* cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
[official-gcc.git] / gcc / ChangeLog
blobf296c376cbcc37d01d0292588c6b9e29018d972a
1 2013-08-26  Jan Hubicka  <jh@suse.cz>
3         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
4         methods can not be called indirectly when their address is not taken.
6 2013-08-26  Jan Hubicka  <jh@suse.cz>
8         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use ctor_for_folding.
10 2013-08-26  Jan Hubicka  <jh@suse.cz>
12         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
13         can be unshared.
15 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
17         * reload.c (find_valid_class): Allow classes that do not include
18         FIRST_PSEUDO_REGISTER - 1.
20 2013-08-26  Jan Hubicka  <jh@suse.cz>
22         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
23         fix edge count/frequency when speculation failed; fix type check
24         for the direct call.
26 2013-08-26  Jan Hubicka  <jh@suse.cz>
28         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
30 2013-08-26  Jan Hubicka  <jh@suse.cz>
32         * ipa-inline-transform.c (inline_transform): Be ready for basic block
33         to be changed by edge redirection.
35 2013-08-26  Jan Hubicka  <jh@suse.cz>
37         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and formating;
38         add sanity check.
39         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
40         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
41         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
42         (resolve_noninline_speculation): Update callee keys, too.
44 2013-08-26  Jan Hubicka  <jh@suse.cz>
46         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
47         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
49 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
51         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
52         into proper place.
54 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
56         * config/i386/i386.c (ix86_debug_options): Remove prototype.
57         (x86_64_elf_select_section): Ditto.
58         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
59         arguments.
60         (ix86_pass_by_reference): Ditto.
61         (output_set_got): Ditto.
62         (ix86_unary_operator_ok): Ditto.
63         (ix86_expand_builtin): Ditto.
65 2013-08-23  Jan Hubicka  <jh@suse.cz>
67         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
69 2013-08-23  Jan Hubicka  <jh@suse.cz>
71         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
72         (tree_decl_with_vis): Add FINAL field.
74 2013-08-23  Jeff Law  <law@redhat.com>
76         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
77         the debugging dump when the expression is fully redundant.
79 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
81         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
82         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
83         * pretty-print.c (pp_formatted_text_data): Likewise.
84         (pp_write_text_to_stream): Likewise.
85         (pp_write_text_as_dot_label_to_stream): Likewise.
86         (pp_append_r): Likewise.
87         (pp_format): Likewise.
88         (pp_flush): Likewise.
89         (pp_clear_output_area): Likewise.
90         (pp_append_text): Likewise.
91         (pp_formatted_text): Likewise.
92         (pp_remaining_character_count_for_line): Likewise.
93         (pp_newline): Likewise.
94         (pp_character): Likewise.
95         (output_buffer::~output_buffer): Define.
96         (pretty_printer::~pretty_printer): Destruct output buffer.
97         * pretty-print.h (output_buffer::~output_buffer): Declare.
98         (pretty_printer::~pretty_printer): Declare virtual.
100 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
102         PR other/57324
103         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
104         HOST_WIDE_INT_M1U): New macros.
105         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
106         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
107         unsigned -1 for lshift.
108         * cse.c (cse_insn): Likewise.
109         * double-int.c (rshift_double, lshift_double): Likewise.
110         * builtins.c (fold_builtin_bitop): Likewise.
111         * combine.c (force_to_mode): Likewise.
112         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
113         * simplify-rtx.c (simplify_const_unary_operation,
114         simplify_const_binary_operation): Likewise.
115         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
116         check_va_list_escapes): Likewise.
117         * rtlanal.c (nonzero_bits1): Likewise.
118         * expmed.c (expand_smod_pow2): Likewise.
119         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
121 2013-08-23  Jan Hubicka  <jh@suse.cz>
123         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
124         as having address taken.
126 2013-08-23  Jan Hubicka  <jh@suse.cz>
128         * ipa-utils.h (method_class_type): Declare.
129         * ipa-devirt.c (method_class_type): Export.
131         * cgraphunit.c (analyze_functions): Do basic devirtualization;
132         do not walk base classes of anonymous types.
134 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
136         PR rtl-optimization/58220
137         PR regression/58221
138         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
139         handle SEQUENCE insns properly.
141 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
143         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
144         definition.
145         (pp_newline_and_indent): Likewise.
146         (pp_separate_with): Likewise.
147         * pretty-print.c (pp_newline_and_flush): Define.
148         (pp_newline_and_indent): Likewise.
149         (pp_separate_with): Likewise.
151 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
153         PR target/58218
154         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
155         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
157 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
159         * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
160         * gcc/config/i386/i386.md (*movti_internal): Use
161         predicate to determine if EVEX is needed.
162         (*movsi_internal): Ditto.
163         (*movdf_internal): Ditto.
164         (*movsf_internal): Ditto.
165         * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
167 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
169         PR tree-optimization/58209
170         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
171         (find_tail_calls): Give up for pointer result types if m is non-NULL.
172         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
173         emit POINTER_PLUS_EXPR.
174         (create_tailcall_accumulator): For pointer result type accumulate in
175         sizetype type.
177 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
179         * configure.ac: Add backslashes missing from the last change.
180         * configure: Regenerate.
182 2013-08-22  Jan Hubicka  <jh@susue.cz>
184         * ipa.c (function_and_variable_visibility): First remember function
185         was global and then make it local.
187 2013-08-22  Julian Brown  <julian@codesourcery.com>
189         * configure.ac: Add aarch64 to list of arches which use "nop" in
190         debug_line test.
191         * configure: Regenerate.
193 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
195         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
196         gnu_libc_has_function.
197         * config/s390/tpf.h: Likewise.
199 2013-08-22  Jan Hubicka  <jh@susue.cz>
201         * timevar.c (validate_phases): Add cast.
203 2013-08-22  Jan Hubicka  <jh@susue.cz>
205         * timevar.c (validate_phases): Use size_t for memory.
206         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
208 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
210         * pretty-print.h (output_buffer::output_buffer): Declare.
211         (pretty_printer::pretty_printer): Likewise.
212         (pp_construct): Remove.
213         * pretty-print.c (output_buffer::output_buffer): Define.
214         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
215         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
216         (print_gimple_expr): Likewise.
217         (print_gimple_seq): Likewise.
218         (gimple_dump_bb): Likewise.
219         * sched-vis.c (dump_value_slim): Likewise.
220         (dump_insn_slim): Likewise.
221         (dump_rtl_slim): Likewise.
222         (str_pattern_slim): Likewise.
223         * tree-mudflap.c (mf_varname_tree): Likewise.
224         * graph.c (print_graph_cfg): Likewise.
225         (start_graph_dump): Likewise.
226         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
227         placement-new.
228         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
229         pretty printer initialization.
230         * coretypes.h (diagnostic_context): Remove superflous type alias
231         declaration.
232         (pretty_printer): Likewise.  Declare directly as a class.
233         (pretty_print_info): Remove declaration as class.
234         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
235         and pp_clear_output_area.
236         (asan_add_global): Likewise.
238 2013-08-22  Jan Hubicka  <jh@suse.cz>
240         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
241         * ipa-utils.h (update_type_inheritance_graph): Declare.
242         (possible_polymorphic_call_target_p): Declare.
243         (possible_polymorphic_call_target_p): New.
244         * ipa-devirt.c: Update toplevel comments.
245         (cached_polymorphic_call_targets): Move up.
246         (odr_type_d): Move ID down.
247         (polymorphic_type_binfo_p): Update comment.
248         (odr_hasher::remove): Likewise;
249         (get_odr_type): Set anonymous_namespace.
250         (dump_odr_type): Dump it.
251         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
252         (maybe_record_node): Record node in cached_polymorphic_call_targets.
253         (record_binfo): Add comment.
254         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
255         built.
256         (devirt_node_removal_hook): Do not iCE when cache is freed.
257         (possible_polymorphic_call_target_p): New predicate.
258         (update_type_inheritance_graph): New function.
260 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
261             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
262             Sergey Lega  <sergey.s.lega@intel.com>
263             Anna Tikhonova  <anna.tikhonova@intel.com>
264             Ilya Tocar  <ilya.tocar@intel.com>
265             Andrey Turetskiy  <andrey.turetskiy@intel.com>
266             Ilya Verbin  <ilya.verbin@intel.com>
267             Kirill Yukhin  <kirill.yukhin@intel.com>
268             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
270         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
271         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
272         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
273         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
274         (OPTION_MASK_ISA_AVX2_UNSET): Update.
275         (OPTION_MASK_ISA_AVX512F_UNSET): New.
276         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
277         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
278         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
279         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
280         OPT_mavx512pf, OPT_mavx512er cases.
281         * config/i386/constraints.md (v): New constraint.
282         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
283         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
284         (bit_AVX512CD): New.
285         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
286         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
287         * config/i386/i386-c.c (ix86_target_macros_internal):
288         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
289         __AVX512PF__.
290         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
291         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
292         * config/i386/i386.c (regclass_map, dbx_register_map)
293         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
294         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
295         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
296         -mavx512pf options.
297         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
298         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
299         -mavx512pf options.  Fix formatting.
300         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
301         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
302         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
303         -mavx512cd, -mavx512pf options.
304         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
305         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
306         (ix86_preferred_output_reload_class): Replace SSE_REGS with
307         ALL_SSE_REGS.
308         (ix86_hard_regno_mode_ok): Support 512-bit registers.
309         (ix86_set_reg_reg_cost): Ditto.
310         (x86_order_regs_for_local_alloc): Ditto.
311         (MAX_VECT_LEN): Extend to 64-byte.
312         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
313         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
314         (TARGET_AVX512ER, TARGET_AVX512CD): New.
315         (BIGGEST_ALIGNMENT): Extend to 512-bits.
316         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
317         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
318         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
319         (SSE_REG_MODE_P): Support new modes.
320         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
321         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
322         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
323         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
324         (REG_CLASS_CONTENTS): Add new registers.
325         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
326         (EXT_REX_SSE_REGNO_P): New.
327         (HI_REGISTER_NAMES): Add new registers.
328         * config/i386/i386.md: Define constants for new registers.
329         (mode): Add new 512-bit modes.
330         (prefix): Support evex prefix.
331         (isa): Support avx512f, noavx512f, fma_avx512f.
332         (ssemodesuffix): Add new 512-bit modes.
333         (movxi): New.
334         (*movxi_internal_avx512f): Ditto.
335         (*movdi_internal): Replace constraint "x" with the new constraint "v".
336         Support MODE_XI.
337         (*movsi_internal): Likewise.
338         (*movdf_internal): Likewise.
339         (*movsf_internal): Likewise.
340         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
341         (<code><mode>3): Likewise.
342         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
343         New.
344         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
345         with the new constraint "v".
346         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
347         modes.
348         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
349         with the new constraint "v".
350         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
351         (<sse2>_storedqu<avxsizesuffix>): Likewise.
352         (*<plusminus_insn><mode>3): Likewise.
353         (<sse>_vm<plusminus_insn><mode>3): Likewise.
354         (*mul<mode>3): Likewise.
355         (<sse>_vmmul<mode>3): Likewise.
356         (<sse>_div<mode>3): Likewise.
357         (<sse>_vmdiv<mode>3): Likewise.
358         (<sse>_sqrt<mode>2): Likewise.
359         (<sse>_vmsqrt<mode>2): Likewise.
360         (*<code><mode>3_finite): Likewise.
361         (*<code><mode>3) <smaxmin>: Likewise.
362         (<sse>_vm<code><mode>3): Likewise.
363         (*<code><mode>3) <any_logic>: Likewise.
364         (*fma_fmadd_<mode>): Likewise.
365         (*fma_fmsub_<mode>): Likewise.
366         (*fma_fnmadd_<mode>): Likewise.
367         (*fma_fnmsub_<mode>): Likewise.
368         (*fma_fmaddsub_<mode>): Likewise.
369         (*fma_fmsubadd_<mode>): Likewise.
370         (*fmai_fmadd_<mode>): Likewise.
371         (*fmai_fmsub_<mode>): Likewise.
372         (*fmai_fnmadd_<mode>): Likewise.
373         (*fmai_fnmsub_<mode>): Likewise.
374         (sse_cvtsi2ss): Likewise.
375         (sse_cvtsi2ssq): Likewise.
376         (sse_cvtss2si): Likewise.
377         (sse_cvtss2si_2): Likewise.
378         (sse_cvtss2siq): Likewise.
379         (sse_cvtss2siq_2): Likewise.
380         (sse_cvttss2si): Likewise.
381         (sse_cvtss2siq_2): Likewise.
382         (float<sseintvecmodelower><mode>2): Likewise.
383         (sse2_cvtsd2si_2): Likewise.
384         (sse2_cvtsd2siq_2): Likewise.
385         (*<plusminus_insn><mode>3): Likewise.
386         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
387         (*<sse4_1_avx2>_mul<mode>3): Likewise.
388         (ashr<mode>3): Likewise.
389         (<shift_insn><mode>3): Likewise.
390         (avx2_<code><mode>3): Likewise.
391         (*avx2_<code><mode>3): Likewise.
392         (*andnot<mode>3): Likewise.
393         (*<code><mode>3) <any_logic>: Likewise.
394         (abs<mode>2): Likewise.
395         (avx2_permvar<mode>): Likewise.
396         (avx2_perm<mode>_1): Likewise.
397         (*avx_vpermilp<mode>): Likewise.
398         (avx_vpermilvar<mode>3): Likewise.
399         (avx2_ashrv<mode>): Likewise.
400         (avx2_<shift_insn>v<mode>): Likewise.
401         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
402         -mavx512cd.
403         * doc/rtl.texi: Document XImode.
405 2013-08-21  Jeff Law  <law@redhat.com>
407         * tree-flow.h (register_jump_thread): Pass vector of edges
408         instead of each important edge.
409         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
410         thread path into a vector and pass that to register_jump_thread.
411         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
412         passed in edge vector to the current 3-edge form.
414         Revert:
415         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
417         * dce.c (fini_dce): Call df_analyze again just in case
418         delete_unmarked_insns removed anything.
420 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
422         * reload.h (struct reg_equivs): Rename to ..
423         (struct reg_equivs_s): .. this.
425 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
427         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
429 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
431         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
433 2013-08-21  Jeff Law  <law@redhat.com>
435         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
436         simplify assignments too.  If the RHS collapses to a singleton
437         range, then return the value for the range.
439 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
441         * config/i386/sse.md (V16): Rename to...
442         (VMOVE): this.
443         (mov<mode>): Update iterator name.
444         (*mov<mode>_internal): Ditto.
445         (push<mode>1): Ditto.
446         (movmisalign<mode>): Ditto.
448 2013-08-20  Jan Hubicka  <jh@suse.cz>
450         PR bootstrap/58186
451         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
452         entry for direct edges.
453         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
455 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
457         Revert my last two changes, r201865 and r201864:
459         Revert r201865:
460         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
462         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
463         instances can own GC refs.
465         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
466         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
467         (gcc::context::gt_pch_nx): Likewise.
468         (gcc::context::gt_pch_nx):  Likewise.
469         * ggc.h (gt_ggc_mx <T>): New.
470         (gt_pch_nx_with_op <T>): New.
471         (gt_pch_nx <T>): New.
472         * passes.c (opt_pass::gt_ggc_mx): New.
473         (opt_pass::gt_pch_nx): New.
474         (opt_pass::gt_pch_nx_with_op): New.
475         (pass_manager::gt_ggc_mx): New.
476         (pass_manager::gt_pch_nx): New.
477         (pass_manager::gt_pch_nx_with_op): New.
478         (pass_manager::operator new): Use
479         ggc_internal_cleared_alloc_stat rather than xcalloc.
480         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
481         (pass_manager::gt_ggc_mx): New.
482         (pass_manager::gt_pch_nx): New.
483         (pass_manager::gt_pch_nx_with_op): New.
484         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
485         (opt_pass::operator new): New.
486         (opt_pass::gt_ggc_mx): New.
487         (opt_pass::gt_pch_nx): New.
488         (opt_pass::gt_pch_nx_with_op): New.
490         Revert r201864:
491         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
493         * Makefile.in (GTFILES): Add context.h.
494         * context.c (gcc::context::operator new): New.
495         (gcc::context::gt_ggc_mx): New.
496         (gcc::context::gt_pch_nx): New.
497         (gcc::context::gt_pch_nx): New.
498         * context.h (gcc::context): Add GTY((user)) marking.
499         (gcc::context::operator new): New.
500         (gcc::context::gt_ggc_mx): New.
501         (gcc::context::gt_pch_nx): New.
502         (gcc::context::gt_pch_nx): New.
503         (g): Add GTY marking.
504         (gt_ggc_mx (gcc::context *)): New.
505         (gt_pch_nx (gcc::context *)): New.
506         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
507         void *cookie)): New.
508         * gengtype.c (open_base_files) <ifiles>: Add context.h.
510 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
512         * dce.c (fini_dce): Call df_analyze again just in case
513         delete_unmarked_insns removed anything.
515 2013-08-20  Teresa Johnson  <tejohnson@google.com>
517         PR rtl-optimizations/57451
518         * final.c (reemit_insn_block_notes): Prevent lexical blocks
519         from crossing split section boundaries.
521 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
523         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
524         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
525         with MULTLIB_DEFAULTS.
527 2013-08-20  Nick Clifton  <nickc@redhat.com>
529         * target.def (narrow_volatile_bitfield): Note that the default
530         value is false, not !TARGET_STRICT_ALIGN.
531         * doc/tm.texi: Regenerate.
533 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
535         Fix LIB_SPEC for systems without libpthread.
537         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
538         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
539         for Android.
540         * config/i386/linux-common.h: Likewise.
541         * config/mips/linux-common.h: Likewise.
543 2013-08-20  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
545         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
546         checks.
548 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
550         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
551         instances can own GC refs.
553         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
554         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
555         (gcc::context::gt_pch_nx): Likewise.
556         (gcc::context::gt_pch_nx):  Likewise.
557         * ggc.h (gt_ggc_mx <T>): New.
558         (gt_pch_nx_with_op <T>): New.
559         (gt_pch_nx <T>): New.
560         * passes.c (opt_pass::gt_ggc_mx): New.
561         (opt_pass::gt_pch_nx): New.
562         (opt_pass::gt_pch_nx_with_op): New.
563         (pass_manager::gt_ggc_mx): New.
564         (pass_manager::gt_pch_nx): New.
565         (pass_manager::gt_pch_nx_with_op): New.
566         (pass_manager::operator new): Use
567         ggc_internal_cleared_alloc_stat rather than xcalloc.
568         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
569         (pass_manager::gt_ggc_mx): New.
570         (pass_manager::gt_pch_nx): New.
571         (pass_manager::gt_pch_nx_with_op): New.
572         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
573         (opt_pass::operator new): New.
574         (opt_pass::gt_ggc_mx): New.
575         (opt_pass::gt_pch_nx): New.
576         (opt_pass::gt_pch_nx_with_op): New.
578 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
580         * Makefile.in (GTFILES): Add context.h.
581         * context.c (gcc::context::operator new): New.
582         (gcc::context::gt_ggc_mx): New.
583         (gcc::context::gt_pch_nx): New.
584         (gcc::context::gt_pch_nx): New.
585         * context.h (gcc::context): Add GTY((user)) marking.
586         (gcc::context::operator new): New.
587         (gcc::context::gt_ggc_mx): New.
588         (gcc::context::gt_pch_nx): New.
589         (gcc::context::gt_pch_nx): New.
590         (g): Add GTY marking.
591         (gt_ggc_mx (gcc::context *)): New.
592         (gt_pch_nx (gcc::context *)): New.
593         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
594         void *cookie)): New.
595         * gengtype.c (open_base_files) <ifiles>: Add context.h.
597 2013-08-20  Alan Modra  <amodra@gmail.com>
599         PR target/57865
600         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
601         (rs6000_emit_epilogue): Likewise.
603 2013-08-19  Dehao Chen (dehao@google.com)
605         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
607 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
608             Jakub Jelinek  <jakub@redhat.com>
610         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
611         (BUILT_IN_FABSD64): Likewise.
612         (BUILT_IN_FABSD128): Likewise.
613         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
614         (fold_builtin_1): Likewise.
615         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
616         destination and source operands.
617         (*abstd2_fpr): Likewise.
618         (*nabstd2_fpr): Likewise.
620 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
622         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
623         JUMP_P and INSN_P.
625 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
627         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
628         incomplete.
630 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
632         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
633         * builtins.c (default_libc_has_function): New.
634         (gnu_libc_has_function): Ditto.
635         (no_c99_libc_has_function): Ditto.
636         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
637         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
638         (fold_builtin_sincos): Likewise.
639         (fold_builtin_cexp): Likewise.
640         * builtins.def (DEF_C94_BUILTIN): Likewise.
641         (DEF_C99_BUILTIN): Likewise.
642         (DEF_C99_C90RES_BUILTIN): Likewise.
643         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
644         definitions to using this define.
645         * config/darwin-protos.h (darwin_libc_has_function): New.
646         * config/darwin.c (darwin_libc_has_function): Ditto.
647         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
648         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
649         * config/darwin.h: Ditto.
650         * config/elfos.h: Ditto.
651         * config/freebsd.h: Ditto.
652         * config/i386/cygming.h: Ditto.
653         * config/i386/djgpp.h: Ditto.
654         * config/i386/i386-interix.h: Ditto.
655         * config/microblaze/microblaze.h: Ditto.
656         * config/mmix/mmix.h: Ditto.
657         * config/gnu-user.h: Ditto.
658         * config/ia64/hpux.h: Ditto.
659         * config/pa/pa-hpux.h: Ditto.
660         * config/pdp11/pdp11.h: Ditto.
661         * config/picochip/picochip.h: Ditto.
662         * config/linux.h: Ditto.
663         * config/netbsd.h: Ditto.
664         * config/openbsd.h: Ditto.
665         * config/rs6000/aix43.h: Ditto.
666         * config/rs6000/aix51.h: Ditto.
667         * config/rs6000/aix52.h: Ditto.
668         * config/rs6000/aix53.h: Ditto.
669         * config/rs6000/aix61.h: Ditto.
670         * config/rs6000/darwin.h: Ditto.
671         * config/rs6000/linux.h: Ditto.
672         * config/rs6000/linux64.h: Ditto.
673         * config/s390/tpf.h: Ditto.
674         * config/sol2-10.h: Ditto.
675         * config/sol2.h: Ditto.
676         * config/vms/vms.h: Ditto.
677         * config/vxworks.h: Ditto.
678         * config/linux-android.c (linux_android_libc_has_function):
679         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
680         * config/linux-protos.h (linux_android_libc_has_function):
681         New declaration.
682         * config/i386/i386.c (ix86_libc_has_function): New.
683         * config/i386/i386-protos.h
684         (ix86_libc_has_function): New declaration.
685         * config/i386/i386.md
686         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
687         ("isinf<mode>2): Likewise.
688         * convert.c (convert_to_integer): Using new target hook
689         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
690         TARGET_C99_FUNCTIONS.
691         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
692         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
693         * coretypes.h (function_class): New enum for different
694         classes of functions.
695         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
696         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
697         (TARGET_HAS_SINCOS): Likewise.
698         (TARGET_LIBC_HAS_FUNCTION): New.
699         * doc/tm.texi: Regenerated.
700         * targhooks.h (default_libc_has_function): New declaration.
701         (no_c99_libc_has_function): Ditto.
702         (gnu_libc_has_function): Ditto.
703         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
704         and TARGET_HAS_SINCOS.
706 2013-08-18  Jan Hubicka  <jh@suse.cz>
708         * Makeifle-in (ipa-devirt.o): New.
709         (GTFILES): Add ipa-utils.h and ipa-devirt.c
710         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
711         (analyze_functions): Look into possible targets of polymorphic call.
712         * dumpfile.c (dump_files): Add type-inheritance dump.
713         * dumpfile.h (TDI_inheritance): New.
714         * ipa-devirt.c: New file.
715         * ipa-utils.h (odr_type_d): Forward declare.
716         (odr_type): New type.
717         (build_type_inheritance_graph): Declare.
718         (possible_polymorphic_call_targets): Declare and introduce inline
719         variant when only edge is pased.
720         (dump_possible_polymorphic_call_targets): Likewise.
721         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
722         * tree.c (type_in_anonymous_namespace_p): Break out from ...
723         (types_same_for_odr): ... here.
724         * tree.h (type_in_anonymous_namespace_p): Declare.
726 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
728         PR tree-optimization/58006
729         * tree-parloops.c (take_address_of): Don't ICE if get_name
730         returns NULL.
731         (eliminate_local_variables_stmt): Remove clobber stmts.
733 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
735         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
736         error message is issued for an alias to undefined symbol.
738 2013-08-18  Jan Hubicka  <jh@suse.cz>
740         * cgraph.c (cgraph_create_indirect_edge): Discover
741         polymorphic calls and record basic info into indirect_info.
742         * gimple-fold.c (gimple_fold_call): When doing BINFO based
743         devirtualization, ignore objc function calls.
744         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
745         call with no parm index info.
746         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
747         * tree.c (virtual_method_call_p): New function.
748         * tree.h (virtual_method_call_p): Declare.
750 2013-08-16  Jan Hubicka  <jh@suse.cz>
752         PR middle-end/58179
753         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
755 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
757         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
758         attribute.
760 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
762         * gengtype.c (type_for_name): Add special-case support for
763         locating types within the "gcc::" namespace.
764         (open_base_files): Emit a "using namespace gcc" directive.
766 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
768         PR target/58160
769         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
770         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
772         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
773         array instead of each individual operand as a separate argument.
774         (emit_fusion_gpr_load): Likewise.
775         (expand_fusion_gpr_load): Add new function declaration.
777         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
778         signature to have the operands passed as an array, instead of as
779         separate arguments.  Allow ZERO_EXTEND to be in the memory
780         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
781         depend on the register live/dead flags when peepholes are run.
782         (expand_fusion_gpr_load): New function to be called from the
783         peephole2 pass, to change the register that addis sets to be the
784         target register.
785         (emit_fusion_gpr_load): Change the calling signature to have the
786         operands passed as an array, instead of as separate arguments.
787         Allow ZERO_EXTEND to be in the memory address, and also
788         SIGN_EXTEND if -mpower8-fusion-sign.
790         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
791         unspec enumeration.
792         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
793         adjust the register addis loads up in the peephole2 pass.  Do not
794         depend on the register live/dead state when the peephole pass is done.
796 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
798         * gengtype.c (create_user_defined_type): Ensure that the kind
799         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
800         declaration is seen before the GTY((user)) marking.
802 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
804         PR target/58105
805         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
807 2013-08-16  Jan Hubicka  <jh@suse.cz>
809         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
810         arugment expected_type.
811         (gimple_fold_call): Use it.
812         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
813         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
814         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
815         (try_make_edge_direct_virtual_call): Likewise.
816         * tree.c (obj_type_ref_class): New.
817         * tree.h (obj_type_ref_class): Use it.
819 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
821         * sched-vis.c (rtl_slim_pp_initialized): Remove.
822         (rtl_slim_pp): Likewise.
823         (init_rtl_slim_pretty_print): Likewise.
824         (dump_value_slim):  Don't call it.  Use local pretty printer.
825         (dump_insn_slim): Likewise.
826         (dump_rtl_slim): Likewise.
827         (str_pattern_slim): Likewise.
828         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
829         Simplify.
831 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
833         PR tree-optimization/58164
834         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
835         walk gimple_goto_dest of GIMPLE_GOTO.
837         PR tree-optimization/58165
838         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
839         bi_call must be the last stmt in a bb, don't split_block, instead
840         use fallthru edge from it and give up if there is none.
841         Release conds vector when returning early.
843 2013-08-14  Xinliang David Li  <davidxl@google.com>
845         * config/i386/i386.c (ix86_option_override_internal):
846         Remove unused variable and field.
848 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
850         PR target/57949
851         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
852         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
853         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
854         and Linux, correct BLKmode alignment when 128-bit alignment is
855         required and compatibility flag is not set.
856         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
857         for zero-size arguments when compatibility flag is not set.
859 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
861         PR tree-optimization/58145
862         * tree-sra.c (build_ref_for_offset): If prev_base has
863         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
865 2013-08-14  Xinliang David Li  <davidxl@google.com>
867         * config/i386/i386.c (ix86_option_override_internal):
868         Fix uninitialized variable error.
870 2013-08-14  Xinliang David Li  <davidxl@google.com>
872         * config/i386/i386.opt: Define two new options.
873         * config/i386/x86-tune.def: Add arch selector field in macros.
874         * config/i386/i386.h: Adjust macro definition.
875         * config/i386/i386.c (ix86_option_override_internal):
876         Refactor the code.
877         (parse_mtune_ctrl_str): New function.
878         (set_ix86_tune_features): New function.
879         (ix86_function_specific_restore): Call the new helper function.
881 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
883         PR rtl-optimization/57662
884         * sel-sched.c (code_motion_process_successors): When the current insn
885         is removed after the recursive traversal, break from the loop.
886         Add comments and debug printouts.
888 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
889             Alexandre Oliva  <aoliva@redhat.com>
891         PR target/58067
892         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
893         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
894         there also UNSPEC_PLTOFF.
896 2013-08-14  Marek Polacek  <polacek@redhat.com>
898         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
899         NUM_CONDITIONS bit positions.
901 2013-08-13  Cary Coutant  <ccoutant@google.com>
903         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
904         (attr_checksum): Hash vector contents instead of pointer.
905         (attr_checksum_ordered): Likewise.
907 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
909         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
910         when Pmode != word_mode.  Add length_address attribute.
911         (sse3_monitor_<mode>): Merge from sse3_monitor and
912         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
913         Pmode != word_mode.  Update insn length attribute.
914         * config/i386/i386.c (ix86_option_override_internal): Update
915         ix86_gen_monitor selection for merged sse3_monitor insn.
917 2013-08-13  Julian Brown  <julian@codesourcery.com>
919         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
920         perform invalid legitimization on greater-than-word-size modes for
921         TARGET_E500_DOUBLE.
923 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
925         * ira.c (setup_class_translate_array): Use aclass instead of cl
926         for classes not fully covered by allocno classes.
928 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
930         PR tree-optimization/57661
931         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
932         * tree-inline.c (tree_function_versioning): Initialize it.
933         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
934         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
935         that is not being copied.
937         PR sanitizer/56417
938         * asan.c (instrument_strlen_call): Fix typo in comment.
939         Use char * type even for the lhs of POINTER_PLUS_EXPR.
941 2013-08-13  Steve Ellcey  <sellcey@mips.com>
943         * config/mips/mips.md (prefetch): Use lw instead of ld on
944         loongson in 32bit mode.
946 2013-08-13  Nick Clifton  <nickc@redhat.com>
948         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
950 2013-08-13  Jan Hubicka  <jh@suse.cz>
952         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
953         introduced edge; fix typo in sanity check.
954         (cgraph_resolve_speculation): Export; improve diagnostic.
955         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
956         speculation at type mismatch.
957         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
958         (cgraph_resolve_speculation): Declare.
959         (symtab_can_be_discarded): New function.
960         * value-prof.c (gimple_ic_transform): Remove actual transform code.
961         * ipa-inline-transform.c (speculation_removed): New global var.
962         (clone_inlined_nodes): See if speculation can be removed.
963         (inline_call): If speculations was removed, we growths may not match.
964         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
965         (speculation_useful_p): New function.
966         (resolve_noninline_speculation): New function.
967         (inline_small_functions): Resolve useless speculations.
968         * ipa-inline.h (speculation_useful_p): Declare
969         * ipa.c (can_replace_by_local_alias): Simplify.
970         (ipa_profile): Produce speculative calls in non-lto, too;
971         add simple cost model; produce local aliases.
973 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
975         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
976         PASS_MANAGER_H.
978 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
980         * config/i386/i386.c (ix86_function_versions): Use error + inform.
982 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
984         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
985         iterator instead of X87MODEF.
987 2013-08-12  Perez Read  <netfirewall@gmail.com>
989         PR target/58132
990         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
991         operand 0 for intel asm alternative.
992         (*movabs<mode>_2): Ditto for operand 1.
994 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
996         * config/aarch64/arm_none.h
997         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
999 2013-08-12  Nick Clifton  <nickc@redhat.com>
1001         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
1003 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
1005         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
1006         expand for QI/HImode operand to produce more effictive code for
1007         unsigned char(short) --> float(double) conversion.
1009 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
1011         * doc/invoke.texi: Mention that -ftls-model does not force the final
1012         model.
1014 2013-08-12  Marek Polacek  <polacek@redhat.com>
1015             Marc Glisse  <marc.glisse@inria.fr>
1017         PR tree-optimization/57980
1018         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
1019         when creating -1 constant.
1021 2013-08-10  Jan Hubicka  <jh@suse.cz>
1023         Workaround binutils PR14342.
1024         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
1025         (gimple_init_edge_profiler): Likewise.
1026         (gimple_gen_ic_func_profiler): Likewise.
1028 2013-08-09  Jan Hubicka  <jh@suse.cz>
1030         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
1032 2013-08-09  Xinliang David Li  <davidxl@google.com>
1034         * config/i386/stringop.def: New file.
1035         * config/i386/stringop.opt: New file.
1036         * config/i386/i386-opts.h: Include stringopt.def.
1037         * config/i386/i386.opt: Include stringopt.opt.
1038         * config/i386/i386.c (ix86_option_override_internal):
1039         Override default size based stringop inline strategies with options.
1040         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
1041         New function.
1043 2013-08-09  Jan Hubicka  <jh@suse.cz>
1045         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
1047 2013-08-09  Jan Hubicka  <jh@suse.cz>
1049         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
1050         CGRAPH_FREQ_MAX.
1051         (dump_cgraph_node): Dump profile-id.
1052         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1053         and common_target_probability.
1054         * lto-cgraph.c (lto_output_edge): Stream common targets.
1055         (lto_output_node): Stream profile ids.
1056         (input_node): Stream profile ids.
1057         (input_edge): Stream common targets.
1058         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1059         * ipa.c: Include value-prof.h
1060         (ipa_profile_generate_summary): Turn indirect call statement histograms
1061         into common targets.
1062         (ipa_profile): Turn common targets into speculative edges.
1064 2013-08-09  Jan Hubicka  <jh@suse.cz>
1066         * cgraph.h (cgraph_node): Add profile_id.
1067         * value-prof.c (cgraph_node_map): Turn into pointer_map.
1068         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1069         (del_node_map): Update.
1070         (find_func_by_funcdef_no): Replace by ...
1071         (find_func_by_profile_id): ... this one.
1072         (gimple_ic_transform): Do not remove useful histograms when
1073         speculation is not done; dump info when indirect call removal
1074         can happen at LTO.
1075         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1076         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1077         (__gcov_indirect_call_profiler_v2): .. this one.
1078         * profile.h (init_node_map): Update.
1079         * coverage.c (coverage_compute_profile_id): New function.
1080         * coverage.h (coverage_compute_profile_id): Declare.
1081         * tree-profile.c (init_ic_make_global_vars): Make
1082         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
1083         (gimple_init_edge_profiler): Update prototype of
1084         __gcov_indirect_call_profiler.
1085         (gimple_gen_ic_func_profiler): Simplify.
1086         (tree_profiling): Use init_node_map
1088 2013-08-09  Jan Hubicka  <jh@suse.cz>
1090         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1091         non-speculative refs.
1092         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1093         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1094         (cgraph_set_call_stmt): Likewise.
1095         (cgraph_create_edge_1): Fix release checking compilatoin;
1096         clear lto_stmt_uid.
1097         (cgraph_free_edge): Free indirect info.
1098         (cgraph_turn_edge_to_speculative): New function.
1099         (cgraph_speculative_call_info): New function.
1100         (cgraph_make_edge_direct): Return direct edge; handle speculation.
1101         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1102         (dump_cgraph_node): Dump speculation.
1103         (verify_edge_count_and_frequency): Accept speculative edges.
1104         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1105         (verify_cgraph_node): Handle speculation.
1106         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1107         (cgraph_set_call_stmt): Update prototype.
1108         (cgraph_make_edge_direct): Update prototype.
1109         (cgraph_speculative_call_info): Declare.
1110         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1111         to change; update call of ipa_find_references.
1112         * ipa-ref.c (ipa_record_reference): Fix return value; clear
1113         lto_stmt_uid and speculative flags.
1114         (ipa_dump_references): Dump speculation.
1115         (ipa_clone_references): Clone speculative flag.
1116         (ipa_clone_referring): Likewise.
1117         (ipa_clone_ref): New function.
1118         (ipa_find_reference): Look into lto_stmt_uids
1119         (ipa_clear_stmts_in_references): Do not clear speculative calls.
1120         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1121         (ipa_find_reference): Update declaration.
1122         (ipa_clone_ref): Declare.
1123         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1124         stream speculative flag.
1125         (lto_output_ref): Stream statements uids and speculation.
1126         (input_ref): Likewise.
1127         (input_edge): Stream speuclation.
1128         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1129         (cgraph_set_call_stmt_including_clones): Handle speculation.
1130         * ipa-inline.c (heap_edge_removal_hook): New function.
1131         (inline_small_functions): Register it.
1132         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1133         also initialize refs.
1134         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1135         edge to change.
1136         (try_make_edge_direct_simple_call): Likewise.
1137         (try_make_edge_direct_simple_call): Likewise.
1138         (update_indirect_edges_after_inlining): Likewise.
1139         (remove_described_reference): Look proper lto_stmt_uid.
1140         (propagate_controlled_uses): Likewise.
1141         (propagate_controlled_uses): Liekwise.
1142         * tree-inline.c (copy_bb): Copy speculative edges.
1143         (redirect_all_calls): New function.
1144         (copy_cfg_body): Do redirection after loop info is updated.
1145         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1147 2013-08-09  Jan Hubicka  <jh@suse.cz>
1149         * lto-streamer-out.c (output_function): Renumber PHIs.
1150         * lto-streamer-in.c (input_function): Likewise.
1152 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
1154         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1155         (get_lane_unsigned): Likewise.
1156         (dup_lane_scalar): Likewise.
1157         (get_lane): enable for VALL.
1158         * config/aarch64/aarch64-simd.md
1159         (aarch64_dup_lane_scalar<mode>): Remove.
1160         (aarch64_get_lane_signed<mode>): Likewise.
1161         (aarch64_get_lane_unsigned<mode>): Likewise.
1162         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1163         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1164         (aarch64_get_lane<mode>): Enable for all vector modes.
1165         (aarch64_get_lanedi): Remove misleading constraints.
1166         * config/aarch64/arm_neon.h
1167         (__aarch64_vget_lane_any): Define.
1168         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1169         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1170         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1171         * config/aarch64/iterators.md (VDQQH): New.
1172         (VDQQHS): Likewise.
1173         (vwcore): Likewise.
1175 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
1177         * configure.ac: Add GAS check for LEON instructions on SPARC.
1178         * configure: Regenerate.
1179         * config.in: Likewise.
1180         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1181         sparc*-*-* block.
1182         * config/sparc/sparc.opt (LEON, LEON3): New masks.
1183         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1184         for LEON or LEON3.
1185         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1186         (AS_LEON_FLAG): New macro.
1187         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1188         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1189         Deal with LEON and LEON3 for the memory model.
1190         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1191         (atomic_compare_and_swap<mode>_1): Likewise.
1192         (*atomic_compare_and_swap<mode>_1): Likewise.
1194 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1196         * config/arm/neon.md (vcond): Fix floating-point vector
1197         comparisons against 0.
1199 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
1201         * lra-constraints.c (emit_spill_move): Remove assert.
1202         (process_alt_operands): Add more debugging
1203         output.  Increase reject for spilling into memory.  Decrease
1204         reject for reloading scratch.
1205         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1207 2013-08-08  Steve Ellcey  <sellcey@mips.com>
1209         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1210         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1211         micromips incompatible.  Add nan2008.
1212         (MULTILIB_DIRNAMES): Add nan2008.
1213         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1214         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1215         and micromips incompatible.  Add nan2008.
1216         (MULTILIB_DIRNAMES): Add nan2008.
1217         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1219 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
1221         PR rtl-optimization/58079
1222         * combine.c (combine_simplify_rtx): Avoid using SUBST if
1223         simplify_comparison has widened a comparison with an integer.
1225 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1227         * config/arm/neon.md (movmisalign<mode>): Disable when we
1228         don't allow unaligned accesses.
1229         (*movmisalign<mode>_neon_store): Likewise.
1230         (*movmisalign<mode>_neon_load): Likewise.
1231         (*movmisalign<mode>_neon_store): Likewise.
1232         (*movmisalign<mode>_neon_load): Likewise.
1234 2013-08-08  Jan Hubicka  <jh@suse.cz>
1236         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1237         (make_pass_rebuild_cgraph_edges): Also clear references.
1238         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1239         * ipa-inline-transform.c (inline_transform): Remove all references
1240         after inlining.
1241         * cgraphunit.c (expand_function): Remove all references after
1242         expansion.
1243         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1244         (ipa_find_reference): Rewrite to iterator.
1245         (remove_stmt_references): Likewise.
1246         (ipa_clear_stmts_in_references): New function.
1247         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
1248         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
1249         clear references.
1250         * ipa-split.c (split_function): Remove references in split function.
1252 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1254         PR target/57431
1255         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
1256         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
1258 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1260         PR target/56979
1261         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
1262         suggested mode for the assignment isn't compatible with the
1263         registers required.
1265 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1267         PR target/58065
1268         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
1270 2013-08-07  Xinliang David Li  <davidxl@google.com>
1272         * config/i386/i386.opt: New option -mtune-ctrl=.
1273         * config/i386/x86-tune.def: New file.
1274         * config/i386/i386.h: include x86-tune.def.
1275         * config/i386/i386.c (ix86_option_override_internal):
1276         Parsing -mtune-ctrl= option and set tune features.
1278 2013-08-07  Oleg Endo <olegendo@gcc.gnu.org>
1280         PR other/12081
1281         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
1282         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
1283         to gen_2arg_fn_t.
1285 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1287         * rtl.h (update_alignments): Declare.
1288         * final.c (grow_label_align): New function extracted from...
1289         (shorten_branches): ...here.  Call it.
1290         (update_alignments): New function.
1291         * reorg.c (sibling_labels): New variable.
1292         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
1293         the new label along with it onto the sibling_labels vector.
1294         (fill_simple_delay_slots): Adjust call to get_label_before.
1295         (fill_slots_from_thread): Likewise.
1296         (relax_delay_slots): Likewise.
1297         (make_return_insns): Likewise.
1298         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
1300 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1302         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
1303         document its semantics.
1304         (diagnostic_report_diagnostic): Adjust accordingly.
1306 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1308         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
1309         (sparc_option_override): ...and port to new C++ pass API.
1310         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
1312 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
1314         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
1316 2013-08-06  Caroline Tice  <cmtice@google.com>
1318         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
1319         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
1320         * tree-pass.h: Add pass_vtable_verify.
1321         * varasm.c (assemble_variable): Add code to properly set the comdat
1322         section and name for the .vtable_map_vars section.
1323         (assemble_vtyv_preinit_initializer): New function.
1324         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
1325         LINK_ONCE flag.
1326         * output.h: Add function decl for assemble_vtv_preinit_initializer.
1327         * vtable-verify.c: New file.
1328         * vtable-verify.h: New file.
1329         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
1330         initialiation levels.
1331         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
1332         * passes.def: Insert pass_vtable_verify.
1333         * aclocal.m4: Reorder includes.
1334         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
1335         -fvtv-counts options.
1336         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
1337         as appropriate, if -fvtable-verify=... is used.
1338         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
1339         -fvtable-verify=... is used.
1340         * Makefile.in (OBJS):  Add vtable-verify.o to list.
1341         (vtable-verify.o): Add new build rule.
1342         (GTFILES): Add vtable-verify.c to list.
1343         * common.opt (fvtable-verify=): New flag.
1344         (vtv_priority): Values for fvtable-verify= flag.
1345         (fvtv-counts): New flag.
1346         (fvtv-debug): New flag.
1347         * tree.h (save_vtable_map_decl): New extern function decl.
1349 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1351         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
1352         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
1353         (pass_data_rl78_devirt): ...new pass_data instance and...
1354         (make_pass_rl78_devirt): ...new function.
1355         (rl78_asm_file_start): Port pass registration to new C++ API.
1357 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1359         * coretypes.h (rtl_opt_pass): Add.
1360         (gcc::context): Add.
1361         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
1362         (epiphany_init): Port to new C++ pass API.
1363         (epiphany_optimize_mode_switching): Likewise.
1364         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
1365         (pass_manager::get_pass_mode_switching): New.
1366         (pass_manager::get_pass_peephole2): New.
1367         * mode-switching.c (pass_mode_switching): Add clone method.
1368         * recog.c (pass_peephole2): Add clone method.
1369         (pass_split_all_insns): Add clone method.
1371 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
1373         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
1374         (mips_option_override): ...here, porting to new C++ API for passes.
1376 2013-08-06  Jan Hubicka  <jh@suse.cz>
1378         * cgraph.c (cgraph_get_body): New function based on lto.c
1379         implementation.
1380         * cgraph.h (cgraph_get_body): Declare.
1381         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
1382         LTO paths.
1383         * cgraphunit.c (expand_function): Get body prior expanding.
1384         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
1385         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
1386         really need.
1387         * passes.c (do_per_function_toporder): Get body.
1388         * tree-inline.c (expand_call_inline): Get body prior inlining it.
1389         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
1391 2013-08-06  Martin Jambor  <mjambor@suse.cz>
1393         PR fortran/57987
1394         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
1395         is not re-finalized.  Rename second parameter to no_collect.
1397 2013-08-06  Martin Jambor  <mjambor@suse.cz>
1399         PR middle-end/58041
1400         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
1401         MEM_REF has proper alignment information.
1403 2013-08-05  Oleg Endo <olegendo@gcc.gnu.org>
1405         PR other/12081
1406         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
1407         class insn_gen_fn.
1408         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
1409         rtx (*) (rtx, ...) with insn_gen_fn.
1410         * genoutput.c (output_insn_data): Cast gen_? function pointers to
1411         insn_gen_fn::stored_funcptr.  Add initializer braces.
1413 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1415         Rewrite how instances of passes are cloned to remove assumptions
1416         about their sizes (thus allowing pass subclasses to have
1417         additional data fields, albeit non-GC-managed ones at this point).
1419         * passes.c (make_pass_instance): Now that passes have clone
1420         methods, rewrite this function to eliminate XNEW and memcpy
1421         calls that used hardcoded sizes.  Since this function no longer
1422         creates pass instances, rename it to...
1423         (add_pass_instance): ...this.  Document the old way that passes were
1424         numbered and flagged, and rework this function to continue using it.
1425         (next_pass_1): Add an initial_pass argument for use by
1426         add_pass_instance.
1427         (position_pass): When adding multiple instances of a pass, use
1428         the pass's clone method, rather than relying on the XNEW/memcpy
1429         within the former make_pass_instance (now add_pass_instance).
1430         (pass_manager::pass_manager): When invoking next_pass_1, also supply
1431         the initial instance of the current pass within the pass manager.
1433 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1435         This is the automated part of the conversion of passes from C
1436         structs to C++ classes.
1438         Patch autogenerated by refactor_passes.py from
1439         https://github.com/davidmalcolm/gcc-refactoring-scripts
1440         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
1442         * asan.c (pass_asan): Convert from a global struct to a subclass of
1443         gimple_opt_pass along with...
1444         (pass_data_asan): ...new pass_data instance and...
1445         (make_pass_asan): ...new function.
1446         (pass_asan_O0): Convert from a global struct to a subclass of
1447         gimple_opt_pass along with...
1448         (pass_data_asan_O0): ...new pass_data instance and...
1449         (make_pass_asan_O0): ...new function.
1450         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
1451         subclass of rtl_opt_pass along with...
1452         (pass_data_inc_dec): ...new pass_data instance and...
1453         (make_pass_inc_dec): ...new function.
1454         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
1455         a subclass of rtl_opt_pass along with...
1456         (pass_data_reorder_blocks): ...new pass_data instance and...
1457         (make_pass_reorder_blocks): ...new function.
1458         (pass_duplicate_computed_gotos): Convert from a global struct to a
1459         subclass of rtl_opt_pass along with...
1460         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
1461         (make_pass_duplicate_computed_gotos): ...new function.
1462         (pass_partition_blocks): Convert from a global struct to a subclass of
1463         rtl_opt_pass along with...
1464         (pass_data_partition_blocks): ...new pass_data instance and...
1465         (make_pass_partition_blocks): ...new function.
1466         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
1467         struct to a subclass of rtl_opt_pass along with...
1468         (pass_data_branch_target_load_optimize1): ...new pass_data instance
1469         and...
1470         (make_pass_branch_target_load_optimize1): ...new function.
1471         (pass_branch_target_load_optimize2): Convert from a global struct to a
1472         subclass of rtl_opt_pass along with...
1473         (pass_data_branch_target_load_optimize2): ...new pass_data instance
1474         and...
1475         (make_pass_branch_target_load_optimize2): ...new function.
1476         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
1477         of rtl_opt_pass along with...
1478         (pass_data_jump): ...new pass_data instance and...
1479         (make_pass_jump): ...new function.
1480         (pass_jump2): Convert from a global struct to a subclass of
1481         rtl_opt_pass along with...
1482         (pass_data_jump2): ...new pass_data instance and...
1483         (make_pass_jump2): ...new function.
1484         * cfgexpand.c (pass_expand): Convert from a global struct to a
1485         subclass of rtl_opt_pass along with...
1486         (pass_data_expand): ...new pass_data instance and...
1487         (make_pass_expand): ...new function.
1488         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
1489         of rtl_opt_pass along with...
1490         (pass_data_free_cfg): ...new pass_data instance and...
1491         (make_pass_free_cfg): ...new function.
1492         (pass_into_cfg_layout_mode): Convert from a global struct to a
1493         subclass of rtl_opt_pass along with...
1494         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
1495         (make_pass_into_cfg_layout_mode): ...new function.
1496         (pass_outof_cfg_layout_mode): Convert from a global struct to a
1497         subclass of rtl_opt_pass along with...
1498         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
1499         (make_pass_outof_cfg_layout_mode): ...new function.
1500         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
1501         struct to a subclass of gimple_opt_pass along with...
1502         (pass_data_build_cgraph_edges): ...new pass_data instance and...
1503         (make_pass_build_cgraph_edges): ...new function.
1504         (pass_rebuild_cgraph_edges): Convert from a global struct to a
1505         subclass of gimple_opt_pass along with...
1506         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
1507         (make_pass_rebuild_cgraph_edges): ...new function.
1508         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
1509         subclass of gimple_opt_pass along with...
1510         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
1511         and...
1512         (make_pass_remove_cgraph_callee_edges): ...new function.
1513         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
1514         struct to a subclass of rtl_opt_pass along with...
1515         (pass_data_stack_adjustments): ...new pass_data instance and...
1516         (make_pass_stack_adjustments): ...new function.
1517         * combine.c (pass_combine): Convert from a global struct to a subclass
1518         of rtl_opt_pass along with...
1519         (pass_data_combine): ...new pass_data instance and...
1520         (make_pass_combine): ...new function.
1521         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
1522         global struct to a subclass of rtl_opt_pass along with...
1523         (pass_data_compare_elim_after_reload): ...new pass_data instance
1524         and...
1525         (make_pass_compare_elim_after_reload): ...new function.
1526         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
1527         of rtl_opt_pass along with...
1528         (pass_data_rtl_cprop): ...new pass_data instance and...
1529         (make_pass_rtl_cprop): ...new function.
1530         * cse.c (pass_cse): Convert from a global struct to a subclass of
1531         rtl_opt_pass along with...
1532         (pass_data_cse): ...new pass_data instance and...
1533         (make_pass_cse): ...new function.
1534         (pass_cse2): Convert from a global struct to a subclass of
1535         rtl_opt_pass along with...
1536         (pass_data_cse2): ...new pass_data instance and...
1537         (make_pass_cse2): ...new function.
1538         (pass_cse_after_global_opts): Convert from a global struct to a
1539         subclass of rtl_opt_pass along with...
1540         (pass_data_cse_after_global_opts): ...new pass_data instance and...
1541         (make_pass_cse_after_global_opts): ...new function.
1542         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
1543         of rtl_opt_pass along with...
1544         (pass_data_ud_rtl_dce): ...new pass_data instance and...
1545         (make_pass_ud_rtl_dce): ...new function.
1546         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
1547         rtl_opt_pass along with...
1548         (pass_data_fast_rtl_dce): ...new pass_data instance and...
1549         (make_pass_fast_rtl_dce): ...new function.
1550         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
1551         a subclass of rtl_opt_pass along with...
1552         (pass_data_df_initialize_opt): ...new pass_data instance and...
1553         (make_pass_df_initialize_opt): ...new function.
1554         (pass_df_initialize_no_opt): Convert from a global struct to a
1555         subclass of rtl_opt_pass along with...
1556         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
1557         (make_pass_df_initialize_no_opt): ...new function.
1558         (pass_df_finish): Convert from a global struct to a subclass of
1559         rtl_opt_pass along with...
1560         (pass_data_df_finish): ...new pass_data instance and...
1561         (make_pass_df_finish): ...new function.
1562         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
1563         rtl_opt_pass along with...
1564         (pass_data_rtl_dse1): ...new pass_data instance and...
1565         (make_pass_rtl_dse1): ...new function.
1566         (pass_rtl_dse2): Convert from a global struct to a subclass of
1567         rtl_opt_pass along with...
1568         (pass_data_rtl_dse2): ...new pass_data instance and...
1569         (make_pass_rtl_dse2): ...new function.
1570         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
1571         subclass of rtl_opt_pass along with...
1572         (pass_data_dwarf2_frame): ...new pass_data instance and...
1573         (make_pass_dwarf2_frame): ...new function.
1574         * except.c (pass_set_nothrow_function_flags): Convert from a global
1575         struct to a subclass of rtl_opt_pass along with...
1576         (pass_data_set_nothrow_function_flags): ...new pass_data instance
1577         and...
1578         (make_pass_set_nothrow_function_flags): ...new function.
1579         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
1580         subclass of rtl_opt_pass along with...
1581         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
1582         and...
1583         (make_pass_convert_to_eh_region_ranges): ...new function.
1584         * final.c (pass_compute_alignments): Convert from a global struct to a
1585         subclass of rtl_opt_pass along with...
1586         (pass_data_compute_alignments): ...new pass_data instance and...
1587         (make_pass_compute_alignments): ...new function.
1588         (pass_final): Convert from a global struct to a subclass of
1589         rtl_opt_pass along with...
1590         (pass_data_final): ...new pass_data instance and...
1591         (make_pass_final): ...new function.
1592         (pass_shorten_branches): Convert from a global struct to a subclass of
1593         rtl_opt_pass along with...
1594         (pass_data_shorten_branches): ...new pass_data instance and...
1595         (make_pass_shorten_branches): ...new function.
1596         (pass_clean_state): Convert from a global struct to a subclass of
1597         rtl_opt_pass along with...
1598         (pass_data_clean_state): ...new pass_data instance and...
1599         (make_pass_clean_state): ...new function.
1600         * function.c (pass_instantiate_virtual_regs): Convert from a global
1601         struct to a subclass of rtl_opt_pass along with...
1602         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
1603         (make_pass_instantiate_virtual_regs): ...new function.
1604         (pass_leaf_regs): Convert from a global struct to a subclass of
1605         rtl_opt_pass along with...
1606         (pass_data_leaf_regs): ...new pass_data instance and...
1607         (make_pass_leaf_regs): ...new function.
1608         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
1609         subclass of rtl_opt_pass along with...
1610         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
1611         and...
1612         (make_pass_thread_prologue_and_epilogue): ...new function.
1613         (pass_match_asm_constraints): Convert from a global struct to a
1614         subclass of rtl_opt_pass along with...
1615         (pass_data_match_asm_constraints): ...new pass_data instance and...
1616         (make_pass_match_asm_constraints): ...new function.
1617         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
1618         subclass of rtl_opt_pass along with...
1619         (pass_data_rtl_fwprop): ...new pass_data instance and...
1620         (make_pass_rtl_fwprop): ...new function.
1621         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
1622         rtl_opt_pass along with...
1623         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
1624         (make_pass_rtl_fwprop_addr): ...new function.
1625         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
1626         rtl_opt_pass along with...
1627         (pass_data_rtl_pre): ...new pass_data instance and...
1628         (make_pass_rtl_pre): ...new function.
1629         (pass_rtl_hoist): Convert from a global struct to a subclass of
1630         rtl_opt_pass along with...
1631         (pass_data_rtl_hoist): ...new pass_data instance and...
1632         (make_pass_rtl_hoist): ...new function.
1633         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
1634         subclass of gimple_opt_pass along with...
1635         (pass_data_lower_cf): ...new pass_data instance and...
1636         (make_pass_lower_cf): ...new function.
1637         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
1638         from a global struct to a subclass of gimple_opt_pass along with...
1639         (pass_data_strength_reduction): ...new pass_data instance and...
1640         (make_pass_strength_reduction): ...new function.
1641         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
1642         of rtl_opt_pass along with...
1643         (pass_data_rtl_ifcvt): ...new pass_data instance and...
1644         (make_pass_rtl_ifcvt): ...new function.
1645         (pass_if_after_combine): Convert from a global struct to a subclass of
1646         rtl_opt_pass along with...
1647         (pass_data_if_after_combine): ...new pass_data instance and...
1648         (make_pass_if_after_combine): ...new function.
1649         (pass_if_after_reload): Convert from a global struct to a subclass of
1650         rtl_opt_pass along with...
1651         (pass_data_if_after_reload): ...new pass_data instance and...
1652         (make_pass_if_after_reload): ...new function.
1653         * init-regs.c (pass_initialize_regs): Convert from a global struct to
1654         a subclass of rtl_opt_pass along with...
1655         (pass_data_initialize_regs): ...new pass_data instance and...
1656         (make_pass_initialize_regs): ...new function.
1657         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
1658         of ipa_opt_pass_d along with...
1659         (pass_data_ipa_cp): ...new pass_data instance and...
1660         (make_pass_ipa_cp): ...new function.
1661         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
1662         global struct to a subclass of gimple_opt_pass along with...
1663         (pass_data_inline_parameters): ...new pass_data instance and...
1664         (make_pass_inline_parameters): ...new function.
1665         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
1666         subclass of gimple_opt_pass along with...
1667         (pass_data_early_inline): ...new pass_data instance and...
1668         (make_pass_early_inline): ...new function.
1669         (pass_ipa_inline): Convert from a global struct to a subclass of
1670         ipa_opt_pass_d along with...
1671         (pass_data_ipa_inline): ...new pass_data instance and...
1672         (make_pass_ipa_inline): ...new function.
1673         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
1674         struct to a subclass of gimple_opt_pass along with...
1675         (pass_data_local_pure_const): ...new pass_data instance and...
1676         (make_pass_local_pure_const): ...new function.
1677         (pass_ipa_pure_const): Convert from a global struct to a subclass of
1678         ipa_opt_pass_d along with...
1679         (pass_data_ipa_pure_const): ...new pass_data instance and...
1680         (make_pass_ipa_pure_const): ...new function.
1681         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
1682         to a subclass of ipa_opt_pass_d along with...
1683         (pass_data_ipa_reference): ...new pass_data instance and...
1684         (make_pass_ipa_reference): ...new function.
1685         * ipa-split.c (pass_split_functions): Convert from a global struct to
1686         a subclass of gimple_opt_pass along with...
1687         (pass_data_split_functions): ...new pass_data instance and...
1688         (make_pass_split_functions): ...new function.
1689         (pass_feedback_split_functions): Convert from a global struct to a
1690         subclass of gimple_opt_pass along with...
1691         (pass_data_feedback_split_functions): ...new pass_data instance and...
1692         (make_pass_feedback_split_functions): ...new function.
1693         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
1694         global struct to a subclass of simple_ipa_opt_pass along with...
1695         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
1696         instance and...
1697         (make_pass_ipa_function_and_variable_visibility): ...new function.
1698         (pass_ipa_free_inline_summary): Convert from a global struct to a
1699         subclass of simple_ipa_opt_pass along with...
1700         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
1701         (make_pass_ipa_free_inline_summary): ...new function.
1702         (pass_ipa_whole_program_visibility): Convert from a global struct to a
1703         subclass of ipa_opt_pass_d along with...
1704         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
1705         and...
1706         (make_pass_ipa_whole_program_visibility): ...new function.
1707         (pass_ipa_profile): Convert from a global struct to a subclass of
1708         ipa_opt_pass_d along with...
1709         (pass_data_ipa_profile): ...new pass_data instance and...
1710         (make_pass_ipa_profile): ...new function.
1711         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
1712         ipa_opt_pass_d along with...
1713         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
1714         (make_pass_ipa_cdtor_merge): ...new function.
1715         * ira.c (pass_ira): Convert from a global struct to a subclass of
1716         rtl_opt_pass along with...
1717         (pass_data_ira): ...new pass_data instance and...
1718         (make_pass_ira): ...new function.
1719         (pass_reload): Convert from a global struct to a subclass of
1720         rtl_opt_pass along with...
1721         (pass_data_reload): ...new pass_data instance and...
1722         (make_pass_reload): ...new function.
1723         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
1724         subclass of rtl_opt_pass along with...
1725         (pass_data_cleanup_barriers): ...new pass_data instance and...
1726         (make_pass_cleanup_barriers): ...new function.
1727         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
1728         of rtl_opt_pass along with...
1729         (pass_data_loop2): ...new pass_data instance and...
1730         (make_pass_loop2): ...new function.
1731         (pass_rtl_loop_init): Convert from a global struct to a subclass of
1732         rtl_opt_pass along with...
1733         (pass_data_rtl_loop_init): ...new pass_data instance and...
1734         (make_pass_rtl_loop_init): ...new function.
1735         (pass_rtl_loop_done): Convert from a global struct to a subclass of
1736         rtl_opt_pass along with...
1737         (pass_data_rtl_loop_done): ...new pass_data instance and...
1738         (make_pass_rtl_loop_done): ...new function.
1739         (pass_rtl_move_loop_invariants): Convert from a global struct to a
1740         subclass of rtl_opt_pass along with...
1741         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
1742         (make_pass_rtl_move_loop_invariants): ...new function.
1743         (pass_rtl_unswitch): Convert from a global struct to a subclass of
1744         rtl_opt_pass along with...
1745         (pass_data_rtl_unswitch): ...new pass_data instance and...
1746         (make_pass_rtl_unswitch): ...new function.
1747         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
1748         subclass of rtl_opt_pass along with...
1749         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
1750         and...
1751         (make_pass_rtl_unroll_and_peel_loops): ...new function.
1752         (pass_rtl_doloop): Convert from a global struct to a subclass of
1753         rtl_opt_pass along with...
1754         (pass_data_rtl_doloop): ...new pass_data instance and...
1755         (make_pass_rtl_doloop): ...new function.
1756         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
1757         a subclass of rtl_opt_pass along with...
1758         (pass_data_lower_subreg): ...new pass_data instance and...
1759         (make_pass_lower_subreg): ...new function.
1760         (pass_lower_subreg2): Convert from a global struct to a subclass of
1761         rtl_opt_pass along with...
1762         (pass_data_lower_subreg2): ...new pass_data instance and...
1763         (make_pass_lower_subreg2): ...new function.
1764         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
1765         struct to a subclass of ipa_opt_pass_d along with...
1766         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
1767         (make_pass_ipa_lto_gimple_out): ...new function.
1768         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
1769         of ipa_opt_pass_d along with...
1770         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
1771         (make_pass_ipa_lto_finish_out): ...new function.
1772         * mode-switching.c (pass_mode_switching): Convert from a global struct
1773         to a subclass of rtl_opt_pass along with...
1774         (pass_data_mode_switching): ...new pass_data instance and...
1775         (make_pass_mode_switching): ...new function.
1776         * modulo-sched.c (pass_sms): Convert from a global struct to a
1777         subclass of rtl_opt_pass along with...
1778         (pass_data_sms): ...new pass_data instance and...
1779         (make_pass_sms): ...new function.
1780         * omp-low.c (pass_expand_omp): Convert from a global struct to a
1781         subclass of gimple_opt_pass along with...
1782         (pass_data_expand_omp): ...new pass_data instance and...
1783         (make_pass_expand_omp): ...new function.
1784         (pass_lower_omp): Convert from a global struct to a subclass of
1785         gimple_opt_pass along with...
1786         (pass_data_lower_omp): ...new pass_data instance and...
1787         (make_pass_lower_omp): ...new function.
1788         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
1789         of gimple_opt_pass along with...
1790         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
1791         (make_pass_diagnose_omp_blocks): ...new function.
1792         * passes.c (pass_early_local_passes): Convert from a global struct to
1793         a subclass of simple_ipa_opt_pass along with...
1794         (pass_data_early_local_passes): ...new pass_data instance and...
1795         (make_pass_early_local_passes): ...new function.
1796         (pass_all_early_optimizations): Convert from a global struct to a
1797         subclass of gimple_opt_pass along with...
1798         (pass_data_all_early_optimizations): ...new pass_data instance and...
1799         (make_pass_all_early_optimizations): ...new function.
1800         (pass_all_optimizations): Convert from a global struct to a subclass
1801         of gimple_opt_pass along with...
1802         (pass_data_all_optimizations): ...new pass_data instance and...
1803         (make_pass_all_optimizations): ...new function.
1804         (pass_all_optimizations_g): Convert from a global struct to a subclass
1805         of gimple_opt_pass along with...
1806         (pass_data_all_optimizations_g): ...new pass_data instance and...
1807         (make_pass_all_optimizations_g): ...new function.
1808         (pass_rest_of_compilation): Convert from a global struct to a subclass
1809         of rtl_opt_pass along with...
1810         (pass_data_rest_of_compilation): ...new pass_data instance and...
1811         (make_pass_rest_of_compilation): ...new function.
1812         (pass_postreload): Convert from a global struct to a subclass of
1813         rtl_opt_pass along with...
1814         (pass_data_postreload): ...new pass_data instance and...
1815         (make_pass_postreload): ...new function.
1816         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
1817         subclass of rtl_opt_pass along with...
1818         (pass_data_gcse2): ...new pass_data instance and...
1819         (make_pass_gcse2): ...new function.
1820         * postreload.c (pass_postreload_cse): Convert from a global struct to
1821         a subclass of rtl_opt_pass along with...
1822         (pass_data_postreload_cse): ...new pass_data instance and...
1823         (make_pass_postreload_cse): ...new function.
1824         * predict.c (pass_profile): Convert from a global struct to a subclass
1825         of gimple_opt_pass along with...
1826         (pass_data_profile): ...new pass_data instance and...
1827         (make_pass_profile): ...new function.
1828         (pass_strip_predict_hints): Convert from a global struct to a subclass
1829         of gimple_opt_pass along with...
1830         (pass_data_strip_predict_hints): ...new pass_data instance and...
1831         (make_pass_strip_predict_hints): ...new function.
1832         * recog.c (pass_peephole2): Convert from a global struct to a subclass
1833         of rtl_opt_pass along with...
1834         (pass_data_peephole2): ...new pass_data instance and...
1835         (make_pass_peephole2): ...new function.
1836         (pass_split_all_insns): Convert from a global struct to a subclass of
1837         rtl_opt_pass along with...
1838         (pass_data_split_all_insns): ...new pass_data instance and...
1839         (make_pass_split_all_insns): ...new function.
1840         (pass_split_after_reload): Convert from a global struct to a subclass
1841         of rtl_opt_pass along with...
1842         (pass_data_split_after_reload): ...new pass_data instance and...
1843         (make_pass_split_after_reload): ...new function.
1844         (pass_split_before_regstack): Convert from a global struct to a
1845         subclass of rtl_opt_pass along with...
1846         (pass_data_split_before_regstack): ...new pass_data instance and...
1847         (make_pass_split_before_regstack): ...new function.
1848         (pass_split_before_sched2): Convert from a global struct to a subclass
1849         of rtl_opt_pass along with...
1850         (pass_data_split_before_sched2): ...new pass_data instance and...
1851         (make_pass_split_before_sched2): ...new function.
1852         (pass_split_for_shorten_branches): Convert from a global struct to a
1853         subclass of rtl_opt_pass along with...
1854         (pass_data_split_for_shorten_branches): ...new pass_data instance
1855         and...
1856         (make_pass_split_for_shorten_branches): ...new function.
1857         * ree.c (pass_ree): Convert from a global struct to a subclass of
1858         rtl_opt_pass along with...
1859         (pass_data_ree): ...new pass_data instance and...
1860         (make_pass_ree): ...new function.
1861         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
1862         subclass of rtl_opt_pass along with...
1863         (pass_data_stack_regs): ...new pass_data instance and...
1864         (make_pass_stack_regs): ...new function.
1865         (pass_stack_regs_run): Convert from a global struct to a subclass of
1866         rtl_opt_pass along with...
1867         (pass_data_stack_regs_run): ...new pass_data instance and...
1868         (make_pass_stack_regs_run): ...new function.
1869         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
1870         subclass of rtl_opt_pass along with...
1871         (pass_data_cprop_hardreg): ...new pass_data instance and...
1872         (make_pass_cprop_hardreg): ...new function.
1873         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
1874         subclass of rtl_opt_pass along with...
1875         (pass_data_reginfo_init): ...new pass_data instance and...
1876         (make_pass_reginfo_init): ...new function.
1877         * regmove.c (pass_regmove): Convert from a global struct to a subclass
1878         of rtl_opt_pass along with...
1879         (pass_data_regmove): ...new pass_data instance and...
1880         (make_pass_regmove): ...new function.
1881         * regrename.c (pass_regrename): Convert from a global struct to a
1882         subclass of rtl_opt_pass along with...
1883         (pass_data_regrename): ...new pass_data instance and...
1884         (make_pass_regrename): ...new function.
1885         * reorg.c (pass_delay_slots): Convert from a global struct to a
1886         subclass of rtl_opt_pass along with...
1887         (pass_data_delay_slots): ...new pass_data instance and...
1888         (make_pass_delay_slots): ...new function.
1889         (pass_machine_reorg): Convert from a global struct to a subclass of
1890         rtl_opt_pass along with...
1891         (pass_data_machine_reorg): ...new pass_data instance and...
1892         (make_pass_machine_reorg): ...new function.
1893         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
1894         of rtl_opt_pass along with...
1895         (pass_data_sched): ...new pass_data instance and...
1896         (make_pass_sched): ...new function.
1897         (pass_sched2): Convert from a global struct to a subclass of
1898         rtl_opt_pass along with...
1899         (pass_data_sched2): ...new pass_data instance and...
1900         (make_pass_sched2): ...new function.
1901         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
1902         to a subclass of rtl_opt_pass along with...
1903         (pass_data_stack_ptr_mod): ...new pass_data instance and...
1904         (make_pass_stack_ptr_mod): ...new function.
1905         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
1906         to a subclass of rtl_opt_pass along with...
1907         (pass_data_rtl_store_motion): ...new pass_data instance and...
1908         (make_pass_rtl_store_motion): ...new function.
1909         * tracer.c (pass_tracer): Convert from a global struct to a subclass
1910         of gimple_opt_pass along with...
1911         (pass_data_tracer): ...new pass_data instance and...
1912         (make_pass_tracer): ...new function.
1913         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
1914         to a subclass of gimple_opt_pass along with...
1915         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
1916         (make_pass_diagnose_tm_blocks): ...new function.
1917         (pass_lower_tm): Convert from a global struct to a subclass of
1918         gimple_opt_pass along with...
1919         (pass_data_lower_tm): ...new pass_data instance and...
1920         (make_pass_lower_tm): ...new function.
1921         (pass_tm_init): Convert from a global struct to a subclass of
1922         gimple_opt_pass along with...
1923         (pass_data_tm_init): ...new pass_data instance and...
1924         (make_pass_tm_init): ...new function.
1925         (pass_tm_mark): Convert from a global struct to a subclass of
1926         gimple_opt_pass along with...
1927         (pass_data_tm_mark): ...new pass_data instance and...
1928         (make_pass_tm_mark): ...new function.
1929         (pass_tm_edges): Convert from a global struct to a subclass of
1930         gimple_opt_pass along with...
1931         (pass_data_tm_edges): ...new pass_data instance and...
1932         (make_pass_tm_edges): ...new function.
1933         (pass_tm_memopt): Convert from a global struct to a subclass of
1934         gimple_opt_pass along with...
1935         (pass_data_tm_memopt): ...new pass_data instance and...
1936         (make_pass_tm_memopt): ...new function.
1937         (pass_ipa_tm): Convert from a global struct to a subclass of
1938         simple_ipa_opt_pass along with...
1939         (pass_data_ipa_tm): ...new pass_data instance and...
1940         (make_pass_ipa_tm): ...new function.
1941         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
1942         subclass of gimple_opt_pass along with...
1943         (pass_data_call_cdce): ...new pass_data instance and...
1944         (make_pass_call_cdce): ...new function.
1945         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
1946         subclass of gimple_opt_pass along with...
1947         (pass_data_build_cfg): ...new pass_data instance and...
1948         (make_pass_build_cfg): ...new function.
1949         (pass_split_crit_edges): Convert from a global struct to a subclass of
1950         gimple_opt_pass along with...
1951         (pass_data_split_crit_edges): ...new pass_data instance and...
1952         (make_pass_split_crit_edges): ...new function.
1953         (pass_warn_function_return): Convert from a global struct to a
1954         subclass of gimple_opt_pass along with...
1955         (pass_data_warn_function_return): ...new pass_data instance and...
1956         (make_pass_warn_function_return): ...new function.
1957         (pass_warn_function_noreturn): Convert from a global struct to a
1958         subclass of gimple_opt_pass along with...
1959         (pass_data_warn_function_noreturn): ...new pass_data instance and...
1960         (make_pass_warn_function_noreturn): ...new function.
1961         (pass_warn_unused_result): Convert from a global struct to a subclass
1962         of gimple_opt_pass along with...
1963         (pass_data_warn_unused_result): ...new pass_data instance and...
1964         (make_pass_warn_unused_result): ...new function.
1965         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
1966         a subclass of gimple_opt_pass along with...
1967         (pass_data_merge_phi): ...new pass_data instance and...
1968         (make_pass_merge_phi): ...new function.
1969         * tree-complex.c (pass_lower_complex): Convert from a global struct to
1970         a subclass of gimple_opt_pass along with...
1971         (pass_data_lower_complex): ...new pass_data instance and...
1972         (make_pass_lower_complex): ...new function.
1973         (pass_lower_complex_O0): Convert from a global struct to a subclass of
1974         gimple_opt_pass along with...
1975         (pass_data_lower_complex_O0): ...new pass_data instance and...
1976         (make_pass_lower_complex_O0): ...new function.
1977         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
1978         subclass of gimple_opt_pass along with...
1979         (pass_data_lower_eh): ...new pass_data instance and...
1980         (make_pass_lower_eh): ...new function.
1981         (pass_refactor_eh): Convert from a global struct to a subclass of
1982         gimple_opt_pass along with...
1983         (pass_data_refactor_eh): ...new pass_data instance and...
1984         (make_pass_refactor_eh): ...new function.
1985         (pass_lower_resx): Convert from a global struct to a subclass of
1986         gimple_opt_pass along with...
1987         (pass_data_lower_resx): ...new pass_data instance and...
1988         (make_pass_lower_resx): ...new function.
1989         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1990         of gimple_opt_pass along with...
1991         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1992         (make_pass_lower_eh_dispatch): ...new function.
1993         (pass_cleanup_eh): Convert from a global struct to a subclass of
1994         gimple_opt_pass along with...
1995         (pass_data_cleanup_eh): ...new pass_data instance and...
1996         (make_pass_cleanup_eh): ...new function.
1997         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1998         to a subclass of simple_ipa_opt_pass along with...
1999         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
2000         (make_pass_ipa_lower_emutls): ...new function.
2001         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
2002         a subclass of gimple_opt_pass along with...
2003         (pass_data_if_conversion): ...new pass_data instance and...
2004         (make_pass_if_conversion): ...new function.
2005         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
2006         subclass of gimple_opt_pass along with...
2007         (pass_data_build_ssa): ...new pass_data instance and...
2008         (make_pass_build_ssa): ...new function.
2009         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
2010         global struct to a subclass of gimple_opt_pass along with...
2011         (pass_data_loop_distribution): ...new pass_data instance and...
2012         (make_pass_loop_distribution): ...new function.
2013         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
2014         subclass of gimple_opt_pass along with...
2015         (pass_data_mudflap_1): ...new pass_data instance and...
2016         (make_pass_mudflap_1): ...new function.
2017         (pass_mudflap_2): Convert from a global struct to a subclass of
2018         gimple_opt_pass along with...
2019         (pass_data_mudflap_2): ...new pass_data instance and...
2020         (make_pass_mudflap_2): ...new function.
2021         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
2022         subclass of gimple_opt_pass along with...
2023         (pass_data_mudflap_1): ...new pass_data instance and...
2024         (make_pass_mudflap_1): ...new function.
2025         (pass_mudflap_2): Convert from a global struct to a subclass of
2026         gimple_opt_pass along with...
2027         (pass_data_mudflap_2): ...new pass_data instance and...
2028         (make_pass_mudflap_2): ...new function.
2029         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
2030         gimple_opt_pass along with...
2031         (pass_data_nrv): ...new pass_data instance and...
2032         (make_pass_nrv): ...new function.
2033         (pass_return_slot): Convert from a global struct to a subclass of
2034         gimple_opt_pass along with...
2035         (pass_data_return_slot): ...new pass_data instance and...
2036         (make_pass_return_slot): ...new function.
2037         * tree-object-size.c (pass_object_sizes): Convert from a global struct
2038         to a subclass of gimple_opt_pass along with...
2039         (pass_data_object_sizes): ...new pass_data instance and...
2040         (make_pass_object_sizes): ...new function.
2041         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
2042         global struct to a subclass of gimple_opt_pass along with...
2043         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
2044         and...
2045         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2046         (pass_fixup_cfg): Convert from a global struct to a subclass of
2047         gimple_opt_pass along with...
2048         (pass_data_fixup_cfg): ...new pass_data instance and...
2049         (make_pass_fixup_cfg): ...new function.
2050         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
2051         (make_pass_mudflap_1): ...new function.
2052         (pass_mudflap_2): Replace declaration with that of...
2053         (make_pass_mudflap_2): ...new function.
2054         (pass_asan): Replace declaration with that of...
2055         (make_pass_asan): ...new function.
2056         (pass_asan_O0): Replace declaration with that of...
2057         (make_pass_asan_O0): ...new function.
2058         (pass_tsan): Replace declaration with that of...
2059         (make_pass_tsan): ...new function.
2060         (pass_tsan_O0): Replace declaration with that of...
2061         (make_pass_tsan_O0): ...new function.
2062         (pass_lower_cf): Replace declaration with that of...
2063         (make_pass_lower_cf): ...new function.
2064         (pass_refactor_eh): Replace declaration with that of...
2065         (make_pass_refactor_eh): ...new function.
2066         (pass_lower_eh): Replace declaration with that of...
2067         (make_pass_lower_eh): ...new function.
2068         (pass_lower_eh_dispatch): Replace declaration with that of...
2069         (make_pass_lower_eh_dispatch): ...new function.
2070         (pass_lower_resx): Replace declaration with that of...
2071         (make_pass_lower_resx): ...new function.
2072         (pass_build_cfg): Replace declaration with that of...
2073         (make_pass_build_cfg): ...new function.
2074         (pass_early_tree_profile): Replace declaration with that of...
2075         (make_pass_early_tree_profile): ...new function.
2076         (pass_cleanup_eh): Replace declaration with that of...
2077         (make_pass_cleanup_eh): ...new function.
2078         (pass_sra): Replace declaration with that of...
2079         (make_pass_sra): ...new function.
2080         (pass_sra_early): Replace declaration with that of...
2081         (make_pass_sra_early): ...new function.
2082         (pass_early_ipa_sra): Replace declaration with that of...
2083         (make_pass_early_ipa_sra): ...new function.
2084         (pass_tail_recursion): Replace declaration with that of...
2085         (make_pass_tail_recursion): ...new function.
2086         (pass_tail_calls): Replace declaration with that of...
2087         (make_pass_tail_calls): ...new function.
2088         (pass_tree_loop): Replace declaration with that of...
