* config/i386/i386.c (ix86_return_in_memory): Do not
[official-gcc.git] / gcc / ChangeLog
blobfd65e13c7f8b68fcd4f12f612dfbaaa48815b154
1 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
3         * config/i386/i386.c (ix86_debug_options): Remove prototype.
4         (x86_64_elf_select_section): Ditto.
5         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
6         arguments.
7         (ix86_pass_by_reference): Ditto.
8         (output_set_got): Ditto.
9         (ix86_unary_operator_ok): Ditto.
10         (ix86_expand_builtin): Ditto.
12 2013-08-23  Jan Hubicka  <jh@suse.cz>
14         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
16 2013-08-23  Jan Hubicka  <jh@suse.cz>
18         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
19         (tree_decl_with_vis): Add FINAL field.
21 2013-08-23  Jeff Law  <law@redhat.com>
23         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
24         the debugging dump when the expression is fully redundant.
26 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
28         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
29         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
30         * pretty-print.c (pp_formatted_text_data): Likewise.
31         (pp_write_text_to_stream): Likewise.
32         (pp_write_text_as_dot_label_to_stream): Likewise.
33         (pp_append_r): Likewise.
34         (pp_format): Likewise.
35         (pp_flush): Likewise.
36         (pp_clear_output_area): Likewise.
37         (pp_append_text): Likewise.
38         (pp_formatted_text): Likewise.
39         (pp_remaining_character_count_for_line): Likewise.
40         (pp_newline): Likewise.
41         (pp_character): Likewise.
42         (output_buffer::~output_buffer): Define.
43         (pretty_printer::~pretty_printer): Destruct output buffer.
44         * pretty-print.h (output_buffer::~output_buffer): Declare.
45         (pretty_printer::~pretty_printer): Declare virtual.
47 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
49         PR other/57324
50         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
51         HOST_WIDE_INT_M1U): New macros.
52         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
53         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
54         unsigned -1 for lshift.
55         * cse.c (cse_insn): Likewise.
56         * double-int.c (rshift_double, lshift_double): Likewise.
57         * builtins.c (fold_builtin_bitop): Likewise.
58         * combine.c (force_to_mode): Likewise.
59         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
60         * simplify-rtx.c (simplify_const_unary_operation,
61         simplify_const_binary_operation): Likewise.
62         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
63         check_va_list_escapes): Likewise.
64         * rtlanal.c (nonzero_bits1): Likewise.
65         * expmed.c (expand_smod_pow2): Likewise.
66         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
68 2013-08-23  Jan Hubicka  <jh@suse.cz>
70         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
71         as having address taken.
73 2013-08-23  Jan Hubicka  <jh@suse.cz>
75         * ipa-utils.h (method_class_type): Declare.
76         * ipa-devirt.c (method_class_type): Export.
78         * cgraphunit.c (analyze_functions): Do basic devirtualization;
79         do not walk base classes of anonymous types.
81 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
83         PR rtl-optimization/58220
84         PR regression/58221
85         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
86         handle SEQUENCE insns properly.
88 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
90         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
91         definition.
92         (pp_newline_and_indent): Likewise.
93         (pp_separate_with): Likewise.
94         * pretty-print.c (pp_newline_and_flush): Define.
95         (pp_newline_and_indent): Likewise.
96         (pp_separate_with): Likewise.
98 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
100         PR target/58218
101         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
102         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
104 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
106         * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
107         * gcc/config/i386/i386.md (*movti_internal): Use
108         predicate to determine if EVEX is needed.
109         (*movsi_internal): Ditto.
110         (*movdf_internal): Ditto.
111         (*movsf_internal): Ditto.
112         * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
114 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
116         PR tree-optimization/58209
117         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
118         (find_tail_calls): Give up for pointer result types if m is non-NULL.
119         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
120         emit POINTER_PLUS_EXPR.
121         (create_tailcall_accumulator): For pointer result type accumulate in
122         sizetype type.
124 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
126         * configure.ac: Add backslashes missing from the last change.
127         * configure: Regenerate.
129 2013-08-22  Jan Hubicka  <jh@susue.cz>
131         * ipa.c (function_and_variable_visibility): First remember function
132         was global and then make it local.
134 2013-08-22  Julian Brown  <julian@codesourcery.com>
136         * configure.ac: Add aarch64 to list of arches which use "nop" in
137         debug_line test.
138         * configure: Regenerate.
140 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
142         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
143         gnu_libc_has_function.
144         * config/s390/tpf.h: Likewise.
146 2013-08-22  Jan Hubicka  <jh@susue.cz>
148         * timevar.c (validate_phases): Add cast.
150 2013-08-22  Jan Hubicka  <jh@susue.cz>
152         * timevar.c (validate_phases): Use size_t for memory.
153         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
155 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
157         * pretty-print.h (output_buffer::output_buffer): Declare.
158         (pretty_printer::pretty_printer): Likewise.
159         (pp_construct): Remove.
160         * pretty-print.c (output_buffer::output_buffer): Define.
161         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
162         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
163         (print_gimple_expr): Likewise.
164         (print_gimple_seq): Likewise.
165         (gimple_dump_bb): Likewise.
166         * sched-vis.c (dump_value_slim): Likewise.
167         (dump_insn_slim): Likewise.
168         (dump_rtl_slim): Likewise.
169         (str_pattern_slim): Likewise.
170         * tree-mudflap.c (mf_varname_tree): Likewise.
171         * graph.c (print_graph_cfg): Likewise.
172         (start_graph_dump): Likewise.
173         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
174         placement-new.
175         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
176         pretty printer initialization.
177         * coretypes.h (diagnostic_context): Remove superflous type alias
178         declaration.
179         (pretty_printer): Likewise.  Declare directly as a class.
180         (pretty_print_info): Remove declaration as class.
181         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
182         and pp_clear_output_area.
183         (asan_add_global): Likewise.
185 2013-08-22  Jan Hubicka  <jh@suse.cz>
187         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
188         * ipa-utils.h (update_type_inheritance_graph): Declare.
189         (possible_polymorphic_call_target_p): Declare.
190         (possible_polymorphic_call_target_p): New.
191         * ipa-devirt.c: Update toplevel comments.
192         (cached_polymorphic_call_targets): Move up.
193         (odr_type_d): Move ID down.
194         (polymorphic_type_binfo_p): Update comment.
195         (odr_hasher::remove): Likewise;
196         (get_odr_type): Set anonymous_namespace.
197         (dump_odr_type): Dump it.
198         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
199         (maybe_record_node): Record node in cached_polymorphic_call_targets.
200         (record_binfo): Add comment.
201         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
202         built.
203         (devirt_node_removal_hook): Do not iCE when cache is freed.
204         (possible_polymorphic_call_target_p): New predicate.
205         (update_type_inheritance_graph): New function.
207 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
208             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
209             Sergey Lega  <sergey.s.lega@intel.com>
210             Anna Tikhonova  <anna.tikhonova@intel.com>
211             Ilya Tocar  <ilya.tocar@intel.com>
212             Andrey Turetskiy  <andrey.turetskiy@intel.com>
213             Ilya Verbin  <ilya.verbin@intel.com>
214             Kirill Yukhin  <kirill.yukhin@intel.com>
215             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
217         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
218         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
219         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
220         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
221         (OPTION_MASK_ISA_AVX2_UNSET): Update.
222         (OPTION_MASK_ISA_AVX512F_UNSET): New.
223         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
224         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
225         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
226         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
227         OPT_mavx512pf, OPT_mavx512er cases.
228         * config/i386/constraints.md (v): New constraint.
229         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
230         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
231         (bit_AVX512CD): New.
232         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
233         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
234         * config/i386/i386-c.c (ix86_target_macros_internal):
235         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
236         __AVX512PF__.
237         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
238         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
239         * config/i386/i386.c (regclass_map, dbx_register_map)
240         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
241         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
242         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
243         -mavx512pf options.
244         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
245         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
246         -mavx512pf options.  Fix formatting.
247         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
248         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
249         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
250         -mavx512cd, -mavx512pf options.
251         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
252         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
253         (ix86_preferred_output_reload_class): Replace SSE_REGS with
254         ALL_SSE_REGS.
255         (ix86_hard_regno_mode_ok): Support 512-bit registers.
256         (ix86_set_reg_reg_cost): Ditto.
257         (x86_order_regs_for_local_alloc): Ditto.
258         (MAX_VECT_LEN): Extend to 64-byte.
259         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
260         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
261         (TARGET_AVX512ER, TARGET_AVX512CD): New.
262         (BIGGEST_ALIGNMENT): Extend to 512-bits.
263         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
264         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
265         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
266         (SSE_REG_MODE_P): Support new modes.
267         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
268         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
269         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
270         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
271         (REG_CLASS_CONTENTS): Add new registers.
272         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
273         (EXT_REX_SSE_REGNO_P): New.
274         (HI_REGISTER_NAMES): Add new registers.
275         * config/i386/i386.md: Define constants for new registers.
276         (mode): Add new 512-bit modes.
277         (prefix): Support evex prefix.
278         (isa): Support avx512f, noavx512f, fma_avx512f.
279         (ssemodesuffix): Add new 512-bit modes.
280         (movxi): New.
281         (*movxi_internal_avx512f): Ditto.
282         (*movdi_internal): Replace constraint "x" with the new constraint "v".
283         Support MODE_XI.
284         (*movsi_internal): Likewise.
285         (*movdf_internal): Likewise.
286         (*movsf_internal): Likewise.
287         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
288         (<code><mode>3): Likewise.
289         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
290         New.
291         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
292         with the new constraint "v".
293         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
294         modes.
295         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
296         with the new constraint "v".
297         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
298         (<sse2>_storedqu<avxsizesuffix>): Likewise.
299         (*<plusminus_insn><mode>3): Likewise.
300         (<sse>_vm<plusminus_insn><mode>3): Likewise.
301         (*mul<mode>3): Likewise.
302         (<sse>_vmmul<mode>3): Likewise.
303         (<sse>_div<mode>3): Likewise.
304         (<sse>_vmdiv<mode>3): Likewise.
305         (<sse>_sqrt<mode>2): Likewise.
306         (<sse>_vmsqrt<mode>2): Likewise.
307         (*<code><mode>3_finite): Likewise.
308         (*<code><mode>3) <smaxmin>: Likewise.
309         (<sse>_vm<code><mode>3): Likewise.
310         (*<code><mode>3) <any_logic>: Likewise.
311         (*fma_fmadd_<mode>): Likewise.
312         (*fma_fmsub_<mode>): Likewise.
313         (*fma_fnmadd_<mode>): Likewise.
314         (*fma_fnmsub_<mode>): Likewise.
315         (*fma_fmaddsub_<mode>): Likewise.
316         (*fma_fmsubadd_<mode>): Likewise.
317         (*fmai_fmadd_<mode>): Likewise.
318         (*fmai_fmsub_<mode>): Likewise.
319         (*fmai_fnmadd_<mode>): Likewise.
320         (*fmai_fnmsub_<mode>): Likewise.
321         (sse_cvtsi2ss): Likewise.
322         (sse_cvtsi2ssq): Likewise.
323         (sse_cvtss2si): Likewise.
324         (sse_cvtss2si_2): Likewise.
325         (sse_cvtss2siq): Likewise.
326         (sse_cvtss2siq_2): Likewise.
327         (sse_cvttss2si): Likewise.
328         (sse_cvtss2siq_2): Likewise.
329         (float<sseintvecmodelower><mode>2): Likewise.
330         (sse2_cvtsd2si_2): Likewise.
331         (sse2_cvtsd2siq_2): Likewise.
332         (*<plusminus_insn><mode>3): Likewise.
333         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
334         (*<sse4_1_avx2>_mul<mode>3): Likewise.
335         (ashr<mode>3): Likewise.
336         (<shift_insn><mode>3): Likewise.
337         (avx2_<code><mode>3): Likewise.
338         (*avx2_<code><mode>3): Likewise.
339         (*andnot<mode>3): Likewise.
340         (*<code><mode>3) <any_logic>: Likewise.
341         (abs<mode>2): Likewise.
342         (avx2_permvar<mode>): Likewise.
343         (avx2_perm<mode>_1): Likewise.
344         (*avx_vpermilp<mode>): Likewise.
345         (avx_vpermilvar<mode>3): Likewise.
346         (avx2_ashrv<mode>): Likewise.
347         (avx2_<shift_insn>v<mode>): Likewise.
348         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
349         -mavx512cd.
350         * doc/rtl.texi: Document XImode.
352 2013-08-21  Jeff Law  <law@redhat.com>
354         * tree-flow.h (register_jump_thread): Pass vector of edges
355         instead of each important edge.
356         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
357         thread path into a vector and pass that to register_jump_thread.
358         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
359         passed in edge vector to the current 3-edge form.
361         Revert:
362         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
364         * dce.c (fini_dce): Call df_analyze again just in case
365         delete_unmarked_insns removed anything.
367 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
369         * reload.h (struct reg_equivs): Rename to ..
370         (struct reg_equivs_s): .. this.
372 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
374         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
376 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
378         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
380 2013-08-21  Jeff Law  <law@redhat.com>
382         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
383         simplify assignments too.  If the RHS collapses to a singleton
384         range, then return the value for the range.
386 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
388         * config/i386/sse.md (V16): Rename to...
389         (VMOVE): this.
390         (mov<mode>): Update iterator name.
391         (*mov<mode>_internal): Ditto.
392         (push<mode>1): Ditto.
393         (movmisalign<mode>): Ditto.
395 2013-08-20  Jan Hubicka  <jh@suse.cz>
397         PR bootstrap/58186
398         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
399         entry for direct edges.
400         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
402 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
404         Revert my last two changes, r201865 and r201864:
406         Revert r201865:
407         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
409         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
410         instances can own GC refs.
412         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
413         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
414         (gcc::context::gt_pch_nx): Likewise.
415         (gcc::context::gt_pch_nx):  Likewise.
416         * ggc.h (gt_ggc_mx <T>): New.
417         (gt_pch_nx_with_op <T>): New.
418         (gt_pch_nx <T>): New.
419         * passes.c (opt_pass::gt_ggc_mx): New.
420         (opt_pass::gt_pch_nx): New.
421         (opt_pass::gt_pch_nx_with_op): New.
422         (pass_manager::gt_ggc_mx): New.
423         (pass_manager::gt_pch_nx): New.
424         (pass_manager::gt_pch_nx_with_op): New.
425         (pass_manager::operator new): Use
426         ggc_internal_cleared_alloc_stat rather than xcalloc.
427         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
428         (pass_manager::gt_ggc_mx): New.
429         (pass_manager::gt_pch_nx): New.
430         (pass_manager::gt_pch_nx_with_op): New.
431         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
432         (opt_pass::operator new): New.
433         (opt_pass::gt_ggc_mx): New.
434         (opt_pass::gt_pch_nx): New.
435         (opt_pass::gt_pch_nx_with_op): New.
437         Revert r201864:
438         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
440         * Makefile.in (GTFILES): Add context.h.
441         * context.c (gcc::context::operator new): New.
442         (gcc::context::gt_ggc_mx): New.
443         (gcc::context::gt_pch_nx): New.
444         (gcc::context::gt_pch_nx): New.
445         * context.h (gcc::context): Add GTY((user)) marking.
446         (gcc::context::operator new): New.
447         (gcc::context::gt_ggc_mx): New.
448         (gcc::context::gt_pch_nx): New.
449         (gcc::context::gt_pch_nx): New.
450         (g): Add GTY marking.
451         (gt_ggc_mx (gcc::context *)): New.
452         (gt_pch_nx (gcc::context *)): New.
453         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
454         void *cookie)): New.
455         * gengtype.c (open_base_files) <ifiles>: Add context.h.
457 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
459         * dce.c (fini_dce): Call df_analyze again just in case
460         delete_unmarked_insns removed anything.
462 2013-08-20  Teresa Johnson  <tejohnson@google.com>
464         PR rtl-optimizations/57451
465         * final.c (reemit_insn_block_notes): Prevent lexical blocks
466         from crossing split section boundaries.
468 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
470         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
471         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
472         with MULTLIB_DEFAULTS.
474 2013-08-20  Nick Clifton  <nickc@redhat.com>
476         * target.def (narrow_volatile_bitfield): Note that the default
477         value is false, not !TARGET_STRICT_ALIGN.
478         * doc/tm.texi: Regenerate.
480 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
482         Fix LIB_SPEC for systems without libpthread.
484         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
485         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
486         for Android.
487         * config/i386/linux-common.h: Likewise.
488         * config/mips/linux-common.h: Likewise.
490 2013-08-20  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
492         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
493         checks.
495 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
497         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
498         instances can own GC refs.
500         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
501         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
502         (gcc::context::gt_pch_nx): Likewise.
503         (gcc::context::gt_pch_nx):  Likewise.
504         * ggc.h (gt_ggc_mx <T>): New.
505         (gt_pch_nx_with_op <T>): New.
506         (gt_pch_nx <T>): New.
507         * passes.c (opt_pass::gt_ggc_mx): New.
508         (opt_pass::gt_pch_nx): New.
509         (opt_pass::gt_pch_nx_with_op): New.
510         (pass_manager::gt_ggc_mx): New.
511         (pass_manager::gt_pch_nx): New.
512         (pass_manager::gt_pch_nx_with_op): New.
513         (pass_manager::operator new): Use
514         ggc_internal_cleared_alloc_stat rather than xcalloc.
515         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
516         (pass_manager::gt_ggc_mx): New.
517         (pass_manager::gt_pch_nx): New.
518         (pass_manager::gt_pch_nx_with_op): New.
519         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
520         (opt_pass::operator new): New.
521         (opt_pass::gt_ggc_mx): New.
522         (opt_pass::gt_pch_nx): New.
523         (opt_pass::gt_pch_nx_with_op): New.
525 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
527         * Makefile.in (GTFILES): Add context.h.
528         * context.c (gcc::context::operator new): New.
529         (gcc::context::gt_ggc_mx): New.
530         (gcc::context::gt_pch_nx): New.
531         (gcc::context::gt_pch_nx): New.
532         * context.h (gcc::context): Add GTY((user)) marking.
533         (gcc::context::operator new): New.
534         (gcc::context::gt_ggc_mx): New.
535         (gcc::context::gt_pch_nx): New.
536         (gcc::context::gt_pch_nx): New.
537         (g): Add GTY marking.
538         (gt_ggc_mx (gcc::context *)): New.
539         (gt_pch_nx (gcc::context *)): New.
540         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
541         void *cookie)): New.
542         * gengtype.c (open_base_files) <ifiles>: Add context.h.
544 2013-08-20  Alan Modra  <amodra@gmail.com>
546         PR target/57865
547         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
548         (rs6000_emit_epilogue): Likewise.
550 2013-08-19  Dehao Chen (dehao@google.com)
552         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
554 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
555             Jakub Jelinek  <jakub@redhat.com>
557         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
558         (BUILT_IN_FABSD64): Likewise.
559         (BUILT_IN_FABSD128): Likewise.
560         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
561         (fold_builtin_1): Likewise.
562         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
563         destination and source operands.
564         (*abstd2_fpr): Likewise.
565         (*nabstd2_fpr): Likewise.
567 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
569         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
570         JUMP_P and INSN_P.
572 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
574         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
575         incomplete.
577 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
579         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
580         * builtins.c (default_libc_has_function): New.
581         (gnu_libc_has_function): Ditto.
582         (no_c99_libc_has_function): Ditto.
583         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
584         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
585         (fold_builtin_sincos): Likewise.
586         (fold_builtin_cexp): Likewise.
587         * builtins.def (DEF_C94_BUILTIN): Likewise.
588         (DEF_C99_BUILTIN): Likewise.
589         (DEF_C99_C90RES_BUILTIN): Likewise.
590         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
591         definitions to using this define.
592         * config/darwin-protos.h (darwin_libc_has_function): New.
593         * config/darwin.c (darwin_libc_has_function): Ditto.
594         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
595         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
596         * config/darwin.h: Ditto.
597         * config/elfos.h: Ditto.
598         * config/freebsd.h: Ditto.
599         * config/i386/cygming.h: Ditto.
600         * config/i386/djgpp.h: Ditto.
601         * config/i386/i386-interix.h: Ditto.
602         * config/microblaze/microblaze.h: Ditto.
603         * config/mmix/mmix.h: Ditto.
604         * config/gnu-user.h: Ditto.
605         * config/ia64/hpux.h: Ditto.
606         * config/pa/pa-hpux.h: Ditto.
607         * config/pdp11/pdp11.h: Ditto.
608         * config/picochip/picochip.h: Ditto.
609         * config/linux.h: Ditto.
610         * config/netbsd.h: Ditto.
611         * config/openbsd.h: Ditto.
612         * config/rs6000/aix43.h: Ditto.
613         * config/rs6000/aix51.h: Ditto.
614         * config/rs6000/aix52.h: Ditto.
615         * config/rs6000/aix53.h: Ditto.
616         * config/rs6000/aix61.h: Ditto.
617         * config/rs6000/darwin.h: Ditto.
618         * config/rs6000/linux.h: Ditto.
619         * config/rs6000/linux64.h: Ditto.
620         * config/s390/tpf.h: Ditto.
621         * config/sol2-10.h: Ditto.
622         * config/sol2.h: Ditto.
623         * config/vms/vms.h: Ditto.
624         * config/vxworks.h: Ditto.
625         * config/linux-android.c (linux_android_libc_has_function):
626         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
627         * config/linux-protos.h (linux_android_libc_has_function):
628         New declaration.
629         * config/i386/i386.c (ix86_libc_has_function): New.
630         * config/i386/i386-protos.h
631         (ix86_libc_has_function): New declaration.
632         * config/i386/i386.md
633         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
634         ("isinf<mode>2): Likewise.
635         * convert.c (convert_to_integer): Using new target hook
636         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
637         TARGET_C99_FUNCTIONS.
638         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
639         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
640         * coretypes.h (function_class): New enum for different
641         classes of functions.
642         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
643         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
644         (TARGET_HAS_SINCOS): Likewise.
645         (TARGET_LIBC_HAS_FUNCTION): New.
646         * doc/tm.texi: Regenerated.
647         * targhooks.h (default_libc_has_function): New declaration.
648         (no_c99_libc_has_function): Ditto.
649         (gnu_libc_has_function): Ditto.
650         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
651         and TARGET_HAS_SINCOS.
653 2013-08-18  Jan Hubicka  <jh@suse.cz>
655         * Makeifle-in (ipa-devirt.o): New.
656         (GTFILES): Add ipa-utils.h and ipa-devirt.c
657         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
658         (analyze_functions): Look into possible targets of polymorphic call.
659         * dumpfile.c (dump_files): Add type-inheritance dump.
660         * dumpfile.h (TDI_inheritance): New.
661         * ipa-devirt.c: New file.
662         * ipa-utils.h (odr_type_d): Forward declare.
663         (odr_type): New type.
664         (build_type_inheritance_graph): Declare.
665         (possible_polymorphic_call_targets): Declare and introduce inline
666         variant when only edge is pased.
667         (dump_possible_polymorphic_call_targets): Likewise.
668         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
669         * tree.c (type_in_anonymous_namespace_p): Break out from ...
670         (types_same_for_odr): ... here.
671         * tree.h (type_in_anonymous_namespace_p): Declare.
673 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
675         PR tree-optimization/58006
676         * tree-parloops.c (take_address_of): Don't ICE if get_name
677         returns NULL.
678         (eliminate_local_variables_stmt): Remove clobber stmts.
680 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
682         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
683         error message is issued for an alias to undefined symbol.
685 2013-08-18  Jan Hubicka  <jh@suse.cz>
687         * cgraph.c (cgraph_create_indirect_edge): Discover
688         polymorphic calls and record basic info into indirect_info.
689         * gimple-fold.c (gimple_fold_call): When doing BINFO based
690         devirtualization, ignore objc function calls.
691         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
692         call with no parm index info.
693         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
694         * tree.c (virtual_method_call_p): New function.
695         * tree.h (virtual_method_call_p): Declare.
697 2013-08-16  Jan Hubicka  <jh@suse.cz>
699         PR middle-end/58179
700         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
702 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
704         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
705         attribute.
707 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
709         * gengtype.c (type_for_name): Add special-case support for
710         locating types within the "gcc::" namespace.
711         (open_base_files): Emit a "using namespace gcc" directive.
713 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
715         PR target/58160
716         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
717         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
719         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
720         array instead of each individual operand as a separate argument.
721         (emit_fusion_gpr_load): Likewise.
722         (expand_fusion_gpr_load): Add new function declaration.
724         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
725         signature to have the operands passed as an array, instead of as
726         separate arguments.  Allow ZERO_EXTEND to be in the memory
727         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
728         depend on the register live/dead flags when peepholes are run.
729         (expand_fusion_gpr_load): New function to be called from the
730         peephole2 pass, to change the register that addis sets to be the
731         target register.
732         (emit_fusion_gpr_load): Change the calling signature to have the
733         operands passed as an array, instead of as separate arguments.
734         Allow ZERO_EXTEND to be in the memory address, and also
735         SIGN_EXTEND if -mpower8-fusion-sign.
737         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
738         unspec enumeration.
739         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
740         adjust the register addis loads up in the peephole2 pass.  Do not
741         depend on the register live/dead state when the peephole pass is done.
743 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
745         * gengtype.c (create_user_defined_type): Ensure that the kind
746         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
747         declaration is seen before the GTY((user)) marking.
749 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
751         PR target/58105
752         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
754 2013-08-16  Jan Hubicka  <jh@suse.cz>
756         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
757         arugment expected_type.
758         (gimple_fold_call): Use it.
759         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
760         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
761         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
762         (try_make_edge_direct_virtual_call): Likewise.
763         * tree.c (obj_type_ref_class): New.
764         * tree.h (obj_type_ref_class): Use it.
766 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
768         * sched-vis.c (rtl_slim_pp_initialized): Remove.
769         (rtl_slim_pp): Likewise.
770         (init_rtl_slim_pretty_print): Likewise.
771         (dump_value_slim):  Don't call it.  Use local pretty printer.
772         (dump_insn_slim): Likewise.
773         (dump_rtl_slim): Likewise.
774         (str_pattern_slim): Likewise.
775         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
776         Simplify.
778 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
780         PR tree-optimization/58164
781         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
782         walk gimple_goto_dest of GIMPLE_GOTO.
784         PR tree-optimization/58165
785         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
786         bi_call must be the last stmt in a bb, don't split_block, instead
787         use fallthru edge from it and give up if there is none.
788         Release conds vector when returning early.
790 2013-08-14  Xinliang David Li  <davidxl@google.com>
792         * config/i386/i386.c (ix86_option_override_internal):
793         Remove unused variable and field.
795 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
797         PR target/57949
798         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
799         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
800         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
801         and Linux, correct BLKmode alignment when 128-bit alignment is
802         required and compatibility flag is not set.
803         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
804         for zero-size arguments when compatibility flag is not set.
806 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
808         PR tree-optimization/58145
809         * tree-sra.c (build_ref_for_offset): If prev_base has
810         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
812 2013-08-14  Xinliang David Li  <davidxl@google.com>
814         * config/i386/i386.c (ix86_option_override_internal):
815         Fix uninitialized variable error.
817 2013-08-14  Xinliang David Li  <davidxl@google.com>
819         * config/i386/i386.opt: Define two new options.
820         * config/i386/x86-tune.def: Add arch selector field in macros.
821         * config/i386/i386.h: Adjust macro definition.
822         * config/i386/i386.c (ix86_option_override_internal):
823         Refactor the code.
824         (parse_mtune_ctrl_str): New function.
825         (set_ix86_tune_features): New function.
826         (ix86_function_specific_restore): Call the new helper function.
828 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
830         PR rtl-optimization/57662
831         * sel-sched.c (code_motion_process_successors): When the current insn
832         is removed after the recursive traversal, break from the loop.
833         Add comments and debug printouts.
835 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
836             Alexandre Oliva  <aoliva@redhat.com>
838         PR target/58067
839         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
840         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
841         there also UNSPEC_PLTOFF.
843 2013-08-14  Marek Polacek  <polacek@redhat.com>
845         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
846         NUM_CONDITIONS bit positions.
848 2013-08-13  Cary Coutant  <ccoutant@google.com>
850         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
851         (attr_checksum): Hash vector contents instead of pointer.
852         (attr_checksum_ordered): Likewise.
854 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
856         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
857         when Pmode != word_mode.  Add length_address attribute.
858         (sse3_monitor_<mode>): Merge from sse3_monitor and
859         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
860         Pmode != word_mode.  Update insn length attribute.
861         * config/i386/i386.c (ix86_option_override_internal): Update
862         ix86_gen_monitor selection for merged sse3_monitor insn.
864 2013-08-13  Julian Brown  <julian@codesourcery.com>
866         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
867         perform invalid legitimization on greater-than-word-size modes for
868         TARGET_E500_DOUBLE.
870 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
872         * ira.c (setup_class_translate_array): Use aclass instead of cl
873         for classes not fully covered by allocno classes.
875 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
877         PR tree-optimization/57661
878         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
879         * tree-inline.c (tree_function_versioning): Initialize it.
880         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
881         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
882         that is not being copied.
884         PR sanitizer/56417
885         * asan.c (instrument_strlen_call): Fix typo in comment.
886         Use char * type even for the lhs of POINTER_PLUS_EXPR.
888 2013-08-13  Steve Ellcey  <sellcey@mips.com>
890         * config/mips/mips.md (prefetch): Use lw instead of ld on
891         loongson in 32bit mode.
893 2013-08-13  Nick Clifton  <nickc@redhat.com>
895         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
897 2013-08-13  Jan Hubicka  <jh@suse.cz>
899         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
900         introduced edge; fix typo in sanity check.
901         (cgraph_resolve_speculation): Export; improve diagnostic.
902         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
903         speculation at type mismatch.
904         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
905         (cgraph_resolve_speculation): Declare.
906         (symtab_can_be_discarded): New function.
907         * value-prof.c (gimple_ic_transform): Remove actual transform code.
908         * ipa-inline-transform.c (speculation_removed): New global var.
909         (clone_inlined_nodes): See if speculation can be removed.
910         (inline_call): If speculations was removed, we growths may not match.
911         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
912         (speculation_useful_p): New function.
913         (resolve_noninline_speculation): New function.
914         (inline_small_functions): Resolve useless speculations.
915         * ipa-inline.h (speculation_useful_p): Declare
916         * ipa.c (can_replace_by_local_alias): Simplify.
917         (ipa_profile): Produce speculative calls in non-lto, too;
918         add simple cost model; produce local aliases.
920 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
922         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
923         PASS_MANAGER_H.
925 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
927         * config/i386/i386.c (ix86_function_versions): Use error + inform.
929 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
931         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
932         iterator instead of X87MODEF.
934 2013-08-12  Perez Read  <netfirewall@gmail.com>
936         PR target/58132
937         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
938         operand 0 for intel asm alternative.
939         (*movabs<mode>_2): Ditto for operand 1.
941 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
943         * config/aarch64/arm_none.h
944         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
946 2013-08-12  Nick Clifton  <nickc@redhat.com>
948         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
950 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
952         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
953         expand for QI/HImode operand to produce more effictive code for
954         unsigned char(short) --> float(double) conversion.
956 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
958         * doc/invoke.texi: Mention that -ftls-model does not force the final
959         model.
961 2013-08-12  Marek Polacek  <polacek@redhat.com>
962             Marc Glisse  <marc.glisse@inria.fr>
964         PR tree-optimization/57980
965         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
966         when creating -1 constant.
968 2013-08-10  Jan Hubicka  <jh@suse.cz>
970         Workaround binutils PR14342.
971         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
972         (gimple_init_edge_profiler): Likewise.
973         (gimple_gen_ic_func_profiler): Likewise.
975 2013-08-09  Jan Hubicka  <jh@suse.cz>
977         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
979 2013-08-09  Xinliang David Li  <davidxl@google.com>
981         * config/i386/stringop.def: New file.
982         * config/i386/stringop.opt: New file.
983         * config/i386/i386-opts.h: Include stringopt.def.
984         * config/i386/i386.opt: Include stringopt.opt.
985         * config/i386/i386.c (ix86_option_override_internal):
986         Override default size based stringop inline strategies with options.
987         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
988         New function.
990 2013-08-09  Jan Hubicka  <jh@suse.cz>
992         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
994 2013-08-09  Jan Hubicka  <jh@suse.cz>
996         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
997         CGRAPH_FREQ_MAX.
998         (dump_cgraph_node): Dump profile-id.
999         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1000         and common_target_probability.
1001         * lto-cgraph.c (lto_output_edge): Stream common targets.
1002         (lto_output_node): Stream profile ids.
1003         (input_node): Stream profile ids.
1004         (input_edge): Stream common targets.
1005         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1006         * ipa.c: Include value-prof.h
1007         (ipa_profile_generate_summary): Turn indirect call statement histograms
1008         into common targets.
1009         (ipa_profile): Turn common targets into speculative edges.
1011 2013-08-09  Jan Hubicka  <jh@suse.cz>
1013         * cgraph.h (cgraph_node): Add profile_id.
1014         * value-prof.c (cgraph_node_map): Turn into pointer_map.
1015         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1016         (del_node_map): Update.
1017         (find_func_by_funcdef_no): Replace by ...
1018         (find_func_by_profile_id): ... this one.
1019         (gimple_ic_transform): Do not remove useful histograms when
1020         speculation is not done; dump info when indirect call removal
1021         can happen at LTO.
1022         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1023         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1024         (__gcov_indirect_call_profiler_v2): .. this one.
1025         * profile.h (init_node_map): Update.
1026         * coverage.c (coverage_compute_profile_id): New function.
1027         * coverage.h (coverage_compute_profile_id): Declare.
1028         * tree-profile.c (init_ic_make_global_vars): Make
1029         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
1030         (gimple_init_edge_profiler): Update prototype of
1031         __gcov_indirect_call_profiler.
1032         (gimple_gen_ic_func_profiler): Simplify.
1033         (tree_profiling): Use init_node_map
1035 2013-08-09  Jan Hubicka  <jh@suse.cz>
1037         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1038         non-speculative refs.
1039         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1040         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1041         (cgraph_set_call_stmt): Likewise.
1042         (cgraph_create_edge_1): Fix release checking compilatoin;
1043         clear lto_stmt_uid.
1044         (cgraph_free_edge): Free indirect info.
1045         (cgraph_turn_edge_to_speculative): New function.
1046         (cgraph_speculative_call_info): New function.
1047         (cgraph_make_edge_direct): Return direct edge; handle speculation.
1048         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1049         (dump_cgraph_node): Dump speculation.
1050         (verify_edge_count_and_frequency): Accept speculative edges.
1051         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1052         (verify_cgraph_node): Handle speculation.
1053         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1054         (cgraph_set_call_stmt): Update prototype.
1055         (cgraph_make_edge_direct): Update prototype.
1056         (cgraph_speculative_call_info): Declare.
1057         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1058         to change; update call of ipa_find_references.
1059         * ipa-ref.c (ipa_record_reference): Fix return value; clear
1060         lto_stmt_uid and speculative flags.
1061         (ipa_dump_references): Dump speculation.
1062         (ipa_clone_references): Clone speculative flag.
1063         (ipa_clone_referring): Likewise.
1064         (ipa_clone_ref): New function.
1065         (ipa_find_reference): Look into lto_stmt_uids
1066         (ipa_clear_stmts_in_references): Do not clear speculative calls.
1067         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1068         (ipa_find_reference): Update declaration.
1069         (ipa_clone_ref): Declare.
1070         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1071         stream speculative flag.
1072         (lto_output_ref): Stream statements uids and speculation.
1073         (input_ref): Likewise.
1074         (input_edge): Stream speuclation.
1075         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1076         (cgraph_set_call_stmt_including_clones): Handle speculation.
1077         * ipa-inline.c (heap_edge_removal_hook): New function.
1078         (inline_small_functions): Register it.
1079         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1080         also initialize refs.
1081         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1082         edge to change.
1083         (try_make_edge_direct_simple_call): Likewise.
1084         (try_make_edge_direct_simple_call): Likewise.
1085         (update_indirect_edges_after_inlining): Likewise.
1086         (remove_described_reference): Look proper lto_stmt_uid.
1087         (propagate_controlled_uses): Likewise.
1088         (propagate_controlled_uses): Liekwise.
1089         * tree-inline.c (copy_bb): Copy speculative edges.
1090         (redirect_all_calls): New function.
1091         (copy_cfg_body): Do redirection after loop info is updated.
1092         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1094 2013-08-09  Jan Hubicka  <jh@suse.cz>
1096         * lto-streamer-out.c (output_function): Renumber PHIs.
1097         * lto-streamer-in.c (input_function): Likewise.
1099 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
1101         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1102         (get_lane_unsigned): Likewise.
1103         (dup_lane_scalar): Likewise.
1104         (get_lane): enable for VALL.
1105         * config/aarch64/aarch64-simd.md
1106         (aarch64_dup_lane_scalar<mode>): Remove.
1107         (aarch64_get_lane_signed<mode>): Likewise.
1108         (aarch64_get_lane_unsigned<mode>): Likewise.
1109         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1110         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1111         (aarch64_get_lane<mode>): Enable for all vector modes.
1112         (aarch64_get_lanedi): Remove misleading constraints.
1113         * config/aarch64/arm_neon.h
1114         (__aarch64_vget_lane_any): Define.
1115         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1116         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1117         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1118         * config/aarch64/iterators.md (VDQQH): New.
1119         (VDQQHS): Likewise.
1120         (vwcore): Likewise.
1122 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
1124         * configure.ac: Add GAS check for LEON instructions on SPARC.
1125         * configure: Regenerate.
1126         * config.in: Likewise.
1127         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1128         sparc*-*-* block.
1129         * config/sparc/sparc.opt (LEON, LEON3): New masks.
1130         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1131         for LEON or LEON3.
1132         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1133         (AS_LEON_FLAG): New macro.
1134         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1135         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1136         Deal with LEON and LEON3 for the memory model.
1137         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1138         (atomic_compare_and_swap<mode>_1): Likewise.
1139         (*atomic_compare_and_swap<mode>_1): Likewise.
1141 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1143         * config/arm/neon.md (vcond): Fix floating-point vector
1144         comparisons against 0.
1146 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
1148         * lra-constraints.c (emit_spill_move): Remove assert.
1149         (process_alt_operands): Add more debugging
1150         output.  Increase reject for spilling into memory.  Decrease
1151         reject for reloading scratch.
1152         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1154 2013-08-08  Steve Ellcey  <sellcey@mips.com>
1156         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1157         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1158         micromips incompatible.  Add nan2008.
1159         (MULTILIB_DIRNAMES): Add nan2008.
1160         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1161         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1162         and micromips incompatible.  Add nan2008.
1163         (MULTILIB_DIRNAMES): Add nan2008.
1164         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1166 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
1168         PR rtl-optimization/58079
1169         * combine.c (combine_simplify_rtx): Avoid using SUBST if
1170         simplify_comparison has widened a comparison with an integer.
1172 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1174         * config/arm/neon.md (movmisalign<mode>): Disable when we
1175         don't allow unaligned accesses.
1176         (*movmisalign<mode>_neon_store): Likewise.
1177         (*movmisalign<mode>_neon_load): Likewise.
1178         (*movmisalign<mode>_neon_store): Likewise.
1179         (*movmisalign<mode>_neon_load): Likewise.
1181 2013-08-08  Jan Hubicka  <jh@suse.cz>
1183         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1184         (make_pass_rebuild_cgraph_edges): Also clear references.
1185         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1186         * ipa-inline-transform.c (inline_transform): Remove all references
1187         after inlining.
1188         * cgraphunit.c (expand_function): Remove all references after
1189         expansion.
1190         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1191         (ipa_find_reference): Rewrite to iterator.
1192         (remove_stmt_references): Likewise.
1193         (ipa_clear_stmts_in_references): New function.
1194         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
1195         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
1196         clear references.
1197         * ipa-split.c (split_function): Remove references in split function.
1199 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1201         PR target/57431
1202         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
1203         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
1205 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1207         PR target/56979
1208         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
1209         suggested mode for the assignment isn't compatible with the
1210         registers required.
1212 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1214         PR target/58065
1215         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
1217 2013-08-07  Xinliang David Li  <davidxl@google.com>
1219         * config/i386/i386.opt: New option -mtune-ctrl=.
1220         * config/i386/x86-tune.def: New file.
1221         * config/i386/i386.h: include x86-tune.def.
1222         * config/i386/i386.c (ix86_option_override_internal):
1223         Parsing -mtune-ctrl= option and set tune features.
1225 2013-08-07  Oleg Endo <olegendo@gcc.gnu.org>
1227         PR other/12081
1228         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
1229         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
1230         to gen_2arg_fn_t.
1232 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1234         * rtl.h (update_alignments): Declare.
1235         * final.c (grow_label_align): New function extracted from...
1236         (shorten_branches): ...here.  Call it.
1237         (update_alignments): New function.
1238         * reorg.c (sibling_labels): New variable.
1239         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
1240         the new label along with it onto the sibling_labels vector.
1241         (fill_simple_delay_slots): Adjust call to get_label_before.
1242         (fill_slots_from_thread): Likewise.
1243         (relax_delay_slots): Likewise.
1244         (make_return_insns): Likewise.
1245         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
1247 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1249         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
1250         document its semantics.
1251         (diagnostic_report_diagnostic): Adjust accordingly.
1253 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1255         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
1256         (sparc_option_override): ...and port to new C++ pass API.
1257         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
1259 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
1261         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
1263 2013-08-06  Caroline Tice  <cmtice@google.com>
1265         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
1266         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
1267         * tree-pass.h: Add pass_vtable_verify.
1268         * varasm.c (assemble_variable): Add code to properly set the comdat
1269         section and name for the .vtable_map_vars section.
1270         (assemble_vtyv_preinit_initializer): New function.
1271         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
1272         LINK_ONCE flag.
1273         * output.h: Add function decl for assemble_vtv_preinit_initializer.
1274         * vtable-verify.c: New file.
1275         * vtable-verify.h: New file.
1276         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
1277         initialiation levels.
1278         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
1279         * passes.def: Insert pass_vtable_verify.
1280         * aclocal.m4: Reorder includes.
1281         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
1282         -fvtv-counts options.
1283         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
1284         as appropriate, if -fvtable-verify=... is used.
1285         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
1286         -fvtable-verify=... is used.
1287         * Makefile.in (OBJS):  Add vtable-verify.o to list.
1288         (vtable-verify.o): Add new build rule.
1289         (GTFILES): Add vtable-verify.c to list.
1290         * common.opt (fvtable-verify=): New flag.
1291         (vtv_priority): Values for fvtable-verify= flag.
1292         (fvtv-counts): New flag.
1293         (fvtv-debug): New flag.
1294         * tree.h (save_vtable_map_decl): New extern function decl.
1296 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1298         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
1299         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
1300         (pass_data_rl78_devirt): ...new pass_data instance and...
1301         (make_pass_rl78_devirt): ...new function.
1302         (rl78_asm_file_start): Port pass registration to new C++ API.
1304 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1306         * coretypes.h (rtl_opt_pass): Add.
1307         (gcc::context): Add.
1308         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
1309         (epiphany_init): Port to new C++ pass API.
1310         (epiphany_optimize_mode_switching): Likewise.
1311         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
1312         (pass_manager::get_pass_mode_switching): New.
1313         (pass_manager::get_pass_peephole2): New.
1314         * mode-switching.c (pass_mode_switching): Add clone method.
1315         * recog.c (pass_peephole2): Add clone method.
1316         (pass_split_all_insns): Add clone method.
1318 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
1320         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
1321         (mips_option_override): ...here, porting to new C++ API for passes.
1323 2013-08-06  Jan Hubicka  <jh@suse.cz>
1325         * cgraph.c (cgraph_get_body): New function based on lto.c
1326         implementation.
1327         * cgraph.h (cgraph_get_body): Declare.
1328         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
1329         LTO paths.
1330         * cgraphunit.c (expand_function): Get body prior expanding.
1331         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
1332         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
1333         really need.
1334         * passes.c (do_per_function_toporder): Get body.
1335         * tree-inline.c (expand_call_inline): Get body prior inlining it.
1336         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
1338 2013-08-06  Martin Jambor  <mjambor@suse.cz>
1340         PR fortran/57987
1341         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
1342         is not re-finalized.  Rename second parameter to no_collect.
1344 2013-08-06  Martin Jambor  <mjambor@suse.cz>
1346         PR middle-end/58041
1347         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
1348         MEM_REF has proper alignment information.
1350 2013-08-05  Oleg Endo <olegendo@gcc.gnu.org>
1352         PR other/12081
1353         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
1354         class insn_gen_fn.
1355         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
1356         rtx (*) (rtx, ...) with insn_gen_fn.
1357         * genoutput.c (output_insn_data): Cast gen_? function pointers to
1358         insn_gen_fn::stored_funcptr.  Add initializer braces.
1360 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1362         Rewrite how instances of passes are cloned to remove assumptions
1363         about their sizes (thus allowing pass subclasses to have
1364         additional data fields, albeit non-GC-managed ones at this point).
1366         * passes.c (make_pass_instance): Now that passes have clone
1367         methods, rewrite this function to eliminate XNEW and memcpy
1368         calls that used hardcoded sizes.  Since this function no longer
1369         creates pass instances, rename it to...
1370         (add_pass_instance): ...this.  Document the old way that passes were
1371         numbered and flagged, and rework this function to continue using it.
1372         (next_pass_1): Add an initial_pass argument for use by
1373         add_pass_instance.
1374         (position_pass): When adding multiple instances of a pass, use
1375         the pass's clone method, rather than relying on the XNEW/memcpy
1376         within the former make_pass_instance (now add_pass_instance).
1377         (pass_manager::pass_manager): When invoking next_pass_1, also supply
1378         the initial instance of the current pass within the pass manager.
1380 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1382         This is the automated part of the conversion of passes from C
1383         structs to C++ classes.
1385         Patch autogenerated by refactor_passes.py from
1386         https://github.com/davidmalcolm/gcc-refactoring-scripts
1387         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
1389         * asan.c (pass_asan): Convert from a global struct to a subclass of
1390         gimple_opt_pass along with...
1391         (pass_data_asan): ...new pass_data instance and...
1392         (make_pass_asan): ...new function.
1393         (pass_asan_O0): Convert from a global struct to a subclass of
1394         gimple_opt_pass along with...
1395         (pass_data_asan_O0): ...new pass_data instance and...
1396         (make_pass_asan_O0): ...new function.
1397         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
1398         subclass of rtl_opt_pass along with...
1399         (pass_data_inc_dec): ...new pass_data instance and...
1400         (make_pass_inc_dec): ...new function.
1401         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
1402         a subclass of rtl_opt_pass along with...
1403         (pass_data_reorder_blocks): ...new pass_data instance and...
1404         (make_pass_reorder_blocks): ...new function.
1405         (pass_duplicate_computed_gotos): Convert from a global struct to a
1406         subclass of rtl_opt_pass along with...
1407         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
1408         (make_pass_duplicate_computed_gotos): ...new function.
1409         (pass_partition_blocks): Convert from a global struct to a subclass of
1410         rtl_opt_pass along with...
1411         (pass_data_partition_blocks): ...new pass_data instance and...
1412         (make_pass_partition_blocks): ...new function.
1413         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
1414         struct to a subclass of rtl_opt_pass along with...
1415         (pass_data_branch_target_load_optimize1): ...new pass_data instance
1416         and...
1417         (make_pass_branch_target_load_optimize1): ...new function.
1418         (pass_branch_target_load_optimize2): Convert from a global struct to a
1419         subclass of rtl_opt_pass along with...
1420         (pass_data_branch_target_load_optimize2): ...new pass_data instance
1421         and...
1422         (make_pass_branch_target_load_optimize2): ...new function.
1423         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
1424         of rtl_opt_pass along with...
1425         (pass_data_jump): ...new pass_data instance and...
1426         (make_pass_jump): ...new function.
1427         (pass_jump2): Convert from a global struct to a subclass of
1428         rtl_opt_pass along with...
1429         (pass_data_jump2): ...new pass_data instance and...
1430         (make_pass_jump2): ...new function.
1431         * cfgexpand.c (pass_expand): Convert from a global struct to a
1432         subclass of rtl_opt_pass along with...
1433         (pass_data_expand): ...new pass_data instance and...
1434         (make_pass_expand): ...new function.
1435         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
1436         of rtl_opt_pass along with...
1437         (pass_data_free_cfg): ...new pass_data instance and...
1438         (make_pass_free_cfg): ...new function.
1439         (pass_into_cfg_layout_mode): Convert from a global struct to a
1440         subclass of rtl_opt_pass along with...
1441         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
1442         (make_pass_into_cfg_layout_mode): ...new function.
1443         (pass_outof_cfg_layout_mode): Convert from a global struct to a
1444         subclass of rtl_opt_pass along with...
1445         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
1446         (make_pass_outof_cfg_layout_mode): ...new function.
1447         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
1448         struct to a subclass of gimple_opt_pass along with...
1449         (pass_data_build_cgraph_edges): ...new pass_data instance and...
1450         (make_pass_build_cgraph_edges): ...new function.
1451         (pass_rebuild_cgraph_edges): Convert from a global struct to a
1452         subclass of gimple_opt_pass along with...
1453         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
1454         (make_pass_rebuild_cgraph_edges): ...new function.
1455         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
1456         subclass of gimple_opt_pass along with...
1457         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
1458         and...
1459         (make_pass_remove_cgraph_callee_edges): ...new function.
1460         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
1461         struct to a subclass of rtl_opt_pass along with...
1462         (pass_data_stack_adjustments): ...new pass_data instance and...
1463         (make_pass_stack_adjustments): ...new function.
1464         * combine.c (pass_combine): Convert from a global struct to a subclass
1465         of rtl_opt_pass along with...
1466         (pass_data_combine): ...new pass_data instance and...
1467         (make_pass_combine): ...new function.
1468         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
1469         global struct to a subclass of rtl_opt_pass along with...
1470         (pass_data_compare_elim_after_reload): ...new pass_data instance
1471         and...
1472         (make_pass_compare_elim_after_reload): ...new function.
1473         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
1474         of rtl_opt_pass along with...
1475         (pass_data_rtl_cprop): ...new pass_data instance and...
1476         (make_pass_rtl_cprop): ...new function.
1477         * cse.c (pass_cse): Convert from a global struct to a subclass of
1478         rtl_opt_pass along with...
1479         (pass_data_cse): ...new pass_data instance and...
1480         (make_pass_cse): ...new function.
1481         (pass_cse2): Convert from a global struct to a subclass of
1482         rtl_opt_pass along with...
1483         (pass_data_cse2): ...new pass_data instance and...
1484         (make_pass_cse2): ...new function.
1485         (pass_cse_after_global_opts): Convert from a global struct to a
1486         subclass of rtl_opt_pass along with...
1487         (pass_data_cse_after_global_opts): ...new pass_data instance and...
1488         (make_pass_cse_after_global_opts): ...new function.
1489         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
1490         of rtl_opt_pass along with...
1491         (pass_data_ud_rtl_dce): ...new pass_data instance and...
1492         (make_pass_ud_rtl_dce): ...new function.
1493         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
1494         rtl_opt_pass along with...
1495         (pass_data_fast_rtl_dce): ...new pass_data instance and...
1496         (make_pass_fast_rtl_dce): ...new function.
1497         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
1498         a subclass of rtl_opt_pass along with...
1499         (pass_data_df_initialize_opt): ...new pass_data instance and...
1500         (make_pass_df_initialize_opt): ...new function.
1501         (pass_df_initialize_no_opt): Convert from a global struct to a
1502         subclass of rtl_opt_pass along with...
1503         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
1504         (make_pass_df_initialize_no_opt): ...new function.
1505         (pass_df_finish): Convert from a global struct to a subclass of
1506         rtl_opt_pass along with...
1507         (pass_data_df_finish): ...new pass_data instance and...
1508         (make_pass_df_finish): ...new function.
1509         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
1510         rtl_opt_pass along with...
1511         (pass_data_rtl_dse1): ...new pass_data instance and...
1512         (make_pass_rtl_dse1): ...new function.
1513         (pass_rtl_dse2): Convert from a global struct to a subclass of
1514         rtl_opt_pass along with...
1515         (pass_data_rtl_dse2): ...new pass_data instance and...
1516         (make_pass_rtl_dse2): ...new function.
1517         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
1518         subclass of rtl_opt_pass along with...
1519         (pass_data_dwarf2_frame): ...new pass_data instance and...
1520         (make_pass_dwarf2_frame): ...new function.
1521         * except.c (pass_set_nothrow_function_flags): Convert from a global
1522         struct to a subclass of rtl_opt_pass along with...
1523         (pass_data_set_nothrow_function_flags): ...new pass_data instance
1524         and...
1525         (make_pass_set_nothrow_function_flags): ...new function.
1526         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
1527         subclass of rtl_opt_pass along with...
1528         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
1529         and...
1530         (make_pass_convert_to_eh_region_ranges): ...new function.
1531         * final.c (pass_compute_alignments): Convert from a global struct to a
1532         subclass of rtl_opt_pass along with...
1533         (pass_data_compute_alignments): ...new pass_data instance and...
1534         (make_pass_compute_alignments): ...new function.
1535         (pass_final): Convert from a global struct to a subclass of
1536         rtl_opt_pass along with...
1537         (pass_data_final): ...new pass_data instance and...
1538         (make_pass_final): ...new function.
1539         (pass_shorten_branches): Convert from a global struct to a subclass of
1540         rtl_opt_pass along with...
1541         (pass_data_shorten_branches): ...new pass_data instance and...
1542         (make_pass_shorten_branches): ...new function.
1543         (pass_clean_state): Convert from a global struct to a subclass of
1544         rtl_opt_pass along with...
1545         (pass_data_clean_state): ...new pass_data instance and...
1546         (make_pass_clean_state): ...new function.
1547         * function.c (pass_instantiate_virtual_regs): Convert from a global
1548         struct to a subclass of rtl_opt_pass along with...
1549         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
1550         (make_pass_instantiate_virtual_regs): ...new function.
1551         (pass_leaf_regs): Convert from a global struct to a subclass of
1552         rtl_opt_pass along with...
1553         (pass_data_leaf_regs): ...new pass_data instance and...
1554         (make_pass_leaf_regs): ...new function.
1555         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
1556         subclass of rtl_opt_pass along with...
1557         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
1558         and...
1559         (make_pass_thread_prologue_and_epilogue): ...new function.
1560         (pass_match_asm_constraints): Convert from a global struct to a
1561         subclass of rtl_opt_pass along with...
1562         (pass_data_match_asm_constraints): ...new pass_data instance and...
1563         (make_pass_match_asm_constraints): ...new function.
1564         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
1565         subclass of rtl_opt_pass along with...
1566         (pass_data_rtl_fwprop): ...new pass_data instance and...
1567         (make_pass_rtl_fwprop): ...new function.
1568         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
1569         rtl_opt_pass along with...
1570         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
1571         (make_pass_rtl_fwprop_addr): ...new function.
1572         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
1573         rtl_opt_pass along with...
1574         (pass_data_rtl_pre): ...new pass_data instance and...
1575         (make_pass_rtl_pre): ...new function.
1576         (pass_rtl_hoist): Convert from a global struct to a subclass of
1577         rtl_opt_pass along with...
1578         (pass_data_rtl_hoist): ...new pass_data instance and...
1579         (make_pass_rtl_hoist): ...new function.
1580         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
1581         subclass of gimple_opt_pass along with...
1582         (pass_data_lower_cf): ...new pass_data instance and...
1583         (make_pass_lower_cf): ...new function.
1584         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
1585         from a global struct to a subclass of gimple_opt_pass along with...
1586         (pass_data_strength_reduction): ...new pass_data instance and...
1587         (make_pass_strength_reduction): ...new function.
1588         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
1589         of rtl_opt_pass along with...
1590         (pass_data_rtl_ifcvt): ...new pass_data instance and...
1591         (make_pass_rtl_ifcvt): ...new function.
1592         (pass_if_after_combine): Convert from a global struct to a subclass of
1593         rtl_opt_pass along with...
1594         (pass_data_if_after_combine): ...new pass_data instance and...
1595         (make_pass_if_after_combine): ...new function.
1596         (pass_if_after_reload): Convert from a global struct to a subclass of
1597         rtl_opt_pass along with...
1598         (pass_data_if_after_reload): ...new pass_data instance and...
1599         (make_pass_if_after_reload): ...new function.
1600         * init-regs.c (pass_initialize_regs): Convert from a global struct to
1601         a subclass of rtl_opt_pass along with...
1602         (pass_data_initialize_regs): ...new pass_data instance and...
1603         (make_pass_initialize_regs): ...new function.
1604         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
1605         of ipa_opt_pass_d along with...
1606         (pass_data_ipa_cp): ...new pass_data instance and...
1607         (make_pass_ipa_cp): ...new function.
1608         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
1609         global struct to a subclass of gimple_opt_pass along with...
1610         (pass_data_inline_parameters): ...new pass_data instance and...
1611         (make_pass_inline_parameters): ...new function.
1612         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
1613         subclass of gimple_opt_pass along with...
1614         (pass_data_early_inline): ...new pass_data instance and...
1615         (make_pass_early_inline): ...new function.
1616         (pass_ipa_inline): Convert from a global struct to a subclass of
1617         ipa_opt_pass_d along with...
1618         (pass_data_ipa_inline): ...new pass_data instance and...
1619         (make_pass_ipa_inline): ...new function.
1620         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
1621         struct to a subclass of gimple_opt_pass along with...
1622         (pass_data_local_pure_const): ...new pass_data instance and...
1623         (make_pass_local_pure_const): ...new function.
1624         (pass_ipa_pure_const): Convert from a global struct to a subclass of
1625         ipa_opt_pass_d along with...
1626         (pass_data_ipa_pure_const): ...new pass_data instance and...
1627         (make_pass_ipa_pure_const): ...new function.
1628         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
1629         to a subclass of ipa_opt_pass_d along with...
1630         (pass_data_ipa_reference): ...new pass_data instance and...
1631         (make_pass_ipa_reference): ...new function.
1632         * ipa-split.c (pass_split_functions): Convert from a global struct to
1633         a subclass of gimple_opt_pass along with...
1634         (pass_data_split_functions): ...new pass_data instance and...
1635         (make_pass_split_functions): ...new function.
1636         (pass_feedback_split_functions): Convert from a global struct to a
1637         subclass of gimple_opt_pass along with...
1638         (pass_data_feedback_split_functions): ...new pass_data instance and...
1639         (make_pass_feedback_split_functions): ...new function.
1640         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
1641         global struct to a subclass of simple_ipa_opt_pass along with...
1642         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
1643         instance and...
1644         (make_pass_ipa_function_and_variable_visibility): ...new function.
1645         (pass_ipa_free_inline_summary): Convert from a global struct to a
1646         subclass of simple_ipa_opt_pass along with...
1647         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
1648         (make_pass_ipa_free_inline_summary): ...new function.
1649         (pass_ipa_whole_program_visibility): Convert from a global struct to a
1650         subclass of ipa_opt_pass_d along with...
1651         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
1652         and...
1653         (make_pass_ipa_whole_program_visibility): ...new function.
1654         (pass_ipa_profile): Convert from a global struct to a subclass of
1655         ipa_opt_pass_d along with...
1656         (pass_data_ipa_profile): ...new pass_data instance and...
1657         (make_pass_ipa_profile): ...new function.
1658         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
1659         ipa_opt_pass_d along with...
1660         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
1661         (make_pass_ipa_cdtor_merge): ...new function.
1662         * ira.c (pass_ira): Convert from a global struct to a subclass of
1663         rtl_opt_pass along with...
1664         (pass_data_ira): ...new pass_data instance and...
1665         (make_pass_ira): ...new function.
1666         (pass_reload): Convert from a global struct to a subclass of
1667         rtl_opt_pass along with...
1668         (pass_data_reload): ...new pass_data instance and...
1669         (make_pass_reload): ...new function.
1670         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
1671         subclass of rtl_opt_pass along with...
1672         (pass_data_cleanup_barriers): ...new pass_data instance and...
1673         (make_pass_cleanup_barriers): ...new function.
1674         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
1675         of rtl_opt_pass along with...
1676         (pass_data_loop2): ...new pass_data instance and...
1677         (make_pass_loop2): ...new function.
1678         (pass_rtl_loop_init): Convert from a global struct to a subclass of
1679         rtl_opt_pass along with...
1680         (pass_data_rtl_loop_init): ...new pass_data instance and...
1681         (make_pass_rtl_loop_init): ...new function.
1682         (pass_rtl_loop_done): Convert from a global struct to a subclass of
1683         rtl_opt_pass along with...
1684         (pass_data_rtl_loop_done): ...new pass_data instance and...
1685         (make_pass_rtl_loop_done): ...new function.
1686         (pass_rtl_move_loop_invariants): Convert from a global struct to a
1687         subclass of rtl_opt_pass along with...
1688         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
1689         (make_pass_rtl_move_loop_invariants): ...new function.
1690         (pass_rtl_unswitch): Convert from a global struct to a subclass of
1691         rtl_opt_pass along with...
1692         (pass_data_rtl_unswitch): ...new pass_data instance and...
1693         (make_pass_rtl_unswitch): ...new function.
1694         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
1695         subclass of rtl_opt_pass along with...
1696         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
1697         and...
1698         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1699         (pass_rtl_doloop): Convert from a global struct to a subclass of
1700         rtl_opt_pass along with...
1701         (pass_data_rtl_doloop): ...new pass_data instance and...
1702         (make_pass_rtl_doloop): ...new function.
1703         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
1704         a subclass of rtl_opt_pass along with...
1705         (pass_data_lower_subreg): ...new pass_data instance and...
1706         (make_pass_lower_subreg): ...new function.
1707         (pass_lower_subreg2): Convert from a global struct to a subclass of
1708         rtl_opt_pass along with...
1709         (pass_data_lower_subreg2): ...new pass_data instance and...
1710         (make_pass_lower_subreg2): ...new function.
1711         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
1712         struct to a subclass of ipa_opt_pass_d along with...
1713         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
1714         (make_pass_ipa_lto_gimple_out): ...new function.
1715         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
1716         of ipa_opt_pass_d along with...
1717         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
1718         (make_pass_ipa_lto_finish_out): ...new function.
1719         * mode-switching.c (pass_mode_switching): Convert from a global struct
1720         to a subclass of rtl_opt_pass along with...
1721         (pass_data_mode_switching): ...new pass_data instance and...
1722         (make_pass_mode_switching): ...new function.
1723         * modulo-sched.c (pass_sms): Convert from a global struct to a
1724         subclass of rtl_opt_pass along with...
1725         (pass_data_sms): ...new pass_data instance and...
1726         (make_pass_sms): ...new function.
1727         * omp-low.c (pass_expand_omp): Convert from a global struct to a
1728         subclass of gimple_opt_pass along with...
1729         (pass_data_expand_omp): ...new pass_data instance and...
1730         (make_pass_expand_omp): ...new function.
1731         (pass_lower_omp): Convert from a global struct to a subclass of
1732         gimple_opt_pass along with...
1733         (pass_data_lower_omp): ...new pass_data instance and...
1734         (make_pass_lower_omp): ...new function.
1735         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
1736         of gimple_opt_pass along with...
1737         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
1738         (make_pass_diagnose_omp_blocks): ...new function.
1739         * passes.c (pass_early_local_passes): Convert from a global struct to
1740         a subclass of simple_ipa_opt_pass along with...
1741         (pass_data_early_local_passes): ...new pass_data instance and...
1742         (make_pass_early_local_passes): ...new function.
1743         (pass_all_early_optimizations): Convert from a global struct to a
1744         subclass of gimple_opt_pass along with...
1745         (pass_data_all_early_optimizations): ...new pass_data instance and...
1746         (make_pass_all_early_optimizations): ...new function.
1747         (pass_all_optimizations): Convert from a global struct to a subclass
1748         of gimple_opt_pass along with...
1749         (pass_data_all_optimizations): ...new pass_data instance and...
1750         (make_pass_all_optimizations): ...new function.
1751         (pass_all_optimizations_g): Convert from a global struct to a subclass
1752         of gimple_opt_pass along with...
1753         (pass_data_all_optimizations_g): ...new pass_data instance and...
1754         (make_pass_all_optimizations_g): ...new function.
1755         (pass_rest_of_compilation): Convert from a global struct to a subclass
1756         of rtl_opt_pass along with...
1757         (pass_data_rest_of_compilation): ...new pass_data instance and...
1758         (make_pass_rest_of_compilation): ...new function.
1759         (pass_postreload): Convert from a global struct to a subclass of
1760         rtl_opt_pass along with...
1761         (pass_data_postreload): ...new pass_data instance and...
1762         (make_pass_postreload): ...new function.
1763         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
1764         subclass of rtl_opt_pass along with...
1765         (pass_data_gcse2): ...new pass_data instance and...
1766         (make_pass_gcse2): ...new function.
1767         * postreload.c (pass_postreload_cse): Convert from a global struct to
1768         a subclass of rtl_opt_pass along with...
1769         (pass_data_postreload_cse): ...new pass_data instance and...
1770         (make_pass_postreload_cse): ...new function.
1771         * predict.c (pass_profile): Convert from a global struct to a subclass
1772         of gimple_opt_pass along with...
1773         (pass_data_profile): ...new pass_data instance and...
1774         (make_pass_profile): ...new function.
1775         (pass_strip_predict_hints): Convert from a global struct to a subclass
1776         of gimple_opt_pass along with...
1777         (pass_data_strip_predict_hints): ...new pass_data instance and...
1778         (make_pass_strip_predict_hints): ...new function.
1779         * recog.c (pass_peephole2): Convert from a global struct to a subclass
1780         of rtl_opt_pass along with...
1781         (pass_data_peephole2): ...new pass_data instance and...
1782         (make_pass_peephole2): ...new function.
1783         (pass_split_all_insns): Convert from a global struct to a subclass of
1784         rtl_opt_pass along with...
1785         (pass_data_split_all_insns): ...new pass_data instance and...
1786         (make_pass_split_all_insns): ...new function.
1787         (pass_split_after_reload): Convert from a global struct to a subclass
1788         of rtl_opt_pass along with...
1789         (pass_data_split_after_reload): ...new pass_data instance and...
1790         (make_pass_split_after_reload): ...new function.
1791         (pass_split_before_regstack): Convert from a global struct to a
1792         subclass of rtl_opt_pass along with...
1793         (pass_data_split_before_regstack): ...new pass_data instance and...
1794         (make_pass_split_before_regstack): ...new function.
1795         (pass_split_before_sched2): Convert from a global struct to a subclass
1796         of rtl_opt_pass along with...
1797         (pass_data_split_before_sched2): ...new pass_data instance and...
1798         (make_pass_split_before_sched2): ...new function.
1799         (pass_split_for_shorten_branches): Convert from a global struct to a
1800         subclass of rtl_opt_pass along with...
1801         (pass_data_split_for_shorten_branches): ...new pass_data instance
1802         and...
1803         (make_pass_split_for_shorten_branches): ...new function.
1804         * ree.c (pass_ree): Convert from a global struct to a subclass of
1805         rtl_opt_pass along with...
1806         (pass_data_ree): ...new pass_data instance and...
1807         (make_pass_ree): ...new function.
1808         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
1809         subclass of rtl_opt_pass along with...
1810         (pass_data_stack_regs): ...new pass_data instance and...
1811         (make_pass_stack_regs): ...new function.
1812         (pass_stack_regs_run): Convert from a global struct to a subclass of
1813         rtl_opt_pass along with...
1814         (pass_data_stack_regs_run): ...new pass_data instance and...
1815         (make_pass_stack_regs_run): ...new function.
1816         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
1817         subclass of rtl_opt_pass along with...
1818         (pass_data_cprop_hardreg): ...new pass_data instance and...
1819         (make_pass_cprop_hardreg): ...new function.
1820         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
1821         subclass of rtl_opt_pass along with...
1822         (pass_data_reginfo_init): ...new pass_data instance and...
1823         (make_pass_reginfo_init): ...new function.
1824         * regmove.c (pass_regmove): Convert from a global struct to a subclass
1825         of rtl_opt_pass along with...
1826         (pass_data_regmove): ...new pass_data instance and...
1827         (make_pass_regmove): ...new function.
1828         * regrename.c (pass_regrename): Convert from a global struct to a
1829         subclass of rtl_opt_pass along with...
1830         (pass_data_regrename): ...new pass_data instance and...
1831         (make_pass_regrename): ...new function.
1832         * reorg.c (pass_delay_slots): Convert from a global struct to a
1833         subclass of rtl_opt_pass along with...
1834         (pass_data_delay_slots): ...new pass_data instance and...
1835         (make_pass_delay_slots): ...new function.
1836         (pass_machine_reorg): Convert from a global struct to a subclass of
1837         rtl_opt_pass along with...
1838         (pass_data_machine_reorg): ...new pass_data instance and...
1839         (make_pass_machine_reorg): ...new function.
1840         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
1841         of rtl_opt_pass along with...
1842         (pass_data_sched): ...new pass_data instance and...
1843         (make_pass_sched): ...new function.
1844         (pass_sched2): Convert from a global struct to a subclass of
1845         rtl_opt_pass along with...
1846         (pass_data_sched2): ...new pass_data instance and...
1847         (make_pass_sched2): ...new function.
1848         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
1849         to a subclass of rtl_opt_pass along with...
1850         (pass_data_stack_ptr_mod): ...new pass_data instance and...
1851         (make_pass_stack_ptr_mod): ...new function.
1852         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
1853         to a subclass of rtl_opt_pass along with...
1854         (pass_data_rtl_store_motion): ...new pass_data instance and...
1855         (make_pass_rtl_store_motion): ...new function.
1856         * tracer.c (pass_tracer): Convert from a global struct to a subclass
1857         of gimple_opt_pass along with...
1858         (pass_data_tracer): ...new pass_data instance and...
1859         (make_pass_tracer): ...new function.
1860         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
1861         to a subclass of gimple_opt_pass along with...
1862         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
1863         (make_pass_diagnose_tm_blocks): ...new function.
1864         (pass_lower_tm): Convert from a global struct to a subclass of
1865         gimple_opt_pass along with...
1866         (pass_data_lower_tm): ...new pass_data instance and...
1867         (make_pass_lower_tm): ...new function.
1868         (pass_tm_init): Convert from a global struct to a subclass of
1869         gimple_opt_pass along with...
1870         (pass_data_tm_init): ...new pass_data instance and...
1871         (make_pass_tm_init): ...new function.
1872         (pass_tm_mark): Convert from a global struct to a subclass of
1873         gimple_opt_pass along with...
1874         (pass_data_tm_mark): ...new pass_data instance and...
1875         (make_pass_tm_mark): ...new function.
1876         (pass_tm_edges): Convert from a global struct to a subclass of
1877         gimple_opt_pass along with...
1878         (pass_data_tm_edges): ...new pass_data instance and...
1879         (make_pass_tm_edges): ...new function.
1880         (pass_tm_memopt): Convert from a global struct to a subclass of
1881         gimple_opt_pass along with...
1882         (pass_data_tm_memopt): ...new pass_data instance and...
1883         (make_pass_tm_memopt): ...new function.
1884         (pass_ipa_tm): Convert from a global struct to a subclass of
1885         simple_ipa_opt_pass along with...
1886         (pass_data_ipa_tm): ...new pass_data instance and...
1887         (make_pass_ipa_tm): ...new function.
1888         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
1889         subclass of gimple_opt_pass along with...
1890         (pass_data_call_cdce): ...new pass_data instance and...
1891         (make_pass_call_cdce): ...new function.
1892         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
1893         subclass of gimple_opt_pass along with...
1894         (pass_data_build_cfg): ...new pass_data instance and...
1895         (make_pass_build_cfg): ...new function.
1896         (pass_split_crit_edges): Convert from a global struct to a subclass of
1897         gimple_opt_pass along with...
1898         (pass_data_split_crit_edges): ...new pass_data instance and...
1899         (make_pass_split_crit_edges): ...new function.
1900         (pass_warn_function_return): Convert from a global struct to a
1901         subclass of gimple_opt_pass along with...
1902         (pass_data_warn_function_return): ...new pass_data instance and...
1903         (make_pass_warn_function_return): ...new function.
1904         (pass_warn_function_noreturn): Convert from a global struct to a
1905         subclass of gimple_opt_pass along with...
1906         (pass_data_warn_function_noreturn): ...new pass_data instance and...
1907         (make_pass_warn_function_noreturn): ...new function.
1908         (pass_warn_unused_result): Convert from a global struct to a subclass
1909         of gimple_opt_pass along with...
1910         (pass_data_warn_unused_result): ...new pass_data instance and...
1911         (make_pass_warn_unused_result): ...new function.
1912         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
1913         a subclass of gimple_opt_pass along with...
1914         (pass_data_merge_phi): ...new pass_data instance and...
1915         (make_pass_merge_phi): ...new function.
1916         * tree-complex.c (pass_lower_complex): Convert from a global struct to
1917         a subclass of gimple_opt_pass along with...
1918         (pass_data_lower_complex): ...new pass_data instance and...
1919         (make_pass_lower_complex): ...new function.
1920         (pass_lower_complex_O0): Convert from a global struct to a subclass of
1921         gimple_opt_pass along with...
1922         (pass_data_lower_complex_O0): ...new pass_data instance and...
1923         (make_pass_lower_complex_O0): ...new function.
1924         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
1925         subclass of gimple_opt_pass along with...
1926         (pass_data_lower_eh): ...new pass_data instance and...
1927         (make_pass_lower_eh): ...new function.
1928         (pass_refactor_eh): Convert from a global struct to a subclass of
1929         gimple_opt_pass along with...
1930         (pass_data_refactor_eh): ...new pass_data instance and...
1931         (make_pass_refactor_eh): ...new function.
1932         (pass_lower_resx): Convert from a global struct to a subclass of
1933         gimple_opt_pass along with...
1934         (pass_data_lower_resx): ...new pass_data instance and...
1935         (make_pass_lower_resx): ...new function.
1936         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1937         of gimple_opt_pass along with...
1938         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1939         (make_pass_lower_eh_dispatch): ...new function.
1940         (pass_cleanup_eh): Convert from a global struct to a subclass of
1941         gimple_opt_pass along with...
1942         (pass_data_cleanup_eh): ...new pass_data instance and...
1943         (make_pass_cleanup_eh): ...new function.
1944         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1945         to a subclass of simple_ipa_opt_pass along with...
1946         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1947         (make_pass_ipa_lower_emutls): ...new function.
1948         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1949         a subclass of gimple_opt_pass along with...
1950         (pass_data_if_conversion): ...new pass_data instance and...
1951         (make_pass_if_conversion): ...new function.
1952         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1953         subclass of gimple_opt_pass along with...
1954         (pass_data_build_ssa): ...new pass_data instance and...
1955         (make_pass_build_ssa): ...new function.
1956         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1957         global struct to a subclass of gimple_opt_pass along with...
1958         (pass_data_loop_distribution): ...new pass_data instance and...
1959         (make_pass_loop_distribution): ...new function.
1960         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1961         subclass of gimple_opt_pass along with...
1962         (pass_data_mudflap_1): ...new pass_data instance and...
1963         (make_pass_mudflap_1): ...new function.
1964         (pass_mudflap_2): Convert from a global struct to a subclass of
1965         gimple_opt_pass along with...
1966         (pass_data_mudflap_2): ...new pass_data instance and...
1967         (make_pass_mudflap_2): ...new function.
1968         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1969         subclass of gimple_opt_pass along with...
1970         (pass_data_mudflap_1): ...new pass_data instance and...
1971         (make_pass_mudflap_1): ...new function.
1972         (pass_mudflap_2): Convert from a global struct to a subclass of
1973         gimple_opt_pass along with...
1974         (pass_data_mudflap_2): ...new pass_data instance and...
1975         (make_pass_mudflap_2): ...new function.
1976         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1977         gimple_opt_pass along with...
1978         (pass_data_nrv): ...new pass_data instance and...
1979         (make_pass_nrv): ...new function.
1980         (pass_return_slot): Convert from a global struct to a subclass of
1981         gimple_opt_pass along with...
1982         (pass_data_return_slot): ...new pass_data instance and...
1983         (make_pass_return_slot): ...new function.
1984         * tree-object-size.c (pass_object_sizes): Convert from a global struct
1985         to a subclass of gimple_opt_pass along with...
1986         (pass_data_object_sizes): ...new pass_data instance and...
1987         (make_pass_object_sizes): ...new function.
1988         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1989         global struct to a subclass of gimple_opt_pass along with...
1990         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1991         and...
1992         (make_pass_cleanup_cfg_post_optimizing): ...new function.
1993         (pass_fixup_cfg): Convert from a global struct to a subclass of
1994         gimple_opt_pass along with...
1995         (pass_data_fixup_cfg): ...new pass_data instance and...
1996         (make_pass_fixup_cfg): ...new function.
1997         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1998         (make_pass_mudflap_1): ...new function.
1999         (pass_mudflap_2): Replace declaration with that of...
2000         (make_pass_mudflap_2): ...new function.
2001         (pass_asan): Replace declaration with that of...
2002         (make_pass_asan): ...new function.
2003         (pass_asan_O0): Replace declaration with that of...
2004         (make_pass_asan_O0): ...new function.
2005         (pass_tsan): Replace declaration with that of...
2006         (make_pass_tsan): ...new function.
2007         (pass_tsan_O0): Replace declaration with that of...
2008         (make_pass_tsan_O0): ...new function.
2009         (pass_lower_cf): Replace declaration with that of...
2010         (make_pass_lower_cf): ...new function.
2011         (pass_refactor_eh): Replace declaration with that of...
2012         (make_pass_refactor_eh): ...new function.
2013         (pass_lower_eh): Replace declaration with that of...
2014         (make_pass_lower_eh): ...new function.
2015         (pass_lower_eh_dispatch): Replace declaration with that of...
2016         (make_pass_lower_eh_dispatch): ...new function.
2017         (pass_lower_resx): Replace declaration with that of...
2018         (make_pass_lower_resx): ...new function.
2019         (pass_build_cfg): Replace declaration with that of...
2020         (make_pass_build_cfg): ...new function.
2021         (pass_early_tree_profile): Replace declaration with that of...
2022         (make_pass_early_tree_profile): ...new function.
2023         (pass_cleanup_eh): Replace declaration with that of...
2024         (make_pass_cleanup_eh): ...new function.
2025         (pass_sra): Replace declaration with that of...
2026         (make_pass_sra): ...new function.
2027         (pass_sra_early): Replace declaration with that of...
2028         (make_pass_sra_early): ...new function.
2029         (pass_early_ipa_sra): Replace declaration with that of...
2030         (make_pass_early_ipa_sra): ...new function.
2031         (pass_tail_recursion): Replace declaration with that of...
2032         (make_pass_tail_recursion): ...new function.
2033         (pass_tail_calls): Replace declaration with that of...
2034         (make_pass_tail_calls): ...new function.
2035         (pass_tree_loop): Replace declaration with that of...
2036         (make_pass_tree_loop): ...new function.
2037         (pass_tree_loop_init): Replace declaration with that of...
2038         (make_pass_tree_loop_init): ...new function.
2039         (pass_lim): Replace declaration with that of...
2040         (make_pass_lim): ...new function.
2041         (pass_tree_unswitch): Replace declaration with that of...
2042         (make_pass_tree_unswitch): ...new function.
2043         (pass_predcom): Replace declaration with that of...
2044         (make_pass_predcom): ...new function.
2045         (pass_iv_canon): Replace declaration with that of...
2046         (make_pass_iv_canon): ...new function.
2047         (pass_scev_cprop): Replace declaration with that of...
2048         (make_pass_scev_cprop): ...new function.
2049         (pass_empty_loop): Replace declaration with that of...
2050         (make_pass_empty_loop): ...new function.
2051         (pass_record_bounds): Replace declaration with that of...
2052         (make_pass_record_bounds): ...new function.
