and result_chain.
15513         PR middle-end/56461
15514         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
15515         pointer_set_destroy on not_executed_last_iteration.
15517         PR middle-end/56461
15518         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
15520         PR middle-end/56461
15521         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
15522         FOR_EACH_DEFINED_FUNCTION when freeing state.
15524         PR middle-end/56461
15525         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
15526         pool_free.
15527         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
15528         overwriting it.
15530         PR middle-end/56461
15531         * ipa-cp.c (decide_whether_version_node): Call vec_free on
15532         known_aggs[i].items and release known_aggs vector.
15534         PR middle-end/56461
15535         * ipa-reference.c (propagate): Free node_info even for alias nodes.
15537 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15539         * config/microblaze/microblaze.c (microblaze_emit_compare):
15540         Use xor for EQ/NE comparisions.
15541         * config/microblaze/microblaze.md (cstoresf4): Add constraints
15542         (cbranchsf4): Adjust operator to comparison_operator.
15544 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15546         PR middle-end/56461
15547         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
15548         vector.
15549         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
15550         vec_safe_push, always update *slot.
15551         (redirect_edge_var_map_clear): Use vec_free.
15552         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
15553         (free_var_map_entry): Use vec_free.
15554         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
15555         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
15557 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
15559         PR middle-end/45472
15560         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
15561         when the may_trap_p bit of the exprs being merged differs.
15562         Reorder tests for speculativeness in the logical and operator.
15564 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15566         * incpath.c (add_standard_paths): Use reconcat instead of concat
15567         where appropriate and avoid leaking memory.
15569         * opts.h: Include obstack.h.
15570         (opts_concat): New prototype.
15571         (opts_obstack): New declaration.
15572         * opts.c (opts_concat): New function.
15573         (opts_obstack): New variable.
15574         (init_options_struct): Call gcc_init_obstack on opts_obstack.
15575         (finish_options): Use opts_concat instead of concat
15576         and XOBNEWVEC instead of XNEWVEC.
15577         * opts-common.c (generate_canonical_option, decode_cmdline_option,
15578         generate_option): Likewise.
15579         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
15580         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
15582         PR target/56455
15583         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
15584         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
15586 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15588         PR middle-end/56461
15589         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
15591 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15593         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
15594         (arm_block_move_unaligned_straight): Likewise.
15595         (arm_adjust_block_mem): Likewise.
15597 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15599         PR target/48901
15600         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
15601         temp, cond and label.
15602         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
15604         PR target/52500
15605         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
15606         * config/c6x/c6x.h (dbx_register_map): Update declaration.
15608         PR target/52501
15609         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
15610         of prologue/epilogue functions.
15612         PR target/52550
15613         * config/tilegx/tilegx.c (tilegx_expand_prologue):
15614         Remove unused variable cfa_offset.
15615         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
15617         PR target/54639
15618         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
15619         type promotion to unsigned.
15621         PR target/54640
15622         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
15623         for HOST_WIDE_INT of 32 bit / same size as int.
15624         (arm_block_move_unaligned_straight): Likewise.
15625         (arm_adjust_block_mem): Likewise.
15627         PR target/54662
15628         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
15629         ALL_CFLAGS.
15631 2013-02-26  Marek Polacek  <polacek@redhat.com>
15633         PR tree-optimization/56426
15634         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
15636 2013-02-26  Richard Biener  <rguenther@suse.de>
15638         PR target/56444
15639         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
15640         unused variable loops.
15642 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15644         PR tree-optimization/56448
15645         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
15646         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
15647         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
15648         later operands of the references, or even first operand for
15649         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
15651         PR tree-optimization/56443
15652         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
15653         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
15654         to type_for_mode langhook.
15656 2013-02-25  Matt Turner  <mattst88@gmail.com>
15658         * doc/invoke.texi: Document r4700.
15660 2013-02-25  Richard Biener  <rguenther@suse.de>
15662         PR tree-optimization/56175
15663         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
15664         split out from ...
15665         (simplify_bitwise_binary): ... here.  Also guard the conversion
15666         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
15668 2013-02-25  Catherine Moore  <clm@codesourcery.com>
15670         Revert:
15671         2013-02-24  Catherine Moore  <clm@codesourcery.com>
15672             Maciej W. Rozycki  <macro@codesourcery.com>
15673             Tom de Vries  <tom@codesourcery.com>
15674             Nathan Sidwell  <nathan@codesourcery.com>
15675             Iain Sandoe  <iain@codesourcery.com>
15676             Nathan Froyd  <froydnj@codesourcery.com>
15677             Chao-ying Fu  <fu@mips.com>
15679         * doc/extend.texi: (micromips, nomicromips, nocompression):
15680         Document new function attributes.
15681         * doc/invoke.texi (minterlink-compressed, mmicromips,
15682         m14k, m14ke, m14kec): Document new options.
15683         (minterlink-mips16): Update documentation.
15684         * doc/md.texi (ZC, ZD): Document new constraints.
15685         * configure.ac (gcc_cv_as_micromips): Check if linker
15686         supports the .set micromips directive.
15687         * configure: Regenerate.
15688         * config.in: Regenerate.
15689         * config/mips/mips-tables.opt: Regenerate.
15690         * config/mips/micromips.md: New file.
15691         * constraints.md (ZC, AD): New constraints.
15692         * config/mips/predicates.md (movep_src_register): New predicate.
15693         (movep_src_operand): New predicate.
15694         (non_volatile_mem_operand): New predicate.
15695         * config/mips/mips.md (multimem): New type.
15696         (length): Differentiate between 17-bit and 18-bit branch offsets.
15697         (MOVEP1, MOVEP2): New mode iterator.
15698         (mov_<load>l): Use ZC constraint.
15699         (mov_<load>r): Likewise.
15700         (mov_<store>l): Likewise.
15701         (mov_<store>r): Likewise.
15702         (*branch_equality<mode>_inverted): Add microMIPS support.
15703         (*branch_equality<mode>): Likewise.
15704         (*jump_absolute): Likewise.
15705         (indirect_jump_<mode>): Likewise.
15706         (tablejump_<mode>): Likewise.
15707         (<optab>_internal): Likewise.
15708         (sibcall_internal): Likewise.
15709         (sibcall_value_internal): Likewise.
15710         (prefetch): Use constraint ZD.
15711         * config/mips/mips.opt (minterlink-compressed): New option.
15712         (minterlink-mips16): Now an alias for minterlink-compressed.
15713         (mmicromips): New option.
15714         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15715         (compare_and_swap_12): Likewise.
15716         (sync_add<mode>): Likewise.
15717         (sync_<optab>_12): Likewise.
15718         (sync_old_<optab>_12): Likewise.
15719         (sync_new_<optab>_12): Likewise.
15720         (sync_nand_12): Likewise.
15721         (sync_old_nand_12): Likewise.
15722         (sync_new_nand_12): Likewise.
15723         (sync_sub<mode>): Likewise.
15724         (sync_old_add<mode>): Likewise.
15725         (sync_old_sub<mode>): Likewise.
15726         (sync_new_add<mode>): Likewise.
15727         (sync_new_sub<mode>): Likewise.
15728         (sync_<optab><mode>): Likewise.
15729         (sync_old_<optab><mode>): Likewise.
15730         (sync_new_<optab><mode>): Likewise.
15731         (sync_nand<mode>): Likewise.
15732         (sync_old_nand<mode>): Likewise.
15733         (sync_new_nand<mode>): Likewise.
15734         (sync_lock_test_and_set<mode>): Likewise.
15735         (test_and_set_12): Likewise.
15736         (atomic_compare_and_swap<mode>): Likewise.
15737         (atomic_exchange<mode>_llsc): Likewise.
15738         (atomic_fetch_add<mode>_llsc): Likewise.
15739         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15740         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15741         (umips_save_restore_pattern_p): Likewise.
15742         (umips_load_store_pair_p): Likewise.
15743         (umips_output_load_store_pair): Likewise.
15744         (umips_movep_target_p): Likewise.
15745         (umips_12bit_offset_address_p): Likewise.
15746         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15747         (mips_base_mips16): Rename this...
15748         (mips_base_compression_flags): ...to this. Update all uses.
15749         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15750         (mips_mips16_decl_p): Delete.
15751         (mips_nomips16_decl_p): Delete.
15752         (mips_get_compress_on_flags): New function.
15753         (mips_get_compress_off_flags): New function.
15754         (mips_get_compress_mode): New function.
15755         (mips_get_compress_on_name): New function.
15756         (mips_get_compress_off_name): New function.
15757         (mips_insert_attributes): Support multiple compression types.
15758         (mips_merge_decl_attributes): Likewise.
15759         (umips_12bit_offset_address_p): New function.
15760         (mips_start_function_definition): Emit .set micromips directive.
15761         (mips_call_may_need_jalx_p): New function.
15762         (mips_function_ok_for_sibcall): Add microMIPS support.
15763         (mips_print_operand_punctuation): Support short delay slots and
15764         compact jumps.
15765         (umips_swm_mask, umips_swm_encoding): New.
15766         (umips_build_save_restore): New function.
15767         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15768         (was_mips16_p): Remove.
15769         (old_compression_mode): New.
15770         (mips_set_compression_mode): New function.
15771         (mips_set_current_function): Add microMIPS support.
15772         (mips_option_override): Likewise.
15773         (umips_save_restore_pattern_p): New function.
15774         (umips_output_save_restore): New function.
15775         (umips_load_store_pair_p_1): New function.
15776         (umips_load_store_pair_p): New function.
15777         (umips_output_load_store_pair_1): New function.
15778         (umips_output_load_store_pair): New function.
15779         (umips_movep_target_p) New function.
15780         (mips_prepare_pch_save): Add microMIPS support.
15781         * config/mips/mips.h (TARGET_COMPRESSION): New.
15782         (TARGET_CPU_CPP_BUILTINS): Update macro
15783         to use new compression flags and to support microMIPS.
15784         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15785         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15786         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15787         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15788         (ASM_SPEC): Support mmicromips and mno-micromips.
15789         (M16STORE_REG_P): New macro.
15790         (MIPS_CALL): Support TARGET_MICROMIPS.
15791         (MICROMIPS_J): New macro.
15792         (mips_base_mips16): Rename this...
15793         (mips_base_compression_flags): ...to this.
15794         (UMIPS_12BIT_OFFSET_P): New macro.
15795         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15796         (MULTILIB_DIRNAMES): Likewise.
15798 2013-02-25  Tom de Vries  <tom@codesourcery.com>
15800         PR rtl-optimization/56131
15801         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
15802         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
15803         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
15805 2013-02-25  Tobias Burnus  <burnus@net-b.de>
15807         * doc/invoke.texi (-fsanitize=): Move from optimization
15808         to debugging options.
15810 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15812         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
15814 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15815             Alexander Monakov  <amonakov@ispras.ru>
15817         PR middle-end/56077
15818         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
15819         flush pending lists also on non-jumps.  Adjust comment.
15821 2013-02-24  Catherine Moore  <clm@codesourcery.com>
15822             Maciej W. Rozycki  <macro@codesourcery.com>
15823             Tom de Vries  <tom@codesourcery.com>
15824             Nathan Sidwell  <nathan@codesourcery.com>
15825             Iain Sandoe  <iain@codesourcery.com>
15826             Nathan Froyd  <froydnj@codesourcery.com>
15827             Chao-ying Fu  <fu@mips.com>
15829         * doc/extend.texi: (micromips, nomicromips, nocompression):
15830         Document new function attributes.
15831         * doc/invoke.texi (minterlink-compressed, mmicromips,
15832         m14k, m14ke, m14kec): Document new options.
15833         (minterlink-mips16): Update documentation.
15834         * doc/md.texi (ZC, ZD): Document new constraints.
15835         * configure.ac (gcc_cv_as_micromips): Check if linker
15836         supports the .set micromips directive.
15837         * configure: Regenerate.
15838         * config.in: Regenerate.
15839         * config/mips/mips-tables.opt: Regenerate.
15840         * config/mips/micromips.md: New file.
15841         * constraints.md (ZC, AD): New constraints.
15842         * config/mips/predicates.md (movep_src_register): New predicate.
15843         (movep_src_operand): New predicate.
15844         (non_volatile_mem_operand): New predicate.
15845         * config/mips/mips.md (multimem): New type.
15846         (length): Differentiate between 17-bit and 18-bit branch offsets.
15847         (MOVEP1, MOVEP2): New mode iterator.
15848         (mov_<load>l): Use ZC constraint.
15849         (mov_<load>r): Likewise.
15850         (mov_<store>l): Likewise.
15851         (mov_<store>r): Likewise.
15852         (*branch_equality<mode>_inverted): Add microMIPS support.
15853         (*branch_equality<mode>): Likewise.
15854         (*jump_absolute): Likewise.
15855         (indirect_jump_<mode>): Likewise.
15856         (tablejump_<mode>): Likewise.
15857         (<optab>_internal): Likewise.