2089         (make_pass_tree_loop): ...new function.
2090         (pass_tree_loop_init): Replace declaration with that of...
2091         (make_pass_tree_loop_init): ...new function.
2092         (pass_lim): Replace declaration with that of...
2093         (make_pass_lim): ...new function.
2094         (pass_tree_unswitch): Replace declaration with that of...
2095         (make_pass_tree_unswitch): ...new function.
2096         (pass_predcom): Replace declaration with that of...
2097         (make_pass_predcom): ...new function.
2098         (pass_iv_canon): Replace declaration with that of...
2099         (make_pass_iv_canon): ...new function.
2100         (pass_scev_cprop): Replace declaration with that of...
2101         (make_pass_scev_cprop): ...new function.
2102         (pass_empty_loop): Replace declaration with that of...
2103         (make_pass_empty_loop): ...new function.
2104         (pass_record_bounds): Replace declaration with that of...
2105         (make_pass_record_bounds): ...new function.
2106         (pass_graphite): Replace declaration with that of...
2107         (make_pass_graphite): ...new function.
2108         (pass_graphite_transforms): Replace declaration with that of...
2109         (make_pass_graphite_transforms): ...new function.
2110         (pass_if_conversion): Replace declaration with that of...
2111         (make_pass_if_conversion): ...new function.
2112         (pass_loop_distribution): Replace declaration with that of...
2113         (make_pass_loop_distribution): ...new function.
2114         (pass_vectorize): Replace declaration with that of...
2115         (make_pass_vectorize): ...new function.
2116         (pass_slp_vectorize): Replace declaration with that of...
2117         (make_pass_slp_vectorize): ...new function.
2118         (pass_complete_unroll): Replace declaration with that of...
2119         (make_pass_complete_unroll): ...new function.
2120         (pass_complete_unrolli): Replace declaration with that of...
2121         (make_pass_complete_unrolli): ...new function.
2122         (pass_parallelize_loops): Replace declaration with that of...
2123         (make_pass_parallelize_loops): ...new function.
2124         (pass_loop_prefetch): Replace declaration with that of...
2125         (make_pass_loop_prefetch): ...new function.
2126         (pass_iv_optimize): Replace declaration with that of...
2127         (make_pass_iv_optimize): ...new function.
2128         (pass_tree_loop_done): Replace declaration with that of...
2129         (make_pass_tree_loop_done): ...new function.
2130         (pass_ch): Replace declaration with that of...
2131         (make_pass_ch): ...new function.
2132         (pass_ccp): Replace declaration with that of...
2133         (make_pass_ccp): ...new function.
2134         (pass_phi_only_cprop): Replace declaration with that of...
2135         (make_pass_phi_only_cprop): ...new function.
2136         (pass_build_ssa): Replace declaration with that of...
2137         (make_pass_build_ssa): ...new function.
2138         (pass_build_alias): Replace declaration with that of...
2139         (make_pass_build_alias): ...new function.
2140         (pass_build_ealias): Replace declaration with that of...
2141         (make_pass_build_ealias): ...new function.
2142         (pass_dominator): Replace declaration with that of...
2143         (make_pass_dominator): ...new function.
2144         (pass_dce): Replace declaration with that of...
2145         (make_pass_dce): ...new function.
2146         (pass_dce_loop): Replace declaration with that of...
2147         (make_pass_dce_loop): ...new function.
2148         (pass_cd_dce): Replace declaration with that of...
2149         (make_pass_cd_dce): ...new function.
2150         (pass_call_cdce): Replace declaration with that of...
2151         (make_pass_call_cdce): ...new function.
2152         (pass_merge_phi): Replace declaration with that of...
2153         (make_pass_merge_phi): ...new function.
2154         (pass_split_crit_edges): Replace declaration with that of...
2155         (make_pass_split_crit_edges): ...new function.
2156         (pass_pre): Replace declaration with that of...
2157         (make_pass_pre): ...new function.
2158         (pass_profile): Replace declaration with that of...
2159         (make_pass_profile): ...new function.
2160         (pass_strip_predict_hints): Replace declaration with that of...
2161         (make_pass_strip_predict_hints): ...new function.
2162         (pass_lower_complex_O0): Replace declaration with that of...
2163         (make_pass_lower_complex_O0): ...new function.
2164         (pass_lower_complex): Replace declaration with that of...
2165         (make_pass_lower_complex): ...new function.
2166         (pass_lower_vector): Replace declaration with that of...
2167         (make_pass_lower_vector): ...new function.
2168         (pass_lower_vector_ssa): Replace declaration with that of...
2169         (make_pass_lower_vector_ssa): ...new function.
2170         (pass_lower_omp): Replace declaration with that of...
2171         (make_pass_lower_omp): ...new function.
2172         (pass_diagnose_omp_blocks): Replace declaration with that of...
2173         (make_pass_diagnose_omp_blocks): ...new function.
2174         (pass_expand_omp): Replace declaration with that of...
2175         (make_pass_expand_omp): ...new function.
2176         (pass_expand_omp_ssa): Replace declaration with that of...
2177         (make_pass_expand_omp_ssa): ...new function.
2178         (pass_object_sizes): Replace declaration with that of...
2179         (make_pass_object_sizes): ...new function.
2180         (pass_strlen): Replace declaration with that of...
2181         (make_pass_strlen): ...new function.
2182         (pass_fold_builtins): Replace declaration with that of...
2183         (make_pass_fold_builtins): ...new function.
2184         (pass_stdarg): Replace declaration with that of...
2185         (make_pass_stdarg): ...new function.
2186         (pass_early_warn_uninitialized): Replace declaration with that of...
2187         (make_pass_early_warn_uninitialized): ...new function.
2188         (pass_late_warn_uninitialized): Replace declaration with that of...
2189         (make_pass_late_warn_uninitialized): ...new function.
2190         (pass_cse_reciprocals): Replace declaration with that of...
2191         (make_pass_cse_reciprocals): ...new function.
2192         (pass_cse_sincos): Replace declaration with that of...
2193         (make_pass_cse_sincos): ...new function.
2194         (pass_optimize_bswap): Replace declaration with that of...
2195         (make_pass_optimize_bswap): ...new function.
2196         (pass_optimize_widening_mul): Replace declaration with that of...
2197         (make_pass_optimize_widening_mul): ...new function.
2198         (pass_warn_function_return): Replace declaration with that of...
2199         (make_pass_warn_function_return): ...new function.
2200         (pass_warn_function_noreturn): Replace declaration with that of...
2201         (make_pass_warn_function_noreturn): ...new function.
2202         (pass_cselim): Replace declaration with that of...
2203         (make_pass_cselim): ...new function.
2204         (pass_phiopt): Replace declaration with that of...
2205         (make_pass_phiopt): ...new function.
2206         (pass_forwprop): Replace declaration with that of...
2207         (make_pass_forwprop): ...new function.
2208         (pass_phiprop): Replace declaration with that of...
2209         (make_pass_phiprop): ...new function.
2210         (pass_tree_ifcombine): Replace declaration with that of...
2211         (make_pass_tree_ifcombine): ...new function.
2212         (pass_dse): Replace declaration with that of...
2213         (make_pass_dse): ...new function.
2214         (pass_nrv): Replace declaration with that of...
2215         (make_pass_nrv): ...new function.
2216         (pass_rename_ssa_copies): Replace declaration with that of...
2217         (make_pass_rename_ssa_copies): ...new function.
2218         (pass_sink_code): Replace declaration with that of...
2219         (make_pass_sink_code): ...new function.
2220         (pass_fre): Replace declaration with that of...
2221         (make_pass_fre): ...new function.
2222         (pass_check_data_deps): Replace declaration with that of...
2223         (make_pass_check_data_deps): ...new function.
2224         (pass_copy_prop): Replace declaration with that of...
2225         (make_pass_copy_prop): ...new function.
2226         (pass_vrp): Replace declaration with that of...
2227         (make_pass_vrp): ...new function.
2228         (pass_uncprop): Replace declaration with that of...
2229         (make_pass_uncprop): ...new function.
2230         (pass_return_slot): Replace declaration with that of...
2231         (make_pass_return_slot): ...new function.
2232         (pass_reassoc): Replace declaration with that of...
2233         (make_pass_reassoc): ...new function.
2234         (pass_rebuild_cgraph_edges): Replace declaration with that of...
2235         (make_pass_rebuild_cgraph_edges): ...new function.
2236         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2237         (make_pass_remove_cgraph_callee_edges): ...new function.
2238         (pass_build_cgraph_edges): Replace declaration with that of...
2239         (make_pass_build_cgraph_edges): ...new function.
2240         (pass_local_pure_const): Replace declaration with that of...
2241         (make_pass_local_pure_const): ...new function.
2242         (pass_tracer): Replace declaration with that of...
2243         (make_pass_tracer): ...new function.
2244         (pass_warn_unused_result): Replace declaration with that of...
2245         (make_pass_warn_unused_result): ...new function.
2246         (pass_diagnose_tm_blocks): Replace declaration with that of...
2247         (make_pass_diagnose_tm_blocks): ...new function.
2248         (pass_lower_tm): Replace declaration with that of...
2249         (make_pass_lower_tm): ...new function.
2250         (pass_tm_init): Replace declaration with that of...
2251         (make_pass_tm_init): ...new function.
2252         (pass_tm_mark): Replace declaration with that of...
2253         (make_pass_tm_mark): ...new function.
2254         (pass_tm_memopt): Replace declaration with that of...
2255         (make_pass_tm_memopt): ...new function.
2256         (pass_tm_edges): Replace declaration with that of...
2257         (make_pass_tm_edges): ...new function.
2258         (pass_split_functions): Replace declaration with that of...
2259         (make_pass_split_functions): ...new function.
2260         (pass_feedback_split_functions): Replace declaration with that of...
2261         (make_pass_feedback_split_functions): ...new function.
2262         (pass_strength_reduction): Replace declaration with that of...
2263         (make_pass_strength_reduction): ...new function.
2264         (pass_ipa_lower_emutls): Replace declaration with that of...
2265         (make_pass_ipa_lower_emutls): ...new function.
2266         (pass_ipa_function_and_variable_visibility): Replace declaration with
2267         that of...
2268         (make_pass_ipa_function_and_variable_visibility): ...new function.
2269         (pass_ipa_tree_profile): Replace declaration with that of...
2270         (make_pass_ipa_tree_profile): ...new function.
2271         (pass_early_local_passes): Replace declaration with that of...
2272         (make_pass_early_local_passes): ...new function.
2273         (pass_ipa_whole_program_visibility): Replace declaration with that
2274         of...
2275         (make_pass_ipa_whole_program_visibility): ...new function.
2276         (pass_ipa_lto_gimple_out): Replace declaration with that of...
2277         (make_pass_ipa_lto_gimple_out): ...new function.
2278         (pass_ipa_increase_alignment): Replace declaration with that of...
2279         (make_pass_ipa_increase_alignment): ...new function.
2280         (pass_ipa_inline): Replace declaration with that of...
2281         (make_pass_ipa_inline): ...new function.
2282         (pass_ipa_free_lang_data): Replace declaration with that of...
2283         (make_pass_ipa_free_lang_data): ...new function.
2284         (pass_ipa_free_inline_summary): Replace declaration with that of...
2285         (make_pass_ipa_free_inline_summary): ...new function.
2286         (pass_ipa_cp): Replace declaration with that of...
2287         (make_pass_ipa_cp): ...new function.
2288         (pass_ipa_reference): Replace declaration with that of...
2289         (make_pass_ipa_reference): ...new function.
2290         (pass_ipa_pure_const): Replace declaration with that of...
2291         (make_pass_ipa_pure_const): ...new function.
2292         (pass_ipa_pta): Replace declaration with that of...
2293         (make_pass_ipa_pta): ...new function.
2294         (pass_ipa_lto_finish_out): Replace declaration with that of...
2295         (make_pass_ipa_lto_finish_out): ...new function.
2296         (pass_ipa_tm): Replace declaration with that of...
2297         (make_pass_ipa_tm): ...new function.
2298         (pass_ipa_profile): Replace declaration with that of...
2299         (make_pass_ipa_profile): ...new function.
2300         (pass_ipa_cdtor_merge): Replace declaration with that of...
2301         (make_pass_ipa_cdtor_merge): ...new function.
2302         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
2303         of...
2304         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2305         (pass_init_datastructures): Replace declaration with that of...
2306         (make_pass_init_datastructures): ...new function.
2307         (pass_fixup_cfg): Replace declaration with that of...
2308         (make_pass_fixup_cfg): ...new function.
2309         (pass_expand): Replace declaration with that of...
2310         (make_pass_expand): ...new function.
2311         (pass_instantiate_virtual_regs): Replace declaration with that of...
2312         (make_pass_instantiate_virtual_regs): ...new function.
2313         (pass_rtl_fwprop): Replace declaration with that of...
2314         (make_pass_rtl_fwprop): ...new function.
2315         (pass_rtl_fwprop_addr): Replace declaration with that of...
2316         (make_pass_rtl_fwprop_addr): ...new function.
2317         (pass_jump): Replace declaration with that of...
2318         (make_pass_jump): ...new function.
2319         (pass_jump2): Replace declaration with that of...
2320         (make_pass_jump2): ...new function.
2321         (pass_lower_subreg): Replace declaration with that of...
2322         (make_pass_lower_subreg): ...new function.
2323         (pass_cse): Replace declaration with that of...
2324         (make_pass_cse): ...new function.
2325         (pass_fast_rtl_dce): Replace declaration with that of...
2326         (make_pass_fast_rtl_dce): ...new function.
2327         (pass_ud_rtl_dce): Replace declaration with that of...
2328         (make_pass_ud_rtl_dce): ...new function.
2329         (pass_rtl_dce): Replace declaration with that of...
2330         (make_pass_rtl_dce): ...new function.
2331         (pass_rtl_dse1): Replace declaration with that of...
2332         (make_pass_rtl_dse1): ...new function.
2333         (pass_rtl_dse2): Replace declaration with that of...
2334         (make_pass_rtl_dse2): ...new function.
2335         (pass_rtl_dse3): Replace declaration with that of...
2336         (make_pass_rtl_dse3): ...new function.
2337         (pass_rtl_cprop): Replace declaration with that of...
2338         (make_pass_rtl_cprop): ...new function.
2339         (pass_rtl_pre): Replace declaration with that of...
2340         (make_pass_rtl_pre): ...new function.
2341         (pass_rtl_hoist): Replace declaration with that of...
2342         (make_pass_rtl_hoist): ...new function.
2343         (pass_rtl_store_motion): Replace declaration with that of...
2344         (make_pass_rtl_store_motion): ...new function.
2345         (pass_cse_after_global_opts): Replace declaration with that of...
2346         (make_pass_cse_after_global_opts): ...new function.
2347         (pass_rtl_ifcvt): Replace declaration with that of...
2348         (make_pass_rtl_ifcvt): ...new function.
2349         (pass_into_cfg_layout_mode): Replace declaration with that of...
2350         (make_pass_into_cfg_layout_mode): ...new function.
2351         (pass_outof_cfg_layout_mode): Replace declaration with that of...
2352         (make_pass_outof_cfg_layout_mode): ...new function.
2353         (pass_loop2): Replace declaration with that of...
2354         (make_pass_loop2): ...new function.
2355         (pass_rtl_loop_init): Replace declaration with that of...
2356         (make_pass_rtl_loop_init): ...new function.
2357         (pass_rtl_move_loop_invariants): Replace declaration with that of...
2358         (make_pass_rtl_move_loop_invariants): ...new function.
2359         (pass_rtl_unswitch): Replace declaration with that of...
2360         (make_pass_rtl_unswitch): ...new function.
2361         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
2362         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2363         (pass_rtl_doloop): Replace declaration with that of...
2364         (make_pass_rtl_doloop): ...new function.
2365         (pass_rtl_loop_done): Replace declaration with that of...
2366         (make_pass_rtl_loop_done): ...new function.
2367         (pass_web): Replace declaration with that of...
2368         (make_pass_web): ...new function.
2369         (pass_cse2): Replace declaration with that of...
2370         (make_pass_cse2): ...new function.
2371         (pass_df_initialize_opt): Replace declaration with that of...
2372         (make_pass_df_initialize_opt): ...new function.
2373         (pass_df_initialize_no_opt): Replace declaration with that of...
2374         (make_pass_df_initialize_no_opt): ...new function.
2375         (pass_reginfo_init): Replace declaration with that of...
2376         (make_pass_reginfo_init): ...new function.
2377         (pass_inc_dec): Replace declaration with that of...
2378         (make_pass_inc_dec): ...new function.
2379         (pass_stack_ptr_mod): Replace declaration with that of...
2380         (make_pass_stack_ptr_mod): ...new function.
2381         (pass_initialize_regs): Replace declaration with that of...
2382         (make_pass_initialize_regs): ...new function.
2383         (pass_combine): Replace declaration with that of...
2384         (make_pass_combine): ...new function.
2385         (pass_if_after_combine): Replace declaration with that of...
2386         (make_pass_if_after_combine): ...new function.
2387         (pass_ree): Replace declaration with that of...
2388         (make_pass_ree): ...new function.
2389         (pass_partition_blocks): Replace declaration with that of...
2390         (make_pass_partition_blocks): ...new function.
2391         (pass_match_asm_constraints): Replace declaration with that of...
2392         (make_pass_match_asm_constraints): ...new function.
2393         (pass_regmove): Replace declaration with that of...
2394         (make_pass_regmove): ...new function.
2395         (pass_split_all_insns): Replace declaration with that of...
2396         (make_pass_split_all_insns): ...new function.
2397         (pass_fast_rtl_byte_dce): Replace declaration with that of...
2398         (make_pass_fast_rtl_byte_dce): ...new function.
2399         (pass_lower_subreg2): Replace declaration with that of...
2400         (make_pass_lower_subreg2): ...new function.
2401         (pass_mode_switching): Replace declaration with that of...
2402         (make_pass_mode_switching): ...new function.
2403         (pass_sms): Replace declaration with that of...
2404         (make_pass_sms): ...new function.
2405         (pass_sched): Replace declaration with that of...
2406         (make_pass_sched): ...new function.
2407         (pass_ira): Replace declaration with that of...
2408         (make_pass_ira): ...new function.
2409         (pass_reload): Replace declaration with that of...
2410         (make_pass_reload): ...new function.
2411         (pass_clean_state): Replace declaration with that of...
2412         (make_pass_clean_state): ...new function.
2413         (pass_branch_prob): Replace declaration with that of...
2414         (make_pass_branch_prob): ...new function.
2415         (pass_value_profile_transformations): Replace declaration with that
2416         of...
2417         (make_pass_value_profile_transformations): ...new function.
2418         (pass_postreload_cse): Replace declaration with that of...
2419         (make_pass_postreload_cse): ...new function.
2420         (pass_gcse2): Replace declaration with that of...
2421         (make_pass_gcse2): ...new function.
2422         (pass_split_after_reload): Replace declaration with that of...
2423         (make_pass_split_after_reload): ...new function.
2424         (pass_branch_target_load_optimize1): Replace declaration with that
2425         of...
2426         (make_pass_branch_target_load_optimize1): ...new function.
2427         (pass_thread_prologue_and_epilogue): Replace declaration with that
2428         of...
2429         (make_pass_thread_prologue_and_epilogue): ...new function.
2430         (pass_stack_adjustments): Replace declaration with that of...
2431         (make_pass_stack_adjustments): ...new function.
2432         (pass_peephole2): Replace declaration with that of...
2433         (make_pass_peephole2): ...new function.
2434         (pass_if_after_reload): Replace declaration with that of...
2435         (make_pass_if_after_reload): ...new function.
2436         (pass_regrename): Replace declaration with that of...
2437         (make_pass_regrename): ...new function.
2438         (pass_cprop_hardreg): Replace declaration with that of...
2439         (make_pass_cprop_hardreg): ...new function.
2440         (pass_reorder_blocks): Replace declaration with that of...
2441         (make_pass_reorder_blocks): ...new function.
2442         (pass_branch_target_load_optimize2): Replace declaration with that
2443         of...
2444         (make_pass_branch_target_load_optimize2): ...new function.
2445         (pass_leaf_regs): Replace declaration with that of...
2446         (make_pass_leaf_regs): ...new function.
2447         (pass_split_before_sched2): Replace declaration with that of...
2448         (make_pass_split_before_sched2): ...new function.
2449         (pass_compare_elim_after_reload): Replace declaration with that of...
2450         (make_pass_compare_elim_after_reload): ...new function.
2451         (pass_sched2): Replace declaration with that of...
2452         (make_pass_sched2): ...new function.
2453         (pass_stack_regs): Replace declaration with that of...
2454         (make_pass_stack_regs): ...new function.
2455         (pass_stack_regs_run): Replace declaration with that of...
2456         (make_pass_stack_regs_run): ...new function.
2457         (pass_df_finish): Replace declaration with that of...
2458         (make_pass_df_finish): ...new function.
2459         (pass_compute_alignments): Replace declaration with that of...
2460         (make_pass_compute_alignments): ...new function.
2461         (pass_duplicate_computed_gotos): Replace declaration with that of...
2462         (make_pass_duplicate_computed_gotos): ...new function.
2463         (pass_variable_tracking): Replace declaration with that of...
2464         (make_pass_variable_tracking): ...new function.
2465         (pass_free_cfg): Replace declaration with that of...
2466         (make_pass_free_cfg): ...new function.
2467         (pass_machine_reorg): Replace declaration with that of...
2468         (make_pass_machine_reorg): ...new function.
2469         (pass_cleanup_barriers): Replace declaration with that of...
2470         (make_pass_cleanup_barriers): ...new function.
2471         (pass_delay_slots): Replace declaration with that of...
2472         (make_pass_delay_slots): ...new function.
2473         (pass_split_for_shorten_branches): Replace declaration with that of...
2474         (make_pass_split_for_shorten_branches): ...new function.
2475         (pass_split_before_regstack): Replace declaration with that of...
2476         (make_pass_split_before_regstack): ...new function.
2477         (pass_convert_to_eh_region_ranges): Replace declaration with that
2478         of...
2479         (make_pass_convert_to_eh_region_ranges): ...new function.
2480         (pass_shorten_branches): Replace declaration with that of...
2481         (make_pass_shorten_branches): ...new function.
2482         (pass_set_nothrow_function_flags): Replace declaration with that of...
2483         (make_pass_set_nothrow_function_flags): ...new function.
2484         (pass_dwarf2_frame): Replace declaration with that of...
2485         (make_pass_dwarf2_frame): ...new function.
2486         (pass_final): Replace declaration with that of...
2487         (make_pass_final): ...new function.
2488         (pass_rtl_seqabstr): Replace declaration with that of...
2489         (make_pass_rtl_seqabstr): ...new function.
2490         (pass_release_ssa_names): Replace declaration with that of...
2491         (make_pass_release_ssa_names): ...new function.
2492         (pass_early_inline): Replace declaration with that of...
2493         (make_pass_early_inline): ...new function.
2494         (pass_inline_parameters): Replace declaration with that of...
2495         (make_pass_inline_parameters): ...new function.
2496         (pass_update_address_taken): Replace declaration with that of...
2497         (make_pass_update_address_taken): ...new function.
2498         (pass_convert_switch): Replace declaration with that of...
2499         (make_pass_convert_switch): ...new function.
2500         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
2501         to a subclass of simple_ipa_opt_pass along with...
2502         (pass_data_ipa_tree_profile): ...new pass_data instance and...
2503         (make_pass_ipa_tree_profile): ...new function.
2504         * tree-sra.c (pass_sra_early): Convert from a global struct to a
2505         subclass of gimple_opt_pass along with...
2506         (pass_data_sra_early): ...new pass_data instance and...
2507         (make_pass_sra_early): ...new function.
2508         (pass_sra): Convert from a global struct to a subclass of
2509         gimple_opt_pass along with...
2510         (pass_data_sra): ...new pass_data instance and...
2511         (make_pass_sra): ...new function.
2512         (pass_early_ipa_sra): Convert from a global struct to a subclass of
2513         gimple_opt_pass along with...
2514         (pass_data_early_ipa_sra): ...new pass_data instance and...
2515         (make_pass_early_ipa_sra): ...new function.
2516         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
2517         subclass of gimple_opt_pass along with...
2518         (pass_data_ccp): ...new pass_data instance and...
2519         (make_pass_ccp): ...new function.
2520         (pass_fold_builtins): Convert from a global struct to a subclass of
2521         gimple_opt_pass along with...
2522         (pass_data_fold_builtins): ...new pass_data instance and...
2523         (make_pass_fold_builtins): ...new function.
2524         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
2525         subclass of gimple_opt_pass along with...
2526         (pass_data_copy_prop): ...new pass_data instance and...
2527         (make_pass_copy_prop): ...new function.
2528         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
2529         global struct to a subclass of gimple_opt_pass along with...
2530         (pass_data_rename_ssa_copies): ...new pass_data instance and...
2531         (make_pass_rename_ssa_copies): ...new function.
2532         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
2533         subclass of gimple_opt_pass along with...
2534         (pass_data_dce): ...new pass_data instance and...
2535         (make_pass_dce): ...new function.
2536         (pass_dce_loop): Convert from a global struct to a subclass of
2537         gimple_opt_pass along with...
2538         (pass_data_dce_loop): ...new pass_data instance and...
2539         (make_pass_dce_loop): ...new function.
2540         (pass_cd_dce): Convert from a global struct to a subclass of
2541         gimple_opt_pass along with...
2542         (pass_data_cd_dce): ...new pass_data instance and...
2543         (make_pass_cd_dce): ...new function.
2544         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
2545         subclass of gimple_opt_pass along with...
2546         (pass_data_dominator): ...new pass_data instance and...
2547         (make_pass_dominator): ...new function.
2548         (pass_phi_only_cprop): Convert from a global struct to a subclass of
2549         gimple_opt_pass along with...
2550         (pass_data_phi_only_cprop): ...new pass_data instance and...
2551         (make_pass_phi_only_cprop): ...new function.
2552         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
2553         subclass of gimple_opt_pass along with...
2554         (pass_data_dse): ...new pass_data instance and...
2555         (make_pass_dse): ...new function.
2556         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
2557         a subclass of gimple_opt_pass along with...
2558         (pass_data_forwprop): ...new pass_data instance and...
2559         (make_pass_forwprop): ...new function.
2560         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
2561         struct to a subclass of gimple_opt_pass along with...
2562         (pass_data_tree_ifcombine): ...new pass_data instance and...
2563         (make_pass_tree_ifcombine): ...new function.
2564         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
2565         subclass of gimple_opt_pass along with...
2566         (pass_data_ch): ...new pass_data instance and...
2567         (make_pass_ch): ...new function.
2568         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
2569         subclass of gimple_opt_pass along with...
2570         (pass_data_tree_loop): ...new pass_data instance and...
2571         (make_pass_tree_loop): ...new function.
2572         (pass_tree_loop_init): Convert from a global struct to a subclass of
2573         gimple_opt_pass along with...
2574         (pass_data_tree_loop_init): ...new pass_data instance and...
2575         (make_pass_tree_loop_init): ...new function.
2576         (pass_lim): Convert from a global struct to a subclass of
2577         gimple_opt_pass along with...
2578         (pass_data_lim): ...new pass_data instance and...
2579         (make_pass_lim): ...new function.
2580         (pass_tree_unswitch): Convert from a global struct to a subclass of
2581         gimple_opt_pass along with...
2582         (pass_data_tree_unswitch): ...new pass_data instance and...
2583         (make_pass_tree_unswitch): ...new function.
2584         (pass_predcom): Convert from a global struct to a subclass of
2585         gimple_opt_pass along with...
2586         (pass_data_predcom): ...new pass_data instance and...
2587         (make_pass_predcom): ...new function.
2588         (pass_vectorize): Convert from a global struct to a subclass of
2589         gimple_opt_pass along with...
2590         (pass_data_vectorize): ...new pass_data instance and...
2591         (make_pass_vectorize): ...new function.
2592         (pass_graphite): Convert from a global struct to a subclass of
2593         gimple_opt_pass along with...
2594         (pass_data_graphite): ...new pass_data instance and...
2595         (make_pass_graphite): ...new function.
2596         (pass_graphite_transforms): Convert from a global struct to a subclass
2597         of gimple_opt_pass along with...
2598         (pass_data_graphite_transforms): ...new pass_data instance and...
2599         (make_pass_graphite_transforms): ...new function.
2600         (pass_check_data_deps): Convert from a global struct to a subclass of
2601         gimple_opt_pass along with...
2602         (pass_data_check_data_deps): ...new pass_data instance and...
2603         (make_pass_check_data_deps): ...new function.
2604         (pass_iv_canon): Convert from a global struct to a subclass of
2605         gimple_opt_pass along with...
2606         (pass_data_iv_canon): ...new pass_data instance and...
2607         (make_pass_iv_canon): ...new function.
2608         (pass_scev_cprop): Convert from a global struct to a subclass of
2609         gimple_opt_pass along with...
2610         (pass_data_scev_cprop): ...new pass_data instance and...
2611         (make_pass_scev_cprop): ...new function.
2612         (pass_record_bounds): Convert from a global struct to a subclass of
2613         gimple_opt_pass along with...
2614         (pass_data_record_bounds): ...new pass_data instance and...
2615         (make_pass_record_bounds): ...new function.
2616         (pass_complete_unroll): Convert from a global struct to a subclass of
2617         gimple_opt_pass along with...
2618         (pass_data_complete_unroll): ...new pass_data instance and...
2619         (make_pass_complete_unroll): ...new function.
2620         (pass_complete_unrolli): Convert from a global struct to a subclass of
2621         gimple_opt_pass along with...
2622         (pass_data_complete_unrolli): ...new pass_data instance and...
2623         (make_pass_complete_unrolli): ...new function.
2624         (pass_parallelize_loops): Convert from a global struct to a subclass
2625         of gimple_opt_pass along with...
2626         (pass_data_parallelize_loops): ...new pass_data instance and...
2627         (make_pass_parallelize_loops): ...new function.
2628         (pass_loop_prefetch): Convert from a global struct to a subclass of
2629         gimple_opt_pass along with...
2630         (pass_data_loop_prefetch): ...new pass_data instance and...
2631         (make_pass_loop_prefetch): ...new function.
2632         (pass_iv_optimize): Convert from a global struct to a subclass of
2633         gimple_opt_pass along with...
2634         (pass_data_iv_optimize): ...new pass_data instance and...
2635         (make_pass_iv_optimize): ...new function.
2636         (pass_tree_loop_done): Convert from a global struct to a subclass of
2637         gimple_opt_pass along with...
2638         (pass_data_tree_loop_done): ...new pass_data instance and...
2639         (make_pass_tree_loop_done): ...new function.
2640         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
2641         struct to a subclass of gimple_opt_pass along with...
2642         (pass_data_cse_reciprocals): ...new pass_data instance and...
2643         (make_pass_cse_reciprocals): ...new function.
2644         (pass_cse_sincos): Convert from a global struct to a subclass of
2645         gimple_opt_pass along with...
2646         (pass_data_cse_sincos): ...new pass_data instance and...
2647         (make_pass_cse_sincos): ...new function.
2648         (pass_optimize_bswap): Convert from a global struct to a subclass of
2649         gimple_opt_pass along with...
2650         (pass_data_optimize_bswap): ...new pass_data instance and...
2651         (make_pass_optimize_bswap): ...new function.
2652         (pass_optimize_widening_mul): Convert from a global struct to a
2653         subclass of gimple_opt_pass along with...
2654         (pass_data_optimize_widening_mul): ...new pass_data instance and...
2655         (make_pass_optimize_widening_mul): ...new function.
2656         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
2657         subclass of gimple_opt_pass along with...
2658         (pass_data_phiopt): ...new pass_data instance and...
2659         (make_pass_phiopt): ...new function.
2660         (pass_cselim): Convert from a global struct to a subclass of
2661         gimple_opt_pass along with...
2662         (pass_data_cselim): ...new pass_data instance and...
2663         (make_pass_cselim): ...new function.
2664         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
2665         subclass of gimple_opt_pass along with...
2666         (pass_data_phiprop): ...new pass_data instance and...
2667         (make_pass_phiprop): ...new function.
2668         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
2669         subclass of gimple_opt_pass along with...
2670         (pass_data_pre): ...new pass_data instance and...
2671         (make_pass_pre): ...new function.
2672         (pass_fre): Convert from a global struct to a subclass of
2673         gimple_opt_pass along with...
2674         (pass_data_fre): ...new pass_data instance and...
2675         (make_pass_fre): ...new function.
2676         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
2677         subclass of gimple_opt_pass along with...
2678         (pass_data_reassoc): ...new pass_data instance and...
2679         (make_pass_reassoc): ...new function.
2680         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
2681         subclass of gimple_opt_pass along with...
2682         (pass_data_sink_code): ...new pass_data instance and...
2683         (make_pass_sink_code): ...new function.
2684         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
2685         subclass of gimple_opt_pass along with...
2686         (pass_data_strlen): ...new pass_data instance and...
2687         (make_pass_strlen): ...new function.
2688         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
2689         struct to a subclass of gimple_opt_pass along with...
2690         (pass_data_build_alias): ...new pass_data instance and...
2691         (make_pass_build_alias): ...new function.
2692         (pass_build_ealias): Convert from a global struct to a subclass of
2693         gimple_opt_pass along with...
2694         (pass_data_build_ealias): ...new pass_data instance and...
2695         (make_pass_build_ealias): ...new function.
2696         (pass_ipa_pta): Convert from a global struct to a subclass of
2697         simple_ipa_opt_pass along with...
2698         (pass_data_ipa_pta): ...new pass_data instance and...
2699         (make_pass_ipa_pta): ...new function.
2700         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
2701         subclass of gimple_opt_pass along with...
2702         (pass_data_uncprop): ...new pass_data instance and...
2703         (make_pass_uncprop): ...new function.
2704         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
2705         global struct to a subclass of gimple_opt_pass along with...
2706         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
2707         (make_pass_late_warn_uninitialized): ...new function.
2708         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
2709         to a subclass of gimple_opt_pass along with...
2710         (pass_data_init_datastructures): ...new pass_data instance and...
2711         (make_pass_init_datastructures): ...new function.
2712         (pass_early_warn_uninitialized): Convert from a global struct to a
2713         subclass of gimple_opt_pass along with...
2714         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
2715         (make_pass_early_warn_uninitialized): ...new function.
2716         (pass_update_address_taken): Convert from a global struct to a
2717         subclass of gimple_opt_pass along with...
2718         (pass_data_update_address_taken): ...new pass_data instance and...
2719         (make_pass_update_address_taken): ...new function.
2720         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
2721         struct to a subclass of gimple_opt_pass along with...
2722         (pass_data_release_ssa_names): ...new pass_data instance and...
2723         (make_pass_release_ssa_names): ...new function.
2724         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
2725         subclass of gimple_opt_pass along with...
2726         (pass_data_stdarg): ...new pass_data instance and...
2727         (make_pass_stdarg): ...new function.
2728         * tree-switch-conversion.c (pass_convert_switch): Convert from a
2729         global struct to a subclass of gimple_opt_pass along with...
2730         (pass_data_convert_switch): ...new pass_data instance and...
2731         (make_pass_convert_switch): ...new function.
2732         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
2733         to a subclass of gimple_opt_pass along with...
2734         (pass_data_tail_recursion): ...new pass_data instance and...
2735         (make_pass_tail_recursion): ...new function.
2736         (pass_tail_calls): Convert from a global struct to a subclass of
2737         gimple_opt_pass along with...
2738         (pass_data_tail_calls): ...new pass_data instance and...
2739         (make_pass_tail_calls): ...new function.
2740         * tree-vect-generic.c (pass_lower_vector): Convert from a global
2741         struct to a subclass of gimple_opt_pass along with...
2742         (pass_data_lower_vector): ...new pass_data instance and...
2743         (make_pass_lower_vector): ...new function.
2744         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
2745         gimple_opt_pass along with...
2746         (pass_data_lower_vector_ssa): ...new pass_data instance and...
2747         (make_pass_lower_vector_ssa): ...new function.
2748         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
2749         to a subclass of gimple_opt_pass along with...
2750         (pass_data_slp_vectorize): ...new pass_data instance and...
2751         (make_pass_slp_vectorize): ...new function.
2752         (pass_ipa_increase_alignment): Convert from a global struct to a
2753         subclass of simple_ipa_opt_pass along with...
2754         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
2755         (make_pass_ipa_increase_alignment): ...new function.
2756         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
2757         gimple_opt_pass along with...
2758         (pass_data_vrp): ...new pass_data instance and...
2759         (make_pass_vrp): ...new function.
2760         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
2761         subclass of simple_ipa_opt_pass along with...
2762         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
2763         (make_pass_ipa_free_lang_data): ...new function.
2764         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
2765         gimple_opt_pass along with...
2766         (pass_data_tsan): ...new pass_data instance and...
2767         (make_pass_tsan): ...new function.
2768         (pass_tsan_O0): Convert from a global struct to a subclass of
2769         gimple_opt_pass along with...
2770         (pass_data_tsan_O0): ...new pass_data instance and...
2771         (make_pass_tsan_O0): ...new function.
2772         * var-tracking.c (pass_variable_tracking): Convert from a global
2773         struct to a subclass of rtl_opt_pass along with...
2774         (pass_data_variable_tracking): ...new pass_data instance and...
2775         (make_pass_variable_tracking): ...new function.
2776         * web.c (pass_web): Convert from a global struct to a subclass of
2777         rtl_opt_pass along with...
2778         (pass_data_web): ...new pass_data instance and...
2779         (make_pass_web): ...new function.
2780         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
2781         declaration with that of...
2782         (make_pass_mode_switch_use): ...new function.
2783         (pass_resolve_sw_modes): Replace declaration with that of...
2784         (make_pass_resolve_sw_modes): ...new function.
2785         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
2786         from a global struct to a subclass of rtl_opt_pass along with...
2787         (pass_data_mode_switch_use): ...new pass_data instance and...
2788         (make_pass_mode_switch_use): ...new function.
2789         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
2790         from a global struct to a subclass of rtl_opt_pass along with...
2791         (pass_data_resolve_sw_modes): ...new pass_data instance and...
2792         (make_pass_resolve_sw_modes): ...new function.
2793         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
2794         struct to a subclass of rtl_opt_pass along with...
2795         (pass_data_insert_vzeroupper): ...new pass_data instance and...
2796         (make_pass_insert_vzeroupper): ...new function.
2797         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
2798         global struct to a subclass of rtl_opt_pass along with...
2799         (pass_data_work_around_errata): ...new pass_data instance and...
2800         (make_pass_work_around_errata): ...new function.
2801         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
2802         struct to a subclass of rtl_opt_pass along with...
2803         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
2804         (make_pass_mips_machine_reorg2): ...new function.
2806 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2808         * passes.c (pass_manager::operator new): New.
2810 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
2812         Handwritten part of conversion of passes to C++ classes.
2814         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
2815         (toplev.o): Add dep on PASS_MANAGER_H.
2816         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
2817         of early local pases to reflect this moving from a global to a
2818         member of gcc::pass_manager.
2819         (cgraph_add_new_function): Likewise.
2820         * lto-cgraph.c (lto_output_node): Update for conversion of
2821         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
2822         * passes.c (opt_pass::clone): New.
2823         (opt_pass::gate): New.
2824         (opt_pass::execute): New.
2825         (opt_pass::opt_pass): New.
2826         (pass_manager::execute_early_local_passes): New.
2827         (pass_manager::execute_pass_mode_switching): new.
2828         (finish_optimization_passes): Convert to...
2829         (pass_manager::finish_optimization_passes): ...this.
2830         (finish_optimization_passes): Update for conversion of passes to
2831         C++ classes.
2832         (register_dump_files_1): Use has_gate since we cannot portably
2833         check a vtable entry against NULL.
2834         (dump_one_pass): Likewise.
2835         (ipa_write_summaries_2): Likewise.
2836         (ipa_write_optimization_summaries_1): Likewise.
2837         (ipa_read_summaries_1): Likewise.
2838         (ipa_read_optimization_summaries_1): Likewise.
2839         (execute_ipa_stmt_fixups): Likewise.
2840         (pass_manager::pass_manager): Rewrite pass-creation, invoking
2841         pass-creation functions rather than wiring up globals, and
2842         storing the results in fields of pass_manager generated using
2843         pass-instances.def.
2844         (pass_manager::dump_profile_report): Update for conversion of
2845         passes to C++ classes.
2846         (pass_manager::execute_ipa_summary_passes): Likewise.
2847         (execute_one_ipa_transform_pass): Likewise.
2848         (execute_one_pass): Use has_gate and has_execute since we cannot
2849         portably check a vtable entry against NULL.
2850         * pass_manager.h (pass_manager::finish_optimization_passes): New.
2851         (pass_manager): Use pass-instances.def to add fields for the
2852         various pass instances.
2853         * toplev.c (finalize): Update for move of
2854         finish_optimization_passes to a method of gcc::pass_manager.
2855         * toplev.h (finish_optimization_passes): Move to method of class
2856         pass_manager.
2857         * tree-pass.h (struct pass_data): New.
2858         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
2859         (opt_pass::gate): Convert to virtual function.
2860         (opt_pass::~opt_pass): New.
2861         (opt_pass::clone): New.
2862         (opt_pass::execute): Convert to virtual function.
2863         (opt_pass::opt_pass): New.
2864         (opt_pass::ctxt_): new.
2865         (gimple_opt_pass): Convert to subclass of opt_pass.
2866         (gimple_opt_pass::gimple_opt_pass): New.
2867         (rtl_opt_pass): Convert to subclass of opt_pass.
2868         (rtl_opt_pass::rtl_opt_pass): New.
2869         (ipa_opt_pass_d): Convert to subclass of opt_pass.
2870         (ipa_opt_pass_d::ipa_opt_pass_d): New.
2871         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
2872         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
2873         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
2874         invocation of pass_mode_switching to reflect this moving from a
2875         global to a member of gcc::pass_manager.
2876         (ix86_option_override): Rework how pass_insert_vzeroupper is
2877         added to the pass_manager to reflect autogenerated changes.
2878         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
2880 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
2882         PR rtl-optimization/57708
2883         * recog.c (peep2_find_free_register): Validate all regs in a
2884         multi-reg mode.
2886 2013-08-05  Jan Hubicka  <jh@suse.cz>
2888         PR lto/57602
2889         * cgraph.c (verify_cgraph_node): Accept local flags from other
2890         partitions.
2891         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
2892         (function_and_variable_visibility): Likewise.
2893         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
2895 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2897         * graph.c (init_graph_slim_pretty_print): Remove.
2898         (print_graph_cfg): Do not call it.  Use local pretty printer.
2899         (start_graph_dump): Likewise.
2901 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2903         * gimple-pretty-print.c (buffer): Remove.
2904         (initialized): Likewise.
2905         (maybe_init_pretty_print): Likewise.
2906         (print_gimple_stmt): Do not call it.  Use non-static local
2907         pretty_printer variable.
2908         (print_gimple_expr): Likewise.
2909         (print_gimple_seq): Likewise.
2910         (gimple_dump_bb): Likewise.
2912 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2914         * asan.c (asan_pp): Remove.
2915         (asan_pp_initialized): Likewise.
2916         (asan_pp_initialize): Likewise.
2917         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
2918         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
2919         (asan_add_global): Likewise.
2921 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2923         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
2924         * diagnostic.h (diagnostic_flush_buffer): Adjust.
2925         * pretty-print.c (pp_formatted_text_data): Likewise.
2926         (pp_indent): Rename from pp_base_indent.
2927         (pp_format): Rename from pp_base_format.
2928         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
2929         (pp_format_verbatim): Rename from pp_base_format_verbatim.
2930         (pp_flush): Rename from pp_base_flush.
2931         (pp_set_line_maximum_length): Rename from
2932         pp_base_set_line_maximum_length.
2933         (pp_clear_output_area): Rename from pp_base_clear_output_area.
2934         (pp_set_prefix): Rename from pp_base_set_prefix.
2935         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
2936         (pp_emit_prefix): Rename from pp_base_emit_prefix.
2937         (pp_append_text): Rename from pp_base_append_text.
2938         (pp_formatted_text): Rename from pp_base_formatted_text.
2939         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
2940         (pp_remaining_character_count_for_line): Rename from
2941         pp_base_remaining_character_count_for_line.
2942         (pp_newline): Rename from pp_base_newline.
2943         (pp_character): Rename from pp_base_character.
2944         (pp_string): Rename from pp_base_string.
2945         (pp_maybe_space): Rename from pp_base_maybe_space.
2946         * asan.c (asan_pp_string): Adjust.
2947         (asan_emit_stack_protection): Likewise.
2948         (asan_add_global): Likewise.
2949         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
2950         * tree-mudflap.c (mf_varname_tree): Likewise.
2951         * tree-pretty-print.c (pp_tree_identifier): Rename from
2952         pp_base_tree_identifier.
2953         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
2954         Declare as function.
2956 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
2958         * pretty-print.h (pp_bar_bar): New.
2959         (pp_ampersand_ampersand): Likewise.
2960         (pp_less_equal): Likewise.
2961         (pp_greater_equal): Likewise.
2962         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
2963         printer functions instead of pp_string or operators and punctuators.
2964         (dump_gimple_call): Likewise.
2965         (dump_gimple_omp_for): Likewise.
2966         (dump_gimple_transaction): Likewise.
2967         (dump_gimple_phi): Likewise.
2968         (pp_gimple_stmt_1): Likewise.
2969         * sched-vis.c (print_insn): Likewise.
2970         * tree-mudflap.c (mf_varname_tree): Likewise.
2971         * tree-pretty-print.c (dump_block_node): Likewise.
2972         (dump_generic_node): Likewise.
2974 2013-08-02  Jan Hubicka  <jh@suse.cz>
2976         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
2977         boundaries.
2978         * lto-streamer-out.c (tree_is_indexable): Results decls and
2979         parm decls are not indexable.
2980         (DFS_write_tree_body): Do not follow args and results.
2981         (hash_tree): Likewise.
2982         (output_functions): Rearrange so struct function is needed
2983         only when real body is output; be able to also ouptut abstract
2984         functions; output DECL_ARGUMENTS and DECL_RESULT.
2985         (lto_output): When not in WPA, ale store abstract functions.
2986         (write_symbol): Do not care about RESULT_DECL.
2987         (output_symbol_p): Handle correctly sbtract decls.
2988         * lto-streamer-in.c (input_function): Rearrange so struct
2989         function can be NULL at entry; allow streaming of
2990         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2991         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2992         sanity check during LTO.
2993         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2994         RESULT_DECl and DECL_ARGUMENTS.
2995         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2996         Likewise.
2998 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3000         * pretty-print.h (pp_underscore): New.
3001         (pp_comma): Tidy.
3002         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
3003         printer functions instead of pp_character.
3004         (dump_binary_rhs): Likewise.
3005         (dump_ternary_rhs): Likewise.
3006         (dump_gimple_call_args): Likewise.
3007         (pp_points_to_solution): Likewise.
3008         (dump_gimple_call): Likewise.
3009         (dump_gimple_switch): Likewise.
3010         (dump_gimple_cond): Likewise.
3011         (dump_gimple_bind): Likewise.
3012         (dump_gimple_try): Likewise.
3013         (dump_gimple_omp_for): Likewise.
3014         (dump_gimple_omp_continue): Likewise.
3015         (dump_gimple_omp_single): Likewise.
3016         (dump_gimple_omp_sections): Likewise.
3017         (dump_gimple_omp_block): Likewise.
3018         (dump_gimple_omp_critical): Likewise.
3019         (dump_gimple_transaction): Likewise.
3020         (dump_gimple_asm): Likewise.
3021         (dump_gimple_phi): Likewise.
3022         (dump_gimple_omp_parallel): Likewise.
3023         (dump_gimple_omp_task): Likewise.
3024         (dump_gimple_omp_atomic_load): Likewise.
3025         (dump_gimple_omp_atomic_store): Likewise.
3026         (dump_gimple_mem_ops): Likewise.
3027         (pp_gimple_stmt_1): Likewise.
3028         (pp_cfg_jump): Likewise.
3029         (dump_implicit_edges): Likewise.
3030         (gimple_dump_bb_for_graph): Likewise.
3031         * graph.c (draw_cfg_node): Likewise.
3032         * langhooks.c (lhd_print_error_function): Likewise.
3033         * sched-vis.c (print_exp): Likewise.
3034         (print_value): Likewise.
3035         (print_pattern): Likewise.
3036         (print_insn): Likewise.
3037         (rtl_dump_bb_for_graph): Likewise.
3038         * tree-pretty-print.c (dump_function_declaration): Likewise.
3039         (dump_array_domain): Likewise.
3040         (dump_omp_clause): Likewise.
3041         (dump_location): Likewise.
3042         (dump_generic_node): Likewise.
3043         (print_struct_decl): Likewise.
3044         * diagnostic.c (diagnostic_show_locus): Use pp_space.
3046 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3048         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
3049         candidate table when replacing a candidate statement.
3050         (replace_rhs_if_not_dup): Likewise.
3051         (replace_one_candidate): Likewise.
3053 2013-08-02  Jan Hubicka  <jh@suse.cz>
3054             Martin Liska  <marxin.liska@gmail.com>
3056         * cgraphunit.c (add_new_function): Fix logic when adding from
3057         late IPA pass.
3058         (assemble_thunk): Rename to ...
3059         (expand_thunk); .. this one; export; get it working with
3060         general functions; make produced gimple valid.
3061         * cgraph.h (expand_thunk): Declare.
3063 2013-08-02  Jan Hubicka  <jh@suse.cz>
3065         * ipa-cp.c (gather_context_independent_values): Use
3066         ipa_get_param_move_cost.
3067         (get_replacement_map): Remove PARAM; move parameter folding
3068         into tree-inline.c
3069         (create_specialized_node): Update.
3070         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3071         assert that we have gimple body; update move_cost.
3072         (count_formal_params): Assert that we have gimple body.
3073         (ipa_dump_param): New function.
3074         (ipa_alloc_node_params): Break out from ...
3075         (ipa_initialize_node_params): ... here.
3076         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3077         (ipa_write_node_info): Stream move costs.
3078         (ipa_read_node_info): Read move costs.
3079         (ipa_update_after_lto_read): Do not recompute node params.
3080         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3081         (ipa_get_param): Check we are not in WPA.
3082         (ipa_get_param_move_cost): New.
3083         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3084         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3085         parm numbers to be present.
3087 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3089         PR rtl-optimization/58048
3090         * lra-constraints.c (process_alt_operands): Don't check asm
3091         operand on register.
3093 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
3095         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3096         the implied StoreLoad barrier for atomic operations if before.
3098 2013-08-02  Jan Hubicka  <jh@suse.cz>
3099             Martin Liska  <marxin.liska@gmail.com>
3101         * cgraph.c (cgraph_function_body_availability): Do not check
3102         cgraph flags.
3103         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3104         symtab_node_availability): Declare.
3105         * ipa.c (can_replace_by_local_alias): New.
3106         (function_and_variable_visibility): Use it.
3107         * symtab.c (symtab_for_node_and_aliases,
3108         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3110 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3112         PR rtl-optimization/57963
3113         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3114         (lra_constraints): Use them.
3116 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
3118         * config/arm/types.md (define_attr "type"): Add "load_acq"
3119         and "store_rel".
3120         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3121         changes.
3122         (cortex_a53_store1): Likewise.
3124 2013-08-01  Jan Hubicka  <jh@suse.cz>
3126         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3127         partitions are not needed.
3129 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
3131         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3132         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3133         MAYBE_NON_Q_CLASS_P where appropriate.
3135 2013-08-01  Jan Hubicka  <jh@suse.cz>
3137         * cgraph.h (release_function_body): Declare.
3138         * tree.c (free_lang_data_in_decl): Free, parameters and return values
3139         of unused delcarations.
3141 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3143         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3144         RTL form when subtracting a constant.
3146 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3148         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3149         Generate canonical plus rtx with negated immediate instead of minus
3150         where appropriate.
3151         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3153 2013-08-01  Jan Hubicka  <jh@suse.cz>
3155         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3156         (cgraph_release_function_body): Likewise.
3157         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3158         * cgraph.h (cgrpah_node): Rename abstract_and_needed
3159         to used_as_abstract_origin.
3160         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3161         symbols used as abstract origins.
3162         * cgraphunit.c (analyze_functions): Update.
3163         * ipa.c (symtab_remove_unreachable_nodes): Recompute
3164         used_as_abstract_origin.
3165         * tree-inline.c (tree_function_versioning): Update
3166         used_as_abstract_origin; be ready for DECL_RESULT and
3167         DECL_ARGUMENTS to be NULL.
3169         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3170         for abstract functions.
3171         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3172         real symbols.
3174 2013-08-01  Jan Hubicka  <jh@suse.cz>
3176         * profile.c (compute_value_histograms): Fix thinko.
3178 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3180         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
3181         aarch-common-protos.h to extra_headers.
3182         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3183         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3184         * config/aarch64/t-aarch64 (aarch-common.o): Define.
3186 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3188         * config/aarch64/aarch64.md (define_attr "type"): Delete.
3189         Include "../arm/types.md".  Define "type" attribute for all patterns.
3190         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3191         attribute changes.
3193 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
3195         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3196         to support power8 load fusion.
3197         (fusion_gpr_mem_load): Likewise.
3199         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3201         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3202         declarations for power8 load fusion.
3203         (emit_fusion_gpr_load): Likewise.
3205         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3206         tuning for power8, turn on fusion mode by default.  Turn on sign
3207         extending fusion mode if normal fusion mode is on, and we are at
3208         -O2 or -O3.
3209         (fusion_gpr_load_p): New function, return true if we can fuse an
3210         addis instruction with a dependent load to a GPR.
3211         (emit_fusion_gpr_load): Emit the instructions for power8 load
3212         fusion to GPRs.
3214         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3215         (VSX load fusion peepholes): New peepholes to fuse together an
3216         addi instruction with a VSX load instruction.
3218         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3219         peepholes to fuse an addis instruction with a load to a GPR base
3220         register.  If we are supporting sign extending fusions, convert
3221         sign extending loads to zero extending loads and add an explicit
3222         sign extension.
3224 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3226         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
3227         aarch-common-protos.h to extra_headers.
3228         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3229         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3230         (arm_early_store_addr_dep): Likewise.
3231         (arm_no_early_alu_shift_dep): Likewise.
3232         (arm_no_early_alu_shift_value_dep): Likewise.
3233         (arm_no_early_mul_dep): Likewise.
3234         (arm_no_early_store_addr_dep): Likewise.
3235         (arm_mac_accumulator_is_mul_result): Likewise.
3236         (arm_mac_accumulator_is_result): Likewise.
3237         * config/arm/aarch-common.c: ... here.  New file.
3238         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3239         here to ...
3240         (arm_early_store_addr_dep): Likewise.
3241         (arm_no_early_alu_shift_dep): Likewise.
3242         (arm_no_early_alu_shift_value_dep): Likewise.
3243         (arm_no_early_mul_dep): Likewise.
3244         (arm_no_early_store_addr_dep): Likewise.
3245         (arm_mac_accumulator_is_mul_result): Likewise.
3246         (arm_mac_accumulator_is_result): Likewise.
3247         * config/arm/aarch-common-protos.h: ... here.  New file.
3248         * config/arm/t-arm (aarch-common.o): Define.
3250 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3252         * config/arm/arm.md: Include new file "types.md".
3253         (define_attr "type"): Move from here to ...
3254         (define_attr "mul32"): Likewise.
3255         (define_attr "mul64"): Likewise.
3256         * config/arm/types.md: ... here.  New file.
3258 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
3260         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
3261         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
3263 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3265         * gen-pass-instances.awk: Fix offset of substr().
3267 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
3269         * Makefile.in (pass-instances.def): New.
3270         (passes.o): Replace dependency on passes.def with one on
3271         pass-instances.def
3273         * gen-pass-instances.awk: New.
3275         * passes.c (pass_manager::pass_manager): Use pass-instances.def
3276         rather than passes.def, updating local definition of NEXT_PASS
3277         macro to add an extra NUM parameter (currently unused).
3279 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
3281         * Makefile.in (PASS_MANAGER_H): New.
3282         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
3283         (passes.o): Likewise.
3284         (statistics.o): Likewise.
3285         (cgraphunit.o): Likewise.
3286         (context.o): Depend on PASS_MANAGER_H.
3288         * pass_manager.h: New.
3290         * cgraphunit.c (cgraph_add_new_function): Update for moves
3291         of globals to fields of pass_manager.
3292         (analyze_function): Likewise.
3293         (expand_function): Likewise.
3294         (ipa_passes): Likewise.
3295         (compile): Likewise.
3297         * context.c (context::context): New.
3298         * context.h  (context::context): New.
3299         (context::get_passes): New.
3300         (context::passes_): New.
3302         * lto-cgraph.c (input_node): Update for moves of globals to
3303         fields of pass_manager.
3305         * passes.c (all_passes): Remove, in favor of a field of the
3306         same name within the new class pass_manager.
3307         (all_small_ipa_passes): Likewise.
3308         (all_lowering_passes): Likewise.
3309         (all_regular_ipa_passes): Likewise.
3310         (all_late_ipa_passes): Likewise.
3311         (all_lto_gen_passes): Likewise.
3312         (passes_by_id): Likewise.
3313         (passes_by_id_size): Likewise.
3314         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3315         the new class pass_manager.
3316         (set_pass_for_id): Convert to...
3317         (pass_manager::set_pass_for_id): ...method.
3318         (get_pass_for_id): Convert to...
3319         (pass_manager::get_pass_for_id): ...method.
3320         (register_one_dump_file): Move body of implementation into...
3321         (pass_manager::register_one_dump_file): ...here.
3322         (register_dump_files_1): Convert to...
3323         (pass_manager::register_dump_files_1): ...method.
3324         (register_dump_files): Convert to...
3325         (pass_manager::register_dump_files): ...method.
3326         (create_pass_tab): Update for moves of globals to fields of
3327         pass_manager.
3328         (dump_passes): Move body of implementation into...
3329         (pass_manager::dump_passes): ...here.
3330         (register_pass): Move body of implementation into...
3331         (pass_manager::register_pass): ...here.
3332         (init_optimization_passes): Convert into...
3333         (pass_manager::pass_manager): ...constructor for new
3334         pass_manager class, and initialize the pass_lists array.
3335         (check_profile_consistency): Update for moves of globals to
3336         fields of pass_manager.
3337         (dump_profile_report): Move body of implementation into...
3338         (pass_manager::dump_profile_report): ...here.
3339         (ipa_write_summaries_1): Update for moves of pass lists from
3340         being globals to fields of pass_manager.
3341         (ipa_write_optimization_summaries): Likewise.
3342         (ipa_read_summaries):  Likewise.
3343         (ipa_read_optimization_summaries): Likewise.
3344         (execute_all_ipa_stmt_fixups): Likewise.
3346         * statistics.c (statistics_fini): Update for moves of globals to
3347         fields of pass_manager.
3349         * toplev.c (general_init): Replace call to
3350         init_optimization_passes with construction of the pass_manager
3351         instance.
3353         * tree-pass.h (all_passes): Remove, in favor of a field of the
3354         same name within the new class pass_manager.
3355         (all_small_ipa_passes): Likewise.
3356         (all_lowering_passes): Likewise.
3357         (all_regular_ipa_passes): Likewise.
3358         (all_lto_gen_passes): Likewise.
3359         (all_late_ipa_passes): Likewise.
3360         (passes_by_id): Likewise.
3361         (passes_by_id_size): Likewise.
3362         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3363         the new class pass_manager.
3364         (get_pass_for_id): Remove.
3366 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
3368         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
3369         configs.
3371 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
3373         * arm.md (mulhi3): New expand pattern.
3375 2013-07-30  Jan Hubicka  <jh@suse.cz>
3376             Martin Liska  <marxin.liska@gmail.com>
3378         * profile.c (compute_value_histograms): Do not ICE when
3379         there is mismatch only on some counters.
3381 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3383         PR rtl-optimization/57637
3384         * function.c (move_insn_for_shrink_wrap): Also check the
3385         GEN set of the LIVE problem for the liveness analysis
3386         if it exists, otherwise give up.
3388 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3390         PR tree-optimization/57993
3391         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
3392         replaced statement in the candidate table.
3393         (phi_add_costs): Return infinite cost when the hidden basis does
3394         not dominate all phis on which the candidate is dependent.
3395         (replace_one_candidate): Record replaced statement in the
3396         candidate table.
3398 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
3400         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
3401         (ashlv2si3): New expander.
3402         (*ashlv2si3_i): New define_insn_and_split.
3403         * predicates.md (float_operation): Allow patterns with three
3404         basic sub-patterns.
3406         PR rtl-optimization/58021
3407         * mode-switching.c (create_pre_exit): Always split off preceding
3408         insns if we are not at the basic block head.
3410 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
3412         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
3413         (UCLIBC_DYNAMIC_LINKER): New macro.
3414         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
3415         `-mnan=2008'.
3416         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
3417         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
3418         `-mnan=2008'.
3419         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
3420         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
3421         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
3422         for SF and DF modes.  Use ieee_quad_format for TF mode.
3423         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
3424         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
3425         (mips_option_override): Handle `-mnan=legacy'.
3426         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
3427         `-mabs=2008' and `-mnan=2008'.
3428         (OPTION_DEFAULT_SPECS): Add "nan" default.
3429         (ASM_SPEC): Handle `-mnan='.
3430         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
3431         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
3432         comment accordingly.
3433         (neg<mode>2): Likewise.
3434         * config/mips/mips.opt (mabs, mnan): New options.
3435         * doc/install.texi (Configuration): Document `--with-nan=' option.
3436         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
3437         `-mnan=' options.
3438         (MIPS Options): Document them.
3439         * config.gcc <mips*-*-*>: Handle `--with-nan='.
3440         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
3441         * configure: Regenerate.
3442         * config.in: Regenerate.
3444 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
3446         * config/i386/i386.md (float post-reload splitters): Do not check
3447         for subregs of SSE registers.
3449 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
3450             H.J. Lu  <hongjiu.lu@intel.com>
3452         PR target/57954
3453         PR target/57988
3454         * config/i386/i386.md (post-reload splitter
3455         to avoid partial SSE reg dependency stalls): New pattern.
3457 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
3459         * config/s390/s390.md ("movcc"): Swap load and store instructions.
3461 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
3463         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
3464         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
3466 2013-07-26  Cary Coutant  <ccoutant@google.com>
3468         * dwarf2out.c (die_checksum_ordered): Don't include template
3469         instantiations in signature.
3470         (is_template_parameter): New function.
3471         (is_template_instantiation): New function.
3472         (generate_skeleton_bottom_up): Don't include template instantiations
3473         in type unit DIE.
3474         (generate_skeleton): Likewise.
3475         (break_out_comdat_types): Move recursive call to break out nested
3476         types earlier.
3477         (prune_unused_types_mark_generic_parms_dies): Call
3478         is_template_parameter.
3480 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
3482         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
3483         uses vector registers.
3484         * config/aarch64/iterators.md: Add attributes rtn and vas.
3486 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3487             Richard Earnshaw  <richard.earnshaw@arm.com>
3489         * combine.c (simplify_comparison): Re-canonicalize operands
3490         where appropriate.
3491         * config/arm/arm.md (movcond_addsi): New splitter.
3493 2013-07-25  Sterling Augustine  <saugustine@google.com>
3495         * dwarf2out.c (size_of_pubnames): Move code to...
3496         (include_pubname_in_output): ...here.  New.
3497         (want_pubnames): Rearrange.
3498         (output_pubnames): Call include_pubname_in_output.  Move assertion.
3500 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
3502         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
3504 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
3506         PR target/38836
3507         * doc/extend.texi: Remove obsolete builtins. Fix
3508         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
3510 2013-07-25  Jan Hubicka  <jh@suse.cz>
3512         * cgraph.c (release_function_body): Break out from ...
3513         (cgraph_release_function_body): ... this one; also release DECL_RESULT
3514         and DECL_ARGUMENTS.
3515         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
3516         old_tree in the map.
3517         (create_specialized_node): Update.
3518         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
3519         into index.
3520         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
3521         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
3522         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
3523         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
3524         DECL_RESULT.
3526 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3528         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
3529         addsi3_carryin_alt2_<optab>): Correct output template.
3531 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3533         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
3534         Adjust for arm_restrict_it.
3535         Remove trailing whitespace.
3537 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
3539         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
3540         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
3542         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
3544 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
3546         PR rtl-optimization/57960
3547         * lra-constraints.c (process_alt_operands): Use the right mode
3548         when checking strict_low.
3550 2013-07-25  Jan Hubicka  <jh@suse.cz>
3552         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
3553         * cgraph.c (cgraph_remove_node): Do not release function body
3554         when in cgraph streaming.
3555         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
3556         in other partitions are not considered reachable; fix handling of
3557         clones.
3559 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3561         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
3563 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3565         PR target/19599
3566         PR target/57731
3567         PR target/57837
3568         * config/arm/arm.md ("*sibcall_insn): Replace use of
3569         Ss with US. Adjust output for v5 and v4t.
3570         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
3572         * config/arm/constraints.md ("Ss"): Rename to US.
3574 2013-07-25  Terry Guo  <terry.guo@arm.com>
3576         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
3577         shift_add/shift_sub0/shift_sub1 RTXs.
3579 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
3580             Anton Blanchard <anton@au1.ibm.com>
3582         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
3583         (altivec_vpks<VI_char>ss): Likewise.
3584         (altivec_vpks<VI_char>us): Likewise.
3585         (altivec_vpku<VI_char>us): Likewise.
3586         (altivec_vpku<VI_char>um): Likewise.
3588 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
3590         Introduce context class.
3592         * Makefile.in (CONTEXT_H): New.
3593         (OBJS): Add context.o.
3594         (toplev.o): Add CONTEXT_H to dependencies.
3595         (context.o): New.
3597         * toplev.c (general_init): Create the singleton gcc::context instance.
3599         * context.c: New.
3601         * context.h: New.
3603 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
3605         PR rtl-optimization/57968
3606         * mode-switching.c (create_pre_exit): Allow instructions that
3607         don't set a return register to need a non-exit mode.
3609 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3610             Anton Blanchard <anton@au1.ibm.com>
3612         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
3613         operands to vperm for little endian.
3614         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
3615         of lvsl to create the control mask for a vperm for little endian.
3617 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3618             Anton Blanchard <anton@au1.ibm.com>
3620         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
3621         two operands for little-endian.
3623 2013-07-23  Steve Ellcey  <sellcey@mips.com>
3625         * config/mips/mips.c (mips_case_values_threshold): New.
3626         (TARGET_CASE_VALUES_THRESHOLD): Define.
3628 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3629             Anton Blanchard <anton@au1.ibm.com>
3631         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
3632         selection of field for vector splat in little endian mode.
3634 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
3636         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
3637         expanders to rs6000.md.
3638         (ior<mode>3): Likewise.
3639         (and<mode>3): Likewise.
3640         (one_cmpl<mode>2): Likewise.
3641         (nor<mode>3): Likewise.
3642         (andc<mode>3): Likewise.
3643         (eqv<mode>3): Likewise.
3644         (nand<mode>3): Likewise.
3645         (orc<mode>3): Likewise.
3647         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
3648         declaration.
3650         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
3651         to split multi-word logical operations.
3652         (rs6000_split_logical_di): Likewise.
3653         (rs6000_split_logical): Likewise.
3655         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
3656         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
3657         and allow TImode operations in 32-bit.
3658         (vsx_and<mode>3_64bit): Likewise.
3659         (vsx_ior<mode>3_32bit): Likewise.
3660         (vsx_ior<mode>3_64bit): Likewise.
3661         (vsx_xor<mode>3_32bit): Likewise.
3662         (vsx_xor<mode>3_64bit): Likewise.
3663         (vsx_one_cmpl<mode>2_32bit): Likewise.
3664         (vsx_one_cmpl<mode>2_64bit): Likewise.
3665         (vsx_nor<mode>3_32bit): Likewise.
3666         (vsx_nor<mode>3_64bit): Likewise.
3667         (vsx_andc<mode>3_32bit): Likewise.
3668         (vsx_andc<mode>3_64bit): Likewise.
3669         (vsx_eqv<mode>3_32bit): Likewise.
3670         (vsx_eqv<mode>3_64bit): Likewise.
3671         (vsx_nand<mode>3_32bit): Likewise.
3672         (vsx_nand<mode>3_64bit): Likewise.
3673         (vsx_orc<mode>3_32bit): Likewise.
3674         (vsx_orc<mode>3_64bit): Likewise.
3676         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
3677         logical types in GPRs.
3679         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
3680         logical insns to rs6000.md, and allow TImode operations in
3681         32-bit.
3682         (altivec_ior<mode>3): Likewise.
3683         (altivec_xor<mode>3): Likewise.
3684         (altivec_one_cmpl<mode>2): Likewise.
3685         (altivec_nor<mode>3): Likewise.
3686         (altivec_andc<mode>3): Likewise.
3688         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
3689         attributes for moving the 128-bit logical operations into
3690         rs6000.md.
3691         (BOOL_REGS_OUTPUT): Likewise.
3692         (BOOL_REGS_OP1): Likewise.
3693         (BOOL_REGS_OP2): Likewise.
3694         (BOOL_REGS_UNARY): Likewise.
3695         (BOOL_REGS_AND_CR0): Likewise.
3696         (one_cmpl<mode>2): Add support for DI logical operations on
3697         32-bit, splitting the operations to 32-bit.
3698         (anddi3): Likewise.
3699         (iordi3): Likewise.
3700         (xordi3): Likewise.
3701         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
3702         changes to combine the 32/64-bit code, allow logical operations on
3703         TI mode in 32-bit, and to use similar match_operator patterns like
3704         scalar mode uses.  Combine the Altivec and VSX code for logical
3705         operations, and move it here.
3706         (ior<mode>3, 128-bit types): Likewise.
3707         (xor<mode>3, 128-bit types): Likewise.
3708         (one_cmpl<mode>3, 128-bit types): Likewise.
3709         (nor<mode>3, 128-bit types): Likewise.
3710         (andc<mode>3, 128-bit types): Likewise.
3711         (eqv<mode>3, 128-bit types): Likewise.
3712         (nand<mode>3, 128-bit types): Likewise.
3713         (orc<mode>3, 128-bit types): Likewise.
3714         (and<mode>3_internal): Likewise.
3715         (bool<mode>3_internal): Likewise.
3716         (boolc<mode>3_internal1): Likewise.
3717         (boolc<mode>3_internal2): Likewise.
3718         (boolcc<mode>3_internal1): Likewise.
3719         (boolcc<mode>3_internal2): Likewise.
3720         (eqv<mode>3_internal1): Likewise.
3721         (eqv<mode>3_internal2): Likewise.
3722         (one_cmpl1<mode>3_internal): Likewise.
3724 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3726         * config/microblaze/microblaze.c (microblaze_expand_prologue):
3727         Rename flag_stack_usage to flag_stack_usage_info.
3729 2013-07-23  David Holsgrove <david.holsgrove@xilinx.com>
3731         * config/microblaze/sync.md: New file.
3732         * config/microblaze/microblaze.md: Include sync.md
3733         * config/microblaze/microblaze.c: Add print_operand 'y'.
3734         * config/microblaze/constraints.md: Add memory_contraint
3735         'Q' which is a single register.
3737 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
3739         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
3741 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3743         * reload.c (find_reloads): Exit loop once we find this operand
3744         cannot be reloaded somehow for this alternative.
3746         * reload.c (find_reloads): Exit loop once we find a hard register.
3748         * rtlanal.c (computed_jump_p): Exit loop once we find label
3749         reference is used.
3751         * i386.c (ix86_pad_returns): Exit loop after setting replace.
3753         * cfgloopmanip.c (remove_path): Exit loop after setting
3754         irred_invalidated.
3756         * gensupport.c (subst_dup): Avoid loop if code is not
3757         MATCH_DUP nor MATCH_OP_DUP.
3759 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
3761         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
3763 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3765         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
3766         true for SP_REGNUM if mode == ptr_mode.
3767         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
3768         with value R0_REGNUM + 31.
3770 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3772         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
3773         pad pointer-typed argument downward.
3775 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3777         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
3778         and __ILP32__ when the ILP32 model is in use.
3780 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3782         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
3783         (aarch64_load_symref_appropriately): In the case of
3784         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
3785         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
3786         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
3787         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
3788         if 'mode' doesn't equal to 'ptr_mode'.
3789         (aarch64_output_mi_thunk): Add an assertion on the alignment of
3790         'vcall_offset'; change to call aarch64_emit_move differently depending
3791         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
3792         to calculate the upper bound of 'vcall_offset'.
3793         (aarch64_cannot_force_const_mem): Change to also return true if
3794         mode != ptr_mode.
3795         (aarch64_legitimize_reload_address): In the case of large
3796         displacements, add new local variable 'xmode' and an assertion
3797         based on it; change to use 'xmode' to generate the new rtx and
3798         reload.
3799         (aarch64_asm_trampoline_template): Change to generate the template
3800         differently depending on TARGET_ILP32 or not; change to use
3801         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
3802         (aarch64_trampoline_size): Removed.
3803         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
3804         and replace immediate literals with it.  Change to use 'ptr_mode'
3805         instead of 'DImode' and call convert_memory_address if the mode
3806         of 'fnaddr' doesn't equal to 'ptr_mode'.
3807         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
3808         to output symbol.
3809         (aarch64_elf_asm_destructor): Likewise.
3810         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
3811         on TARGET_ILP32 instead of aarch64_trampoline_size.
3812         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
3813         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
3814         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
3815         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
3816         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
3817         (storewb_pair<GPI:mode>_<P:mode>): ... this.
3818         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
3819         depending on the value of 'mode'.
3820         (add_losym_<mode>): New.
3821         (ldr_got_small_<mode>): New, based on ldr_got_small.
3822         (ldr_got_small): Remove.
3823         (ldr_got_small_sidi): New.
3824         * config/aarch64/iterators.md (P): New.
3825         (PTR): Change to 'ptr_mode' in the condition.
3827 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
3829         * config.gcc (aarch64*-*-*): Support --with-abi.
3830         (aarch64*-*-elf): Support --with-multilib-list.
3831         (aarch64*-*-linux*): Likewise.
3832         (supported_defaults): Add abi to aarch64*-*-*.
3833         * configure.ac: Mention AArch64 for --with-multilib-list.
3834         * configure: Re-generated.
3835         * config/aarch64/biarchilp32.h: New file.
3836         * config/aarch64/biarchlp64.h: New file.
3837         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
3838         (ABI_SPEC): Ditto.
3839         (MULTILIB_DEFAULTS): Ditto.
3840         (DRIVER_SELF_SPECS): Ditto.
3841         (ASM_SPEC): Update to also substitute -mabi.
3842         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
3843         file whose name depends on -mabi= and -mbig-endian.
3844         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
3845         TARGET_ILP32.
3846         (POINTER_SIZE): New define.
3847         (POINTERS_EXTEND_UNSIGNED): Ditto.
3848         (enum aarch64_abi_type): New enumeration tag.
3849         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
3850         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
3851         (TARGET_ILP32): New define.
3852         * config/aarch64/aarch64.opt (mabi): New.
3853         (aarch64_abi): New.
3854         (ilp32, lp64): New values for -mabi.
3855         * config/aarch64/t-aarch64 (comma): New define.
3856         (MULTILIB_OPTIONS): Ditto.
3857         (MULTILIB_DIRNAMES): Ditto.
3858         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
3859         * doc/invoke.texi: Document -mabi for AArch64.
3861 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
3863         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
3865 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3866             Anton Blanchard <anton@au1.ibm.com>
3868         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
3869         endianness when selecting field to splat.
3871 2013-07-22  Eric Christopher  <echristo@gmail.com>
3873         * dwarf2out.c (die_odr_checksum): New function to use
3874         CHECKSUM_ macros and ULEB128 for DIE tag.
3875         (generate_type_signature): Use.
3877 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
3879         * config.gcc (sparc*-*-*): Accept leon3 processor.
3880         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
3881         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
3882         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
3883         * config/sparc/sparc.opt (enum processor_type): Add leon3.
3884         (mfix-ut699): Adjust comment.
3885         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
3886         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
3887         (CPP_CPU_SPEC): Likewise.
3888         (ASM_CPU_SPEC): Likewise.
3889         * config/sparc/sparc.c (leon3_cost): New constant.
3890         (sparc_option_override): Add leon3 support.
3891         (mem_ref): New function.
3892         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
3893         (sparc_do_work_around_errata): Look into the instruction in the delay
3894         slot and adjust accordingly.  Add fix for the data cache nullify issues
3895         of the UT699.  Change insertion position for the NOP.
3896         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
3897         (leon3_load): New reservation.
3898         (leon_store): Bump latency to 2.
3899         (grfpu): New automaton.
3900         (grfpu_alu): New unit.
3901         (grfpu_ds): Likewise.
3902         (leon_fp_alu): Adjust.
3903         (leon_fp_mult): Delete.
3904         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
3905         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
3906         * config/sparc/sparc.md (cpu): Add leon3.
3907         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
3908         (swapsi): Likewise.
3909         (atomic_test_and_set): Likewise.
3910         (ldstub): Likewise.
3912 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
3914         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
3915         default for R5900 targets.
3916         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
3917         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
3918         * config/mips/mips.c (mips_option_override): Report an error for
3919         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
3920         for -march=r5900 -mhard-float.
3922 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
3924         * df-problems.c (can_move_insns_across): Exit loop once we
3925         find a non-fixed, non-global register.
3927         * ipa-pure-const.c (propagate_nothrow): Exit loop after
3928         setting can_throw.
3930         * omega.c (omega_eliminate_red): Break after setting red_found.
3931         (omega_problem_has_red_equations): Similarly after setting found.
3932         (omega_query_variable): Similarly after setting coupled.
3934 2013-07-22  Marek Polacek  <polacek@redhat.com>
3936         * gimplify.c: Don't include gimple.h twice.
3938 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3940         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
3941         instead of TARGET_THUMB1.
3942         (Pz): New constraint.
3943         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
3944         encodings.
3945         (compare_negsi_si): Likewise.
3946         (compare_addsi2_op0): Likewise.
3947         (compare_addsi2_op1): Likewise.
3948         (addsi3_carryin_<optab>): Likewise.
3949         (addsi3_carryin_alt2_<optab>): Likewise.
3950         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
3951         for arm_restrict_it.
3952         (subsi3_carryin): Likewise.
3953         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
3954         (minmax_arithsi): Disable for arm_restrict_it.
3955         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
3956         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
3957         (satsi_<SAT:code>_shift): Likewise.
3958         (arm_shiftsi3): Add alternative for 16-bit encoding.
3959         (arm32_movhf): Disable for arm_restrict_it.
3960         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
3961         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
3963 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
3965         * config/arm/arm.md (attribute "insn"): Delete.
3966         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
3967         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
3968         (not_shiftsi): Update for attribute change.
3969         (not_shiftsi_compare0): Likewise.
3970         (not_shiftsi_compare0_scratch): Likewise.
3971         (arm_one_cmplsi2): Likewise.
3972         (thumb1_one_cmplsi2): Likewise.
3973         (notsi_compare0): Likewise.
3974         (notsi_compare0_scratch): Likewise.
3975         (thumb1_movdi_insn): Likewise.
3976         (arm_movsi_insn): Likewise.
3977         (movhi_insn_arch4): Likewise.
3978         (movhi_bytes): Likewise.
3979         (arm_movqi_insn): Likewise.
3980         (thumb1_movqi_insn): Likewise.
3981         (arm32_movhf): Likewise.
3982         (thumb1_movhf): Likewise.
3983         (arm_movsf_soft_insn): Likewise.
3984         (thumb1_movsf_insn): Likewise.
3985         (thumb_movdf_insn): Likewise.
3986         (movsicc_insn): Likewise.
3987         (movsfcc_soft_insn): Likewise.
3988         (and_scc): Likewise.
3989         (cond_move): Likewise.
3990         (if_move_not): Likewise.
3991         (if_not_move): Likewise.
3992         (if_shift_move): Likewise.
3993         (if_move_shift): Likewise.
3994         (if_shift_shift): Likewise.
3995         (if_not_arith): Likewise.
