1 2013-08-22 Jan Hubicka <jh@suse.cz>
3 * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
4 * ipa-utils.h (update_type_inheritance_graph): Declare.
5 (possible_polymorphic_call_target_p): Declare.
6 (possible_polymorphic_call_target_p): New.
7 * ipa-devirt.c: Update toplevel comments.
8 (cached_polymorphic_call_targets): Move up.
9 (odr_type_d): Move ID down.
10 (polymorphic_type_binfo_p): Update comment.
11 (odr_hasher::remove): Likewise;
12 (get_odr_type): Set anonymous_namespace.
13 (dump_odr_type): Dump it.
14 (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
15 (maybe_record_node): Record node in cached_polymorphic_call_targets.
16 (record_binfo): Add comment.
17 (free_polymorphic_call_targets_hash): Do not ICE when cache is not built.
18 (devirt_node_removal_hook): Do not iCE when cache is freed.
19 (possible_polymorphic_call_target_p): New predicate.
20 (update_type_inheritance_graph): New function.
22 2013-08-22 Alexander Ivchenko <alexander.ivchenko@intel.com>
23 Maxim Kuznetsov <maxim.kuznetsov@intel.com>
24 Sergey Lega <sergey.s.lega@intel.com>
25 Anna Tikhonova <anna.tikhonova@intel.com>
26 Ilya Tocar <ilya.tocar@intel.com>
27 Andrey Turetskiy <andrey.turetskiy@intel.com>
28 Ilya Verbin <ilya.verbin@intel.com>
29 Kirill Yukhin <kirill.yukhin@intel.com>
30 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
32 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
33 (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
34 (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
35 (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
36 (OPTION_MASK_ISA_AVX2_UNSET): Update.
37 (OPTION_MASK_ISA_AVX512F_UNSET): New.
38 (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
39 (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
40 (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
41 (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
42 OPT_mavx512pf, OPT_mavx512er cases.
43 * config/i386/constraints.md (v): New constraint.
44 (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
45 * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
47 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
48 AVX512F, AVX512ER, AVX512PF, AVX512CD features.
49 * config/i386/i386-c.c (ix86_target_macros_internal):
50 Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
52 * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
53 (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
54 * config/i386/i386.c (regclass_map, dbx_register_map)
55 (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
56 (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
57 (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
59 (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
60 PTA_AVX512PF, PTA_AVX512CD. Handle -mavx512f, -mavx512er, -mavx512cd,
61 -mavx512pf options. Fix formatting.
62 (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
63 targets. Squash EVEX_SSE_REGS if AVX512F is disabled.
64 (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
65 -mavx512cd, -mavx512pf options.
66 (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
67 (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
68 (ix86_preferred_output_reload_class): Replace SSE_REGS with
70 (ix86_hard_regno_mode_ok): Support 512-bit registers.
71 (ix86_set_reg_reg_cost): Ditto.
72 (x86_order_regs_for_local_alloc): Ditto.
73 (MAX_VECT_LEN): Extend to 64-byte.
74 (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
75 * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
76 (TARGET_AVX512ER, TARGET_AVX512CD): New.
77 (BIGGEST_ALIGNMENT): Extend to 512-bits.
78 (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
79 (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
80 (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
81 (SSE_REG_MODE_P): Support new modes.
82 (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
83 (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
84 (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
85 (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
86 (REG_CLASS_CONTENTS): Add new registers.
87 (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
88 (EXT_REX_SSE_REGNO_P): New.
89 (HI_REGISTER_NAMES): Add new registers.
90 * config/i386/i386.md: Define constants for new registers.
91 (mode): Add new 512-bit modes.
92 (prefix): Support evex prefix.
93 (isa): Support avx512f, noavx512f, fma_avx512f.
94 (ssemodesuffix): Add new 512-bit modes.
96 (*movxi_internal_avx512f): Ditto.
97 (*movdi_internal): Replace constraint "x" with the new constraint "v".
99 (*movsi_internal): Likewise.
100 (*movdf_internal): Likewise.
101 (*movsf_internal): Likewise.
102 (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
103 (<code><mode>3): Likewise.
104 * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd): New.
105 * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
106 with the new constraint "v".
107 * config/i386/sse.md (*mov<mode>_internal): Support new registers and
109 (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
110 with the new constraint "v".
111 (<sse2>_loaddqu<avxsizesuffix>): Likewise.
112 (<sse2>_storedqu<avxsizesuffix>): Likewise.
113 (*<plusminus_insn><mode>3): Likewise.
114 (<sse>_vm<plusminus_insn><mode>3): Likewise.
115 (*mul<mode>3): Likewise.
116 (<sse>_vmmul<mode>3): Likewise.
117 (<sse>_div<mode>3): Likewise.
118 (<sse>_vmdiv<mode>3): Likewise.
119 (<sse>_sqrt<mode>2): Likewise.
120 (<sse>_vmsqrt<mode>2): Likewise.
121 (*<code><mode>3_finite): Likewise.
122 (*<code><mode>3) <smaxmin>: Likewise.
123 (<sse>_vm<code><mode>3): Likewise.
124 (*<code><mode>3) <any_logic>: Likewise.
125 (*fma_fmadd_<mode>): Likewise.
126 (*fma_fmsub_<mode>): Likewise.
127 (*fma_fnmadd_<mode>): Likewise.
128 (*fma_fnmsub_<mode>): Likewise.
129 (*fma_fmaddsub_<mode>): Likewise.
130 (*fma_fmsubadd_<mode>): Likewise.
131 (*fmai_fmadd_<mode>): Likewise.
132 (*fmai_fmsub_<mode>): Likewise.
133 (*fmai_fnmadd_<mode>): Likewise.
134 (*fmai_fnmsub_<mode>): Likewise.
135 (sse_cvtsi2ss): Likewise.
136 (sse_cvtsi2ssq): Likewise.
137 (sse_cvtss2si): Likewise.
138 (sse_cvtss2si_2): Likewise.
139 (sse_cvtss2siq): Likewise.
140 (sse_cvtss2siq_2): Likewise.
141 (sse_cvttss2si): Likewise.
142 (sse_cvtss2siq_2): Likewise.
143 (float<sseintvecmodelower><mode>2): Likewise.
144 (sse2_cvtsd2si_2): Likewise.
145 (sse2_cvtsd2siq_2): Likewise.
146 (*<plusminus_insn><mode>3): Likewise.
147 (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
148 (*<sse4_1_avx2>_mul<mode>3): Likewise.