2053         (pass_graphite): Replace declaration with that of...
2054         (make_pass_graphite): ...new function.
2055         (pass_graphite_transforms): Replace declaration with that of...
2056         (make_pass_graphite_transforms): ...new function.
2057         (pass_if_conversion): Replace declaration with that of...
2058         (make_pass_if_conversion): ...new function.
2059         (pass_loop_distribution): Replace declaration with that of...
2060         (make_pass_loop_distribution): ...new function.
2061         (pass_vectorize): Replace declaration with that of...
2062         (make_pass_vectorize): ...new function.
2063         (pass_slp_vectorize): Replace declaration with that of...
2064         (make_pass_slp_vectorize): ...new function.
2065         (pass_complete_unroll): Replace declaration with that of...
2066         (make_pass_complete_unroll): ...new function.
2067         (pass_complete_unrolli): Replace declaration with that of...
2068         (make_pass_complete_unrolli): ...new function.
2069         (pass_parallelize_loops): Replace declaration with that of...
2070         (make_pass_parallelize_loops): ...new function.
2071         (pass_loop_prefetch): Replace declaration with that of...
2072         (make_pass_loop_prefetch): ...new function.
2073         (pass_iv_optimize): Replace declaration with that of...
2074         (make_pass_iv_optimize): ...new function.
2075         (pass_tree_loop_done): Replace declaration with that of...
2076         (make_pass_tree_loop_done): ...new function.
2077         (pass_ch): Replace declaration with that of...
2078         (make_pass_ch): ...new function.
2079         (pass_ccp): Replace declaration with that of...
2080         (make_pass_ccp): ...new function.
2081         (pass_phi_only_cprop): Replace declaration with that of...
2082         (make_pass_phi_only_cprop): ...new function.
2083         (pass_build_ssa): Replace declaration with that of...
2084         (make_pass_build_ssa): ...new function.
2085         (pass_build_alias): Replace declaration with that of...
2086         (make_pass_build_alias): ...new function.
2087         (pass_build_ealias): Replace declaration with that of...
2088         (make_pass_build_ealias): ...new function.
2089         (pass_dominator): Replace declaration with that of...
2090         (make_pass_dominator): ...new function.
2091         (pass_dce): Replace declaration with that of...
2092         (make_pass_dce): ...new function.
2093         (pass_dce_loop): Replace declaration with that of...
2094         (make_pass_dce_loop): ...new function.
2095         (pass_cd_dce): Replace declaration with that of...
2096         (make_pass_cd_dce): ...new function.
2097         (pass_call_cdce): Replace declaration with that of...
2098         (make_pass_call_cdce): ...new function.
2099         (pass_merge_phi): Replace declaration with that of...
2100         (make_pass_merge_phi): ...new function.
2101         (pass_split_crit_edges): Replace declaration with that of...
2102         (make_pass_split_crit_edges): ...new function.
2103         (pass_pre): Replace declaration with that of...
2104         (make_pass_pre): ...new function.
2105         (pass_profile): Replace declaration with that of...
2106         (make_pass_profile): ...new function.
2107         (pass_strip_predict_hints): Replace declaration with that of...
2108         (make_pass_strip_predict_hints): ...new function.
2109         (pass_lower_complex_O0): Replace declaration with that of...
2110         (make_pass_lower_complex_O0): ...new function.
2111         (pass_lower_complex): Replace declaration with that of...
2112         (make_pass_lower_complex): ...new function.
2113         (pass_lower_vector): Replace declaration with that of...
2114         (make_pass_lower_vector): ...new function.
2115         (pass_lower_vector_ssa): Replace declaration with that of...
2116         (make_pass_lower_vector_ssa): ...new function.
2117         (pass_lower_omp): Replace declaration with that of...
2118         (make_pass_lower_omp): ...new function.
2119         (pass_diagnose_omp_blocks): Replace declaration with that of...
2120         (make_pass_diagnose_omp_blocks): ...new function.
2121         (pass_expand_omp): Replace declaration with that of...
2122         (make_pass_expand_omp): ...new function.
2123         (pass_expand_omp_ssa): Replace declaration with that of...
2124         (make_pass_expand_omp_ssa): ...new function.
2125         (pass_object_sizes): Replace declaration with that of...
2126         (make_pass_object_sizes): ...new function.
2127         (pass_strlen): Replace declaration with that of...
2128         (make_pass_strlen): ...new function.
2129         (pass_fold_builtins): Replace declaration with that of...
2130         (make_pass_fold_builtins): ...new function.
2131         (pass_stdarg): Replace declaration with that of...
2132         (make_pass_stdarg): ...new function.
2133         (pass_early_warn_uninitialized): Replace declaration with that of...
2134         (make_pass_early_warn_uninitialized): ...new function.
2135         (pass_late_warn_uninitialized): Replace declaration with that of...
2136         (make_pass_late_warn_uninitialized): ...new function.
2137         (pass_cse_reciprocals): Replace declaration with that of...
2138         (make_pass_cse_reciprocals): ...new function.
2139         (pass_cse_sincos): Replace declaration with that of...
2140         (make_pass_cse_sincos): ...new function.
2141         (pass_optimize_bswap): Replace declaration with that of...
2142         (make_pass_optimize_bswap): ...new function.
2143         (pass_optimize_widening_mul): Replace declaration with that of...
2144         (make_pass_optimize_widening_mul): ...new function.
2145         (pass_warn_function_return): Replace declaration with that of...
2146         (make_pass_warn_function_return): ...new function.
2147         (pass_warn_function_noreturn): Replace declaration with that of...
2148         (make_pass_warn_function_noreturn): ...new function.
2149         (pass_cselim): Replace declaration with that of...
2150         (make_pass_cselim): ...new function.
2151         (pass_phiopt): Replace declaration with that of...
2152         (make_pass_phiopt): ...new function.
2153         (pass_forwprop): Replace declaration with that of...
2154         (make_pass_forwprop): ...new function.
2155         (pass_phiprop): Replace declaration with that of...
2156         (make_pass_phiprop): ...new function.
2157         (pass_tree_ifcombine): Replace declaration with that of...
2158         (make_pass_tree_ifcombine): ...new function.
2159         (pass_dse): Replace declaration with that of...
2160         (make_pass_dse): ...new function.
2161         (pass_nrv): Replace declaration with that of...
2162         (make_pass_nrv): ...new function.
2163         (pass_rename_ssa_copies): Replace declaration with that of...
2164         (make_pass_rename_ssa_copies): ...new function.
2165         (pass_sink_code): Replace declaration with that of...
2166         (make_pass_sink_code): ...new function.
2167         (pass_fre): Replace declaration with that of...
2168         (make_pass_fre): ...new function.
2169         (pass_check_data_deps): Replace declaration with that of...
2170         (make_pass_check_data_deps): ...new function.
2171         (pass_copy_prop): Replace declaration with that of...
2172         (make_pass_copy_prop): ...new function.
2173         (pass_vrp): Replace declaration with that of...
2174         (make_pass_vrp): ...new function.
2175         (pass_uncprop): Replace declaration with that of...
2176         (make_pass_uncprop): ...new function.
2177         (pass_return_slot): Replace declaration with that of...
2178         (make_pass_return_slot): ...new function.
2179         (pass_reassoc): Replace declaration with that of...
2180         (make_pass_reassoc): ...new function.
2181         (pass_rebuild_cgraph_edges): Replace declaration with that of...
2182         (make_pass_rebuild_cgraph_edges): ...new function.
2183         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2184         (make_pass_remove_cgraph_callee_edges): ...new function.
2185         (pass_build_cgraph_edges): Replace declaration with that of...
2186         (make_pass_build_cgraph_edges): ...new function.
2187         (pass_local_pure_const): Replace declaration with that of...
2188         (make_pass_local_pure_const): ...new function.
2189         (pass_tracer): Replace declaration with that of...
2190         (make_pass_tracer): ...new function.
2191         (pass_warn_unused_result): Replace declaration with that of...
2192         (make_pass_warn_unused_result): ...new function.
2193         (pass_diagnose_tm_blocks): Replace declaration with that of...
2194         (make_pass_diagnose_tm_blocks): ...new function.
2195         (pass_lower_tm): Replace declaration with that of...
2196         (make_pass_lower_tm): ...new function.
2197         (pass_tm_init): Replace declaration with that of...
2198         (make_pass_tm_init): ...new function.
2199         (pass_tm_mark): Replace declaration with that of...
2200         (make_pass_tm_mark): ...new function.
2201         (pass_tm_memopt): Replace declaration with that of...
2202         (make_pass_tm_memopt): ...new function.
2203         (pass_tm_edges): Replace declaration with that of...
2204         (make_pass_tm_edges): ...new function.
2205         (pass_split_functions): Replace declaration with that of...
2206         (make_pass_split_functions): ...new function.
2207         (pass_feedback_split_functions): Replace declaration with that of...
2208         (make_pass_feedback_split_functions): ...new function.
2209         (pass_strength_reduction): Replace declaration with that of...
2210         (make_pass_strength_reduction): ...new function.
2211         (pass_ipa_lower_emutls): Replace declaration with that of...
2212         (make_pass_ipa_lower_emutls): ...new function.
2213         (pass_ipa_function_and_variable_visibility): Replace declaration with
2214         that of...
2215         (make_pass_ipa_function_and_variable_visibility): ...new function.
2216         (pass_ipa_tree_profile): Replace declaration with that of...
2217         (make_pass_ipa_tree_profile): ...new function.
2218         (pass_early_local_passes): Replace declaration with that of...
2219         (make_pass_early_local_passes): ...new function.
2220         (pass_ipa_whole_program_visibility): Replace declaration with that
2221         of...
2222         (make_pass_ipa_whole_program_visibility): ...new function.
2223         (pass_ipa_lto_gimple_out): Replace declaration with that of...
2224         (make_pass_ipa_lto_gimple_out): ...new function.
2225         (pass_ipa_increase_alignment): Replace declaration with that of...
2226         (make_pass_ipa_increase_alignment): ...new function.
2227         (pass_ipa_inline): Replace declaration with that of...
2228         (make_pass_ipa_inline): ...new function.
2229         (pass_ipa_free_lang_data): Replace declaration with that of...
2230         (make_pass_ipa_free_lang_data): ...new function.
2231         (pass_ipa_free_inline_summary): Replace declaration with that of...
2232         (make_pass_ipa_free_inline_summary): ...new function.
2233         (pass_ipa_cp): Replace declaration with that of...
2234         (make_pass_ipa_cp): ...new function.
2235         (pass_ipa_reference): Replace declaration with that of...
2236         (make_pass_ipa_reference): ...new function.
2237         (pass_ipa_pure_const): Replace declaration with that of...
2238         (make_pass_ipa_pure_const): ...new function.
2239         (pass_ipa_pta): Replace declaration with that of...
2240         (make_pass_ipa_pta): ...new function.
2241         (pass_ipa_lto_finish_out): Replace declaration with that of...
2242         (make_pass_ipa_lto_finish_out): ...new function.
2243         (pass_ipa_tm): Replace declaration with that of...
2244         (make_pass_ipa_tm): ...new function.
2245         (pass_ipa_profile): Replace declaration with that of...
2246         (make_pass_ipa_profile): ...new function.
2247         (pass_ipa_cdtor_merge): Replace declaration with that of...
2248         (make_pass_ipa_cdtor_merge): ...new function.
2249         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
2250         of...
2251         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2252         (pass_init_datastructures): Replace declaration with that of...
2253         (make_pass_init_datastructures): ...new function.
2254         (pass_fixup_cfg): Replace declaration with that of...
2255         (make_pass_fixup_cfg): ...new function.
2256         (pass_expand): Replace declaration with that of...
2257         (make_pass_expand): ...new function.
2258         (pass_instantiate_virtual_regs): Replace declaration with that of...
2259         (make_pass_instantiate_virtual_regs): ...new function.
2260         (pass_rtl_fwprop): Replace declaration with that of...
2261         (make_pass_rtl_fwprop): ...new function.
2262         (pass_rtl_fwprop_addr): Replace declaration with that of...
2263         (make_pass_rtl_fwprop_addr): ...new function.
2264         (pass_jump): Replace declaration with that of...
2265         (make_pass_jump): ...new function.
2266         (pass_jump2): Replace declaration with that of...
2267         (make_pass_jump2): ...new function.
2268         (pass_lower_subreg): Replace declaration with that of...
2269         (make_pass_lower_subreg): ...new function.
2270         (pass_cse): Replace declaration with that of...
2271         (make_pass_cse): ...new function.
2272         (pass_fast_rtl_dce): Replace declaration with that of...
2273         (make_pass_fast_rtl_dce): ...new function.
2274         (pass_ud_rtl_dce): Replace declaration with that of...
2275         (make_pass_ud_rtl_dce): ...new function.
2276         (pass_rtl_dce): Replace declaration with that of...
2277         (make_pass_rtl_dce): ...new function.
2278         (pass_rtl_dse1): Replace declaration with that of...
2279         (make_pass_rtl_dse1): ...new function.
2280         (pass_rtl_dse2): Replace declaration with that of...
2281         (make_pass_rtl_dse2): ...new function.
2282         (pass_rtl_dse3): Replace declaration with that of...
2283         (make_pass_rtl_dse3): ...new function.
2284         (pass_rtl_cprop): Replace declaration with that of...
2285         (make_pass_rtl_cprop): ...new function.
2286         (pass_rtl_pre): Replace declaration with that of...
2287         (make_pass_rtl_pre): ...new function.
2288         (pass_rtl_hoist): Replace declaration with that of...
2289         (make_pass_rtl_hoist): ...new function.
2290         (pass_rtl_store_motion): Replace declaration with that of...
2291         (make_pass_rtl_store_motion): ...new function.
2292         (pass_cse_after_global_opts): Replace declaration with that of...
2293         (make_pass_cse_after_global_opts): ...new function.
2294         (pass_rtl_ifcvt): Replace declaration with that of...
2295         (make_pass_rtl_ifcvt): ...new function.
2296         (pass_into_cfg_layout_mode): Replace declaration with that of...
2297         (make_pass_into_cfg_layout_mode): ...new function.
2298         (pass_outof_cfg_layout_mode): Replace declaration with that of...
2299         (make_pass_outof_cfg_layout_mode): ...new function.
2300         (pass_loop2): Replace declaration with that of...
2301         (make_pass_loop2): ...new function.
2302         (pass_rtl_loop_init): Replace declaration with that of...
2303         (make_pass_rtl_loop_init): ...new function.
2304         (pass_rtl_move_loop_invariants): Replace declaration with that of...
2305         (make_pass_rtl_move_loop_invariants): ...new function.
2306         (pass_rtl_unswitch): Replace declaration with that of...
2307         (make_pass_rtl_unswitch): ...new function.
2308         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
2309         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2310         (pass_rtl_doloop): Replace declaration with that of...
2311         (make_pass_rtl_doloop): ...new function.
2312         (pass_rtl_loop_done): Replace declaration with that of...
2313         (make_pass_rtl_loop_done): ...new function.
2314         (pass_web): Replace declaration with that of...
2315         (make_pass_web): ...new function.
2316         (pass_cse2): Replace declaration with that of...
2317         (make_pass_cse2): ...new function.
2318         (pass_df_initialize_opt): Replace declaration with that of...
2319         (make_pass_df_initialize_opt): ...new function.
2320         (pass_df_initialize_no_opt): Replace declaration with that of...
2321         (make_pass_df_initialize_no_opt): ...new function.
2322         (pass_reginfo_init): Replace declaration with that of...
2323         (make_pass_reginfo_init): ...new function.
2324         (pass_inc_dec): Replace declaration with that of...
2325         (make_pass_inc_dec): ...new function.
2326         (pass_stack_ptr_mod): Replace declaration with that of...
2327         (make_pass_stack_ptr_mod): ...new function.
2328         (pass_initialize_regs): Replace declaration with that of...
2329         (make_pass_initialize_regs): ...new function.
2330         (pass_combine): Replace declaration with that of...
2331         (make_pass_combine): ...new function.
2332         (pass_if_after_combine): Replace declaration with that of...
2333         (make_pass_if_after_combine): ...new function.
2334         (pass_ree): Replace declaration with that of...
2335         (make_pass_ree): ...new function.
2336         (pass_partition_blocks): Replace declaration with that of...
2337         (make_pass_partition_blocks): ...new function.
2338         (pass_match_asm_constraints): Replace declaration with that of...
2339         (make_pass_match_asm_constraints): ...new function.
2340         (pass_regmove): Replace declaration with that of...
2341         (make_pass_regmove): ...new function.
2342         (pass_split_all_insns): Replace declaration with that of...
2343         (make_pass_split_all_insns): ...new function.
2344         (pass_fast_rtl_byte_dce): Replace declaration with that of...
2345         (make_pass_fast_rtl_byte_dce): ...new function.
2346         (pass_lower_subreg2): Replace declaration with that of...
2347         (make_pass_lower_subreg2): ...new function.
2348         (pass_mode_switching): Replace declaration with that of...
2349         (make_pass_mode_switching): ...new function.
2350         (pass_sms): Replace declaration with that of...
2351         (make_pass_sms): ...new function.
2352         (pass_sched): Replace declaration with that of...
2353         (make_pass_sched): ...new function.
2354         (pass_ira): Replace declaration with that of...
2355         (make_pass_ira): ...new function.
2356         (pass_reload): Replace declaration with that of...
2357         (make_pass_reload): ...new function.
2358         (pass_clean_state): Replace declaration with that of...
2359         (make_pass_clean_state): ...new function.
2360         (pass_branch_prob): Replace declaration with that of...
2361         (make_pass_branch_prob): ...new function.
2362         (pass_value_profile_transformations): Replace declaration with that
2363         of...
2364         (make_pass_value_profile_transformations): ...new function.
2365         (pass_postreload_cse): Replace declaration with that of...
2366         (make_pass_postreload_cse): ...new function.
2367         (pass_gcse2): Replace declaration with that of...
2368         (make_pass_gcse2): ...new function.
2369         (pass_split_after_reload): Replace declaration with that of...
2370         (make_pass_split_after_reload): ...new function.
2371         (pass_branch_target_load_optimize1): Replace declaration with that
2372         of...
2373         (make_pass_branch_target_load_optimize1): ...new function.
2374         (pass_thread_prologue_and_epilogue): Replace declaration with that
2375         of...
2376         (make_pass_thread_prologue_and_epilogue): ...new function.
2377         (pass_stack_adjustments): Replace declaration with that of...
2378         (make_pass_stack_adjustments): ...new function.
2379         (pass_peephole2): Replace declaration with that of...
2380         (make_pass_peephole2): ...new function.
2381         (pass_if_after_reload): Replace declaration with that of...
2382         (make_pass_if_after_reload): ...new function.
2383         (pass_regrename): Replace declaration with that of...
2384         (make_pass_regrename): ...new function.
2385         (pass_cprop_hardreg): Replace declaration with that of...
2386         (make_pass_cprop_hardreg): ...new function.
2387         (pass_reorder_blocks): Replace declaration with that of...
2388         (make_pass_reorder_blocks): ...new function.
2389         (pass_branch_target_load_optimize2): Replace declaration with that
2390         of...
2391         (make_pass_branch_target_load_optimize2): ...new function.
2392         (pass_leaf_regs): Replace declaration with that of...
2393         (make_pass_leaf_regs): ...new function.
2394         (pass_split_before_sched2): Replace declaration with that of...
2395         (make_pass_split_before_sched2): ...new function.
2396         (pass_compare_elim_after_reload): Replace declaration with that of...
2397         (make_pass_compare_elim_after_reload): ...new function.
2398         (pass_sched2): Replace declaration with that of...
2399         (make_pass_sched2): ...new function.
2400         (pass_stack_regs): Replace declaration with that of...
2401         (make_pass_stack_regs): ...new function.
2402         (pass_stack_regs_run): Replace declaration with that of...
2403         (make_pass_stack_regs_run): ...new function.
2404         (pass_df_finish): Replace declaration with that of...
2405         (make_pass_df_finish): ...new function.
2406         (pass_compute_alignments): Replace declaration with that of...
2407         (make_pass_compute_alignments): ...new function.
2408         (pass_duplicate_computed_gotos): Replace declaration with that of...
2409         (make_pass_duplicate_computed_gotos): ...new function.
2410         (pass_variable_tracking): Replace declaration with that of...
2411         (make_pass_variable_tracking): ...new function.
2412         (pass_free_cfg): Replace declaration with that of...
2413         (make_pass_free_cfg): ...new function.
2414         (pass_machine_reorg): Replace declaration with that of...
2415         (make_pass_machine_reorg): ...new function.
2416         (pass_cleanup_barriers): Replace declaration with that of...
2417         (make_pass_cleanup_barriers): ...new function.
2418         (pass_delay_slots): Replace declaration with that of...
2419         (make_pass_delay_slots): ...new function.
2420         (pass_split_for_shorten_branches): Replace declaration with that of...
2421         (make_pass_split_for_shorten_branches): ...new function.
2422         (pass_split_before_regstack): Replace declaration with that of...
2423         (make_pass_split_before_regstack): ...new function.
2424         (pass_convert_to_eh_region_ranges): Replace declaration with that
2425         of...
2426         (make_pass_convert_to_eh_region_ranges): ...new function.
2427         (pass_shorten_branches): Replace declaration with that of...
2428         (make_pass_shorten_branches): ...new function.
2429         (pass_set_nothrow_function_flags): Replace declaration with that of...
2430         (make_pass_set_nothrow_function_flags): ...new function.
2431         (pass_dwarf2_frame): Replace declaration with that of...
2432         (make_pass_dwarf2_frame): ...new function.
2433         (pass_final): Replace declaration with that of...
2434         (make_pass_final): ...new function.
2435         (pass_rtl_seqabstr): Replace declaration with that of...
2436         (make_pass_rtl_seqabstr): ...new function.
2437         (pass_release_ssa_names): Replace declaration with that of...
2438         (make_pass_release_ssa_names): ...new function.
2439         (pass_early_inline): Replace declaration with that of...
2440         (make_pass_early_inline): ...new function.
2441         (pass_inline_parameters): Replace declaration with that of...
2442         (make_pass_inline_parameters): ...new function.
2443         (pass_update_address_taken): Replace declaration with that of...
2444         (make_pass_update_address_taken): ...new function.
2445         (pass_convert_switch): Replace declaration with that of...
2446         (make_pass_convert_switch): ...new function.
2447         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
2448         to a subclass of simple_ipa_opt_pass along with...
2449         (pass_data_ipa_tree_profile): ...new pass_data instance and...
2450         (make_pass_ipa_tree_profile): ...new function.
2451         * tree-sra.c (pass_sra_early): Convert from a global struct to a
2452         subclass of gimple_opt_pass along with...
2453         (pass_data_sra_early): ...new pass_data instance and...
2454         (make_pass_sra_early): ...new function.
2455         (pass_sra): Convert from a global struct to a subclass of
2456         gimple_opt_pass along with...
2457         (pass_data_sra): ...new pass_data instance and...
2458         (make_pass_sra): ...new function.
2459         (pass_early_ipa_sra): Convert from a global struct to a subclass of
2460         gimple_opt_pass along with...
2461         (pass_data_early_ipa_sra): ...new pass_data instance and...
2462         (make_pass_early_ipa_sra): ...new function.
2463         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
2464         subclass of gimple_opt_pass along with...
2465         (pass_data_ccp): ...new pass_data instance and...
2466         (make_pass_ccp): ...new function.
2467         (pass_fold_builtins): Convert from a global struct to a subclass of
2468         gimple_opt_pass along with...
2469         (pass_data_fold_builtins): ...new pass_data instance and...
2470         (make_pass_fold_builtins): ...new function.
2471         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
2472         subclass of gimple_opt_pass along with...
2473         (pass_data_copy_prop): ...new pass_data instance and...
2474         (make_pass_copy_prop): ...new function.
2475         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
2476         global struct to a subclass of gimple_opt_pass along with...
2477         (pass_data_rename_ssa_copies): ...new pass_data instance and...
2478         (make_pass_rename_ssa_copies): ...new function.
2479         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
2480         subclass of gimple_opt_pass along with...
2481         (pass_data_dce): ...new pass_data instance and...
2482         (make_pass_dce): ...new function.
2483         (pass_dce_loop): Convert from a global struct to a subclass of
2484         gimple_opt_pass along with...
2485         (pass_data_dce_loop): ...new pass_data instance and...
2486         (make_pass_dce_loop): ...new function.
2487         (pass_cd_dce): Convert from a global struct to a subclass of
2488         gimple_opt_pass along with...
2489         (pass_data_cd_dce): ...new pass_data instance and...
2490         (make_pass_cd_dce): ...new function.
2491         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
2492         subclass of gimple_opt_pass along with...
2493         (pass_data_dominator): ...new pass_data instance and...
2494         (make_pass_dominator): ...new function.
2495         (pass_phi_only_cprop): Convert from a global struct to a subclass of
2496         gimple_opt_pass along with...
2497         (pass_data_phi_only_cprop): ...new pass_data instance and...
2498         (make_pass_phi_only_cprop): ...new function.
2499         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
2500         subclass of gimple_opt_pass along with...
2501         (pass_data_dse): ...new pass_data instance and...
2502         (make_pass_dse): ...new function.
2503         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
2504         a subclass of gimple_opt_pass along with...
2505         (pass_data_forwprop): ...new pass_data instance and...
2506         (make_pass_forwprop): ...new function.
2507         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
2508         struct to a subclass of gimple_opt_pass along with...
2509         (pass_data_tree_ifcombine): ...new pass_data instance and...
2510         (make_pass_tree_ifcombine): ...new function.
2511         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
2512         subclass of gimple_opt_pass along with...
2513         (pass_data_ch): ...new pass_data instance and...
2514         (make_pass_ch): ...new function.
2515         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
2516         subclass of gimple_opt_pass along with...
2517         (pass_data_tree_loop): ...new pass_data instance and...
2518         (make_pass_tree_loop): ...new function.
2519         (pass_tree_loop_init): Convert from a global struct to a subclass of
2520         gimple_opt_pass along with...
2521         (pass_data_tree_loop_init): ...new pass_data instance and...
2522         (make_pass_tree_loop_init): ...new function.
2523         (pass_lim): Convert from a global struct to a subclass of
2524         gimple_opt_pass along with...
2525         (pass_data_lim): ...new pass_data instance and...
2526         (make_pass_lim): ...new function.
2527         (pass_tree_unswitch): Convert from a global struct to a subclass of
2528         gimple_opt_pass along with...
2529         (pass_data_tree_unswitch): ...new pass_data instance and...
2530         (make_pass_tree_unswitch): ...new function.
2531         (pass_predcom): Convert from a global struct to a subclass of
2532         gimple_opt_pass along with...
2533         (pass_data_predcom): ...new pass_data instance and...
2534         (make_pass_predcom): ...new function.
2535         (pass_vectorize): Convert from a global struct to a subclass of
2536         gimple_opt_pass along with...
2537         (pass_data_vectorize): ...new pass_data instance and...
2538         (make_pass_vectorize): ...new function.
2539         (pass_graphite): Convert from a global struct to a subclass of
2540         gimple_opt_pass along with...
2541         (pass_data_graphite): ...new pass_data instance and...
2542         (make_pass_graphite): ...new function.
2543         (pass_graphite_transforms): Convert from a global struct to a subclass
2544         of gimple_opt_pass along with...
2545         (pass_data_graphite_transforms): ...new pass_data instance and...
2546         (make_pass_graphite_transforms): ...new function.
2547         (pass_check_data_deps): Convert from a global struct to a subclass of
2548         gimple_opt_pass along with...
2549         (pass_data_check_data_deps): ...new pass_data instance and...
2550         (make_pass_check_data_deps): ...new function.
2551         (pass_iv_canon): Convert from a global struct to a subclass of
2552         gimple_opt_pass along with...
2553         (pass_data_iv_canon): ...new pass_data instance and...
2554         (make_pass_iv_canon): ...new function.
2555         (pass_scev_cprop): Convert from a global struct to a subclass of
2556         gimple_opt_pass along with...
2557         (pass_data_scev_cprop): ...new pass_data instance and...
2558         (make_pass_scev_cprop): ...new function.
2559         (pass_record_bounds): Convert from a global struct to a subclass of
2560         gimple_opt_pass along with...
2561         (pass_data_record_bounds): ...new pass_data instance and...
2562         (make_pass_record_bounds): ...new function.
2563         (pass_complete_unroll): Convert from a global struct to a subclass of
2564         gimple_opt_pass along with...
2565         (pass_data_complete_unroll): ...new pass_data instance and...
2566         (make_pass_complete_unroll): ...new function.
2567         (pass_complete_unrolli): Convert from a global struct to a subclass of
2568         gimple_opt_pass along with...
2569         (pass_data_complete_unrolli): ...new pass_data instance and...
2570         (make_pass_complete_unrolli): ...new function.
2571         (pass_parallelize_loops): Convert from a global struct to a subclass
2572         of gimple_opt_pass along with...
2573         (pass_data_parallelize_loops): ...new pass_data instance and...
2574         (make_pass_parallelize_loops): ...new function.
2575         (pass_loop_prefetch): Convert from a global struct to a subclass of
2576         gimple_opt_pass along with...
2577         (pass_data_loop_prefetch): ...new pass_data instance and...
2578         (make_pass_loop_prefetch): ...new function.
2579         (pass_iv_optimize): Convert from a global struct to a subclass of
2580         gimple_opt_pass along with...
2581         (pass_data_iv_optimize): ...new pass_data instance and...
2582         (make_pass_iv_optimize): ...new function.
2583         (pass_tree_loop_done): Convert from a global struct to a subclass of
2584         gimple_opt_pass along with...
2585         (pass_data_tree_loop_done): ...new pass_data instance and...
2586         (make_pass_tree_loop_done): ...new function.
2587         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
2588         struct to a subclass of gimple_opt_pass along with...
2589         (pass_data_cse_reciprocals): ...new pass_data instance and...
2590         (make_pass_cse_reciprocals): ...new function.
2591         (pass_cse_sincos): Convert from a global struct to a subclass of
2592         gimple_opt_pass along with...
2593         (pass_data_cse_sincos): ...new pass_data instance and...
2594         (make_pass_cse_sincos): ...new function.
2595         (pass_optimize_bswap): Convert from a global struct to a subclass of
2596         gimple_opt_pass along with...
2597         (pass_data_optimize_bswap): ...new pass_data instance and...
2598         (make_pass_optimize_bswap): ...new function.
2599         (pass_optimize_widening_mul): Convert from a global struct to a
2600         subclass of gimple_opt_pass along with...
2601         (pass_data_optimize_widening_mul): ...new pass_data instance and...
2602         (make_pass_optimize_widening_mul): ...new function.
2603         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
2604         subclass of gimple_opt_pass along with...
2605         (pass_data_phiopt): ...new pass_data instance and...
2606         (make_pass_phiopt): ...new function.
2607         (pass_cselim): Convert from a global struct to a subclass of
2608         gimple_opt_pass along with...
2609         (pass_data_cselim): ...new pass_data instance and...
2610         (make_pass_cselim): ...new function.
2611         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
2612         subclass of gimple_opt_pass along with...
2613         (pass_data_phiprop): ...new pass_data instance and...
2614         (make_pass_phiprop): ...new function.
2615         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
2616         subclass of gimple_opt_pass along with...
2617         (pass_data_pre): ...new pass_data instance and...
2618         (make_pass_pre): ...new function.
2619         (pass_fre): Convert from a global struct to a subclass of
2620         gimple_opt_pass along with...
2621         (pass_data_fre): ...new pass_data instance and...
2622         (make_pass_fre): ...new function.
2623         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
2624         subclass of gimple_opt_pass along with...
2625         (pass_data_reassoc): ...new pass_data instance and...
2626         (make_pass_reassoc): ...new function.
2627         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
2628         subclass of gimple_opt_pass along with...
2629         (pass_data_sink_code): ...new pass_data instance and...
2630         (make_pass_sink_code): ...new function.
2631         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
2632         subclass of gimple_opt_pass along with...
2633         (pass_data_strlen): ...new pass_data instance and...
2634         (make_pass_strlen): ...new function.
2635         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
2636         struct to a subclass of gimple_opt_pass along with...
2637         (pass_data_build_alias): ...new pass_data instance and...
2638         (make_pass_build_alias): ...new function.
2639         (pass_build_ealias): Convert from a global struct to a subclass of
2640         gimple_opt_pass along with...
2641         (pass_data_build_ealias): ...new pass_data instance and...
2642         (make_pass_build_ealias): ...new function.
2643         (pass_ipa_pta): Convert from a global struct to a subclass of
2644         simple_ipa_opt_pass along with...
2645         (pass_data_ipa_pta): ...new pass_data instance and...
2646         (make_pass_ipa_pta): ...new function.
2647         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
2648         subclass of gimple_opt_pass along with...
2649         (pass_data_uncprop): ...new pass_data instance and...
2650         (make_pass_uncprop): ...new function.
2651         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
2652         global struct to a subclass of gimple_opt_pass along with...
2653         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
2654         (make_pass_late_warn_uninitialized): ...new function.
2655         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
2656         to a subclass of gimple_opt_pass along with...
2657         (pass_data_init_datastructures): ...new pass_data instance and...
2658         (make_pass_init_datastructures): ...new function.
2659         (pass_early_warn_uninitialized): Convert from a global struct to a
2660         subclass of gimple_opt_pass along with...
2661         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
2662         (make_pass_early_warn_uninitialized): ...new function.
2663         (pass_update_address_taken): Convert from a global struct to a
2664         subclass of gimple_opt_pass along with...
2665         (pass_data_update_address_taken): ...new pass_data instance and...
2666         (make_pass_update_address_taken): ...new function.
2667         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
2668         struct to a subclass of gimple_opt_pass along with...
2669         (pass_data_release_ssa_names): ...new pass_data instance and...
2670         (make_pass_release_ssa_names): ...new function.
2671         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
2672         subclass of gimple_opt_pass along with...
2673         (pass_data_stdarg): ...new pass_data instance and...
2674         (make_pass_stdarg): ...new function.
2675         * tree-switch-conversion.c (pass_convert_switch): Convert from a
2676         global struct to a subclass of gimple_opt_pass along with...
2677         (pass_data_convert_switch): ...new pass_data instance and...
2678         (make_pass_convert_switch): ...new function.
2679         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
2680         to a subclass of gimple_opt_pass along with...
2681         (pass_data_tail_recursion): ...new pass_data instance and...
2682         (make_pass_tail_recursion): ...new function.
2683         (pass_tail_calls): Convert from a global struct to a subclass of
2684         gimple_opt_pass along with...
2685         (pass_data_tail_calls): ...new pass_data instance and...
2686         (make_pass_tail_calls): ...new function.
2687         * tree-vect-generic.c (pass_lower_vector): Convert from a global
2688         struct to a subclass of gimple_opt_pass along with...
2689         (pass_data_lower_vector): ...new pass_data instance and...
2690         (make_pass_lower_vector): ...new function.
2691         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
2692         gimple_opt_pass along with...
2693         (pass_data_lower_vector_ssa): ...new pass_data instance and...
2694         (make_pass_lower_vector_ssa): ...new function.
2695         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
2696         to a subclass of gimple_opt_pass along with...
2697         (pass_data_slp_vectorize): ...new pass_data instance and...
2698         (make_pass_slp_vectorize): ...new function.
2699         (pass_ipa_increase_alignment): Convert from a global struct to a
2700         subclass of simple_ipa_opt_pass along with...
2701         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
2702         (make_pass_ipa_increase_alignment): ...new function.
2703         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
2704         gimple_opt_pass along with...
2705         (pass_data_vrp): ...new pass_data instance and...
2706         (make_pass_vrp): ...new function.
2707         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
2708         subclass of simple_ipa_opt_pass along with...
2709         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
2710         (make_pass_ipa_free_lang_data): ...new function.
2711         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
2712         gimple_opt_pass along with...
2713         (pass_data_tsan): ...new pass_data instance and...
2714         (make_pass_tsan): ...new function.
2715         (pass_tsan_O0): Convert from a global struct to a subclass of
2716         gimple_opt_pass along with...
2717         (pass_data_tsan_O0): ...new pass_data instance and...
2718         (make_pass_tsan_O0): ...new function.
2719         * var-tracking.c (pass_variable_tracking): Convert from a global
2720         struct to a subclass of rtl_opt_pass along with...
2721         (pass_data_variable_tracking): ...new pass_data instance and...
2722         (make_pass_variable_tracking): ...new function.
2723         * web.c (pass_web): Convert from a global struct to a subclass of
2724         rtl_opt_pass along with...
2725         (pass_data_web): ...new pass_data instance and...
2726         (make_pass_web): ...new function.
2727         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
2728         declaration with that of...
2729         (make_pass_mode_switch_use): ...new function.
2730         (pass_resolve_sw_modes): Replace declaration with that of...
2731         (make_pass_resolve_sw_modes): ...new function.
2732         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
2733         from a global struct to a subclass of rtl_opt_pass along with...
2734         (pass_data_mode_switch_use): ...new pass_data instance and...
2735         (make_pass_mode_switch_use): ...new function.
2736         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
2737         from a global struct to a subclass of rtl_opt_pass along with...
2738         (pass_data_resolve_sw_modes): ...new pass_data instance and...
2739         (make_pass_resolve_sw_modes): ...new function.
2740         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
2741         struct to a subclass of rtl_opt_pass along with...
2742         (pass_data_insert_vzeroupper): ...new pass_data instance and...
2743         (make_pass_insert_vzeroupper): ...new function.
2744         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
2745         global struct to a subclass of rtl_opt_pass along with...
2746         (pass_data_work_around_errata): ...new pass_data instance and...
2747         (make_pass_work_around_errata): ...new function.
2748         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
2749         struct to a subclass of rtl_opt_pass along with...
2750         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
2751         (make_pass_mips_machine_reorg2): ...new function.
2753 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2755         * passes.c (pass_manager::operator new): New.
2757 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2759         Handwritten part of conversion of passes to C++ classes.
2761         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
2762         (toplev.o): Add dep on PASS_MANAGER_H.
2763         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
2764         of early local pases to reflect this moving from a global to a
2765         member of gcc::pass_manager.
2766         (cgraph_add_new_function): Likewise.
2767         * lto-cgraph.c (lto_output_node): Update for conversion of
2768         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
2769         * passes.c (opt_pass::clone): New.
2770         (opt_pass::gate): New.
2771         (opt_pass::execute): New.
2772         (opt_pass::opt_pass): New.
2773         (pass_manager::execute_early_local_passes): New.
2774         (pass_manager::execute_pass_mode_switching): new.
2775         (finish_optimization_passes): Convert to...
2776         (pass_manager::finish_optimization_passes): ...this.
2777         (finish_optimization_passes): Update for conversion of passes to
2778         C++ classes.
2779         (register_dump_files_1): Use has_gate since we cannot portably
2780         check a vtable entry against NULL.
2781         (dump_one_pass): Likewise.
2782         (ipa_write_summaries_2): Likewise.
2783         (ipa_write_optimization_summaries_1): Likewise.
2784         (ipa_read_summaries_1): Likewise.
2785         (ipa_read_optimization_summaries_1): Likewise.
2786         (execute_ipa_stmt_fixups): Likewise.
2787         (pass_manager::pass_manager): Rewrite pass-creation, invoking
2788         pass-creation functions rather than wiring up globals, and
2789         storing the results in fields of pass_manager generated using
2790         pass-instances.def.
2791         (pass_manager::dump_profile_report): Update for conversion of
2792         passes to C++ classes.
2793         (pass_manager::execute_ipa_summary_passes): Likewise.
2794         (execute_one_ipa_transform_pass): Likewise.
2795         (execute_one_pass): Use has_gate and has_execute since we cannot
2796         portably check a vtable entry against NULL.
2797         * pass_manager.h (pass_manager::finish_optimization_passes): New.
2798         (pass_manager): Use pass-instances.def to add fields for the
2799         various pass instances.
2800         * toplev.c (finalize): Update for move of
2801         finish_optimization_passes to a method of gcc::pass_manager.
2802         * toplev.h (finish_optimization_passes): Move to method of class
2803         pass_manager.
2804         * tree-pass.h (struct pass_data): New.
2805         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
2806         (opt_pass::gate): Convert to virtual function.
2807         (opt_pass::~opt_pass): New.
2808         (opt_pass::clone): New.
2809         (opt_pass::execute): Convert to virtual function.
2810         (opt_pass::opt_pass): New.
2811         (opt_pass::ctxt_): new.
2812         (gimple_opt_pass): Convert to subclass of opt_pass.
2813         (gimple_opt_pass::gimple_opt_pass): New.
2814         (rtl_opt_pass): Convert to subclass of opt_pass.
2815         (rtl_opt_pass::rtl_opt_pass): New.
2816         (ipa_opt_pass_d): Convert to subclass of opt_pass.
2817         (ipa_opt_pass_d::ipa_opt_pass_d): New.
2818         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
2819         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
2820         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
2821         invocation of pass_mode_switching to reflect this moving from a
2822         global to a member of gcc::pass_manager.
2823         (ix86_option_override): Rework how pass_insert_vzeroupper is
2824         added to the pass_manager to reflect autogenerated changes.
2825         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
2827 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
2829         PR rtl-optimization/57708
2830         * recog.c (peep2_find_free_register): Validate all regs in a
2831         multi-reg mode.
2833 2013-08-05  Jan Hubicka  <jh@suse.cz>
2835         PR lto/57602
2836         * cgraph.c (verify_cgraph_node): Accept local flags from other
2837         partitions.
2838         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
2839         (function_and_variable_visibility): Likewise.
2840         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
2842 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2844         * graph.c (init_graph_slim_pretty_print): Remove.
2845         (print_graph_cfg): Do not call it.  Use local pretty printer.
2846         (start_graph_dump): Likewise.
2848 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2850         * gimple-pretty-print.c (buffer): Remove.
2851         (initialized): Likewise.
2852         (maybe_init_pretty_print): Likewise.
2853         (print_gimple_stmt): Do not call it.  Use non-static local
2854         pretty_printer variable.
2855         (print_gimple_expr): Likewise.
2856         (print_gimple_seq): Likewise.
2857         (gimple_dump_bb): Likewise.
2859 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2861         * asan.c (asan_pp): Remove.
2862         (asan_pp_initialized): Likewise.
2863         (asan_pp_initialize): Likewise.
2864         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
2865         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
2866         (asan_add_global): Likewise.
2868 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2870         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
2871         * diagnostic.h (diagnostic_flush_buffer): Adjust.
2872         * pretty-print.c (pp_formatted_text_data): Likewise.
2873         (pp_indent): Rename from pp_base_indent.
2874         (pp_format): Rename from pp_base_format.
2875         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
2876         (pp_format_verbatim): Rename from pp_base_format_verbatim.
2877         (pp_flush): Rename from pp_base_flush.
2878         (pp_set_line_maximum_length): Rename from
2879         pp_base_set_line_maximum_length.
2880         (pp_clear_output_area): Rename from pp_base_clear_output_area.
2881         (pp_set_prefix): Rename from pp_base_set_prefix.
2882         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
2883         (pp_emit_prefix): Rename from pp_base_emit_prefix.
2884         (pp_append_text): Rename from pp_base_append_text.
2885         (pp_formatted_text): Rename from pp_base_formatted_text.
2886         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
2887         (pp_remaining_character_count_for_line): Rename from
2888         pp_base_remaining_character_count_for_line.
2889         (pp_newline): Rename from pp_base_newline.
2890         (pp_character): Rename from pp_base_character.
2891         (pp_string): Rename from pp_base_string.
2892         (pp_maybe_space): Rename from pp_base_maybe_space.
2893         * asan.c (asan_pp_string): Adjust.
2894         (asan_emit_stack_protection): Likewise.
2895         (asan_add_global): Likewise.
2896         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
2897         * tree-mudflap.c (mf_varname_tree): Likewise.
2898         * tree-pretty-print.c (pp_tree_identifier): Rename from
2899         pp_base_tree_identifier.
2900         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
2901         Declare as function.
2903 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2905         * pretty-print.h (pp_bar_bar): New.
2906         (pp_ampersand_ampersand): Likewise.
2907         (pp_less_equal): Likewise.
2908         (pp_greater_equal): Likewise.
2909         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
2910         printer functions instead of pp_string or operators and punctuators.
2911         (dump_gimple_call): Likewise.
2912         (dump_gimple_omp_for): Likewise.
2913         (dump_gimple_transaction): Likewise.
2914         (dump_gimple_phi): Likewise.
2915         (pp_gimple_stmt_1): Likewise.
2916         * sched-vis.c (print_insn): Likewise.
2917         * tree-mudflap.c (mf_varname_tree): Likewise.
2918         * tree-pretty-print.c (dump_block_node): Likewise.
2919         (dump_generic_node): Likewise.
2921 2013-08-02  Jan Hubicka  <jh@suse.cz>
2923         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
2924         boundaries.
2925         * lto-streamer-out.c (tree_is_indexable): Results decls and
2926         parm decls are not indexable.
2927         (DFS_write_tree_body): Do not follow args and results.
2928         (hash_tree): Likewise.
2929         (output_functions): Rearrange so struct function is needed
2930         only when real body is output; be able to also ouptut abstract
2931         functions; output DECL_ARGUMENTS and DECL_RESULT.
2932         (lto_output): When not in WPA, ale store abstract functions.
2933         (write_symbol): Do not care about RESULT_DECL.
2934         (output_symbol_p): Handle correctly sbtract decls.
2935         * lto-streamer-in.c (input_function): Rearrange so struct
2936         function can be NULL at entry; allow streaming of
2937         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2938         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2939         sanity check during LTO.
2940         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2941         RESULT_DECl and DECL_ARGUMENTS.
2942         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2943         Likewise.
2945 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2947         * pretty-print.h (pp_underscore): New.
2948         (pp_comma): Tidy.
2949         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2950         printer functions instead of pp_character.
2951         (dump_binary_rhs): Likewise.
2952         (dump_ternary_rhs): Likewise.
2953         (dump_gimple_call_args): Likewise.
2954         (pp_points_to_solution): Likewise.
2955         (dump_gimple_call): Likewise.
2956         (dump_gimple_switch): Likewise.
2957         (dump_gimple_cond): Likewise.
2958         (dump_gimple_bind): Likewise.
2959         (dump_gimple_try): Likewise.
2960         (dump_gimple_omp_for): Likewise.
2961         (dump_gimple_omp_continue): Likewise.
2962         (dump_gimple_omp_single): Likewise.
2963         (dump_gimple_omp_sections): Likewise.
2964         (dump_gimple_omp_block): Likewise.
2965         (dump_gimple_omp_critical): Likewise.
2966         (dump_gimple_transaction): Likewise.
2967         (dump_gimple_asm): Likewise.
2968         (dump_gimple_phi): Likewise.
2969         (dump_gimple_omp_parallel): Likewise.
2970         (dump_gimple_omp_task): Likewise.
2971         (dump_gimple_omp_atomic_load): Likewise.
2972         (dump_gimple_omp_atomic_store): Likewise.
2973         (dump_gimple_mem_ops): Likewise.
2974         (pp_gimple_stmt_1): Likewise.
2975         (pp_cfg_jump): Likewise.
2976         (dump_implicit_edges): Likewise.
2977         (gimple_dump_bb_for_graph): Likewise.
2978         * graph.c (draw_cfg_node): Likewise.
2979         * langhooks.c (lhd_print_error_function): Likewise.
2980         * sched-vis.c (print_exp): Likewise.
2981         (print_value): Likewise.
2982         (print_pattern): Likewise.
2983         (print_insn): Likewise.
2984         (rtl_dump_bb_for_graph): Likewise.
2985         * tree-pretty-print.c (dump_function_declaration): Likewise.
2986         (dump_array_domain): Likewise.
2987         (dump_omp_clause): Likewise.
2988         (dump_location): Likewise.
2989         (dump_generic_node): Likewise.
2990         (print_struct_decl): Likewise.
2991         * diagnostic.c (diagnostic_show_locus): Use pp_space.
2993 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
2995         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2996         candidate table when replacing a candidate statement.
2997         (replace_rhs_if_not_dup): Likewise.
2998         (replace_one_candidate): Likewise.
3000 2013-08-02  Jan Hubicka  <jh@suse.cz>
3001             Martin Liska  <marxin.liska@gmail.com>
3003         * cgraphunit.c (add_new_function): Fix logic when adding from
3004         late IPA pass.
3005         (assemble_thunk): Rename to ...
3006         (expand_thunk); .. this one; export; get it working with
3007         general functions; make produced gimple valid.
3008         * cgraph.h (expand_thunk): Declare.
3010 2013-08-02  Jan Hubicka  <jh@suse.cz>
3012         * ipa-cp.c (gather_context_independent_values): Use
3013         ipa_get_param_move_cost.
3014         (get_replacement_map): Remove PARAM; move parameter folding
3015         into tree-inline.c
3016         (create_specialized_node): Update.
3017         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3018         assert that we have gimple body; update move_cost.
3019         (count_formal_params): Assert that we have gimple body.
3020         (ipa_dump_param): New function.
3021         (ipa_alloc_node_params): Break out from ...
3022         (ipa_initialize_node_params): ... here.
3023         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3024         (ipa_write_node_info): Stream move costs.
3025         (ipa_read_node_info): Read move costs.
3026         (ipa_update_after_lto_read): Do not recompute node params.
3027         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3028         (ipa_get_param): Check we are not in WPA.
3029         (ipa_get_param_move_cost): New.
3030         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3031         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3032         parm numbers to be present.
3034 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3036         PR rtl-optimization/58048
3037         * lra-constraints.c (process_alt_operands): Don't check asm
3038         operand on register.
3040 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
3042         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3043         the implied StoreLoad barrier for atomic operations if before.
3045 2013-08-02  Jan Hubicka  <jh@suse.cz>
3046             Martin Liska  <marxin.liska@gmail.com>
3048         * cgraph.c (cgraph_function_body_availability): Do not check
3049         cgraph flags.
3050         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3051         symtab_node_availability): Declare.
3052         * ipa.c (can_replace_by_local_alias): New.
3053         (function_and_variable_visibility): Use it.
3054         * symtab.c (symtab_for_node_and_aliases,
3055         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3057 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3059         PR rtl-optimization/57963
3060         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3061         (lra_constraints): Use them.
3063 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
3065         * config/arm/types.md (define_attr "type"): Add "load_acq"
3066         and "store_rel".
3067         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3068         changes.
3069         (cortex_a53_store1): Likewise.
3071 2013-08-01  Jan Hubicka  <jh@suse.cz>
3073         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3074         partitions are not needed.
3076 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
3078         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3079         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3080         MAYBE_NON_Q_CLASS_P where appropriate.
3082 2013-08-01  Jan Hubicka  <jh@suse.cz>
3084         * cgraph.h (release_function_body): Declare.
3085         * tree.c (free_lang_data_in_decl): Free, parameters and return values
3086         of unused delcarations.
3088 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3090         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3091         RTL form when subtracting a constant.
3093 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3095         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3096         Generate canonical plus rtx with negated immediate instead of minus
3097         where appropriate.
3098         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3100 2013-08-01  Jan Hubicka  <jh@suse.cz>
3102         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3103         (cgraph_release_function_body): Likewise.
3104         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3105         * cgraph.h (cgrpah_node): Rename abstract_and_needed
3106         to used_as_abstract_origin.
3107         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3108         symbols used as abstract origins.
3109         * cgraphunit.c (analyze_functions): Update.
3110         * ipa.c (symtab_remove_unreachable_nodes): Recompute
3111         used_as_abstract_origin.
3112         * tree-inline.c (tree_function_versioning): Update
3113         used_as_abstract_origin; be ready for DECL_RESULT and
3114         DECL_ARGUMENTS to be NULL.
3116         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3117         for abstract functions.
3118         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3119         real symbols.
3121 2013-08-01  Jan Hubicka  <jh@suse.cz>
3123         * profile.c (compute_value_histograms): Fix thinko.
3125 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3127         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
3128         aarch-common-protos.h to extra_headers.
3129         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3130         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3131         * config/aarch64/t-aarch64 (aarch-common.o): Define.
3133 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3135         * config/aarch64/aarch64.md (define_attr "type"): Delete.
3136         Include "../arm/types.md".  Define "type" attribute for all patterns.
3137         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3138         attribute changes.
3140 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
3142         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3143         to support power8 load fusion.
3144         (fusion_gpr_mem_load): Likewise.
3146         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3148         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3149         declarations for power8 load fusion.
3150         (emit_fusion_gpr_load): Likewise.
3152         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3153         tuning for power8, turn on fusion mode by default.  Turn on sign
3154         extending fusion mode if normal fusion mode is on, and we are at
3155         -O2 or -O3.
3156         (fusion_gpr_load_p): New function, return true if we can fuse an
3157         addis instruction with a dependent load to a GPR.
3158         (emit_fusion_gpr_load): Emit the instructions for power8 load
3159         fusion to GPRs.
3161         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3162         (VSX load fusion peepholes): New peepholes to fuse together an
3163         addi instruction with a VSX load instruction.
3165         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3166         peepholes to fuse an addis instruction with a load to a GPR base
3167         register.  If we are supporting sign extending fusions, convert
3168         sign extending loads to zero extending loads and add an explicit
3169         sign extension.
3171 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3173         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
3174         aarch-common-protos.h to extra_headers.
3175         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3176         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3177         (arm_early_store_addr_dep): Likewise.
3178         (arm_no_early_alu_shift_dep): Likewise.
3179         (arm_no_early_alu_shift_value_dep): Likewise.
3180         (arm_no_early_mul_dep): Likewise.
3181         (arm_no_early_store_addr_dep): Likewise.
3182         (arm_mac_accumulator_is_mul_result): Likewise.
3183         (arm_mac_accumulator_is_result): Likewise.
3184         * config/arm/aarch-common.c: ... here.  New file.
3185         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3186         here to ...
3187         (arm_early_store_addr_dep): Likewise.
3188         (arm_no_early_alu_shift_dep): Likewise.
3189         (arm_no_early_alu_shift_value_dep): Likewise.
3190         (arm_no_early_mul_dep): Likewise.
3191         (arm_no_early_store_addr_dep): Likewise.
3192         (arm_mac_accumulator_is_mul_result): Likewise.
3193         (arm_mac_accumulator_is_result): Likewise.
3194         * config/arm/aarch-common-protos.h: ... here.  New file.
3195         * config/arm/t-arm (aarch-common.o): Define.
3197 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3199         * config/arm/arm.md: Include new file "types.md".
3200         (define_attr "type"): Move from here to ...
3201         (define_attr "mul32"): Likewise.
3202         (define_attr "mul64"): Likewise.
3203         * config/arm/types.md: ... here.  New file.
3205 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
3207         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
3208         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
3210 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3212         * gen-pass-instances.awk: Fix offset of substr().
3214 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
3216         * Makefile.in (pass-instances.def): New.
3217         (passes.o): Replace dependency on passes.def with one on
3218         pass-instances.def
3220         * gen-pass-instances.awk: New.
3222         * passes.c (pass_manager::pass_manager): Use pass-instances.def
3223         rather than passes.def, updating local definition of NEXT_PASS
3224         macro to add an extra NUM parameter (currently unused).
3226 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
3228         * Makefile.in (PASS_MANAGER_H): New.
3229         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
3230         (passes.o): Likewise.
3231         (statistics.o): Likewise.
3232         (cgraphunit.o): Likewise.
3233         (context.o): Depend on PASS_MANAGER_H.
3235         * pass_manager.h: New.
3237         * cgraphunit.c (cgraph_add_new_function): Update for moves
3238         of globals to fields of pass_manager.
3239         (analyze_function): Likewise.
3240         (expand_function): Likewise.
3241         (ipa_passes): Likewise.
3242         (compile): Likewise.
3244         * context.c (context::context): New.
3245         * context.h  (context::context): New.
3246         (context::get_passes): New.
3247         (context::passes_): New.
3249         * lto-cgraph.c (input_node): Update for moves of globals to
3250         fields of pass_manager.
3252         * passes.c (all_passes): Remove, in favor of a field of the
3253         same name within the new class pass_manager.
3254         (all_small_ipa_passes): Likewise.
3255         (all_lowering_passes): Likewise.
3256         (all_regular_ipa_passes): Likewise.
3257         (all_late_ipa_passes): Likewise.
3258         (all_lto_gen_passes): Likewise.
3259         (passes_by_id): Likewise.
3260         (passes_by_id_size): Likewise.
3261         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3262         the new class pass_manager.
3263         (set_pass_for_id): Convert to...
3264         (pass_manager::set_pass_for_id): ...method.
3265         (get_pass_for_id): Convert to...
3266         (pass_manager::get_pass_for_id): ...method.
3267         (register_one_dump_file): Move body of implementation into...
3268         (pass_manager::register_one_dump_file): ...here.
3269         (register_dump_files_1): Convert to...
3270         (pass_manager::register_dump_files_1): ...method.
3271         (register_dump_files): Convert to...
3272         (pass_manager::register_dump_files): ...method.
3273         (create_pass_tab): Update for moves of globals to fields of
3274         pass_manager.
3275         (dump_passes): Move body of implementation into...
3276         (pass_manager::dump_passes): ...here.
3277         (register_pass): Move body of implementation into...
3278         (pass_manager::register_pass): ...here.
3279         (init_optimization_passes): Convert into...
3280         (pass_manager::pass_manager): ...constructor for new
3281         pass_manager class, and initialize the pass_lists array.
3282         (check_profile_consistency): Update for moves of globals to
3283         fields of pass_manager.
3284         (dump_profile_report): Move body of implementation into...
3285         (pass_manager::dump_profile_report): ...here.
3286         (ipa_write_summaries_1): Update for moves of pass lists from
3287         being globals to fields of pass_manager.
3288         (ipa_write_optimization_summaries): Likewise.
3289         (ipa_read_summaries):  Likewise.
3290         (ipa_read_optimization_summaries): Likewise.
3291         (execute_all_ipa_stmt_fixups): Likewise.
3293         * statistics.c (statistics_fini): Update for moves of globals to
3294         fields of pass_manager.
3296         * toplev.c (general_init): Replace call to
3297         init_optimization_passes with construction of the pass_manager
3298         instance.
3300         * tree-pass.h (all_passes): Remove, in favor of a field of the
3301         same name within the new class pass_manager.
3302         (all_small_ipa_passes): Likewise.
3303         (all_lowering_passes): Likewise.
3304         (all_regular_ipa_passes): Likewise.
3305         (all_lto_gen_passes): Likewise.
3306         (all_late_ipa_passes): Likewise.
3307         (passes_by_id): Likewise.
3308         (passes_by_id_size): Likewise.
3309         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3310         the new class pass_manager.
3311         (get_pass_for_id): Remove.
3313 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
3315         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
3316         configs.
3318 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
3320         * arm.md (mulhi3): New expand pattern.
3322 2013-07-30  Jan Hubicka  <jh@suse.cz>
3323             Martin Liska  <marxin.liska@gmail.com>
3325         * profile.c (compute_value_histograms): Do not ICE when
3326         there is mismatch only on some counters.
3328 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3330         PR rtl-optimization/57637
3331         * function.c (move_insn_for_shrink_wrap): Also check the
3332         GEN set of the LIVE problem for the liveness analysis
3333         if it exists, otherwise give up.
3335 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3337         PR tree-optimization/57993
3338         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
3339         replaced statement in the candidate table.
3340         (phi_add_costs): Return infinite cost when the hidden basis does
3341         not dominate all phis on which the candidate is dependent.
3342         (replace_one_candidate): Record replaced statement in the
3343         candidate table.
3345 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
3347         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
3348         (ashlv2si3): New expander.
3349         (*ashlv2si3_i): New define_insn_and_split.
3350         * predicates.md (float_operation): Allow patterns with three
3351         basic sub-patterns.
3353         PR rtl-optimization/58021
3354         * mode-switching.c (create_pre_exit): Always split off preceding
3355         insns if we are not at the basic block head.
3357 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
3359         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
3360         (UCLIBC_DYNAMIC_LINKER): New macro.
3361         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
3362         `-mnan=2008'.
3363         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
3364         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
3365         `-mnan=2008'.
3366         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
3367         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
3368         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
3369         for SF and DF modes.  Use ieee_quad_format for TF mode.
3370         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
3371         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
3372         (mips_option_override): Handle `-mnan=legacy'.
3373         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
3374         `-mabs=2008' and `-mnan=2008'.
3375         (OPTION_DEFAULT_SPECS): Add "nan" default.
3376         (ASM_SPEC): Handle `-mnan='.
3377         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
3378         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
3379         comment accordingly.
3380         (neg<mode>2): Likewise.
3381         * config/mips/mips.opt (mabs, mnan): New options.
3382         * doc/install.texi (Configuration): Document `--with-nan=' option.
3383         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
3384         `-mnan=' options.
3385         (MIPS Options): Document them.
3386         * config.gcc <mips*-*-*>: Handle `--with-nan='.
3387         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
3388         * configure: Regenerate.
3389         * config.in: Regenerate.
3391 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
3393         * config/i386/i386.md (float post-reload splitters): Do not check
3394         for subregs of SSE registers.
3396 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
3397             H.J. Lu  <hongjiu.lu@intel.com>
3399         PR target/57954
3400         PR target/57988
3401         * config/i386/i386.md (post-reload splitter
3402         to avoid partial SSE reg dependency stalls): New pattern.
3404 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
3406         * config/s390/s390.md ("movcc"): Swap load and store instructions.
3408 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
3410         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
3411         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
3413 2013-07-26  Cary Coutant  <ccoutant@google.com>
3415         * dwarf2out.c (die_checksum_ordered): Don't include template
3416         instantiations in signature.
3417         (is_template_parameter): New function.
3418         (is_template_instantiation): New function.
3419         (generate_skeleton_bottom_up): Don't include template instantiations
3420         in type unit DIE.
3421         (generate_skeleton): Likewise.
3422         (break_out_comdat_types): Move recursive call to break out nested
3423         types earlier.
3424         (prune_unused_types_mark_generic_parms_dies): Call
3425         is_template_parameter.
3427 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
3429         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
3430         uses vector registers.
3431         * config/aarch64/iterators.md: Add attributes rtn and vas.
3433 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3434             Richard Earnshaw  <richard.earnshaw@arm.com>
3436         * combine.c (simplify_comparison): Re-canonicalize operands
3437         where appropriate.
3438         * config/arm/arm.md (movcond_addsi): New splitter.
3440 2013-07-25  Sterling Augustine  <saugustine@google.com>
3442         * dwarf2out.c (size_of_pubnames): Move code to...
3443         (include_pubname_in_output): ...here.  New.
3444         (want_pubnames): Rearrange.
3445         (output_pubnames): Call include_pubname_in_output.  Move assertion.
3447 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
3449         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
3451 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
3453         PR target/38836
3454         * doc/extend.texi: Remove obsolete builtins. Fix
3455         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
3457 2013-07-25  Jan Hubicka  <jh@suse.cz>
3459         * cgraph.c (release_function_body): Break out from ...
3460         (cgraph_release_function_body): ... this one; also release DECL_RESULT
3461         and DECL_ARGUMENTS.
3462         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
3463         old_tree in the map.
3464         (create_specialized_node): Update.
3465         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
3466         into index.
3467         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
3468         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
3469         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
3470         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
3471         DECL_RESULT.
3473 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3475         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
3476         addsi3_carryin_alt2_<optab>): Correct output template.
3478 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3480         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
3481         Adjust for arm_restrict_it.
3482         Remove trailing whitespace.
3484 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
3486         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
3487         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
3489         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
3491 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
3493         PR rtl-optimization/57960
3494         * lra-constraints.c (process_alt_operands): Use the right mode
3495         when checking strict_low.
3497 2013-07-25  Jan Hubicka  <jh@suse.cz>
3499         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
3500         * cgraph.c (cgraph_remove_node): Do not release function body
3501         when in cgraph streaming.
3502         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
3503         in other partitions are not considered reachable; fix handling of
3504         clones.
3506 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3508         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
3510 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3512         PR target/19599
3513         PR target/57731
3514         PR target/57837
3515         * config/arm/arm.md ("*sibcall_insn): Replace use of
3516         Ss with US. Adjust output for v5 and v4t.
3517         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
3519         * config/arm/constraints.md ("Ss"): Rename to US.
3521 2013-07-25  Terry Guo  <terry.guo@arm.com>
3523         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
3524         shift_add/shift_sub0/shift_sub1 RTXs.
3526 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
3527             Anton Blanchard <anton@au1.ibm.com>
3529         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
3530         (altivec_vpks<VI_char>ss): Likewise.
3531         (altivec_vpks<VI_char>us): Likewise.
3532         (altivec_vpku<VI_char>us): Likewise.
3533         (altivec_vpku<VI_char>um): Likewise.
3535 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
3537         Introduce context class.
3539         * Makefile.in (CONTEXT_H): New.
3540         (OBJS): Add context.o.
3541         (toplev.o): Add CONTEXT_H to dependencies.
3542         (context.o): New.
3544         * toplev.c (general_init): Create the singleton gcc::context instance.
3546         * context.c: New.
3548         * context.h: New.
3550 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
3552         PR rtl-optimization/57968
3553         * mode-switching.c (create_pre_exit): Allow instructions that
3554         don't set a return register to need a non-exit mode.
3556 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3557             Anton Blanchard <anton@au1.ibm.com>
3559         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
3560         operands to vperm for little endian.
3561         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
3562         of lvsl to create the control mask for a vperm for little endian.
3564 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3565             Anton Blanchard <anton@au1.ibm.com>
3567         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
3568         two operands for little-endian.
3570 2013-07-23  Steve Ellcey  <sellcey@mips.com>
3572         * config/mips/mips.c (mips_case_values_threshold): New.
3573         (TARGET_CASE_VALUES_THRESHOLD): Define.
3575 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3576             Anton Blanchard <anton@au1.ibm.com>
3578         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
3579         selection of field for vector splat in little endian mode.
3581 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
3583         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
3584         expanders to rs6000.md.
3585         (ior<mode>3): Likewise.
3586         (and<mode>3): Likewise.
3587         (one_cmpl<mode>2): Likewise.
3588         (nor<mode>3): Likewise.
3589         (andc<mode>3): Likewise.
3590         (eqv<mode>3): Likewise.
3591         (nand<mode>3): Likewise.
3592         (orc<mode>3): Likewise.
3594         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
3595         declaration.
3597         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
3598         to split multi-word logical operations.
3599         (rs6000_split_logical_di): Likewise.
3600         (rs6000_split_logical): Likewise.
3602         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
3603         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
3604         and allow TImode operations in 32-bit.
3605         (vsx_and<mode>3_64bit): Likewise.
3606         (vsx_ior<mode>3_32bit): Likewise.
3607         (vsx_ior<mode>3_64bit): Likewise.
3608         (vsx_xor<mode>3_32bit): Likewise.
3609         (vsx_xor<mode>3_64bit): Likewise.
3610         (vsx_one_cmpl<mode>2_32bit): Likewise.
3611         (vsx_one_cmpl<mode>2_64bit): Likewise.
3612         (vsx_nor<mode>3_32bit): Likewise.
3613         (vsx_nor<mode>3_64bit): Likewise.
3614         (vsx_andc<mode>3_32bit): Likewise.
3615         (vsx_andc<mode>3_64bit): Likewise.
3616         (vsx_eqv<mode>3_32bit): Likewise.
3617         (vsx_eqv<mode>3_64bit): Likewise.
3618         (vsx_nand<mode>3_32bit): Likewise.
3619         (vsx_nand<mode>3_64bit): Likewise.
3620         (vsx_orc<mode>3_32bit): Likewise.
3621         (vsx_orc<mode>3_64bit): Likewise.
3623         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
3624         logical types in GPRs.
3626         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
3627         logical insns to rs6000.md, and allow TImode operations in
3628         32-bit.
3629         (altivec_ior<mode>3): Likewise.
3630         (altivec_xor<mode>3): Likewise.
3631         (altivec_one_cmpl<mode>2): Likewise.
3632         (altivec_nor<mode>3): Likewise.
3633         (altivec_andc<mode>3): Likewise.
3635         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
3636         attributes for moving the 128-bit logical operations into
3637         rs6000.md.
3638         (BOOL_REGS_OUTPUT): Likewise.
3639         (BOOL_REGS_OP1): Likewise.
3640         (BOOL_REGS_OP2): Likewise.
3641         (BOOL_REGS_UNARY): Likewise.
3642         (BOOL_REGS_AND_CR0): Likewise.
3643         (one_cmpl<mode>2): Add support for DI logical operations on
3644         32-bit, splitting the operations to 32-bit.
3645         (anddi3): Likewise.
3646         (iordi3): Likewise.
3647         (xordi3): Likewise.
3648         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
3649         changes to combine the 32/64-bit code, allow logical operations on
3650         TI mode in 32-bit, and to use similar match_operator patterns like
3651         scalar mode uses.  Combine the Altivec and VSX code for logical
3652         operations, and move it here.
3653         (ior<mode>3, 128-bit types): Likewise.
3654         (xor<mode>3, 128-bit types): Likewise.
3655         (one_cmpl<mode>3, 128-bit types): Likewise.
3656         (nor<mode>3, 128-bit types): Likewise.
3657         (andc<mode>3, 128-bit types): Likewise.
3658         (eqv<mode>3, 128-bit types): Likewise.
3659         (nand<mode>3, 128-bit types): Likewise.
3660         (orc<mode>3, 128-bit types): Likewise.
3661         (and<mode>3_internal): Likewise.
3662         (bool<mode>3_internal): Likewise.
3663         (boolc<mode>3_internal1): Likewise.
3664         (boolc<mode>3_internal2): Likewise.
3665         (boolcc<mode>3_internal1): Likewise.
3666         (boolcc<mode>3_internal2): Likewise.
3667         (eqv<mode>3_internal1): Likewise.
3668         (eqv<mode>3_internal2): Likewise.
3669         (one_cmpl1<mode>3_internal): Likewise.
3671 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3673         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3674         Rename flag_stack_usage to flag_stack_usage_info.
3676 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3678         * config/microblaze/sync.md: New file.
3679         * config/microblaze/microblaze.md: Include sync.md
3680         * config/microblaze/microblaze.c: Add print_operand 'y'.
3681         * config/microblaze/constraints.md: Add memory_contraint
3682         'Q' which is a single register.
3684 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
3686         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
3688 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3690         * reload.c (find_reloads): Exit loop once we find this operand
3691         cannot be reloaded somehow for this alternative.
3693         * reload.c (find_reloads): Exit loop once we find a hard register.
3695         * rtlanal.c (computed_jump_p): Exit loop once we find label
3696         reference is used.
3698         * i386.c (ix86_pad_returns): Exit loop after setting replace.
3700         * cfgloopmanip.c (remove_path): Exit loop after setting
3701         irred_invalidated.
3703         * gensupport.c (subst_dup): Avoid loop if code is not
3704         MATCH_DUP nor MATCH_OP_DUP.
3706 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
3708         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
3710 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3712         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
3713         true for SP_REGNUM if mode == ptr_mode.
3714         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
3715         with value R0_REGNUM + 31.
3717 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3719         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
3720         pad pointer-typed argument downward.
3722 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3724         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
3725         and __ILP32__ when the ILP32 model is in use.
3727 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3729         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
3730         (aarch64_load_symref_appropriately): In the case of
3731         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
3732         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
3733         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
3734         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
3735         if 'mode' doesn't equal to 'ptr_mode'.
3736         (aarch64_output_mi_thunk): Add an assertion on the alignment of
3737         'vcall_offset'; change to call aarch64_emit_move differently depending
3738         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
3739         to calculate the upper bound of 'vcall_offset'.
3740         (aarch64_cannot_force_const_mem): Change to also return true if
3741         mode != ptr_mode.
3742         (aarch64_legitimize_reload_address): In the case of large
3743         displacements, add new local variable 'xmode' and an assertion
3744         based on it; change to use 'xmode' to generate the new rtx and
3745         reload.
3746         (aarch64_asm_trampoline_template): Change to generate the template
3747         differently depending on TARGET_ILP32 or not; change to use
3748         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
3749         (aarch64_trampoline_size): Removed.
3750         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
3751         and replace immediate literals with it.  Change to use 'ptr_mode'
3752         instead of 'DImode' and call convert_memory_address if the mode
3753         of 'fnaddr' doesn't equal to 'ptr_mode'.
3754         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
3755         to output symbol.
3756         (aarch64_elf_asm_destructor): Likewise.
3757         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
3758         on TARGET_ILP32 instead of aarch64_trampoline_size.
3759         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
3760         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
3761         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
3762         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
3763         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
3764         (storewb_pair<GPI:mode>_<P:mode>): ... this.
3765         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
3766         depending on the value of 'mode'.
3767         (add_losym_<mode>): New.
3768         (ldr_got_small_<mode>): New, based on ldr_got_small.
3769         (ldr_got_small): Remove.
3770         (ldr_got_small_sidi): New.
3771         * config/aarch64/iterators.md (P): New.
3772         (PTR): Change to 'ptr_mode' in the condition.
3774 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3776         * config.gcc (aarch64*-*-*): Support --with-abi.
3777         (aarch64*-*-elf): Support --with-multilib-list.
3778         (aarch64*-*-linux*): Likewise.
3779         (supported_defaults): Add abi to aarch64*-*-*.
3780         * configure.ac: Mention AArch64 for --with-multilib-list.
3781         * configure: Re-generated.
3782         * config/aarch64/biarchilp32.h: New file.
3783         * config/aarch64/biarchlp64.h: New file.
3784         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
3785         (ABI_SPEC): Ditto.
3786         (MULTILIB_DEFAULTS): Ditto.
3787         (DRIVER_SELF_SPECS): Ditto.
3788         (ASM_SPEC): Update to also substitute -mabi.
3789         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
3790         file whose name depends on -mabi= and -mbig-endian.
3791         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
3792         TARGET_ILP32.
3793         (POINTER_SIZE): New define.
3794         (POINTERS_EXTEND_UNSIGNED): Ditto.
3795         (enum aarch64_abi_type): New enumeration tag.
3796         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
3797         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
3798         (TARGET_ILP32): New define.
3799         * config/aarch64/aarch64.opt (mabi): New.
3800         (aarch64_abi): New.
3801         (ilp32, lp64): New values for -mabi.
3802         * config/aarch64/t-aarch64 (comma): New define.
3803         (MULTILIB_OPTIONS): Ditto.
3804         (MULTILIB_DIRNAMES): Ditto.
3805         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
3806         * doc/invoke.texi: Document -mabi for AArch64.
3808 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
3810         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
3812 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3813             Anton Blanchard <anton@au1.ibm.com>
3815         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
3816         endianness when selecting field to splat.
3818 2013-07-22  Eric Christopher  <echristo@gmail.com>
3820         * dwarf2out.c (die_odr_checksum): New function to use
3821         CHECKSUM_ macros and ULEB128 for DIE tag.
3822         (generate_type_signature): Use.
3824 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
3826         * config.gcc (sparc*-*-*): Accept leon3 processor.
3827         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
3828         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
3829         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
3830         * config/sparc/sparc.opt (enum processor_type): Add leon3.
3831         (mfix-ut699): Adjust comment.
3832         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
3833         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
3834         (CPP_CPU_SPEC): Likewise.
3835         (ASM_CPU_SPEC): Likewise.
3836         * config/sparc/sparc.c (leon3_cost): New constant.
3837         (sparc_option_override): Add leon3 support.
3838         (mem_ref): New function.
3839         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
3840         (sparc_do_work_around_errata): Look into the instruction in the delay
3841         slot and adjust accordingly.  Add fix for the data cache nullify issues
3842         of the UT699.  Change insertion position for the NOP.
3843         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
3844         (leon3_load): New reservation.
3845         (leon_store): Bump latency to 2.
3846         (grfpu): New automaton.
3847         (grfpu_alu): New unit.
3848         (grfpu_ds): Likewise.
3849         (leon_fp_alu): Adjust.
3850         (leon_fp_mult): Delete.
3851         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
3852         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
3853         * config/sparc/sparc.md (cpu): Add leon3.
3854         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
3855         (swapsi): Likewise.
3856         (atomic_test_and_set): Likewise.
3857         (ldstub): Likewise.
3859 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
3861         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
3862         default for R5900 targets.
3863         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
3864         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
3865         * config/mips/mips.c (mips_option_override): Report an error for
3866         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
3867         for -march=r5900 -mhard-float.
3869 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3871         * df-problems.c (can_move_insns_across): Exit loop once we
3872         find a non-fixed, non-global register.
3874         * ipa-pure-const.c (propagate_nothrow): Exit loop after
3875         setting can_throw.
3877         * omega.c (omega_eliminate_red): Break after setting red_found.
3878         (omega_problem_has_red_equations): Similarly after setting found.
3879         (omega_query_variable): Similarly after setting coupled.
3881 2013-07-22  Marek Polacek  <polacek@redhat.com>
3883         * gimplify.c: Don't include gimple.h twice.
3885 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3887         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
3888         instead of TARGET_THUMB1.
3889         (Pz): New constraint.
3890         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
3891         encodings.
3892         (compare_negsi_si): Likewise.
3893         (compare_addsi2_op0): Likewise.
3894         (compare_addsi2_op1): Likewise.
3895         (addsi3_carryin_<optab>): Likewise.
3896         (addsi3_carryin_alt2_<optab>): Likewise.
3897         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
3898         for arm_restrict_it.
3899         (subsi3_carryin): Likewise.
3900         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
3901         (minmax_arithsi): Disable for arm_restrict_it.
3902         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
3903         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
3904         (satsi_<SAT:code>_shift): Likewise.
3905         (arm_shiftsi3): Add alternative for 16-bit encoding.
3906         (arm32_movhf): Disable for arm_restrict_it.
3907         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
3908         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
3910 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
3912         * config/arm/arm.md (attribute "insn"): Delete.
3913         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
3914         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
3915         (not_shiftsi): Update for attribute change.
3916         (not_shiftsi_compare0): Likewise.
3917         (not_shiftsi_compare0_scratch): Likewise.
3918         (arm_one_cmplsi2): Likewise.
3919         (thumb1_one_cmplsi2): Likewise.
3920         (notsi_compare0): Likewise.
3921         (notsi_compare0_scratch): Likewise.
3922         (thumb1_movdi_insn): Likewise.
3923         (arm_movsi_insn): Likewise.
3924         (movhi_insn_arch4): Likewise.
3925         (movhi_bytes): Likewise.
3926         (arm_movqi_insn): Likewise.
3927         (thumb1_movqi_insn): Likewise.
3928         (arm32_movhf): Likewise.
3929         (thumb1_movhf): Likewise.
3930         (arm_movsf_soft_insn): Likewise.
3931         (thumb1_movsf_insn): Likewise.
3932         (thumb_movdf_insn): Likewise.
3933         (movsicc_insn): Likewise.
3934         (movsfcc_soft_insn): Likewise.
3935         (and_scc): Likewise.
3936         (cond_move): Likewise.
3937         (if_move_not): Likewise.
3938         (if_not_move): Likewise.
3939         (if_shift_move): Likewise.
3940         (if_move_shift): Likewise.
3941         (if_shift_shift): Likewise.
3942         (if_not_arith): Likewise.
3943         (if_arith_not): Likewise.
3944         (cond_move_not): Likewise.
3945         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3946         (neon_mov<mode>): Likewise.
3947         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3948         (thumb2_movsi_vfp): Likewise.
3949         (movsf_vfp): Likewise.
3950         (thumb2_movsf_vfp): Likewise.
3951         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3952         change.
3953         (cortexa7_older_only): Likewise.
3954         (cortexa7_younger): Likewise.
3955         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3956         (1020alu_shift_op): Likewise.
3957         (1020alu_shift_reg_op): Likewise.
3958         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3959         (alu_shift_op): Likewise.
3960         (alu_shift_reg_op): Likewise.
3961         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3962         (11_alu_shift_op): Likewise.
3963         (11_alu_shift_reg_op): Likewise.
3964         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3965         (9_alu_shift_reg_op): Likewise.