3996         (if_arith_not): Likewise.
3997         (cond_move_not): Likewise.
3998         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3999         (neon_mov<mode>): Likewise.
4000         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
4001         (thumb2_movsi_vfp): Likewise.
4002         (movsf_vfp): Likewise.
4003         (thumb2_movsf_vfp): Likewise.
4004         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4005         change.
4006         (cortexa7_older_only): Likewise.
4007         (cortexa7_younger): Likewise.
4008         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
4009         (1020alu_shift_op): Likewise.
4010         (1020alu_shift_reg_op): Likewise.
4011         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
4012         (alu_shift_op): Likewise.
4013         (alu_shift_reg_op): Likewise.
4014         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
4015         (11_alu_shift_op): Likewise.
4016         (11_alu_shift_reg_op): Likewise.
4017         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
4018         (9_alu_shift_reg_op): Likewise.
4019         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4020         change.
4021         (cortex_a15_alu_shift): Likewise.
4022         (cortex_a15_alu_shift_reg): Likewise.
4023         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
4024         (cortex_a5_alu_shift): Likewise.
4025         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
4026         change.
4027         (cortex_a53_alu_shift): Likewise.
4028         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4029         change.
4030         (cortex_a7_alu_reg): Likewise.
4031         (cortex_a7_alu_shift): Likewise.
4032         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4033         (cortex_a8_alu_shift): Likewise.
4034         (cortex_a8_alu_shift_reg): Likewise.
4035         (cortex_a8_mov): Likewise.
4036         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
4037         (cortex_a9_dp_shift): Likewise.
4038         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
4039         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
4040         (cortex_r4_mov): Likewise.
4041         (cortex_r4_alu_shift): Likewise.
4042         (cortex_r4_alu_shift_reg): Likewise.
4043         * config/arm/fa526.md (526_alu_op): Update for attribute change.
4044         (526_alu_shift_op): Likewise.
4045         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
4046         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
4047         (626te_alu_shift_op): Likewise.
4048         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
4049         (726te_alu_op): Likewise.
4050         (726te_alu_shift_op): Likewise.
4051         (726te_alu_shift_reg_op): Likewise.
4052         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4053         (mp626_alu_shift_op): Likewise.
4054         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4055         (pj4_alu_e1_conds): Likewise.
4056         (pj4_alu): Likewise.
4057         (pj4_alu_conds): Likewise.
4058         (pj4_shift): Likewise.
4059         (pj4_shift_conds): Likewise.
4060         (pj4_alu_shift): Likewise.
4061         (pj4_alu_shift_conds): Likewise.
4063 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4065         * config/arm/predicates.md (shiftable_operator_strict_it):
4066         New predicate.
4067         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4068         Disable cond_exec version for arm_restrict_it.
4069         (thumb2_smaxsi3): Convert to generate cond_exec.
4070         (thumb2_sminsi3): Likewise.
4071         (thumb32_umaxsi3): Likewise.
4072         (thumb2_uminsi3): Likewise.
4073         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4074         (thumb2_neg_abssi2): Likewise.
4075         (thumb2_mov_scc): Add alternative for 16-bit encoding.
4076         (thumb2_movsicc_insn): Adjust alternatives.
4077         (thumb2_mov_negscc): Disable for arm_restrict_it.
4078         (thumb2_mov_negscc_strict_it): New pattern.
4079         (thumb2_mov_notscc_strict_it): New pattern.
4080         (thumb2_mov_notscc): Disable for arm_restrict_it.
4081         (thumb2_ior_scc): Likewise.
4082         (thumb2_ior_scc_strict_it): New pattern.
4083         (thumb2_cond_move): Adjust for arm_restrict_it.
4084         (thumb2_cond_arith): Disable for arm_restrict_it.
4085         (thumb2_cond_arith_strict_it): New pattern.
4086         (thumb2_cond_sub): Adjust for arm_restrict_it.
4087         (thumb2_movcond): Likewise.
4088         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4089         (thumb2_zero_extendhisi2_v6): Likewise.
4090         (thumb2_zero_extendqisi2_v6): Likewise.
4091         (orsi_notsi_si): Likewise.
4092         (orsi_not_shiftsi_si): Likewise.
4094 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
4096         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4097         instruction sequence is 1 byte shorter.
4099 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
4101         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4102         it is not needed after split.
4104 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
4106         PR target/51784
4107         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4108         second label for nonlocal goto receivers. Don't output pic base labels
4109         unless we're producing PIC; mark that action unreachable().
4110         (ix86_save_reg): If the function contains a nonlocal label, save the
4111         PIC base reg.
4112         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4113         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4114         (update_pic_label_number_if_needed): New.
4115         (machopic_output_function_base_name): Adjust for nonlocal receiver
4116         case.
4117         (machopic_should_output_picbase_label): New.
4118         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4119         (nonlocal_goto_receiver): New insn and split.
4121 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
4123         * config/aarch64/aarch64-builtins.c
4124         (aarch64_fold_builtin): Fold abs in all modes.
4125         * config/aarch64/aarch64-simd-builtins.def
4126         (abs): Enable for all modes.
4127         * config/aarch64/arm_neon.h
4128         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4129         (vabs_f64): Add missing intrinsic.
4131 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
4133         * config/aarch64/arm_neon.h (vabs_s64): New function
4135 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
4137         PR target/57516
4138         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4139         * config/avr/avr.md (adjust_len): Add `round'.
4140         * config/avr/avr-protos.h (avr_out_round): New prototype.
4141         (avr_out_plus): Add `out_label' argument.
4142         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4143         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4144         Handle the case where `insn' is just a pattern.
4145         (avr_out_bitop): Handle the case where `insn' is just a pattern.
4146         (avr_out_round): New function.
4147         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4149 2013-07-18  David Holsgrove <david.holsgrove@xilinx.com>
4151         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4152         Add check for flag_stack_usage to handle -fstack-usage support
4154 2013-07-18  Pat Haugen <pthaugen@us.ibm.com>
4156         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4157         interaction for new Power8 flags and VSX.
4159 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
4161         PR middle-end/57698
4162         * tree-inline.c (expand_call_inline): Emit errors during
4163         early_inlining only if optimization is not turned on.
4165 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4167         * passes.def: New.
4169         * passes.c (init_optimization_passes): Move the construction of
4170         the pass hierarchy into a new passes.def file.
4172         * Makefile.in (passes.o): Add dependency on passes.def.
4174 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4176         * passes.c (init_optimization_passes): Introduce macros for
4177         constructing the tree of passes (INSERT_PASSES_AFTER,
4178         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4179         TERMINATE_PASS_LIST).
4181 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
4182             Wei Mi  <wmi@google.com>
4184         PR rtl-optimization/57878
4185         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4186         top.
4187         (reload_pseudo_compare_func): Check nregs first for reload
4188         pseudos.
4190 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4192         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4194 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
4196         * read-rtl.c (validate_const_int): Once an invalid character is
4197         seen, quit the loop.
4199         * gengtype.c (write_roots): Similarly once we find the "deletable"
4200         or "if_marked" option.
4202 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4204         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4205         "xtab" and "sat".  Move value "clz" from here to ...
4206         (attriubte "type"): ... here.
4207         (satsi_<SAT:code>): Delete "insn" attribute.
4208         (satsi_<SAT:code>_shift): Likewise.
4209         (arm_zero_extendqisi2addsi): Likewise.
4210         (arm_extendqisi2addsi): Likewise.
4211         (clzsi2): Update for attribute changes.
4212         (rbitsi2): Likewise.
4213         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4214         attribute.
4215         (arm_usatsihi): Likewise.
4216         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4218 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4220         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4221         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
4222         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
4223         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
4224         in alphabetical order.
4225         (attribute "core_cycles"): Update for attribute changes.
4226         (arm_addsi3): Likewise.
4227         (addsi3_compare0): Likewise.
4228         (addsi3_compare0_scratch): Likewise.
4229         (addsi3_compare_op1): Likewise.
4230         (addsi3_compare_op2): Likewise.
4231         (compare_addsi2_op0): Likewise.
4232         (compare_addsi2_op1): Likewise.
4233         (addsi3_carryin_shift_<optab>): Likewise.
4234         (subsi3_carryin_shift): Likewise.
4235         (rsbsi3_carryin_shift): Likewise.
4236         (arm_subsi3_insn): Likewise.
4237         (subsi3_compare0): Likewise.
4238         (subsi3_compare): Likewise.
4239         (arm_andsi3_insn): Likewise.
4240         (thumb1_andsi3_insn): Likewise.
4241         (andsi3_compare0): Likewise.
4242         (andsi3_compare0_scratch): Likewise.
4243         (zeroextractsi_compare0_scratch
4244         (andsi_not_shiftsi_si): Likewise.
4245         (iorsi3_insn): Likewise.
4246         (iorsi3_compare0): Likewise.
4247         (iorsi3_compare0_scratch): Likewise.
4248         (arm_xorsi3): Likewise.
4249         (thumb1_xorsi3_insn): Likewise.
4250         (xorsi3_compare0): Likewise.
4251         (xorsi3_compare0_scratch): Likewise.
4252         (satsi_<SAT:code>_shift): Likewise.
4253         (rrx): Likewise.
4254         (arm_shiftsi3): Likewise.
4255         (shiftsi3_compare0): Likewise.
4256         (not_shiftsi): Likewise.
4257         (not_shiftsi_compare0): Likewise.
4258         (not_shiftsi_compare0_scratch): Likewise.
4259         (arm_one_cmplsi2): Likewise.
4260         (thumb_one_complsi2): Likewise.
4261         (notsi_compare0): Likewise.
4262         (notsi_compare0_scratch): Likewise.
4263         (thumb1_zero_extendhisi2): Likewise.
4264         (arm_zero_extendhisi2): Likewise.
4265         (arm_zero_extendhisi2_v6): Likewise.
4266         (arm_zero_extendhisi2addsi): Likewise.
4267         (thumb1_zero_extendqisi2): Likewise.
4268         (thumb1_zero_extendqisi2_v6): Likewise.
4269         (arm_zero_extendqisi2): Likewise.
4270         (arm_zero_extendqisi2_v6): Likewise.
4271         (arm_zero_extendqisi2addsi): Likewise.
4272         (thumb1_extendhisi2): Likewise.
4273         (arm_extendhisi2): Likewise.
4274         (arm_extendhisi2_v6): Likewise.
4275         (arm_extendqisi): Likewise.
4276         (arm_extendqisi_v6): Likewise.
4277         (arm_extendqisi2addsi): Likewise.
4278         (thumb1_extendqisi2): Likewise.
4279         (thumb1_movdi_insn): Likewise.
4280         (arm_movsi_insn): Likewise.
4281         (movsi_compare0): Likewise.
4282         (movhi_insn_arch4): Likewise.
4283         (movhi_bytes): Likewise.
4284         (arm_movqi_insn): Likewise.
4285         (thumb1_movqi_insn): Likewise.
4286         (arm32_movhf): Likewise.
4287         (thumb1_movhf): Likewise.
4288         (arm_movsf_soft_insn): Likewise.
4289         (thumb1_movsf_insn): Likewise.
4290         (movdf_soft_insn): Likewise.
4291         (thumb_movdf_insn): Likewise.
4292         (arm_cmpsi_insn): Likewise.
4293         (cmpsi_shiftsi): Likewise.
4294         (cmpsi_shiftsi_swp): Likewise.
4295         (arm_cmpsi_negshiftsi_si): Likewise.
4296         (movsicc_insn): Likewise.
4297         (movsfcc_soft_insn): Likewise.
4298         (arith_shiftsi): Likewise.
4299         (arith_shiftsi_compare0
4300         (arith_shiftsi_compare0_scratch
4301         (sub_shiftsi): Likewise.
4302         (sub_shiftsi_compare0
4303         (sub_shiftsi_compare0_scratch
4304         (and_scc): Likewise.
4305         (cond_move): Likewise.
4306         (if_plus_move): Likewise.
4307         (if_move_plus): Likewise.
4308         (if_move_not): Likewise.
4309         (if_not_move): Likewise.
4310         (if_shift_move): Likewise.
4311         (if_move_shift): Likewise.
4312         (if_shift_shift): Likewise.
4313         (if_not_arith): Likewise.
4314         (if_arith_not): Likewise.
4315         (cond_move_not): Likewise.
4316         (thumb1_ashlsi3): Set type attribute.
4317         (thumb1_ashrsi3): Likewise.
4318         (thumb1_lshrsi3): Likewise.
4319         (thumb1_rotrsi3): Likewise.
4320         (shiftsi3_compare0_scratch): Likewise.
4321         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
4322         (neon_mov<mode>): Likewise.
4323         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
4324         attribute changes.
4325         (thumb2_movsi_insn): Likewise.
4326         (thumb2_cmpsi_neg_shiftsi): Likewise.
4327         (thumb2_extendqisi_v6): Likewise.
4328         (thumb2_zero_extendhisi2_v6): Likewise.
4329         (thumb2_zero_extendqisi2_v6): Likewise.
4330         (thumb2_shiftsi3_short): Likewise.
4331         (thumb2_addsi3_compare0_scratch): Likewise.
4332         (orsi_not_shiftsi_si): Likewise.
4333         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
4334         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
4335         changes.
4336         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
4337         (1020alu_shift_op): Likewise.
4338         (1020alu_shift_reg_op): Likewise.
4339         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
4340         (alu_shift_op): Likewise.
4341         (alu_shift_reg_op): Likewise.
4342         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
4343         (11_alu_shift_op): Likewise.
4344         (11_alu_shift_reg_op): Likewise.
4345         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
4346         (9_alu_shift_reg_op): Likewise.
4347         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4348         changes.
4349         (cortex_a15_alu_shift): Likewise.
4350         (cortex_a15_alu_shift_reg): Likewise.
4351         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
4352         changes.
4353         (cortex_a5_alu_shift): Likewise.
4354         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
4355         changes.
4356         (cortex_a53_alu_shift): Likewise.
4357         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4358         changes.
4359         (cortex_a7_alu_reg): Likewise.
4360         (cortex_a7_alu_shift): Likewise.
4361         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
4362         changes.
4363         (cortex_a8_alu_shift): Likewise.
4364         (cortex_a8_alu_shift_reg): Likewise.
4365         (cortex_a8_mov): Likewise.
4366         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
4367         (cortex_a9_dp_shift): Likewise.
4368         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
4369         changes.
4370         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
4371         changes.
4372         (cortex_r4_mov): Likewise.
4373         (cortex_r4_alu_shift): Likewise.
4374         (cortex_r4_alu_shift_reg): Likewise.
4375         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
4376         (526_alu_shift_op): Likewise.
4377         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
4378         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
4379         (626te_alu_shift_op): Likewise.
4380         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
4381         (726te_alu_op): Likewise.
4382         (726te_alu_shift_op): Likewise.
4383         (726te_alu_shift_reg_op): Likewise.
4384         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
4385         (mp626_alu_shift_op): Likewise.
4386         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
4387         (pj4_alu_e1_conds): Likewise.
4388         (pj4_alu): Likewise.
4389         (pj4_alu_conds): Likewise.
4390         (pj4_shift): Likewise.
4391         (pj4_shift_conds): Likewise.
4392         (pj4_alu_shift): Likewise.
4393         (pj4_alu_shift_conds): Likewise.
4394         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4395         changes.
4396         (cortexa7_older_only): Likewise.
4397         (cortexa7_younger): Likewise.
4399 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4401         * ipa-pure-const.c (generate_summary): Rename to...
4402         (pure_const_generate_summary): ... this.
4404 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
4406         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
4408 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
4410         PR target/57909
4411         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
4412         usage in HI mode.
4414 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4416         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
4417         enabled without -march=zEC12.
4418         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
4419         flags to be set.
4421 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4423         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
4424         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
4425         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
4426         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
4427         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
4428         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
4429         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
4430         ISA_HAS_FP4.
4431         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
4432         and ISA_HAS_NMADD3_NMSUB3.
4433         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
4434         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
4435         (nmsub4<mode>, nmsub3<mode>): Likewise.
4436         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
4438 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4440         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
4441         TARGET_MIPS5400 checking.
4443 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
4444             Peter Bergner  <bergner@vnet.ibm.com>
4446         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
4447         registers in the comment.
4448         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
4449         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
4450         rather than FIRST_PSEUDO_REGISTERS.
4452 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
4454         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
4455         enable extra ISA flags with TARGET_HTM.
4457 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4459         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
4460         Fix comment typos.
4462 2013-07-15  Cong Hou  <congh@google.com>
4464         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
4465         in compare function for sorting.
4467 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
4469         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
4470         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
4471         * config/rs6000/rs6000.opt: Add -mhtm option.
4472         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
4473         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
4474         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
4475         __HTM__ if the HTM instructions are available.
4476         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
4477         htm_spr_reg_operand): New define_predicates.
4478         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
4479         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
4480         Include htm.md.
4481         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
4482         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
4483         HTM builtin functions.
4484         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
4485         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
4486         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
4487         (rs6000_builtin_mask_calculate): Likewise.
4488         (rs6000_option_override_internal): Likewise.
4489         (bdesc_htm): Add new HTM builtin support.
4490         (htm_spr_num): New function.
4491         (htm_spr_regno): Likewise.
4492         (rs6000_htm_spr_icode): Likewise.
4493         (htm_expand_builtin): Likewise.
4494         (htm_init_builtins): Likewise.
4495         (rs6000_expand_builtin): Add support for HTM builtin functions.
4496         (rs6000_init_builtins): Likewise.
4497         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
4498         option.
4499         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
4500         (TARGET_HTM, MASK_HTM): Define macros.
4501         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
4502         (FIXED_REGISTERS): Likewise.
4503         (CALL_USED_REGISTERS): Likewise.
4504         (CALL_REALLY_USED_REGISTERS): Likewise.
4505         (REG_ALLOC_ORDER): Likewise.
4506         (enum reg_class): Likewise.
4507         (REG_CLASS_NAMES): Likewise.
4508         (REG_CLASS_CONTENTS): Likewise.
4509         (REGISTER_NAMES): Likewise.
4510         (ADDITIONAL_REGISTER_NAMES): Likewise.
4511         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
4512         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
4513         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
4514         * config/rs6000/htm.md: New file.
4515         * config/rs6000/htmintrin.h: New file.
4516         * config/rs6000/htmxlintrin.h: New file.
4518 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4520         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4521         Define SYMBOL_TINY_GOT, update comment.
4522         * config/aarch64/aarch64.c
4523         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
4524         (aarch64_expand_mov_immediate): Likewise.
4525         (aarch64_print_operand): Likewise.
4526         (aarch64_classify_symbol): Likewise.
4527         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
4528         (ldr_got_tiny): Define.
4530 2013-07-13  Tobias Grosser  <tobias@grosser.es>
4532         PR tree-optimization/54094
4533         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
4534           scheduling dimension for the parallelism check from the polyhedral
4535           information in the AST.
4536         * graphite-dependences.c (carries_deps): Do not assume the schedule is
4537           in 2D + 1 form.
4539 2013-07-13  Jason Merrill  <jason@redhat.com>
4541         * print-tree.c (debug_vec_tree): Use debug_raw.
4542         (debug_raw (vec<tree, va_gc> &)): New.
4543         (debug_raw (vec<tree, va_gc> *)): New.
4544         * tree.h: Declare them.
4546 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
4548         * ifcvt.c (ifcvt_after_combine): New static variable.
4549         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
4550         for size.
4551         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
4552         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
4553         rest_of_handle_if_after_reload): Pass new argument for if_convert.
4555 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
4557         * config/mips/mips.c (mips_expand_call): Remove empty statement.
4559 2013-07-12  Michael Matz  <matz@suse.de>
4561         PR middle-end/55771
4562         * convert.c (convert_to_real): Reject non-float inner types.
4564 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
4566         * config/aarch64/aarch64-protos.h
4567         (aarch64_simd_immediate_valid_for_move): Remove.
4568         * config/aarch64/aarch64.c (simd_immediate_info): New member.
4569         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
4570         cases.
4571         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
4573 2013-07-11  Steve Ellcey  <sellcey@mips.com>
4575         * config/mips/mips.c (mips_conditional_register_usage): Do not
4576         use t[0-7] registers in MIPS16 mode when optimizing for size.
4578 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
4580         * config/i386/i386.c (dispatch_function_versions): Fix array
4581         indexing of function_version_info to match actual_versions.
4583 2013-07-11  Teresa Johnson  <tejohnson@google.com>
4585         * vec.h (struct va_gc): Move release out-of-line.
4586         (va_gc::release): Call ggc_free on released vec.
4588 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4590         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
4591         Require GOT register as additional operand in UNSPEC.
4592         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
4593         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4594         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4595         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
4596         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
4597         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
4598         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
4600 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
4602         PR target/57631
4603         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
4604         name seen by assembler/linker rather if available.
4606 2013-07-11  Andreas Schwab  <schwab@suse.de>
4608         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
4610 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
4612         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
4614 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
4616         * doc/tm.texi.in: Move hook documentation to ....
4617         * target.def: ... here.
4619         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
4620         text on @hook line.
4621         * doc/tm.texi: Regenerate.
4623 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
4625         PR c++/57869
4626         * doc/invoke.texi: Document -Wconditionally-supported.
4628 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4630         PR target/57844
4631         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
4632         of my_fp.
4634 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4636         PR target/57506
4637         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
4638         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
4639         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
4640         Remove duplicate devices.
4641         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
4642         * config/avr/t-multilib: Regenerate.
4643         * config/avr/avr-tables.opt: Regenerate.
4644         * doc/avr-mmcu.texi: Regenerate.
4646 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
4648         PR target/56987
4649         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
4651 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
4653         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
4654         the cost of MULT when optimizing for size.
4656 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
4658         * config/cr16/cr16-protos.h: Don't include target.h.
4660 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
4662         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
4663         adjust register size for TDmode and TFmode for VSX registers.
4665 2013-07-08  Kai Tietz  <ktietz@redhat.com>
4667         PR target/56892
4668         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
4669         hook_bool_const_tree_true.
4671 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4673         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
4674         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
4675         * config/s390/s390.md: Define FPR*_REGNUM constants.
4676         Fix FPR2_REGNUM constant (18 -> 17).
4677         ("*trunc<BFP:mode><DFP_ALL:mode>2")
4678         ("*trunc<DFP_ALL:mode><BFP:mode>2")
4679         ("trunc<BFP:mode><DFP_ALL:mode>2")
4680         ("trunc<DFP_ALL:mode><BFP:mode>2")
4681         ("*extend<BFP:mode><DFP_ALL:mode>2")
4682         ("*extend<DFP_ALL:mode><BFP:mode>2")
4683         ("extend<BFP:mode><DFP_ALL:mode>2")
4684         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
4685         FPR4_REGNUM.
4687 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
4689         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
4691 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4693         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
4694         and cfun_fpr_bit_p to cfun_fpr_save_p.
4695         (s390_frame_area, s390_register_info, s390_frame_info)
4696         (s390_emit_prologue, s390_emit_epilogue)
4697         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
4698         register numbers.
4699         * config/s390/s390.h: Define *_REGNUM macros for floating point
4700         register numbers.
4702 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
4704         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
4706 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
4708         PR rtl-optimization/57786
4709         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
4710         and break out of the loop when it is set to false.
4712 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
4714         PR target/57819
4715         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
4716         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
4717         (const_int 63)) 0)).
4718         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
4719         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
4720         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
4722         PR rtl-optimization/57829
4723         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
4724         mask bits outside of mode are just sign-extension from mode to HWI.
4726 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
4728         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
4729         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
4730         adjust_address instead of change_address to keep info about alignment.
4731         (emit_strmov): Remove.
4732         (emit_memmov): New function.
4733         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
4734         (expand_movmem_epilogue): Likewise and return updated rtx for
4735         destination.
4736         (expand_constant_movmem_prologue): Likewise and return updated rtx for
4737         destination and source.
4738         (decide_alignment): Refactor, handle vector_loop.
4739         (ix86_expand_movmem): Likewise.
4740         (ix86_expand_setmem): Likewise.
4741         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
4743 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
4745         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
4746         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
4748 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4750         * config/i386/sse.md (sse_movlhps): Change alternative 3
4751         of operand 2 to "m".
4753 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
4755         PR target/57807
4756         * config/i386/sse.md (iptr): New mode attribute.
4757         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
4758         (<sse>_vm<plusminus_insn><mode>3): Ditto.
4759         (<sse>_vmmul<mode>3): Ditto.
4760         (<sse>_vmdiv<mode>3): Ditto.
4761         (sse_vmrcpv4sf2): Ditto.
4762         (<sse>_vmsqrt<mode>2): Ditto.
4763         (sse_vmrsqrtv4sf2): Ditto.
4764         (<sse>_vm<code><mode>3): Ditto.
4765         (avx_vmcmp<mode>3): Ditto.
4766         (<sse>_vmmaskcmp<mode>3): Ditto.
4767         (<sse>_comi): Ditto.
4768         (<sse>_ucomi): Ditto.
4769         (*xop_vmfrcz_<mode>): Ditto.
4770         (*fmai_fmadd_<mode>): Ditto.
4771         (*fmai_fmsub_<mode>): Ditto.
4772         (*fmai_fnmadd_<mode>): Ditto.
4773         (*fmai_fnmsub_<mode>): Ditto.
4774         (*fma4i_vmfmadd_<mode>): Ditto.
4775         (*fma4i_vmfmsub_<mode>): Ditto.
4776         (*fma4i_vmfnmadd_<mode>): Ditto.
4777         (*fma4i_vmfnmsub_<mode>): Ditto.
4778         (*xop_vmfrcz_<mode>): Ditto.
4779         (sse_cvtps2pi): Ditto.
4780         (sse_cvttps2pi): Ditto.
4781         (sse_cvtss2si): Ditto.
4782         (sse_cvtss2si_2): Ditto.
4783         (sse_cvtss2siq_2): Ditto.
4784         (sse_cvttss2si): Ditto.
4785         (sse_cvttss2siq): Ditto.
4786         (sse_cvtsd2si): Ditto.
4787         (sse_cvtsd2si_2): Ditto.
4788         (sse_cvtsd2siq_2): Ditto.
4789         (sse_cvttsd2si): Ditto.
4790         (sse_cvttsd2siq): Ditto.
4791         (sse_cvtsd2ss): Ditto.
4792         (sse_cvtss2sd): Ditto.
4793         (avx2_pbroadcast<mode>): Ditto.
4794         (avx2_pbroadcast<mode>_1): Ditto.
4795         (*avx_vperm_broadcast_v4sf): Ditto.
4797         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
4798         (sse_movlhps): Ditto.
4799         (sse_storehps): Ditto.
4800         (sse_loadhps): Ditto.
4801         (sse_storelps): Ditto.
4802         (sse_loadlps): Ditto.
4803         (*vec_concatv4sf): Ditto.
4804         (*vec_interleave_highv2df): Ditto.
4805         (*vec_interleave_lowv2df): Ditto.
4806         (*vec_extractv2df_1_sse): Ditto.
4807         (*vec_extractv2df_0_sse): Ditto.
4808         (sse2_storelpd): Ditto.
4809         (sse2_loadlpd): Ditto.
4810         (sse2_movsd): Ditto.
4811         (*vec_concatv4si): Ditto.
4812         (vec_concatv2di): Ditto.
4814         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
4815         for Intel asm dialect.
4816         (mmx_punpcklwd): Ditto.
4817         (mmx_punpckldq): Ditto.
4819         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
4820         for intel assembler dialect.
4822 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
4824         PR target/29776
4825         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
4826         for BUILT_IN_C{LZ,LRSB}*.
4827         * tree.h (CASE_INT_FN): Add FN##IMAX case.
4828         * tree-vrp.c (extract_range_basic): Handle
4829         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
4830         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
4831         fall thru to code calling set_value*.
4832         * builtins.c (expand_builtin): Remove *IMAX cases.
4833         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
4834         if width is bigger than 2*HWI.
4836 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
4838         PR rtl-optimization/55342
4839         * lra-int.h (lra_subreg_reload_pseudos): New.
4840         * lra.c: Add undoing optional reloads to the block diagram.
4841         (lra_subreg_reload_pseudos): New.
4842         (lra_optional_reload_pseudos): Change comments.
4843         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
4844         lra_optional_reload_pseudos after undo transformations.
4845         * lra-assigns.c (pseudo_prefix_title): New.
4846         (lra_setup_reg_renumber): Use it.
4847         (spill_for): Ditto.  Check subreg reload pseudos too.
4848         (assign_by_spills): Consider subreg reload pseudos too.
4849         * lra-constraints.c (simplify_operand_subreg): Use
4850         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
4851         (curr_insn_transform): Recognize and do optional reloads.
4852         (undo_optional_reloads): New.
4853         (lra_undo_inheritance): Call undo_optional_reloads.
4855 2013-07-05  Thomas Quinot  <quinot@adacore.com>
4857         * tree-complex.c (expand_complex_operations_1): Fix typo.
4859 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
4861         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
4862         (tune_params): New member 'const vec_costs'.
4863         * config/aarch64/aarch64.c (generic_vector_cost): New.
4864         (generic_tunings): New member 'generic_vector_cost'.
4865         (aarch64_builtin_vectorization_cost): New.
4866         (aarch64_add_stmt_cost): New.
4867         (TARGET_VECTORIZE_ADD_STMT_COST): New.
4868         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
4870 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
4872         PR target/57777
4873         * config/i386/predicates.md (vsib_address_operand): Disallow
4874         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
4876 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
4878         PR middle-end/55030
4879         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
4880         expand_builtin_setjmp_receiver.
4881         (expand_label): Adjust, call expand_builtin_setjmp_receiver
4882         with NULL for the label parameter.
4883         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
4884         the frame-pointer.  Adjust comments.
4885         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
4886         only if LABEL is non-NULL.
4888 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
4890         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
4891         (ARM_ABI_AAPCS64): Ditto.
4892         (arm_abi): Ditto.
4893         (ARM_DEFAULT_ABI): Ditto.
4895 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
4897         * config/aarch64/aarch64-builtins.c
4898         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
4899         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
4900         (st1): Likewise.
4901         * config/aarch64/aarch64-simd.md
4902         (aarch64_ld1<VALL:mode>): New.
4903         (aarch64_st1<VALL:mode>): Likewise.
4904         * config/aarch64/arm_neon.h
4905         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
4907 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
4909         * config/i386/i386.c (gate_insert_vzeroupper): Check if
4910         target ISA is AVX.
4911         (ix86_option_override_internal):Turn on all -mavx target flags by
4912         default as they are dependent on AVX anyway.
4914 2013-07-02  Cary Coutant  <ccoutant@google.com>
4916         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
4917         deterministic hash.
4918         (loc_checksum_ordered): Likewise.
4919         (hash_loc_operands): Remove inline keyword.
4921 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
4923         PR tree-optimization/57741
4924         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
4925         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
4926         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
4927         Allow REAL_CST step_exprs if flag_associative_math.
4928         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
4930 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4932         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
4934 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
4936         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
4938 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4940         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
4941         encoding.
4942         (iorsi3_insn): Likewise.
4943         (arm_xorsi3): Likewise.
4945 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
4947         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
4948         to ...
4949         (attribute "type"): ... here, and prefix with "wmmx_".
4950         (attribute "core_cycles"): Update for attribute changes.
4951         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
4952         (tbcstv4hi): Likewise.
4953         (tbcstv2si): Likewise.
4954         (iwmmxt_iordi3): Likewise.
4955         (iwmmxt_xordi3): Likewise.
4956         (iwmmxt_anddi3): Likewise.
4957         (iwmmxt_nanddi3): Likewise.
4958         (iwmmxt_arm_movdi): Likewise.
4959         (iwmmxt_movsi_insn): Likewise.
4960         (mov<mode>_internal): Likewise.
4961         (and<mode>3_iwmmxt): Likewise.
4962         (ior<mode>3_iwmmxt): Likewise.
4963         (xor<mode>3_iwmmxt): Likewise.
4964         (add<mode>3_iwmmxt): Likewise.
4965         (ssaddv8qi3): Likewise.
4966         (ssaddv4hi3): Likewise.
4967         (ssaddv2si3): Likewise.
4968         (usaddv8qi3): Likewise.
4969         (usaddv4hi3): Likewise.
4970         (usaddv2si3): Likewise.
4971         (sub<mode>3_iwmmxt): Likewise.
4972         (sssubv8qi3): Likewise.
4973         (sssubv4hi3): Likewise.
4974         (sssubv2si3): Likewise.
4975         (ussubv8qi3): Likewise.
4976         (ussubv4hi3): Likewise.
4977         (ussubv2si3): Likewise.
4978         (mulv4hi3_iwmmxt): Likewise.
4979         (smulv4hi3_highpart): Likewise.
4980         (umulv4hi3_highpart): Likewise.
4981         (iwmmxt_wmacs): Likewise.
4982         (iwmmxt_wmacsz): Likewise.
4983         (iwmmxt_wmacu): Likewise.
4984         (iwmmxt_wmacuz): Likewise.
4985         (iwmmxt_clrdi): Likewise.
4986         (iwmmxt_clrv8qi): Likewise.
4987         (iwmmxt_clr4hi): Likewise.
4988         (iwmmxt_clr2si): Likewise.
4989         (iwmmxt_uavgrndv8qi3): Likewise.
4990         (iwmmxt_uavgrndv4hi3): Likewise.
4991         (iwmmxt_uavgv8qi3): Likewise.
4992         (iwmmxt_uavgv4hi3): Likewise.
4993         (iwmmxt_tinsrb): Likewise.
4994         (iwmmxt_tinsrh): Likewise.
4995         (iwmmxt_tinsrw): Likewise.
4996         (iwmmxt_textrmub): Likewise.
4997         (iwmmxt_textrmsb): Likewise.
4998         (iwmmxt_textrmuh): Likewise.
4999         (iwmmxt_textrmsh): Likewise.
5000         (iwmmxt_textrmw): Likewise.
5001         (iwmxxt_wshufh): Likewise.
5002         (eqv8qi3): Likewise.
5003         (eqv4hi3): Likewise.
5004         (eqv2si3): Likewise.
5005         (gtuv8qi3): Likewise.
5006         (gtuv4hi3): Likewise.
5007         (gtuv2si3): Likewise.
5008         (gtv8qi3): Likewise.
5009         (gtv4hi3): Likewise.
5010         (gtv2si3): Likewise.
5011         (smax<mode>3_iwmmxt): Likewise.
5012         (umax<mode>3_iwmmxt): Likewise.
5013         (smin<mode>3_iwmmxt): Likewise.
5014         (umin<mode>3_iwmmxt): Likewise.
5015         (iwmmxt_wpackhss): Likewise.
5016         (iwmmxt_wpackwss): Likewise.
5017         (iwmmxt_wpackdss): Likewise.
5018         (iwmmxt_wpackhus): Likewise.
5019         (iwmmxt_wpackwus): Likewise.
5020         (iwmmxt_wpackdus): Likewise.
5021         (iwmmxt_wunpckihb): Likewise.
5022         (iwmmxt_wunpckihh): Likewise.
5023         (iwmmxt_wunpckihw): Likewise.
5024         (iwmmxt_wunpckilb): Likewise.
5025         (iwmmxt_wunpckilh): Likewise.
5026         (iwmmxt_wunpckilw): Likewise.
5027         (iwmmxt_wunpckehub): Likewise.
5028         (iwmmxt_wunpckehuh): Likewise.
5029         (iwmmxt_wunpckehuw): Likewise.
5030         (iwmmxt_wunpckehsb): Likewise.
5031         (iwmmxt_wunpckehsh): Likewise.
5032         (iwmmxt_wunpckehsw): Likewise.
5033         (iwmmxt_wunpckelub): Likewise.
5034         (iwmmxt_wunpckeluh): Likewise.
5035         (iwmmxt_wunpckeluw): Likewise.
5036         (iwmmxt_wunpckelsb): Likewise.
5037         (iwmmxt_wunpckelsh): Likewise.
5038         (iwmmxt_wunpckelsw): Likewise.
5039         (ror<mode>3): Likewise.
5040         (ashr<mode>3_iwmmxt): Likewise.
5041         (lshr<mode>3_iwmmxt): Likewise.
5042         (ashl<mode>3_iwmmxt): Likewise.
5043         (ror<mode>3_di): Likewise.
5044         (ashr<mode>3_di): Likewise.
5045         (lshr<mode>3_di): Likewise.
5046         (ashl<mode>3_di): Likewise.
5047         (iwmmxt_wmadds): Likewise.
5048         (iwmmxt_wmaddu): Likewise.
5049         (iwmmxt_tmia): Likewise.
5050         (iwmmxt_tmiaph): Likewise.
5051         (iwmmxt_tmiabb): Likewise.
5052         (iwmmxt_tmiatb): Likewise.
5053         (iwmmxt_tmiabt): Likewise.
5054         (iwmmxt_tmiatt): Likewise.
5055         (iwmmxt_tmovmskb): Likewise.
5056         (iwmmxt_tmovmskh): Likewise.
5057         (iwmmxt_tmovmskw): Likewise.
5058         (iwmmxt_waccb): Likewise.
5059         (iwmmxt_wacch): Likewise.
5060         (iwmmxt_waccw): Likewise.
5061         (iwmmxt_waligni): Likewise.
5062         (iwmmxt_walignr): Likewise.
5063         (iwmmxt_walignr0): Likewise.
5064         (iwmmxt_walignr1): Likewise.
5065         (iwmmxt_walignr2): Likewise.
5066         (iwmmxt_walignr3): Likewise.
5067         (iwmmxt_wsadb): Likewise.
5068         (iwmmxt_wsadh): Likewise.
5069         (iwmmxt_wsadbz): Likewise.
5070         (iwmmxt_wsadhz): Likewise.
5071         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5072         (iwmmxt_wabsdiffb): Likewise.
5073         (iwmmxt_wabsdiffh): Likewise.
5074         (iwmmxt_wabsdiffw): Likewise.
5075         (iwmmxt_waddsubhx): Likewise
5076         (iwmmxt_wsubaddhx): Likewise.
5077         (addc<mode>3): Likewise.
5078         (iwmmxt_avg4): Likewise.
5079         (iwmmxt_avg4r): Likewise.
5080         (iwmmxt_wmaddsx): Likewise.
5081         (iwmmxt_wmaddux): Likewise.
5082         (iwmmxt_wmaddsn): Likewise.
5083         (iwmmxt_wmaddun): Likewise.
5084         (iwmmxt_wmulwsm): Likewise.
5085         (iwmmxt_wmulwum): Likewise.
5086         (iwmmxt_wmulsmr): Likewise.
5087         (iwmmxt_wmulumr): Likewise.
5088         (iwmmxt_wmulwsmr): Likewise.
5089         (iwmmxt_wmulwumr): Likewise.
5090         (iwmmxt_wmulwl): Likewise.
5091         (iwmmxt_wqmulm): Likewise.
5092         (iwmmxt_wqmulwm): Likewise.
5093         (iwmmxt_wqmulmr): Likewise.
5094         (iwmmxt_wqmulwmr): Likewise.
5095         (iwmmxt_waddbhusm): Likewise.
5096         (iwmmxt_waddbhusl): Likewise.
5097         (iwmmxt_wqmiabb): Likewise.
5098         (iwmmxt_wqmiabt): Likewise.
5099         (iwmmxt_wqmiatb): Likewise.
5100         (iwmmxt_wqmiatt): Likewise.
5101         (iwmmxt_wqmiabbn): Likewise.
5102         (iwmmxt_wqmiabtn): Likewise.
5103         (iwmmxt_wqmiatbn): Likewise.
5104         (iwmmxt_wqmiattn): Likewise.
5105         (iwmmxt_wmiabb): Likewise.
5106         (iwmmxt_wmiabt): Likewise.
5107         (iwmmxt_wmiatb): Likewise.
5108         (iwmmxt_wmiatt): Likewise.
5109         (iwmmxt_wmiabbn): Likewise.
5110         (iwmmxt_wmiabtn): Likewise.
5111         (iwmmxt_wmiatbn): Likewise.
5112         (iwmmxt_wmiattn): Likewise.
5113         (iwmmxt_wmiawbb): Likewise.
5114         (iwmmxt_wmiawbt): Likewise.
5115         (iwmmxt_wmiawtb): Likewise.
5116         (iwmmxt_wmiawtt): Likewise.
5117         (iwmmxt_wmiawbbn): Likewise.
5118         (iwmmxt_wmiawbtn): Likewise.
5119         (iwmmxt_wmiawtbn): Likewise.
5120         (iwmmxt_wmiawttn): Likewise.
5121         (iwmmxt_wmerge): Likewise.
5122         (iwmmxt_tandc<mode>3): Likewise.
5123         (iwmmxt_torc<mode>3): Likewise.
5124         (iwmmxt_torvsc<mode>3): Likewise.
5125         (iwmmxt_textrc<mode>3): Likewise.
5126         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5127         (wmmxt_pack): Likewise.
5128         (wmmxt_mult_c1): Likewise.
5129         (wmmxt_mult_c2): Likewise.
5130         (wmmxt_alu_c1): Likewise.
5131         (wmmxt_alu_c2): Likewise.
5132         (wmmxt_alu_c3): Likewise.
5133         (wmmxt_transfer_c1): Likewise.
5134         (wmmxt_transfer_c2): Likewise.
5135         (wmmxt_transfer_c3): Likewise.
5136         (marvell_f_iwmmxt_wstr): Likewise.
5137         (marvell_f_iwmmxt_wldr): Likewise.
5139 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
5141         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5143 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5145         Revert:
5146         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5147         * lra-constraints.c (need_for_split_p): Check call used hard regs
5148         living through calls.
5150         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5151         call used regs for call insn.
5153 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
5155         PR target/57736
5156         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5157         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5159 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
5161         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5163 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5165         * lra-constraints.c (need_for_split_p): Check call used hard regs
5166         living through calls.
5168 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
5170         PR target/57744
5171         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5172         to tie with any other modes.  Eliminate Altivec vector mode tests,
5173         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
5174         code, to return 0 if testing MODE2 for a condition, if we've
5175         already tested MODE1 for the same condition.
5177 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5179         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5180         layout.
5182 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5184         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5185         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5187 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5189         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5190         Define.
5191         (aarch64_symbolic_constant_p): Remove.
5192         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5193         static.  Fix line length and white space.
5194         (aarch64_symbolic_constant_p): Remove.
5195         * config/aarch64/predicates.md (aarch64_valid_symref):
5196         Use aarch64_classify_symbol_expression.
5198 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5200         * config/arm/constraints.md (Ts): New constraint.
5201         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5202         16-bit encodings.
5203         (compare_scc): Use "Ts" constraint for operand 0.
5204         (ior_scc_scc): Likewise.
5205         (and_scc_scc): Likewise.
5206         (and_scc_scc_nodom): Likewise.
5207         (ior_scc_scc_cmp): Likewise for operand 7.
5208         (and_scc_scc_cmp): Likewise.
5209         * config/arm/thumb2.md (thumb2_movsi_insn):
5210         Add alternatives for 16-bit encodings.
5211         (thumb2_movhi_insn): Likewise.
5212         (thumb2_movsicc_insn): Likewise.
5213         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
5214         (thumb2_negscc): Use "Ts" constraint.
5215         Move mvn instruction outside cond_exec block.
5216         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5217         for 16-bit encodings.
5219 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5221         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5222         encoding.
5223         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5224         (mulsi3subsi): Likewise.
5225         (mulsidi3adddi): Likewise.
5226         (mulsidi3_v6): Likewise.
5227         (umulsidi3_v6): Likewise.
5228         (umulsidi3adddi_v6): Likewise.
5229         (smulsi3_highpart_v6): Likewise.
5230         (umulsi3_highpart_v6): Likewise.
5231         (mulhisi3tb): Likewise.
5232         (mulhisi3bt): Likewise.
5233         (mulhisi3tt): Likewise.
5234         (maddhisi4): Likewise.
5235         (maddhisi4tb): Likewise.
5236         (maddhisi4tt): Likewise.
5237         (maddhidi4): Likewise.
5238         (maddhidi4tb): Likewise.
5239         (maddhidi4tt): Likewise.
5240         (zeroextractsi_compare0_scratch): Likewise.
5241         (insv_zero): Likewise.
5242         (insv_t2): Likewise.
5243         (anddi_notzesidi_di): Likewise.
5244         (anddi_notsesidi_di): Likewise.
5245         (andsi_notsi_si): Likewise.
5246         (iordi_zesidi_di): Likewise.
5247         (xordi_zesidi_di): Likewise.
5248         (andsi_iorsi3_notsi): Likewise.
5249         (smax_0): Likewise.
5250         (smax_m1): Likewise.
5251         (smin_0): Likewise.
5252         (not_shiftsi): Likewise.
5253         (unaligned_loadsi): Likewise.
5254         (unaligned_loadhis): Likewise.
5255         (unaligned_loadhiu): Likewise.
5256         (unaligned_storesi): Likewise.
5257         (unaligned_storehi): Likewise.
5258         (extv_reg): Likewise.
5259         (extzv_t2): Likewise.
5260         (divsi3): Likewise.
5261         (udivsi3): Likewise.
5262         (arm_zero_extendhisi2addsi): Likewise.
5263         (arm_zero_extendqisi2addsi): Likewise.
5264         (compareqi_eq0): Likewise.
5265         (arm_extendhisi2_v6): Likewise.
5266         (arm_extendqisi2addsi): Likewise.
5267         (arm_movt): Likewise.
5268         (thumb2_ldrd): Likewise.
5269         (thumb2_ldrd_base): Likewise.
5270         (thumb2_ldrd_base_neg): Likewise.
5271         (thumb2_strd): Likewise.
5272         (thumb2_strd_base): Likewise.
5273         (thumb2_strd_base_neg): Likewise.
5274         (arm_negsi2): Add alternative for 16-bit encoding.
5275         (arm_one_cmplsi2): Likewise.
5277 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5279         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
5280         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
5281         (movdfcc): Likewise.
5282         * config/arm/vfp.md (*thumb2_movsf_vfp):
5283         Disable predication for arm_restrict_it.
5284         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
5285         (*thumb2_movdfcc_vfp): Likewise.
5286         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
5287         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
5288         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
5289         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
5290         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
5291         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
5292         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
5293         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
5294         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
5295         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
5296         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
5297         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
5298         Disable predication for arm_restrict_it.
5300 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
5302         * config/i386/bmiintrin.h (_bextr_u32): New.
5303         (_bextr_u64): Ditto.
5305 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
5307         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
5308         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
5309         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
5310         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
5311         * config/mips/n32-elf.h: ...this new file.
5313 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
5315         PR target/57224
5316         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
5317         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
5319 2013-06-27  Catherine Moore <clm@codesourcery.com>
5321         * config/mips/mips-tables.opt: Regenerate.
5322         * config/mips/mips-cpus.def: Add m14ke and m14kec.
5323         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
5324         * doc/invoke.texi: Add -m14kc.
5326 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5328         PR target/57623
5329         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
5330         constraints of operand 1 and 2.
5332         PR target/57623
5333         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
5334         to match RTL canonicalization.  Swap predicates and
5335         constraints of operand 1 and 2.
5337 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
5339         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
5340         Process OP_INOUT regs for splitting too.
5342 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5344         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
5345         decl before the loop, initialize to NULL.
5346         (vectorizable_load): Initialize ptr_incr to NULL.
5348 2013-06-27  Martin Jambor  <mjambor@suse.cz>
5350         PR lto/57208
5351         * ipa-ref.h (ipa_maybe_record_reference): Declare.
5352         * ipa-ref.c (ipa_maybe_record_reference): New function.
5353         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
5354         * ipa-cp.c (create_specialized_node): Record potential references from
5355         aggvals.
5356         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
5358 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5360         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
5361         parameter 'mode' of type 'enum machine_mode mode'; change to pass
5362         'mode' to force_reg.
5363         (aarch64_add_offset): Update calls to aarch64_force_temporary.
5364         (aarch64_expand_mov_immediate): Likewise.
5366 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5368         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
5369         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
5371 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5373         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
5374         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5375         (struct machine_function): Add tbegin_p.
5376         (s390_canonicalize_comparison): Fold CC mode compares to
5377         conditional jump if possible.
5378         (s390_emit_jump): Return the emitted jump.
5379         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
5380         Handle CCRAWmode compares.
5381         (s390_option_override): Default to -mhtm if available.
5382         (s390_reg_clobbered_rtx): Handle floating point regs as well.
5383         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
5384         FPRs instead of df_regs_ever_live_p.
5385         (s390_optimize_nonescaping_tx): New function.
5386         (s390_init_frame_layout): Extend clobbered_regs array to cover
5387         FPRs as well.
5388         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
5389         (s390_expand_tbegin): New function.
5390         (enum s390_builtin): New enum definition.
5391         (code_for_builtin): New array definition.
5392         (s390_init_builtins): New function.
5393         (s390_expand_builtin): New function.
5394         (TARGET_INIT_BUILTINS): Define.
5395         (TARGET_EXPAND_BUILTIN): Define.
5396         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
5397         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
5398         (s390_alc_comparison): Likewise.
5399         * config/s390/s390-modes.def: Add CCRAWmode.
5400         * config/s390/s390.h (processor_flags): Add PF_TX.
5401         (TARGET_CPU_HTM): Define macro.
5402         (TARGET_HTM): Define macro.
5403         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
5404         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
5405         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5406         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
5407         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
5408         (TBEGIN_MASK, TBEGINC_MASK): New constants.
5409         ("*cc_to_int"): Move up.
5410         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
5411         constants other than 0.
5412         ("*ccraw_to_int"): New insn and splitter definition.
5413         ("tbegin", "tbegin_nofloat", "tbegin_retry")
5414         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
5415         ("tx_assist"): New expander.
5416         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
5417         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
5418         * config/s390/s390.opt: Add -mhtm option.
5419         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
5420         * config/s390/htmxlintrin.h: New file.
5421         * config/s390/htmintrin.h: New file.
5422         * config/s390/s390intrin.h: New file.
5423         * doc/extend.texi: Document htm builtins.
5424         * config.gcc: Add the new header files to extra_headers.
5426 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
5428         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
5429         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
5431 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
5432             Pat Haugen <pthaugen@us.ibm.com>
5433             Peter Bergner <bergner@vnet.ibm.com>
5435         * config/rs6000/power8.md: New.
5436         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
5437         setting for power8 entry.
5438         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
5439         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
5440         test for Power4/Power5 only.
5441         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
5442         support.
5443         (force_new_group): Adjust comment.
5444         * config/rs6000/rs6000.md: Include power8.md.
5446 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
5448         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
5449         * config/arm/arm-protos.h (arm_max_conditional_execute): New
5450         declaration.
5451         (tune_params): Update comment.
5452         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
5453         (arm_max_conditional_execute): New function.
5454         (thumb2_final_prescan_insn): Use max_insn_skipped and
5455         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
5457 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
5459         PR tree-optimization/57705
5460         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
5461         SSA_NAME step, provided that it is not defined inside the loop.
5462         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
5463         (get_initial_def_for_induction): Handle SSA_NAME IV step.
5465 2013-06-25  Martin Jambor  <mjambor@suse.cz>
5467         PR middle-end/57670
5468         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
5469         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
5470         calls in the dump.
5471         (ipa_note_param_call): Initialize member_ptr flag.
5472         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
5473         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
5474         (ipa_write_indirect_edge_info): Stream member_ptr flag.
5475         (ipa_read_indirect_edge_info): Likewise.
5477 2013-06-25  Richard Biener  <rguenther@suse.de>
5479         PR middle-end/56977
5480         * passes.c (init_optimization_passes): Move pass_fold_builtins
5481         and pass_dce earlier with -Og.
5483 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
5485         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
5486         <BIT_FIELD_REF>: Remove trailing TAB.
5487         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
5488         remove blank line.
5490 2013-06-24  Martin Jambor  <mjambor@suse.cz>
5492         PR tree-optimization/57358
5493         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
5494         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
5495         (ipa_analyze_params_uses): Generate pessimistic info when true.
5497 2013-06-24  Martin Jambor  <mjambor@suse.cz>
5499         PR tree-optimization/57539
5500         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
5501         global.inlined_to of the new node to it.  All callers changed.
5502         * ipa-inline-transform.c (clone_inlined_nodes): New variable
5503         inlining_into, pass it to cgraph_clone_node.
5504         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
5505         ipa_free_edge_args_substructures.
5506         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
5507         rdesc linked list.  Do not assert rdesc edges have inlined caller.
5508         Assert we have found an rdesc in the rdesc list.
5510 2013-06-24  Richard Biener  <rguenther@suse.de>
5512         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
5513         (pointer_set_lookup): Declare.
5514         (class pointer_map): New template class implementing a
5515         generic pointer to T map.
5516         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
5517         pointer_map<T>::contains, pointer_map<T>::insert,
5518         pointer_map<T>::traverse): New functions.
5519         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
5520         (pointer_set_lookup): New function.
5521         (pointer_set_contains): Use pointer_set_lookup.
5522         (pointer_set_insert): Likewise.
5523         (insert_aux): Remove.
5524         (struct pointer_map_t): Embed a pointer_set_t.
5525         (pointer_map_create): Adjust.
5526         (pointer_map_destroy): Likewise.
5527         (pointer_map_contains): Likewise.
5528         (pointer_map_insert): Likewise.
5529         (pointer_map_traverse): Likewise.
5530         * tree-streamer.h (struct streamer_tree_cache_d): Use a
5531         pointer_map<unsigned> instead of a pointer_map_t.
5532         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
5533         (streamer_tree_cache_lookup): Likewise.
5534         (streamer_tree_cache_create): Likewise.
5535         (streamer_tree_cache_delete): Likewise.
5536         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
5537         pointer_map<unsigned> instead of a pointer_map_t.
5538         (lto_init_tree_ref_encoder): Adjust.
5539         (lto_destroy_tree_ref_encoder): Likewise.
5540         * lto-section-out.c (lto_output_decl_index): Likewise.
5541         (lto_record_function_out_decl_state): Likewise.
5542         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
5544 2013-06-24  Richard Biener  <rguenther@suse.de>
5546         PR tree-optimization/57488
5547         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
5549 2013-06-24  Alan Modra  <amodra@gmail.com>
5551         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
5552         (gen_easy_altivec_constant): Likewise.
5553         * config/rs6000/predicates.md (easy_vector_constant_add_self,
5554         easy_vector_constant_msb): Likewise.
5556 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
5558         PR target/57688
5559         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
5560         add missing return true.
5562 2013-06-23  Oleg Endo <olegendo@gcc.gnu.org>
5564         PR target/52483
5565         * config/sh/predicates.md (general_extend_operand): Invoke
5566         general_movsrc_operand for memory operands.
5567         (general_movsrc_operand): Allow reg+reg addressing, do not use
5568         general_operand for memory operands.
5570 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
5572         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
5573         when current target options does not apply.
5574         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
5575         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
5576         * config/i386/bmiintrin.h: Pass appropriate target
5577         attributes to header.
5578         * config/i386/mmintrin.h: Ditto.
5579         * config/i386/nmmintrin.h: Ditto.
5580         * config/i386/avx2intrin.h: Ditto.
5581         * config/i386/fxsrintrin.h: Ditto.
5582         * config/i386/tbmintrin.h: Ditto.
5583         * config/i386/xsaveintrin.h: Ditto.
5584         * config/i386/f16cintrin.h: Ditto.
5585         * config/i386/xtestintrin.h: Ditto.
5586         * config/i386/xsaveoptintrin.h: Ditto.
5587         * config/i386/bmi2intrin.h: Ditto.
5588         * config/i386/lzcntintrin.h: Ditto.
5589         * config/i386/smmintrin.h: Ditto.
5590         * config/i386/wmmintrin.h: Ditto.
5591         * config/i386/x86intrin.h: Remove all header include guards.
5592         * config/i386/prfchwintrin.h: Ditto.
5593         * config/i386/pmmintrin.h: Ditto.
5594         * config/i386/tmmintrin.h: Ditto.
5595         * config/i386/xmmintrin.h: Ditto.
5596         * config/i386/popcntintrin.h: Ditto.
5597         * config/i386/rdseedintrin.h: Ditto.
5598         * config/i386/ammintrin.h: Ditto.
5599         * config/i386/emmintrin.h: Ditto.
5600         * config/i386/immintrin.h: Remove all header include guards.
5601         * config/i386/fma4intrin.h: Ditto.
5602         * config/i386/lwpintrin.h: Ditto.
5603         * config/i386/xopintrin.h: Ditto.
5604         * config/i386/ia32intrin.h: Ditto.
5605         * config/i386/avxintrin.h: Ditto.
5606         * config/i386/rtmintrin.h: Ditto.
5607         * config/i386/fmaintrin.h: Ditto.
5608         * config/i386/mm3dnow.h: Ditto.
5610 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
5612         * common/config/i386/i386-common.c: Handle LZCNT.
5614 2013-06-22  Andi Kleen  <ak@linux.intel.com>
5616         * doc/extend.texi: Use __atomic_store_n instead of
5617         __atomic_store in HLE example.
5619 2013-06-22  Oleg Endo <olegendo@gcc.gnu.org>
5621         * config/sh/sh.c: Remove <cstdlib> workaround.
5623 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5625         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
5627 2013-06-21  Andi Kleen  <ak@linux.intel.com>
5629         * doc/extend.texi: Document that __atomic_clear and
5630           __atomic_test_and_set should only be used with bool.
5632 2013-06-20  Jan Hubicka  <jh@suse.cz>
5634         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
5635         types_same_for_odr.
5636         * tree.c (decls_same_for_odr): New function.
5637         (same_for_edr): New function.
5638         (types_same_for_odr): New function.
5639         (get_binfo_at_offset): Use it.
5640         * tree.h (types_same_for_odr): Declare.
5642 2013-06-20  Oleg Endo <olegendo@gcc.gnu.org>
5643             Jason Merrill  <jason@redhat.com>
5645         * system.h: Include <cstdlib> as well as <stdlib.h>.
5647 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
5649         PR target/57655
5650         * config/i386/i386.c (construct_container): Report error if
5651         long double is used with disabled x87 float returns.
5653 2013-06-20  Jan Hubicka  <jh@suse.cz>
5655         * lto-cgraph.c (input_symtab): Do not set cgraph state.
5657 2013-06-20  Joern Rennecke <joern.rennecke@embecosm.com>
5659         PR rtl-optimization/57425
5660         PR rtl-optimization/57569
5661         * alias.c (write_dependence_p): Remove parameters mem_mode and
5662         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
5663         Changed all callers.
5664         (canon_anti_dependence): Get comments and semantics in sync.
5665         Add parameter mem_canonicalized.  Changed all callers.
5666         * rtl.h (canon_anti_dependence): Update prototype.
5668 2013-06-20  Richard Biener  <rguenther@suse.de>
5670         * data-streamer-in.c (streamer_read_uhwi): Optimize single
5671         byte case, inline streamer_read_uchar and defer section
5672         overrun check.
5674 2013-06-20  Richard Biener  <rguenther@suse.de>
5676         PR tree-optimization/57584
5677         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
5678         SSA names into the expanded expression that take part in
5679         abnormal coalescing.
5681 2013-06-19  Sharad Singhai  <singhai@google.com>
5683         * gcov.c (print_usage): Handle new option.
5684         (process_args): Ditto.
5685         (get_gcov_intermediate_filename): New function.
5686         (output_intermediate_file): New function.
5687         (output_gcov_file): New function
5688         (generate_results): Handle new option.
5689         (release_function): Relase demangled name.
5690         (read_graph_file): Handle demangled name.
5691         (output_lines): Ditto.
5692         * doc/gcov.texi: Document gcov intermediate format.
5694 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
5696         PR bootstrap/57604
5697         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
5698         (lra_emit_add): Use the functions.  Add comment about Y as an
5699         address segment.
5701 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
5703         PR driver/57652
5704         * collect2.c (collect_atexit): New.
5705         (collect_exit): Delete.
5706         (main): Register collect_atexit with atexit.
5707         (collect_wait): Change collect_exit to exit.
5708         (do_wait): Same.
5709         * collect2.h (collect_exit): Delete.
5710         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
5712 2013-06-19  Wei Mi  <wmi@google.com>
5714         PR rtl-optimization/57518
5715         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
5716         if regno is used in paradoxical subreg.
5717         (update_equiv_regs): Check pdx_subregs[regno] before
5718         set a reg to be equivalent with a mem.
5720 2013-06-19  Matthias Klose  <doko@ubuntu.com>
5722         PR driver/57651
5723         * file-find.h (find_a_file): Add a mode parameter.
5724         * file-find.c (find_a_file): Likewise.
5725         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
5726         with X_OK for the executables.
5727         * collect2.c (main): Call find_a_file with X_OK.
5729 2013-06-19  Steve Ellcey  <sellcey@mips.com>
5731         PR target/56942
5732         * config/mips/mips.md (casesi_internal_mips16_<mode>):
5733         Use NEXT_INSN instead of next_real_insn.
5735 2013-06-19  Jan Hubicka  <jh@suse.cz>
5737         * cgraph.h (const_value_known_p): Replace by ...
5738         (ctor_for_folding): .. this one.
5739         * cgraphunit.c (process_function_and_variable_attributes): Use it.
5740         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
5741         * expr.c (expand_expr_real_1): Likewise.
5742         (string_constant): Likewise.
5743         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
5744         * ipa.c (process_references): Likewise.
5745         (symtab_remove_unreachable_nodes): Likewise.
5746         * ipa-inline-analysis.c (param_change_prob): Likewise.
5747         * gimple-fold.c (canonicalize_constructor_val): Likewise.
5748         (get_base_constructor): Likwise.
5749         * varpool.c (varpool_remove_node): Likewise.
5750         (varpool_remove_initializer): LIkewise.
5751         (dump_varpool_node): LIkwise.
5752         (const_value_known_p): Rewrite to ...
5753         (ctor_for_folding): ... this one.
5755 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
5757         PR driver/57651
5758         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
5759         PERSONALITY in $PATH derived prefixes.
5761 2013-06-19  Jeff Law  <law@redhat.com>
5763         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
5764         in comment.
5766         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
5767         (simplify_bitwise_binary): Use it to simpify certain binary ops on
5768         booleans.
5770 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
5772         * config/arm/vfp.md: Move VFP instruction classification documentation
5773         to ...
5774         * config/arm/arm.md: ... here.  Update instruction classification
5775         documentation.
5777 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
5779         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
5780         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
5781         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
5783 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
5785         PR target/57609
5786         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5787         with NEXT_INSN.  Use tablejump_p to check for jump table data
5788         insns.
5790 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
5792         PR c++/56544
5793         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
5794         that now in C++ the value is correct per the C++ standards.
5796 2013-06-19  Richard Biener  <rguenther@suse.de>
5798         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
5799         for global context.
5801 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5803         Revert:
5804         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5806         PR target/57609
5807         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5808         with next_active_insn.
5810 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
5812         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
5813         functions are inlined during failures to flag an error.
5814         * tree-inline.c (expand_call_inline): Allow the error to be flagged
5815         in early inline pass.
5817 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
5819         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
5820         in comments.
5822 2013-06-18  Julian Brown  <julian@codesourcery.com>
5824         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
5825         Permit virtual register pre-reload if !strict.
5826         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
5827         change.
5828         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
5829         prototype.
5830         * config/arm/neon.md (movmisalign<mode>): Use
5831         neon_perm_struct_or_reg_operand instead of
5832         neon_struct_or_register_operand.
5833         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
5834         neon_permissive_struct_operand instead of neon_struct_operand.
5835         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
5836         neon_vector_mem_operand.
5837         * config/arm/predicates.md (neon_struct_operand): Adjust call to
5838         neon_vector_mem_operand.
5839         (neon_permissive_struct_operand): New.
5840         (neon_struct_or_register_operand): Rename to...
5841         (neon_perm_struct_or_reg_operand): This. Adjust call to
5842         neon_vector_mem_operand.
5844 2013-06-18  Richard Biener  <rguenther@suse.de>
5846         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
5847         * lto-streamer.h: Include pointer-set.h.
5848         (struct lto_decl_slot): Remove.
5849         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
5850         Remove next_index entry.
5851         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5852         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5853         (lto_init_tree_ref_encoder): Adjust.
5854         (lto_destroy_tree_ref_encoder): Likewise.
5855         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5856         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5857         (lto_output_decl_index): Adjust.
5858         (lto_new_out_decl_state): Likewise.
5859         (lto_record_function_out_decl_state): Likewise.
5860         * lto-streamer-out.c (copy_function): Likewise.
5862 2013-06-18  Richard Biener  <rguenther@suse.de>
5864         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
5865         * cgraphunit.c: Include cfgloop.h.
5866         (init_lowered_empty_function): Initialize the loop tree.
5867         (assemble_thunk): Insert new BBs into loops.
5869 2013-06-18  Richard Biener  <rguenther@suse.de>
5871         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
5872         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
5873         the map from cache entry to cache index optional.
5874         (streamer_tree_cache_replace_tree): Adjust accordingly.
5875         (streamer_tree_cache_append): Likewise.
5876         (streamer_tree_cache_delete): Likewise.
5877         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
5878         streamer cache map from cache entry to cache index.
5879         * lto-streamer-out.c (create_output_block): Adjust.
5881 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
5883         * config/arm/arm.md (attribute "insn"): Move multiplication and
5884         division attributes to...
5885         (attribute "type"): ... here.  Remove mult.
5886         (attribute "mul32"): New attribute.
5887         (attribute "mul64"): Add umaal.
5888         (*arm_mulsi3): Update attributes.
5889         (*arm_mulsi3_v6): Likewise.
5890         (*thumb_mulsi3): Likewise.
5891         (*thumb_mulsi3_v6): Likewise.
5892         (*mulsi3_compare0): Likewise.
5893         (*mulsi3_compare0_v6): Likewise.
5894         (*mulsi_compare0_scratch): Likewise.
5895         (*mulsi_compare0_scratch_v6): Likewise.
5896         (*mulsi3addsi): Likewise.
5897         (*mulsi3addsi_v6): Likewise.
5898         (*mulsi3addsi_compare0): Likewise.
5899         (*mulsi3addsi_compare0_v6): Likewise.
5900         (*mulsi3addsi_compare0_scratch): Likewise.
5901         (*mulsi3addsi_compare0_scratch_v6): Likewise.
5902         (*mulsi3subsi): Likewise.
5903         (*mulsidi3adddi): Likewise.
5904         (*mulsi3addsi_v6): Likewise.
5905         (*mulsidi3adddi_v6): Likewise.
5906         (*mulsidi3_nov6): Likewise.
5907         (*mulsidi3_v6): Likewise.
5908         (*umulsidi3_nov6): Likewise.
5909         (*umulsidi3_v6): Likewise.
5910         (*umulsidi3adddi): Likewise.
5911         (*umulsidi3adddi_v6): Likewise.
5912         (*smulsi3_highpart_nov6): Likewise.
5913         (*smulsi3_highpart_v6): Likewise.
5914         (*umulsi3_highpart_nov6): Likewise.
5915         (*umulsi3_highpart_v6): Likewise.
5916         (mulhisi3): Likewise.
5917         (*mulhisi3tb): Likewise.
5918         (*mulhisi3bt): Likewise.
5919         (*mulhisi3tt): Likewise.
5920         (maddhisi4): Likewise.
5921         (*maddhisi4tb): Likewise.
5922         (*maddhisi4tt): Likewise.
5923         (maddhidi4): Likewise.
5924         (*maddhidi4tb): Likewise.
5925         (*maddhidi4tt): Likewise.
5926         (divsi3): Likewise.
5927         (udivsi3): Likewise.
5928         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
5929         (thumb2_mulsi_short_compare0): Likewise.
5930         (thumb2_mulsi_short_compare0_scratch): Likewise.
5931         * config/arm/arm1020e.md (1020mult1): Update attribute change.
5932         (1020mult2): Likewise.
5933         (1020mult3): Likewise.
5934         (1020mult4): Likewise.
5935         (1020mult5): Likewise.
5936         (1020mult6): Likewise.
5937         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
5938         change.
5939         (cortex_a15_mult64): Likewise.
5940         (cortex_a15_sdiv): Likewise.
5941         (cortex_a15_udiv): Likewise.
5942         * config/arm/arm1026ejs.md (mult1): Update attribute change.
5943         (mult2): Likewise.
5944         (mult3): Likewise.
5945         (mult4): Likewise.
5946         (mult5): Likewise.
5947         (mult6): Likewise.
5948         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
5949         (pj4_ir_div): Likewise.
5950         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
5951         (11_mult2): Likewise.
5952         (11_mult3): Likewise.
5953         (11_mult4): Likewise.
5954         (11_mult5): Likewise.
5955         (11_mult6): Likewise.
5956         (11_mult7): Likewise.
5957         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
5958         (cortex_a8_mla): Likewise.
5959         (cortex_a8_mull): Likewise.
5960         (cortex_a8_smulwy): Likewise.
5961         (cortex_a8_smlald): Likewise.
5962         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
5963         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
5964         (cortex_r4_mul_3): Likewise.
5965         (cortex_r4_mla_4): Likewise.
5966         (cortex_r4_mla_3): Likewise.
5967         (cortex_r4_smlald): Likewise.
5968         (cortex_r4_mull): Likewise.
5969         (cortex_r4_sdiv): Likewise.
5970         (cortex_r4_udiv): Likewise.
5971         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
5972         (cortex_a7_idiv): Likewise.
5973         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
5974         (9_mult2): Likewise.
5975         (9_mult3): Likewise.
5976         (9_mult4): Likewise.
5977         (9_mult5): Likewise.
5978         (9_mult6): Likewise.
5979         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
5980         (cortex_a53_sdiv): Likewise.
5981         (cortex_a53_udiv): Likewise.
5982         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5983         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5984         (mp626_mult2): Likewise.
5985         (mp626_mult3): Likewise.
5986         (mp626_mult4): Likewise.
5987         * config/arm/fa526.md (526_mult1): Update attribute change.
5988         (526_mult2): Likewise.
5989         * config/arm/arm-generic.md (mult): Update attribute change.
5990         (mult_ldsched_strongarm): Likewise.
5991         (mult_ldsched): Likewise.
5992         (multi_cycle): Likewise.
5993         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5994         * config/arm/fa606te.md (606te_mult1): Update attribute change.
5995         (606te_mult2): Likewise.
5996         (606te_mult3): Likewise.
5997         (606te_mult4): Likewise.
5998         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5999         (cortex_a9_mac16): Likewise.
6000         (cortex_a9_multiply): Likewise.
6001         (cortex_a9_mac): Likewise.
6002         (cortex_a9_multiply_long): Likewise.
6003         * config/arm/fa626te.md (626te_mult1): Update attribute change.
6004         (626te_mult2): Likewise.
6005         (626te_mult3): Likewise.
6006         (626te_mult4): Likewise.
6008 2013-06-18  Richard Biener  <rguenther@suse.de>
6010         PR lto/57334
6011         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6013 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6015         PR target/57609
6016         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6017         with next_active_insn.
6019 2013-06-18  Alan Modra  <amodra@gmail.com>
6021         * config/rs6000/rs6000.h (enum data_align): New.
6022         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
6023         (DATA_ABI_ALIGNMENT): Define.
6024         (CONSTANT_ALIGNMENT): Correct comment.
6025         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
6026         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
6028 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
6030         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
6031         ATTRIBUTE_UNUSED marking.
6033 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
6035         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
6036         alternative and update.
6037         (aarch64_dup_lanedi): Delete.
6038         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
6039         * config/aarch64/aarch64-simd-builtins.def: Update.
6041 2013-06-17  Richard Biener  <rguenther@suse.de>
6043         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
6044         (lto_input_scc): Declare.
6045         (lto_input_tree_1): Likewise.
6046         (struct lto_stats_d): Add num_tree_bodies_output and
6047         num_pickle_refs_output.
6048         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
6049         (lto_read_tree_1): Split out from ...
6050         (lto_read_tree): ... this.
6051         (lto_input_scc): New function.
6052         (lto_input_tree_1): Split out from ...
6053         (lto_input_tree): ... this.  Handle LTO_tree_scc.
6054         (lto_data_in_create): Create the streamer cache without hashes.
6055         * lto-streamer-out.c (create_output_block): Create the streamer
6056         cache with hashes when not doing WPA.
6057         (lto_write_tree_1): Split out from ...
6058         (lto_write_tree): ... this.
6059         (get_symbol_initial_value): New function.
6060         (lto_output_tree_1): Split out from ...
6061         (lto_output_tree): ... this.  Write trees as series of SCCs
6062         using a DFS walk via DFS_write_tree.
6063         (struct sccs, struct scc_entry): New types.
6064         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6065         (DFS_write_tree_body): New function.
6066         (DFS_write_tree): Likewise.
6067         (hash_tree): Likewise.
6068         (scc_entry_compare): Likewise.
6069         (hash_scc): Likewise.
6070         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6071         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6072         TREE_CHAIN as regular reference.
6073         (streamer_read_integer_cst): Remove.
6074         (streamer_get_pickled_tree): Adjust.
6075         * tree-streamer-out.c (streamer_write_chain): Disable streaming
6076         of DECL_EXTERNALs in BLOCK_VARS for now.
6077         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6078         reference.
6079         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6080         Add hash value argument and record that if hashes are recorded
6081         in the cache.
6082         (streamer_tree_cache_insert_1): Adjust.
6083         (streamer_tree_cache_insert): Likewise.
6084         (streamer_tree_cache_insert_at): Rename to ...
6085         (streamer_tree_cache_replace_tree): ... this and adjust.
6086         (streamer_tree_cache_append): Adjust.
6087         (record_common_node): Likewise.
6088         (streamer_tree_cache_create): Add argument whether to
6089         record hash values together with trees.
6090         (streamer_tree_cache_delete): Adjust.
6091         * tree-streamer.h (struct streamer_tree_cache_d): Add
6092         vector of hashes.
6093         (streamer_read_integer_cst): Remove.
6094         (streamer_tree_cache_insert): Adjust.
6095         (streamer_tree_cache_append): Likewise.
6096         (streamer_tree_cache_insert_at): Rename to ...
6097         (streamer_tree_cache_replace_tree): ... this and adjust.
6098         (streamer_tree_cache_create): Add argument whether to record hashes.
6099         (streamer_tree_cache_get): Rename to ...
6100         (streamer_tree_cache_get_tree): ... this.
6101         (streamer_tree_cache_get_hash): New function.
6102         * tree.c (cache_integer_cst): New function.
6103         * tree.h (cache_integer_cst): Declare.
6104         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6105         * lto-symtab.c (lto_varpool_replace_node): Only release
6106         DECL_INITIAL of non-prevailing decls.
6107         * varpool.c (varpool_remove_initializer): Do not release
6108         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6110 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6112         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6113         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6114         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6115         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6116         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6117         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6118         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6119         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6120         instead of TARGET_64BIT.
6121         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6122         Require ISA_HAS_<D>DIV.
6124 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
6126         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6127         (mips*-*-linux*): Move default with_llsc setting to where other
6128         defaults are set.
6129         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6130         with_arch block.
6131         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6132         Likewise.  Remove default with_tune setting.  Move default float
6133         setting to its own block.  Handle with_llsc in the same block as above.
6135 2013-06-16  Joern Rennecke <joern.rennecke@embecosm.com>
6137         PR rtl-optimization/57425
6138         PR rtl-optimization/57569
6139         * alias.c (write_dependence_p): Add new parameters mem_mode,
6140         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
6141         Changed all callers.
6142         (canon_anti_dependence): New function.
6143         * cse.c (check_dependence): Use canon_anti_dependence.
6144         * cselib.c (cselib_invalidate_mem): Likewise.
6145         * rtl.h (canon_anti_dependence): Declare.
6147 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6149         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6150         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6151         ".set mips3" for 64-bit targets.
6153 2013-06-15  Dehao Chen  <dehao@google.com>
6155         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6156         * gimple-low.c (gimple_check_call_matching_types): Likewise.
6157         (gimple_check_call_args): Likewise.
6158         * value-prof.c (check_ic_target): Likewise.
6159         * ipa-inline.c (early_inliner): Likewise.
6160         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6161         * cgraph.c (cgraph_create_edge_1): Likewise.
6162         (cgraph_make_edge_direct): Likewise.
6164 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
6166         PR target/57615
6167         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6168         rs6000_output_move_128bit to handle emitting quad memory
6169         operations.  Set attribute length to 8 bytes.
6171 2013-06-14  Vidya Praveen <vidyapraveen@arm.com>
6173         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6174         New pattern.
6175         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6176         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6177         (aarch64_<su>mlsl<mode>): Likewise.
6179 2013-06-14  Mike Stump  <mikestump@comcast.net>
6181         * Makefile.in (TARGET_H): Add insn-codes.h.
6183 2013-06-14  Alan Modra  <amodra@gmail.com>
6185         PR middle-end/57134
6186         PR middle-end/57586
6187         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6188         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
6189         bitfield expansion when EXPAND_MEMORY.
6190         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6192 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
6194         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6195         test for clearing quad memory on 32-bit later.
6197 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6199         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6200         (fold_negate_expr): Likewise.
6201         (fold_real_zero_addition_p): Handle vectors.
6202         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6204 2013-06-14  Alan Modra  <amodra@gmail.com>
6206         * varasm.c (force_const_mem): Revert 2013-06-07 change.
6208 2013-06-13  Jan Hubicka  <jh@suse.cz>
6210         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6211         Local comdats are not externally visible.
6212         * symtab.c (dump_symtab_base): Dump externally visible.
6213         (verify_symtab_base): Verify back links in the symtab hash.
6215 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6217         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6218         CONVERT_EXPR as equal nodes.
6220 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6222         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6224 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6226         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6227         Generalize to complex and vector.
6228         * tree.c (build_all_ones_cst): New function.
6229         * tree.h (build_all_ones_cst): Declare it.
6231 2013-06-13  Alan Modra  <amodra@gmail.com>
6233         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6234         * config/rs6000/rs6000.md (signbittf2): New insn.
6235         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6236         (abstf2_internal, cmptf_internal2): Likewise.
6237         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6239 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
6240             Pat Haugen <pthaugen@us.ibm.com>
6241             Peter Bergner <bergner@vnet.ibm.com>
6243         * config/rs6000/rs6000.c (emit_load_locked): Add support for
6244         power8 byte, half-word, and quad-word atomic instructions.
6245         (emit_store_conditional): Likewise.
6246         (rs6000_expand_atomic_compare_and_swap): Likewise.
6247         (rs6000_expand_atomic_op): Likewise.
6249         * config/rs6000/sync.md (larx): Add new modes for power8.
6250         (stcx): Likewise.
6251         (AINT): New mode iterator to include TImode as well as normal
6252         integer modes on power8.
6253         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
6254         that VSX registers are not considered.  Use AINT mode iterator
6255         instead of INT1 to allow inclusion of quad word atomic operations
6256         on power8.
6257         (load_locked<mode>): Likewise.
6258         (store_conditional<mode>): Likewise.
6259         (atomic_compare_and_swap<mode>): Likewise.
6260         (atomic_exchange<mode>): Likewise.
6261         (atomic_nand<mode>): Likewise.
6262         (atomic_fetch_<fetchop_name><mode>): Likewise.
6263         (atomic_nand_fetch<mode>): Likewise.
6264         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
6265         each type.
6266         (ATOMIC): On power8, add QImode, HImode modes.
6267         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
6268         modes that promote to SImode.
6269         (load_lockedti): Convert TImode arguments to PTImode, so that we
6270         get a guaranteed even/odd register pair.
6271         (load_lockedpti): Likewise.
6272         (store_conditionalti): Likewise.
6273         (store_conditionalpti): Likewise.
6275         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
6276         atomic load/store instructions.
6277         (HSI): Likewise.
6279 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
6281         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
6282         loads.
6283         (insn_count): New attribute, with most cases extracted from...
6284         (length): ...here.  Redefine most cases in terms of insn_count.
6285         (single_insn): Delete.
6286         (can_delay): Use insn_count to check for single instructions.
6287         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
6288         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
6289         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
6290         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
6291         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
6292         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
6293         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
6294         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
6295         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
6296         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
6297         rather than "length".
6298         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
6299         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
6300         Use "insn_count" rather than "length".
6301         * config/mips/mips-dsp.md
6302         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
6303         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
6304         length attributes.
6306 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
6308         PR tree-optimization/57361
6309         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
6311 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
6313         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
6314         to split.