149 (ashr<mode>3): Likewise.
150 (<shift_insn><mode>3): Likewise.
151 (avx2_<code><mode>3): Likewise.
152 (*avx2_<code><mode>3): Likewise.
153 (*andnot<mode>3): Likewise.
154 (*<code><mode>3) <any_logic>: Likewise.
155 (abs<mode>2): Likewise.
156 (avx2_permvar<mode>): Likewise.
157 (avx2_perm<mode>_1): Likewise.
158 (*avx_vpermilp<mode>): Likewise.
159 (avx_vpermilvar<mode>3): Likewise.
160 (avx2_ashrv<mode>): Likewise.
161 (avx2_<shift_insn>v<mode>): Likewise.
162 * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
164 * doc/rtl.texi: Document XImode.
166 2013-08-21 Jeff Law <law@redhat.com>
168 * tree-flow.h (register_jump_thread): Pass vector of edges
169 instead of each important edge.
170 * tree-ssa-threadedge.c (thread_across_edge): Build the jump
171 thread path into a vector and pass that to register_jump_thread.
172 * tree-ssa-threadupdate.c (register_jump_thread): Conver the
173 passed in edge vector to the current 3-edge form.
176 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
178 * dce.c (fini_dce): Call df_analyze again just in case
179 delete_unmarked_insns removed anything.
181 2013-08-21 Joern Rennecke <joern.rennecke@embecosm.com>
183 * reload.h (struct reg_equivs): Rename to ..
184 (struct reg_equivs_s): .. this.
186 2013-08-20 Martin Liska <marxin.liska@gmail.com>
188 * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
190 2013-08-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
192 * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest
195 2013-08-21 Jeff Law <law@redhat.com>
197 * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
198 simplify assignments too. If the RHS collapses to a singleton
199 range, then return the value for the range.
201 2013-08-21 Kirill Yukhin <kirill.yukhin@intel.com>
203 * config/i386/sse.md (V16): Rename to...
205 (mov<mode>): Update iterator name.
206 (*mov<mode>_internal): Ditto.
207 (push<mode>1): Ditto.
208 (movmisalign<mode>): Ditto.
210 2013-08-20 Jan Hubicka <jh@suse.cz>
213 * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
214 entry for direct edges.
215 (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
217 2013-08-20 David Malcolm <dmalcolm@redhat.com>
219 Revert my last two changes, r201865 and r201864:
222 2013-08-20 David Malcolm <dmalcolm@redhat.com>
224 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
225 instances can own GC refs.
227 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
228 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
229 (gcc::context::gt_pch_nx): Likewise.
230 (gcc::context::gt_pch_nx): Likewise.
231 * ggc.h (gt_ggc_mx <T>): New.
232 (gt_pch_nx_with_op <T>): New.
233 (gt_pch_nx <T>): New.
234 * passes.c (opt_pass::gt_ggc_mx): New.
235 (opt_pass::gt_pch_nx): New.
236 (opt_pass::gt_pch_nx_with_op): New.
237 (pass_manager::gt_ggc_mx): New.
238 (pass_manager::gt_pch_nx): New.
239 (pass_manager::gt_pch_nx_with_op): New.
240 (pass_manager::operator new): Use
241 ggc_internal_cleared_alloc_stat rather than xcalloc.
242 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
243 (pass_manager::gt_ggc_mx): New.
244 (pass_manager::gt_pch_nx): New.
245 (pass_manager::gt_pch_nx_with_op): New.
246 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
247 (opt_pass::operator new): New.
248 (opt_pass::gt_ggc_mx): New.
249 (opt_pass::gt_pch_nx): New.
250 (opt_pass::gt_pch_nx_with_op): New.
253 2013-08-20 David Malcolm <dmalcolm@redhat.com>
255 * Makefile.in (GTFILES): Add context.h.
256 * context.c (gcc::context::operator new): New.
257 (gcc::context::gt_ggc_mx): New.
258 (gcc::context::gt_pch_nx): New.
259 (gcc::context::gt_pch_nx): New.
260 * context.h (gcc::context): Add GTY((user)) marking.
261 (gcc::context::operator new): New.
262 (gcc::context::gt_ggc_mx): New.
263 (gcc::context::gt_pch_nx): New.
264 (gcc::context::gt_pch_nx): New.
265 (g): Add GTY marking.
266 (gt_ggc_mx (gcc::context *)): New.
267 (gt_pch_nx (gcc::context *)): New.
268 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
270 * gengtype.c (open_base_files) <ifiles>: Add context.h.
272 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
274 * dce.c (fini_dce): Call df_analyze again just in case
275 delete_unmarked_insns removed anything.
277 2013-08-20 Teresa Johnson <tejohnson@google.com>
279 PR rtl-optimizations/57451
280 * final.c (reemit_insn_block_notes): Prevent lexical blocks
281 from crossing split section boundaries.
283 2013-08-20 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
285 * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
286 * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
287 with MULTLIB_DEFAULTS.
289 2013-08-20 Nick Clifton <nickc@redhat.com>
291 * target.def (narrow_volatile_bitfield): Note that the default
292 value is false, not !TARGET_STRICT_ALIGN.
293 * doc/tm.texi: Regenerate.
295 2013-08-20 Pavel Chupin <pavel.v.chupin@intel.com>
297 Fix LIB_SPEC for systems without libpthread.
299 * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
300 * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
302 * config/i386/linux-common.h: Likewise.
303 * config/mips/linux-common.h: Likewise.
305 2013-08-20 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
307 * tree-ssa-ccp.c (get_default_value): Remove redundant condition
310 2013-08-20 David Malcolm <dmalcolm@redhat.com>
312 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
313 instances can own GC refs.
315 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
316 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
317 (gcc::context::gt_pch_nx): Likewise.
318 (gcc::context::gt_pch_nx): Likewise.
319 * ggc.h (gt_ggc_mx <T>): New.
320 (gt_pch_nx_with_op <T>): New.
321 (gt_pch_nx <T>): New.
322 * passes.c (opt_pass::gt_ggc_mx): New.
323 (opt_pass::gt_pch_nx): New.
324 (opt_pass::gt_pch_nx_with_op): New.
325 (pass_manager::gt_ggc_mx): New.
326 (pass_manager::gt_pch_nx): New.
327 (pass_manager::gt_pch_nx_with_op): New.
328 (pass_manager::operator new): Use
329 ggc_internal_cleared_alloc_stat rather than xcalloc.