3966         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3967         change.
3968         (cortex_a15_alu_shift): Likewise.
3969         (cortex_a15_alu_shift_reg): Likewise.
3970         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3971         (cortex_a5_alu_shift): Likewise.
3972         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3973         change.
3974         (cortex_a53_alu_shift): Likewise.
3975         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3976         change.
3977         (cortex_a7_alu_reg): Likewise.
3978         (cortex_a7_alu_shift): Likewise.
3979         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3980         (cortex_a8_alu_shift): Likewise.
3981         (cortex_a8_alu_shift_reg): Likewise.
3982         (cortex_a8_mov): Likewise.
3983         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3984         (cortex_a9_dp_shift): Likewise.
3985         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3986         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3987         (cortex_r4_mov): Likewise.
3988         (cortex_r4_alu_shift): Likewise.
3989         (cortex_r4_alu_shift_reg): Likewise.
3990         * config/arm/fa526.md (526_alu_op): Update for attribute change.
3991         (526_alu_shift_op): Likewise.
3992         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3993         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3994         (626te_alu_shift_op): Likewise.
3995         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3996         (726te_alu_op): Likewise.
3997         (726te_alu_shift_op): Likewise.
3998         (726te_alu_shift_reg_op): Likewise.
3999         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4000         (mp626_alu_shift_op): Likewise.
4001         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4002         (pj4_alu_e1_conds): Likewise.
4003         (pj4_alu): Likewise.
4004         (pj4_alu_conds): Likewise.
4005         (pj4_shift): Likewise.
4006         (pj4_shift_conds): Likewise.
4007         (pj4_alu_shift): Likewise.
4008         (pj4_alu_shift_conds): Likewise.
4010 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4012         * config/arm/predicates.md (shiftable_operator_strict_it):
4013         New predicate.
4014         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4015         Disable cond_exec version for arm_restrict_it.
4016         (thumb2_smaxsi3): Convert to generate cond_exec.
4017         (thumb2_sminsi3): Likewise.
4018         (thumb32_umaxsi3): Likewise.
4019         (thumb2_uminsi3): Likewise.
4020         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4021         (thumb2_neg_abssi2): Likewise.
4022         (thumb2_mov_scc): Add alternative for 16-bit encoding.
4023         (thumb2_movsicc_insn): Adjust alternatives.
4024         (thumb2_mov_negscc): Disable for arm_restrict_it.
4025         (thumb2_mov_negscc_strict_it): New pattern.
4026         (thumb2_mov_notscc_strict_it): New pattern.
4027         (thumb2_mov_notscc): Disable for arm_restrict_it.
4028         (thumb2_ior_scc): Likewise.
4029         (thumb2_ior_scc_strict_it): New pattern.
4030         (thumb2_cond_move): Adjust for arm_restrict_it.
4031         (thumb2_cond_arith): Disable for arm_restrict_it.
4032         (thumb2_cond_arith_strict_it): New pattern.
4033         (thumb2_cond_sub): Adjust for arm_restrict_it.
4034         (thumb2_movcond): Likewise.
4035         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4036         (thumb2_zero_extendhisi2_v6): Likewise.
4037         (thumb2_zero_extendqisi2_v6): Likewise.
4038         (orsi_notsi_si): Likewise.
4039         (orsi_not_shiftsi_si): Likewise.
4041 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
4043         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4044         instruction sequence is 1 byte shorter.
4046 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
4048         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4049         it is not needed after split.
4051 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
4053         PR target/51784
4054         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4055         second label for nonlocal goto receivers. Don't output pic base labels
4056         unless we're producing PIC; mark that action unreachable().
4057         (ix86_save_reg): If the function contains a nonlocal label, save the
4058         PIC base reg.
4059         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4060         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4061         (update_pic_label_number_if_needed): New.
4062         (machopic_output_function_base_name): Adjust for nonlocal receiver
4063         case.
4064         (machopic_should_output_picbase_label): New.
4065         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4066         (nonlocal_goto_receiver): New insn and split.
4068 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
4070         * config/aarch64/aarch64-builtins.c
4071         (aarch64_fold_builtin): Fold abs in all modes.
4072         * config/aarch64/aarch64-simd-builtins.def
4073         (abs): Enable for all modes.
4074         * config/aarch64/arm_neon.h
4075         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4076         (vabs_f64): Add missing intrinsic.
4078 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
4080         * config/aarch64/arm_neon.h (vabs_s64): New function
4082 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
4084         PR target/57516
4085         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4086         * config/avr/avr.md (adjust_len): Add `round'.
4087         * config/avr/avr-protos.h (avr_out_round): New prototype.
4088         (avr_out_plus): Add `out_label' argument.
4089         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4090         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4091         Handle the case where `insn' is just a pattern.
4092         (avr_out_bitop): Handle the case where `insn' is just a pattern.
4093         (avr_out_round): New function.
4094         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4096 2013-07-18  David Holsgrove <david.holsgrove@xilinx.com>
4098         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4099         Add check for flag_stack_usage to handle -fstack-usage support
4101 2013-07-18  Pat Haugen <pthaugen@us.ibm.com>
4103         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4104         interaction for new Power8 flags and VSX.
4106 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
4108         PR middle-end/57698
4109         * tree-inline.c (expand_call_inline): Emit errors during
4110         early_inlining only if optimization is not turned on.
4112 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4114         * passes.def: New.
4116         * passes.c (init_optimization_passes): Move the construction of
4117         the pass hierarchy into a new passes.def file.
4119         * Makefile.in (passes.o): Add dependency on passes.def.
4121 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4123         * passes.c (init_optimization_passes): Introduce macros for
4124         constructing the tree of passes (INSERT_PASSES_AFTER,
4125         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4126         TERMINATE_PASS_LIST).
4128 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
4129             Wei Mi  <wmi@google.com>
4131         PR rtl-optimization/57878
4132         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4133         top.
4134         (reload_pseudo_compare_func): Check nregs first for reload
4135         pseudos.
4137 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4139         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4141 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
4143         * read-rtl.c (validate_const_int): Once an invalid character is
4144         seen, quit the loop.
4146         * gengtype.c (write_roots): Similarly once we find the "deletable"
4147         or "if_marked" option.
4149 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4151         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4152         "xtab" and "sat".  Move value "clz" from here to ...
4153         (attriubte "type"): ... here.
4154         (satsi_<SAT:code>): Delete "insn" attribute.
4155         (satsi_<SAT:code>_shift): Likewise.
4156         (arm_zero_extendqisi2addsi): Likewise.
4157         (arm_extendqisi2addsi): Likewise.
4158         (clzsi2): Update for attribute changes.
4159         (rbitsi2): Likewise.
4160         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4161         attribute.
4162         (arm_usatsihi): Likewise.
4163         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4165 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4167         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4168         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
4169         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
4170         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
4171         in alphabetical order.
4172         (attribute "core_cycles"): Update for attribute changes.
4173         (arm_addsi3): Likewise.
4174         (addsi3_compare0): Likewise.
4175         (addsi3_compare0_scratch): Likewise.
4176         (addsi3_compare_op1): Likewise.
4177         (addsi3_compare_op2): Likewise.
4178         (compare_addsi2_op0): Likewise.
4179         (compare_addsi2_op1): Likewise.
4180         (addsi3_carryin_shift_<optab>): Likewise.
4181         (subsi3_carryin_shift): Likewise.
4182         (rsbsi3_carryin_shift): Likewise.
4183         (arm_subsi3_insn): Likewise.
4184         (subsi3_compare0): Likewise.
4185         (subsi3_compare): Likewise.
4186         (arm_andsi3_insn): Likewise.
4187         (thumb1_andsi3_insn): Likewise.
4188         (andsi3_compare0): Likewise.
4189         (andsi3_compare0_scratch): Likewise.
4190         (zeroextractsi_compare0_scratch
4191         (andsi_not_shiftsi_si): Likewise.
4192         (iorsi3_insn): Likewise.
4193         (iorsi3_compare0): Likewise.
4194         (iorsi3_compare0_scratch): Likewise.
4195         (arm_xorsi3): Likewise.
4196         (thumb1_xorsi3_insn): Likewise.
4197         (xorsi3_compare0): Likewise.
4198         (xorsi3_compare0_scratch): Likewise.
4199         (satsi_<SAT:code>_shift): Likewise.
4200         (rrx): Likewise.
4201         (arm_shiftsi3): Likewise.
4202         (shiftsi3_compare0): Likewise.
4203         (not_shiftsi): Likewise.
4204         (not_shiftsi_compare0): Likewise.
4205         (not_shiftsi_compare0_scratch): Likewise.
4206         (arm_one_cmplsi2): Likewise.
4207         (thumb_one_complsi2): Likewise.
4208         (notsi_compare0): Likewise.
4209         (notsi_compare0_scratch): Likewise.
4210         (thumb1_zero_extendhisi2): Likewise.
4211         (arm_zero_extendhisi2): Likewise.
4212         (arm_zero_extendhisi2_v6): Likewise.
4213         (arm_zero_extendhisi2addsi): Likewise.
4214         (thumb1_zero_extendqisi2): Likewise.
4215         (thumb1_zero_extendqisi2_v6): Likewise.
4216         (arm_zero_extendqisi2): Likewise.
4217         (arm_zero_extendqisi2_v6): Likewise.
4218         (arm_zero_extendqisi2addsi): Likewise.
4219         (thumb1_extendhisi2): Likewise.
4220         (arm_extendhisi2): Likewise.
4221         (arm_extendhisi2_v6): Likewise.
4222         (arm_extendqisi): Likewise.
4223         (arm_extendqisi_v6): Likewise.
4224         (arm_extendqisi2addsi): Likewise.
4225         (thumb1_extendqisi2): Likewise.
4226         (thumb1_movdi_insn): Likewise.
4227         (arm_movsi_insn): Likewise.
4228         (movsi_compare0): Likewise.
4229         (movhi_insn_arch4): Likewise.
4230         (movhi_bytes): Likewise.
4231         (arm_movqi_insn): Likewise.
4232         (thumb1_movqi_insn): Likewise.
4233         (arm32_movhf): Likewise.
4234         (thumb1_movhf): Likewise.
4235         (arm_movsf_soft_insn): Likewise.
4236         (thumb1_movsf_insn): Likewise.
4237         (movdf_soft_insn): Likewise.
4238         (thumb_movdf_insn): Likewise.
4239         (arm_cmpsi_insn): Likewise.
4240         (cmpsi_shiftsi): Likewise.
4241         (cmpsi_shiftsi_swp): Likewise.
4242         (arm_cmpsi_negshiftsi_si): Likewise.
4243         (movsicc_insn): Likewise.
4244         (movsfcc_soft_insn): Likewise.
4245         (arith_shiftsi): Likewise.
4246         (arith_shiftsi_compare0
4247         (arith_shiftsi_compare0_scratch
4248         (sub_shiftsi): Likewise.
4249         (sub_shiftsi_compare0
4250         (sub_shiftsi_compare0_scratch
4251         (and_scc): Likewise.
4252         (cond_move): Likewise.
4253         (if_plus_move): Likewise.
4254         (if_move_plus): Likewise.
4255         (if_move_not): Likewise.
4256         (if_not_move): Likewise.
4257         (if_shift_move): Likewise.
4258         (if_move_shift): Likewise.
4259         (if_shift_shift): Likewise.
4260         (if_not_arith): Likewise.
4261         (if_arith_not): Likewise.
4262         (cond_move_not): Likewise.
4263         (thumb1_ashlsi3): Set type attribute.
4264         (thumb1_ashrsi3): Likewise.
4265         (thumb1_lshrsi3): Likewise.
4266         (thumb1_rotrsi3): Likewise.
4267         (shiftsi3_compare0_scratch): Likewise.
4268         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
4269         (neon_mov<mode>): Likewise.
4270         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
4271         attribute changes.
4272         (thumb2_movsi_insn): Likewise.
4273         (thumb2_cmpsi_neg_shiftsi): Likewise.
4274         (thumb2_extendqisi_v6): Likewise.
4275         (thumb2_zero_extendhisi2_v6): Likewise.
4276         (thumb2_zero_extendqisi2_v6): Likewise.
4277         (thumb2_shiftsi3_short): Likewise.
4278         (thumb2_addsi3_compare0_scratch): Likewise.
4279         (orsi_not_shiftsi_si): Likewise.
4280         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
4281         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
4282         changes.
4283         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
4284         (1020alu_shift_op): Likewise.
4285         (1020alu_shift_reg_op): Likewise.
4286         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
4287         (alu_shift_op): Likewise.
4288         (alu_shift_reg_op): Likewise.
4289         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
4290         (11_alu_shift_op): Likewise.
4291         (11_alu_shift_reg_op): Likewise.
4292         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
4293         (9_alu_shift_reg_op): Likewise.
4294         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4295         changes.
4296         (cortex_a15_alu_shift): Likewise.
4297         (cortex_a15_alu_shift_reg): Likewise.
4298         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
4299         changes.
4300         (cortex_a5_alu_shift): Likewise.
4301         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
4302         changes.
4303         (cortex_a53_alu_shift): Likewise.
4304         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4305         changes.
4306         (cortex_a7_alu_reg): Likewise.
4307         (cortex_a7_alu_shift): Likewise.
4308         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
4309         changes.
4310         (cortex_a8_alu_shift): Likewise.
4311         (cortex_a8_alu_shift_reg): Likewise.
4312         (cortex_a8_mov): Likewise.
4313         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
4314         (cortex_a9_dp_shift): Likewise.
4315         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
4316         changes.
4317         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
4318         changes.
4319         (cortex_r4_mov): Likewise.
4320         (cortex_r4_alu_shift): Likewise.
4321         (cortex_r4_alu_shift_reg): Likewise.
4322         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
4323         (526_alu_shift_op): Likewise.
4324         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
4325         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
4326         (626te_alu_shift_op): Likewise.
4327         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
4328         (726te_alu_op): Likewise.
4329         (726te_alu_shift_op): Likewise.
4330         (726te_alu_shift_reg_op): Likewise.
4331         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
4332         (mp626_alu_shift_op): Likewise.
4333         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
4334         (pj4_alu_e1_conds): Likewise.
4335         (pj4_alu): Likewise.
4336         (pj4_alu_conds): Likewise.
4337         (pj4_shift): Likewise.
4338         (pj4_shift_conds): Likewise.
4339         (pj4_alu_shift): Likewise.
4340         (pj4_alu_shift_conds): Likewise.
4341         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4342         changes.
4343         (cortexa7_older_only): Likewise.
4344         (cortexa7_younger): Likewise.
4346 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4348         * ipa-pure-const.c (generate_summary): Rename to...
4349         (pure_const_generate_summary): ... this.
4351 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
4353         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
4355 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
4357         PR target/57909
4358         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
4359         usage in HI mode.
4361 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4363         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
4364         enabled without -march=zEC12.
4365         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
4366         flags to be set.
4368 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4370         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
4371         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
4372         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
4373         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
4374         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
4375         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
4376         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
4377         ISA_HAS_FP4.
4378         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
4379         and ISA_HAS_NMADD3_NMSUB3.
4380         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
4381         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
4382         (nmsub4<mode>, nmsub3<mode>): Likewise.
4383         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
4385 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4387         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
4388         TARGET_MIPS5400 checking.
4390 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
4391             Peter Bergner  <bergner@vnet.ibm.com>
4393         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
4394         registers in the comment.
4395         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
4396         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
4397         rather than FIRST_PSEUDO_REGISTERS.
4399 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
4401         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
4402         enable extra ISA flags with TARGET_HTM.
4404 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4406         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
4407         Fix comment typos.
4409 2013-07-15  Cong Hou  <congh@google.com>
4411         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
4412         in compare function for sorting.
4414 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
4416         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
4417         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
4418         * config/rs6000/rs6000.opt: Add -mhtm option.
4419         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
4420         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
4421         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
4422         __HTM__ if the HTM instructions are available.
4423         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
4424         htm_spr_reg_operand): New define_predicates.
4425         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
4426         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
4427         Include htm.md.
4428         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
4429         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
4430         HTM builtin functions.
4431         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
4432         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
4433         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
4434         (rs6000_builtin_mask_calculate): Likewise.
4435         (rs6000_option_override_internal): Likewise.
4436         (bdesc_htm): Add new HTM builtin support.
4437         (htm_spr_num): New function.
4438         (htm_spr_regno): Likewise.
4439         (rs6000_htm_spr_icode): Likewise.
4440         (htm_expand_builtin): Likewise.
4441         (htm_init_builtins): Likewise.
4442         (rs6000_expand_builtin): Add support for HTM builtin functions.
4443         (rs6000_init_builtins): Likewise.
4444         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
4445         option.
4446         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
4447         (TARGET_HTM, MASK_HTM): Define macros.
4448         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
4449         (FIXED_REGISTERS): Likewise.
4450         (CALL_USED_REGISTERS): Likewise.
4451         (CALL_REALLY_USED_REGISTERS): Likewise.
4452         (REG_ALLOC_ORDER): Likewise.
4453         (enum reg_class): Likewise.
4454         (REG_CLASS_NAMES): Likewise.
4455         (REG_CLASS_CONTENTS): Likewise.
4456         (REGISTER_NAMES): Likewise.
4457         (ADDITIONAL_REGISTER_NAMES): Likewise.
4458         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
4459         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
4460         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
4461         * config/rs6000/htm.md: New file.
4462         * config/rs6000/htmintrin.h: New file.
4463         * config/rs6000/htmxlintrin.h: New file.
4465 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4467         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4468         Define SYMBOL_TINY_GOT, update comment.
4469         * config/aarch64/aarch64.c
4470         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
4471         (aarch64_expand_mov_immediate): Likewise.
4472         (aarch64_print_operand): Likewise.
4473         (aarch64_classify_symbol): Likewise.
4474         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
4475         (ldr_got_tiny): Define.
4477 2013-07-13  Tobias Grosser  <tobias@grosser.es>
4479         PR tree-optimization/54094
4480         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
4481           scheduling dimension for the parallelism check from the polyhedral
4482           information in the AST.
4483         * graphite-dependences.c (carries_deps): Do not assume the schedule is
4484           in 2D + 1 form.
4486 2013-07-13  Jason Merrill  <jason@redhat.com>
4488         * print-tree.c (debug_vec_tree): Use debug_raw.
4489         (debug_raw (vec<tree, va_gc> &)): New.
4490         (debug_raw (vec<tree, va_gc> *)): New.
4491         * tree.h: Declare them.
4493 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
4495         * ifcvt.c (ifcvt_after_combine): New static variable.
4496         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
4497         for size.
4498         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
4499         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
4500         rest_of_handle_if_after_reload): Pass new argument for if_convert.
4502 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
4504         * config/mips/mips.c (mips_expand_call): Remove empty statement.
4506 2013-07-12  Michael Matz  <matz@suse.de>
4508         PR middle-end/55771
4509         * convert.c (convert_to_real): Reject non-float inner types.
4511 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
4513         * config/aarch64/aarch64-protos.h
4514         (aarch64_simd_immediate_valid_for_move): Remove.
4515         * config/aarch64/aarch64.c (simd_immediate_info): New member.
4516         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
4517         cases.
4518         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
4520 2013-07-11  Steve Ellcey  <sellcey@mips.com>
4522         * config/mips/mips.c (mips_conditional_register_usage): Do not
4523         use t[0-7] registers in MIPS16 mode when optimizing for size.
4525 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
4527         * config/i386/i386.c (dispatch_function_versions): Fix array
4528         indexing of function_version_info to match actual_versions.
4530 2013-07-11  Teresa Johnson  <tejohnson@google.com>
4532         * vec.h (struct va_gc): Move release out-of-line.
4533         (va_gc::release): Call ggc_free on released vec.
4535 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4537         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
4538         Require GOT register as additional operand in UNSPEC.
4539         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
4540         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4541         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4542         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
4543         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
4544         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
4545         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
4547 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
4549         PR target/57631
4550         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
4551         name seen by assembler/linker rather if available.
4553 2013-07-11  Andreas Schwab  <schwab@suse.de>
4555         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
4557 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
4559         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
4561 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
4563         * doc/tm.texi.in: Move hook documentation to ....
4564         * target.def: ... here.
4566         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
4567         text on @hook line.
4568         * doc/tm.texi: Regenerate.
4570 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
4572         PR c++/57869
4573         * doc/invoke.texi: Document -Wconditionally-supported.
4575 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4577         PR target/57844
4578         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
4579         of my_fp.
4581 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4583         PR target/57506
4584         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
4585         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
4586         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
4587         Remove duplicate devices.
4588         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
4589         * config/avr/t-multilib: Regenerate.
4590         * config/avr/avr-tables.opt: Regenerate.
4591         * doc/avr-mmcu.texi: Regenerate.
4593 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4595         PR target/56987
4596         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
4598 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
4600         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
4601         the cost of MULT when optimizing for size.
4603 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4605         * config/cr16/cr16-protos.h: Don't include target.h.
4607 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
4609         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
4610         adjust register size for TDmode and TFmode for VSX registers.
4612 2013-07-08  Kai Tietz  <ktietz@redhat.com>
4614         PR target/56892
4615         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
4616         hook_bool_const_tree_true.
4618 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4620         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
4621         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
4622         * config/s390/s390.md: Define FPR*_REGNUM constants.
4623         Fix FPR2_REGNUM constant (18 -> 17).
4624         ("*trunc<BFP:mode><DFP_ALL:mode>2")
4625         ("*trunc<DFP_ALL:mode><BFP:mode>2")
4626         ("trunc<BFP:mode><DFP_ALL:mode>2")
4627         ("trunc<DFP_ALL:mode><BFP:mode>2")
4628         ("*extend<BFP:mode><DFP_ALL:mode>2")
4629         ("*extend<DFP_ALL:mode><BFP:mode>2")
4630         ("extend<BFP:mode><DFP_ALL:mode>2")
4631         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
4632         FPR4_REGNUM.
4634 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
4636         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
4638 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4640         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
4641         and cfun_fpr_bit_p to cfun_fpr_save_p.
4642         (s390_frame_area, s390_register_info, s390_frame_info)
4643         (s390_emit_prologue, s390_emit_epilogue)
4644         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
4645         register numbers.
4646         * config/s390/s390.h: Define *_REGNUM macros for floating point
4647         register numbers.
4649 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
4651         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
4653 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
4655         PR rtl-optimization/57786
4656         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
4657         and break out of the loop when it is set to false.
4659 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
4661         PR target/57819
4662         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
4663         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
4664         (const_int 63)) 0)).
4665         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
4666         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
4667         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
4669         PR rtl-optimization/57829
4670         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
4671         mask bits outside of mode are just sign-extension from mode to HWI.
4673 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
4675         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
4676         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
4677         adjust_address instead of change_address to keep info about alignment.
4678         (emit_strmov): Remove.
4679         (emit_memmov): New function.
4680         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
4681         (expand_movmem_epilogue): Likewise and return updated rtx for
4682         destination.
4683         (expand_constant_movmem_prologue): Likewise and return updated rtx for
4684         destination and source.
4685         (decide_alignment): Refactor, handle vector_loop.
4686         (ix86_expand_movmem): Likewise.
4687         (ix86_expand_setmem): Likewise.
4688         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
4690 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
4692         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
4693         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
4695 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4697         * config/i386/sse.md (sse_movlhps): Change alternative 3
4698         of operand 2 to "m".
4700 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4702         PR target/57807
4703         * config/i386/sse.md (iptr): New mode attribute.
4704         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
4705         (<sse>_vm<plusminus_insn><mode>3): Ditto.
4706         (<sse>_vmmul<mode>3): Ditto.
4707         (<sse>_vmdiv<mode>3): Ditto.
4708         (sse_vmrcpv4sf2): Ditto.
4709         (<sse>_vmsqrt<mode>2): Ditto.
4710         (sse_vmrsqrtv4sf2): Ditto.
4711         (<sse>_vm<code><mode>3): Ditto.
4712         (avx_vmcmp<mode>3): Ditto.
4713         (<sse>_vmmaskcmp<mode>3): Ditto.
4714         (<sse>_comi): Ditto.
4715         (<sse>_ucomi): Ditto.
4716         (*xop_vmfrcz_<mode>): Ditto.
4717         (*fmai_fmadd_<mode>): Ditto.
4718         (*fmai_fmsub_<mode>): Ditto.
4719         (*fmai_fnmadd_<mode>): Ditto.
4720         (*fmai_fnmsub_<mode>): Ditto.
4721         (*fma4i_vmfmadd_<mode>): Ditto.
4722         (*fma4i_vmfmsub_<mode>): Ditto.
4723         (*fma4i_vmfnmadd_<mode>): Ditto.
4724         (*fma4i_vmfnmsub_<mode>): Ditto.
4725         (*xop_vmfrcz_<mode>): Ditto.
4726         (sse_cvtps2pi): Ditto.
4727         (sse_cvttps2pi): Ditto.
4728         (sse_cvtss2si): Ditto.
4729         (sse_cvtss2si_2): Ditto.
4730         (sse_cvtss2siq_2): Ditto.
4731         (sse_cvttss2si): Ditto.
4732         (sse_cvttss2siq): Ditto.
4733         (sse_cvtsd2si): Ditto.
4734         (sse_cvtsd2si_2): Ditto.
4735         (sse_cvtsd2siq_2): Ditto.
4736         (sse_cvttsd2si): Ditto.
4737         (sse_cvttsd2siq): Ditto.
4738         (sse_cvtsd2ss): Ditto.
4739         (sse_cvtss2sd): Ditto.
4740         (avx2_pbroadcast<mode>): Ditto.
4741         (avx2_pbroadcast<mode>_1): Ditto.
4742         (*avx_vperm_broadcast_v4sf): Ditto.
4744         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
4745         (sse_movlhps): Ditto.
4746         (sse_storehps): Ditto.
4747         (sse_loadhps): Ditto.
4748         (sse_storelps): Ditto.
4749         (sse_loadlps): Ditto.
4750         (*vec_concatv4sf): Ditto.
4751         (*vec_interleave_highv2df): Ditto.
4752         (*vec_interleave_lowv2df): Ditto.
4753         (*vec_extractv2df_1_sse): Ditto.
4754         (*vec_extractv2df_0_sse): Ditto.
4755         (sse2_storelpd): Ditto.
4756         (sse2_loadlpd): Ditto.
4757         (sse2_movsd): Ditto.
4758         (*vec_concatv4si): Ditto.
4759         (vec_concatv2di): Ditto.
4761         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
4762         for Intel asm dialect.
4763         (mmx_punpcklwd): Ditto.
4764         (mmx_punpckldq): Ditto.
4766         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
4767         for intel assembler dialect.
4769 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
4771         PR target/29776
4772         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
4773         for BUILT_IN_C{LZ,LRSB}*.
4774         * tree.h (CASE_INT_FN): Add FN##IMAX case.
4775         * tree-vrp.c (extract_range_basic): Handle
4776         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
4777         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
4778         fall thru to code calling set_value*.
4779         * builtins.c (expand_builtin): Remove *IMAX cases.
4780         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
4781         if width is bigger than 2*HWI.
4783 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
4785         PR rtl-optimization/55342
4786         * lra-int.h (lra_subreg_reload_pseudos): New.
4787         * lra.c: Add undoing optional reloads to the block diagram.
4788         (lra_subreg_reload_pseudos): New.
4789         (lra_optional_reload_pseudos): Change comments.
4790         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
4791         lra_optional_reload_pseudos after undo transformations.
4792         * lra-assigns.c (pseudo_prefix_title): New.
4793         (lra_setup_reg_renumber): Use it.
4794         (spill_for): Ditto.  Check subreg reload pseudos too.
4795         (assign_by_spills): Consider subreg reload pseudos too.
4796         * lra-constraints.c (simplify_operand_subreg): Use
4797         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
4798         (curr_insn_transform): Recognize and do optional reloads.
4799         (undo_optional_reloads): New.
4800         (lra_undo_inheritance): Call undo_optional_reloads.
4802 2013-07-05  Thomas Quinot  <quinot@adacore.com>
4804         * tree-complex.c (expand_complex_operations_1): Fix typo.
4806 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
4808         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
4809         (tune_params): New member 'const vec_costs'.
4810         * config/aarch64/aarch64.c (generic_vector_cost): New.
4811         (generic_tunings): New member 'generic_vector_cost'.
4812         (aarch64_builtin_vectorization_cost): New.
4813         (aarch64_add_stmt_cost): New.
4814         (TARGET_VECTORIZE_ADD_STMT_COST): New.
4815         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
4817 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
4819         PR target/57777
4820         * config/i386/predicates.md (vsib_address_operand): Disallow
4821         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
4823 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
4825         PR middle-end/55030
4826         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
4827         expand_builtin_setjmp_receiver.
4828         (expand_label): Adjust, call expand_builtin_setjmp_receiver
4829         with NULL for the label parameter.
4830         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
4831         the frame-pointer.  Adjust comments.
4832         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
4833         only if LABEL is non-NULL.
4835 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
4837         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
4838         (ARM_ABI_AAPCS64): Ditto.
4839         (arm_abi): Ditto.
4840         (ARM_DEFAULT_ABI): Ditto.
4842 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
4844         * config/aarch64/aarch64-builtins.c
4845         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
4846         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
4847         (st1): Likewise.
4848         * config/aarch64/aarch64-simd.md
4849         (aarch64_ld1<VALL:mode>): New.
4850         (aarch64_st1<VALL:mode>): Likewise.
4851         * config/aarch64/arm_neon.h
4852         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
4854 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
4856         * config/i386/i386.c (gate_insert_vzeroupper): Check if
4857         target ISA is AVX.
4858         (ix86_option_override_internal):Turn on all -mavx target flags by
4859         default as they are dependent on AVX anyway.
4861 2013-07-02  Cary Coutant  <ccoutant@google.com>
4863         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
4864         deterministic hash.
4865         (loc_checksum_ordered): Likewise.
4866         (hash_loc_operands): Remove inline keyword.
4868 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
4870         PR tree-optimization/57741
4871         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
4872         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
4873         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
4874         Allow REAL_CST step_exprs if flag_associative_math.
4875         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
4877 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4879         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
4881 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4883         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
4885 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4887         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
4888         encoding.
4889         (iorsi3_insn): Likewise.
4890         (arm_xorsi3): Likewise.
4892 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
4894         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
4895         to ...
4896         (attribute "type"): ... here, and prefix with "wmmx_".
4897         (attribute "core_cycles"): Update for attribute changes.
4898         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
4899         (tbcstv4hi): Likewise.
4900         (tbcstv2si): Likewise.
4901         (iwmmxt_iordi3): Likewise.
4902         (iwmmxt_xordi3): Likewise.
4903         (iwmmxt_anddi3): Likewise.
4904         (iwmmxt_nanddi3): Likewise.
4905         (iwmmxt_arm_movdi): Likewise.
4906         (iwmmxt_movsi_insn): Likewise.
4907         (mov<mode>_internal): Likewise.
4908         (and<mode>3_iwmmxt): Likewise.
4909         (ior<mode>3_iwmmxt): Likewise.
4910         (xor<mode>3_iwmmxt): Likewise.
4911         (add<mode>3_iwmmxt): Likewise.
4912         (ssaddv8qi3): Likewise.
4913         (ssaddv4hi3): Likewise.
4914         (ssaddv2si3): Likewise.
4915         (usaddv8qi3): Likewise.
4916         (usaddv4hi3): Likewise.
4917         (usaddv2si3): Likewise.
4918         (sub<mode>3_iwmmxt): Likewise.
4919         (sssubv8qi3): Likewise.
4920         (sssubv4hi3): Likewise.
4921         (sssubv2si3): Likewise.
4922         (ussubv8qi3): Likewise.
4923         (ussubv4hi3): Likewise.
4924         (ussubv2si3): Likewise.
4925         (mulv4hi3_iwmmxt): Likewise.
4926         (smulv4hi3_highpart): Likewise.
4927         (umulv4hi3_highpart): Likewise.
4928         (iwmmxt_wmacs): Likewise.
4929         (iwmmxt_wmacsz): Likewise.
4930         (iwmmxt_wmacu): Likewise.
4931         (iwmmxt_wmacuz): Likewise.
4932         (iwmmxt_clrdi): Likewise.
4933         (iwmmxt_clrv8qi): Likewise.
4934         (iwmmxt_clr4hi): Likewise.
4935         (iwmmxt_clr2si): Likewise.
4936         (iwmmxt_uavgrndv8qi3): Likewise.
4937         (iwmmxt_uavgrndv4hi3): Likewise.
4938         (iwmmxt_uavgv8qi3): Likewise.
4939         (iwmmxt_uavgv4hi3): Likewise.
4940         (iwmmxt_tinsrb): Likewise.
4941         (iwmmxt_tinsrh): Likewise.
4942         (iwmmxt_tinsrw): Likewise.
4943         (iwmmxt_textrmub): Likewise.
4944         (iwmmxt_textrmsb): Likewise.
4945         (iwmmxt_textrmuh): Likewise.
4946         (iwmmxt_textrmsh): Likewise.
4947         (iwmmxt_textrmw): Likewise.
4948         (iwmxxt_wshufh): Likewise.
4949         (eqv8qi3): Likewise.
4950         (eqv4hi3): Likewise.
4951         (eqv2si3): Likewise.
4952         (gtuv8qi3): Likewise.
4953         (gtuv4hi3): Likewise.
4954         (gtuv2si3): Likewise.
4955         (gtv8qi3): Likewise.
4956         (gtv4hi3): Likewise.
4957         (gtv2si3): Likewise.
4958         (smax<mode>3_iwmmxt): Likewise.
4959         (umax<mode>3_iwmmxt): Likewise.
4960         (smin<mode>3_iwmmxt): Likewise.
4961         (umin<mode>3_iwmmxt): Likewise.
4962         (iwmmxt_wpackhss): Likewise.
4963         (iwmmxt_wpackwss): Likewise.
4964         (iwmmxt_wpackdss): Likewise.
4965         (iwmmxt_wpackhus): Likewise.
4966         (iwmmxt_wpackwus): Likewise.
4967         (iwmmxt_wpackdus): Likewise.
4968         (iwmmxt_wunpckihb): Likewise.
4969         (iwmmxt_wunpckihh): Likewise.
4970         (iwmmxt_wunpckihw): Likewise.
4971         (iwmmxt_wunpckilb): Likewise.
4972         (iwmmxt_wunpckilh): Likewise.
4973         (iwmmxt_wunpckilw): Likewise.
4974         (iwmmxt_wunpckehub): Likewise.
4975         (iwmmxt_wunpckehuh): Likewise.
4976         (iwmmxt_wunpckehuw): Likewise.
4977         (iwmmxt_wunpckehsb): Likewise.
4978         (iwmmxt_wunpckehsh): Likewise.
4979         (iwmmxt_wunpckehsw): Likewise.
4980         (iwmmxt_wunpckelub): Likewise.
4981         (iwmmxt_wunpckeluh): Likewise.
4982         (iwmmxt_wunpckeluw): Likewise.
4983         (iwmmxt_wunpckelsb): Likewise.
4984         (iwmmxt_wunpckelsh): Likewise.
4985         (iwmmxt_wunpckelsw): Likewise.
4986         (ror<mode>3): Likewise.
4987         (ashr<mode>3_iwmmxt): Likewise.
4988         (lshr<mode>3_iwmmxt): Likewise.
4989         (ashl<mode>3_iwmmxt): Likewise.
4990         (ror<mode>3_di): Likewise.
4991         (ashr<mode>3_di): Likewise.
4992         (lshr<mode>3_di): Likewise.
4993         (ashl<mode>3_di): Likewise.
4994         (iwmmxt_wmadds): Likewise.
4995         (iwmmxt_wmaddu): Likewise.
4996         (iwmmxt_tmia): Likewise.
4997         (iwmmxt_tmiaph): Likewise.
4998         (iwmmxt_tmiabb): Likewise.
4999         (iwmmxt_tmiatb): Likewise.
5000         (iwmmxt_tmiabt): Likewise.
5001         (iwmmxt_tmiatt): Likewise.
5002         (iwmmxt_tmovmskb): Likewise.
5003         (iwmmxt_tmovmskh): Likewise.
5004         (iwmmxt_tmovmskw): Likewise.
5005         (iwmmxt_waccb): Likewise.
5006         (iwmmxt_wacch): Likewise.
5007         (iwmmxt_waccw): Likewise.
5008         (iwmmxt_waligni): Likewise.
5009         (iwmmxt_walignr): Likewise.
5010         (iwmmxt_walignr0): Likewise.
5011         (iwmmxt_walignr1): Likewise.
5012         (iwmmxt_walignr2): Likewise.
5013         (iwmmxt_walignr3): Likewise.
5014         (iwmmxt_wsadb): Likewise.
5015         (iwmmxt_wsadh): Likewise.
5016         (iwmmxt_wsadbz): Likewise.
5017         (iwmmxt_wsadhz): Likewise.
5018         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5019         (iwmmxt_wabsdiffb): Likewise.
5020         (iwmmxt_wabsdiffh): Likewise.
5021         (iwmmxt_wabsdiffw): Likewise.
5022         (iwmmxt_waddsubhx): Likewise
5023         (iwmmxt_wsubaddhx): Likewise.
5024         (addc<mode>3): Likewise.
5025         (iwmmxt_avg4): Likewise.
5026         (iwmmxt_avg4r): Likewise.
5027         (iwmmxt_wmaddsx): Likewise.
5028         (iwmmxt_wmaddux): Likewise.
5029         (iwmmxt_wmaddsn): Likewise.
5030         (iwmmxt_wmaddun): Likewise.
5031         (iwmmxt_wmulwsm): Likewise.
5032         (iwmmxt_wmulwum): Likewise.
5033         (iwmmxt_wmulsmr): Likewise.
5034         (iwmmxt_wmulumr): Likewise.
5035         (iwmmxt_wmulwsmr): Likewise.
5036         (iwmmxt_wmulwumr): Likewise.
5037         (iwmmxt_wmulwl): Likewise.
5038         (iwmmxt_wqmulm): Likewise.
5039         (iwmmxt_wqmulwm): Likewise.
5040         (iwmmxt_wqmulmr): Likewise.
5041         (iwmmxt_wqmulwmr): Likewise.
5042         (iwmmxt_waddbhusm): Likewise.
5043         (iwmmxt_waddbhusl): Likewise.
5044         (iwmmxt_wqmiabb): Likewise.
5045         (iwmmxt_wqmiabt): Likewise.
5046         (iwmmxt_wqmiatb): Likewise.
5047         (iwmmxt_wqmiatt): Likewise.
5048         (iwmmxt_wqmiabbn): Likewise.
5049         (iwmmxt_wqmiabtn): Likewise.
5050         (iwmmxt_wqmiatbn): Likewise.
5051         (iwmmxt_wqmiattn): Likewise.
5052         (iwmmxt_wmiabb): Likewise.
5053         (iwmmxt_wmiabt): Likewise.
5054         (iwmmxt_wmiatb): Likewise.
5055         (iwmmxt_wmiatt): Likewise.
5056         (iwmmxt_wmiabbn): Likewise.
5057         (iwmmxt_wmiabtn): Likewise.
5058         (iwmmxt_wmiatbn): Likewise.
5059         (iwmmxt_wmiattn): Likewise.
5060         (iwmmxt_wmiawbb): Likewise.
5061         (iwmmxt_wmiawbt): Likewise.
5062         (iwmmxt_wmiawtb): Likewise.
5063         (iwmmxt_wmiawtt): Likewise.
5064         (iwmmxt_wmiawbbn): Likewise.
5065         (iwmmxt_wmiawbtn): Likewise.
5066         (iwmmxt_wmiawtbn): Likewise.
5067         (iwmmxt_wmiawttn): Likewise.
5068         (iwmmxt_wmerge): Likewise.
5069         (iwmmxt_tandc<mode>3): Likewise.
5070         (iwmmxt_torc<mode>3): Likewise.
5071         (iwmmxt_torvsc<mode>3): Likewise.
5072         (iwmmxt_textrc<mode>3): Likewise.
5073         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5074         (wmmxt_pack): Likewise.
5075         (wmmxt_mult_c1): Likewise.
5076         (wmmxt_mult_c2): Likewise.
5077         (wmmxt_alu_c1): Likewise.
5078         (wmmxt_alu_c2): Likewise.
5079         (wmmxt_alu_c3): Likewise.
5080         (wmmxt_transfer_c1): Likewise.
5081         (wmmxt_transfer_c2): Likewise.
5082         (wmmxt_transfer_c3): Likewise.
5083         (marvell_f_iwmmxt_wstr): Likewise.
5084         (marvell_f_iwmmxt_wldr): Likewise.
5086 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
5088         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5090 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5092         Revert:
5093         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5094         * lra-constraints.c (need_for_split_p): Check call used hard regs
5095         living through calls.
5097         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5098         call used regs for call insn.
5100 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
5102         PR target/57736
5103         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5104         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5106 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
5108         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5110 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5112         * lra-constraints.c (need_for_split_p): Check call used hard regs
5113         living through calls.
5115 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
5117         PR target/57744
5118         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5119         to tie with any other modes.  Eliminate Altivec vector mode tests,
5120         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
5121         code, to return 0 if testing MODE2 for a condition, if we've
5122         already tested MODE1 for the same condition.
5124 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5126         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5127         layout.
5129 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5131         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5132         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5134 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5136         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5137         Define.
5138         (aarch64_symbolic_constant_p): Remove.
5139         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5140         static.  Fix line length and white space.
5141         (aarch64_symbolic_constant_p): Remove.
5142         * config/aarch64/predicates.md (aarch64_valid_symref):
5143         Use aarch64_classify_symbol_expression.
5145 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5147         * config/arm/constraints.md (Ts): New constraint.
5148         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5149         16-bit encodings.
5150         (compare_scc): Use "Ts" constraint for operand 0.
5151         (ior_scc_scc): Likewise.
5152         (and_scc_scc): Likewise.
5153         (and_scc_scc_nodom): Likewise.
5154         (ior_scc_scc_cmp): Likewise for operand 7.
5155         (and_scc_scc_cmp): Likewise.
5156         * config/arm/thumb2.md (thumb2_movsi_insn):
5157         Add alternatives for 16-bit encodings.
5158         (thumb2_movhi_insn): Likewise.
5159         (thumb2_movsicc_insn): Likewise.
5160         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
5161         (thumb2_negscc): Use "Ts" constraint.
5162         Move mvn instruction outside cond_exec block.
5163         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5164         for 16-bit encodings.
5166 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5168         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5169         encoding.
5170         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5171         (mulsi3subsi): Likewise.
5172         (mulsidi3adddi): Likewise.
5173         (mulsidi3_v6): Likewise.
5174         (umulsidi3_v6): Likewise.
5175         (umulsidi3adddi_v6): Likewise.
5176         (smulsi3_highpart_v6): Likewise.
5177         (umulsi3_highpart_v6): Likewise.
5178         (mulhisi3tb): Likewise.
5179         (mulhisi3bt): Likewise.
5180         (mulhisi3tt): Likewise.
5181         (maddhisi4): Likewise.
5182         (maddhisi4tb): Likewise.
5183         (maddhisi4tt): Likewise.
5184         (maddhidi4): Likewise.
5185         (maddhidi4tb): Likewise.
5186         (maddhidi4tt): Likewise.
5187         (zeroextractsi_compare0_scratch): Likewise.
5188         (insv_zero): Likewise.
5189         (insv_t2): Likewise.
5190         (anddi_notzesidi_di): Likewise.
5191         (anddi_notsesidi_di): Likewise.
5192         (andsi_notsi_si): Likewise.
5193         (iordi_zesidi_di): Likewise.
5194         (xordi_zesidi_di): Likewise.
5195         (andsi_iorsi3_notsi): Likewise.
5196         (smax_0): Likewise.
5197         (smax_m1): Likewise.
5198         (smin_0): Likewise.
5199         (not_shiftsi): Likewise.
5200         (unaligned_loadsi): Likewise.
5201         (unaligned_loadhis): Likewise.
5202         (unaligned_loadhiu): Likewise.
5203         (unaligned_storesi): Likewise.
5204         (unaligned_storehi): Likewise.
5205         (extv_reg): Likewise.
5206         (extzv_t2): Likewise.
5207         (divsi3): Likewise.
5208         (udivsi3): Likewise.
5209         (arm_zero_extendhisi2addsi): Likewise.
5210         (arm_zero_extendqisi2addsi): Likewise.
5211         (compareqi_eq0): Likewise.
5212         (arm_extendhisi2_v6): Likewise.
5213         (arm_extendqisi2addsi): Likewise.
5214         (arm_movt): Likewise.
5215         (thumb2_ldrd): Likewise.
5216         (thumb2_ldrd_base): Likewise.
5217         (thumb2_ldrd_base_neg): Likewise.
5218         (thumb2_strd): Likewise.
5219         (thumb2_strd_base): Likewise.
5220         (thumb2_strd_base_neg): Likewise.
5221         (arm_negsi2): Add alternative for 16-bit encoding.
5222         (arm_one_cmplsi2): Likewise.
5224 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5226         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
5227         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
5228         (movdfcc): Likewise.
5229         * config/arm/vfp.md (*thumb2_movsf_vfp):
5230         Disable predication for arm_restrict_it.
5231         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
5232         (*thumb2_movdfcc_vfp): Likewise.
5233         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
5234         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
5235         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
5236         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
5237         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
5238         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
5239         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
5240         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
5241         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
5242         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
5243         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
5244         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
5245         Disable predication for arm_restrict_it.
5247 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
5249         * config/i386/bmiintrin.h (_bextr_u32): New.
5250         (_bextr_u64): Ditto.
5252 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
5254         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
5255         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
5256         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
5257         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
5258         * config/mips/n32-elf.h: ...this new file.
5260 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
5262         PR target/57224
5263         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
5264         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
5266 2013-06-27  Catherine Moore <clm@codesourcery.com>
5268         * config/mips/mips-tables.opt: Regenerate.
5269         * config/mips/mips-cpus.def: Add m14ke and m14kec.
5270         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
5271         * doc/invoke.texi: Add -m14kc.
5273 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5275         PR target/57623
5276         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
5277         constraints of operand 1 and 2.
5279         PR target/57623
5280         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
5281         to match RTL canonicalization.  Swap predicates and
5282         constraints of operand 1 and 2.
5284 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
5286         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
5287         Process OP_INOUT regs for splitting too.
5289 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5291         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
5292         decl before the loop, initialize to NULL.
5293         (vectorizable_load): Initialize ptr_incr to NULL.
5295 2013-06-27  Martin Jambor  <mjambor@suse.cz>
5297         PR lto/57208
5298         * ipa-ref.h (ipa_maybe_record_reference): Declare.
5299         * ipa-ref.c (ipa_maybe_record_reference): New function.
5300         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
5301         * ipa-cp.c (create_specialized_node): Record potential references from
5302         aggvals.
5303         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
5305 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5307         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
5308         parameter 'mode' of type 'enum machine_mode mode'; change to pass
5309         'mode' to force_reg.
5310         (aarch64_add_offset): Update calls to aarch64_force_temporary.
5311         (aarch64_expand_mov_immediate): Likewise.
5313 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5315         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
5316         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
5318 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5320         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
5321         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5322         (struct machine_function): Add tbegin_p.
5323         (s390_canonicalize_comparison): Fold CC mode compares to
5324         conditional jump if possible.
5325         (s390_emit_jump): Return the emitted jump.
5326         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
5327         Handle CCRAWmode compares.
5328         (s390_option_override): Default to -mhtm if available.
5329         (s390_reg_clobbered_rtx): Handle floating point regs as well.
5330         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
5331         FPRs instead of df_regs_ever_live_p.
5332         (s390_optimize_nonescaping_tx): New function.
5333         (s390_init_frame_layout): Extend clobbered_regs array to cover
5334         FPRs as well.
5335         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
5336         (s390_expand_tbegin): New function.
5337         (enum s390_builtin): New enum definition.
5338         (code_for_builtin): New array definition.
5339         (s390_init_builtins): New function.
5340         (s390_expand_builtin): New function.
5341         (TARGET_INIT_BUILTINS): Define.
5342         (TARGET_EXPAND_BUILTIN): Define.
5343         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
5344         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
5345         (s390_alc_comparison): Likewise.
5346         * config/s390/s390-modes.def: Add CCRAWmode.
5347         * config/s390/s390.h (processor_flags): Add PF_TX.
5348         (TARGET_CPU_HTM): Define macro.
5349         (TARGET_HTM): Define macro.
5350         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
5351         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
5352         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5353         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
5354         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
5355         (TBEGIN_MASK, TBEGINC_MASK): New constants.
5356         ("*cc_to_int"): Move up.
5357         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
5358         constants other than 0.
5359         ("*ccraw_to_int"): New insn and splitter definition.
5360         ("tbegin", "tbegin_nofloat", "tbegin_retry")
5361         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
5362         ("tx_assist"): New expander.
5363         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
5364         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
5365         * config/s390/s390.opt: Add -mhtm option.
5366         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
5367         * config/s390/htmxlintrin.h: New file.
5368         * config/s390/htmintrin.h: New file.
5369         * config/s390/s390intrin.h: New file.
5370         * doc/extend.texi: Document htm builtins.
5371         * config.gcc: Add the new header files to extra_headers.
5373 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
5375         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
5376         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
5378 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
5379             Pat Haugen <pthaugen@us.ibm.com>
5380             Peter Bergner <bergner@vnet.ibm.com>
5382         * config/rs6000/power8.md: New.
5383         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
5384         setting for power8 entry.
5385         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
5386         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
5387         test for Power4/Power5 only.
5388         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
5389         support.
5390         (force_new_group): Adjust comment.
5391         * config/rs6000/rs6000.md: Include power8.md.
5393 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
5395         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
5396         * config/arm/arm-protos.h (arm_max_conditional_execute): New
5397         declaration.
5398         (tune_params): Update comment.
5399         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
5400         (arm_max_conditional_execute): New function.
5401         (thumb2_final_prescan_insn): Use max_insn_skipped and
5402         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
5404 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
5406         PR tree-optimization/57705
5407         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
5408         SSA_NAME step, provided that it is not defined inside the loop.
5409         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
5410         (get_initial_def_for_induction): Handle SSA_NAME IV step.
5412 2013-06-25  Martin Jambor  <mjambor@suse.cz>
5414         PR middle-end/57670
5415         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
5416         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
5417         calls in the dump.
5418         (ipa_note_param_call): Initialize member_ptr flag.
5419         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
5420         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
5421         (ipa_write_indirect_edge_info): Stream member_ptr flag.
5422         (ipa_read_indirect_edge_info): Likewise.
5424 2013-06-25  Richard Biener  <rguenther@suse.de>
5426         PR middle-end/56977
5427         * passes.c (init_optimization_passes): Move pass_fold_builtins
5428         and pass_dce earlier with -Og.
5430 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
5432         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
5433         <BIT_FIELD_REF>: Remove trailing TAB.
5434         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
5435         remove blank line.
5437 2013-06-24  Martin Jambor  <mjambor@suse.cz>
5439         PR tree-optimization/57358
5440         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
5441         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
5442         (ipa_analyze_params_uses): Generate pessimistic info when true.
5444 2013-06-24  Martin Jambor  <mjambor@suse.cz>
5446         PR tree-optimization/57539
5447         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
5448         global.inlined_to of the new node to it.  All callers changed.
5449         * ipa-inline-transform.c (clone_inlined_nodes): New variable
5450         inlining_into, pass it to cgraph_clone_node.
5451         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
5452         ipa_free_edge_args_substructures.
5453         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
5454         rdesc linked list.  Do not assert rdesc edges have inlined caller.
5455         Assert we have found an rdesc in the rdesc list.
5457 2013-06-24  Richard Biener  <rguenther@suse.de>
5459         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
5460         (pointer_set_lookup): Declare.
5461         (class pointer_map): New template class implementing a
5462         generic pointer to T map.
5463         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
5464         pointer_map<T>::contains, pointer_map<T>::insert,
5465         pointer_map<T>::traverse): New functions.
5466         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
5467         (pointer_set_lookup): New function.
5468         (pointer_set_contains): Use pointer_set_lookup.
5469         (pointer_set_insert): Likewise.
5470         (insert_aux): Remove.
5471         (struct pointer_map_t): Embed a pointer_set_t.
5472         (pointer_map_create): Adjust.
5473         (pointer_map_destroy): Likewise.
5474         (pointer_map_contains): Likewise.
5475         (pointer_map_insert): Likewise.
5476         (pointer_map_traverse): Likewise.
5477         * tree-streamer.h (struct streamer_tree_cache_d): Use a
5478         pointer_map<unsigned> instead of a pointer_map_t.
5479         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
5480         (streamer_tree_cache_lookup): Likewise.
5481         (streamer_tree_cache_create): Likewise.
5482         (streamer_tree_cache_delete): Likewise.
5483         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
5484         pointer_map<unsigned> instead of a pointer_map_t.
5485         (lto_init_tree_ref_encoder): Adjust.
5486         (lto_destroy_tree_ref_encoder): Likewise.
5487         * lto-section-out.c (lto_output_decl_index): Likewise.
5488         (lto_record_function_out_decl_state): Likewise.
5489         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
5491 2013-06-24  Richard Biener  <rguenther@suse.de>
5493         PR tree-optimization/57488
5494         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
5496 2013-06-24  Alan Modra  <amodra@gmail.com>
5498         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
5499         (gen_easy_altivec_constant): Likewise.
5500         * config/rs6000/predicates.md (easy_vector_constant_add_self,
5501         easy_vector_constant_msb): Likewise.
5503 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
5505         PR target/57688
5506         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
5507         add missing return true.
5509 2013-06-23  Oleg Endo <olegendo@gcc.gnu.org>
5511         PR target/52483
5512         * config/sh/predicates.md (general_extend_operand): Invoke
5513         general_movsrc_operand for memory operands.
5514         (general_movsrc_operand): Allow reg+reg addressing, do not use
5515         general_operand for memory operands.
5517 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
5519         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
5520         when current target options does not apply.
5521         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
5522         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
5523         * config/i386/bmiintrin.h: Pass appropriate target
5524         attributes to header.
5525         * config/i386/mmintrin.h: Ditto.
5526         * config/i386/nmmintrin.h: Ditto.
5527         * config/i386/avx2intrin.h: Ditto.
5528         * config/i386/fxsrintrin.h: Ditto.
5529         * config/i386/tbmintrin.h: Ditto.
5530         * config/i386/xsaveintrin.h: Ditto.
5531         * config/i386/f16cintrin.h: Ditto.
5532         * config/i386/xtestintrin.h: Ditto.
5533         * config/i386/xsaveoptintrin.h: Ditto.
5534         * config/i386/bmi2intrin.h: Ditto.
5535         * config/i386/lzcntintrin.h: Ditto.
5536         * config/i386/smmintrin.h: Ditto.
5537         * config/i386/wmmintrin.h: Ditto.
5538         * config/i386/x86intrin.h: Remove all header include guards.
5539         * config/i386/prfchwintrin.h: Ditto.
5540         * config/i386/pmmintrin.h: Ditto.
5541         * config/i386/tmmintrin.h: Ditto.
5542         * config/i386/xmmintrin.h: Ditto.
5543         * config/i386/popcntintrin.h: Ditto.
5544         * config/i386/rdseedintrin.h: Ditto.
5545         * config/i386/ammintrin.h: Ditto.
5546         * config/i386/emmintrin.h: Ditto.
5547         * config/i386/immintrin.h: Remove all header include guards.
5548         * config/i386/fma4intrin.h: Ditto.
5549         * config/i386/lwpintrin.h: Ditto.
5550         * config/i386/xopintrin.h: Ditto.
5551         * config/i386/ia32intrin.h: Ditto.
5552         * config/i386/avxintrin.h: Ditto.
5553         * config/i386/rtmintrin.h: Ditto.
5554         * config/i386/fmaintrin.h: Ditto.
5555         * config/i386/mm3dnow.h: Ditto.
5557 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
5559         * common/config/i386/i386-common.c: Handle LZCNT.
5561 2013-06-22  Andi Kleen  <ak@linux.intel.com>
5563         * doc/extend.texi: Use __atomic_store_n instead of
5564         __atomic_store in HLE example.
5566 2013-06-22  Oleg Endo <olegendo@gcc.gnu.org>
5568         * config/sh/sh.c: Remove <cstdlib> workaround.
5570 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5572         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
5574 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5576         * doc/extend.texi: Document that __atomic_clear and
5577           __atomic_test_and_set should only be used with bool.
5579 2013-06-20  Jan Hubicka  <jh@suse.cz>
5581         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
5582         types_same_for_odr.
5583         * tree.c (decls_same_for_odr): New function.
5584         (same_for_edr): New function.
5585         (types_same_for_odr): New function.
5586         (get_binfo_at_offset): Use it.
5587         * tree.h (types_same_for_odr): Declare.
5589 2013-06-20  Oleg Endo <olegendo@gcc.gnu.org>
5590             Jason Merrill  <jason@redhat.com>
5592         * system.h: Include <cstdlib> as well as <stdlib.h>.
5594 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
5596         PR target/57655
5597         * config/i386/i386.c (construct_container): Report error if
5598         long double is used with disabled x87 float returns.
5600 2013-06-20  Jan Hubicka  <jh@suse.cz>
5602         * lto-cgraph.c (input_symtab): Do not set cgraph state.
5604 2013-06-20  Joern Rennecke <joern.rennecke@embecosm.com>
5606         PR rtl-optimization/57425
5607         PR rtl-optimization/57569
5608         * alias.c (write_dependence_p): Remove parameters mem_mode and
5609         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
5610         Changed all callers.
5611         (canon_anti_dependence): Get comments and semantics in sync.
5612         Add parameter mem_canonicalized.  Changed all callers.
5613         * rtl.h (canon_anti_dependence): Update prototype.
5615 2013-06-20  Richard Biener  <rguenther@suse.de>
5617         * data-streamer-in.c (streamer_read_uhwi): Optimize single
5618         byte case, inline streamer_read_uchar and defer section
5619         overrun check.
5621 2013-06-20  Richard Biener  <rguenther@suse.de>
5623         PR tree-optimization/57584
5624         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
5625         SSA names into the expanded expression that take part in
5626         abnormal coalescing.
5628 2013-06-19  Sharad Singhai  <singhai@google.com>
5630         * gcov.c (print_usage): Handle new option.
5631         (process_args): Ditto.
5632         (get_gcov_intermediate_filename): New function.
5633         (output_intermediate_file): New function.
5634         (output_gcov_file): New function
5635         (generate_results): Handle new option.
5636         (release_function): Relase demangled name.
5637         (read_graph_file): Handle demangled name.
5638         (output_lines): Ditto.
5639         * doc/gcov.texi: Document gcov intermediate format.
5641 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
5643         PR bootstrap/57604
5644         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
5645         (lra_emit_add): Use the functions.  Add comment about Y as an
5646         address segment.
5648 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
5650         PR driver/57652
5651         * collect2.c (collect_atexit): New.
5652         (collect_exit): Delete.
5653         (main): Register collect_atexit with atexit.
5654         (collect_wait): Change collect_exit to exit.
5655         (do_wait): Same.
5656         * collect2.h (collect_exit): Delete.
5657         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
5659 2013-06-19  Wei Mi  <wmi@google.com>
5661         PR rtl-optimization/57518
5662         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
5663         if regno is used in paradoxical subreg.
5664         (update_equiv_regs): Check pdx_subregs[regno] before
5665         set a reg to be equivalent with a mem.
5667 2013-06-19  Matthias Klose  <doko@ubuntu.com>
5669         PR driver/57651
5670         * file-find.h (find_a_file): Add a mode parameter.
5671         * file-find.c (find_a_file): Likewise.
5672         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
5673         with X_OK for the executables.
5674         * collect2.c (main): Call find_a_file with X_OK.
5676 2013-06-19  Steve Ellcey  <sellcey@mips.com>
5678         PR target/56942
5679         * config/mips/mips.md (casesi_internal_mips16_<mode>):
5680         Use NEXT_INSN instead of next_real_insn.
5682 2013-06-19  Jan Hubicka  <jh@suse.cz>
5684         * cgraph.h (const_value_known_p): Replace by ...
5685         (ctor_for_folding): .. this one.
5686         * cgraphunit.c (process_function_and_variable_attributes): Use it.
5687         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
5688         * expr.c (expand_expr_real_1): Likewise.
5689         (string_constant): Likewise.
5690         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
5691         * ipa.c (process_references): Likewise.
5692         (symtab_remove_unreachable_nodes): Likewise.
5693         * ipa-inline-analysis.c (param_change_prob): Likewise.
5694         * gimple-fold.c (canonicalize_constructor_val): Likewise.
5695         (get_base_constructor): Likwise.
5696         * varpool.c (varpool_remove_node): Likewise.
5697         (varpool_remove_initializer): LIkewise.
5698         (dump_varpool_node): LIkwise.
5699         (const_value_known_p): Rewrite to ...
5700         (ctor_for_folding): ... this one.
5702 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
5704         PR driver/57651
5705         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
5706         PERSONALITY in $PATH derived prefixes.
5708 2013-06-19  Jeff Law  <law@redhat.com>
5710         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
5711         in comment.
5713         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
5714         (simplify_bitwise_binary): Use it to simpify certain binary ops on
5715         booleans.
5717 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
5719         * config/arm/vfp.md: Move VFP instruction classification documentation
5720         to ...
5721         * config/arm/arm.md: ... here.  Update instruction classification
5722         documentation.
5724 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
5726         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
5727         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
5728         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
5730 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
5732         PR target/57609
5733         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5734         with NEXT_INSN.  Use tablejump_p to check for jump table data
5735         insns.
5737 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
5739         PR c++/56544
5740         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
5741         that now in C++ the value is correct per the C++ standards.
5743 2013-06-19  Richard Biener  <rguenther@suse.de>
5745         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
5746         for global context.
5748 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5750         Revert:
5751         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5753         PR target/57609
5754         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5755         with next_active_insn.
5757 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
5759         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
5760         functions are inlined during failures to flag an error.
5761         * tree-inline.c (expand_call_inline): Allow the error to be flagged
5762         in early inline pass.
5764 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
5766         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
5767         in comments.
5769 2013-06-18  Julian Brown  <julian@codesourcery.com>
5771         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
5772         Permit virtual register pre-reload if !strict.
5773         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
5774         change.
5775         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
5776         prototype.
5777         * config/arm/neon.md (movmisalign<mode>): Use
5778         neon_perm_struct_or_reg_operand instead of
5779         neon_struct_or_register_operand.
5780         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
5781         neon_permissive_struct_operand instead of neon_struct_operand.
5782         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
5783         neon_vector_mem_operand.
5784         * config/arm/predicates.md (neon_struct_operand): Adjust call to
5785         neon_vector_mem_operand.
5786         (neon_permissive_struct_operand): New.
5787         (neon_struct_or_register_operand): Rename to...
5788         (neon_perm_struct_or_reg_operand): This. Adjust call to
5789         neon_vector_mem_operand.
5791 2013-06-18  Richard Biener  <rguenther@suse.de>
5793         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
5794         * lto-streamer.h: Include pointer-set.h.
5795         (struct lto_decl_slot): Remove.
5796         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
5797         Remove next_index entry.
5798         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5799         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5800         (lto_init_tree_ref_encoder): Adjust.
5801         (lto_destroy_tree_ref_encoder): Likewise.
5802         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5803         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5804         (lto_output_decl_index): Adjust.
5805         (lto_new_out_decl_state): Likewise.
5806         (lto_record_function_out_decl_state): Likewise.
5807         * lto-streamer-out.c (copy_function): Likewise.
5809 2013-06-18  Richard Biener  <rguenther@suse.de>
5811         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
5812         * cgraphunit.c: Include cfgloop.h.
5813         (init_lowered_empty_function): Initialize the loop tree.
5814         (assemble_thunk): Insert new BBs into loops.
5816 2013-06-18  Richard Biener  <rguenther@suse.de>
5818         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
5819         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
5820         the map from cache entry to cache index optional.
5821         (streamer_tree_cache_replace_tree): Adjust accordingly.
5822         (streamer_tree_cache_append): Likewise.
5823         (streamer_tree_cache_delete): Likewise.
5824         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
5825         streamer cache map from cache entry to cache index.
5826         * lto-streamer-out.c (create_output_block): Adjust.
5828 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
5830         * config/arm/arm.md (attribute "insn"): Move multiplication and
5831         division attributes to...
5832         (attribute "type"): ... here.  Remove mult.
5833         (attribute "mul32"): New attribute.
5834         (attribute "mul64"): Add umaal.
5835         (*arm_mulsi3): Update attributes.
5836         (*arm_mulsi3_v6): Likewise.
5837         (*thumb_mulsi3): Likewise.
5838         (*thumb_mulsi3_v6): Likewise.
5839         (*mulsi3_compare0): Likewise.
5840         (*mulsi3_compare0_v6): Likewise.
5841         (*mulsi_compare0_scratch): Likewise.
5842         (*mulsi_compare0_scratch_v6): Likewise.
5843         (*mulsi3addsi): Likewise.
5844         (*mulsi3addsi_v6): Likewise.
5845         (*mulsi3addsi_compare0): Likewise.
5846         (*mulsi3addsi_compare0_v6): Likewise.
5847         (*mulsi3addsi_compare0_scratch): Likewise.
5848         (*mulsi3addsi_compare0_scratch_v6): Likewise.
5849         (*mulsi3subsi): Likewise.
5850         (*mulsidi3adddi): Likewise.
5851         (*mulsi3addsi_v6): Likewise.
5852         (*mulsidi3adddi_v6): Likewise.
5853         (*mulsidi3_nov6): Likewise.
5854         (*mulsidi3_v6): Likewise.
5855         (*umulsidi3_nov6): Likewise.
5856         (*umulsidi3_v6): Likewise.
5857         (*umulsidi3adddi): Likewise.
5858         (*umulsidi3adddi_v6): Likewise.
5859         (*smulsi3_highpart_nov6): Likewise.
5860         (*smulsi3_highpart_v6): Likewise.
5861         (*umulsi3_highpart_nov6): Likewise.
5862         (*umulsi3_highpart_v6): Likewise.
5863         (mulhisi3): Likewise.
5864         (*mulhisi3tb): Likewise.
5865         (*mulhisi3bt): Likewise.
5866         (*mulhisi3tt): Likewise.
5867         (maddhisi4): Likewise.
5868         (*maddhisi4tb): Likewise.
5869         (*maddhisi4tt): Likewise.
5870         (maddhidi4): Likewise.
5871         (*maddhidi4tb): Likewise.
5872         (*maddhidi4tt): Likewise.
5873         (divsi3): Likewise.
5874         (udivsi3): Likewise.
5875         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
5876         (thumb2_mulsi_short_compare0): Likewise.
5877         (thumb2_mulsi_short_compare0_scratch): Likewise.
5878         * config/arm/arm1020e.md (1020mult1): Update attribute change.
5879         (1020mult2): Likewise.
5880         (1020mult3): Likewise.
5881         (1020mult4): Likewise.
5882         (1020mult5): Likewise.
5883         (1020mult6): Likewise.
5884         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
5885         change.
5886         (cortex_a15_mult64): Likewise.
5887         (cortex_a15_sdiv): Likewise.
5888         (cortex_a15_udiv): Likewise.
5889         * config/arm/arm1026ejs.md (mult1): Update attribute change.
5890         (mult2): Likewise.
5891         (mult3): Likewise.
5892         (mult4): Likewise.
5893         (mult5): Likewise.
5894         (mult6): Likewise.
5895         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
5896         (pj4_ir_div): Likewise.
5897         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
5898         (11_mult2): Likewise.
5899         (11_mult3): Likewise.
5900         (11_mult4): Likewise.
5901         (11_mult5): Likewise.
5902         (11_mult6): Likewise.
5903         (11_mult7): Likewise.
5904         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
5905         (cortex_a8_mla): Likewise.
5906         (cortex_a8_mull): Likewise.
5907         (cortex_a8_smulwy): Likewise.
5908         (cortex_a8_smlald): Likewise.
5909         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
5910         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
5911         (cortex_r4_mul_3): Likewise.
5912         (cortex_r4_mla_4): Likewise.
5913         (cortex_r4_mla_3): Likewise.
5914         (cortex_r4_smlald): Likewise.
5915         (cortex_r4_mull): Likewise.
5916         (cortex_r4_sdiv): Likewise.
5917         (cortex_r4_udiv): Likewise.
5918         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
5919         (cortex_a7_idiv): Likewise.
5920         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
5921         (9_mult2): Likewise.
5922         (9_mult3): Likewise.
5923         (9_mult4): Likewise.
5924         (9_mult5): Likewise.
5925         (9_mult6): Likewise.
5926         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
5927         (cortex_a53_sdiv): Likewise.
5928         (cortex_a53_udiv): Likewise.
5929         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5930         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5931         (mp626_mult2): Likewise.
5932         (mp626_mult3): Likewise.
5933         (mp626_mult4): Likewise.
5934         * config/arm/fa526.md (526_mult1): Update attribute change.
5935         (526_mult2): Likewise.
5936         * config/arm/arm-generic.md (mult): Update attribute change.
5937         (mult_ldsched_strongarm): Likewise.
5938         (mult_ldsched): Likewise.
5939         (multi_cycle): Likewise.
5940         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5941         * config/arm/fa606te.md (606te_mult1): Update attribute change.
5942         (606te_mult2): Likewise.
5943         (606te_mult3): Likewise.
5944         (606te_mult4): Likewise.
5945         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5946         (cortex_a9_mac16): Likewise.
5947         (cortex_a9_multiply): Likewise.
5948         (cortex_a9_mac): Likewise.
5949         (cortex_a9_multiply_long): Likewise.
5950         * config/arm/fa626te.md (626te_mult1): Update attribute change.
5951         (626te_mult2): Likewise.
5952         (626te_mult3): Likewise.
5953         (626te_mult4): Likewise.
5955 2013-06-18  Richard Biener  <rguenther@suse.de>
5957         PR lto/57334
5958         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5960 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5962         PR target/57609
5963         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5964         with next_active_insn.
5966 2013-06-18  Alan Modra  <amodra@gmail.com>
5968         * config/rs6000/rs6000.h (enum data_align): New.
5969         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5970         (DATA_ABI_ALIGNMENT): Define.
5971         (CONSTANT_ALIGNMENT): Correct comment.
5972         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5973         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5975 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
5977         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5978         ATTRIBUTE_UNUSED marking.
5980 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
5982         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5983         alternative and update.
5984         (aarch64_dup_lanedi): Delete.
5985         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5986         * config/aarch64/aarch64-simd-builtins.def: Update.
5988 2013-06-17  Richard Biener  <rguenther@suse.de>
5990         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5991         (lto_input_scc): Declare.
5992         (lto_input_tree_1): Likewise.
5993         (struct lto_stats_d): Add num_tree_bodies_output and
5994         num_pickle_refs_output.
5995         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5996         (lto_read_tree_1): Split out from ...
5997         (lto_read_tree): ... this.
5998         (lto_input_scc): New function.
5999         (lto_input_tree_1): Split out from ...
6000         (lto_input_tree): ... this.  Handle LTO_tree_scc.
6001         (lto_data_in_create): Create the streamer cache without hashes.
6002         * lto-streamer-out.c (create_output_block): Create the streamer
6003         cache with hashes when not doing WPA.
6004         (lto_write_tree_1): Split out from ...
6005         (lto_write_tree): ... this.
6006         (get_symbol_initial_value): New function.
6007         (lto_output_tree_1): Split out from ...
6008         (lto_output_tree): ... this.  Write trees as series of SCCs
6009         using a DFS walk via DFS_write_tree.
6010         (struct sccs, struct scc_entry): New types.
6011         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6012         (DFS_write_tree_body): New function.
6013         (DFS_write_tree): Likewise.
6014         (hash_tree): Likewise.
6015         (scc_entry_compare): Likewise.
6016         (hash_scc): Likewise.
6017         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6018         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6019         TREE_CHAIN as regular reference.
6020         (streamer_read_integer_cst): Remove.
6021         (streamer_get_pickled_tree): Adjust.
6022         * tree-streamer-out.c (streamer_write_chain): Disable streaming
6023         of DECL_EXTERNALs in BLOCK_VARS for now.
6024         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6025         reference.
6026         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6027         Add hash value argument and record that if hashes are recorded
6028         in the cache.
6029         (streamer_tree_cache_insert_1): Adjust.
6030         (streamer_tree_cache_insert): Likewise.
6031         (streamer_tree_cache_insert_at): Rename to ...
6032         (streamer_tree_cache_replace_tree): ... this and adjust.
6033         (streamer_tree_cache_append): Adjust.
6034         (record_common_node): Likewise.
6035         (streamer_tree_cache_create): Add argument whether to
6036         record hash values together with trees.
6037         (streamer_tree_cache_delete): Adjust.
6038         * tree-streamer.h (struct streamer_tree_cache_d): Add
6039         vector of hashes.
6040         (streamer_read_integer_cst): Remove.
6041         (streamer_tree_cache_insert): Adjust.
6042         (streamer_tree_cache_append): Likewise.
6043         (streamer_tree_cache_insert_at): Rename to ...
6044         (streamer_tree_cache_replace_tree): ... this and adjust.
6045         (streamer_tree_cache_create): Add argument whether to record hashes.
6046         (streamer_tree_cache_get): Rename to ...
6047         (streamer_tree_cache_get_tree): ... this.
6048         (streamer_tree_cache_get_hash): New function.
6049         * tree.c (cache_integer_cst): New function.
6050         * tree.h (cache_integer_cst): Declare.
6051         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6052         * lto-symtab.c (lto_varpool_replace_node): Only release
6053         DECL_INITIAL of non-prevailing decls.
6054         * varpool.c (varpool_remove_initializer): Do not release
6055         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6057 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6059         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6060         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6061         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6062         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6063         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6064         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6065         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6066         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6067         instead of TARGET_64BIT.
6068         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6069         Require ISA_HAS_<D>DIV.
6071 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
6073         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6074         (mips*-*-linux*): Move default with_llsc setting to where other
6075         defaults are set.
6076         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6077         with_arch block.
6078         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6079         Likewise.  Remove default with_tune setting.  Move default float
6080         setting to its own block.  Handle with_llsc in the same block as above.
6082 2013-06-16  Joern Rennecke <joern.rennecke@embecosm.com>
6084         PR rtl-optimization/57425
6085         PR rtl-optimization/57569
6086         * alias.c (write_dependence_p): Add new parameters mem_mode,
6087         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
6088         Changed all callers.
6089         (canon_anti_dependence): New function.
6090         * cse.c (check_dependence): Use canon_anti_dependence.
6091         * cselib.c (cselib_invalidate_mem): Likewise.
6092         * rtl.h (canon_anti_dependence): Declare.
6094 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6096         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6097         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6098         ".set mips3" for 64-bit targets.
6100 2013-06-15  Dehao Chen  <dehao@google.com>
6102         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6103         * gimple-low.c (gimple_check_call_matching_types): Likewise.
6104         (gimple_check_call_args): Likewise.
6105         * value-prof.c (check_ic_target): Likewise.
6106         * ipa-inline.c (early_inliner): Likewise.
6107         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6108         * cgraph.c (cgraph_create_edge_1): Likewise.
6109         (cgraph_make_edge_direct): Likewise.
6111 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
6113         PR target/57615
6114         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6115         rs6000_output_move_128bit to handle emitting quad memory
6116         operations.  Set attribute length to 8 bytes.
6118 2013-06-14  Vidya Praveen <vidyapraveen@arm.com>
6120         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6121         New pattern.
6122         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6123         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6124         (aarch64_<su>mlsl<mode>): Likewise.
6126 2013-06-14  Mike Stump  <mikestump@comcast.net>
6128         * Makefile.in (TARGET_H): Add insn-codes.h.
6130 2013-06-14  Alan Modra  <amodra@gmail.com>
6132         PR middle-end/57134
6133         PR middle-end/57586
6134         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6135         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
6136         bitfield expansion when EXPAND_MEMORY.
6137         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6139 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
6141         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6142         test for clearing quad memory on 32-bit later.
6144 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6146         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6147         (fold_negate_expr): Likewise.
6148         (fold_real_zero_addition_p): Handle vectors.
6149         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6151 2013-06-14  Alan Modra  <amodra@gmail.com>
6153         * varasm.c (force_const_mem): Revert 2013-06-07 change.
6155 2013-06-13  Jan Hubicka  <jh@suse.cz>
6157         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6158         Local comdats are not externally visible.
6159         * symtab.c (dump_symtab_base): Dump externally visible.
6160         (verify_symtab_base): Verify back links in the symtab hash.
6162 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6164         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6165         CONVERT_EXPR as equal nodes.
6167 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6169         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6171 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6173         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6174         Generalize to complex and vector.
6175         * tree.c (build_all_ones_cst): New function.
6176         * tree.h (build_all_ones_cst): Declare it.
6178 2013-06-13  Alan Modra  <amodra@gmail.com>
6180         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6181         * config/rs6000/rs6000.md (signbittf2): New insn.
6182         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6183         (abstf2_internal, cmptf_internal2): Likewise.
6184         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6186 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
6187             Pat Haugen <pthaugen@us.ibm.com>
6188             Peter Bergner <bergner@vnet.ibm.com>
6190         * config/rs6000/rs6000.c (emit_load_locked): Add support for
6191         power8 byte, half-word, and quad-word atomic instructions.
6192         (emit_store_conditional): Likewise.
6193         (rs6000_expand_atomic_compare_and_swap): Likewise.
6194         (rs6000_expand_atomic_op): Likewise.
6196         * config/rs6000/sync.md (larx): Add new modes for power8.
6197         (stcx): Likewise.
6198         (AINT): New mode iterator to include TImode as well as normal
6199         integer modes on power8.
6200         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
6201         that VSX registers are not considered.  Use AINT mode iterator
6202         instead of INT1 to allow inclusion of quad word atomic operations
6203         on power8.
6204         (load_locked<mode>): Likewise.
6205         (store_conditional<mode>): Likewise.
6206         (atomic_compare_and_swap<mode>): Likewise.
6207         (atomic_exchange<mode>): Likewise.
6208         (atomic_nand<mode>): Likewise.
6209         (atomic_fetch_<fetchop_name><mode>): Likewise.
6210         (atomic_nand_fetch<mode>): Likewise.
6211         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
6212         each type.
6213         (ATOMIC): On power8, add QImode, HImode modes.
6214         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
6215         modes that promote to SImode.
6216         (load_lockedti): Convert TImode arguments to PTImode, so that we
6217         get a guaranteed even/odd register pair.
6218         (load_lockedpti): Likewise.
6219         (store_conditionalti): Likewise.
6220         (store_conditionalpti): Likewise.
6222         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
6223         atomic load/store instructions.
6224         (HSI): Likewise.
6226 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
6228         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
6229         loads.
6230         (insn_count): New attribute, with most cases extracted from...
6231         (length): ...here.  Redefine most cases in terms of insn_count.
6232         (single_insn): Delete.
6233         (can_delay): Use insn_count to check for single instructions.
6234         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
6235         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
6236         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
6237         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
6238         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
6239         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
6240         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
6241         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
6242         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
6243         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
6244         rather than "length".
6245         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
6246         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
6247         Use "insn_count" rather than "length".
6248         * config/mips/mips-dsp.md
6249         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
6250         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
6251         length attributes.
6253 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
6255         PR tree-optimization/57361
6256         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
6258 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
6260         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
6261         to split.
6262         (aarch64_simd_combine<mode>): New instruction expansion.
6263         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
6264         function prototype.
6265         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
6266         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
6268 2013-06-12  Jan Hubicka  <jh@suse.cz>
6270         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
6271         decl has when in streaming stage.
6272         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
6273         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
6275 2013-06-12  Roland Stigge <stigge@antcom.de>
6277         PR target/57578
6278         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
6280 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6282         PR tree-optimization/57537
6283         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
6284         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
6286 2013-06-12  Richard Biener  <rguenther@suse.de>
6288         * data-streamer.h (streamer_write_char_stream): CSE
6289         obs->current_pointer.