6315         (aarch64_simd_combine<mode>): New instruction expansion.
6316         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
6317         function prototype.
6318         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
6319         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
6321 2013-06-12  Jan Hubicka  <jh@suse.cz>
6323         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
6324         decl has when in streaming stage.
6325         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
6326         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
6328 2013-06-12  Roland Stigge <stigge@antcom.de>
6330         PR target/57578
6331         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
6333 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6335         PR tree-optimization/57537
6336         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
6337         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
6339 2013-06-12  Richard Biener  <rguenther@suse.de>
6341         * data-streamer.h (streamer_write_char_stream): CSE
6342         obs->current_pointer.
6343         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
6344         streamer_write_char_stream manually and optimize the resulting loop.
6345         (streamer_write_hwi_stream): Likewise.
6347 2013-06-12  Jan Hubicka  <jh@suse.cz>
6349         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
6350         * cgraph.h (varpool_create_empty_node): Declare.
6351         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
6352         duplicated nodes.
6353         * symtab.c (symtab_unregister_node): Be lax about missin entries
6354         in node hash.
6355         (symtab_get_node): Update comment.
6356         * varpool.c (varpool_create_empty_node): Break out from ...
6357         (varpool_node_for_decl): ... here.
6358         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
6360 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
6362         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
6363         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
6364         part.  Use straight-line flow at the end.
6365         <COMPONENT_REF>: Remove superfluous else.
6366         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
6368 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6370         PR target/56564
6371         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
6372         target hook even for !TREE_PUBLIC decls.  If no resolution info
6373         is available, return false for common and external decls.
6375 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6377         * config/rl78/constraints.md (U): New constraint.
6378         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
6379         valloc attribute.
6381 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
6383         PR target/57589
6384         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
6385         to allow returning address to AT_PLATFORM name.
6387 2013-06-11  Jan Hubicka  <jh@suse.cz>
6389         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
6390         * cgraph.h (symtab_node_base): Add weakref flag.
6391         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
6392         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
6393         (output_weakrefs): Use weakref flag.
6394         * fold-const.c (simple_operand_p): Handle WEAK.
6395         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
6396         * ipa.c (varpool_externally_visible_p): Drop weakref.
6397         (function_and_variable_visibility): Update comment; fix weakref
6398         sanity checks; do not clear DECL_WEAK on them.
6399         * lto-cgraph.c (lto_output_node): update.
6400         (lto_output_varpool_node): Update.
6401         (input_overwrite_node): Update.
6402         (input_node): Update.
6403         (input_varpool_node): Update.
6404         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
6405         (lto_symtab_merge_symbols): Add sanity check.
6406         (lto_symtab_prevailing_decl): Do not special case weakrefs.
6407         * passes.c (rest_of_decl_compilation): Set static flag, too.
6408         * symtab.c (dump_symtab_base): Dump weakref.
6409         (verify_symtab_base): Sanity check weakrefs.
6410         (symtab_make_decl_local): Remove duplicated code.
6411         (symtab_alias_ultimate_target): Simplify.
6412         * varpool.c (varpool_create_variable_alias): Set weakref flag.
6414 2013-06-11  DJ Delorie  <dj@redhat.com>
6416         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
6417         (rl78_unwind_word_mode): New.
6419 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
6421         * final.c (debug_prefix_maps): Make static.
6423 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
6425         * function.c (initial_trampoline): Remove stray copy.
6427 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
6429         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
6431 2013-06-11  Martin Jambor  <mjambor@suse.cz>
6433         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
6434         within bounds at the beginning of the function.
6436 2013-06-11  Alan Modra  <amodra@gmail.com>
6438         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
6439         reporting.
6440         (get_named_section): Don't NULL !DECL_P decl.
6442 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
6444         * doc/invoke.texi (core-avx2): Document.
6445         (slm): Likewise.
6446         (atom): Updated with MOVBE.
6448 2013-06-11  Richard Biener  <rguenther@suse.de>
6450         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
6452 2013-06-11  Anton Blanchard  <anton@samba.org>
6454         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
6455         correct shift value in little-endian mode.
6457 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
6459         PR target/56564
6460         * varasm.c (get_variable_align): Move #endif to the right place.
6462 2013-06-10  Cary Coutant  <ccoutant@google.com>
6464         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
6465         for hash so that hash table traversal order is deterministic.
6467 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
6468             Pat Haugen <pthaugen@us.ibm.com>
6469             Peter Bergner <bergner@vnet.ibm.com>
6471         * config/rs6000/vector.md (GPR move splitter): Do not split moves
6472         of vectors in GPRS if they are direct moves or quad word load or
6473         store moves.
6475         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
6476         declaration.
6477         (direct_move_p): Likewise.
6478         (quad_load_store_p): Likewise.
6480         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
6481         classes into bins based on the physical register type.
6482         (reg_class_to_reg_type): Likewise.
6483         (IS_STD_REG_TYPE): Likewise.
6484         (IS_FP_VECT_REG_TYPE): Likewise.
6485         (reload_fpr_gpr): Arrays to determine what insn to use if we can
6486         use direct move instructions.
6487         (reload_gpr_vsx): Likewise.
6488         (reload_vsx_gpr): Likewise.
6489         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
6490         information that is a simplification of register classes.  Also
6491         precalculate direct move reload helpers.
6492         (direct_move_p): New function to return true if the operation can
6493         be done as a direct move instruciton.
6494         (quad_load_store_p): New function to return true if the operation
6495         is a quad memory operation.
6496         (rs6000_legitimize_address): If quad memory, only allow register
6497         indirect for TImode addresses.
6498         (rs6000_legitimate_address_p): Likewise.
6499         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
6500         (rs6000_reload_register_type): Likewise.
6501         (register_to_reg_type): Return register type.
6502         (rs6000_secondary_reload_simple_move): New helper function for
6503         secondary reload and secondary memory needed to identify anything
6504         that is a simple move, and does not need reloading.
6505         (rs6000_secondary_reload_direct_move): New helper function for
6506         secondary reload to identify cases that can be done with several
6507         instructions via the direct move instructions.
6508         (rs6000_secondary_reload_move): New helper function for secondary
6509         reload to identify moves between register types that can be done.
6510         (rs6000_secondary_reload): Add support for quad memory operations
6511         and for direct move.
6512         (rs6000_secondary_memory_needed): Likewise.
6513         (rs6000_debug_secondary_memory_needed): Change argument names.
6514         (rs6000_output_move_128bit): New function to return the move to
6515         use for 128-bit moves, including knowing about the various
6516         limitations of quad memory operations.
6518         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
6519         memory operations.  call rs6000_output_move_128bit for the actual
6520         instruciton(s) to generate.
6521         (vsx_movti_64bit): Likewise.
6523         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
6524         (UNSPEC_P8V_MTVSRWZ): Likewise.
6525         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
6526         (UNSPEC_P8V_MTVSRD): Likewise.
6527         (UNSPEC_P8V_XXPERMDI): Likewise.
6528         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
6529         (UNSPEC_FUSION_GPR): Likewise.
6530         (FMOVE128_GPR): New iterator for direct move.
6531         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
6532         (f32_sv): Likewise.
6533         (f32_dm): Likewise.
6534         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
6535         loads and direct move instructions.
6536         (zero_extendsidi2_lfiwzx): Likewise.
6537         (extendsidi2_lfiwax): Likewise.
6538         (extendsidi2_nocell): Likewise.
6539         (floatsi<mode>2_lfiwax): Likewise.
6540         (lfiwax): Likewise.
6541         (floatunssi<mode>2_lfiwzx): Likewise.
6542         (lfiwzx): Likewise.
6543         (fix_trunc<mode>_stfiwx): Likewise.
6544         (fixuns_trunc<mode>_stfiwx): Likewise.
6545         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
6546         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
6547         (parity<mode>2_cmpb): Set length/type attr.
6548         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
6549         for 'mr.' to fast_compare.
6550         (bpermd_<mode>): Change type attr to popcnt.
6551         (p8_fmrgow_<mode>): New insns for power8 direct move support.
6552         (p8_mtvsrwz_1): Likewise.
6553         (p8_mtvsrwz_2): Likewise.
6554         (reload_fpr_from_gpr<mode>): Likewise.
6555         (p8_mtvsrd_1): Likewise.
6556         (p8_mtvsrd_2): Likewise.
6557         (p8_xxpermdi_<mode>): Likewise.
6558         (reload_vsx_from_gpr<mode>): Likewise.
6559         (reload_vsx_from_gprsf): Likewise.
6560         (p8_mfvsrd_3_<mode>): LIkewise.
6561         (reload_gpr_from_vsx<mode>): Likewise.
6562         (reload_gpr_from_vsxsf): Likewise.
6563         (p8_mfvsrd_4_disf): Likewise.
6564         (multi-word GPR splits): Do not split direct moves or quad memory
6565         operations.
6567 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
6569         * tree-into-ssa.c (interesting_blocks): Make static.
6571 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
6573         PR target/56564
6574         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
6575         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
6576         Use DATA_ABI_ALIGNMENT for that case instead if defined.
6577         (get_variable_align): New function.
6578         (get_variable_section, emit_bss, emit_common,
6579         assemble_variable_contents, place_block_symbol): Use
6580         get_variable_align instead of DECL_ALIGN.
6581         (assemble_noswitch_variable): Add align argument, use it
6582         instead of DECL_ALIGN.
6583         (assemble_variable): Adjust caller.  Use get_variable_align
6584         instead of DECL_ALIGN.
6585         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
6586         caller.
6587         (DATA_ABI_ALIGNMENT): Define.
6588         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
6589         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
6590         opt is false, only return the psABI mandated alignment increase.
6591         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
6592         (DATA_ABI_ALIGNMENT): ... this.
6593         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
6594         (DATA_ABI_ALIGNMENT): ... this.
6595         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
6596         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
6597         (DATA_ABI_ALIGNMENT): ... this.
6598         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
6599         * doc/tm.texi: Regenerated.
6601 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
6603         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
6604         cmp_code to construct REG_EQUAL note.
6606 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
6608         PR target/57568
6609         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
6610         that operands[2] doesn't overlap with operands[0].
6612 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
6613             Jan Hubicka  <jh@suse.cz>
6615         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
6616         hack to mark symbols as used.
6618 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
6620         PR rtl-optimization/57559
6621         * lra-constraints.c (process_alt_operands): Don't discourage
6622         memory with known offset for offsetable memory constraint.
6623         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
6625 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
6627         * varasm.c (struct oc_local_state): Reorder fields.
6628         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
6629         and adjust accordingly.
6630         (output_constructor): Reorder initialization code and adjust call to
6631         output_constructor_bitfield.
6633 2013-06-07  Jan Hubicka  <jh@suse.cz>
6635         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
6637 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
6639         * tree-object-size.c (unknown): Make const.
6641 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6643         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
6644         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
6645         for last alternative in the cpu_facility attribute.
6647 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6649         PR target/56315
6650         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
6651         (xordi3): Change operand 2 constraint to arm_xordi_operand.
6652         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
6653         * config/arm/constraints.md (Dg): New constraint.
6654         * config/arm/neon.md (xordi3_neon): Remove.
6655         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
6656         * config/arm/predicates.md (arm_xordi_operand): New predicate.
6658 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6660         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
6661         Clean up alternatives.
6663 2013-06-07  Alan Modra  <amodra@gmail.com>
6665         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
6666         va_list_gpr_size.
6668 2013-06-07  Alan Modra  <amodra@gmail.com>
6670         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
6672 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6674         * config/arm/constraints.md (Df): New constraint.
6675         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
6676         Correct length attribute for last two alternatives.
6678 2013-06-07  Alan Modra  <amodra@gmail.com>
6680         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6681         override user -mfp-in-toc.
6682         (offsettable_ok_by_alignment): Consider just the current access
6683         rather than the whole object, unless BLKmode.  Handle
6684         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
6685         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
6686         for -mcmodel=medium.
6687         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
6688         override user -mfp-in-toc or -msum-in-toc.  Default to
6689         -mno-fp-in-toc for -mcmodel=medium.
6691 2013-06-06  DJ Delorie  <dj@redhat.com>
6693         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
6694         TARGET_VALID_POINTER_MODE.
6696 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
6697             Pat Haugen <pthaugen@us.ibm.com>
6698             Peter Bergner <bergner@vnet.ibm.com>
6700         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6701         Document new power8 builtins.
6703         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
6704         condition code register, to allow 128-bit logical operations to be
6705         done in the VSX or GPR registers.
6706         (nor<mode>3): Use the canonical form for nor.
6707         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
6708         vclz*, and vpopcnt* vector instructions.
6709         (nand<mode>3): Likewise.
6710         (orc<mode>3): Likewise.
6711         (clz<mode>2): LIkewise.
6712         (popcount<mode>2): Likewise.
6714         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
6715         that only the GPRs are recognized.
6717         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6718         support for new power8 builtins.
6720         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
6721         builtin functions.
6722         (xscvdpspn): Likewise.
6723         (vclz): Likewise.
6724         (vclzb): Likewise.
6725         (vclzh): Likewise.
6726         (vclzw): Likewise.
6727         (vclzd): Likewise.
6728         (vpopcnt): Likewise.
6729         (vpopcntb): Likewise.
6730         (vpopcnth): Likewise.
6731         (vpopcntw): Likewise.
6732         (vpopcntd): Likewise.
6733         (vgbbd): Likewise.
6734         (vmrgew): Likewise.
6735         (vmrgow): Likewise.
6736         (eqv): Likewise.
6737         (eqv_v16qi3): Likewise.
6738         (eqv_v8hi3): Likewise.
6739         (eqv_v4si3): Likewise.
6740         (eqv_v2di3): Likewise.
6741         (eqv_v4sf3): Likewise.
6742         (eqv_v2df3): Likewise.
6743         (nand): Likewise.
6744         (nand_v16qi3): Likewise.
6745         (nand_v8hi3): Likewise.
6746         (nand_v4si3): Likewise.
6747         (nand_v2di3): Likewise.
6748         (nand_v4sf3): Likewise.
6749         (nand_v2df3): Likewise.
6750         (orc): Likewise.
6751         (orc_v16qi3): Likewise.
6752         (orc_v8hi3): Likewise.
6753         (orc_v4si3): Likewise.
6754         (orc_v2di3): Likewise.
6755         (orc_v4sf3): Likewise.
6756         (orc_v2df3): Likewise.
6758         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
6759         allow power8 quad mode in 64-bit.
6760         (rs6000_builtin_vectorized_function): Add support to vectorize
6761         ISA 2.07 count leading zeros, population count builtins.
6762         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
6763         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
6764         (builtin_function_type): Add vgbbd builtin function which takes an
6765         unsigned argument.
6766         (altivec_expand_vec_perm_const): Add support for new power8 merge
6767         instructions.
6769         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
6770         that does not include TImdoe for use with 32-bit.
6771         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
6772         instructions.
6773         (UNSPEC_VSX_CVDPSPN): Likewise.
6774         (vsx_xscvdpspn): Likewise.
6775         (vsx_xscvspdpn): Likewise.
6776         (vsx_xscvdpspn_scalar): Likewise.
6777         (vsx_xscvspdpn_directmove): Likewise.
6778         (vsx_and<mode>3): Split logical operations into 32-bit and
6779         64-bit. Add support to do logical operations on TImode as well as
6780         VSX vector types.  Allow logical operations to be done in either
6781         VSX registers or in general purpose registers in 64-bit mode.  Add
6782         splitters if GPRs were used. For AND, add clobber of CCmode to
6783         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
6784         encoding.
6785         (vsx_and<mode>3_32bit): Likewise.
6786         (vsx_and<mode>3_64bit): Likewise.
6787         (vsx_ior<mode>3): Likewise.
6788         (vsx_ior<mode>3_32bit): Likewise.
6789         (vsx_ior<mode>3_64bit): Likewise.
6790         (vsx_xor<mode>3): Likewise.
6791         (vsx_xor<mode>3_32bit): Likewise.
6792         (vsx_xor<mode>3_64bit): Likewise.
6793         (vsx_one_cmpl<mode>2): Likewise.
6794         (vsx_one_cmpl<mode>2_32bit): Likewise.
6795         (vsx_one_cmpl<mode>2_64bit): Likewise.
6796         (vsx_nor<mode>3): Likewise.
6797         (vsx_nor<mode>3_32bit): Likewise.
6798         (vsx_nor<mode>3_64bit): Likewise.
6799         (vsx_andc<mode>3): Likewise.
6800         (vsx_andc<mode>3_32bit): Likewise.
6801         (vsx_andc<mode>3_64bit): Likewise.
6802         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
6803         and xxlorc instructions.
6804         (vsx_eqv<mode>3_64bit): Likewise.
6805         (vsx_nand<mode>3_32bit): Likewise.
6806         (vsx_nand<mode>3_64bit): Likewise.
6807         (vsx_orc<mode>3_32bit): Likewise.
6808         (vsx_orc<mode>3_64bit): Likewise.
6810         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
6812         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
6813         instruction.
6814         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
6815         (p8_vmrgow): Likewise.
6816         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
6817         GPRs to be split under VSX.
6818         (p8v_clz<mode>2): Add power8 count leading zero support.
6819         (p8v_popcount<mode>2): Add power8 population count support.
6820         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
6821         support.
6823         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
6824         instruction.
6826         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
6827         builtin functions.
6828         (vec_nand): Likewise.
6829         (vec_vclz): Likewise.
6830         (vec_vclzb): Likewise.
6831         (vec_vclzd): Likewise.
6832         (vec_vclzh): Likewise.
6833         (vec_vclzw): Likewise.
6834         (vec_vgbbd): Likewise.
6835         (vec_vmrgew): Likewise.
6836         (vec_vmrgow): Likewise.
6837         (vec_vpopcnt): Likewise.
6838         (vec_vpopcntb): Likewise.
6839         (vec_vpopcntd): Likewise.
6840         (vec_vpopcnth): Likewise.
6841         (vec_vpopcntw): Likewise.
6843 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6845         PR rtl-optimization/57468
6846         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
6847         spilled pseudos.
6849 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6851         PR rtl-optimization/57459
6852         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
6853         type when setting live regs.
6855 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
6857         * config/s390/s390.opt (mlra): New option.
6858         * config/s390/s390.c (s390_decompose_address): Check displacement
6859         for all registers for LRA.
6860         (s390_secondary_reload): Don't used secondary reloads for LRA.
6861         (s390_lra_p): New function.
6862         (TARGET_LRA_P): Define.
6863         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
6864         of attribute cpu_facility to zarch for the last alternative.
6865         (*cmpmem_short): Ditto.
6867 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
6869         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
6870         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
6871         (arm_expand_prologue): Likewise.
6873 2013-06-06  Teresa Johnson  <tejohnson@google.com>
6875         PR c++/53743
6876         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
6877         as this is now done by redirect_edge_and_branch_force.
6878         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
6879         barriers, and fix interaction with splitting.
6880         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
6881         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
6882         reflect changes made in the routine.
6883         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
6884         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
6885         since this is called in cfglayout mode, and replace partition fixup
6886         with assert as that is now done by force_nonfallthru_and_redirect.
6887         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
6888         already be marked with region crossing note.
6889         (insert_section_boundary_note): Make non-static, gate on flag
6890         has_bb_partition, rewrite to also check for multiple partitions.
6891         (rest_of_handle_reorder_blocks): Remove call to
6892         insert_section_boundary_note, now done later during free_cfg.
6893         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
6894         * bb-reorder.h (insert_section_boundary_note): Declare.
6895         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
6896         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
6897         invoke insert_section_boundary_note.
6898         (try_redirect_by_replacing_jump): Remove unnecessary
6899         check for region crossing note.
6900         (fixup_partition_crossing): New function.
6901         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
6902         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
6903         in non-cfglayout mode.
6904         (force_nonfallthru_and_redirect): Fixup partition boundaries,
6905         remove old code that tried to do this. Emit barrier correctly
6906         when we are in cfglayout mode.
6907         (last_bb_in_partition): New function.
6908         (rtl_split_edge): Correctly fixup partition boundaries.
6909         (commit_one_edge_insertion): Remove old code that tried to
6910         fixup region crossing edge since this is now handled in
6911         split_block, and set up insertion point correctly since
6912         block may now end in a jump.
6913         (verify_hot_cold_block_grouping): Guard against checking when not in
6914         linearized RTL mode.
6915         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
6916         notes.
6917         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
6918         rtl_verify_flow_info, so not called in cfglayout mode.
6919         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
6920         (fixup_reorder_chain): Remove old code that attempted to fixup region
6921         crossing note as this is now handled in force_nonfallthru_and_redirect.
6922         (duplicate_insn_chain): Don't duplicate switch section notes.
6923         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
6924         note.
6925         * basic-block.h (emit_barrier_after_bb): Declare.
6927 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6929         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
6930         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
6931         arm_usatsihi): Adjust alternatives for arm_restrict_it.
6933 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6935         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
6936         where appropriate.
6937         * config/arm/ldmstm.md: Regenerate.
6939 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6941         * config/arm/sync.md (atomic_loaddi_1):
6942         Disable predication for arm_restrict_it.
6943         (arm_load_exclusive<mode>): Likewise.
6944         (arm_load_exclusivesi): Likewise.
6945         (arm_load_exclusivedi): Likewise.
6946         (arm_load_acquire_exclusive<mode>): Likewise.
6947         (arm_load_acquire_exclusivesi): Likewise.
6948         (arm_load_acquire_exclusivedi): Likewise.
6949         (arm_store_exclusive<mode>): Likewise.
6950         (arm_store_exclusive<mode>): Likewise.
6951         (arm_store_release_exclusivedi): Likewise.
6952         (arm_store_release_exclusive<mode>): Likewise.
6954 2013-06-06  Richard Biener  <rguenther@suse.de>
6956         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
6957         after LTO_null.
6958         (lto_tag_is_tree_code_p): Adjust.
6959         (lto_tag_is_gimple_code_p): Likewise.
6960         (lto_gimple_code_to_tag): Likewise.
6961         (lto_tag_to_gimple_code): Likewise.
6962         (lto_tree_code_to_tag): Likewise.
6963         (lto_tag_to_tree_code): Likewise.
6964         * data-streamer.h (streamer_write_hwi_in_range): Use
6965         uhwi streaming to stream the normalized range.
6966         (streamer_read_hwi_in_range): Likewise.
6968 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6970         * config/arm/arm.md (enabled_for_depr_it): New attribute.
6971         (predicable_short_it): Likewise.
6972         (predicated): Likewise.
6973         (enabled): Handle above.
6974         (define_cond_exec): Set predicated attribute to yes.
6976 2013-06-05  Mike Stump  <mikestump@comcast.net>
6978         * gdbinit.in (__FUNCTION__): Add.
6980 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
6982         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6983         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6985 2013-06-05  Jan Hubicka  <jh@suse.cz>
6987         * varasm.c (mark_decl_referenced): Revert the removal until targets
6988         are fixed.
6990 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
6992         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6993         instead of mark_decl_referenced.
6995 2013-06-05  Jan Hubicka  <jh@suse.cz>
6997         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6998         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6999         and symtab_used_from_object_file_p.
7000         (cgraph_make_node_local_1): Clear forced_by_abi.
7001         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
7002         * cgraph.h (symtab_node_base): Add forced_by_abi.
7003         (decide_is_variable_needed): Remove.
7004         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
7005         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
7006         (decide_is_symbol_needed): ... this one; handle symbols in general;
7007         always analyze virtuals; honnor forced_by_abi.
7008         (cgraph_finalize_function): Update.
7009         (varpool_finalize_decl): Update.
7010         (symbol_defined_and_needed): Remove.
7011         (analyze_functions): Update.
7012         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
7013         output_refs, input_overwrite_node): Handle forced_by_abi.
7014         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
7015         (address_taken_from_non_vtable_p): ... this one.
7016         (comdat_can_be_unshared_p_1): New function.
7017         (cgraph_comdat_can_be_unshared_p): Rename to ...
7018         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
7019         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
7020         (function_and_variable_visibility): Clear forced_by_abi as needed.
7021         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
7022         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
7023         * symtab.c (dump_symtab_base): Dump forced_by_abi.
7024         * varpool.c (decide_is_variable_needed): Remove.
7026 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7028         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
7029         (arm_option_override): Override arm_restrict_it where appropriate.
7030         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
7031         * config/arm/arm.opt (mrestrict-it): New command-line option.
7032         * doc/invoke.texi: Document -mrestrict-it.
7034 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7036         * tsan.c (tsan_atomic_table): Make const.
7038 2013-06-05  Richard Biener  <rguenther@suse.de>
7040         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
7041         index associated with the tree we are supposed to replace.
7042         * tree-streamer-out.c (pack_ts_base_value_fields): Output
7043         TREE_ASM_WRITTEN as zero for everything but SSA names.
7045 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7047         * tree-ssa-structalias.c (call_stmt_vars): Make static.
7049 2013-06-04  Jan Hubicka  <jh@suse.cz>
7051         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
7052         (input_node, input_varpool_node): Handle correctly external same
7053         body aliases.
7054         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7055         nodes at ltrans stage.
7057 2013-06-04  Jan Hubicka  <jh@suse.cz>
7059         * ipa-inline.c (update_caller_keys): Fix availability test.
7060         (update_callee_keys): Likewise.
7061         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7062         to follow ELF standard.
7064 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
7066         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7067         (mips64r5900el-*-elf*): New configurations.
7068         * config/mips/mips-cpus.def (r5900): New processor.
7069         * config/mips/mips-tables.opt: Regenerate.
7070         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7071         (mips_issue_rate): Handle PROCESSOR_R5900.
7072         (mips_reorg_process_insns): Force reorder mode for the R5900.
7073         * config/mips/mips.h (TARGET_MIPS5900): Define.
7074         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7075         TARGET_MIPS5900.
7076         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7077         TARGET_MIPS5900.
7078         * config/mips/mips.md (processor): Add r5900.
7079         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7081 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7083         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7084         into function to generate MOVI instruction.
7085         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7086         (aarch64_preferred_simd_mode): Turn into wrapper.
7087         (aarch64_output_scalar_simd_mov_immediate): New function.
7088         * config/aarch64/aarch64-protos.h: Add prototype for above.
7090 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7092         * config/aarch64/aarch64.c (simd_immediate_info): Remove
7093         element_char member.
7094         (sizetochar): Return signed char.
7095         (aarch64_simd_valid_immediate): Remove elchar and other
7096         unnecessary variables.
7097         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7098         Calculate element_char as required.
7099         * config/aarch64/aarch64-protos.h: Update and move prototype
7100         for aarch64_output_simd_mov_immediate.
7101         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7102         Update arguments.
7104 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7106         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7107         information completed by aarch64_simd_valid_immediate.
7108         (aarch64_legitimate_constant_p): Update arguments.
7109         (aarch64_simd_valid_immediate): Work with struct rather than many
7110         pointers.
7111         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7112         (aarch64_simd_make_constant): Update arguments.
7113         (aarch64_output_simd_mov_immediate): Work with struct rather than
7114         many pointers.  Output immediate directly rather than as operand.
7115         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7116         Update prototype.
7117         * config/aarch64/constraints.md (Dn): Update arguments.
7119 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7121         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7122         longer static.
7123         (aarch64_simd_immediate_valid_for_move): Remove.
7124         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7125         (aarch64_simd_make_constant): Update call.
7126         (aarch64_output_simd_mov_immediate): Update call.
7127         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7128         Add prototype.
7129         * config/aarch64/constraints.md (Dn): Update call.
7131 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7133         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7134         return type to bool for prototype.
7135         (aarch64_legitimate_constant_p): Check for true instead of not -1.
7136         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7137         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7139 2013-06-04  Catherine Moore  <clm@codesourcery.com>
7141         * config/mips/mips.opt (meva): New.
7142         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7143         (ASM_SPEC): Handle -meva.
7144         * doc/invoke.texi (meva):  Document.
7146 2013-06-04  Alan Modra  <amodra@gmail.com>
7148         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7149         constant output.
7151 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7153         * rtl.def: Add extra fourth optional field to define_cond_exec.
7154         * gensupport.c (process_one_cond_exec): Process attributes from
7155         define_cond_exec.
7156         * doc/md.texi: Document fourth field in define_cond_exec.
7158 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
7160         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7161         out the processing order as in store_bit_field_1.
7163 2013-06-04  Jan Hubicka  <jh@suse.cz>
7165         PR middle-end/57500
7166         * cgraphunit.c (cgraph_process_same_body_aliases): Create
7167         non-VAR_DECL node if it does not exist yet.
7169 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
7171         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7172         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7173         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7174         target_cpu_default setting.
7176 2013-06-03  Teresa Johnson  <tejohnson@google.com>
7178         * dumpfile.c (opt_info_switch_p): Change -fopt-info
7179         default to -fopt-info=optimized instead of all.
7180         * doc/invoke.texi: Ditto.
7181         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7182         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7183         (execute_vect_slp): Emit BB vectorization success under
7184         MSG_OPTIMIZED_LOCATIONS.
7185         * tree-vect-slp.c (vect_slp_transform_bb): Change
7186         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7187         * tree-vect-loop.c (vect_transform_loop): Ditto.
7189 2013-06-03  Jason Merrill  <jason@redhat.com>
7191         PR c++/57415
7192         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7193         Use TARGET_EXPR for C++.
7195 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
7197         PR rtl-optimization/57268
7198         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7199         if DEBUG_INSN_P (insn).
7201         Reapply
7202         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7204         PR rtl-optimization/57268
7205         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7206         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7208 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7210         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7211         (ix86_avoid_lea_for_addr): Likewise.
7212         (exact_dependency_1): Likewise.
7213         (ix86_adjust_cost): Likewise.
7214         (swap_top_of_ready_list): Fix formatting and !reload_completed check
7215         removed.
7216         (do_reorder_for_imul): Fix typo, formatting and
7217         !reload_completed check removed.
7218         (ix86_sched_reorder): Fix typo and formatting.
7219         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7220         list.
7222 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
7224         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7226 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
7228         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7229         <STRING_CST>: Likewise.
7230         <VECTOR_CST>: Likewise.
7232 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
7233             Mikael Morin  <mikael@gcc.gnu.org>
7235         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7236         * config.in: Regenerated.
7237         * configure: Regenerated.
7239 2013-06-01  Jan Hubicka  <jh@suse.cz>
7241         PR middle-end/57366
7242         * cgraphunit.c (compile): When weakref is not supported,
7243         set up transparent aliases before final output pass.
7244         * varasm.c (assemble_alias): Do not try to do it here.
7246 2013-06-01  Jan Hubicka  <jh@suse.cz>
7248         PR middle-end/57467
7249         * passes.c (for_per_function): Skip unanalyzed functions.
7251 2013-06-01  Jan Hubicka  <jh@suse.cz>
7253         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
7254         (lto_symtab_merge_symbols_1): ... this one.
7255         (lto_symtab_merge_cgraph_nodes): Rename to ...
7256         (lto_symtab_merge_symbols): ... this one; simplify.
7257         * cgraph.c (same_body_aliases_done): Rename to ...
7258         (cpp_implicit_aliases_done): ... this one.
7259         (cgraph_create_function_alias): Update.
7260         (cgraph_same_body_alias): Update.
7261         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
7262         (verify_edge_corresponds_to_fndecl): Simplify.
7263         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
7264         (cgraph_node): Remove same_body_alias.
7265         (varpool_node): Remove alias_of and extra_name_alias.
7266         (same_body_aliases_done): Rename to ..
7267         (cpp_implicit_aliases_done): ... this one.
7268         (symtab_alias_ultimate_target): Add default parameter.
7269         (symtab_resolve_alias): New function.
7270         (fixup_same_cpp_alias_visibility): Declare.
7271         (cgraph_function_node): Add default parameter.
7272         (cgraph_node_asm_name): Likewise.
7273         (cgraph_function_or_thunk_node): Add default parameter; do
7274         not ICE when it is NULL.
7275         (varpool_variable_node): Likewise.
7276         * tree-emutls.c (create_emultls_var): Update.
7277         (ipa_lower_emutls): Update.
7278         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
7279         (cgraph_reset_node): Reset alias info.
7280         (cgraph_finalize_function): Update.
7281         (fixup_same_cpp_alias_visibility): Move to symtab.c.
7282         (analyze_function): Simplify.
7283         (cgraph_process_same_body_aliases): Simplify.
7284         (analyze_functions): Fixup same body aliases.
7285         (handle_alias_pairs): Simplify.
7286         (assemble_thunk): Update.
7287         (assemble_thunks_and_aliases): Update.
7288         (output_weakrefs): Rewrite.
7289         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
7290         (lto_output_varpool_node): Likewise.
7291         (compute_ltrans_boundary): Remve assert.
7292         (get_alias_symbol): New functoin.
7293         (input_node): Rewrite alias handling.
7294         (input_varpool_node): Likewise.
7295         * ipa-pure-const.c (propagate_pure_const): Fix formating.
7296         * ipa.c (process_references): Handle weakrefs correctly.
7297         (symtab_remove_unreachable_nodes): Likewise.
7298         * trans-mem.c (get_cg_data): Update.
7299         (ipa_tm_create_version_alias): Update.
7300         (ipa_tm_execute): Update.
7301         * symtab.c (dump_symtab_base): Dump aliases.
7302         (verify_symtab_base): Verify aliases.
7303         (symtab_node_availability): New function.
7304         (symtab_alias_ultimate_target): Simplify.
7305         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
7306         handle all the fixup cases.
7307         (symtab_resolve_alias): New function.
7308         * passes.c (ipa_write_summaries): Handle weakrefs.
7309         * varpool.c (varpool_analyze_node): Simplify.
7310         (assemble_aliases): Update.
7311         (varpool_create_variable_alias): Simplify.
7312         (varpool_extra_name_alias): Simplify.
7313         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
7314         (lto_symtab_merge_symbols): ... this one.
7316 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
7318         Revert
7319         PR rtl-optimization/57268
7320         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7321         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7323 2013-06-01  Tobias Burnus  <burnus@net-b.de>
7325         Partially reverted:
7326         2013-05-31  Tobias Burnus  <burnus@net-b.de>
7328         PR middle-end/57073
7329         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
7330         further up.
7332 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7334         PR rtl-optimization/57268
7335         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7336         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7338 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7340         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
7341         unordered comparison operators when -fno-trapping-math is in effect
7342         on the e500.
7343         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
7344         and implement unordered comparison operators properly on the e500.
7346 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7348         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
7349         for constant scalar integers.
7350         (simplify_relational_operation_1): Likewise.
7352 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
7354         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
7355         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
7356         Fix comment.
7358 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7359             Igor Zamyatin  <igor.zamyatin@intel.com>
7361         Silvermont (SLM) architecture performance tuning.
7362         * config/i386/i386.h (enum ix86_tune_indices): Add
7363         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
7364         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
7366         * config/i386/i386.c (initial_ix86_tune_features)
7367         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
7368         (ix86_lea_outperforms): Handle Silvermont tuning.
7369         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
7370         call.
7371         (ix86_use_lea_for_mov): Likewise.
7372         (ix86_avoid_lea_for_addr): Likewise.
7373         (ix86_lea_for_add_ok): Likewise.
7374         (exact_dependency_1): New function.
7375         (exact_store_load_dependency): Likewise.
7376         (ix86_adjust_cost): Handle Silvermont tuning.
7377         (do_reoder_for_imul): Likewise.
7378         (swap_top_of_ready_list): New function.
7379         (ix86_sched_reorder): Changed to handle Silvermont tuning.
7381         * config/i386/i386.md (peepholes that split memory operand in fp
7382         converts): New.
7384 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7386         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7387         Remove un-necessary braces.
7389 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7391         * config/aarch64/aarch64.c (aarch64_classify_symbol):
7392         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
7394 2013-05-31  Tobias Burnus  <burnus@net-b.de>
7396         PR middle-end/57073
7397         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
7399 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7401         PR target/56315
7402         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
7403         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
7404         * config/arm/neon.md (iordi3_neon): Remove.
7405         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
7406         * config/arm/predicates.md (imm_for_neon_logic_operand):
7407         Move to earlier in the file.
7408         (neon_logic_op2): Likewise.
7409         (arm_iordi_operand_neon): New predicate.
7411 2013-05-31  Richard Biener  <rguenther@suse.de>
7413         PR tree-optimization/57478
7414         PR tree-optimization/57453
7415         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
7416         are life as well.
7418 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
7420         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
7421         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
7423 2013-05-30  Tobias Burnus  <burnus@net-b.de>
7424             Thomas Koenig  <tkoenig@gcc.gnu.org>
7426         PR middle-end/57073
7427         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
7428         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
7430 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
7432         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
7434 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
7436         * target.def (register_usage_leveling_p): New hook.
7437         * targhooks.c (default_register_usage_leveling_p): New.
7438         * targhooks.h (default_register_usage_leveling_p): New prototype.
7439         * lra-assigns.c (register_usage_leveling_p): Use the hook.
7440         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
7441         * doc/tm.texi: Update.
7442         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
7444 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
7446         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
7447         (*insv_reg<mode>): New define_insn.
7449 2013-05-30  Joern Rennecke <joern.rennecke@embecosm.com>
7451         PR rtl-optimization/57439
7452         * postreload.c (move2add_valid_value_p): Check that we have
7453         a zero subreg_regno_offset when accessing the register in
7454         the requested mode.
7456 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7457             Igor Zamyatin  <igor.zamyatin@intel.com>
7459         Silvermont (SLM) architecture pipeline model, tuning and
7460         insn selection.
7461         * config.gcc: Add slm config options and target.
7463         * config/i386/slm.md: New.
7465         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
7467         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
7468         PROCESSOR_SLM.
7469         (ix86_target_macros_internal): Likewise.
7471         * gcc/config/i386/i386.c (slm_cost): New cost.
7472         (m_SLM): New macro flag.
7473         (initial_ix86_tune_features): Set m_SLM.
7474         (x86_accumulate_outgoing_args): Likewise.
7475         (x86_arch_always_fancy_math_387): Likewise.
7476         (processor_target_table): Add slm cost.
7477         (cpu_names): Add slm cpu name.
7478         (x86_option_override_internal): Set SLM ISA.
7479         (ix86_issue_rate): New case PROCESSOR_SLM.
7480         (ia32_multipass_dfa_lookahead): Likewise.
7481         (fold_builtin_cpu): Add slm.
7483         * config/i386/i386.h (TARGET_SLM): New target macro.
7484         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
7485         (processor_type): Add PROCESSOR_SLM.
7487         * config/i386/i386.md (cpu): Add new value "slm".
7488         (slm.md): Include slm.md.
7490 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
7491             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7493         * config/arm/arm-protos.h: Add and update function protos.
7494         * config/arm/arm.c (use_simple_return_p): New added.
7495         (thumb2_expand_return): Check simple_return flag.
7496         * config/arm/arm.md: Add simple_return and conditional simple_return.
7497         * config/arm/iterators.md: Add iterator for return and simple_return.
7499 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7501         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
7502         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
7503         (arm_emit_vfp_multi_reg_pop): Likewise.
7504         (thumb2_emit_ldrd_pop): Likewise.
7505         (arm_expand_epilogue): Add misc REG_CFA notes.
7506         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
7508 2013-05-29  Lawrence Crowl  <crowl@google.com>
7510         * config/arm/t-arm: Update for below.
7512         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
7513         Change type to hash_table.  Update dependent calls and types.
7515         * config/i386/t-cygming: Update for below.
7517         * config/i386/t-interix: Update for below.
7519         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
7520         Change type to hash_table.  Update dependent calls and types.
7521         (i386_find_on_wrapper_list::wrappers): Likewise.
7523         * config/ia64/t-ia64: Update for below.
7525         * config/ia64/ia64.c (bundle_state_table):
7526         Change type to hash_table.  Update dependent calls and types.
7528         * config/mips/mips.c (mips_reorg_process_insns::htab):
7529         Change type to hash_table.  Update dependent calls and types.
7531         * config/sol2.c (solaris_comdat_htab):
7532         Change type to hash_table.  Update dependent calls and types.
7534         * config/t-sol2: Update for above.
7536 2013-05-29  Teresa Johnson  <tejohnson@google.com>
7538         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
7539         functions are not yet marked as defined.
7541 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
7542             Pat Haugen <pthaugen@us.ibm.com>
7543             Peter Bergner <bergner@vnet.ibm.com>
7545         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
7546         instructions.
7547         (VEC_A): Likewise.
7548         (VEC_C): Likewise.
7549         (vrotl<mode>3): Likewise.
7550         (vashl<mode>3): Likewise.
7551         (vlshr<mode>3): Likewise.
7552         (vashr<mode>3): Likewise.
7554         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7555         support for power8 V2DI builtins.
7557         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
7558         power8 V2DI builtins.
7559         (vupkhsw): Likewise.
7560         (vupklsw): Likewise.
7561         (vaddudm): Likewise.
7562         (vminsd): Likewise.
7563         (vmaxsd): Likewise.
7564         (vminud): Likewise.
7565         (vmaxud): Likewise.
7566         (vpkudum): Likewise.
7567         (vpksdss): Likewise.
7568         (vpkudus): Likewise.
7569         (vpksdus): Likewise.
7570         (vrld): Likewise.
7571         (vsld): Likewise.
7572         (vsrd): Likewise.
7573         (vsrad): Likewise.
7574         (vsubudm): Likewise.
7575         (vcmpequd): Likewise.
7576         (vcmpgtsd): Likewise.
7577         (vcmpgtud): Likewise.
7578         (vcmpequd_p): Likewise.
7579         (vcmpgtsd_p): Likewise.
7580         (vcmpgtud_p): Likewise.
7581         (vupkhsw): Likewise.
7582         (vupklsw): Likewise.
7583         (vaddudm): Likewise.
7584         (vmaxsd): Likewise.
7585         (vmaxud): Likewise.
7586         (vminsd): Likewise.
7587         (vminud): Likewise.
7588         (vpksdss): Likewise.
7589         (vpksdus): Likewise.
7590         (vpkudum): Likewise.
7591         (vpkudus): Likewise.
7592         (vrld): Likewise.
7593         (vsld): Likewise.
7594         (vsrad): Likewise.
7595         (vsrd): Likewise.
7596         (vsubudm): Likewise.
7598         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
7599         support for power8 V2DI instructions.
7601         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
7602         power8 V2DI instructions.  Combine pack and unpack insns to use an
7603         iterator for each mode.  Check whether a particular mode supports
7604         Altivec instructions instead of just checking TARGET_ALTIVEC.
7605         (UNSPEC_VPKUWUM): Likewise.
7606         (UNSPEC_VPKSHSS): Likewise.
7607         (UNSPEC_VPKSWSS): Likewise.
7608         (UNSPEC_VPKUHUS): Likewise.
7609         (UNSPEC_VPKSHUS): Likewise.
7610         (UNSPEC_VPKUWUS): Likewise.
7611         (UNSPEC_VPKSWUS): Likewise.
7612         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
7613         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
7614         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
7615         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
7616         (UNSPEC_VUPKHSB): Likewise.
7617         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
7618         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
7619         (UNSPEC_VUPKHSH): Likewise.
7620         (UNSPEC_VUPKLSB): Likewise.
7621         (UNSPEC_VUPKLSH): Likewise.
7622         (VI2): Likewise.
7623         (VI_char): Likewise.
7624         (VI_scalar): Likewise.
7625         (VI_unit): Likewise.
7626         (VP): Likewise.
7627         (VP_small): Likewise.
7628         (VP_small_lc): Likewise.
7629         (VU_char): Likewise.
7630         (add<mode>3): Likewise.
7631         (altivec_vaddcuw): Likewise.
7632         (altivec_vaddu<VI_char>s): Likewise.
7633         (altivec_vadds<VI_char>s): Likewise.
7634         (sub<mode>3): Likewise.
7635         (altivec_vsubcuw): Likewise.
7636         (altivec_vsubu<VI_char>s): Likewise.
7637         (altivec_vsubs<VI_char>s): Likewise.
7638         (altivec_vavgs<VI_char>): Likewise.
7639         (altivec_vcmpbfp): Likewise.
7640         (altivec_eq<mode>): Likewise.
7641         (altivec_gt<mode>): Likewise.
7642         (altivec_gtu<mode>): Likewise.
7643         (umax<mode>3): Likewise.
7644         (smax<mode>3): Likewise.
7645         (umin<mode>3): Likewise.
7646         (smin<mode>3): Likewise.
7647         (altivec_vpkuhum): Likewise.
7648         (altivec_vpkuwum): Likewise.
7649         (altivec_vpkshss): Likewise.
7650         (altivec_vpkswss): Likewise.
7651         (altivec_vpkuhus): Likewise.
7652         (altivec_vpkshus): Likewise.
7653         (altivec_vpkuwus): Likewise.
7654         (altivec_vpkswus): Likewise.
7655         (altivec_vpks<VI_char>ss): Likewise.
7656         (altivec_vpks<VI_char>us): Likewise.
7657         (altivec_vpku<VI_char>us): Likewise.
7658         (altivec_vpku<VI_char>um): Likewise.
7659         (altivec_vrl<VI_char>): Likewise.
7660         (altivec_vsl<VI_char>): Likewise.
7661         (altivec_vsr<VI_char>): Likewise.
7662         (altivec_vsra<VI_char>): Likewise.
7663         (altivec_vsldoi_<mode>): Likewise.
7664         (altivec_vupkhsb): Likewise.
7665         (altivec_vupkhs<VU_char>): Likewise.
7666         (altivec_vupkls<VU_char>): Likewise.
7667         (altivec_vupkhsh): Likewise.
7668         (altivec_vupklsb): Likewise.
7669         (altivec_vupklsh): Likewise.
7670         (altivec_vcmpequ<VI_char>_p): Likewise.
7671         (altivec_vcmpgts<VI_char>_p): Likewise.
7672         (altivec_vcmpgtu<VI_char>_p): Likewise.
7673         (abs<mode>2): Likewise.
7674         (vec_unpacks_hi_v16qi): Likewise.
7675         (vec_unpacks_hi_v8hi): Likewise.
7676         (vec_unpacks_lo_v16qi): Likewise.
7677         (vec_unpacks_hi_<VP_small_lc>): Likewise.
7678         (vec_unpacks_lo_v8hi): Likewise.
7679         (vec_unpacks_lo_<VP_small_lc>): Likewise.
7680         (vec_pack_trunc_v8h): Likewise.
7681         (vec_pack_trunc_v4si): Likewise.
7682         (vec_pack_trunc_<mode>): Likewise.
7684         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
7685         V2DI builtins.
7686         (vec_vmaxsd): Likewise.
7687         (vec_vmaxud): Likewise.
7688         (vec_vminsd): Likewise.
7689         (vec_vminud): Likewise.
7690         (vec_vpksdss): Likewise.
7691         (vec_vpksdus): Likewise.
7692         (vec_vpkudum): Likewise.
7693         (vec_vpkudus): Likewise.
7694         (vec_vrld): Likewise.
7695         (vec_vsld): Likewise.
7696         (vec_vsrad): Likewise.
7697         (vec_vsrd): Likewise.
7698         (vec_vsubudm): Likewise.
7699         (vec_vupkhsw): Likewise.
7700         (vec_vupklsw): Likewise.
7702 2013-05-29  Jan Hubicka  <jh@suse.cz>
7704         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
7705         flags; reorder rest of fields in more consistent way.
7706         (varpool_node): Remove analyzed, finalized and alias.
7707         (cgraph_ndoe): Likewise.
7708         (symtab_alias_ultimate_target): New function.
7709         (cgraph_function_node): Move offline.
7710         (cgraph_reset_node): Declare.
7711         (cgraph_comdat_can_be_unshared_p): Remove.
7712         (varpool_remove_initializer): Declare.
7713         (varpool_first_defined_variable, varpool_next_defined_variable
7714         cgraph_first_defined_function, cgraph_next_defined_function): Update.
7715         (cgraph_function_with_gimple_body_p): Update.
7716         (varpool_all_refs_explicit_p): Update.
7717         (symtab_alias_target): New function.
7718         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
7719         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
7720         (cgraph_function_or_thunk_node): Simplify using
7721         symtab_alias_ultimate_target.
7722         (varpool_variable_node): Likewise.
7723         * cgraph.c (cgraph_create_function_alias): Update.
7724         (cgraph_add_thunk): Update.
7725         (cgraph_remove_node): Update.
7726         (dump_cgraph_node): Do not dump removed flags.
7727         (cgraph_function_body_availability): Update.
7728         (cgraph_propagate_frequency): Update.
7729         (verify_cgraph_node): Check sanity of local flag.
7730         (cgraph_function_node): Move here from cgraph.h; revamp for
7731         cgraph_function_or_thunk_node.
7732         * lto-symtab.c (lto_varpool_replace_node): Update.
7733         (lto_symtab_resolve_can_prevail_p): Update.
7734         (lto_symtab_merge_cgraph_nodes): Update.
7735         * ipa-cp.c (determine_versionability, initialize_node_lattices,
7736         propagate_constants_accross_call, devirtualization_time_bonus,
7737         ipcp_propagate_stage): Update.
7738         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
7739         * ipa-inline-transform.c (clone_inlined_nodes,
7740         preserve_function_body_p): Update.
7741         * ipa-reference.c (propagate): Update.
7742         (write_node_summary_p): Update.
7743         * toplev.c (wrapup_global_declaration_2): Update.
7744         * cgraphunit.c (cgraph_analyze_function): Rename to ...
7745         (analyze_function) ... this one.
7746         (cgraph_process_new_functions): Update.
7747         (cgraph_reset_node): Export.
7748         (cgraph_finalize_function): Update.
7749         (cgraph_add_new_function): Update.
7750         (process_function_and_variable_attributes): Update.
7751         (varpool_finalize_decl): Update.
7752         (symbol_finalized): Remove.
7753         (symbol_finalized_and_needed): Rename to ...
7754         (symbol_defined_and_needed): ... update.
7755         (cgraph_analyze_functions): Update.
7756         (handle_alias_pairs): Update.
7757         (mark_functions_to_output): Update.
7758         (assemble_thunk): Update.
7759         (output_in_order): Update.
7760         (output_weakrefs): Update.
7761         (finalize_compilation_unit): Update.
7762         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
7763         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
7764         input_node, input_varpool_node): Update.
7765         * dbxout.c (dbxout_expand_expr): Update.
7766         * cgraphclones.c (cgraph_clone_node): Update.
7767         (cgraph_copy_node_for_versioning): Update.
7768         (cgraph_materialize_clone): Update.
7769         (cgraph_materialize_all_clones): Update.
7770         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
7771         propagate_pure_const, propagate_nothrow): Update.
7772         * lto-streamer-out.c (lto_output, write_symbol): Update.
7773         * ipa-utils.c (ipa_reverse_postorder): Update.
7774         * ipa-inline.c (can_inline_edge_p): Update.
7775         (update_caller_keys, ipa_inline): Update.
7776         * dwarf2out.c (reference_to_unused,
7777         premark_types_used_by_global_vars_helper): Update.
7778         * tree-eh.c (tree_could_trap_p): Update.
7779         * ipa-split.c (consider_split, execute_split_functions): Update.
7780         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
7781         has_addr_references_p): Update; move ahead in file for better
7782         readability.
7783         (process_references): Simplify.
7784         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
7785         bodies are removed.
7786         (cgraph_comdat_can_be_unshared_p): Make static.
7787         (cgraph_externally_visible_p): Update.
7788         (varpool_externally_visible_p): Update.
7789         (function_and_variable_visibility): Update.
7790         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
7791         ipa_tm_mark_force_output_node): Update.
7792         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
7793         estimate_edge_devirt_benefit, inline_generate_summary,
7794         inline_write_summary): Update.
7795         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
7796         * ipa-prop.c (ipa_compute_jump_functions): Update.
7797         (ipa_print_node_params, ipa_prop_read_section,
7798         ipa_update_after_lto_read, read_replacements_section): Update.
7799         * varasm.c (mark_decl_referenced): Update.
7800         (assemble_alias, dump_tm_clone_pairs): Update.
7801         * tree-inline.c (copy_bb): Update.
7802         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
7803         Update.
7804         * symtab.c (dump_symtab_base): Print new flags.
7805         (verify_symtab_base): Verify new flags.
7806         (symtab_alias_ultimate_target): New function.
7807         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
7808         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
7809         Update.
7810         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
7811         Update.
7812         * i386.c (ix86_get_function_versions_dispatcher,
7813         ix86_generate_version_dispatcher_body): Update.
7814         (fold_builtin_cpu): Use varpool_add_new_variable.
7815         * varpool.c (varpool_remove_initializer): Break out from ...
7816         (varpool_remove_node): ... this one.
7817         (dump_varpool_node, varpool_node_for_asm,
7818         cgraph_variable_initializer_availability, varpool_analyze_node,
7819         varpool_assemble_decl, varpool_remove_unreferenced_decls,
7820         varpool_finalize_named_section_flags, varpool_create_variable_alias):
7821         Update.
7823 2013-05-29  Jan Hubicka  <jh@suse.cz>
7825         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
7827 2013-05-29  Easwaran Raman  <eraman@google.com>
7829         PR tree-optimization/57442
7830         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
7831         when control exits the main loop.
7833 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
7835         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
7836         and RX600.
7837         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
7838         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
7839         * rx/t-rx: Add rx100 under multi library matches option for nofpu
7840         option.
7842 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
7844         PR tree-optimization/57441
7845         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
7846         Don't limit size of incr_vec to number of candidates.
7848 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
7850         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
7851         and mips16 directories.
7852         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
7853         (MULTILIB_DIRNAMES): Ditto.
7854         (MULTILIB_EXCEPTIONS): Add new exceptions.
7855         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
7856         (MULTILIB_DIRNAMES): Ditto.
7857         (MULTILIB_EXCEPTIONS): Add new exceptions.
7859 2012-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7860             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7862         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
7863         SYMBOL_TINY_ABSOLUTE.
7864         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
7865         SYMBOL_TINY_ABSOLUTE.
7866         (aarch64_expand_mov_immediate): Likewise.
7867         (aarch64_classify_symbol): Likewise.
7868         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
7869         Permit SYMBOL_TINY_ABSOLUTE.
7870         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
7872 2013-05-29  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7873             Marcus Shawcroft  <marcus.shawcroft@arm.com>
7875         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
7876         Refactor if/switch.  Replace gcc_assert with if.
7878 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
7880         * config/i386/i386.c (initial_ix86_tune_features): Enable
7881         FP Reassociation for AMD bdver1 and bdver2.
7883 2013-05-29  Martin Jambor  <mjambor@suse.cz>
7885         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
7886         and IMAGPART_EXPR do not occur within other handled_components.
7888 2013-05-29  Richard Biener  <rguenther@suse.de>
7890         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
7891         access on whether the use is in the BB we currently try to
7892         vectorize.
7893         (vect_bb_vectorization_profitable_p): Pass the BB we currently
7894         vectorize to vect_bb_slp_scalar_cost.
7896 2013-05-29  Richard Biener  <rguenther@suse.de>
7898         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
7899         computing scalar cost offsetted by stmts that are kept live
7900         by scalar uses.
7901         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
7902         for computation of scalar cost.
7904 2013-05-28  Steve Ellcey  <sellcey@mips.com>
7906         * config/mips/mips-cpus.def (mips32r2): Change processor type.
7908 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
7910         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
7911         array notation built-in reduction functions.
7912         * doc/passes.texi (Passes): Added documentation about changes done
7913         for Cilk Plus.
7914         * doc/invoke.texi (C Dialect Options): Added documentation about
7915         the -fcilkplus flag.
7916         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
7917         (BUILTINS_DEF): Depend on cilkplus.def.
7918         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
7919         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
7920         * cilkplus.def: New file.
7922 2013-05-28  Joern Rennecke <joern.rennecke@embecosm.com>
7924         PR rtl-optimization/57439
7925         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
7927 2013-05-28  Easwaran Raman  <eraman@google.com>
7929         PR tree-optimization/57337
7930         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
7931         (find_insert_point): Correctly identify the insertion point
7932         when two statements with the same UID is compared.
7934 2013-05-28  Richard Biener  <rguenther@suse.de>
7936         PR tree-optimization/56787
7937         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
7938         from the list of data references.
7939         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
7940         clobbers.
7941         (vect_analyze_loop_operations): Likewise.
7942         (vect_transform_loop): Remove clobbers.
7944 2013-05-28  Martin Jambor  <mjambor@suse.cz>
7946         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
7947         and REALPART_EXPRs have scalar type.
7949 2013-05-28  Richard Biener  <rguenther@suse.de>
7951         PR tree-optimization/57411
7952         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
7953         virtual operands.
7954         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
7955         virtual operand propagation.
7957 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7959         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
7960         destination register for bmasksi_vis.
7961         (vector_init_bshuffle): Likewise.
7962         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
7964 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
7966         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
7967         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
7968         mode if the instruction isn't available in the original mode.
7969         * config/sparc/sparc.opt (mfix-ut699): New option.
7970         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
7971         (divdf3): Turn into expander.
7972         (divdf3_nofix): New insn.
7973         (divdf3_fix): Likewise.
7974         (divsf3): Disable if -mfix-ut699.
7975         (sqrtdf2): Turn into expander.
7976         (sqrtdf2_nofix): New insn.
7977         (sqrtdf2_fix): Likewise.
7978         (sqrtsf2): Disable if -mfix-ut699.
7980 2013-05-27  Richard Biener  <rguenther@suse.de>
7982         PR middle-end/57412
7983         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7984         block for the new loop.
7986 2013-05-27  Richard Biener  <rguenther@suse.de>
7988         PR tree-optimization/57343
7989         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7990         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7991         (number_of_iterations_cond): Do not build the folded tree.
7993 2013-05-27  Richard Biener  <rguenther@suse.de>
7995         Revert
7996         PR middle-end/57381
7997         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7998         OEP_CONSTANT_ADDRESS_OF retained.
8000         PR tree-optimization/57417
8001         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
8002         for unchanged base.
8003         (set_ssa_val_to): Compare addresses using
8004         get_addr_base_and_unit_offset.
8006 2013-05-27  Joern Rennecke <joern.rennecke@embecosm.com>
8008         PR rtl-optimization/56833
8009         * postreload.c (move2add_record_mode): New function.
8010         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
8011         (move2add_use_add2_insn): Use move2add_record_sym_value.
8012         (move2add_use_add3_insn): Likewise.
8013         (reload_cse_move2add): Use move2add_valid_value_p and
8014         move2add_record_mode.  Invalidate call-clobbered and REG_INC
8015         affected regs by setting reg_mode to VOIDmode.
8016         (move2add_note_store): Don't pretend the inside of a SUBREG is
8017         the actual destination.  Invalidate single/leading registers by
8018         setting reg_mode to VOIDmode.
8019         Use move2add_record_sym_value, move2add_valid_value_p and
8020         move2add_record_mode.
8022 2013-05-27  Richard Biener  <rguenther@suse.de>
8024         PR tree-optimization/57396
8025         * tree-affine.c (double_int_constant_multiple_p): Properly
8026         return false for val == 0 and div != 0.
8028 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8030         * config/mips/mips.h: Use #elif in preprocessor conditions.
8032 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8034         PR target/53916
8035         * config/mips/constraints.md (kl): New constraint.
8036         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
8037         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
8038         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
8039         from LO for MIPS16.
8040         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8042 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8044         PR target/55777
8045         * config/mips/mips.c (mips_can_inline_p): New function.
8046         (TARGET_CAN_INLINE_P): Define.
8048 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
8050         * sched-int.h (ds_t, dw_t): Make unsigned int.
8051         Fix documentation that describes how all the ds_t bits are used.
8052         Reserve the last bit for delayed-branch scheduling.
8053         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8054         (BITS_PER_DEP_WEAK): Fix definition and documentation.
8055         (gen_dep_weak_1): Remove prototype.
8056         * sched-deps.c (get_dep_weak_1): Make static.
8057         * target.def (speculate_insn, needs_block_p, gen_spec_check,
8058         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8059         * doc/tm.texi: Regenerate.
8060         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8062 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
8064         PR debug/56950
8065         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8067 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
8068             Sandra Loosemore  <sandra@codesourcery.com>
8070         * config.gcc (powerpc-*): Allow native for with-cpu.
8072 2013-05-24  Jeff Law  <law@redhat.com>
8074         PR tree-optimization/57124
8075         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8076         conversion feeding a condition if the range has an overflow
8077         if -fstrict-overflow.  Add warnings for when we do make the
8078         transformation.
8080 2013-05-24  Dehao Chen  <dehao@google.com>
8082         * tree-cfg.c (locus_discrim_map): Fix the typo.
8083         (locus_discrim_hasher): Likewise.
8084         (locus_discrim_hasher::hash): Likewise.
8085         (locus_discrim_hasher::equal): Likewise.
8087 2013-05-24  Martin Jambor  <mjambor@suse.cz>
8089         PR tree-optimization/57294
8090         * cgraph.h (ipa_record_stmt_references): Declare.
8091         * cgraphbuild.c (ipa_record_stmt_references): New function.
8092         (build_cgraph_edges): Use ipa_record_stmt_references.
8093         (rebuild_cgraph_edges): Likewise.
8094         (cgraph_rebuild_references): Likewise.
8095         * ipa-prop.c (ipa_modify_call_arguments): Discard references
8096         associated with the old statement and build references from the
8097         newly built statements.
8098         * ipa-ref.c (ipa_remove_stmt_references): New function.
8099         * ipa-ref.h (ipa_remove_stmt_references): Declare.
8101 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
8103         * lra-constraints.c (emit_spill_move): Use smaller mode for
8104         mem-mem moves.
8105         (check_and_process_move): Consider mem-reg moves for secondary
8106         too.
8107         (curr_insn_transform): Don't lose insns emitted before for
8108         secondary memory moves.
8109         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
8110         reg set up in the current insn.
8112 2013-05-24  Dehao Chen  <dehao@google.com>
8114         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8115         hash function.
8116         (locus_descrim_hasher::equal): Likewise.
8117         (build_gimple_cfg): New discriminator assignment algorithm.
8118         (make_edges): Likewise.
8119         (next_discriminator_for_locus): Likewise.
8120         (same_line_p): Likewise.
8121         (assign_discriminators): Likewise.
8122         (make_cond_expr_edges): Likewise.
8123         (make_gimple_switch_edges): Likewise.
8124         (make_goto_expr_edges): Likewise.
8125         (make_gimple_asm_edges): Likewise.
8127 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
8129         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8130         X format specifier to only display bottom 16 bits.
8131         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8132         immediate to match for operand 2, since it will be masked.
8134 2013-05-24  Richard Biener  <rguenther@suse.de>
8136         PR tree-optimization/57287
8137         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8138         all SSA names that occur in abnormal PHIs.
8140 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8142         PR tree-ssa/57385
8143         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8144         that index is not negative.
8146 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
8148         PR rtl-optimization/55177
8149         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8150         (simplify_byte_swapping_operation): New.
8151         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8152         (simplify_relational_operation_1): Deal with BSWAP.
8154 2013-05-23  Richard Henderson  <rth@redhat.com>
8156         PR target/56742
8157         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8158         (ix86_reorg): Call it.
8160 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
8162         PR target/57379
8163         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8164         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8165         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8167 2013-05-23  Christian Bruel  <christian.bruel@st.com>
8169         PR debug/57351
8170         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8172 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8173             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8175         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8176         * config/aarch64/constraints.md (Usa): Remove.
8177         * doc/md.texi (AArch64 Usa): Remove.
8179 2013-05-23  Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8180             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8182         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8183         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8184         * config/aarch64/predicates.md (aarch64_const_address): Remove.
8185         (aarch64_mov_operand): Use aarch64_mov_operand_p.
8187 2013-05-23  Vidya Praveen <vidyapraveen@arm.com>
8189         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8190         instruction (AdvSIMD).
8191         * config/aarch64/aarch64-builtins.c
8192         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8193         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8195 2013-05-23  Martin Jambor  <mjambor@suse.cz>
8197         PR middle-end/57347
8198         * tree.h (contains_bitfld_component_ref_p): Declare.
8199         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8200         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
8201         caller.
8202         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8203         not access a bit-field.  Assert all final offsets are byte-aligned.
8205 2013-05-23  Richard Biener  <rguenther@suse.de>
8207         PR tree-optimization/57380
8208         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8209         least one invariant or re-used load.
8210         * passes.c (init_optimization_passes): Move pass_phiprop before
8211         pass_forwprop.
8213 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
8215         * config/aarch64/aarch64-simd.md
8216         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8218 2013-05-23  Richard Biener  <rguenther@suse.de>
8220         PR middle-end/57381
8221         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8222         OEP_CONSTANT_ADDRESS_OF retained.
8224 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
8226         PR middle-end/57344
8227         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8228         don't lower unit.  Handle unit not being always BITS_PER_WORD.
8230 2013-05-23  Richard Biener  <rguenther@suse.de>
8232         PR rtl-optimization/57341
8233         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8234         instead of true_dependence.
8236 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
8238         * bb-reorder.c (branch_threshold): Make const.
8239         (exec_threshold): Ditto.
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/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8246         documentation for the power8 crypto builtins.
8248         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
8250         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
8251         macros for defining power8 builtin functions.
8252         (BU_P8V_AV_2): Likewise.
8253         (BU_P8V_AV_P): Likewise.
8254         (BU_P8V_VSX_1): Likewise.
8255         (BU_P8V_OVERLOAD_1): Likewise.
8256         (BU_P8V_OVERLOAD_2): Likewise.
8257         (BU_CRYPTO_1): Likewise.
8258         (BU_CRYPTO_2): Likewise.
8259         (BU_CRYPTO_3): Likewise.
8260         (BU_CRYPTO_OVERLOAD_1): Likewise.
8261         (BU_CRYPTO_OVERLOAD_2): Likewise.
8262         (XSCVSPDP): Fix typo, point to the correct instruction.
8263         (VCIPHER): Add power8 crypto builtins.
8264         (VCIPHERLAST): Likewise.
8265         (VNCIPHER): Likewise.
8266         (VNCIPHERLAST): Likewise.
8267         (VPMSUMB): Likewise.
8268         (VPMSUMH): Likewise.
8269         (VPMSUMW): Likewise.
8270         (VPERMXOR_V2DI): Likewise.
8271         (VPERMXOR_V4SI: Likewise.
8272         (VPERMXOR_V8HI: Likewise.
8273         (VPERMXOR_V16QI: Likewise.
8274         (VSHASIGMAW): Likewise.
8275         (VSHASIGMAD): Likewise.
8276         (VPMSUM): Likewise.
8277         (VPERMXOR): Likewise.
8278         (VSHASIGMA): Likewise.
8280         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8281         __CRYPTO__ if the crypto instructions are available.
8282         (altivec_overloaded_builtins): Add support for overloaded power8
8283         builtins.
8285         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
8286         support for power8 crypto builtins.
8287         (builtin_function_type): Likewise.
8288         (altivec_init_builtins): Add support for builtins that take vector
8289         long long (V2DI) arguments.
8291         * config/rs6000/crypto.md: New file, define power8 crypto
8292         instructions.
8294 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8295             Pat Haugen <pthaugen@us.ibm.com>
8296             Peter Bergner <bergner@vnet.ibm.com>
8298         * doc/invoke.texi (Option Summary): Add power8 options.
8299         (RS/6000 and PowerPC Options): Likewise.
8301         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
8302         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
8303         wm, wn, wr documentation.
8305         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
8306         registers if direct move instructions are enabled.
8307         (wn): New constraint for no registers.
8308         (wq): New constraint for quad word even GPR registers.
8309         (wr): New constraint if 64-bit instructions are enabled.
8310         (wv): New constraint if power8 vector instructions are enabled.
8311         (wQ): New constraint for quad word memory locations.
8313         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
8314         constraint for 0..15 for crypto instructions.
8315         (gpc_reg_operand): If VSX allow registers in VSX registers as well
8316         as GPR and floating point registers.
8317         (int_reg_operand): New predicate to match only GPR registers.
8318         (base_reg_operand): New predicate to match base registers.
8319         (quad_int_reg_operand): New predicate to match even GPR registers
8320         for quad memory operations.
8321         (vsx_reg_or_cint_operand): New predicate to allow vector logical
8322         operations in both GPR and VSX registers.
8323         (quad_memory_operand): New predicate for quad memory operations.
8324         (reg_or_indexed_operand): New predicate for direct move support.
8326         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
8327         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
8328         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
8329         (POWERPC_MASKS): Add power8 options.
8330         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
8331         various options.
8333         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8334         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
8336         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
8337         (-mpower8-fusion): New power8 options.
8338         (-mpower8-fusion-sign): Likewise.
8339         (-mpower8-vector): Likewise.
8340         (-mcrypto): Likewise.
8341         (-mdirect-move): Likewise.
8342         (-mquad-memory): Likewise.
8344         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
8345         power8.
8346         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
8347         registers.
8348         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
8349         (rs6000_debug_vector_unit): Add p8_vector.
8350         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
8351         definitions.  Also print fusion state.
8352         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
8353         (rs6000_builtin_mask_calculate): Add power8 builtin support.
8354         (rs6000_option_override_internal): Add support for power8.
8355         (rs6000_common_init_builtins): Add debugging for skipped builtins
8356         if -mdebug=builtin.
8357         (rs6000_adjust_cost): Add power8 support.
8358         (rs6000_issue_rate): Likewise.
8359         (insn_must_be_first_in_group): Likewise.
8360         (insn_must_be_last_in_group): Likewise.
8361         (force_new_group): Likewise.
8362         (rs6000_register_move_cost): Likewise.
8363         (rs6000_opt_masks): Likewise.
8365         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
8366         power8 capable assembler, default to power7 options.
8367         (TARGET_DIRECT_MOVE): Likewise.
8368         (TARGET_CRYPTO): Likewise.
8369         (TARGET_P8_VECTOR): Likewise.
8370         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
8371         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
8372         (VECTOR_MEM_P8_VECTOR_P): Likewise.
8373         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
8374         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
8375         (TARGET_XSCVDPSPN): Likewise.
8376         (TARGET_XSCVSPDPN): Likewsie.
8377         (TARGET_SYNC_HI_QI): Likewise.
8378         (TARGET_SYNC_TI): Likewise.
8379         (MASK_CRYPTO): Likewise.
8380         (MASK_DIRECT_MOVE): Likewise.
8381         (MASK_P8_FUSION): Likewise.
8382         (MASK_P8_VECTOR): Likewise.
8383         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
8384         temporary used by some of the direct move instructions to get two FP
8385         temporary registers does not force creation of a stack frame.
8386         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
8387         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
8388         that any VSX registers are tieable, even if they are also an
8389         Altivec vector mode.
8390         (r6000_reg_class_enum): Add wm, wr, wv constraints.
8391         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
8392         (RS6000_BTM_CRYPTO): Likewise.
8393         (RS6000_BTM_COMMON): Likewise.
8395         * config/rs6000/rs6000.md (cpu attribute): Add power8.
8396         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
8397         (enum rs6000_vector): Add power8 vector support.
8399 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8401         PR target/19599
8402         PR target/57340
8403         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
8404         (any_sibcall_could_use_r3): this and handle indirect calls.
8405         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
8407 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8409         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
8411 2013-05-22  Richard Biener  <rguenther@suse.de>
8413         PR middle-end/57349
8414         * profile.c (branch_prob): Do not split blocks that are
8415         abnormally receiving from ECF_RETURNS_TWICE functions.
8417 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
8419         * recog.c (offsettable_address_addr_space_p): Fix calculation of
8420         address mode.  Move pointer mode initialization to the same place.
8422 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8424         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
8425         while it has any effect.
8427 2013-05-21  Easwaran Raman  <eraman@google.com>
8429         PR tree-optimization/57322
8430         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
8431         UID of the statement added to the BB to be 1.
8433 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
8435         PR tree-optimization/57331
8436         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
8437         of conversion from pointer type to integral type with integer.
8439 2013-05-21  Martin Jambor  <mjambor@suse.cz>
8441         PR lto/57289
8442         * ipa-prop.c (ipa_read_node_info): Process param_used and
8443         controlled_uses in the same order as when writing.
8445 2013-05-21  Magnus Granberg  <baldrick@free.fr>
8447         PR plugins/56754
8448         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
8450 2013-05-21  Richard Biener  <rguenther@suse.de>
8452         PR tree-optimization/57318
8453         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
8454         estimate stmts with side-effects as likely eliminated.
8456 2013-05-21  Richard Biener  <rguenther@suse.de>
8458         PR tree-optimization/57330
8459         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
8460         preserve the call stmts fntype.
8462 2013-05-21  Richard Biener  <rguenther@suse.de>
8464         PR tree-optimization/57303
8465         * tree-ssa-sink.c (statement_sink_location): Improve killing
8466         stmt detection and properly handle self-assignments.
8468 2013-05-21  Christian Bruel  <christian.bruel@st.com>
8470         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
8471         spanning registers. LEAF_REG_REMAP is supported only for contiguous
8472         registers. Set register size out of the PARALLEL loop.
8474 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
8476         PR target/56547
8477         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
8478         (*fmasf4, *fmasf4_media): New insns.
8480 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8482         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
8483         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
8484         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
8485         (mips_idiv_insns): Update the comments to say that the returned
8486         instruction counts are in units of BASE_INSN_LENGTH.
8487         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
8488         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
8489         using 2 rather than 4 as the length of indirect MIPS16 and
8490         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
8491         length of a NOP.  Don't divide MIPS16 lengths by 2.
8492         (mips16_split_long_branches): Assume a branch is long if the
8493         length is greater than 4 rather than 8.
8494         * config/mips/mips.md (length): Give MIPS16 lengths directly,
8495         rather than multiplying them by 2.  Multiply instruction counts
8496         by BASE_INSN_LENGTH rather than 4.
8497         (*jump_mips16, tls_get_tp_mips16_<mode>)
8498         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
8500 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8502         * config/mips/mips.md (extended_mips16): Remove branch case.
8503         (length): Remove duplicated extended_mips16 test.
8505 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8507         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
8509 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
8511         * recog.h (Recog_data): Rename to...
8512         (recog_data_d): ...this.
8513         (recog_data): Update accordingly.
8514         * recog.c (recog_data): Likewise.
8515         * reload.c (save_recog_data): Likewise.
8516         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
8517         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
8519 2013-05-17  Julian Brown  <julian@codesourcery.com>
8521         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
8522         found in a REG_EQUAL note, invalidate it.
8524 2013-05-17  Easwaran Raman  <eraman@google.com>
8526         * tree-ssa-reassoc.c (find_insert_point): New function.
8527         (insert_stmt_after): Likewise.
8528         (get_def_stmt): Likewise.
8529         (ensure_ops_are_available): Likewise.
8530         (not_dominated_by): Likewise.
8531         (rewrite_expr_tree): Do not move statements beyond what is
8532         necessary. Remove call to swap_ops_for_binary_stmt...
8533         (reassociate_bb): ... and move it here.
8534         (build_and_add_sum): Assign UIDs for new statements.
8535         (linearize_expr): Likewise.
8536         (do_reassoc): Renumber gimple statement UIDs.
8538 2013-05-17  Jan Hubicka  <jh@suse.cz>
8540         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
8541         weakrefs.
8542         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
8543         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
8544         weakrefs.
8545         (output_weakrefs): Update.
8547 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
8548             Martin Jambor  <mjambor@suse.cz>
8550         PR middle-end/57276
8551         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
8552         value that corresponds to the given aggval is found in values vector.
8554 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
8556         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
8557         sse, sse2, sse3, ssse3 and sse4a flags to options.
8559 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
8561         * gengtype-state.c: (s_expr_writer): New class, to handle
8562         prettifying of output layout of s-expressions.
8563         (state_writer): New class, to write out gtype.state.
8564         (state_written_type_count): Move this variable into member data of
8565         state_writer.
8566         (s_expr_writer::s_expr_writer): New code: constructor for new class
8567         (state_writer::state_writer(): ditto
8568         (s_expr_writer::write_new_line): New function
8569         (s_expr_writer::write_any_indent): ditto
8570         (s_expr_writer::begin_s_expr): ditto
8571         (s_expr_writer::end_s_expr): ditto
8572         (write_state_fileloc): convert to method of state_writer...
8573         (state_writer:: write_state_fileloc): ...and use methods of
8574         s_expr_writer to write indentation into the gtype.state output file
8575         to visually represent the hierarchical structure of the list
8576         structures
8577         (write_state_fields): ditto, renaming to...
8578         (state_writer::write_state_fields)
8579         (write_state_a_string): ditto, renaming to...
8580         (state_writer::write_state_a_string)
8581         (write_state_string_option): ditto, renaming to...
8582         (state_writer::write_state_string_option)
8583         (write_state_type_option): ditto, renaming to...
8584         (state_writer::write_state_type_option)
8585         (write_state_nested_option): ditto, renaming to...
8586         (state_writer::write_state_nested_option)
8587         (write_state_option): ditto, renaming to...
8588         (state_writer::write_state_option)
8589         (write_state_options): ditto, renaming to...
8590         (state_writer::write_state_options)
8591         (write_state_lang_bitmap): ditto, renaming to...
8592         (state_writer::write_state_lang_bitmap)
8593         (write_state_version): ditto, renaming to...
8594         (state_writer::write_state_version)
8595         (write_state_scalar_type): ditto, renaming to...
8596         (state_writer::write_state_scalar_type)
8597         (write_state_string_type): ditto, renaming to...
8598         (state_writer::write_state_string_type)
8599         (write_state_undefined_type): ditto, renaming to...
8600         (state_writer::write_state_undefined_type)
8601         (write_state_struct_union_type): ditto, renaming to...
8602         (state_writer::write_state_struct_union_type)
8603         (write_state_struct_type): ditto, renaming to...
8604         (state_writer::write_state_struct_type)
8605         (write_state_user_struct_type): ditto, renaming to...
8606         (state_writer::write_state_user_struct_type)
8607         (write_state_lang_struct_type): ditto, renaming to...
8608         (state_writer::write_state_lang_struct_type)
8609         (write_state_param_struct_type): ditto, renaming to...
8610         (state_writer::write_state_param_struct_type)
8611         (write_state_pointer_type): ditto, renaming to...
8612         (state_writer::write_state_pointer_type)
8613         (write_state_array_type): ditto, renaming to...
8614         (state_writer::write_state_array_type)
8615         (write_state_gc_used): ditto, renaming to...
8616         (state_writer::write_state_gc_used)
8617         (write_state_common_type_content): ditto, renaming to...
8618         (state_writer::write_state_common_type_content)
8619         (write_state_type): ditto, renaming to...
8620         (state_writer::write_state_type)
8621         (write_state_pair_list): ditto, renaming to...
8622         (state_writer::write_state_pair_list)
8623         (write_state_pair): ditto, renaming to...
8624         (state_writer::write_state_pair)
8625         (write_state_typedefs): ditto, renaming to...
8626         (state_writer::write_state_typedefs)
8627         (write_state_structures): ditto, renaming to...
8628         (state_writer::write_state_structures)
8629         (write_state_param_structs): ditto, renaming to...
8630         (state_writer::write_state_param_structs)
8631         (write_state_variables): ditto, renaming to...
8632         (state_writer::write_state_variables)
8633         (write_state_srcdir): ditto, renaming to...
8634         (state_writer::write_state_srcdir)
8635         (write_state_files_list): ditto, renaming to...
8636         (state_writer::write_state_files_list)
8637         (write_state_languages): ditto, renaming to...
8638         (state_writer::write_state_languages)
8639         (write_state): create a state_writer instance and use it when
8640         writing out the state file
8642 2013-05-17  Mike Stump  <mikestump@comcast.net>
8644         PR rtl-optimization/57304
8645         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
8646         accessing DF_REF_REAL_LOC.
8648 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8650         PR rtl-optimization/57281
8651         PR rtl-optimization/57300
8652         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
8653         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
8654         what the other splitter did if the registers are dead.
8656 2013-05-17  Richard Biener  <rguenther@suse.de>
8658         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
8659         MEM_REF offsets.
8661 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8663         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
8664         linking.
8666 2013-05-17  Marek Polacek  <polacek@redhat.com>
8668         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
8669         length when doing non-zero store of storing '\0' to '\0'.
8671 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
8673         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
8674         vect_external_def oprnd1 with loop_vinfo, try to emit
8675         optional cast, negation and and stmts on the loop preheader
8676         edge instead of into the pattern def seq.