330 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
331 (pass_manager::gt_ggc_mx): New.
332 (pass_manager::gt_pch_nx): New.
333 (pass_manager::gt_pch_nx_with_op): New.
334 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
335 (opt_pass::operator new): New.
336 (opt_pass::gt_ggc_mx): New.
337 (opt_pass::gt_pch_nx): New.
338 (opt_pass::gt_pch_nx_with_op): New.
340 2013-08-20 David Malcolm <dmalcolm@redhat.com>
342 * Makefile.in (GTFILES): Add context.h.
343 * context.c (gcc::context::operator new): New.
344 (gcc::context::gt_ggc_mx): New.
345 (gcc::context::gt_pch_nx): New.
346 (gcc::context::gt_pch_nx): New.
347 * context.h (gcc::context): Add GTY((user)) marking.
348 (gcc::context::operator new): New.
349 (gcc::context::gt_ggc_mx): New.
350 (gcc::context::gt_pch_nx): New.
351 (gcc::context::gt_pch_nx): New.
352 (g): Add GTY marking.
353 (gt_ggc_mx (gcc::context *)): New.
354 (gt_pch_nx (gcc::context *)): New.
355 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
357 * gengtype.c (open_base_files) <ifiles>: Add context.h.
359 2013-08-20 Alan Modra <amodra@gmail.com>
362 * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
363 (rs6000_emit_epilogue): Likewise.
365 2013-08-19 Dehao Chen (dehao@google.com)
367 * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
369 2013-08-19 Peter Bergner <bergner@vnet.ibm.com>
370 Jakub Jelinek <jakub@redhat.com>
372 * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
373 (BUILT_IN_FABSD64): Likewise.
374 (BUILT_IN_FABSD128): Likewise.
375 * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
376 (fold_builtin_1): Likewise.
377 * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping destination
379 (*abstd2_fpr): Likewise.
380 (*nabstd2_fpr): Likewise.
382 2013-08-19 Richard Sandiford <rdsandiford@googlemail.com>
384 * config/mips/mips.c (mips_adjust_insn_length): Add checks for
387 2013-08-19 Aldy Hernandez <aldyh@redhat.com>
389 * doc/invoke.texi (-fcilkplus): Clarify that implementation is
392 2013-08-19 Alexander Ivchenko <alexander.ivchenko@intel.com>
394 * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
395 * builtins.c (default_libc_has_function): New.
396 (gnu_libc_has_function): Ditto.
397 (no_c99_libc_has_function): Ditto.
398 (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
399 instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
400 (fold_builtin_sincos): Likewise.
401 (fold_builtin_cexp): Likewise.
402 * builtins.def (DEF_C94_BUILTIN): Likewise.
403 (DEF_C99_BUILTIN): Likewise.
404 (DEF_C99_C90RES_BUILTIN): Likewise.
405 (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
406 definitions to using this define.
407 * config/darwin-protos.h (darwin_libc_has_function): New.
408 * config/darwin.c: (darwin_libc_has_function: Ditto.
409 * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
410 TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
411 * config/darwin.h: Ditto.
412 * config/elfos.h: Ditto.
413 * config/freebsd.h: Ditto.
414 * config/i386/cygming.h: Ditto.
415 * config/i386/djgpp.h: Ditto.
416 * config/i386/i386-interix.h: Ditto.
417 * config/microblaze/microblaze.h: Ditto.
418 * config/mmix/mmix.h: Ditto.
419 * config/gnu-user.h: Ditto.
420 * config/ia64/hpux.h: Ditto.
421 * config/pa/pa-hpux.h: Ditto.
422 * config/pdp11/pdp11.h: Ditto.
423 * config/picochip/picochip.h: Ditto.
424 * config/linux.h: Ditto.
425 * config/netbsd.h: Ditto.
426 * config/openbsd.h: Ditto.
427 * config/rs6000/aix43.h: Ditto.
428 * config/rs6000/aix51.h: Ditto.
429 * config/rs6000/aix52.h: Ditto.
430 * config/rs6000/aix53.h: Ditto.
431 * config/rs6000/aix61.h: Ditto.
432 * config/rs6000/darwin.h: Ditto.
433 * config/rs6000/linux.h: Ditto.
434 * config/rs6000/linux64.h: Ditto.
435 * config/s390/tpf.h: Ditto.
436 * config/sol2-10.h: Ditto.
437 * config/sol2.h: Ditto.
438 * config/vms/vms.h: Ditto.
439 * config/vxworks.h: Ditto.
440 * config/linux-android.c (linux_android_libc_has_function):
441 New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
442 * config/linux-protos.h (linux_android_libc_has_function):
444 * config/i386/i386.c (ix86_libc_has_function): New.
445 * config/i386/i386-protos.h
446 (ix86_libc_has_function): New declaration.
447 * config/i386/i386.md
448 ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
449 ("isinf<mode>2): Likewise.
450 * convert.c (convert_to_integer): Using new target hook
451 TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
452 TARGET_C99_FUNCTIONS.
453 * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
454 * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
455 * coretypes.h (function_class): New enum for different
456 classes of functions.
457 * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
458 * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
459 (TARGET_HAS_SINCOS): Likewise.
460 (TARGET_LIBC_HAS_FUNCTION): New.
461 * doc/tm.texi: Regenerated.
462 * targhooks.h (default_libc_has_function): New declaration.
463 (no_c99_libc_has_function): Ditto.
464 (gnu_libc_has_function): Ditto.
465 * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
466 and TARGET_HAS_SINCOS.
468 2013-08-18 Jan Hubicka <jh@suse.cz>
470 * Makeifle-in (ipa-devirt.o): New.
471 (GTFILES): Add ipa-utils.h and ipa-devirt.c
472 * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
473 (analyze_functions): Look into possible targets of polymorphic call.
474 * dumpfile.c (dump_files): Add type-inheritance dump.
475 * dumpfile.h (TDI_inheritance): New.
476 * ipa-devirt.c: New file.
477 * ipa-utils.h (odr_type_d): Forward declare.
478 (odr_type): New type.
479 (build_type_inheritance_graph): Declare.
480 (possible_polymorphic_call_targets): Declare and introduce inline
481 variant when only edge is pased.
482 (dump_possible_polymorphic_call_targets): Likewise.
483 * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
484 * tree.c (type_in_anonymous_namespace_p): Break out from ...
485 (types_same_for_odr): ... here.