15858         (sibcall_internal): Likewise.
15859         (sibcall_value_internal): Likewise.
15860         (prefetch): Use constraint ZD.
15861         * config/mips/mips.opt (minterlink-compressed): New option.
15862         (minterlink-mips16): Now an alias for minterlink-compressed.
15863         (mmicromips): New option.
15864         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15865         (compare_and_swap_12): Likewise.
15866         (sync_add<mode>): Likewise.
15867         (sync_<optab>_12): Likewise.
15868         (sync_old_<optab>_12): Likewise.
15869         (sync_new_<optab>_12): Likewise.
15870         (sync_nand_12): Likewise.
15871         (sync_old_nand_12): Likewise.
15872         (sync_new_nand_12): Likewise.
15873         (sync_sub<mode>): Likewise.
15874         (sync_old_add<mode>): Likewise.
15875         (sync_old_sub<mode>): Likewise.
15876         (sync_new_add<mode>): Likewise.
15877         (sync_new_sub<mode>): Likewise.
15878         (sync_<optab><mode>): Likewise.
15879         (sync_old_<optab><mode>): Likewise.
15880         (sync_new_<optab><mode>): Likewise.
15881         (sync_nand<mode>): Likewise.
15882         (sync_old_nand<mode>): Likewise.
15883         (sync_new_nand<mode>): Likewise.
15884         (sync_lock_test_and_set<mode>): Likewise.
15885         (test_and_set_12): Likewise.
15886         (atomic_compare_and_swap<mode>): Likewise.
15887         (atomic_exchange<mode>_llsc): Likewise.
15888         (atomic_fetch_add<mode>_llsc): Likewise.
15889         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15890         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15891         (umips_save_restore_pattern_p): Likewise.
15892         (umips_load_store_pair_p): Likewise.
15893         (umips_output_load_store_pair): Likewise.
15894         (umips_movep_target_p): Likewise.
15895         (umips_12bit_offset_address_p): Likewise.
15896         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15897         (mips_base_mips16): Rename this...
15898         (mips_base_compression_flags): ...to this. Update all uses.
15899         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15900         (mips_mips16_decl_p): Delete.
15901         (mips_nomips16_decl_p): Delete.
15902         (mips_get_compress_on_flags): New function.
15903         (mips_get_compress_off_flags): New function.
15904         (mips_get_compress_mode): New function.
15905         (mips_get_compress_on_name): New function.
15906         (mips_get_compress_off_name): New function.
15907         (mips_insert_attributes): Support multiple compression types.
15908         (mips_merge_decl_attributes): Likewise.
15909         (umips_12bit_offset_address_p): New function.
15910         (mips_start_function_definition): Emit .set micromips directive.
15911         (mips_call_may_need_jalx_p): New function.
15912         (mips_function_ok_for_sibcall): Add microMIPS support.
15913         (mips_print_operand_punctuation): Support short delay slots and
15914         compact jumps.
15915         (umips_swm_mask, umips_swm_encoding): New.
15916         (umips_build_save_restore): New function.
15917         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15918         (was_mips16_p): Remove.
15919         (old_compression_mode): New.
15920         (mips_set_compression_mode): New function.
15921         (mips_set_current_function): Add microMIPS support.
15922         (mips_option_override): Likewise.
15923         (umips_save_restore_pattern_p): New function.
15924         (umips_output_save_restore): New function.
15925         (umips_load_store_pair_p_1): New function.
15926         (umips_load_store_pair_p): New function.
15927         (umips_output_load_store_pair_1): New function.
15928         (umips_output_load_store_pair): New function.
15929         (umips_movep_target_p) New function.
15930         (mips_prepare_pch_save): Add microMIPS support.
15931         * config/mips/mips.h (TARGET_COMPRESSION): New.
15932         (TARGET_CPU_CPP_BUILTINS): Update macro
15933         to use new compression flags and to support microMIPS.
15934         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15935         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15936         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15937         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15938         (ASM_SPEC): Support mmicromips and mno-micromips.
15939         (M16STORE_REG_P): New macro.
15940         (MIPS_CALL): Support TARGET_MICROMIPS.
15941         (MICROMIPS_J): New macro.
15942         (mips_base_mips16): Rename this...
15943         (mips_base_compression_flags): ...to this.
15944         (UMIPS_12BIT_OFFSET_P): New macro.
15945         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15946         (MULTILIB_DIRNAMES): Likewise.
15948 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
15950         PR target/52555
15951         * target-globals.c (save_target_globals): For init_reg_sets and
15952         target_reinit remporarily set this_fn_optabs to this_target_optabs.
15954 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
15956         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
15957         * config/aarch64/t-aarch64
15958         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
15960 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
15962         PR inline-asm/56148
15963         * lra-constraints.c (process_alt_operands): Reload operand
15964         conflicting with earlier clobber only if no more other conflicting
15965         operands.
15967 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
15969         PR sanitizer/56393
15970         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
15971         if not linking a shared library.
15973 2013-02-22  Seth LaForge  <sethml@google.com>
15975         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
15977 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
15979         * config/arm/arm.md (split for extendsidi): Update condition.
15980         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
15981         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
15982         (qhs_zextenddi_cstr): Likewise.
15984 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
15986         PR middle-end/56420
15987         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
15988         avoid signed wrapping.
15989         (expand_mult): Handle properly multiplication by
15990         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
15991         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
15992         in the compiler if coeff is HOST_WIDE_INT_MIN.
15993         (expand_divmod): Don't make ext_op1 static, change it's type to
15994         uhwi.  Avoid undefined behavior in -INTVAL (op1).
15996         PR rtl-optimization/50339
15997         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
15998         field.
15999         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16000         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16001         into splitting_ashiftrt field.
16002         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16003         ASHIFTRT.
16004         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16005         choices.
16007 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
16009         PR middle-end/56108
16010         * trans-mem.c (execute_tm_mark): Do not expand transactions that
16011         are sure to go irrevocable.
16013 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
16015         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16016         scalars are valid operands.
16018 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16020         PR tree-optimization/56310
16021         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16022         only matching indices and non-negative final offsets.
16023         (intersect_aggregates_with_edge): Pass src_idx to
16024         agg_replacements_to_vector.  Pass src_idx insstead of index to
16025         intersect_with_agg_replacements.
16027 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16029         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16030         instead of hard-wired defaults.
16032 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
16034         * doc/invoke.texi (MIPS Options): Update documentation of the
16035         floating-point multiply-accumulate instruction restrictions.
16037 2013-02-21  Kostya Serebryany  <kcc@google.com>
16039         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16040         asan_shadow_offset on x86_64 linux.
16042 2013-02-21  Richard Biener  <rguenther@suse.de>
16044         PR tree-optimization/56415
16045         Revert
16046         2013-02-11  Richard Biener  <rguenther@suse.de>
16048         PR tree-optimization/56273
16049         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16050         first VRP run.
16052 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16054         PR bootstrap/56258
16055         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16056         instead of @itemx.
16058         PR inline-asm/56405
16059         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16060         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16062 2013-02-20  Jan Hubicka  <jh@suse.cz>
16064         PR tree-optimization/56265
16065         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16066         when target is referenced for first time.
16068 2013-02-20  Richard Biener  <rguenther@suse.de>
16070         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16071         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16072         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16073         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16074         not return anything.
16075         (rename_ssa_copies): Do not remove unused locals.
16076         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16077         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16078         * passes.c (execute_function_todo): Do not schedule unused locals
16079         removal if cleanup_tree_cfg did something.
16080         * tree-ssa-live.c (remove_unused_locals): Dump statistics
16081         about the number of removed locals.
16083 2013-02-20  Richard Biener  <rguenther@suse.de>
16085         PR tree-optimization/56398
16086         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16088 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16090         PR tree-optimization/55334
16091         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16092         restricted pointers to arrays.
16094 2013-02-20  Richard Biener  <rguenther@suse.de>
16095             Jakub Jelinek  <jakub@redhat.com>
16097         PR tree-optimization/56396
16098         * tree-ssa-ccp.c (n_const_val): New static variable.
16099         (get_value): Return NULL for SSA names we don't have a lattice
16100         entry for.
16101         (ccp_initialize): Initialize n_const_val.
16102         * tree-ssa-copy.c (n_copy_of): New static variable.
16103         (init_copy_prop): Initialize n_copy_of.
16104         (get_value): Return NULL_TREE for SSA names we don't have a
16105         lattice entry for.
16107 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16109         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16111 2013-02-20  Richard Biener  <rguenther@suse.de>
16113         * genpreds.c (write_lookup_constraint): Do not compare first
16114         letter of the constraint again.
16116 2013-02-20  Richard Biener  <rguenther@suse.de>
16118         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16119         and ceil_log2.
16120         (get_use_iv_cost): Terminate hashtable walk when coming across
16121         an empty entry.
16123 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
16125         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
16126         reassociation for avx2 targets.
16128 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16130         * config/microblaze/microblaze.c: microblaze_has_clz = 0
16131         Add version check for v8.10.a to enable microblaze_has_clz
16132         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
16133         version and TARGET_PATTERN_COMPARE check
16134         * config/microblaze/microblaze.md: New clzsi2 instruction
16136 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16138         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
16139         function before branching.
16141 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16143         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
16144         DUMP_INSN_RTX_UID.
16145         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
16147 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16149         PR middle-end/55889
16150         * sel-sched.c: Include ira.h.
16151         (implicit_clobber_conflict_p): New function.
16152         (moveup_expr): Use it.
16153         * Makefile.in (sel-sched.o): Depend on ira.h.
16155 2013-02-19  Richard Biener  <rguenther@suse.de>
16157         PR tree-optimization/56384
16158         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
16159         (vn_hash_type): Split out from ...
16160         (vn_hash_constant_with_type): ... here.
16161         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
16162         (vn_phi_eq): Compare types from vn_phi_s structure.
16163         (vn_phi_lookup): Populate vn_phi_s type.
16164         (vn_phi_insert): Likewise.
16166 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
16168         PR tree-optimization/56350
16169         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
16170         if haven't found reduction or nested cycle operand, rather than
16171         asserting we must find it.
16173         PR tree-optimization/56381
16174         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
16175         to fold_build3.
16177 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
16178             Jakub Jelinek  <jakub@redhat.com>
16180         PR target/52555
16181         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
16182         (swap_optab_enable): Same.
16183         (init_all_optabs): Use argument instead of global.
16184         * tree.h (struct tree_optimization_option): New field target_optabs.
16185         * expr.h (init_all_optabs): Add argument to prototype.
16186         (TREE_OPTIMIZATION_OPTABS): New.
16187         (save_optabs_if_changed): Protoize.
16188         * optabs.h: Declare this_fn_optabs.
16189         * optabs.c (save_optabs_if_changed): New.
16190         Declare this_fn_optabs.
16191         (init_optabs): Add argument to init_all_optabs() call.
16192         * function.c (invoke_set_current_function_hook): Handle per
16193         function optabs.
16194         * function.h (struct function): New field optabs.
16195         * config/mips/mips.c (mips_set_mips16_mode): Handle when
16196         optimization_current_node has changed.
16197         * target-globals.h (save_target_globals_default_opts): Protoize.
16198         * target-globals.c (save_target_globals_default_opts): New.
16200 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16202         PR target/56347
16203         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
16204         registers %fr12 and %fr12R as call used.
16206         PR target/56214
16207         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
16208         and HImode, require all displacements to be an integer multiple of
16209         their mode size.
16210         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
16211         only allow QImode and HImode when reload is in progress and strict is
16212         true.  Likewise for symbolic addresses.  Use base14_operand to check
16213         displacements in REG+BASE addresses.
16215 2013-02-18  Richard Biener  <rguenther@suse.de>
16217         PR tree-optimization/56366
16218         * tree-vect-loop.c (get_initial_def_for_induction): Properly
16219         handle sign-conversion of outer-loop initial induction value.
16221 2013-02-18  Richard Biener  <rguenther@suse.de>
16223         PR middle-end/56349
16224         * cfghooks.c (merge_blocks): If we merge a latch into another
16225         block adjust references to it.
16226         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
16227         (verify_loop_structure): Verify that a recorded latch is in fact
16228         a latch.
16230 2013-02-18  Richard Biener  <rguenther@suse.de>
16232         PR tree-optimization/56321
16233         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
16234         order SSA name release and virtual operand unlinking.
16236 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16238         * config/microblaze/microblaze.md (save_stack_block): Define.
16239         (restore_stack_block): Likewise.
16241 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16243         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
16244         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
16245         * config/microblaze/microblaze.c (microblaze_option_override):
16246         Bail out early for PIC modes when target does not support PIC.
16248 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16250         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
16251         Replace with a microblaze version.
16252         (microblaze_trampoline_init): Adapt for microblaze.
16253         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
16254         microblaze.