8678         PR tree-optimization/57051
8679         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
8680         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
8682 2013-05-16  Nick Clifton  <nickc@redhat.com>
8684         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
8685         (rl78_is_naked_func): New function.
8686         (rl78_expand_prologue): Skip prologue generation for naked functions.
8687         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
8688         * doc/extend.texi (naked): Add RL78 to the list of processors
8689         that supports this attribute.
8691 2013-05-16  Jeff Law  <law@redhat.com>
8693         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
8695 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8697         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
8698         cache parameters using detect_caches_amd also for CYRIX,
8699         NSC and TM2 signatures.
8701 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
8702             Dzianis Kahanovich  <mahatma@eu.by>
8704         PR target/45359
8705         PR target/46396
8706         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8707         VIA/Centaur processors and determine their cache parameters
8708         using detect_caches_amd.
8710 2013-05-16  Teresa Johnson  <tejohnson@google.com>
8712         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
8713         (rtl_verify_edges): New function.
8714         (rtl_verify_bb_insns): Ditto.
8715         (rtl_verify_bb_pointers): Ditto.
8716         (rtl_verify_bb_insn_chain): Ditto.
8717         (rtl_verify_fallthru): Ditto.
8718         (rtl_verify_bb_layout): Ditto.
8719         (rtl_verify_flow_info_1): Outline checks into new functions.
8720         (rtl_verify_flow_info): Ditto.
8722 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
8724         * cfghooks.c (copy_bbs): Add update_dominance argument.
8725         * cfghooks.h (copy_bbs): Update prototype.
8726         * tree-cfg.c (gimple_duplicate_sese_region):
8727         Add update_dominance argument.
8728         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
8729         * tree-ssa-loop-ch.c (copy_loop_headers): Update
8730         gimple_duplicate_sese_region call.
8731         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
8732         Update copy_bbs call.
8733         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
8734         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
8736 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8738         * tree-vectorizer.h (NUM_PATTERNS): Increment.
8739         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
8740         vect_recog_rotate_pattern.
8741         (vect_recog_rotate_pattern): New function.
8743 2013-05-16  Jason Merrill  <jason@redhat.com>
8745         * Makefile.in (LLINKER): New variable.
8746         (mostlyclean): Remove link mutex.
8747         * configure.ac: Handle --enable-link-mutex.
8748         * lock-and-run.sh: New script.
8750 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8752         PR target/19599
8753         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
8754         for NULL decl.
8756 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8758         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
8760 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
8762         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
8763         * config/arm/arm.c (next_consecutive_mem): New function.
8764         (gen_movmem_ldrd_strd): Likewise.
8765         * config/arm/arm.md (movmemqi): Update condition and code.
8766         (unaligned_loaddi, unaligned_storedi): New patterns.
8768 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8770         * config.gcc: Obsolete *-*-solaris2.9*.
8771         * doc/install.texi (Specific, *-*-solaris2*): Document it.
8773 2013-05-16  Richard Biener  <rguenther@suse.de>
8775         * passes.c (init_optimization_passes): Move pass_parallelize_loops
8776         earlier, after GRAPHITE transforms and IV canonicalization.
8778 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
8780         * omp-low.c (extract_omp_for_data): For collapsed loops,
8781         if at least one of the loops is known at compile time to
8782         iterate zero times, set count to 0.
8783         (expand_omp_regimplify_p): New function.
8784         (expand_omp_for_generic): For collapsed loops, if at least
8785         one of the loops isn't known to iterate at least once,
8786         add runtime check with setting count to 0.
8787         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
8788         For unsigned types if it isn't known at compile time that
8789         the loop will iterate at least once, add runtime check to bypass
8790         the whole loop if initial condition isn't true.
8792 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
8794         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
8796 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
8798         PR middle-end/57286
8799         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
8800         transformations to avoid an infinite loop.
8802 2013-05-16  Marek Polacek  <polacek@redhat.com>
8804         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
8806 2013-05-15  Leif Ekblad  <leif@rdos.net>
8808         * config/i386/i386.c (ix86_decompose_address): Use
8809         DEFAULT_TLS_SEG_REG to access TLS segment register.
8810         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
8811         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
8812         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
8814 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
8816         PR target/57260
8817         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
8818         sibling calls to functions that would normally be lazily bound,
8819         unless $gp is call-clobbered.
8821 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
8823         * config/i386/i386.c (ix86_option_override_internal): Update
8824         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
8825         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
8826         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
8827         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
8828         of TARGET_3DNOW.
8829         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
8831 2013-05-15  Andreas Schwab  <schwab@suse.de>
8833         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
8834         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
8835         third operand.
8837 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8839         * loop-unroll.c (report_unroll_peel): Check decision before
8840         emitting unroll/peel message.
8842 2013-05-15  Teresa Johnson  <tejohnson@google.com>
8844         * function.h (has_bb_partition): New rtl_data flag.
8845         (bb_reorder_complete): Ditto.
8846         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
8847         instead of flag_reorder_blocks_and_partition.
8848         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
8849         with some enhancements.
8850         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
8851         * bb-reorder.c (connect_traces): Check for has_bb_partition
8852         instead of flag_reorder_blocks_and_partition.
8853         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
8854         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
8855         verify_hot_cold_block_grouping.
8856         (partition_hot_cold_basic_blocks): Set has_bb_partition.
8858 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8860         PR target/19599
8861         * config/arm/predicates.md (call_insn_operand): New predicate.
8862         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
8863         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
8864         if insn is not a tail call.
8865         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
8866         registers.
8867         * config/arm/arm.h (enum reg_class): New caller save register class.
8868         (REG_CLASS_NAMES): Likewise.
8869         (REG_CLASS_CONTENTS): Likewise.
8870         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
8871         without decls.
8873 2013-05-15  Richard Biener  <rguenther@suse.de>
8875         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
8876         of MSG_OPTIMIZED_LOCATIONS.
8877         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
8878         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
8879         message.
8880         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
8881         of MSG_OPTIMIZED_LOCATIONS.
8882         (execute_vect_slp): Likewise.
8883         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
8884         (vect_create_cond_for_alias_checks): Likewise.
8885         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8886         (vect_recog_widen_mult_pattern): Likewise.
8887         (vect_recog_widen_sum_pattern): Likewise.
8888         (vect_recog_over_widening_pattern): Likewise.
8889         (vect_recog_widen_shift_pattern): Likewise.
8890         (vect_recog_vector_vector_shift_pattern): Likewise.
8891         (vect_recog_divmod_pattern): Likewise.
8892         (vect_recog_mixed_size_cond_pattern): Likewise.
8893         (vect_recog_bool_pattern): Likewise.
8894         (vect_pattern_recog_1): Likewise.
8896 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8898         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
8899         non-functions to builtin_unreachable.
8900         * ipa-inline-transform.c (inline_call): Do not assert estimates were
8901         correct when new direct edges were discovered.
8903 2013-05-15  Martin Jambor  <mjambor@suse.cz>
8905         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
8906         header, print symbol order instead of node uid, print more information
8907         about indirect edge targets.
8908         (ipa_make_edge_direct_to_target): Print symbol order instead of node
8909         uids.
8910         (ipa_make_edge_direct_to_target): Likewise.
8911         (remove_described_reference): Likewise.
8912         (propagate_controlled_uses): Likewise.
8913         (ipa_print_node_params): Also print symbol order.
8914         (ipcp_transform_function): Print symbol order instead of node uids.
8915         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
8916         (cgraph_get_create_real_symbol_node): Likewise.
8917         * ipa-cp.c (print_lattice): Likewise.
8918         (print_all_lattices): Likewise.
8919         (determine_versionability): Likewise.
8920         (initialize_node_lattices): Likewise.
8921         (estimate_local_effects): Likewise.
8922         (update_profiling_info): Likewise.
8923         (create_specialized_node): Likewise.
8924         (perhaps_add_new_callers): Likewise.
8925         (decide_about_value): Likewise.
8926         (decide_whether_version_node): Likewise.
8927         (identify_dead_nodes): Likewise.
8928         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
8929         (dump_inline_summary): Likewise.
8930         (estimate_node_size_and_time): Likewise.
8931         (inline_analyze_function): Likewise.
8932         * ipa-inline.c (report_inline_failed_reason): Likewise.
8933         (want_early_inline_function_p): Likewise.
8934         (edge_badness): Likewise.
8935         (update_edge_key): Likewise.
8936         (inline_small_functions): Likewise.  Add dumping of order to two other
8937         dumps.
8938         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
8939         instead of node uids.
8940         (propagate_pure_const): Likewise.
8941         (propagate_pure_const): Likewise.
8942         * ipa-utils.c (dump_cgraph_node_set): Likewise.
8943         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
8944         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
8945         of node uids.
8946         * tree-pretty-print.c (dump_function_header): Likewise.
8947         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
8948         Print symbol order instead of node uids.
8950 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8952         * config/s390/s390.c (s390_register_move_cost): Don't impose the
8953         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
8955 2013-05-15  Richard Biener  <rguenther@suse.de>
8957         PR tree-optimization/57275
8958         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
8959         return value for fail to do runtime alias checks for gather loads.
8961 2013-05-15  Jan Hubicka  <jh@suse.cz>
8963         PR lto/57038
8964         PR lto/47375
8965         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
8966         weakrefs are not external.
8967         (lto_symtab_merge_decls): Fix thinko when dealing with
8968         non-lto_symtab decls.
8969         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
8970         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
8971         * varpool.c (dump_varpool_node): Dump more flags.
8973 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8975         * config/i386/i386.c (processor_alias_table): Add instruction
8976         FSGSBASE for AMD bdver3 architecture.
8978 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
8980         * tree.c (warn_deprecated_use): Print file:line using locus color.
8981         * diagnostic.c (diagnostic_report_current_module): Print file:line
8982         and file:line:column using locus color.
8984 2013-05-14  Mike Stump  <mikestump@comcast.net>
8986         * gdbinit.in: Add __null.
8988 2013-05-14  Mike Stump  <mikestump@comcast.net>
8990         * recog.h: Rename struct recog_data to Recog_data.
8991         * recog.c: Likewise.
8992         * reload.c (can_reload_into): Likewise.
8993         * config/picochip/picochip.c: Likewise.
8995 2013-05-14  Mike Stump  <mikestump@comcast.net>
8997         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8999 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
9001         * resource.h (struct resources): Remove unch_memory member.
9002         (CLEAR_RESOURCE): Don't clear unch_memory.
9003         * resource.c (mark_referenced_resources): Don't set it.
9004         (mark_set_resources): Likewise.
9005         (mark_target_live_regs): Don't clear it.
9006         (init_resource_info): Likewise.
9007         * reorg.c (resource_conflicts_p): Don't compare it.
9008         (redundant_insn): Don't set it.
9010         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
9011         Remove prototypes.
9012         * emit-rtl.c (next_label): Remove unused function.
9013         (skip_consecutive_labels, link_cc0_insns): Move to ...
9014         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
9015         only place where these functions are used, and make them static.
9017 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9019         * fold-const.c (fold_negate_expr): Handle vectors.
9020         (fold_truth_not_expr): Make it static.
9021         (fold_invert_truthvalue): New static function.
9022         (invert_truthvalue_loc): Handle vectors. Do not call
9023         fold_truth_not_expr directly.
9024         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
9025         <TRUTH_NOT_EXPR>: Do not cast to boolean.
9026         (fold_comparison): Handle vector constants.
9027         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
9028         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
9029         * tree.h (fold_truth_not_expr): Remove declaration.
9031 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
9033         * config/aarch64/aarch64-simd.md
9034         (aarch64_vcond_internal<mode>): Rename to...
9035         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
9036         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
9037         float modes. Clarify all iterator modes.
9038         (vcond<mode><mode>): Use new name for vcond expanders.
9039         (vcond<v_cmp_result><mode>): Likewise.
9040         (vcondu<mode><mode>: Likewise.
9041         * config/aarch64/iterators.md (VDQF_COND): New.
9043 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9045         PR bootstrap/57266
9046         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9047         variable for the shift amount. Check that we shift by non-negative
9048         amounts.
9050 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
9052         PR target/42017
9053         * config/arm/arm.h (EPILOGUE_USES): Only return true
9054         for LR_REGNUM after epilogue_completed.
9056 2013-05-14  Joern Rennecke <joern.rennecke@embecosm.com>
9058         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9059         is error_mark_node.
9061 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9063         PR target/57261
9064         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9065         and Solaris 11+/x86 with gld.
9066         * configure: Regenerate.
9068 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9070         * expmed.c (expand_shift_1): Canonicalize rotates by
9071         constant bitsize / 2 to bitsize - 1.
9072         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9073         case ROTATERT>: Likewise.
9075         Revert:
9076         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9078         * config/i386/i386.md (rotateinv): New code attr.
9079         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9080         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9081         roll $31, %eax, etc.
9083 2013-05-14  Richard Biener  <rguenther@suse.de>
9085         PR middle-end/57235
9086         * tree-eh.c (sink_clobbers): Give up for successors with
9087         multiple predecessors and no virtual uses.
9089 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
9091         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9092         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9094 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9096         PR middle-end/57251
9097         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9098         the case when both op0 and op1 have VOIDmode.
9100 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9102         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9103         in multiply-accumulate mode.
9105 2013-05-13  Guozhi Wei  <carrot@google.com>
9107         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9109 2013-05-13  Kai Tietz  <ktietz@redhat.com>
9111         PR target/56975
9112         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9113         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9114         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9115         * config/i386/i386.c (ix86_option_override_internal): Likewise.
9116         (ix86_expand_prologue): Likewise.
9117         (ix86_expand_split_stack_prologue): Likewise.
9118         (legitimate_pic_address_disp_p): Likewise.
9119         (legitimize_pic_address): Likewise.
9120         (legitimize_tls_address): Likewise.
9121         (legitimize_pe_coff_symbol): Likewise.
9122         (output_pic_addr_const): Likewise.
9123         (construct_plt_address): Likewise.
9124         (ix86_expand_call): Likewise.
9125         (x86_output_mi_thunk): Likewise.
9126         (x86_function_profiler): Likewise.
9128 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
9130         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9131         similar switch cases.
9132         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9133         (aarch64_simd_mov_to_<mode>low): Delete.
9134         (aarch64_simd_mov_to_<mode>high): Delete.
9135         (move_lo_quad_<mode>): Add w<-r alternative.
9136         (aarch64_simd_move_hi_quad_<mode>): Likewise.
9137         (aarch64_simd_mov_from_*): Update type attribute.
9138         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9139         statement.
9141 2013-05-13  Jan Hubicka  <jh@suse.cz>
9143         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9144         * config/i386/i386.c (ix86_compute_frame_layout,
9145         ix86_expand_epilogue, emit_i387_cw_initialization,
9146         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9147         ix86_local_alignment): Fix use of size/speed predicates.
9149 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9151         PR tree-optimization/45216
9152         PR tree-optimization/57157
9153         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9154         the (-Y) & (B - 1) variant if OP is |.
9155         * expmed.c (expand_shift_1): For rotations by const0_rtx just
9156         return shifted.  Use (-op1) & (prec - 1) as other_amount
9157         instead of prec - op1.
9159 2013-05-13  Martin Jambor  <mjambor@suse.cz>
9161         PR middle-end/42371
9162         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9163         (ipa_constant_data): New type.
9164         (ipa_jump_func): Use ipa_constant_data to hold information about
9165         constant jump functions.
9166         (ipa_get_jf_constant): Adjust to jump function type changes.
9167         (ipa_get_jf_constant_rdesc): New function.
9168         (ipa_param_descriptor): New field controlled_uses.
9169         (ipa_get_controlled_uses): New function.
9170         (ipa_set_controlled_uses): Likewise.
9171         * ipa-ref.h (ipa_find_reference): Declare.
9172         * ipa-prop.c (ipa_cst_ref_desc): New type.
9173         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9174         changes.
9175         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
9176         New parameter cs.  Adjust all callers.
9177         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9178         (remove_described_reference): New function.
9179         (jfunc_rdesc_usable): Likewise.
9180         (try_make_edge_direct_simple_call): Decrement controlled use count,
9181         attempt to remove reference if it hits zero.
9182         (combine_controlled_uses_counters): New function.
9183         (propagate_controlled_uses): Likewise.
9184         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9185         (ipa_edge_duplication_hook): Duplicate reference descriptions.
9186         (ipa_print_node_params): Print described use counter.
9187         (ipa_write_jump_function): Adjust to jump function type changes.
9188         (ipa_read_jump_function): New parameter CS, pass it to
9189         ipa_set_jf_constant.  Adjust caller.
9190         (ipa_write_node_info): Stream controlled use count
9191         (ipa_read_node_info): Likewise.
9192         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9193         asserting.
9194         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9195         count.  Remove cloning-added reference if it reaches zero.
9196         * ipa-ref.c (ipa_find_reference): New function.
9198 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9200         * config/i386/i386.c (processor_target_table): Modified default
9201         alignment values for AMD BD and BT architectures.
9203 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
9205         * tree-vect-generic.c (uniform_vector_p): Move ...
9206         * tree.c (uniform_vector_p): ... here.
9207         * tree.h (uniform_vector_p): Declare it.
9208         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9209         into a scalar.
9211 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9213         PR tree-optimization/57230
9214         * tree-ssa-strlen.c (handle_char_store): Record length for
9215         array store from STRING_CST.
9217         PR tree-optimization/57230
9218         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9219         check.
9221 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
9223         * config/epiphany/epiphany.c (epiphany_init): Check size of
9224         NUM_MODES_FOR_MODE_SWITCHING.
9225         (epiphany_expand_prologue):
9226         Remove CONFIG_REGNUM initial value handling code.
9227         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9228         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9229         (emit_set_fp_mode, epiphany_mode_after): Likewise.
9230         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9231         Don't return 1 for FP_MODE_NONE.
9232         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9233         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9234         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9235         * config/epiphany/epiphany.md (save_config): New pattern.
9237 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
9239         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9241 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
9243         * config/i386/i386.md (memory): Handle sseishft1.
9244         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9245         (*vec_extractv2di_1): Ditto.
9247 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
9249         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
9250         saved registers.
9252 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
9254         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
9255         Add mthumb/march=armv7-a multilib.
9256         Add mthumb/march=armv7-r multilib.
9257         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
9259 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
9261         * config/v850/t-rtems: Add more multilibs.
9263 2013-05-10  Richard Biener  <rguenther@suse.de>
9265         PR tree-optimization/57214
9266         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
9267         not propagate from SSA names that occur in abnormal PHI nodes.
9269 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
9271         * stor-layout.c (element_precision): New function.
9272         * machmode.h (element_precision): Declare it.
9273         * tree.c (build_minus_one_cst): New function.
9274         (element_precision): Likewise.
9275         * tree.h (build_minus_one_cst): Declare new function.
9276         (element_precision): Likewise.
9277         * fold-const.c (operand_equal_p): Use element_precision.
9278         (fold_binary_loc): Handle vector types.
9279         * convert.c (convert_to_integer): Use element_precision.
9280         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
9281         separately.
9283 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
9285         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
9286         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
9287         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
9288         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9289         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
9290         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
9291         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
9292         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9293         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
9294         (Uuw8): New constraints.
9295         (Usb4): Move into alphabetical order.
9296         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
9297         (sd8_operand, ub8_operand, uw8_operand): New predicates.
9298         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
9299         previously unnamed patterns.
9300         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
9301         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
9302         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
9303         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
9304         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
9305         of specific lengths.
9307 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9309         * config/i386/i386.md (rotateinv): New code attr.
9310         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9311         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9312         roll $31, %eax, etc.
9314         PR tree-optimization/45216
9315         PR tree-optimization/57157
9316         * tree-ssa-forwprop.c (simplify_rotate): New function.
9317         (ssa_forward_propagate_and_combine): Call it.
9319 2013-05-10  Richard Biener  <rguenther@suse.de>
9321         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
9322         disable peeling when we version for aliasing.
9323         (vector_alignment_reachable_p): Honor explicit user alignment.
9324         (vect_supportable_dr_alignment): Likewise.
9325         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
9326         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
9327         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
9328         then peeling to arrange for the cost-model check to come first.
9330 2013-05-10  Alan Modra  <amodra@gmail.com>
9332         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
9333         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
9334         * configure: Regenerate.
9336 2013-05-10  Alan Modra  <amodra@gmail.com>
9338         PR target/55033
9339         * varasm.c (default_elf_select_section): Move !DECL_P check..
9340         (get_named_section): ..to here before calling get_section_name.
9341         Adjust assertion.
9342         (default_section_type_flags): Add DECL_P check.
9343         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
9344         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
9346 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
9348         * config/epiphany/epiphany.c (epiphany_expand_prologue):
9349         When using gen_stack_adjust_str with a register offset, add a
9350         REG_FRAME_RELATED_EXPR note.
9352 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
9354         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
9355         (*vec_extractv4si_zext_mem): Ditto.
9356         (*vec_extractv2di): Add 0->x and x->x alternatives.
9357         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
9358         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
9360 2013-05-09  Jason Merrill  <jason@redhat.com>
9362         N3639 C++1y VLA support
9363         * gimplify.c (gimplify_vla_decl): Don't touch an existing
9364         DECL_VALUE_EXPR.
9366         * tree.c (build_constructor_va): New.
9367         * tree.h: Declare it.
9369 2013-05-09  Martin Jambor  <mjambor@suse.cz>
9371         PR lto/57084
9372         * gimple-fold.c (canonicalize_constructor_val): Call
9373         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
9375 2013-05-09  Jan Hubicka  <jh@suse.cz>
9376             Richard Biener  <rguenther@suse.de>
9378         PR lto/54095
9379         * symtab.c (symtab_make_decl_local): Do not add private names.
9381 2013-05-09  Jan Hubicka  <jh@suse.cz>
9383         PR lto/54095
9384         * symtab.c (insert_to_assembler_name_hash): Handle clones.
9385         (unlink_from_assembler_name_hash): Likewise.
9386         (symtab_prevail_in_asm_name_hash, symtab_register_node,
9387         symtab_unregister_node, symtab_initialize_asm_name_hash,
9388         change_decl_assembler_name): Update.
9390 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
9392         * config/aarch64/aarch64.md: New movtf split.
9393         (*movtf_aarch64): Update.
9394         (aarch64_movdi_tilow): Handle TF modes and rename to
9395         aarch64_movdi_<mode>low.
9396         (aarch64_movdi_tihigh): Handle TF modes and rename to
9397         aarch64_movdi_<mode>high
9398         (aarch64_movtihigh_di): Handle TF modes and rename to
9399         aarch64_mov<mode>high_di
9400         (aarch64_movtilow_di): Handle TF modes and rename to
9401         aarch64_mov<mode>low_di
9402         (aarch64_movtilow_tilow): Remove spurious whitespace.
9403         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
9404         splits.
9405         (aarch64_print_operand): Update.
9407 2013-05-09  Alan Modra  <amodra@gmail.com>
9409         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
9410         powerpc64le.
9411         * configure: Regenerate.
9413 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
9415         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
9416         splitter preparation statements.
9417         * config/i386/sse.md (*vec_extract* splitters): Ditto.
9418         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
9419         adjust_address_nv.
9421 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9423         * gimple-ssa-strength-reduction.c (count_candidates): Change
9424         return value to int.
9425         (analyze_candidates_and_replace): Change type of length to int.
9427 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
9429         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
9430         (*vec_extract<mode>): Use VI12_128 mode iterator.
9431         (*vec_extract<mode>_mem): Ditto.
9432         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
9433         attribute.
9435 2013-05-08  Diego Novillo  <dnovillo@google.com>
9437         PR bootstrap/54659
9439         Revert:
9440         2012-08-17  Diego Novillo  <dnovillo@google.com>
9442         PR bootstrap/54281
9443         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
9444         * config.in: Regenerate.
9445         * configure: Regenerate.
9446         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
9448 2013-05-08  Jan Hubicka  <jh@suse.cz>
9450         PR lto/54095
9451         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
9452         * cgraph.h (symtab_node_base): Add unique_name.
9453         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9454         input_overwrite_node, input_varpool_node): Stream unique_name.
9455         * cgraphclones.c (cgraph_create_virtual_clone,
9456         cgraph_function_versioning): Set unique_name.
9457         * ipa.c (function_and_variable_visibility): Set unique_name.
9459 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9461         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
9462         (alloc_cand_and_find_basis): Restrict conditional candidate
9463         processing to CAND_MULTs.
9465 2013-05-08  Jan Hubicka  <jh@suse.cz>
9467         PR lto/54095
9468         lto-symtab.c (lto_symtab_symbol_p): New function.
9469         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
9470         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
9471         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
9472         Skip static symbols.
9474 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
9476         PR tree-optimization/57200
9477         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
9478         Only call inform if the preceding warning_at returns true.
9480 2013-05-07  Han Shen  <shenhan@google.com>
9482         * cfgexpand.c (record_or_union_type_has_array_p): New function.
9483         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
9484         * common.opt (fstack-protector-strong): New option.
9485         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
9486         * doc/invoke.texi (Optimization Options): Document
9487         "-fstack-protector-strong".
9488         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
9490 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9492         * config/mips/mips.c (mips_machine_reorg2): Return 0.
9494 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
9496         * ira.c (update_equiv_regs): Add insn having equiv memory even if
9497         it is not lhs of the insn.
9498         (setup_reg_equiv): Remove insn having equiv memory which it is not
9499         lhs of the insn.
9500         * lra-constraints.c (process_address): Try to improve generation
9501         code for address base + disp.
9502         (lra_constraints): Make correct the code for checking insn setting
9503         up backward equivalence.  Remove insn only if it is in the init
9504         insn list.
9505         * lra-eliminations.c (update_reg_eliminate): Change return value.
9506         (lra_eliminate): Use the result.
9508 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
9510         * config/i386/sse.md (ssescalarnummask): New mode attribute.
9511         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
9512         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
9513         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
9514         register target operands.
9515         (*vec_extractv8hi_sse2): New pattern.
9516         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
9517         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
9518         (*vec_extract<mode>_mem): New insn and split pattern.
9520 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
9522         * config/arm/arm.c (arm_asan_shadow_offset): New function.
9523         (TARGET_ASAN_SHADOW_OFFSET): Define.
9524         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
9525         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
9527 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9529         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
9530         (incr_vec_index): Return -1 if increment not found.
9531         (create_add_on_incoming_edge): Assert if increment not found.
9532         (record_increment): Limit number of increments recorded.
9533         (all_phi_incrs_profitable): Return false if an increment not found.
9534         (replace_profitable_candidates): Don't process increments that were
9535         not recorded.
9536         (analyze_candidates_and_replace): Limit size of incr_vec.
9538 2013-05-07  Richard Biener  <rguenther@suse.de>
9540         * calls.c (special_function_p): setjmp-like functions are leaf.
9541         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
9542         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
9544 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
9546         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
9547         (aarch64_simd_mov<mode>): New expander.
9548         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
9549         (aarch64_simd_mov_to_<mode>high): Likewise.
9550         (aarch64_simd_mov_from_<mode>low): Likewise.
9551         (aarch64_simd_mov_from_<mode>high): Likewise.
9552         (aarch64_dup_lane<mode>): Update.
9553         (aarch64_dup_lanedi): New instruction pattern.
9554         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
9555         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
9557 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9559         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
9560         (replace_mult_candidate): Remove unnecessary argument; remove
9561         unnecessary parameter from call to introduce_cast_before_cand.
9562         (replace_unconditional_candidate): Remove unnecessary parameter
9563         from call to replace_mult_candidate.
9564         (replace_conditional_candidate): Likewise.
9565         (insert_initializers): Use make_temp_ssa_name.
9566         (introduce_cast_before_cand): Remove unnecessary argument; use
9567         make_temp_ssa_name.
9568         (replace_one_candidate): Remove unnecessary argument; remove
9569         unnecessary parameter from calls to introduce_cast_before_cand.
9570         (replace_profitable_candidates): Remove unnecessary parameters
9571         from calls to replace_one_candidate.
9573 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9575         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9576         phi def as possibly hiding a basis for a CAND_ADD whose operands
9577         have been commuted in the analysis.
9578         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
9580 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
9582         * config/aarch64/aarch64.md
9583         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
9584         shift value between 0-4.
9586 2013-05-07  Richard Biener  <rguenther@suse.de>
9588         * double-int.h (rshift): New overload.
9589         * double-int.c (rshift): New function.
9590         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
9591         (create_reference_ops_from_ref): Remove.
9592         (vn_reference_insert): Use shared ops for constructing the
9593         reference and copy it.
9595 2013-05-07  Richard Biener  <rguenther@suse.de>
9597         PR middle-end/57190
9598         * tree-eh.c (sink_clobbers): Properly propagate
9599         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
9601 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
9603         PR tree-optimization/57149
9604         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
9605         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
9606         collect_phi_def_edges, execute_late_warn_uninitialized): Use
9607         uninit_undefined_value_p instead of ssa_undefined_value_p.
9609         PR debug/57184
9610         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
9611         for modifier == EXPAND_INITIALIZER.
9613 2013-05-07  Anton Blanchard  <anton@samba.org>
9615         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
9616         for powerpc64 little endian.
9617         * configure: Regenerate.
9619 2013-05-06  Graham Stott  <grahams@btinternet.com>
9621         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
9622         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
9623         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
9624         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
9626 2013-05-06  Graham Stott  <grahams@btinternet.com>
9628         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
9629         codes which allow non-lvalues.
9631 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
9633         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
9634         components are all 1s.
9635         (integer_minus_onep): New function.
9636         * tree.h (integer_minus_onep): Declare it.
9637         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
9638         integer_minus_onep instead of integer_all_onesp.
9640 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9642         PR target/52933
9643         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
9644         variations of these patterns.
9646 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9648         * config/i386/i386.md (isa): Add x64_sse4 member.
9649         (enabled): Handle x64_sse4.
9650         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
9651         instruction for 64bit SSE4_1 targets.  Update insn attributes.
9652         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
9653         instruction for SSE4_1 targets.  Update insn attributes.
9654         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9655         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
9656         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
9657         const_1 selector.
9658         (*vec_extractv4si): Rename from *sse4_1_pextrd.
9659         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
9660         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
9662 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9664         PR target/57108
9665         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
9667 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
9669         * final.c (do_assembler_dialects): Don't handle curly braces and
9670         vertical bar escaped by % as dialect delimiters.
9671         (output_asm_insn): Print curly braces and vertical bar if escaped
9672         by % and ASSEMBLER_DIALECT defined.
9673         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
9674         * doc/tm.texi: Regenerated.
9676 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9678         * config/mips/mips.c: Include tree-pass.h.
9679         (mips_reorg): Split in pre- and post-dbr_schedule parts.
9680         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
9681         (pass_mips_machine_reorg2): New machine specific pass.
9682         (insert_pass_mips_machine_reorg2): New pass plugin definition.
9683         (mips_option_override): Register the new pass.
9684         * rtl.h (cleanup_barriers): Remove prototype.
9685         (dbr_schedule): Likewise.
9686         * jump.c (cleanup_barriers): Make static.
9687         * reorg.c (dbr_schedule): Likewise.
9689 2013-05-06  Richard Biener  <rguenther@suse.de>
9691         PR tree-optimization/57185
9692         * tree-parloops.c (add_field_for_reduction): Handle anonymous
9693         SSA names properly.
9695 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
9697         PR target/57106
9698         * config/i386/i386.c (add_parameter_dependencies): Add dependence
9699         between "first_arg" and "insn", not "last" and "insn".
9701 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9703         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
9704         (find_candidates_in_block): Re-enable slsr_process_phi.
9705         (create_phi_basis): Fix double counting of candidate adjustment.
9707 2013-05-06  Richard Biener  <rguenther@suse.de>
9709         PR middle-end/57147
9710         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
9711         the edge is also fallthru, preserve it and just clear the
9712         abnormal flag.
9713         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
9714         also complex, preserve that and just clear the fallthru flag.
9715         * tree-inline.c (update_ssa_across_abnormal_edges): Also
9716         update virtual operands.
9718 2013-05-06  Alan Modra  <amodra@gmail.com>
9720         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
9721         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
9722         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
9723         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
9724         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9725         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
9727 2013-05-06  Alan Modra  <amodra@gmail.com>
9729         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
9730         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
9731         (DEFAULT_ASM_ENDIAN): Define.
9732         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9733         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
9734         Update -K PIC clause from sysv4.h.
9735         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
9736         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
9738 2013-05-06  Alan Modra  <amodra@gmail.com>
9740         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
9741         twice for little-endian.
9742         (ashrdi3_no_power, ashrdi3): Support little-endian.
9744 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
9746         PR target/55303
9747         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
9748         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
9749         related expanders.
9750         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
9751         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
9752         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
9753         New predicates.
9755 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
9756             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
9758         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
9759         * config/pa/pa.opt: Make mbig-switch a no-op.
9760         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
9761         (CASE_VECTOR_MODE): Always return SImode.
9762         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
9763         for the !TARGET_BIG_SWITCH case.
9764         * config/pa/pa-linux.h: Likewise.
9765         * config/pa/pa-openbsd.h: Likewise.
9766         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
9767         * config/pa/pa.md (short_jump): Remove define_insn.
9768         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
9769         (casesi0): Remove define_insn.
9770         (type): Remove btable_branch.
9771         (pa_combine_type): Likewise.
9772         (in_nullified_branch_delay): Likewise.
9773         (in_call_delay): Likewise.
9774         (define_delay): Likewise.
9775         (define_insn_reservation "Z3"): Likewise.
9776         (define_insn_reservation "Z4"): Likewise.
9777         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
9778         (pa_adjust_insn_length): Remove adjustment for btable branches.
9779         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
9780         and mno-big-switch
9782 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
9784         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9785         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
9786         Add m->r,x alternatives.
9787         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
9788         splitters using SWI48x mode iterator.
9789         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
9790         TARGET_64BIT.  Add m->x alternative.
9791         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
9792         Add o->x alternative.  Enable for TARGET_SSE.
9793         (sse_storeq): Remove expander.
9794         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
9795         with memory input operand.
9796         (*vec_extractv2di_1 splitter): New.
9797         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
9798         * config/i386/i386.md (ssevecmodelower): New mode attribute.
9800 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9802         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
9803         (INT_LOWPART): Delete.
9804         (extract_MB): Adjust.
9805         (extract_ME): Adjust.
9806         (print_operand): Adjust.
9808 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
9810         * config/rs6000/predicates.md (reg_or_add_cint_operand,
9811         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9812         (reg_or_logical_cint_operand, easy_fp_constant,
9813         logical_const_operand): Delete "CONST_DOUBLE" case.
9814         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
9815         "HOST_BITS_PER_WIDE_INT == 64" test.
9816         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
9817         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
9818         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
9819         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
9820         test.
9821         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
9822         CONST_DOUBLE DImode/VOIDmode case.
9823         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
9824         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9825         CONST_DOUBLE VOIDmode case.
9826         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9827         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
9828         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
9829         Delete CONST_DOUBLE case.
9830         (splitters for mov FMOVE64 const_double): Delete
9831         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
9832         "HOST_BITS_PER_WIDE_INT >= 64" test.
9833         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
9834         case.
9835         (mov DI const_double): Delete.
9837 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
9839         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
9840         on op shows all bits zero in mode of a lowpart subreg, return zero.
9842 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
9844         PR target/57150
9845         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
9846         to save TFmode registers and DImode to save TImode registers for
9847         caller save operations.
9848         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
9849         mark being partially clobbered since they only use the first
9850         double word.
9852         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
9853         and TDmode only use the upper 64-bits of each VSX register.
9855 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9857         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
9858         (find_candidates_in_block): Disable slsr_process_phi.
9860 2013-05-03  Guozhi Wei  <carrot@google.com>
9862         * coverage.c (coverage_obj_init): Move the construction of gcov
9863         constructor to ...
9864         (build_init_ctor): ... here.
9866 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9868         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
9869         (slsr_cand_d): Redefine def_phi.
9870         (stride_status, phi_adjust_status, count_phis_status): New enums.
9871         (find_phi_def): New.
9872         (find_basis_for_base_expr): New.
9873         (find_basis_for_candidate): Handle hidden bases.
9874         (alloc_cand_and_find_basis): Handle phi candidates.
9875         (slsr_process_phi): New.
9876         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
9877         (create_mul_imm_cand): Likewise.
9878         (create_add_ssa_cand): Exclude phi base candidates.
9879         (create_add_imm_cand): Likewise.
9880         (slsr_process_cast): Likewise.
9881         (slsr_process_copy): Likewise.
9882         (find_candidates_in_block): Handle phi candidates.
9883         (dump_candidate): Likewise.
9884         (unconditional_cands): Delete.
9885         (unconditional_cands_with_known_stride_p): Delete.
9886         (phi_dependent_cand_p): New.
9887         (cand_increment): Handle phi-dependent candidates.
9888         (replace_dependent): Delete.
9889         (replace_mult_candidate): New.
9890         (replace_unconditional_candidate): New.
9891         (incr_vec_index): Move to avoid forward reference.
9892         (create_add_on_incoming_edge): New.
9893         (create_phi_basis): New.
9894         (replace_dependents): Delete.
9895         (replace_conditional_candidate): New.
9896         (phi_add_costs): New.
9897         (replace_uncond_cands_and_profitable_phis): New.
9898         (record_increment): Handle phi adjustments.
9899         (record_phi_increments): New.
9900         (record_increments): Handle phi adjustments.
9901         (phi_incr_cost): New.
9902         (lowest_cost_path): Handle phis.
9903         (total_savings): Likewise.
9904         (analyze_increments): Likewise.
9905         (ncd_with_phi): New.
9906         (ncd_of_cand_and_phis): New.
9907         (nearest_common_dominator_for_cands): Handle phi increments.
9908         (all_phi_incrs_profitable): New.
9909         (replace_profitable_candidates): Handle phi-dependent candidates.
9910         (analyze_candidates_and_replace): Likewise.
9912 2013-05-03  Teresa Johnson  <tejohnson@google.com>
9914         PR bootstrap/57154
9915         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
9916         do not exceed REG_BR_PROB_BASE.
9918 2013-05-03  Jeff Law  <law@redhat.com>
9920         PR tree-optimization/57144
9921         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
9922         operand of the condition will bit into the new type when eliminating
9923         a cast feeding a condition.
9925 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
9927         PR rtl-optimization/57130
9928         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
9929         of COMPARE as in_code to the recursive call if needed.
9931 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
9933         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
9934         (enabled): Handle new members.
9935         * config/i386/sse.md (*vec_concatv2si): Merge from
9936         *vec_concatv2si_sse2 and vec_concatv2si_sse.
9937         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
9939 2013-05-03  Joern Rennecke <joern.rennecke@embecosm.com>
9941         PR tree-optimization/57027
9942         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
9943         for fnms opportunity, check we got the prerequisite kind
9944         of tree / gimple before using accessor functions.
9946 2013-05-03  Richard Biener  <rguenther@suse.de>
9948         * double-int.h (lshift): New overload without precision
9949         and arith argument.
9950         (operator *=, operator +=, operator -=): Move ...
9951         * double-int.c (operator *=, operator +=, operator -=): ... here
9952         and implement more efficiently.
9953         (mul_double_with_sign): Remove.
9954         (lshift_double): Adjust to take unsinged shift argument, push
9955         dispatching code to callers.
9956         (mul_double_wide_with_sign): Add early out for callers that
9957         are not interested in high parts or overflow.
9958         (lshift): New function.
9959         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
9960         dispatch code here.
9961         (lrotate, rrotate): Use logical shifts.
9962         * expr.c (get_inner_reference): Use lshift.
9963         * fixed-value.c (do_fixed_divide): Likewise.
9964         * tree-dfa.c (get_ref_base_and_extent): Likewise.
9965         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
9966         (indirect_refs_may_alias_p): Likewise.
9967         (stmt_kills_ref_p_1): Likewise.
9969 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9971         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
9973 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
9975         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
9976         scalar form of FABD instruction.
9978 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
9980         * lra-constraints.c (process_alt_operands): Add checking alt
9981         number to choose the best alternative.
9983 2013-05-02  Richard Biener  <rguenther@suse.de>
9985         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9986         bitmap and its handling.
9987         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9989 2013-05-02  Richard Biener  <rguenther@suse.de>
9991         PR middle-end/57140
9992         * tree-inline.c (copy_loops): Properly handle removed loops.
9993         (copy_cfg_body): Mark destination loops for fixup if source
9994         loops needed fixup.
9996 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
9998         PR target/56732
9999         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
10000         generating simple_return for naked functions.
10002 2013-05-02  Martin Jambor  <mjambor@suse.cz>
10004         PR middle-end/56988
10005         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
10006         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
10007         flags match.
10008         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
10009         ipa_agg_replacement_value structures.
10010         (known_aggs_to_agg_replacement_list): Likewise.
10011         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
10012         (read_agg_replacement_chain): Likewise.
10013         (ipcp_transform_function): Also check that by_ref flags match.
10015 2013-05-02  Richard Biener  <rguenther@suse.de>
10017         * graphds.h (struct graph): Add obstack member.
10018         * graphds.c (new_graph): Initialize obstack and allocate
10019         vertices from it.
10020         (add_edge): Allocate edge from the obstack.
10021         (free_graph): Free the obstack instead of all edges and vertices.
10023 2013-05-02  Teresa Johnson  <tejohnson@google.com>
10025         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
10026         divides.
10027         * cfg.c (update_bb_profile_for_threading): Ditto.
10028         * tree-inline.c (copy_bb): Ditto.
10029         (copy_edges_for_bb): Ditto.
10030         (initialize_cfun): Ditto.
10031         (copy_cfg_body): Ditto.
10032         (expand_call_inline): Ditto.
10033         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
10034         (estimate_node_size_and_time): Ditto.
10035         (inline_merge_summary): Ditto.
10036         * cgraphclones.c (cgraph_clone_edge): Ditto.
10037         (cgraph_clone_node): Ditto.
10038         * sched-rgn.c (compute_dom_prob_ps): Ditto.
10039         (compute_trg_info): Ditto.
10041 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10043         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
10044         S reg when fp attribute set.
10045         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
10047 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10049         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
10050         New pattern.
10051         (*and_one_cmplsi3_compare0_uxtw): Likewise.
10052         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10053         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10055 2013-05-02  Richard Biener  <rguenther@suse.de>
10057         * tree-scalar-evolution.c (scev_info_hasher): Remove.
10058         (struct instantiate_cache_entry): New type.
10059         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10060         (struct instantiate_cache_type): New type.
10061         (set_instantiated_value, get_instantiated_value): Remove.
10062         (get_instantiated_value_entry): New function.
10063         (instantiate_scev_name): Use the new cache and adjust.
10064         (instantiate_scev_poly): Adjust.
10065         (instantiate_scev_binary): Likewise.
10066         (instantiate_array_ref): Likewise.
10067         (instantiate_scev_convert): Likewise.
10068         (instantiate_scev_not): Likewise.
10069         (instantiate_scev_3): Likewise.
10070         (instantiate_scev_2): Likewise.
10071         (instantiate_scev_r): Likewise.
10072         (instantiate_scev): Likewise.
10073         (resolve_mixers): Likewise.
10075 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
10077         PR target/57091
10078         * lra-constraints.c (best_small_class_operands_num): Remove.
10079         (process_alt_operands): Remove small_class_operands_num.  Take
10080         small classes operands into losers and only if the operand is not
10081         matched.  Modify debugging output.
10082         (curr_insn_transform): Remove best_small_class_operands_num.
10083         Print insn name.
10085 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10087         * config/aarch64/aarch64-builtins.c
10088         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10089         * config/aarch64/aarch64-simd-builtins.def
10090         (reduc_splus_): Add new modes.
10091         (reduc_uplus_): New.
10092         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10093         (reduc_uplus_v4sf): Likewise.
10094         (reduc_splus_v4sf): Likewise.
10095         (aarch64_addv<mode>): Likewise.
10096         (reduc_uplus_<mode>): Likewise.
10097         (reduc_splus_<mode>): Likewise.
10098         (aarch64_addvv2di): Likewise.
10099         (reduc_uplus_v2di): Likewise.
10100         (reduc_splus_v2di): Likewise.
10101         (aarch64_addvv2si): Likewise.
10102         (reduc_uplus_v2si): Likewise.
10103         (reduc_splus_v2si): Likewise.
10104         (reduc_<sur>plus_<mode>): New.
10105         (reduc_<sur>plus_v2di): Likewise.
10106         (reduc_<sur>plus_v2si): Likewise.
10107         (reduc_<sur>plus_v4sf): Likewise.
10108         (aarch64_addpv4sf): Likewise.
10109         * config/aarch64/arm_neon.h
10110         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10111         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10112         add UNSPEC_SADDV, UNSPEC_UADDV.
10113         (SUADDV): New.
10114         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10116 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10118         * config/aarch64/arm_neon.h
10119         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10121 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10123         * config/aarch64/aarch64-builtins
10124         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10126 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10128         * config/aarch64/aarch64-simd-builtins.def
10129         (reduc_smax_): New.
10130         (reduc_smin_): Likewise.
10131         (reduc_umax_): Likewise.
10132         (reduc_umin_): Likewise.
10133         (reduc_smax_nan_): Likewise.
10134         (reduc_smin_nan_): Likewise.
10135         (fmax): Remove.
10136         (fmin): Likewise.
10137         (smax): Update for V2SF, V4SF and V2DF modes.
10138         (smin): Likewise.
10139         (smax_nan): New.
10140         (smin_nan): Likewise.
10141         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10142         (<su><maxmin><mode>3): ...This, refactor.
10143         (s<maxmin><mode>3): New.
10144         (<maxmin_uns><mode>3): Likewise.
10145         (reduc_<maxmin_uns>_<mode>): Refactor.
10146         (reduc_<maxmin_uns>_v4sf): Likewise.
10147         (reduc_<maxmin_uns>_v2si): Likewise.
10148         (aarch64_<fmaxmin><mode>: Remove.
10149         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10150         new builtin names.
10151         (vmin<q>_f<32,64>): Likewise.
10152         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10153         (FMAXMIN): New.
10154         (su): Add mappings for smax, smin, umax, umin.
10155         (maxmin): New.
10156         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10157         (FMAXMIN): Rename as...
10158         (FMAXMIN_UNS): ...This.
10159         (maxminv): Remove.
10160         (fmaxminv): Likewise.
10161         (fmaxmin): Likewise.
10162         (maxmin_uns): New.
10163         (maxmin_uns_op): Likewise.
10165 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10167         * config/aarch64/arm_neon.h
10168         (vac<ge, gt><sd>_f<32, 64>): Rename to...
10169         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10170         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10172 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10174         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10175         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10177 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10179         * config/aarch64/aarch64-simd.md
10180         (vcond<mode>_internal): Handle special cases for constant masks.
10181         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10182         (vcondu<mode><mode>): Likewise.
10183         (vcond<v_cmp_result><mode>): New.
10185 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10187         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10188         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10189         * config/aarch64/aarch64-simd-builtins.def
10190         (cmeq): Update to BUILTIN_VALLDI.
10191         (cmgt): Likewise.
10192         (cmge): Likewise.
10193         (cmle): Likewise.
10194         (cmlt): Likewise.
10195         * config/aarch64/arm_neon.h
10196         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10197         to builtins or C as appropriate.
10199 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10201         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10202         (cmgeu): ...This.
10203         (cmhi): Rename to...
10204         (cmgtu): ...This.
10205         * config/aarch64/aarch64-simd.md
10206         (simd_mode): Add SF.
10207         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10208         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10209         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10210         (cstore<mode>_neg): ...This.
10211         * config/aarch64/iterators.md
10212         (VALLF): new.
10213         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10214         (COMPARISONS): New.
10215         (UCOMPARISONS): Likewise.
10216         (optab): Add missing comparisons.
10217         (n_optab): New.
10218         (cmp_1): Likewise.
10219         (cmp_2): Likewise.
10220         (CMP): Likewise.
10221         (cmp): Remove.
10222         (VCMP_S): Likewise.
10223         (VCMP_U): Likewise.
10224         (V_cmp_result): Add DF, SF modes.
10225         (v_cmp_result): Likewise.
10226         (v): Likewise.
10227         (vmtype): Likewise.
10228         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10230 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
10232         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10233         define_insn to define_insn_and_split.
10234         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10235         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10236         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10237         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10238         (thumb2_negscc): Likewise.
10240 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10242         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10244 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10246         * config/arm/thumb2.md: Remove trailing whitespaces.
10248 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
10250         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
10251         Use gen_int_mode rather than GEN_INT.
10253 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
10255         * value-prof.c (stream_in_histogram_value): Remove the strayed
10256         debug_gimple_stmt.
10258 2013-04-30  Richard Biener  <rguenther@suse.de>
10260         PR middle-end/57122
10261         * cfghooks.c (split_edge): Properly check for the loop latch edge.
10263 2013-04-30  Richard Biener  <rguenther@suse.de>
10265         PR middle-end/57107
10266         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
10268 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
10270         PR rtl-optimization/56957
10271         PR rtl-optimization/57105
10272         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
10273         variable.  Use just INSN_UID for determining whether an insn
10274         should be only disconnected from the insn stream.
10275         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
10277 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
10279         PR tree-optimization/57104
10280         * tsan.c (instrument_expr): Don't instrument accesses to
10281         DECL_HARD_REGISTER VAR_DECLs.
10283 2013-04-30  Richard Biener  <rguenther@suse.de>
10285         * function.h (loops_for_fn): New inline function.
10286         (set_loops_for_fn): Likewise.
10287         * cfgloop.h (place_new_loop): Add struct function parameter.
10288         (get_loop): Likewise.
10289         (get_loops): Likewise.
10290         (number_of_loops): Likewise.
10291         (fel_next): Adjust.
10292         (fel_init): Likewise.
10293         * cfg.c (get_loop_copy): Adjust.
10294         * cfgloop.c (flow_loops_dump): Likewise.
10295         (record_loop_exits): Likewise.
10296         (verify_loop_structure): Likewise.
10297         * cfgloopanal.c (mark_irreducible_loops): Likewise.
10298         (estimate_reg_pressure_cost): Likewise.
10299         (mark_loop_exit_edges): Likewise.
10300         * cfgloopmanip.c (place_new_loop): Likewise.
10301         (add_loop): Likewise.
10302         (duplicate_loop): Likewise.
10303         * graph.c (draw_cfg_nodes): Likewise.
10304         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
10305         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
10306         (extract_affine_chrec): Likewise.
10307         (build_scop_iteration_domain): Likewise.
10308         * graphite.c (graphite_initialize): Likewise.
10309         * ira-build.c (create_loop_tree_nodes): Likewise.
10310         (more_one_region_p): Likewise.
10311         (rebuild_regno_allocno_maps): Likewise.
10312         (mark_loops_for_removal): Likewise.
10313         (mark_all_loops_for_removal): Likewise.
10314         (remove_unnecessary_regions): Likewise.
10315         (ira_build): Likewise.
10316         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
10317         * loop-init.c (fix_loop_structure): Likewise.
10318         (gate_rtl_move_loop_invariants): Likewise.
10319         (gate_rtl_unswitch): Likewise.
10320         (gate_rtl_unroll_and_peel_loops): Likewise.
10321         (rtl_doloop): Likewise.
10322         * lto-streamer-in.c (input_cfg): Likewise.
10323         * lto-streamer-out.c (output_cfg): Likewise.
10324         * modulo-sched.c (sms_schedule): Likewise.
10325         * predict.c (tree_estimate_probability): Likewise.
10326         (tree_estimate_probability_driver): Likewise.
10327         (estimate_loops): Likewise.
10328         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
10329         (move_sese_region_to_fn): Likewise.
10330         (debug_loop_num): Likewise.
10331         * tree-chrec.c (chrec_evaluate): Likewise.
10332         (hide_evolution_in_other_loops_than_loop): Likewise.
10333         (chrec_component_in_loop_num): Likewise.
10334         (reset_evolution_in_loop): Likewise.
10335         (evolution_function_is_invariant_rec_p): Likewise.
10336         * tree-if-conv.c (main_tree_if_conversion): Likewise.
10337         * tree-inline.c (copy_loops): Likewise.
10338         (copy_cfg_body): Likewise.
10339         (tree_function_versioning): Likewise.
10340         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
10341         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
10342         Likewise.
10343         (add_to_evolution_1): Likewise.
10344         (scev_const_prop): Likewise.
10345         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
10346         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
10347         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
10348         (tree_ssa_lim_initialize): Likewise.
10349         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
10350         (verify_loop_closed_ssa): Likewise.
10351         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
10352         (tree_ssa_loop_im): Likewise.
10353         (tree_ssa_loop_unswitch): Likewise.
10354         (tree_vectorize): Likewise.
10355         (check_data_deps): Likewise.
10356         (tree_ssa_loop_ivcanon): Likewise.
10357         (tree_ssa_loop_bounds): Likewise.
10358         (tree_complete_unroll): Likewise.
10359         (tree_complete_unroll_inner): Likewise.
10360         (tree_parallelize_loops): Likewise.
10361         (tree_ssa_loop_prefetch): Likewise.
10362         (tree_ssa_loop_ivopts): Likewise.
10363         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
10364         * tree-vectorizer.c (vectorize_loops): Likewise.
10366 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
10368         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
10369         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
10370         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
10371         with EABI_LINK_SPEC.
10373 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10375         PR target/44578
10376         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
10377         alternative.
10379 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
10381         PR target/57097
10382         * lra-constraints.c (process_alt_operands): Discourage a bit more
10383         using memory for pseudos.  Print cost dump for alternatives.
10384         Modify cost values for conflicts with early clobbers.
10385         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
10387 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10389         PR target/57098
10390         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
10392 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
10394         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
10395         from/to S register.
10396         (movdi_aarch64): Support LDR/STR from/to D register.
10398 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
10400         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
10401         or higher by default.
10403 2013-04-29  Richard Biener  <rguenther@suse.de>
10405         PR middle-end/57075
10406         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
10407         even if not adding abnormal edges for calls that can make
10408         abnormal gotos.
10410 2013-04-29  Richard Biener  <rguenther@suse.de>
10412         PR middle-end/57103
10413         * tree-cfg.c (move_stmt_op): Fix condition under which to update
10414         TREE_BLOCK.
10415         (move_stmt_r): Remove redundant checking.
10417 2013-04-29  Teresa Johnson  <tejohnson@google.com>
10419         PR bootstrap/57077
10420         * basic-block.h (apply_scale): New function.
10421         (apply_probability): Use apply_scale.
10422         * gimple-streamer-in.c (input_bb): Ditto.
10423         * lto-streamer-in.c (input_cfg): Ditto.
10424         * lto-cgraph.c (merge_profile_summaries): Ditto.
10425         * tree-optimize.c (execute_fixup_cfg): Ditto.
10426         * tree-inline.c (copy_bb): Update comment to use apply_scale.
10427         (copy_edges_for_bb): Ditto.
10428         (copy_cfg_body): Ditto.
10430 2013-04-29  Tom de Vries  <tom@codesourcery.com>
10432         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
10433         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
10434         (tail_merge_optimize): Handle current_loops == NULL.
10436 2013-04-26  Jeff Law  <law@redhat.com>
10438         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
10439         (simplify_cond_using_ranges): Generalize code to simplify
10440         COND_EXPRs where one argument is a constant and the other
10441         is an SSA_NAME created by an integral type conversion.
10443 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10445         * config/arm/arm.md (store_minmaxsi): Use only when
10446         optimize_insn_for_size_p.
10448 2013-04-29  Christian Bruel  <christian.bruel@st.com>
10450         PR target/57108
10451         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
10453 2013-04-29  Richard Biener  <rguenther@suse.de>
10455         PR middle-end/57089
10456         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
10457         loop tree make sure to schedule a fixup for the child as well.
10458         (expand_omp_for_generic): Properly add loops.
10459         (expand_omp_for_static_nochunk): Likewise.
10460         (expand_omp_for_static_chunk): Likewise.
10461         (expand_omp_for): For the degenerate case fixup loops.
10462         (expand_omp_sections): Fix default bb placement in loops.
10463         (expand_omp_atomic_pipeline): Properly add loops.
10465 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10467         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
10469 2013-04-29  Tom de Vries  <tom@codesourcery.com>
10471         * tree-ssa-tail-merge.c: Update header comment.
10473 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10475         * config/aarch64/arm_neon.h
10476         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
10477         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
10478         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
10479         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
10480         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
10481         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
10482         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
10483         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
10485 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10487         * config/aarch64/aarch64-simd.md
10488         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
10489         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
10490         fix_trunc, fixuns_trunc.
10491         (ftrunc<VDQF:mode>2): New.
10492         * config/aarch64/iterators.md (optab): Add fix, fixuns.
10493         (fix_trunc_optab): New.
10495 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10497         * config/aarch64/aarch64-builtins.c
10498         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
10499         iceilf, lround, iroundf.
10501 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10503         PR target/54349
10504         * config/i386/i386.h (enum ix86_tune_indices)
10505         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
10506         New, split from X86_TUNE_INTER_UNIT_MOVES.
10507         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
10508         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
10509         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
10510         (TARGET_INTER_UNIT_MOVES): Remove.
10511         * config/i386/i386.c (initial_ix86_tune_features): Update.
10512         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
10513         (ix86_expand_convert_uns_didf_sse): Use
10514         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10515         (ix86_expand_vector_init_one_nonzero): Ditto.
10516         (ix86_expand_vector_init_interleave): Ditto.
10517         (inline_secondary_memory_needed): Return true for moves from SSE class
10518         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
10519         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
10520         * config/i386/constraints.md (Yi, Ym): Depend on
10521         TARGET_INTER_UNIT_MOVES_TO_VEC.
10522         (Yj, Yn): New constraints.
10523         * config/i386/i386.md (*movdi_internal): Change constraints of
10524         operand 1 from Yi to Yj and from Ym to Yn.
10525         (*movsi_internal): Ditto.
10526         (*movdf_internal): Ditto.
10527         (*movsf_internal): Ditto.
10528         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
10529         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10530         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
10531         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
10532         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
10533         * config/i386/sse.md (movdi_to_sse): Ditto.
10534         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
10535         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
10536         TARGET_INTER_UNIT_MOVES.
10537         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
10538         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
10539         instead of TARGET_INTER_UNIT_MOVES.
10540         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
10541         operand 1 from Yi to Yj and from Ym to Yn.
10543 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10545         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
10546         (float_truncate_hi_): Likewise.
10547         (float_extend_lo_): Likewise.
10548         (float_truncate_lo_): Likewise.
10549         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
10550         (aarch64_float_extend_lo_v2df): Likewise.
10551         (vec_unpacks_hi_v4sf): Likewise.
10552         (aarch64_float_truncate_lo_v2sf): Likewise.
10553         (aarch64_float_truncate_hi_v4sf): Likewise.
10554         (vec_pack_trunc_v2df): Likewise.
10555         (vec_pack_trunc_df): Likewise.
10557 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10559         * config/aarch64/aarch64-builtins.c
10560         (aarch64_fold_builtin): Fold float conversions.
10561         * config/aarch64/aarch64-simd-builtins.def
10562         (floatv2si, floatv4si, floatv2di): New.
10563         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
10564         * config/aarch64/aarch64-simd.md
10565         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
10566         * config/aarch64/iterators.md (FLOATUORS): New.
10567         (optab): Add float, floatuns.
10568         (su_optab): Likewise.
10570 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10572         * config/aarch64/aarch64-builtins.c
10573         (aarch64_builtin_vectorized_function): Use new names for
10574         fcvt builtins.
10575         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
10576         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
10577         (fcvtzu): Split as...
10578         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
10579         (fcvtas): Split as...
10580         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
10581         (fcvtau): Split as...
10582         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
10583         (fcvtps): Split as...
10584         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
10585         (fcvtpu): Split as...
10586         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
10587         (fcvtms): Split as...
10588         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
10589         (fcvtmu): Split as...
10590         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
10591         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
10592         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
10593         (lfrintnusf, lfrintnudf): Likewise.
10594         * config/aarch64/aarch64-simd.md
10595         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
10596         define_insn.
10597         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
10598         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
10599         (fcvt_pattern): Likewise.
10601 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10603         * config/aarch64/aarch64-simd.md
10604         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
10605         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
10607 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10609         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
10610         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
10611         (vrnd<a,m,n,p>_f32): Implement using builtins.
10612         (vrnd<i,x><q>_f<32, 64>): New.
10614 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10616         * config/aarch64/aarch64-builtins.c
10617         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
10618         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
10619         (frintz): Rename to...
10620         (btrunc): ...this.
10621         (frintp): Rename to...
10622         (ceil): ...this.
10623         (frintm): Rename to...
10624         (floor): ...this.
10625         (frinti): Rename to...
10626         (nearbyint): ...this.
10627         (frintx): Rename to...
10628         (rint): ...this.
10629         (frinta): Rename to...
10630         (round): ...this.
10631         * config/aarch64/aarch64-simd.md
10632         (aarch64_frint<frint_suffix><mode>): Delete.
10633         (<frint_pattern><mode>2): Convert to insn.
10634         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
10635         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
10636         (frint_pattern): Likewise.
10637         (frint_suffix): Likewise.
10639 2013-04-29  Richard Biener  <rguenther@suse.de>
10641         PR tree-optimization/57081
10642         * loop-init.c: Include tree-flow.h.
10643         (loop_optimizer_finalize): Free number of iteration estimates.
10644         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
10646 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
10648         PR tree-optimization/57083
10649         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
10650         non-singleton shift count range, zero extend low_bound for uns case.
10652         * config/i386/predicates.md (general_vector_operand): New predicate.
10653         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
10654         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
10655         if they aren't nonimmediate operands.  If their original values
10656         satisfy const_vector_equal_evenodd_p, don't shift them.
10657         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
10658         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
10659         if not nonimmediate_operand.
10660         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
10661         instead of register_operand.
10662         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
10664 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
10666         * stor-layout.c (finalize_size_functions): Allocate a structure and
10667         reset cfun before dumping the functions.
10669 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
10671         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
10673         PR target/56866
10674         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
10675         use xop_pmacsdqh if uns_p.
10676         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
10677         the immediate rotate count.
10679 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
10681         * rtl.h (struct rtx_def): Add comment for field jump.
10682         (LRA_SUBREG_P): New macro.
10683         * recog.c (register_operand): Check LRA_SUBREG_P.
10684         * lra.c (lra): Add note at the end of RTL code. Align non-empty
10685         stack frame.
10686         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10687         (lra_final_code_change): Skip subreg change for operators.
10688         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10689         if there are no operand changes.
10690         * lra-constraints.c (curr_insn_set): New.
10691         (match_reload): Set LRA_SUBREG_P.
10692         (emit_spill_move): Ditto.
10693         (check_and_process_move): Use curr_insn_set. Process only single
10694         set insns.  Don't initialize sec_mem_p and change_p.
10695         (simplify_operand_subreg): Use LRA_SUBREG_P.
10696         (reg_in_class_p): New function.
10697         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
10698         of #ifdef.  Add code to remove cycling.
10699         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10700         non-null disp.  Reload inner instead of disp when base and index
10701         are null.  Try to put lo_sum into register.
10702         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10703         (check_and_process_move): Move code for move cost check to
10704         simple_move_p.  Remove equiv_substitution.
10705         (simple_move_p): New function.
10706         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
10707         curr_insn_set.  Call check_and_process_move only for single set
10708         insns.  Use the new function.  Move call of check_and_process_move
10709         after operand equiv substitution and address process.
10711 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10713         PR go/57045
10714         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
10715         with nonlocal goto receivers or returns twice calls, ignore
10716         unininitialized values from abnormal edges to nl goto receiver
10717         or returns twice call.
10719 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10721         PR tree-optimization/57051
10722         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
10723         and VEC_RSHIFT_EXPR if shift count is a multiple of element
10724         bitsize.
10726 2013-04-26  Richard Biener  <rguenther@suse.de>
10728         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
10729         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
10730         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
10731         (fixup_loop_arrays_after_move): New function.
10732         (move_sese_region_to_fn): Properly outline the loop tree parts
10733         of the SESE region.
10735 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
10737         * config/i386/i386.md (type, unit): Fix long lines.
10739 2013-04-26  Richard Biener  <rguenther@suse.de>
10741         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
10742         (lto-streamer-out.o): Likewise.
10743         * cfgloop.c (init_loops_structure): Export, add struct function
10744         argument and adjust.
10745         (flow_loops_find): Adjust.
10746         * cfgloop.h (enum loop_estimation): Add EST_LAST.
10747         (init_loops_structure): Declare.
10748         * lto-streamer-in.c: Include cfgloop.h.
10749         (input_cfg): Input the loop tree.
10750         * lto-streamer-out.c: Include cfgloop.h.
10751         (output_cfg): Output the loop tree.
10752         (output_struct_function_base): Do not drop PROP_loops.
10754 2013-03-26  Richard Biener  <rguenther@suse.de>
10756         * tree-cfg.c (execute_build_cfg): Build the loop tree.
10757         (pass_build_cfg): Provide PROP_loops.
10758         (move_sese_region_to_fn): Remove loops that are outlined into fn
10759         for now.
10760         * tree-inline.c: Include cfgloop.h.
10761         (initialize_cfun): Do not drop PROP_loops.
10762         (copy_loops): New function.
10763         (copy_cfg_body): Copy loop structure.
10764         (tree_function_versioning): Initialize destination loop tree.
10765         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
10766         (pass_parallelize_loops): Do IL verification.
10767         * loop-init.c (loop_optimizer_init): Fixup loops if required.
10768         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
10769         the CFG make sure we fixup loops as well.
10770         * tree-ssa-tail-merge.c: Include cfgloop.h.
10771         (replace_block_by): When merging loop latches mark loops for fixup.
10772         * lto-streamer-out.c (output_struct_function_base): Drop
10773         PROP_loops for now.
10774         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
10775         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
10776         * ipa-split.c: Include cfgloop.h.
10777         (split_function): Add the new return block to the loop tree root.
10778         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
10779         whether we have removed the forwarder block.
10780         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
10781         * cfgloop.h (place_new_loop): Declare.
10782         * cfgloopmanip.c (place_new_loop): Export.
10783         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
10784         (tree-switch-conversion.o): Likewise.
10785         (tree-complex.o): Likewise.
10786         (tree-inline.o): Likewise.
10787         (tree-ssa-tailmerge.o): Likewise.
10788         (ipa-split.o): Likewise.
10789         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
10790         (tree-ssa-copy.o): Likewise.
10791         * tree-switch-conversion.c: Include cfgloop.h
10792         (process_switch): If we emit a bit-test cascade, schedule loops
10793         for fixup.
10794         * tree-complex.c: Include cfgloop.h.
10795         (expand_complex_div_wide): Properly add new basic-blocks to loops.
10796         * asan.c: Include cfgloop.h.
10797         (create_cond_insert_point): Properly add new basic-blocks to
10798         loops, schedule loop fixup.
10799         * cfgloop.c (verify_loop_structure): Check that looks are not
10800         marked for fixup.
10801         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
10802         to loops.
10803         (expand_omp_for_generic): Likewise.
10804         (expand_omp_sections): Likewise.
10805         (expand_omp_atomic_pipeline): Schedule loops for fixup.
10806         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
10807         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
10808         is initialized, not when loops are present.
10809         * tree-parloops.c (parallelize_loops): Remove checking here.
10810         * passes.c (init_optimization_passes): Schedule a copy-propagation
10811         pass before complete unrolling of inner loops.
10813 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
10815         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
10816         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
10817         (colorize_init): Add argument to _WIN32 version.
10818         * toplev.c: Include diagnostic-color.h.
10819         (process_options): Default to -fdiagnostics-color=auto if
10820         GCC_COLORS env var is in the environment.
10821         * common.opt (fdiagnostics-color=): Add Var and Init.
10822         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
10823         env var is in the environment, the default is auto rather than never.
10825         * diagnostic.h (file_name_as_prefix): Add context argument.
10826         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
10827         the string as locus.
10828         * langhooks.c (lhd_print_error_function): Adjust caller.
10830 2013-04-25  Lawrence Crowl  <crowl@google.com>
10832         * var-tracking.c (shared_hash_def::htab):
10833         Change type to hash_table.  Update dependent calls and types.
10835 2013-04-25  Lawrence Crowl  <crowl@google.com>
10837         * Makefile.in: Update as needed below.
10839         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
10840         Move declaration to after the type's method definitons.
10842         * attribs.c (htab_t scoped_attributes::attribute_hash):
10843         Change type to hash_table.  Update dependent calls and types.
10845         * bitmap.c (htab_t bitmap_desc_hash):
10846         Change type to hash_table.  Update dependent calls and types.
10848         * cselib.c (htab_t cselib_hash_table):
10849         Change type to hash_table.  Update dependent calls and types.
10851         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
10852         (hash_string_slot_node): Move implementation into lto-streamer.h
10853         struct string_slot_hasher.
10854         (eq_string_slot_node): Likewise.
10856         * data-streamer-out.c: Update output_block::string_hash_table
10857         dependent calls and types.
10859         * dwarf2cfi.c (htab_t trace_index):
10860         Change type to hash_table.  Update dependent calls and types.
10862         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
10863         Change type to hash_table.  Update dependent calls and types.
10864         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
10865         (htab_t optimize_external_refs::map): Likewise.
10866         (htab_t output_comp_unit::extern_map): Likewise.
10867         (htab_t output_comdat_type_unit::extern_map): Likewise.
10868         (htab_t output_macinfo::macinfo_htab): Likewise.
10869         (htab_t optimize_location_lists::htab): Likewise.
10870         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
10872         * except.c (htab_t ehspec_hash_type):
10873         Change type to hash_table.  Update dependent calls and types.
10874         (assign_filter_values::ttypes): Likewise.
10875         (assign_filter_values::ehspec): Likewise.
10876         (sjlj_assign_call_site_values::ar_hash): Likewise.
10877         (convert_to_eh_region_ranges::ar_hash): Likewise.
10879         * gcse.c (htab_t pre_ldst_table):
10880         Change type to hash_table.  Update dependent calls and types.
10882         * ggc-common.c (htab_t saving_htab):
10883         Change type to hash_table.  Update dependent calls and types.
10884         (htab_t loc_hash): Likewise.
10885         (htab_t ptr_hash): Likewise.
10886         (call_count): Rename ggc_call_count.
10887         (call_alloc): Rename ggc_call_alloc.
10888         (loc_descriptor): Rename make_loc_descriptor.
10889         (add_statistics): Rename ggc_add_statistics.
10891         * ggc-common.c (saving_htab):
10892         Change type to hash_table.  Update dependent calls and types.
10894         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
10895         (push_gimplify_context): Likewise.
10896         (pop_gimplify_context): Likewise.
10897         (struct gimple_temp_hash_elt): Added.
10898         (struct gimplify_hasher): Likewise.
10899         (struct gimplify_ctx.temp_htab):
10900         Change type to hash_table.  Update dependent calls and types.
10902         * gimple-fold.c: Include gimplify-ctx.h.
10904         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
10905         Change type to hash_table.  Update dependent calls and types.
10906         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
10907         avoid potential global name collision.
10909         * gimplify.c: Include gimplify-ctx.h.
10910         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
10911         (htab_t gimplify_ctx::temp_htab):
10912         Update dependent calls and types for new type hash_table.
10913         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
10914         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
10916         * gimplify-ctx.h: New.
10917         (struct gimple_temp_hash_elt): Move from gimplify.c.
10918         (class gimplify_hasher): New.
10919         (struct gimplify_ctx): Move from gimple.h.
10920         (htab_t gimplify_ctx::temp_htab):
10921         Change type to hash_table.  Update dependent calls and types.
10923         * graphite-clast-to-gimple.c: Include graphite-htab.h.
10924         (htab_t ivs_params::newivs_index):
10925         Change type to hash_table.  Update dependent calls and types.
10926         (htab_t ivs_params::params_index): Likewise.
10927         (htab_t print_generated_program::params_index): Likewise.
10928         (htab_t gloog::newivs_index): Likewise.
10929         (htab_t gloog::params_index): Likewise.
10931         * graphite.c: Include graphite-htab.h.
10932         4htab_t graphite_transform_loops::bb_pbb_mapping):
10933         Change type to hash_table.  Update dependent calls and types.
10935         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
10936         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
10937         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
10939         * graphite-dependences.c: Include graphite-htab.h.
10940         (loop_is_parallel_p): Change hash table type of parameter.
10942         * graphite-htab.h: New.
10943         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
10944         (extern find_pbb_via_hash): Move from graphite-poly.h.
10945         (extern loop_is_parallel_p): Move from graphite-poly.h.
10946         (extern get_loop_body_pbbs): Move from graphite-poly.h.
10948         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
10949         (extern loop_is_parallel_p): Move to graphite-htab.h.
10950         (extern get_loop_body_pbbs): Move to graphite-htab.h.
10952         * haifa-sched.c (htab_t delay_htab):
10953         Change type to hash_table.  Update dependent calls and types.
10954         (htab_t delay_htab_i2): Likewise.
10956         * ira-color.c (htab_t allocno_hard_regs_htab):
10957         Change type to hash_table.  Update dependent calls and types.
10959         * ira-costs.c (htab_t cost_classes_htab):
10960         Change type to hash_table.  Update dependent calls and types.
10962         * loop-invariant.c (htab_t merge_identical_invariants::eq):
10963         Change type to hash_table.  Update dependent calls and types.
10965         * loop-iv.c (htab_t bivs):
10966         Change type to hash_table.  Update dependent calls and types.
10968         * loop-unroll.c (htab_t opt_info::insns_to_split):
10969         Change type to hash_table.  Update dependent calls and types.
10970         (htab_t opt_info::insns_with_var_to_expand): Likewise.
10972         * lto-streamer.h (struct string_slot): Move from data-streamer.h
10973         (struct string_slot_hasher): New.
10974         (htab_t output_block::string_hash_table):
10975         Change type to hash_table.  Update dependent calls and types.
10977         * lto-streamer-in.c (freeing_string_slot_hasher): New.
10978         (htab_t file_name_hash_table):
10979         Change type to hash_table.  Update dependent calls and types.
10981         * lto-streamer-out.c: Update output_block::string_hash_table dependent
10982         calls and types.
10984         * lto-streamer.c (htab_t tree_htab):
10985         Change type to hash_table.  Update dependent calls and types.
10987         * omp-low.c: Include gimplify-ctx.h.
10989         * passes.c (htab_t name_to_pass_map):
10990         Change type to hash_table.  Update dependent calls and types.
10991         (pass_traverse): Rename to passes_pass_traverse.
10993         * plugin.c (htab_t event_tab):
10994         Change type to hash_table.  Update dependent calls and types.
10996         * postreload-gcse.c (htab_t expr_table):
10997         Change type to hash_table.  Update dependent calls and types.
10998         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
11000         * sese.c (debug_rename_map_1): Make extern.
11001         (htab_t copy_bb_and_scalar_dependences::rename_map):
11002         Change type to hash_table.  Update dependent calls and types.
11004         * sese.h (extern debug_rename_map): Move to .c file.
11006         * store-motion.c (htab_t store_motion_mems_table):
11007         Change type to hash_table.  Update dependent calls and types.
11009         * trans-mem.c (htab_t tm_new_mem_hash):
11010         Change type to hash_table.  Update dependent calls and types.
11012         * tree-browser.c (htab_t TB_up_ht):
11013         Change type to hash_table.  Update dependent calls and types.
11015         * tree-cfg.c (htab_t discriminator_per_locus):
11016         Change type to hash_table.  Update dependent calls and types.
11018         * tree-complex.c: Include tree-hasher.h
11019         (htab_t complex_variable_components):
11020         Change type to hash_table.  Update dependent calls and types.
11022         * tree-eh.c (htab_t finally_tree):
11023         Change type to hash_table.  Update dependent calls and types.
11025         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
11026         struct int_tree_hasher.
11027         (extern int_tree_map_eq): Likewise.
11028         (uid_decl_map_hash): Removed.
11029         (extern decl_tree_map_eq): Likewise.
11031         * tree-hasher.h: New.
11032         (struct int_tree_hasher): New.
11033         (typedef int_tree_htab_type): New.
11035         * tree-inline.c: Include gimplify-ctx.h.
11037         * tree-mudflap.c: Include gimplify-ctx.h.
11039         * tree-parloops.c: Include tree-hasher.h.
11040         (htab_t eliminate_local_variables_stmt::decl_address):
11041         Change type to hash_table.  Update dependent calls and types.
11042         (htab_t separate_decls_in_region::decl_copies): Likewise.
11044         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
11045         Change type to hash_table.  Update dependent calls and types.
11047         * tree-sra.c (candidates):
11048         Change type to hash_table.  Update dependent calls and types.
11050         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
11051         in tree-flow.h.
11052         (int_tree_map_hash): Likewise.
11054         * tree-ssa-dom.c (htab_t avail_exprs):
11055         Change type to hash_table.  Update dependent calls and types.
11057         * tree-ssa-live.c (var_map_base_init::tree_to_index):
11058         Change type to hash_table.  Update dependent calls and types.
11060         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11061         Change type to hash_table.  Update dependent calls and types.
11063         * tree-ssa-phiopt.c (seen_ssa_names):
11064         Change type to hash_table.  Update dependent calls and types.
11066         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11067         Change type to hash_table.  Update dependent calls and types.
11069         * tree-ssa-uncprop.c (equiv):
11070         Change type to hash_table.  Update dependent calls and types.
11072 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
11074         PR rtl-optimization/57003
11075         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11076         call note_stores with kill_clobbered_value callback again after
11077         killing regs_invalidated_by_call.
11079 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11081         * config/aarch64/aarch64-simd.md
11082         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11083         (aarch64_simd_bsl<mode>): Likewise.
11084         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11086 2013-04-25  Marek Polacek  <polacek@redhat.com>
11088         PR tree-optimization/57066
11089         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11091 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
11093         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11095 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11097         * config/aarch64/aarch64-builtins.c
11098         (aarch64_fold_builtin): New.
11099         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11100         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11101         * config/aarch64/aarch64-simd-builtins.def (abs): New.
11102         * config/aarch64/arm_neon.h
11103         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11105 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11106             Tejas Belagod  <tejas.belagod@arm.com>
11108         * config/aarch64/aarch64-builtins.c
11109         (aarch64_gimple_fold_builtin): New.
11110         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11111         * config/aarch64/aarch64-simd-builtins.def (addv): New.
11112         * config/aarch64/aarch64-simd.md (addpv4sf): New.
11113         (addvv4sf): Update.
11114         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11116 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11118         * config/aarch64/aarch64.md
11119         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11121 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11123         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11124         (*ngcsi_uxtw): New pattern.
11126 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11127             Julian Brown  <julian@codesourcery.com>
11129         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11130         (TB_DREG): Add T_V4HF.
11131         (v4hf_UP): New macro.
11132         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11133         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11134         Handle initialisation of V4HF. Adjust initialisation of reinterpret
11135         built-ins.
11136         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11137         (arm_vector_mode_supported_p): Handle V4HF.
11138         (arm_mangle_map): Handle V4HFmode.
11139         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11140         * config/arm/arm_neon_builtins.def: Add entries for
11141         vcvtv4hfv4sf, vcvtv4sfv4hf.
11142         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11143         (neon_vcvtv4hfv4sf): Likewise.
11144         * config/arm/neon-gen.ml: Handle half-precision floating point
11145         features.
11146         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11147         * config/arm/arm_neon.h: Regenerate.
11148         * config/arm/neon.ml (type elts): Add F16.
11149         (type vectype): Add T_float16x4, T_floatHF.
11150         (type vecmode): Add V4HF.
11151         (type features): Add Requires_FP_bit feature.
11152         (elt_width): Handle F16.
11153         (elt_class): Likewise.
11154         (elt_of_class_width): Likewise.
11155         (mode_of_elt): Refactor.
11156         (type_for_elt): Handle F16, fix error messages.
11157         (vectype_size): Handle T_float16x4.
11158         (vcvt_sh): New function.
11159         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11160         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11161         (string_of_mode): Handle V4HF.
11162         * doc/arm-neon-intrinsics.texi: Regenerate.
11164 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11166         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11167         format specifier in 'X' case.
11169 2013-04-25  Alan Modra  <amodra@gmail.com>
11171         PR target/57052
11172         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11173         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11174         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11175         Repeat for many other rotate/shift and mask patterns using subregs.