486 * tree.h (type_in_anonymous_namespace_p): Declare.
488 2013-08-18 Jakub Jelinek <jakub@redhat.com>
490 PR tree-optimization/58006
491 * tree-parloops.c (take_address_of): Don't ICE if get_name
493 (eliminate_local_variables_stmt): Remove clobber stmts.
495 2013-08-18 Eric Botcazou <ebotcazou@adacore.com>
497 * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
498 error message is issued for an alias to undefined symbol.
500 2013-08-18 Jan Hubicka <jh@suse.cz>
502 * cgraph.c (cgraph_create_indirect_edge): Discover
503 polymorphic calls and record basic info into indirect_info.
504 * gimple-fold.c (gimple_fold_call): When doing BINFO based
505 devirtualization, ignore objc function calls.
506 * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
507 call with no parm index info.
508 * ipa-prop.c (ipa_analyze_call_uses): Likewise.
509 * tree.c (virtual_method_call_p): New function.
510 * tree.h (virtual_method_call_p): Declare.
512 2013-08-16 Jan Hubicka <jh@suse.cz>
515 * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
517 2013-08-16 David Edelsohn <dje.gcc@gmail.com>
519 * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
522 2013-08-16 David Malcolm <dmalcolm@redhat.com>
524 * gengtype.c (type_for_name): Add special-case support for
525 locating types within the "gcc::" namespace.
526 (open_base_files): Emit a "using namespace gcc" directive.
528 2013-08-16 Michael Meissner <meissner@linux.vnet.ibm.com>
531 * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
532 memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
534 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
535 array instead of each individual operand as a separate argument.
536 (emit_fusion_gpr_load): Likewise.
537 (expand_fusion_gpr_load): Add new function declaration.
539 * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
540 signature to have the operands passed as an array, instead of as
541 separate arguments. Allow ZERO_EXTEND to be in the memory
542 address, and also SIGN_EXTEND if -mpower8-fusion-sign. Do not
543 depend on the register live/dead flags when peepholes are run.
544 (expand_fusion_gpr_load): New function to be called from the
545 peephole2 pass, to change the register that addis sets to be the
547 (emit_fusion_gpr_load): Change the calling signature to have the
548 operands passed as an array, instead of as separate arguments.
549 Allow ZERO_EXTEND to be in the memory address, and also
550 SIGN_EXTEND if -mpower8-fusion-sign.
552 * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
554 (power8 fusion peephole/peephole2): Rework the fusion peepholes to
555 adjust the register addis loads up in the peephole2 pass. Do not
556 depend on the register live/dead state when the peephole pass is
559 2013-08-16 David Malcolm <dmalcolm@redhat.com>
561 * gengtype.c (create_user_defined_type): Ensure that the kind
562 is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
563 declaration is seen before the GTY((user)) marking.
565 2013-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
568 * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
570 2013-08-16 Jan Hubicka <jh@suse.cz>
572 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
573 arugment expected_type.
574 (gimple_fold_call): Use it.
575 * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
576 * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
577 * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
578 (try_make_edge_direct_virtual_call): Likewise.
579 * tree.c (obj_type_ref_class): New.
580 * tree.h (obj_type_ref_class): Use it.
582 2013-08-16 Gabriel Dos Reis <gdr@integrable-solutions.net>
584 * sched-vis.c (rtl_slim_pp_initialized): Remove.
585 (rtl_slim_pp): Likewise.
586 (init_rtl_slim_pretty_print): Likewise.
587 (dump_value_slim): Don't call it. Use local pretty printer.
588 (dump_insn_slim): Likewise.
589 (dump_rtl_slim): Likewise.
590 (str_pattern_slim): Likewise.
591 * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
594 2013-08-16 Jakub Jelinek <jakub@redhat.com>
596 PR tree-optimization/58164
597 * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
598 walk gimple_goto_dest of GIMPLE_GOTO.
600 PR tree-optimization/58165
601 * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
602 bi_call must be the last stmt in a bb, don't split_block, instead
603 use fallthru edge from it and give up if there is none.
604 Release conds vector when returning early.
606 2013-08-14 Xinliang David Li <davidxl@google.com>
608 * config/i386/i386.c (ix86_option_override_internal):
609 Remove unused variable and field.
611 2013-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
614 * doc/invoke.texi: Add documentation of mcompat-align-parm option.
615 * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
616 * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
617 and Linux, correct BLKmode alignment when 128-bit alignment is
618 required and compatibility flag is not set.
619 (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
620 for zero-size arguments when compatibility flag is not set.
622 2013-08-14 Jakub Jelinek <jakub@redhat.com>
624 PR tree-optimization/58145
625 * tree-sra.c (build_ref_for_offset): If prev_base has
626 TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
628 2013-08-14 Xinliang David Li <davidxl@google.com>
630 * config/i386/i386.c (ix86_option_override_internal):
631 Fix uninitialized variable error.
633 2013-08-14 Xinliang David Li <davidxl@google.com>
635 * config/i386/i386.opt: Define two new options.
636 * config/i386/x86-tune.def: Add arch selector field in macros.
637 * config/i386/i386.h: Adjust macro definition.
638 * config/i386/i386.c (ix86_option_override_internal):
640 (parse_mtune_ctrl_str): New function.
641 (set_ix86_tune_features): New function.
642 (ix86_function_specific_restore): Call the new helper function.
644 2013-08-14 Andrey Belevantsev <abel@ispras.ru>
646 PR rtl-optimization/57662
647 * sel-sched.c (code_motion_process_successors): When the current insn
648 is removed after the recursive traversal, break from the loop.
649 Add comments and debug printouts.
651 2013-08-14 Jakub Jelinek <jakub@redhat.com>
652 Alexandre Oliva <aoliva@redhat.com>
655 * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
656 and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
657 there also UNSPEC_PLTOFF.
659 2013-08-14 Marek Polacek <polacek@redhat.com>
661 * ipa-inline-analysis.c (add_clause): Avoid shifting integer
662 NUM_CONDITIONS bit positions.
664 2013-08-13 Cary Coutant <ccoutant@google.com>
666 * dwarf2out.c (CHECKSUM_BLOCK): New macro.
667 (attr_checksum): Hash vector contents instead of pointer.
668 (attr_checksum_ordered): Likewise.
670 2013-08-13 Uros Bizjak <ubizjak@gmail.com>
672 * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
673 when Pmode != word_mode. Add length_address attribute.