16256 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
16257             Dodji Seketeli  <dodji@redhat.com>
16259         PR asan/56330
16260         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
16261         (instrument_mem_region_access): Do not forget to always put
16262         instrumentation of the of 'base' and 'base + len' in a "if (len !=
16263         0) statement, even for cases where either 'base' or 'base + len'
16264         are not instrumented -- because they have been previously
16265         instrumented.  Simplify the logic by putting all the statements
16266         instrument 'base + len' inside a sequence, and then insert that
16267         sequence right before the current insertion point.  Then, to
16268         instrument 'base + len', just get an iterator on that statement.
16269         And do not forget to update the pointer to iterator the function
16270         received as argument.
16272 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
16274         PR rtl-optimization/56348
16275         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
16277 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
16279         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
16280         (clean_graph_dump_file): Pass base to start_graph_dump.
16282 2013-02-14  Richard Henderson  <rth@redhat.com>
16284         PR target/55941
16285         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
16287 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
16289         * collect2-aix.h: Define F_LOADONLY.
16291 2013-02-14  Richard Biener  <rguenther@suse.de>
16293         PR lto/50494
16294         * varasm.c (output_constant_def_1): Get the decl representing
16295         the constant as argument.
16296         (output_constant_def): Wrap output_constant_def_1.
16297         (make_decl_rtl): Use output_constant_def_1 with the decl
16298         representing the constant.
16299         (build_constant_desc): Optionally re-use a decl already
16300         representing the constant.
16301         (tree_output_constant_def): Adjust.
16303 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
16305         Fix an asan crash
16306         * asan.c (instrument_builtin_call):  Really put the length of the
16307         second source argument into src1_len.
16309 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
16311         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
16312         argument.  If it is false, don't create edge from then_bb to
16313         fallthru_bb.
16314         (insert_if_then_before_iter): Pass true to it.
16315         (build_check_stmt): Pass false to it.
16316         (transform_statements): Flush hash table only on extended basic
16317         block boundaries, rather than at the beginning of every bb.
16318         Don't flush hash table on nonfreeing_call_p calls.
16319         * tree-flow.h (nonfreeing_call_p): New prototype.
16320         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
16322 2013-02-13  David S. Miller  <davem@davemloft.net>
16324         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
16326 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
16328         PR target/56184
16329         * ira.c (max_regno_before_ira): Move from ...
16330         (ira): ... here.
16331         (fix_reg_equiv_init): Use max_regno_before_ira instead of
16332         vec_safe_length.
16334 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
16336         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
16338 2013-02-13  Richard Biener  <rguenther@suse.de>
16340         PR lto/56295
16341         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
16342         globals in MEM_REFs.
16344 2013-02-13  Richard Biener  <rguenther@suse.de>
16346         * loop-init.c (loop_optimizer_init): Clear loop state when
16347         re-initializing preserved loops.
16348         * loop-unswitch.c (unswitch_single_loop): Return whether
16349         we unswitched the loop.  Do not verify loop state here.
16350         (unswitch_loops): When we unswitched a loop discover new loops.
16352 2013-02-13  Kostya Serebryany  <kcc@google.com>
16354         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
16355         on x86_64 linux.
16356         * sanitizer.def: Rename __asan_init to __asan_init_v1.
16358 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
16360         Avoid instrumenting duplicated memory access in the same basic block
16361         * Makefile.in (asan.o): Add new dependency on hash-table.h
16362         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
16363         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
16364         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
16365         (free_mem_ref_resources, has_mem_ref_been_instrumented)
16366         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
16367         (get_mem_ref_of_assignment): New functions.
16368         (get_mem_refs_of_builtin_call): Extract from
16369         instrument_builtin_call and tweak a little bit to make it fit with
16370         the new signature.
16371         (instrument_builtin_call): Use the new
16372         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
16373         of is_gimple_builtin_call.
16374         (instrument_derefs, instrument_mem_region_access): Insert the
16375         instrumented memory reference into the hash table.
16376         (maybe_instrument_assignment): Renamed instrument_assignment into
16377         this, and change it to advance the iterator when instrumentation
16378         actually happened and return true in that case.  This makes it
16379         homogeneous with maybe_instrument_assignment, and thus give a
16380         chance to callers to be more 'regular'.
16381         (transform_statements): Clear the memory reference hash table
16382         whenever we enter a new BB, when we cross a function call, or when
16383         we are done transforming statements.  Use
16384         maybe_instrument_assignment instead of instrumentation.  No more
16385         need to special case maybe_instrument_assignment and advance the
16386         iterator after calling it; it's now handled just like
16387         maybe_instrument_call.  Update comment.
16389 2013-02-13  Richard Biener  <rguenther@suse.de>
16391         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
16392         Fix loop discovery code.
16394 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
16396         PR inline-asm/56148
16397         * lra-constraints.c (process_alt_operands): Match early clobber
16398         operand with itself.  Check conflicts with earlyclobber only if
16399         the operand is not reloaded.  Prefer to reload conflicting operand
16400         if earlyclobber and matching operands are the same.
16402 2013-02-12  Richard Biener  <rguenther@suse.de>
16404         PR lto/56297
16405         * lto-streamer-out.c (write_symbol): Do not output symbols
16406         for hard register variables.
16408 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
16410         PR target/54222
16411         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
16412         (umulsidi3_insn, mulsidi3_insn): New insns.
16414 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
16416         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
16417         (struct tune_params): Add vec_costs field.
16418         * config/arm/arm.c (arm_builtin_vectorization_cost)
16419         (arm_add_stmt_cost): New functions.
16420         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
16421         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
16422         (arm_default_vec_cost): New struct of type cpu_vec_costs.
16423         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
16424         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
16425         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
16426         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
16428 2013-02-12  Richard Biener  <rguenther@suse.de>
16430         PR lto/56295
16431         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
16432         decls again if possible.
16434 2013-02-12  Richard Biener  <rguenther@suse.de>
16436         PR middle-end/56288
16437         * tree-ssa.c (verify_ssa_name): Fix check, move
16438         SSA_NAME_IN_FREE_LIST check up.
16440 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
16441             Steven Bosscher   <steven@gcc.gnu.org>
16443         PR rtl-optimization/56151
16444         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
16445         equal to op0 or op1, and last_insn pattern is CODE operation
16446         with MEM dest and one of the operands matches that MEM.
16448 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
16450         * doc/extend.texi: Document Function Multiversioning and "default"
16451         parameter string to target attribute.
16452         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
16453         target attribute parameter is "default".
16454         (ix86_compare_version_priority): Remove checks for target attribute.
16455         (ix86_mangle_function_version_assembler_name): Change error to sorry.
16456         Remove check for target attribute equal to NULL. Add assert.
16457         (ix86_generate_version_dispatcher_body): Change error to sorry.
16459 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
16460             Jack Howarth  <howarth@bromo.med.uc.edu>
16461             Patrick Marlier  <patrick.marlier@gmail.com>
16463         PR libitm/55693
16464         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
16465         define ENDFILE_SPEC as TM_DESTRUCTOR.
16466         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
16468 2013-02-11  Alexander Potapenko  <glider@google.com>
16469             Jack Howarth  <howarth@bromo.med.uc.edu>
16470             Jakub Jelinek  <jakub@redhat.com>
16472         PR sanitizer/55617
16473         * config/darwin.c (cdtor_record): Rename ctor_record.
16474         (sort_cdtor_records): Rename sort_ctor_records.
16475         (finalize_dtors): New routine to sort destructors by
16476         priority before use in assemble_integer.
16477         (machopic_asm_out_destructor): Use finalize_dtors if needed.
16479 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
16481         PR rtl-optimization/56275
16482         * simplify-rtx.c (avoid_constant_pool_reference): Check that
16483         offset is non-negative and less than cmode size before
16484         calling simplify_subreg.
16486 2013-02-11  Richard Biener  <rguenther@suse.de>
16488         PR tree-optimization/56264
16489         * cfgloop.h (fix_loop_structure): Adjust prototype.
16490         * loop-init.c (fix_loop_structure): Return the number of
16491         newly discovered loops.
16492         * tree-cfgcleanup.c (repair_loop_structures): When new loops
16493         are discovered, do a full loop-closed SSA rewrite.
16495 2013-02-11  Richard Biener  <rguenther@suse.de>
16497         PR tree-optimization/56273
16498         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16499         first VRP run.
16500         (check_array_ref): Fix missing newline in dumps.
16501         (search_for_addr_array): Likewise.
16503 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
16505         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
16507 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
16509         PR target/56256
16510         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
16512 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
16514         PR rtl-optimization/56246
16515         * lra-constraints.c (simplify_operand_subreg): Try to reuse
16516         reload pseudo.
16517         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
16518         constraints are satisfied.
16520 2013-02-08  Jeff Law  <law@redhat.com>
16522         PR debug/53948
16523         * emit-rtl.c (reg_is_parm_p): New function.
16524         * regs.h (reg_is_parm_p): New prototype.
16525         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
16526         callee-clobbered registers.
16528 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
16530         PR target/56043
16531         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
16532         If there is no implicit builtin declaration, just return NULL.
16534 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16536         * config/i386/sse.md (FMAMODEM): New mode iterator.
16537         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
16538         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
16540 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16542         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
16543         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
16544         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
16546 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16548         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
16549         (microblaze*-*-elf): Likewise.
16550         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
16551         LINK_SPEC.
16552         * config/microblaze/microblaze-c.c: Add builtin defines for
16553         _LITTLE_ENDIAN and _BIG_ENDIAN.
16554         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
16555         add to TARGET_DEFAULT flags.
16556         Expand ASM_SPEC and LINK_SPEC.
16557         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
16558         * config/microblaze/microblaze.md: Update extendsidi2 and
16559         movdi_internal instructions to use low-order / high-order reg
16560         print_operands.
16561         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
16562         options and inversemask / mask of LITTLE_ENDIAN.
16563         * config/microblaze/t-microblaze: Expand multilib options to
16564         include mlittle-endian (le) and update exceptions patterns.
16566 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
16568         PR rtl-optimization/56195
16569         * lra-constraints.c (get_reload_reg): Don't reuse regs
16570         if they have smaller mode than requested, if they have
16571         wider mode than requested, try to return a SUBREG.
16573         PR tree-optimization/56250
16574         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
16575         if type is unsigned and code isn't MULT_EXPR.
16577 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16579         PR tree-optimization/56064
16580         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
16581         bits according to mode.
16582         * fixed-value.h (fixed_from_double_int)
16583         (const_fixed_from_double_int): Adjust comments.
16585 2013-02-08  Richard Biener  <rguenther@suse.de>
16587         PR lto/56231
16588         * lto-streamer.h (struct data_in): Remove current_file, current_line
16589         and current_col members.
16590         * lto-streamer-out.c (lto_output_location): Stream changed bits
16591         en-block for efficiency.
16592         * lto-streamer-in.c (clear_line_info): Remove.
16593         (lto_input_location): Cache current file, line and column
16594         globally via local statics.  Read changed bits en-block.
16595         (input_function): Do not call clear_line_info.
16596         (lto_read_body): Likewise.
16597         (lto_input_toplevel_asms): Likewise.
16599 2013-02-08  Michael Matz  <matz@suse.de>
16601         PR tree-optimization/52448
16602         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
16603         (nt_call_phase): New static.
16604         (add_or_mark_expr): Only mark accesses with newer phase than any
16605         call seen.
16606         (nonfreeing_call_p): New.
16607         (nt_init_block): Update nt_call_phase, mark blocks as visited.
16608         (nt_fini_block): Keep blocks marked as visited.
16609         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
16611 2013-02-08  Richard Biener  <rguenther@suse.de>
16613         * ira.c (ira): Free broken dominator information.
16615 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16617         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
16619 2013-02-08  Marek Polacek  <polacek@redhat.com>
16621         * cfgloop.c (verify_loop_structure): Add more checking of headers.
16623 2013-02-08  Richard Biener  <rguenther@suse.de>
16625         PR middle-end/56181
16626         * cfgloop.h (flow_loops_find): Adjust.
16627         (bb_loop_header_p): Declare.
16628         * cfgloop.c (bb_loop_header_p): New function split out from ...
16629         (flow_loops_find): ... here.  Adjust function signature,
16630         support incremental loop structure update.
16631         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
16632         * cfgloopmanip.c (fix_loop_structure): Move ...
16633         * loop-init.c (fix_loop_structure): ... here.
16634         (apply_loop_flags): Split out from ...
16635         (loop_optimizer_init): ... here.
16636         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
16637         in incremental mode, only remove dead loops here.
16639 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16641         PR target/54222
16642         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
16643         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
16644         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
16645         (*round<mode>3.libgcc): New insns for fixed-modes.
16646         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
16647         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
16648         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
16649         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
16650         implementations.  Define to __builtin_avr_absFX,
16651         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
16652         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
16653         __builtin_avr_countlsFX, respectively.
16654         * config/avr/avr-c.c (target.h): Include it.
16655         (enum avr_builtin_id): New enum.
16656         (avr_resolve_overloaded_builtin): New static function.