11176         Name lshiftrt insns.
11177         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11178         on WORDS_BIG_ENDIAN.
11180 2013-04-25  Alan Modra  <amodra@gmail.com>
11182         * config.gcc: Support little-endian powerpc-linux targets.
11183         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11184         (LINK_OS_LINUX_SPEC): Define.
11185         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11186         Preserve MASK_LITTLE_ENDIAN.
11187         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11188         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11189         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11190         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11191         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11192         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11193         Correct fp word order for little-endian.  Don't shift toc entries
11194         smaller than a word for little-endian.
11195         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11196         (bswapdi2 splits): Correct low-part subreg for little-endian.
11197         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11198         low/high where such is correct only for be.
11199         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11200         little-endian for -mcall-aixdesc.
11202 2013-04-25  Alan Modra  <amodra@gmail.com>
11204         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11205         replace_equiv_address_nv.
11207 2013-04-25  Alan Modra  <amodra@gmail.com>
11209         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11211 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11213         Revert:
11214         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11215         * rtl.h (struct rtx_def): ...
11217 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11219         PR rtl-optimizations/57046
11220         * lra-constraints (split_reg): Set up lra_risky_transformations_p
11221         for multi-reg splits.
11223 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
11225         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11227 2013-04-24  Sterling Augustine  <saugustine@google.com>
11229         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11230         (comp_dir_string, debug_str_dwo_section): New.
11231         (DEBUG_STR_DWO_SECTION): Rename to ...
11232         (DEBUG_DWO_STR_SECTION): ... this.
11233         (DEBUG_NORM_STR_SECTION): Delete.
11234         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11235         (DEBUG_STR_DWO_SECTION_FLAGS): New.
11236         (find_AT_string): Move most logic to ...
11237         (find_AT_string_in_table): ... here.  New.
11238         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11239         add_skeleton_AT_string.  Delete logic.
11240         (output_skeleton_debug_sections): Remove call to
11241         add_top_level_skeleton_die_attrs.
11242         (add_comp_dir_attribute): Move logic to comp_dir_string.
11243         (dwarf2out_init): Initialize debug_str_dwo_section.
11244         (output_indirect_string): Call find_string_form.
11245         (output_indirect_strings): Rewrite.
11246         (prune_unused_types): Empty skeleton_debug_str_hash.
11247         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
11248         (dwarf2out_finish):  Call output_indirect_strings.
11250 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11252         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
11254 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11256         * rtl.h (struct rtx_def): Add comment for field jump.
11257         (LRA_SUBREG_P): New macro.
11258         * recog.c (register_operand): Check LRA_SUBREG_P.
11259         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11260         stack frame.
11261         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11262         (lra_final_code_change): Skip subreg change for operators.
11263         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11264         if there are no operand changes.
11265         * lra-constraints.c (curr_insn_set): New.
11266         (match_reload): Set LRA_SUBREG_P.
11267         (emit_spill_move): Ditto.
11268         (check_and_process_move): Use curr_insn_set. Process only single
11269         set insns.  Don't initialize sec_mem_p and change_p.
11270         (simplify_operand_subreg): Use LRA_SUBREG_P.
11271         (reg_in_class_p): New function.
11272         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11273         of #ifdef.  Add code to remove cycling.
11274         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11275         non-null disp.  Reload inner instead of disp when base and index
11276         are null.  Try to put lo_sum into register.
11277         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11278         (check_and_process_move): Move code for move cost check to
11279         simple_move_p.  Remove equiv_substitution.
11280         (simple_move_p): New function.
11281         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11282         curr_insn_set.  Call check_and_process_move only for single set
11283         insns.  Use the new function.  Move call of check_and_process_move
11284         after operand equiv substitution and address process.
11286 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
11288         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
11289         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
11290         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
11292 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11294         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
11296 2013-04-24  Marek Polacek  <polacek@redhat.com>
11298         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
11299         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
11300         (select_loops_exit_conditions): Likewise.
11301         (number_of_iterations_for_all_loops): Likewise.
11302         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
11303         (scev_analysis): Likewise.
11305 2013-04-02  Catherine Moore  <clm@codesourcery.com>
11306             Chao-ying Fu <fu@mips.com>
11308         * config/mips/micromips.md (jraddiusp): New pattern.
11309         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
11310         instruction if possible.
11312 2013-04-24  Alan Modra  <amodra@gmail.com>
11314         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
11316 2013-04-24  Julian Brown  <julian@codesourcery.com>
11317             Chung-Lin Tang  <cltang@codesourcery.com>
11319         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
11320         dependency behavior in enumeration type DIE generation. Add TODO note
11321         to comments about future DW_FORM_sdata/udata re-work of related code.
11323 2013-04-23  Lawrence Crowl  <crowl@google.com>
11325         * Makefile.in: Update as needed below.
11327         * hash-table.h (class hash_table):
11328         Correct many methods with parameter types compare_type to the correct
11329         value_type.  (Correct code was unlikely to notice the change.)
11330         (hash_table::elements_with_deleted) New.
11331         (class hashtable::iterator): New.
11332         (hashtable::begin()): New.
11333         (hashtable::end()): New.
11334         (FOR_EACH_HASH_TABLE_ELEMENT): New.
11336         * statistics.c (statistics_hashes):
11337         Change type to hash_table.  Update dependent calls and types.
11339         * tree-into-ssa.c (var_infos):
11340         Change type to hash_table.  Update dependent calls and types.
11342         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
11343         Change type to hash_table.  Update dependent calls and types.
11345         * tree-ssa-loop-im.c (struct mem_ref.refs):
11346         Change type to hash_table.  Update dependent calls and types.
11348         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
11349         Change type to hash_table.  Update dependent calls and types.
11351         * tree-ssa-sccvn.c (vn_tables_s::nary):
11352         Change type to hash_table.  Update dependent calls and types.
11353         (vn_tables_s::phis): Likewise.
11354         (vn_tables_s::references): Likewise.
11356         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
11357         (vn_reference_eq): Update parameter and return types.
11359         * tree-ssa-structalias.c (pointer_equiv_class_table):
11360         Change type to hash_table.  Update dependent calls and types.
11361         (location_equiv_class_table): Likewise.
11363         * tree-vect-data-refs.c: Consequential changes for making
11364         peeling a hash_table.
11366         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
11367         (destroy_loop_vec_info): Dependent hash_table update.
11369         * tree-vectorizer.h (peeling_htab):
11370         Change type to hash_table.  Update dependent calls and types.
11372 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
11374         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
11375         to check the register content is equal or not.
11376         * lra-constraints.c (match_reload): Use lra_assign_reg_val
11377         to assign register content record.
11378         * lra-eliminations.c (update_reg_eliminate): Use
11379         lra_update_reg_val_offset to update register content offset.
11380         * lra-int.h (struct lra_reg): Add offset member.
11381         (lra_reg_val_equal_p): New static inline function.
11382         (lra_update_reg_val_offset): New static inline function.
11383         (lra_assign_reg_val): New static inline function.
11384         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
11385         to assign register content record.
11386         (initialize_lra_reg_info_element): Initial offset to zero.
11388 2013-04-23  Catherine Moore  <clm@codesourcery.com>
11390         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
11391         operands.  Record compression.
11393 2013-04-23  Xinliang David Li  <davidxl@google.com>
11395         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
11397 2013-04-23  Richard Biener  <rguenther@suse.de>
11399         PR middle-end/57036
11400         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
11401         parameter, only add abnormal goto edges from the copied body
11402         if the call could perform abnormal gotos.
11403         (copy_cfg_body): Adjust.
11405 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
11407         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
11409 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
11411         * coretypes.h (gimple_stmt_iterator): Add struct to make
11412         compatible with C.
11414 2013-04-23  Richard Biener  <rguenther@suse.de>
11416         PR tree-optimization/57026
11417         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
11418         from SSA names occuring in abnormal PHI nodes.
11420 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11422         * lto/lto.c (print_lto_report_1): Fix LTO report names.
11424 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11426         * lto/lto.c (print_lto_report_1): Declare early.
11427         (read_cgraph_and_symbols): Call print_lto_report_1 early.
11429 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11431         * common.opt (-flto-report-wpa): Add.
11432         * doc/invoke.texi (-flto-report-wpa): Add.
11433         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
11434         (lto_main): dito.
11436 2013-04-22  Xinliang David Li  <davidxl@google.com>
11438         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
11439         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
11440         * Makefile.in: New dependency
11442             David Daney <ddaney.cavm@gmail.com>
11444         * configure.ac (gcc_cv_as_micromips_support):  Use the
11445         --fatal-warnings option.
11446         * configure: Regenerate.
11448 2013-04-22  Marek Polacek  <polacek@redhat.com>
11450         PR sanitizer/56990
11451         * tsan.c (instrument_expr): Don't instrument expression
11452         in case its size is zero.
11454 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
11456         PR target/57032
11457         Revert:
11458         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
11460         * config/alpha/alpha.c (TARGET_LRA_P): New define.
11462 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11464         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
11465         (gimple_stmt_iterator): New typedef.
11466         * gimple.h (gimple_stmt_iterator): Rename to...
11467         (gimple_stmt_iterator_d): ... This.
11468         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
11469         trees be valid for GIMPLE and GENERIC.
11470         (TARGET_GIMPLE_FOLD_BUILTIN): New.
11471         * gimple-fold.c (gimple_fold_call): Call target hook
11472         gimple_fold_builtin.
11473         * hooks.c (hook_bool_gsiptr_false): New.
11474         * hooks.h (hook_bool_gsiptr_false): New.
11475         * target.def (fold_stmt): New.
11476         * doc/tm.texi: Regenerate.
11478 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
11480         PR target/57018
11481         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
11482         a set sp if no stack realignment.
11484 2013-04-22  Nick Clifton  <nickc@redhat.com>
11486         * config.gcc (tilegx-linux): Extend extra_objs rather than
11487         overwriting it.
11488         (tilepro-linux): Likewise.
11490 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11492         * config/aarch64/aarch64-builtins.c
11493         (CF): Remove.
11494         (CF0, CF1, CF2, CF3, CF4, CF10): New.
11495         (VAR<1-12>): Add MAP parameter.
11496         (BUILTIN_*): Likewise.
11497         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
11498         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
11499         (aarch64_ushl_n<mode>): Likewise.
11500         (aarch64_sshr_n<mode>): Likewise.
11501         (aarch64_ushr_n<mode>): Likewise.
11502         (aarch64_<maxmin><mode>): Likewise.
11503         (aarch64_sqrt<mode>): Likewise.
11504         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
11505         (vshr<q>_n_*): Likewise.
11507 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11509         * config/aarch64/aarch64-builtins.c
11510         (aarch64_simd_builtin_type_mode): Handle SF types.
11511         (sf_UP): Define.
11512         (BUILTIN_GPF): Define.
11513         (aarch64_init_simd_builtins): Handle SF types.
11514         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
11515         (frecps): Likewise.
11516         (frecpx): Likewise.
11517         * config/aarch64/aarch64-simd.md
11518         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
11519         (aarch64_frecpe<mode>): New.
11520         (aarch64_frecps<mode>): Likewise.
11521         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
11522         (v8type): Add frecp<esx>.
11523         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
11524         (aarch64_frecps<mode>): Likewise.
11525         * config/aarch64/iterators.md (FRECP): New.
11526         (frecp_suffix): Likewise.
11527         * config/aarch64/arm_neon.h
11528         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
11530 2013-04-22  Christian Bruel  <christian.bruel@st.com>
11532         PR target/56995
11533         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
11534         (REG_CLASS_NAMES): Idem.
11535         (REG_CLASS_CONTENTS): Idem.
11536         (REGCLASS_HAS_FP_REG): Idem.
11537         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
11538         (sh_conditional_register_usage): Idem.
11540 2013-04-21  Jeff Law  <law@redhat.com>
11542         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
11543         (ssa_forward_propagate_and_combine): Use it.
11545 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11547         * lra.c: Update the flow chart diagram.
11549 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11551         PR rtl-optimization/56847
11552         * lra-constraints.c (process_alt_operands): Discourage alternative
11553         with non-matche doffsettable memory constraint fro memory with
11554         known offset.
11556 2013-04-19  Richard Biener  <rguenther@suse.de>
11558         PR tree-optimization/56982
11559         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
11560         function.
11561         * gimplify.c (gimplify_call_expr): Notice special calls.
11562         (gimplify_modify_expr): Likewise.
11563         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
11564         abnormal control flow receivers.
11565         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
11566         in the same way as cfun->has_nonlocal_labels.
11567         (gimple_purge_dead_abnormal_call_edges): Likewise.
11568         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
11569         receivers start a basic-block.
11571 2013-04-19  Richard Biener  <rguenther@suse.de>
11573         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
11574         member ...
11575         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
11576         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
11577         (SLP_TREE_LOAD_PERMUTATION): Add.
11578         (vect_transform_slp_perm_load): Adjust prototype.
11579         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
11580         (vect_free_slp_instance): Likewise.
11581         (vect_create_new_slp_node): Likewise.
11582         (vect_supported_slp_permutation_p): Remove.
11583         (vect_slp_rearrange_stmts): Adjust.
11584         (vect_supported_load_permutation_p): Likewise.  Inline
11585         vect_supported_slp_permutation_p here.
11586         (vect_analyze_slp_instance): Compute load permutations per
11587         slp node instead of per instance.
11588         (vect_get_slp_defs): Adjust.
11589         (vect_transform_slp_perm_load): Likewise.
11590         (vect_schedule_slp_instance): Remove redundant code.
11591         (vect_schedule_slp): Remove hack for PR56270, add it ...
11592         * tree-vect-stmts.c (vectorizable_load): ... here, do not
11593         CSE loads for SLP.  Adjust.
11595 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11597         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
11598         spelling in two comments.
11600 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11602         PR target/56797
11603         * config/arm/arm.c (load_multiple_sequence): Require SP
11604         as base register for loads if SP is in the register list.
11606 2013-04-19  Martin Jambor  <mjambor@suse.cz>
11608         PR tree-optimization/56718
11609         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
11610         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
11611         and made public.  Adjusted all callers.
11612         (ipa_intraprocedural_devirtualization): New function.
11613         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
11614         (ipa_intraprocedural_devirtualization): Likewise.
11615         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
11617 2013-04-19  Richard Biener  <rguenther@suse.de>
11619         PR tree-optimization/57000
11620         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
11622 2013-04-19  Terry Guo  <terry.guo@arm.com>
11624         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
11625         Replace with ...
11626         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
11627         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
11628         (cortex_m4_fmacs): Use new reservations.
11629         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
11631 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
11633         PR rtl-optimization/56999
11634         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
11635         related code.
11636         (lra_coalesce): Remove split_origin_bitmap and related code.
11637         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
11638         ranges if necessary.
11640 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
11642         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
11643         New array.
11644         (ix86_expand_call): Remove clobbered_registers array and use
11645         x86_64_ms_sysv_extra_clobbered_registers instead.
11646         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
11647         Declare here.
11648         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
11649         predicate.
11650         * config/i386/i386.md (*call_rex64_ms_sysv): Use
11651         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
11652         (*call_value_rex64_ms_sysv): Ditto.
11654 2013-04-18  Cary Coutant  <ccoutant@google.com>
11656         * dwarf2out.c (output_pubnames): Check die_perennial_p of
11657         parent instead of die_mark.
11659 2013-04-18  Diego Novillo  <dnovillo@google.com>
11661         * gimple.c (create_gimple_tmp): New.
11662         (get_expr_type): New.
11663         (build_assign): New.
11664         (build_type_cast): New.
11665         * gimple.h (enum ssa_mode): Define.
11666         (gimple_seq_set_location): New.
11667         * asan.c (build_check_stmt): Change some gimple_build_* calls
11668         to use build_assign and build_type_cast.
11670 2013-04-18  Richard Biener  <rguenther@suse.de>
11672         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
11673         handle negative step.  Remove redundant checks.
11674         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
11675         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
11676         for negative step and grouped loads fail to vectorize.
11678 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11680         * emit-rtl.c (reset_insn_used_flags): New function.
11681         (reset_all_used_flags): Use it.
11682         (verify_insn_sharing): New function.
11683         (verify_rtl_sharing): Fix verification for SEQUENCEs.
11685 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
11687         PR tree-optimization/56984
11688         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
11689         and (x >> M) >= N don't register any assertion if N << M is the
11690         minimum value.
11692 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
11694         * lower-subreg.c (resolve_simple_move): If called self-recursive,
11695         do not delete_insn insns that have not yet been emitted, only
11696         unlink them with remove_insn.
11697         * df-scan.c (df_insn_delete): Revert r197492.
11699 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
11701         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
11702         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
11704 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11706         * config/arm/arm.md (movsicc_insn): Convert define_insn into
11707         define_insn_and_split.
11708         (and_scc,ior_scc,negscc): Likewise.
11709         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
11711 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
11713         * config/arm/arm.c (use_return_insn): Return 0 for targets that
11714         can benefit from using a sequence of LDRD instructions in epilogue
11715         instead of a single LDM instruction.
11717 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
11719         PR 45688
11720         * doc/extend.texi: Fix typo.
11722 2013-04-17  Richard Biener  <rguenther@suse.de>
11724         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
11725         (vect_build_slp_tree): ... here.
11726         (vect_build_slp_tree_1): Compute which stmts of the SLP group
11727         match.  Remove special-casing of mismatched complex loads.
11728         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
11729         re-try the match with swapped commutative operands.
11730         (vect_supported_load_permutation_p): Remove special-casing of
11731         mismatched complex loads.
11732         (vect_analyze_slp_instance): Adjust.
11734 2013-04-17  Richard Biener  <rguenther@suse.de>
11736         PR rtl-optimization/56921
11737         * cfgloop.h (struct loop): Add simple_loop_desc member.
11738         (struct niter_desc): Mark with GTY(()).
11739         (simple_loop_desc): Do not use aux field but simple_loop_desc.
11740         * loop-iv.c (get_simple_loop_desc): Likewise.
11741         (free_simple_loop_desc): Likewise.
11743         Revert
11744         2013-04-16  Richard Biener  <rguenther@suse.de>
11746         PR rtl-optimization/56921
11747         * loop-init.c (pass_rtl_move_loop_invariants): Add
11748         TODO_do_not_ggc_collect to todo_flags_finish.
11749         (pass_rtl_unswitch): Same.
11750         (pass_rtl_unroll_and_peel_loops): Same.
11751         (pass_rtl_doloop): Same.
11753 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
11755         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
11756         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
11757         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
11758         references.
11759         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
11760         * tree-streamer.c (record_common_node): Adjust reference in comment.
11762 2013-04-17  Terry Guo  <terry.guo@arm.com>
11764         * config/arm/cortex-m4.md: Add a new bypass.
11766 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11768         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
11769         New pattern.
11770         (*subs_<optab><mode>_multp2): New pattern.
11771         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11772         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11774 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11776         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
11777         (*subs_mul_imm_<mode>): New pattern.
11779 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
11781         PR target/56948
11782         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
11783         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
11784         (vsx_movti_32bit): Change j->wa to O->wa.
11786 2013-04-16  Richard Biener  <rguenther@suse.de>
11788         PR rtl-optimization/56921
11789         * loop-init.c (pass_rtl_move_loop_invariants): Add
11790         TODO_do_not_ggc_collect to todo_flags_finish.
11791         (pass_rtl_unswitch): Same.
11792         (pass_rtl_unroll_and_peel_loops): Same.
11793         (pass_rtl_doloop): Same.
11795 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
11797         * config/arm/arm.c (emit_multi_reg_push): New declaration
11798         for an existing function.
11799         (arm_emit_strd_push): New function.
11800         (arm_expand_prologue): Used here.
11801         (arm_emit_ldrd_pop): New function.
11802         (arm_expand_epilogue): Used here.
11803         (arm_get_frame_offsets): Update condition.
11804         (arm_emit_multi_reg_pop): Add a special case for load of a single
11805         register with writeback.
11807 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
11809         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
11810         description.
11812 2013-04-16  Richard Biener  <rguenther@suse.de>
11814         PR tree-optimization/56756
11815         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
11816         (first_mem_ref_loc): New.
11817         (execute_sm): Place the load temporarily before a previous
11818         access instead of in the latch edge to ensure its SSA dependencies
11819         are defined at points dominating the load.
11821 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
11823         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
11824         correct fix by moving header and footer insn to the footer of
11825         the merged basic block.  Clear BB_END of the merged-away block.
11827         PR middle-end/43631
11828         * emit-rtl.c (make_note_raw): New function.
11829         (link_insn_into_chain): New static inline function.
11830         (add_insn): Use it.
11831         (add_insn_before, add_insn_after): Factor insn chain linking code...
11832         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
11833         using link_insn_into_chain.
11834         (note_outside_basic_block_p): New helper function for emit_note_after
11835         and emit_note_before.
11836         (emit_note_after): Use nobb variant of add_insn_after if the note
11837         should not be contained in a basic block.
11838         (emit_note_before): Use nobb variant of add_insn_before if the note
11839         should not be contained in a basic block.
11840         (emit_note_copy): Use make_note_raw.
11841         (emit_note): Likewise.
11842         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
11843         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
11844         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
11845         the moved barrier the tail of the basic block it follows.
11846         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
11848 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
11850         PR tree-optimization/56962
11851         * gimple-ssa-strength-reduction.c (record_increment): Only set
11852         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
11853         either rhs1 or rhs2 is equal to c->base_expr.
11855 2013-04-15  Richard Biener  <rguenther@suse.de>
11857         PR tree-optimization/56933
11858         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
11859         member.
11860         (GROUP_READ_WRITE_DEPENDENCE): Remove.
11861         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
11862         * tree-vect-data-refs.c (vect_analyze_group_access): Move
11863         dependence check ...
11864         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
11865         ... here.
11866         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
11867         GROUP_READ_WRITE_DEPENDENCE.
11869 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11871         * emit-rtl.c (reset_all_used_flags): New function.
11872         (verify_rtl_sharing): Call reset_all_used_flags before and after
11873         performing the checks.
11875 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11877         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
11878         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
11879         * config/arm/constraints.md (De): New constraint.
11880         * config/arm/neon.md (anddi3_neon): Delete.
11881         (neon_vand<mode>): Expand to standard anddi3 pattern.
11882         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
11883         Move earlier in the file.
11884         (neon_inv_logic_op2): Likewise.
11885         (arm_anddi_operand_neon): New predicate.
11887 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11889         * configure.ac (gcc_cv_ld_as_needed): Set
11890         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
11891         Use -z ignore, -z record on *-*-solaris2*.
11892         (HAVE_LD_AS_NEEDED): Update comment.
11893         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
11894         * configure: Regenerate.
11895         * config.in: Regenerate.
11896         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
11897         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
11898         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
11899         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
11900         equivalents.  Fix markup.
11901         * doc/tm.texi: Regenerate.
11903 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
11905         * config/i386/i386.opt: New option mstack-protector-guard=.
11906         * config/i386/i386-opts.h: Add enum stack_protector_guard.
11907         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
11908         TARGET_SSP_TLS_GUARD.
11909         * config/i386/i386.c (ix86_option_override_internal): Set
11910         ix86_stack_protector_guard.
11911         * config/i386/i386.md (stack_protect_set): Enable for
11912         TARGET_SSP_TLS_GUARD only.
11913         (stack_protect_set_<mode>): Ditto.
11914         (stack_protect_test): Ditto.
11915         (stack_protect_test_<mode>): Ditto.
11916         * doc/invoke.texi (i386 Option): Document.
11918 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
11920         PR target/56890
11921         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
11922         (S_MODES): Set H_MODE bit.
11923         (SF_MODES): Set only S_MODE and SF_MODE bits.
11924         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
11925         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
11926         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
11927         <MODE_FLOAT>: Likewise.
11929 2013-04-15  Joey Ye  <joey.ye@arm.com>
11931         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
11933 2013-04-15  Joey Ye  <joey.ye@arm.com>
11935         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
11936         for real far jump.
11937         (thumb_far_jump_used_p): Count instruction size and set
11938         far_jump_used.
11940 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
11942         * reorg.c (fill_simple_delay_slots): Reindent block of code.
11943         * resource.c (mark_target_live_regs): Reformat conditional block.
11945 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11947         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
11948         notes, they are emitted only just before final.
11949         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
11951 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
11953         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
11954         * cfgrtl.c (delete_insn): Call it here instead.
11955         * lra-spills.c (lra_final_code_change): Use delete_insn.
11956         * haifa-sched.c (sched_remove_insn): Likewise.
11957         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
11958         returning to the nop pool.
11959         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
11960         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
11962 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11964         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
11965         * doc/tm.texi: Regenerated.
11967 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
11969         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
11970         QImode checks.
11972 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
11974         * df-core.c (df_find_def): Compare register numbers.
11975         (df_find_use): Likewise.
11977 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
11979         PR target/56903
11980         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11981         lra_in_progress for return.
11983 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11985         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11986         define_insn into define_insn_and_split and emit movsicc patterns.
11988 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
11990         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11992 2013-04-12  Richard Biener  <rguenther@suse.de>
11994         * tree-pass.h (TODO_do_not_ggc_collect): New.
11995         * passes.c (execute_one_ipa_transform_pass): Honor
11996         TODO_do_not_ggc_collect.
11997         (execute_one_pass): Likewise.
11999         Revert
12000         2013-04-10  Richard Biener  <rguenther@suse.de>
12002         * passes.c (init_optimization_passes): Remove reload pass.
12003         * ira.c (do_reload): Merge into ...
12004         (ira): ... this.
12005         (rest_of_handle_reload): Remove.
12006         (pass_reload): Likewise.
12007         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12008         of reload for vzeroupper pass placement.
12010 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
12012         PR tree-optimization/56918
12013         PR tree-optimization/56920
12014         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
12015         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
12016         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
12017         use wide_mul_with_sign method.
12019 2013-04-12  Richard Biener  <rguenther@suse.de>
12021         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
12022         not be considered a gimple constant.
12024 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
12026         * fold-const.c (const_binop): Handle vector shifts by a scalar.
12027         (fold_binary_loc): Call const_binop also for mixed vector-scalar
12028         operations.
12030 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12031             Jakub Jelinek  <jakub@redhat.com>
12033         * opts.c: Include diagnostic-color.h.
12034         (common_handle_option): Handle OPT_fdiagnostics_color_.
12035         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
12036         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
12037         (diagnostic-color.o): New.
12038         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
12039         (diagnostic_color_rule): New enum.
12040         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
12041         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
12042         the location string.
12043         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
12044         either NULL, or color kind.
12045         * diagnostic-color.c: New file.
12046         * diagnostic-color.h: New file.
12047         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
12048         arguments.
12049         * doc/invoke.texi (-fdiagnostics-color): Document.
12050         * pretty-print.h (pp_show_color): Define.
12051         (struct pretty_print_info): Add show_color field.
12052         * diagnostic.c: Include diagnostic-color.h.
12053         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12054         macros.  Colorize error:, warning: etc. strings and also the location
12055         string.
12056         (diagnostic_show_locus): Colorize the caret line.
12057         * pretty-print.c: Include diagnostic-color.h.
12058         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
12059         inside of %< %> quotes or quoted through q format modifier.
12061 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12063         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12065 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12067         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12068         code in CC_NZ mode.
12069         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12070         pattern.
12072 2013-04-11  Marek Polacek  <polacek@redhat.com>
12074         PR tree-optimization/48184
12075         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12077 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
12079         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12080         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12081         (skip_simple_arithmetic): Tidy up.
12082         * tree.h (skip_simple_constant_arithmetic): Declare.
12084 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12086         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12088 2013-04-11  Richard Biener  <rguenther@suse.de>
12090         * tree-vect-loop.c (get_initial_def_for_induction): Properly
12091         generate vector constants.
12093 2013-04-11  Richard Biener  <rguenther@suse.de>
12095         PR tree-optimization/56878
12096         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12097         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12098         New function.
12099         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12100         Prefer to align the DR with the most invariant base address.
12102 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12104         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12105         comment.
12107 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
12109         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12110         floating-point vector comparisons against 0.
12112 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
12114         PR tree-optimization/56899
12115         * fold-const.c (extract_muldiv_1): Apply distributive law
12116         only if TYPE_OVERFLOW_WRAPS (ctype).
12118 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
12120         PR target/56124
12121         * ira-costs.c (scan_one_insn): Check whether the source rtx of
12122         loading has side effect.
12124 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
12126         * config/sparc/sparc.c: Include tree-pass.h.
12127         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12128         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
12129         head of file.  Change return type.  Split off gate function.
12130         (sparc_gate_work_around_errata): New function.
12131         (pass_work_around_errata): New pass definition.
12132         (insert_pass_work_around_errata) New pass insert definition to
12133         insert pass_work_around_errata just after delayed-branch scheduling.
12134         (sparc_option_override): Insert the pass.
12135         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12137 2013-04-10  David S. Miller  <davem@davemloft.net>
12139         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12140         or -mcpu=hypersparc.
12142         * target.def (cstore_mode): New hook.
12143         * target.h: Include insn-codes.h
12144         * targhooks.c: Likewise.
12145         (default_cstore_mode): New function.
12146         * targhooks.h: Declare it.
12147         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12148         * doc/tm.texi: Rebuild.
12149         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12150         target hook, rather than inspecting the insn_data.
12151         * config/sparc/sparc.c (sparc_cstore_mode): New function.
12152         (TARGET_CSTORE_MODE): Redefine.
12153         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12154         result patterns.
12155         * config/sparc/predicates.md (cstore_result_operand): New special
12156         predicate.
12157         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12158         Use it for operand 0.
12159         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12160         (*snesi_special): Likewise.
12161         (*snesi_zero): Likewise.
12162         (*seqsi_zero): Likewise.
12163         (*sltu_insn): Likewise.
12164         (*sgeu_insn): Likewise.
12165         (*seqdi_special): Make operand 0 and comparison operation be of
12166         DImode.
12167         (*snedi_special): Likewise.
12168         (*snedi_special_vis3): Likewise.
12169         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12170         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12171         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12172         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12173         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12174         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12175         (*sltu_extend_sp64): Likewise.
12176         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12177         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12178         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12179         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12180         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12182 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
12184         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12185         (aarch64_start_file): Use the new function.
12187 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12188             Jason Merrill  <jason@redhat.com>
12190         * common.opt: Add -gdwarf.
12191         * opts.c (common_handle_option): Handle it.
12192         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12194 2013-04-10  Richard Biener  <rguenther@suse.de>
12196         * passes.c (execute_todo): Do not call ggc_collect conditional here.
12197         (execute_one_ipa_transform_pass): But unconditionally here.
12198         (execute_one_pass): And here.
12199         (init_optimization_passes): Remove reload pass.
12200         * tree-pass.h (TODO_ggc_collect): Remove.
12201         (pass_reload): Likewise.
12202         * ira.c (do_reload): Merge into ...
12203         (ira): ... this.
12204         (rest_of_handle_reload): Remove.
12205         (pass_reload): Likewise.
12206         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12207         of reload for vzeroupper pass placement.
12208         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12209         and todo_flags_finish of all passes.
12211 2013-04-10  Richard Biener  <rguenther@suse.de>
12213         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12214         first_const_oprnd field, rename first_def_type to first_op_type.
12215         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12216         (vect_get_and_check_slp_defs): Always use the type of the
12217         operand.  Allow mixed vect_external_def, vect_constant_def types.
12218         (vect_get_constant_vectors): Handle mixed vect_external_def,
12219         vect_constant_def types.
12221 2013-04-10  Joern Rennecke <joern.rennecke@embecosm.com>
12223         PR tree-optimization/55524
12224         * tree-ssa-math-opts.c
12225         (convert_mult_to_fma): Don't use an fms construct
12226         when we don't have an fms operation, but fnma, and it looks
12227         likely that we'll be able to use the latter.
12229 2013-04-10  Zhouyi Zhou <yizhouzhou@ict.ac.cn>
12231         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12232         function.
12233         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12234         inline fail caused by overwritable functions.
12236 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
12238         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12239         unnecessary bits in the constant power of two case.
12241 2013-04-10  Richard Biener  <rguenther@suse.de>
12243         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12244         broken code swapping operands.
12245         (vect_build_slp_tree): Do not compute load permutations here.
12246         (vect_analyze_slp_instance): Compute load permutations here,
12247         after building the SLP tree.
12249 2013-04-09  Christian Bruel  <christian.bruel@st.com>
12251         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
12252         of next/prev_real_insn.
12254 2013-04-09  Jan Hubicka  <jh@suse.cz>
12256         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
12257         Drop aliased parameter.
12258         (function_and_variable_visibility): Do not handle alias pairs.
12259         * cgraph.c (varpool_externally_visible_p): Update prototype.
12260         * varpool.c (varpool_add_new_variable): Update.
12262 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12264         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
12266 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
12268         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
12270         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
12272 2013-04-09  Marek Polacek  <polacek@redhat.com>
12274         PR tree-optimization/48762
12275         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
12277 2013-04-09  Richard Biener  <rguenther@suse.de>
12279         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
12280         dealing with cost.
12281         (vect_build_slp_tree): Likewise.
12282         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
12283         calculating the cost of a SLP instance.
12284         (vect_analyze_slp_instance): Use it from here, after building
12285         the SLP tree.
12287 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12289         PR middle-end/56883
12290         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
12291         expand_omp_for_static_chunk): Use simple_p = true in
12292         force_gimple_operand_gsi calls when assigning to addressable decls.
12294 2013-04-09  Jeff Law  <law@redhat.com>
12296         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
12297         when the boolean was created by converting a wider object which
12298         had a boolean range.
12300 2013-04-09  Richard Biener  <rguenther@suse.de>
12302         * tree-vectorizer.h (slp_void_p): Remove.
12303         (slp_tree): Typedef before _slp_tree declaration.
12304         (struct _slp_tree): Use a vector of slp_tree as children.
12305         (vect_get_place_in_interleaving_chain): Remove.
12306         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
12307         Move ...
12308         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
12309         and make static.
12310         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
12311         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
12312         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
12313         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
12314         Use slp_node instead of slp_void_p and adjust.
12316 2013-04-09  Richard Biener  <rguenther@suse.de>
12318         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
12319         work that is not necessary.
12321 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12323         PR tree-optimization/56854
12324         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
12325         forward into clobber stmts if it would change MEM_REF lhs into
12326         non-MEM_REF.
12328 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
12330         * tree.c (type_hash_lookup, type_hash_add): Make static.
12331         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
12333 2013-04-09  Richard Biener  <rguenther@suse.de>
12335         * tree.h (unsave_expr_now): Remove.
12336         * tree-inline.c (mark_local_for_remap_r): Remove.
12337         (unsave_expr_1): Likewise.
12338         (unsave_r): Likewise.
12339         (unsave_expr_now): Likewise.
12340         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
12341         (propagate_tree_value): Likewise.
12343 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
12345         * doc/rtl.texi (sequence): Rewrite documentation to match the
12346         current use of SEQUENCE rtl objects.
12347         * rtl.def (SEQUENCE): Likewise.
12349         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
12350         Update documentation.
12351         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
12352         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
12354         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
12356 2013-04-08  Teresa Johnson  <tejohnson@google.com>
12358         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
12359         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
12360         methods.
12361         (estimate_edge_size_and_time): Add comment to suggest using rounding
12362         methods.
12363         (estimate_node_size_and_time): Ditto.
12364         (remap_edge_change_prob): Use helper rounding divide methods.
12365         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
12366         (gimple_mod_pow2_value_transform): Ditto.
12367         (gimple_mod_subtract_transform): Ditto.
12368         (gimple_ic_transform): Ditto.
12369         (gimple_stringops_transform): Ditto.
12370         * stmt.c (conditional_probability): Ditto.
12371         (emit_case_dispatch_table): Ditto.
12372         * lto-cgraph.c (merge_profile_summaries): Ditto.
12373         * tree-optimize.c (execute_fixup_cfg): Ditto.
12374         * cfgcleanup.c (try_forward_edges): Ditto.
12375         * cfgloopmanip.c (scale_loop_profile): Ditto.
12376         (loopify): Ditto.
12377         (duplicate_loop_to_header_edge): Ditto.
12378         (lv_adjust_loop_entry_edge): Ditto.
12379         * tree-vect-loop.c (vect_transform_loop): Ditto.
12380         * profile.c (compute_branch_probabilities): Ditto.
12381         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
12382         * lto-streamer-in.c (input_cfg): Ditto.
12383         * gimple-streamer-in.c (input_bb): Ditto.
12384         * ipa-cp.c (update_profiling_info): Ditto.
12385         (update_specialized_profile): Ditto.
12386         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
12387         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
12388         rounding methods.
12389         * sched-rgn.c (compute_dom_prob_ps): Ditto.
12390         (compute_trg_info): Ditto.
12391         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
12392         (purge_dead_edges): Ditto.
12393         * loop-unswitch.c (unswitch_loop): Ditto.
12394         * cgraphclones.c (cgraph_clone_edge): Ditto.
12395         (cgraph_clone_node): Ditto.
12396         * tree-inline.c (copy_bb): Ditto.
12397         (copy_edges_for_bb): Ditto.
12398         (initialize_cfun): Ditto.
12399         (copy_cfg_body): Ditto.
12400         (expand_call_inline): Ditto.
12402 2013-04-08  Kai Tietz  <ktietz@redhat.com>
12404         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
12405         TARGET_CYGWIN64 by TARGET_64BIT.
12407 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
12409         * config/epiphany/epiphany.md (GPR_1): New constant.
12410         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
12411         * config/epiphany/epiphany.c (gen_compare_reg):
12412         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
12413         is already in place.
12414         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
12415         Don't require being called during rtl expansion; If y operlaps r0,
12416         return 0.
12417         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
12418         (epiphany_expand_epilogue): Likewise.
12420         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
12421         Don't use CC_FPmode for ORDERED / UNORDERED.
12422         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
12424         * config/epiphany/constraints.md (CnL): New constraint.
12425         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
12426         * config/epiphany/predicates.md (add_operand): Allow 1024.
12428         * config/epiphany/epiphany.md (logical_op): New code iterator.
12429         (op_mnc): New code attribute.
12430         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
12431         (mov_f+1, mov_f+2): New peephole2 patterns.
12433         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
12434         (cstoresi4): Also allow re-use of zero result when doing a NE
12435         comparison to a non-zero operand.
12436         Use (clobber (scratch)) for first insn if the gpr output is not needed.
12438         * config/epiphany/epiphany.md (<insn_opname>v2si3):
12439         Use gen_addsi3_i / gen_subsi3_i.
12441 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
12443         PR c++/34949
12444         PR c++/50243
12445         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
12446         contain anything but clobbers, at most one __builtin_stack_restore,
12447         optionally debug stmts and final resx, and if it has at least one
12448         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
12449         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
12450         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
12451         which isn't defaut definition, remove them.
12452         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
12453         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
12454         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
12455         with MEM_REF LHS with SSA_NAME address.
12457 2013-04-08  Jeff Law  <law@redhat.com>
12459         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
12461 2013-04-08  Richard Biener  <rguenther@suse.de>
12463         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
12464         extra newline.
12465         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
12466         determined vector type.
12467         (vect_analyze_data_refs): Likewise.
12468         (vect_get_new_vect_var): Adjust.
12469         (vect_create_destination_var): Preserve SSA name versions.
12470         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
12471         not dump anything here.
12473 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
12475         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
12476         Add member lr_slot_known.
12477         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
12478         if necessary.
12479         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
12480         Remove code that sets lr_slot_offset according to what a previous
12481         version of epiphany_emit_save_restore used to do.
12482         (epiphany_emit_save_restore): When doing an lr save or restore,
12483         set/verify lr_slot_known and lr_slot_offset.
12485 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
12487         PR target/54338
12488         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
12489         in ALL_REGS.
12491 2013-04-08  Richard Biener  <rguenther@suse.de>
12493         * alias.c (find_base_term): Fix thinko in previous change.
12495 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
12497         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
12498         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
12499         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
12500         if possible to compute val.
12501         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
12502         For QImode integers don't require anything about precision.  Use
12503         const_with_all_bytes_same to find out if the constant doesn't have
12504         repeated bytes in it.
12506 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12508         * config/s390/s390.c (s390_expand_insv): Only accept insertions
12509         within mode size.
12511 2013-04-08  Marek Polacek  <polacek@redhat.com>
12513         PR rtl-optimization/48182
12514         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
12515         value to 1.
12517 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
12519         PR target/55487
12520         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
12521         nuses, make sure we have a label.
12523 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12525         PR target/56843
12526         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
12527         (rs6000_emit_swdiv_low_precision): Remove.
12528         (rs6000_emit_swdiv): Rewrite to handle between one and four
12529         iterations of Newton-Raphson generally; modify required number of
12530         iterations for some cases.
12531         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
12533 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
12535         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
12536         set-but-unused variable.
12538         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
12539         basic blocks of released function bodies garbage-collectable.
12541         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
12542         (struct rtl_opt_pass): Add TODO_df_finish.
12544         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
12546 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12548         * config/arm/constraints.md (q): New constraint.
12549         * config/arm/ldrdstrd.md: New file.
12550         * config/arm/arm.md (ldrdstrd.md) New include.
12551         (arm_movdi): Use "q" instead of "r" constraint
12552         for double-word memory access.
12553         (movdf_soft_insn): Likewise.
12554         * config/arm/vfp.md (movdi_vfp): Likewise.
12555         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
12556         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
12557         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
12558         (mem_ok_for_ldrd_strd): Likewise.
12559         (output_move_double): Update assertion.
12561 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12563         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
12565 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12567         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
12568         define_insn_and_split.
12569         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
12571 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12573         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
12574         define_insn_and_split.
12575         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
12576         (shiftsi3_compare): New pattern.
12577         (rrx): New pattern.
12578         * config/arm/unspecs.md (UNSPEC_RRX): New.
12580 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12582         * config/arm/arm.md (negdi_extendsidi): New pattern.
12583         (negdi_zero_extendsidi): Likewise.
12585 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12587         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
12588         define_insn_and_split.
12589         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
12590         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
12592 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12594         * config/arm/arm.md (arm_subdi3): Convert define_insn into
12595         define_insn_and_split.
12596         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
12597         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
12599 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12601         * config/arm/arm.md (subsi3_carryin): New pattern.
12602         (subsi3_carryin_const): Likewise.
12603         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
12604         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
12606 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12608         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
12610 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12612         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
12613         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
12615 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12617         * config/arm/arm.c (arm_expand_builtin): Change fcode
12618         type to unsigned int.
12620 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12622         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
12624 2013-04-04  Ian Lance Taylor  <iant@google.com>
12626         * doc/standards.texi (Standards): The Go frontend supports the Go 1
12627         language standard.
12629 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
12631         PR middle-end/56729
12632         * df-scan.c (df_insn_delete): Disable failing assert.
12634 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12636         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
12637         New function prototype.
12638         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
12639         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
12640         (arm_builtin_vectorized_function): New function.
12642 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12644         * config/arm/arm_neon_builtins.def: New file.
12645         * config/arm/arm.c (neon_builtin_data): Move contents to
12646         arm_neon_builtins.def.
12647         (enum arm_builtins): Include neon builtin definitions.
12648         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
12649         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
12651 2013-04-04  Marek Polacek  <polacek@redhat.com>
12653         PR tree-optimization/48186
12654         * predict.c (maybe_hot_frequency_p): Return false if
12655         HOT_BB_FREQUENCY_FRACTION is 0.
12656         (cgraph_maybe_hot_edge_p): Likewise.
12658 2013-04-04  Richard Biener  <rguenther@suse.de>
12660         PR tree-optimization/56826
12661         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
12662         more accurately.
12664 2013-04-04  Richard Biener  <rguenther@suse.de>
12666         PR tree-optimization/56213
12667         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
12668         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
12670 2013-04-04  Richard Biener  <rguenther@suse.de>
12672         PR tree-optimization/56837
12673         * tree-loop-distribution.c (classify_partition): For non-zero
12674         values require that the value has the same precision as its
12675         mode to be useful as memset value.
12677 2013-04-03  Nick Clifton  <nickc@redhat.com>
12679         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
12680         (fmssf4): Use fmsf.s on E3V5 architectures.
12681         (fnmasf4): Use fnmaf.s on E3V5 architectures.
12682         (fnmssf4): Use fnmsf.s on E3V5 architectures.
12684 2013-04-03  Jeff Law  <law@redhat.com>
12686         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
12687         (lra-eliminations.o): Likewise.
12689 2013-04-03  Teresa Johnson  <tejohnson@google.com>
12691         * gcov-io.c (compute_working_sets): Moved most of body of old
12692         compute_working_sets here from profile.c.
12693         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
12694         (gcov_working_set_t): Moved typedef here from basic-block.h
12695         (compute_working_set): Declare.
12696         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
12697         (get_working_sets): Renamed from compute_working_set,
12698         replace most of body with call to new compute_working_sets.
12699         (get_exec_counts): Replace call to compute_working_sets
12700         to get_working_sets.
12701         * profile.h (get_working_sets): Renamed from compute_working_set.
12702         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
12703         to get_working_sets.
12704         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
12705         * gcov-dump.c (dump_working_sets): New function.
12707 2013-04-03  Kenneth Zadeck <zadeck@naturalbridge.com>
12709         * hwint.c (sext_hwi, zext_hwi): New functions.
12710         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
12711         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
12712         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
12713         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
12714         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
12715         (sext_hwi, zext_hwi): New functions.
12717 2013-04-03  Jeff Law  <law@redhat.com>
12719         PR tree-optimization/56799
12720         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
12721         back test for widening conversion erroneously dropped in prior change.
12723 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12725         PR target/56809
12726         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
12727         instead of next_real_insn.
12729 2013-04-03  Marek Polacek  <polacek@redhat.com>
12731         PR sanitizer/55702
12732         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
12734 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12736         PR target/56809
12737         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
12738         next_real_insn.
12739         (thumb1_output_casesi): Likewise.
12740         (thumb2_output_casesi): Likewise.
12742 2013-04-03  Richard Biener  <rguenther@suse.de>
12744         PR tree-optimization/56817
12745         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
12746         Split out ...
12747         (tree_unroll_loops_completely_1): ... new function to manually
12748         walk the loop tree, properly defering outer loops of unrolled
12749         loops to later iterations.
12751 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12753         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
12754         (vectorizable_load): Likewise.
12755         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
12756         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
12758 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12760         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
12761         BIT_FIELD_REF.
12763 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12765         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
12767 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
12769         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
12771 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12773         PR tree-optimization/56790
12774         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
12775         folding.
12777 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
12779         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
12780         Handle VEC_MERGE.
12781         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
12782         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
12783         equal arguments.
12785 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
12787         PR c/19449
12788         * tree.h (force_folding_builtin_constant_p): New decl.
12789         * builtins.c (force_folding_builtin_constant_p): New variable.
12790         (fold_builtin_constant_p): Fold immediately also if
12791         force_folding_builtin_constant_p.
12793 2013-04-03  Richard Biener  <rguenther@suse.de>
12795         PR tree-optimization/56812
12796         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
12797         DRs of the same interleaving chain are independent.
12799 2013-04-02  Jason Merrill  <jason@redhat.com>
12801         * gdbinit.in (pbb): Use debug fn.
12803 2013-04-02  Lawrence Crowl  <crowl@google.com>
12805         * sese.h (struct ivtype_map_elt_s): Remove unused.
12806         (extern debug_ivtype_map): Remove unused.
12807         (extern eq_ivtype_map_elts): Remove unused.
12808         * sese.c (debug_ivtype_map): Removed unused.
12809         (debug_ivtype_map_1): Removed unused.
12810         (debug_ivtype_elt): Remove unused.
12811         (eq_ivtype_map_elts): Remove unused.
12814 2013-04-02  Kai Tietz  <ktietz@redhat.com>
12816         PR target/52790
12817         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
12818         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
12819         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
12820         function.
12821         (legitimize_pe_coff_symbol): Likewise.
12822         (is_imported_p): New helper-function.
12823         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
12824         for Windows x64 targets.
12825         (ix86_expand_prologue): Optimize for pe-coff targets.
12826         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
12827         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
12828         medium/large code-model.
12829         (legitimize_pic_address): Likewise.
12830         (legitimize_tls_address): Likewise.
12831         (ix86_expand_call): Likewise.
12832         (x86_output_mi_thunk): Likewise.
12833         (get_dllimport_decl): Add new beimport argument.
12834         (construct_plt_address): Don't assert for x64 pe-coff targets.
12835         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
12836         targets.
12837         (SYMBOL_FLAG_STUBVAR): New macro.
12838         (SYMBOL_REF_STUBVAR_P): Likewise.
12839         * config/i386/winnt.c (stub_list): New structure.
12840         (stub_head): New local variable.
12841         (i386_pe_record_stub): New function.
12842         (i386_pe_file_end): Emit refptr-stubs.
12844 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
12846         PR rtl-optimization/56745
12847         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
12848         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
12850         PR c++/34949
12851         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
12852         and both of them are MEM_REFs, just compare first argument for
12853         equality and attempt to deal even with differing offsets.
12855         PR c++/34949
12856         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
12857         of gimple_clobber_p to be MEM_REF.
12858         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
12859         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
12860         after gimplification.
12861         * asan.c (get_mem_ref_of_assignment): Don't instrument
12862         gimple_clobber_p stmts.
12863         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
12864         gimple_clobber_p stmt if they have MEM_REF lhs and
12865         are dead because of another gimple_clobber_p stmt.
12866         * tree-ssa-live.c (clear_unused_block_pointer): Treat
12867         gimple_clobber_p stmts like debug stmts.
12868         (remove_unused_locals): Remove clobbers with MEM_REF lhs
12869         that refer to unused VAR_DECLs or uninitialized values.
12870         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
12871         gimple_clobber_p stmts if they refer to removed parameters.
12872         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
12873         formatting.
12875 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
12877         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
12878         using SWI48 mode attribute.
12880 2013-04-02  Wei Mi  <wmi@google.com>
12882         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
12883         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
12884         *<rotate_insn><mode>3_mask in i386.md.
12886 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
12888         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
12890 2013-04-02  Richard Biener  <rguenther@suse.de>
12892         PR tree-optimization/56778
12893         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
12894         Runtime alias tests are not supported for gather loads.
12895         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
12896         stmts referenced from SSA operands before updating SSA form.
12898 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
12899             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12901         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
12902         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
12903         * config/arm/cortex-a53.md: New file.
12904         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
12905         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
12906         * config/arm/arm.c (arm_issue_rate): Likewise.
12907         * config/arm/arm-tune.md: Regenerate
12908         * config/arm/arm-tables.opt: Regenerate.
12909         * config/arm/arm-cores.def: Add cortex-a53.
12911 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
12913         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
12914         non-static link.
12916 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12918         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
12919         scalar load/store operations using B/H registers.
12920         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
12922 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
12924         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
12925         scalar move.
12926         * config/aarch64/aarch64.c
12927         (aarch64_simd_scalar_immediate_valid_for_move): New.
12928         * config/aarch64/aarch64-protos.h
12929         (aarch64_simd_scalar_immediate_valid_for_move): New.
12930         * config/aarch64/constraints.md (Dh, Dq): New.
12931         * config/aarch64/iterators.md (hq): New.
12933 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
12935         * reorg.c (get_branch_condition): Deal with conditional returns.
12936         (fill_simple_delay_slots): Remove dead code dealing with jumps.
12938 2013-04-01  Wei Mi  <wmi@google.com>
12940         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
12941         Truncate operand 2 using %b asm operand modifier.
12942         (*<shift_insn><mode>3_mask): Ditto.
12943         (*<rotate_insn><mode>3_mask): Ditto.
12945 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
12947         PR middle-end/56798
12948         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
12950 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
12952         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
12953         of next_real_insn.
12954         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
12956 2013-03-30  Lawrence Crowl  <crowl@google.com>
12958         * dse.c (clear_alias_sets): Remove never set.
12959         (disqualified_clear_alias_sets): Remove never set.
12960         (clear_alias_mode_pool): Remove never set.
12961         (dse_step0): Remove condition that is never true.
12962         (canon_address): Remove condition that is never true.
12963         (dse_step7): Remove condition that is never true.
12964         (rest_of_handle_dse): Remove condition that is never true.
12965         (rest_of_handle_dse::did_global): Remove never read from above.
12966         (dse_step2_spill): Remove never called from above.
12967         (dse_step5_spill): Remove never called from above.
12969 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
12971         * doc/md.texi (Standard Names) <casesi>: Update documentation for
12972         JUMP_TABLE_DATA changes.
12973         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
12974         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
12975         (Insns) <jump_table_data>: New entry.
12976         * doc/tm.texi: Regenerate.
12978         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
12980         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12981         for table jump at the end of a basic block using tablejump_p.
12982         * targhooks.c (default_invalid_within_doloop): Likewise.
12983         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12984         target hook implementation that is identical to the default hook.
12985         (rs6000_invalid_within_doloop): Remove.
12987         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12988         unused variable from tablejump_p call.
12990         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12991         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12992         (INSN_DELETED_P): Likewise.
12993         (emit_jump_table_data): New prototype.
12994         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12995         after 4th as unused.
12996         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12997         * sched-vis.c (print_insn): Likewise.
12998         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12999         insn for compatibility with back ends that use next_active_insn to
13000         identify jump table data.
13001         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
13002         (remove_insn): Likewise.
13003         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
13004         to be emitted.
13005         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
13006         (emit_jump_table_data): New function.
13008         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
13009         basic block, a JUMP_TABLE_DATA never is.
13010         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
13011         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
13012         off from code handling real insns.
13013         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
13014         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
13015         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
13016         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
13017         is not a NONDEBUG_INSN_P.
13018         * ira-costs.c (scan_one_insn): Likewise.
13019         * jump.c (mark_all_labels): Likewise.
13020         (mark_jump_label_1): Likewise.
13021         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13022         * lra.c (get_insn_freq): Expect all insns reaching here to be in
13023         a basic block.
13024         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
13025         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
13026         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
13027         JUMP_TABLE_DATA_P insns.
13028         (calculate_elim_costs_all_insns): Likewise.
13029         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
13030         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
13031         (delete_output_reload): Code style fixups.
13032         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
13033         insn flags on this non-insn.
13034         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
13035         as scheduling barriers, for pre-change compatibility.
13036         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
13037         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
13039         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
13040         redundant JUMP_TABLE_DATA_P test.
13041         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
13042         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13043         (frv_for_each_packet): Likewise.
13044         * config/i386/i386.c (min_insn_size): Likewise.
13045         (ix86_avoid_jump_mispredicts): Likewise.
13046         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13047         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
13048         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13049         (mips16_insn_length): Robustify.
13050         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
13051         (mips16_split_long_branches): Likewise.
13052         * config/pa/pa.c (pa_combine_instructions): Likewise.
13053         * config/rs6000/rs6000.c (get_next_active_insn): Treat
13054         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13055         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13056         as contributing to pool range lengths.
13057         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13058         Remove redundant JUMP_TABLE_DATA_P test.
13059         (sh_loop_align): Likewise.
13060         (split_branches): Likewise.
13061         (sh_insn_length_adjustment): Likewise.
13062         * config/spu/spu.c (get_branch_target): Likewise.
13064 2013-03-29  Jan Hubicka  <jh@suse.cz>
13066         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13067         gcov streaming; stream hot bb threshold to ltrans.
13068         * predict.c (get_hot_bb_threshold): Break out from ....
13069         (maybe_hot_count_p): ... here.
13070         (set_hot_bb_threshold): New function.
13071         * lto-section-in.c (lto_section_name): Add profile.
13072         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13073         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13074         and data-streamer.h
13075         (histogram_entry): New structure.
13076         (histogram, histogram_pool): New global vars.
13077         (histogram_hash): New structure.
13078         (histogram_hash::hash): New method.
13079         (histogram_hash::equal): Likewise.
13080         (account_time_size): New function.
13081         (cmp_counts): New function.
13082         (dump_histogram): New function.
13083         (ipa_profile_generate_summary): New function.
13084         (ipa_profile_write_summary): New function.
13085         (ipa_profile_read_summary): New function.
13086         (ipa_profile): Decide on threshold.
13087         (pass_ipa_profile): Add ipa_profile_write_summary and
13088         ipa_profile_read_summary.
13089         * Makefile.in (ipa.o): Update dependencies.
13090         * lto-streamer.h (LTO_section_ipa_profile): New section.
13092 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13094         * tree.h (VAR_P): New.
13096 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
13098         PR lto/56777
13099         * doc/invoke.texi ([-fwhole-program]): Fix typo.
13101 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
13103         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13104         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13105         (control_flow_insn_p): Likewise.
13106         * cfgrtl.c (duplicate_insn_chain): Likewise.
13107         * final.c (get_attr_length_1): Likewise.
13108         (shorten_branches): Likewise.
13109         (final_scan_insn): Likewise.
13110         * function.c (instantiate_virtual_regs): Likewise.
13111         * gcse.c (insert_insn_end_basic_block): Likewise.
13112         * ira-costs.c (scan_one_insn): Likewise.
13113         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13114         * lra.c (check_rtl): Likewise.
13115         * reload1.c (elimination_costs_in_insn): Likewise.
13116         * reorg.c (follow_jumps): Likewise.
13118         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13119         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13120         (thumb_far_jump_used_p): Likewise.
13121         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13122         (workaround_speculation): Likewise.
13123         (add_sched_insns_for_speculation): Likewise.
13124         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13125         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13126         (frv_for_each_packet): Likewise.
13127         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13128         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13129         (final_emit_insn_group_barriers): Likewise.
13130         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13131         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13132         (mips16_insn_length): Likewise.
13133         * config/pa/pa.c (pa_reorg): Likewise.
13134         (pa_combine_instructions): Likewise.
13135         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13136         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13137         (sh_reorg): Likewise.
13138         (split_branches): Likewise.
13139         * config/spu/spu.c (get_branch_target): Likewise.
13141         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13142         JUMP_TABLE_DATA_P.
13144 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
13146         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13147         Fix declaration name.
13149 2013-03-28  Lawrence Crowl  <crowl@google.com>
13151         * graphds.h (struct graph.indicies): Remove unused.
13152         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13153         (SCOP_ORIGINAL_PDDRS): Remove unused.
13154         * sese.h (extern insert_loop_close_phis): Removed unused.
13155         (extern insert_guard_phis): Removed unused.
13156         (extern ivtype_map_elt_info): Removed unused.
13157         (new_ivtype_map_elt): Removed unused.
13158         * sese.c (ivtype_map_elt_info): Removed unused.
13160 2013-03-28  Lawrence Crowl  <crowl@google.com>
13162         * Makefile.in: Add several missing include dependences.
13163         (DUMPFILE_H): New.
13164         (test-dump.o): New.  This object is not added to any executable,
13165         but is present for ad-hoc testing.
13166         * bitmap.c
13167         (debug (const bitmap_head_def &)): New.
13168         (debug (const bitmap_head_def *)): New.
13169         * bitmap.h
13170         (extern debug (const bitmap_head_def &)): New.
13171         (extern debug (const bitmap_head_def *)): New.
13172         * cfg.c
13173         (debug (edge_def &)): New.
13174         (debug (edge_def *)): New.
13175         * cfghooks.c
13176         (debug (basic_block_def &)): New.
13177         (debug (basic_block_def *)): New.
13178         * dumpfile.h
13179         (dump_node (const_tree, int, FILE *)): Correct source file.
13180         * dwarf2out.c
13181         (debug (die_struct &)): New.
13182         (debug (die_struct *)): New.
13183         * dwarf2out.h
13184         (extern debug (die_struct &)): New.
13185         (extern debug (die_struct *)): New.
13186         * gimple-pretty-print.c
13187         (debug (gimple_statement_d &)): New.
13188         (debug (gimple_statement_d *)): New.
13189         * gimple-pretty-print.h
13190         (extern debug (gimple_statement_d &)): New.
13191         (extern debug (gimple_statement_d *)): New.
13192         * ira-build.c
13193         (debug (ira_allocno_copy &)): New.
13194         (debug (ira_allocno_copy *)): New.
13195         (debug (ira_allocno &)): New.
13196         (debug (ira_allocno *)): New.
13197         * ira-int.h
13198         (extern debug (ira_allocno_copy &)): New.
13199         (extern debug (ira_allocno_copy *)): New.
13200         (extern debug (ira_allocno &)): New.
13201         (extern debug (ira_allocno *)): New.
13202         * ira-lives.c
13203         (debug (live_range &)): New.
13204         (debug (live_range *)): New.
13205         * lra-int.h
13206         (debug (lra_live_range &)): New.
13207         (debug (lra_live_range *)): New.
13208         * lra-lives.c
13209         (debug (lra_live_range &)): New.
13210         (debug (lra_live_range *)): New.
13211         * omega.c
13212         (debug (omega_pb_d &)): New.
13213         (debug (omega_pb_d *)): New.
13214         * omega.h
13215         (extern debug (omega_pb_d &)): New.
13216         (extern debug (omega_pb_d *)): New.
13217         * print-rtl.c
13218         (debug (const rtx_def &)): New.
13219         (debug (const rtx_def *)): New.
13220         * print-tree.c
13221         (debug_tree (tree): Move within file.
13222         (debug_raw (const tree_node &)): New.
13223         (debug_raw (const tree_node *)): New.
13224         (dump_tree_via_hooks (const tree_node *, int)): New.
13225         (debug (const tree_node &)): New.
13226         (debug (const tree_node *)): New.
13227         (debug_verbose (const tree_node &)): New.
13228         (debug_verbose (const tree_node *)): New.
13229         (debug_head (const tree_node &)): New.
13230         (debug_head (const tree_node *)): New.
13231         (debug_body (const tree_node &)): New.
13232         (debug_body (const tree_node *)): New.
13233         (debug_vec_tree (tree): Move and reimplement in terms of dump.
13234         (debug (vec<tree, va_gc> &)): New.
13235         (debug (vec<tree, va_gc> *)): New.
13236         * rtl.h
13237         (extern debug (const rtx_def &)): New.
13238         (extern debug (const rtx_def *)): New.
13239         * sbitmap.c
13240         (debug_raw (simple_bitmap_def &)): New.
13241         (debug_raw (simple_bitmap_def *)): New.
13242         (debug (simple_bitmap_def &)): New.
13243         (debug (simple_bitmap_def *)): New.
13244         * sbitmap.h
13245         (extern debug (simple_bitmap_def &)): New.
13246         (extern debug (simple_bitmap_def *)): New.
13247         (extern debug_raw (simple_bitmap_def &)): New.
13248         (extern debug_raw (simple_bitmap_def *)): New.
13249         * sel-sched-dump.c
13250         (debug (vinsn_def &)): New.
13251         (debug (vinsn_def *)): New.
13252         (debug_verbose (vinsn_def &)): New.
13253         (debug_verbose (vinsn_def *)): New.
13254         (debug (expr_def &)): New.
13255         (debug (expr_def *)): New.
13256         (debug_verbose (expr_def &)): New.
13257         (debug_verbose (expr_def *)): New.
13258         (debug (vec<rtx> &)): New.
13259         (debug (vec<rtx> *)): New.
13260         * sel-sched-dump.h
13261         (extern debug (vinsn_def &)): New.
13262         (extern debug (vinsn_def *)): New.
13263         (extern debug_verbose (vinsn_def &)): New.
13264         (extern debug_verbose (vinsn_def *)): New.
13265         (extern debug (expr_def &)): New.
13266         (extern debug (expr_def *)): New.
13267         (extern debug_verbose (expr_def &)): New.
13268         (extern debug_verbose (expr_def *)): New.
13269         (extern debug (vec<rtx> &)): New.
13270         (extern debug (vec<rtx> *)): New.
13271         * sel-sched-ir.h
13272         (_list_iter_cond_expr): Make inline instead of static.
13273         * sreal.c
13274         (debug (sreal &)): New.
13275         (debug (sreal *)): New.
13276         * sreal.h
13277         (extern debug (sreal &)): New.
13278         (extern debug (sreal *)): New.
13279         * tree.h
13280         (extern debug_raw (const tree_node &)): New.
13281         (extern debug_raw (const tree_node *)): New.
13282         (extern debug (const tree_node &)): New.
13283         (extern debug (const tree_node *)): New.
13284         (extern debug_verbose (const tree_node &)): New.
13285         (extern debug_verbose (const tree_node *)): New.
13286         (extern debug_head (const tree_node &)): New.
13287         (extern debug_head (const tree_node *)): New.
13288         (extern debug_body (const tree_node &)): New.
13289         (extern debug_body (const tree_node *)): New.
13290         (extern debug (vec<tree, va_gc> &)): New.
13291         (extern debug (vec<tree, va_gc> *)): New.
13292         * tree-cfg.c
13293         (debug (struct loop &)): New.
13294         (debug (struct loop *)): New.
13295         (debug_verbose (struct loop &)): New.
13296         (debug_verbose (struct loop *)): New.
13297         * tree-dump.c: Add header dependence.
13298         * tree-flow.h
13299         (extern debug (struct loop &)): New.
13300         (extern debug (struct loop *)): New.
13301         (extern debug_verbose (struct loop &)): New.
13302         (extern debug_verbose (struct loop *)): New.
13303         * tree-data-ref.c
13304         (debug (data_reference &)): New.
13305         (debug (data_reference *)): New.
13306         (debug (vec<data_reference_p> &)): New.
13307         (debug (vec<data_reference_p> *)): New.
13308         (debug (vec<ddr_p> &)): New.
13309         (debug (vec<ddr_p> *)): New.
13310         * tree-data-ref.h
13311         (extern debug (data_reference &)): New.
13312         (extern debug (data_reference *)): New.
13313         (extern debug (vec<data_reference_p> &)): New.
13314         (extern debug (vec<data_reference_p> *)): New.
13315         (extern debug (vec<ddr_p> &)): New.
13316         (extern debug (vec<ddr_p> *)): New.
13317         * tree-ssa-alias.c
13318         (debug (pt_solution &)): New.
13319         (debug (pt_solution *)): New.
13320         * tree-ssa-alias.h
13321         (extern debug (pt_solution &)): New.
13322         (extern debug (pt_solution *)): New.
13323         * tree-ssa-alias.c
13324         (debug (_var_map &)): New.
13325         (debug (_var_map *)): New.
13326         (debug (tree_live_info_d &)): New.
13327         (debug (tree_live_info_d *)): New.
13328         * tree-ssa-alias.h
13329         (extern debug (_var_map &)): New.
13330         (extern debug (_var_map *)): New.
13331         (extern debug (tree_live_info_d &)): New.
13332         (extern debug (tree_live_info_d *)): New.
13334 2013-03-28  Jan Hubicka  <jh@suse.cz>
13336         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
13338 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
13340         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
13341         record only when desired or required.
13343 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
13345         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
13346         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
13348 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13350         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
13351         (*andsi3_compare0_uxtw): New pattern.
13352         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
13353         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
13355 2013-03-28  Jan Hubicka  <jh@suse.cz>
13357         * data-streamer-in.c (streamer_read_gcov_count): New function.
13358         * gimple-streamer-out.c: Include value-prof.h.
13359         (output_gimple_stmt): Output histogram.
13360         (output_bb): Use streamer_write_gcov_count.
13361         * value-prof.c: Include data-streamer.h
13362         (dump_histogram_value): Add HIST_TYPE_MAX.
13363         (stream_out_histogram_value): New function.
13364         (stream_in_histogram_value): New function.
13365         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
13366         (stream_out_histogram_value, stream_in_histogram_value): Declare.
13367         * data-streamer-out.c (streamer_write_gcov_count): New function.
13368         (streamer_write_gcov_count_stream): New function.
13369         * lto-cgraph.c (lto_output_edge): Update counter streaming.
13370         (lto_output_node): Likewise.
13371         (input_node, input_edge): Likewise.
13372         * lto-streamer-out.c (output_cfg): Update streaming.
13373         * lto-streamer-in.c (input_cfg): Likewise.
13374         * data-streamer.h (streamer_write_gcov_count,
13375         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
13376         * gimple-streamer-in.c: Include value-prof.h
13377         (input_gimple_stmt): Input histograms.
13378         (input_bb): Update profile streaming.
13380 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
13382         * genmodes.c (emit_max_int): New function.
13383         (emit_insn_modes_h): Added call to emit_max_function.
13384         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
13385         Added doc.
13386         * machmode.def: Fixed comment.
13388 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
13390         * combine.c (try_combine): Removed useless assert.
13391         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
13393 2013-03-28  Marek Polacek  <polacek@redhat.com>
13394             Richard Biener  <rguenther@suse.de>
13396         PR tree-optimization/56695
13397         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
13398         build signed result of a vector comparison.
13399         * tree-cfg.c (verify_gimple_comparison): Check that a result
13400         of a vector comparison has signed type.
13402 2013-03-28  Richard Biener  <rguenther@suse.de>
13404         PR tree-optimization/37021
13405         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
13406         do not restrict gaps between groups.
13407         * tree-vect-stmts.c (vectorizable_load): Properly account for
13408         a gap between groups.
13410 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
13412         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
13413         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
13414         is not enabled.
13416 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
13418         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
13419         * doc/extend.texi (Named Address Spaces): Ditto.
13420         (Variable Attributes): Ditto.
13422 2013-03-27  Kai Tietz  <ktietz@redhat.com>
13424         * config.build: Add support for cygwin x64 target.
13425         * config.gcc: Likewise.
13426         * config.host: Likewise.
13427         * configure.ac: Likewise
13428         * configure: Regenerated.
13430 2013-03-27  Kai Tietz  <ktietz@redhat.com>
13432         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
13433         * config/i386/t-cygwin-w64: New file.
13434         * config/i386/cygwin-w64.h: New file.
13435         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
13436         and add support for x64-cygwin target.
13437         (CPP_SPEC): Likewise.
13438         (CXX_WRAP_SPEC_LIST): Undefine before define.
13439         (LIBGCJ_SONAME): Use 15 as version.
13441 2013-03-27  Richard Biener  <rguenther@suse.de>
13443         PR tree-optimization/56716
13444         * tree-ssa-structalias.c (perform_var_substitution): Adjust
13445         dumping for ref nodes.
13447 2013-03-27  Martin Jambor  <mjambor@suse.cz>
13449         PR tree-optimization/55334
13450         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
13451         restricted pointers to arrays.
13453 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13455         * Makefile.in (.SUFFIXES): Add .cc.
13456         (.c.o): Apply same recipe for implicit rule .cc.o.
13458 2013-03-27  Richard Biener  <rguenther@suse.de>
13460         PR tree-optimization/37021
13461         * tree-vect-data-refs.c (vect_check_strided_load): Allow
13462         REALPART/IMAGPART_EXPRs around the supported refs.
13463         * tree-ssa-structalias.c (find_func_aliases): Assume that
13464         floating-point values are not used to transfer pointers.
13466 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
13468         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
13469         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
13470         * gcc/doc/tm.texi: Regenerate.
13471         * gcc/targhooks.h (default_has_ifunc_p): New.
13472         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
13473         * gcc/config/linux-protos.h: New file.
13474         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
13475         this hook for linux which disables support of indirect functions in
13476         android.
13477         * gcc/config/linux-android.c: New file.
13478         * gcc/config/t-linux-android.c: Ditto.
13479         * gcc/config.gcc: Added new object file linux-android.o.
13480         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
13481         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
13482         * gcc/varasm.c (do_assemble_alias): Likewise.
13483         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
13484         doesn't support indirect functions.
13485         * configure: Regenerate.
13487 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
13489         PR target/56102
13490         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
13491         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
13492         mult-word mode.
13494 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13496         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
13498 2013-03-27  Terry Guo  <terry.guo@arm.com>
13500         * config/arm/arm-cores.def: Added core cortex-r7.
13501         * config/arm/arm-tune.md: Regenerated.
13502         * config/arm/arm-tables.opt: Regenerated.
13503         * doc/invoke.texi: Added entry for core cortex-r7.
13505 2013-03-27  Walter Lee  <walt@tilera.com>
13507         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
13508         double-decrement of next_scratch_regno.
13510 2013-03-27  Walter Lee  <walt@tilera.com>
13512         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
13513         input operands.
13514         (insn_v1mulus): Ditto.
13515         (insn_v2muls): Ditto.
13517 2013-03-27  Walter Lee  <walt@tilera.com>
13519         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
13520         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
13522 2013-03-27  Walter Lee  <walt@tilera.com>
13524         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
13525         (*sibcall_value): Ditto.
13527 2013-03-27  Walter Lee  <walt@tilera.com>
13529         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
13530         (insn_mnz_v8qi): ... this ...
13531         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
13532         vector equivalent.
13533         (insn_v<n>mnz): Replaced by ...
13534         (insn_v1mnz): ... this ...
13535         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
13536         equivalent.
13537         (insn_mz_<mode>): Replaced by ...
13538         (insn_mz_v8qi): ... this ...
13539         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
13540         vector equivalent.
13541         (insn_v<n>mz): Replaced by ...
13542         (insn_v1mz): ... this ...
13543         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
13544         equivalent.
13546 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
13548         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
13550 2013-03-26  Roland McGrath  <mcgrathr@google.com>
13552         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
13553         than fprintf with a non-constant, non-format string.
13555 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13557         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
13558         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
13559         operand 0 predicate.
13560         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
13561         attribute.  Use general_x64nomem_operand as operand 1 predicate.
13562         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
13563         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13564         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
13565         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13566         (mov<mode>_insv_1): Remove expander.  Merge insn with
13567         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
13568         Use general_x64nomem_operand as operand 1 predicate.
13569         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
13570         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
13571         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13572         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
13573         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
13574         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
13575         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13576         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
13577         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
13578         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
13579         (general_x64nomem_operand): Ditto.
13581 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
13583         * config/rtems.opt: Add -pthread option.
13585 2013-03-26  Richard Biener  <rguenther@suse.de>
13587         * alias.c (find_base_term): Avoid redundant and not used recursion.
13588         (base_alias_check): Get the initial base term from the caller.
13589         (true_dependence_1): Compute and pass base terms to base_alias_check.
13590         (write_dependence_p): Likewise.
13591         (may_alias_p): Likewise.
13593 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
13595         * config/aarch64/aarch64.c (aarch64_classify_address): Support
13596         PC-relative load in SI modes and above only.
13598 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
13600         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
13601         * config/arm/iwmmxt.md (WCGR0): Update.
13602         (WCGR1, WCGR2, WCGR3): Likewise.
13604 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13606         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
13607         Use x64 and nox64 isa attributes.
13609 2013-03-26  Richard Biener  <rguenther@suse.de>
13611         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
13612         alignment computations and rely on get_object_alignment_1
13613         for the !TYPE_P case.
13614         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
13616 2013-03-26  Walter Lee  <walt@tilera.com>
13618         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
13619         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
13621 2013-03-25  Jeff Law  <law@redhat.com>
13623         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
13624         check for INTEGRAL_TYPE_P that was missing due to checking in
13625         wrong version of prior patch.
13627 2013-03-25  Walter Lee  <walt@tilera.com>
13629         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
13630         TILEGX_INSN_SHUFFLEBYTES1.
13631         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
13632         shufflebytes1.
13633         (tilegx_builtins): Ditto.
13634         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
13636 2013-03-25  Walter Lee  <walt@tilera.com>
13638         * config/tilegx/tilegx.md (floatsisf2): New pattern.
13639         (floatunssisf2): New pattern.
13640         (floatsidf2): New pattern.
13641         (floatunssidf2): New pattern.
13643 2013-03-25  Walter Lee  <walt@tilera.com>
13645         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
13646         tests for constraint J, K, N, P.
13648 2013-03-25  Walter Lee  <walt@tilera.com>
13650         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
13651         Use indirect/pcrel encoding.
13652         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
13653         Ditto.
13655 2013-03-25  Steve Ellcey  <sellcey@mips.com>
13657         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
13658         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
13659         * config/mips/mips.c (mips_option_override): Set IMADD default.
13660         * config/mips/mips.h (PTF_AVOID_IMADD): New.
13661         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
13662         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
13663         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
13664         * doc/invoke.texi (-mimadd/-mno-imadd): New.
13666 2013-03-25  Jeff Law  <law@redhat.com>
13668         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
13669         slightly to avoid creating and folding useless trees.  Simplify
13670         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
13672 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
13674         * config/i386/i386.md (*zero_extendsidi2): Merge with
13675         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
13676         * config/i386/predicates.md (x86_64_zext_operand): Rename from
13677         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
13678         targets.  Clarify comment.
13680 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13682         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
13683         pass-through jump functions differently.
13684         (ipa_read_jump_function): Likewise.  Also use setter functions to set
13685         up jump functions.
13687 2013-03-25  Martin Jambor  <mjambor@suse.cz>
13689         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
13690         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
13691         process it.
13692         (ipa_get_indirect_edge_target): New function.
13693         (devirtualization_time_bonus): New parameter known_aggs, pass it to
13694         ipa_get_indirect_edge_target.  Update all callers.
13695         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
13696         ipa_get_indirect_edge_target_1 instead of calling
13697         ipa_get_indirect_edge_target.
13698         (create_specialized_node): Pass aggvlas to
13699         ipcp_discover_new_direct_edges.
13701 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13703         * config/arm/arm.md (f_sels, f_seld): New types.
13704         (*cmov<mode>): New pattern.
13705         * config/arm/predicates.md (arm_vsel_comparison_operator): New
13706         predicate.
13708 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13710         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
13711         POSIX-printf for mingw-hosted  builds.
13713 2013-03-25  Richard Biener  <rguenther@suse.de>
13715         PR middle-end/56694
13716         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
13717         must-not-throw stmt location.
13719 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13721         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
13722         Emit load-acquire versions when acq is true.
13723         (arm_emit_store_exclusive): Add rel parameter.
13724         Emit store-release versions when rel is true.
13725         (arm_split_compare_and_swap): Use acquire-release instructions
13726         instead.
13727         of barriers when appropriate.
13728         (arm_split_atomic_op): Likewise.
13729         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
13730         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
13731         (VUNSPEC_SLX): Likewise.
13732         (VUNSPEC_LDA): Likewise.
13733         (VUNSPEC_STL): Likewise.
13734         * config/arm/sync.md (atomic_load<mode>): New pattern.
13735         (atomic_store<mode>): Likewise.
13736         (arm_load_acquire_exclusive<mode>): Likewise.
13737         (arm_load_acquire_exclusivesi): Likewise.
13738         (arm_load_acquire_exclusivedi): Likewise.
13739         (arm_store_release_exclusive<mode>): Likewise.
13741 2013-03-25  Catherine Moore  <clm@codesourcery.com>
13743         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
13744         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
13745         * config/mip/predicates.md (lwsp_swsp_operand,
13746         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
13747         sb16_operand, db4_operand, db7_operand, ib3_operand,
13748         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
13749         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
13750         andi16_operand): New predicates.
13751         * config/mips/mips.md (compression): New attribute.
13752         (enabled): New attribute.
13753         (length): Consider compression in computing length.
13754         (shift_compression): New code attribute.
13755         (*add<mode>3): New operands. Record compression.
13756         (sub<mode>3): Likewise.
13757         (one_cmpl<mode>2): Likewise.
13758         (*and<mode>3): Likewise.
13759         (*ior<mode>3): Likewise.
13760         (unnamed pattern for xor): Likewise.
13761         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
13762         (*<optab><mode>3): Likewise.
13763         (*mov<mode>_internal: Likewise.
13764         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
13765         (mips_unsigned_immediate_p): New.
13766         (umips_lwsp_swsp_address_p): New.
13767         (m16_based_address_p): New.
13768         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
13769         (mips_unsigned_immediate_p): New prototype.
13770         (lwsp_swsp_address_p): New prototype.
13771         (m16_based_address_p): New prototype.
13772         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
13773         (mips_signed_immediate_p): New function.
13774         (m16_based_address_p): New function.
13775         (lwsp_swsp_address_p): New function.
13776         (mips_print_operand_punctuation): Recognize short delay slot insns
13777         for microMIPS.add<mode>3"
13779 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13781         PR target/56720
13782         * config/arm/iterators.md (v_cmp_result): New mode attribute.
13783         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
13785 2013-03-25  Richard Biener  <rguenther@suse.de>
13787         PR tree-optimization/56689
13788         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
13789         any edge.
13791 2013-03-25  Richard Biener  <rguenther@suse.de>
13793         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
13794         of bitmap.
13795         (memory_references): Likewise.
13796         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
13797         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
13798         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
13799         (gather_mem_refs_in_loops): Fold into ...