674 (sse3_monitor_<mode>): Merge from sse3_monitor and
675 sse3_monitor64_<mode> insn patterns. Emit addr32 prefix when
676 Pmode != word_mode. Update insn length attribute.
677 * config/i386/i386.c (ix86_option_override_internal): Update
678 ix86_gen_monitor selection for merged sse3_monitor insn.
680 2013-08-13 Julian Brown <julian@codesourcery.com>
682 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
683 perform invalid legitimization on greater-than-word-size modes for
686 2013-08-13 Vladimir Makarov <vmakarov@redhat.com>
688 * ira.c (setup_class_translate_array): Use aclass instead of cl
689 for classes not fully covered by allocno classes.
691 2013-08-13 Jakub Jelinek <jakub@redhat.com>
693 PR tree-optimization/57661
694 * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
695 * tree-inline.c (tree_function_versioning): Initialize it.
696 (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
697 if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
698 that is not being copied.
701 * asan.c (instrument_strlen_call): Fix typo in comment.
702 Use char * type even for the lhs of POINTER_PLUS_EXPR.
704 2013-08-13 Steve Ellcey <sellcey@mips.com>
706 * config/mips/mips.md (prefetch): Use lw instead of ld on
707 loongson in 32bit mode.
709 2013-08-13 Nick Clifton <nickc@redhat.com>
711 * config.gcc: (avr-linux): Allow for tmake_file not being empty.
713 2013-08-13 Jan Hubicka <jh@suse.cz>
715 * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
716 introduced edge; fix typo in sanity check.
717 (cgraph_resolve_speculation): Export; improve diagnostic.
718 (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
719 speculation at type mismatch.
720 * cgraph.h (cgraph_turn_edge_to_speculative): Update.
721 (cgraph_resolve_speculation): Declare.
722 (symtab_can_be_discarded): New function.
723 * value-prof.c (gimple_ic_transform): Remove actual transform code.
724 * ipa-inline-transform.c (speculation_removed): New global var.
725 (clone_inlined_nodes): See if speculation can be removed.
726 (inline_call): If speculations was removed, we growths may not match.
727 * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
728 (speculation_useful_p): New function.
729 (resolve_noninline_speculation): New function.
730 (inline_small_functions): Resolve useless speculations.
731 * ipa-inline.h (speculation_useful_p): Declare
732 * ipa.c (can_replace_by_local_alias): Simplify.
733 (ipa_profile): Produce speculative calls in non-lto, too;
734 add simple cost model; produce local aliases.
736 2013-08-13 David Malcolm <dmalcolm@redhat.com>
738 * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
741 2013-08-12 Paolo Carlini <paolo.carlini@oracle.com>
743 * config/i386/i386.c (ix86_function_versions): Use error + inform.
745 2013-08-12 Uros Bizjak <ubizjak@gmail.com>
747 * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
748 iterator instead of X87MODEF.
750 2013-08-12 Perez Read <netfirewall@gmail.com>
753 * config/i386/i386.md (*movabs<mode>_1): Add <ptrsize> PTR before
754 operand 0 for intel asm alternative.
755 (*movabs<mode>_2): Ditto for operand 1.
757 2013-08-12 James Greenhalgh <james.greenhalgh@arm.com>
759 * config/aarch64/arm_none.h
760 (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
762 2013-08-12 Nick Clifton <nickc@redhat.com>
764 * config.gcc (m32r-linux): Allow for tmake_file not being empty.
766 2013-08-12 Yuri Rumyantsev <ysrumyan@gmail.com>
768 * config/i386/i386.md (floatunssi<mode>2 expand): Add new
769 expand for QI/HImode operand to produce more effictive code for
770 unsigned char(short) --> float(double) conversion.
772 2013-08-12 Alexander Monakov <amonakov@ispras.ru>
774 * doc/invoke.texi: Mention that -ftls-model does not force the final
777 2013-08-12 Marek Polacek <polacek@redhat.com>
778 Marc Glisse <marc.glisse@inria.fr>
780 PR tree-optimization/57980
781 * tree-tailcall.c (process_assignment): Call build_minus_one_cst
782 when creating -1 constant.
784 2013-08-10 Jan Hubicka <jh@suse.cz>
786 Workaround binutils PR14342.
787 * tree-profile.c (init_ic_make_global_vars): Add LTO path.
788 (gimple_init_edge_profiler): Likewise.
789 (gimple_gen_ic_func_profiler): Likewise.
791 2013-08-09 Jan Hubicka <jh@suse.cz>
793 * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
795 2013-08-09 Xinliang David Li <davidxl@google.com>
797 * config/i386/stringop.def: New file.
798 * config/i386/stringop.opt: New file.
799 * config/i386/i386-opts.h: Include stringopt.def.
800 * config/i386/i386.opt: Include stringopt.opt.
801 * config/i386/i386.c (ix86_option_override_internal):
802 Override default size based stringop inline strategies with options.
803 * config/i386/i386.c (ix86_parse_stringop_strategy_string):
806 2013-08-09 Jan Hubicka <jh@suse.cz>
808 * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
810 2013-08-09 Jan Hubicka <jh@suse.cz>
812 * cgraph.c (cgraph_resolve_speculation): Cut frequency to
814 (dump_cgraph_node): Dump profile-id.
815 * cgraph.h (cgraph_indirect_call_info): Add common_target_id
816 and common_target_probability.
817 * lto-cgraph.c (lto_output_edge): Stream common targets.
818 (lto_output_node): Stream profile ids.
819 (input_node): Stream profile ids.
820 (input_edge): Stream common targets.
821 * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
822 * ipa.c: Include value-prof.h
823 (ipa_profile_generate_summary): Turn indirect call statement histograms
825 (ipa_profile): Turn common targets into speculative edges.
827 2013-08-09 Jan Hubicka <jh@suse.cz>
829 * cgraph.h (cgraph_node): Add profile_id.
830 * value-prof.c (cgraph_node_map): Turn into pointer_map.
831 (init_node_map): Rewrite to handle hashes increas of incremental IDs.
832 (del_node_map): Update.
833 (find_func_by_funcdef_no): Replace by ...
834 (find_func_by_profile_id): ... this one.
835 (gimple_ic_transform): Do not remove useful histograms when
836 speculation is not done; dump info when indirect call removal
838 * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
839 * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
840 (__gcov_indirect_call_profiler_v2): .. this one.
841 * profile.h (init_node_map): Update.
842 * coverage.c (coverage_compute_profile_id): New function.