16657         (avr_register_target_pragmas): Use it to set
16658         targetm.resolve_overloaded_builtin.
16659         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
16660         tree nodes used by DEF_BUILTIN.
16661         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
16662         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
16663         <AVR_BUILTIN_xxBITS>: Same.
16665 2013-02-08  Richard Biener  <rguenther@suse.de>
16667         * cfgloop.c (verify_loop_structure): Properly handle
16668         a loop exiting to another loop header.
16669         * ira-int.h (ira_loops): Remove.
16670         * ira.c (ira_loops): Remove.
16671         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
16672         (do_reload): Use loop_optimizer_finalize.
16673         * ira-build.c (create_loop_tree_nodes): Use get_loops and
16674         number_of_loops to access the loop tree.
16675         (more_one_region_p): Likewise.
16676         (finish_loop_tree_nodes): Likewise.
16677         (rebuild_regno_allocno_maps): Likewise.
16678         (mark_loops_for_removal): Likewise.
16679         (mark_all_loops_for_removal): Likewise.
16680         (remove_unnecessary_regions): Likewise.
16681         (ira_build): Likewise.
16682         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
16684 2013-02-08  Richard Biener  <rguenther@suse.de>
16686         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
16687         * ipa-pure-const.c (analyze_function): Avoid calling
16688         mark_irreducible_loops twice.
16689         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
16691 2013-02-07  David S. Miller  <davem@davemloft.net>
16693         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
16694         on 'reg'.
16695         * var-tracking.c (vt_add_function_parameter): Test the presence of
16696         HAVE_window_save properly and do not remap argument registers when
16697         we have a leaf function.
16699 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
16701         PR bootstrap/56227
16702         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
16703         instead of "ll".
16704         * config/i386/i386.c (ix86_print_operand): Ditto.
16706 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16708         * lra-constraints.c (process_alt_operands): Fix recently added comment.
16710 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16712         PR rtl-optimization/56225
16713         * lra-constraints.c (process_alt_operands): Check that reload hard
16714         reg can hold value for strict_low_part.
16716 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16718         PR debug/56154
16719         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
16720         dwarf2out_end_function.
16721         (in_first_function_p, maybe_at_text_label_p,
16722         first_loclabel_num_not_at_text_label): New variables.
16723         (dwarf2out_var_location): In the first function find out
16724         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
16725         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
16726         functions.
16728 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
16730         PR rtl-optimization/56178
16731         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
16732         SUBREG of a register.  Tidy up related block of code.
16733         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
16734         note if the source is a register or a SUBREG of a register.
16736 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16738         PR target/56228
16739         * config/rs6000/rs6000.md (ptrm): New mode attr.
16740         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
16741         call_value_indirect_aix<pttrsize>,
16742         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
16743         m in constraints.
16745 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16747         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
16748         if -bnortl. Convert to strcmp and strncmp.
16750 2013-02-07  Alan Modra  <amodra@gmail.com>
16752         PR target/54009
16753         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
16754         addresses won't wrap when offsetting.
16755         (rs6000_secondary_reload): Provide secondary reloads needed for
16756         wrapping LO_SUM addresses.
16758 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
16760         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
16761         MACH, just __MACH__.
16763 2013-02-06  Richard Biener  <rguenther@suse.de>
16765         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
16766         instead of calling fix_loop_structure.
16768 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
16770         PR middle-end/56217
16771         * omp-low.c (use_pointer_for_field): Return false if
16772         lower_send_shared_vars doesn't generate any copy-out code.
16774 2013-02-06  Tom de Vries  <tom@codesourcery.com>
16776         PR rtl-optimization/56131
16777         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
16778         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
16779         of the label is NULL.  Add comment.
16781 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
16783         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
16785         PR sanitizer/55374
16786         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
16787         (STATIC_LIBTSAN_LIBS): Likewise.
16788         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
16789         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
16790         is defined, don't add anything else beyond that.
16791         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
16792         (LINK_COMMAND_SPEC): Use them.
16794         PR tree-optimization/56205
16795         * tree-stdarg.c (check_all_va_list_escapes): Return true if
16796         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
16797         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
16799 2013-02-05  Richard Biener  <rguenther@suse.de>
16801         PR tree-optimization/53342
16802         PR tree-optimization/53185
16803         * tree-vectorizer.h (vect_check_strided_load): Remove.
16804         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
16805         not disallow peeling for vectorized strided loads.
16806         (vect_check_strided_load): Make static and simplify.
16807         (vect_analyze_data_refs): Adjust.
16808         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
16809         correctly when vectorizing strided loads.
16811 2013-02-05  Richard Biener  <rguenther@suse.de>
16813         * doc/install.texi: Refer to ISL, not PPL.
16815 2013-02-05  Jan Hubicka  <jh@suse.cz>
16817         PR tree-optimization/55789
16818         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
16820 2013-02-05  Jan Hubicka  <jh@suse.cz>
16822         PR tree-optimization/55789
16823         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
16824         the dead call anyway.
16826 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
16828         PR sanitizer/55374
16829         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
16831 2013-02-04  Alexander Potapenko  <glider@google.com>
16832             Jack Howarth  <howarth@bromo.med.uc.edu>
16833             Jakub Jelinek  <jakub@redhat.com>
16835         PR sanitizer/55617
16836         * config/darwin.c (sort_ctor_records): Stabilized qsort
16837         on constructor priority by using original position.
16838         (finalize_ctors): New routine to sort constructors by
16839         priority before use in assemble_integer.
16840         (machopic_asm_out_constructor): Use finalize_ctors if needed.
16842 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
16844         PR libstdc++/54314
16845         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
16846         about visibility on artificial decls.
16847         * config/sol2.c (solaris_assemble_visibility): Likewise.
16849 2013-02-04  Kai Tietz  <ktietz@redhat.com>
16851         PR target/56186
16852         * config/i386/i386.c (function_value_ms_64): Add additional valtype
16853         argument and improve checking of return-argument types for 16-byte
16854         modes.
16855         (ix86_function_value_1): Add additional valtype argument on call
16856         of function_value_64.
16857         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16858         handling infunction_value_64 function.
16860 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
16862         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16864 2013-02-04  Richard Biener  <rguenther@suse.de>
16866         PR tree-optimization/56188
16867         * tree-ssa-structalias.c (label_visit): Consider case with
16868         initially non-empty points-to set.
16869         (perform_var_substitution): Dump node mapping and clean up.
16871 2013-02-04  Richard Guenther  <rguenther@suse.de>
16873         PR lto/56168
16874         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16875         node prevail as last resort.
16876         (lto_symtab_merge_decls): Remove guard on LTRANS here.
16877         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16879 2013-02-04  Richard Biener  <rguenther@suse.de>
16881         PR tree-optimization/56113
16882         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16883         Merge into ...
16884         (equiv_class_lookup_or_add): ... this.
16885         (label_visit): Adjust and fix error in previous patch.
16886         (perform_var_substitution): Adjust.
16888 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
16890         * config/sh/divtab.c: Fix formatting and comments throughout the file.
16891         * config/sh/sh4-300.md: Likewise.
16892         * config/sh/sh4a.md: Likewise.
16893         * config/sh/constraints.md: Likewise.
16894         * config/sh/sh.md: Likewise.
16895         * config/sh/netbsd-elf.h: Likewise.
16896         * config/sh/predicates.md: Likewise.
16897         * config/sh/sh-protos.h: Likewise.
16898         * config/sh/ushmedia.h: Likewise.
16899         * config/sh/linux.h: Likewise.
16900         * config/sh/sh.c: Likewise.
16901         * config/sh/superh.h: Likewise.
16902         * config/sh/elf.h: Likewise.
16903         * config/sh/sh4.md: Likewise.
16904         * config/sh/sh.h: Likewise.
16906 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16908         * config/pa/constraints.md: Adjust unused letters.  Change "T"
16909         constraint to match_test floating_point_store_memory_operand().
16910         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16911         (base14_operand): New.
16912         (floating_point_store_memory_operand): New.
16913         (integer_store_memory_operand): Revise to use base14_operand and
16914         reg_plus_base_memory_operand.
16915         (move_dest_operand): Allow symbolic_memory_operands.
16916         (symbolic_memory_operand): Check for LO_SOM.
16917         (symbolic_operand): Change default case to break.
16918         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16919         CONST_DOUBLE values to be reloaded by putting them into memory when
16920         the destination is a floating point register.
16921         (movdf): Remove code to handle CONST_DOUBLE.
16922         (movsf): Likewise.
16923         (reload_indf_r1): New.
16924         (reload_insf_r1): New.
16925         Consistently use "Q" and "T" constraints with integer and floating
16926         point move instructions, respectively.
16927         (movdi): Remove FAIL.
16928         Change predicate for source operand unamed DImode move from
16929         general_operand to move_src_operand.
16930         (umulsidi3): Change predicate for destination operand to
16931         register_operand.
16932         Likewise for similar unamed patterns.
16933         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16934         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16935         (hppa_legitimize_address): Simplify mask calculation.
16936         (pa_emit_move_sequence): Revised handling of secondary reloads from
16937         REG+D addresses for floating point loads and stores.  Directly handle
16938         loading CONST0_RTX (mode) to a floating point register.
16939         (pa_secondary_reload): Handle reloading DF and SFmode constant values
16940         to floating point registers.  Don't restrict secondary reloads to
16941         floating point registers to integer modes.  Revise some comments and
16942         cleanup some code.
16943         (TARGET_LEGITIMATE_ADDRESS_P): Define.
16944         (pa_legitimate_address_p): New.
16945         (pa_legitimize_reload_address): New.
16946         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
16947         (STRICT_REG_OK_FOR_BASE_P): New.
16948         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
16949         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
16951 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
16952             Andrew Dixie  <andrewd@gentrack.com>
16954         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
16955         flag set.
16957 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
16959         * expmed.c (extract_bit_field_1): Pass the full width of the
16960         structure to get_best_reg_extraction_insn.
16962 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16964         PR target/54601
16965         * configure.ac (use_cxa_atexit): Add AIX.
16966         * configure: Regenerate.
16968         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
16970 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
16972         PR debug/54793
16973         * final.c (need_profile_function): New variable.
16974         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
16975         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
16976         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
16977         notes, targetm.asm_out.function_prologue doesn't emit anything,
16978         HAVE_prologue and profiler should be emitted before prologue,
16979         set need_profile_function instead of emitting it.
16980         (final_scan_insn): If need_profile_function, emit
16981         profile_function on the first NOTE_INSN_BASIC_BLOCK or
16982         NOTE_INSN_FUNCTION_BEG note.
16984 2013-02-01  Richard Henderson  <rth@redhat.com>
16986         * config/rs6000/rs6000.md (smulditi3): New.
16987         (umulditi3): New.
16989         * config/alpha/alpha.md (umulditi3): New.
16991 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
16993         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
16994         (ASM_OUTPUT_ALIGNED_LOCAL): New.
16996 2013-02-01  Richard Biener  <rguenther@suse.de>
16998         PR tree-optimization/56113
16999         * tree-ssa-structalias.c (label_visit): Reduce work for
17000         single-predecessor nodes.
17002 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
17004         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17005         range isn't testing for zero.
17007 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
17009         PR middle-end/56113
17010         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17012 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
17013             Nick Clifton  <nickc@redhat.com>
17015         * config/v850/constraints.md (Q): Define as a memory constraint.
17016         * config/v850/predicates.md (label_ref_operand): New predicate.
17017         (e3v5_shift_operand): New predicate.
17018         (ior_operator): New predicate.
17019         * config/v850/t-v850: Add e3v5 multilib.
17020         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17021         (v850_gen_movdi): Prototype.
17022         * config/v850/v850.c: Add support for e3v5 architecture.
17023         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17024         TARGET_V850E_UP.
17025         (construct_save_jarl): Add e3v5 long JARL support.
17026         (v850_adjust_insn_length): New function.  Adjust length of call
17027         insns when using e3v5 instructions.
17028         (v850_gen_movdi): New function: Generate instructions to move a
17029         DImode value.
17030         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17031         (CPP_SPEC): Define __v850e3v5__ as appropriate.
17032         (TARGET_USE_FPU): Enable for e3v5.
17033         (CONST_OK_FOR_W): New macro.
17034         (ADJUST_INSN_LENGTH): Define.
17035         * config/v850/v850.md (UNSPEC_LOOP): Define.
17036         (attr cpu): Add v850e3v5.
17037         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17038         (movdi): New pattern.
17039         (movdi_internal): New pattern.
17040         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17041         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17042         (cstoresf4): Likewise.
17043         (cstoredf4): Likewise.
17044         (insv): New pattern.
17045         (rotlso3_a): New pattern.
17046         (rotlsi3_b): New pattern
17047         (rotlsi3_v850e3v5): New pattern.
17048         (doloop_begin): New pattern.
17049         (fix_loop_counter): New pattern.
17050         (doloop_end): New pattern.
17051         (branch_normal): Add e3v5 long branch support.