13800         (analyze_memory_references): ... this.  Move initialization
13801         to tree_ssa_lim_initialize.
13802         (fill_always_executed_in): Rename to ...
13803         (fill_always_executed_in_1): ... this.
13804         (fill_always_executed_in): Move contains_call computation to
13805         this new function from ...
13806         (tree_ssa_lim_initialize): ... here.
13807         (tree_ssa_lim): Call fill_always_executed_in.
13809 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13811         * postreload.c (reload_combine): Fix code detecting returns.
13813 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
13815         * function.c (emit_use_return_register_into_block): On cc0 targets,
13816         do not emit the sequence between cc0 setter and user.
13818 2013-03-25  Kai Tietz  <ktietz@redhat.com>
13820         * config/i386/predicates.md (local_symbolic_operand): Interpret
13821         dll-imported symbols as none-local.
13823 2013-03-25  Richard Biener  <rguenther@suse.de>
13825         * tree-ssa-loop-im.c (struct depend): Remove.
13826         (struct lim_aux_data): Make depends a vec of gimples.
13827         (free_lim_aux_data): Adjust.
13828         (add_dependency): Likewise.
13829         (set_level): Likewise.
13831 2013-03-25  Richard Biener  <rguenther@suse.de>
13833         PR middle-end/56434
13834         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
13835         the pointer returned by calls with ECF_MALLOC set.
13837 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13839         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
13841 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
13843         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
13844         using MMXMODE mode iterator.
13845         (*move<mode>_internal): Merge with *movv2sf_internal and
13846         *movv2sf_internal_rex64 using MMXMODE mode iterator.
13848 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13850         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
13851         (record_last_mem_set_info): Likewise.
13853         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
13854         of XNEWVEC followed by memset.
13855         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
13857 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
13859         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
13860         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
13861         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
13862         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
13863         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
13864         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
13865         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
13866         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
13867         BARRIER_P instead of GET_CODE.
13869 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13871         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
13872         inaccuracy in the probing code.
13874         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
13875         (ctrapdi4): Likewise.
13877 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13879         * calls.c (expand_call): Add missing guard to code handling return
13880         of non-BLKmode structures in MSB.
13881         * function.c (expand_function_end): Likewise.
13883 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
13885         * combine.c (try_combine): Adjust comment.  Do not add the set of
13886         insn #0 if the destination indirectly is set or dies in insn #2.
13887         Tidy up code to distribute a new note.
13889 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13891         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
13892         also for alternatives 16 and 17.
13894 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
13896         * config/i386/sse.md (*mov<mode>_internal): Merge with
13897         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
13898         Emit insn template depending on type attribute.  Use
13899         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13900         movd instead of movq mnemonic for interunit moves.  Rewrite mode
13901         attribute calculation.  Remove unit attribute calculation.
13902         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
13903         Set prefix_data16 attribute for DImode ssemov types.
13904         Use Ym instead of y for SSE-MMX conversion alternatives.
13905         Reorder operand constraints.
13907 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
13909         * df.h (df_insn_delete): Adjust prototype.
13910         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
13911         and let it decide whether mark the basic block dirty.
13912         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
13913         * df-scan.c (df_insn_info_delete): New helper function, split
13914         off from df_insn_delete.
13915         (df_scan_free_bb_info): Use it.
13916         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
13917         Likewise.
13918         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
13919         that the insn is actually an insn and it has a non-NULL basic block.
13920         Do not mark basic block dirty if only deleting a DEBUG_INSN.
13922 2013-03-22  Richard Biener  <rguenther@suse.de>
13924         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
13925         dep_ref members.
13926         (mem_ref_alloc): Do not allocate them.
13927         (refs_independent_p): Do not query or maintain a cache.
13929 2013-03-22  Richard Biener  <rguenther@suse.de>
13931         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
13932         (gather_mem_refs_in_loops): Do not compute it.
13933         (analyze_memory_references): Do not allocate it.
13934         (tree_ssa_lim_finalize): Do not free it.
13935         (for_all_locs_in_loop): Do not query all_refs_in_loop.
13937 2013-03-22  Richard Biener  <rguenther@suse.de>
13939         * is-a.h (as_a): Use gcc_checking_assert.
13941 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
13943         * config/aarch64/aarch64.c (aarch64_print_operand): New
13944         format specifier for printing a constant in hex.
13945         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
13946         format specifier for printing second operand.
13948 2013-03-22  Richard Biener  <rguenther@suse.de>
13950         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
13951         bitmaps.
13952         (gather_mem_refs_in_loops): Perform store accumulation here.
13953         (create_vop_ref_mapping_loop): Remove.
13954         (create_vop_ref_mapping): Likewise.
13955         (analyze_memory_references): Initialize refs_stored_in_loop.
13956         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
13957         (record_indep_loop): Remove.
13958         (record_dep_loop): New function.
13959         (ref_indep_loop_p_1): Adjust to only walk over references
13960         in the loop, not its subloops.
13961         (ref_indep_loop_p): Rename to ...
13962         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
13963         maintaining a more fine-grained cache.
13964         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
13965         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
13967 2013-03-22  Richard Biener  <rguenther@suse.de>
13969         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
13970         (struct mem_ref): Make accesses_in_loop a vec of a vec of
13971         aggregate mem_ref_loc.
13972         (free_mem_ref_locs): Inline into ...
13973         (memref_free): ... this and adjust.
13974         (mem_ref_alloc): Adjust.
13975         (mem_ref_locs_alloc): Remove.
13976         (record_mem_ref_loc): Adjust.
13977         (get_all_locs_in_loop): Rewrite into ...
13978         (for_all_locs_in_loop): ... this iterator.
13979         (rewrite_mem_ref_loc): New functor.
13980         (rewrite_mem_refs): Use for_all_locs_in_loop.
13981         (sm_set_flag_if_changed): New functor.
13982         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13983         (ref_always_accessed): New functor.
13984         (ref_always_accessed_p): Use for_all_locs_in_loop.
13986 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
13988         * tree-pass.h (PROP_gimple_lvec): New.
13989         * passes.c (dump_properties): Handle PROP_gimple_lvec.
13990         (init_optimization_passes): Move pass_lower_vector.
13991         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13992         PROP_gimple_lvec.
13993         (pass_lower_vector): Provide PROP_gimple_lvec.
13994         (pass_lower_vector_ssa): Likewise.
13995         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13997 2013-03-21  Mark Wielaard  <mjw@redhat.com>
13999         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
14001 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14003         * config/i386/i386.md (*movdi_internal): Disparage slightly
14004         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
14005         conversion alternatives.
14007 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
14009         PR middle-end/48087
14010         * diagnostic.def (DK_WERROR): New kind.
14011         * diagnostic.h (werrorcount): Define.
14012         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
14013         promoted to DK_ERROR, increment DK_WERROR counter instead of
14014         DK_ERROR counter.
14015         * toplev.c (toplev_main): Call print_ignored_options even if
14016         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
14017         even if just werrorcount is non-zero.
14019         PR debug/55608
14020         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
14021         on failure.
14022         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
14023         (string_cst_pool_decl): New function.
14024         (optimize_one_addr_into_implicit_ptr): New function.
14025         (resolve_addr_in_expr): Optimize DWARF location expression
14026         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
14027         which doesn't live in memory, but has DW_AT_location or
14028         DW_AT_const_value, or refers to a string literal, into
14029         DW_OP_GNU_implicit_pointer.
14030         (optimize_location_into_implicit_ptr): New function.
14031         (resolve_addr): If removing DW_AT_location of a variable because
14032         it was DW_OP_addr of address of the variable, but the variable doesn't
14033         live in memory, try to emit const value attribute for the initializer.
14035 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14037         * tree.h (VECTOR_TYPE_P): New macro.
14038         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
14039         TYPE_MODE): Use it.
14040         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
14041         VEC_COND_EXPR cannot be lvalues.
14042         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
14044 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14046         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
14047         Restrict the transformation to equal modes.
14049 2013-03-21  Richard Biener  <rguenther@suse.de>
14051         PR tree-optimization/39326
14052         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14053         (MEM_ANALYZABLE): Adjust.
14054         (record_mem_ref_loc): Move bitmap ops ...
14055         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
14056         unanalyzable refs, do not record locations for it.
14057         (analyze_memory_references): Allocate ref zero as shared
14058         unanalyzable ref.
14059         (refs_independent_p): Do not test for unanalyzed mems here.
14060         (ref_indep_loop_p_1): Special-case disambiguation against
14061         the unanalyzed ref.
14062         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14064 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
14066         * config/arm/arm-protos.h (tune_params): Add
14067         prefer_neon_for_64bits field.
14068         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14069         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14070         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14071         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14072         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14073         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14074         (arm_option_override): Handle -mneon-for-64bits new option.
14075         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14076         (prefer_neon_for_64bits): Declare new variable.
14077         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14078         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14079         (arch_enabled): Handle new arch types. Remove support for onlya8
14080         and nota8.
14081         (one_cmpldi2): Use new arch names.
14082         (zero_extend<mode>di2, extend<mode>di2): Ditto.
14083         * config/arm/arm.opt (mneon-for-64bits): Add option.
14084         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14085         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14086         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14087         of onlya8.
14088         * doc/invoke.texi (-mneon-for-64bits): Document.
14090 2013-03-21  Richard Biener  <rguenther@suse.de>
14092         PR tree-optimization/39326
14093         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14094         (sort_bbs_in_loop_postorder_cmp): New function.
14095         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14097 2013-03-21  Richard Biener  <rguenther@suse.de>
14099         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14100         (vect_insert_into_interleaving_chain): Likewise.
14101         (vect_drs_dependent_in_basic_block): Inline ...
14102         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
14103         split out from ...
14104         (vect_analyze_data_ref_dependence): ... here.  Simplify.
14105         (vect_check_interleaving): Simplify.
14106         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
14107         (vect_slp_analyze_data_ref_dependences): ... this new function.
14108         (dr_group_sort_cmp): New function.
14109         (vect_analyze_data_ref_accesses): Compute data-reference groups
14110         here instead of in vect_analyze_data_ref_dependence.  Use
14111         a more efficient algorithm.
14112         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14113         vect_slp_analyze_data_ref_dependences.  Call
14114         vect_analyze_data_ref_accesses earlier.
14115         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14116         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14117         (vect_slp_analyze_data_ref_dependences): New prototype.
14119 2013-03-21  Richard Biener  <rguenther@suse.de>
14121         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14122         ref is stored in the loop.
14123         (find_refs_for_sm): Walk only over all stores.
14124         (store_motion_loop): Allocate from lim_bitmap_obstack.
14125         (store_motion): Likewise.
14127 2013-03-21  Richard Biener  <rguenther@suse.de>
14129         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14130         Update virtual SSA form.
14132 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14134         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14135         * configure: Regenerate.
14136         * config.in: Regenerate.
14137         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14138         if !HAVE_LD_EH_FRAME_CIEV3.
14140 2013-03-21  Richard Biener  <rguenther@suse.de>
14142         * tree-cfg.c (verify_expr_no_block): New function.
14143         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14144         nor DECL_VALUE_EXPR have locations with associated blocks.
14145         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14146         (clear_unused_block_pointer): Remove code dealing with
14147         blocks in DECL_DEBUG_EXPR locations.
14149 2013-03-21  Richard Biener  <rguenther@suse.de>
14151         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14152         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
14153         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14154         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14155         instead of DECL_DEBUG_EXPR_IS_FROM.
14156         * gimplify.c (gimplify_modify_expr): Likewise.
14157         * tree-cfg.c (verify_expr_location_1): Likewise.
14158         * tree-complex.c (create_one_component_var): Likewise.
14159         * tree-sra.c (create_access_replacement): Likewise.
14160         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14161         (clear_unused_block_pointer): Likewise.
14162         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14163         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14164         * var-tracking.c (var_debug_decl): Likewise.
14165         (track_expr_p): Likewise.
14166         * tree-inline.c (add_local_variables): Likewise.  Set
14167         DECL_HAS_DEBUG_EXPR_P after copying it.
14168         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14169         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
14171 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14173         PR bootstrap/56656
14174         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14175         * configure: Regenerate.
14176         * config.in: Regenerate.
14177         * config/i386/i386.md (*movdf_internal): Use
14178         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14179         movd instead of movq mnemonic for interunit moves.
14180         (*movdi_internal): Ditto.
14182 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14184         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14185         (abd<mode>_3): New pattern.
14186         (aba<mode>_3): New pattern.
14187         (fabd<mode>_3): New pattern.
14189 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14191         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14192         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14193         occurrence of REGISTER_PREFIX as its empty string.
14195 2013-03-20  Jeff Law  <law@redhat.com>
14197         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14198         addititional equivalences for equality comparisons between an SSA_NAME
14199         and a constant where the SSA_NAME was set from a widening conversion.
14201 2013-03-20  Walter Lee  <walt@tilera.com>
14203         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14205 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14207         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14208         depending on type attribute.
14209         (*movti_internal): Ditto.
14210         (*movtf_internal): Ditto.
14211         (*movxf_internal): Ditto.
14212         (*movdf_internal): Ditto.
14213         (*movsf_internal): Ditto.
14215 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14217         * config/i386/i386.md (*movti_internal): Set prefix attribute to
14218         maybe_vex for sselog1 and ssemov types.
14219         (*movdi_internal): Reorder operand constraints.
14220         (*movsi_internal): Ditto.  Set prefix attribute to
14221         maybe_vex for sselog1 and ssemov types.
14222         (*movtf_internal): Set prefix attribute to maybe_vex
14223         for sselog1 and ssemov types.
14224         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
14225         DImode ssemov types.  Reorder operand constraints.
14226         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
14227         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
14228         attribute for SImode ssemov types.  Reorder operand constraints.
14230 2013-03-20  Martin Jambor  <mjambor@suse.cz>
14232         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14233         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14235 2013-03-20  Pat Haugen <pthaugen@us.ibm.com>
14237         * config/rs6000/predicates.md (indexed_address, update_address_mem
14238         update_indexed_address_mem): New predicates.
14239         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14240         attribute for load/store instructions.
14241         * config/rs6000/dfp.md (movsd_store): Likewise.
14242         (movsd_load): Likewise.
14243         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14244         (unnamed HI->DI extend define_insn): Likewise.
14245         (unnamed SI->DI extend define_insn): Likewise.
14246         (unnamed QI->SI extend define_insn): Likewise.
14247         (unnamed QI->HI extend define_insn): Likewise.
14248         (unnamed HI->SI extend define_insn): Likewise.
14249         (unnamed HI->SI extend define_insn): Likewise.
14250         (extendsfdf2_fpr): Likewise.
14251         (movsi_internal1): Likewise.
14252         (movsi_internal1_single): Likewise.
14253         (movhi_internal): Likewise.
14254         (movqi_internal): Likewise.
14255         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
14256         attribute for load/store instructions.
14257         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
14258         instructions.
14259         (mov<mode>_softfloat): Likewise.
14260         (mov<mode>_hardfloat32): Likewise.
14261         (mov<mode>_hardfloat64): Likewise.
14262         (mov<mode>_softfloat64): Likewise.
14263         (movdi_internal32): Likewise.
14264         (movdi_internal64): Likewise.
14265         (probe_stack_<mode>): Likewise.
14267 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
14269         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
14270         floating point, and decimal floating point to reload iterator.
14272         * config/rs6000/constraints.md (wl constraint): New constraints to
14273         return FLOAT_REGS if certain options are used to reduce the number
14274         of separate patterns that exist in the file.
14275         (wx constraint): Likewise.
14276         (wz constraint): Likewise.
14278         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14279         -mdebug=reg, print wg, wl, wx, and wz constraints.
14280         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14281         Initialize the reload functions for 64-bit binary/decimal floating
14282         point types.
14283         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14284         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14285         create the buffer on the stack to overcome not having a 32-bit
14286         load and store.
14287         (rs6000_emit_move): Likewise.
14288         (rs6000_secondary_memory_needed_rtx): Likewise.
14289         (rs6000_alloc_sdmode_stack_slot): Likewise.
14290         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14291         via xxlxor, just like DFmode 0.0.
14293         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
14294         define as 1 if we are running on a power7 or newer.
14295         (enum r6000_reg_class_enum): Add new constraints.
14297         * config/rs6000/dfp.md (movsd): Delete, combine with binary
14298         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14299         with other moves by using conditional constraits (wg).  Use LFIWZX
14300         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
14301         (movsd splitter): Likewise.
14302         (movsd_hardfloat): Likewise.
14303         (movsd_softfloat): Likewise.
14305         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
14306         binary and decimal floating point moves.
14307         (fmove_ok): New attributes to combine binary and decimal floating
14308         point moves, and to combine power6x (mfpgpr) moves along normal
14309         floating moves.
14310         (real_value_to_target): Likewise.
14311         (f32_lr): Likewise.
14312         (f32_lm): Likewise.
14313         (f32_li): Likewise.
14314         (f32_sr): Likewise.
14315         (f32_sm): Likewise.
14316         (f32_si): Likewise.
14317         (movsf): Combine binary and decimal floating point moves.  Combine
14318         power6x (mfpgpr) moves with other moves by using conditional
14319         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
14320         (mov<mode> for SFmode/SDmode); Likewise.
14321         (SFmode/SDmode splitters): Likewise.
14322         (movsf_hardfloat): Likewise.
14323         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
14324         (movsf_softfloat): Likewise.
14325         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
14327         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
14328         wx and wz constraints.
14330         * config/rs6000/constraints.md (wg constraint): New constraint to
14331         return FLOAT_REGS if -mmfpgpr (power6x) was used.
14333         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
14334         constraint.
14336         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14337         -mdebug=reg, print wg, wl, wx, and wz constraints.
14338         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14339         Initialize the reload functions for 64-bit binary/decimal floating
14340         point types.
14341         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14342         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14343         create the buffer on the stack to overcome not having a 32-bit
14344         load and store.
14345         (rs6000_emit_move): Likewise.
14346         (rs6000_secondary_memory_needed_rtx): Likewise.
14347         (rs6000_alloc_sdmode_stack_slot): Likewise.
14348         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14349         via xxlxor, just like DFmode 0.0.
14351         * config/rs6000/dfp.md (movdd): Delete, combine with binary
14352         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14353         with other moves by using conditional constraits (wg).  Use LFIWZX
14354         and STFIWX for loading SDmode on power7.
14355         (movdd splitters): Likewise.
14356         (movdd_hardfloat32): Likewise.
14357         (movdd_softfloat32): Likewise.
14358         (movdd_hardfloat64_mfpgpr): Likewise.
14359         (movdd_hardfloat64): Likewise.
14360         (movdd_softfloat64): Likewise.
14362         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
14363         64-bit binary and decimal floating point moves.
14364         (FMOVE64X): Likewise.
14365         (movdf): Combine 64-bit binary and decimal floating point moves.
14366         Combine power6x (mfpgpr) moves with other moves by using
14367         conditional constraits (wg).
14368         (mov<mode> for DFmode/DDmode): Likewise.
14369         (DFmode/DDmode splitters): Likewise.
14370         (movdf_hardfloat32): Likewise.
14371         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
14372         (movdf_softfloat32): Likewise.
14373         (movdf_hardfloat64_mfpgpr): Likewise.
14374         (movdf_hardfloat64): Likewise.
14375         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
14376         (movdf_softfloat64): Likewise.
14377         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
14378         (reload_<mode>_load): Move to later in the file so they aren't in
14379         the middle of the floating point move insns.
14380         (reload_<mode>_store): Likewise.
14382         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
14383         constraint.
14385         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
14386         constraint if -mdebug=reg.
14387         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
14388         Enable using dd reload support if needed.
14390         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
14391         binary and decimal floating point moves in rs6000.md.
14392         (movtd_internal): Likewise.
14394         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
14395         decimal floating point moves.
14396         (movtf): Likewise.
14397         (movtf_internal): Likewise.
14398         (mov<mode>_internal, TDmode/TFmode): Likewise.
14399         (movtf_softfloat): Likewise.
14400         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
14402         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
14403         movdi_internal64, using wg constraint for move direct operations.
14404         (movdi_internal64): Likewise.
14406         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
14407         MODES_TIEABLE_P for selected modes.  Print the numerical value of
14408         the various virtual registers. Use GPR/FPR first/last values,
14409         instead of hard coding the register numbers.  Print which modes
14410         have reload functions registered.
14411         (rs6000_option_override_internal): If -mdebug=reg, trace the options
14412         settings before/after setting cpu, target and subtarget settings.
14413         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
14414         and for secondary reload failures in rs6000_secondary_reload_inner.
14415         (rs6000_secondary_reload_fail): Likewise.
14416         (rs6000_secondary_reload_inner): Likewise.
14418         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
14419         macros for first/last GPR and FPR registers.
14420         (LAST_GPR_REGNO): Likewise.
14421         (FIRST_FPR_REGNO): Likewise.
14422         (LAST_FPR_REGNO): Likewise.
14424         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
14425         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
14426         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
14427         (vcond<mode><mode>): Likewise.
14428         (vcondu<mode><mode>): Likewise.
14429         (vector_gtu<mode>): Likewise.
14430         (vector_gte<mode>): Likewise.
14431         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
14432         to prevent the compiler from converting DImode operations to TImode.
14433         (ior<mode>3): Likewise.
14434         (and<mode>3): Likewise.
14435         (one_cmpl<mode>2): Likewise.
14436         (nor<mode>3): Likewise.
14437         (andc<mode>3): Likewise.
14439         * config/rs6000/constraints.md (wt constraint): New constraint
14440         that returns VSX_REGS if TImode is allowed in VSX registers.
14442         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
14443         constant under VSX.
14445         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
14446         similar to TImode, but it is restricted to being in the GPRs.
14448         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
14449         TImode to occupy a single VSX register.
14451         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
14452         -mvsx-timode for power7/power8.
14453         (power7 cpu): Likewise.
14454         (power8 cpu): Likewise.
14456         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
14457         sure that TFmode/TDmode take up two registers if they are ever
14458         allowed in the upper VSX registers.
14459         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
14460         registers.
14461         (rs6000_init_hard_regno_mode_ok): Likewise.
14462         (rs6000_debug_reg_global): Add debugging for PTImode and wt
14463         constraint.  Print if LRA is turned on.
14464         (rs6000_option_override_internal): Give an error if -mvsx-timode
14465         and VSX is not enabled.
14466         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
14467         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
14468         to reg+offset addressing.  Use PTImode when checking offset
14469         addresses for validity.
14470         (reg_offset_addressing_ok_p): Likewise.
14471         (rs6000_legitimate_offset_address_p): Likewise.
14472         (rs6000_legitimize_address): Likewise.
14473         (rs6000_legitimize_reload_address): Likewise.
14474         (rs6000_legitimate_address_p): Likewise.
14475         (rs6000_eliminate_indexed_memrefs): Likewise.
14476         (rs6000_emit_move): Likewise.
14477         (rs6000_secondary_reload): Likewise.
14478         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
14479         reloads to fpr registers to continue to use reg+offset addressing,
14480         but 64-bit reloads to altivec registers need reg+reg addressing.
14481         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
14482         it.  Treat LO_SUM like a PLUS operation.
14483         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
14484         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
14485         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
14486         registers to share a register with a smaller sized type, since VSX
14487         puts scalars in the upper 64-bits.
14488         (print_operand): Add support for PTImode.
14489         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
14490         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
14491         registers, but don't have arithmetic support.
14492         (rs6000_memory_move_cost): Add test for VSX.
14493         (rs6000_opt_masks): Add -mvsx-timode.
14495         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
14496         for TImode.
14497         (VSs): Likewise.
14498         (VSr): Use wt constraint for TImode.
14499         (VSv): Drop TImode support.
14500         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
14501         (vsx_movti_64bit): Likewise.
14502         (vsx_movti_32bit): Likewise.
14503         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
14504         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
14505         one '?' on the appropriate output constraint.  Do not allow TImode
14506         logical operations on 32-bit systems.
14507         (vsx_ior<mode>3): Likewise.
14508         (vsx_xor<mode>3): Likewise.
14509         (vsx_one_cmpl<mode>2): Likewise.
14510         (vsx_nor<mode>3): Likewise.
14511         (vsx_andc<mode>3): Likewise.
14512         (vsx_concat_<mode>): Likewise.
14513         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
14515         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
14516         OPTION_MASK_VSX_TIMODE.
14517         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
14518         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
14520         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
14521         (TI2 iterator): New iterator for TImode, PTImode.
14522         (wd mode attribute): Add values for vector types.
14523         (movti_string): Replace TI move operations with operations for TImode
14524         and PTImode.  Add support for TImode being allowed in VSX registers.
14525         (mov<mode>_string, TImode/PTImode): Likewise.
14526         (movti_ppc64): Likewise.
14527         (mov<mode>_ppc64, TImode/PTImode): Likewise.
14528         (TI mode splitters): Likewise.
14530         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
14531         constraint.
14533 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
14535         PR tree-optimization/56355
14536         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
14537         Also handle integers with undefined overflow.
14539 2013-03-20  Catherine Moore  <clm@codesourcery.com>
14540             Maciej W. Rozycki  <macro@codesourcery.com>
14541             Tom de Vries  <tom@codesourcery.com>
14542             Nathan Sidwell <nathan@codesourcery.com>
14543             Iain Sandoe  <iain@codesourcery.com>
14544             Nathan Froyd  <froydnj@codesourcery.com>
14545             Chao-ying Fu <fu@mips.com>
14547         * doc/extend.texi: (micromips, nomicromips, nocompression):
14548         Document new function attributes.
14549         * doc/invoke.texi (minterlink-compressed, mmicromips,
14550         m14k, m14ke, m14kec): Document new options.
14551         (minterlink-mips16): Update documentation.
14552         * doc/md.texi (ZC, ZD): Document new constraints.
14553         * configure.ac (gcc_cv_as_micromips): Check if linker
14554         supports the .set micromips directive.
14555         * configure: Regenerate.
14556         * config.in: Regenerate.
14557         * config/mips/mips-tables.opt: Regenerate.
14558         * config/mips/micromips.md: New file.
14559         * constraints.md (ZC, ZD): New constraints.
14560         * config/mips/predicates.md (movep_src_register): New predicate.
14561         (movep_src_operand): New predicate.
14562         (non_volatile_mem_operand): New predicate.
14563         * config/mips/mips.md (multimem): New type.
14564         (length): Differentiate between 17-bit and 18-bit branch offsets.
14565         (MOVEP1, MOVEP2): New mode iterator.
14566         (mov_<load>l): Use ZC constraint.
14567         (mov_<load>r): Likewise.
14568         (mov_<store>l): Likewise.
14569         (mov_<store>r): Likewise.
14570         (*branch_equality<mode>_inverted): Add microMIPS support.
14571         (*branch_equality<mode>): Likewise.
14572         (*jump_absolute): Likewise.
14573         (indirect_jump_<mode>): Likewise.
14574         (tablejump_<mode>): Likewise.
14575         (<optab>_internal): Likewise.
14576         (sibcall_internal): Likewise.
14577         (sibcall_value_internal): Likewise.
14578         (prefetch): Use constraint ZD.
14579         * config/mips/mips.opt (minterlink-compressed): New option.
14580         (minterlink-mips16): Now an alias for minterlink-compressed.
14581         (mmicromips): New option.
14582         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14583         (compare_and_swap_12): Likewise.
14584         (sync_add<mode>): Likewise.
14585         (sync_<optab>_12): Likewise.
14586         (sync_old_<optab>_12): Likewise.
14587         (sync_new_<optab>_12): Likewise.
14588         (sync_nand_12): Likewise.
14589         (sync_old_nand_12): Likewise.
14590         (sync_new_nand_12): Likewise.
14591         (sync_sub<mode>): Likewise.
14592         (sync_old_add<mode>): Likewise.
14593         (sync_old_sub<mode>): Likewise.
14594         (sync_new_add<mode>): Likewise.
14595         (sync_new_sub<mode>): Likewise.
14596         (sync_<optab><mode>): Likewise.
14597         (sync_old_<optab><mode>): Likewise.
14598         (sync_new_<optab><mode>): Likewise.
14599         (sync_nand<mode>): Likewise.
14600         (sync_old_nand<mode>): Likewise.
14601         (sync_new_nand<mode>): Likewise.
14602         (sync_lock_test_and_set<mode>): Likewise.
14603         (test_and_set_12): Likewise.
14604         (atomic_compare_and_swap<mode>): Likewise.
14605         (atomic_exchange<mode>_llsc): Likewise.
14606         (atomic_fetch_add<mode>_llsc): Likewise.
14607         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14608         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14609         (umips_save_restore_pattern_p): Likewise.
14610         (umips_load_store_pair_p): Likewise.
14611         (umips_output_load_store_pair): Likewise.
14612         (umips_movep_target_p): Likewise.
14613         (umips_12bit_offset_address_p): Likewise.
14614         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14615         (mips_base_mips16): Rename this...
14616         (mips_base_compression_flags): ...to this. Update all uses.
14617         (mips_attribute_table): Add micromips, nomicromips and nocompression.
14618         (mips_mips16_decl_p): Delete.
14619         (mips_nomips16_decl_p): Delete.
14620         (mips_get_compress_on_flags): New function.
14621         (mips_get_compress_off_flags): New function.
14622         (mips_get_compress_mode): New function.
14623         (mips_get_compress_on_name): New function.
14624         (mips_get_compress_off_name): New function.
14625         (mips_insert_attributes): Support multiple compression types.
14626         (mips_merge_decl_attributes): Likewise.
14627         (umips_12bit_offset_address_p): New function.
14628         (mips_start_function_definition): Emit .set micromips directive.
14629         (mips_call_may_need_jalx_p): New function.
14630         (mips_function_ok_for_sibcall): Add microMIPS support.
14631         (mips_print_operand_punctuation): Support short delay slots and
14632         compact jumps.
14633         (umips_swm_mask, umips_swm_encoding): New.
14634         (umips_build_save_restore): New function.
14635         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14636         (was_mips16_p): Remove.
14637         (old_compression_mode): New.
14638         (mips_set_compression_mode): New function.
14639         (mips_set_current_function): Add microMIPS support.
14640         (mips_option_override): Likewise.
14641         (umips_save_restore_pattern_p): New function.
14642         (umips_output_save_restore): New function.
14643         (umips_load_store_pair_p_1): New function.
14644         (umips_load_store_pair_p): New function.
14645         (umips_output_load_store_pair_1): New function.
14646         (umips_output_load_store_pair): New function.
14647         (umips_movep_target_p) New function.
14648         (mips_prepare_pch_save): Add microMIPS support.
14649         * config/mips/mips.h (TARGET_COMPRESSION): New.
14650         (TARGET_CPU_CPP_BUILTINS): Update macro
14651         to use new compression flags and to support microMIPS.
14652         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14653         (MIPS_ARCH_FLOAT_SPEC): Likewise.
14654         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14655         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14656         (ASM_SPEC): Support mmicromips and mno-micromips.
14657         (M16STORE_REG_P): New macro.
14658         (MIPS_CALL): Support TARGET_MICROMIPS.
14659         (MICROMIPS_J): New macro.
14660         (mips_base_mips16): Rename this...
14661         (mips_base_compression_flags): ...to this.
14662         (UMIPS_12BIT_OFFSET_P): New macro.
14663         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14664         (MULTILIB_DIRNAMES): Likewise.
14665 2013-03-20  Richard Biener  <rguenther@suse.de>
14667         PR tree-optimization/56661
14668         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
14669         the result does not have to be distinct.
14671 2013-03-20  Richard Biener  <rguenther@suse.de>
14673         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
14674         remap_gimple_op_r.
14676 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14677             Steven Bosscher <steven@gcc.gnu.org>
14679         PR rtl-optimization/56605
14680         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
14682 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14684         PR bootstrap/56656
14685         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
14686         that require movd instead of movq.
14688 2013-03-20  Richard Biener  <rguenther@suse.de>
14690         * tree-ssa-structalias.c (struct variable_info): Add pointer
14691         to the first field of an aggregate with sub-vars.  Make
14692         this and the pointer to the next subfield its ID.
14693         (vi_next): New function.
14694         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
14695         storedanything_id, integer_id): Increment by one.
14696         (new_var_info, get_call_vi, lookup_call_clobber_vi,
14697         get_call_clobber_vi): Adjust.
14698         (solution_set_expand): Simplify and speedup.
14699         (solution_set_add): Inline into ...
14700         (set_union_with_increment): ... this.  Adjust accordingly.
14701         (do_sd_constraint): Likewise.
14702         (do_ds_constraint): Likewise.
14703         (do_complex_constraint): Simplify.
14704         (build_pred_graph): Adjust.
14705         (solve_graph): Likewise.  Simplify and speedup.
14706         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
14707         get_constraint_for_component_ref, get_constraint_for_1,
14708         first_vi_for_offset, first_or_preceding_vi_for_offset,
14709         create_function_info_for, create_variable_info_for_1,
14710         create_variable_info_for, intra_create_variable_infos): Adjust.
14711         (init_base_vars): Push NULL for ID zero.
14712         (compute_points_to_sets): Adjust.
14714 2013-03-20  Richard Biener  <rguenther@suse.de>
14716         * cfgloop.c (verify_loop_structure): Streamline and avoid
14717         ICEing on corrupt loop tree.
14718         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
14719         loop tree.
14721 2013-03-20  Richard Biener  <rguenther@suse.de>
14723         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
14724         check whether an SSA update is needed.
14726 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
14728         * config/mips/constraints.md (T): Rename to...
14729         (Yf): ...this.
14730         (U): Rename to...
14731         (Yd): ...this.
14732         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
14733         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
14735 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14737         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
14738         (*subsi3_carryin_uxtw): Likewise.
14740 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14742         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
14743         (*rorsi3_insn_uxtw): Likewise.
14745 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
14747         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
14748         (*extrsi5_insn_uxtw): Likewise.
14750 2013-03-19  Richard Biener  <rguenther@suse.de>
14752         PR tree-optimization/56273
14753         * passes.c (init_optimization_passes): Move second VRP after DOM.
14755 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
14757         * config/i386/i386.md (*movti_internal): Merge from
14758         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
14759         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
14760         nox64 isa attributes.
14762 2013-03-18  Richard Biener  <rguenther@suse.de>
14764         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
14765         (unite): Likewise.
14766         (merge_node_constraints): Likewise.
14767         (build_succ_graph): Likewise.
14768         (valid_graph_edge): Inline into single caller.
14769         (unify_nodes): Likewise.  Use bitmap_set_bit return value
14770         and cache varinfo.
14771         (scc_visit): Fix formatting and variable use.
14772         (do_sd_constraint): Use gcc_checking_assert.
14773         (do_ds_constraint): Likewise.
14774         (do_complex_constraint): Likewise.
14775         (condense_visit): Likewise.  Cleanup.
14776         (dump_pred_graph): New function.
14777         (perform_var_substitution): Dump the pred-graph before
14778         variable substitution.
14779         (find_equivalent_node): Use gcc_checking_assert.
14780         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
14782 2013-03-18  Richard Biener  <rguenther@suse.de>
14784         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
14785         Remove cond_expr_stmt_list argument and do not gimplify the
14786         built expression.
14787         (vect_loop_versioning): Adjust.
14788         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
14789         Cleanup to use less temporaries.
14790         (vect_create_data_ref_ptr): Cleanup.
14792 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14794         PR tree-optimization/56635
14795         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
14796         require types_compatible_p types.
14798 2013-03-18  Nick Clifton  <nickc@redhat.com>
14800         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
14801         spurious backslash.
14803         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
14804         Add missing line to comment describing function.
14806 2013-03-18  Richard Biener  <rguenther@suse.de>
14808         PR tree-optimization/56210
14809         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14810         Handle string / character search functions.
14811         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
14813 2013-03-18  Richard Biener  <rguenther@suse.de>
14815         PR middle-end/56483
14816         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
14817         and implement properly.
14818         * gimple.h (gimple_cond_single_var_p): Remove.
14820 2013-03-18  Richard Biener  <rguenther@suse.de>
14822         * tree-data-ref.h (find_data_references_in_loop): Declare.
14823         * tree-data-ref.c (get_references_in_stmt): Use a stack
14824         vector pre-allocated in the callers.
14825         (find_data_references_in_stmt): Adjust.
14826         (graphite_find_data_references_in_stmt): Likewise.
14827         (create_rdg_vertices): Likewise.
14828         (find_data_references_in_loop): Export.
14829         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14830         Compute dependences here...
14831         (vect_analyze_data_refs): ...not here.  When we encounter
14832         a non-vectorizable data reference in basic-block vectorization
14833         truncate the data reference vector.  Do not bother to
14834         fixup data-dependence information for gather loads.
14835         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
14836         of data references, as reported.
14838 2013-03-18  Richard Biener  <rguenther@suse.de>
14840         PR tree-optimization/3713
14841         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
14842         has_constants and expr.
14843         (stmt_has_constants): Properly valueize SSA names when deciding
14844         whether the stmt has constants.
14846 2013-03-18  Richard Biener  <rguenther@suse.de>
14848         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
14849         whole function when there is nothing to do.
14850         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
14851         * tree-vectorizer.c (vectorize_loops): Update virtual and
14852         loop-closed SSA once.
14853         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
14855 2013-03-18  Richard Biener  <rguenther@suse.de>
14857         PR middle-end/56113
14858         * domwalk.c (bb_postorder): New global static.
14859         (cmp_bb_postorder): New function.
14860         (walk_dominator_tree): Replace scheme imposing an order for
14861         visiting dominator sons by one sorting them at the time they
14862         are pushed on the stack.
14864 2013-03-18  Richard Biener  <rguenther@suse.de>
14866         PR tree-optimization/39326
14867         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
14868         (struct mem_ref): Replace mem member with ao_ref typed member.
14869         (MEM_ANALYZABLE): Adjust.
14870         (memref_eq): Likewise.
14871         (mem_ref_alloc): Likewise.
14872         (gather_mem_refs_stmt): Likewise.
14873         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
14874         (execute_sm_if_changed_flag_set): Adjust.
14875         (execute_sm): Likewise.
14876         (ref_always_accessed_p): Likewise.
14877         (refs_independent_p): Likewise.
14878         (can_sm_ref_p): Likewise.
14880 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
14882         PR c/56566
14883         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
14884         return 1 even for !unsignedp.
14886 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14888         * config/i386/i386.md (isa): Add x64 and nox64.
14889         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
14890         (*pushtf): Enable *roF alternative for x64 isa only.
14891         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
14892         mode attribute of integer alternatives to DImode for TARGET_64BIT.
14893         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
14894         (*movtf_internal): Merge from *movtf_internal_rex64 and
14895         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
14896         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
14897         nox64 isa attributes.
14898         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
14899         nox64 isa attributes.
14900         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
14902 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
14904         * config/alpha/alpha.c (TARGET_LRA_P): New define.
14906 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
14908         PR target/56640
14909         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
14910         class names.  Remove trailing comma after "ALL_REGS".
14912 2013-03-16  Jan Hubicka  <jh@suse.cz>
14914         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
14915         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
14916         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
14917         of cgraph_get_create_node.
14918         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14920 2013-03-16  Jason Merrill  <jason@redhat.com>
14922         PR debug/49090
14923         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
14924         with DW_AT_default_value.
14926 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
14928         * BASE-VER: Set to 4.9.0.
14930 2013-03-14  Andi Kleen  <ak@linux.intel.com>
14932         PR target/56619
14933         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
14934         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
14935         Document _x* TSX intrinsics.
14937 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
14938             David Holsgrove  <david.holsgrove@xilinx.com>
14940         * configure.ac: Add MicroBlaze TLS support detection.
14941         * configure: Regenerate.
14942         * config/microblaze/microblaze-protos.h
14943         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
14944         symbol_mentioned_p, label_mentioned_p): Add prototypes.
14945         * config/microblaze/microblaze.c (microblaze_address_type): Add
14946         ADDRESS_TLS and tls_reloc address types.
14947         (microblaze_address_info): Add tls_reloc.
14948         (TARGET_HAVE_TLS): Define.
14949         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
14950          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
14951          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
14952         load_tls_operand,  microblaze_call_tls_get_addr,
14953         microblaze_legitimize_tls_address): New functions.
14954         (microblaze_classify_unspec): Handle UNSPEC_TLS.
14955         (get_base_reg): Use microblaze_tls_symbol_p.
14956         (microblaze_classify_address): Handle TLS.
14957         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
14958         label_mentioned_p and microblaze_tls_referenced_p.
14959         (microblaze_legitimize_address): Handle TLS.
14960         (microblaze_address_insns): Handle ADDRESS_TLS.
14961         (pic_address_needs_scratch): Handle TLS.
14962         (print_operand_address): Handle TLS.
14963         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
14964         (microblaze_expand_move): Handle TLS.
14965         (microblaze_legitimate_constant_p): Check
14966         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
14967         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14968         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
14969         (PIC_OFFSET_TABLE_REGNUM): Set.
14970         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
14971         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
14972         (addsi3, movsi_internal2, movdf_internal): Update constraints
14973         * config/microblaze/predicates.md (arith_plus_operand): Define
14974         (move_operand): Redefine as move_src_operand,
14975         check microblaze_tls_referenced_p.
14977 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14979         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
14980         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14982 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
14984         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14985         CC mode for AND.
14987 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
14989         PR tree-optimization/53265
14990         * common.opt (Waggressive-loop-optimizations): New option.
14991         * tree-ssa-loop-niter.c: Include tree-pass.h.
14992         (do_warn_aggressive_loop_optimizations): New function.
14993         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
14994         if number_of_latch_executions returned constant.
14995         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14996         early.  If number_of_latch_executions returned constant, set
14997         nb_iterations_upper_bound back to it.
14998         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14999         field.
15000         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
15001         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
15003         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
15004         (MULTILIB_OSDIRNAMES): Set.
15005         * genmultilib: If defaultosdirname doesn't start with :: , set
15006         defaultosdirname2 instead, clear it and emit two . multilib_raw
15007         entries instead of just one.
15009 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
15011         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15012         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
15013         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15014         (SUBTARGET_OVERRIDE_OPTIONS): New.
15016 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
15018         PR target/49880
15019         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
15020         (musermode): Convert to Var(TARGET_USERMODE).
15021         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
15022         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
15023         * config/sh/sh.c (sh_option_override): Use
15024         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
15025         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
15026         condition.
15027         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
15028         TARGET_SH4.
15029         (udivsi3_i4_single, divsi3_i4_single): Use
15030         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
15032 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
15034         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
15035         default setting.
15037 2013-03-13  Richard Biener  <rguenther@suse.de>
15039         PR tree-optimization/56608
15040         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
15041         calls when vectorizing basic-blocks.
15043 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
15045         PR plugins/45078
15046         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
15047         tm_file.
15049 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
15051         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15053 2013-03-11  Jan Hubicka  <jh@suse.cz>
15055         PR lto/56557
15056         * lto-streamer-out.c (output_symbol_p): Skip references from
15057         constructors of external variables.
15059 2013-03-11  Jan Hubicka  <jh@suse.cz>
15061         PR middle-end/56571
15062         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15063         from pseudos.
15064         * emit-rtl.c (verify_rtx_sharing): Likewise.
15065         (copy_insn_1): Likewise.
15066         * rtl.c (copy_rtx): Likewise.
15068 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
15070         PR target/56591
15071         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15072         output_operand_lossage message.
15074 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
15076         PR target/56470
15077         * arm.c (shift_op): Validate RTL pattern on the fly.
15078         (arm_print_operand, case 'S'): Don't use shift_operator to validate
15079         the RTL.
15081 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15083         PR target/56347
15084         * config/pa/pa.md (call_value): Check for calls to powf and direct to
15085         new call patterns that clobber %fr12.
15086         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15087         split and postreload patterns.
15088         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15089         registers %fr12 and %fr12R as call used.
15091 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
15093         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15094         (canon_address, record_store, replace_read, check_mem_read_rtx,
15095         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15096         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15097         rest_of_handle_dse): Likewise.
15099 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
15101         PR middle-end/56524
15102         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15103         Add base_optabs.
15104         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15105         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15106         (save_optabs_if_changed): Replace with...
15107         (init_tree_optimization_optabs): ...this.
15108         * optabs.c (save_optabs_if_changed): Rename to...
15109         (init_tree_optimization_optabs): ...this.  Take the optimization node
15110         as argument.  Do nothing if the base optabs are already correct.
15111         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15112         to recompute optabs.
15113         * function.h (function): Remove optabs field.
15114         * function.c (invoke_set_current_function_hook): Call
15115         init_tree_optimization_optabs.  Use the result to initialize
15116         this_fn_optabs.
15118 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
15120         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15121         if GTMA_HAS_NO_INSTRUMENTATION.
15122         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15123         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15124         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15125         * gimple-pretty-print.c (dump_gimple_transaction): Handle
15126         GTMA_HAS_NO_INSTRUMENTATION.
15128 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
15130         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15131         libasan_preinit.o.
15133 2013-03-08  Marek Polacek  <polacek@redhat.com>
15134             Jakub Jelinek  <jakub@redhat.com>
15136         PR tree-optimization/56478
15137         * predict.c (is_comparison_with_loop_invariant_p): Change the
15138         type of loop_step to tree.
15139         (predict_loops): Adjust.
15140         (predict_iv_comparison): Perform the computations on double_ints.
15142 2013-03-08  Richard Biener  <rguenther@suse.de>
15144         PR tree-optimization/56570
15145         * tree-cfg.c (verify_expr_location_1): Verify locations for
15146         DECL_DEBUG_EXPR.
15147         * tree-sra.c (create_access_replacement): Strip locations
15148         from DECL_DEBUG_EXPRs.
15150 2013-03-08  Richard Biener  <rguenther@suse.de>
15152         * tree-inline.c (expand_call_inline): Do not associate
15153         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15154         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15156 2013-03-08  Richard Biener  <rguenther@suse.de>
15158         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15159         or block changes with -Og.  Fix for location / block encoding
15160         changes and PHI arguments with locations.
15162 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
15164         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15165         for all counters.
15166         (struct output_info): Likewise.
15167         (register_overhead): Remove bad gcc_assert.
15168         (bitmap_find_bit): If there is only a single bitmap element, do not
15169         count a miss as a search.
15170         (print_statistics): Update for counter type changes.
15171         (dump_bitmap_statistics): Likewise.  Print headers such that they
15172         are properly lined up with the printed counters.
15174 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
15176         PR tree-optimization/56559
15177         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15178         check that it has only a single use.
15180 2013-03-07  Richard Biener  <rguenther@suse.de>
15182         * doc/invoke.texi (fwhole-program): Discourage use in combination
15183         with -flto.
15185 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
15187         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15189         PR tree-optimization/56539
15190         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15191         instead of GSI_CONTINUE_LINKING as last argument to
15192         force_gimple_operand_gsi.  Adjust function comment.
15194         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15195         aarch64-cores.def.
15197         PR middle-end/56548
15198         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15199         promoted mode, convert the result back to the original mode.
15201 2013-03-06  Richard Biener  <rguenther@suse.de>
15203         PR middle-end/56294
15204         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15205         (insert_updated_phi_nodes_compare_uids): New function.
15206         (update_ssa): Sort symbols_to_rename after UID before
15207         traversing it to insert PHI nodes.
15209 2013-03-06  Richard Biener  <rguenther@suse.de>
15211         PR middle-end/50494
15212         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15213         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15215         Revert
15216         2013-02-13  Richard Biener  <rguenther@suse.de>
15218         PR lto/50494
15219         * varasm.c (output_constant_def_1): Get the decl representing
15220         the constant as argument.
15221         (output_constant_def): Wrap output_constant_def_1.
15222         (make_decl_rtl): Use output_constant_def_1 with the decl
15223         representing the constant.
15224         (build_constant_desc): Optionally re-use a decl already
15225         representing the constant.
15226         (tree_output_constant_def): Adjust.
15228 2013-03-06  Joey Ye  <joey.ye@arm.com>
15230         PR lto/50293
15231         * gcc.c (convert_white_space): New function.
15232         (main): Handles white space in function name.
15234 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
15236         PR target/56529
15237         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15238         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
15239         to SH_DIV_CALL_TABLE for TARGET_SH2.
15240         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15241         list.
15242         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15243         call-table options.
15245 2013-03-05  Sterling Augustine  <saugustine@google.com>
15246             Cary Coutant  <ccoutant@google.com>
15248         PR debug/55364
15249         * dwarf2out.c (resolve_addr): Don't call
15250         remove_loc_list_addr_table_entries a second time for the same
15251         expression.
15253 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15255         PR debug/56510
15256         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
15257         (avoid_complex_debug_insns): New function.
15258         (expand_debug_locations): Call it.
15260         PR rtl-optimization/56484
15261         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
15262         lifetimes of hard registers on small register class machines.
15264 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15266         * config/microblaze/microblaze-protos.h: Rename
15267         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
15268         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
15269         fast_interrupt.
15270         (microblaze_fast_interrupt_function_p): New function.
15271         (microblaze_is_interrupt_handler): Rename to
15272         microblaze_is_interrupt_variant and add fast_interrupt check.
15273         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
15274         (save_restore_insns): Likewise.
15275         (compute_frame_size): Likewise.
15276         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
15277         (microblaze_globalize_label): Likewise.
15278         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
15279         * config/microblaze/microblaze.md: Use wrapper
15280         microblaze_is_interrupt_variant.
15282 2013-03-05  Kai Tietz  <ktietz@redhat.com>
15284         * sdbout.c (sdbout_one_type): Switch to current function's section
15285         supporting cold/hot.
15287 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15289         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
15290         -mxl-reorder.
15292 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15294         PR middle-end/56461
15295         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
15296         if VALGRIND_GET_VBITS is defined, temporarily make object
15297         memory all defined, and restore previous valgrind addressability
15298         and definability afterwards.  Free this_object at the end.
15300         PR middle-end/56461
15301         * lra.c (lra): Call lra_clear_live_ranges if live_p,
15302         right before calling lra_create_live_ranges, also call it
15303         when clearing live_p.  Only call lra_clear_live_ranges
15304         at the end if live_p.
15306         PR middle-end/56461
15307         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
15309 2013-03-05  Richard Biener  <rguenther@suse.de>
15311         PR tree-optimization/56521
15312         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
15313         value-id.
15315 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
15317         PR c++/55135
15318         * except.h (remove_unreachable_eh_regions): New prototype.
15319         * except.c (remove_eh_handler_splicer): New function, split out
15320         of remove_eh_handler.
15321         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
15322         warning about running it on many EH regions one at a time.
15323         (remove_unreachable_eh_regions_worker): New function, walk the
15324         EH tree in depth-first order and remove non-marked regions.
15325         (remove_unreachable_eh_regions): New function.
15326         * tree-eh.c (mark_reachable_handlers): New function, split out
15327         from remove_unreachable_handlers.
15328         (remove_unreachable_handlers): Use mark_reachable_handlers and
15329         remove_unreachable_eh_regions.
15330         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
15331         and remove_unreachable_eh_regions.
15333 2013-03-05  Richard Biener  <rguenther@suse.de>
15335         PR middle-end/56525
15336         * loop-init.c (fix_loop_structure): Remove loops in two stages,
15337         not freeing them until the end.
15339 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15341         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
15343 2013-03-05  Richard Biener  <rguenther@suse.de>
15345         PR tree-optimization/56270
15346         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
15347         of loads after scheduling an SLP instance.
15349 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15351         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
15352         tic6x.exp.
15353         (check_gcc_parallelize): Run guality.exp as a separate job from
15354         vect.exp with unsorted.exp and $(dg_target_exps) separately from
15355         struct-layout-1.exp with stackalign.exp.
15357         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
15359         PR middle-end/56461
15360         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
15361         load_index sbitmap even if some bit in it isn't set.
15363         PR middle-end/56461
15364         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
15365         (discover_iteration_bound_by_body_walk): Change queues to
15366         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
15367         spelling in comment.  Call safe_push on queues[bound_index] directly.
15368         Release queues[queue_index] in every iteration unconditionally.
15369         Release bounds vector.
15371         PR middle-end/56461
15372         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
15373         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
15374         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
15375         inner_phis vector.
15377 2013-03-05  Richard Biener  <rguenther@suse.de>
15379         PR lto/56515
15380         * tree-inline.c (remap_blocks_to_null): New function.
15381         (expand_call_inline): When expanding a call stmt without
15382         an associated block inline remap all callee blocks to NULL.
15384 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15386         PR rtl-optimization/56494
15387         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
15388         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
15389         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
15391         PR middle-end/56461
15392         * sel-sched-ir.c (free_sched_pools): Release
15393         succs_info_pool.stack[succs_info_pool.max_top] vectors too
15394         if succs_info_pool.max_top isn't -1.
15396         PR bootstrap/56509
15397         * opts.c (opts_obstack, opts_concat): Moved to...
15398         * opts-common.c (opts_obstack, opts_concat): ... here.
15400 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
15402         PR middle-end/56461
15403         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
15405 2013-03-04  Martin Jambor  <mjambor@suse.cz>
15407         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
15408         all appropriate places.
15410 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
15412         PR tree-optimization/56424
15413         * ipa-split.c (split_function): Do not set the RSO flag if result is
15414         not by reference and its type is a register type.
15416 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
15418         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
15419         (microblaze_legitimate_pic_operand): Likewise
15420         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
15421         new function microblaze_legitimate_pic_operand
15422         * config/microblaze/microblaze-protos.h
15423         (microblaze_legitimate_pic_operand): Declare.
15425 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15427         * config/microblaze/predicates.md (call_insn_simple_operand):
15428         New predicate for supported rtx code types.
15429         * config/microblaze/microblaze.md (call_internal1): Use
15430         call_insn_simple_operand predicate.
15432 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
15434         PR middle-end/56461
15435         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
15436         partitions.ordered_remove.
15438         PR middle-end/56461
15439         * tree-vect-stmts.c (vectorizable_conversion): Don't call
15440         vec_oprnds0.create (1) for modifier == NONE.
15442         PR middle-end/56461
15443         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
15444         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
15445         vec_oprnds1 right before pushing anything to it for
15446         scalar_shift_arg.
15448         PR middle-end/56461
15449         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
15450         set nbbs to 0 instead of having separate code path.
15451         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
15452         instead of false as last argument if returning NULL.
15454 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
15456         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
15457         the attribute is now called "target" instead of "option".
15458         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
15459         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
15460         attribute/pragma name for TARGET_OPTION_VALID_P and
15461         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
15462         * doc/tm.texi: Regenerated.
15464 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
15466         * config/microblaze/microblaze.c:
15467         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
15468         * config/microblaze/microblaze.h: Add -mxl-reorder to
15469         DRIVER_SELF_SPECS.
15470         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
15471         instructions emitted if TARGET_REORDER.
15472         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
15473         or 0 for -m/-mno case, but initialises as 2 to detect default use case
15474         separately.
15476 2013-03-01  Xinliang David Li  <davidxl@google.com>
15478         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
15479         walk length.
15481 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
15483         PR middle-end/56461
15484         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
15485         vector even when returning true.  Fix up function comment formatting.
15487         PR middle-end/56461
15488         * ira-build.c (ira_loop_nodes_count): New variable.
15489         (create_loop_tree_nodes): Initialize it.
15490         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
15492         PR middle-end/56461
15493         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
15494         method on dr_chain and result_chain.
15495         * tree-vect-stmts.c (vectorizable_store): Only call
15496         result_chain.create if j == 0.
15498         PR middle-end/56461
15499         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
15500         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
15501         before overwriting it.
15503 2013-03-01  Tobias Burnus  <burnus@net-b.de>
15505         * doc/extended.texi (C Extensions): Change order in @menu
15506         to match @node.
15507         (Other MIPS Built-in Functions): Move last MIPS entry before
15508         "picoChip Built-in Functions".
15509         (SH Built-in Functions): Move after RX Built-in Functions.
15510         * doc/gcc.texi (Introduction): Change order in @menu
15511         to match @node.
15512         * doc/md.texi (Constraints): Ditto.
15513         * gty.texi (Type Information): Ditto.
15514         (User-provided marking routines for template types): Make
15515         subsection.
15516         * doc/invoke.texi (AArch64 Options): Move before
15517         "Adapteva Epiphany Options".
15519 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
15520             Jakub Jelinek  <jakub@redhat.com>
15522         PR sanitizer/56454
15523         * asan.c (gate_asan): Lookup no_sanitize_address instead of
15524         no_address_safety_analysis attribute.
15525         * doc/extend.texi (no_address_safety_attribute): Rename to
15526         no_sanitize_address attribute, mention no_address_safety_analysis
15527         attribute as deprecated alias.
15529 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
15531         PR middle-end/56461
15532         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
15533         type to vec<vec<tree> > *.
15534         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
15535         to be vec<tree> instead of vec<tree> *, set vec_defs
15536         to vNULL and call vec_defs.create (number_of_vects), adjust other
15537         uses of vec_defs.
15538         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
15539         vectorizable_condition): Adjust vect_get_slp_defs callers.
15541 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15543         * config/aarch64/aarch64.c
15544         (aarch64_float_const_representable): Remove unused variable.
15546 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15548         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
15550 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15552         * config/aarch64/aarch64-builtins.c
15553         (aarch64_init_simd_builtins): Make static.
15555 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15557         * config/aarch64/aarch64.c
15558         (aarch64_simd_make_constant): Make static.
15560 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15562         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
15563         with no initialization to the RHS of debug statements.
15565 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15567         PR tree-optimization/56294
15568         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
15569         Adjust dumping.
15570         (get_access_replacement): Do not call create_access_replacement.
15571         Assert a replacement exists.
15572         (get_repl_default_def_ssa_name): Create the replacement declaration
15573         itself.
15575 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15577         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
15578         final_end_function.
15580 2013-02-28  Marek Polacek  <polacek@redhat.com>
15582         PR rtl-optimization/56466
15583         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
15584         if we're changing a loop.
15585         (peel_loops_completely): Likewise.
15587 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
15589         PR c++/55813
15590         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
15592 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15594         PR target/56445
15595         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
15596         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
15597         INTX_FTYPE_FX, FX_FTYPE_INTX.
15598         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
15600 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15602         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
15603         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
15604         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
15605         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
15606         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
15607         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
15608         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
15609         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
15610         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
15611         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
15612         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
15613         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15614         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
15615         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
15616         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
15617         (avrxmega6): Increase max flash segments from 5 to 6.
15618         * config/avr/t-multilib: Regenerate.
15619         * config/avr/avr-tables.opt: Regenerate.
15620         * doc/avr-mmcu.texi: Regenerate.
15622 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15624         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
15625         (avr_device_to_arch): Rename to avr_device_to_ld.
15626         (avr_device_to_as): New prototype.
15627         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
15628         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
15629         * config/avr/driver-avr.c (avr_device_to_as): New.
15630         (avr_device_to_arch): Rename to avr_device_to_ld.
15632 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15634         PR middle-end/56461
15635         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
15636         method on dr_chain and result_chain.
15638         PR middle-end/56461
15639         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
15640         pointer_set_destroy on not_executed_last_iteration.
15642         PR middle-end/56461
15643         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
15645         PR middle-end/56461
15646         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
15647         FOR_EACH_DEFINED_FUNCTION when freeing state.
15649         PR middle-end/56461
15650         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
15651         pool_free.
15652         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
15653         overwriting it.
15655         PR middle-end/56461
15656         * ipa-cp.c (decide_whether_version_node): Call vec_free on
15657         known_aggs[i].items and release known_aggs vector.
15659         PR middle-end/56461
15660         * ipa-reference.c (propagate): Free node_info even for alias nodes.
15662 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15664         * config/microblaze/microblaze.c (microblaze_emit_compare):
15665         Use xor for EQ/NE comparisions.
15666         * config/microblaze/microblaze.md (cstoresf4): Add constraints
15667         (cbranchsf4): Adjust operator to comparison_operator.
15669 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15671         PR middle-end/56461
15672         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
15673         vector.
15674         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
15675         vec_safe_push, always update *slot.
15676         (redirect_edge_var_map_clear): Use vec_free.
15677         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
15678         (free_var_map_entry): Use vec_free.
15679         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
15680         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
15682 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
15684         PR middle-end/45472
15685         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
15686         when the may_trap_p bit of the exprs being merged differs.
15687         Reorder tests for speculativeness in the logical and operator.
15689 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
15691         * incpath.c (add_standard_paths): Use reconcat instead of concat
15692         where appropriate and avoid leaking memory.
15694         * opts.h: Include obstack.h.
15695         (opts_concat): New prototype.
15696         (opts_obstack): New declaration.
15697         * opts.c (opts_concat): New function.
15698         (opts_obstack): New variable.
15699         (init_options_struct): Call gcc_init_obstack on opts_obstack.
15700         (finish_options): Use opts_concat instead of concat
15701         and XOBNEWVEC instead of XNEWVEC.
15702         * opts-common.c (generate_canonical_option, decode_cmdline_option,
15703         generate_option): Likewise.
15704         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
15705         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
15707         PR target/56455
15708         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
15709         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
15711 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15713         PR middle-end/56461
15714         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
15716 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15718         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
15719         (arm_block_move_unaligned_straight): Likewise.
15720         (arm_adjust_block_mem): Likewise.
15722 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
15724         PR target/48901
15725         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
15726         temp, cond and label.
15727         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
15729         PR target/52500
15730         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
15731         * config/c6x/c6x.h (dbx_register_map): Update declaration.
15733         PR target/52501
15734         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
15735         of prologue/epilogue functions.
15737         PR target/52550
15738         * config/tilegx/tilegx.c (tilegx_expand_prologue):
15739         Remove unused variable cfa_offset.
15740         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
15742         PR target/54639
15743         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
15744         type promotion to unsigned.
15746         PR target/54640
15747         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
15748         for HOST_WIDE_INT of 32 bit / same size as int.
15749         (arm_block_move_unaligned_straight): Likewise.
15750         (arm_adjust_block_mem): Likewise.
15752         PR target/54662
15753         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
15754         ALL_CFLAGS.
15756 2013-02-26  Marek Polacek  <polacek@redhat.com>
15758         PR tree-optimization/56426
15759         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
15761 2013-02-26  Richard Biener  <rguenther@suse.de>
15763         PR target/56444
15764         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
15765         unused variable loops.
15767 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
15769         PR tree-optimization/56448
15770         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
15771         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
15772         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
15773         later operands of the references, or even first operand for
15774         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
15776         PR tree-optimization/56443
15777         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
15778         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
15779         to type_for_mode langhook.
15781 2013-02-25  Matt Turner  <mattst88@gmail.com>
15783         * doc/invoke.texi: Document r4700.
15785 2013-02-25  Richard Biener  <rguenther@suse.de>
15787         PR tree-optimization/56175
15788         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
15789         split out from ...
15790         (simplify_bitwise_binary): ... here.  Also guard the conversion
15791         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
15793 2013-02-25  Catherine Moore  <clm@codesourcery.com>
15795         Revert:
15796         2013-02-24  Catherine Moore  <clm@codesourcery.com>
15797             Maciej W. Rozycki  <macro@codesourcery.com>
15798             Tom de Vries  <tom@codesourcery.com>
15799             Nathan Sidwell  <nathan@codesourcery.com>
15800             Iain Sandoe  <iain@codesourcery.com>
15801             Nathan Froyd  <froydnj@codesourcery.com>
15802             Chao-ying Fu  <fu@mips.com>
15804         * doc/extend.texi: (micromips, nomicromips, nocompression):
15805         Document new function attributes.
15806         * doc/invoke.texi (minterlink-compressed, mmicromips,
15807         m14k, m14ke, m14kec): Document new options.
15808         (minterlink-mips16): Update documentation.
15809         * doc/md.texi (ZC, ZD): Document new constraints.
15810         * configure.ac (gcc_cv_as_micromips): Check if linker
15811         supports the .set micromips directive.
15812         * configure: Regenerate.
15813         * config.in: Regenerate.
15814         * config/mips/mips-tables.opt: Regenerate.
15815         * config/mips/micromips.md: New file.
15816         * constraints.md (ZC, AD): New constraints.
15817         * config/mips/predicates.md (movep_src_register): New predicate.
15818         (movep_src_operand): New predicate.
15819         (non_volatile_mem_operand): New predicate.
15820         * config/mips/mips.md (multimem): New type.
15821         (length): Differentiate between 17-bit and 18-bit branch offsets.
15822         (MOVEP1, MOVEP2): New mode iterator.
15823         (mov_<load>l): Use ZC constraint.
15824         (mov_<load>r): Likewise.
15825         (mov_<store>l): Likewise.
15826         (mov_<store>r): Likewise.
15827         (*branch_equality<mode>_inverted): Add microMIPS support.
15828         (*branch_equality<mode>): Likewise.
15829         (*jump_absolute): Likewise.
15830         (indirect_jump_<mode>): Likewise.
15831         (tablejump_<mode>): Likewise.
15832         (<optab>_internal): Likewise.
15833         (sibcall_internal): Likewise.
15834         (sibcall_value_internal): Likewise.
15835         (prefetch): Use constraint ZD.
15836         * config/mips/mips.opt (minterlink-compressed): New option.
15837         (minterlink-mips16): Now an alias for minterlink-compressed.
15838         (mmicromips): New option.
15839         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15840         (compare_and_swap_12): Likewise.
15841         (sync_add<mode>): Likewise.
15842         (sync_<optab>_12): Likewise.
15843         (sync_old_<optab>_12): Likewise.
15844         (sync_new_<optab>_12): Likewise.
15845         (sync_nand_12): Likewise.
15846         (sync_old_nand_12): Likewise.
15847         (sync_new_nand_12): Likewise.
15848         (sync_sub<mode>): Likewise.
15849         (sync_old_add<mode>): Likewise.
15850         (sync_old_sub<mode>): Likewise.
15851         (sync_new_add<mode>): Likewise.
15852         (sync_new_sub<mode>): Likewise.
15853         (sync_<optab><mode>): Likewise.
15854         (sync_old_<optab><mode>): Likewise.
15855         (sync_new_<optab><mode>): Likewise.
15856         (sync_nand<mode>): Likewise.
15857         (sync_old_nand<mode>): Likewise.
15858         (sync_new_nand<mode>): Likewise.
15859         (sync_lock_test_and_set<mode>): Likewise.
15860         (test_and_set_12): Likewise.
15861         (atomic_compare_and_swap<mode>): Likewise.
15862         (atomic_exchange<mode>_llsc): Likewise.
15863         (atomic_fetch_add<mode>_llsc): Likewise.
15864         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15865         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15866         (umips_save_restore_pattern_p): Likewise.
15867         (umips_load_store_pair_p): Likewise.
15868         (umips_output_load_store_pair): Likewise.
15869         (umips_movep_target_p): Likewise.
15870         (umips_12bit_offset_address_p): Likewise.
15871         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15872         (mips_base_mips16): Rename this...
15873         (mips_base_compression_flags): ...to this. Update all uses.
15874         (mips_attribute_table): Add micromips, nomicromips and nocompression.
15875         (mips_mips16_decl_p): Delete.
15876         (mips_nomips16_decl_p): Delete.
15877         (mips_get_compress_on_flags): New function.
15878         (mips_get_compress_off_flags): New function.
15879         (mips_get_compress_mode): New function.
15880         (mips_get_compress_on_name): New function.
15881         (mips_get_compress_off_name): New function.
15882         (mips_insert_attributes): Support multiple compression types.
15883         (mips_merge_decl_attributes): Likewise.
15884         (umips_12bit_offset_address_p): New function.
15885         (mips_start_function_definition): Emit .set micromips directive.
15886         (mips_call_may_need_jalx_p): New function.
15887         (mips_function_ok_for_sibcall): Add microMIPS support.
15888         (mips_print_operand_punctuation): Support short delay slots and
15889         compact jumps.
15890         (umips_swm_mask, umips_swm_encoding): New.
15891         (umips_build_save_restore): New function.
15892         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15893         (was_mips16_p): Remove.
15894         (old_compression_mode): New.
15895         (mips_set_compression_mode): New function.
15896         (mips_set_current_function): Add microMIPS support.
15897         (mips_option_override): Likewise.
15898         (umips_save_restore_pattern_p): New function.
15899         (umips_output_save_restore): New function.
15900         (umips_load_store_pair_p_1): New function.
15901         (umips_load_store_pair_p): New function.
15902         (umips_output_load_store_pair_1): New function.
15903         (umips_output_load_store_pair): New function.
15904         (umips_movep_target_p) New function.
15905         (mips_prepare_pch_save): Add microMIPS support.
15906         * config/mips/mips.h (TARGET_COMPRESSION): New.
15907         (TARGET_CPU_CPP_BUILTINS): Update macro
15908         to use new compression flags and to support microMIPS.
15909         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15910         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15911         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15912         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15913         (ASM_SPEC): Support mmicromips and mno-micromips.
15914         (M16STORE_REG_P): New macro.
15915         (MIPS_CALL): Support TARGET_MICROMIPS.
15916         (MICROMIPS_J): New macro.
15917         (mips_base_mips16): Rename this...
15918         (mips_base_compression_flags): ...to this.
15919         (UMIPS_12BIT_OFFSET_P): New macro.
15920         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15921         (MULTILIB_DIRNAMES): Likewise.
15923 2013-02-25  Tom de Vries  <tom@codesourcery.com>
15925         PR rtl-optimization/56131
15926         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
15927         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
15928         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
15930 2013-02-25  Tobias Burnus  <burnus@net-b.de>
15932         * doc/invoke.texi (-fsanitize=): Move from optimization
15933         to debugging options.
15935 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15937         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
15939 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
15940             Alexander Monakov  <amonakov@ispras.ru>
15942         PR middle-end/56077
15943         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
15944         flush pending lists also on non-jumps.  Adjust comment.
15946 2013-02-24  Catherine Moore  <clm@codesourcery.com>
15947             Maciej W. Rozycki  <macro@codesourcery.com>
15948             Tom de Vries  <tom@codesourcery.com>
15949             Nathan Sidwell  <nathan@codesourcery.com>
15950             Iain Sandoe  <iain@codesourcery.com>
15951             Nathan Froyd  <froydnj@codesourcery.com>
15952             Chao-ying Fu  <fu@mips.com>
15954         * doc/extend.texi: (micromips, nomicromips, nocompression):
15955         Document new function attributes.
15956         * doc/invoke.texi (minterlink-compressed, mmicromips,
15957         m14k, m14ke, m14kec): Document new options.
15958         (minterlink-mips16): Update documentation.
15959         * doc/md.texi (ZC, ZD): Document new constraints.
15960         * configure.ac (gcc_cv_as_micromips): Check if linker
15961         supports the .set micromips directive.
15962         * configure: Regenerate.
15963         * config.in: Regenerate.
15964         * config/mips/mips-tables.opt: Regenerate.
15965         * config/mips/micromips.md: New file.
15966         * constraints.md (ZC, AD): New constraints.
15967         * config/mips/predicates.md (movep_src_register): New predicate.
15968         (movep_src_operand): New predicate.
15969         (non_volatile_mem_operand): New predicate.
15970         * config/mips/mips.md (multimem): New type.
15971         (length): Differentiate between 17-bit and 18-bit branch offsets.
15972         (MOVEP1, MOVEP2): New mode iterator.
15973         (mov_<load>l): Use ZC constraint.
15974         (mov_<load>r): Likewise.
15975         (mov_<store>l): Likewise.
15976         (mov_<store>r): Likewise.
15977         (*branch_equality<mode>_inverted): Add microMIPS support.
15978         (*branch_equality<mode>): Likewise.
15979         (*jump_absolute): Likewise.
15980         (indirect_jump_<mode>): Likewise.
15981         (tablejump_<mode>): Likewise.
15982         (<optab>_internal): Likewise.
15983         (sibcall_internal): Likewise.
15984         (sibcall_value_internal): Likewise.
15985         (prefetch): Use constraint ZD.
15986         * config/mips/mips.opt (minterlink-compressed): New option.
15987         (minterlink-mips16): Now an alias for minterlink-compressed.
15988         (mmicromips): New option.
15989         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15990         (compare_and_swap_12): Likewise.
15991         (sync_add<mode>): Likewise.
15992         (sync_<optab>_12): Likewise.
15993         (sync_old_<optab>_12): Likewise.
15994         (sync_new_<optab>_12): Likewise.
15995         (sync_nand_12): Likewise.
15996         (sync_old_nand_12): Likewise.
15997         (sync_new_nand_12): Likewise.
15998         (sync_sub<mode>): Likewise.
15999         (sync_old_add<mode>): Likewise.
16000         (sync_old_sub<mode>): Likewise.
16001         (sync_new_add<mode>): Likewise.
16002         (sync_new_sub<mode>): Likewise.
16003         (sync_<optab><mode>): Likewise.
16004         (sync_old_<optab><mode>): Likewise.
16005         (sync_new_<optab><mode>): Likewise.
16006         (sync_nand<mode>): Likewise.
16007         (sync_old_nand<mode>): Likewise.
16008         (sync_new_nand<mode>): Likewise.
16009         (sync_lock_test_and_set<mode>): Likewise.
16010         (test_and_set_12): Likewise.
16011         (atomic_compare_and_swap<mode>): Likewise.
16012         (atomic_exchange<mode>_llsc): Likewise.
16013         (atomic_fetch_add<mode>_llsc): Likewise.
16014         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16015         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16016         (umips_save_restore_pattern_p): Likewise.
16017         (umips_load_store_pair_p): Likewise.
16018         (umips_output_load_store_pair): Likewise.
16019         (umips_movep_target_p): Likewise.
16020         (umips_12bit_offset_address_p): Likewise.
16021         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16022         (mips_base_mips16): Rename this...
16023         (mips_base_compression_flags): ...to this. Update all uses.
16024         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16025         (mips_mips16_decl_p): Delete.
16026         (mips_nomips16_decl_p): Delete.
16027         (mips_get_compress_on_flags): New function.
16028         (mips_get_compress_off_flags): New function.
16029         (mips_get_compress_mode): New function.
16030         (mips_get_compress_on_name): New function.
16031         (mips_get_compress_off_name): New function.
16032         (mips_insert_attributes): Support multiple compression types.
16033         (mips_merge_decl_attributes): Likewise.
16034         (umips_12bit_offset_address_p): New function.
16035         (mips_start_function_definition): Emit .set micromips directive.
16036         (mips_call_may_need_jalx_p): New function.
16037         (mips_function_ok_for_sibcall): Add microMIPS support.
16038         (mips_print_operand_punctuation): Support short delay slots and
16039         compact jumps.
16040         (umips_swm_mask, umips_swm_encoding): New.
16041         (umips_build_save_restore): New function.
16042         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16043         (was_mips16_p): Remove.
16044         (old_compression_mode): New.
16045         (mips_set_compression_mode): New function.
16046         (mips_set_current_function): Add microMIPS support.
16047         (mips_option_override): Likewise.
16048         (umips_save_restore_pattern_p): New function.
16049         (umips_output_save_restore): New function.
16050         (umips_load_store_pair_p_1): New function.
16051         (umips_load_store_pair_p): New function.
16052         (umips_output_load_store_pair_1): New function.
16053         (umips_output_load_store_pair): New function.
16054         (umips_movep_target_p) New function.
16055         (mips_prepare_pch_save): Add microMIPS support.
16056         * config/mips/mips.h (TARGET_COMPRESSION): New.
16057         (TARGET_CPU_CPP_BUILTINS): Update macro
16058         to use new compression flags and to support microMIPS.
16059         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16060         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16061         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16062         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16063         (ASM_SPEC): Support mmicromips and mno-micromips.
16064         (M16STORE_REG_P): New macro.
16065         (MIPS_CALL): Support TARGET_MICROMIPS.
16066         (MICROMIPS_J): New macro.
16067         (mips_base_mips16): Rename this...
16068         (mips_base_compression_flags): ...to this.
16069         (UMIPS_12BIT_OFFSET_P): New macro.
16070         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16071         (MULTILIB_DIRNAMES): Likewise.
16073 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
16075         PR target/52555
16076         * target-globals.c (save_target_globals): For init_reg_sets and
16077         target_reinit remporarily set this_fn_optabs to this_target_optabs.
16079 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
16081         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16082         * config/aarch64/t-aarch64
16083         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16085 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
16087         PR inline-asm/56148
16088         * lra-constraints.c (process_alt_operands): Reload operand
16089         conflicting with earlier clobber only if no more other conflicting
16090         operands.
16092 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
16094         PR sanitizer/56393
16095         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16096         if not linking a shared library.
16098 2013-02-22  Seth LaForge  <sethml@google.com>
16100         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16102 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
16104         * config/arm/arm.md (split for extendsidi): Update condition.
16105         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16106         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16107         (qhs_zextenddi_cstr): Likewise.
16109 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16111         PR middle-end/56420
16112         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16113         avoid signed wrapping.
16114         (expand_mult): Handle properly multiplication by
16115         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
16116         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
16117         in the compiler if coeff is HOST_WIDE_INT_MIN.
16118         (expand_divmod): Don't make ext_op1 static, change it's type to
16119         uhwi.  Avoid undefined behavior in -INTVAL (op1).
16121         PR rtl-optimization/50339
16122         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16123         field.
16124         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16125         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16126         into splitting_ashiftrt field.
16127         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16128         ASHIFTRT.
16129         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16130         choices.
16132 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
16134         PR middle-end/56108
16135         * trans-mem.c (execute_tm_mark): Do not expand transactions that
16136         are sure to go irrevocable.
16138 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
16140         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16141         scalars are valid operands.
16143 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16145         PR tree-optimization/56310
16146         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16147         only matching indices and non-negative final offsets.
16148         (intersect_aggregates_with_edge): Pass src_idx to
16149         agg_replacements_to_vector.  Pass src_idx insstead of index to
16150         intersect_with_agg_replacements.
16152 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16154         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16155         instead of hard-wired defaults.
16157 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
16159         * doc/invoke.texi (MIPS Options): Update documentation of the
16160         floating-point multiply-accumulate instruction restrictions.
16162 2013-02-21  Kostya Serebryany  <kcc@google.com>
16164         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16165         asan_shadow_offset on x86_64 linux.
16167 2013-02-21  Richard Biener  <rguenther@suse.de>
16169         PR tree-optimization/56415
16170         Revert
16171         2013-02-11  Richard Biener  <rguenther@suse.de>
16173         PR tree-optimization/56273
16174         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16175         first VRP run.
16177 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16179         PR bootstrap/56258
16180         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16181         instead of @itemx.
16183         PR inline-asm/56405
16184         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16185         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16187 2013-02-20  Jan Hubicka  <jh@suse.cz>
16189         PR tree-optimization/56265
16190         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16191         when target is referenced for first time.
16193 2013-02-20  Richard Biener  <rguenther@suse.de>
16195         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16196         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16197         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16198         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16199         not return anything.
16200         (rename_ssa_copies): Do not remove unused locals.
16201         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16202         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16203         * passes.c (execute_function_todo): Do not schedule unused locals
16204         removal if cleanup_tree_cfg did something.
16205         * tree-ssa-live.c (remove_unused_locals): Dump statistics
16206         about the number of removed locals.
16208 2013-02-20  Richard Biener  <rguenther@suse.de>
16210         PR tree-optimization/56398
16211         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16213 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16215         PR tree-optimization/55334
16216         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16217         restricted pointers to arrays.
16219 2013-02-20  Richard Biener  <rguenther@suse.de>
16220             Jakub Jelinek  <jakub@redhat.com>
16222         PR tree-optimization/56396
16223         * tree-ssa-ccp.c (n_const_val): New static variable.
16224         (get_value): Return NULL for SSA names we don't have a lattice
16225         entry for.
16226         (ccp_initialize): Initialize n_const_val.
16227         * tree-ssa-copy.c (n_copy_of): New static variable.
16228         (init_copy_prop): Initialize n_copy_of.
16229         (get_value): Return NULL_TREE for SSA names we don't have a
16230         lattice entry for.
16232 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16234         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16236 2013-02-20  Richard Biener  <rguenther@suse.de>
16238         * genpreds.c (write_lookup_constraint): Do not compare first
16239         letter of the constraint again.
16241 2013-02-20  Richard Biener  <rguenther@suse.de>
16243         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16244         and ceil_log2.
16245         (get_use_iv_cost): Terminate hashtable walk when coming across
16246         an empty entry.
16248 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
16250         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
16251         reassociation for avx2 targets.