6290         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
6291         streamer_write_char_stream manually and optimize the resulting loop.
6292         (streamer_write_hwi_stream): Likewise.
6294 2013-06-12  Jan Hubicka  <jh@suse.cz>
6296         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
6297         * cgraph.h (varpool_create_empty_node): Declare.
6298         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
6299         duplicated nodes.
6300         * symtab.c (symtab_unregister_node): Be lax about missin entries
6301         in node hash.
6302         (symtab_get_node): Update comment.
6303         * varpool.c (varpool_create_empty_node): Break out from ...
6304         (varpool_node_for_decl): ... here.
6305         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
6307 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
6309         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
6310         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
6311         part.  Use straight-line flow at the end.
6312         <COMPONENT_REF>: Remove superfluous else.
6313         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
6315 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6317         PR target/56564
6318         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
6319         target hook even for !TREE_PUBLIC decls.  If no resolution info
6320         is available, return false for common and external decls.
6322 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6324         * config/rl78/constraints.md (U): New constraint.
6325         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
6326         valloc attribute.
6328 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
6330         PR target/57589
6331         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
6332         to allow returning address to AT_PLATFORM name.
6334 2013-06-11  Jan Hubicka  <jh@suse.cz>
6336         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
6337         * cgraph.h (symtab_node_base): Add weakref flag.
6338         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
6339         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
6340         (output_weakrefs): Use weakref flag.
6341         * fold-const.c (simple_operand_p): Handle WEAK.
6342         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
6343         * ipa.c (varpool_externally_visible_p): Drop weakref.
6344         (function_and_variable_visibility): Update comment; fix weakref
6345         sanity checks; do not clear DECL_WEAK on them.
6346         * lto-cgraph.c (lto_output_node): update.
6347         (lto_output_varpool_node): Update.
6348         (input_overwrite_node): Update.
6349         (input_node): Update.
6350         (input_varpool_node): Update.
6351         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
6352         (lto_symtab_merge_symbols): Add sanity check.
6353         (lto_symtab_prevailing_decl): Do not special case weakrefs.
6354         * passes.c (rest_of_decl_compilation): Set static flag, too.
6355         * symtab.c (dump_symtab_base): Dump weakref.
6356         (verify_symtab_base): Sanity check weakrefs.
6357         (symtab_make_decl_local): Remove duplicated code.
6358         (symtab_alias_ultimate_target): Simplify.
6359         * varpool.c (varpool_create_variable_alias): Set weakref flag.
6361 2013-06-11  DJ Delorie  <dj@redhat.com>
6363         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
6364         (rl78_unwind_word_mode): New.
6366 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
6368         * final.c (debug_prefix_maps): Make static.
6370 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
6372         * function.c (initial_trampoline): Remove stray copy.
6374 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
6376         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
6378 2013-06-11  Martin Jambor  <mjambor@suse.cz>
6380         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
6381         within bounds at the beginning of the function.
6383 2013-06-11  Alan Modra  <amodra@gmail.com>
6385         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
6386         reporting.
6387         (get_named_section): Don't NULL !DECL_P decl.
6389 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
6391         * doc/invoke.texi (core-avx2): Document.
6392         (slm): Likewise.
6393         (atom): Updated with MOVBE.
6395 2013-06-11  Richard Biener  <rguenther@suse.de>
6397         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
6399 2013-06-11  Anton Blanchard  <anton@samba.org>
6401         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
6402         correct shift value in little-endian mode.
6404 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
6406         PR target/56564
6407         * varasm.c (get_variable_align): Move #endif to the right place.
6409 2013-06-10  Cary Coutant  <ccoutant@google.com>
6411         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
6412         for hash so that hash table traversal order is deterministic.
6414 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
6415             Pat Haugen <pthaugen@us.ibm.com>
6416             Peter Bergner <bergner@vnet.ibm.com>
6418         * config/rs6000/vector.md (GPR move splitter): Do not split moves
6419         of vectors in GPRS if they are direct moves or quad word load or
6420         store moves.
6422         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
6423         declaration.
6424         (direct_move_p): Likewise.
6425         (quad_load_store_p): Likewise.
6427         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
6428         classes into bins based on the physical register type.
6429         (reg_class_to_reg_type): Likewise.
6430         (IS_STD_REG_TYPE): Likewise.
6431         (IS_FP_VECT_REG_TYPE): Likewise.
6432         (reload_fpr_gpr): Arrays to determine what insn to use if we can
6433         use direct move instructions.
6434         (reload_gpr_vsx): Likewise.
6435         (reload_vsx_gpr): Likewise.
6436         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
6437         information that is a simplification of register classes.  Also
6438         precalculate direct move reload helpers.
6439         (direct_move_p): New function to return true if the operation can
6440         be done as a direct move instruciton.
6441         (quad_load_store_p): New function to return true if the operation
6442         is a quad memory operation.
6443         (rs6000_legitimize_address): If quad memory, only allow register
6444         indirect for TImode addresses.
6445         (rs6000_legitimate_address_p): Likewise.
6446         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
6447         (rs6000_reload_register_type): Likewise.
6448         (register_to_reg_type): Return register type.
6449         (rs6000_secondary_reload_simple_move): New helper function for
6450         secondary reload and secondary memory needed to identify anything
6451         that is a simple move, and does not need reloading.
6452         (rs6000_secondary_reload_direct_move): New helper function for
6453         secondary reload to identify cases that can be done with several
6454         instructions via the direct move instructions.
6455         (rs6000_secondary_reload_move): New helper function for secondary
6456         reload to identify moves between register types that can be done.
6457         (rs6000_secondary_reload): Add support for quad memory operations
6458         and for direct move.
6459         (rs6000_secondary_memory_needed): Likewise.
6460         (rs6000_debug_secondary_memory_needed): Change argument names.
6461         (rs6000_output_move_128bit): New function to return the move to
6462         use for 128-bit moves, including knowing about the various
6463         limitations of quad memory operations.
6465         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
6466         memory operations.  call rs6000_output_move_128bit for the actual
6467         instruciton(s) to generate.
6468         (vsx_movti_64bit): Likewise.
6470         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
6471         (UNSPEC_P8V_MTVSRWZ): Likewise.
6472         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
6473         (UNSPEC_P8V_MTVSRD): Likewise.
6474         (UNSPEC_P8V_XXPERMDI): Likewise.
6475         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
6476         (UNSPEC_FUSION_GPR): Likewise.
6477         (FMOVE128_GPR): New iterator for direct move.
6478         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
6479         (f32_sv): Likewise.
6480         (f32_dm): Likewise.
6481         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
6482         loads and direct move instructions.
6483         (zero_extendsidi2_lfiwzx): Likewise.
6484         (extendsidi2_lfiwax): Likewise.
6485         (extendsidi2_nocell): Likewise.
6486         (floatsi<mode>2_lfiwax): Likewise.
6487         (lfiwax): Likewise.
6488         (floatunssi<mode>2_lfiwzx): Likewise.
6489         (lfiwzx): Likewise.
6490         (fix_trunc<mode>_stfiwx): Likewise.
6491         (fixuns_trunc<mode>_stfiwx): Likewise.
6492         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
6493         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
6494         (parity<mode>2_cmpb): Set length/type attr.
6495         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
6496         for 'mr.' to fast_compare.
6497         (bpermd_<mode>): Change type attr to popcnt.
6498         (p8_fmrgow_<mode>): New insns for power8 direct move support.
6499         (p8_mtvsrwz_1): Likewise.
6500         (p8_mtvsrwz_2): Likewise.
6501         (reload_fpr_from_gpr<mode>): Likewise.
6502         (p8_mtvsrd_1): Likewise.
6503         (p8_mtvsrd_2): Likewise.
6504         (p8_xxpermdi_<mode>): Likewise.
6505         (reload_vsx_from_gpr<mode>): Likewise.
6506         (reload_vsx_from_gprsf): Likewise.
6507         (p8_mfvsrd_3_<mode>): LIkewise.
6508         (reload_gpr_from_vsx<mode>): Likewise.
6509         (reload_gpr_from_vsxsf): Likewise.
6510         (p8_mfvsrd_4_disf): Likewise.
6511         (multi-word GPR splits): Do not split direct moves or quad memory
6512         operations.
6514 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
6516         * tree-into-ssa.c (interesting_blocks): Make static.
6518 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
6520         PR target/56564
6521         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
6522         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
6523         Use DATA_ABI_ALIGNMENT for that case instead if defined.
6524         (get_variable_align): New function.
6525         (get_variable_section, emit_bss, emit_common,
6526         assemble_variable_contents, place_block_symbol): Use
6527         get_variable_align instead of DECL_ALIGN.
6528         (assemble_noswitch_variable): Add align argument, use it
6529         instead of DECL_ALIGN.
6530         (assemble_variable): Adjust caller.  Use get_variable_align
6531         instead of DECL_ALIGN.
6532         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
6533         caller.
6534         (DATA_ABI_ALIGNMENT): Define.
6535         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
6536         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
6537         opt is false, only return the psABI mandated alignment increase.
6538         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
6539         (DATA_ABI_ALIGNMENT): ... this.
6540         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
6541         (DATA_ABI_ALIGNMENT): ... this.
6542         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
6543         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
6544         (DATA_ABI_ALIGNMENT): ... this.
6545         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
6546         * doc/tm.texi: Regenerated.
6548 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
6550         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
6551         cmp_code to construct REG_EQUAL note.
6553 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
6555         PR target/57568
6556         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
6557         that operands[2] doesn't overlap with operands[0].
6559 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
6560             Jan Hubicka  <jh@suse.cz>
6562         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
6563         hack to mark symbols as used.
6565 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
6567         PR rtl-optimization/57559
6568         * lra-constraints.c (process_alt_operands): Don't discourage
6569         memory with known offset for offsetable memory constraint.
6570         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
6572 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
6574         * varasm.c (struct oc_local_state): Reorder fields.
6575         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
6576         and adjust accordingly.
6577         (output_constructor): Reorder initialization code and adjust call to
6578         output_constructor_bitfield.
6580 2013-06-07  Jan Hubicka  <jh@suse.cz>
6582         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
6584 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
6586         * tree-object-size.c (unknown): Make const.
6588 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6590         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
6591         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
6592         for last alternative in the cpu_facility attribute.
6594 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6596         PR target/56315
6597         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
6598         (xordi3): Change operand 2 constraint to arm_xordi_operand.
6599         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
6600         * config/arm/constraints.md (Dg): New constraint.
6601         * config/arm/neon.md (xordi3_neon): Remove.
6602         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
6603         * config/arm/predicates.md (arm_xordi_operand): New predicate.
6605 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6607         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
6608         Clean up alternatives.
6610 2013-06-07  Alan Modra  <amodra@gmail.com>
6612         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
6613         va_list_gpr_size.
6615 2013-06-07  Alan Modra  <amodra@gmail.com>
6617         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
6619 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6621         * config/arm/constraints.md (Df): New constraint.
6622         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
6623         Correct length attribute for last two alternatives.
6625 2013-06-07  Alan Modra  <amodra@gmail.com>
6627         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6628         override user -mfp-in-toc.
6629         (offsettable_ok_by_alignment): Consider just the current access
6630         rather than the whole object, unless BLKmode.  Handle
6631         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
6632         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
6633         for -mcmodel=medium.
6634         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
6635         override user -mfp-in-toc or -msum-in-toc.  Default to
6636         -mno-fp-in-toc for -mcmodel=medium.
6638 2013-06-06  DJ Delorie  <dj@redhat.com>
6640         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
6641         TARGET_VALID_POINTER_MODE.
6643 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
6644             Pat Haugen <pthaugen@us.ibm.com>
6645             Peter Bergner <bergner@vnet.ibm.com>
6647         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6648         Document new power8 builtins.
6650         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
6651         condition code register, to allow 128-bit logical operations to be
6652         done in the VSX or GPR registers.
6653         (nor<mode>3): Use the canonical form for nor.
6654         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
6655         vclz*, and vpopcnt* vector instructions.
6656         (nand<mode>3): Likewise.
6657         (orc<mode>3): Likewise.
6658         (clz<mode>2): LIkewise.
6659         (popcount<mode>2): Likewise.
6661         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
6662         that only the GPRs are recognized.
6664         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6665         support for new power8 builtins.
6667         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
6668         builtin functions.
6669         (xscvdpspn): Likewise.
6670         (vclz): Likewise.
6671         (vclzb): Likewise.
6672         (vclzh): Likewise.
6673         (vclzw): Likewise.
6674         (vclzd): Likewise.
6675         (vpopcnt): Likewise.
6676         (vpopcntb): Likewise.
6677         (vpopcnth): Likewise.
6678         (vpopcntw): Likewise.
6679         (vpopcntd): Likewise.
6680         (vgbbd): Likewise.
6681         (vmrgew): Likewise.
6682         (vmrgow): Likewise.
6683         (eqv): Likewise.
6684         (eqv_v16qi3): Likewise.
6685         (eqv_v8hi3): Likewise.
6686         (eqv_v4si3): Likewise.
6687         (eqv_v2di3): Likewise.
6688         (eqv_v4sf3): Likewise.
6689         (eqv_v2df3): Likewise.
6690         (nand): Likewise.
6691         (nand_v16qi3): Likewise.
6692         (nand_v8hi3): Likewise.
6693         (nand_v4si3): Likewise.
6694         (nand_v2di3): Likewise.
6695         (nand_v4sf3): Likewise.
6696         (nand_v2df3): Likewise.
6697         (orc): Likewise.
6698         (orc_v16qi3): Likewise.
6699         (orc_v8hi3): Likewise.
6700         (orc_v4si3): Likewise.
6701         (orc_v2di3): Likewise.
6702         (orc_v4sf3): Likewise.
6703         (orc_v2df3): Likewise.
6705         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
6706         allow power8 quad mode in 64-bit.
6707         (rs6000_builtin_vectorized_function): Add support to vectorize
6708         ISA 2.07 count leading zeros, population count builtins.
6709         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
6710         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
6711         (builtin_function_type): Add vgbbd builtin function which takes an
6712         unsigned argument.
6713         (altivec_expand_vec_perm_const): Add support for new power8 merge
6714         instructions.
6716         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
6717         that does not include TImdoe for use with 32-bit.
6718         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
6719         instructions.
6720         (UNSPEC_VSX_CVDPSPN): Likewise.
6721         (vsx_xscvdpspn): Likewise.
6722         (vsx_xscvspdpn): Likewise.
6723         (vsx_xscvdpspn_scalar): Likewise.
6724         (vsx_xscvspdpn_directmove): Likewise.
6725         (vsx_and<mode>3): Split logical operations into 32-bit and
6726         64-bit. Add support to do logical operations on TImode as well as
6727         VSX vector types.  Allow logical operations to be done in either
6728         VSX registers or in general purpose registers in 64-bit mode.  Add
6729         splitters if GPRs were used. For AND, add clobber of CCmode to
6730         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
6731         encoding.
6732         (vsx_and<mode>3_32bit): Likewise.
6733         (vsx_and<mode>3_64bit): Likewise.
6734         (vsx_ior<mode>3): Likewise.
6735         (vsx_ior<mode>3_32bit): Likewise.
6736         (vsx_ior<mode>3_64bit): Likewise.
6737         (vsx_xor<mode>3): Likewise.
6738         (vsx_xor<mode>3_32bit): Likewise.
6739         (vsx_xor<mode>3_64bit): Likewise.
6740         (vsx_one_cmpl<mode>2): Likewise.
6741         (vsx_one_cmpl<mode>2_32bit): Likewise.
6742         (vsx_one_cmpl<mode>2_64bit): Likewise.
6743         (vsx_nor<mode>3): Likewise.
6744         (vsx_nor<mode>3_32bit): Likewise.
6745         (vsx_nor<mode>3_64bit): Likewise.
6746         (vsx_andc<mode>3): Likewise.
6747         (vsx_andc<mode>3_32bit): Likewise.
6748         (vsx_andc<mode>3_64bit): Likewise.
6749         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
6750         and xxlorc instructions.
6751         (vsx_eqv<mode>3_64bit): Likewise.
6752         (vsx_nand<mode>3_32bit): Likewise.
6753         (vsx_nand<mode>3_64bit): Likewise.
6754         (vsx_orc<mode>3_32bit): Likewise.
6755         (vsx_orc<mode>3_64bit): Likewise.
6757         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
6759         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
6760         instruction.
6761         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
6762         (p8_vmrgow): Likewise.
6763         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
6764         GPRs to be split under VSX.
6765         (p8v_clz<mode>2): Add power8 count leading zero support.
6766         (p8v_popcount<mode>2): Add power8 population count support.
6767         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
6768         support.
6770         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
6771         instruction.
6773         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
6774         builtin functions.
6775         (vec_nand): Likewise.
6776         (vec_vclz): Likewise.
6777         (vec_vclzb): Likewise.
6778         (vec_vclzd): Likewise.
6779         (vec_vclzh): Likewise.
6780         (vec_vclzw): Likewise.
6781         (vec_vgbbd): Likewise.
6782         (vec_vmrgew): Likewise.
6783         (vec_vmrgow): Likewise.
6784         (vec_vpopcnt): Likewise.
6785         (vec_vpopcntb): Likewise.
6786         (vec_vpopcntd): Likewise.
6787         (vec_vpopcnth): Likewise.
6788         (vec_vpopcntw): Likewise.
6790 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6792         PR rtl-optimization/57468
6793         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
6794         spilled pseudos.
6796 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6798         PR rtl-optimization/57459
6799         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
6800         type when setting live regs.
6802 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6804         * config/s390/s390.opt (mlra): New option.
6805         * config/s390/s390.c (s390_decompose_address): Check displacement
6806         for all registers for LRA.
6807         (s390_secondary_reload): Don't used secondary reloads for LRA.
6808         (s390_lra_p): New function.
6809         (TARGET_LRA_P): Define.
6810         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
6811         of attribute cpu_facility to zarch for the last alternative.
6812         (*cmpmem_short): Ditto.
6814 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
6816         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
6817         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
6818         (arm_expand_prologue): Likewise.
6820 2013-06-06  Teresa Johnson  <tejohnson@google.com>
6822         PR c++/53743
6823         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
6824         as this is now done by redirect_edge_and_branch_force.
6825         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
6826         barriers, and fix interaction with splitting.
6827         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
6828         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
6829         reflect changes made in the routine.
6830         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
6831         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
6832         since this is called in cfglayout mode, and replace partition fixup
6833         with assert as that is now done by force_nonfallthru_and_redirect.
6834         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
6835         already be marked with region crossing note.
6836         (insert_section_boundary_note): Make non-static, gate on flag
6837         has_bb_partition, rewrite to also check for multiple partitions.
6838         (rest_of_handle_reorder_blocks): Remove call to
6839         insert_section_boundary_note, now done later during free_cfg.
6840         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
6841         * bb-reorder.h (insert_section_boundary_note): Declare.
6842         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
6843         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
6844         invoke insert_section_boundary_note.
6845         (try_redirect_by_replacing_jump): Remove unnecessary
6846         check for region crossing note.
6847         (fixup_partition_crossing): New function.
6848         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
6849         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
6850         in non-cfglayout mode.
6851         (force_nonfallthru_and_redirect): Fixup partition boundaries,
6852         remove old code that tried to do this. Emit barrier correctly
6853         when we are in cfglayout mode.
6854         (last_bb_in_partition): New function.
6855         (rtl_split_edge): Correctly fixup partition boundaries.
6856         (commit_one_edge_insertion): Remove old code that tried to
6857         fixup region crossing edge since this is now handled in
6858         split_block, and set up insertion point correctly since
6859         block may now end in a jump.
6860         (verify_hot_cold_block_grouping): Guard against checking when not in
6861         linearized RTL mode.
6862         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
6863         notes.
6864         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
6865         rtl_verify_flow_info, so not called in cfglayout mode.
6866         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
6867         (fixup_reorder_chain): Remove old code that attempted to fixup region
6868         crossing note as this is now handled in force_nonfallthru_and_redirect.
6869         (duplicate_insn_chain): Don't duplicate switch section notes.
6870         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
6871         note.
6872         * basic-block.h (emit_barrier_after_bb): Declare.
6874 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6876         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
6877         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
6878         arm_usatsihi): Adjust alternatives for arm_restrict_it.
6880 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6882         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
6883         where appropriate.
6884         * config/arm/ldmstm.md: Regenerate.
6886 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6888         * config/arm/sync.md (atomic_loaddi_1):
6889         Disable predication for arm_restrict_it.
6890         (arm_load_exclusive<mode>): Likewise.
6891         (arm_load_exclusivesi): Likewise.
6892         (arm_load_exclusivedi): Likewise.
6893         (arm_load_acquire_exclusive<mode>): Likewise.
6894         (arm_load_acquire_exclusivesi): Likewise.
6895         (arm_load_acquire_exclusivedi): Likewise.
6896         (arm_store_exclusive<mode>): Likewise.
6897         (arm_store_exclusive<mode>): Likewise.
6898         (arm_store_release_exclusivedi): Likewise.
6899         (arm_store_release_exclusive<mode>): Likewise.
6901 2013-06-06  Richard Biener  <rguenther@suse.de>
6903         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
6904         after LTO_null.
6905         (lto_tag_is_tree_code_p): Adjust.
6906         (lto_tag_is_gimple_code_p): Likewise.
6907         (lto_gimple_code_to_tag): Likewise.
6908         (lto_tag_to_gimple_code): Likewise.
6909         (lto_tree_code_to_tag): Likewise.
6910         (lto_tag_to_tree_code): Likewise.
6911         * data-streamer.h (streamer_write_hwi_in_range): Use
6912         uhwi streaming to stream the normalized range.
6913         (streamer_read_hwi_in_range): Likewise.
6915 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6917         * config/arm/arm.md (enabled_for_depr_it): New attribute.
6918         (predicable_short_it): Likewise.
6919         (predicated): Likewise.
6920         (enabled): Handle above.
6921         (define_cond_exec): Set predicated attribute to yes.
6923 2013-06-05  Mike Stump  <mikestump@comcast.net>
6925         * gdbinit.in (__FUNCTION__): Add.
6927 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
6929         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6930         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6932 2013-06-05  Jan Hubicka  <jh@suse.cz>
6934         * varasm.c (mark_decl_referenced): Revert the removal until targets
6935         are fixed.
6937 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
6939         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6940         instead of mark_decl_referenced.
6942 2013-06-05  Jan Hubicka  <jh@suse.cz>
6944         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6945         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6946         and symtab_used_from_object_file_p.
6947         (cgraph_make_node_local_1): Clear forced_by_abi.
6948         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6949         * cgraph.h (symtab_node_base): Add forced_by_abi.
6950         (decide_is_variable_needed): Remove.
6951         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6952         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6953         (decide_is_symbol_needed): ... this one; handle symbols in general;
6954         always analyze virtuals; honnor forced_by_abi.
6955         (cgraph_finalize_function): Update.
6956         (varpool_finalize_decl): Update.
6957         (symbol_defined_and_needed): Remove.
6958         (analyze_functions): Update.
6959         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6960         output_refs, input_overwrite_node): Handle forced_by_abi.
6961         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6962         (address_taken_from_non_vtable_p): ... this one.
6963         (comdat_can_be_unshared_p_1): New function.
6964         (cgraph_comdat_can_be_unshared_p): Rename to ...
6965         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6966         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6967         (function_and_variable_visibility): Clear forced_by_abi as needed.
6968         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6969         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6970         * symtab.c (dump_symtab_base): Dump forced_by_abi.
6971         * varpool.c (decide_is_variable_needed): Remove.
6973 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6975         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6976         (arm_option_override): Override arm_restrict_it where appropriate.
6977         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6978         * config/arm/arm.opt (mrestrict-it): New command-line option.
6979         * doc/invoke.texi: Document -mrestrict-it.
6981 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6983         * tsan.c (tsan_atomic_table): Make const.
6985 2013-06-05  Richard Biener  <rguenther@suse.de>
6987         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6988         index associated with the tree we are supposed to replace.
6989         * tree-streamer-out.c (pack_ts_base_value_fields): Output
6990         TREE_ASM_WRITTEN as zero for everything but SSA names.
6992 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
6994         * tree-ssa-structalias.c (call_stmt_vars): Make static.
6996 2013-06-04  Jan Hubicka  <jh@suse.cz>
6998         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6999         (input_node, input_varpool_node): Handle correctly external same
7000         body aliases.
7001         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7002         nodes at ltrans stage.
7004 2013-06-04  Jan Hubicka  <jh@suse.cz>
7006         * ipa-inline.c (update_caller_keys): Fix availability test.
7007         (update_callee_keys): Likewise.
7008         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7009         to follow ELF standard.
7011 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
7013         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7014         (mips64r5900el-*-elf*): New configurations.
7015         * config/mips/mips-cpus.def (r5900): New processor.
7016         * config/mips/mips-tables.opt: Regenerate.
7017         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7018         (mips_issue_rate): Handle PROCESSOR_R5900.
7019         (mips_reorg_process_insns): Force reorder mode for the R5900.
7020         * config/mips/mips.h (TARGET_MIPS5900): Define.
7021         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7022         TARGET_MIPS5900.
7023         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7024         TARGET_MIPS5900.
7025         * config/mips/mips.md (processor): Add r5900.
7026         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7028 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7030         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7031         into function to generate MOVI instruction.
7032         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7033         (aarch64_preferred_simd_mode): Turn into wrapper.
7034         (aarch64_output_scalar_simd_mov_immediate): New function.
7035         * config/aarch64/aarch64-protos.h: Add prototype for above.
7037 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7039         * config/aarch64/aarch64.c (simd_immediate_info): Remove
7040         element_char member.
7041         (sizetochar): Return signed char.
7042         (aarch64_simd_valid_immediate): Remove elchar and other
7043         unnecessary variables.
7044         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7045         Calculate element_char as required.
7046         * config/aarch64/aarch64-protos.h: Update and move prototype
7047         for aarch64_output_simd_mov_immediate.
7048         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7049         Update arguments.
7051 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7053         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7054         information completed by aarch64_simd_valid_immediate.
7055         (aarch64_legitimate_constant_p): Update arguments.
7056         (aarch64_simd_valid_immediate): Work with struct rather than many
7057         pointers.
7058         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7059         (aarch64_simd_make_constant): Update arguments.
7060         (aarch64_output_simd_mov_immediate): Work with struct rather than
7061         many pointers.  Output immediate directly rather than as operand.
7062         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7063         Update prototype.
7064         * config/aarch64/constraints.md (Dn): Update arguments.
7066 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7068         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7069         longer static.
7070         (aarch64_simd_immediate_valid_for_move): Remove.
7071         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7072         (aarch64_simd_make_constant): Update call.
7073         (aarch64_output_simd_mov_immediate): Update call.
7074         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7075         Add prototype.
7076         * config/aarch64/constraints.md (Dn): Update call.
7078 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7080         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7081         return type to bool for prototype.
7082         (aarch64_legitimate_constant_p): Check for true instead of not -1.
7083         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7084         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7086 2013-06-04  Catherine Moore  <clm@codesourcery.com>
7088         * config/mips/mips.opt (meva): New.
7089         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7090         (ASM_SPEC): Handle -meva.
7091         * doc/invoke.texi (meva):  Document.
7093 2013-06-04  Alan Modra  <amodra@gmail.com>
7095         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7096         constant output.
7098 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7100         * rtl.def: Add extra fourth optional field to define_cond_exec.
7101         * gensupport.c (process_one_cond_exec): Process attributes from
7102         define_cond_exec.
7103         * doc/md.texi: Document fourth field in define_cond_exec.
7105 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
7107         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7108         out the processing order as in store_bit_field_1.
7110 2013-06-04  Jan Hubicka  <jh@suse.cz>
7112         PR middle-end/57500
7113         * cgraphunit.c (cgraph_process_same_body_aliases): Create
7114         non-VAR_DECL node if it does not exist yet.
7116 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
7118         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7119         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7120         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7121         target_cpu_default setting.
7123 2013-06-03  Teresa Johnson  <tejohnson@google.com>
7125         * dumpfile.c (opt_info_switch_p): Change -fopt-info
7126         default to -fopt-info=optimized instead of all.
7127         * doc/invoke.texi: Ditto.
7128         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7129         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7130         (execute_vect_slp): Emit BB vectorization success under
7131         MSG_OPTIMIZED_LOCATIONS.
7132         * tree-vect-slp.c (vect_slp_transform_bb): Change
7133         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7134         * tree-vect-loop.c (vect_transform_loop): Ditto.
7136 2013-06-03  Jason Merrill  <jason@redhat.com>
7138         PR c++/57415
7139         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7140         Use TARGET_EXPR for C++.
7142 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
7144         PR rtl-optimization/57268
7145         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7146         if DEBUG_INSN_P (insn).
7148         Reapply
7149         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7151         PR rtl-optimization/57268
7152         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7153         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7155 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7157         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7158         (ix86_avoid_lea_for_addr): Likewise.
7159         (exact_dependency_1): Likewise.
7160         (ix86_adjust_cost): Likewise.
7161         (swap_top_of_ready_list): Fix formatting and !reload_completed check
7162         removed.
7163         (do_reorder_for_imul): Fix typo, formatting and
7164         !reload_completed check removed.
7165         (ix86_sched_reorder): Fix typo and formatting.
7166         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7167         list.
7169 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
7171         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7173 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
7175         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7176         <STRING_CST>: Likewise.
7177         <VECTOR_CST>: Likewise.
7179 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
7180             Mikael Morin  <mikael@gcc.gnu.org>
7182         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7183         * config.in: Regenerated.
7184         * configure: Regenerated.
7186 2013-06-01  Jan Hubicka  <jh@suse.cz>
7188         PR middle-end/57366
7189         * cgraphunit.c (compile): When weakref is not supported,
7190         set up transparent aliases before final output pass.
7191         * varasm.c (assemble_alias): Do not try to do it here.
7193 2013-06-01  Jan Hubicka  <jh@suse.cz>
7195         PR middle-end/57467
7196         * passes.c (for_per_function): Skip unanalyzed functions.
7198 2013-06-01  Jan Hubicka  <jh@suse.cz>
7200         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
7201         (lto_symtab_merge_symbols_1): ... this one.
7202         (lto_symtab_merge_cgraph_nodes): Rename to ...
7203         (lto_symtab_merge_symbols): ... this one; simplify.
7204         * cgraph.c (same_body_aliases_done): Rename to ...
7205         (cpp_implicit_aliases_done): ... this one.
7206         (cgraph_create_function_alias): Update.
7207         (cgraph_same_body_alias): Update.
7208         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
7209         (verify_edge_corresponds_to_fndecl): Simplify.
7210         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
7211         (cgraph_node): Remove same_body_alias.
7212         (varpool_node): Remove alias_of and extra_name_alias.
7213         (same_body_aliases_done): Rename to ..
7214         (cpp_implicit_aliases_done): ... this one.
7215         (symtab_alias_ultimate_target): Add default parameter.
7216         (symtab_resolve_alias): New function.
7217         (fixup_same_cpp_alias_visibility): Declare.
7218         (cgraph_function_node): Add default parameter.
7219         (cgraph_node_asm_name): Likewise.
7220         (cgraph_function_or_thunk_node): Add default parameter; do
7221         not ICE when it is NULL.
7222         (varpool_variable_node): Likewise.
7223         * tree-emutls.c (create_emultls_var): Update.
7224         (ipa_lower_emutls): Update.
7225         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
7226         (cgraph_reset_node): Reset alias info.
7227         (cgraph_finalize_function): Update.
7228         (fixup_same_cpp_alias_visibility): Move to symtab.c.
7229         (analyze_function): Simplify.
7230         (cgraph_process_same_body_aliases): Simplify.
7231         (analyze_functions): Fixup same body aliases.
7232         (handle_alias_pairs): Simplify.
7233         (assemble_thunk): Update.
7234         (assemble_thunks_and_aliases): Update.
7235         (output_weakrefs): Rewrite.
7236         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
7237         (lto_output_varpool_node): Likewise.
7238         (compute_ltrans_boundary): Remve assert.
7239         (get_alias_symbol): New functoin.
7240         (input_node): Rewrite alias handling.
7241         (input_varpool_node): Likewise.
7242         * ipa-pure-const.c (propagate_pure_const): Fix formating.
7243         * ipa.c (process_references): Handle weakrefs correctly.
7244         (symtab_remove_unreachable_nodes): Likewise.
7245         * trans-mem.c (get_cg_data): Update.
7246         (ipa_tm_create_version_alias): Update.
7247         (ipa_tm_execute): Update.
7248         * symtab.c (dump_symtab_base): Dump aliases.
7249         (verify_symtab_base): Verify aliases.
7250         (symtab_node_availability): New function.
7251         (symtab_alias_ultimate_target): Simplify.
7252         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
7253         handle all the fixup cases.
7254         (symtab_resolve_alias): New function.
7255         * passes.c (ipa_write_summaries): Handle weakrefs.
7256         * varpool.c (varpool_analyze_node): Simplify.
7257         (assemble_aliases): Update.
7258         (varpool_create_variable_alias): Simplify.
7259         (varpool_extra_name_alias): Simplify.
7260         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
7261         (lto_symtab_merge_symbols): ... this one.
7263 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
7265         Revert
7266         PR rtl-optimization/57268
7267         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7268         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7270 2013-06-01  Tobias Burnus  <burnus@net-b.de>
7272         Partially reverted:
7273         2013-05-31  Tobias Burnus  <burnus@net-b.de>
7275         PR middle-end/57073
7276         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
7277         further up.
7279 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7281         PR rtl-optimization/57268
7282         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7283         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7285 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7287         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
7288         unordered comparison operators when -fno-trapping-math is in effect
7289         on the e500.
7290         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
7291         and implement unordered comparison operators properly on the e500.
7293 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7295         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
7296         for constant scalar integers.
7297         (simplify_relational_operation_1): Likewise.
7299 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
7301         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
7302         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
7303         Fix comment.
7305 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7306             Igor Zamyatin  <igor.zamyatin@intel.com>
7308         Silvermont (SLM) architecture performance tuning.
7309         * config/i386/i386.h (enum ix86_tune_indices): Add
7310         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
7311         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
7313         * config/i386/i386.c (initial_ix86_tune_features)
7314         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
7315         (ix86_lea_outperforms): Handle Silvermont tuning.
7316         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
7317         call.
7318         (ix86_use_lea_for_mov): Likewise.
7319         (ix86_avoid_lea_for_addr): Likewise.
7320         (ix86_lea_for_add_ok): Likewise.
7321         (exact_dependency_1): New function.
7322         (exact_store_load_dependency): Likewise.
7323         (ix86_adjust_cost): Handle Silvermont tuning.
7324         (do_reoder_for_imul): Likewise.
7325         (swap_top_of_ready_list): New function.
7326         (ix86_sched_reorder): Changed to handle Silvermont tuning.
7328         * config/i386/i386.md (peepholes that split memory operand in fp
7329         converts): New.
7331 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7333         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7334         Remove un-necessary braces.
7336 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7338         * config/aarch64/aarch64.c (aarch64_classify_symbol):
7339         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
7341 2013-05-31  Tobias Burnus  <burnus@net-b.de>
7343         PR middle-end/57073
7344         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
7346 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7348         PR target/56315
7349         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
7350         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
7351         * config/arm/neon.md (iordi3_neon): Remove.
7352         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
7353         * config/arm/predicates.md (imm_for_neon_logic_operand):
7354         Move to earlier in the file.
7355         (neon_logic_op2): Likewise.
7356         (arm_iordi_operand_neon): New predicate.
7358 2013-05-31  Richard Biener  <rguenther@suse.de>
7360         PR tree-optimization/57478
7361         PR tree-optimization/57453
7362         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
7363         are life as well.
7365 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
7367         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
7368         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
7370 2013-05-30  Tobias Burnus  <burnus@net-b.de>
7371             Thomas Koenig  <tkoenig@gcc.gnu.org>
7373         PR middle-end/57073
7374         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
7375         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
7377 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
7379         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
7381 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
7383         * target.def (register_usage_leveling_p): New hook.
7384         * targhooks.c (default_register_usage_leveling_p): New.
7385         * targhooks.h (default_register_usage_leveling_p): New prototype.
7386         * lra-assigns.c (register_usage_leveling_p): Use the hook.
7387         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
7388         * doc/tm.texi: Update.
7389         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
7391 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
7393         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
7394         (*insv_reg<mode>): New define_insn.
7396 2013-05-30  Joern Rennecke <joern.rennecke@embecosm.com>
7398         PR rtl-optimization/57439
7399         * postreload.c (move2add_valid_value_p): Check that we have
7400         a zero subreg_regno_offset when accessing the register in
7401         the requested mode.
7403 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7404             Igor Zamyatin  <igor.zamyatin@intel.com>
7406         Silvermont (SLM) architecture pipeline model, tuning and
7407         insn selection.
7408         * config.gcc: Add slm config options and target.
7410         * config/i386/slm.md: New.
7412         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
7414         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
7415         PROCESSOR_SLM.
7416         (ix86_target_macros_internal): Likewise.
7418         * gcc/config/i386/i386.c (slm_cost): New cost.
7419         (m_SLM): New macro flag.
7420         (initial_ix86_tune_features): Set m_SLM.
7421         (x86_accumulate_outgoing_args): Likewise.
7422         (x86_arch_always_fancy_math_387): Likewise.
7423         (processor_target_table): Add slm cost.
7424         (cpu_names): Add slm cpu name.
7425         (x86_option_override_internal): Set SLM ISA.
7426         (ix86_issue_rate): New case PROCESSOR_SLM.
7427         (ia32_multipass_dfa_lookahead): Likewise.
7428         (fold_builtin_cpu): Add slm.
7430         * config/i386/i386.h (TARGET_SLM): New target macro.
7431         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
7432         (processor_type): Add PROCESSOR_SLM.
7434         * config/i386/i386.md (cpu): Add new value "slm".
7435         (slm.md): Include slm.md.
7437 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
7438             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7440         * config/arm/arm-protos.h: Add and update function protos.
7441         * config/arm/arm.c (use_simple_return_p): New added.
7442         (thumb2_expand_return): Check simple_return flag.
7443         * config/arm/arm.md: Add simple_return and conditional simple_return.
7444         * config/arm/iterators.md: Add iterator for return and simple_return.
7446 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7448         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
7449         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
7450         (arm_emit_vfp_multi_reg_pop): Likewise.
7451         (thumb2_emit_ldrd_pop): Likewise.
7452         (arm_expand_epilogue): Add misc REG_CFA notes.
7453         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
7455 2013-05-29  Lawrence Crowl  <crowl@google.com>
7457         * config/arm/t-arm: Update for below.
7459         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
7460         Change type to hash_table.  Update dependent calls and types.
7462         * config/i386/t-cygming: Update for below.
7464         * config/i386/t-interix: Update for below.
7466         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
7467         Change type to hash_table.  Update dependent calls and types.
7468         (i386_find_on_wrapper_list::wrappers): Likewise.
7470         * config/ia64/t-ia64: Update for below.
7472         * config/ia64/ia64.c (bundle_state_table):
7473         Change type to hash_table.  Update dependent calls and types.
7475         * config/mips/mips.c (mips_reorg_process_insns::htab):
7476         Change type to hash_table.  Update dependent calls and types.
7478         * config/sol2.c (solaris_comdat_htab):
7479         Change type to hash_table.  Update dependent calls and types.
7481         * config/t-sol2: Update for above.
7483 2013-05-29  Teresa Johnson  <tejohnson@google.com>
7485         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
7486         functions are not yet marked as defined.
7488 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
7489             Pat Haugen <pthaugen@us.ibm.com>
7490             Peter Bergner <bergner@vnet.ibm.com>
7492         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
7493         instructions.
7494         (VEC_A): Likewise.
7495         (VEC_C): Likewise.
7496         (vrotl<mode>3): Likewise.
7497         (vashl<mode>3): Likewise.
7498         (vlshr<mode>3): Likewise.
7499         (vashr<mode>3): Likewise.
7501         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7502         support for power8 V2DI builtins.
7504         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
7505         power8 V2DI builtins.
7506         (vupkhsw): Likewise.
7507         (vupklsw): Likewise.
7508         (vaddudm): Likewise.
7509         (vminsd): Likewise.
7510         (vmaxsd): Likewise.
7511         (vminud): Likewise.
7512         (vmaxud): Likewise.
7513         (vpkudum): Likewise.
7514         (vpksdss): Likewise.
7515         (vpkudus): Likewise.
7516         (vpksdus): Likewise.
7517         (vrld): Likewise.
7518         (vsld): Likewise.
7519         (vsrd): Likewise.
7520         (vsrad): Likewise.
7521         (vsubudm): Likewise.
7522         (vcmpequd): Likewise.
7523         (vcmpgtsd): Likewise.
7524         (vcmpgtud): Likewise.
7525         (vcmpequd_p): Likewise.
7526         (vcmpgtsd_p): Likewise.
7527         (vcmpgtud_p): Likewise.
7528         (vupkhsw): Likewise.
7529         (vupklsw): Likewise.
7530         (vaddudm): Likewise.
7531         (vmaxsd): Likewise.
7532         (vmaxud): Likewise.
7533         (vminsd): Likewise.
7534         (vminud): Likewise.
7535         (vpksdss): Likewise.
7536         (vpksdus): Likewise.
7537         (vpkudum): Likewise.
7538         (vpkudus): Likewise.
7539         (vrld): Likewise.
7540         (vsld): Likewise.
7541         (vsrad): Likewise.
7542         (vsrd): Likewise.
7543         (vsubudm): Likewise.
7545         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
7546         support for power8 V2DI instructions.
7548         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
7549         power8 V2DI instructions.  Combine pack and unpack insns to use an
7550         iterator for each mode.  Check whether a particular mode supports
7551         Altivec instructions instead of just checking TARGET_ALTIVEC.
7552         (UNSPEC_VPKUWUM): Likewise.
7553         (UNSPEC_VPKSHSS): Likewise.
7554         (UNSPEC_VPKSWSS): Likewise.
7555         (UNSPEC_VPKUHUS): Likewise.
7556         (UNSPEC_VPKSHUS): Likewise.
7557         (UNSPEC_VPKUWUS): Likewise.
7558         (UNSPEC_VPKSWUS): Likewise.
7559         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
7560         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
7561         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
7562         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
7563         (UNSPEC_VUPKHSB): Likewise.
7564         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
7565         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
7566         (UNSPEC_VUPKHSH): Likewise.
7567         (UNSPEC_VUPKLSB): Likewise.
7568         (UNSPEC_VUPKLSH): Likewise.
7569         (VI2): Likewise.
7570         (VI_char): Likewise.
7571         (VI_scalar): Likewise.
7572         (VI_unit): Likewise.
7573         (VP): Likewise.
7574         (VP_small): Likewise.
7575         (VP_small_lc): Likewise.
7576         (VU_char): Likewise.
7577         (add<mode>3): Likewise.
7578         (altivec_vaddcuw): Likewise.
7579         (altivec_vaddu<VI_char>s): Likewise.
7580         (altivec_vadds<VI_char>s): Likewise.
7581         (sub<mode>3): Likewise.
7582         (altivec_vsubcuw): Likewise.
7583         (altivec_vsubu<VI_char>s): Likewise.
7584         (altivec_vsubs<VI_char>s): Likewise.
7585         (altivec_vavgs<VI_char>): Likewise.
7586         (altivec_vcmpbfp): Likewise.
7587         (altivec_eq<mode>): Likewise.
7588         (altivec_gt<mode>): Likewise.
7589         (altivec_gtu<mode>): Likewise.
7590         (umax<mode>3): Likewise.
7591         (smax<mode>3): Likewise.
7592         (umin<mode>3): Likewise.
7593         (smin<mode>3): Likewise.
7594         (altivec_vpkuhum): Likewise.
7595         (altivec_vpkuwum): Likewise.
7596         (altivec_vpkshss): Likewise.
7597         (altivec_vpkswss): Likewise.
7598         (altivec_vpkuhus): Likewise.
7599         (altivec_vpkshus): Likewise.
7600         (altivec_vpkuwus): Likewise.
7601         (altivec_vpkswus): Likewise.
7602         (altivec_vpks<VI_char>ss): Likewise.
7603         (altivec_vpks<VI_char>us): Likewise.
7604         (altivec_vpku<VI_char>us): Likewise.
7605         (altivec_vpku<VI_char>um): Likewise.
7606         (altivec_vrl<VI_char>): Likewise.
7607         (altivec_vsl<VI_char>): Likewise.
7608         (altivec_vsr<VI_char>): Likewise.
7609         (altivec_vsra<VI_char>): Likewise.
7610         (altivec_vsldoi_<mode>): Likewise.
7611         (altivec_vupkhsb): Likewise.
7612         (altivec_vupkhs<VU_char>): Likewise.
7613         (altivec_vupkls<VU_char>): Likewise.
7614         (altivec_vupkhsh): Likewise.
7615         (altivec_vupklsb): Likewise.
7616         (altivec_vupklsh): Likewise.
7617         (altivec_vcmpequ<VI_char>_p): Likewise.
7618         (altivec_vcmpgts<VI_char>_p): Likewise.
7619         (altivec_vcmpgtu<VI_char>_p): Likewise.
7620         (abs<mode>2): Likewise.
7621         (vec_unpacks_hi_v16qi): Likewise.
7622         (vec_unpacks_hi_v8hi): Likewise.
7623         (vec_unpacks_lo_v16qi): Likewise.
7624         (vec_unpacks_hi_<VP_small_lc>): Likewise.
7625         (vec_unpacks_lo_v8hi): Likewise.
7626         (vec_unpacks_lo_<VP_small_lc>): Likewise.
7627         (vec_pack_trunc_v8h): Likewise.
7628         (vec_pack_trunc_v4si): Likewise.
7629         (vec_pack_trunc_<mode>): Likewise.
7631         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
7632         V2DI builtins.
7633         (vec_vmaxsd): Likewise.
7634         (vec_vmaxud): Likewise.
7635         (vec_vminsd): Likewise.
7636         (vec_vminud): Likewise.
7637         (vec_vpksdss): Likewise.
7638         (vec_vpksdus): Likewise.
7639         (vec_vpkudum): Likewise.
7640         (vec_vpkudus): Likewise.
7641         (vec_vrld): Likewise.
7642         (vec_vsld): Likewise.
7643         (vec_vsrad): Likewise.
7644         (vec_vsrd): Likewise.
7645         (vec_vsubudm): Likewise.
7646         (vec_vupkhsw): Likewise.
7647         (vec_vupklsw): Likewise.
7649 2013-05-29  Jan Hubicka  <jh@suse.cz>
7651         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
7652         flags; reorder rest of fields in more consistent way.
7653         (varpool_node): Remove analyzed, finalized and alias.
7654         (cgraph_ndoe): Likewise.
7655         (symtab_alias_ultimate_target): New function.
7656         (cgraph_function_node): Move offline.
7657         (cgraph_reset_node): Declare.
7658         (cgraph_comdat_can_be_unshared_p): Remove.
7659         (varpool_remove_initializer): Declare.
7660         (varpool_first_defined_variable, varpool_next_defined_variable
7661         cgraph_first_defined_function, cgraph_next_defined_function): Update.
7662         (cgraph_function_with_gimple_body_p): Update.
7663         (varpool_all_refs_explicit_p): Update.
7664         (symtab_alias_target): New function.
7665         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
7666         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
7667         (cgraph_function_or_thunk_node): Simplify using
7668         symtab_alias_ultimate_target.
7669         (varpool_variable_node): Likewise.
7670         * cgraph.c (cgraph_create_function_alias): Update.
7671         (cgraph_add_thunk): Update.
7672         (cgraph_remove_node): Update.
7673         (dump_cgraph_node): Do not dump removed flags.
7674         (cgraph_function_body_availability): Update.
7675         (cgraph_propagate_frequency): Update.
7676         (verify_cgraph_node): Check sanity of local flag.
7677         (cgraph_function_node): Move here from cgraph.h; revamp for
7678         cgraph_function_or_thunk_node.
7679         * lto-symtab.c (lto_varpool_replace_node): Update.
7680         (lto_symtab_resolve_can_prevail_p): Update.
7681         (lto_symtab_merge_cgraph_nodes): Update.
7682         * ipa-cp.c (determine_versionability, initialize_node_lattices,
7683         propagate_constants_accross_call, devirtualization_time_bonus,
7684         ipcp_propagate_stage): Update.
7685         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
7686         * ipa-inline-transform.c (clone_inlined_nodes,
7687         preserve_function_body_p): Update.
7688         * ipa-reference.c (propagate): Update.
7689         (write_node_summary_p): Update.
7690         * toplev.c (wrapup_global_declaration_2): Update.
7691         * cgraphunit.c (cgraph_analyze_function): Rename to ...
7692         (analyze_function) ... this one.
7693         (cgraph_process_new_functions): Update.
7694         (cgraph_reset_node): Export.
7695         (cgraph_finalize_function): Update.
7696         (cgraph_add_new_function): Update.
7697         (process_function_and_variable_attributes): Update.
7698         (varpool_finalize_decl): Update.
7699         (symbol_finalized): Remove.
7700         (symbol_finalized_and_needed): Rename to ...
7701         (symbol_defined_and_needed): ... update.
7702         (cgraph_analyze_functions): Update.
7703         (handle_alias_pairs): Update.
7704         (mark_functions_to_output): Update.
7705         (assemble_thunk): Update.
7706         (output_in_order): Update.
7707         (output_weakrefs): Update.
7708         (finalize_compilation_unit): Update.
7709         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
7710         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
7711         input_node, input_varpool_node): Update.
7712         * dbxout.c (dbxout_expand_expr): Update.
7713         * cgraphclones.c (cgraph_clone_node): Update.
7714         (cgraph_copy_node_for_versioning): Update.
7715         (cgraph_materialize_clone): Update.
7716         (cgraph_materialize_all_clones): Update.
7717         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
7718         propagate_pure_const, propagate_nothrow): Update.
7719         * lto-streamer-out.c (lto_output, write_symbol): Update.
7720         * ipa-utils.c (ipa_reverse_postorder): Update.
7721         * ipa-inline.c (can_inline_edge_p): Update.
7722         (update_caller_keys, ipa_inline): Update.
7723         * dwarf2out.c (reference_to_unused,
7724         premark_types_used_by_global_vars_helper): Update.
7725         * tree-eh.c (tree_could_trap_p): Update.
7726         * ipa-split.c (consider_split, execute_split_functions): Update.
7727         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
7728         has_addr_references_p): Update; move ahead in file for better
7729         readability.
7730         (process_references): Simplify.
7731         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
7732         bodies are removed.
7733         (cgraph_comdat_can_be_unshared_p): Make static.
7734         (cgraph_externally_visible_p): Update.
7735         (varpool_externally_visible_p): Update.
7736         (function_and_variable_visibility): Update.
7737         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
7738         ipa_tm_mark_force_output_node): Update.
7739         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
7740         estimate_edge_devirt_benefit, inline_generate_summary,
7741         inline_write_summary): Update.
7742         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
7743         * ipa-prop.c (ipa_compute_jump_functions): Update.
7744         (ipa_print_node_params, ipa_prop_read_section,
7745         ipa_update_after_lto_read, read_replacements_section): Update.
7746         * varasm.c (mark_decl_referenced): Update.
7747         (assemble_alias, dump_tm_clone_pairs): Update.
7748         * tree-inline.c (copy_bb): Update.
7749         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
7750         Update.
7751         * symtab.c (dump_symtab_base): Print new flags.
7752         (verify_symtab_base): Verify new flags.
7753         (symtab_alias_ultimate_target): New function.
7754         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
7755         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
7756         Update.
7757         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
7758         Update.
7759         * i386.c (ix86_get_function_versions_dispatcher,
7760         ix86_generate_version_dispatcher_body): Update.
7761         (fold_builtin_cpu): Use varpool_add_new_variable.
7762         * varpool.c (varpool_remove_initializer): Break out from ...
7763         (varpool_remove_node): ... this one.
7764         (dump_varpool_node, varpool_node_for_asm,
7765         cgraph_variable_initializer_availability, varpool_analyze_node,
7766         varpool_assemble_decl, varpool_remove_unreferenced_decls,
7767         varpool_finalize_named_section_flags, varpool_create_variable_alias):
7768         Update.
7770 2013-05-29  Jan Hubicka  <jh@suse.cz>
7772         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
7774 2013-05-29  Easwaran Raman  <eraman@google.com>
7776         PR tree-optimization/57442
7777         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
7778         when control exits the main loop.
7780 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
7782         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
7783         and RX600.
7784         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
7785         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
7786         * rx/t-rx: Add rx100 under multi library matches option for nofpu
7787         option.
7789 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7791         PR tree-optimization/57441
7792         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
7793         Don't limit size of incr_vec to number of candidates.
7795 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
7797         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
7798         and mips16 directories.
7799         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
7800         (MULTILIB_DIRNAMES): Ditto.
7801         (MULTILIB_EXCEPTIONS): Add new exceptions.
7802         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
7803         (MULTILIB_DIRNAMES): Ditto.
7804         (MULTILIB_EXCEPTIONS): Add new exceptions.
7806 2012-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7807             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7809         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
7810         SYMBOL_TINY_ABSOLUTE.
7811         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
7812         SYMBOL_TINY_ABSOLUTE.
7813         (aarch64_expand_mov_immediate): Likewise.
7814         (aarch64_classify_symbol): Likewise.
7815         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
7816         Permit SYMBOL_TINY_ABSOLUTE.
7817         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
7819 2013-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7820             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7822         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
7823         Refactor if/switch.  Replace gcc_assert with if.
7825 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
7827         * config/i386/i386.c (initial_ix86_tune_features): Enable
7828         FP Reassociation for AMD bdver1 and bdver2.
7830 2013-05-29  Martin Jambor  <mjambor@suse.cz>
7832         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
7833         and IMAGPART_EXPR do not occur within other handled_components.
7835 2013-05-29  Richard Biener  <rguenther@suse.de>
7837         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
7838         access on whether the use is in the BB we currently try to
7839         vectorize.
7840         (vect_bb_vectorization_profitable_p): Pass the BB we currently
7841         vectorize to vect_bb_slp_scalar_cost.
7843 2013-05-29  Richard Biener  <rguenther@suse.de>
7845         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
7846         computing scalar cost offsetted by stmts that are kept live
7847         by scalar uses.
7848         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
7849         for computation of scalar cost.
7851 2013-05-28  Steve Ellcey  <sellcey@mips.com>
7853         * config/mips/mips-cpus.def (mips32r2): Change processor type.
7855 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7857         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
7858         array notation built-in reduction functions.
7859         * doc/passes.texi (Passes): Added documentation about changes done
7860         for Cilk Plus.
7861         * doc/invoke.texi (C Dialect Options): Added documentation about
7862         the -fcilkplus flag.
7863         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
7864         (BUILTINS_DEF): Depend on cilkplus.def.
7865         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
7866         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
7867         * cilkplus.def: New file.
7869 2013-05-28  Joern Rennecke <joern.rennecke@embecosm.com>
7871         PR rtl-optimization/57439
7872         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
7874 2013-05-28  Easwaran Raman  <eraman@google.com>
7876         PR tree-optimization/57337
7877         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
7878         (find_insert_point): Correctly identify the insertion point
7879         when two statements with the same UID is compared.
7881 2013-05-28  Richard Biener  <rguenther@suse.de>
7883         PR tree-optimization/56787
7884         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
7885         from the list of data references.
7886         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
7887         clobbers.
7888         (vect_analyze_loop_operations): Likewise.
7889         (vect_transform_loop): Remove clobbers.
7891 2013-05-28  Martin Jambor  <mjambor@suse.cz>
7893         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
7894         and REALPART_EXPRs have scalar type.
7896 2013-05-28  Richard Biener  <rguenther@suse.de>
7898         PR tree-optimization/57411
7899         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
7900         virtual operands.
7901         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
7902         virtual operand propagation.
7904 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7906         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
7907         destination register for bmasksi_vis.
7908         (vector_init_bshuffle): Likewise.
7909         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
7911 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7913         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
7914         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
7915         mode if the instruction isn't available in the original mode.
7916         * config/sparc/sparc.opt (mfix-ut699): New option.
7917         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
7918         (divdf3): Turn into expander.
7919         (divdf3_nofix): New insn.
7920         (divdf3_fix): Likewise.
7921         (divsf3): Disable if -mfix-ut699.
7922         (sqrtdf2): Turn into expander.
7923         (sqrtdf2_nofix): New insn.
7924         (sqrtdf2_fix): Likewise.
7925         (sqrtsf2): Disable if -mfix-ut699.
7927 2013-05-27  Richard Biener  <rguenther@suse.de>
7929         PR middle-end/57412
7930         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7931         block for the new loop.
7933 2013-05-27  Richard Biener  <rguenther@suse.de>
7935         PR tree-optimization/57343
7936         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7937         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7938         (number_of_iterations_cond): Do not build the folded tree.
7940 2013-05-27  Richard Biener  <rguenther@suse.de>
7942         Revert
7943         PR middle-end/57381
7944         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7945         OEP_CONSTANT_ADDRESS_OF retained.
7947         PR tree-optimization/57417
7948         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7949         for unchanged base.
7950         (set_ssa_val_to): Compare addresses using
7951         get_addr_base_and_unit_offset.
7953 2013-05-27  Joern Rennecke <joern.rennecke@embecosm.com>
7955         PR rtl-optimization/56833
7956         * postreload.c (move2add_record_mode): New function.
7957         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7958         (move2add_use_add2_insn): Use move2add_record_sym_value.
7959         (move2add_use_add3_insn): Likewise.
7960         (reload_cse_move2add): Use move2add_valid_value_p and
7961         move2add_record_mode.  Invalidate call-clobbered and REG_INC
7962         affected regs by setting reg_mode to VOIDmode.
7963         (move2add_note_store): Don't pretend the inside of a SUBREG is
7964         the actual destination.  Invalidate single/leading registers by
7965         setting reg_mode to VOIDmode.
7966         Use move2add_record_sym_value, move2add_valid_value_p and
7967         move2add_record_mode.
7969 2013-05-27  Richard Biener  <rguenther@suse.de>
7971         PR tree-optimization/57396
7972         * tree-affine.c (double_int_constant_multiple_p): Properly
7973         return false for val == 0 and div != 0.
7975 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7977         * config/mips/mips.h: Use #elif in preprocessor conditions.
7979 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7981         PR target/53916
7982         * config/mips/constraints.md (kl): New constraint.
7983         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7984         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
7985         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
7986         from LO for MIPS16.
7987         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7989 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
7991         PR target/55777
7992         * config/mips/mips.c (mips_can_inline_p): New function.
7993         (TARGET_CAN_INLINE_P): Define.
7995 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
7997         * sched-int.h (ds_t, dw_t): Make unsigned int.
7998         Fix documentation that describes how all the ds_t bits are used.
7999         Reserve the last bit for delayed-branch scheduling.
8000         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8001         (BITS_PER_DEP_WEAK): Fix definition and documentation.
8002         (gen_dep_weak_1): Remove prototype.
8003         * sched-deps.c (get_dep_weak_1): Make static.
8004         * target.def (speculate_insn, needs_block_p, gen_spec_check,
8005         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8006         * doc/tm.texi: Regenerate.
8007         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8009 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
8011         PR debug/56950
8012         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8014 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
8015             Sandra Loosemore  <sandra@codesourcery.com>
8017         * config.gcc (powerpc-*): Allow native for with-cpu.
8019 2013-05-24  Jeff Law  <law@redhat.com>
8021         PR tree-optimization/57124
8022         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8023         conversion feeding a condition if the range has an overflow
8024         if -fstrict-overflow.  Add warnings for when we do make the
8025         transformation.
8027 2013-05-24  Dehao Chen  <dehao@google.com>
8029         * tree-cfg.c (locus_discrim_map): Fix the typo.
8030         (locus_discrim_hasher): Likewise.
8031         (locus_discrim_hasher::hash): Likewise.
8032         (locus_discrim_hasher::equal): Likewise.
8034 2013-05-24  Martin Jambor  <mjambor@suse.cz>
8036         PR tree-optimization/57294
8037         * cgraph.h (ipa_record_stmt_references): Declare.
8038         * cgraphbuild.c (ipa_record_stmt_references): New function.
8039         (build_cgraph_edges): Use ipa_record_stmt_references.
8040         (rebuild_cgraph_edges): Likewise.
8041         (cgraph_rebuild_references): Likewise.
8042         * ipa-prop.c (ipa_modify_call_arguments): Discard references
8043         associated with the old statement and build references from the
8044         newly built statements.
8045         * ipa-ref.c (ipa_remove_stmt_references): New function.
8046         * ipa-ref.h (ipa_remove_stmt_references): Declare.
8048 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
8050         * lra-constraints.c (emit_spill_move): Use smaller mode for
8051         mem-mem moves.
8052         (check_and_process_move): Consider mem-reg moves for secondary
8053         too.
8054         (curr_insn_transform): Don't lose insns emitted before for
8055         secondary memory moves.
8056         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
8057         reg set up in the current insn.
8059 2013-05-24  Dehao Chen  <dehao@google.com>
8061         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8062         hash function.
8063         (locus_descrim_hasher::equal): Likewise.
8064         (build_gimple_cfg): New discriminator assignment algorithm.
8065         (make_edges): Likewise.
8066         (next_discriminator_for_locus): Likewise.
8067         (same_line_p): Likewise.
8068         (assign_discriminators): Likewise.
8069         (make_cond_expr_edges): Likewise.
8070         (make_gimple_switch_edges): Likewise.
8071         (make_goto_expr_edges): Likewise.
8072         (make_gimple_asm_edges): Likewise.
8074 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
8076         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8077         X format specifier to only display bottom 16 bits.
8078         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8079         immediate to match for operand 2, since it will be masked.
8081 2013-05-24  Richard Biener  <rguenther@suse.de>
8083         PR tree-optimization/57287
8084         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8085         all SSA names that occur in abnormal PHIs.
8087 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8089         PR tree-ssa/57385
8090         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8091         that index is not negative.
8093 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
8095         PR rtl-optimization/55177
8096         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8097         (simplify_byte_swapping_operation): New.
8098         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8099         (simplify_relational_operation_1): Deal with BSWAP.
8101 2013-05-23  Richard Henderson  <rth@redhat.com>
8103         PR target/56742
8104         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8105         (ix86_reorg): Call it.
8107 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
8109         PR target/57379
8110         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8111         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8112         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8114 2013-05-23  Christian Bruel  <christian.bruel@st.com>
8116         PR debug/57351
8117         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8119 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8120             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8122         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8123         * config/aarch64/constraints.md (Usa): Remove.
8124         * doc/md.texi (AArch64 Usa): Remove.
8126 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8127             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8129         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8130         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8131         * config/aarch64/predicates.md (aarch64_const_address): Remove.
8132         (aarch64_mov_operand): Use aarch64_mov_operand_p.
8134 2013-05-23  Vidya Praveen <vidyapraveen@arm.com>
8136         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8137         instruction (AdvSIMD).
8138         * config/aarch64/aarch64-builtins.c
8139         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8140         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8142 2013-05-23  Martin Jambor  <mjambor@suse.cz>
8144         PR middle-end/57347
8145         * tree.h (contains_bitfld_component_ref_p): Declare.
8146         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8147         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
8148         caller.
8149         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8150         not access a bit-field.  Assert all final offsets are byte-aligned.
8152 2013-05-23  Richard Biener  <rguenther@suse.de>
8154         PR tree-optimization/57380
8155         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8156         least one invariant or re-used load.
8157         * passes.c (init_optimization_passes): Move pass_phiprop before
8158         pass_forwprop.
8160 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
8162         * config/aarch64/aarch64-simd.md
8163         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8165 2013-05-23  Richard Biener  <rguenther@suse.de>
8167         PR middle-end/57381
8168         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8169         OEP_CONSTANT_ADDRESS_OF retained.
8171 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
8173         PR middle-end/57344
8174         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8175         don't lower unit.  Handle unit not being always BITS_PER_WORD.
8177 2013-05-23  Richard Biener  <rguenther@suse.de>
8179         PR rtl-optimization/57341
8180         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8181         instead of true_dependence.
8183 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
8185         * bb-reorder.c (branch_threshold): Make const.
8186         (exec_threshold): Ditto.
8188 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8189             Pat Haugen <pthaugen@us.ibm.com>
8190             Peter Bergner <bergner@vnet.ibm.com>
8192         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8193         documentation for the power8 crypto builtins.
8195         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
8197         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
8198         macros for defining power8 builtin functions.
8199         (BU_P8V_AV_2): Likewise.
8200         (BU_P8V_AV_P): Likewise.
8201         (BU_P8V_VSX_1): Likewise.
8202         (BU_P8V_OVERLOAD_1): Likewise.
8203         (BU_P8V_OVERLOAD_2): Likewise.
8204         (BU_CRYPTO_1): Likewise.
8205         (BU_CRYPTO_2): Likewise.
8206         (BU_CRYPTO_3): Likewise.
8207         (BU_CRYPTO_OVERLOAD_1): Likewise.
8208         (BU_CRYPTO_OVERLOAD_2): Likewise.
8209         (XSCVSPDP): Fix typo, point to the correct instruction.
8210         (VCIPHER): Add power8 crypto builtins.
8211         (VCIPHERLAST): Likewise.
8212         (VNCIPHER): Likewise.
8213         (VNCIPHERLAST): Likewise.
8214         (VPMSUMB): Likewise.
8215         (VPMSUMH): Likewise.
8216         (VPMSUMW): Likewise.
8217         (VPERMXOR_V2DI): Likewise.
8218         (VPERMXOR_V4SI: Likewise.
8219         (VPERMXOR_V8HI: Likewise.
8220         (VPERMXOR_V16QI: Likewise.
8221         (VSHASIGMAW): Likewise.
8222         (VSHASIGMAD): Likewise.
8223         (VPMSUM): Likewise.
8224         (VPERMXOR): Likewise.
8225         (VSHASIGMA): Likewise.
8227         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8228         __CRYPTO__ if the crypto instructions are available.
8229         (altivec_overloaded_builtins): Add support for overloaded power8
8230         builtins.
8232         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
8233         support for power8 crypto builtins.
8234         (builtin_function_type): Likewise.
8235         (altivec_init_builtins): Add support for builtins that take vector
8236         long long (V2DI) arguments.
8238         * config/rs6000/crypto.md: New file, define power8 crypto
8239         instructions.
8241 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8242             Pat Haugen <pthaugen@us.ibm.com>
8243             Peter Bergner <bergner@vnet.ibm.com>
8245         * doc/invoke.texi (Option Summary): Add power8 options.
8246         (RS/6000 and PowerPC Options): Likewise.
8248         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
8249         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
8250         wm, wn, wr documentation.
8252         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
8253         registers if direct move instructions are enabled.
8254         (wn): New constraint for no registers.
8255         (wq): New constraint for quad word even GPR registers.
8256         (wr): New constraint if 64-bit instructions are enabled.
8257         (wv): New constraint if power8 vector instructions are enabled.
8258         (wQ): New constraint for quad word memory locations.
8260         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
8261         constraint for 0..15 for crypto instructions.
8262         (gpc_reg_operand): If VSX allow registers in VSX registers as well
8263         as GPR and floating point registers.
8264         (int_reg_operand): New predicate to match only GPR registers.
8265         (base_reg_operand): New predicate to match base registers.
8266         (quad_int_reg_operand): New predicate to match even GPR registers
8267         for quad memory operations.
8268         (vsx_reg_or_cint_operand): New predicate to allow vector logical
8269         operations in both GPR and VSX registers.
8270         (quad_memory_operand): New predicate for quad memory operations.
8271         (reg_or_indexed_operand): New predicate for direct move support.
8273         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
8274         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
8275         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
8276         (POWERPC_MASKS): Add power8 options.
8277         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
8278         various options.
8280         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8281         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
8283         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
8284         (-mpower8-fusion): New power8 options.
8285         (-mpower8-fusion-sign): Likewise.
8286         (-mpower8-vector): Likewise.
8287         (-mcrypto): Likewise.
8288         (-mdirect-move): Likewise.
8289         (-mquad-memory): Likewise.
8291         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
8292         power8.
8293         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
8294         registers.
8295         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
8296         (rs6000_debug_vector_unit): Add p8_vector.
8297         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
8298         definitions.  Also print fusion state.
8299         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
8300         (rs6000_builtin_mask_calculate): Add power8 builtin support.
8301         (rs6000_option_override_internal): Add support for power8.
8302         (rs6000_common_init_builtins): Add debugging for skipped builtins
8303         if -mdebug=builtin.
8304         (rs6000_adjust_cost): Add power8 support.
8305         (rs6000_issue_rate): Likewise.
8306         (insn_must_be_first_in_group): Likewise.
8307         (insn_must_be_last_in_group): Likewise.
8308         (force_new_group): Likewise.
8309         (rs6000_register_move_cost): Likewise.
8310         (rs6000_opt_masks): Likewise.
8312         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
8313         power8 capable assembler, default to power7 options.
8314         (TARGET_DIRECT_MOVE): Likewise.
8315         (TARGET_CRYPTO): Likewise.
8316         (TARGET_P8_VECTOR): Likewise.
8317         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
8318         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
8319         (VECTOR_MEM_P8_VECTOR_P): Likewise.
8320         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
8321         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
8322         (TARGET_XSCVDPSPN): Likewise.
8323         (TARGET_XSCVSPDPN): Likewsie.
8324         (TARGET_SYNC_HI_QI): Likewise.
8325         (TARGET_SYNC_TI): Likewise.
8326         (MASK_CRYPTO): Likewise.
8327         (MASK_DIRECT_MOVE): Likewise.
8328         (MASK_P8_FUSION): Likewise.
8329         (MASK_P8_VECTOR): Likewise.
8330         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
8331         temporary used by some of the direct move instructions to get two FP
8332         temporary registers does not force creation of a stack frame.
8333         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
8334         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
8335         that any VSX registers are tieable, even if they are also an
8336         Altivec vector mode.
8337         (r6000_reg_class_enum): Add wm, wr, wv constraints.
8338         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
8339         (RS6000_BTM_CRYPTO): Likewise.
8340         (RS6000_BTM_COMMON): Likewise.
8342         * config/rs6000/rs6000.md (cpu attribute): Add power8.
8343         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
8344         (enum rs6000_vector): Add power8 vector support.
8346 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8348         PR target/19599
8349         PR target/57340
8350         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
8351         (any_sibcall_could_use_r3): this and handle indirect calls.
8352         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
8354 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8356         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
8358 2013-05-22  Richard Biener  <rguenther@suse.de>
8360         PR middle-end/57349
8361         * profile.c (branch_prob): Do not split blocks that are
8362         abnormally receiving from ECF_RETURNS_TWICE functions.
8364 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
8366         * recog.c (offsettable_address_addr_space_p): Fix calculation of
8367         address mode.  Move pointer mode initialization to the same place.
8369 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8371         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
8372         while it has any effect.
8374 2013-05-21  Easwaran Raman  <eraman@google.com>
8376         PR tree-optimization/57322
8377         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
8378         UID of the statement added to the BB to be 1.
8380 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
8382         PR tree-optimization/57331
8383         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
8384         of conversion from pointer type to integral type with integer.
8386 2013-05-21  Martin Jambor  <mjambor@suse.cz>
8388         PR lto/57289
8389         * ipa-prop.c (ipa_read_node_info): Process param_used and
8390         controlled_uses in the same order as when writing.
8392 2013-05-21  Magnus Granberg  <baldrick@free.fr>
8394         PR plugins/56754
8395         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
8397 2013-05-21  Richard Biener  <rguenther@suse.de>
8399         PR tree-optimization/57318
8400         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
8401         estimate stmts with side-effects as likely eliminated.
8403 2013-05-21  Richard Biener  <rguenther@suse.de>
8405         PR tree-optimization/57330
8406         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
8407         preserve the call stmts fntype.
8409 2013-05-21  Richard Biener  <rguenther@suse.de>
8411         PR tree-optimization/57303
8412         * tree-ssa-sink.c (statement_sink_location): Improve killing
8413         stmt detection and properly handle self-assignments.
8415 2013-05-21  Christian Bruel  <christian.bruel@st.com>
8417         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
8418         spanning registers. LEAF_REG_REMAP is supported only for contiguous
8419         registers. Set register size out of the PARALLEL loop.
8421 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
8423         PR target/56547
8424         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
8425         (*fmasf4, *fmasf4_media): New insns.
8427 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8429         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
8430         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
8431         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
8432         (mips_idiv_insns): Update the comments to say that the returned
8433         instruction counts are in units of BASE_INSN_LENGTH.
8434         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
8435         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
8436         using 2 rather than 4 as the length of indirect MIPS16 and
8437         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
8438         length of a NOP.  Don't divide MIPS16 lengths by 2.
8439         (mips16_split_long_branches): Assume a branch is long if the
8440         length is greater than 4 rather than 8.
8441         * config/mips/mips.md (length): Give MIPS16 lengths directly,
8442         rather than multiplying them by 2.  Multiply instruction counts
8443         by BASE_INSN_LENGTH rather than 4.
8444         (*jump_mips16, tls_get_tp_mips16_<mode>)
8445         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
8447 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8449         * config/mips/mips.md (extended_mips16): Remove branch case.
8450         (length): Remove duplicated extended_mips16 test.
8452 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8454         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
8456 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
8458         * recog.h (Recog_data): Rename to...
8459         (recog_data_d): ...this.
8460         (recog_data): Update accordingly.
8461         * recog.c (recog_data): Likewise.
8462         * reload.c (save_recog_data): Likewise.
8463         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
8464         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
8466 2013-05-17  Julian Brown  <julian@codesourcery.com>
8468         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
8469         found in a REG_EQUAL note, invalidate it.
8471 2013-05-17  Easwaran Raman  <eraman@google.com>
8473         * tree-ssa-reassoc.c (find_insert_point): New function.
8474         (insert_stmt_after): Likewise.
8475         (get_def_stmt): Likewise.
8476         (ensure_ops_are_available): Likewise.
8477         (not_dominated_by): Likewise.
8478         (rewrite_expr_tree): Do not move statements beyond what is
8479         necessary. Remove call to swap_ops_for_binary_stmt...
8480         (reassociate_bb): ... and move it here.
8481         (build_and_add_sum): Assign UIDs for new statements.
8482         (linearize_expr): Likewise.
8483         (do_reassoc): Renumber gimple statement UIDs.
8485 2013-05-17  Jan Hubicka  <jh@suse.cz>
8487         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
8488         weakrefs.
8489         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
8490         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
8491         weakrefs.
8492         (output_weakrefs): Update.
8494 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
8495             Martin Jambor  <mjambor@suse.cz>
8497         PR middle-end/57276
8498         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
8499         value that corresponds to the given aggval is found in values vector.
8501 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
8503         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
8504         sse, sse2, sse3, ssse3 and sse4a flags to options.
8506 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
8508         * gengtype-state.c: (s_expr_writer): New class, to handle
8509         prettifying of output layout of s-expressions.
8510         (state_writer): New class, to write out gtype.state.
8511         (state_written_type_count): Move this variable into member data of
8512         state_writer.
8513         (s_expr_writer::s_expr_writer): New code: constructor for new class
8514         (state_writer::state_writer(): ditto
8515         (s_expr_writer::write_new_line): New function
8516         (s_expr_writer::write_any_indent): ditto
8517         (s_expr_writer::begin_s_expr): ditto
8518         (s_expr_writer::end_s_expr): ditto
8519         (write_state_fileloc): convert to method of state_writer...
8520         (state_writer:: write_state_fileloc): ...and use methods of
8521         s_expr_writer to write indentation into the gtype.state output file
8522         to visually represent the hierarchical structure of the list
8523         structures
8524         (write_state_fields): ditto, renaming to...
8525         (state_writer::write_state_fields)
8526         (write_state_a_string): ditto, renaming to...
8527         (state_writer::write_state_a_string)
8528         (write_state_string_option): ditto, renaming to...
8529         (state_writer::write_state_string_option)
8530         (write_state_type_option): ditto, renaming to...
8531         (state_writer::write_state_type_option)
8532         (write_state_nested_option): ditto, renaming to...
8533         (state_writer::write_state_nested_option)
8534         (write_state_option): ditto, renaming to...
8535         (state_writer::write_state_option)
8536         (write_state_options): ditto, renaming to...
8537         (state_writer::write_state_options)
8538         (write_state_lang_bitmap): ditto, renaming to...
8539         (state_writer::write_state_lang_bitmap)
8540         (write_state_version): ditto, renaming to...
8541         (state_writer::write_state_version)
8542         (write_state_scalar_type): ditto, renaming to...
8543         (state_writer::write_state_scalar_type)
8544         (write_state_string_type): ditto, renaming to...
8545         (state_writer::write_state_string_type)
8546         (write_state_undefined_type): ditto, renaming to...
8547         (state_writer::write_state_undefined_type)
8548         (write_state_struct_union_type): ditto, renaming to...
8549         (state_writer::write_state_struct_union_type)
8550         (write_state_struct_type): ditto, renaming to...
8551         (state_writer::write_state_struct_type)
8552         (write_state_user_struct_type): ditto, renaming to...
8553         (state_writer::write_state_user_struct_type)
8554         (write_state_lang_struct_type): ditto, renaming to...
8555         (state_writer::write_state_lang_struct_type)
8556         (write_state_param_struct_type): ditto, renaming to...
8557         (state_writer::write_state_param_struct_type)
8558         (write_state_pointer_type): ditto, renaming to...
8559         (state_writer::write_state_pointer_type)
8560         (write_state_array_type): ditto, renaming to...
8561         (state_writer::write_state_array_type)
8562         (write_state_gc_used): ditto, renaming to...
8563         (state_writer::write_state_gc_used)
8564         (write_state_common_type_content): ditto, renaming to...
8565         (state_writer::write_state_common_type_content)
8566         (write_state_type): ditto, renaming to...
8567         (state_writer::write_state_type)
8568         (write_state_pair_list): ditto, renaming to...
8569         (state_writer::write_state_pair_list)
8570         (write_state_pair): ditto, renaming to...
8571         (state_writer::write_state_pair)
8572         (write_state_typedefs): ditto, renaming to...
8573         (state_writer::write_state_typedefs)
8574         (write_state_structures): ditto, renaming to...
8575         (state_writer::write_state_structures)
8576         (write_state_param_structs): ditto, renaming to...
8577         (state_writer::write_state_param_structs)
8578         (write_state_variables): ditto, renaming to...
8579         (state_writer::write_state_variables)
8580         (write_state_srcdir): ditto, renaming to...
8581         (state_writer::write_state_srcdir)
8582         (write_state_files_list): ditto, renaming to...
8583         (state_writer::write_state_files_list)
8584         (write_state_languages): ditto, renaming to...
8585         (state_writer::write_state_languages)
8586         (write_state): create a state_writer instance and use it when
8587         writing out the state file
8589 2013-05-17  Mike Stump  <mikestump@comcast.net>
8591         PR rtl-optimization/57304
8592         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
8593         accessing DF_REF_REAL_LOC.
8595 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8597         PR rtl-optimization/57281
8598         PR rtl-optimization/57300
8599         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
8600         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
8601         what the other splitter did if the registers are dead.
8603 2013-05-17  Richard Biener  <rguenther@suse.de>
8605         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
8606         MEM_REF offsets.
8608 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8610         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
8611         linking.
8613 2013-05-17  Marek Polacek  <polacek@redhat.com>
8615         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
8616         length when doing non-zero store of storing '\0' to '\0'.
8618 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8620         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
8621         vect_external_def oprnd1 with loop_vinfo, try to emit
8622         optional cast, negation and and stmts on the loop preheader
8623         edge instead of into the pattern def seq.
8625         PR tree-optimization/57051
8626         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
8627         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
8629 2013-05-16  Nick Clifton  <nickc@redhat.com>
8631         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
8632         (rl78_is_naked_func): New function.
8633         (rl78_expand_prologue): Skip prologue generation for naked functions.
8634         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
8635         * doc/extend.texi (naked): Add RL78 to the list of processors
8636         that supports this attribute.