17052         (branch_invert): Likewise.
17053         (branch_z_normal): Likewise.
17054         (branch_z_invert): Likewise.
17055         (branch_nz_normal): Likewise.
17056         (branch_nz_invert): Likewise.
17057         (call_internal_short): Add e3v5 register-indirect JARL support.
17058         (call_internal_long): Likewise.
17059         (call_value_internal_short): Likewise.
17060         (call_value_internal_long): Likewise.
17061         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17062         (mloop): New option.
17063         * config.gcc: Add support for configuring v840e3v5 target.
17064         * doc/invoke.texi: Document new v850 specific command line options.
17066 2013-01-31  Paul Koning  <ni1d@arrl.net>
17068         PR debug/55059
17069         PR debug/54508
17070         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17071         children if parent is a class.
17072         (prune_unused_types_prune): Don't add DW_AT_declaration.
17074 2013-01-31  Richard Biener  <rguenther@suse.de>
17076         PR tree-optimization/56157
17077         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17078         match up operand with SLP child.
17080 2013-01-31  Jason Merrill  <jason@redhat.com>
17082         PR debug/54410
17083         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17084         parameters the first time.
17085         (gen_scheduled_generic_parms_dies): Check completeness here.
17087 2013-01-31  Richard Biener  <rguenther@suse.de>
17089         PR middle-end/53073
17090         * common.opt (faggressive-loop-optimizations): New flag,
17091         enabled by default.
17092         * doc/invoke.texi (faggressive-loop-optimizations): Document.
17093         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17094         infer_loop_bounds_from_undefined by it.
17096 2013-01-31  Richard Biener  <rguenther@suse.de>
17098         PR tree-optimization/56150
17099         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17100         visit virtual operands.
17101         (find_uses_to_rename_bb): Likewise.
17103 2013-01-31  Richard Biener  <rguenther@suse.de>
17105         PR tree-optimization/56150
17106         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17107         mixed store non-store stmts.
17109 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
17111         PR sanitizer/55374
17112         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17113         LIBASAN_EARLY_SPEC is defined.
17114         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17115         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17116         before %o.
17117         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17119         PR c++/55742
17120         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17121         invalid args instead of ICEing on it.
17122         (ix86_valid_target_attribute_tree): Return error_mark_node if
17123         ix86_valid_target_attribute_inner_p failed.
17124         (ix86_valid_target_attribute_p): Return false only if
17125         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
17126         target("default") attribute.
17127         (sorted_attr_string): Change argument from const char * to tree,
17128         merge in all target attribute arguments rather than just one.
17129         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
17130         instead of free.  Avoid using strcat.
17131         (ix86_mangle_function_version_assembler_name): Mangle
17132         target("default") as if no target attribute is present.  Adjust
17133         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
17134         instead of xmalloc and XDELETEVEC instead of free.
17135         (ix86_function_versions): Don't return true if one of the decls
17136         doesn't have target attribute.  If they don't and one of the decls
17137         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
17138         sorted_attr_string caller.  Use XDELETEVEC instead of free.
17139         (ix86_supports_function_versions): Remove.
17140         (make_name): Fix up formatting.
17141         (make_dispatcher_decl): Remove resolver_name and its initialization.
17142         Avoid leaking memory.
17143         (is_function_default_version): Return true if there is
17144         target("default") attribute rather than no target attribute at all.
17145         (make_resolver_func): Avoid leaking memory.
17146         (ix86_generate_version_dispatcher_body): Likewise.
17147         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
17148         * target.def (supports_function_versions): Remove.
17149         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
17150         * doc/tm.texi: Regenerated.
17152 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
17154         PR rtl-optimization/56144
17155         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
17156         for values with side effects.
17158 2013-01-30  Richard Biener  <rguenther@suse.de>
17160         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
17161         (sparseset_pop): Likewise.
17162         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
17163         to be able to use quick_push in the worker loop.
17165 2013-01-30  Marek Polacek  <polacek@redhat.com>
17167         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
17169 2013-01-30  Richard Biener  <rguenther@suse.de>
17171         PR lto/56147
17172         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
17174 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
17176         PR tree-optimization/56064
17177         * fixed-value.c (fixed_from_double_int): New function.
17178         * fixed-value.h (fixed_from_double_int): New prototype.
17179         (const_fixed_from_double_int): New static inline function.
17180         * fold-const.c (native_interpret_fixed): New static function.
17181         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
17182         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
17183         (native_encode_fixed): New static function.
17184         (native_encode_expr) <FIXED_CST>: Use it.
17185         (native_interpret_int): Move double_int worker code to...
17186         * double-int.c (double_int::from_buffer): ...this new static method.
17187         * double-int.h (double_int::from_buffer): Prototype it.
17189 2013-01-30  Richard Biener  <rguenther@suse.de>
17191         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
17192         New pointer-map and obstack.
17193         (init_alias_vars): Allocate pointer-map and obstack.
17194         (delete_points_to_sets): Free them.
17195         (find_what_var_points_to): Cache result.
17196         (find_what_p_points_to): Adjust for changed interface of
17197         find_what_var_points_to.
17198         (compute_points_to_sets): Likewise.
17199         (ipa_pta_execute): Likewise.
17201 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17203         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
17204         * configure: Regenerate.
17205         * config.in: Regenerate.
17206         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
17207         #nobits/#progbits if supported.
17209 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
17211         PR target/56121
17212         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
17213         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
17214         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
17216 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17218         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
17219         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
17221 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17223         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
17224         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
17226 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17228         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
17229         declaration.
17230         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
17231         * config/arm/cortex-a7.md: New bypasses using
17232         arm_mac_accumulator_is_result.
17234 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17236         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
17237         (cortex_a7_neon_mla): Likewise.
17238         (cortex_a7_fpfmad): New reservation.
17239         (cortex_a7_fpmacs): Use ffmas and update required units.
17240         (cortex_a7_fpmuld): Update required units and latency.
17241         (cortex_a7_fpmacd): Likewise.
17242         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
17243         (cortex_a7_neon). Likewise.
17244         (bypass) Update participating units.
17246 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17248         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
17249         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
17250         from fmac to ffma.
17251         * config/arm/vfp11.md (vfp_farith): Use ffmas.
17252         (vfp_fmul): Use ffmad.
17253         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
17254         (cortex_r4_fmacd): Use ffmad.
17255         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
17256         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
17257         (cortex_a9_fmacd): Use ffmad.
17258         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
17259         (cortex_a8_vfp_macd): Use ffmad.
17260         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
17261         (cortex_a5_fpmacd): Use ffmad.
17262         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
17263         (cortex_a15_vfp_macd): Use ffmad.
17264         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
17266 2013-01-29  Jason Merrill  <jason@redhat.com>
17268         PR libstdc++/54314
17269         * varasm.c (default_assemble_visibility): Don't warn about
17270         visibility on artificial decls.
17272 2013-01-29  Richard Biener  <rguenther@suse.de>
17274         PR tree-optimization/56113
17275         * tree-ssa-structalias.c (equiv_class_lookup): Also return
17276         the bitmap leader.
17277         (label_visit): Free duplicate bitmaps and record the leader instead.
17278         (perform_var_substitution): Adjust.
17280 2013-01-29  Richard Biener  <rguenther@suse.de>
17282         PR tree-optimization/55270
17283         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
17284         the CFG, schedule loops for fixup.
17286 2013-01-29  Nick Clifton  <nickc@redhat.com>
17288         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
17289         SP_REG.
17291 2013-01-28  Leif Ekblad  <leif@rdos.net>
17293         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
17294         * config/i386/i386.h (TARGET_RDOS): New macro.
17295         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
17296         * config/i386/i386.c (ix86_option_override_internal): For 64bit
17297         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
17298         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
17299         DEFAULT_LARGE_SECTION_THRESHOLD.
17300         * config/i386/i386.md (R14_REG, R15_REG): New constants.
17301         * config/i386/rdos.h: New file.
17302         * config/i386/rdos64.h: New file.
17304 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
17306         PR other/54814
17307         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
17308         TEST_HARD_REG_BIT.
17310 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
17312         PR rtl-optimization/56117
17313         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
17314         call cselib_lookup_from_insn on the MEM before calling
17315         add_insn_mem_dependence.
17317 2013-01-28  Richard Biener  <rguenther@suse.de>
17319         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
17320         to a stmt that didn't have one.
17321         (copy_phis_for_bb): Likewise for PHI arguments.
17322         (copy_debug_stmt): Likewise for debug stmts.
17324 2013-01-28  Richard Biener  <rguenther@suse.de>
17326         PR tree-optimization/56034
17327         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
17328         (partition_builtin_p): Adjust.
17329         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
17330         it is the last partition.
17331         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
17332         up the vertex for the definition.
17333         (classify_partition): Classify whether a partition is a
17334         PKIND_REDUCTION, thus has uses outside of the loop.
17335         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
17336         Merge all PKIND_REDUCTION partitions into the last partition.
17337         (tree_loop_distribution): Seed partitions from reductions as well.
17339 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
17341         PR tree-optimization/56125
17342         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
17343         pow(x,c) into sqrt(x) * powi(x, n/2) or
17344         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
17345         optimizing for size.
17346         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
17347         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
17348         integer.
17350         PR tree-optimization/56094
17351         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
17352         to UNKNOWN_LOCATION while gimplifying expr.
17354 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
17356         PR target/56114
17357         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
17358         operand 0 in movabs insn template for -masm=intel asm alternative.
17359         (*movabs<mode>_2): Ditto for operand 1.
17361 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
17363         PR target/54663
17364         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
17365         of microblaze-c.o
17367 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17369         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
17370         tm_file.
17372 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17374         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
17375         Undef to avoid warning.
17377 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
17379         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
17380         * configure: Regenerate.
17382 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
17384         PR tree-optimization/56098
17385         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
17386         for stmts with volatile ops.
17387         (cond_store_replacement): Don't optimize if assign has volatile ops.
17388         (cond_if_else_store_replacement_1): Don't optimize if either
17389         then_assign or else_assign have volatile ops.
17390         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
17391         volatile ops.
17393 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17395         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
17397 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17399         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
17400         missing ':' in asm example.
17402 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
17404         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
17405         entries into lane and laneq entries.
17406         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
17407         Remove AdvSIMD scalar modes.
17408         (aarch64_sq<r>dmulh_laneq<mode>): New.
17409         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
17410         modes.
17411         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
17412         builtin implementations to relfect changes in RTL in aarch64-simd.md.
17413         * config/aarch64/iterators.md (VCOND): New.
17414         (VCONQ): New.
17416 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17418         PR target/54222
17419         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
17420         Add NULL LIBNAME argument to existing definitions.
17421         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
17422         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
17423         * config/avr/avr.c (DEF_BUILTIN): Same.
17424         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
17425         (avr_expand_builtin): Expand to a vanilla call if a libgcc
17426         implementation is available (DECL_ASSEMBLER_NAME is set).
17427         (avr_fold_absfx): New static function.
17428         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
17429         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
17430         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
17431         AVR_BUILTIN_ABSLLK.
17432         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
17433         (abshk, absk, abslk, absllk): Provide as static inline functions.
17435 2013-01-25  Marek Polacek  <polacek@redhat.com>
17437         PR tree-optimization/56035
17438         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
17440 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
17442         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
17443         (*movtf_internal_rex64): Add (!o,C) alternative
17444         (*movxf_internal_rex64): Ditto.
17445         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
17447 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
17449         * doc/invoke.texi: fix typo.
17450         * doc/objc.texi: fix typo.
17452 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
17454         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
17455         for the first two alternatives.
17457 2013-01-24  Diego Novillo  <dnovillo@google.com>
17459         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
17460         (ggc-zone.o): Remove.
17461         * configure.ac: Remove option --with-gc.
17462         * configure: Re-generate.
17463         * doc/install.texi: Remove documentation for --with-gc.
17464         * gengtype.c (write_enum_defn): Remove.  Update all users.
17465         (write_Types_process_field): Remove generation of gt_e_* argument.
17466         (output_type_enum): Remove.  Update all users.
17467         (write_enum_defn): Remove.  Update all users.
17468         (enum alloc_zone): Remove.  Update all users.
17469         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
17470         * ggc-common.c (ggc_splay_alloc): Remove first argument.
17471         Update all callers.
17472         (struct ptr_data): Remove field TYPE.  Update all users.
17473         (gt_pch_note_object): Remove argument TYPE.  Update all users.
17474         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
17475         Update all users.
17476         * ggc-none.c (ggc_alloc_typed_stat): Remove.
17477         (struct alloc_zone): Remove.
17478         (ggc_internal_alloc_zone_stat): Remove.
17479         (ggc_internal_cleared_alloc_zone_stat): Remove.
17480         * ggc-page.c (ggc_alloc_typed_stat): Remove.
17481         (ggc_pch_count_object): Remove last argument.  Update all users.
17482         (ggc_pch_alloc_object): Remove last argument.  Update all users.