16253 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16255         * config/microblaze/microblaze.c: microblaze_has_clz = 0
16256         Add version check for v8.10.a to enable microblaze_has_clz
16257         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
16258         version and TARGET_PATTERN_COMPARE check
16259         * config/microblaze/microblaze.md: New clzsi2 instruction
16261 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16263         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
16264         function before branching.
16266 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16268         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
16269         DUMP_INSN_RTX_UID.
16270         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
16272 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16274         PR middle-end/55889
16275         * sel-sched.c: Include ira.h.
16276         (implicit_clobber_conflict_p): New function.
16277         (moveup_expr): Use it.
16278         * Makefile.in (sel-sched.o): Depend on ira.h.
16280 2013-02-19  Richard Biener  <rguenther@suse.de>
16282         PR tree-optimization/56384
16283         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
16284         (vn_hash_type): Split out from ...
16285         (vn_hash_constant_with_type): ... here.
16286         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
16287         (vn_phi_eq): Compare types from vn_phi_s structure.
16288         (vn_phi_lookup): Populate vn_phi_s type.
16289         (vn_phi_insert): Likewise.
16291 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
16293         PR tree-optimization/56350
16294         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
16295         if haven't found reduction or nested cycle operand, rather than
16296         asserting we must find it.
16298         PR tree-optimization/56381
16299         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
16300         to fold_build3.
16302 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
16303             Jakub Jelinek  <jakub@redhat.com>
16305         PR target/52555
16306         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
16307         (swap_optab_enable): Same.
16308         (init_all_optabs): Use argument instead of global.
16309         * tree.h (struct tree_optimization_option): New field target_optabs.
16310         * expr.h (init_all_optabs): Add argument to prototype.
16311         (TREE_OPTIMIZATION_OPTABS): New.
16312         (save_optabs_if_changed): Protoize.
16313         * optabs.h: Declare this_fn_optabs.
16314         * optabs.c (save_optabs_if_changed): New.
16315         Declare this_fn_optabs.
16316         (init_optabs): Add argument to init_all_optabs() call.
16317         * function.c (invoke_set_current_function_hook): Handle per
16318         function optabs.
16319         * function.h (struct function): New field optabs.
16320         * config/mips/mips.c (mips_set_mips16_mode): Handle when
16321         optimization_current_node has changed.
16322         * target-globals.h (save_target_globals_default_opts): Protoize.
16323         * target-globals.c (save_target_globals_default_opts): New.
16325 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16327         PR target/56347
16328         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
16329         registers %fr12 and %fr12R as call used.
16331         PR target/56214
16332         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
16333         and HImode, require all displacements to be an integer multiple of
16334         their mode size.
16335         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
16336         only allow QImode and HImode when reload is in progress and strict is
16337         true.  Likewise for symbolic addresses.  Use base14_operand to check
16338         displacements in REG+BASE addresses.
16340 2013-02-18  Richard Biener  <rguenther@suse.de>
16342         PR tree-optimization/56366
16343         * tree-vect-loop.c (get_initial_def_for_induction): Properly
16344         handle sign-conversion of outer-loop initial induction value.
16346 2013-02-18  Richard Biener  <rguenther@suse.de>
16348         PR middle-end/56349
16349         * cfghooks.c (merge_blocks): If we merge a latch into another
16350         block adjust references to it.
16351         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
16352         (verify_loop_structure): Verify that a recorded latch is in fact
16353         a latch.
16355 2013-02-18  Richard Biener  <rguenther@suse.de>
16357         PR tree-optimization/56321
16358         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
16359         order SSA name release and virtual operand unlinking.
16361 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16363         * config/microblaze/microblaze.md (save_stack_block): Define.
16364         (restore_stack_block): Likewise.
16366 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16368         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
16369         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
16370         * config/microblaze/microblaze.c (microblaze_option_override):
16371         Bail out early for PIC modes when target does not support PIC.
16373 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16375         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
16376         Replace with a microblaze version.
16377         (microblaze_trampoline_init): Adapt for microblaze.
16378         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
16379         microblaze.
16381 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
16382             Dodji Seketeli  <dodji@redhat.com>
16384         PR asan/56330
16385         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
16386         (instrument_mem_region_access): Do not forget to always put
16387         instrumentation of the of 'base' and 'base + len' in a "if (len !=
16388         0) statement, even for cases where either 'base' or 'base + len'
16389         are not instrumented -- because they have been previously
16390         instrumented.  Simplify the logic by putting all the statements
16391         instrument 'base + len' inside a sequence, and then insert that
16392         sequence right before the current insertion point.  Then, to
16393         instrument 'base + len', just get an iterator on that statement.
16394         And do not forget to update the pointer to iterator the function
16395         received as argument.
16397 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
16399         PR rtl-optimization/56348
16400         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
16402 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
16404         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
16405         (clean_graph_dump_file): Pass base to start_graph_dump.
16407 2013-02-14  Richard Henderson  <rth@redhat.com>
16409         PR target/55941
16410         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
16412 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
16414         * collect2-aix.h: Define F_LOADONLY.
16416 2013-02-14  Richard Biener  <rguenther@suse.de>
16418         PR lto/50494
16419         * varasm.c (output_constant_def_1): Get the decl representing
16420         the constant as argument.
16421         (output_constant_def): Wrap output_constant_def_1.
16422         (make_decl_rtl): Use output_constant_def_1 with the decl
16423         representing the constant.
16424         (build_constant_desc): Optionally re-use a decl already
16425         representing the constant.
16426         (tree_output_constant_def): Adjust.
16428 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
16430         Fix an asan crash
16431         * asan.c (instrument_builtin_call):  Really put the length of the
16432         second source argument into src1_len.
16434 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
16436         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
16437         argument.  If it is false, don't create edge from then_bb to
16438         fallthru_bb.
16439         (insert_if_then_before_iter): Pass true to it.
16440         (build_check_stmt): Pass false to it.
16441         (transform_statements): Flush hash table only on extended basic
16442         block boundaries, rather than at the beginning of every bb.
16443         Don't flush hash table on nonfreeing_call_p calls.
16444         * tree-flow.h (nonfreeing_call_p): New prototype.
16445         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
16447 2013-02-13  David S. Miller  <davem@davemloft.net>
16449         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
16451 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
16453         PR target/56184
16454         * ira.c (max_regno_before_ira): Move from ...
16455         (ira): ... here.
16456         (fix_reg_equiv_init): Use max_regno_before_ira instead of
16457         vec_safe_length.
16459 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
16461         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
16463 2013-02-13  Richard Biener  <rguenther@suse.de>
16465         PR lto/56295
16466         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
16467         globals in MEM_REFs.
16469 2013-02-13  Richard Biener  <rguenther@suse.de>
16471         * loop-init.c (loop_optimizer_init): Clear loop state when
16472         re-initializing preserved loops.
16473         * loop-unswitch.c (unswitch_single_loop): Return whether
16474         we unswitched the loop.  Do not verify loop state here.
16475         (unswitch_loops): When we unswitched a loop discover new loops.
16477 2013-02-13  Kostya Serebryany  <kcc@google.com>
16479         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
16480         on x86_64 linux.
16481         * sanitizer.def: Rename __asan_init to __asan_init_v1.
16483 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
16485         Avoid instrumenting duplicated memory access in the same basic block
16486         * Makefile.in (asan.o): Add new dependency on hash-table.h
16487         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
16488         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
16489         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
16490         (free_mem_ref_resources, has_mem_ref_been_instrumented)
16491         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
16492         (get_mem_ref_of_assignment): New functions.
16493         (get_mem_refs_of_builtin_call): Extract from
16494         instrument_builtin_call and tweak a little bit to make it fit with
16495         the new signature.
16496         (instrument_builtin_call): Use the new
16497         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
16498         of is_gimple_builtin_call.
16499         (instrument_derefs, instrument_mem_region_access): Insert the
16500         instrumented memory reference into the hash table.
16501         (maybe_instrument_assignment): Renamed instrument_assignment into
16502         this, and change it to advance the iterator when instrumentation
16503         actually happened and return true in that case.  This makes it
16504         homogeneous with maybe_instrument_assignment, and thus give a
16505         chance to callers to be more 'regular'.
16506         (transform_statements): Clear the memory reference hash table
16507         whenever we enter a new BB, when we cross a function call, or when
16508         we are done transforming statements.  Use
16509         maybe_instrument_assignment instead of instrumentation.  No more
16510         need to special case maybe_instrument_assignment and advance the
16511         iterator after calling it; it's now handled just like
16512         maybe_instrument_call.  Update comment.
16514 2013-02-13  Richard Biener  <rguenther@suse.de>
16516         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
16517         Fix loop discovery code.
16519 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
16521         PR inline-asm/56148
16522         * lra-constraints.c (process_alt_operands): Match early clobber
16523         operand with itself.  Check conflicts with earlyclobber only if
16524         the operand is not reloaded.  Prefer to reload conflicting operand
16525         if earlyclobber and matching operands are the same.
16527 2013-02-12  Richard Biener  <rguenther@suse.de>
16529         PR lto/56297
16530         * lto-streamer-out.c (write_symbol): Do not output symbols
16531         for hard register variables.
16533 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
16535         PR target/54222
16536         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
16537         (umulsidi3_insn, mulsidi3_insn): New insns.
16539 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
16541         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
16542         (struct tune_params): Add vec_costs field.
16543         * config/arm/arm.c (arm_builtin_vectorization_cost)
16544         (arm_add_stmt_cost): New functions.
16545         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
16546         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
16547         (arm_default_vec_cost): New struct of type cpu_vec_costs.
16548         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
16549         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
16550         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
16551         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
16553 2013-02-12  Richard Biener  <rguenther@suse.de>
16555         PR lto/56295
16556         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
16557         decls again if possible.
16559 2013-02-12  Richard Biener  <rguenther@suse.de>
16561         PR middle-end/56288
16562         * tree-ssa.c (verify_ssa_name): Fix check, move
16563         SSA_NAME_IN_FREE_LIST check up.
16565 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
16566             Steven Bosscher   <steven@gcc.gnu.org>
16568         PR rtl-optimization/56151
16569         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
16570         equal to op0 or op1, and last_insn pattern is CODE operation
16571         with MEM dest and one of the operands matches that MEM.
16573 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
16575         * doc/extend.texi: Document Function Multiversioning and "default"
16576         parameter string to target attribute.
16577         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
16578         target attribute parameter is "default".
16579         (ix86_compare_version_priority): Remove checks for target attribute.
16580         (ix86_mangle_function_version_assembler_name): Change error to sorry.
16581         Remove check for target attribute equal to NULL. Add assert.
16582         (ix86_generate_version_dispatcher_body): Change error to sorry.
16584 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
16585             Jack Howarth  <howarth@bromo.med.uc.edu>
16586             Patrick Marlier  <patrick.marlier@gmail.com>
16588         PR libitm/55693
16589         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
16590         define ENDFILE_SPEC as TM_DESTRUCTOR.
16591         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
16593 2013-02-11  Alexander Potapenko  <glider@google.com>
16594             Jack Howarth  <howarth@bromo.med.uc.edu>
16595             Jakub Jelinek  <jakub@redhat.com>
16597         PR sanitizer/55617
16598         * config/darwin.c (cdtor_record): Rename ctor_record.
16599         (sort_cdtor_records): Rename sort_ctor_records.
16600         (finalize_dtors): New routine to sort destructors by
16601         priority before use in assemble_integer.
16602         (machopic_asm_out_destructor): Use finalize_dtors if needed.
16604 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
16606         PR rtl-optimization/56275
16607         * simplify-rtx.c (avoid_constant_pool_reference): Check that
16608         offset is non-negative and less than cmode size before
16609         calling simplify_subreg.
16611 2013-02-11  Richard Biener  <rguenther@suse.de>
16613         PR tree-optimization/56264
16614         * cfgloop.h (fix_loop_structure): Adjust prototype.
16615         * loop-init.c (fix_loop_structure): Return the number of
16616         newly discovered loops.
16617         * tree-cfgcleanup.c (repair_loop_structures): When new loops
16618         are discovered, do a full loop-closed SSA rewrite.
16620 2013-02-11  Richard Biener  <rguenther@suse.de>
16622         PR tree-optimization/56273
16623         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16624         first VRP run.
16625         (check_array_ref): Fix missing newline in dumps.
16626         (search_for_addr_array): Likewise.
16628 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
16630         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
16632 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
16634         PR target/56256
16635         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
16637 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
16639         PR rtl-optimization/56246
16640         * lra-constraints.c (simplify_operand_subreg): Try to reuse
16641         reload pseudo.
16642         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
16643         constraints are satisfied.
16645 2013-02-08  Jeff Law  <law@redhat.com>
16647         PR debug/53948
16648         * emit-rtl.c (reg_is_parm_p): New function.
16649         * regs.h (reg_is_parm_p): New prototype.
16650         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
16651         callee-clobbered registers.
16653 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
16655         PR target/56043
16656         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
16657         If there is no implicit builtin declaration, just return NULL.
16659 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16661         * config/i386/sse.md (FMAMODEM): New mode iterator.
16662         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
16663         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
16665 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16667         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
16668         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
16669         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
16671 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16673         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
16674         (microblaze*-*-elf): Likewise.
16675         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
16676         LINK_SPEC.
16677         * config/microblaze/microblaze-c.c: Add builtin defines for
16678         _LITTLE_ENDIAN and _BIG_ENDIAN.
16679         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
16680         add to TARGET_DEFAULT flags.
16681         Expand ASM_SPEC and LINK_SPEC.
16682         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
16683         * config/microblaze/microblaze.md: Update extendsidi2 and
16684         movdi_internal instructions to use low-order / high-order reg
16685         print_operands.
16686         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
16687         options and inversemask / mask of LITTLE_ENDIAN.
16688         * config/microblaze/t-microblaze: Expand multilib options to
16689         include mlittle-endian (le) and update exceptions patterns.
16691 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
16693         PR rtl-optimization/56195
16694         * lra-constraints.c (get_reload_reg): Don't reuse regs
16695         if they have smaller mode than requested, if they have
16696         wider mode than requested, try to return a SUBREG.
16698         PR tree-optimization/56250
16699         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
16700         if type is unsigned and code isn't MULT_EXPR.
16702 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16704         PR tree-optimization/56064
16705         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
16706         bits according to mode.
16707         * fixed-value.h (fixed_from_double_int)
16708         (const_fixed_from_double_int): Adjust comments.
16710 2013-02-08  Richard Biener  <rguenther@suse.de>
16712         PR lto/56231
16713         * lto-streamer.h (struct data_in): Remove current_file, current_line
16714         and current_col members.
16715         * lto-streamer-out.c (lto_output_location): Stream changed bits
16716         en-block for efficiency.
16717         * lto-streamer-in.c (clear_line_info): Remove.
16718         (lto_input_location): Cache current file, line and column
16719         globally via local statics.  Read changed bits en-block.
16720         (input_function): Do not call clear_line_info.
16721         (lto_read_body): Likewise.
16722         (lto_input_toplevel_asms): Likewise.
16724 2013-02-08  Michael Matz  <matz@suse.de>
16726         PR tree-optimization/52448
16727         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
16728         (nt_call_phase): New static.
16729         (add_or_mark_expr): Only mark accesses with newer phase than any
16730         call seen.
16731         (nonfreeing_call_p): New.
16732         (nt_init_block): Update nt_call_phase, mark blocks as visited.
16733         (nt_fini_block): Keep blocks marked as visited.
16734         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
16736 2013-02-08  Richard Biener  <rguenther@suse.de>
16738         * ira.c (ira): Free broken dominator information.
16740 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
16742         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
16744 2013-02-08  Marek Polacek  <polacek@redhat.com>
16746         * cfgloop.c (verify_loop_structure): Add more checking of headers.
16748 2013-02-08  Richard Biener  <rguenther@suse.de>
16750         PR middle-end/56181
16751         * cfgloop.h (flow_loops_find): Adjust.
16752         (bb_loop_header_p): Declare.
16753         * cfgloop.c (bb_loop_header_p): New function split out from ...
16754         (flow_loops_find): ... here.  Adjust function signature,
16755         support incremental loop structure update.
16756         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
16757         * cfgloopmanip.c (fix_loop_structure): Move ...
16758         * loop-init.c (fix_loop_structure): ... here.
16759         (apply_loop_flags): Split out from ...
16760         (loop_optimizer_init): ... here.
16761         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
16762         in incremental mode, only remove dead loops here.
16764 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
16766         PR target/54222
16767         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
16768         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
16769         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
16770         (*round<mode>3.libgcc): New insns for fixed-modes.
16771         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
16772         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
16773         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
16774         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
16775         implementations.  Define to __builtin_avr_absFX,
16776         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
16777         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
16778         __builtin_avr_countlsFX, respectively.
16779         * config/avr/avr-c.c (target.h): Include it.
16780         (enum avr_builtin_id): New enum.
16781         (avr_resolve_overloaded_builtin): New static function.
16782         (avr_register_target_pragmas): Use it to set
16783         targetm.resolve_overloaded_builtin.
16784         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
16785         tree nodes used by DEF_BUILTIN.
16786         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
16787         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
16788         <AVR_BUILTIN_xxBITS>: Same.
16790 2013-02-08  Richard Biener  <rguenther@suse.de>
16792         * cfgloop.c (verify_loop_structure): Properly handle
16793         a loop exiting to another loop header.
16794         * ira-int.h (ira_loops): Remove.
16795         * ira.c (ira_loops): Remove.
16796         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
16797         (do_reload): Use loop_optimizer_finalize.
16798         * ira-build.c (create_loop_tree_nodes): Use get_loops and
16799         number_of_loops to access the loop tree.
16800         (more_one_region_p): Likewise.
16801         (finish_loop_tree_nodes): Likewise.
16802         (rebuild_regno_allocno_maps): Likewise.
16803         (mark_loops_for_removal): Likewise.
16804         (mark_all_loops_for_removal): Likewise.
16805         (remove_unnecessary_regions): Likewise.
16806         (ira_build): Likewise.
16807         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
16809 2013-02-08  Richard Biener  <rguenther@suse.de>
16811         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
16812         * ipa-pure-const.c (analyze_function): Avoid calling
16813         mark_irreducible_loops twice.
16814         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
16816 2013-02-07  David S. Miller  <davem@davemloft.net>
16818         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
16819         on 'reg'.
16820         * var-tracking.c (vt_add_function_parameter): Test the presence of
16821         HAVE_window_save properly and do not remap argument registers when
16822         we have a leaf function.
16824 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
16826         PR bootstrap/56227
16827         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
16828         instead of "ll".
16829         * config/i386/i386.c (ix86_print_operand): Ditto.
16831 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16833         * lra-constraints.c (process_alt_operands): Fix recently added comment.
16835 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
16837         PR rtl-optimization/56225
16838         * lra-constraints.c (process_alt_operands): Check that reload hard
16839         reg can hold value for strict_low_part.
16841 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16843         PR debug/56154
16844         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
16845         dwarf2out_end_function.
16846         (in_first_function_p, maybe_at_text_label_p,
16847         first_loclabel_num_not_at_text_label): New variables.
16848         (dwarf2out_var_location): In the first function find out
16849         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
16850         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
16851         functions.
16853 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
16855         PR rtl-optimization/56178
16856         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
16857         SUBREG of a register.  Tidy up related block of code.
16858         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
16859         note if the source is a register or a SUBREG of a register.
16861 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
16863         PR target/56228
16864         * config/rs6000/rs6000.md (ptrm): New mode attr.
16865         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
16866         call_value_indirect_aix<pttrsize>,
16867         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
16868         m in constraints.
16870 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
16872         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
16873         if -bnortl. Convert to strcmp and strncmp.
16875 2013-02-07  Alan Modra  <amodra@gmail.com>
16877         PR target/54009
16878         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
16879         addresses won't wrap when offsetting.
16880         (rs6000_secondary_reload): Provide secondary reloads needed for
16881         wrapping LO_SUM addresses.
16883 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
16885         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
16886         MACH, just __MACH__.
16888 2013-02-06  Richard Biener  <rguenther@suse.de>
16890         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
16891         instead of calling fix_loop_structure.
16893 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
16895         PR middle-end/56217
16896         * omp-low.c (use_pointer_for_field): Return false if
16897         lower_send_shared_vars doesn't generate any copy-out code.
16899 2013-02-06  Tom de Vries  <tom@codesourcery.com>
16901         PR rtl-optimization/56131
16902         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
16903         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
16904         of the label is NULL.  Add comment.
16906 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
16908         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
16910         PR sanitizer/55374
16911         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
16912         (STATIC_LIBTSAN_LIBS): Likewise.
16913         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
16914         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
16915         is defined, don't add anything else beyond that.
16916         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
16917         (LINK_COMMAND_SPEC): Use them.
16919         PR tree-optimization/56205
16920         * tree-stdarg.c (check_all_va_list_escapes): Return true if
16921         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
16922         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
16924 2013-02-05  Richard Biener  <rguenther@suse.de>
16926         PR tree-optimization/53342
16927         PR tree-optimization/53185
16928         * tree-vectorizer.h (vect_check_strided_load): Remove.
16929         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
16930         not disallow peeling for vectorized strided loads.
16931         (vect_check_strided_load): Make static and simplify.
16932         (vect_analyze_data_refs): Adjust.
16933         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
16934         correctly when vectorizing strided loads.
16936 2013-02-05  Richard Biener  <rguenther@suse.de>
16938         * doc/install.texi: Refer to ISL, not PPL.
16940 2013-02-05  Jan Hubicka  <jh@suse.cz>
16942         PR tree-optimization/55789
16943         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
16945 2013-02-05  Jan Hubicka  <jh@suse.cz>
16947         PR tree-optimization/55789
16948         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
16949         the dead call anyway.
16951 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
16953         PR sanitizer/55374
16954         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
16956 2013-02-04  Alexander Potapenko  <glider@google.com>
16957             Jack Howarth  <howarth@bromo.med.uc.edu>
16958             Jakub Jelinek  <jakub@redhat.com>
16960         PR sanitizer/55617
16961         * config/darwin.c (sort_ctor_records): Stabilized qsort
16962         on constructor priority by using original position.
16963         (finalize_ctors): New routine to sort constructors by
16964         priority before use in assemble_integer.
16965         (machopic_asm_out_constructor): Use finalize_ctors if needed.
16967 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
16969         PR libstdc++/54314
16970         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
16971         about visibility on artificial decls.
16972         * config/sol2.c (solaris_assemble_visibility): Likewise.
16974 2013-02-04  Kai Tietz  <ktietz@redhat.com>
16976         PR target/56186
16977         * config/i386/i386.c (function_value_ms_64): Add additional valtype
16978         argument and improve checking of return-argument types for 16-byte
16979         modes.
16980         (ix86_function_value_1): Add additional valtype argument on call
16981         of function_value_64.
16982         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16983         handling infunction_value_64 function.
16985 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
16987         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16989 2013-02-04  Richard Biener  <rguenther@suse.de>
16991         PR tree-optimization/56188
16992         * tree-ssa-structalias.c (label_visit): Consider case with
16993         initially non-empty points-to set.
16994         (perform_var_substitution): Dump node mapping and clean up.
16996 2013-02-04  Richard Guenther  <rguenther@suse.de>
16998         PR lto/56168
16999         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
17000         node prevail as last resort.
17001         (lto_symtab_merge_decls): Remove guard on LTRANS here.
17002         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
17004 2013-02-04  Richard Biener  <rguenther@suse.de>
17006         PR tree-optimization/56113
17007         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
17008         Merge into ...
17009         (equiv_class_lookup_or_add): ... this.
17010         (label_visit): Adjust and fix error in previous patch.
17011         (perform_var_substitution): Adjust.
17013 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
17015         * config/sh/divtab.c: Fix formatting and comments throughout the file.
17016         * config/sh/sh4-300.md: Likewise.
17017         * config/sh/sh4a.md: Likewise.
17018         * config/sh/constraints.md: Likewise.
17019         * config/sh/sh.md: Likewise.
17020         * config/sh/netbsd-elf.h: Likewise.
17021         * config/sh/predicates.md: Likewise.
17022         * config/sh/sh-protos.h: Likewise.
17023         * config/sh/ushmedia.h: Likewise.
17024         * config/sh/linux.h: Likewise.
17025         * config/sh/sh.c: Likewise.
17026         * config/sh/superh.h: Likewise.
17027         * config/sh/elf.h: Likewise.
17028         * config/sh/sh4.md: Likewise.
17029         * config/sh/sh.h: Likewise.
17031 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17033         * config/pa/constraints.md: Adjust unused letters.  Change "T"
17034         constraint to match_test floating_point_store_memory_operand().
17035         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
17036         (base14_operand): New.
17037         (floating_point_store_memory_operand): New.
17038         (integer_store_memory_operand): Revise to use base14_operand and
17039         reg_plus_base_memory_operand.
17040         (move_dest_operand): Allow symbolic_memory_operands.
17041         (symbolic_memory_operand): Check for LO_SOM.
17042         (symbolic_operand): Change default case to break.
17043         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
17044         CONST_DOUBLE values to be reloaded by putting them into memory when
17045         the destination is a floating point register.
17046         (movdf): Remove code to handle CONST_DOUBLE.
17047         (movsf): Likewise.
17048         (reload_indf_r1): New.
17049         (reload_insf_r1): New.
17050         Consistently use "Q" and "T" constraints with integer and floating
17051         point move instructions, respectively.
17052         (movdi): Remove FAIL.
17053         Change predicate for source operand unamed DImode move from
17054         general_operand to move_src_operand.
17055         (umulsidi3): Change predicate for destination operand to
17056         register_operand.
17057         Likewise for similar unamed patterns.
17058         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17059         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17060         (hppa_legitimize_address): Simplify mask calculation.
17061         (pa_emit_move_sequence): Revised handling of secondary reloads from
17062         REG+D addresses for floating point loads and stores.  Directly handle
17063         loading CONST0_RTX (mode) to a floating point register.
17064         (pa_secondary_reload): Handle reloading DF and SFmode constant values
17065         to floating point registers.  Don't restrict secondary reloads to
17066         floating point registers to integer modes.  Revise some comments and
17067         cleanup some code.
17068         (TARGET_LEGITIMATE_ADDRESS_P): Define.
17069         (pa_legitimate_address_p): New.
17070         (pa_legitimize_reload_address): New.
17071         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17072         (STRICT_REG_OK_FOR_BASE_P): New.
17073         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
17074         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17076 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
17077             Andrew Dixie  <andrewd@gentrack.com>
17079         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17080         flag set.
17082 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
17084         * expmed.c (extract_bit_field_1): Pass the full width of the
17085         structure to get_best_reg_extraction_insn.
17087 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17089         PR target/54601
17090         * configure.ac (use_cxa_atexit): Add AIX.
17091         * configure: Regenerate.
17093         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17095 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
17097         PR debug/54793
17098         * final.c (need_profile_function): New variable.
17099         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17100         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17101         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17102         notes, targetm.asm_out.function_prologue doesn't emit anything,
17103         HAVE_prologue and profiler should be emitted before prologue,
17104         set need_profile_function instead of emitting it.
17105         (final_scan_insn): If need_profile_function, emit
17106         profile_function on the first NOTE_INSN_BASIC_BLOCK or
17107         NOTE_INSN_FUNCTION_BEG note.
17109 2013-02-01  Richard Henderson  <rth@redhat.com>
17111         * config/rs6000/rs6000.md (smulditi3): New.
17112         (umulditi3): New.
17114         * config/alpha/alpha.md (umulditi3): New.
17116 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17118         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17119         (ASM_OUTPUT_ALIGNED_LOCAL): New.
17121 2013-02-01  Richard Biener  <rguenther@suse.de>
17123         PR tree-optimization/56113
17124         * tree-ssa-structalias.c (label_visit): Reduce work for
17125         single-predecessor nodes.
17127 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
17129         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17130         range isn't testing for zero.
17132 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
17134         PR middle-end/56113
17135         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17137 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
17138             Nick Clifton  <nickc@redhat.com>
17140         * config/v850/constraints.md (Q): Define as a memory constraint.
17141         * config/v850/predicates.md (label_ref_operand): New predicate.
17142         (e3v5_shift_operand): New predicate.
17143         (ior_operator): New predicate.
17144         * config/v850/t-v850: Add e3v5 multilib.
17145         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17146         (v850_gen_movdi): Prototype.
17147         * config/v850/v850.c: Add support for e3v5 architecture.
17148         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17149         TARGET_V850E_UP.
17150         (construct_save_jarl): Add e3v5 long JARL support.
17151         (v850_adjust_insn_length): New function.  Adjust length of call
17152         insns when using e3v5 instructions.
17153         (v850_gen_movdi): New function: Generate instructions to move a
17154         DImode value.
17155         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17156         (CPP_SPEC): Define __v850e3v5__ as appropriate.
17157         (TARGET_USE_FPU): Enable for e3v5.
17158         (CONST_OK_FOR_W): New macro.
17159         (ADJUST_INSN_LENGTH): Define.
17160         * config/v850/v850.md (UNSPEC_LOOP): Define.
17161         (attr cpu): Add v850e3v5.
17162         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17163         (movdi): New pattern.
17164         (movdi_internal): New pattern.
17165         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17166         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17167         (cstoresf4): Likewise.
17168         (cstoredf4): Likewise.
17169         (insv): New pattern.
17170         (rotlso3_a): New pattern.
17171         (rotlsi3_b): New pattern
17172         (rotlsi3_v850e3v5): New pattern.
17173         (doloop_begin): New pattern.
17174         (fix_loop_counter): New pattern.
17175         (doloop_end): New pattern.
17176         (branch_normal): Add e3v5 long branch support.
17177         (branch_invert): Likewise.
17178         (branch_z_normal): Likewise.
17179         (branch_z_invert): Likewise.
17180         (branch_nz_normal): Likewise.
17181         (branch_nz_invert): Likewise.
17182         (call_internal_short): Add e3v5 register-indirect JARL support.
17183         (call_internal_long): Likewise.
17184         (call_value_internal_short): Likewise.
17185         (call_value_internal_long): Likewise.
17186         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17187         (mloop): New option.
17188         * config.gcc: Add support for configuring v840e3v5 target.
17189         * doc/invoke.texi: Document new v850 specific command line options.
17191 2013-01-31  Paul Koning  <ni1d@arrl.net>
17193         PR debug/55059
17194         PR debug/54508
17195         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17196         children if parent is a class.
17197         (prune_unused_types_prune): Don't add DW_AT_declaration.
17199 2013-01-31  Richard Biener  <rguenther@suse.de>
17201         PR tree-optimization/56157
17202         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17203         match up operand with SLP child.
17205 2013-01-31  Jason Merrill  <jason@redhat.com>
17207         PR debug/54410
17208         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17209         parameters the first time.
17210         (gen_scheduled_generic_parms_dies): Check completeness here.
17212 2013-01-31  Richard Biener  <rguenther@suse.de>
17214         PR middle-end/53073
17215         * common.opt (faggressive-loop-optimizations): New flag,
17216         enabled by default.
17217         * doc/invoke.texi (faggressive-loop-optimizations): Document.
17218         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17219         infer_loop_bounds_from_undefined by it.
17221 2013-01-31  Richard Biener  <rguenther@suse.de>
17223         PR tree-optimization/56150
17224         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17225         visit virtual operands.
17226         (find_uses_to_rename_bb): Likewise.
17228 2013-01-31  Richard Biener  <rguenther@suse.de>
17230         PR tree-optimization/56150
17231         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17232         mixed store non-store stmts.
17234 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
17236         PR sanitizer/55374
17237         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17238         LIBASAN_EARLY_SPEC is defined.
17239         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17240         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17241         before %o.
17242         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17244         PR c++/55742
17245         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17246         invalid args instead of ICEing on it.
17247         (ix86_valid_target_attribute_tree): Return error_mark_node if
17248         ix86_valid_target_attribute_inner_p failed.
17249         (ix86_valid_target_attribute_p): Return false only if
17250         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
17251         target("default") attribute.
17252         (sorted_attr_string): Change argument from const char * to tree,
17253         merge in all target attribute arguments rather than just one.
17254         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
17255         instead of free.  Avoid using strcat.
17256         (ix86_mangle_function_version_assembler_name): Mangle
17257         target("default") as if no target attribute is present.  Adjust
17258         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
17259         instead of xmalloc and XDELETEVEC instead of free.
17260         (ix86_function_versions): Don't return true if one of the decls
17261         doesn't have target attribute.  If they don't and one of the decls
17262         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
17263         sorted_attr_string caller.  Use XDELETEVEC instead of free.
17264         (ix86_supports_function_versions): Remove.
17265         (make_name): Fix up formatting.
17266         (make_dispatcher_decl): Remove resolver_name and its initialization.
17267         Avoid leaking memory.
17268         (is_function_default_version): Return true if there is
17269         target("default") attribute rather than no target attribute at all.
17270         (make_resolver_func): Avoid leaking memory.
17271         (ix86_generate_version_dispatcher_body): Likewise.
17272         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
17273         * target.def (supports_function_versions): Remove.
17274         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
17275         * doc/tm.texi: Regenerated.
17277 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
17279         PR rtl-optimization/56144
17280         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
17281         for values with side effects.
17283 2013-01-30  Richard Biener  <rguenther@suse.de>
17285         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
17286         (sparseset_pop): Likewise.
17287         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
17288         to be able to use quick_push in the worker loop.
17290 2013-01-30  Marek Polacek  <polacek@redhat.com>
17292         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
17294 2013-01-30  Richard Biener  <rguenther@suse.de>
17296         PR lto/56147
17297         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
17299 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
17301         PR tree-optimization/56064
17302         * fixed-value.c (fixed_from_double_int): New function.
17303         * fixed-value.h (fixed_from_double_int): New prototype.
17304         (const_fixed_from_double_int): New static inline function.
17305         * fold-const.c (native_interpret_fixed): New static function.
17306         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
17307         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
17308         (native_encode_fixed): New static function.
17309         (native_encode_expr) <FIXED_CST>: Use it.
17310         (native_interpret_int): Move double_int worker code to...
17311         * double-int.c (double_int::from_buffer): ...this new static method.
17312         * double-int.h (double_int::from_buffer): Prototype it.
17314 2013-01-30  Richard Biener  <rguenther@suse.de>
17316         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
17317         New pointer-map and obstack.
17318         (init_alias_vars): Allocate pointer-map and obstack.
17319         (delete_points_to_sets): Free them.
17320         (find_what_var_points_to): Cache result.
17321         (find_what_p_points_to): Adjust for changed interface of
17322         find_what_var_points_to.
17323         (compute_points_to_sets): Likewise.
17324         (ipa_pta_execute): Likewise.
17326 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17328         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
17329         * configure: Regenerate.
17330         * config.in: Regenerate.
17331         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
17332         #nobits/#progbits if supported.
17334 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
17336         PR target/56121
17337         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
17338         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
17339         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
17341 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17343         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
17344         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
17346 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17348         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
17349         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
17351 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17353         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
17354         declaration.
17355         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
17356         * config/arm/cortex-a7.md: New bypasses using
17357         arm_mac_accumulator_is_result.
17359 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17361         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
17362         (cortex_a7_neon_mla): Likewise.
17363         (cortex_a7_fpfmad): New reservation.
17364         (cortex_a7_fpmacs): Use ffmas and update required units.
17365         (cortex_a7_fpmuld): Update required units and latency.
17366         (cortex_a7_fpmacd): Likewise.
17367         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
17368         (cortex_a7_neon). Likewise.
17369         (bypass) Update participating units.
17371 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17373         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
17374         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
17375         from fmac to ffma.
17376         * config/arm/vfp11.md (vfp_farith): Use ffmas.
17377         (vfp_fmul): Use ffmad.
17378         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
17379         (cortex_r4_fmacd): Use ffmad.
17380         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
17381         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
17382         (cortex_a9_fmacd): Use ffmad.
17383         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
17384         (cortex_a8_vfp_macd): Use ffmad.
17385         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
17386         (cortex_a5_fpmacd): Use ffmad.
17387         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
17388         (cortex_a15_vfp_macd): Use ffmad.
17389         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
17391 2013-01-29  Jason Merrill  <jason@redhat.com>
17393         PR libstdc++/54314
17394         * varasm.c (default_assemble_visibility): Don't warn about
17395         visibility on artificial decls.
17397 2013-01-29  Richard Biener  <rguenther@suse.de>
17399         PR tree-optimization/56113
17400         * tree-ssa-structalias.c (equiv_class_lookup): Also return
17401         the bitmap leader.
17402         (label_visit): Free duplicate bitmaps and record the leader instead.
17403         (perform_var_substitution): Adjust.
17405 2013-01-29  Richard Biener  <rguenther@suse.de>
17407         PR tree-optimization/55270
17408         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
17409         the CFG, schedule loops for fixup.
17411 2013-01-29  Nick Clifton  <nickc@redhat.com>
17413         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
17414         SP_REG.
17416 2013-01-28  Leif Ekblad  <leif@rdos.net>
17418         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
17419         * config/i386/i386.h (TARGET_RDOS): New macro.
17420         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
17421         * config/i386/i386.c (ix86_option_override_internal): For 64bit
17422         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
17423         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
17424         DEFAULT_LARGE_SECTION_THRESHOLD.
17425         * config/i386/i386.md (R14_REG, R15_REG): New constants.
17426         * config/i386/rdos.h: New file.
17427         * config/i386/rdos64.h: New file.
17429 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
17431         PR other/54814
17432         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
17433         TEST_HARD_REG_BIT.
17435 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
17437         PR rtl-optimization/56117
17438         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
17439         call cselib_lookup_from_insn on the MEM before calling
17440         add_insn_mem_dependence.
17442 2013-01-28  Richard Biener  <rguenther@suse.de>
17444         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
17445         to a stmt that didn't have one.
17446         (copy_phis_for_bb): Likewise for PHI arguments.
17447         (copy_debug_stmt): Likewise for debug stmts.
17449 2013-01-28  Richard Biener  <rguenther@suse.de>
17451         PR tree-optimization/56034
17452         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
17453         (partition_builtin_p): Adjust.
17454         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
17455         it is the last partition.
17456         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
17457         up the vertex for the definition.
17458         (classify_partition): Classify whether a partition is a
17459         PKIND_REDUCTION, thus has uses outside of the loop.
17460         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
17461         Merge all PKIND_REDUCTION partitions into the last partition.
17462         (tree_loop_distribution): Seed partitions from reductions as well.
17464 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
17466         PR tree-optimization/56125
17467         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
17468         pow(x,c) into sqrt(x) * powi(x, n/2) or
17469         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
17470         optimizing for size.
17471         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
17472         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
17473         integer.
17475         PR tree-optimization/56094
17476         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
17477         to UNKNOWN_LOCATION while gimplifying expr.
17479 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
17481         PR target/56114
17482         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
17483         operand 0 in movabs insn template for -masm=intel asm alternative.
17484         (*movabs<mode>_2): Ditto for operand 1.
17486 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
17488         PR target/54663
17489         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
17490         of microblaze-c.o
17492 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17494         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
17495         tm_file.
17497 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17499         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
17500         Undef to avoid warning.
17502 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
17504         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
17505         * configure: Regenerate.
17507 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
17509         PR tree-optimization/56098
17510         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
17511         for stmts with volatile ops.
17512         (cond_store_replacement): Don't optimize if assign has volatile ops.
17513         (cond_if_else_store_replacement_1): Don't optimize if either
17514         then_assign or else_assign have volatile ops.
17515         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
17516         volatile ops.
17518 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17520         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
17522 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17524         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
17525         missing ':' in asm example.
17527 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
17529         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
17530         entries into lane and laneq entries.
17531         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
17532         Remove AdvSIMD scalar modes.
17533         (aarch64_sq<r>dmulh_laneq<mode>): New.
17534         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
17535         modes.
17536         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
17537         builtin implementations to relfect changes in RTL in aarch64-simd.md.
17538         * config/aarch64/iterators.md (VCOND): New.
17539         (VCONQ): New.
17541 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17543         PR target/54222
17544         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
17545         Add NULL LIBNAME argument to existing definitions.
17546         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
17547         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
17548         * config/avr/avr.c (DEF_BUILTIN): Same.
17549         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
17550         (avr_expand_builtin): Expand to a vanilla call if a libgcc
17551         implementation is available (DECL_ASSEMBLER_NAME is set).
17552         (avr_fold_absfx): New static function.
17553         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
17554         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
17555         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
17556         AVR_BUILTIN_ABSLLK.
17557         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
17558         (abshk, absk, abslk, absllk): Provide as static inline functions.
17560 2013-01-25  Marek Polacek  <polacek@redhat.com>
17562         PR tree-optimization/56035
17563         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
17565 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
17567         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
17568         (*movtf_internal_rex64): Add (!o,C) alternative
17569         (*movxf_internal_rex64): Ditto.
17570         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
17572 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
17574         * doc/invoke.texi: fix typo.
17575         * doc/objc.texi: fix typo.
17577 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
17579         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
17580         for the first two alternatives.
17582 2013-01-24  Diego Novillo  <dnovillo@google.com>
17584         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
17585         (ggc-zone.o): Remove.
17586         * configure.ac: Remove option --with-gc.
17587         * configure: Re-generate.
17588         * doc/install.texi: Remove documentation for --with-gc.
17589         * gengtype.c (write_enum_defn): Remove.  Update all users.
17590         (write_Types_process_field): Remove generation of gt_e_* argument.
17591         (output_type_enum): Remove.  Update all users.
17592         (write_enum_defn): Remove.  Update all users.
17593         (enum alloc_zone): Remove.  Update all users.
17594         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
17595         * ggc-common.c (ggc_splay_alloc): Remove first argument.
17596         Update all callers.
17597         (struct ptr_data): Remove field TYPE.  Update all users.
17598         (gt_pch_note_object): Remove argument TYPE.  Update all users.
17599         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
17600         Update all users.
17601         * ggc-none.c (ggc_alloc_typed_stat): Remove.
17602         (struct alloc_zone): Remove.
17603         (ggc_internal_alloc_zone_stat): Remove.
17604         (ggc_internal_cleared_alloc_zone_stat): Remove.
17605         * ggc-page.c (ggc_alloc_typed_stat): Remove.
17606         (ggc_pch_count_object): Remove last argument.  Update all users.
17607         (ggc_pch_alloc_object): Remove last argument.  Update all users.
17608         (struct alloc_zone): Remove.
17609         * ggc-zone.c: Remove.
17610         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
17611         (struct alloc_zone): Remove.
17612         (ggc_alloc_typed_stat): Remove.
17613         (ggc_alloc_typed): Remove.
17614         (ggc_splay_alloc): Remove first argument.
17615         (rtl_zone): Remove.  Update all users.
17616         (tree_zone): Remove.  Update all users.
17617         (tree_id_zone): Remove.  Update all users.
17618         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
17619         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
17620         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
17621         * tree-ssanames.c: Remove references to zone allocator in comments.
17623 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
17625         * config/avr/avr.c (avr_out_fract): Make register numbers that
17626         might be outside of source operand signed.
17628 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
17630         * config/i386/constraints.md (Yf): New constraint.
17631         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
17632         of f constraint to conditionaly disable x87 register preferences.
17633         (*movdf_internal): Ditto.
17634         (*movsf_internal): Ditto.
17636 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
17638         PR inline-asm/55934
17639         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
17640         that have operands with impossible constraints.
17641         Add a FIXME for a speed-up opportunity.
17642         * lra-constraints.c (process_alt_operands): Verify that a class
17643         selected from constraints on asms is valid for the operand mode.
17644         (curr_insn_transform): Remove incorrect comment.
17646 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
17648         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
17649         TOC operand is a valid symbol ref in the constant pool.
17651 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17653         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
17655 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
17657         PR target/54222
17658         * config/avr/stdfix.h: New file.
17659         * t-avr (stdfix-gcc.h): New rule to build it.
17660         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
17662 2013-01-23  Kostya Serebryany  <kcc@google.com>
17664         * config/darwin.h: remove dependency on
17665         CoreFoundation (asan on Mac OS).
17667 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
17669         PR target/49069
17670         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
17671         instead of cmpdi_operand for first comparison operand.
17672         Don't assert that comparison operands aren't both constants.
17674 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17676         * doc/install.texi (Downloading the Source): Update references to
17677         downloading separate components.
17679 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
17681         * doc/extend.texi (__int128): Improve grammar.
17683 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
17685         PR target/56028
17686         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
17687         alternative to (o,r).
17688         (*movdi_internal_rex64): Remove (!o,n) alternative.
17689         (DImode immediate->memory splitter): Remove.
17690         (DImode immediate->memory peephole2): Remove.
17691         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
17692         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
17693         alternative to (!o,*r).
17694         (*movtf_internal_sse): New pattern.
17695         (*movxf_internal_rex64): New pattern.
17696         (*movxf_internal): Disable for TARGET_64BIT.
17697         (*movdf_internal_rex64): Remove (!o,F) alternative.
17699 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
17701         PR middle-end/56074
17702         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
17703         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
17704         * tree-vect-loop-manip.c (find_loop_location): Also ignore
17705         stmt locations where LOCATION_LOCUS of the stmt location is
17706         UNKNOWN_LOCATION or BUILTINS_LOCATION.
17708         PR target/55686
17709         * config/i386/i386.md (UNSPEC_STOS): New.
17710         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
17711         *strsetqi_1): Add UNSPEC_STOS.
17713 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
17715         PR c++/56067
17716         * doc/invoke.texi: Remove left over -Wsynth example.
17718 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17720         PR tree-optimization/56051
17721         * fold-const.c (fold_binary_loc): Don't fold
17722         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
17723         a narrowing conversion, or widening conversion from signed
17724         to unsigned.
17726 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17728         PR rtl-optimization/56023
17729         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
17730         dependent on debug instruction.
17732 2013-01-21  Martin Jambor  <mjambor@suse.cz>
17734         PR middle-end/56022
17735         * function.c (allocate_struct_function): Call
17736         invoke_set_current_function_hook earlier.
17738 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
17740         * reload1.c (init_reload): Only initialize reload_obstack
17741         during the first call.
17743 2013-01-21  Marek Polacek  <polacek@redhat.com>
17745         * cfgloop.c (verify_loop_structure): Fix up grammar.
17747 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
17749         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
17750         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
17752 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17754         PR target/56058
17755         * config/arm/marvell-pj4.md: Update copyright year.
17756         Fix up use of alu to alu_reg and simple_alu_imm.
17758 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
17760         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
17762 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
17764         PR target/55433
17765         * lra-constraints.c (curr_insn_transform): Don't reuse original
17766         insn for secondary memory move when memory mode should be different.
17768 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17770         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
17771         atomic_storedi_1): New patterns.
17773 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
17775         btver2 pipeline descriptions.
17776         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
17777         descriptions.
17778         * config/i386/i386.md (btver2_decode): New type attributes.
17779         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
17780         type attributes.
17781         * config/i386/btver2.md: New file describing btver2 pipelines.
17783 2013-01-19  Andrew Pinski  <apinski@cavium.com>
17785         PR tree-optimization/52631
17786         * tree-ssa-sccvn (visit_use): Before looking up the original
17787         statement, try looking up the simplified expression.
17789 2013-01-19  Anthony Green  <green@moxielogic.com>
17791         * config/moxie/moxie.c (moxie_expand_prologue): Set
17792         current_function_static_stack_size.
17794 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17796         PR tree-optimization/56029
17797         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
17798         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
17800 2013-01-18  Sharad Singhai  <singhai@google.com>
17802         PR tree-optimization/55995
17803         * dumpfile.c (dump_loc): Print location only if available.
17804         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
17806 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
17808         PR target/55433
17809         * lra-constraints.c (curr_insn_transform): Reuse original insn for
17810         secondary memory move.
17811         (inherit_reload_reg): Use rclass instead of cl for
17812         check_secondary_memory_needed_p.
17814 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
17816         PR middle-end/56015
17817         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
17818         the case where writing real complex part of target modifies op1.
17820 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
17822         * config/aarch64/aarch64-simd.md
17823         (aarch64_vcond_internal<mode>): Handle unordered cases.
17824         * config/aarch64/iterators.md (v_cmp_result): New.
17826 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
17827             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17829         * config/arm/marvell-pj4.md: New file.
17830         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
17831         * config/arm/arm.md (generic_sched): Add marvell_pj4.
17832         (generic_vfp): Likewise.
17833         * config/arm/arm-cores.def: Add marvell-pj4.
17834         * config/arm/arm-tune.md: Regenerate.
17835         * config/arm/arm-tables.opt: Regenerate.
17836         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
17837         * doc/invoke.texi: Document marvell-pj4.
17839 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
17841         * config/aarch64/arm_neon.h: Map scalar types to standard types.
17843 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17845         PR debug/54114
17846         PR debug/54402
17847         PR debug/49888
17848         * var-tracking.c (negative_power_of_two_p): New.
17849         (global_get_addr_cache, local_get_addr_cache): New.
17850         (get_addr_from_global_cache, get_addr_from_local_cache): New.
17851         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
17852         heading comment.
17853         (vt_stack_offset_p): Remove.
17854         (vt_canon_true_dep): Always canonicalize loc's address.
17855         (clobber_overlapping_mems): Make sure we have a MEM.
17856         (local_get_addr_clear_given_value): New.
17857         (val_reset): Clear local cached entries.
17858         (compute_bb_dataflow): Create and release the local cache.
17859         Disable duplicate MEMs clobbering.
17860         (emit_notes_in_bb): Clobber MEMs likewise.
17861         (vt_emit_notes): Create and release the local cache.
17862         (vt_initialize, vt_finalize): Create and release the global
17863         cache, respectively.
17864         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
17866 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17868         PR libmudflap/53359
17869         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
17870         not found in the symtab.
17872 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
17874         PR debug/56006
17875         PR rtl-optimization/55547
17876         PR rtl-optimization/53827
17877         PR debug/53671
17878         PR debug/49888
17879         * alias.c (offset_overlap_p): New, factored out of...
17880         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
17881         the conservative special case for symbolic constants.  Don't
17882         adjust zero sizes on alignment.
17884 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
17886         PR rtl-optimization/52573
17887         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
17888         REG_UNUSED for the same register.
17890 2013-01-17  Richard Biener  <rguenther@suse.de>
17891             Marek Polacek  <polacek@redhat.com>
17893         PR rtl-optimization/55833
17894         * loop-unswitch.c (unswitch_loops): Move loop verification...
17895         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
17896         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
17897         Set it to true when we're removing a loop from hierarchy tree in
17898         an irreducible region.
17899         (fix_bb_placements): Adjust caller.
17900         (fix_loop_placements): Likewise.
17902 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
17904         * config/avr/builtins.def (DEF_BUILTIN): Factor out
17905         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
17906         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
17907         Remove ID.  Adjust comments.
17908         * config/avr/avr-c.c (avr_builtin_name): Remove.
17909         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
17910         * config/avr/avr.c (avr_tolower): New static function.
17911         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
17912         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
17913         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
17914         default expansion.
17916 2013-01-17  Jan Hubicka  <jh@suse.cz>
17918         PR tree-optimization/55273
17919         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
17921 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
17923         PR target/55981
17924         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
17925         store through atomic_store<mode>_1.
17926         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
17928 2013-01-17  Martin Jambor  <mjambor@suse.cz>
17930         PR tree-optimizations/55264
17931         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
17932         for virtual methods.
17933         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
17934         virtual methods before inlining is over.
17935         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
17936         virtual functions.
17937         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
17938         non-virtual.
17940 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
17942         PR rtl-optimization/56005
17943         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
17944         pending reads for prefetch.
17946 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
17948         * config/aarch64/aarch64.md
17949         (*cstoresi_neg_uxtw): New pattern.
17950         (*cmovsi_insn_uxtw): New pattern.
17951         (*<optab>si3_uxtw): New pattern.
17952         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
17953         (*<optab>si3_insn_uxtw): New pattern.
17954         (*bswapsi2_uxtw): New pattern.
17956 2013-01-16  Richard Biener  <rguenther@suse.de>
17958         * tree-inline.c (tree_function_versioning): Remove set but
17959         never used variable.
17961 2013-01-16  Richard Biener  <rguenther@suse.de>
17963         PR tree-optimization/55964
17964         * tree-flow.h (rename_variables_in_loop): Remove.
17965         (rename_variables_in_bb): Likewise.
17966         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
17967         (copy_loop_before): Adjust and delete update-ssa status.
17968         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
17969         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
17970         (rename_variables_in_loop): Remove.
17971         (slpeel_update_phis_for_duplicate_loop): Likewise.
17972         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
17973         use available cfg machinery instead of duplicating it.
17974         Update PHI nodes and perform poor-mans SSA update here.
17975         (slpeel_tree_peel_loop_to_edge): Adjust.
17977 2013-01-16  Richard Biener  <rguenther@suse.de>
17979         PR tree-optimization/54767
17980         PR tree-optimization/53465
17981         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17982         (vrp_visit_phi_node): For PHI arguments coming via backedges
17983         drop all symbolical range information.
17984         (execute_vrp): Compute backedges.
17986 2013-01-16  Richard Biener  <rguenther@suse.de>
17988         * doc/install.texi: Update CLooG and ISL requirements to
17989         0.18.0 and 0.11.1.
17991 2013-01-16  Christian Bruel  <christian.bruel@st.com>
17993         PR target/55301
17994         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17995         (broken_move): Handle UNSPECV_SP_SWITCH_B.
17996         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17998 2013-01-16  DJ Delorie  <dj@redhat.com>
18000         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
18001         (UNSPECV_SP_SWITCH_E): New.
18002         (sp_switch_1): Change to an unspec.
18003         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
18004         replace $r15.
18006 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
18008         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
18009         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
18010         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
18011         (expand_mem_thread_fence): Ditto.
18012         (expand_mem_signal_fence): Ditto.
18013         (expand_atomic_load): Ditto.
18014         (expand_atomic_store): Ditto.
18016 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
18018         PR rtl-optimization/55547
18019         PR rtl-optimization/53827
18020         PR debug/53671
18021         PR debug/49888
18022         * alias.c (memrefs_conflict_p): Set sizes to negative after
18023         AND adjustments.
18025 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18027         PR target/55940
18028         * function.c (thread_prologue_and_epilogue_insns): Always
18029         add crtl->drap_reg to set_up_by_prologue.set, even if
18030         stack_realign_drap is false.
18032 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
18034         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
18035         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
18036         *call): Fix indention.
18038 2013-01-15  Tom de Vries  <tom@codesourcery.com>
18040         PR target/55876
18041         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
18042         Update comment.
18044 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
18046         PR rtl-optimization/55153
18047         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
18049 2013-01-15  Martin Jambor  <mjambor@suse.cz>
18051         PR tree-optimization/55920
18052         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18053         accesses as grp_to_be_debug_replaced.
18055 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18057         PR tree-optimization/55920
18058         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18059         there is non-useless type conversion needed from debug rhs to lhs,
18060         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18062 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
18063             Mikael Pettersson  <mikpe@it.uu.se>
18065         PR target/43961
18066         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18067         Thumb.
18068         (ASM_OUTPUT_CASE_LABEL): Remove.
18069         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18070         * final.c (shorten_branches): Update alignment of labels before
18071         jump tables if CASE_VECTOR_SHORTEN_MODE.
18073 2013-01-15  Richard Biener  <rguenther@suse.de>
18075         PR bootstrap/55961
18076         * system.h: Do not include gmp.h for building host tools.
18078 2013-01-15  Richard Biener  <rguenther@suse.de>
18080         PR middle-end/55882
18081         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18082         account for bitpos when computing alignment.
18084 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
18086         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18087         (ix86_target_macros_internal): Likewise.
18089         * config/i386/i386.c (m_CORE2I7): Removed.
18090         (m_CORE_HASWELL): New macro.
18091         (m_CORE_ALL): Likewise.
18092         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18093         (initial_ix86_arch_features): Likewise.
18094         (processor_target_table): Initializations for Core avx2.
18095         (cpu_names): New names "core-avx2".
18096         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18097         PROCESSOR_CORE_HASWELL.
18098         (ix86_issue_rate): New case.
18099         (ia32_multipass_dfa_lookahead): Likewise.
18100         (ix86_sched_init_global): Likewise.
18102         * config/i386/i386.h (TARGET_HASWELL): New macro.
18103         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18104         (processor_type): New PROCESSOR_HASWELL.
18106 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18108         PR tree-optimization/55955
18109         * tree-vect-loop.c (vectorizable_reduction): Give up early on
18110         *SHIFT_EXPR and *ROTATE_EXPR codes.
18112         PR tree-optimization/48766
18113         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18114         -ftrapv disable -fwrapv.
18116 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18118         PR target/55974
18119         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18120         etc. to 1 and not to __flash.
18121         Use LL suffix for __INT24_MAX__ with -mint8.
18122         Use ULL suffix for __UINT24_MAX__ with -mint8.
18124 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18126         * config/avr/avr-arch.h
18127         (struct base_arch_s): Use typedef avr_arch_t instead.
18128         (struct arch_info_s): Use typedef avr_arch_info_t instead.
18129         (struct mcu_type_s): Use typedef avr_mcu_t instead.
18130         * config/avr/avr.c: Same.
18131         * config/avr/avr-devices.c: Same.
18132         * config/avr/driver-avr.c: Same.
18133         * config/avr/gen-avr-mmcu-texi.c: Same.
18134         * config/avr/avr-mcus.def: Adjust comment.
18136 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
18138         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18139         * config/aarch64/iterators.md (VALLDI): New.
18141 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
18142             Andi Kleen  <ak@linux.intel.com>
18144         PR target/55948
18145         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18146         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18147         memmodel flag.
18149 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18151         * config/avr/avr-stdint.h: Remove trailing blanks.
18152         * config/avr/avr-log.h: Same.
18153         * config/avr/avr-arch.h: Same.
18154         * config/avr/avr-devices.c: Same.
18155         * config/avr/avr-dimode.md: Same.
18156         * config/avr/predicates.md: Same.
18157         * config/avr/avr-c.c: Same.  And fix typo.
18159         * config/avr/avr-protos.h: Same.  And:
18160         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18161         (init_cumulative_args): Rename to avr_init_cumulative_args.
18162         (expand_prologue): Rename to avr_expand_prologue.
18163         (expand_epilogue): Rename to avr_expand_epilogue.
18164         (adjust_insn_length): Rename to avr_adjust_insn_length.
18165         (notice_update_cc): Rename to avr_notice_update_cc.
18166         (final_prescan_insn): Rename to avr_final_prescan_insn.
18167         * config/avr/avr.c: Same.
18168         * config/avr/avr.h: Same.
18169         * config/avr/avr.md: Remove trailing blanks.
18170         (prologue): Use avr_expand_prologue.
18171         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18173 2013-01-14  Richard Biener  <rguenther@suse.de>
18175         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18176         verify_location, collect_subblocks): New functions.
18177         (verify_gimple_in_cfg): Verify that locations only reference
18178         BLOCKs in the functions BLOCK tree.
18180 2013-01-14  Richard Biener  <rguenther@suse.de>
18182         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18183         PHI argument.
18184         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18185         unshare reference.
18186         (insert_out_of_ssa_copy_on_edge): Likewise.
18187         (rewrite_close_phi_out_of_ssa): Likewise.
18188         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18189         debug expressions.
18190         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18191         propagated constants.
18192         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18193         can not be shared.
18195 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18197         * config/avr/avr-modes.def: Add GPL copyright notice.
18199 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
18201         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18202         MEMMODEL_MASK to determine memory model.
18203         (atomic_store<mode>): Ditto from operands[2].
18204         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18206 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
18208         PR fortran/55935
18209         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18210         (fold_gimple_assign): Don't call unshare_expr here.
18211         (fold_ctor_reference): Call unshare_expr.
18213 2013-01-13  Terry Guo  <terry.guo@arm.com>
18215         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18216         * doc/fragments.texi: Document MULTILIB_REUSE.
18217         * gcc.c (multilib_reuse): New internal spec.
18218         (set_multilib_dir): Also search multilib from multilib_reuse.
18219         * genmultilib (tmpmultilib3): Refactor code.
18220         (tmpmultilib4): Ditto.
18221         (multilib_reuse): New multilib argument.
18223 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
18225         * Makefile.in: Update copyright.
18227 2013-01-12  Tom de Vries  <tom@codesourcery.com>
18229         PR middle-end/55890
18230         * calls.c (expand_call): Check if arg_nr is valid.
18232 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
18234         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18235         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18236         documentation.  Add missing '__' in front of
18237         __builtin_ia32_packssdw256.
18239 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18241         PR target/55719
18242         * config/s390/s390.c (s390_preferred_reload_class): Do not return
18243         NO_REGS for larl operands.
18244         (s390_reload_larl_operand): Use s390_load_address instead of
18245         emit_move_insn.
18247 2013-01-11  Richard Biener  <rguenther@suse.de>
18249         * tree-cfg.c (verify_node_sharing_1): Split out from ...
18250         (verify_node_sharing): ... here.
18251         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
18253 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
18255         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
18256         Substitute TREECHECKING.
18257         * configure: Regenerate.
18258         * Makefile.in (TREECHECKING): New.
18260 2013-01-11  Richard Guenther  <rguenther@suse.de>
18262         PR tree-optimization/44061
18263         * tree-vrp.c (extract_range_basic): Compute zero as
18264         value-range for __builtin_constant_p of function parameters.
18266 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
18268         Update copyright years.
18270 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
18272         PR rtl-optimization/55672
18273         * lra-eliminations.c (mark_not_eliminable): Permit addition with
18274         const to be eliminable.
18276 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
18278         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
18279         * configure: Regenerate.
18281 2013-01-10  Richard Biener  <rguenther@suse.de>
18283         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
18285 2013-01-10  Richard Biener  <rguenther@suse.de>
18287         PR bootstrap/55792
18288         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
18289         locations for virtual PHI arguments.
18290         (rewrite_update_phi_arguments): Likewise.
18292 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
18294         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
18295         on to assembler.
18297 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
18299         PR tree-optimization/55921
18300         * tree-complex.c (expand_complex_asm): New function.
18301         (expand_complex_operations_1): Call it for GIMPLE_ASM.
18303 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18305         PR target/55718
18306         * config/s390/s390.c (s390_symref_operand_p)
18307         (s390_loadrelative_operand_p): Merge the two functions.
18308         (s390_check_qrst_address, print_operand_address): Add parameters
18309         to s390_loadrelative_operand_p invokation.
18310         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
18311         (s390_reload_larl_operand, s390_secondary_reload): Use
18312         s390_loadrelative_operand_p instead of s390_symref_operand_p.
18313         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
18315 2013-01-09  Mike Stump  <mikestump@comcast.net>
18317         * dse.c (record_store): Remove unnecessary assert.
18319 2013-01-09  Jan Hubicka  <jh@suse.cz>
18321         PR tree-optimization/55569
18322         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
18323         * cfgloop.h (scale_loop_profile): Likewise.
18325 2013-01-09  Jan Hubicka  <jh@suse.cz>
18327         PR lto/45375
18328         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
18329         functions.
18330         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
18332 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
18334         PR middle-end/55114
18335         * expr.h (maybe_emit_group_store): Declare.
18336         * expr.c (maybe_emit_group_store): New function.
18337         * builtins.c (expand_builtin_int_roundingfn): Call it.
18338         (expand_builtin_int_roundingfn_2): Likewise.
18340 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
18342         PR rtl-optimization/55829
18343         * lra-constraints.c (match_reload): Add code for absent output.
18344         (curr_insn_transform): Add code for reloads of matched inputs
18345         without output.
18347 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
18349         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
18350         attribute of movddup insn to DF.
18351         (*vec_interleave_lowv2df): Ditto.
18352         (vec_dupv2df): Ditto.
18354 2013-01-09  Jan Hubicka  <jh@suse.cz>
18356         PR tree-optimiation/55875
18357         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
18358         EVERY_ITERATION parameter.
18359         (number_of_iterations_exit): Check if exit is executed every iteration.
18360         (idx_infer_loop_bounds): Similarly here.
18361         (n_of_executions_at_most): Simplify
18362         to only test for cases where statement is dominated by the
18363         particular bound; handle correctly the "postdominance" test.
18364         (scev_probably_wraps_p): Use max loop iterations info
18365         as a global bound first.
18367 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
18368             Nick Clifton  <nickc@redhat.com>
18370         * config/v850/v850.md (cbranchsf4): New pattern.
18371         (cstoresf4): New pattern.
18372         (cbranchdf4): New pattern.
18373         (cstoredf4): New pattern.
18374         (movsicc): Disallow floating point comparisons.
18375         (cmpsf_le_insn): Fix order of operators.
18376         (cmpsf_lt_insn): Likewise.
18377         (cmpsf_eq_insn): Likewise.
18378         (cmpdf_le_insn): Likewise.
18379         (cmpdf_lt_insn): Likewise.
18380         (cmpdf_eq_insn): Likewise.
18381         (cmpsf_ge_insn): Use LE comparison.
18382         (cmpdf_ge_insn): Likewise.
18383         (cmpsf_gt_insn): Use LT comparison.
18384         (cmpdf_gt_insn): Likewise.
18385         (cmpsf_ne_insn): Delete pattern.
18386         (cmpdf_ne_insn): Delete pattern.
18387         * config/v850/v850.c (v850_gen_float_compare): Use
18388         gen_cmpdf_eq_insn for NE comparison.
18389         (v850_float_z_comparison_operator)
18390         (v850_float_nz_comparison_operator): Move from here ...
18391         * config/v850/predicates.md: ... to here.  Move GT and GE
18392         comparisons into v850_float_z_comparison_operator.
18393         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
18394         Delete prototype.
18395         (v850_float_nz_comparison_operator): Likewise.
18397 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18399         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
18400         with calls to gen_insvsi/gen_insvdi.
18402 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
18404         * config/i386/i386.c (initial_ix86_tune_features): Set up
18405         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
18407 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
18408             Jakub Jelinek  <jakub@redhat.com>
18410         PR tree-optimization/48189
18411         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
18412         If nitercst is 0, don't predict the exit edge.
18414 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18416         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
18417         in asm_fprintf with reg_names.
18418         (aarch64_print_operand_address): Likewise.
18419         (aarch64_return_addr): Likewise.
18420         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
18422 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18424         * config/pa/pa.h (VAL_U6_BITS_P): Define.
18425         (INT_U6_BITS): Likewise.
18426         * config/pa/predicates.md (uint6_operand): New predicate.
18427         (shift5_operand, shift6_operand): Likewise.
18428         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
18429         arith32_operand.
18430         (lshrdi3): Use shift6_operand.
18431         (shrpsi4, shrpdi4): New insn patterns.
18432         (extzv): Delete expander.
18433         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
18434         predicates in unamed zero extract patterns.  Tighten common constraint.
18435         (extv): Delete expander.
18436         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
18437         predicates in unamed sign extract patterns.  Tighten common constraint.
18438         (insv): Delete expander.
18439         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
18440         predicates in unamed insert patterns.  Tighten common constraint.
18441         Change uint32_operand predicate to uint6_operand predicate in unamed
18442         DImode pattern to insert constant values of type 1...1xxxx.
18444 2013-01-04  Jan Hubicka  <jh@suse.cz>
18446         PR tree-optimization/55823
18447         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
18448         issue.
18450 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18451             Uros Bizjak  <ubizjak@gmail.com>
18453         PR rtl-optimization/55845
18454         * df-problems.c (can_move_insns_across): Stop scanning at
18455         volatile_insn_p source instruction or give up if
18456         across_from .. across_to range contains any volatile_insn_p
18457         instructions.
18459 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
18461         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
18462         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
18463         Declare.
18464         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
18465         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
18467 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18469         PR fortran/55341
18470         * asan.c (asan_clear_shadow): New function.
18471         (asan_emit_stack_protection): Use it.
18473 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
18475         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
18476         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
18477         with tab instead of space.
18479 2013-01-08  Nick Clifton  <nickc@redhat.com>
18481         * config/rl78/rl78.c (rl78_expand_prologue): Always select
18482         register bank 0 at the start of an interrupt handler.
18483         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
18484         MDBH registers.
18486 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
18488         * config/aarch64/aarch64-simd.md
18489         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
18490         (aarch64_simd_bsl): Likewise.
18491         (aarch64_vcond_internal<mode>): Likewise.
18492         (vcond<mode><mode>): Likewise.
18493         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
18494         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
18496 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
18498         * config/aarch64/aarch64-builtins.c
18499         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
18501 2013-01-08  Martin Jambor  <mjambor@suse.cz>
18503         PR debug/55579
18504         * tree-sra.c (analyze_access_subtree): Return true also after
18505         potentially creating a debug-only replacement.
18507 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18509         PR middle-end/55890
18510         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
18512         PR tree-optimization/54120
18513         * tree-vrp.c (range_fits_type_p): Don't allow
18514         src_precision < precision from signed vr to unsigned_p
18515         if vr->min or vr->max is negative.
18516         (simplify_float_conversion_using_ranges): Test can_float_p
18517         against CODE_FOR_nothing.
18519 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18520             Richard Biener  <rguenther@suse.de>
18522         PR middle-end/55851
18523         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
18524         types instead of just INTEGER_TYPE types.
18526 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
18528         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
18529         TF_SIZE): Define.
18531 2013-01-07  Steve Ellcey  <sellcey@mips.com>
18533         PR target/42661
18534         * config/mips/mips.opt: Change mad to mmad to match documentation.
18536 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18538         PR target/55897
18539         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
18540         .progmemx.data now.
18542 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18544         PR target/55897
18545         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
18546         (avr_addrspace_t): Add .section_name field.
18547         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
18548         array size.
18549         (avr_addrspace): Same.  Initialize .section_name.  Remove last
18550         NULL entry.  Put __memx into .progmemx.data.
18551         (progmem_section_prefix): Remove.
18552         (avr_asm_init_sections): No need to initialize progmem_section.
18553         (avr_asm_named_section): Use avr_addrspace[].section_name to get
18554         section name prefix.
18555         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
18556         retrieve the progmem section.
18557         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
18558         boundary to run over avr_addrspace[].
18559         (avr_register_target_pragmas): Ditto.
18561 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
18563         * varasm.c (output_constant_def_contents): For asan_protect_global
18564         protected strings, adjust DECL_ALIGN if needed, before testing for
18565         anchored symbols.
18566         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
18567         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
18568         normal decls.
18569         (output_object_block): For asan protected decls, emit asan padding
18570         after their contents.
18571         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
18572         (asan_finish_file): Test it here instead.
18574 2013-01-07  Nick Clifton  <nickc@redhat.com>
18575             Matthias Klose  <doko@debian.org>
18576             Doug Kwan  <dougkwan@google.com>
18577             H.J. Lu  <hongjiu.lu@intel.com>
18579         PR driver/55470
18580         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
18582         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
18584         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
18586         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
18588         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
18590 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18592         PR target/54461
18593         * doc/install.texi (Cross-Compiler-Specific Options): Document
18594         --with-avrlibc.
18596 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
18598         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
18599         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
18600         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
18601         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
18602         vqmovun_high_s64): Fix source operand number and update copyright.
18604 2013-01-07  Richard Biener  <rguenther@suse.de>
18606         PR middle-end/55890
18607         * gimple.h (gimple_call_builtin_p): New overload.
18608         * gimple.c (validate_call): New function.
18609         (gimple_call_builtin_p): Likewise.
18610         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18611         Use gimple_call_builtin_p.
18612         (find_func_clobbers): Likewise.
18613         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
18614         (strlen_optimize_stmt): Likewise.
18616 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18618         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
18619         (vld1q_dup_*): Likewise.
18620         (vld1_*): Likewise.
18621         (vld1q_*): Likewise.
18622         (vld1_lane_*): Likewise.
18623         (vld1q_lane_*): Likewise.
18625 2013-01-07  Richard Biener  <rguenther@suse.de>
18627         * lto-streamer.h (LTO_minor_version): Bump to 2.
18629 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18631         * config/aarch64/aarch64-protos.h
18632         (aarch64_const_double_zero_rtx_p): Rename to...
18633         (aarch64_float_const_zero_rtx_p): ...this.
18634         (aarch64_float_const_representable_p): New.
18635         (aarch64_output_simd_mov_immediate): Likewise.
18636         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
18637         move immediate case.
18638         * config/aarch64/aarch64.c
18639         (aarch64_const_double_zero_rtx_p): Rename to...
18640         (aarch64_float_const_zero_rtx_p): ...this.
18641         (aarch64_print_operand): Allow printing of new constants.
18642         (aarch64_valid_floating_const): New.
18643         (aarch64_legitimate_constant_p): Check for valid floating-point
18644         constants.
18645         (aarch64_simd_valid_immediate): Likewise.
18646         (aarch64_vect_float_const_representable_p): New.
18647         (aarch64_float_const_representable_p): Likewise.
18648         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
18649         (aarch64_output_simd_mov_immediate): New.
18650         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
18651         (*movdf_aarch64): Likewise.
18652         * config/aarch64/constraints.md (Ufc): New.
18653         (Y): call aarch64_float_const_zero_rtx.
18654         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
18656 2013-01-07  Richard Biener  <rguenther@suse.de>
18658         PR tree-optimization/55888
18659         PR tree-optimization/55862
18660         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
18661         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
18662         not if it is contained therein.
18664 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18666         * config/avr/t-avr: Typo.
18668 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18670         PR55243
18671         * config/avr/t-avr: Don't automatically rebuild
18672         $(srcdir)/config/avr/t-multilib
18673         $(srcdir)/config/avr/avr-tables.opt
18674         $(srcdir)/doc/avr-mmcu.texi
18675         (avr-mcus): New phony target to build them on request.
18676         (s-avr-mlib, s-avr-mmcu-texi): Remove.
18677         * avr/avr-mcus.def: Adjust comments.
18679 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
18681         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
18683 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18685         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
18687 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
18689         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
18691 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
18693         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
18694         to generate profiling.
18695         * config/rs6000/aix64.h (LIB_SPEC): Same.
18697 2013-01-04  Andrew Pinski  <apinski@cavium.com>
18699         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
18700         New function.
18701         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
18703 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
18705         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
18706         unconditionally.
18707         (ix86_expand_move): Ditto.
18708         (ix86_zero_extend_to_Pmode): Ditto.
18709         (ix86_expand_call): Ditto.
18710         (ix86_expand_special_args_builtin): Ditto.
18711         (ix86_expand_builtin): Ditto.
18713 2013-01-04  Richard Biener  <rguenther@suse.de>
18715         PR tree-optimization/55862
18716         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
18717         translating them through PHI nodes.
18719 2013-01-04  Martin Jambor  <mjambor@suse.cz>
18721         PR tree-optimization/55755
18722         * tree-sra.c (sra_modify_assign): Do not check that an access has no
18723         children when trying to avoid producing a VIEW_CONVERT_EXPR.
18725 2013-01-04  Marek Polacek  <polacek@redhat.com>
18727         PR middle-end/55859
18728         * opts.c (default_options_optimization): Clarify error message.
18730 2013-01-04  Richard Biener  <rguenther@suse.de>
18732         PR middle-end/55863
18733         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
18734         reassociation.
18736 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18738         PR target/53789
18739         * config/pa/pa.md (movsi): Revert previous change.
18740         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
18741         references.
18743 2013-01-03  Richard Henderson  <rth@redhat.com>
18745         * config/i386/i386.c (ix86_expand_move): Always assign to op1
18746         after eliminating TLS symbols.
18748 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
18750         PR bootstrap/50167
18751         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
18752         * graphite-poly.c (debug_gmp_value): Likewise.
18754 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
18756         PR target/55712
18757         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
18758         selected code model, define __code_mode_small__, __code_model_medium__,
18759         __code_model_large__, __code_model_32__ or __code_model_kernel__.
18760         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
18761         xchg temporary register with %k.  Declare temporary register as
18762         early clobbered.
18763         [__x86_64__]: For medium and large code models, preserve %rbx register.
18765 2013-01-03  Richard Biener  <rguenther@suse.de>
18767         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
18768         (dump_subscript): Adjust.
18769         (finalize_ddr_dependent): Do not dump redundant info.
18770         (analyze_siv_subscript): Adjust.
18771         (subscript_dependence_tester): Likewise.
18772         (compute_affine_dependence): Likewise.
18774 2013-01-03  Richard Biener  <rguenther@suse.de>
18776         Revert
18777         2013-01-03  Richard Biener  <rguenther@suse.de>
18779         PR tree-optimization/55857
18780         * tree-vect-stmts.c (vectorizable_load): Do not setup
18781         re-alignment for invariant loads.
18783         2013-01-02  Richard Biener  <rguenther@suse.de>
18785         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18786         invariant load do not generate a vector load from the scalar location.
18788 2013-01-03  Richard Biener  <rguenther@suse.de>
18790         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
18791         for not vectorizing.
18792         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
18793         not build INDIRECT_REFs, call get_name once only.
18794         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
18795         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
18797 2013-01-03  Richard Biener  <rguenther@suse.de>
18799         PR tree-optimization/55857
18800         * tree-vect-stmts.c (vectorizable_load): Do not setup
18801         re-alignment for invariant loads.
18803 2013-01-03  Richard Biener  <rguenther@suse.de>
18805         PR lto/55848
18806         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
18807         prefer a built-in decl.
18809 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18811         * gcc.c (process_command): Update copyright notice dates.
18812         * gcov.c (print_version): Likewise.
18813         * gcov-dump.c (print_version): Likewise.
18815         PR rtl-optimization/55838
18816         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
18817         iv0.step, iv1.step and step.
18819 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18820             Marc Glisse  <marc.glisse@inria.fr>
18822         PR tree-optimization/55832
18823         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
18824         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
18825         integer_{one,zero}_node.
18827 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
18829         PR debug/54402
18830         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
18831         * var-tracking.c (reverse_op): Don't add reverse ops to
18832         VALUEs that have already
18833         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
18835 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
18837         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
18839 2013-01-02  Teresa Johnson  <tejohnson@google.com>
18841         * dumpfile.c (dump_loc): Print filename with location.
18842         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
18843         new location_t parameter to emit complete unroll message with
18844         new dump framework.
18845         (canonicalize_loop_induction_variables): Compute loops location
18846         and pass to try_unroll_loop_completely.
18847         * loop-unroll.c (report_unroll_peel): New function.
18848         (peel_loops_completely): Use new dump format with location
18849         for main dumpfile message, and invoke report_unroll_peel on success.
18850         (decide_unrolling_and_peeling): Ditto.
18851         (decide_peel_once_rolling): Remove old dumpfile message subsumed
18852         by report_unroll_peel.
18853         (decide_peel_completely): Ditto.
18854         (decide_unroll_constant_iterations): Ditto.
18855         (decide_unroll_runtime_iterations): Ditto.
18856         (decide_peel_simple): Ditto.
18857         (decide_unroll_stupid): Ditto.
18858         * cfgloop.c (get_loop_location): New function.
18859         * cfgloop.h (get_loop_location): Declare.
18861 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18863         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
18864         NULL.
18866 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18868         PR middle-end/55198
18869         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
18870         BLKmode objects when EXPAND_MEMORY is specified.
18872 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
18874         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
18875         in loop predicate.
18876         (fold_builtin_cpu): Do not share cpu model decls across statements.
18878 2013-01-02  Jason Merrill  <jason@redhat.com>
18880         PR c++/55804
18881         * tree.c (build_array_type_1): Revert earlier change.
18883 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
18885         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
18886         "cortex-a57".
18887         * config/aarch64/aarch64-tune.md: Re-generate.
18889 2013-01-02  Richard Biener  <rguenther@suse.de>
18891         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18892         invariant load do not generate a vector load from the scalar location.
18894 2013-01-02  Richard Biener  <rguenther@suse.de>
18896         PR bootstrap/55784
18897         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
18898         * configure: Regenerate.
18900 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18902         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
18903         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
18904         (expand_builtin_int_roundingfn_2): Keep the original target around
18905         for the fallback case.
18907 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
18909         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
18910         to be clear for sign changes.
18912 2013-01-01  Jan Hubicka  <jh@suse.cz>
18914         * ipa-inline-analysis.c: Fix formatting.
18916 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
18918         PR tree-optimization/55831
18919         * tree-vect-loop.c (get_initial_def_for_induction): Use
18920         gsi_after_labels instead of gsi_start_bb.
18922 Copyright (C) 2013 Free Software Foundation, Inc.
18924 Copying and distribution of this file, with or without modification,
18925 are permitted in any medium without royalty provided the copyright
18926 notice and this notice are preserved.