8638 2013-05-16  Jeff Law  <law@redhat.com>
8640         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
8642 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8644         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
8645         cache parameters using detect_caches_amd also for CYRIX,
8646         NSC and TM2 signatures.
8648 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8649             Dzianis Kahanovich  <mahatma@eu.by>
8651         PR target/45359
8652         PR target/46396
8653         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8654         VIA/Centaur processors and determine their cache parameters
8655         using detect_caches_amd.
8657 2013-05-16  Teresa Johnson  <tejohnson@google.com>
8659         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
8660         (rtl_verify_edges): New function.
8661         (rtl_verify_bb_insns): Ditto.
8662         (rtl_verify_bb_pointers): Ditto.
8663         (rtl_verify_bb_insn_chain): Ditto.
8664         (rtl_verify_fallthru): Ditto.
8665         (rtl_verify_bb_layout): Ditto.
8666         (rtl_verify_flow_info_1): Outline checks into new functions.
8667         (rtl_verify_flow_info): Ditto.
8669 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
8671         * cfghooks.c (copy_bbs): Add update_dominance argument.
8672         * cfghooks.h (copy_bbs): Update prototype.
8673         * tree-cfg.c (gimple_duplicate_sese_region):
8674         Add update_dominance argument.
8675         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
8676         * tree-ssa-loop-ch.c (copy_loop_headers): Update
8677         gimple_duplicate_sese_region call.
8678         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
8679         Update copy_bbs call.
8680         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
8681         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
8683 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8685         * tree-vectorizer.h (NUM_PATTERNS): Increment.
8686         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
8687         vect_recog_rotate_pattern.
8688         (vect_recog_rotate_pattern): New function.
8690 2013-05-16  Jason Merrill  <jason@redhat.com>
8692         * Makefile.in (LLINKER): New variable.
8693         (mostlyclean): Remove link mutex.
8694         * configure.ac: Handle --enable-link-mutex.
8695         * lock-and-run.sh: New script.
8697 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8699         PR target/19599
8700         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
8701         for NULL decl.
8703 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8705         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
8707 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
8709         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
8710         * config/arm/arm.c (next_consecutive_mem): New function.
8711         (gen_movmem_ldrd_strd): Likewise.
8712         * config/arm/arm.md (movmemqi): Update condition and code.
8713         (unaligned_loaddi, unaligned_storedi): New patterns.
8715 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8717         * config.gcc: Obsolete *-*-solaris2.9*.
8718         * doc/install.texi (Specific, *-*-solaris2*): Document it.
8720 2013-05-16  Richard Biener  <rguenther@suse.de>
8722         * passes.c (init_optimization_passes): Move pass_parallelize_loops
8723         earlier, after GRAPHITE transforms and IV canonicalization.
8725 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8727         * omp-low.c (extract_omp_for_data): For collapsed loops,
8728         if at least one of the loops is known at compile time to
8729         iterate zero times, set count to 0.
8730         (expand_omp_regimplify_p): New function.
8731         (expand_omp_for_generic): For collapsed loops, if at least
8732         one of the loops isn't known to iterate at least once,
8733         add runtime check with setting count to 0.
8734         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
8735         For unsigned types if it isn't known at compile time that
8736         the loop will iterate at least once, add runtime check to bypass
8737         the whole loop if initial condition isn't true.
8739 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
8741         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
8743 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
8745         PR middle-end/57286
8746         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
8747         transformations to avoid an infinite loop.
8749 2013-05-16  Marek Polacek  <polacek@redhat.com>
8751         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
8753 2013-05-15  Leif Ekblad  <leif@rdos.net>
8755         * config/i386/i386.c (ix86_decompose_address): Use
8756         DEFAULT_TLS_SEG_REG to access TLS segment register.
8757         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
8758         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
8759         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
8761 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
8763         PR target/57260
8764         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
8765         sibling calls to functions that would normally be lazily bound,
8766         unless $gp is call-clobbered.
8768 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
8770         * config/i386/i386.c (ix86_option_override_internal): Update
8771         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
8772         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
8773         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
8774         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
8775         of TARGET_3DNOW.
8776         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
8778 2013-05-15  Andreas Schwab  <schwab@suse.de>
8780         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
8781         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
8782         third operand.
8784 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8786         * loop-unroll.c (report_unroll_peel): Check decision before
8787         emitting unroll/peel message.
8789 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8791         * function.h (has_bb_partition): New rtl_data flag.
8792         (bb_reorder_complete): Ditto.
8793         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
8794         instead of flag_reorder_blocks_and_partition.
8795         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
8796         with some enhancements.
8797         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
8798         * bb-reorder.c (connect_traces): Check for has_bb_partition
8799         instead of flag_reorder_blocks_and_partition.
8800         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
8801         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
8802         verify_hot_cold_block_grouping.
8803         (partition_hot_cold_basic_blocks): Set has_bb_partition.
8805 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8807         PR target/19599
8808         * config/arm/predicates.md (call_insn_operand): New predicate.
8809         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
8810         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
8811         if insn is not a tail call.
8812         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
8813         registers.
8814         * config/arm/arm.h (enum reg_class): New caller save register class.
8815         (REG_CLASS_NAMES): Likewise.
8816         (REG_CLASS_CONTENTS): Likewise.
8817         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
8818         without decls.
8820 2013-05-15  Richard Biener  <rguenther@suse.de>
8822         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
8823         of MSG_OPTIMIZED_LOCATIONS.
8824         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
8825         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
8826         message.
8827         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
8828         of MSG_OPTIMIZED_LOCATIONS.
8829         (execute_vect_slp): Likewise.
8830         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
8831         (vect_create_cond_for_alias_checks): Likewise.
8832         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8833         (vect_recog_widen_mult_pattern): Likewise.
8834         (vect_recog_widen_sum_pattern): Likewise.
8835         (vect_recog_over_widening_pattern): Likewise.
8836         (vect_recog_widen_shift_pattern): Likewise.
8837         (vect_recog_vector_vector_shift_pattern): Likewise.
8838         (vect_recog_divmod_pattern): Likewise.
8839         (vect_recog_mixed_size_cond_pattern): Likewise.
8840         (vect_recog_bool_pattern): Likewise.
8841         (vect_pattern_recog_1): Likewise.
8843 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8845         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
8846         non-functions to builtin_unreachable.
8847         * ipa-inline-transform.c (inline_call): Do not assert estimates were
8848         correct when new direct edges were discovered.
8850 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8852         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
8853         header, print symbol order instead of node uid, print more information
8854         about indirect edge targets.
8855         (ipa_make_edge_direct_to_target): Print symbol order instead of node
8856         uids.
8857         (ipa_make_edge_direct_to_target): Likewise.
8858         (remove_described_reference): Likewise.
8859         (propagate_controlled_uses): Likewise.
8860         (ipa_print_node_params): Also print symbol order.
8861         (ipcp_transform_function): Print symbol order instead of node uids.
8862         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
8863         (cgraph_get_create_real_symbol_node): Likewise.
8864         * ipa-cp.c (print_lattice): Likewise.
8865         (print_all_lattices): Likewise.
8866         (determine_versionability): Likewise.
8867         (initialize_node_lattices): Likewise.
8868         (estimate_local_effects): Likewise.
8869         (update_profiling_info): Likewise.
8870         (create_specialized_node): Likewise.
8871         (perhaps_add_new_callers): Likewise.
8872         (decide_about_value): Likewise.
8873         (decide_whether_version_node): Likewise.
8874         (identify_dead_nodes): Likewise.
8875         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
8876         (dump_inline_summary): Likewise.
8877         (estimate_node_size_and_time): Likewise.
8878         (inline_analyze_function): Likewise.
8879         * ipa-inline.c (report_inline_failed_reason): Likewise.
8880         (want_early_inline_function_p): Likewise.
8881         (edge_badness): Likewise.
8882         (update_edge_key): Likewise.
8883         (inline_small_functions): Likewise.  Add dumping of order to two other
8884         dumps.
8885         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
8886         instead of node uids.
8887         (propagate_pure_const): Likewise.
8888         (propagate_pure_const): Likewise.
8889         * ipa-utils.c (dump_cgraph_node_set): Likewise.
8890         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
8891         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
8892         of node uids.
8893         * tree-pretty-print.c (dump_function_header): Likewise.
8894         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
8895         Print symbol order instead of node uids.
8897 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8899         * config/s390/s390.c (s390_register_move_cost): Don't impose the
8900         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
8902 2013-05-15  Richard Biener  <rguenther@suse.de>
8904         PR tree-optimization/57275
8905         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
8906         return value for fail to do runtime alias checks for gather loads.
8908 2013-05-15  Jan Hubicka  <jh@suse.cz>
8910         PR lto/57038
8911         PR lto/47375
8912         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
8913         weakrefs are not external.
8914         (lto_symtab_merge_decls): Fix thinko when dealing with
8915         non-lto_symtab decls.
8916         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
8917         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
8918         * varpool.c (dump_varpool_node): Dump more flags.
8920 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8922         * config/i386/i386.c (processor_alias_table): Add instruction
8923         FSGSBASE for AMD bdver3 architecture.
8925 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8927         * tree.c (warn_deprecated_use): Print file:line using locus color.
8928         * diagnostic.c (diagnostic_report_current_module): Print file:line
8929         and file:line:column using locus color.
8931 2013-05-14  Mike Stump  <mikestump@comcast.net>
8933         * gdbinit.in: Add __null.
8935 2013-05-14  Mike Stump  <mikestump@comcast.net>
8937         * recog.h: Rename struct recog_data to Recog_data.
8938         * recog.c: Likewise.
8939         * reload.c (can_reload_into): Likewise.
8940         * config/picochip/picochip.c: Likewise.
8942 2013-05-14  Mike Stump  <mikestump@comcast.net>
8944         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8946 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
8948         * resource.h (struct resources): Remove unch_memory member.
8949         (CLEAR_RESOURCE): Don't clear unch_memory.
8950         * resource.c (mark_referenced_resources): Don't set it.
8951         (mark_set_resources): Likewise.
8952         (mark_target_live_regs): Don't clear it.
8953         (init_resource_info): Likewise.
8954         * reorg.c (resource_conflicts_p): Don't compare it.
8955         (redundant_insn): Don't set it.
8957         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8958         Remove prototypes.
8959         * emit-rtl.c (next_label): Remove unused function.
8960         (skip_consecutive_labels, link_cc0_insns): Move to ...
8961         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8962         only place where these functions are used, and make them static.
8964 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8966         * fold-const.c (fold_negate_expr): Handle vectors.
8967         (fold_truth_not_expr): Make it static.
8968         (fold_invert_truthvalue): New static function.
8969         (invert_truthvalue_loc): Handle vectors. Do not call
8970         fold_truth_not_expr directly.
8971         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8972         <TRUTH_NOT_EXPR>: Do not cast to boolean.
8973         (fold_comparison): Handle vector constants.
8974         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8975         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8976         * tree.h (fold_truth_not_expr): Remove declaration.
8978 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
8980         * config/aarch64/aarch64-simd.md
8981         (aarch64_vcond_internal<mode>): Rename to...
8982         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8983         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8984         float modes. Clarify all iterator modes.
8985         (vcond<mode><mode>): Use new name for vcond expanders.
8986         (vcond<v_cmp_result><mode>): Likewise.
8987         (vcondu<mode><mode>: Likewise.
8988         * config/aarch64/iterators.md (VDQF_COND): New.
8990 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
8992         PR bootstrap/57266
8993         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8994         variable for the shift amount. Check that we shift by non-negative
8995         amounts.
8997 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
8999         PR target/42017
9000         * config/arm/arm.h (EPILOGUE_USES): Only return true
9001         for LR_REGNUM after epilogue_completed.
9003 2013-05-14  Joern Rennecke <joern.rennecke@embecosm.com>
9005         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9006         is error_mark_node.
9008 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9010         PR target/57261
9011         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9012         and Solaris 11+/x86 with gld.
9013         * configure: Regenerate.
9015 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9017         * expmed.c (expand_shift_1): Canonicalize rotates by
9018         constant bitsize / 2 to bitsize - 1.
9019         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9020         case ROTATERT>: Likewise.
9022         Revert:
9023         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9025         * config/i386/i386.md (rotateinv): New code attr.
9026         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9027         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9028         roll $31, %eax, etc.
9030 2013-05-14  Richard Biener  <rguenther@suse.de>
9032         PR middle-end/57235
9033         * tree-eh.c (sink_clobbers): Give up for successors with
9034         multiple predecessors and no virtual uses.
9036 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
9038         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9039         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9041 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9043         PR middle-end/57251
9044         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9045         the case when both op0 and op1 have VOIDmode.
9047 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9049         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9050         in multiply-accumulate mode.
9052 2013-05-13  Guozhi Wei  <carrot@google.com>
9054         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9056 2013-05-13  Kai Tietz  <ktietz@redhat.com>
9058         PR target/56975
9059         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9060         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9061         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9062         * config/i386/i386.c (ix86_option_override_internal): Likewise.
9063         (ix86_expand_prologue): Likewise.
9064         (ix86_expand_split_stack_prologue): Likewise.
9065         (legitimate_pic_address_disp_p): Likewise.
9066         (legitimize_pic_address): Likewise.
9067         (legitimize_tls_address): Likewise.
9068         (legitimize_pe_coff_symbol): Likewise.
9069         (output_pic_addr_const): Likewise.
9070         (construct_plt_address): Likewise.
9071         (ix86_expand_call): Likewise.
9072         (x86_output_mi_thunk): Likewise.
9073         (x86_function_profiler): Likewise.
9075 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
9077         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9078         similar switch cases.
9079         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9080         (aarch64_simd_mov_to_<mode>low): Delete.
9081         (aarch64_simd_mov_to_<mode>high): Delete.
9082         (move_lo_quad_<mode>): Add w<-r alternative.
9083         (aarch64_simd_move_hi_quad_<mode>): Likewise.
9084         (aarch64_simd_mov_from_*): Update type attribute.
9085         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9086         statement.
9088 2013-05-13  Jan Hubicka  <jh@suse.cz>
9090         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9091         * config/i386/i386.c (ix86_compute_frame_layout,
9092         ix86_expand_epilogue, emit_i387_cw_initialization,
9093         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9094         ix86_local_alignment): Fix use of size/speed predicates.
9096 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9098         PR tree-optimization/45216
9099         PR tree-optimization/57157
9100         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9101         the (-Y) & (B - 1) variant if OP is |.
9102         * expmed.c (expand_shift_1): For rotations by const0_rtx just
9103         return shifted.  Use (-op1) & (prec - 1) as other_amount
9104         instead of prec - op1.
9106 2013-05-13  Martin Jambor  <mjambor@suse.cz>
9108         PR middle-end/42371
9109         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9110         (ipa_constant_data): New type.
9111         (ipa_jump_func): Use ipa_constant_data to hold information about
9112         constant jump functions.
9113         (ipa_get_jf_constant): Adjust to jump function type changes.
9114         (ipa_get_jf_constant_rdesc): New function.
9115         (ipa_param_descriptor): New field controlled_uses.
9116         (ipa_get_controlled_uses): New function.
9117         (ipa_set_controlled_uses): Likewise.
9118         * ipa-ref.h (ipa_find_reference): Declare.
9119         * ipa-prop.c (ipa_cst_ref_desc): New type.
9120         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9121         changes.
9122         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
9123         New parameter cs.  Adjust all callers.
9124         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9125         (remove_described_reference): New function.
9126         (jfunc_rdesc_usable): Likewise.
9127         (try_make_edge_direct_simple_call): Decrement controlled use count,
9128         attempt to remove reference if it hits zero.
9129         (combine_controlled_uses_counters): New function.
9130         (propagate_controlled_uses): Likewise.
9131         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9132         (ipa_edge_duplication_hook): Duplicate reference descriptions.
9133         (ipa_print_node_params): Print described use counter.
9134         (ipa_write_jump_function): Adjust to jump function type changes.
9135         (ipa_read_jump_function): New parameter CS, pass it to
9136         ipa_set_jf_constant.  Adjust caller.
9137         (ipa_write_node_info): Stream controlled use count
9138         (ipa_read_node_info): Likewise.
9139         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9140         asserting.
9141         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9142         count.  Remove cloning-added reference if it reaches zero.
9143         * ipa-ref.c (ipa_find_reference): New function.
9145 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9147         * config/i386/i386.c (processor_target_table): Modified default
9148         alignment values for AMD BD and BT architectures.
9150 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
9152         * tree-vect-generic.c (uniform_vector_p): Move ...
9153         * tree.c (uniform_vector_p): ... here.
9154         * tree.h (uniform_vector_p): Declare it.
9155         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9156         into a scalar.
9158 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9160         PR tree-optimization/57230
9161         * tree-ssa-strlen.c (handle_char_store): Record length for
9162         array store from STRING_CST.
9164         PR tree-optimization/57230
9165         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9166         check.
9168 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
9170         * config/epiphany/epiphany.c (epiphany_init): Check size of
9171         NUM_MODES_FOR_MODE_SWITCHING.
9172         (epiphany_expand_prologue):
9173         Remove CONFIG_REGNUM initial value handling code.
9174         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9175         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9176         (emit_set_fp_mode, epiphany_mode_after): Likewise.
9177         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9178         Don't return 1 for FP_MODE_NONE.
9179         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9180         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9181         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9182         * config/epiphany/epiphany.md (save_config): New pattern.
9184 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
9186         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9188 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
9190         * config/i386/i386.md (memory): Handle sseishft1.
9191         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9192         (*vec_extractv2di_1): Ditto.
9194 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
9196         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
9197         saved registers.
9199 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
9201         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
9202         Add mthumb/march=armv7-a multilib.
9203         Add mthumb/march=armv7-r multilib.
9204         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
9206 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
9208         * config/v850/t-rtems: Add more multilibs.
9210 2013-05-10  Richard Biener  <rguenther@suse.de>
9212         PR tree-optimization/57214
9213         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
9214         not propagate from SSA names that occur in abnormal PHI nodes.
9216 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
9218         * stor-layout.c (element_precision): New function.
9219         * machmode.h (element_precision): Declare it.
9220         * tree.c (build_minus_one_cst): New function.
9221         (element_precision): Likewise.
9222         * tree.h (build_minus_one_cst): Declare new function.
9223         (element_precision): Likewise.
9224         * fold-const.c (operand_equal_p): Use element_precision.
9225         (fold_binary_loc): Handle vector types.
9226         * convert.c (convert_to_integer): Use element_precision.
9227         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
9228         separately.
9230 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
9232         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
9233         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
9234         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
9235         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9236         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
9237         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
9238         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
9239         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9240         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
9241         (Uuw8): New constraints.
9242         (Usb4): Move into alphabetical order.
9243         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
9244         (sd8_operand, ub8_operand, uw8_operand): New predicates.
9245         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
9246         previously unnamed patterns.
9247         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
9248         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
9249         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
9250         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
9251         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
9252         of specific lengths.
9254 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9256         * config/i386/i386.md (rotateinv): New code attr.
9257         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9258         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9259         roll $31, %eax, etc.
9261         PR tree-optimization/45216
9262         PR tree-optimization/57157
9263         * tree-ssa-forwprop.c (simplify_rotate): New function.
9264         (ssa_forward_propagate_and_combine): Call it.
9266 2013-05-10  Richard Biener  <rguenther@suse.de>
9268         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
9269         disable peeling when we version for aliasing.
9270         (vector_alignment_reachable_p): Honor explicit user alignment.
9271         (vect_supportable_dr_alignment): Likewise.
9272         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
9273         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
9274         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
9275         then peeling to arrange for the cost-model check to come first.
9277 2013-05-10  Alan Modra  <amodra@gmail.com>
9279         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
9280         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
9281         * configure: Regenerate.
9283 2013-05-10  Alan Modra  <amodra@gmail.com>
9285         PR target/55033
9286         * varasm.c (default_elf_select_section): Move !DECL_P check..
9287         (get_named_section): ..to here before calling get_section_name.
9288         Adjust assertion.
9289         (default_section_type_flags): Add DECL_P check.
9290         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
9291         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
9293 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
9295         * config/epiphany/epiphany.c (epiphany_expand_prologue):
9296         When using gen_stack_adjust_str with a register offset, add a
9297         REG_FRAME_RELATED_EXPR note.
9299 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
9301         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
9302         (*vec_extractv4si_zext_mem): Ditto.
9303         (*vec_extractv2di): Add 0->x and x->x alternatives.
9304         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
9305         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
9307 2013-05-09  Jason Merrill  <jason@redhat.com>
9309         N3639 C++1y VLA support
9310         * gimplify.c (gimplify_vla_decl): Don't touch an existing
9311         DECL_VALUE_EXPR.
9313         * tree.c (build_constructor_va): New.
9314         * tree.h: Declare it.
9316 2013-05-09  Martin Jambor  <mjambor@suse.cz>
9318         PR lto/57084
9319         * gimple-fold.c (canonicalize_constructor_val): Call
9320         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
9322 2013-05-09  Jan Hubicka  <jh@suse.cz>
9323             Richard Biener  <rguenther@suse.de>
9325         PR lto/54095
9326         * symtab.c (symtab_make_decl_local): Do not add private names.
9328 2013-05-09  Jan Hubicka  <jh@suse.cz>
9330         PR lto/54095
9331         * symtab.c (insert_to_assembler_name_hash): Handle clones.
9332         (unlink_from_assembler_name_hash): Likewise.
9333         (symtab_prevail_in_asm_name_hash, symtab_register_node,
9334         symtab_unregister_node, symtab_initialize_asm_name_hash,
9335         change_decl_assembler_name): Update.
9337 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
9339         * config/aarch64/aarch64.md: New movtf split.
9340         (*movtf_aarch64): Update.
9341         (aarch64_movdi_tilow): Handle TF modes and rename to
9342         aarch64_movdi_<mode>low.
9343         (aarch64_movdi_tihigh): Handle TF modes and rename to
9344         aarch64_movdi_<mode>high
9345         (aarch64_movtihigh_di): Handle TF modes and rename to
9346         aarch64_mov<mode>high_di
9347         (aarch64_movtilow_di): Handle TF modes and rename to
9348         aarch64_mov<mode>low_di
9349         (aarch64_movtilow_tilow): Remove spurious whitespace.
9350         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
9351         splits.
9352         (aarch64_print_operand): Update.
9354 2013-05-09  Alan Modra  <amodra@gmail.com>
9356         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
9357         powerpc64le.
9358         * configure: Regenerate.
9360 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
9362         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
9363         splitter preparation statements.
9364         * config/i386/sse.md (*vec_extract* splitters): Ditto.
9365         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
9366         adjust_address_nv.
9368 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9370         * gimple-ssa-strength-reduction.c (count_candidates): Change
9371         return value to int.
9372         (analyze_candidates_and_replace): Change type of length to int.
9374 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
9376         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
9377         (*vec_extract<mode>): Use VI12_128 mode iterator.
9378         (*vec_extract<mode>_mem): Ditto.
9379         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
9380         attribute.
9382 2013-05-08  Diego Novillo  <dnovillo@google.com>
9384         PR bootstrap/54659
9386         Revert:
9387         2012-08-17  Diego Novillo  <dnovillo@google.com>
9389         PR bootstrap/54281
9390         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
9391         * config.in: Regenerate.
9392         * configure: Regenerate.
9393         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
9395 2013-05-08  Jan Hubicka  <jh@suse.cz>
9397         PR lto/54095
9398         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
9399         * cgraph.h (symtab_node_base): Add unique_name.
9400         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9401         input_overwrite_node, input_varpool_node): Stream unique_name.
9402         * cgraphclones.c (cgraph_create_virtual_clone,
9403         cgraph_function_versioning): Set unique_name.
9404         * ipa.c (function_and_variable_visibility): Set unique_name.
9406 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9408         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
9409         (alloc_cand_and_find_basis): Restrict conditional candidate
9410         processing to CAND_MULTs.
9412 2013-05-08  Jan Hubicka  <jh@suse.cz>
9414         PR lto/54095
9415         lto-symtab.c (lto_symtab_symbol_p): New function.
9416         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
9417         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
9418         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
9419         Skip static symbols.
9421 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
9423         PR tree-optimization/57200
9424         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
9425         Only call inform if the preceding warning_at returns true.
9427 2013-05-07  Han Shen  <shenhan@google.com>
9429         * cfgexpand.c (record_or_union_type_has_array_p): New function.
9430         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
9431         * common.opt (fstack-protector-strong): New option.
9432         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
9433         * doc/invoke.texi (Optimization Options): Document
9434         "-fstack-protector-strong".
9435         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
9437 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9439         * config/mips/mips.c (mips_machine_reorg2): Return 0.
9441 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
9443         * ira.c (update_equiv_regs): Add insn having equiv memory even if
9444         it is not lhs of the insn.
9445         (setup_reg_equiv): Remove insn having equiv memory which it is not
9446         lhs of the insn.
9447         * lra-constraints.c (process_address): Try to improve generation
9448         code for address base + disp.
9449         (lra_constraints): Make correct the code for checking insn setting
9450         up backward equivalence.  Remove insn only if it is in the init
9451         insn list.
9452         * lra-eliminations.c (update_reg_eliminate): Change return value.
9453         (lra_eliminate): Use the result.
9455 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
9457         * config/i386/sse.md (ssescalarnummask): New mode attribute.
9458         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
9459         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
9460         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
9461         register target operands.
9462         (*vec_extractv8hi_sse2): New pattern.
9463         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
9464         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
9465         (*vec_extract<mode>_mem): New insn and split pattern.
9467 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
9469         * config/arm/arm.c (arm_asan_shadow_offset): New function.
9470         (TARGET_ASAN_SHADOW_OFFSET): Define.
9471         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
9472         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
9474 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9476         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
9477         (incr_vec_index): Return -1 if increment not found.
9478         (create_add_on_incoming_edge): Assert if increment not found.
9479         (record_increment): Limit number of increments recorded.
9480         (all_phi_incrs_profitable): Return false if an increment not found.
9481         (replace_profitable_candidates): Don't process increments that were
9482         not recorded.
9483         (analyze_candidates_and_replace): Limit size of incr_vec.
9485 2013-05-07  Richard Biener  <rguenther@suse.de>
9487         * calls.c (special_function_p): setjmp-like functions are leaf.
9488         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
9489         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
9491 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
9493         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
9494         (aarch64_simd_mov<mode>): New expander.
9495         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
9496         (aarch64_simd_mov_to_<mode>high): Likewise.
9497         (aarch64_simd_mov_from_<mode>low): Likewise.
9498         (aarch64_simd_mov_from_<mode>high): Likewise.
9499         (aarch64_dup_lane<mode>): Update.
9500         (aarch64_dup_lanedi): New instruction pattern.
9501         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
9502         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
9504 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9506         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
9507         (replace_mult_candidate): Remove unnecessary argument; remove
9508         unnecessary parameter from call to introduce_cast_before_cand.
9509         (replace_unconditional_candidate): Remove unnecessary parameter
9510         from call to replace_mult_candidate.
9511         (replace_conditional_candidate): Likewise.
9512         (insert_initializers): Use make_temp_ssa_name.
9513         (introduce_cast_before_cand): Remove unnecessary argument; use
9514         make_temp_ssa_name.
9515         (replace_one_candidate): Remove unnecessary argument; remove
9516         unnecessary parameter from calls to introduce_cast_before_cand.
9517         (replace_profitable_candidates): Remove unnecessary parameters
9518         from calls to replace_one_candidate.
9520 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9522         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9523         phi def as possibly hiding a basis for a CAND_ADD whose operands
9524         have been commuted in the analysis.
9525         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
9527 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
9529         * config/aarch64/aarch64.md
9530         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
9531         shift value between 0-4.
9533 2013-05-07  Richard Biener  <rguenther@suse.de>
9535         * double-int.h (rshift): New overload.
9536         * double-int.c (rshift): New function.
9537         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
9538         (create_reference_ops_from_ref): Remove.
9539         (vn_reference_insert): Use shared ops for constructing the
9540         reference and copy it.
9542 2013-05-07  Richard Biener  <rguenther@suse.de>
9544         PR middle-end/57190
9545         * tree-eh.c (sink_clobbers): Properly propagate
9546         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
9548 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
9550         PR tree-optimization/57149
9551         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
9552         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
9553         collect_phi_def_edges, execute_late_warn_uninitialized): Use
9554         uninit_undefined_value_p instead of ssa_undefined_value_p.
9556         PR debug/57184
9557         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
9558         for modifier == EXPAND_INITIALIZER.
9560 2013-05-07  Anton Blanchard  <anton@samba.org>
9562         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
9563         for powerpc64 little endian.
9564         * configure: Regenerate.
9566 2013-05-06  Graham Stott  <grahams@btinternet.com>
9568         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
9569         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
9570         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
9571         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
9573 2013-05-06  Graham Stott  <grahams@btinternet.com>
9575         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
9576         codes which allow non-lvalues.
9578 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
9580         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
9581         components are all 1s.
9582         (integer_minus_onep): New function.
9583         * tree.h (integer_minus_onep): Declare it.
9584         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
9585         integer_minus_onep instead of integer_all_onesp.
9587 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9589         PR target/52933
9590         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
9591         variations of these patterns.
9593 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9595         * config/i386/i386.md (isa): Add x64_sse4 member.
9596         (enabled): Handle x64_sse4.
9597         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
9598         instruction for 64bit SSE4_1 targets.  Update insn attributes.
9599         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
9600         instruction for SSE4_1 targets.  Update insn attributes.
9601         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9602         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
9603         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
9604         const_1 selector.
9605         (*vec_extractv4si): Rename from *sse4_1_pextrd.
9606         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
9607         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
9609 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9611         PR target/57108
9612         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
9614 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
9616         * final.c (do_assembler_dialects): Don't handle curly braces and
9617         vertical bar escaped by % as dialect delimiters.
9618         (output_asm_insn): Print curly braces and vertical bar if escaped
9619         by % and ASSEMBLER_DIALECT defined.
9620         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
9621         * doc/tm.texi: Regenerated.
9623 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9625         * config/mips/mips.c: Include tree-pass.h.
9626         (mips_reorg): Split in pre- and post-dbr_schedule parts.
9627         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
9628         (pass_mips_machine_reorg2): New machine specific pass.
9629         (insert_pass_mips_machine_reorg2): New pass plugin definition.
9630         (mips_option_override): Register the new pass.
9631         * rtl.h (cleanup_barriers): Remove prototype.
9632         (dbr_schedule): Likewise.
9633         * jump.c (cleanup_barriers): Make static.
9634         * reorg.c (dbr_schedule): Likewise.
9636 2013-05-06  Richard Biener  <rguenther@suse.de>
9638         PR tree-optimization/57185
9639         * tree-parloops.c (add_field_for_reduction): Handle anonymous
9640         SSA names properly.
9642 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9644         PR target/57106
9645         * config/i386/i386.c (add_parameter_dependencies): Add dependence
9646         between "first_arg" and "insn", not "last" and "insn".
9648 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9650         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
9651         (find_candidates_in_block): Re-enable slsr_process_phi.
9652         (create_phi_basis): Fix double counting of candidate adjustment.
9654 2013-05-06  Richard Biener  <rguenther@suse.de>
9656         PR middle-end/57147
9657         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
9658         the edge is also fallthru, preserve it and just clear the
9659         abnormal flag.
9660         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
9661         also complex, preserve that and just clear the fallthru flag.
9662         * tree-inline.c (update_ssa_across_abnormal_edges): Also
9663         update virtual operands.
9665 2013-05-06  Alan Modra  <amodra@gmail.com>
9667         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
9668         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
9669         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
9670         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
9671         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9672         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
9674 2013-05-06  Alan Modra  <amodra@gmail.com>
9676         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
9677         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
9678         (DEFAULT_ASM_ENDIAN): Define.
9679         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9680         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
9681         Update -K PIC clause from sysv4.h.
9682         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
9683         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
9685 2013-05-06  Alan Modra  <amodra@gmail.com>
9687         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
9688         twice for little-endian.
9689         (ashrdi3_no_power, ashrdi3): Support little-endian.
9691 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9693         PR target/55303
9694         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
9695         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
9696         related expanders.
9697         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
9698         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
9699         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
9700         New predicates.
9702 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
9703             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
9705         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
9706         * config/pa/pa.opt: Make mbig-switch a no-op.
9707         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
9708         (CASE_VECTOR_MODE): Always return SImode.
9709         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
9710         for the !TARGET_BIG_SWITCH case.
9711         * config/pa/pa-linux.h: Likewise.
9712         * config/pa/pa-openbsd.h: Likewise.
9713         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
9714         * config/pa/pa.md (short_jump): Remove define_insn.
9715         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
9716         (casesi0): Remove define_insn.
9717         (type): Remove btable_branch.
9718         (pa_combine_type): Likewise.
9719         (in_nullified_branch_delay): Likewise.
9720         (in_call_delay): Likewise.
9721         (define_delay): Likewise.
9722         (define_insn_reservation "Z3"): Likewise.
9723         (define_insn_reservation "Z4"): Likewise.
9724         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
9725         (pa_adjust_insn_length): Remove adjustment for btable branches.
9726         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
9727         and mno-big-switch
9729 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
9731         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9732         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
9733         Add m->r,x alternatives.
9734         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
9735         splitters using SWI48x mode iterator.
9736         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
9737         TARGET_64BIT.  Add m->x alternative.
9738         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
9739         Add o->x alternative.  Enable for TARGET_SSE.
9740         (sse_storeq): Remove expander.
9741         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
9742         with memory input operand.
9743         (*vec_extractv2di_1 splitter): New.
9744         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
9745         * config/i386/i386.md (ssevecmodelower): New mode attribute.
9747 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9749         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
9750         (INT_LOWPART): Delete.
9751         (extract_MB): Adjust.
9752         (extract_ME): Adjust.
9753         (print_operand): Adjust.
9755 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9757         * config/rs6000/predicates.md (reg_or_add_cint_operand,
9758         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9759         (reg_or_logical_cint_operand, easy_fp_constant,
9760         logical_const_operand): Delete "CONST_DOUBLE" case.
9761         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
9762         "HOST_BITS_PER_WIDE_INT == 64" test.
9763         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
9764         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
9765         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
9766         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
9767         test.
9768         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
9769         CONST_DOUBLE DImode/VOIDmode case.
9770         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
9771         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9772         CONST_DOUBLE VOIDmode case.
9773         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9774         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
9775         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
9776         Delete CONST_DOUBLE case.
9777         (splitters for mov FMOVE64 const_double): Delete
9778         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9779         "HOST_BITS_PER_WIDE_INT >= 64" test.
9780         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
9781         case.
9782         (mov DI const_double): Delete.
9784 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
9786         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
9787         on op shows all bits zero in mode of a lowpart subreg, return zero.
9789 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
9791         PR target/57150
9792         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
9793         to save TFmode registers and DImode to save TImode registers for
9794         caller save operations.
9795         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
9796         mark being partially clobbered since they only use the first
9797         double word.
9799         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
9800         and TDmode only use the upper 64-bits of each VSX register.
9802 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9804         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
9805         (find_candidates_in_block): Disable slsr_process_phi.
9807 2013-05-03  Guozhi Wei  <carrot@google.com>
9809         * coverage.c (coverage_obj_init): Move the construction of gcov
9810         constructor to ...
9811         (build_init_ctor): ... here.
9813 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9815         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
9816         (slsr_cand_d): Redefine def_phi.
9817         (stride_status, phi_adjust_status, count_phis_status): New enums.
9818         (find_phi_def): New.
9819         (find_basis_for_base_expr): New.
9820         (find_basis_for_candidate): Handle hidden bases.
9821         (alloc_cand_and_find_basis): Handle phi candidates.
9822         (slsr_process_phi): New.
9823         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
9824         (create_mul_imm_cand): Likewise.
9825         (create_add_ssa_cand): Exclude phi base candidates.
9826         (create_add_imm_cand): Likewise.
9827         (slsr_process_cast): Likewise.
9828         (slsr_process_copy): Likewise.
9829         (find_candidates_in_block): Handle phi candidates.
9830         (dump_candidate): Likewise.
9831         (unconditional_cands): Delete.
9832         (unconditional_cands_with_known_stride_p): Delete.
9833         (phi_dependent_cand_p): New.
9834         (cand_increment): Handle phi-dependent candidates.
9835         (replace_dependent): Delete.
9836         (replace_mult_candidate): New.
9837         (replace_unconditional_candidate): New.
9838         (incr_vec_index): Move to avoid forward reference.
9839         (create_add_on_incoming_edge): New.
9840         (create_phi_basis): New.
9841         (replace_dependents): Delete.
9842         (replace_conditional_candidate): New.
9843         (phi_add_costs): New.
9844         (replace_uncond_cands_and_profitable_phis): New.
9845         (record_increment): Handle phi adjustments.
9846         (record_phi_increments): New.
9847         (record_increments): Handle phi adjustments.
9848         (phi_incr_cost): New.
9849         (lowest_cost_path): Handle phis.
9850         (total_savings): Likewise.
9851         (analyze_increments): Likewise.
9852         (ncd_with_phi): New.
9853         (ncd_of_cand_and_phis): New.
9854         (nearest_common_dominator_for_cands): Handle phi increments.
9855         (all_phi_incrs_profitable): New.
9856         (replace_profitable_candidates): Handle phi-dependent candidates.
9857         (analyze_candidates_and_replace): Likewise.
9859 2013-05-03  Teresa Johnson  <tejohnson@google.com>
9861         PR bootstrap/57154
9862         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
9863         do not exceed REG_BR_PROB_BASE.
9865 2013-05-03  Jeff Law  <law@redhat.com>
9867         PR tree-optimization/57144
9868         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
9869         operand of the condition will bit into the new type when eliminating
9870         a cast feeding a condition.
9872 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
9874         PR rtl-optimization/57130
9875         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
9876         of COMPARE as in_code to the recursive call if needed.
9878 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
9880         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
9881         (enabled): Handle new members.
9882         * config/i386/sse.md (*vec_concatv2si): Merge from
9883         *vec_concatv2si_sse2 and vec_concatv2si_sse.
9884         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
9886 2013-05-03  Joern Rennecke <joern.rennecke@embecosm.com>
9888         PR tree-optimization/57027
9889         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
9890         for fnms opportunity, check we got the prerequisite kind
9891         of tree / gimple before using accessor functions.
9893 2013-05-03  Richard Biener  <rguenther@suse.de>
9895         * double-int.h (lshift): New overload without precision
9896         and arith argument.
9897         (operator *=, operator +=, operator -=): Move ...
9898         * double-int.c (operator *=, operator +=, operator -=): ... here
9899         and implement more efficiently.
9900         (mul_double_with_sign): Remove.
9901         (lshift_double): Adjust to take unsinged shift argument, push
9902         dispatching code to callers.
9903         (mul_double_wide_with_sign): Add early out for callers that
9904         are not interested in high parts or overflow.
9905         (lshift): New function.
9906         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
9907         dispatch code here.
9908         (lrotate, rrotate): Use logical shifts.
9909         * expr.c (get_inner_reference): Use lshift.
9910         * fixed-value.c (do_fixed_divide): Likewise.
9911         * tree-dfa.c (get_ref_base_and_extent): Likewise.
9912         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
9913         (indirect_refs_may_alias_p): Likewise.
9914         (stmt_kills_ref_p_1): Likewise.
9916 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9918         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
9920 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9922         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
9923         scalar form of FABD instruction.
9925 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
9927         * lra-constraints.c (process_alt_operands): Add checking alt
9928         number to choose the best alternative.
9930 2013-05-02  Richard Biener  <rguenther@suse.de>
9932         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9933         bitmap and its handling.
9934         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9936 2013-05-02  Richard Biener  <rguenther@suse.de>
9938         PR middle-end/57140
9939         * tree-inline.c (copy_loops): Properly handle removed loops.
9940         (copy_cfg_body): Mark destination loops for fixup if source
9941         loops needed fixup.
9943 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
9945         PR target/56732
9946         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9947         generating simple_return for naked functions.
9949 2013-05-02  Martin Jambor  <mjambor@suse.cz>
9951         PR middle-end/56988
9952         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9953         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9954         flags match.
9955         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9956         ipa_agg_replacement_value structures.
9957         (known_aggs_to_agg_replacement_list): Likewise.
9958         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9959         (read_agg_replacement_chain): Likewise.
9960         (ipcp_transform_function): Also check that by_ref flags match.
9962 2013-05-02  Richard Biener  <rguenther@suse.de>
9964         * graphds.h (struct graph): Add obstack member.
9965         * graphds.c (new_graph): Initialize obstack and allocate
9966         vertices from it.
9967         (add_edge): Allocate edge from the obstack.
9968         (free_graph): Free the obstack instead of all edges and vertices.
9970 2013-05-02  Teresa Johnson  <tejohnson@google.com>
9972         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9973         divides.
9974         * cfg.c (update_bb_profile_for_threading): Ditto.
9975         * tree-inline.c (copy_bb): Ditto.
9976         (copy_edges_for_bb): Ditto.
9977         (initialize_cfun): Ditto.
9978         (copy_cfg_body): Ditto.
9979         (expand_call_inline): Ditto.
9980         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9981         (estimate_node_size_and_time): Ditto.
9982         (inline_merge_summary): Ditto.
9983         * cgraphclones.c (cgraph_clone_edge): Ditto.
9984         (cgraph_clone_node): Ditto.
9985         * sched-rgn.c (compute_dom_prob_ps): Ditto.
9986         (compute_trg_info): Ditto.
9988 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9990         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9991         S reg when fp attribute set.
9992         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9994 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
9996         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9997         New pattern.
9998         (*and_one_cmplsi3_compare0_uxtw): Likewise.
9999         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10000         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10002 2013-05-02  Richard Biener  <rguenther@suse.de>
10004         * tree-scalar-evolution.c (scev_info_hasher): Remove.
10005         (struct instantiate_cache_entry): New type.
10006         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10007         (struct instantiate_cache_type): New type.
10008         (set_instantiated_value, get_instantiated_value): Remove.
10009         (get_instantiated_value_entry): New function.
10010         (instantiate_scev_name): Use the new cache and adjust.
10011         (instantiate_scev_poly): Adjust.
10012         (instantiate_scev_binary): Likewise.
10013         (instantiate_array_ref): Likewise.
10014         (instantiate_scev_convert): Likewise.
10015         (instantiate_scev_not): Likewise.
10016         (instantiate_scev_3): Likewise.
10017         (instantiate_scev_2): Likewise.
10018         (instantiate_scev_r): Likewise.
10019         (instantiate_scev): Likewise.
10020         (resolve_mixers): Likewise.
10022 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
10024         PR target/57091
10025         * lra-constraints.c (best_small_class_operands_num): Remove.
10026         (process_alt_operands): Remove small_class_operands_num.  Take
10027         small classes operands into losers and only if the operand is not
10028         matched.  Modify debugging output.
10029         (curr_insn_transform): Remove best_small_class_operands_num.
10030         Print insn name.
10032 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10034         * config/aarch64/aarch64-builtins.c
10035         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10036         * config/aarch64/aarch64-simd-builtins.def
10037         (reduc_splus_): Add new modes.
10038         (reduc_uplus_): New.
10039         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10040         (reduc_uplus_v4sf): Likewise.
10041         (reduc_splus_v4sf): Likewise.
10042         (aarch64_addv<mode>): Likewise.
10043         (reduc_uplus_<mode>): Likewise.
10044         (reduc_splus_<mode>): Likewise.
10045         (aarch64_addvv2di): Likewise.
10046         (reduc_uplus_v2di): Likewise.
10047         (reduc_splus_v2di): Likewise.
10048         (aarch64_addvv2si): Likewise.
10049         (reduc_uplus_v2si): Likewise.
10050         (reduc_splus_v2si): Likewise.
10051         (reduc_<sur>plus_<mode>): New.
10052         (reduc_<sur>plus_v2di): Likewise.
10053         (reduc_<sur>plus_v2si): Likewise.
10054         (reduc_<sur>plus_v4sf): Likewise.
10055         (aarch64_addpv4sf): Likewise.
10056         * config/aarch64/arm_neon.h
10057         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10058         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10059         add UNSPEC_SADDV, UNSPEC_UADDV.
10060         (SUADDV): New.
10061         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10063 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10065         * config/aarch64/arm_neon.h
10066         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10068 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10070         * config/aarch64/aarch64-builtins
10071         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10073 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10075         * config/aarch64/aarch64-simd-builtins.def
10076         (reduc_smax_): New.
10077         (reduc_smin_): Likewise.
10078         (reduc_umax_): Likewise.
10079         (reduc_umin_): Likewise.
10080         (reduc_smax_nan_): Likewise.
10081         (reduc_smin_nan_): Likewise.
10082         (fmax): Remove.
10083         (fmin): Likewise.
10084         (smax): Update for V2SF, V4SF and V2DF modes.
10085         (smin): Likewise.
10086         (smax_nan): New.
10087         (smin_nan): Likewise.
10088         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10089         (<su><maxmin><mode>3): ...This, refactor.
10090         (s<maxmin><mode>3): New.
10091         (<maxmin_uns><mode>3): Likewise.
10092         (reduc_<maxmin_uns>_<mode>): Refactor.
10093         (reduc_<maxmin_uns>_v4sf): Likewise.
10094         (reduc_<maxmin_uns>_v2si): Likewise.
10095         (aarch64_<fmaxmin><mode>: Remove.
10096         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10097         new builtin names.
10098         (vmin<q>_f<32,64>): Likewise.
10099         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10100         (FMAXMIN): New.
10101         (su): Add mappings for smax, smin, umax, umin.
10102         (maxmin): New.
10103         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10104         (FMAXMIN): Rename as...
10105         (FMAXMIN_UNS): ...This.
10106         (maxminv): Remove.
10107         (fmaxminv): Likewise.
10108         (fmaxmin): Likewise.
10109         (maxmin_uns): New.
10110         (maxmin_uns_op): Likewise.
10112 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10114         * config/aarch64/arm_neon.h
10115         (vac<ge, gt><sd>_f<32, 64>): Rename to...
10116         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10117         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10119 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10121         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10122         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10124 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10126         * config/aarch64/aarch64-simd.md
10127         (vcond<mode>_internal): Handle special cases for constant masks.
10128         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10129         (vcondu<mode><mode>): Likewise.
10130         (vcond<v_cmp_result><mode>): New.
10132 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10134         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10135         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10136         * config/aarch64/aarch64-simd-builtins.def
10137         (cmeq): Update to BUILTIN_VALLDI.
10138         (cmgt): Likewise.
10139         (cmge): Likewise.
10140         (cmle): Likewise.
10141         (cmlt): Likewise.
10142         * config/aarch64/arm_neon.h
10143         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10144         to builtins or C as appropriate.
10146 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10148         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10149         (cmgeu): ...This.
10150         (cmhi): Rename to...
10151         (cmgtu): ...This.
10152         * config/aarch64/aarch64-simd.md
10153         (simd_mode): Add SF.
10154         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10155         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10156         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10157         (cstore<mode>_neg): ...This.
10158         * config/aarch64/iterators.md
10159         (VALLF): new.
10160         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10161         (COMPARISONS): New.
10162         (UCOMPARISONS): Likewise.
10163         (optab): Add missing comparisons.
10164         (n_optab): New.
10165         (cmp_1): Likewise.
10166         (cmp_2): Likewise.
10167         (CMP): Likewise.
10168         (cmp): Remove.
10169         (VCMP_S): Likewise.
10170         (VCMP_U): Likewise.
10171         (V_cmp_result): Add DF, SF modes.
10172         (v_cmp_result): Likewise.
10173         (v): Likewise.
10174         (vmtype): Likewise.
10175         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10177 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
10179         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10180         define_insn to define_insn_and_split.
10181         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10182         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10183         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10184         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10185         (thumb2_negscc): Likewise.
10187 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10189         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10191 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10193         * config/arm/thumb2.md: Remove trailing whitespaces.
10195 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
10197         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
10198         Use gen_int_mode rather than GEN_INT.
10200 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
10202         * value-prof.c (stream_in_histogram_value): Remove the strayed
10203         debug_gimple_stmt.
10205 2013-04-30  Richard Biener  <rguenther@suse.de>
10207         PR middle-end/57122
10208         * cfghooks.c (split_edge): Properly check for the loop latch edge.
10210 2013-04-30  Richard Biener  <rguenther@suse.de>
10212         PR middle-end/57107
10213         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
10215 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
10217         PR rtl-optimization/56957
10218         PR rtl-optimization/57105
10219         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
10220         variable.  Use just INSN_UID for determining whether an insn
10221         should be only disconnected from the insn stream.
10222         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
10224 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
10226         PR tree-optimization/57104
10227         * tsan.c (instrument_expr): Don't instrument accesses to
10228         DECL_HARD_REGISTER VAR_DECLs.
10230 2013-04-30  Richard Biener  <rguenther@suse.de>
10232         * function.h (loops_for_fn): New inline function.
10233         (set_loops_for_fn): Likewise.
10234         * cfgloop.h (place_new_loop): Add struct function parameter.
10235         (get_loop): Likewise.
10236         (get_loops): Likewise.
10237         (number_of_loops): Likewise.
10238         (fel_next): Adjust.
10239         (fel_init): Likewise.
10240         * cfg.c (get_loop_copy): Adjust.
10241         * cfgloop.c (flow_loops_dump): Likewise.
10242         (record_loop_exits): Likewise.
10243         (verify_loop_structure): Likewise.
10244         * cfgloopanal.c (mark_irreducible_loops): Likewise.
10245         (estimate_reg_pressure_cost): Likewise.
10246         (mark_loop_exit_edges): Likewise.
10247         * cfgloopmanip.c (place_new_loop): Likewise.
10248         (add_loop): Likewise.
10249         (duplicate_loop): Likewise.
10250         * graph.c (draw_cfg_nodes): Likewise.
10251         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
10252         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
10253         (extract_affine_chrec): Likewise.
10254         (build_scop_iteration_domain): Likewise.
10255         * graphite.c (graphite_initialize): Likewise.
10256         * ira-build.c (create_loop_tree_nodes): Likewise.
10257         (more_one_region_p): Likewise.
10258         (rebuild_regno_allocno_maps): Likewise.
10259         (mark_loops_for_removal): Likewise.
10260         (mark_all_loops_for_removal): Likewise.
10261         (remove_unnecessary_regions): Likewise.
10262         (ira_build): Likewise.
10263         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
10264         * loop-init.c (fix_loop_structure): Likewise.
10265         (gate_rtl_move_loop_invariants): Likewise.
10266         (gate_rtl_unswitch): Likewise.
10267         (gate_rtl_unroll_and_peel_loops): Likewise.
10268         (rtl_doloop): Likewise.
10269         * lto-streamer-in.c (input_cfg): Likewise.
10270         * lto-streamer-out.c (output_cfg): Likewise.
10271         * modulo-sched.c (sms_schedule): Likewise.
10272         * predict.c (tree_estimate_probability): Likewise.
10273         (tree_estimate_probability_driver): Likewise.
10274         (estimate_loops): Likewise.
10275         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
10276         (move_sese_region_to_fn): Likewise.
10277         (debug_loop_num): Likewise.
10278         * tree-chrec.c (chrec_evaluate): Likewise.
10279         (hide_evolution_in_other_loops_than_loop): Likewise.
10280         (chrec_component_in_loop_num): Likewise.
10281         (reset_evolution_in_loop): Likewise.
10282         (evolution_function_is_invariant_rec_p): Likewise.
10283         * tree-if-conv.c (main_tree_if_conversion): Likewise.
10284         * tree-inline.c (copy_loops): Likewise.
10285         (copy_cfg_body): Likewise.
10286         (tree_function_versioning): Likewise.
10287         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
10288         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
10289         Likewise.
10290         (add_to_evolution_1): Likewise.
10291         (scev_const_prop): Likewise.
10292         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
10293         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
10294         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
10295         (tree_ssa_lim_initialize): Likewise.
10296         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
10297         (verify_loop_closed_ssa): Likewise.
10298         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
10299         (tree_ssa_loop_im): Likewise.
10300         (tree_ssa_loop_unswitch): Likewise.
10301         (tree_vectorize): Likewise.
10302         (check_data_deps): Likewise.
10303         (tree_ssa_loop_ivcanon): Likewise.
10304         (tree_ssa_loop_bounds): Likewise.
10305         (tree_complete_unroll): Likewise.
10306         (tree_complete_unroll_inner): Likewise.
10307         (tree_parallelize_loops): Likewise.
10308         (tree_ssa_loop_prefetch): Likewise.
10309         (tree_ssa_loop_ivopts): Likewise.
10310         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
10311         * tree-vectorizer.c (vectorize_loops): Likewise.
10313 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
10315         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
10316         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
10317         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
10318         with EABI_LINK_SPEC.
10320 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10322         PR target/44578
10323         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
10324         alternative.
10326 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
10328         PR target/57097
10329         * lra-constraints.c (process_alt_operands): Discourage a bit more
10330         using memory for pseudos.  Print cost dump for alternatives.
10331         Modify cost values for conflicts with early clobbers.
10332         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
10334 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10336         PR target/57098
10337         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
10339 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
10341         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
10342         from/to S register.
10343         (movdi_aarch64): Support LDR/STR from/to D register.
10345 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
10347         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
10348         or higher by default.
10350 2013-04-29  Richard Biener  <rguenther@suse.de>
10352         PR middle-end/57075
10353         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
10354         even if not adding abnormal edges for calls that can make
10355         abnormal gotos.
10357 2013-04-29  Richard Biener  <rguenther@suse.de>
10359         PR middle-end/57103
10360         * tree-cfg.c (move_stmt_op): Fix condition under which to update
10361         TREE_BLOCK.
10362         (move_stmt_r): Remove redundant checking.
10364 2013-04-29  Teresa Johnson  <tejohnson@google.com>
10366         PR bootstrap/57077
10367         * basic-block.h (apply_scale): New function.
10368         (apply_probability): Use apply_scale.
10369         * gimple-streamer-in.c (input_bb): Ditto.
10370         * lto-streamer-in.c (input_cfg): Ditto.
10371         * lto-cgraph.c (merge_profile_summaries): Ditto.
10372         * tree-optimize.c (execute_fixup_cfg): Ditto.
10373         * tree-inline.c (copy_bb): Update comment to use apply_scale.
10374         (copy_edges_for_bb): Ditto.
10375         (copy_cfg_body): Ditto.
10377 2013-04-29  Tom de Vries  <tom@codesourcery.com>
10379         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
10380         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
10381         (tail_merge_optimize): Handle current_loops == NULL.
10383 2013-04-26  Jeff Law  <law@redhat.com>
10385         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
10386         (simplify_cond_using_ranges): Generalize code to simplify
10387         COND_EXPRs where one argument is a constant and the other
10388         is an SSA_NAME created by an integral type conversion.
10390 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10392         * config/arm/arm.md (store_minmaxsi): Use only when
10393         optimize_insn_for_size_p.
10395 2013-04-29  Christian Bruel  <christian.bruel@st.com>
10397         PR target/57108
10398         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
10400 2013-04-29  Richard Biener  <rguenther@suse.de>
10402         PR middle-end/57089
10403         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
10404         loop tree make sure to schedule a fixup for the child as well.
10405         (expand_omp_for_generic): Properly add loops.
10406         (expand_omp_for_static_nochunk): Likewise.
10407         (expand_omp_for_static_chunk): Likewise.
10408         (expand_omp_for): For the degenerate case fixup loops.
10409         (expand_omp_sections): Fix default bb placement in loops.
10410         (expand_omp_atomic_pipeline): Properly add loops.
10412 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10414         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
10416 2013-04-29  Tom de Vries  <tom@codesourcery.com>
10418         * tree-ssa-tail-merge.c: Update header comment.
10420 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10422         * config/aarch64/arm_neon.h
10423         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
10424         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
10425         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
10426         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
10427         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
10428         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
10429         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
10430         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
10432 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10434         * config/aarch64/aarch64-simd.md
10435         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
10436         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
10437         fix_trunc, fixuns_trunc.
10438         (ftrunc<VDQF:mode>2): New.
10439         * config/aarch64/iterators.md (optab): Add fix, fixuns.
10440         (fix_trunc_optab): New.
10442 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10444         * config/aarch64/aarch64-builtins.c
10445         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
10446         iceilf, lround, iroundf.
10448 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10450         PR target/54349
10451         * config/i386/i386.h (enum ix86_tune_indices)
10452         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
10453         New, split from X86_TUNE_INTER_UNIT_MOVES.
10454         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
10455         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
10456         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
10457         (TARGET_INTER_UNIT_MOVES): Remove.
10458         * config/i386/i386.c (initial_ix86_tune_features): Update.
10459         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
10460         (ix86_expand_convert_uns_didf_sse): Use
10461         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10462         (ix86_expand_vector_init_one_nonzero): Ditto.
10463         (ix86_expand_vector_init_interleave): Ditto.
10464         (inline_secondary_memory_needed): Return true for moves from SSE class
10465         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
10466         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
10467         * config/i386/constraints.md (Yi, Ym): Depend on
10468         TARGET_INTER_UNIT_MOVES_TO_VEC.
10469         (Yj, Yn): New constraints.
10470         * config/i386/i386.md (*movdi_internal): Change constraints of
10471         operand 1 from Yi to Yj and from Ym to Yn.
10472         (*movsi_internal): Ditto.
10473         (*movdf_internal): Ditto.
10474         (*movsf_internal): Ditto.
10475         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
10476         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10477         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
10478         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
10479         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
10480         * config/i386/sse.md (movdi_to_sse): Ditto.
10481         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
10482         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
10483         TARGET_INTER_UNIT_MOVES.
10484         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
10485         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
10486         instead of TARGET_INTER_UNIT_MOVES.
10487         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
10488         operand 1 from Yi to Yj and from Ym to Yn.
10490 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10492         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
10493         (float_truncate_hi_): Likewise.
10494         (float_extend_lo_): Likewise.
10495         (float_truncate_lo_): Likewise.
10496         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
10497         (aarch64_float_extend_lo_v2df): Likewise.
10498         (vec_unpacks_hi_v4sf): Likewise.
10499         (aarch64_float_truncate_lo_v2sf): Likewise.
10500         (aarch64_float_truncate_hi_v4sf): Likewise.
10501         (vec_pack_trunc_v2df): Likewise.
10502         (vec_pack_trunc_df): Likewise.
10504 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10506         * config/aarch64/aarch64-builtins.c
10507         (aarch64_fold_builtin): Fold float conversions.
10508         * config/aarch64/aarch64-simd-builtins.def
10509         (floatv2si, floatv4si, floatv2di): New.
10510         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
10511         * config/aarch64/aarch64-simd.md
10512         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
10513         * config/aarch64/iterators.md (FLOATUORS): New.
10514         (optab): Add float, floatuns.
10515         (su_optab): Likewise.
10517 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10519         * config/aarch64/aarch64-builtins.c
10520         (aarch64_builtin_vectorized_function): Use new names for
10521         fcvt builtins.
10522         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
10523         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
10524         (fcvtzu): Split as...
10525         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
10526         (fcvtas): Split as...
10527         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
10528         (fcvtau): Split as...
10529         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
10530         (fcvtps): Split as...
10531         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
10532         (fcvtpu): Split as...
10533         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
10534         (fcvtms): Split as...
10535         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
10536         (fcvtmu): Split as...
10537         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
10538         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
10539         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
10540         (lfrintnusf, lfrintnudf): Likewise.
10541         * config/aarch64/aarch64-simd.md
10542         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
10543         define_insn.
10544         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
10545         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
10546         (fcvt_pattern): Likewise.
10548 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10550         * config/aarch64/aarch64-simd.md
10551         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
10552         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
10554 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10556         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
10557         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
10558         (vrnd<a,m,n,p>_f32): Implement using builtins.
10559         (vrnd<i,x><q>_f<32, 64>): New.
10561 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10563         * config/aarch64/aarch64-builtins.c
10564         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
10565         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
10566         (frintz): Rename to...
10567         (btrunc): ...this.
10568         (frintp): Rename to...
10569         (ceil): ...this.
10570         (frintm): Rename to...
10571         (floor): ...this.
10572         (frinti): Rename to...
10573         (nearbyint): ...this.
10574         (frintx): Rename to...
10575         (rint): ...this.
10576         (frinta): Rename to...
10577         (round): ...this.
10578         * config/aarch64/aarch64-simd.md
10579         (aarch64_frint<frint_suffix><mode>): Delete.
10580         (<frint_pattern><mode>2): Convert to insn.
10581         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
10582         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
10583         (frint_pattern): Likewise.
10584         (frint_suffix): Likewise.
10586 2013-04-29  Richard Biener  <rguenther@suse.de>
10588         PR tree-optimization/57081
10589         * loop-init.c: Include tree-flow.h.
10590         (loop_optimizer_finalize): Free number of iteration estimates.
10591         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
10593 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
10595         PR tree-optimization/57083
10596         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
10597         non-singleton shift count range, zero extend low_bound for uns case.
10599         * config/i386/predicates.md (general_vector_operand): New predicate.
10600         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
10601         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
10602         if they aren't nonimmediate operands.  If their original values
10603         satisfy const_vector_equal_evenodd_p, don't shift them.
10604         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
10605         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
10606         if not nonimmediate_operand.
10607         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
10608         instead of register_operand.
10609         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
10611 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
10613         * stor-layout.c (finalize_size_functions): Allocate a structure and
10614         reset cfun before dumping the functions.
10616 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
10618         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
10620         PR target/56866
10621         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
10622         use xop_pmacsdqh if uns_p.
10623         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
10624         the immediate rotate count.
10626 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
10628         * rtl.h (struct rtx_def): Add comment for field jump.
10629         (LRA_SUBREG_P): New macro.
10630         * recog.c (register_operand): Check LRA_SUBREG_P.
10631         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10632         stack frame.
10633         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10634         (lra_final_code_change): Skip subreg change for operators.
10635         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10636         if there are no operand changes.
10637         * lra-constraints.c (curr_insn_set): New.
10638         (match_reload): Set LRA_SUBREG_P.
10639         (emit_spill_move): Ditto.
10640         (check_and_process_move): Use curr_insn_set. Process only single
10641         set insns.  Don't initialize sec_mem_p and change_p.
10642         (simplify_operand_subreg): Use LRA_SUBREG_P.
10643         (reg_in_class_p): New function.
10644         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10645         of #ifdef.  Add code to remove cycling.
10646         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10647         non-null disp.  Reload inner instead of disp when base and index
10648         are null.  Try to put lo_sum into register.
10649         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10650         (check_and_process_move): Move code for move cost check to
10651         simple_move_p.  Remove equiv_substitution.
10652         (simple_move_p): New function.
10653         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10654         curr_insn_set.  Call check_and_process_move only for single set
10655         insns.  Use the new function.  Move call of check_and_process_move
10656         after operand equiv substitution and address process.
10658 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10660         PR go/57045
10661         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
10662         with nonlocal goto receivers or returns twice calls, ignore
10663         unininitialized values from abnormal edges to nl goto receiver
10664         or returns twice call.
10666 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10668         PR tree-optimization/57051
10669         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
10670         and VEC_RSHIFT_EXPR if shift count is a multiple of element
10671         bitsize.
10673 2013-04-26  Richard Biener  <rguenther@suse.de>
10675         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
10676         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
10677         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
10678         (fixup_loop_arrays_after_move): New function.
10679         (move_sese_region_to_fn): Properly outline the loop tree parts
10680         of the SESE region.
10682 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
10684         * config/i386/i386.md (type, unit): Fix long lines.
10686 2013-04-26  Richard Biener  <rguenther@suse.de>
10688         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
10689         (lto-streamer-out.o): Likewise.
10690         * cfgloop.c (init_loops_structure): Export, add struct function
10691         argument and adjust.
10692         (flow_loops_find): Adjust.
10693         * cfgloop.h (enum loop_estimation): Add EST_LAST.
10694         (init_loops_structure): Declare.
10695         * lto-streamer-in.c: Include cfgloop.h.
10696         (input_cfg): Input the loop tree.
10697         * lto-streamer-out.c: Include cfgloop.h.
10698         (output_cfg): Output the loop tree.
10699         (output_struct_function_base): Do not drop PROP_loops.
10701 2013-03-26  Richard Biener  <rguenther@suse.de>
10703         * tree-cfg.c (execute_build_cfg): Build the loop tree.
10704         (pass_build_cfg): Provide PROP_loops.
10705         (move_sese_region_to_fn): Remove loops that are outlined into fn
10706         for now.
10707         * tree-inline.c: Include cfgloop.h.
10708         (initialize_cfun): Do not drop PROP_loops.
10709         (copy_loops): New function.
10710         (copy_cfg_body): Copy loop structure.
10711         (tree_function_versioning): Initialize destination loop tree.
10712         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
10713         (pass_parallelize_loops): Do IL verification.
10714         * loop-init.c (loop_optimizer_init): Fixup loops if required.
10715         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
10716         the CFG make sure we fixup loops as well.
10717         * tree-ssa-tail-merge.c: Include cfgloop.h.
10718         (replace_block_by): When merging loop latches mark loops for fixup.
10719         * lto-streamer-out.c (output_struct_function_base): Drop
10720         PROP_loops for now.
10721         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
10722         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
10723         * ipa-split.c: Include cfgloop.h.
10724         (split_function): Add the new return block to the loop tree root.
10725         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
10726         whether we have removed the forwarder block.
10727         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
10728         * cfgloop.h (place_new_loop): Declare.
10729         * cfgloopmanip.c (place_new_loop): Export.
10730         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
10731         (tree-switch-conversion.o): Likewise.
10732         (tree-complex.o): Likewise.
10733         (tree-inline.o): Likewise.
10734         (tree-ssa-tailmerge.o): Likewise.
10735         (ipa-split.o): Likewise.
10736         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
10737         (tree-ssa-copy.o): Likewise.
10738         * tree-switch-conversion.c: Include cfgloop.h
10739         (process_switch): If we emit a bit-test cascade, schedule loops
10740         for fixup.
10741         * tree-complex.c: Include cfgloop.h.
10742         (expand_complex_div_wide): Properly add new basic-blocks to loops.
10743         * asan.c: Include cfgloop.h.
10744         (create_cond_insert_point): Properly add new basic-blocks to
10745         loops, schedule loop fixup.
10746         * cfgloop.c (verify_loop_structure): Check that looks are not
10747         marked for fixup.
10748         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
10749         to loops.
10750         (expand_omp_for_generic): Likewise.
10751         (expand_omp_sections): Likewise.
10752         (expand_omp_atomic_pipeline): Schedule loops for fixup.
10753         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
10754         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
10755         is initialized, not when loops are present.
10756         * tree-parloops.c (parallelize_loops): Remove checking here.
10757         * passes.c (init_optimization_passes): Schedule a copy-propagation
10758         pass before complete unrolling of inner loops.
10760 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10762         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
10763         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
10764         (colorize_init): Add argument to _WIN32 version.
10765         * toplev.c: Include diagnostic-color.h.
10766         (process_options): Default to -fdiagnostics-color=auto if
10767         GCC_COLORS env var is in the environment.
10768         * common.opt (fdiagnostics-color=): Add Var and Init.
10769         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
10770         env var is in the environment, the default is auto rather than never.
10772         * diagnostic.h (file_name_as_prefix): Add context argument.
10773         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
10774         the string as locus.
10775         * langhooks.c (lhd_print_error_function): Adjust caller.
10777 2013-04-25  Lawrence Crowl  <crowl@google.com>
10779         * var-tracking.c (shared_hash_def::htab):
10780         Change type to hash_table.  Update dependent calls and types.
10782 2013-04-25  Lawrence Crowl  <crowl@google.com>
10784         * Makefile.in: Update as needed below.
10786         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
10787         Move declaration to after the type's method definitons.
10789         * attribs.c (htab_t scoped_attributes::attribute_hash):
10790         Change type to hash_table.  Update dependent calls and types.
10792         * bitmap.c (htab_t bitmap_desc_hash):
10793         Change type to hash_table.  Update dependent calls and types.
10795         * cselib.c (htab_t cselib_hash_table):
10796         Change type to hash_table.  Update dependent calls and types.
10798         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
10799         (hash_string_slot_node): Move implementation into lto-streamer.h
10800         struct string_slot_hasher.
10801         (eq_string_slot_node): Likewise.
10803         * data-streamer-out.c: Update output_block::string_hash_table
10804         dependent calls and types.
10806         * dwarf2cfi.c (htab_t trace_index):
10807         Change type to hash_table.  Update dependent calls and types.
10809         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
10810         Change type to hash_table.  Update dependent calls and types.
10811         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
10812         (htab_t optimize_external_refs::map): Likewise.
10813         (htab_t output_comp_unit::extern_map): Likewise.
10814         (htab_t output_comdat_type_unit::extern_map): Likewise.
10815         (htab_t output_macinfo::macinfo_htab): Likewise.
10816         (htab_t optimize_location_lists::htab): Likewise.
10817         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
10819         * except.c (htab_t ehspec_hash_type):
10820         Change type to hash_table.  Update dependent calls and types.
10821         (assign_filter_values::ttypes): Likewise.
10822         (assign_filter_values::ehspec): Likewise.
10823         (sjlj_assign_call_site_values::ar_hash): Likewise.
10824         (convert_to_eh_region_ranges::ar_hash): Likewise.
10826         * gcse.c (htab_t pre_ldst_table):
10827         Change type to hash_table.  Update dependent calls and types.
10829         * ggc-common.c (htab_t saving_htab):
10830         Change type to hash_table.  Update dependent calls and types.
10831         (htab_t loc_hash): Likewise.
10832         (htab_t ptr_hash): Likewise.
10833         (call_count): Rename ggc_call_count.
10834         (call_alloc): Rename ggc_call_alloc.
10835         (loc_descriptor): Rename make_loc_descriptor.
10836         (add_statistics): Rename ggc_add_statistics.
10838         * ggc-common.c (saving_htab):
10839         Change type to hash_table.  Update dependent calls and types.
10841         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
10842         (push_gimplify_context): Likewise.
10843         (pop_gimplify_context): Likewise.
10844         (struct gimple_temp_hash_elt): Added.
10845         (struct gimplify_hasher): Likewise.
10846         (struct gimplify_ctx.temp_htab):
10847         Change type to hash_table.  Update dependent calls and types.
10849         * gimple-fold.c: Include gimplify-ctx.h.
10851         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
10852         Change type to hash_table.  Update dependent calls and types.
10853         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
10854         avoid potential global name collision.
10856         * gimplify.c: Include gimplify-ctx.h.
10857         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
10858         (htab_t gimplify_ctx::temp_htab):
10859         Update dependent calls and types for new type hash_table.
10860         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
10861         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
10863         * gimplify-ctx.h: New.
10864         (struct gimple_temp_hash_elt): Move from gimplify.c.
10865         (class gimplify_hasher): New.
10866         (struct gimplify_ctx): Move from gimple.h.
10867         (htab_t gimplify_ctx::temp_htab):
10868         Change type to hash_table.  Update dependent calls and types.
10870         * graphite-clast-to-gimple.c: Include graphite-htab.h.
10871         (htab_t ivs_params::newivs_index):
10872         Change type to hash_table.  Update dependent calls and types.
10873         (htab_t ivs_params::params_index): Likewise.
10874         (htab_t print_generated_program::params_index): Likewise.
10875         (htab_t gloog::newivs_index): Likewise.
10876         (htab_t gloog::params_index): Likewise.
10878         * graphite.c: Include graphite-htab.h.
10879         4htab_t graphite_transform_loops::bb_pbb_mapping):
10880         Change type to hash_table.  Update dependent calls and types.
10882         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
10883         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
10884         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
10886         * graphite-dependences.c: Include graphite-htab.h.
10887         (loop_is_parallel_p): Change hash table type of parameter.
10889         * graphite-htab.h: New.
10890         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
10891         (extern find_pbb_via_hash): Move from graphite-poly.h.
10892         (extern loop_is_parallel_p): Move from graphite-poly.h.
10893         (extern get_loop_body_pbbs): Move from graphite-poly.h.
10895         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
10896         (extern loop_is_parallel_p): Move to graphite-htab.h.
10897         (extern get_loop_body_pbbs): Move to graphite-htab.h.
10899         * haifa-sched.c (htab_t delay_htab):
10900         Change type to hash_table.  Update dependent calls and types.
10901         (htab_t delay_htab_i2): Likewise.
10903         * ira-color.c (htab_t allocno_hard_regs_htab):
10904         Change type to hash_table.  Update dependent calls and types.
10906         * ira-costs.c (htab_t cost_classes_htab):
10907         Change type to hash_table.  Update dependent calls and types.
10909         * loop-invariant.c (htab_t merge_identical_invariants::eq):
10910         Change type to hash_table.  Update dependent calls and types.
10912         * loop-iv.c (htab_t bivs):
10913         Change type to hash_table.  Update dependent calls and types.
10915         * loop-unroll.c (htab_t opt_info::insns_to_split):
10916         Change type to hash_table.  Update dependent calls and types.
10917         (htab_t opt_info::insns_with_var_to_expand): Likewise.
10919         * lto-streamer.h (struct string_slot): Move from data-streamer.h
10920         (struct string_slot_hasher): New.
10921         (htab_t output_block::string_hash_table):
10922         Change type to hash_table.  Update dependent calls and types.
10924         * lto-streamer-in.c (freeing_string_slot_hasher): New.
10925         (htab_t file_name_hash_table):
10926         Change type to hash_table.  Update dependent calls and types.
10928         * lto-streamer-out.c: Update output_block::string_hash_table dependent
10929         calls and types.
10931         * lto-streamer.c (htab_t tree_htab):
10932         Change type to hash_table.  Update dependent calls and types.
10934         * omp-low.c: Include gimplify-ctx.h.
10936         * passes.c (htab_t name_to_pass_map):
10937         Change type to hash_table.  Update dependent calls and types.
10938         (pass_traverse): Rename to passes_pass_traverse.
10940         * plugin.c (htab_t event_tab):
10941         Change type to hash_table.  Update dependent calls and types.
10943         * postreload-gcse.c (htab_t expr_table):
10944         Change type to hash_table.  Update dependent calls and types.
10945         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10947         * sese.c (debug_rename_map_1): Make extern.
10948         (htab_t copy_bb_and_scalar_dependences::rename_map):
10949         Change type to hash_table.  Update dependent calls and types.
10951         * sese.h (extern debug_rename_map): Move to .c file.
10953         * store-motion.c (htab_t store_motion_mems_table):
10954         Change type to hash_table.  Update dependent calls and types.
10956         * trans-mem.c (htab_t tm_new_mem_hash):
10957         Change type to hash_table.  Update dependent calls and types.
10959         * tree-browser.c (htab_t TB_up_ht):
10960         Change type to hash_table.  Update dependent calls and types.
10962         * tree-cfg.c (htab_t discriminator_per_locus):
10963         Change type to hash_table.  Update dependent calls and types.
10965         * tree-complex.c: Include tree-hasher.h
10966         (htab_t complex_variable_components):
10967         Change type to hash_table.  Update dependent calls and types.
10969         * tree-eh.c (htab_t finally_tree):
10970         Change type to hash_table.  Update dependent calls and types.
10972         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10973         struct int_tree_hasher.
10974         (extern int_tree_map_eq): Likewise.
10975         (uid_decl_map_hash): Removed.
10976         (extern decl_tree_map_eq): Likewise.
10978         * tree-hasher.h: New.
10979         (struct int_tree_hasher): New.
10980         (typedef int_tree_htab_type): New.
10982         * tree-inline.c: Include gimplify-ctx.h.
10984         * tree-mudflap.c: Include gimplify-ctx.h.
10986         * tree-parloops.c: Include tree-hasher.h.
10987         (htab_t eliminate_local_variables_stmt::decl_address):
10988         Change type to hash_table.  Update dependent calls and types.
10989         (htab_t separate_decls_in_region::decl_copies): Likewise.
10991         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10992         Change type to hash_table.  Update dependent calls and types.
10994         * tree-sra.c (candidates):
10995         Change type to hash_table.  Update dependent calls and types.
10997         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10998         in tree-flow.h.
10999         (int_tree_map_hash): Likewise.
11001         * tree-ssa-dom.c (htab_t avail_exprs):
11002         Change type to hash_table.  Update dependent calls and types.
11004         * tree-ssa-live.c (var_map_base_init::tree_to_index):
11005         Change type to hash_table.  Update dependent calls and types.
11007         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11008         Change type to hash_table.  Update dependent calls and types.
11010         * tree-ssa-phiopt.c (seen_ssa_names):
11011         Change type to hash_table.  Update dependent calls and types.
11013         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11014         Change type to hash_table.  Update dependent calls and types.
11016         * tree-ssa-uncprop.c (equiv):
11017         Change type to hash_table.  Update dependent calls and types.
11019 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
11021         PR rtl-optimization/57003
11022         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11023         call note_stores with kill_clobbered_value callback again after
11024         killing regs_invalidated_by_call.
11026 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11028         * config/aarch64/aarch64-simd.md
11029         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11030         (aarch64_simd_bsl<mode>): Likewise.
11031         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11033 2013-04-25  Marek Polacek  <polacek@redhat.com>
11035         PR tree-optimization/57066
11036         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11038 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
11040         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11042 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11044         * config/aarch64/aarch64-builtins.c
11045         (aarch64_fold_builtin): New.
11046         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11047         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11048         * config/aarch64/aarch64-simd-builtins.def (abs): New.
11049         * config/aarch64/arm_neon.h
11050         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11052 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11053             Tejas Belagod  <tejas.belagod@arm.com>
11055         * config/aarch64/aarch64-builtins.c
11056         (aarch64_gimple_fold_builtin): New.
11057         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11058         * config/aarch64/aarch64-simd-builtins.def (addv): New.
11059         * config/aarch64/aarch64-simd.md (addpv4sf): New.
11060         (addvv4sf): Update.
11061         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11063 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11065         * config/aarch64/aarch64.md
11066         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11068 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11070         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11071         (*ngcsi_uxtw): New pattern.
11073 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11074             Julian Brown  <julian@codesourcery.com>
11076         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11077         (TB_DREG): Add T_V4HF.
11078         (v4hf_UP): New macro.
11079         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11080         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11081         Handle initialisation of V4HF. Adjust initialisation of reinterpret
11082         built-ins.
11083         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11084         (arm_vector_mode_supported_p): Handle V4HF.
11085         (arm_mangle_map): Handle V4HFmode.
11086         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11087         * config/arm/arm_neon_builtins.def: Add entries for
11088         vcvtv4hfv4sf, vcvtv4sfv4hf.
11089         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11090         (neon_vcvtv4hfv4sf): Likewise.
11091         * config/arm/neon-gen.ml: Handle half-precision floating point
11092         features.
11093         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11094         * config/arm/arm_neon.h: Regenerate.
11095         * config/arm/neon.ml (type elts): Add F16.
11096         (type vectype): Add T_float16x4, T_floatHF.
11097         (type vecmode): Add V4HF.
11098         (type features): Add Requires_FP_bit feature.
11099         (elt_width): Handle F16.
11100         (elt_class): Likewise.
11101         (elt_of_class_width): Likewise.
11102         (mode_of_elt): Refactor.
11103         (type_for_elt): Handle F16, fix error messages.
11104         (vectype_size): Handle T_float16x4.
11105         (vcvt_sh): New function.
11106         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11107         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11108         (string_of_mode): Handle V4HF.
11109         * doc/arm-neon-intrinsics.texi: Regenerate.
11111 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11113         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11114         format specifier in 'X' case.
11116 2013-04-25  Alan Modra  <amodra@gmail.com>
11118         PR target/57052
11119         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11120         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11121         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11122         Repeat for many other rotate/shift and mask patterns using subregs.
11123         Name lshiftrt insns.
11124         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11125         on WORDS_BIG_ENDIAN.
11127 2013-04-25  Alan Modra  <amodra@gmail.com>
11129         * config.gcc: Support little-endian powerpc-linux targets.
11130         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11131         (LINK_OS_LINUX_SPEC): Define.
11132         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11133         Preserve MASK_LITTLE_ENDIAN.