17483         (struct alloc_zone): Remove.
17484         * ggc-zone.c: Remove.
17485         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
17486         (struct alloc_zone): Remove.
17487         (ggc_alloc_typed_stat): Remove.
17488         (ggc_alloc_typed): Remove.
17489         (ggc_splay_alloc): Remove first argument.
17490         (rtl_zone): Remove.  Update all users.
17491         (tree_zone): Remove.  Update all users.
17492         (tree_id_zone): Remove.  Update all users.
17493         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
17494         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
17495         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
17496         * tree-ssanames.c: Remove references to zone allocator in comments.
17498 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
17500         * config/avr/avr.c (avr_out_fract): Make register numbers that
17501         might be outside of source operand signed.
17503 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
17505         * config/i386/constraints.md (Yf): New constraint.
17506         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
17507         of f constraint to conditionaly disable x87 register preferences.
17508         (*movdf_internal): Ditto.
17509         (*movsf_internal): Ditto.
17511 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
17513         PR inline-asm/55934
17514         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
17515         that have operands with impossible constraints.
17516         Add a FIXME for a speed-up opportunity.
17517         * lra-constraints.c (process_alt_operands): Verify that a class
17518         selected from constraints on asms is valid for the operand mode.
17519         (curr_insn_transform): Remove incorrect comment.
17521 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
17523         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
17524         TOC operand is a valid symbol ref in the constant pool.
17526 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17528         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
17530 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
17532         PR target/54222
17533         * config/avr/stdfix.h: New file.
17534         * t-avr (stdfix-gcc.h): New rule to build it.
17535         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
17537 2013-01-23  Kostya Serebryany  <kcc@google.com>
17539         * config/darwin.h: remove dependency on
17540         CoreFoundation (asan on Mac OS).
17542 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
17544         PR target/49069
17545         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
17546         instead of cmpdi_operand for first comparison operand.
17547         Don't assert that comparison operands aren't both constants.
17549 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17551         * doc/install.texi (Downloading the Source): Update references to
17552         downloading separate components.
17554 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17556         * doc/extend.texi (__int128): Improve grammar.
17558 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
17560         PR target/56028
17561         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
17562         alternative to (o,r).
17563         (*movdi_internal_rex64): Remove (!o,n) alternative.
17564         (DImode immediate->memory splitter): Remove.
17565         (DImode immediate->memory peephole2): Remove.
17566         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
17567         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
17568         alternative to (!o,*r).
17569         (*movtf_internal_sse): New pattern.
17570         (*movxf_internal_rex64): New pattern.
17571         (*movxf_internal): Disable for TARGET_64BIT.
17572         (*movdf_internal_rex64): Remove (!o,F) alternative.
17574 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
17576         PR middle-end/56074
17577         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
17578         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
17579         * tree-vect-loop-manip.c (find_loop_location): Also ignore
17580         stmt locations where LOCATION_LOCUS of the stmt location is
17581         UNKNOWN_LOCATION or BUILTINS_LOCATION.
17583         PR target/55686
17584         * config/i386/i386.md (UNSPEC_STOS): New.
17585         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
17586         *strsetqi_1): Add UNSPEC_STOS.
17588 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
17590         PR c++/56067
17591         * doc/invoke.texi: Remove left over -Wsynth example.
17593 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17595         PR tree-optimization/56051
17596         * fold-const.c (fold_binary_loc): Don't fold
17597         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
17598         a narrowing conversion, or widening conversion from signed
17599         to unsigned.
17601 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17603         PR rtl-optimization/56023
17604         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
17605         dependent on debug instruction.
17607 2013-01-21  Martin Jambor  <mjambor@suse.cz>
17609         PR middle-end/56022
17610         * function.c (allocate_struct_function): Call
17611         invoke_set_current_function_hook earlier.
17613 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17615         * reload1.c (init_reload): Only initialize reload_obstack
17616         during the first call.
17618 2013-01-21  Marek Polacek  <polacek@redhat.com>
17620         * cfgloop.c (verify_loop_structure): Fix up grammar.
17622 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
17624         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
17625         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
17627 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17629         PR target/56058
17630         * config/arm/marvell-pj4.md: Update copyright year.
17631         Fix up use of alu to alu_reg and simple_alu_imm.
17633 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17635         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
17637 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
17639         PR target/55433
17640         * lra-constraints.c (curr_insn_transform): Don't reuse original
17641         insn for secondary memory move when memory mode should be different.
17643 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17645         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
17646         atomic_storedi_1): New patterns.
17648 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17650         btver2 pipeline descriptions.
17651         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
17652         descriptions.
17653         * config/i386/i386.md (btver2_decode): New type attributes.
17654         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
17655         type attributes.
17656         * config/i386/btver2.md: New file describing btver2 pipelines.
17658 2013-01-19  Andrew Pinski  <apinski@cavium.com>
17660         PR tree-optimization/52631
17661         * tree-ssa-sccvn (visit_use): Before looking up the original
17662         statement, try looking up the simplified expression.
17664 2013-01-19  Anthony Green  <green@moxielogic.com>
17666         * config/moxie/moxie.c (moxie_expand_prologue): Set
17667         current_function_static_stack_size.
17669 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17671         PR tree-optimization/56029
17672         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
17673         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
17675 2013-01-18  Sharad Singhai  <singhai@google.com>
17677         PR tree-optimization/55995
17678         * dumpfile.c (dump_loc): Print location only if available.
17679         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
17681 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
17683         PR target/55433
17684         * lra-constraints.c (curr_insn_transform): Reuse original insn for
17685         secondary memory move.
17686         (inherit_reload_reg): Use rclass instead of cl for
17687         check_secondary_memory_needed_p.
17689 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17691         PR middle-end/56015
17692         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
17693         the case where writing real complex part of target modifies op1.
17695 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
17697         * config/aarch64/aarch64-simd.md
17698         (aarch64_vcond_internal<mode>): Handle unordered cases.
17699         * config/aarch64/iterators.md (v_cmp_result): New.
17701 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
17702             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17704         * config/arm/marvell-pj4.md: New file.
17705         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
17706         * config/arm/arm.md (generic_sched): Add marvell_pj4.
17707         (generic_vfp): Likewise.
17708         * config/arm/arm-cores.def: Add marvell-pj4.
17709         * config/arm/arm-tune.md: Regenerate.
17710         * config/arm/arm-tables.opt: Regenerate.
17711         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
17712         * doc/invoke.texi: Document marvell-pj4.
17714 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
17716         * config/aarch64/arm_neon.h: Map scalar types to standard types.
17718 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17720         PR debug/54114
17721         PR debug/54402
17722         PR debug/49888
17723         * var-tracking.c (negative_power_of_two_p): New.
17724         (global_get_addr_cache, local_get_addr_cache): New.
17725         (get_addr_from_global_cache, get_addr_from_local_cache): New.
17726         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
17727         heading comment.
17728         (vt_stack_offset_p): Remove.
17729         (vt_canon_true_dep): Always canonicalize loc's address.
17730         (clobber_overlapping_mems): Make sure we have a MEM.
17731         (local_get_addr_clear_given_value): New.
17732         (val_reset): Clear local cached entries.
17733         (compute_bb_dataflow): Create and release the local cache.
17734         Disable duplicate MEMs clobbering.
17735         (emit_notes_in_bb): Clobber MEMs likewise.
17736         (vt_emit_notes): Create and release the local cache.
17737         (vt_initialize, vt_finalize): Create and release the global
17738         cache, respectively.
17739         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
17741 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17743         PR libmudflap/53359
17744         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
17745         not found in the symtab.
17747 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17749         PR debug/56006
17750         PR rtl-optimization/55547
17751         PR rtl-optimization/53827
17752         PR debug/53671
17753         PR debug/49888
17754         * alias.c (offset_overlap_p): New, factored out of...
17755         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
17756         the conservative special case for symbolic constants.  Don't
17757         adjust zero sizes on alignment.
17759 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
17761         PR rtl-optimization/52573
17762         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
17763         REG_UNUSED for the same register.
17765 2013-01-17  Richard Biener  <rguenther@suse.de>
17766             Marek Polacek  <polacek@redhat.com>
17768         PR rtl-optimization/55833
17769         * loop-unswitch.c (unswitch_loops): Move loop verification...
17770         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
17771         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
17772         Set it to true when we're removing a loop from hierarchy tree in
17773         an irreducible region.
17774         (fix_bb_placements): Adjust caller.
17775         (fix_loop_placements): Likewise.
17777 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
17779         * config/avr/builtins.def (DEF_BUILTIN): Factor out
17780         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
17781         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
17782         Remove ID.  Adjust comments.
17783         * config/avr/avr-c.c (avr_builtin_name): Remove.
17784         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
17785         * config/avr/avr.c (avr_tolower): New static function.
17786         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
17787         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
17788         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
17789         default expansion.
17791 2013-01-17  Jan Hubicka  <jh@suse.cz>
17793         PR tree-optimization/55273
17794         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
17796 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
17798         PR target/55981
17799         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
17800         store through atomic_store<mode>_1.
17801         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
17803 2013-01-17  Martin Jambor  <mjambor@suse.cz>
17805         PR tree-optimizations/55264
17806         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
17807         for virtual methods.
17808         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
17809         virtual methods before inlining is over.
17810         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
17811         virtual functions.
17812         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
17813         non-virtual.
17815 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
17817         PR rtl-optimization/56005
17818         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
17819         pending reads for prefetch.
17821 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
17823         * config/aarch64/aarch64.md
17824         (*cstoresi_neg_uxtw): New pattern.
17825         (*cmovsi_insn_uxtw): New pattern.
17826         (*<optab>si3_uxtw): New pattern.
17827         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
17828         (*<optab>si3_insn_uxtw): New pattern.
17829         (*bswapsi2_uxtw): New pattern.
17831 2013-01-16  Richard Biener  <rguenther@suse.de>
17833         * tree-inline.c (tree_function_versioning): Remove set but
17834         never used variable.
17836 2013-01-16  Richard Biener  <rguenther@suse.de>
17838         PR tree-optimization/55964
17839         * tree-flow.h (rename_variables_in_loop): Remove.
17840         (rename_variables_in_bb): Likewise.
17841         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
17842         (copy_loop_before): Adjust and delete update-ssa status.
17843         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
17844         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
17845         (rename_variables_in_loop): Remove.
17846         (slpeel_update_phis_for_duplicate_loop): Likewise.
17847         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
17848         use available cfg machinery instead of duplicating it.
17849         Update PHI nodes and perform poor-mans SSA update here.
17850         (slpeel_tree_peel_loop_to_edge): Adjust.
17852 2013-01-16  Richard Biener  <rguenther@suse.de>
17854         PR tree-optimization/54767
17855         PR tree-optimization/53465
17856         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17857         (vrp_visit_phi_node): For PHI arguments coming via backedges
17858         drop all symbolical range information.
17859         (execute_vrp): Compute backedges.
17861 2013-01-16  Richard Biener  <rguenther@suse.de>
17863         * doc/install.texi: Update CLooG and ISL requirements to
17864         0.18.0 and 0.11.1.
17866 2013-01-16  Christian Bruel  <christian.bruel@st.com>
17868         PR target/55301
17869         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17870         (broken_move): Handle UNSPECV_SP_SWITCH_B.
17871         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17873 2013-01-16  DJ Delorie  <dj@redhat.com>
17875         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17876         (UNSPECV_SP_SWITCH_E): New.
17877         (sp_switch_1): Change to an unspec.
17878         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
17879         replace $r15.
17881 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
17883         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17884         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17885         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17886         (expand_mem_thread_fence): Ditto.
17887         (expand_mem_signal_fence): Ditto.
17888         (expand_atomic_load): Ditto.
17889         (expand_atomic_store): Ditto.
17891 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
17893         PR rtl-optimization/55547
17894         PR rtl-optimization/53827
17895         PR debug/53671
17896         PR debug/49888
17897         * alias.c (memrefs_conflict_p): Set sizes to negative after
17898         AND adjustments.
17900 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17902         PR target/55940
17903         * function.c (thread_prologue_and_epilogue_insns): Always
17904         add crtl->drap_reg to set_up_by_prologue.set, even if
17905         stack_realign_drap is false.
17907 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
17909         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17910         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17911         *call): Fix indention.
17913 2013-01-15  Tom de Vries  <tom@codesourcery.com>
17915         PR target/55876
17916         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17917         Update comment.
17919 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
17921         PR rtl-optimization/55153
17922         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17924 2013-01-15  Martin Jambor  <mjambor@suse.cz>
17926         PR tree-optimization/55920
17927         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17928         accesses as grp_to_be_debug_replaced.
17930 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17932         PR tree-optimization/55920
17933         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17934         there is non-useless type conversion needed from debug rhs to lhs,
17935         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17937 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
17938             Mikael Pettersson  <mikpe@it.uu.se>
17940         PR target/43961
17941         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
17942         Thumb.
17943         (ASM_OUTPUT_CASE_LABEL): Remove.