843 * coverage.h (coverage_compute_profile_id): Declare.
844 * tree-profile.c (init_ic_make_global_vars): Make
845 __gcov_indirect_call_callee and __gcov_indirect_call_counters global.
846 (gimple_init_edge_profiler): Update prototype of
847 __gcov_indirect_call_profiler.
848 (gimple_gen_ic_func_profiler): Simplify.
849 (tree_profiling): Use init_node_map
851 2013-08-09 Jan Hubicka <jh@suse.cz>
853 * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
854 non-speculative refs.
855 * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
856 (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
857 (cgraph_set_call_stmt): Likewise.
858 (cgraph_create_edge_1): Fix release checking compilatoin;
860 (cgraph_free_edge): Free indirect info.
861 (cgraph_turn_edge_to_speculative): New function.
862 (cgraph_speculative_call_info): New function.
863 (cgraph_make_edge_direct): Return direct edge; handle speculation.
864 (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
865 (dump_cgraph_node): Dump speculation.
866 (verify_edge_count_and_frequency): Accept speculative edges.
867 (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
868 (verify_cgraph_node): Handle speculation.
869 * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
870 (cgraph_set_call_stmt): Update prototype.
871 (cgraph_make_edge_direct): Update prototype.
872 (cgraph_speculative_call_info): Declare.
873 * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
874 to change; update call of ipa_find_references.
875 * ipa-ref.c (ipa_record_reference): Fix return value; clear
876 lto_stmt_uid and speculative flags.
877 (ipa_dump_references): Dump speculation.
878 (ipa_clone_references): Clone speculative flag.
879 (ipa_clone_referring): Likewise.
880 (ipa_clone_ref): New function.
881 (ipa_find_reference): Look into lto_stmt_uids
882 (ipa_clear_stmts_in_references): Do not clear speculative calls.
883 * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
884 (ipa_find_reference): Update declaration.
885 (ipa_clone_ref): Declare.
886 * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
887 stream speculative flag.
888 (lto_output_ref): Stream statements uids and speculation.
889 (input_ref): Likewise.
890 (input_edge): Stream speuclation.
891 * cgraphclones.c (cgraph_clone_edge): Clone speculation.
892 (cgraph_set_call_stmt_including_clones): Handle speculation.
893 * ipa-inline.c (heap_edge_removal_hook): New function.
894 (inline_small_functions): Register it.
895 * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
896 also initialize refs.
897 * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
899 (try_make_edge_direct_simple_call): Likewise.
900 (try_make_edge_direct_simple_call): Likewise.
901 (update_indirect_edges_after_inlining): Likewise.
902 (remove_described_reference): Look proper lto_stmt_uid.
903 (propagate_controlled_uses): Likewise.
904 (propagate_controlled_uses): Liekwise.
905 * tree-inline.c (copy_bb): Copy speculative edges.
906 (redirect_all_calls): New function.
907 (copy_cfg_body): Do redirection after loop info is updated.
908 (delete_unreachable_blocks_update_callgraph): Updadte speculation.
910 2013-08-09 Jan Hubicka <jh@suse.cz>
912 * lto-streamer-out.c (output_function): Renumber PHIs.
913 * lto-streamer-in.c (input_function): Likewise.
915 2013-08-09 James Greenhalgh <james.greenhalgh@arm.com>
917 * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
918 (get_lane_unsigned): Likewise.
919 (dup_lane_scalar): Likewise.
920 (get_lane): enable for VALL.
921 * config/aarch64/aarch64-simd.md
922 (aarch64_dup_lane_scalar<mode>): Remove.
923 (aarch64_get_lane_signed<mode>): Likewise.
924 (aarch64_get_lane_unsigned<mode>): Likewise.
925 (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
926 (aarch64_get_lane_zero_extendsi<mode>): Likewise.
927 (aarch64_get_lane<mode>): Enable for all vector modes.
928 (aarch64_get_lanedi): Remove misleading constraints.
929 * config/aarch64/arm_neon.h
930 (__aarch64_vget_lane_any): Define.
931 (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
932 (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
933 (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
934 * config/aarch64/iterators.md (VDQQH): New.
938 2013-08-09 Eric Botcazou <ebotcazou@adacore.com>
940 * configure.ac: Add GAS check for LEON instructions on SPARC.
941 * configure: Regenerate.
942 * config.in: Likewise.
943 * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
945 * config/sparc/sparc.opt (LEON, LEON3): New masks.
946 * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
948 (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
949 (AS_LEON_FLAG): New macro.
950 * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
951 and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
952 Deal with LEON and LEON3 for the memory model.
953 * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
954 (atomic_compare_and_swap<mode>_1): Likewise.
955 (*atomic_compare_and_swap<mode>_1): Likewise.
957 2013-08-09 Zhenqiang Chen <zhenqiang.chen@linaro.org>
959 * config/arm/neon.md (vcond): Fix floating-point vector
960 comparisons against 0.
962 2013-08-08 Vladimir Makarov <vmakarov@redhat.com>
964 * lra-constraints.c (emit_spill_move): Remove assert.
965 (process_alt_operands): Add more debugging
966 output. Increase reject for spilling into memory. Decrease
967 reject for reloading scratch.
968 (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
970 2013-08-08 Steve Ellcey <sellcey@mips.com>
972 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
973 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
974 micromips incompatible. Add nan2008.
975 (MULTILIB_DIRNAMES): Add nan2008.
976 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
977 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
978 and micromips incompatible. Add nan2008.
979 (MULTILIB_DIRNAMES): Add nan2008.
980 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
982 2013-08-08 Richard Sandiford <rdsandiford@googlemail.com>
984 PR rtl-optimization/58079
985 * combine.c (combine_simplify_rtx): Avoid using SUBST if
986 simplify_comparison has widened a comparison with an integer.
988 2013-08-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
990 * config/arm/neon.md (movmisalign<mode>): Disable when we
991 don't allow unaligned accesses.
992 (*movmisalign<mode>_neon_store): Likewise.
993 (*movmisalign<mode>_neon_load): Likewise.
994 (*movmisalign<mode>_neon_store): Likewise.
995 (*movmisalign<mode>_neon_load): Likewise.
997 2013-08-08 Jan Hubicka <jh@suse.cz>
999 * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1000 (make_pass_rebuild_cgraph_edges): Also clear references.
1001 * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1002 * ipa-inline-transform.c (inline_transform): Remove all references
1004 * cgraphunit.c (expand_function): Remove all references after
1006 * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1007 (ipa_find_reference): Rewrite to iterator.