11134         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11135         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11136         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11137         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11138         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11139         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11140         Correct fp word order for little-endian.  Don't shift toc entries
11141         smaller than a word for little-endian.
11142         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11143         (bswapdi2 splits): Correct low-part subreg for little-endian.
11144         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11145         low/high where such is correct only for be.
11146         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11147         little-endian for -mcall-aixdesc.
11149 2013-04-25  Alan Modra  <amodra@gmail.com>
11151         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11152         replace_equiv_address_nv.
11154 2013-04-25  Alan Modra  <amodra@gmail.com>
11156         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11158 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11160         Revert:
11161         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11162         * rtl.h (struct rtx_def): ...
11164 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11166         PR rtl-optimizations/57046
11167         * lra-constraints (split_reg): Set up lra_risky_transformations_p
11168         for multi-reg splits.
11170 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
11172         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11174 2013-04-24  Sterling Augustine  <saugustine@google.com>
11176         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11177         (comp_dir_string, debug_str_dwo_section): New.
11178         (DEBUG_STR_DWO_SECTION): Rename to ...
11179         (DEBUG_DWO_STR_SECTION): ... this.
11180         (DEBUG_NORM_STR_SECTION): Delete.
11181         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11182         (DEBUG_STR_DWO_SECTION_FLAGS): New.
11183         (find_AT_string): Move most logic to ...
11184         (find_AT_string_in_table): ... here.  New.
11185         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11186         add_skeleton_AT_string.  Delete logic.
11187         (output_skeleton_debug_sections): Remove call to
11188         add_top_level_skeleton_die_attrs.
11189         (add_comp_dir_attribute): Move logic to comp_dir_string.
11190         (dwarf2out_init): Initialize debug_str_dwo_section.
11191         (output_indirect_string): Call find_string_form.
11192         (output_indirect_strings): Rewrite.
11193         (prune_unused_types): Empty skeleton_debug_str_hash.
11194         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
11195         (dwarf2out_finish):  Call output_indirect_strings.
11197 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11199         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
11201 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11203         * rtl.h (struct rtx_def): Add comment for field jump.
11204         (LRA_SUBREG_P): New macro.
11205         * recog.c (register_operand): Check LRA_SUBREG_P.
11206         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11207         stack frame.
11208         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11209         (lra_final_code_change): Skip subreg change for operators.
11210         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11211         if there are no operand changes.
11212         * lra-constraints.c (curr_insn_set): New.
11213         (match_reload): Set LRA_SUBREG_P.
11214         (emit_spill_move): Ditto.
11215         (check_and_process_move): Use curr_insn_set. Process only single
11216         set insns.  Don't initialize sec_mem_p and change_p.
11217         (simplify_operand_subreg): Use LRA_SUBREG_P.
11218         (reg_in_class_p): New function.
11219         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11220         of #ifdef.  Add code to remove cycling.
11221         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11222         non-null disp.  Reload inner instead of disp when base and index
11223         are null.  Try to put lo_sum into register.
11224         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11225         (check_and_process_move): Move code for move cost check to
11226         simple_move_p.  Remove equiv_substitution.
11227         (simple_move_p): New function.
11228         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11229         curr_insn_set.  Call check_and_process_move only for single set
11230         insns.  Use the new function.  Move call of check_and_process_move
11231         after operand equiv substitution and address process.
11233 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
11235         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
11236         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
11237         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
11239 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11241         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
11243 2013-04-24  Marek Polacek  <polacek@redhat.com>
11245         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
11246         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
11247         (select_loops_exit_conditions): Likewise.
11248         (number_of_iterations_for_all_loops): Likewise.
11249         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
11250         (scev_analysis): Likewise.
11252 2013-04-02  Catherine Moore  <clm@codesourcery.com>
11253             Chao-ying Fu <fu@mips.com>
11255         * config/mips/micromips.md (jraddiusp): New pattern.
11256         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
11257         instruction if possible.
11259 2013-04-24  Alan Modra  <amodra@gmail.com>
11261         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
11263 2013-04-24  Julian Brown  <julian@codesourcery.com>
11264             Chung-Lin Tang  <cltang@codesourcery.com>
11266         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
11267         dependency behavior in enumeration type DIE generation. Add TODO note
11268         to comments about future DW_FORM_sdata/udata re-work of related code.
11270 2013-04-23  Lawrence Crowl  <crowl@google.com>
11272         * Makefile.in: Update as needed below.
11274         * hash-table.h (class hash_table):
11275         Correct many methods with parameter types compare_type to the correct
11276         value_type.  (Correct code was unlikely to notice the change.)
11277         (hash_table::elements_with_deleted) New.
11278         (class hashtable::iterator): New.
11279         (hashtable::begin()): New.
11280         (hashtable::end()): New.
11281         (FOR_EACH_HASH_TABLE_ELEMENT): New.
11283         * statistics.c (statistics_hashes):
11284         Change type to hash_table.  Update dependent calls and types.
11286         * tree-into-ssa.c (var_infos):
11287         Change type to hash_table.  Update dependent calls and types.
11289         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
11290         Change type to hash_table.  Update dependent calls and types.
11292         * tree-ssa-loop-im.c (struct mem_ref.refs):
11293         Change type to hash_table.  Update dependent calls and types.
11295         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
11296         Change type to hash_table.  Update dependent calls and types.
11298         * tree-ssa-sccvn.c (vn_tables_s::nary):
11299         Change type to hash_table.  Update dependent calls and types.
11300         (vn_tables_s::phis): Likewise.
11301         (vn_tables_s::references): Likewise.
11303         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
11304         (vn_reference_eq): Update parameter and return types.
11306         * tree-ssa-structalias.c (pointer_equiv_class_table):
11307         Change type to hash_table.  Update dependent calls and types.
11308         (location_equiv_class_table): Likewise.
11310         * tree-vect-data-refs.c: Consequential changes for making
11311         peeling a hash_table.
11313         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
11314         (destroy_loop_vec_info): Dependent hash_table update.
11316         * tree-vectorizer.h (peeling_htab):
11317         Change type to hash_table.  Update dependent calls and types.
11319 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
11321         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
11322         to check the register content is equal or not.
11323         * lra-constraints.c (match_reload): Use lra_assign_reg_val
11324         to assign register content record.
11325         * lra-eliminations.c (update_reg_eliminate): Use
11326         lra_update_reg_val_offset to update register content offset.
11327         * lra-int.h (struct lra_reg): Add offset member.
11328         (lra_reg_val_equal_p): New static inline function.
11329         (lra_update_reg_val_offset): New static inline function.
11330         (lra_assign_reg_val): New static inline function.
11331         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
11332         to assign register content record.
11333         (initialize_lra_reg_info_element): Initial offset to zero.
11335 2013-04-23  Catherine Moore  <clm@codesourcery.com>
11337         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
11338         operands.  Record compression.
11340 2013-04-23  Xinliang David Li  <davidxl@google.com>
11342         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
11344 2013-04-23  Richard Biener  <rguenther@suse.de>
11346         PR middle-end/57036
11347         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
11348         parameter, only add abnormal goto edges from the copied body
11349         if the call could perform abnormal gotos.
11350         (copy_cfg_body): Adjust.
11352 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
11354         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
11356 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
11358         * coretypes.h (gimple_stmt_iterator): Add struct to make
11359         compatible with C.
11361 2013-04-23  Richard Biener  <rguenther@suse.de>
11363         PR tree-optimization/57026
11364         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
11365         from SSA names occuring in abnormal PHI nodes.
11367 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11369         * lto/lto.c (print_lto_report_1): Fix LTO report names.
11371 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11373         * lto/lto.c (print_lto_report_1): Declare early.
11374         (read_cgraph_and_symbols): Call print_lto_report_1 early.
11376 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11378         * common.opt (-flto-report-wpa): Add.
11379         * doc/invoke.texi (-flto-report-wpa): Add.
11380         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
11381         (lto_main): dito.
11383 2013-04-22  Xinliang David Li  <davidxl@google.com>
11385         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
11386         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
11387         * Makefile.in: New dependency
11389             David Daney <ddaney.cavm@gmail.com>
11391         * configure.ac (gcc_cv_as_micromips_support):  Use the
11392         --fatal-warnings option.
11393         * configure: Regenerate.
11395 2013-04-22  Marek Polacek  <polacek@redhat.com>
11397         PR sanitizer/56990
11398         * tsan.c (instrument_expr): Don't instrument expression
11399         in case its size is zero.
11401 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
11403         PR target/57032
11404         Revert:
11405         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
11407         * config/alpha/alpha.c (TARGET_LRA_P): New define.
11409 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11411         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
11412         (gimple_stmt_iterator): New typedef.
11413         * gimple.h (gimple_stmt_iterator): Rename to...
11414         (gimple_stmt_iterator_d): ... This.
11415         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
11416         trees be valid for GIMPLE and GENERIC.
11417         (TARGET_GIMPLE_FOLD_BUILTIN): New.
11418         * gimple-fold.c (gimple_fold_call): Call target hook
11419         gimple_fold_builtin.
11420         * hooks.c (hook_bool_gsiptr_false): New.
11421         * hooks.h (hook_bool_gsiptr_false): New.
11422         * target.def (fold_stmt): New.
11423         * doc/tm.texi: Regenerate.
11425 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
11427         PR target/57018
11428         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
11429         a set sp if no stack realignment.
11431 2013-04-22  Nick Clifton  <nickc@redhat.com>
11433         * config.gcc (tilegx-linux): Extend extra_objs rather than
11434         overwriting it.
11435         (tilepro-linux): Likewise.
11437 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11439         * config/aarch64/aarch64-builtins.c
11440         (CF): Remove.
11441         (CF0, CF1, CF2, CF3, CF4, CF10): New.
11442         (VAR<1-12>): Add MAP parameter.
11443         (BUILTIN_*): Likewise.
11444         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
11445         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
11446         (aarch64_ushl_n<mode>): Likewise.
11447         (aarch64_sshr_n<mode>): Likewise.
11448         (aarch64_ushr_n<mode>): Likewise.
11449         (aarch64_<maxmin><mode>): Likewise.
11450         (aarch64_sqrt<mode>): Likewise.
11451         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
11452         (vshr<q>_n_*): Likewise.
11454 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11456         * config/aarch64/aarch64-builtins.c
11457         (aarch64_simd_builtin_type_mode): Handle SF types.
11458         (sf_UP): Define.
11459         (BUILTIN_GPF): Define.
11460         (aarch64_init_simd_builtins): Handle SF types.
11461         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
11462         (frecps): Likewise.
11463         (frecpx): Likewise.
11464         * config/aarch64/aarch64-simd.md
11465         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
11466         (aarch64_frecpe<mode>): New.
11467         (aarch64_frecps<mode>): Likewise.
11468         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
11469         (v8type): Add frecp<esx>.
11470         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
11471         (aarch64_frecps<mode>): Likewise.
11472         * config/aarch64/iterators.md (FRECP): New.
11473         (frecp_suffix): Likewise.
11474         * config/aarch64/arm_neon.h
11475         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
11477 2013-04-22  Christian Bruel  <christian.bruel@st.com>
11479         PR target/56995
11480         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
11481         (REG_CLASS_NAMES): Idem.
11482         (REG_CLASS_CONTENTS): Idem.
11483         (REGCLASS_HAS_FP_REG): Idem.
11484         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
11485         (sh_conditional_register_usage): Idem.
11487 2013-04-21  Jeff Law  <law@redhat.com>
11489         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
11490         (ssa_forward_propagate_and_combine): Use it.
11492 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11494         * lra.c: Update the flow chart diagram.
11496 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11498         PR rtl-optimization/56847
11499         * lra-constraints.c (process_alt_operands): Discourage alternative
11500         with non-matche doffsettable memory constraint fro memory with
11501         known offset.
11503 2013-04-19  Richard Biener  <rguenther@suse.de>
11505         PR tree-optimization/56982
11506         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
11507         function.
11508         * gimplify.c (gimplify_call_expr): Notice special calls.
11509         (gimplify_modify_expr): Likewise.
11510         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
11511         abnormal control flow receivers.
11512         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
11513         in the same way as cfun->has_nonlocal_labels.
11514         (gimple_purge_dead_abnormal_call_edges): Likewise.
11515         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
11516         receivers start a basic-block.
11518 2013-04-19  Richard Biener  <rguenther@suse.de>
11520         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
11521         member ...
11522         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
11523         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
11524         (SLP_TREE_LOAD_PERMUTATION): Add.
11525         (vect_transform_slp_perm_load): Adjust prototype.
11526         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
11527         (vect_free_slp_instance): Likewise.
11528         (vect_create_new_slp_node): Likewise.
11529         (vect_supported_slp_permutation_p): Remove.
11530         (vect_slp_rearrange_stmts): Adjust.
11531         (vect_supported_load_permutation_p): Likewise.  Inline
11532         vect_supported_slp_permutation_p here.
11533         (vect_analyze_slp_instance): Compute load permutations per
11534         slp node instead of per instance.
11535         (vect_get_slp_defs): Adjust.
11536         (vect_transform_slp_perm_load): Likewise.
11537         (vect_schedule_slp_instance): Remove redundant code.
11538         (vect_schedule_slp): Remove hack for PR56270, add it ...
11539         * tree-vect-stmts.c (vectorizable_load): ... here, do not
11540         CSE loads for SLP.  Adjust.
11542 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11544         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
11545         spelling in two comments.
11547 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11549         PR target/56797
11550         * config/arm/arm.c (load_multiple_sequence): Require SP
11551         as base register for loads if SP is in the register list.
11553 2013-04-19  Martin Jambor  <mjambor@suse.cz>
11555         PR tree-optimization/56718
11556         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
11557         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
11558         and made public.  Adjusted all callers.
11559         (ipa_intraprocedural_devirtualization): New function.
11560         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
11561         (ipa_intraprocedural_devirtualization): Likewise.
11562         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
11564 2013-04-19  Richard Biener  <rguenther@suse.de>
11566         PR tree-optimization/57000
11567         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
11569 2013-04-19  Terry Guo  <terry.guo@arm.com>
11571         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
11572         Replace with ...
11573         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
11574         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
11575         (cortex_m4_fmacs): Use new reservations.
11576         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
11578 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
11580         PR rtl-optimization/56999
11581         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
11582         related code.
11583         (lra_coalesce): Remove split_origin_bitmap and related code.
11584         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
11585         ranges if necessary.
11587 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
11589         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
11590         New array.
11591         (ix86_expand_call): Remove clobbered_registers array and use
11592         x86_64_ms_sysv_extra_clobbered_registers instead.
11593         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
11594         Declare here.
11595         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
11596         predicate.
11597         * config/i386/i386.md (*call_rex64_ms_sysv): Use
11598         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
11599         (*call_value_rex64_ms_sysv): Ditto.
11601 2013-04-18  Cary Coutant  <ccoutant@google.com>
11603         * dwarf2out.c (output_pubnames): Check die_perennial_p of
11604         parent instead of die_mark.
11606 2013-04-18  Diego Novillo  <dnovillo@google.com>
11608         * gimple.c (create_gimple_tmp): New.
11609         (get_expr_type): New.
11610         (build_assign): New.
11611         (build_type_cast): New.
11612         * gimple.h (enum ssa_mode): Define.
11613         (gimple_seq_set_location): New.
11614         * asan.c (build_check_stmt): Change some gimple_build_* calls
11615         to use build_assign and build_type_cast.
11617 2013-04-18  Richard Biener  <rguenther@suse.de>
11619         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
11620         handle negative step.  Remove redundant checks.
11621         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
11622         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
11623         for negative step and grouped loads fail to vectorize.
11625 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11627         * emit-rtl.c (reset_insn_used_flags): New function.
11628         (reset_all_used_flags): Use it.
11629         (verify_insn_sharing): New function.
11630         (verify_rtl_sharing): Fix verification for SEQUENCEs.
11632 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
11634         PR tree-optimization/56984
11635         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
11636         and (x >> M) >= N don't register any assertion if N << M is the
11637         minimum value.
11639 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11641         * lower-subreg.c (resolve_simple_move): If called self-recursive,
11642         do not delete_insn insns that have not yet been emitted, only
11643         unlink them with remove_insn.
11644         * df-scan.c (df_insn_delete): Revert r197492.
11646 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
11648         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
11649         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
11651 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11653         * config/arm/arm.md (movsicc_insn): Convert define_insn into
11654         define_insn_and_split.
11655         (and_scc,ior_scc,negscc): Likewise.
11656         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
11658 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11660         * config/arm/arm.c (use_return_insn): Return 0 for targets that
11661         can benefit from using a sequence of LDRD instructions in epilogue
11662         instead of a single LDM instruction.
11664 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11666         PR 45688
11667         * doc/extend.texi: Fix typo.
11669 2013-04-17  Richard Biener  <rguenther@suse.de>
11671         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
11672         (vect_build_slp_tree): ... here.
11673         (vect_build_slp_tree_1): Compute which stmts of the SLP group
11674         match.  Remove special-casing of mismatched complex loads.
11675         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
11676         re-try the match with swapped commutative operands.
11677         (vect_supported_load_permutation_p): Remove special-casing of
11678         mismatched complex loads.
11679         (vect_analyze_slp_instance): Adjust.
11681 2013-04-17  Richard Biener  <rguenther@suse.de>
11683         PR rtl-optimization/56921
11684         * cfgloop.h (struct loop): Add simple_loop_desc member.
11685         (struct niter_desc): Mark with GTY(()).
11686         (simple_loop_desc): Do not use aux field but simple_loop_desc.
11687         * loop-iv.c (get_simple_loop_desc): Likewise.
11688         (free_simple_loop_desc): Likewise.
11690         Revert
11691         2013-04-16  Richard Biener  <rguenther@suse.de>
11693         PR rtl-optimization/56921
11694         * loop-init.c (pass_rtl_move_loop_invariants): Add
11695         TODO_do_not_ggc_collect to todo_flags_finish.
11696         (pass_rtl_unswitch): Same.
11697         (pass_rtl_unroll_and_peel_loops): Same.
11698         (pass_rtl_doloop): Same.
11700 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
11702         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
11703         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
11704         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
11705         references.
11706         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
11707         * tree-streamer.c (record_common_node): Adjust reference in comment.
11709 2013-04-17  Terry Guo  <terry.guo@arm.com>
11711         * config/arm/cortex-m4.md: Add a new bypass.
11713 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11715         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
11716         New pattern.
11717         (*subs_<optab><mode>_multp2): New pattern.
11718         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11719         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11721 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11723         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
11724         (*subs_mul_imm_<mode>): New pattern.
11726 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
11728         PR target/56948
11729         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
11730         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
11731         (vsx_movti_32bit): Change j->wa to O->wa.
11733 2013-04-16  Richard Biener  <rguenther@suse.de>
11735         PR rtl-optimization/56921
11736         * loop-init.c (pass_rtl_move_loop_invariants): Add
11737         TODO_do_not_ggc_collect to todo_flags_finish.
11738         (pass_rtl_unswitch): Same.
11739         (pass_rtl_unroll_and_peel_loops): Same.
11740         (pass_rtl_doloop): Same.
11742 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11744         * config/arm/arm.c (emit_multi_reg_push): New declaration
11745         for an existing function.
11746         (arm_emit_strd_push): New function.
11747         (arm_expand_prologue): Used here.
11748         (arm_emit_ldrd_pop): New function.
11749         (arm_expand_epilogue): Used here.
11750         (arm_get_frame_offsets): Update condition.
11751         (arm_emit_multi_reg_pop): Add a special case for load of a single
11752         register with writeback.
11754 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
11756         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
11757         description.
11759 2013-04-16  Richard Biener  <rguenther@suse.de>
11761         PR tree-optimization/56756
11762         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
11763         (first_mem_ref_loc): New.
11764         (execute_sm): Place the load temporarily before a previous
11765         access instead of in the latch edge to ensure its SSA dependencies
11766         are defined at points dominating the load.
11768 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
11770         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
11771         correct fix by moving header and footer insn to the footer of
11772         the merged basic block.  Clear BB_END of the merged-away block.
11774         PR middle-end/43631
11775         * emit-rtl.c (make_note_raw): New function.
11776         (link_insn_into_chain): New static inline function.
11777         (add_insn): Use it.
11778         (add_insn_before, add_insn_after): Factor insn chain linking code...
11779         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
11780         using link_insn_into_chain.
11781         (note_outside_basic_block_p): New helper function for emit_note_after
11782         and emit_note_before.
11783         (emit_note_after): Use nobb variant of add_insn_after if the note
11784         should not be contained in a basic block.
11785         (emit_note_before): Use nobb variant of add_insn_before if the note
11786         should not be contained in a basic block.
11787         (emit_note_copy): Use make_note_raw.
11788         (emit_note): Likewise.
11789         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
11790         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
11791         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
11792         the moved barrier the tail of the basic block it follows.
11793         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
11795 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
11797         PR tree-optimization/56962
11798         * gimple-ssa-strength-reduction.c (record_increment): Only set
11799         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
11800         either rhs1 or rhs2 is equal to c->base_expr.
11802 2013-04-15  Richard Biener  <rguenther@suse.de>
11804         PR tree-optimization/56933
11805         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
11806         member.
11807         (GROUP_READ_WRITE_DEPENDENCE): Remove.
11808         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
11809         * tree-vect-data-refs.c (vect_analyze_group_access): Move
11810         dependence check ...
11811         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
11812         ... here.
11813         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
11814         GROUP_READ_WRITE_DEPENDENCE.
11816 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11818         * emit-rtl.c (reset_all_used_flags): New function.
11819         (verify_rtl_sharing): Call reset_all_used_flags before and after
11820         performing the checks.
11822 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11824         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
11825         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
11826         * config/arm/constraints.md (De): New constraint.
11827         * config/arm/neon.md (anddi3_neon): Delete.
11828         (neon_vand<mode>): Expand to standard anddi3 pattern.
11829         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
11830         Move earlier in the file.
11831         (neon_inv_logic_op2): Likewise.
11832         (arm_anddi_operand_neon): New predicate.
11834 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11836         * configure.ac (gcc_cv_ld_as_needed): Set
11837         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
11838         Use -z ignore, -z record on *-*-solaris2*.
11839         (HAVE_LD_AS_NEEDED): Update comment.
11840         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
11841         * configure: Regenerate.
11842         * config.in: Regenerate.
11843         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
11844         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
11845         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
11846         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
11847         equivalents.  Fix markup.
11848         * doc/tm.texi: Regenerate.
11850 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
11852         * config/i386/i386.opt: New option mstack-protector-guard=.
11853         * config/i386/i386-opts.h: Add enum stack_protector_guard.
11854         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
11855         TARGET_SSP_TLS_GUARD.
11856         * config/i386/i386.c (ix86_option_override_internal): Set
11857         ix86_stack_protector_guard.
11858         * config/i386/i386.md (stack_protect_set): Enable for
11859         TARGET_SSP_TLS_GUARD only.
11860         (stack_protect_set_<mode>): Ditto.
11861         (stack_protect_test): Ditto.
11862         (stack_protect_test_<mode>): Ditto.
11863         * doc/invoke.texi (i386 Option): Document.
11865 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
11867         PR target/56890
11868         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
11869         (S_MODES): Set H_MODE bit.
11870         (SF_MODES): Set only S_MODE and SF_MODE bits.
11871         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
11872         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
11873         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
11874         <MODE_FLOAT>: Likewise.
11876 2013-04-15  Joey Ye  <joey.ye@arm.com>
11878         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
11880 2013-04-15  Joey Ye  <joey.ye@arm.com>
11882         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
11883         for real far jump.
11884         (thumb_far_jump_used_p): Count instruction size and set
11885         far_jump_used.
11887 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
11889         * reorg.c (fill_simple_delay_slots): Reindent block of code.
11890         * resource.c (mark_target_live_regs): Reformat conditional block.
11892 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11894         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
11895         notes, they are emitted only just before final.
11896         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
11898 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11900         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
11901         * cfgrtl.c (delete_insn): Call it here instead.
11902         * lra-spills.c (lra_final_code_change): Use delete_insn.
11903         * haifa-sched.c (sched_remove_insn): Likewise.
11904         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
11905         returning to the nop pool.
11906         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
11907         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
11909 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11911         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
11912         * doc/tm.texi: Regenerated.
11914 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
11916         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
11917         QImode checks.
11919 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11921         * df-core.c (df_find_def): Compare register numbers.
11922         (df_find_use): Likewise.
11924 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
11926         PR target/56903
11927         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11928         lra_in_progress for return.
11930 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11932         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11933         define_insn into define_insn_and_split and emit movsicc patterns.
11935 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11937         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11939 2013-04-12  Richard Biener  <rguenther@suse.de>
11941         * tree-pass.h (TODO_do_not_ggc_collect): New.
11942         * passes.c (execute_one_ipa_transform_pass): Honor
11943         TODO_do_not_ggc_collect.
11944         (execute_one_pass): Likewise.
11946         Revert
11947         2013-04-10  Richard Biener  <rguenther@suse.de>
11949         * passes.c (init_optimization_passes): Remove reload pass.
11950         * ira.c (do_reload): Merge into ...
11951         (ira): ... this.
11952         (rest_of_handle_reload): Remove.
11953         (pass_reload): Likewise.
11954         * config/i386/i386.c (ix86_option_override): Refer to ira instead
11955         of reload for vzeroupper pass placement.
11957 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
11959         PR tree-optimization/56918
11960         PR tree-optimization/56920
11961         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11962         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
11963         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
11964         use wide_mul_with_sign method.
11966 2013-04-12  Richard Biener  <rguenther@suse.de>
11968         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11969         not be considered a gimple constant.
11971 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
11973         * fold-const.c (const_binop): Handle vector shifts by a scalar.
11974         (fold_binary_loc): Call const_binop also for mixed vector-scalar
11975         operations.
11977 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11978             Jakub Jelinek  <jakub@redhat.com>
11980         * opts.c: Include diagnostic-color.h.
11981         (common_handle_option): Handle OPT_fdiagnostics_color_.
11982         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11983         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11984         (diagnostic-color.o): New.
11985         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11986         (diagnostic_color_rule): New enum.
11987         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11988         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11989         the location string.
11990         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11991         either NULL, or color kind.
11992         * diagnostic-color.c: New file.
11993         * diagnostic-color.h: New file.
11994         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11995         arguments.
11996         * doc/invoke.texi (-fdiagnostics-color): Document.
11997         * pretty-print.h (pp_show_color): Define.
11998         (struct pretty_print_info): Add show_color field.
11999         * diagnostic.c: Include diagnostic-color.h.
12000         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12001         macros.  Colorize error:, warning: etc. strings and also the location
12002         string.
12003         (diagnostic_show_locus): Colorize the caret line.
12004         * pretty-print.c: Include diagnostic-color.h.
12005         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
12006         inside of %< %> quotes or quoted through q format modifier.
12008 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12010         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12012 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12014         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12015         code in CC_NZ mode.
12016         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12017         pattern.
12019 2013-04-11  Marek Polacek  <polacek@redhat.com>
12021         PR tree-optimization/48184
12022         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12024 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
12026         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12027         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12028         (skip_simple_arithmetic): Tidy up.
12029         * tree.h (skip_simple_constant_arithmetic): Declare.
12031 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12033         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12035 2013-04-11  Richard Biener  <rguenther@suse.de>
12037         * tree-vect-loop.c (get_initial_def_for_induction): Properly
12038         generate vector constants.
12040 2013-04-11  Richard Biener  <rguenther@suse.de>
12042         PR tree-optimization/56878
12043         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12044         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12045         New function.
12046         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12047         Prefer to align the DR with the most invariant base address.
12049 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12051         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12052         comment.
12054 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
12056         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12057         floating-point vector comparisons against 0.
12059 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
12061         PR tree-optimization/56899
12062         * fold-const.c (extract_muldiv_1): Apply distributive law
12063         only if TYPE_OVERFLOW_WRAPS (ctype).
12065 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
12067         PR target/56124
12068         * ira-costs.c (scan_one_insn): Check whether the source rtx of
12069         loading has side effect.
12071 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
12073         * config/sparc/sparc.c: Include tree-pass.h.
12074         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12075         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
12076         head of file.  Change return type.  Split off gate function.
12077         (sparc_gate_work_around_errata): New function.
12078         (pass_work_around_errata): New pass definition.
12079         (insert_pass_work_around_errata) New pass insert definition to
12080         insert pass_work_around_errata just after delayed-branch scheduling.
12081         (sparc_option_override): Insert the pass.
12082         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12084 2013-04-10  David S. Miller  <davem@davemloft.net>
12086         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12087         or -mcpu=hypersparc.
12089         * target.def (cstore_mode): New hook.
12090         * target.h: Include insn-codes.h
12091         * targhooks.c: Likewise.
12092         (default_cstore_mode): New function.
12093         * targhooks.h: Declare it.
12094         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12095         * doc/tm.texi: Rebuild.
12096         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12097         target hook, rather than inspecting the insn_data.
12098         * config/sparc/sparc.c (sparc_cstore_mode): New function.
12099         (TARGET_CSTORE_MODE): Redefine.
12100         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12101         result patterns.
12102         * config/sparc/predicates.md (cstore_result_operand): New special
12103         predicate.
12104         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12105         Use it for operand 0.
12106         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12107         (*snesi_special): Likewise.
12108         (*snesi_zero): Likewise.
12109         (*seqsi_zero): Likewise.
12110         (*sltu_insn): Likewise.
12111         (*sgeu_insn): Likewise.
12112         (*seqdi_special): Make operand 0 and comparison operation be of
12113         DImode.
12114         (*snedi_special): Likewise.
12115         (*snedi_special_vis3): Likewise.
12116         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12117         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12118         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12119         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12120         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12121         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12122         (*sltu_extend_sp64): Likewise.
12123         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12124         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12125         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12126         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12127         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12129 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
12131         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12132         (aarch64_start_file): Use the new function.
12134 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12135             Jason Merrill  <jason@redhat.com>
12137         * common.opt: Add -gdwarf.
12138         * opts.c (common_handle_option): Handle it.
12139         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12141 2013-04-10  Richard Biener  <rguenther@suse.de>
12143         * passes.c (execute_todo): Do not call ggc_collect conditional here.
12144         (execute_one_ipa_transform_pass): But unconditionally here.
12145         (execute_one_pass): And here.
12146         (init_optimization_passes): Remove reload pass.
12147         * tree-pass.h (TODO_ggc_collect): Remove.
12148         (pass_reload): Likewise.
12149         * ira.c (do_reload): Merge into ...
12150         (ira): ... this.
12151         (rest_of_handle_reload): Remove.
12152         (pass_reload): Likewise.
12153         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12154         of reload for vzeroupper pass placement.
12155         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12156         and todo_flags_finish of all passes.
12158 2013-04-10  Richard Biener  <rguenther@suse.de>
12160         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12161         first_const_oprnd field, rename first_def_type to first_op_type.
12162         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12163         (vect_get_and_check_slp_defs): Always use the type of the
12164         operand.  Allow mixed vect_external_def, vect_constant_def types.
12165         (vect_get_constant_vectors): Handle mixed vect_external_def,
12166         vect_constant_def types.
12168 2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
12170         PR tree-optimization/55524
12171         * tree-ssa-math-opts.c
12172         (convert_mult_to_fma): Don't use an fms construct
12173         when we don't have an fms operation, but fnma, and it looks
12174         likely that we'll be able to use the latter.
12176 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
12178         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12179         function.
12180         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12181         inline fail caused by overwritable functions.
12183 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
12185         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12186         unnecessary bits in the constant power of two case.
12188 2013-04-10  Richard Biener  <rguenther@suse.de>
12190         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12191         broken code swapping operands.
12192         (vect_build_slp_tree): Do not compute load permutations here.
12193         (vect_analyze_slp_instance): Compute load permutations here,
12194         after building the SLP tree.
12196 2013-04-09  Christian Bruel  <christian.bruel@st.com>
12198         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
12199         of next/prev_real_insn.
12201 2013-04-09  Jan Hubicka  <jh@suse.cz>
12203         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
12204         Drop aliased parameter.
12205         (function_and_variable_visibility): Do not handle alias pairs.
12206         * cgraph.c (varpool_externally_visible_p): Update prototype.
12207         * varpool.c (varpool_add_new_variable): Update.
12209 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12211         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
12213 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
12215         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
12217         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
12219 2013-04-09  Marek Polacek  <polacek@redhat.com>
12221         PR tree-optimization/48762
12222         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
12224 2013-04-09  Richard Biener  <rguenther@suse.de>
12226         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
12227         dealing with cost.
12228         (vect_build_slp_tree): Likewise.
12229         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
12230         calculating the cost of a SLP instance.
12231         (vect_analyze_slp_instance): Use it from here, after building
12232         the SLP tree.
12234 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12236         PR middle-end/56883
12237         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
12238         expand_omp_for_static_chunk): Use simple_p = true in
12239         force_gimple_operand_gsi calls when assigning to addressable decls.
12241 2013-04-09  Jeff Law  <law@redhat.com>
12243         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
12244         when the boolean was created by converting a wider object which
12245         had a boolean range.
12247 2013-04-09  Richard Biener  <rguenther@suse.de>
12249         * tree-vectorizer.h (slp_void_p): Remove.
12250         (slp_tree): Typedef before _slp_tree declaration.
12251         (struct _slp_tree): Use a vector of slp_tree as children.
12252         (vect_get_place_in_interleaving_chain): Remove.
12253         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
12254         Move ...
12255         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
12256         and make static.
12257         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
12258         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
12259         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
12260         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
12261         Use slp_node instead of slp_void_p and adjust.
12263 2013-04-09  Richard Biener  <rguenther@suse.de>
12265         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
12266         work that is not necessary.
12268 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12270         PR tree-optimization/56854
12271         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
12272         forward into clobber stmts if it would change MEM_REF lhs into
12273         non-MEM_REF.
12275 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
12277         * tree.c (type_hash_lookup, type_hash_add): Make static.
12278         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
12280 2013-04-09  Richard Biener  <rguenther@suse.de>
12282         * tree.h (unsave_expr_now): Remove.
12283         * tree-inline.c (mark_local_for_remap_r): Remove.
12284         (unsave_expr_1): Likewise.
12285         (unsave_r): Likewise.
12286         (unsave_expr_now): Likewise.
12287         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
12288         (propagate_tree_value): Likewise.
12290 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
12292         * doc/rtl.texi (sequence): Rewrite documentation to match the
12293         current use of SEQUENCE rtl objects.
12294         * rtl.def (SEQUENCE): Likewise.
12296         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
12297         Update documentation.
12298         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
12299         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
12301         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
12303 2013-04-08  Teresa Johnson  <tejohnson@google.com>
12305         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
12306         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
12307         methods.
12308         (estimate_edge_size_and_time): Add comment to suggest using rounding
12309         methods.
12310         (estimate_node_size_and_time): Ditto.
12311         (remap_edge_change_prob): Use helper rounding divide methods.
12312         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
12313         (gimple_mod_pow2_value_transform): Ditto.
12314         (gimple_mod_subtract_transform): Ditto.
12315         (gimple_ic_transform): Ditto.
12316         (gimple_stringops_transform): Ditto.
12317         * stmt.c (conditional_probability): Ditto.
12318         (emit_case_dispatch_table): Ditto.
12319         * lto-cgraph.c (merge_profile_summaries): Ditto.
12320         * tree-optimize.c (execute_fixup_cfg): Ditto.
12321         * cfgcleanup.c (try_forward_edges): Ditto.
12322         * cfgloopmanip.c (scale_loop_profile): Ditto.
12323         (loopify): Ditto.
12324         (duplicate_loop_to_header_edge): Ditto.
12325         (lv_adjust_loop_entry_edge): Ditto.
12326         * tree-vect-loop.c (vect_transform_loop): Ditto.
12327         * profile.c (compute_branch_probabilities): Ditto.
12328         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
12329         * lto-streamer-in.c (input_cfg): Ditto.
12330         * gimple-streamer-in.c (input_bb): Ditto.
12331         * ipa-cp.c (update_profiling_info): Ditto.
12332         (update_specialized_profile): Ditto.
12333         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
12334         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
12335         rounding methods.
12336         * sched-rgn.c (compute_dom_prob_ps): Ditto.
12337         (compute_trg_info): Ditto.
12338         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
12339         (purge_dead_edges): Ditto.
12340         * loop-unswitch.c (unswitch_loop): Ditto.
12341         * cgraphclones.c (cgraph_clone_edge): Ditto.
12342         (cgraph_clone_node): Ditto.
12343         * tree-inline.c (copy_bb): Ditto.
12344         (copy_edges_for_bb): Ditto.
12345         (initialize_cfun): Ditto.
12346         (copy_cfg_body): Ditto.
12347         (expand_call_inline): Ditto.
12349 2013-04-08  Kai Tietz  <ktietz@redhat.com>
12351         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
12352         TARGET_CYGWIN64 by TARGET_64BIT.
12354 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
12356         * config/epiphany/epiphany.md (GPR_1): New constant.
12357         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
12358         * config/epiphany/epiphany.c (gen_compare_reg):
12359         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
12360         is already in place.
12361         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
12362         Don't require being called during rtl expansion; If y operlaps r0,
12363         return 0.
12364         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
12365         (epiphany_expand_epilogue): Likewise.
12367         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
12368         Don't use CC_FPmode for ORDERED / UNORDERED.
12369         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
12371         * config/epiphany/constraints.md (CnL): New constraint.
12372         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
12373         * config/epiphany/predicates.md (add_operand): Allow 1024.
12375         * config/epiphany/epiphany.md (logical_op): New code iterator.
12376         (op_mnc): New code attribute.
12377         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
12378         (mov_f+1, mov_f+2): New peephole2 patterns.
12380         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
12381         (cstoresi4): Also allow re-use of zero result when doing a NE
12382         comparison to a non-zero operand.
12383         Use (clobber (scratch)) for first insn if the gpr output is not needed.
12385         * config/epiphany/epiphany.md (<insn_opname>v2si3):
12386         Use gen_addsi3_i / gen_subsi3_i.
12388 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
12390         PR c++/34949
12391         PR c++/50243
12392         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
12393         contain anything but clobbers, at most one __builtin_stack_restore,
12394         optionally debug stmts and final resx, and if it has at least one
12395         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
12396         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
12397         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
12398         which isn't defaut definition, remove them.
12399         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
12400         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
12401         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
12402         with MEM_REF LHS with SSA_NAME address.
12404 2013-04-08  Jeff Law  <law@redhat.com>
12406         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
12408 2013-04-08  Richard Biener  <rguenther@suse.de>
12410         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
12411         extra newline.
12412         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
12413         determined vector type.
12414         (vect_analyze_data_refs): Likewise.
12415         (vect_get_new_vect_var): Adjust.
12416         (vect_create_destination_var): Preserve SSA name versions.
12417         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
12418         not dump anything here.
12420 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
12422         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
12423         Add member lr_slot_known.
12424         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
12425         if necessary.
12426         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
12427         Remove code that sets lr_slot_offset according to what a previous
12428         version of epiphany_emit_save_restore used to do.
12429         (epiphany_emit_save_restore): When doing an lr save or restore,
12430         set/verify lr_slot_known and lr_slot_offset.
12432 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
12434         PR target/54338
12435         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
12436         in ALL_REGS.
12438 2013-04-08  Richard Biener  <rguenther@suse.de>
12440         * alias.c (find_base_term): Fix thinko in previous change.
12442 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
12444         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
12445         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
12446         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
12447         if possible to compute val.
12448         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
12449         For QImode integers don't require anything about precision.  Use
12450         const_with_all_bytes_same to find out if the constant doesn't have
12451         repeated bytes in it.
12453 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12455         * config/s390/s390.c (s390_expand_insv): Only accept insertions
12456         within mode size.
12458 2013-04-08  Marek Polacek  <polacek@redhat.com>
12460         PR rtl-optimization/48182
12461         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
12462         value to 1.
12464 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
12466         PR target/55487
12467         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
12468         nuses, make sure we have a label.
12470 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12472         PR target/56843
12473         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
12474         (rs6000_emit_swdiv_low_precision): Remove.
12475         (rs6000_emit_swdiv): Rewrite to handle between one and four
12476         iterations of Newton-Raphson generally; modify required number of
12477         iterations for some cases.
12478         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
12480 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
12482         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
12483         set-but-unused variable.
12485         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
12486         basic blocks of released function bodies garbage-collectable.
12488         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
12489         (struct rtl_opt_pass): Add TODO_df_finish.
12491         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
12493 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12495         * config/arm/constraints.md (q): New constraint.
12496         * config/arm/ldrdstrd.md: New file.
12497         * config/arm/arm.md (ldrdstrd.md) New include.
12498         (arm_movdi): Use "q" instead of "r" constraint
12499         for double-word memory access.
12500         (movdf_soft_insn): Likewise.
12501         * config/arm/vfp.md (movdi_vfp): Likewise.
12502         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
12503         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
12504         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
12505         (mem_ok_for_ldrd_strd): Likewise.
12506         (output_move_double): Update assertion.
12508 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12510         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
12512 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12514         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
12515         define_insn_and_split.
12516         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
12518 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12520         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
12521         define_insn_and_split.
12522         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
12523         (shiftsi3_compare): New pattern.
12524         (rrx): New pattern.
12525         * config/arm/unspecs.md (UNSPEC_RRX): New.
12527 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12529         * config/arm/arm.md (negdi_extendsidi): New pattern.
12530         (negdi_zero_extendsidi): Likewise.
12532 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12534         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
12535         define_insn_and_split.
12536         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
12537         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
12539 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12541         * config/arm/arm.md (arm_subdi3): Convert define_insn into
12542         define_insn_and_split.
12543         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
12544         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
12546 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12548         * config/arm/arm.md (subsi3_carryin): New pattern.
12549         (subsi3_carryin_const): Likewise.
12550         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
12551         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
12553 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12555         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
12557 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12559         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
12560         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
12562 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12564         * config/arm/arm.c (arm_expand_builtin): Change fcode
12565         type to unsigned int.
12567 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12569         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
12571 2013-04-04  Ian Lance Taylor  <iant@google.com>
12573         * doc/standards.texi (Standards): The Go frontend supports the Go 1
12574         language standard.
12576 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
12578         PR middle-end/56729
12579         * df-scan.c (df_insn_delete): Disable failing assert.
12581 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12583         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
12584         New function prototype.
12585         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
12586         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
12587         (arm_builtin_vectorized_function): New function.
12589 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12591         * config/arm/arm_neon_builtins.def: New file.
12592         * config/arm/arm.c (neon_builtin_data): Move contents to
12593         arm_neon_builtins.def.
12594         (enum arm_builtins): Include neon builtin definitions.
12595         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
12596         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
12598 2013-04-04  Marek Polacek  <polacek@redhat.com>
12600         PR tree-optimization/48186
12601         * predict.c (maybe_hot_frequency_p): Return false if
12602         HOT_BB_FREQUENCY_FRACTION is 0.
12603         (cgraph_maybe_hot_edge_p): Likewise.
12605 2013-04-04  Richard Biener  <rguenther@suse.de>
12607         PR tree-optimization/56826
12608         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
12609         more accurately.
12611 2013-04-04  Richard Biener  <rguenther@suse.de>
12613         PR tree-optimization/56213
12614         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
12615         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
12617 2013-04-04  Richard Biener  <rguenther@suse.de>
12619         PR tree-optimization/56837
12620         * tree-loop-distribution.c (classify_partition): For non-zero
12621         values require that the value has the same precision as its
12622         mode to be useful as memset value.
12624 2013-04-03  Nick Clifton  <nickc@redhat.com>
12626         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
12627         (fmssf4): Use fmsf.s on E3V5 architectures.
12628         (fnmasf4): Use fnmaf.s on E3V5 architectures.
12629         (fnmssf4): Use fnmsf.s on E3V5 architectures.
12631 2013-04-03  Jeff Law  <law@redhat.com>
12633         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
12634         (lra-eliminations.o): Likewise.
12636 2013-04-03  Teresa Johnson  <tejohnson@google.com>
12638         * gcov-io.c (compute_working_sets): Moved most of body of old
12639         compute_working_sets here from profile.c.
12640         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
12641         (gcov_working_set_t): Moved typedef here from basic-block.h
12642         (compute_working_set): Declare.
12643         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
12644         (get_working_sets): Renamed from compute_working_set,
12645         replace most of body with call to new compute_working_sets.
12646         (get_exec_counts): Replace call to compute_working_sets
12647         to get_working_sets.
12648         * profile.h (get_working_sets): Renamed from compute_working_set.
12649         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
12650         to get_working_sets.
12651         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
12652         * gcov-dump.c (dump_working_sets): New function.
12654 2013-04-03  Kenneth Zadeck <zadeck@naturalbridge.com>
12656         * hwint.c (sext_hwi, zext_hwi): New functions.
12657         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
12658         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
12659         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
12660         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
12661         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
12662         (sext_hwi, zext_hwi): New functions.
12664 2013-04-03  Jeff Law  <law@redhat.com>
12666         PR tree-optimization/56799
12667         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
12668         back test for widening conversion erroneously dropped in prior change.
12670 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12672         PR target/56809
12673         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
12674         instead of next_real_insn.
12676 2013-04-03  Marek Polacek  <polacek@redhat.com>
12678         PR sanitizer/55702
12679         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
12681 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12683         PR target/56809
12684         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
12685         next_real_insn.
12686         (thumb1_output_casesi): Likewise.
12687         (thumb2_output_casesi): Likewise.
12689 2013-04-03  Richard Biener  <rguenther@suse.de>
12691         PR tree-optimization/56817
12692         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
12693         Split out ...
12694         (tree_unroll_loops_completely_1): ... new function to manually
12695         walk the loop tree, properly defering outer loops of unrolled
12696         loops to later iterations.
12698 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12700         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
12701         (vectorizable_load): Likewise.
12702         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
12703         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
12705 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12707         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
12708         BIT_FIELD_REF.
12710 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12712         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
12714 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
12716         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
12718 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12720         PR tree-optimization/56790
12721         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
12722         folding.
12724 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12726         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
12727         Handle VEC_MERGE.
12728         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
12729         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
12730         equal arguments.
12732 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
12734         PR c/19449
12735         * tree.h (force_folding_builtin_constant_p): New decl.
12736         * builtins.c (force_folding_builtin_constant_p): New variable.
12737         (fold_builtin_constant_p): Fold immediately also if
12738         force_folding_builtin_constant_p.
12740 2013-04-03  Richard Biener  <rguenther@suse.de>
12742         PR tree-optimization/56812
12743         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
12744         DRs of the same interleaving chain are independent.
12746 2013-04-02  Jason Merrill  <jason@redhat.com>
12748         * gdbinit.in (pbb): Use debug fn.
12750 2013-04-02  Lawrence Crowl  <crowl@google.com>
12752         * sese.h (struct ivtype_map_elt_s): Remove unused.
12753         (extern debug_ivtype_map): Remove unused.
12754         (extern eq_ivtype_map_elts): Remove unused.
12755         * sese.c (debug_ivtype_map): Removed unused.
12756         (debug_ivtype_map_1): Removed unused.
12757         (debug_ivtype_elt): Remove unused.
12758         (eq_ivtype_map_elts): Remove unused.
12761 2013-04-02  Kai Tietz  <ktietz@redhat.com>
12763         PR target/52790
12764         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
12765         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
12766         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
12767         function.
12768         (legitimize_pe_coff_symbol): Likewise.
12769         (is_imported_p): New helper-function.
12770         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
12771         for Windows x64 targets.
12772         (ix86_expand_prologue): Optimize for pe-coff targets.
12773         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
12774         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
12775         medium/large code-model.
12776         (legitimize_pic_address): Likewise.
12777         (legitimize_tls_address): Likewise.
12778         (ix86_expand_call): Likewise.
12779         (x86_output_mi_thunk): Likewise.
12780         (get_dllimport_decl): Add new beimport argument.
12781         (construct_plt_address): Don't assert for x64 pe-coff targets.
12782         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
12783         targets.
12784         (SYMBOL_FLAG_STUBVAR): New macro.
12785         (SYMBOL_REF_STUBVAR_P): Likewise.
12786         * config/i386/winnt.c (stub_list): New structure.
12787         (stub_head): New local variable.
12788         (i386_pe_record_stub): New function.
12789         (i386_pe_file_end): Emit refptr-stubs.
12791 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
12793         PR rtl-optimization/56745
12794         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
12795         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
12797         PR c++/34949
12798         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
12799         and both of them are MEM_REFs, just compare first argument for
12800         equality and attempt to deal even with differing offsets.
12802         PR c++/34949
12803         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
12804         of gimple_clobber_p to be MEM_REF.
12805         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
12806         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
12807         after gimplification.
12808         * asan.c (get_mem_ref_of_assignment): Don't instrument
12809         gimple_clobber_p stmts.
12810         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
12811         gimple_clobber_p stmt if they have MEM_REF lhs and
12812         are dead because of another gimple_clobber_p stmt.
12813         * tree-ssa-live.c (clear_unused_block_pointer): Treat
12814         gimple_clobber_p stmts like debug stmts.
12815         (remove_unused_locals): Remove clobbers with MEM_REF lhs
12816         that refer to unused VAR_DECLs or uninitialized values.
12817         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
12818         gimple_clobber_p stmts if they refer to removed parameters.
12819         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
12820         formatting.
12822 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
12824         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
12825         using SWI48 mode attribute.
12827 2013-04-02  Wei Mi  <wmi@google.com>
12829         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
12830         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
12831         *<rotate_insn><mode>3_mask in i386.md.
12833 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12835         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
12837 2013-04-02  Richard Biener  <rguenther@suse.de>
12839         PR tree-optimization/56778
12840         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
12841         Runtime alias tests are not supported for gather loads.
12842         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
12843         stmts referenced from SSA operands before updating SSA form.
12845 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
12846             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12848         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
12849         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
12850         * config/arm/cortex-a53.md: New file.
12851         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
12852         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
12853         * config/arm/arm.c (arm_issue_rate): Likewise.
12854         * config/arm/arm-tune.md: Regenerate
12855         * config/arm/arm-tables.opt: Regenerate.
12856         * config/arm/arm-cores.def: Add cortex-a53.
12858 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
12860         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
12861         non-static link.
12863 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12865         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
12866         scalar load/store operations using B/H registers.
12867         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
12869 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12871         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
12872         scalar move.
12873         * config/aarch64/aarch64.c
12874         (aarch64_simd_scalar_immediate_valid_for_move): New.
12875         * config/aarch64/aarch64-protos.h
12876         (aarch64_simd_scalar_immediate_valid_for_move): New.
12877         * config/aarch64/constraints.md (Dh, Dq): New.
12878         * config/aarch64/iterators.md (hq): New.
12880 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
12882         * reorg.c (get_branch_condition): Deal with conditional returns.
12883         (fill_simple_delay_slots): Remove dead code dealing with jumps.
12885 2013-04-01  Wei Mi  <wmi@google.com>
12887         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
12888         Truncate operand 2 using %b asm operand modifier.
12889         (*<shift_insn><mode>3_mask): Ditto.
12890         (*<rotate_insn><mode>3_mask): Ditto.
12892 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
12894         PR middle-end/56798
12895         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
12897 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
12899         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
12900         of next_real_insn.
12901         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
12903 2013-03-30  Lawrence Crowl  <crowl@google.com>
12905         * dse.c (clear_alias_sets): Remove never set.
12906         (disqualified_clear_alias_sets): Remove never set.
12907         (clear_alias_mode_pool): Remove never set.
12908         (dse_step0): Remove condition that is never true.
12909         (canon_address): Remove condition that is never true.
12910         (dse_step7): Remove condition that is never true.
12911         (rest_of_handle_dse): Remove condition that is never true.
12912         (rest_of_handle_dse::did_global): Remove never read from above.
12913         (dse_step2_spill): Remove never called from above.
12914         (dse_step5_spill): Remove never called from above.
12916 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
12918         * doc/md.texi (Standard Names) <casesi>: Update documentation for
12919         JUMP_TABLE_DATA changes.
12920         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
12921         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
12922         (Insns) <jump_table_data>: New entry.
12923         * doc/tm.texi: Regenerate.
12925         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
12927         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12928         for table jump at the end of a basic block using tablejump_p.
12929         * targhooks.c (default_invalid_within_doloop): Likewise.
12930         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12931         target hook implementation that is identical to the default hook.
12932         (rs6000_invalid_within_doloop): Remove.
12934         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12935         unused variable from tablejump_p call.
12937         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12938         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12939         (INSN_DELETED_P): Likewise.
12940         (emit_jump_table_data): New prototype.
12941         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12942         after 4th as unused.
12943         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12944         * sched-vis.c (print_insn): Likewise.
12945         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12946         insn for compatibility with back ends that use next_active_insn to
12947         identify jump table data.
12948         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12949         (remove_insn): Likewise.
12950         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12951         to be emitted.
12952         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12953         (emit_jump_table_data): New function.
12955         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12956         basic block, a JUMP_TABLE_DATA never is.
12957         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12958         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12959         off from code handling real insns.
12960         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12961         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12962         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12963         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12964         is not a NONDEBUG_INSN_P.
12965         * ira-costs.c (scan_one_insn): Likewise.
12966         * jump.c (mark_all_labels): Likewise.
12967         (mark_jump_label_1): Likewise.
12968         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12969         * lra.c (get_insn_freq): Expect all insns reaching here to be in
12970         a basic block.
12971         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12972         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12973         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12974         JUMP_TABLE_DATA_P insns.
12975         (calculate_elim_costs_all_insns): Likewise.
12976         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12977         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12978         (delete_output_reload): Code style fixups.
12979         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12980         insn flags on this non-insn.
12981         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12982         as scheduling barriers, for pre-change compatibility.
12983         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12984         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12986         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12987         redundant JUMP_TABLE_DATA_P test.
12988         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12989         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12990         (frv_for_each_packet): Likewise.
12991         * config/i386/i386.c (min_insn_size): Likewise.
12992         (ix86_avoid_jump_mispredicts): Likewise.
12993         * config/m32r/m32r.c (m32r_is_insn): Likewise.
12994         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12995         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12996         (mips16_insn_length): Robustify.
12997         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12998         (mips16_split_long_branches): Likewise.
12999         * config/pa/pa.c (pa_combine_instructions): Likewise.
13000         * config/rs6000/rs6000.c (get_next_active_insn): Treat
13001         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13002         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13003         as contributing to pool range lengths.
13004         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13005         Remove redundant JUMP_TABLE_DATA_P test.
13006         (sh_loop_align): Likewise.
13007         (split_branches): Likewise.
13008         (sh_insn_length_adjustment): Likewise.
13009         * config/spu/spu.c (get_branch_target): Likewise.
13011 2013-03-29  Jan Hubicka  <jh@suse.cz>
13013         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13014         gcov streaming; stream hot bb threshold to ltrans.
13015         * predict.c (get_hot_bb_threshold): Break out from ....
13016         (maybe_hot_count_p): ... here.
13017         (set_hot_bb_threshold): New function.
13018         * lto-section-in.c (lto_section_name): Add profile.
13019         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13020         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13021         and data-streamer.h
13022         (histogram_entry): New structure.
13023         (histogram, histogram_pool): New global vars.
13024         (histogram_hash): New structure.
13025         (histogram_hash::hash): New method.
13026         (histogram_hash::equal): Likewise.
13027         (account_time_size): New function.
13028         (cmp_counts): New function.
13029         (dump_histogram): New function.
13030         (ipa_profile_generate_summary): New function.
13031         (ipa_profile_write_summary): New function.
13032         (ipa_profile_read_summary): New function.
13033         (ipa_profile): Decide on threshold.
13034         (pass_ipa_profile): Add ipa_profile_write_summary and
13035         ipa_profile_read_summary.
13036         * Makefile.in (ipa.o): Update dependencies.
13037         * lto-streamer.h (LTO_section_ipa_profile): New section.
13039 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13041         * tree.h (VAR_P): New.
13043 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
13045         PR lto/56777
13046         * doc/invoke.texi ([-fwhole-program]): Fix typo.
13048 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
13050         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13051         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13052         (control_flow_insn_p): Likewise.
13053         * cfgrtl.c (duplicate_insn_chain): Likewise.
13054         * final.c (get_attr_length_1): Likewise.
13055         (shorten_branches): Likewise.
13056         (final_scan_insn): Likewise.
13057         * function.c (instantiate_virtual_regs): Likewise.
13058         * gcse.c (insert_insn_end_basic_block): Likewise.
13059         * ira-costs.c (scan_one_insn): Likewise.
13060         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13061         * lra.c (check_rtl): Likewise.
13062         * reload1.c (elimination_costs_in_insn): Likewise.
13063         * reorg.c (follow_jumps): Likewise.
13065         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13066         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13067         (thumb_far_jump_used_p): Likewise.
13068         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13069         (workaround_speculation): Likewise.
13070         (add_sched_insns_for_speculation): Likewise.
13071         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13072         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13073         (frv_for_each_packet): Likewise.
13074         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13075         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13076         (final_emit_insn_group_barriers): Likewise.
13077         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13078         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13079         (mips16_insn_length): Likewise.
13080         * config/pa/pa.c (pa_reorg): Likewise.
13081         (pa_combine_instructions): Likewise.
13082         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13083         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13084         (sh_reorg): Likewise.
13085         (split_branches): Likewise.
13086         * config/spu/spu.c (get_branch_target): Likewise.
13088         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13089         JUMP_TABLE_DATA_P.
13091 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
13093         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13094         Fix declaration name.
13096 2013-03-28  Lawrence Crowl  <crowl@google.com>
13098         * graphds.h (struct graph.indicies): Remove unused.
13099         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13100         (SCOP_ORIGINAL_PDDRS): Remove unused.
13101         * sese.h (extern insert_loop_close_phis): Removed unused.
13102         (extern insert_guard_phis): Removed unused.
13103         (extern ivtype_map_elt_info): Removed unused.
13104         (new_ivtype_map_elt): Removed unused.
13105         * sese.c (ivtype_map_elt_info): Removed unused.
13107 2013-03-28  Lawrence Crowl  <crowl@google.com>
13109         * Makefile.in: Add several missing include dependences.
13110         (DUMPFILE_H): New.
13111         (test-dump.o): New.  This object is not added to any executable,
13112         but is present for ad-hoc testing.
13113         * bitmap.c
13114         (debug (const bitmap_head_def &)): New.
13115         (debug (const bitmap_head_def *)): New.
13116         * bitmap.h
13117         (extern debug (const bitmap_head_def &)): New.
13118         (extern debug (const bitmap_head_def *)): New.
13119         * cfg.c
13120         (debug (edge_def &)): New.
13121         (debug (edge_def *)): New.
13122         * cfghooks.c
13123         (debug (basic_block_def &)): New.
13124         (debug (basic_block_def *)): New.
13125         * dumpfile.h
13126         (dump_node (const_tree, int, FILE *)): Correct source file.
13127         * dwarf2out.c
13128         (debug (die_struct &)): New.
13129         (debug (die_struct *)): New.
13130         * dwarf2out.h
13131         (extern debug (die_struct &)): New.
13132         (extern debug (die_struct *)): New.
13133         * gimple-pretty-print.c
13134         (debug (gimple_statement_d &)): New.
13135         (debug (gimple_statement_d *)): New.
13136         * gimple-pretty-print.h
13137         (extern debug (gimple_statement_d &)): New.
13138         (extern debug (gimple_statement_d *)): New.
13139         * ira-build.c
13140         (debug (ira_allocno_copy &)): New.
13141         (debug (ira_allocno_copy *)): New.
13142         (debug (ira_allocno &)): New.
13143         (debug (ira_allocno *)): New.
13144         * ira-int.h
13145         (extern debug (ira_allocno_copy &)): New.
13146         (extern debug (ira_allocno_copy *)): New.
13147         (extern debug (ira_allocno &)): New.
13148         (extern debug (ira_allocno *)): New.
13149         * ira-lives.c
13150         (debug (live_range &)): New.
13151         (debug (live_range *)): New.
13152         * lra-int.h
13153         (debug (lra_live_range &)): New.
13154         (debug (lra_live_range *)): New.
13155         * lra-lives.c
13156         (debug (lra_live_range &)): New.
13157         (debug (lra_live_range *)): New.
13158         * omega.c
13159         (debug (omega_pb_d &)): New.
13160         (debug (omega_pb_d *)): New.
13161         * omega.h
13162         (extern debug (omega_pb_d &)): New.
13163         (extern debug (omega_pb_d *)): New.
13164         * print-rtl.c
13165         (debug (const rtx_def &)): New.
13166         (debug (const rtx_def *)): New.
13167         * print-tree.c
13168         (debug_tree (tree): Move within file.
13169         (debug_raw (const tree_node &)): New.
13170         (debug_raw (const tree_node *)): New.
13171         (dump_tree_via_hooks (const tree_node *, int)): New.
13172         (debug (const tree_node &)): New.
13173         (debug (const tree_node *)): New.
13174         (debug_verbose (const tree_node &)): New.
13175         (debug_verbose (const tree_node *)): New.
13176         (debug_head (const tree_node &)): New.
13177         (debug_head (const tree_node *)): New.
13178         (debug_body (const tree_node &)): New.
13179         (debug_body (const tree_node *)): New.
13180         (debug_vec_tree (tree): Move and reimplement in terms of dump.
13181         (debug (vec<tree, va_gc> &)): New.
13182         (debug (vec<tree, va_gc> *)): New.
13183         * rtl.h
13184         (extern debug (const rtx_def &)): New.
13185         (extern debug (const rtx_def *)): New.
13186         * sbitmap.c
13187         (debug_raw (simple_bitmap_def &)): New.
13188         (debug_raw (simple_bitmap_def *)): New.
13189         (debug (simple_bitmap_def &)): New.
13190         (debug (simple_bitmap_def *)): New.
13191         * sbitmap.h
13192         (extern debug (simple_bitmap_def &)): New.
13193         (extern debug (simple_bitmap_def *)): New.
13194         (extern debug_raw (simple_bitmap_def &)): New.
13195         (extern debug_raw (simple_bitmap_def *)): New.
13196         * sel-sched-dump.c
13197         (debug (vinsn_def &)): New.
13198         (debug (vinsn_def *)): New.
13199         (debug_verbose (vinsn_def &)): New.
13200         (debug_verbose (vinsn_def *)): New.
13201         (debug (expr_def &)): New.
13202         (debug (expr_def *)): New.
13203         (debug_verbose (expr_def &)): New.
13204         (debug_verbose (expr_def *)): New.
13205         (debug (vec<rtx> &)): New.
13206         (debug (vec<rtx> *)): New.
13207         * sel-sched-dump.h
13208         (extern debug (vinsn_def &)): New.
13209         (extern debug (vinsn_def *)): New.
13210         (extern debug_verbose (vinsn_def &)): New.
13211         (extern debug_verbose (vinsn_def *)): New.
13212         (extern debug (expr_def &)): New.
13213         (extern debug (expr_def *)): New.
13214         (extern debug_verbose (expr_def &)): New.
13215         (extern debug_verbose (expr_def *)): New.
13216         (extern debug (vec<rtx> &)): New.
13217         (extern debug (vec<rtx> *)): New.
13218         * sel-sched-ir.h
13219         (_list_iter_cond_expr): Make inline instead of static.
13220         * sreal.c
13221         (debug (sreal &)): New.
13222         (debug (sreal *)): New.
13223         * sreal.h
13224         (extern debug (sreal &)): New.
13225         (extern debug (sreal *)): New.
13226         * tree.h
13227         (extern debug_raw (const tree_node &)): New.
13228         (extern debug_raw (const tree_node *)): New.
13229         (extern debug (const tree_node &)): New.
13230         (extern debug (const tree_node *)): New.
13231         (extern debug_verbose (const tree_node &)): New.
13232         (extern debug_verbose (const tree_node *)): New.
13233         (extern debug_head (const tree_node &)): New.
13234         (extern debug_head (const tree_node *)): New.
13235         (extern debug_body (const tree_node &)): New.
13236         (extern debug_body (const tree_node *)): New.
13237         (extern debug (vec<tree, va_gc> &)): New.
13238         (extern debug (vec<tree, va_gc> *)): New.
13239         * tree-cfg.c
13240         (debug (struct loop &)): New.
13241         (debug (struct loop *)): New.
13242         (debug_verbose (struct loop &)): New.
13243         (debug_verbose (struct loop *)): New.
13244         * tree-dump.c: Add header dependence.
13245         * tree-flow.h
13246         (extern debug (struct loop &)): New.
13247         (extern debug (struct loop *)): New.
13248         (extern debug_verbose (struct loop &)): New.
13249         (extern debug_verbose (struct loop *)): New.
13250         * tree-data-ref.c
13251         (debug (data_reference &)): New.
13252         (debug (data_reference *)): New.
13253         (debug (vec<data_reference_p> &)): New.
13254         (debug (vec<data_reference_p> *)): New.
13255         (debug (vec<ddr_p> &)): New.
13256         (debug (vec<ddr_p> *)): New.
13257         * tree-data-ref.h
13258         (extern debug (data_reference &)): New.
13259         (extern debug (data_reference *)): New.
13260         (extern debug (vec<data_reference_p> &)): New.
13261         (extern debug (vec<data_reference_p> *)): New.
13262         (extern debug (vec<ddr_p> &)): New.
13263         (extern debug (vec<ddr_p> *)): New.
13264         * tree-ssa-alias.c
13265         (debug (pt_solution &)): New.
13266         (debug (pt_solution *)): New.
13267         * tree-ssa-alias.h
13268         (extern debug (pt_solution &)): New.
13269         (extern debug (pt_solution *)): New.
13270         * tree-ssa-alias.c
13271         (debug (_var_map &)): New.
13272         (debug (_var_map *)): New.
13273         (debug (tree_live_info_d &)): New.
13274         (debug (tree_live_info_d *)): New.
13275         * tree-ssa-alias.h
13276         (extern debug (_var_map &)): New.
13277         (extern debug (_var_map *)): New.
13278         (extern debug (tree_live_info_d &)): New.
13279         (extern debug (tree_live_info_d *)): New.
13281 2013-03-28  Jan Hubicka  <jh@suse.cz>
13283         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
13285 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
13287         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
13288         record only when desired or required.
13290 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
13292         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
13293         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
13295 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13297         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
13298         (*andsi3_compare0_uxtw): New pattern.
13299         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
13300         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
13302 2013-03-28  Jan Hubicka  <jh@suse.cz>
13304         * data-streamer-in.c (streamer_read_gcov_count): New function.
13305         * gimple-streamer-out.c: Include value-prof.h.
13306         (output_gimple_stmt): Output histogram.
13307         (output_bb): Use streamer_write_gcov_count.
13308         * value-prof.c: Include data-streamer.h
13309         (dump_histogram_value): Add HIST_TYPE_MAX.
13310         (stream_out_histogram_value): New function.
13311         (stream_in_histogram_value): New function.
13312         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
13313         (stream_out_histogram_value, stream_in_histogram_value): Declare.
13314         * data-streamer-out.c (streamer_write_gcov_count): New function.
13315         (streamer_write_gcov_count_stream): New function.
13316         * lto-cgraph.c (lto_output_edge): Update counter streaming.
13317         (lto_output_node): Likewise.
13318         (input_node, input_edge): Likewise.
13319         * lto-streamer-out.c (output_cfg): Update streaming.
13320         * lto-streamer-in.c (input_cfg): Likewise.
13321         * data-streamer.h (streamer_write_gcov_count,
13322         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
13323         * gimple-streamer-in.c: Include value-prof.h
13324         (input_gimple_stmt): Input histograms.
13325         (input_bb): Update profile streaming.
13327 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
13329         * genmodes.c (emit_max_int): New function.
13330         (emit_insn_modes_h): Added call to emit_max_function.
13331         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
13332         Added doc.
13333         * machmode.def: Fixed comment.
13335 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
13337         * combine.c (try_combine): Removed useless assert.
13338         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
13340 2013-03-28  Marek Polacek  <polacek@redhat.com>
13341             Richard Biener  <rguenther@suse.de>
13343         PR tree-optimization/56695
13344         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
13345         build signed result of a vector comparison.
13346         * tree-cfg.c (verify_gimple_comparison): Check that a result
13347         of a vector comparison has signed type.
13349 2013-03-28  Richard Biener  <rguenther@suse.de>
13351         PR tree-optimization/37021
13352         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
13353         do not restrict gaps between groups.
13354         * tree-vect-stmts.c (vectorizable_load): Properly account for
13355         a gap between groups.
13357 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
13359         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
13360         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
13361         is not enabled.
13363 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
13365         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
13366         * doc/extend.texi (Named Address Spaces): Ditto.
13367         (Variable Attributes): Ditto.
13369 2013-03-27  Kai Tietz  <ktietz@redhat.com>
13371         * config.build: Add support for cygwin x64 target.
13372         * config.gcc: Likewise.
13373         * config.host: Likewise.
13374         * configure.ac: Likewise
13375         * configure: Regenerated.
13377 2013-03-27  Kai Tietz  <ktietz@redhat.com>
13379         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
13380         * config/i386/t-cygwin-w64: New file.
13381         * config/i386/cygwin-w64.h: New file.
13382         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
13383         and add support for x64-cygwin target.
13384         (CPP_SPEC): Likewise.
13385         (CXX_WRAP_SPEC_LIST): Undefine before define.
13386         (LIBGCJ_SONAME): Use 15 as version.
13388 2013-03-27  Richard Biener  <rguenther@suse.de>
13390         PR tree-optimization/56716
13391         * tree-ssa-structalias.c (perform_var_substitution): Adjust
13392         dumping for ref nodes.
13394 2013-03-27  Martin Jambor  <mjambor@suse.cz>
13396         PR tree-optimization/55334
13397         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
13398         restricted pointers to arrays.
13400 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13402         * Makefile.in (.SUFFIXES): Add .cc.
13403         (.c.o): Apply same recipe for implicit rule .cc.o.
13405 2013-03-27  Richard Biener  <rguenther@suse.de>
13407         PR tree-optimization/37021
13408         * tree-vect-data-refs.c (vect_check_strided_load): Allow
13409         REALPART/IMAGPART_EXPRs around the supported refs.
13410         * tree-ssa-structalias.c (find_func_aliases): Assume that
13411         floating-point values are not used to transfer pointers.
13413 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
13415         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
13416         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
13417         * gcc/doc/tm.texi: Regenerate.
13418         * gcc/targhooks.h (default_has_ifunc_p): New.
13419         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
13420         * gcc/config/linux-protos.h: New file.
13421         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
13422         this hook for linux which disables support of indirect functions in
13423         android.
13424         * gcc/config/linux-android.c: New file.
13425         * gcc/config/t-linux-android.c: Ditto.
13426         * gcc/config.gcc: Added new object file linux-android.o.
13427         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
13428         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
13429         * gcc/varasm.c (do_assemble_alias): Likewise.
13430         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
13431         doesn't support indirect functions.
13432         * configure: Regenerate.
13434 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
13436         PR target/56102
13437         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
13438         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
13439         mult-word mode.
13441 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13443         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
13445 2013-03-27  Terry Guo  <terry.guo@arm.com>
13447         * config/arm/arm-cores.def: Added core cortex-r7.
13448         * config/arm/arm-tune.md: Regenerated.
13449         * config/arm/arm-tables.opt: Regenerated.
13450         * doc/invoke.texi: Added entry for core cortex-r7.
13452 2013-03-27  Walter Lee  <walt@tilera.com>
13454         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
13455         double-decrement of next_scratch_regno.
13457 2013-03-27  Walter Lee  <walt@tilera.com>
13459         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
13460         input operands.
13461         (insn_v1mulus): Ditto.
13462         (insn_v2muls): Ditto.
13464 2013-03-27  Walter Lee  <walt@tilera.com>
13466         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
13467         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
13469 2013-03-27  Walter Lee  <walt@tilera.com>
13471         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
13472         (*sibcall_value): Ditto.
13474 2013-03-27  Walter Lee  <walt@tilera.com>
13476         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
13477         (insn_mnz_v8qi): ... this ...
13478         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
13479         vector equivalent.
13480         (insn_v<n>mnz): Replaced by ...
13481         (insn_v1mnz): ... this ...
13482         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
13483         equivalent.
13484         (insn_mz_<mode>): Replaced by ...
13485         (insn_mz_v8qi): ... this ...
13486         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
13487         vector equivalent.
13488         (insn_v<n>mz): Replaced by ...
13489         (insn_v1mz): ... this ...
13490         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
13491         equivalent.
13493 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
13495         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
13497 2013-03-26  Roland McGrath  <mcgrathr@google.com>
13499         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
13500         than fprintf with a non-constant, non-format string.
13502 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13504         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
13505         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
13506         operand 0 predicate.
13507         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
13508         attribute.  Use general_x64nomem_operand as operand 1 predicate.
13509         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
13510         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13511         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
13512         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13513         (mov<mode>_insv_1): Remove expander.  Merge insn with
13514         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
13515         Use general_x64nomem_operand as operand 1 predicate.
13516         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
13517         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
13518         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13519         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
13520         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
13521         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
13522         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13523         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
13524         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
13525         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
13526         (general_x64nomem_operand): Ditto.
13528 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
13530         * config/rtems.opt: Add -pthread option.
13532 2013-03-26  Richard Biener  <rguenther@suse.de>
13534         * alias.c (find_base_term): Avoid redundant and not used recursion.
13535         (base_alias_check): Get the initial base term from the caller.
13536         (true_dependence_1): Compute and pass base terms to base_alias_check.
13537         (write_dependence_p): Likewise.
13538         (may_alias_p): Likewise.
13540 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
13542         * config/aarch64/aarch64.c (aarch64_classify_address): Support
13543         PC-relative load in SI modes and above only.
13545 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
13547         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
13548         * config/arm/iwmmxt.md (WCGR0): Update.
13549         (WCGR1, WCGR2, WCGR3): Likewise.
13551 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13553         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
13554         Use x64 and nox64 isa attributes.
13556 2013-03-26  Richard Biener  <rguenther@suse.de>
13558         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
13559         alignment computations and rely on get_object_alignment_1
13560         for the !TYPE_P case.
13561         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
13563 2013-03-26  Walter Lee  <walt@tilera.com>
13565         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
13566         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
13568 2013-03-25  Jeff Law  <law@redhat.com>
13570         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
13571         check for INTEGRAL_TYPE_P that was missing due to checking in
13572         wrong version of prior patch.
13574 2013-03-25  Walter Lee  <walt@tilera.com>
13576         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
13577         TILEGX_INSN_SHUFFLEBYTES1.
13578         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
13579         shufflebytes1.
13580         (tilegx_builtins): Ditto.
13581         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
13583 2013-03-25  Walter Lee  <walt@tilera.com>
13585         * config/tilegx/tilegx.md (floatsisf2): New pattern.
13586         (floatunssisf2): New pattern.
13587         (floatsidf2): New pattern.
13588         (floatunssidf2): New pattern.
13590 2013-03-25  Walter Lee  <walt@tilera.com>
13592         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
13593         tests for constraint J, K, N, P.
13595 2013-03-25  Walter Lee  <walt@tilera.com>
13597         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
13598         Use indirect/pcrel encoding.
13599         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
13600         Ditto.
13602 2013-03-25  Steve Ellcey  <sellcey@mips.com>
13604         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
13605         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
13606         * config/mips/mips.c (mips_option_override): Set IMADD default.
13607         * config/mips/mips.h (PTF_AVOID_IMADD): New.
13608         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
13609         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
13610         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
13611         * doc/invoke.texi (-mimadd/-mno-imadd): New.
13613 2013-03-25  Jeff Law  <law@redhat.com>
13615         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
13616         slightly to avoid creating and folding useless trees.  Simplify
13617         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
13619 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
13621         * config/i386/i386.md (*zero_extendsidi2): Merge with
13622         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
13623         * config/i386/predicates.md (x86_64_zext_operand): Rename from
13624         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
13625         targets.  Clarify comment.
13627 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13629         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
13630         pass-through jump functions differently.
13631         (ipa_read_jump_function): Likewise.  Also use setter functions to set
13632         up jump functions.
13634 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13636         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
13637         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
13638         process it.
13639         (ipa_get_indirect_edge_target): New function.
13640         (devirtualization_time_bonus): New parameter known_aggs, pass it to
13641         ipa_get_indirect_edge_target.  Update all callers.
13642         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
13643         ipa_get_indirect_edge_target_1 instead of calling
13644         ipa_get_indirect_edge_target.
13645         (create_specialized_node): Pass aggvlas to
13646         ipcp_discover_new_direct_edges.
13648 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13650         * config/arm/arm.md (f_sels, f_seld): New types.
13651         (*cmov<mode>): New pattern.
13652         * config/arm/predicates.md (arm_vsel_comparison_operator): New
13653         predicate.
13655 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13657         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
13658         POSIX-printf for mingw-hosted  builds.
13660 2013-03-25  Richard Biener  <rguenther@suse.de>
13662         PR middle-end/56694
13663         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
13664         must-not-throw stmt location.
13666 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13668         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
13669         Emit load-acquire versions when acq is true.
13670         (arm_emit_store_exclusive): Add rel parameter.
13671         Emit store-release versions when rel is true.
13672         (arm_split_compare_and_swap): Use acquire-release instructions
13673         instead.
13674         of barriers when appropriate.
13675         (arm_split_atomic_op): Likewise.
13676         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
13677         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
13678         (VUNSPEC_SLX): Likewise.
13679         (VUNSPEC_LDA): Likewise.
13680         (VUNSPEC_STL): Likewise.
13681         * config/arm/sync.md (atomic_load<mode>): New pattern.
13682         (atomic_store<mode>): Likewise.
13683         (arm_load_acquire_exclusive<mode>): Likewise.
13684         (arm_load_acquire_exclusivesi): Likewise.
13685         (arm_load_acquire_exclusivedi): Likewise.
13686         (arm_store_release_exclusive<mode>): Likewise.
13688 2013-03-25  Catherine Moore  <clm@codesourcery.com>
13690         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
13691         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
13692         * config/mip/predicates.md (lwsp_swsp_operand,
13693         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
13694         sb16_operand, db4_operand, db7_operand, ib3_operand,
13695         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
13696         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
13697         andi16_operand): New predicates.
13698         * config/mips/mips.md (compression): New attribute.
13699         (enabled): New attribute.
13700         (length): Consider compression in computing length.
13701         (shift_compression): New code attribute.
13702         (*add<mode>3): New operands. Record compression.
13703         (sub<mode>3): Likewise.
13704         (one_cmpl<mode>2): Likewise.
13705         (*and<mode>3): Likewise.
13706         (*ior<mode>3): Likewise.
13707         (unnamed pattern for xor): Likewise.
13708         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
13709         (*<optab><mode>3): Likewise.
13710         (*mov<mode>_internal: Likewise.
13711         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
13712         (mips_unsigned_immediate_p): New.
13713         (umips_lwsp_swsp_address_p): New.
13714         (m16_based_address_p): New.
13715         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
13716         (mips_unsigned_immediate_p): New prototype.
13717         (lwsp_swsp_address_p): New prototype.
13718         (m16_based_address_p): New prototype.
13719         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
13720         (mips_signed_immediate_p): New function.
13721         (m16_based_address_p): New function.
13722         (lwsp_swsp_address_p): New function.
13723         (mips_print_operand_punctuation): Recognize short delay slot insns
13724         for microMIPS.add<mode>3"
13726 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13728         PR target/56720
13729         * config/arm/iterators.md (v_cmp_result): New mode attribute.
13730         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
13732 2013-03-25  Richard Biener  <rguenther@suse.de>
13734         PR tree-optimization/56689
13735         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
13736         any edge.
13738 2013-03-25  Richard Biener  <rguenther@suse.de>
13740         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
13741         of bitmap.
13742         (memory_references): Likewise.
13743         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
13744         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
13745         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
13746         (gather_mem_refs_in_loops): Fold into ...
13747         (analyze_memory_references): ... this.  Move initialization
13748         to tree_ssa_lim_initialize.
13749         (fill_always_executed_in): Rename to ...
13750         (fill_always_executed_in_1): ... this.