17944         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
17945         * final.c (shorten_branches): Update alignment of labels before
17946         jump tables if CASE_VECTOR_SHORTEN_MODE.
17948 2013-01-15  Richard Biener  <rguenther@suse.de>
17950         PR bootstrap/55961
17951         * system.h: Do not include gmp.h for building host tools.
17953 2013-01-15  Richard Biener  <rguenther@suse.de>
17955         PR middle-end/55882
17956         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
17957         account for bitpos when computing alignment.
17959 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
17961         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
17962         (ix86_target_macros_internal): Likewise.
17964         * config/i386/i386.c (m_CORE2I7): Removed.
17965         (m_CORE_HASWELL): New macro.
17966         (m_CORE_ALL): Likewise.
17967         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
17968         (initial_ix86_arch_features): Likewise.
17969         (processor_target_table): Initializations for Core avx2.
17970         (cpu_names): New names "core-avx2".
17971         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
17972         PROCESSOR_CORE_HASWELL.
17973         (ix86_issue_rate): New case.
17974         (ia32_multipass_dfa_lookahead): Likewise.
17975         (ix86_sched_init_global): Likewise.
17977         * config/i386/i386.h (TARGET_HASWELL): New macro.
17978         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
17979         (processor_type): New PROCESSOR_HASWELL.
17981 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17983         PR tree-optimization/55955
17984         * tree-vect-loop.c (vectorizable_reduction): Give up early on
17985         *SHIFT_EXPR and *ROTATE_EXPR codes.
17987         PR tree-optimization/48766
17988         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
17989         -ftrapv disable -fwrapv.
17991 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
17993         PR target/55974
17994         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
17995         etc. to 1 and not to __flash.
17996         Use LL suffix for __INT24_MAX__ with -mint8.
17997         Use ULL suffix for __UINT24_MAX__ with -mint8.
17999 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18001         * config/avr/avr-arch.h
18002         (struct base_arch_s): Use typedef avr_arch_t instead.
18003         (struct arch_info_s): Use typedef avr_arch_info_t instead.
18004         (struct mcu_type_s): Use typedef avr_mcu_t instead.
18005         * config/avr/avr.c: Same.
18006         * config/avr/avr-devices.c: Same.
18007         * config/avr/driver-avr.c: Same.
18008         * config/avr/gen-avr-mmcu-texi.c: Same.
18009         * config/avr/avr-mcus.def: Adjust comment.
18011 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
18013         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18014         * config/aarch64/iterators.md (VALLDI): New.
18016 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
18017             Andi Kleen  <ak@linux.intel.com>
18019         PR target/55948
18020         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18021         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18022         memmodel flag.
18024 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18026         * config/avr/avr-stdint.h: Remove trailing blanks.
18027         * config/avr/avr-log.h: Same.
18028         * config/avr/avr-arch.h: Same.
18029         * config/avr/avr-devices.c: Same.
18030         * config/avr/avr-dimode.md: Same.
18031         * config/avr/predicates.md: Same.
18032         * config/avr/avr-c.c: Same.  And fix typo.
18034         * config/avr/avr-protos.h: Same.  And:
18035         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18036         (init_cumulative_args): Rename to avr_init_cumulative_args.
18037         (expand_prologue): Rename to avr_expand_prologue.
18038         (expand_epilogue): Rename to avr_expand_epilogue.
18039         (adjust_insn_length): Rename to avr_adjust_insn_length.
18040         (notice_update_cc): Rename to avr_notice_update_cc.
18041         (final_prescan_insn): Rename to avr_final_prescan_insn.
18042         * config/avr/avr.c: Same.
18043         * config/avr/avr.h: Same.
18044         * config/avr/avr.md: Remove trailing blanks.
18045         (prologue): Use avr_expand_prologue.
18046         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18048 2013-01-14  Richard Biener  <rguenther@suse.de>
18050         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18051         verify_location, collect_subblocks): New functions.
18052         (verify_gimple_in_cfg): Verify that locations only reference
18053         BLOCKs in the functions BLOCK tree.
18055 2013-01-14  Richard Biener  <rguenther@suse.de>
18057         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18058         PHI argument.
18059         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18060         unshare reference.
18061         (insert_out_of_ssa_copy_on_edge): Likewise.
18062         (rewrite_close_phi_out_of_ssa): Likewise.
18063         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18064         debug expressions.
18065         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18066         propagated constants.
18067         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18068         can not be shared.
18070 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18072         * config/avr/avr-modes.def: Add GPL copyright notice.
18074 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
18076         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18077         MEMMODEL_MASK to determine memory model.
18078         (atomic_store<mode>): Ditto from operands[2].
18079         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18081 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
18083         PR fortran/55935
18084         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18085         (fold_gimple_assign): Don't call unshare_expr here.
18086         (fold_ctor_reference): Call unshare_expr.
18088 2013-01-13  Terry Guo  <terry.guo@arm.com>
18090         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18091         * doc/fragments.texi: Document MULTILIB_REUSE.
18092         * gcc.c (multilib_reuse): New internal spec.
18093         (set_multilib_dir): Also search multilib from multilib_reuse.
18094         * genmultilib (tmpmultilib3): Refactor code.
18095         (tmpmultilib4): Ditto.
18096         (multilib_reuse): New multilib argument.
18098 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
18100         * Makefile.in: Update copyright.
18102 2013-01-12  Tom de Vries  <tom@codesourcery.com>
18104         PR middle-end/55890
18105         * calls.c (expand_call): Check if arg_nr is valid.
18107 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
18109         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18110         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18111         documentation.  Add missing '__' in front of
18112         __builtin_ia32_packssdw256.
18114 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18116         PR target/55719
18117         * config/s390/s390.c (s390_preferred_reload_class): Do not return
18118         NO_REGS for larl operands.
18119         (s390_reload_larl_operand): Use s390_load_address instead of
18120         emit_move_insn.
18122 2013-01-11  Richard Biener  <rguenther@suse.de>
18124         * tree-cfg.c (verify_node_sharing_1): Split out from ...
18125         (verify_node_sharing): ... here.
18126         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
18128 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
18130         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
18131         Substitute TREECHECKING.
18132         * configure: Regenerate.
18133         * Makefile.in (TREECHECKING): New.
18135 2013-01-11  Richard Guenther  <rguenther@suse.de>
18137         PR tree-optimization/44061
18138         * tree-vrp.c (extract_range_basic): Compute zero as
18139         value-range for __builtin_constant_p of function parameters.
18141 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
18143         Update copyright years.
18145 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
18147         PR rtl-optimization/55672
18148         * lra-eliminations.c (mark_not_eliminable): Permit addition with
18149         const to be eliminable.
18151 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
18153         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
18154         * configure: Regenerate.
18156 2013-01-10  Richard Biener  <rguenther@suse.de>
18158         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
18160 2013-01-10  Richard Biener  <rguenther@suse.de>
18162         PR bootstrap/55792
18163         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
18164         locations for virtual PHI arguments.
18165         (rewrite_update_phi_arguments): Likewise.
18167 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
18169         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
18170         on to assembler.
18172 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
18174         PR tree-optimization/55921
18175         * tree-complex.c (expand_complex_asm): New function.
18176         (expand_complex_operations_1): Call it for GIMPLE_ASM.
18178 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18180         PR target/55718
18181         * config/s390/s390.c (s390_symref_operand_p)
18182         (s390_loadrelative_operand_p): Merge the two functions.
18183         (s390_check_qrst_address, print_operand_address): Add parameters
18184         to s390_loadrelative_operand_p invokation.
18185         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
18186         (s390_reload_larl_operand, s390_secondary_reload): Use
18187         s390_loadrelative_operand_p instead of s390_symref_operand_p.
18188         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
18190 2013-01-09  Mike Stump  <mikestump@comcast.net>
18192         * dse.c (record_store): Remove unnecessary assert.
18194 2013-01-09  Jan Hubicka  <jh@suse.cz>
18196         PR tree-optimization/55569
18197         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
18198         * cfgloop.h (scale_loop_profile): Likewise.
18200 2013-01-09  Jan Hubicka  <jh@suse.cz>
18202         PR lto/45375
18203         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
18204         functions.
18205         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
18207 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
18209         PR middle-end/55114
18210         * expr.h (maybe_emit_group_store): Declare.
18211         * expr.c (maybe_emit_group_store): New function.
18212         * builtins.c (expand_builtin_int_roundingfn): Call it.
18213         (expand_builtin_int_roundingfn_2): Likewise.
18215 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
18217         PR rtl-optimization/55829
18218         * lra-constraints.c (match_reload): Add code for absent output.
18219         (curr_insn_transform): Add code for reloads of matched inputs
18220         without output.
18222 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
18224         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
18225         attribute of movddup insn to DF.
18226         (*vec_interleave_lowv2df): Ditto.
18227         (vec_dupv2df): Ditto.
18229 2013-01-09  Jan Hubicka  <jh@suse.cz>
18231         PR tree-optimiation/55875
18232         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
18233         EVERY_ITERATION parameter.
18234         (number_of_iterations_exit): Check if exit is executed every iteration.
18235         (idx_infer_loop_bounds): Similarly here.
18236         (n_of_executions_at_most): Simplify
18237         to only test for cases where statement is dominated by the
18238         particular bound; handle correctly the "postdominance" test.
18239         (scev_probably_wraps_p): Use max loop iterations info
18240         as a global bound first.
18242 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
18243             Nick Clifton  <nickc@redhat.com>
18245         * config/v850/v850.md (cbranchsf4): New pattern.
18246         (cstoresf4): New pattern.
18247         (cbranchdf4): New pattern.
18248         (cstoredf4): New pattern.
18249         (movsicc): Disallow floating point comparisons.
18250         (cmpsf_le_insn): Fix order of operators.
18251         (cmpsf_lt_insn): Likewise.
18252         (cmpsf_eq_insn): Likewise.
18253         (cmpdf_le_insn): Likewise.
18254         (cmpdf_lt_insn): Likewise.
18255         (cmpdf_eq_insn): Likewise.
18256         (cmpsf_ge_insn): Use LE comparison.
18257         (cmpdf_ge_insn): Likewise.
18258         (cmpsf_gt_insn): Use LT comparison.
18259         (cmpdf_gt_insn): Likewise.
18260         (cmpsf_ne_insn): Delete pattern.
18261         (cmpdf_ne_insn): Delete pattern.
18262         * config/v850/v850.c (v850_gen_float_compare): Use
18263         gen_cmpdf_eq_insn for NE comparison.
18264         (v850_float_z_comparison_operator)
18265         (v850_float_nz_comparison_operator): Move from here ...
18266         * config/v850/predicates.md: ... to here.  Move GT and GE
18267         comparisons into v850_float_z_comparison_operator.
18268         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
18269         Delete prototype.
18270         (v850_float_nz_comparison_operator): Likewise.
18272 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18274         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
18275         with calls to gen_insvsi/gen_insvdi.
18277 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
18279         * config/i386/i386.c (initial_ix86_tune_features): Set up
18280         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
18282 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
18283             Jakub Jelinek  <jakub@redhat.com>
18285         PR tree-optimization/48189
18286         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
18287         If nitercst is 0, don't predict the exit edge.
18289 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18291         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
18292         in asm_fprintf with reg_names.
18293         (aarch64_print_operand_address): Likewise.
18294         (aarch64_return_addr): Likewise.
18295         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
18297 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18299         * config/pa/pa.h (VAL_U6_BITS_P): Define.
18300         (INT_U6_BITS): Likewise.
18301         * config/pa/predicates.md (uint6_operand): New predicate.
18302         (shift5_operand, shift6_operand): Likewise.
18303         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
18304         arith32_operand.
18305         (lshrdi3): Use shift6_operand.
18306         (shrpsi4, shrpdi4): New insn patterns.
18307         (extzv): Delete expander.
18308         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
18309         predicates in unamed zero extract patterns.  Tighten common constraint.
18310         (extv): Delete expander.
18311         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
18312         predicates in unamed sign extract patterns.  Tighten common constraint.
18313         (insv): Delete expander.
18314         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
18315         predicates in unamed insert patterns.  Tighten common constraint.
18316         Change uint32_operand predicate to uint6_operand predicate in unamed
18317         DImode pattern to insert constant values of type 1...1xxxx.
18319 2013-01-04  Jan Hubicka  <jh@suse.cz>
18321         PR tree-optimization/55823
18322         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
18323         issue.
18325 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18326             Uros Bizjak  <ubizjak@gmail.com>
18328         PR rtl-optimization/55845
18329         * df-problems.c (can_move_insns_across): Stop scanning at
18330         volatile_insn_p source instruction or give up if
18331         across_from .. across_to range contains any volatile_insn_p
18332         instructions.
18334 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
18336         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
18337         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
18338         Declare.
18339         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
18340         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
18342 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18344         PR fortran/55341
18345         * asan.c (asan_clear_shadow): New function.
18346         (asan_emit_stack_protection): Use it.