1008 (remove_stmt_references): Likewise.
1009 (ipa_clear_stmts_in_references): New function.
1010 * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
1011 * cgraphclones.c (cgraph_materialize_all_clones): Remove or
1013 * ipa-split.c (split_function): Remove references in split function.
1015 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
1018 * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
1019 (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
1021 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
1024 * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
1025 suggested mode for the assignment isn't compatible with the
1028 2013-08-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1031 * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
1033 2013-08-07 Xinliang David Li <davidxl@google.com>
1035 * config/i386/i386.opt: New option -mtune-ctrl=.
1036 * config/i386/x86-tune.def: New file.
1037 * config/i386/i386.h: include x86-tune.def.
1038 * config/i386/i386.c (ix86_option_override_internal):
1039 Parsing -mtune-ctrl= option and set tune features.
1041 2013-08-07 Oleg Endo <olegendo@gcc.gnu.org>
1044 * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
1045 (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
1048 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
1050 * rtl.h (update_alignments): Declare.
1051 * final.c (grow_label_align): New function extracted from...
1052 (shorten_branches): ...here. Call it.
1053 (update_alignments): New function.
1054 * reorg.c (sibling_labels): New variable.
1055 (get_label_before): Add SIBLING parameter. If it is non-zero, push
1056 the new label along with it onto the sibling_labels vector.
1057 (fill_simple_delay_slots): Adjust call to get_label_before.
1058 (fill_slots_from_thread): Likewise.
1059 (relax_delay_slots): Likewise.
1060 (make_return_insns): Likewise.
1061 (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
1063 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
1065 * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
1066 document its semantics.
1067 (diagnostic_report_diagnostic): Adjust accordingly.
1069 2013-08-07 David Malcolm <dmalcolm@redhat.com>
1071 * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
1072 (sparc_option_override): ...and port to new C++ pass API.
1073 * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
1075 2013-08-07 Peter Bergner <bergner@vnet.ibm.com>
1077 * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
1079 2013-08-06 Caroline Tice <cmtice@google.com>
1081 * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
1082 (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
1083 * tree-pass.h: Add pass_vtable_verify.
1084 * varasm.c (assemble_variable): Add code to properly set the comdat
1085 section and name for the .vtable_map_vars section.
1086 (assemble_vtyv_preinit_initializer): New function.
1087 (default_sectin_type_flags): Make sure .vtable_map_vars section has
1089 * output.h: Add function decl for assemble_vtv_preinit_initializer.
1090 * vtable-verify.c: New file.
1091 * vtable-verify.h: New file.
1092 * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
1093 initialiation levels.
1094 * timevar.def (TV_VTABLE_VERIFICATION): New definition.
1095 * passes.def: Insert pass_vtable_verify.
1096 * aclocal.m4: Reorder includes.
1097 * doc/invoke.texi: Document the -fvtable-verify=, -fvtv-debug, and
1098 -fvtv-counts options.
1099 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
1100 as appropriate, if -fvtable-verify=... is used.
1101 (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
1102 -fvtable-verify=... is used.
1103 * Makefile.in (OBJS): Add vtable-verify.o to list.
1104 (vtable-verify.o): Add new build rule.
1105 (GTFILES): Add vtable-verify.c to list.
1106 * common.opt (fvtable-verify=): New flag.
1107 (vtv_priority): Values for fvtable-verify= flag.
1108 (fvtv-counts): New flag.
1109 (fvtv-debug): New flag.
1110 * tree.h (save_vtable_map_decl): New extern function decl.
1112 2013-08-07 David Malcolm <dmalcolm@redhat.com>
1114 * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
1115 (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
1116 (pass_data_rl78_devirt): ...new pass_data instance and...
1117 (make_pass_rl78_devirt): ...new function.
1118 (rl78_asm_file_start): Port pass registration to new C++ API.
1120 2013-08-07 David Malcolm <dmalcolm@redhat.com>
1122 * coretypes.h (rtl_opt_pass): Add.
1123 (gcc::context): Add.
1124 * config/epiphany/epiphany.c (pass_mode_switch_use): New.
1125 (epiphany_init): Port to new C++ pass API.
1126 (epiphany_optimize_mode_switching): Likewise.
1127 * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
1128 (pass_manager::get_pass_mode_switching): New.
1129 (pass_manager::get_pass_peephole2): New.
1130 * mode-switching.c (pass_mode_switching): Add clone method.
1131 * recog.c (pass_peephole2): Add clone method.
1132 (pass_split_all_insns): Add clone method.
1134 2013-08-06 David Malcolm <dmalcolm@redhat.com>
1136 * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
1137 (mips_option_override): ...here, porting to new C++ API for passes.
1139 2013-08-06 Jan Hubicka <jh@suse.cz>
1141 * cgraph.c (cgraph_get_body): New function based on lto.c
1143 * cgraph.h (cgraph_get_body): Declare.
1144 * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
1146 * cgraphunit.c (expand_function): Get body prior expanding.
1147 * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
1148 * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
1150 * passes.c (do_per_function_toporder): Get body.
1151 * tree-inline.c (expand_call_inline): Get body prior inlining it.
1152 * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
1154 2013-08-06 Martin Jambor <mjambor@suse.cz>
1157 * cgraphunit.c (cgraph_finalize_function): Assert that nested function
1158 is not re-finalized. Rename second parameter to no_collect.
1160 2013-08-06 Martin Jambor <mjambor@suse.cz>
1163 * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
1164 MEM_REF has proper alignment information.
1166 2013-08-05 Oleg Endo <olegendo@gcc.gnu.org>
1169 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
1171 * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
1172 rtx (*) (rtx, ...) with insn_gen_fn.
1173 * genoutput.c (output_insn_data): Cast gen_? function pointers to
1174 insn_gen_fn::stored_funcptr. Add initializer braces.
1176 2013-08-05 David Malcolm <dmalcolm@redhat.com>
1178 Rewrite how instances of passes are cloned to remove assumptions
1179 about their sizes (thus allowing pass subclasses to have
1180 additional data fields, albeit non-GC-managed ones at this point).
1182 * passes.c (make_pass_instance): Now that passes have clone
1183 methods, rewrite this function to eliminate XNEW and memcpy
1184 calls that used hardcoded sizes. Since this function no longer
1185 creates pass instances, rename it to...