13751         (fill_always_executed_in): Move contains_call computation to
13752         this new function from ...
13753         (tree_ssa_lim_initialize): ... here.
13754         (tree_ssa_lim): Call fill_always_executed_in.
13756 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13758         * postreload.c (reload_combine): Fix code detecting returns.
13760 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13762         * function.c (emit_use_return_register_into_block): On cc0 targets,
13763         do not emit the sequence between cc0 setter and user.
13765 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13767         * config/i386/predicates.md (local_symbolic_operand): Interpret
13768         dll-imported symbols as none-local.
13770 2013-03-25  Richard Biener  <rguenther@suse.de>
13772         * tree-ssa-loop-im.c (struct depend): Remove.
13773         (struct lim_aux_data): Make depends a vec of gimples.
13774         (free_lim_aux_data): Adjust.
13775         (add_dependency): Likewise.
13776         (set_level): Likewise.
13778 2013-03-25  Richard Biener  <rguenther@suse.de>
13780         PR middle-end/56434
13781         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
13782         the pointer returned by calls with ECF_MALLOC set.
13784 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13786         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
13788 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13790         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
13791         using MMXMODE mode iterator.
13792         (*move<mode>_internal): Merge with *movv2sf_internal and
13793         *movv2sf_internal_rex64 using MMXMODE mode iterator.
13795 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13797         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
13798         (record_last_mem_set_info): Likewise.
13800         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
13801         of XNEWVEC followed by memset.
13802         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
13804 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13806         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
13807         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
13808         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
13809         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
13810         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
13811         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
13812         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
13813         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
13814         BARRIER_P instead of GET_CODE.
13816 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13818         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
13819         inaccuracy in the probing code.
13821         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
13822         (ctrapdi4): Likewise.
13824 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13826         * calls.c (expand_call): Add missing guard to code handling return
13827         of non-BLKmode structures in MSB.
13828         * function.c (expand_function_end): Likewise.
13830 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13832         * combine.c (try_combine): Adjust comment.  Do not add the set of
13833         insn #0 if the destination indirectly is set or dies in insn #2.
13834         Tidy up code to distribute a new note.
13836 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13838         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
13839         also for alternatives 16 and 17.
13841 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13843         * config/i386/sse.md (*mov<mode>_internal): Merge with
13844         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
13845         Emit insn template depending on type attribute.  Use
13846         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13847         movd instead of movq mnemonic for interunit moves.  Rewrite mode
13848         attribute calculation.  Remove unit attribute calculation.
13849         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
13850         Set prefix_data16 attribute for DImode ssemov types.
13851         Use Ym instead of y for SSE-MMX conversion alternatives.
13852         Reorder operand constraints.
13854 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
13856         * df.h (df_insn_delete): Adjust prototype.
13857         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
13858         and let it decide whether mark the basic block dirty.
13859         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
13860         * df-scan.c (df_insn_info_delete): New helper function, split
13861         off from df_insn_delete.
13862         (df_scan_free_bb_info): Use it.
13863         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
13864         Likewise.
13865         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
13866         that the insn is actually an insn and it has a non-NULL basic block.
13867         Do not mark basic block dirty if only deleting a DEBUG_INSN.
13869 2013-03-22  Richard Biener  <rguenther@suse.de>
13871         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
13872         dep_ref members.
13873         (mem_ref_alloc): Do not allocate them.
13874         (refs_independent_p): Do not query or maintain a cache.
13876 2013-03-22  Richard Biener  <rguenther@suse.de>
13878         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
13879         (gather_mem_refs_in_loops): Do not compute it.
13880         (analyze_memory_references): Do not allocate it.
13881         (tree_ssa_lim_finalize): Do not free it.
13882         (for_all_locs_in_loop): Do not query all_refs_in_loop.
13884 2013-03-22  Richard Biener  <rguenther@suse.de>
13886         * is-a.h (as_a): Use gcc_checking_assert.
13888 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
13890         * config/aarch64/aarch64.c (aarch64_print_operand): New
13891         format specifier for printing a constant in hex.
13892         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
13893         format specifier for printing second operand.
13895 2013-03-22  Richard Biener  <rguenther@suse.de>
13897         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
13898         bitmaps.
13899         (gather_mem_refs_in_loops): Perform store accumulation here.
13900         (create_vop_ref_mapping_loop): Remove.
13901         (create_vop_ref_mapping): Likewise.
13902         (analyze_memory_references): Initialize refs_stored_in_loop.
13903         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
13904         (record_indep_loop): Remove.
13905         (record_dep_loop): New function.
13906         (ref_indep_loop_p_1): Adjust to only walk over references
13907         in the loop, not its subloops.
13908         (ref_indep_loop_p): Rename to ...
13909         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
13910         maintaining a more fine-grained cache.
13911         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
13912         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
13914 2013-03-22  Richard Biener  <rguenther@suse.de>
13916         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
13917         (struct mem_ref): Make accesses_in_loop a vec of a vec of
13918         aggregate mem_ref_loc.
13919         (free_mem_ref_locs): Inline into ...
13920         (memref_free): ... this and adjust.
13921         (mem_ref_alloc): Adjust.
13922         (mem_ref_locs_alloc): Remove.
13923         (record_mem_ref_loc): Adjust.
13924         (get_all_locs_in_loop): Rewrite into ...
13925         (for_all_locs_in_loop): ... this iterator.
13926         (rewrite_mem_ref_loc): New functor.
13927         (rewrite_mem_refs): Use for_all_locs_in_loop.
13928         (sm_set_flag_if_changed): New functor.
13929         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13930         (ref_always_accessed): New functor.
13931         (ref_always_accessed_p): Use for_all_locs_in_loop.
13933 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13935         * tree-pass.h (PROP_gimple_lvec): New.
13936         * passes.c (dump_properties): Handle PROP_gimple_lvec.
13937         (init_optimization_passes): Move pass_lower_vector.
13938         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13939         PROP_gimple_lvec.
13940         (pass_lower_vector): Provide PROP_gimple_lvec.
13941         (pass_lower_vector_ssa): Likewise.
13942         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13944 2013-03-21  Mark Wielaard  <mjw@redhat.com>
13946         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13948 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
13950         * config/i386/i386.md (*movdi_internal): Disparage slightly
13951         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
13952         conversion alternatives.
13954 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
13956         PR middle-end/48087
13957         * diagnostic.def (DK_WERROR): New kind.
13958         * diagnostic.h (werrorcount): Define.
13959         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13960         promoted to DK_ERROR, increment DK_WERROR counter instead of
13961         DK_ERROR counter.
13962         * toplev.c (toplev_main): Call print_ignored_options even if
13963         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
13964         even if just werrorcount is non-zero.
13966         PR debug/55608
13967         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13968         on failure.
13969         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13970         (string_cst_pool_decl): New function.
13971         (optimize_one_addr_into_implicit_ptr): New function.
13972         (resolve_addr_in_expr): Optimize DWARF location expression
13973         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13974         which doesn't live in memory, but has DW_AT_location or
13975         DW_AT_const_value, or refers to a string literal, into
13976         DW_OP_GNU_implicit_pointer.
13977         (optimize_location_into_implicit_ptr): New function.
13978         (resolve_addr): If removing DW_AT_location of a variable because
13979         it was DW_OP_addr of address of the variable, but the variable doesn't
13980         live in memory, try to emit const value attribute for the initializer.
13982 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13984         * tree.h (VECTOR_TYPE_P): New macro.
13985         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13986         TYPE_MODE): Use it.
13987         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13988         VEC_COND_EXPR cannot be lvalues.
13989         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13991 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13993         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13994         Restrict the transformation to equal modes.
13996 2013-03-21  Richard Biener  <rguenther@suse.de>
13998         PR tree-optimization/39326
13999         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14000         (MEM_ANALYZABLE): Adjust.
14001         (record_mem_ref_loc): Move bitmap ops ...
14002         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
14003         unanalyzable refs, do not record locations for it.
14004         (analyze_memory_references): Allocate ref zero as shared
14005         unanalyzable ref.
14006         (refs_independent_p): Do not test for unanalyzed mems here.
14007         (ref_indep_loop_p_1): Special-case disambiguation against
14008         the unanalyzed ref.
14009         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14011 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
14013         * config/arm/arm-protos.h (tune_params): Add
14014         prefer_neon_for_64bits field.
14015         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14016         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14017         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14018         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14019         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14020         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14021         (arm_option_override): Handle -mneon-for-64bits new option.
14022         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14023         (prefer_neon_for_64bits): Declare new variable.
14024         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14025         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14026         (arch_enabled): Handle new arch types. Remove support for onlya8
14027         and nota8.
14028         (one_cmpldi2): Use new arch names.
14029         (zero_extend<mode>di2, extend<mode>di2): Ditto.
14030         * config/arm/arm.opt (mneon-for-64bits): Add option.
14031         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14032         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14033         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14034         of onlya8.
14035         * doc/invoke.texi (-mneon-for-64bits): Document.
14037 2013-03-21  Richard Biener  <rguenther@suse.de>
14039         PR tree-optimization/39326
14040         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14041         (sort_bbs_in_loop_postorder_cmp): New function.
14042         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14044 2013-03-21  Richard Biener  <rguenther@suse.de>
14046         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14047         (vect_insert_into_interleaving_chain): Likewise.
14048         (vect_drs_dependent_in_basic_block): Inline ...
14049         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
14050         split out from ...
14051         (vect_analyze_data_ref_dependence): ... here.  Simplify.
14052         (vect_check_interleaving): Simplify.
14053         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
14054         (vect_slp_analyze_data_ref_dependences): ... this new function.
14055         (dr_group_sort_cmp): New function.
14056         (vect_analyze_data_ref_accesses): Compute data-reference groups
14057         here instead of in vect_analyze_data_ref_dependence.  Use
14058         a more efficient algorithm.
14059         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14060         vect_slp_analyze_data_ref_dependences.  Call
14061         vect_analyze_data_ref_accesses earlier.
14062         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14063         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14064         (vect_slp_analyze_data_ref_dependences): New prototype.
14066 2013-03-21  Richard Biener  <rguenther@suse.de>
14068         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14069         ref is stored in the loop.
14070         (find_refs_for_sm): Walk only over all stores.
14071         (store_motion_loop): Allocate from lim_bitmap_obstack.
14072         (store_motion): Likewise.
14074 2013-03-21  Richard Biener  <rguenther@suse.de>
14076         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14077         Update virtual SSA form.
14079 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14081         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14082         * configure: Regenerate.
14083         * config.in: Regenerate.
14084         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14085         if !HAVE_LD_EH_FRAME_CIEV3.
14087 2013-03-21  Richard Biener  <rguenther@suse.de>
14089         * tree-cfg.c (verify_expr_no_block): New function.
14090         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14091         nor DECL_VALUE_EXPR have locations with associated blocks.
14092         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14093         (clear_unused_block_pointer): Remove code dealing with
14094         blocks in DECL_DEBUG_EXPR locations.
14096 2013-03-21  Richard Biener  <rguenther@suse.de>
14098         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14099         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
14100         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14101         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14102         instead of DECL_DEBUG_EXPR_IS_FROM.
14103         * gimplify.c (gimplify_modify_expr): Likewise.
14104         * tree-cfg.c (verify_expr_location_1): Likewise.
14105         * tree-complex.c (create_one_component_var): Likewise.
14106         * tree-sra.c (create_access_replacement): Likewise.
14107         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14108         (clear_unused_block_pointer): Likewise.
14109         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14110         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14111         * var-tracking.c (var_debug_decl): Likewise.
14112         (track_expr_p): Likewise.
14113         * tree-inline.c (add_local_variables): Likewise.  Set
14114         DECL_HAS_DEBUG_EXPR_P after copying it.
14115         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14116         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
14118 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14120         PR bootstrap/56656
14121         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14122         * configure: Regenerate.
14123         * config.in: Regenerate.
14124         * config/i386/i386.md (*movdf_internal): Use
14125         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14126         movd instead of movq mnemonic for interunit moves.
14127         (*movdi_internal): Ditto.
14129 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14131         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14132         (abd<mode>_3): New pattern.
14133         (aba<mode>_3): New pattern.
14134         (fabd<mode>_3): New pattern.
14136 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14138         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14139         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14140         occurrence of REGISTER_PREFIX as its empty string.
14142 2013-03-20  Jeff Law  <law@redhat.com>
14144         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14145         addititional equivalences for equality comparisons between an SSA_NAME
14146         and a constant where the SSA_NAME was set from a widening conversion.
14148 2013-03-20  Walter Lee  <walt@tilera.com>
14150         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14152 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14154         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14155         depending on type attribute.
14156         (*movti_internal): Ditto.
14157         (*movtf_internal): Ditto.
14158         (*movxf_internal): Ditto.
14159         (*movdf_internal): Ditto.
14160         (*movsf_internal): Ditto.
14162 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14164         * config/i386/i386.md (*movti_internal): Set prefix attribute to
14165         maybe_vex for sselog1 and ssemov types.
14166         (*movdi_internal): Reorder operand constraints.
14167         (*movsi_internal): Ditto.  Set prefix attribute to
14168         maybe_vex for sselog1 and ssemov types.
14169         (*movtf_internal): Set prefix attribute to maybe_vex
14170         for sselog1 and ssemov types.
14171         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
14172         DImode ssemov types.  Reorder operand constraints.
14173         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
14174         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
14175         attribute for SImode ssemov types.  Reorder operand constraints.
14177 2013-03-20  Martin Jambor  <mjambor@suse.cz>
14179         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14180         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14182 2013-03-20  Pat Haugen <pthaugen@us.ibm.com>
14184         * config/rs6000/predicates.md (indexed_address, update_address_mem
14185         update_indexed_address_mem): New predicates.
14186         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14187         attribute for load/store instructions.
14188         * config/rs6000/dfp.md (movsd_store): Likewise.
14189         (movsd_load): Likewise.
14190         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14191         (unnamed HI->DI extend define_insn): Likewise.
14192         (unnamed SI->DI extend define_insn): Likewise.
14193         (unnamed QI->SI extend define_insn): Likewise.
14194         (unnamed QI->HI extend define_insn): Likewise.
14195         (unnamed HI->SI extend define_insn): Likewise.
14196         (unnamed HI->SI extend define_insn): Likewise.
14197         (extendsfdf2_fpr): Likewise.
14198         (movsi_internal1): Likewise.
14199         (movsi_internal1_single): Likewise.
14200         (movhi_internal): Likewise.
14201         (movqi_internal): Likewise.
14202         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
14203         attribute for load/store instructions.
14204         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
14205         instructions.
14206         (mov<mode>_softfloat): Likewise.
14207         (mov<mode>_hardfloat32): Likewise.
14208         (mov<mode>_hardfloat64): Likewise.
14209         (mov<mode>_softfloat64): Likewise.
14210         (movdi_internal32): Likewise.
14211         (movdi_internal64): Likewise.
14212         (probe_stack_<mode>): Likewise.
14214 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
14216         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
14217         floating point, and decimal floating point to reload iterator.
14219         * config/rs6000/constraints.md (wl constraint): New constraints to
14220         return FLOAT_REGS if certain options are used to reduce the number
14221         of separate patterns that exist in the file.
14222         (wx constraint): Likewise.
14223         (wz constraint): Likewise.
14225         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14226         -mdebug=reg, print wg, wl, wx, and wz constraints.
14227         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14228         Initialize the reload functions for 64-bit binary/decimal floating
14229         point types.
14230         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14231         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14232         create the buffer on the stack to overcome not having a 32-bit
14233         load and store.
14234         (rs6000_emit_move): Likewise.
14235         (rs6000_secondary_memory_needed_rtx): Likewise.
14236         (rs6000_alloc_sdmode_stack_slot): Likewise.
14237         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14238         via xxlxor, just like DFmode 0.0.
14240         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
14241         define as 1 if we are running on a power7 or newer.
14242         (enum r6000_reg_class_enum): Add new constraints.
14244         * config/rs6000/dfp.md (movsd): Delete, combine with binary
14245         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14246         with other moves by using conditional constraits (wg).  Use LFIWZX
14247         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
14248         (movsd splitter): Likewise.
14249         (movsd_hardfloat): Likewise.
14250         (movsd_softfloat): Likewise.
14252         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
14253         binary and decimal floating point moves.
14254         (fmove_ok): New attributes to combine binary and decimal floating
14255         point moves, and to combine power6x (mfpgpr) moves along normal
14256         floating moves.
14257         (real_value_to_target): Likewise.
14258         (f32_lr): Likewise.
14259         (f32_lm): Likewise.
14260         (f32_li): Likewise.
14261         (f32_sr): Likewise.
14262         (f32_sm): Likewise.
14263         (f32_si): Likewise.
14264         (movsf): Combine binary and decimal floating point moves.  Combine
14265         power6x (mfpgpr) moves with other moves by using conditional
14266         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
14267         (mov<mode> for SFmode/SDmode); Likewise.
14268         (SFmode/SDmode splitters): Likewise.
14269         (movsf_hardfloat): Likewise.
14270         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
14271         (movsf_softfloat): Likewise.
14272         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
14274         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
14275         wx and wz constraints.
14277         * config/rs6000/constraints.md (wg constraint): New constraint to
14278         return FLOAT_REGS if -mmfpgpr (power6x) was used.
14280         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
14281         constraint.
14283         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14284         -mdebug=reg, print wg, wl, wx, and wz constraints.
14285         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14286         Initialize the reload functions for 64-bit binary/decimal floating
14287         point types.
14288         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14289         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14290         create the buffer on the stack to overcome not having a 32-bit
14291         load and store.
14292         (rs6000_emit_move): Likewise.
14293         (rs6000_secondary_memory_needed_rtx): Likewise.
14294         (rs6000_alloc_sdmode_stack_slot): Likewise.
14295         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14296         via xxlxor, just like DFmode 0.0.
14298         * config/rs6000/dfp.md (movdd): Delete, combine with binary
14299         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14300         with other moves by using conditional constraits (wg).  Use LFIWZX
14301         and STFIWX for loading SDmode on power7.
14302         (movdd splitters): Likewise.
14303         (movdd_hardfloat32): Likewise.
14304         (movdd_softfloat32): Likewise.
14305         (movdd_hardfloat64_mfpgpr): Likewise.
14306         (movdd_hardfloat64): Likewise.
14307         (movdd_softfloat64): Likewise.
14309         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
14310         64-bit binary and decimal floating point moves.
14311         (FMOVE64X): Likewise.
14312         (movdf): Combine 64-bit binary and decimal floating point moves.
14313         Combine power6x (mfpgpr) moves with other moves by using
14314         conditional constraits (wg).
14315         (mov<mode> for DFmode/DDmode): Likewise.
14316         (DFmode/DDmode splitters): Likewise.
14317         (movdf_hardfloat32): Likewise.
14318         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
14319         (movdf_softfloat32): Likewise.
14320         (movdf_hardfloat64_mfpgpr): Likewise.
14321         (movdf_hardfloat64): Likewise.
14322         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
14323         (movdf_softfloat64): Likewise.
14324         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
14325         (reload_<mode>_load): Move to later in the file so they aren't in
14326         the middle of the floating point move insns.
14327         (reload_<mode>_store): Likewise.
14329         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
14330         constraint.
14332         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
14333         constraint if -mdebug=reg.
14334         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
14335         Enable using dd reload support if needed.
14337         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
14338         binary and decimal floating point moves in rs6000.md.
14339         (movtd_internal): Likewise.
14341         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
14342         decimal floating point moves.
14343         (movtf): Likewise.
14344         (movtf_internal): Likewise.
14345         (mov<mode>_internal, TDmode/TFmode): Likewise.
14346         (movtf_softfloat): Likewise.
14347         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
14349         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
14350         movdi_internal64, using wg constraint for move direct operations.
14351         (movdi_internal64): Likewise.
14353         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
14354         MODES_TIEABLE_P for selected modes.  Print the numerical value of
14355         the various virtual registers. Use GPR/FPR first/last values,
14356         instead of hard coding the register numbers.  Print which modes
14357         have reload functions registered.
14358         (rs6000_option_override_internal): If -mdebug=reg, trace the options
14359         settings before/after setting cpu, target and subtarget settings.
14360         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
14361         and for secondary reload failures in rs6000_secondary_reload_inner.
14362         (rs6000_secondary_reload_fail): Likewise.
14363         (rs6000_secondary_reload_inner): Likewise.
14365         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
14366         macros for first/last GPR and FPR registers.
14367         (LAST_GPR_REGNO): Likewise.
14368         (FIRST_FPR_REGNO): Likewise.
14369         (LAST_FPR_REGNO): Likewise.
14371         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
14372         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
14373         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
14374         (vcond<mode><mode>): Likewise.
14375         (vcondu<mode><mode>): Likewise.
14376         (vector_gtu<mode>): Likewise.
14377         (vector_gte<mode>): Likewise.
14378         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
14379         to prevent the compiler from converting DImode operations to TImode.
14380         (ior<mode>3): Likewise.
14381         (and<mode>3): Likewise.
14382         (one_cmpl<mode>2): Likewise.
14383         (nor<mode>3): Likewise.
14384         (andc<mode>3): Likewise.
14386         * config/rs6000/constraints.md (wt constraint): New constraint
14387         that returns VSX_REGS if TImode is allowed in VSX registers.
14389         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
14390         constant under VSX.
14392         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
14393         similar to TImode, but it is restricted to being in the GPRs.
14395         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
14396         TImode to occupy a single VSX register.
14398         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
14399         -mvsx-timode for power7/power8.
14400         (power7 cpu): Likewise.
14401         (power8 cpu): Likewise.
14403         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
14404         sure that TFmode/TDmode take up two registers if they are ever
14405         allowed in the upper VSX registers.
14406         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
14407         registers.
14408         (rs6000_init_hard_regno_mode_ok): Likewise.
14409         (rs6000_debug_reg_global): Add debugging for PTImode and wt
14410         constraint.  Print if LRA is turned on.
14411         (rs6000_option_override_internal): Give an error if -mvsx-timode
14412         and VSX is not enabled.
14413         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
14414         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
14415         to reg+offset addressing.  Use PTImode when checking offset
14416         addresses for validity.
14417         (reg_offset_addressing_ok_p): Likewise.
14418         (rs6000_legitimate_offset_address_p): Likewise.
14419         (rs6000_legitimize_address): Likewise.
14420         (rs6000_legitimize_reload_address): Likewise.
14421         (rs6000_legitimate_address_p): Likewise.
14422         (rs6000_eliminate_indexed_memrefs): Likewise.
14423         (rs6000_emit_move): Likewise.
14424         (rs6000_secondary_reload): Likewise.
14425         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
14426         reloads to fpr registers to continue to use reg+offset addressing,
14427         but 64-bit reloads to altivec registers need reg+reg addressing.
14428         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
14429         it.  Treat LO_SUM like a PLUS operation.
14430         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
14431         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
14432         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
14433         registers to share a register with a smaller sized type, since VSX
14434         puts scalars in the upper 64-bits.
14435         (print_operand): Add support for PTImode.
14436         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
14437         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
14438         registers, but don't have arithmetic support.
14439         (rs6000_memory_move_cost): Add test for VSX.
14440         (rs6000_opt_masks): Add -mvsx-timode.
14442         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
14443         for TImode.
14444         (VSs): Likewise.
14445         (VSr): Use wt constraint for TImode.
14446         (VSv): Drop TImode support.
14447         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
14448         (vsx_movti_64bit): Likewise.
14449         (vsx_movti_32bit): Likewise.
14450         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
14451         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
14452         one '?' on the appropriate output constraint.  Do not allow TImode
14453         logical operations on 32-bit systems.
14454         (vsx_ior<mode>3): Likewise.
14455         (vsx_xor<mode>3): Likewise.
14456         (vsx_one_cmpl<mode>2): Likewise.
14457         (vsx_nor<mode>3): Likewise.
14458         (vsx_andc<mode>3): Likewise.
14459         (vsx_concat_<mode>): Likewise.
14460         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
14462         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
14463         OPTION_MASK_VSX_TIMODE.
14464         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
14465         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
14467         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
14468         (TI2 iterator): New iterator for TImode, PTImode.
14469         (wd mode attribute): Add values for vector types.
14470         (movti_string): Replace TI move operations with operations for TImode
14471         and PTImode.  Add support for TImode being allowed in VSX registers.
14472         (mov<mode>_string, TImode/PTImode): Likewise.
14473         (movti_ppc64): Likewise.
14474         (mov<mode>_ppc64, TImode/PTImode): Likewise.
14475         (TI mode splitters): Likewise.
14477         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
14478         constraint.
14480 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
14482         PR tree-optimization/56355
14483         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
14484         Also handle integers with undefined overflow.
14486 2013-03-20  Catherine Moore  <clm@codesourcery.com>
14487             Maciej W. Rozycki  <macro@codesourcery.com>
14488             Tom de Vries  <tom@codesourcery.com>
14489             Nathan Sidwell <nathan@codesourcery.com>
14490             Iain Sandoe  <iain@codesourcery.com>
14491             Nathan Froyd  <froydnj@codesourcery.com>
14492             Chao-ying Fu <fu@mips.com>
14494         * doc/extend.texi: (micromips, nomicromips, nocompression):
14495         Document new function attributes.
14496         * doc/invoke.texi (minterlink-compressed, mmicromips,
14497         m14k, m14ke, m14kec): Document new options.
14498         (minterlink-mips16): Update documentation.
14499         * doc/md.texi (ZC, ZD): Document new constraints.
14500         * configure.ac (gcc_cv_as_micromips): Check if linker
14501         supports the .set micromips directive.
14502         * configure: Regenerate.
14503         * config.in: Regenerate.
14504         * config/mips/mips-tables.opt: Regenerate.
14505         * config/mips/micromips.md: New file.
14506         * constraints.md (ZC, ZD): New constraints.
14507         * config/mips/predicates.md (movep_src_register): New predicate.
14508         (movep_src_operand): New predicate.
14509         (non_volatile_mem_operand): New predicate.
14510         * config/mips/mips.md (multimem): New type.
14511         (length): Differentiate between 17-bit and 18-bit branch offsets.
14512         (MOVEP1, MOVEP2): New mode iterator.
14513         (mov_<load>l): Use ZC constraint.
14514         (mov_<load>r): Likewise.
14515         (mov_<store>l): Likewise.
14516         (mov_<store>r): Likewise.
14517         (*branch_equality<mode>_inverted): Add microMIPS support.
14518         (*branch_equality<mode>): Likewise.
14519         (*jump_absolute): Likewise.
14520         (indirect_jump_<mode>): Likewise.
14521         (tablejump_<mode>): Likewise.
14522         (<optab>_internal): Likewise.
14523         (sibcall_internal): Likewise.
14524         (sibcall_value_internal): Likewise.
14525         (prefetch): Use constraint ZD.
14526         * config/mips/mips.opt (minterlink-compressed): New option.
14527         (minterlink-mips16): Now an alias for minterlink-compressed.
14528         (mmicromips): New option.
14529         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14530         (compare_and_swap_12): Likewise.
14531         (sync_add<mode>): Likewise.
14532         (sync_<optab>_12): Likewise.
14533         (sync_old_<optab>_12): Likewise.
14534         (sync_new_<optab>_12): Likewise.
14535         (sync_nand_12): Likewise.
14536         (sync_old_nand_12): Likewise.
14537         (sync_new_nand_12): Likewise.
14538         (sync_sub<mode>): Likewise.
14539         (sync_old_add<mode>): Likewise.
14540         (sync_old_sub<mode>): Likewise.
14541         (sync_new_add<mode>): Likewise.
14542         (sync_new_sub<mode>): Likewise.
14543         (sync_<optab><mode>): Likewise.
14544         (sync_old_<optab><mode>): Likewise.
14545         (sync_new_<optab><mode>): Likewise.
14546         (sync_nand<mode>): Likewise.
14547         (sync_old_nand<mode>): Likewise.
14548         (sync_new_nand<mode>): Likewise.
14549         (sync_lock_test_and_set<mode>): Likewise.
14550         (test_and_set_12): Likewise.
14551         (atomic_compare_and_swap<mode>): Likewise.
14552         (atomic_exchange<mode>_llsc): Likewise.
14553         (atomic_fetch_add<mode>_llsc): Likewise.
14554         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14555         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14556         (umips_save_restore_pattern_p): Likewise.
14557         (umips_load_store_pair_p): Likewise.
14558         (umips_output_load_store_pair): Likewise.
14559         (umips_movep_target_p): Likewise.
14560         (umips_12bit_offset_address_p): Likewise.
14561         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14562         (mips_base_mips16): Rename this...
14563         (mips_base_compression_flags): ...to this. Update all uses.
14564         (mips_attribute_table): Add micromips, nomicromips and nocompression.
14565         (mips_mips16_decl_p): Delete.
14566         (mips_nomips16_decl_p): Delete.
14567         (mips_get_compress_on_flags): New function.
14568         (mips_get_compress_off_flags): New function.
14569         (mips_get_compress_mode): New function.
14570         (mips_get_compress_on_name): New function.
14571         (mips_get_compress_off_name): New function.
14572         (mips_insert_attributes): Support multiple compression types.
14573         (mips_merge_decl_attributes): Likewise.
14574         (umips_12bit_offset_address_p): New function.
14575         (mips_start_function_definition): Emit .set micromips directive.
14576         (mips_call_may_need_jalx_p): New function.
14577         (mips_function_ok_for_sibcall): Add microMIPS support.
14578         (mips_print_operand_punctuation): Support short delay slots and
14579         compact jumps.
14580         (umips_swm_mask, umips_swm_encoding): New.
14581         (umips_build_save_restore): New function.
14582         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14583         (was_mips16_p): Remove.
14584         (old_compression_mode): New.
14585         (mips_set_compression_mode): New function.
14586         (mips_set_current_function): Add microMIPS support.
14587         (mips_option_override): Likewise.
14588         (umips_save_restore_pattern_p): New function.
14589         (umips_output_save_restore): New function.
14590         (umips_load_store_pair_p_1): New function.
14591         (umips_load_store_pair_p): New function.
14592         (umips_output_load_store_pair_1): New function.
14593         (umips_output_load_store_pair): New function.
14594         (umips_movep_target_p) New function.
14595         (mips_prepare_pch_save): Add microMIPS support.
14596         * config/mips/mips.h (TARGET_COMPRESSION): New.
14597         (TARGET_CPU_CPP_BUILTINS): Update macro
14598         to use new compression flags and to support microMIPS.
14599         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14600         (MIPS_ARCH_FLOAT_SPEC): Likewise.
14601         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14602         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14603         (ASM_SPEC): Support mmicromips and mno-micromips.
14604         (M16STORE_REG_P): New macro.
14605         (MIPS_CALL): Support TARGET_MICROMIPS.
14606         (MICROMIPS_J): New macro.
14607         (mips_base_mips16): Rename this...
14608         (mips_base_compression_flags): ...to this.
14609         (UMIPS_12BIT_OFFSET_P): New macro.
14610         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14611         (MULTILIB_DIRNAMES): Likewise.
14612 2013-03-20  Richard Biener  <rguenther@suse.de>
14614         PR tree-optimization/56661
14615         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
14616         the result does not have to be distinct.
14618 2013-03-20  Richard Biener  <rguenther@suse.de>
14620         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
14621         remap_gimple_op_r.
14623 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14624             Steven Bosscher <steven@gcc.gnu.org>
14626         PR rtl-optimization/56605
14627         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
14629 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14631         PR bootstrap/56656
14632         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
14633         that require movd instead of movq.
14635 2013-03-20  Richard Biener  <rguenther@suse.de>
14637         * tree-ssa-structalias.c (struct variable_info): Add pointer
14638         to the first field of an aggregate with sub-vars.  Make
14639         this and the pointer to the next subfield its ID.
14640         (vi_next): New function.
14641         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
14642         storedanything_id, integer_id): Increment by one.
14643         (new_var_info, get_call_vi, lookup_call_clobber_vi,
14644         get_call_clobber_vi): Adjust.
14645         (solution_set_expand): Simplify and speedup.
14646         (solution_set_add): Inline into ...
14647         (set_union_with_increment): ... this.  Adjust accordingly.
14648         (do_sd_constraint): Likewise.
14649         (do_ds_constraint): Likewise.
14650         (do_complex_constraint): Simplify.
14651         (build_pred_graph): Adjust.
14652         (solve_graph): Likewise.  Simplify and speedup.
14653         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
14654         get_constraint_for_component_ref, get_constraint_for_1,
14655         first_vi_for_offset, first_or_preceding_vi_for_offset,
14656         create_function_info_for, create_variable_info_for_1,
14657         create_variable_info_for, intra_create_variable_infos): Adjust.
14658         (init_base_vars): Push NULL for ID zero.
14659         (compute_points_to_sets): Adjust.
14661 2013-03-20  Richard Biener  <rguenther@suse.de>
14663         * cfgloop.c (verify_loop_structure): Streamline and avoid
14664         ICEing on corrupt loop tree.
14665         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
14666         loop tree.
14668 2013-03-20  Richard Biener  <rguenther@suse.de>
14670         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
14671         check whether an SSA update is needed.
14673 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
14675         * config/mips/constraints.md (T): Rename to...
14676         (Yf): ...this.
14677         (U): Rename to...
14678         (Yd): ...this.
14679         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
14680         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
14682 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14684         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
14685         (*subsi3_carryin_uxtw): Likewise.
14687 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14689         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
14690         (*rorsi3_insn_uxtw): Likewise.
14692 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14694         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
14695         (*extrsi5_insn_uxtw): Likewise.
14697 2013-03-19  Richard Biener  <rguenther@suse.de>
14699         PR tree-optimization/56273
14700         * passes.c (init_optimization_passes): Move second VRP after DOM.
14702 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
14704         * config/i386/i386.md (*movti_internal): Merge from
14705         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
14706         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
14707         nox64 isa attributes.
14709 2013-03-18  Richard Biener  <rguenther@suse.de>
14711         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
14712         (unite): Likewise.
14713         (merge_node_constraints): Likewise.
14714         (build_succ_graph): Likewise.
14715         (valid_graph_edge): Inline into single caller.
14716         (unify_nodes): Likewise.  Use bitmap_set_bit return value
14717         and cache varinfo.
14718         (scc_visit): Fix formatting and variable use.
14719         (do_sd_constraint): Use gcc_checking_assert.
14720         (do_ds_constraint): Likewise.
14721         (do_complex_constraint): Likewise.
14722         (condense_visit): Likewise.  Cleanup.
14723         (dump_pred_graph): New function.
14724         (perform_var_substitution): Dump the pred-graph before
14725         variable substitution.
14726         (find_equivalent_node): Use gcc_checking_assert.
14727         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
14729 2013-03-18  Richard Biener  <rguenther@suse.de>
14731         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
14732         Remove cond_expr_stmt_list argument and do not gimplify the
14733         built expression.
14734         (vect_loop_versioning): Adjust.
14735         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
14736         Cleanup to use less temporaries.
14737         (vect_create_data_ref_ptr): Cleanup.
14739 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14741         PR tree-optimization/56635
14742         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
14743         require types_compatible_p types.
14745 2013-03-18  Nick Clifton  <nickc@redhat.com>
14747         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
14748         spurious backslash.
14750         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
14751         Add missing line to comment describing function.
14753 2013-03-18  Richard Biener  <rguenther@suse.de>
14755         PR tree-optimization/56210
14756         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14757         Handle string / character search functions.
14758         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
14760 2013-03-18  Richard Biener  <rguenther@suse.de>
14762         PR middle-end/56483
14763         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
14764         and implement properly.
14765         * gimple.h (gimple_cond_single_var_p): Remove.
14767 2013-03-18  Richard Biener  <rguenther@suse.de>
14769         * tree-data-ref.h (find_data_references_in_loop): Declare.
14770         * tree-data-ref.c (get_references_in_stmt): Use a stack
14771         vector pre-allocated in the callers.
14772         (find_data_references_in_stmt): Adjust.
14773         (graphite_find_data_references_in_stmt): Likewise.
14774         (create_rdg_vertices): Likewise.
14775         (find_data_references_in_loop): Export.
14776         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14777         Compute dependences here...
14778         (vect_analyze_data_refs): ...not here.  When we encounter
14779         a non-vectorizable data reference in basic-block vectorization
14780         truncate the data reference vector.  Do not bother to
14781         fixup data-dependence information for gather loads.
14782         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
14783         of data references, as reported.
14785 2013-03-18  Richard Biener  <rguenther@suse.de>
14787         PR tree-optimization/3713
14788         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
14789         has_constants and expr.
14790         (stmt_has_constants): Properly valueize SSA names when deciding
14791         whether the stmt has constants.
14793 2013-03-18  Richard Biener  <rguenther@suse.de>
14795         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
14796         whole function when there is nothing to do.
14797         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
14798         * tree-vectorizer.c (vectorize_loops): Update virtual and
14799         loop-closed SSA once.
14800         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
14802 2013-03-18  Richard Biener  <rguenther@suse.de>
14804         PR middle-end/56113
14805         * domwalk.c (bb_postorder): New global static.
14806         (cmp_bb_postorder): New function.
14807         (walk_dominator_tree): Replace scheme imposing an order for
14808         visiting dominator sons by one sorting them at the time they
14809         are pushed on the stack.
14811 2013-03-18  Richard Biener  <rguenther@suse.de>
14813         PR tree-optimization/39326
14814         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
14815         (struct mem_ref): Replace mem member with ao_ref typed member.
14816         (MEM_ANALYZABLE): Adjust.
14817         (memref_eq): Likewise.
14818         (mem_ref_alloc): Likewise.
14819         (gather_mem_refs_stmt): Likewise.
14820         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
14821         (execute_sm_if_changed_flag_set): Adjust.
14822         (execute_sm): Likewise.
14823         (ref_always_accessed_p): Likewise.
14824         (refs_independent_p): Likewise.
14825         (can_sm_ref_p): Likewise.
14827 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14829         PR c/56566
14830         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
14831         return 1 even for !unsignedp.
14833 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14835         * config/i386/i386.md (isa): Add x64 and nox64.
14836         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
14837         (*pushtf): Enable *roF alternative for x64 isa only.
14838         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
14839         mode attribute of integer alternatives to DImode for TARGET_64BIT.
14840         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
14841         (*movtf_internal): Merge from *movtf_internal_rex64 and
14842         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
14843         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
14844         nox64 isa attributes.
14845         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
14846         nox64 isa attributes.
14847         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
14849 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14851         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14853 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
14855         PR target/56640
14856         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
14857         class names.  Remove trailing comma after "ALL_REGS".
14859 2013-03-16  Jan Hubicka  <jh@suse.cz>
14861         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
14862         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
14863         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
14864         of cgraph_get_create_node.
14865         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14867 2013-03-16  Jason Merrill  <jason@redhat.com>
14869         PR debug/49090
14870         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
14871         with DW_AT_default_value.
14873 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
14875         * BASE-VER: Set to 4.9.0.
14877 2013-03-14  Andi Kleen  <ak@linux.intel.com>
14879         PR target/56619
14880         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
14881         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
14882         Document _x* TSX intrinsics.
14884 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
14885             David Holsgrove  <david.holsgrove@xilinx.com>
14887         * configure.ac: Add MicroBlaze TLS support detection.
14888         * configure: Regenerate.
14889         * config/microblaze/microblaze-protos.h
14890         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
14891         symbol_mentioned_p, label_mentioned_p): Add prototypes.
14892         * config/microblaze/microblaze.c (microblaze_address_type): Add
14893         ADDRESS_TLS and tls_reloc address types.
14894         (microblaze_address_info): Add tls_reloc.
14895         (TARGET_HAVE_TLS): Define.
14896         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
14897          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
14898          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
14899         load_tls_operand,  microblaze_call_tls_get_addr,
14900         microblaze_legitimize_tls_address): New functions.
14901         (microblaze_classify_unspec): Handle UNSPEC_TLS.
14902         (get_base_reg): Use microblaze_tls_symbol_p.
14903         (microblaze_classify_address): Handle TLS.
14904         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
14905         label_mentioned_p and microblaze_tls_referenced_p.
14906         (microblaze_legitimize_address): Handle TLS.
14907         (microblaze_address_insns): Handle ADDRESS_TLS.
14908         (pic_address_needs_scratch): Handle TLS.
14909         (print_operand_address): Handle TLS.
14910         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
14911         (microblaze_expand_move): Handle TLS.
14912         (microblaze_legitimate_constant_p): Check
14913         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
14914         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14915         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
14916         (PIC_OFFSET_TABLE_REGNUM): Set.
14917         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
14918         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
14919         (addsi3, movsi_internal2, movdf_internal): Update constraints
14920         * config/microblaze/predicates.md (arith_plus_operand): Define
14921         (move_operand): Redefine as move_src_operand,
14922         check microblaze_tls_referenced_p.
14924 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14926         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
14927         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14929 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14931         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14932         CC mode for AND.
14934 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
14936         PR tree-optimization/53265
14937         * common.opt (Waggressive-loop-optimizations): New option.
14938         * tree-ssa-loop-niter.c: Include tree-pass.h.
14939         (do_warn_aggressive_loop_optimizations): New function.
14940         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
14941         if number_of_latch_executions returned constant.
14942         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14943         early.  If number_of_latch_executions returned constant, set
14944         nb_iterations_upper_bound back to it.
14945         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14946         field.
14947         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14948         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14950         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14951         (MULTILIB_OSDIRNAMES): Set.
14952         * genmultilib: If defaultosdirname doesn't start with :: , set
14953         defaultosdirname2 instead, clear it and emit two . multilib_raw
14954         entries instead of just one.
14956 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
14958         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14959         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14960         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14961         (SUBTARGET_OVERRIDE_OPTIONS): New.
14963 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
14965         PR target/49880
14966         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14967         (musermode): Convert to Var(TARGET_USERMODE).
14968         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14969         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14970         * config/sh/sh.c (sh_option_override): Use
14971         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14972         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14973         condition.
14974         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14975         TARGET_SH4.
14976         (udivsi3_i4_single, divsi3_i4_single): Use
14977         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14979 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
14981         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14982         default setting.
14984 2013-03-13  Richard Biener  <rguenther@suse.de>
14986         PR tree-optimization/56608
14987         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14988         calls when vectorizing basic-blocks.
14990 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
14992         PR plugins/45078
14993         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14994         tm_file.
14996 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
14998         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15000 2013-03-11  Jan Hubicka  <jh@suse.cz>
15002         PR lto/56557
15003         * lto-streamer-out.c (output_symbol_p): Skip references from
15004         constructors of external variables.
15006 2013-03-11  Jan Hubicka  <jh@suse.cz>
15008         PR middle-end/56571
15009         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15010         from pseudos.
15011         * emit-rtl.c (verify_rtx_sharing): Likewise.
15012         (copy_insn_1): Likewise.
15013         * rtl.c (copy_rtx): Likewise.
15015 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
15017         PR target/56591
15018         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15019         output_operand_lossage message.
15021 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
15023         PR target/56470
15024         * arm.c (shift_op): Validate RTL pattern on the fly.
15025         (arm_print_operand, case 'S'): Don't use shift_operator to validate
15026         the RTL.
15028 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15030         PR target/56347
15031         * config/pa/pa.md (call_value): Check for calls to powf and direct to
15032         new call patterns that clobber %fr12.
15033         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15034         split and postreload patterns.
15035         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15036         registers %fr12 and %fr12R as call used.
15038 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
15040         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15041         (canon_address, record_store, replace_read, check_mem_read_rtx,
15042         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15043         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15044         rest_of_handle_dse): Likewise.
15046 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
15048         PR middle-end/56524
15049         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15050         Add base_optabs.
15051         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15052         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15053         (save_optabs_if_changed): Replace with...
15054         (init_tree_optimization_optabs): ...this.
15055         * optabs.c (save_optabs_if_changed): Rename to...
15056         (init_tree_optimization_optabs): ...this.  Take the optimization node
15057         as argument.  Do nothing if the base optabs are already correct.
15058         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15059         to recompute optabs.
15060         * function.h (function): Remove optabs field.
15061         * function.c (invoke_set_current_function_hook): Call
15062         init_tree_optimization_optabs.  Use the result to initialize
15063         this_fn_optabs.
15065 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
15067         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15068         if GTMA_HAS_NO_INSTRUMENTATION.
15069         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15070         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15071         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15072         * gimple-pretty-print.c (dump_gimple_transaction): Handle
15073         GTMA_HAS_NO_INSTRUMENTATION.
15075 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
15077         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15078         libasan_preinit.o.
15080 2013-03-08  Marek Polacek  <polacek@redhat.com>
15081             Jakub Jelinek  <jakub@redhat.com>
15083         PR tree-optimization/56478
15084         * predict.c (is_comparison_with_loop_invariant_p): Change the
15085         type of loop_step to tree.
15086         (predict_loops): Adjust.
15087         (predict_iv_comparison): Perform the computations on double_ints.
15089 2013-03-08  Richard Biener  <rguenther@suse.de>
15091         PR tree-optimization/56570
15092         * tree-cfg.c (verify_expr_location_1): Verify locations for
15093         DECL_DEBUG_EXPR.
15094         * tree-sra.c (create_access_replacement): Strip locations
15095         from DECL_DEBUG_EXPRs.
15097 2013-03-08  Richard Biener  <rguenther@suse.de>
15099         * tree-inline.c (expand_call_inline): Do not associate
15100         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15101         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15103 2013-03-08  Richard Biener  <rguenther@suse.de>
15105         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15106         or block changes with -Og.  Fix for location / block encoding
15107         changes and PHI arguments with locations.
15109 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
15111         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15112         for all counters.
15113         (struct output_info): Likewise.
15114         (register_overhead): Remove bad gcc_assert.
15115         (bitmap_find_bit): If there is only a single bitmap element, do not
15116         count a miss as a search.
15117         (print_statistics): Update for counter type changes.
15118         (dump_bitmap_statistics): Likewise.  Print headers such that they
15119         are properly lined up with the printed counters.
15121 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
15123         PR tree-optimization/56559
15124         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15125         check that it has only a single use.
15127 2013-03-07  Richard Biener  <rguenther@suse.de>
15129         * doc/invoke.texi (fwhole-program): Discourage use in combination
15130         with -flto.
15132 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
15134         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15136         PR tree-optimization/56539
15137         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15138         instead of GSI_CONTINUE_LINKING as last argument to
15139         force_gimple_operand_gsi.  Adjust function comment.
15141         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15142         aarch64-cores.def.
15144         PR middle-end/56548
15145         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15146         promoted mode, convert the result back to the original mode.
15148 2013-03-06  Richard Biener  <rguenther@suse.de>
15150         PR middle-end/56294
15151         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15152         (insert_updated_phi_nodes_compare_uids): New function.
15153         (update_ssa): Sort symbols_to_rename after UID before
15154         traversing it to insert PHI nodes.
15156 2013-03-06  Richard Biener  <rguenther@suse.de>
15158         PR middle-end/50494
15159         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15160         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15162         Revert
15163         2013-02-13  Richard Biener  <rguenther@suse.de>
15165         PR lto/50494
15166         * varasm.c (output_constant_def_1): Get the decl representing
15167         the constant as argument.
15168         (output_constant_def): Wrap output_constant_def_1.
15169         (make_decl_rtl): Use output_constant_def_1 with the decl
15170         representing the constant.
15171         (build_constant_desc): Optionally re-use a decl already
15172         representing the constant.
15173         (tree_output_constant_def): Adjust.
15175 2013-03-06  Joey Ye  <joey.ye@arm.com>
15177         PR lto/50293
15178         * gcc.c (convert_white_space): New function.
15179         (main): Handles white space in function name.
15181 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
15183         PR target/56529
15184         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15185         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
15186         to SH_DIV_CALL_TABLE for TARGET_SH2.
15187         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15188         list.
15189         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15190         call-table options.
15192 2013-03-05  Sterling Augustine  <saugustine@google.com>
15193             Cary Coutant  <ccoutant@google.com>
15195         PR debug/55364
15196         * dwarf2out.c (resolve_addr): Don't call
15197         remove_loc_list_addr_table_entries a second time for the same
15198         expression.
15200 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15202         PR debug/56510
15203         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
15204         (avoid_complex_debug_insns): New function.
15205         (expand_debug_locations): Call it.
15207         PR rtl-optimization/56484
15208         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
15209         lifetimes of hard registers on small register class machines.
15211 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15213         * config/microblaze/microblaze-protos.h: Rename
15214         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
15215         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
15216         fast_interrupt.
15217         (microblaze_fast_interrupt_function_p): New function.
15218         (microblaze_is_interrupt_handler): Rename to
15219         microblaze_is_interrupt_variant and add fast_interrupt check.
15220         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
15221         (save_restore_insns): Likewise.
15222         (compute_frame_size): Likewise.
15223         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
15224         (microblaze_globalize_label): Likewise.
15225         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
15226         * config/microblaze/microblaze.md: Use wrapper
15227         microblaze_is_interrupt_variant.
15229 2013-03-05  Kai Tietz  <ktietz@redhat.com>
15231         * sdbout.c (sdbout_one_type): Switch to current function's section
15232         supporting cold/hot.
15234 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15236         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
15237         -mxl-reorder.
15239 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15241         PR middle-end/56461
15242         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
15243         if VALGRIND_GET_VBITS is defined, temporarily make object
15244         memory all defined, and restore previous valgrind addressability
15245         and definability afterwards.  Free this_object at the end.
15247         PR middle-end/56461
15248         * lra.c (lra): Call lra_clear_live_ranges if live_p,
15249         right before calling lra_create_live_ranges, also call it
15250         when clearing live_p.  Only call lra_clear_live_ranges
15251         at the end if live_p.
15253         PR middle-end/56461
15254         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
15256 2013-03-05  Richard Biener  <rguenther@suse.de>
15258         PR tree-optimization/56521
15259         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
15260         value-id.
15262 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
15264         PR c++/55135
15265         * except.h (remove_unreachable_eh_regions): New prototype.
15266         * except.c (remove_eh_handler_splicer): New function, split out
15267         of remove_eh_handler.
15268         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
15269         warning about running it on many EH regions one at a time.
15270         (remove_unreachable_eh_regions_worker): New function, walk the
15271         EH tree in depth-first order and remove non-marked regions.
15272         (remove_unreachable_eh_regions): New function.
15273         * tree-eh.c (mark_reachable_handlers): New function, split out
15274         from remove_unreachable_handlers.
15275         (remove_unreachable_handlers): Use mark_reachable_handlers and
15276         remove_unreachable_eh_regions.
15277         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
15278         and remove_unreachable_eh_regions.
15280 2013-03-05  Richard Biener  <rguenther@suse.de>
15282         PR middle-end/56525
15283         * loop-init.c (fix_loop_structure): Remove loops in two stages,
15284         not freeing them until the end.
15286 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15288         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
15290 2013-03-05  Richard Biener  <rguenther@suse.de>
15292         PR tree-optimization/56270
15293         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
15294         of loads after scheduling an SLP instance.
15296 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15298         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
15299         tic6x.exp.
15300         (check_gcc_parallelize): Run guality.exp as a separate job from
15301         vect.exp with unsorted.exp and $(dg_target_exps) separately from
15302         struct-layout-1.exp with stackalign.exp.
15304         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
15306         PR middle-end/56461
15307         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
15308         load_index sbitmap even if some bit in it isn't set.
15310         PR middle-end/56461
15311         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
15312         (discover_iteration_bound_by_body_walk): Change queues to
15313         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
15314         spelling in comment.  Call safe_push on queues[bound_index] directly.
15315         Release queues[queue_index] in every iteration unconditionally.
15316         Release bounds vector.
15318         PR middle-end/56461
15319         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
15320         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
15321         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
15322         inner_phis vector.
15324 2013-03-05  Richard Biener  <rguenther@suse.de>
15326         PR lto/56515
15327         * tree-inline.c (remap_blocks_to_null): New function.
15328         (expand_call_inline): When expanding a call stmt without
15329         an associated block inline remap all callee blocks to NULL.
15331 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15333         PR rtl-optimization/56494
15334         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
15335         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
15336         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
15338         PR middle-end/56461
15339         * sel-sched-ir.c (free_sched_pools): Release
15340         succs_info_pool.stack[succs_info_pool.max_top] vectors too
15341         if succs_info_pool.max_top isn't -1.
15343         PR bootstrap/56509
15344         * opts.c (opts_obstack, opts_concat): Moved to...
15345         * opts-common.c (opts_obstack, opts_concat): ... here.
15347 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
15349         PR middle-end/56461
15350         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
15352 2013-03-04  Martin Jambor  <mjambor@suse.cz>
15354         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
15355         all appropriate places.
15357 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
15359         PR tree-optimization/56424
15360         * ipa-split.c (split_function): Do not set the RSO flag if result is
15361         not by reference and its type is a register type.
15363 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
15365         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
15366         (microblaze_legitimate_pic_operand): Likewise
15367         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
15368         new function microblaze_legitimate_pic_operand
15369         * config/microblaze/microblaze-protos.h
15370         (microblaze_legitimate_pic_operand): Declare.
15372 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15374         * config/microblaze/predicates.md (call_insn_simple_operand):
15375         New predicate for supported rtx code types.
15376         * config/microblaze/microblaze.md (call_internal1): Use
15377         call_insn_simple_operand predicate.
15379 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
15381         PR middle-end/56461
15382         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
15383         partitions.ordered_remove.
15385         PR middle-end/56461
15386         * tree-vect-stmts.c (vectorizable_conversion): Don't call
15387         vec_oprnds0.create (1) for modifier == NONE.
15389         PR middle-end/56461
15390         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
15391         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
15392         vec_oprnds1 right before pushing anything to it for
15393         scalar_shift_arg.
15395         PR middle-end/56461
15396         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
15397         set nbbs to 0 instead of having separate code path.
15398         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
15399         instead of false as last argument if returning NULL.
15401 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
15403         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
15404         the attribute is now called "target" instead of "option".
15405         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
15406         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
15407         attribute/pragma name for TARGET_OPTION_VALID_P and
15408         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
15409         * doc/tm.texi: Regenerated.
15411 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
15413         * config/microblaze/microblaze.c:
15414         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
15415         * config/microblaze/microblaze.h: Add -mxl-reorder to
15416         DRIVER_SELF_SPECS.
15417         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
15418         instructions emitted if TARGET_REORDER.
15419         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
15420         or 0 for -m/-mno case, but initialises as 2 to detect default use case
15421         separately.
15423 2013-03-01  Xinliang David Li  <davidxl@google.com>
15425         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
15426         walk length.
15428 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
15430         PR middle-end/56461
15431         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
15432         vector even when returning true.  Fix up function comment formatting.
15434         PR middle-end/56461
15435         * ira-build.c (ira_loop_nodes_count): New variable.
15436         (create_loop_tree_nodes): Initialize it.
15437         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
15439         PR middle-end/56461
15440         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
15441         method on dr_chain and result_chain.
15442         * tree-vect-stmts.c (vectorizable_store): Only call
15443         result_chain.create if j == 0.
15445         PR middle-end/56461
15446         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
15447         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
15448         before overwriting it.
15450 2013-03-01  Tobias Burnus  <burnus@net-b.de>
15452         * doc/extended.texi (C Extensions): Change order in @menu
15453         to match @node.
15454         (Other MIPS Built-in Functions): Move last MIPS entry before
15455         "picoChip Built-in Functions".
15456         (SH Built-in Functions): Move after RX Built-in Functions.
15457         * doc/gcc.texi (Introduction): Change order in @menu
15458         to match @node.
15459         * doc/md.texi (Constraints): Ditto.
15460         * gty.texi (Type Information): Ditto.
15461         (User-provided marking routines for template types): Make
15462         subsection.
15463         * doc/invoke.texi (AArch64 Options): Move before
15464         "Adapteva Epiphany Options".
15466 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
15467             Jakub Jelinek  <jakub@redhat.com>
15469         PR sanitizer/56454
15470         * asan.c (gate_asan): Lookup no_sanitize_address instead of
15471         no_address_safety_analysis attribute.
15472         * doc/extend.texi (no_address_safety_attribute): Rename to
15473         no_sanitize_address attribute, mention no_address_safety_analysis
15474         attribute as deprecated alias.
15476 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
15478         PR middle-end/56461
15479         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
15480         type to vec<vec<tree> > *.
15481         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
15482         to be vec<tree> instead of vec<tree> *, set vec_defs
15483         to vNULL and call vec_defs.create (number_of_vects), adjust other
15484         uses of vec_defs.
15485         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
15486         vectorizable_condition): Adjust vect_get_slp_defs callers.
15488 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15490         * config/aarch64/aarch64.c
15491         (aarch64_float_const_representable): Remove unused variable.
15493 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15495         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
15497 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15499         * config/aarch64/aarch64-builtins.c
15500         (aarch64_init_simd_builtins): Make static.
15502 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15504         * config/aarch64/aarch64.c
15505         (aarch64_simd_make_constant): Make static.
15507 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15509         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
15510         with no initialization to the RHS of debug statements.
15512 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15514         PR tree-optimization/56294
15515         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
15516         Adjust dumping.
15517         (get_access_replacement): Do not call create_access_replacement.
15518         Assert a replacement exists.
15519         (get_repl_default_def_ssa_name): Create the replacement declaration
15520         itself.
15522 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15524         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
15525         final_end_function.
15527 2013-02-28  Marek Polacek  <polacek@redhat.com>
15529         PR rtl-optimization/56466
15530         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
15531         if we're changing a loop.
15532         (peel_loops_completely): Likewise.
15534 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
15536         PR c++/55813
15537         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
15539 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15541         PR target/56445
15542         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
15543         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
15544         INTX_FTYPE_FX, FX_FTYPE_INTX.
15545         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
15547 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15549         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
15550         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
15551         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
15552         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
15553         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
15554         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
15555         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
15556         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
15557         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
15558         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
15559         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
15560         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15561         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
15562         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
15563         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
15564         (avrxmega6): Increase max flash segments from 5 to 6.
15565         * config/avr/t-multilib: Regenerate.
15566         * config/avr/avr-tables.opt: Regenerate.
15567         * doc/avr-mmcu.texi: Regenerate.
15569 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15571         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
15572         (avr_device_to_arch): Rename to avr_device_to_ld.
15573         (avr_device_to_as): New prototype.
15574         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
15575         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
15576         * config/avr/driver-avr.c (avr_device_to_as): New.
15577         (avr_device_to_arch): Rename to avr_device_to_ld.
15579 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15581         PR middle-end/56461
15582         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
15583         method on dr_chain and result_chain.
15585         PR middle-end/56461
15586         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
15587         pointer_set_destroy on not_executed_last_iteration.
15589         PR middle-end/56461
15590         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
15592         PR middle-end/56461
15593         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
15594         FOR_EACH_DEFINED_FUNCTION when freeing state.
15596         PR middle-end/56461
15597         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
15598         pool_free.
15599         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
15600         overwriting it.
15602         PR middle-end/56461
15603         * ipa-cp.c (decide_whether_version_node): Call vec_free on
15604         known_aggs[i].items and release known_aggs vector.
15606         PR middle-end/56461
15607         * ipa-reference.c (propagate): Free node_info even for alias nodes.
15609 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15611         * config/microblaze/microblaze.c (microblaze_emit_compare):
15612         Use xor for EQ/NE comparisions.
15613         * config/microblaze/microblaze.md (cstoresf4): Add constraints
15614         (cbranchsf4): Adjust operator to comparison_operator.
15616 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15618         PR middle-end/56461
15619         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
15620         vector.
15621         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
15622         vec_safe_push, always update *slot.
15623         (redirect_edge_var_map_clear): Use vec_free.
15624         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
15625         (free_var_map_entry): Use vec_free.
15626         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
15627         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
15629 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
15631         PR middle-end/45472
15632         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
15633         when the may_trap_p bit of the exprs being merged differs.
15634         Reorder tests for speculativeness in the logical and operator.
15636 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15638         * incpath.c (add_standard_paths): Use reconcat instead of concat
15639         where appropriate and avoid leaking memory.
15641         * opts.h: Include obstack.h.
15642         (opts_concat): New prototype.
15643         (opts_obstack): New declaration.
15644         * opts.c (opts_concat): New function.
15645         (opts_obstack): New variable.
15646         (init_options_struct): Call gcc_init_obstack on opts_obstack.
15647         (finish_options): Use opts_concat instead of concat
15648         and XOBNEWVEC instead of XNEWVEC.
15649         * opts-common.c (generate_canonical_option, decode_cmdline_option,
15650         generate_option): Likewise.
15651         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
15652         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
15654         PR target/56455
15655         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
15656         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
15658 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15660         PR middle-end/56461
15661         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
15663 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15665         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
15666         (arm_block_move_unaligned_straight): Likewise.
15667         (arm_adjust_block_mem): Likewise.
15669 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15671         PR target/48901
15672         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
15673         temp, cond and label.
15674         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
15676         PR target/52500
15677         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
15678         * config/c6x/c6x.h (dbx_register_map): Update declaration.
15680         PR target/52501
15681         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
15682         of prologue/epilogue functions.
15684         PR target/52550
15685         * config/tilegx/tilegx.c (tilegx_expand_prologue):
15686         Remove unused variable cfa_offset.
15687         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
15689         PR target/54639
15690         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
15691         type promotion to unsigned.
15693         PR target/54640
15694         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
15695         for HOST_WIDE_INT of 32 bit / same size as int.
15696         (arm_block_move_unaligned_straight): Likewise.
15697         (arm_adjust_block_mem): Likewise.
15699         PR target/54662
15700         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
15701         ALL_CFLAGS.
15703 2013-02-26  Marek Polacek  <polacek@redhat.com>
15705         PR tree-optimization/56426
15706         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
15708 2013-02-26  Richard Biener  <rguenther@suse.de>
15710         PR target/56444
15711         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
15712         unused variable loops.
15714 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15716         PR tree-optimization/56448
15717         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
15718         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
15719         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
15720         later operands of the references, or even first operand for
15721         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
15723         PR tree-optimization/56443
15724         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
15725         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
15726         to type_for_mode langhook.
15728 2013-02-25  Matt Turner  <mattst88@gmail.com>
15730         * doc/invoke.texi: Document r4700.
15732 2013-02-25  Richard Biener  <rguenther@suse.de>
15734         PR tree-optimization/56175
15735         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
15736         split out from ...
15737         (simplify_bitwise_binary): ... here.  Also guard the conversion
15738         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
15740 2013-02-25  Catherine Moore  <clm@codesourcery.com>
15742         Revert:
15743         2013-02-24  Catherine Moore  <clm@codesourcery.com>
15744             Maciej W. Rozycki  <macro@codesourcery.com>
15745             Tom de Vries  <tom@codesourcery.com>
15746             Nathan Sidwell  <nathan@codesourcery.com>
15747             Iain Sandoe  <iain@codesourcery.com>
15748             Nathan Froyd  <froydnj@codesourcery.com>
15749             Chao-ying Fu  <fu@mips.com>
15751         * doc/extend.texi: (micromips, nomicromips, nocompression):
15752         Document new function attributes.
15753         * doc/invoke.texi (minterlink-compressed, mmicromips,
15754         m14k, m14ke, m14kec): Document new options.
15755         (minterlink-mips16): Update documentation.
15756         * doc/md.texi (ZC, ZD): Document new constraints.
15757         * configure.ac (gcc_cv_as_micromips): Check if linker
15758         supports the .set micromips directive.
15759         * configure: Regenerate.
15760         * config.in: Regenerate.
15761         * config/mips/mips-tables.opt: Regenerate.
15762         * config/mips/micromips.md: New file.
15763         * constraints.md (ZC, AD): New constraints.
15764         * config/mips/predicates.md (movep_src_register): New predicate.
15765         (movep_src_operand): New predicate.
15766         (non_volatile_mem_operand): New predicate.
15767         * config/mips/mips.md (multimem): New type.
15768         (length): Differentiate between 17-bit and 18-bit branch offsets.
15769         (MOVEP1, MOVEP2): New mode iterator.
15770         (mov_<load>l): Use ZC constraint.
15771         (mov_<load>r): Likewise.
15772         (mov_<store>l): Likewise.
15773         (mov_<store>r): Likewise.
15774         (*branch_equality<mode>_inverted): Add microMIPS support.
15775         (*branch_equality<mode>): Likewise.
15776         (*jump_absolute): Likewise.
15777         (indirect_jump_<mode>): Likewise.
15778         (tablejump_<mode>): Likewise.
15779         (<optab>_internal): Likewise.
15780         (sibcall_internal): Likewise.
15781         (sibcall_value_internal): Likewise.
15782         (prefetch): Use constraint ZD.
15783         * config/mips/mips.opt (minterlink-compressed): New option.
15784         (minterlink-mips16): Now an alias for minterlink-compressed.
15785         (mmicromips): New option.
15786         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15787         (compare_and_swap_12): Likewise.
15788         (sync_add<mode>): Likewise.
15789         (sync_<optab>_12): Likewise.
15790         (sync_old_<optab>_12): Likewise.
15791         (sync_new_<optab>_12): Likewise.
15792         (sync_nand_12): Likewise.
15793         (sync_old_nand_12): Likewise.
15794         (sync_new_nand_12): Likewise.
15795         (sync_sub<mode>): Likewise.
15796         (sync_old_add<mode>): Likewise.
15797         (sync_old_sub<mode>): Likewise.
15798         (sync_new_add<mode>): Likewise.
15799         (sync_new_sub<mode>): Likewise.
15800         (sync_<optab><mode>): Likewise.
15801         (sync_old_<optab><mode>): Likewise.
15802         (sync_new_<optab><mode>): Likewise.
15803         (sync_nand<mode>): Likewise.
15804         (sync_old_nand<mode>): Likewise.
15805         (sync_new_nand<mode>): Likewise.
15806         (sync_lock_test_and_set<mode>): Likewise.
15807         (test_and_set_12): Likewise.
15808         (atomic_compare_and_swap<mode>): Likewise.
15809         (atomic_exchange<mode>_llsc): Likewise.
15810         (atomic_fetch_add<mode>_llsc): Likewise.
15811         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15812         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15813         (umips_save_restore_pattern_p): Likewise.
15814         (umips_load_store_pair_p): Likewise.
15815         (umips_output_load_store_pair): Likewise.
15816         (umips_movep_target_p): Likewise.
15817         (umips_12bit_offset_address_p): Likewise.
15818         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15819         (mips_base_mips16): Rename this...
15820         (mips_base_compression_flags): ...to this. Update all uses.
15821         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15822         (mips_mips16_decl_p): Delete.
15823         (mips_nomips16_decl_p): Delete.
15824         (mips_get_compress_on_flags): New function.
15825         (mips_get_compress_off_flags): New function.
15826         (mips_get_compress_mode): New function.
15827         (mips_get_compress_on_name): New function.
15828         (mips_get_compress_off_name): New function.
15829         (mips_insert_attributes): Support multiple compression types.
15830         (mips_merge_decl_attributes): Likewise.
15831         (umips_12bit_offset_address_p): New function.
15832         (mips_start_function_definition): Emit .set micromips directive.
15833         (mips_call_may_need_jalx_p): New function.
15834         (mips_function_ok_for_sibcall): Add microMIPS support.
15835         (mips_print_operand_punctuation): Support short delay slots and
15836         compact jumps.
15837         (umips_swm_mask, umips_swm_encoding): New.
15838         (umips_build_save_restore): New function.
15839         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15840         (was_mips16_p): Remove.
15841         (old_compression_mode): New.
15842         (mips_set_compression_mode): New function.
15843         (mips_set_current_function): Add microMIPS support.
15844         (mips_option_override): Likewise.
15845         (umips_save_restore_pattern_p): New function.
15846         (umips_output_save_restore): New function.
15847         (umips_load_store_pair_p_1): New function.
15848         (umips_load_store_pair_p): New function.
15849         (umips_output_load_store_pair_1): New function.
15850         (umips_output_load_store_pair): New function.
15851         (umips_movep_target_p) New function.
15852         (mips_prepare_pch_save): Add microMIPS support.
15853         * config/mips/mips.h (TARGET_COMPRESSION): New.
15854         (TARGET_CPU_CPP_BUILTINS): Update macro
15855         to use new compression flags and to support microMIPS.
15856         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15857         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15858         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15859         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15860         (ASM_SPEC): Support mmicromips and mno-micromips.
15861         (M16STORE_REG_P): New macro.
15862         (MIPS_CALL): Support TARGET_MICROMIPS.
15863         (MICROMIPS_J): New macro.
15864         (mips_base_mips16): Rename this...
15865         (mips_base_compression_flags): ...to this.
15866         (UMIPS_12BIT_OFFSET_P): New macro.
15867         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15868         (MULTILIB_DIRNAMES): Likewise.
15870 2013-02-25  Tom de Vries  <tom@codesourcery.com>
15872         PR rtl-optimization/56131
15873         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
15874         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
15875         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
15877 2013-02-25  Tobias Burnus  <burnus@net-b.de>
15879         * doc/invoke.texi (-fsanitize=): Move from optimization
15880         to debugging options.
15882 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15884         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
15886 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15887             Alexander Monakov  <amonakov@ispras.ru>
15889         PR middle-end/56077
15890         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
15891         flush pending lists also on non-jumps.  Adjust comment.
15893 2013-02-24  Catherine Moore  <clm@codesourcery.com>
15894             Maciej W. Rozycki  <macro@codesourcery.com>
15895             Tom de Vries  <tom@codesourcery.com>
15896             Nathan Sidwell  <nathan@codesourcery.com>
15897             Iain Sandoe  <iain@codesourcery.com>
15898             Nathan Froyd  <froydnj@codesourcery.com>
15899             Chao-ying Fu  <fu@mips.com>
15901         * doc/extend.texi: (micromips, nomicromips, nocompression):
15902         Document new function attributes.
15903         * doc/invoke.texi (minterlink-compressed, mmicromips,
15904         m14k, m14ke, m14kec): Document new options.
15905         (minterlink-mips16): Update documentation.
15906         * doc/md.texi (ZC, ZD): Document new constraints.
15907         * configure.ac (gcc_cv_as_micromips): Check if linker
15908         supports the .set micromips directive.
15909         * configure: Regenerate.
15910         * config.in: Regenerate.
15911         * config/mips/mips-tables.opt: Regenerate.
15912         * config/mips/micromips.md: New file.
15913         * constraints.md (ZC, AD): New constraints.
15914         * config/mips/predicates.md (movep_src_register): New predicate.
15915         (movep_src_operand): New predicate.
15916         (non_volatile_mem_operand): New predicate.
15917         * config/mips/mips.md (multimem): New type.
15918         (length): Differentiate between 17-bit and 18-bit branch offsets.
15919         (MOVEP1, MOVEP2): New mode iterator.
15920         (mov_<load>l): Use ZC constraint.
15921         (mov_<load>r): Likewise.
15922         (mov_<store>l): Likewise.
15923         (mov_<store>r): Likewise.
15924         (*branch_equality<mode>_inverted): Add microMIPS support.
15925         (*branch_equality<mode>): Likewise.
15926         (*jump_absolute): Likewise.
15927         (indirect_jump_<mode>): Likewise.
15928         (tablejump_<mode>): Likewise.
15929         (<optab>_internal): Likewise.
15930         (sibcall_internal): Likewise.
15931         (sibcall_value_internal): Likewise.
15932         (prefetch): Use constraint ZD.
15933         * config/mips/mips.opt (minterlink-compressed): New option.
15934         (minterlink-mips16): Now an alias for minterlink-compressed.
15935         (mmicromips): New option.
15936         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15937         (compare_and_swap_12): Likewise.
15938         (sync_add<mode>): Likewise.
15939         (sync_<optab>_12): Likewise.
15940         (sync_old_<optab>_12): Likewise.
15941         (sync_new_<optab>_12): Likewise.
15942         (sync_nand_12): Likewise.
15943         (sync_old_nand_12): Likewise.
15944         (sync_new_nand_12): Likewise.
15945         (sync_sub<mode>): Likewise.
15946         (sync_old_add<mode>): Likewise.
15947         (sync_old_sub<mode>): Likewise.
15948         (sync_new_add<mode>): Likewise.
15949         (sync_new_sub<mode>): Likewise.
15950         (sync_<optab><mode>): Likewise.
15951         (sync_old_<optab><mode>): Likewise.
15952         (sync_new_<optab><mode>): Likewise.
15953         (sync_nand<mode>): Likewise.
15954         (sync_old_nand<mode>): Likewise.
15955         (sync_new_nand<mode>): Likewise.
15956         (sync_lock_test_and_set<mode>): Likewise.
15957         (test_and_set_12): Likewise.
15958         (atomic_compare_and_swap<mode>): Likewise.
15959         (atomic_exchange<mode>_llsc): Likewise.
15960         (atomic_fetch_add<mode>_llsc): Likewise.
15961         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15962         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15963         (umips_save_restore_pattern_p): Likewise.
15964         (umips_load_store_pair_p): Likewise.
15965         (umips_output_load_store_pair): Likewise.
15966         (umips_movep_target_p): Likewise.
15967         (umips_12bit_offset_address_p): Likewise.
15968         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15969         (mips_base_mips16): Rename this...
15970         (mips_base_compression_flags): ...to this. Update all uses.
15971         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15972         (mips_mips16_decl_p): Delete.
15973         (mips_nomips16_decl_p): Delete.
15974         (mips_get_compress_on_flags): New function.
15975         (mips_get_compress_off_flags): New function.
15976         (mips_get_compress_mode): New function.
15977         (mips_get_compress_on_name): New function.
15978         (mips_get_compress_off_name): New function.
15979         (mips_insert_attributes): Support multiple compression types.
15980         (mips_merge_decl_attributes): Likewise.
15981         (umips_12bit_offset_address_p): New function.
15982         (mips_start_function_definition): Emit .set micromips directive.
15983         (mips_call_may_need_jalx_p): New function.
15984         (mips_function_ok_for_sibcall): Add microMIPS support.
15985         (mips_print_operand_punctuation): Support short delay slots and
15986         compact jumps.
15987         (umips_swm_mask, umips_swm_encoding): New.
15988         (umips_build_save_restore): New function.
15989         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15990         (was_mips16_p): Remove.
15991         (old_compression_mode): New.
15992         (mips_set_compression_mode): New function.
15993         (mips_set_current_function): Add microMIPS support.
15994         (mips_option_override): Likewise.
15995         (umips_save_restore_pattern_p): New function.
15996         (umips_output_save_restore): New function.
15997         (umips_load_store_pair_p_1): New function.
15998         (umips_load_store_pair_p): New function.
15999         (umips_output_load_store_pair_1): New function.
16000         (umips_output_load_store_pair): New function.
16001         (umips_movep_target_p) New function.
16002         (mips_prepare_pch_save): Add microMIPS support.
16003         * config/mips/mips.h (TARGET_COMPRESSION): New.
16004         (TARGET_CPU_CPP_BUILTINS): Update macro
16005         to use new compression flags and to support microMIPS.
16006         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16007         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16008         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16009         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16010         (ASM_SPEC): Support mmicromips and mno-micromips.
16011         (M16STORE_REG_P): New macro.
16012         (MIPS_CALL): Support TARGET_MICROMIPS.
16013         (MICROMIPS_J): New macro.
16014         (mips_base_mips16): Rename this...
16015         (mips_base_compression_flags): ...to this.
16016         (UMIPS_12BIT_OFFSET_P): New macro.
16017         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16018         (MULTILIB_DIRNAMES): Likewise.
16020 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
16022         PR target/52555
16023         * target-globals.c (save_target_globals): For init_reg_sets and
16024         target_reinit remporarily set this_fn_optabs to this_target_optabs.
16026 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
16028         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16029         * config/aarch64/t-aarch64
16030         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16032 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
16034         PR inline-asm/56148
16035         * lra-constraints.c (process_alt_operands): Reload operand
16036         conflicting with earlier clobber only if no more other conflicting
16037         operands.
16039 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
16041         PR sanitizer/56393
16042         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16043         if not linking a shared library.
16045 2013-02-22  Seth LaForge  <sethml@google.com>
16047         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16049 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
16051         * config/arm/arm.md (split for extendsidi): Update condition.
16052         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16053         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16054         (qhs_zextenddi_cstr): Likewise.
16056 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16058         PR middle-end/56420
16059         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16060         avoid signed wrapping.
16061         (expand_mult): Handle properly multiplication by
16062         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
16063         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
16064         in the compiler if coeff is HOST_WIDE_INT_MIN.
16065         (expand_divmod): Don't make ext_op1 static, change it's type to
16066         uhwi.  Avoid undefined behavior in -INTVAL (op1).
16068         PR rtl-optimization/50339
16069         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16070         field.
16071         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16072         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16073         into splitting_ashiftrt field.
16074         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16075         ASHIFTRT.
16076         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16077         choices.
16079 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
16081         PR middle-end/56108
16082         * trans-mem.c (execute_tm_mark): Do not expand transactions that
16083         are sure to go irrevocable.
16085 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
16087         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16088         scalars are valid operands.
16090 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16092         PR tree-optimization/56310
16093         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16094         only matching indices and non-negative final offsets.
16095         (intersect_aggregates_with_edge): Pass src_idx to
16096         agg_replacements_to_vector.  Pass src_idx insstead of index to
16097         intersect_with_agg_replacements.
16099 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16101         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16102         instead of hard-wired defaults.
16104 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
16106         * doc/invoke.texi (MIPS Options): Update documentation of the
16107         floating-point multiply-accumulate instruction restrictions.
16109 2013-02-21  Kostya Serebryany  <kcc@google.com>
16111         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16112         asan_shadow_offset on x86_64 linux.
16114 2013-02-21  Richard Biener  <rguenther@suse.de>
16116         PR tree-optimization/56415
16117         Revert
16118         2013-02-11  Richard Biener  <rguenther@suse.de>
16120         PR tree-optimization/56273
16121         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16122         first VRP run.
16124 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16126         PR bootstrap/56258
16127         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16128         instead of @itemx.
16130         PR inline-asm/56405
16131         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16132         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16134 2013-02-20  Jan Hubicka  <jh@suse.cz>
16136         PR tree-optimization/56265
16137         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16138         when target is referenced for first time.
16140 2013-02-20  Richard Biener  <rguenther@suse.de>
16142         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16143         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16144         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16145         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16146         not return anything.
16147         (rename_ssa_copies): Do not remove unused locals.
16148         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16149         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16150         * passes.c (execute_function_todo): Do not schedule unused locals
16151         removal if cleanup_tree_cfg did something.
16152         * tree-ssa-live.c (remove_unused_locals): Dump statistics
16153         about the number of removed locals.
16155 2013-02-20  Richard Biener  <rguenther@suse.de>
16157         PR tree-optimization/56398
16158         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16160 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16162         PR tree-optimization/55334
16163         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16164         restricted pointers to arrays.
16166 2013-02-20  Richard Biener  <rguenther@suse.de>
16167             Jakub Jelinek  <jakub@redhat.com>
16169         PR tree-optimization/56396
16170         * tree-ssa-ccp.c (n_const_val): New static variable.
16171         (get_value): Return NULL for SSA names we don't have a lattice
16172         entry for.
16173         (ccp_initialize): Initialize n_const_val.
16174         * tree-ssa-copy.c (n_copy_of): New static variable.
16175         (init_copy_prop): Initialize n_copy_of.
16176         (get_value): Return NULL_TREE for SSA names we don't have a
16177         lattice entry for.
16179 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16181         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16183 2013-02-20  Richard Biener  <rguenther@suse.de>
16185         * genpreds.c (write_lookup_constraint): Do not compare first
16186         letter of the constraint again.
16188 2013-02-20  Richard Biener  <rguenther@suse.de>
16190         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16191         and ceil_log2.
16192         (get_use_iv_cost): Terminate hashtable walk when coming across
16193         an empty entry.
16195 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
16197         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
16198         reassociation for avx2 targets.
16200 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16202         * config/microblaze/microblaze.c: microblaze_has_clz = 0
16203         Add version check for v8.10.a to enable microblaze_has_clz
16204         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
16205         version and TARGET_PATTERN_COMPARE check
16206         * config/microblaze/microblaze.md: New clzsi2 instruction
16208 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16210         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
16211         function before branching.