18348 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
18350         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
18351         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
18352         with tab instead of space.
18354 2013-01-08  Nick Clifton  <nickc@redhat.com>
18356         * config/rl78/rl78.c (rl78_expand_prologue): Always select
18357         register bank 0 at the start of an interrupt handler.
18358         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
18359         MDBH registers.
18361 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
18363         * config/aarch64/aarch64-simd.md
18364         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
18365         (aarch64_simd_bsl): Likewise.
18366         (aarch64_vcond_internal<mode>): Likewise.
18367         (vcond<mode><mode>): Likewise.
18368         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
18369         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
18371 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
18373         * config/aarch64/aarch64-builtins.c
18374         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
18376 2013-01-08  Martin Jambor  <mjambor@suse.cz>
18378         PR debug/55579
18379         * tree-sra.c (analyze_access_subtree): Return true also after
18380         potentially creating a debug-only replacement.
18382 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18384         PR middle-end/55890
18385         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
18387         PR tree-optimization/54120
18388         * tree-vrp.c (range_fits_type_p): Don't allow
18389         src_precision < precision from signed vr to unsigned_p
18390         if vr->min or vr->max is negative.
18391         (simplify_float_conversion_using_ranges): Test can_float_p
18392         against CODE_FOR_nothing.
18394 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18395             Richard Biener  <rguenther@suse.de>
18397         PR middle-end/55851
18398         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
18399         types instead of just INTEGER_TYPE types.
18401 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
18403         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
18404         TF_SIZE): Define.
18406 2013-01-07  Steve Ellcey  <sellcey@mips.com>
18408         PR target/42661
18409         * config/mips/mips.opt: Change mad to mmad to match documentation.
18411 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18413         PR target/55897
18414         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
18415         .progmemx.data now.
18417 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18419         PR target/55897
18420         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
18421         (avr_addrspace_t): Add .section_name field.
18422         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
18423         array size.
18424         (avr_addrspace): Same.  Initialize .section_name.  Remove last
18425         NULL entry.  Put __memx into .progmemx.data.
18426         (progmem_section_prefix): Remove.
18427         (avr_asm_init_sections): No need to initialize progmem_section.
18428         (avr_asm_named_section): Use avr_addrspace[].section_name to get
18429         section name prefix.
18430         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
18431         retrieve the progmem section.
18432         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
18433         boundary to run over avr_addrspace[].
18434         (avr_register_target_pragmas): Ditto.
18436 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
18438         * varasm.c (output_constant_def_contents): For asan_protect_global
18439         protected strings, adjust DECL_ALIGN if needed, before testing for
18440         anchored symbols.
18441         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
18442         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
18443         normal decls.
18444         (output_object_block): For asan protected decls, emit asan padding
18445         after their contents.
18446         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
18447         (asan_finish_file): Test it here instead.
18449 2013-01-07  Nick Clifton  <nickc@redhat.com>
18450             Matthias Klose  <doko@debian.org>
18451             Doug Kwan  <dougkwan@google.com>
18452             H.J. Lu  <hongjiu.lu@intel.com>
18454         PR driver/55470
18455         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
18457         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
18459         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
18461         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
18463         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
18465 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18467         PR target/54461
18468         * doc/install.texi (Cross-Compiler-Specific Options): Document
18469         --with-avrlibc.
18471 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
18473         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
18474         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
18475         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
18476         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
18477         vqmovun_high_s64): Fix source operand number and update copyright.
18479 2013-01-07  Richard Biener  <rguenther@suse.de>
18481         PR middle-end/55890
18482         * gimple.h (gimple_call_builtin_p): New overload.
18483         * gimple.c (validate_call): New function.
18484         (gimple_call_builtin_p): Likewise.
18485         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18486         Use gimple_call_builtin_p.
18487         (find_func_clobbers): Likewise.
18488         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
18489         (strlen_optimize_stmt): Likewise.
18491 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18493         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
18494         (vld1q_dup_*): Likewise.
18495         (vld1_*): Likewise.
18496         (vld1q_*): Likewise.
18497         (vld1_lane_*): Likewise.
18498         (vld1q_lane_*): Likewise.
18500 2013-01-07  Richard Biener  <rguenther@suse.de>
18502         * lto-streamer.h (LTO_minor_version): Bump to 2.
18504 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18506         * config/aarch64/aarch64-protos.h
18507         (aarch64_const_double_zero_rtx_p): Rename to...
18508         (aarch64_float_const_zero_rtx_p): ...this.
18509         (aarch64_float_const_representable_p): New.
18510         (aarch64_output_simd_mov_immediate): Likewise.
18511         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
18512         move immediate case.
18513         * config/aarch64/aarch64.c
18514         (aarch64_const_double_zero_rtx_p): Rename to...
18515         (aarch64_float_const_zero_rtx_p): ...this.
18516         (aarch64_print_operand): Allow printing of new constants.
18517         (aarch64_valid_floating_const): New.
18518         (aarch64_legitimate_constant_p): Check for valid floating-point
18519         constants.
18520         (aarch64_simd_valid_immediate): Likewise.
18521         (aarch64_vect_float_const_representable_p): New.
18522         (aarch64_float_const_representable_p): Likewise.
18523         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
18524         (aarch64_output_simd_mov_immediate): New.
18525         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
18526         (*movdf_aarch64): Likewise.
18527         * config/aarch64/constraints.md (Ufc): New.
18528         (Y): call aarch64_float_const_zero_rtx.
18529         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
18531 2013-01-07  Richard Biener  <rguenther@suse.de>
18533         PR tree-optimization/55888
18534         PR tree-optimization/55862
18535         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
18536         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
18537         not if it is contained therein.
18539 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18541         * config/avr/t-avr: Typo.
18543 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18545         PR55243
18546         * config/avr/t-avr: Don't automatically rebuild
18547         $(srcdir)/config/avr/t-multilib
18548         $(srcdir)/config/avr/avr-tables.opt
18549         $(srcdir)/doc/avr-mmcu.texi
18550         (avr-mcus): New phony target to build them on request.
18551         (s-avr-mlib, s-avr-mmcu-texi): Remove.
18552         * avr/avr-mcus.def: Adjust comments.
18554 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
18556         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
18558 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18560         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
18562 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18564         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
18566 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
18568         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
18569         to generate profiling.
18570         * config/rs6000/aix64.h (LIB_SPEC): Same.
18572 2013-01-04  Andrew Pinski  <apinski@cavium.com>
18574         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
18575         New function.
18576         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
18578 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
18580         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
18581         unconditionally.
18582         (ix86_expand_move): Ditto.
18583         (ix86_zero_extend_to_Pmode): Ditto.
18584         (ix86_expand_call): Ditto.
18585         (ix86_expand_special_args_builtin): Ditto.
18586         (ix86_expand_builtin): Ditto.
18588 2013-01-04  Richard Biener  <rguenther@suse.de>
18590         PR tree-optimization/55862
18591         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
18592         translating them through PHI nodes.
18594 2013-01-04  Martin Jambor  <mjambor@suse.cz>
18596         PR tree-optimization/55755
18597         * tree-sra.c (sra_modify_assign): Do not check that an access has no
18598         children when trying to avoid producing a VIEW_CONVERT_EXPR.
18600 2013-01-04  Marek Polacek  <polacek@redhat.com>
18602         PR middle-end/55859
18603         * opts.c (default_options_optimization): Clarify error message.
18605 2013-01-04  Richard Biener  <rguenther@suse.de>
18607         PR middle-end/55863
18608         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
18609         reassociation.
18611 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18613         PR target/53789
18614         * config/pa/pa.md (movsi): Revert previous change.
18615         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
18616         references.
18618 2013-01-03  Richard Henderson  <rth@redhat.com>
18620         * config/i386/i386.c (ix86_expand_move): Always assign to op1
18621         after eliminating TLS symbols.
18623 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
18625         PR bootstrap/50167
18626         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
18627         * graphite-poly.c (debug_gmp_value): Likewise.
18629 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
18631         PR target/55712
18632         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
18633         selected code model, define __code_mode_small__, __code_model_medium__,
18634         __code_model_large__, __code_model_32__ or __code_model_kernel__.
18635         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
18636         xchg temporary register with %k.  Declare temporary register as
18637         early clobbered.
18638         [__x86_64__]: For medium and large code models, preserve %rbx register.
18640 2013-01-03  Richard Biener  <rguenther@suse.de>
18642         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
18643         (dump_subscript): Adjust.
18644         (finalize_ddr_dependent): Do not dump redundant info.
18645         (analyze_siv_subscript): Adjust.
18646         (subscript_dependence_tester): Likewise.
18647         (compute_affine_dependence): Likewise.
18649 2013-01-03  Richard Biener  <rguenther@suse.de>
18651         Revert
18652         2013-01-03  Richard Biener  <rguenther@suse.de>
18654         PR tree-optimization/55857
18655         * tree-vect-stmts.c (vectorizable_load): Do not setup
18656         re-alignment for invariant loads.
18658         2013-01-02  Richard Biener  <rguenther@suse.de>
18660         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18661         invariant load do not generate a vector load from the scalar location.
18663 2013-01-03  Richard Biener  <rguenther@suse.de>
18665         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
18666         for not vectorizing.
18667         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
18668         not build INDIRECT_REFs, call get_name once only.
18669         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
18670         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
18672 2013-01-03  Richard Biener  <rguenther@suse.de>
18674         PR tree-optimization/55857
18675         * tree-vect-stmts.c (vectorizable_load): Do not setup
18676         re-alignment for invariant loads.
18678 2013-01-03  Richard Biener  <rguenther@suse.de>
18680         PR lto/55848
18681         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
18682         prefer a built-in decl.
18684 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18686         * gcc.c (process_command): Update copyright notice dates.
18687         * gcov.c (print_version): Likewise.
18688         * gcov-dump.c (print_version): Likewise.
18690         PR rtl-optimization/55838
18691         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
18692         iv0.step, iv1.step and step.
18694 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18695             Marc Glisse  <marc.glisse@inria.fr>
18697         PR tree-optimization/55832
18698         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
18699         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
18700         integer_{one,zero}_node.
18702 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18704         PR debug/54402
18705         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
18706         * var-tracking.c (reverse_op): Don't add reverse ops to
18707         VALUEs that have already
18708         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
18710 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
18712         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
18714 2013-01-02  Teresa Johnson  <tejohnson@google.com>
18716         * dumpfile.c (dump_loc): Print filename with location.
18717         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
18718         new location_t parameter to emit complete unroll message with
18719         new dump framework.
18720         (canonicalize_loop_induction_variables): Compute loops location
18721         and pass to try_unroll_loop_completely.
18722         * loop-unroll.c (report_unroll_peel): New function.
18723         (peel_loops_completely): Use new dump format with location
18724         for main dumpfile message, and invoke report_unroll_peel on success.
18725         (decide_unrolling_and_peeling): Ditto.
18726         (decide_peel_once_rolling): Remove old dumpfile message subsumed
18727         by report_unroll_peel.
18728         (decide_peel_completely): Ditto.
18729         (decide_unroll_constant_iterations): Ditto.
18730         (decide_unroll_runtime_iterations): Ditto.
18731         (decide_peel_simple): Ditto.
18732         (decide_unroll_stupid): Ditto.
18733         * cfgloop.c (get_loop_location): New function.
18734         * cfgloop.h (get_loop_location): Declare.
18736 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18738         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
18739         NULL.
18741 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18743         PR middle-end/55198
18744         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
18745         BLKmode objects when EXPAND_MEMORY is specified.
18747 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18749         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
18750         in loop predicate.
18751         (fold_builtin_cpu): Do not share cpu model decls across statements.
18753 2013-01-02  Jason Merrill  <jason@redhat.com>
18755         PR c++/55804
18756         * tree.c (build_array_type_1): Revert earlier change.
18758 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
18760         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
18761         "cortex-a57".
18762         * config/aarch64/aarch64-tune.md: Re-generate.
18764 2013-01-02  Richard Biener  <rguenther@suse.de>
18766         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18767         invariant load do not generate a vector load from the scalar location.
18769 2013-01-02  Richard Biener  <rguenther@suse.de>
18771         PR bootstrap/55784
18772         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
18773         * configure: Regenerate.
18775 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18777         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
18778         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
18779         (expand_builtin_int_roundingfn_2): Keep the original target around
18780         for the fallback case.
18782 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18784         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
18785         to be clear for sign changes.
18787 2013-01-01  Jan Hubicka  <jh@suse.cz>
18789         * ipa-inline-analysis.c: Fix formatting.
18791 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
18793         PR tree-optimization/55831
18794         * tree-vect-loop.c (get_initial_def_for_induction): Use
18795         gsi_after_labels instead of gsi_start_bb.
18797 Copyright (C) 2013 Free Software Foundation, Inc.
18799 Copying and distribution of this file, with or without modification,
18800 are permitted in any medium without royalty provided the copyright
18801 notice and this notice are preserved.