16213 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16215         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
16216         DUMP_INSN_RTX_UID.
16217         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
16219 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16221         PR middle-end/55889
16222         * sel-sched.c: Include ira.h.
16223         (implicit_clobber_conflict_p): New function.
16224         (moveup_expr): Use it.
16225         * Makefile.in (sel-sched.o): Depend on ira.h.
16227 2013-02-19  Richard Biener  <rguenther@suse.de>
16229         PR tree-optimization/56384
16230         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
16231         (vn_hash_type): Split out from ...
16232         (vn_hash_constant_with_type): ... here.
16233         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
16234         (vn_phi_eq): Compare types from vn_phi_s structure.
16235         (vn_phi_lookup): Populate vn_phi_s type.
16236         (vn_phi_insert): Likewise.
16238 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
16240         PR tree-optimization/56350
16241         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
16242         if haven't found reduction or nested cycle operand, rather than
16243         asserting we must find it.
16245         PR tree-optimization/56381
16246         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
16247         to fold_build3.
16249 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
16250             Jakub Jelinek  <jakub@redhat.com>
16252         PR target/52555
16253         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
16254         (swap_optab_enable): Same.
16255         (init_all_optabs): Use argument instead of global.
16256         * tree.h (struct tree_optimization_option): New field target_optabs.
16257         * expr.h (init_all_optabs): Add argument to prototype.
16258         (TREE_OPTIMIZATION_OPTABS): New.
16259         (save_optabs_if_changed): Protoize.
16260         * optabs.h: Declare this_fn_optabs.
16261         * optabs.c (save_optabs_if_changed): New.
16262         Declare this_fn_optabs.
16263         (init_optabs): Add argument to init_all_optabs() call.
16264         * function.c (invoke_set_current_function_hook): Handle per
16265         function optabs.
16266         * function.h (struct function): New field optabs.
16267         * config/mips/mips.c (mips_set_mips16_mode): Handle when
16268         optimization_current_node has changed.
16269         * target-globals.h (save_target_globals_default_opts): Protoize.
16270         * target-globals.c (save_target_globals_default_opts): New.
16272 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16274         PR target/56347
16275         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
16276         registers %fr12 and %fr12R as call used.
16278         PR target/56214
16279         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
16280         and HImode, require all displacements to be an integer multiple of
16281         their mode size.
16282         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
16283         only allow QImode and HImode when reload is in progress and strict is
16284         true.  Likewise for symbolic addresses.  Use base14_operand to check
16285         displacements in REG+BASE addresses.
16287 2013-02-18  Richard Biener  <rguenther@suse.de>
16289         PR tree-optimization/56366
16290         * tree-vect-loop.c (get_initial_def_for_induction): Properly
16291         handle sign-conversion of outer-loop initial induction value.
16293 2013-02-18  Richard Biener  <rguenther@suse.de>
16295         PR middle-end/56349
16296         * cfghooks.c (merge_blocks): If we merge a latch into another
16297         block adjust references to it.
16298         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
16299         (verify_loop_structure): Verify that a recorded latch is in fact
16300         a latch.
16302 2013-02-18  Richard Biener  <rguenther@suse.de>
16304         PR tree-optimization/56321
16305         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
16306         order SSA name release and virtual operand unlinking.
16308 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16310         * config/microblaze/microblaze.md (save_stack_block): Define.
16311         (restore_stack_block): Likewise.
16313 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16315         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
16316         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
16317         * config/microblaze/microblaze.c (microblaze_option_override):
16318         Bail out early for PIC modes when target does not support PIC.
16320 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16322         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
16323         Replace with a microblaze version.
16324         (microblaze_trampoline_init): Adapt for microblaze.
16325         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
16326         microblaze.
16328 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
16329             Dodji Seketeli  <dodji@redhat.com>
16331         PR asan/56330
16332         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
16333         (instrument_mem_region_access): Do not forget to always put
16334         instrumentation of the of 'base' and 'base + len' in a "if (len !=
16335         0) statement, even for cases where either 'base' or 'base + len'
16336         are not instrumented -- because they have been previously
16337         instrumented.  Simplify the logic by putting all the statements
16338         instrument 'base + len' inside a sequence, and then insert that
16339         sequence right before the current insertion point.  Then, to
16340         instrument 'base + len', just get an iterator on that statement.
16341         And do not forget to update the pointer to iterator the function
16342         received as argument.
16344 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
16346         PR rtl-optimization/56348
16347         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
16349 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
16351         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
16352         (clean_graph_dump_file): Pass base to start_graph_dump.
16354 2013-02-14  Richard Henderson  <rth@redhat.com>
16356         PR target/55941
16357         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
16359 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
16361         * collect2-aix.h: Define F_LOADONLY.
16363 2013-02-14  Richard Biener  <rguenther@suse.de>
16365         PR lto/50494
16366         * varasm.c (output_constant_def_1): Get the decl representing
16367         the constant as argument.
16368         (output_constant_def): Wrap output_constant_def_1.
16369         (make_decl_rtl): Use output_constant_def_1 with the decl
16370         representing the constant.
16371         (build_constant_desc): Optionally re-use a decl already
16372         representing the constant.
16373         (tree_output_constant_def): Adjust.
16375 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
16377         Fix an asan crash
16378         * asan.c (instrument_builtin_call):  Really put the length of the
16379         second source argument into src1_len.
16381 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
16383         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
16384         argument.  If it is false, don't create edge from then_bb to
16385         fallthru_bb.
16386         (insert_if_then_before_iter): Pass true to it.
16387         (build_check_stmt): Pass false to it.
16388         (transform_statements): Flush hash table only on extended basic
16389         block boundaries, rather than at the beginning of every bb.
16390         Don't flush hash table on nonfreeing_call_p calls.
16391         * tree-flow.h (nonfreeing_call_p): New prototype.
16392         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
16394 2013-02-13  David S. Miller  <davem@davemloft.net>
16396         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
16398 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
16400         PR target/56184
16401         * ira.c (max_regno_before_ira): Move from ...
16402         (ira): ... here.
16403         (fix_reg_equiv_init): Use max_regno_before_ira instead of
16404         vec_safe_length.
16406 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
16408         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
16410 2013-02-13  Richard Biener  <rguenther@suse.de>
16412         PR lto/56295
16413         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
16414         globals in MEM_REFs.
16416 2013-02-13  Richard Biener  <rguenther@suse.de>
16418         * loop-init.c (loop_optimizer_init): Clear loop state when
16419         re-initializing preserved loops.
16420         * loop-unswitch.c (unswitch_single_loop): Return whether
16421         we unswitched the loop.  Do not verify loop state here.
16422         (unswitch_loops): When we unswitched a loop discover new loops.
16424 2013-02-13  Kostya Serebryany  <kcc@google.com>
16426         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
16427         on x86_64 linux.
16428         * sanitizer.def: Rename __asan_init to __asan_init_v1.
16430 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
16432         Avoid instrumenting duplicated memory access in the same basic block
16433         * Makefile.in (asan.o): Add new dependency on hash-table.h
16434         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
16435         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
16436         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
16437         (free_mem_ref_resources, has_mem_ref_been_instrumented)
16438         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
16439         (get_mem_ref_of_assignment): New functions.
16440         (get_mem_refs_of_builtin_call): Extract from
16441         instrument_builtin_call and tweak a little bit to make it fit with
16442         the new signature.
16443         (instrument_builtin_call): Use the new
16444         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
16445         of is_gimple_builtin_call.
16446         (instrument_derefs, instrument_mem_region_access): Insert the
16447         instrumented memory reference into the hash table.
16448         (maybe_instrument_assignment): Renamed instrument_assignment into
16449         this, and change it to advance the iterator when instrumentation
16450         actually happened and return true in that case.  This makes it
16451         homogeneous with maybe_instrument_assignment, and thus give a
16452         chance to callers to be more 'regular'.
16453         (transform_statements): Clear the memory reference hash table
16454         whenever we enter a new BB, when we cross a function call, or when
16455         we are done transforming statements.  Use
16456         maybe_instrument_assignment instead of instrumentation.  No more
16457         need to special case maybe_instrument_assignment and advance the
16458         iterator after calling it; it's now handled just like
16459         maybe_instrument_call.  Update comment.
16461 2013-02-13  Richard Biener  <rguenther@suse.de>
16463         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
16464         Fix loop discovery code.
16466 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
16468         PR inline-asm/56148
16469         * lra-constraints.c (process_alt_operands): Match early clobber
16470         operand with itself.  Check conflicts with earlyclobber only if
16471         the operand is not reloaded.  Prefer to reload conflicting operand
16472         if earlyclobber and matching operands are the same.
16474 2013-02-12  Richard Biener  <rguenther@suse.de>
16476         PR lto/56297
16477         * lto-streamer-out.c (write_symbol): Do not output symbols
16478         for hard register variables.
16480 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
16482         PR target/54222
16483         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
16484         (umulsidi3_insn, mulsidi3_insn): New insns.
16486 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
16488         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
16489         (struct tune_params): Add vec_costs field.
16490         * config/arm/arm.c (arm_builtin_vectorization_cost)
16491         (arm_add_stmt_cost): New functions.
16492         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
16493         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
16494         (arm_default_vec_cost): New struct of type cpu_vec_costs.
16495         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
16496         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
16497         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
16498         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
16500 2013-02-12  Richard Biener  <rguenther@suse.de>
16502         PR lto/56295
16503         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
16504         decls again if possible.
16506 2013-02-12  Richard Biener  <rguenther@suse.de>
16508         PR middle-end/56288
16509         * tree-ssa.c (verify_ssa_name): Fix check, move
16510         SSA_NAME_IN_FREE_LIST check up.
16512 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
16513             Steven Bosscher   <steven@gcc.gnu.org>
16515         PR rtl-optimization/56151
16516         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
16517         equal to op0 or op1, and last_insn pattern is CODE operation
16518         with MEM dest and one of the operands matches that MEM.
16520 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
16522         * doc/extend.texi: Document Function Multiversioning and "default"
16523         parameter string to target attribute.
16524         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
16525         target attribute parameter is "default".
16526         (ix86_compare_version_priority): Remove checks for target attribute.
16527         (ix86_mangle_function_version_assembler_name): Change error to sorry.
16528         Remove check for target attribute equal to NULL. Add assert.
16529         (ix86_generate_version_dispatcher_body): Change error to sorry.
16531 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
16532             Jack Howarth  <howarth@bromo.med.uc.edu>
16533             Patrick Marlier  <patrick.marlier@gmail.com>
16535         PR libitm/55693
16536         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
16537         define ENDFILE_SPEC as TM_DESTRUCTOR.
16538         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
16540 2013-02-11  Alexander Potapenko  <glider@google.com>
16541             Jack Howarth  <howarth@bromo.med.uc.edu>
16542             Jakub Jelinek  <jakub@redhat.com>
16544         PR sanitizer/55617
16545         * config/darwin.c (cdtor_record): Rename ctor_record.
16546         (sort_cdtor_records): Rename sort_ctor_records.
16547         (finalize_dtors): New routine to sort destructors by
16548         priority before use in assemble_integer.
16549         (machopic_asm_out_destructor): Use finalize_dtors if needed.
16551 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
16553         PR rtl-optimization/56275
16554         * simplify-rtx.c (avoid_constant_pool_reference): Check that
16555         offset is non-negative and less than cmode size before
16556         calling simplify_subreg.
16558 2013-02-11  Richard Biener  <rguenther@suse.de>
16560         PR tree-optimization/56264
16561         * cfgloop.h (fix_loop_structure): Adjust prototype.
16562         * loop-init.c (fix_loop_structure): Return the number of
16563         newly discovered loops.
16564         * tree-cfgcleanup.c (repair_loop_structures): When new loops
16565         are discovered, do a full loop-closed SSA rewrite.
16567 2013-02-11  Richard Biener  <rguenther@suse.de>
16569         PR tree-optimization/56273
16570         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16571         first VRP run.
16572         (check_array_ref): Fix missing newline in dumps.
16573         (search_for_addr_array): Likewise.
16575 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
16577         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
16579 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
16581         PR target/56256
16582         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
16584 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
16586         PR rtl-optimization/56246
16587         * lra-constraints.c (simplify_operand_subreg): Try to reuse
16588         reload pseudo.
16589         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
16590         constraints are satisfied.
16592 2013-02-08  Jeff Law  <law@redhat.com>
16594         PR debug/53948
16595         * emit-rtl.c (reg_is_parm_p): New function.
16596         * regs.h (reg_is_parm_p): New prototype.
16597         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
16598         callee-clobbered registers.
16600 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
16602         PR target/56043
16603         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
16604         If there is no implicit builtin declaration, just return NULL.
16606 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16608         * config/i386/sse.md (FMAMODEM): New mode iterator.
16609         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
16610         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
16612 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16614         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
16615         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
16616         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
16618 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16620         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
16621         (microblaze*-*-elf): Likewise.
16622         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
16623         LINK_SPEC.
16624         * config/microblaze/microblaze-c.c: Add builtin defines for
16625         _LITTLE_ENDIAN and _BIG_ENDIAN.
16626         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
16627         add to TARGET_DEFAULT flags.
16628         Expand ASM_SPEC and LINK_SPEC.
16629         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
16630         * config/microblaze/microblaze.md: Update extendsidi2 and
16631         movdi_internal instructions to use low-order / high-order reg
16632         print_operands.
16633         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
16634         options and inversemask / mask of LITTLE_ENDIAN.
16635         * config/microblaze/t-microblaze: Expand multilib options to
16636         include mlittle-endian (le) and update exceptions patterns.
16638 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
16640         PR rtl-optimization/56195
16641         * lra-constraints.c (get_reload_reg): Don't reuse regs
16642         if they have smaller mode than requested, if they have
16643         wider mode than requested, try to return a SUBREG.
16645         PR tree-optimization/56250
16646         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
16647         if type is unsigned and code isn't MULT_EXPR.
16649 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16651         PR tree-optimization/56064
16652         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
16653         bits according to mode.
16654         * fixed-value.h (fixed_from_double_int)
16655         (const_fixed_from_double_int): Adjust comments.
16657 2013-02-08  Richard Biener  <rguenther@suse.de>
16659         PR lto/56231
16660         * lto-streamer.h (struct data_in): Remove current_file, current_line
16661         and current_col members.
16662         * lto-streamer-out.c (lto_output_location): Stream changed bits
16663         en-block for efficiency.
16664         * lto-streamer-in.c (clear_line_info): Remove.
16665         (lto_input_location): Cache current file, line and column
16666         globally via local statics.  Read changed bits en-block.
16667         (input_function): Do not call clear_line_info.
16668         (lto_read_body): Likewise.
16669         (lto_input_toplevel_asms): Likewise.
16671 2013-02-08  Michael Matz  <matz@suse.de>
16673         PR tree-optimization/52448
16674         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
16675         (nt_call_phase): New static.
16676         (add_or_mark_expr): Only mark accesses with newer phase than any
16677         call seen.
16678         (nonfreeing_call_p): New.
16679         (nt_init_block): Update nt_call_phase, mark blocks as visited.
16680         (nt_fini_block): Keep blocks marked as visited.
16681         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
16683 2013-02-08  Richard Biener  <rguenther@suse.de>
16685         * ira.c (ira): Free broken dominator information.
16687 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16689         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
16691 2013-02-08  Marek Polacek  <polacek@redhat.com>
16693         * cfgloop.c (verify_loop_structure): Add more checking of headers.
16695 2013-02-08  Richard Biener  <rguenther@suse.de>
16697         PR middle-end/56181
16698         * cfgloop.h (flow_loops_find): Adjust.
16699         (bb_loop_header_p): Declare.
16700         * cfgloop.c (bb_loop_header_p): New function split out from ...
16701         (flow_loops_find): ... here.  Adjust function signature,
16702         support incremental loop structure update.
16703         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
16704         * cfgloopmanip.c (fix_loop_structure): Move ...
16705         * loop-init.c (fix_loop_structure): ... here.
16706         (apply_loop_flags): Split out from ...
16707         (loop_optimizer_init): ... here.
16708         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
16709         in incremental mode, only remove dead loops here.
16711 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16713         PR target/54222
16714         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
16715         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
16716         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
16717         (*round<mode>3.libgcc): New insns for fixed-modes.
16718         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
16719         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
16720         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
16721         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
16722         implementations.  Define to __builtin_avr_absFX,
16723         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
16724         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
16725         __builtin_avr_countlsFX, respectively.
16726         * config/avr/avr-c.c (target.h): Include it.
16727         (enum avr_builtin_id): New enum.
16728         (avr_resolve_overloaded_builtin): New static function.
16729         (avr_register_target_pragmas): Use it to set
16730         targetm.resolve_overloaded_builtin.
16731         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
16732         tree nodes used by DEF_BUILTIN.
16733         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
16734         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
16735         <AVR_BUILTIN_xxBITS>: Same.
16737 2013-02-08  Richard Biener  <rguenther@suse.de>
16739         * cfgloop.c (verify_loop_structure): Properly handle
16740         a loop exiting to another loop header.
16741         * ira-int.h (ira_loops): Remove.
16742         * ira.c (ira_loops): Remove.
16743         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
16744         (do_reload): Use loop_optimizer_finalize.
16745         * ira-build.c (create_loop_tree_nodes): Use get_loops and
16746         number_of_loops to access the loop tree.
16747         (more_one_region_p): Likewise.
16748         (finish_loop_tree_nodes): Likewise.
16749         (rebuild_regno_allocno_maps): Likewise.
16750         (mark_loops_for_removal): Likewise.
16751         (mark_all_loops_for_removal): Likewise.
16752         (remove_unnecessary_regions): Likewise.
16753         (ira_build): Likewise.
16754         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
16756 2013-02-08  Richard Biener  <rguenther@suse.de>
16758         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
16759         * ipa-pure-const.c (analyze_function): Avoid calling
16760         mark_irreducible_loops twice.
16761         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
16763 2013-02-07  David S. Miller  <davem@davemloft.net>
16765         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
16766         on 'reg'.
16767         * var-tracking.c (vt_add_function_parameter): Test the presence of
16768         HAVE_window_save properly and do not remap argument registers when
16769         we have a leaf function.
16771 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
16773         PR bootstrap/56227
16774         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
16775         instead of "ll".
16776         * config/i386/i386.c (ix86_print_operand): Ditto.
16778 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16780         * lra-constraints.c (process_alt_operands): Fix recently added comment.
16782 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16784         PR rtl-optimization/56225
16785         * lra-constraints.c (process_alt_operands): Check that reload hard
16786         reg can hold value for strict_low_part.
16788 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16790         PR debug/56154
16791         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
16792         dwarf2out_end_function.
16793         (in_first_function_p, maybe_at_text_label_p,
16794         first_loclabel_num_not_at_text_label): New variables.
16795         (dwarf2out_var_location): In the first function find out
16796         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
16797         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
16798         functions.
16800 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
16802         PR rtl-optimization/56178
16803         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
16804         SUBREG of a register.  Tidy up related block of code.
16805         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
16806         note if the source is a register or a SUBREG of a register.
16808 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16810         PR target/56228
16811         * config/rs6000/rs6000.md (ptrm): New mode attr.
16812         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
16813         call_value_indirect_aix<pttrsize>,
16814         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
16815         m in constraints.
16817 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16819         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
16820         if -bnortl. Convert to strcmp and strncmp.
16822 2013-02-07  Alan Modra  <amodra@gmail.com>
16824         PR target/54009
16825         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
16826         addresses won't wrap when offsetting.
16827         (rs6000_secondary_reload): Provide secondary reloads needed for
16828         wrapping LO_SUM addresses.
16830 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
16832         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
16833         MACH, just __MACH__.
16835 2013-02-06  Richard Biener  <rguenther@suse.de>
16837         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
16838         instead of calling fix_loop_structure.
16840 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
16842         PR middle-end/56217
16843         * omp-low.c (use_pointer_for_field): Return false if
16844         lower_send_shared_vars doesn't generate any copy-out code.
16846 2013-02-06  Tom de Vries  <tom@codesourcery.com>
16848         PR rtl-optimization/56131
16849         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
16850         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
16851         of the label is NULL.  Add comment.
16853 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
16855         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
16857         PR sanitizer/55374
16858         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
16859         (STATIC_LIBTSAN_LIBS): Likewise.
16860         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
16861         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
16862         is defined, don't add anything else beyond that.
16863         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
16864         (LINK_COMMAND_SPEC): Use them.
16866         PR tree-optimization/56205
16867         * tree-stdarg.c (check_all_va_list_escapes): Return true if
16868         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
16869         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
16871 2013-02-05  Richard Biener  <rguenther@suse.de>
16873         PR tree-optimization/53342
16874         PR tree-optimization/53185
16875         * tree-vectorizer.h (vect_check_strided_load): Remove.
16876         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
16877         not disallow peeling for vectorized strided loads.
16878         (vect_check_strided_load): Make static and simplify.
16879         (vect_analyze_data_refs): Adjust.
16880         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
16881         correctly when vectorizing strided loads.
16883 2013-02-05  Richard Biener  <rguenther@suse.de>
16885         * doc/install.texi: Refer to ISL, not PPL.
16887 2013-02-05  Jan Hubicka  <jh@suse.cz>
16889         PR tree-optimization/55789
16890         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
16892 2013-02-05  Jan Hubicka  <jh@suse.cz>
16894         PR tree-optimization/55789
16895         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
16896         the dead call anyway.
16898 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
16900         PR sanitizer/55374
16901         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
16903 2013-02-04  Alexander Potapenko  <glider@google.com>
16904             Jack Howarth  <howarth@bromo.med.uc.edu>
16905             Jakub Jelinek  <jakub@redhat.com>
16907         PR sanitizer/55617
16908         * config/darwin.c (sort_ctor_records): Stabilized qsort
16909         on constructor priority by using original position.
16910         (finalize_ctors): New routine to sort constructors by
16911         priority before use in assemble_integer.
16912         (machopic_asm_out_constructor): Use finalize_ctors if needed.
16914 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
16916         PR libstdc++/54314
16917         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
16918         about visibility on artificial decls.
16919         * config/sol2.c (solaris_assemble_visibility): Likewise.
16921 2013-02-04  Kai Tietz  <ktietz@redhat.com>
16923         PR target/56186
16924         * config/i386/i386.c (function_value_ms_64): Add additional valtype
16925         argument and improve checking of return-argument types for 16-byte
16926         modes.
16927         (ix86_function_value_1): Add additional valtype argument on call
16928         of function_value_64.
16929         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16930         handling infunction_value_64 function.
16932 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
16934         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16936 2013-02-04  Richard Biener  <rguenther@suse.de>
16938         PR tree-optimization/56188
16939         * tree-ssa-structalias.c (label_visit): Consider case with
16940         initially non-empty points-to set.
16941         (perform_var_substitution): Dump node mapping and clean up.
16943 2013-02-04  Richard Guenther  <rguenther@suse.de>
16945         PR lto/56168
16946         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16947         node prevail as last resort.
16948         (lto_symtab_merge_decls): Remove guard on LTRANS here.
16949         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16951 2013-02-04  Richard Biener  <rguenther@suse.de>
16953         PR tree-optimization/56113
16954         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16955         Merge into ...
16956         (equiv_class_lookup_or_add): ... this.
16957         (label_visit): Adjust and fix error in previous patch.
16958         (perform_var_substitution): Adjust.
16960 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
16962         * config/sh/divtab.c: Fix formatting and comments throughout the file.
16963         * config/sh/sh4-300.md: Likewise.
16964         * config/sh/sh4a.md: Likewise.
16965         * config/sh/constraints.md: Likewise.
16966         * config/sh/sh.md: Likewise.
16967         * config/sh/netbsd-elf.h: Likewise.
16968         * config/sh/predicates.md: Likewise.
16969         * config/sh/sh-protos.h: Likewise.
16970         * config/sh/ushmedia.h: Likewise.
16971         * config/sh/linux.h: Likewise.
16972         * config/sh/sh.c: Likewise.
16973         * config/sh/superh.h: Likewise.
16974         * config/sh/elf.h: Likewise.
16975         * config/sh/sh4.md: Likewise.
16976         * config/sh/sh.h: Likewise.
16978 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16980         * config/pa/constraints.md: Adjust unused letters.  Change "T"
16981         constraint to match_test floating_point_store_memory_operand().
16982         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16983         (base14_operand): New.
16984         (floating_point_store_memory_operand): New.
16985         (integer_store_memory_operand): Revise to use base14_operand and
16986         reg_plus_base_memory_operand.
16987         (move_dest_operand): Allow symbolic_memory_operands.
16988         (symbolic_memory_operand): Check for LO_SOM.
16989         (symbolic_operand): Change default case to break.
16990         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16991         CONST_DOUBLE values to be reloaded by putting them into memory when
16992         the destination is a floating point register.
16993         (movdf): Remove code to handle CONST_DOUBLE.
16994         (movsf): Likewise.
16995         (reload_indf_r1): New.
16996         (reload_insf_r1): New.
16997         Consistently use "Q" and "T" constraints with integer and floating
16998         point move instructions, respectively.
16999         (movdi): Remove FAIL.
17000         Change predicate for source operand unamed DImode move from
17001         general_operand to move_src_operand.
17002         (umulsidi3): Change predicate for destination operand to
17003         register_operand.
17004         Likewise for similar unamed patterns.
17005         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17006         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17007         (hppa_legitimize_address): Simplify mask calculation.
17008         (pa_emit_move_sequence): Revised handling of secondary reloads from
17009         REG+D addresses for floating point loads and stores.  Directly handle
17010         loading CONST0_RTX (mode) to a floating point register.
17011         (pa_secondary_reload): Handle reloading DF and SFmode constant values
17012         to floating point registers.  Don't restrict secondary reloads to
17013         floating point registers to integer modes.  Revise some comments and
17014         cleanup some code.
17015         (TARGET_LEGITIMATE_ADDRESS_P): Define.
17016         (pa_legitimate_address_p): New.
17017         (pa_legitimize_reload_address): New.
17018         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17019         (STRICT_REG_OK_FOR_BASE_P): New.
17020         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
17021         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17023 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
17024             Andrew Dixie  <andrewd@gentrack.com>
17026         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17027         flag set.
17029 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
17031         * expmed.c (extract_bit_field_1): Pass the full width of the
17032         structure to get_best_reg_extraction_insn.
17034 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17036         PR target/54601
17037         * configure.ac (use_cxa_atexit): Add AIX.
17038         * configure: Regenerate.
17040         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17042 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
17044         PR debug/54793
17045         * final.c (need_profile_function): New variable.
17046         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17047         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17048         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17049         notes, targetm.asm_out.function_prologue doesn't emit anything,
17050         HAVE_prologue and profiler should be emitted before prologue,
17051         set need_profile_function instead of emitting it.
17052         (final_scan_insn): If need_profile_function, emit
17053         profile_function on the first NOTE_INSN_BASIC_BLOCK or
17054         NOTE_INSN_FUNCTION_BEG note.
17056 2013-02-01  Richard Henderson  <rth@redhat.com>
17058         * config/rs6000/rs6000.md (smulditi3): New.
17059         (umulditi3): New.
17061         * config/alpha/alpha.md (umulditi3): New.
17063 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17065         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17066         (ASM_OUTPUT_ALIGNED_LOCAL): New.
17068 2013-02-01  Richard Biener  <rguenther@suse.de>
17070         PR tree-optimization/56113
17071         * tree-ssa-structalias.c (label_visit): Reduce work for
17072         single-predecessor nodes.
17074 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
17076         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17077         range isn't testing for zero.
17079 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
17081         PR middle-end/56113
17082         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17084 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
17085             Nick Clifton  <nickc@redhat.com>
17087         * config/v850/constraints.md (Q): Define as a memory constraint.
17088         * config/v850/predicates.md (label_ref_operand): New predicate.
17089         (e3v5_shift_operand): New predicate.
17090         (ior_operator): New predicate.
17091         * config/v850/t-v850: Add e3v5 multilib.
17092         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17093         (v850_gen_movdi): Prototype.
17094         * config/v850/v850.c: Add support for e3v5 architecture.
17095         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17096         TARGET_V850E_UP.
17097         (construct_save_jarl): Add e3v5 long JARL support.
17098         (v850_adjust_insn_length): New function.  Adjust length of call
17099         insns when using e3v5 instructions.
17100         (v850_gen_movdi): New function: Generate instructions to move a
17101         DImode value.
17102         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17103         (CPP_SPEC): Define __v850e3v5__ as appropriate.
17104         (TARGET_USE_FPU): Enable for e3v5.
17105         (CONST_OK_FOR_W): New macro.
17106         (ADJUST_INSN_LENGTH): Define.
17107         * config/v850/v850.md (UNSPEC_LOOP): Define.
17108         (attr cpu): Add v850e3v5.
17109         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17110         (movdi): New pattern.
17111         (movdi_internal): New pattern.
17112         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17113         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17114         (cstoresf4): Likewise.
17115         (cstoredf4): Likewise.
17116         (insv): New pattern.
17117         (rotlso3_a): New pattern.
17118         (rotlsi3_b): New pattern
17119         (rotlsi3_v850e3v5): New pattern.
17120         (doloop_begin): New pattern.
17121         (fix_loop_counter): New pattern.
17122         (doloop_end): New pattern.
17123         (branch_normal): Add e3v5 long branch support.
17124         (branch_invert): Likewise.
17125         (branch_z_normal): Likewise.
17126         (branch_z_invert): Likewise.
17127         (branch_nz_normal): Likewise.
17128         (branch_nz_invert): Likewise.
17129         (call_internal_short): Add e3v5 register-indirect JARL support.
17130         (call_internal_long): Likewise.
17131         (call_value_internal_short): Likewise.
17132         (call_value_internal_long): Likewise.
17133         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17134         (mloop): New option.
17135         * config.gcc: Add support for configuring v840e3v5 target.
17136         * doc/invoke.texi: Document new v850 specific command line options.
17138 2013-01-31  Paul Koning  <ni1d@arrl.net>
17140         PR debug/55059
17141         PR debug/54508
17142         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17143         children if parent is a class.
17144         (prune_unused_types_prune): Don't add DW_AT_declaration.
17146 2013-01-31  Richard Biener  <rguenther@suse.de>
17148         PR tree-optimization/56157
17149         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17150         match up operand with SLP child.
17152 2013-01-31  Jason Merrill  <jason@redhat.com>
17154         PR debug/54410
17155         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17156         parameters the first time.
17157         (gen_scheduled_generic_parms_dies): Check completeness here.
17159 2013-01-31  Richard Biener  <rguenther@suse.de>
17161         PR middle-end/53073
17162         * common.opt (faggressive-loop-optimizations): New flag,
17163         enabled by default.
17164         * doc/invoke.texi (faggressive-loop-optimizations): Document.
17165         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17166         infer_loop_bounds_from_undefined by it.
17168 2013-01-31  Richard Biener  <rguenther@suse.de>
17170         PR tree-optimization/56150
17171         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17172         visit virtual operands.
17173         (find_uses_to_rename_bb): Likewise.
17175 2013-01-31  Richard Biener  <rguenther@suse.de>
17177         PR tree-optimization/56150
17178         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17179         mixed store non-store stmts.
17181 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
17183         PR sanitizer/55374
17184         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17185         LIBASAN_EARLY_SPEC is defined.
17186         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17187         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17188         before %o.
17189         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17191         PR c++/55742
17192         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17193         invalid args instead of ICEing on it.
17194         (ix86_valid_target_attribute_tree): Return error_mark_node if
17195         ix86_valid_target_attribute_inner_p failed.
17196         (ix86_valid_target_attribute_p): Return false only if
17197         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
17198         target("default") attribute.
17199         (sorted_attr_string): Change argument from const char * to tree,
17200         merge in all target attribute arguments rather than just one.
17201         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
17202         instead of free.  Avoid using strcat.
17203         (ix86_mangle_function_version_assembler_name): Mangle
17204         target("default") as if no target attribute is present.  Adjust
17205         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
17206         instead of xmalloc and XDELETEVEC instead of free.
17207         (ix86_function_versions): Don't return true if one of the decls
17208         doesn't have target attribute.  If they don't and one of the decls
17209         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
17210         sorted_attr_string caller.  Use XDELETEVEC instead of free.
17211         (ix86_supports_function_versions): Remove.
17212         (make_name): Fix up formatting.
17213         (make_dispatcher_decl): Remove resolver_name and its initialization.
17214         Avoid leaking memory.
17215         (is_function_default_version): Return true if there is
17216         target("default") attribute rather than no target attribute at all.
17217         (make_resolver_func): Avoid leaking memory.
17218         (ix86_generate_version_dispatcher_body): Likewise.
17219         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
17220         * target.def (supports_function_versions): Remove.
17221         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
17222         * doc/tm.texi: Regenerated.
17224 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
17226         PR rtl-optimization/56144
17227         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
17228         for values with side effects.
17230 2013-01-30  Richard Biener  <rguenther@suse.de>
17232         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
17233         (sparseset_pop): Likewise.
17234         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
17235         to be able to use quick_push in the worker loop.
17237 2013-01-30  Marek Polacek  <polacek@redhat.com>
17239         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
17241 2013-01-30  Richard Biener  <rguenther@suse.de>
17243         PR lto/56147
17244         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
17246 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
17248         PR tree-optimization/56064
17249         * fixed-value.c (fixed_from_double_int): New function.
17250         * fixed-value.h (fixed_from_double_int): New prototype.
17251         (const_fixed_from_double_int): New static inline function.
17252         * fold-const.c (native_interpret_fixed): New static function.
17253         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
17254         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
17255         (native_encode_fixed): New static function.
17256         (native_encode_expr) <FIXED_CST>: Use it.
17257         (native_interpret_int): Move double_int worker code to...
17258         * double-int.c (double_int::from_buffer): ...this new static method.
17259         * double-int.h (double_int::from_buffer): Prototype it.
17261 2013-01-30  Richard Biener  <rguenther@suse.de>
17263         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
17264         New pointer-map and obstack.
17265         (init_alias_vars): Allocate pointer-map and obstack.
17266         (delete_points_to_sets): Free them.
17267         (find_what_var_points_to): Cache result.
17268         (find_what_p_points_to): Adjust for changed interface of
17269         find_what_var_points_to.
17270         (compute_points_to_sets): Likewise.
17271         (ipa_pta_execute): Likewise.
17273 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17275         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
17276         * configure: Regenerate.
17277         * config.in: Regenerate.
17278         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
17279         #nobits/#progbits if supported.
17281 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
17283         PR target/56121
17284         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
17285         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
17286         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
17288 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17290         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
17291         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
17293 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17295         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
17296         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
17298 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17300         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
17301         declaration.
17302         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
17303         * config/arm/cortex-a7.md: New bypasses using
17304         arm_mac_accumulator_is_result.
17306 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17308         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
17309         (cortex_a7_neon_mla): Likewise.
17310         (cortex_a7_fpfmad): New reservation.
17311         (cortex_a7_fpmacs): Use ffmas and update required units.
17312         (cortex_a7_fpmuld): Update required units and latency.
17313         (cortex_a7_fpmacd): Likewise.
17314         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
17315         (cortex_a7_neon). Likewise.
17316         (bypass) Update participating units.
17318 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17320         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
17321         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
17322         from fmac to ffma.
17323         * config/arm/vfp11.md (vfp_farith): Use ffmas.
17324         (vfp_fmul): Use ffmad.
17325         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
17326         (cortex_r4_fmacd): Use ffmad.
17327         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
17328         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
17329         (cortex_a9_fmacd): Use ffmad.
17330         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
17331         (cortex_a8_vfp_macd): Use ffmad.
17332         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
17333         (cortex_a5_fpmacd): Use ffmad.
17334         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
17335         (cortex_a15_vfp_macd): Use ffmad.
17336         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
17338 2013-01-29  Jason Merrill  <jason@redhat.com>
17340         PR libstdc++/54314
17341         * varasm.c (default_assemble_visibility): Don't warn about
17342         visibility on artificial decls.
17344 2013-01-29  Richard Biener  <rguenther@suse.de>
17346         PR tree-optimization/56113
17347         * tree-ssa-structalias.c (equiv_class_lookup): Also return
17348         the bitmap leader.
17349         (label_visit): Free duplicate bitmaps and record the leader instead.
17350         (perform_var_substitution): Adjust.
17352 2013-01-29  Richard Biener  <rguenther@suse.de>
17354         PR tree-optimization/55270
17355         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
17356         the CFG, schedule loops for fixup.
17358 2013-01-29  Nick Clifton  <nickc@redhat.com>
17360         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
17361         SP_REG.
17363 2013-01-28  Leif Ekblad  <leif@rdos.net>
17365         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
17366         * config/i386/i386.h (TARGET_RDOS): New macro.
17367         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
17368         * config/i386/i386.c (ix86_option_override_internal): For 64bit
17369         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
17370         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
17371         DEFAULT_LARGE_SECTION_THRESHOLD.
17372         * config/i386/i386.md (R14_REG, R15_REG): New constants.
17373         * config/i386/rdos.h: New file.
17374         * config/i386/rdos64.h: New file.
17376 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
17378         PR other/54814
17379         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
17380         TEST_HARD_REG_BIT.
17382 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
17384         PR rtl-optimization/56117
17385         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
17386         call cselib_lookup_from_insn on the MEM before calling
17387         add_insn_mem_dependence.
17389 2013-01-28  Richard Biener  <rguenther@suse.de>
17391         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
17392         to a stmt that didn't have one.
17393         (copy_phis_for_bb): Likewise for PHI arguments.
17394         (copy_debug_stmt): Likewise for debug stmts.
17396 2013-01-28  Richard Biener  <rguenther@suse.de>
17398         PR tree-optimization/56034
17399         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
17400         (partition_builtin_p): Adjust.
17401         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
17402         it is the last partition.
17403         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
17404         up the vertex for the definition.
17405         (classify_partition): Classify whether a partition is a
17406         PKIND_REDUCTION, thus has uses outside of the loop.
17407         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
17408         Merge all PKIND_REDUCTION partitions into the last partition.
17409         (tree_loop_distribution): Seed partitions from reductions as well.
17411 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
17413         PR tree-optimization/56125
17414         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
17415         pow(x,c) into sqrt(x) * powi(x, n/2) or
17416         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
17417         optimizing for size.
17418         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
17419         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
17420         integer.
17422         PR tree-optimization/56094
17423         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
17424         to UNKNOWN_LOCATION while gimplifying expr.
17426 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
17428         PR target/56114
17429         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
17430         operand 0 in movabs insn template for -masm=intel asm alternative.
17431         (*movabs<mode>_2): Ditto for operand 1.
17433 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
17435         PR target/54663
17436         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
17437         of microblaze-c.o
17439 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17441         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
17442         tm_file.
17444 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17446         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
17447         Undef to avoid warning.
17449 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
17451         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
17452         * configure: Regenerate.
17454 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
17456         PR tree-optimization/56098
17457         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
17458         for stmts with volatile ops.
17459         (cond_store_replacement): Don't optimize if assign has volatile ops.
17460         (cond_if_else_store_replacement_1): Don't optimize if either
17461         then_assign or else_assign have volatile ops.
17462         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
17463         volatile ops.
17465 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17467         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
17469 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17471         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
17472         missing ':' in asm example.
17474 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
17476         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
17477         entries into lane and laneq entries.
17478         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
17479         Remove AdvSIMD scalar modes.
17480         (aarch64_sq<r>dmulh_laneq<mode>): New.
17481         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
17482         modes.
17483         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
17484         builtin implementations to relfect changes in RTL in aarch64-simd.md.
17485         * config/aarch64/iterators.md (VCOND): New.
17486         (VCONQ): New.
17488 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17490         PR target/54222
17491         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
17492         Add NULL LIBNAME argument to existing definitions.
17493         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
17494         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
17495         * config/avr/avr.c (DEF_BUILTIN): Same.
17496         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
17497         (avr_expand_builtin): Expand to a vanilla call if a libgcc
17498         implementation is available (DECL_ASSEMBLER_NAME is set).
17499         (avr_fold_absfx): New static function.
17500         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
17501         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
17502         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
17503         AVR_BUILTIN_ABSLLK.
17504         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
17505         (abshk, absk, abslk, absllk): Provide as static inline functions.
17507 2013-01-25  Marek Polacek  <polacek@redhat.com>
17509         PR tree-optimization/56035
17510         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
17512 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
17514         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
17515         (*movtf_internal_rex64): Add (!o,C) alternative
17516         (*movxf_internal_rex64): Ditto.
17517         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
17519 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
17521         * doc/invoke.texi: fix typo.
17522         * doc/objc.texi: fix typo.
17524 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
17526         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
17527         for the first two alternatives.
17529 2013-01-24  Diego Novillo  <dnovillo@google.com>
17531         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
17532         (ggc-zone.o): Remove.
17533         * configure.ac: Remove option --with-gc.
17534         * configure: Re-generate.
17535         * doc/install.texi: Remove documentation for --with-gc.
17536         * gengtype.c (write_enum_defn): Remove.  Update all users.
17537         (write_Types_process_field): Remove generation of gt_e_* argument.
17538         (output_type_enum): Remove.  Update all users.
17539         (write_enum_defn): Remove.  Update all users.
17540         (enum alloc_zone): Remove.  Update all users.
17541         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
17542         * ggc-common.c (ggc_splay_alloc): Remove first argument.
17543         Update all callers.
17544         (struct ptr_data): Remove field TYPE.  Update all users.
17545         (gt_pch_note_object): Remove argument TYPE.  Update all users.
17546         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
17547         Update all users.
17548         * ggc-none.c (ggc_alloc_typed_stat): Remove.
17549         (struct alloc_zone): Remove.
17550         (ggc_internal_alloc_zone_stat): Remove.
17551         (ggc_internal_cleared_alloc_zone_stat): Remove.
17552         * ggc-page.c (ggc_alloc_typed_stat): Remove.
17553         (ggc_pch_count_object): Remove last argument.  Update all users.
17554         (ggc_pch_alloc_object): Remove last argument.  Update all users.
17555         (struct alloc_zone): Remove.
17556         * ggc-zone.c: Remove.
17557         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
17558         (struct alloc_zone): Remove.
17559         (ggc_alloc_typed_stat): Remove.
17560         (ggc_alloc_typed): Remove.
17561         (ggc_splay_alloc): Remove first argument.
17562         (rtl_zone): Remove.  Update all users.
17563         (tree_zone): Remove.  Update all users.
17564         (tree_id_zone): Remove.  Update all users.
17565         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
17566         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
17567         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
17568         * tree-ssanames.c: Remove references to zone allocator in comments.
17570 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
17572         * config/avr/avr.c (avr_out_fract): Make register numbers that
17573         might be outside of source operand signed.
17575 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
17577         * config/i386/constraints.md (Yf): New constraint.
17578         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
17579         of f constraint to conditionaly disable x87 register preferences.
17580         (*movdf_internal): Ditto.
17581         (*movsf_internal): Ditto.
17583 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
17585         PR inline-asm/55934
17586         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
17587         that have operands with impossible constraints.
17588         Add a FIXME for a speed-up opportunity.
17589         * lra-constraints.c (process_alt_operands): Verify that a class
17590         selected from constraints on asms is valid for the operand mode.
17591         (curr_insn_transform): Remove incorrect comment.
17593 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
17595         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
17596         TOC operand is a valid symbol ref in the constant pool.
17598 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17600         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
17602 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
17604         PR target/54222
17605         * config/avr/stdfix.h: New file.
17606         * t-avr (stdfix-gcc.h): New rule to build it.
17607         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
17609 2013-01-23  Kostya Serebryany  <kcc@google.com>
17611         * config/darwin.h: remove dependency on
17612         CoreFoundation (asan on Mac OS).
17614 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
17616         PR target/49069
17617         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
17618         instead of cmpdi_operand for first comparison operand.
17619         Don't assert that comparison operands aren't both constants.
17621 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17623         * doc/install.texi (Downloading the Source): Update references to
17624         downloading separate components.
17626 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17628         * doc/extend.texi (__int128): Improve grammar.
17630 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
17632         PR target/56028
17633         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
17634         alternative to (o,r).
17635         (*movdi_internal_rex64): Remove (!o,n) alternative.
17636         (DImode immediate->memory splitter): Remove.
17637         (DImode immediate->memory peephole2): Remove.
17638         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
17639         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
17640         alternative to (!o,*r).
17641         (*movtf_internal_sse): New pattern.
17642         (*movxf_internal_rex64): New pattern.
17643         (*movxf_internal): Disable for TARGET_64BIT.
17644         (*movdf_internal_rex64): Remove (!o,F) alternative.
17646 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
17648         PR middle-end/56074
17649         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
17650         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
17651         * tree-vect-loop-manip.c (find_loop_location): Also ignore
17652         stmt locations where LOCATION_LOCUS of the stmt location is
17653         UNKNOWN_LOCATION or BUILTINS_LOCATION.
17655         PR target/55686
17656         * config/i386/i386.md (UNSPEC_STOS): New.
17657         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
17658         *strsetqi_1): Add UNSPEC_STOS.
17660 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
17662         PR c++/56067
17663         * doc/invoke.texi: Remove left over -Wsynth example.
17665 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17667         PR tree-optimization/56051
17668         * fold-const.c (fold_binary_loc): Don't fold
17669         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
17670         a narrowing conversion, or widening conversion from signed
17671         to unsigned.
17673 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17675         PR rtl-optimization/56023
17676         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
17677         dependent on debug instruction.
17679 2013-01-21  Martin Jambor  <mjambor@suse.cz>
17681         PR middle-end/56022
17682         * function.c (allocate_struct_function): Call
17683         invoke_set_current_function_hook earlier.
17685 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17687         * reload1.c (init_reload): Only initialize reload_obstack
17688         during the first call.
17690 2013-01-21  Marek Polacek  <polacek@redhat.com>
17692         * cfgloop.c (verify_loop_structure): Fix up grammar.
17694 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
17696         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
17697         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
17699 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17701         PR target/56058
17702         * config/arm/marvell-pj4.md: Update copyright year.
17703         Fix up use of alu to alu_reg and simple_alu_imm.
17705 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17707         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
17709 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
17711         PR target/55433
17712         * lra-constraints.c (curr_insn_transform): Don't reuse original
17713         insn for secondary memory move when memory mode should be different.
17715 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17717         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
17718         atomic_storedi_1): New patterns.
17720 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17722         btver2 pipeline descriptions.
17723         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
17724         descriptions.
17725         * config/i386/i386.md (btver2_decode): New type attributes.
17726         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
17727         type attributes.
17728         * config/i386/btver2.md: New file describing btver2 pipelines.
17730 2013-01-19  Andrew Pinski  <apinski@cavium.com>
17732         PR tree-optimization/52631
17733         * tree-ssa-sccvn (visit_use): Before looking up the original
17734         statement, try looking up the simplified expression.
17736 2013-01-19  Anthony Green  <green@moxielogic.com>
17738         * config/moxie/moxie.c (moxie_expand_prologue): Set
17739         current_function_static_stack_size.
17741 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17743         PR tree-optimization/56029
17744         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
17745         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
17747 2013-01-18  Sharad Singhai  <singhai@google.com>
17749         PR tree-optimization/55995
17750         * dumpfile.c (dump_loc): Print location only if available.
17751         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
17753 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
17755         PR target/55433
17756         * lra-constraints.c (curr_insn_transform): Reuse original insn for
17757         secondary memory move.
17758         (inherit_reload_reg): Use rclass instead of cl for
17759         check_secondary_memory_needed_p.
17761 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17763         PR middle-end/56015
17764         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
17765         the case where writing real complex part of target modifies op1.
17767 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
17769         * config/aarch64/aarch64-simd.md
17770         (aarch64_vcond_internal<mode>): Handle unordered cases.
17771         * config/aarch64/iterators.md (v_cmp_result): New.
17773 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
17774             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17776         * config/arm/marvell-pj4.md: New file.
17777         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
17778         * config/arm/arm.md (generic_sched): Add marvell_pj4.
17779         (generic_vfp): Likewise.
17780         * config/arm/arm-cores.def: Add marvell-pj4.
17781         * config/arm/arm-tune.md: Regenerate.
17782         * config/arm/arm-tables.opt: Regenerate.
17783         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
17784         * doc/invoke.texi: Document marvell-pj4.
17786 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
17788         * config/aarch64/arm_neon.h: Map scalar types to standard types.
17790 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17792         PR debug/54114
17793         PR debug/54402
17794         PR debug/49888
17795         * var-tracking.c (negative_power_of_two_p): New.
17796         (global_get_addr_cache, local_get_addr_cache): New.
17797         (get_addr_from_global_cache, get_addr_from_local_cache): New.
17798         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
17799         heading comment.
17800         (vt_stack_offset_p): Remove.
17801         (vt_canon_true_dep): Always canonicalize loc's address.
17802         (clobber_overlapping_mems): Make sure we have a MEM.
17803         (local_get_addr_clear_given_value): New.
17804         (val_reset): Clear local cached entries.
17805         (compute_bb_dataflow): Create and release the local cache.
17806         Disable duplicate MEMs clobbering.
17807         (emit_notes_in_bb): Clobber MEMs likewise.
17808         (vt_emit_notes): Create and release the local cache.
17809         (vt_initialize, vt_finalize): Create and release the global
17810         cache, respectively.
17811         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
17813 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17815         PR libmudflap/53359
17816         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
17817         not found in the symtab.
17819 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17821         PR debug/56006
17822         PR rtl-optimization/55547
17823         PR rtl-optimization/53827
17824         PR debug/53671
17825         PR debug/49888
17826         * alias.c (offset_overlap_p): New, factored out of...
17827         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
17828         the conservative special case for symbolic constants.  Don't
17829         adjust zero sizes on alignment.
17831 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
17833         PR rtl-optimization/52573
17834         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
17835         REG_UNUSED for the same register.
17837 2013-01-17  Richard Biener  <rguenther@suse.de>
17838             Marek Polacek  <polacek@redhat.com>
17840         PR rtl-optimization/55833
17841         * loop-unswitch.c (unswitch_loops): Move loop verification...
17842         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
17843         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
17844         Set it to true when we're removing a loop from hierarchy tree in
17845         an irreducible region.
17846         (fix_bb_placements): Adjust caller.
17847         (fix_loop_placements): Likewise.
17849 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
17851         * config/avr/builtins.def (DEF_BUILTIN): Factor out
17852         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
17853         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
17854         Remove ID.  Adjust comments.
17855         * config/avr/avr-c.c (avr_builtin_name): Remove.
17856         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
17857         * config/avr/avr.c (avr_tolower): New static function.
17858         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
17859         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
17860         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
17861         default expansion.
17863 2013-01-17  Jan Hubicka  <jh@suse.cz>
17865         PR tree-optimization/55273
17866         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
17868 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
17870         PR target/55981
17871         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
17872         store through atomic_store<mode>_1.
17873         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
17875 2013-01-17  Martin Jambor  <mjambor@suse.cz>
17877         PR tree-optimizations/55264
17878         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
17879         for virtual methods.
17880         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
17881         virtual methods before inlining is over.
17882         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
17883         virtual functions.
17884         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
17885         non-virtual.
17887 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
17889         PR rtl-optimization/56005
17890         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
17891         pending reads for prefetch.
17893 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
17895         * config/aarch64/aarch64.md
17896         (*cstoresi_neg_uxtw): New pattern.
17897         (*cmovsi_insn_uxtw): New pattern.
17898         (*<optab>si3_uxtw): New pattern.
17899         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
17900         (*<optab>si3_insn_uxtw): New pattern.
17901         (*bswapsi2_uxtw): New pattern.
17903 2013-01-16  Richard Biener  <rguenther@suse.de>
17905         * tree-inline.c (tree_function_versioning): Remove set but
17906         never used variable.
17908 2013-01-16  Richard Biener  <rguenther@suse.de>
17910         PR tree-optimization/55964
17911         * tree-flow.h (rename_variables_in_loop): Remove.
17912         (rename_variables_in_bb): Likewise.
17913         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
17914         (copy_loop_before): Adjust and delete update-ssa status.
17915         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
17916         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
17917         (rename_variables_in_loop): Remove.
17918         (slpeel_update_phis_for_duplicate_loop): Likewise.
17919         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
17920         use available cfg machinery instead of duplicating it.
17921         Update PHI nodes and perform poor-mans SSA update here.
17922         (slpeel_tree_peel_loop_to_edge): Adjust.
17924 2013-01-16  Richard Biener  <rguenther@suse.de>
17926         PR tree-optimization/54767
17927         PR tree-optimization/53465
17928         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17929         (vrp_visit_phi_node): For PHI arguments coming via backedges
17930         drop all symbolical range information.
17931         (execute_vrp): Compute backedges.
17933 2013-01-16  Richard Biener  <rguenther@suse.de>
17935         * doc/install.texi: Update CLooG and ISL requirements to
17936         0.18.0 and 0.11.1.
17938 2013-01-16  Christian Bruel  <christian.bruel@st.com>
17940         PR target/55301
17941         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17942         (broken_move): Handle UNSPECV_SP_SWITCH_B.
17943         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17945 2013-01-16  DJ Delorie  <dj@redhat.com>
17947         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17948         (UNSPECV_SP_SWITCH_E): New.
17949         (sp_switch_1): Change to an unspec.
17950         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
17951         replace $r15.
17953 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
17955         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17956         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17957         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17958         (expand_mem_thread_fence): Ditto.
17959         (expand_mem_signal_fence): Ditto.
17960         (expand_atomic_load): Ditto.
17961         (expand_atomic_store): Ditto.
17963 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
17965         PR rtl-optimization/55547
17966         PR rtl-optimization/53827
17967         PR debug/53671
17968         PR debug/49888
17969         * alias.c (memrefs_conflict_p): Set sizes to negative after
17970         AND adjustments.
17972 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
17974         PR target/55940
17975         * function.c (thread_prologue_and_epilogue_insns): Always
17976         add crtl->drap_reg to set_up_by_prologue.set, even if
17977         stack_realign_drap is false.
17979 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
17981         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17982         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17983         *call): Fix indention.
17985 2013-01-15  Tom de Vries  <tom@codesourcery.com>
17987         PR target/55876
17988         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17989         Update comment.
17991 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
17993         PR rtl-optimization/55153
17994         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17996 2013-01-15  Martin Jambor  <mjambor@suse.cz>
17998         PR tree-optimization/55920
17999         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18000         accesses as grp_to_be_debug_replaced.
18002 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18004         PR tree-optimization/55920
18005         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18006         there is non-useless type conversion needed from debug rhs to lhs,
18007         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18009 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
18010             Mikael Pettersson  <mikpe@it.uu.se>
18012         PR target/43961
18013         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18014         Thumb.
18015         (ASM_OUTPUT_CASE_LABEL): Remove.
18016         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18017         * final.c (shorten_branches): Update alignment of labels before
18018         jump tables if CASE_VECTOR_SHORTEN_MODE.
18020 2013-01-15  Richard Biener  <rguenther@suse.de>
18022         PR bootstrap/55961
18023         * system.h: Do not include gmp.h for building host tools.
18025 2013-01-15  Richard Biener  <rguenther@suse.de>
18027         PR middle-end/55882
18028         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18029         account for bitpos when computing alignment.
18031 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
18033         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18034         (ix86_target_macros_internal): Likewise.
18036         * config/i386/i386.c (m_CORE2I7): Removed.
18037         (m_CORE_HASWELL): New macro.
18038         (m_CORE_ALL): Likewise.
18039         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18040         (initial_ix86_arch_features): Likewise.
18041         (processor_target_table): Initializations for Core avx2.
18042         (cpu_names): New names "core-avx2".
18043         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18044         PROCESSOR_CORE_HASWELL.
18045         (ix86_issue_rate): New case.
18046         (ia32_multipass_dfa_lookahead): Likewise.
18047         (ix86_sched_init_global): Likewise.
18049         * config/i386/i386.h (TARGET_HASWELL): New macro.
18050         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18051         (processor_type): New PROCESSOR_HASWELL.
18053 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18055         PR tree-optimization/55955
18056         * tree-vect-loop.c (vectorizable_reduction): Give up early on
18057         *SHIFT_EXPR and *ROTATE_EXPR codes.
18059         PR tree-optimization/48766
18060         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18061         -ftrapv disable -fwrapv.
18063 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18065         PR target/55974
18066         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18067         etc. to 1 and not to __flash.
18068         Use LL suffix for __INT24_MAX__ with -mint8.
18069         Use ULL suffix for __UINT24_MAX__ with -mint8.
18071 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18073         * config/avr/avr-arch.h
18074         (struct base_arch_s): Use typedef avr_arch_t instead.
18075         (struct arch_info_s): Use typedef avr_arch_info_t instead.
18076         (struct mcu_type_s): Use typedef avr_mcu_t instead.
18077         * config/avr/avr.c: Same.
18078         * config/avr/avr-devices.c: Same.
18079         * config/avr/driver-avr.c: Same.
18080         * config/avr/gen-avr-mmcu-texi.c: Same.
18081         * config/avr/avr-mcus.def: Adjust comment.
18083 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
18085         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18086         * config/aarch64/iterators.md (VALLDI): New.
18088 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
18089             Andi Kleen  <ak@linux.intel.com>
18091         PR target/55948
18092         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18093         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18094         memmodel flag.
18096 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18098         * config/avr/avr-stdint.h: Remove trailing blanks.
18099         * config/avr/avr-log.h: Same.
18100         * config/avr/avr-arch.h: Same.
18101         * config/avr/avr-devices.c: Same.
18102         * config/avr/avr-dimode.md: Same.
18103         * config/avr/predicates.md: Same.
18104         * config/avr/avr-c.c: Same.  And fix typo.
18106         * config/avr/avr-protos.h: Same.  And:
18107         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18108         (init_cumulative_args): Rename to avr_init_cumulative_args.
18109         (expand_prologue): Rename to avr_expand_prologue.
18110         (expand_epilogue): Rename to avr_expand_epilogue.
18111         (adjust_insn_length): Rename to avr_adjust_insn_length.
18112         (notice_update_cc): Rename to avr_notice_update_cc.
18113         (final_prescan_insn): Rename to avr_final_prescan_insn.
18114         * config/avr/avr.c: Same.
18115         * config/avr/avr.h: Same.
18116         * config/avr/avr.md: Remove trailing blanks.
18117         (prologue): Use avr_expand_prologue.
18118         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18120 2013-01-14  Richard Biener  <rguenther@suse.de>
18122         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18123         verify_location, collect_subblocks): New functions.
18124         (verify_gimple_in_cfg): Verify that locations only reference
18125         BLOCKs in the functions BLOCK tree.
18127 2013-01-14  Richard Biener  <rguenther@suse.de>
18129         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18130         PHI argument.
18131         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18132         unshare reference.
18133         (insert_out_of_ssa_copy_on_edge): Likewise.
18134         (rewrite_close_phi_out_of_ssa): Likewise.
18135         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18136         debug expressions.
18137         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18138         propagated constants.
18139         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18140         can not be shared.
18142 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18144         * config/avr/avr-modes.def: Add GPL copyright notice.
18146 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
18148         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18149         MEMMODEL_MASK to determine memory model.
18150         (atomic_store<mode>): Ditto from operands[2].
18151         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18153 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
18155         PR fortran/55935
18156         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18157         (fold_gimple_assign): Don't call unshare_expr here.
18158         (fold_ctor_reference): Call unshare_expr.
18160 2013-01-13  Terry Guo  <terry.guo@arm.com>
18162         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18163         * doc/fragments.texi: Document MULTILIB_REUSE.
18164         * gcc.c (multilib_reuse): New internal spec.
18165         (set_multilib_dir): Also search multilib from multilib_reuse.
18166         * genmultilib (tmpmultilib3): Refactor code.
18167         (tmpmultilib4): Ditto.
18168         (multilib_reuse): New multilib argument.
18170 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
18172         * Makefile.in: Update copyright.
18174 2013-01-12  Tom de Vries  <tom@codesourcery.com>
18176         PR middle-end/55890
18177         * calls.c (expand_call): Check if arg_nr is valid.
18179 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
18181         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18182         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18183         documentation.  Add missing '__' in front of
18184         __builtin_ia32_packssdw256.
18186 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18188         PR target/55719
18189         * config/s390/s390.c (s390_preferred_reload_class): Do not return
18190         NO_REGS for larl operands.
18191         (s390_reload_larl_operand): Use s390_load_address instead of
18192         emit_move_insn.
18194 2013-01-11  Richard Biener  <rguenther@suse.de>
18196         * tree-cfg.c (verify_node_sharing_1): Split out from ...
18197         (verify_node_sharing): ... here.
18198         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
18200 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
18202         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
18203         Substitute TREECHECKING.
18204         * configure: Regenerate.
18205         * Makefile.in (TREECHECKING): New.
18207 2013-01-11  Richard Guenther  <rguenther@suse.de>
18209         PR tree-optimization/44061
18210         * tree-vrp.c (extract_range_basic): Compute zero as
18211         value-range for __builtin_constant_p of function parameters.
18213 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
18215         Update copyright years.
18217 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
18219         PR rtl-optimization/55672
18220         * lra-eliminations.c (mark_not_eliminable): Permit addition with
18221         const to be eliminable.
18223 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
18225         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
18226         * configure: Regenerate.
18228 2013-01-10  Richard Biener  <rguenther@suse.de>
18230         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
18232 2013-01-10  Richard Biener  <rguenther@suse.de>
18234         PR bootstrap/55792
18235         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
18236         locations for virtual PHI arguments.
18237         (rewrite_update_phi_arguments): Likewise.
18239 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
18241         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
18242         on to assembler.
18244 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
18246         PR tree-optimization/55921
18247         * tree-complex.c (expand_complex_asm): New function.
18248         (expand_complex_operations_1): Call it for GIMPLE_ASM.
18250 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18252         PR target/55718
18253         * config/s390/s390.c (s390_symref_operand_p)
18254         (s390_loadrelative_operand_p): Merge the two functions.
18255         (s390_check_qrst_address, print_operand_address): Add parameters
18256         to s390_loadrelative_operand_p invokation.
18257         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
18258         (s390_reload_larl_operand, s390_secondary_reload): Use
18259         s390_loadrelative_operand_p instead of s390_symref_operand_p.
18260         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
18262 2013-01-09  Mike Stump  <mikestump@comcast.net>
18264         * dse.c (record_store): Remove unnecessary assert.
18266 2013-01-09  Jan Hubicka  <jh@suse.cz>
18268         PR tree-optimization/55569
18269         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
18270         * cfgloop.h (scale_loop_profile): Likewise.
18272 2013-01-09  Jan Hubicka  <jh@suse.cz>
18274         PR lto/45375
18275         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
18276         functions.
18277         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
18279 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
18281         PR middle-end/55114
18282         * expr.h (maybe_emit_group_store): Declare.
18283         * expr.c (maybe_emit_group_store): New function.
18284         * builtins.c (expand_builtin_int_roundingfn): Call it.
18285         (expand_builtin_int_roundingfn_2): Likewise.
18287 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
18289         PR rtl-optimization/55829
18290         * lra-constraints.c (match_reload): Add code for absent output.
18291         (curr_insn_transform): Add code for reloads of matched inputs
18292         without output.
18294 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
18296         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
18297         attribute of movddup insn to DF.
18298         (*vec_interleave_lowv2df): Ditto.
18299         (vec_dupv2df): Ditto.
18301 2013-01-09  Jan Hubicka  <jh@suse.cz>
18303         PR tree-optimiation/55875
18304         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
18305         EVERY_ITERATION parameter.
18306         (number_of_iterations_exit): Check if exit is executed every iteration.
18307         (idx_infer_loop_bounds): Similarly here.
18308         (n_of_executions_at_most): Simplify
18309         to only test for cases where statement is dominated by the
18310         particular bound; handle correctly the "postdominance" test.
18311         (scev_probably_wraps_p): Use max loop iterations info
18312         as a global bound first.
18314 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
18315             Nick Clifton  <nickc@redhat.com>
18317         * config/v850/v850.md (cbranchsf4): New pattern.
18318         (cstoresf4): New pattern.
18319         (cbranchdf4): New pattern.
18320         (cstoredf4): New pattern.
18321         (movsicc): Disallow floating point comparisons.
18322         (cmpsf_le_insn): Fix order of operators.
18323         (cmpsf_lt_insn): Likewise.
18324         (cmpsf_eq_insn): Likewise.
18325         (cmpdf_le_insn): Likewise.
18326         (cmpdf_lt_insn): Likewise.
18327         (cmpdf_eq_insn): Likewise.
18328         (cmpsf_ge_insn): Use LE comparison.
18329         (cmpdf_ge_insn): Likewise.
18330         (cmpsf_gt_insn): Use LT comparison.
18331         (cmpdf_gt_insn): Likewise.
18332         (cmpsf_ne_insn): Delete pattern.
18333         (cmpdf_ne_insn): Delete pattern.
18334         * config/v850/v850.c (v850_gen_float_compare): Use
18335         gen_cmpdf_eq_insn for NE comparison.
18336         (v850_float_z_comparison_operator)
18337         (v850_float_nz_comparison_operator): Move from here ...
18338         * config/v850/predicates.md: ... to here.  Move GT and GE
18339         comparisons into v850_float_z_comparison_operator.
18340         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
18341         Delete prototype.
18342         (v850_float_nz_comparison_operator): Likewise.
18344 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18346         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
18347         with calls to gen_insvsi/gen_insvdi.
18349 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
18351         * config/i386/i386.c (initial_ix86_tune_features): Set up
18352         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
18354 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
18355             Jakub Jelinek  <jakub@redhat.com>
18357         PR tree-optimization/48189
18358         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
18359         If nitercst is 0, don't predict the exit edge.
18361 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18363         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
18364         in asm_fprintf with reg_names.
18365         (aarch64_print_operand_address): Likewise.
18366         (aarch64_return_addr): Likewise.
18367         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
18369 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18371         * config/pa/pa.h (VAL_U6_BITS_P): Define.
18372         (INT_U6_BITS): Likewise.
18373         * config/pa/predicates.md (uint6_operand): New predicate.
18374         (shift5_operand, shift6_operand): Likewise.
18375         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
18376         arith32_operand.
18377         (lshrdi3): Use shift6_operand.
18378         (shrpsi4, shrpdi4): New insn patterns.
18379         (extzv): Delete expander.
18380         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
18381         predicates in unamed zero extract patterns.  Tighten common constraint.
18382         (extv): Delete expander.
18383         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
18384         predicates in unamed sign extract patterns.  Tighten common constraint.
18385         (insv): Delete expander.
18386         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
18387         predicates in unamed insert patterns.  Tighten common constraint.
18388         Change uint32_operand predicate to uint6_operand predicate in unamed
18389         DImode pattern to insert constant values of type 1...1xxxx.
18391 2013-01-04  Jan Hubicka  <jh@suse.cz>
18393         PR tree-optimization/55823
18394         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
18395         issue.
18397 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18398             Uros Bizjak  <ubizjak@gmail.com>
18400         PR rtl-optimization/55845
18401         * df-problems.c (can_move_insns_across): Stop scanning at
18402         volatile_insn_p source instruction or give up if
18403         across_from .. across_to range contains any volatile_insn_p
18404         instructions.
18406 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
18408         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
18409         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
18410         Declare.
18411         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
18412         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
18414 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18416         PR fortran/55341
18417         * asan.c (asan_clear_shadow): New function.
18418         (asan_emit_stack_protection): Use it.
18420 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
18422         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
18423         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
18424         with tab instead of space.
18426 2013-01-08  Nick Clifton  <nickc@redhat.com>
18428         * config/rl78/rl78.c (rl78_expand_prologue): Always select
18429         register bank 0 at the start of an interrupt handler.
18430         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
18431         MDBH registers.
18433 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
18435         * config/aarch64/aarch64-simd.md
18436         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
18437         (aarch64_simd_bsl): Likewise.
18438         (aarch64_vcond_internal<mode>): Likewise.
18439         (vcond<mode><mode>): Likewise.
18440         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
18441         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
18443 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
18445         * config/aarch64/aarch64-builtins.c
18446         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
18448 2013-01-08  Martin Jambor  <mjambor@suse.cz>
18450         PR debug/55579
18451         * tree-sra.c (analyze_access_subtree): Return true also after
18452         potentially creating a debug-only replacement.
18454 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18456         PR middle-end/55890
18457         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
18459         PR tree-optimization/54120
18460         * tree-vrp.c (range_fits_type_p): Don't allow
18461         src_precision < precision from signed vr to unsigned_p
18462         if vr->min or vr->max is negative.
18463         (simplify_float_conversion_using_ranges): Test can_float_p
18464         against CODE_FOR_nothing.
18466 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18467             Richard Biener  <rguenther@suse.de>
18469         PR middle-end/55851
18470         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
18471         types instead of just INTEGER_TYPE types.
18473 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
18475         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
18476         TF_SIZE): Define.
18478 2013-01-07  Steve Ellcey  <sellcey@mips.com>
18480         PR target/42661
18481         * config/mips/mips.opt: Change mad to mmad to match documentation.
18483 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18485         PR target/55897
18486         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
18487         .progmemx.data now.
18489 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18491         PR target/55897
18492         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
18493         (avr_addrspace_t): Add .section_name field.
18494         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
18495         array size.
18496         (avr_addrspace): Same.  Initialize .section_name.  Remove last
18497         NULL entry.  Put __memx into .progmemx.data.
18498         (progmem_section_prefix): Remove.
18499         (avr_asm_init_sections): No need to initialize progmem_section.
18500         (avr_asm_named_section): Use avr_addrspace[].section_name to get
18501         section name prefix.
18502         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
18503         retrieve the progmem section.
18504         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
18505         boundary to run over avr_addrspace[].
18506         (avr_register_target_pragmas): Ditto.
18508 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
18510         * varasm.c (output_constant_def_contents): For asan_protect_global
18511         protected strings, adjust DECL_ALIGN if needed, before testing for
18512         anchored symbols.
18513         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
18514         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
18515         normal decls.
18516         (output_object_block): For asan protected decls, emit asan padding
18517         after their contents.
18518         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
18519         (asan_finish_file): Test it here instead.
18521 2013-01-07  Nick Clifton  <nickc@redhat.com>
18522             Matthias Klose  <doko@debian.org>
18523             Doug Kwan  <dougkwan@google.com>
18524             H.J. Lu  <hongjiu.lu@intel.com>
18526         PR driver/55470
18527         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
18529         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
18531         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
18533         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
18535         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
18537 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18539         PR target/54461
18540         * doc/install.texi (Cross-Compiler-Specific Options): Document
18541         --with-avrlibc.
18543 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
18545         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
18546         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
18547         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
18548         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
18549         vqmovun_high_s64): Fix source operand number and update copyright.
18551 2013-01-07  Richard Biener  <rguenther@suse.de>
18553         PR middle-end/55890
18554         * gimple.h (gimple_call_builtin_p): New overload.
18555         * gimple.c (validate_call): New function.
18556         (gimple_call_builtin_p): Likewise.
18557         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18558         Use gimple_call_builtin_p.
18559         (find_func_clobbers): Likewise.
18560         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
18561         (strlen_optimize_stmt): Likewise.
18563 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18565         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
18566         (vld1q_dup_*): Likewise.
18567         (vld1_*): Likewise.
18568         (vld1q_*): Likewise.
18569         (vld1_lane_*): Likewise.
18570         (vld1q_lane_*): Likewise.
18572 2013-01-07  Richard Biener  <rguenther@suse.de>
18574         * lto-streamer.h (LTO_minor_version): Bump to 2.
18576 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18578         * config/aarch64/aarch64-protos.h
18579         (aarch64_const_double_zero_rtx_p): Rename to...
18580         (aarch64_float_const_zero_rtx_p): ...this.
18581         (aarch64_float_const_representable_p): New.
18582         (aarch64_output_simd_mov_immediate): Likewise.
18583         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
18584         move immediate case.
18585         * config/aarch64/aarch64.c
18586         (aarch64_const_double_zero_rtx_p): Rename to...
18587         (aarch64_float_const_zero_rtx_p): ...this.
18588         (aarch64_print_operand): Allow printing of new constants.
18589         (aarch64_valid_floating_const): New.
18590         (aarch64_legitimate_constant_p): Check for valid floating-point
18591         constants.
18592         (aarch64_simd_valid_immediate): Likewise.
18593         (aarch64_vect_float_const_representable_p): New.
18594         (aarch64_float_const_representable_p): Likewise.
18595         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
18596         (aarch64_output_simd_mov_immediate): New.
18597         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
18598         (*movdf_aarch64): Likewise.
18599         * config/aarch64/constraints.md (Ufc): New.
18600         (Y): call aarch64_float_const_zero_rtx.
18601         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
18603 2013-01-07  Richard Biener  <rguenther@suse.de>
18605         PR tree-optimization/55888
18606         PR tree-optimization/55862
18607         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
18608         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
18609         not if it is contained therein.
18611 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18613         * config/avr/t-avr: Typo.
18615 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18617         PR55243
18618         * config/avr/t-avr: Don't automatically rebuild
18619         $(srcdir)/config/avr/t-multilib
18620         $(srcdir)/config/avr/avr-tables.opt
18621         $(srcdir)/doc/avr-mmcu.texi
18622         (avr-mcus): New phony target to build them on request.
18623         (s-avr-mlib, s-avr-mmcu-texi): Remove.
18624         * avr/avr-mcus.def: Adjust comments.
18626 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
18628         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
18630 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18632         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
18634 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18636         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
18638 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
18640         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
18641         to generate profiling.
18642         * config/rs6000/aix64.h (LIB_SPEC): Same.
18644 2013-01-04  Andrew Pinski  <apinski@cavium.com>
18646         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
18647         New function.
18648         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
18650 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
18652         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
18653         unconditionally.
18654         (ix86_expand_move): Ditto.
18655         (ix86_zero_extend_to_Pmode): Ditto.
18656         (ix86_expand_call): Ditto.
18657         (ix86_expand_special_args_builtin): Ditto.
18658         (ix86_expand_builtin): Ditto.
18660 2013-01-04  Richard Biener  <rguenther@suse.de>
18662         PR tree-optimization/55862
18663         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
18664         translating them through PHI nodes.
18666 2013-01-04  Martin Jambor  <mjambor@suse.cz>
18668         PR tree-optimization/55755
18669         * tree-sra.c (sra_modify_assign): Do not check that an access has no
18670         children when trying to avoid producing a VIEW_CONVERT_EXPR.
18672 2013-01-04  Marek Polacek  <polacek@redhat.com>
18674         PR middle-end/55859
18675         * opts.c (default_options_optimization): Clarify error message.
18677 2013-01-04  Richard Biener  <rguenther@suse.de>
18679         PR middle-end/55863
18680         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
18681         reassociation.
18683 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18685         PR target/53789
18686         * config/pa/pa.md (movsi): Revert previous change.
18687         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
18688         references.
18690 2013-01-03  Richard Henderson  <rth@redhat.com>
18692         * config/i386/i386.c (ix86_expand_move): Always assign to op1
18693         after eliminating TLS symbols.
18695 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
18697         PR bootstrap/50167
18698         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
18699         * graphite-poly.c (debug_gmp_value): Likewise.
18701 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
18703         PR target/55712
18704         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
18705         selected code model, define __code_mode_small__, __code_model_medium__,
18706         __code_model_large__, __code_model_32__ or __code_model_kernel__.
18707         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
18708         xchg temporary register with %k.  Declare temporary register as
18709         early clobbered.
18710         [__x86_64__]: For medium and large code models, preserve %rbx register.
18712 2013-01-03  Richard Biener  <rguenther@suse.de>
18714         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
18715         (dump_subscript): Adjust.
18716         (finalize_ddr_dependent): Do not dump redundant info.
18717         (analyze_siv_subscript): Adjust.
18718         (subscript_dependence_tester): Likewise.
18719         (compute_affine_dependence): Likewise.
18721 2013-01-03  Richard Biener  <rguenther@suse.de>
18723         Revert
18724         2013-01-03  Richard Biener  <rguenther@suse.de>
18726         PR tree-optimization/55857
18727         * tree-vect-stmts.c (vectorizable_load): Do not setup
18728         re-alignment for invariant loads.
18730         2013-01-02  Richard Biener  <rguenther@suse.de>
18732         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18733         invariant load do not generate a vector load from the scalar location.
18735 2013-01-03  Richard Biener  <rguenther@suse.de>
18737         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
18738         for not vectorizing.
18739         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
18740         not build INDIRECT_REFs, call get_name once only.
18741         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
18742         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
18744 2013-01-03  Richard Biener  <rguenther@suse.de>
18746         PR tree-optimization/55857
18747         * tree-vect-stmts.c (vectorizable_load): Do not setup
18748         re-alignment for invariant loads.
18750 2013-01-03  Richard Biener  <rguenther@suse.de>
18752         PR lto/55848
18753         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
18754         prefer a built-in decl.
18756 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18758         * gcc.c (process_command): Update copyright notice dates.
18759         * gcov.c (print_version): Likewise.
18760         * gcov-dump.c (print_version): Likewise.
18762         PR rtl-optimization/55838
18763         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
18764         iv0.step, iv1.step and step.
18766 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18767             Marc Glisse  <marc.glisse@inria.fr>
18769         PR tree-optimization/55832
18770         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
18771         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
18772         integer_{one,zero}_node.
18774 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18776         PR debug/54402
18777         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
18778         * var-tracking.c (reverse_op): Don't add reverse ops to
18779         VALUEs that have already
18780         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
18782 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
18784         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
18786 2013-01-02  Teresa Johnson  <tejohnson@google.com>
18788         * dumpfile.c (dump_loc): Print filename with location.
18789         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
18790         new location_t parameter to emit complete unroll message with
18791         new dump framework.
18792         (canonicalize_loop_induction_variables): Compute loops location
18793         and pass to try_unroll_loop_completely.
18794         * loop-unroll.c (report_unroll_peel): New function.
18795         (peel_loops_completely): Use new dump format with location
18796         for main dumpfile message, and invoke report_unroll_peel on success.
18797         (decide_unrolling_and_peeling): Ditto.
18798         (decide_peel_once_rolling): Remove old dumpfile message subsumed
18799         by report_unroll_peel.
18800         (decide_peel_completely): Ditto.
18801         (decide_unroll_constant_iterations): Ditto.
18802         (decide_unroll_runtime_iterations): Ditto.
18803         (decide_peel_simple): Ditto.
18804         (decide_unroll_stupid): Ditto.
18805         * cfgloop.c (get_loop_location): New function.
18806         * cfgloop.h (get_loop_location): Declare.
18808 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18810         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
18811         NULL.
18813 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18815         PR middle-end/55198
18816         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
18817         BLKmode objects when EXPAND_MEMORY is specified.
18819 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18821         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
18822         in loop predicate.
18823         (fold_builtin_cpu): Do not share cpu model decls across statements.
18825 2013-01-02  Jason Merrill  <jason@redhat.com>
18827         PR c++/55804
18828         * tree.c (build_array_type_1): Revert earlier change.
18830 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
18832         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
18833         "cortex-a57".
18834         * config/aarch64/aarch64-tune.md: Re-generate.
18836 2013-01-02  Richard Biener  <rguenther@suse.de>
18838         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18839         invariant load do not generate a vector load from the scalar location.
18841 2013-01-02  Richard Biener  <rguenther@suse.de>
18843         PR bootstrap/55784
18844         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
18845         * configure: Regenerate.
18847 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18849         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
18850         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
18851         (expand_builtin_int_roundingfn_2): Keep the original target around
18852         for the fallback case.
18854 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18856         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
18857         to be clear for sign changes.
18859 2013-01-01  Jan Hubicka  <jh@suse.cz>
18861         * ipa-inline-analysis.c: Fix formatting.
18863 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
18865         PR tree-optimization/55831
18866         * tree-vect-loop.c (get_initial_def_for_induction): Use
18867         gsi_after_labels instead of gsi_start_bb.
18869 Copyright (C) 2013 Free Software Foundation, Inc.
18871 Copying and distribution of this file, with or without modification,
18872 are permitted in any medium without royalty provided the copyright
18873 notice and this notice are preserved.