1186 (add_pass_instance): ...this. Document the old way that passes were
1187 numbered and flagged, and rework this function to continue using it.
1188 (next_pass_1): Add an initial_pass argument for use by
1190 (position_pass): When adding multiple instances of a pass, use
1191 the pass's clone method, rather than relying on the XNEW/memcpy
1192 within the former make_pass_instance (now add_pass_instance).
1193 (pass_manager::pass_manager): When invoking next_pass_1, also supply
1194 the initial instance of the current pass within the pass manager.
1196 2013-08-05 David Malcolm <dmalcolm@redhat.com>
1198 This is the automated part of the conversion of passes from C
1199 structs to C++ classes.
1201 Patch autogenerated by refactor_passes.py from
1202 https://github.com/davidmalcolm/gcc-refactoring-scripts
1203 revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
1205 * asan.c (pass_asan): Convert from a global struct to a subclass of
1206 gimple_opt_pass along with...
1207 (pass_data_asan): ...new pass_data instance and...
1208 (make_pass_asan): ...new function.
1209 (pass_asan_O0): Convert from a global struct to a subclass of
1210 gimple_opt_pass along with...
1211 (pass_data_asan_O0): ...new pass_data instance and...
1212 (make_pass_asan_O0): ...new function.
1213 * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
1214 subclass of rtl_opt_pass along with...
1215 (pass_data_inc_dec): ...new pass_data instance and...
1216 (make_pass_inc_dec): ...new function.
1217 * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
1218 a subclass of rtl_opt_pass along with...
1219 (pass_data_reorder_blocks): ...new pass_data instance and...
1220 (make_pass_reorder_blocks): ...new function.
1221 (pass_duplicate_computed_gotos): Convert from a global struct to a
1222 subclass of rtl_opt_pass along with...
1223 (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
1224 (make_pass_duplicate_computed_gotos): ...new function.
1225 (pass_partition_blocks): Convert from a global struct to a subclass of
1226 rtl_opt_pass along with...
1227 (pass_data_partition_blocks): ...new pass_data instance and...
1228 (make_pass_partition_blocks): ...new function.
1229 * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
1230 struct to a subclass of rtl_opt_pass along with...
1231 (pass_data_branch_target_load_optimize1): ...new pass_data instance
1233 (make_pass_branch_target_load_optimize1): ...new function.
1234 (pass_branch_target_load_optimize2): Convert from a global struct to a
1235 subclass of rtl_opt_pass along with...
1236 (pass_data_branch_target_load_optimize2): ...new pass_data instance
1238 (make_pass_branch_target_load_optimize2): ...new function.
1239 * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
1240 of rtl_opt_pass along with...
1241 (pass_data_jump): ...new pass_data instance and...
1242 (make_pass_jump): ...new function.
1243 (pass_jump2): Convert from a global struct to a subclass of
1244 rtl_opt_pass along with...
1245 (pass_data_jump2): ...new pass_data instance and...
1246 (make_pass_jump2): ...new function.
1247 * cfgexpand.c (pass_expand): Convert from a global struct to a
1248 subclass of rtl_opt_pass along with...
1249 (pass_data_expand): ...new pass_data instance and...
1250 (make_pass_expand): ...new function.
1251 * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
1252 of rtl_opt_pass along with...
1253 (pass_data_free_cfg): ...new pass_data instance and...
1254 (make_pass_free_cfg): ...new function.
1255 (pass_into_cfg_layout_mode): Convert from a global struct to a
1256 subclass of rtl_opt_pass along with...
1257 (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
1258 (make_pass_into_cfg_layout_mode): ...new function.
1259 (pass_outof_cfg_layout_mode): Convert from a global struct to a
1260 subclass of rtl_opt_pass along with...
1261 (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
1262 (make_pass_outof_cfg_layout_mode): ...new function.
1263 * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
1264 struct to a subclass of gimple_opt_pass along with...
1265 (pass_data_build_cgraph_edges): ...new pass_data instance and...
1266 (make_pass_build_cgraph_edges): ...new function.
1267 (pass_rebuild_cgraph_edges): Convert from a global struct to a
1268 subclass of gimple_opt_pass along with...
1269 (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
1270 (make_pass_rebuild_cgraph_edges): ...new function.
1271 (pass_remove_cgraph_callee_edges): Convert from a global struct to a
1272 subclass of gimple_opt_pass along with...
1273 (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
1275 (make_pass_remove_cgraph_callee_edges): ...new function.
1276 * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
1277 struct to a subclass of rtl_opt_pass along with...
1278 (pass_data_stack_adjustments): ...new pass_data instance and...
1279 (make_pass_stack_adjustments): ...new function.
1280 * combine.c (pass_combine): Convert from a global struct to a subclass
1281 of rtl_opt_pass along with...
1282 (pass_data_combine): ...new pass_data instance and...
1283 (make_pass_combine): ...new function.
1284 * compare-elim.c (pass_compare_elim_after_reload): Convert from a
1285 global struct to a subclass of rtl_opt_pass along with...
1286 (pass_data_compare_elim_after_reload): ...new pass_data instance
1288 (make_pass_compare_elim_after_reload): ...new function.
1289 * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
1290 of rtl_opt_pass along with...
1291 (pass_data_rtl_cprop): ...new pass_data instance and...
1292 (make_pass_rtl_cprop): ...new function.
1293 * cse.c (pass_cse): Convert from a global struct to a subclass of
1294 rtl_opt_pass along with...
1295 (pass_data_cse): ...new pass_data instance and...
1296 (make_pass_cse): ...new function.
1297 (pass_cse2): Convert from a global struct to a subclass of
1298 rtl_opt_pass along with...
1299 (pass_data_cse2): ...new pass_data instance and...
1300 (make_pass_cse2): ...new function.
1301 (pass_cse_after_global_opts): Convert from a global struct to a
1302 subclass of rtl_opt_pass along with...
1303 (pass_data_cse_after_global_opts): ...new pass_data instance and...
1304 (make_pass_cse_after_global_opts): ...new function.
1305 * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
1306 of rtl_opt_pass along with...
1307 (pass_data_ud_rtl_dce): ...new pass_data instance and...
1308 (make_pass_ud_rtl_dce): ...new function.
1309 (pass_fast_rtl_dce): Convert from a global struct to a subclass of
1310 rtl_opt_pass along with...
1311 (pass_data_fast_rtl_dce): ...new pass_data instance and...
1312 (make_pass_fast_rtl_dce): ...new function.
1313 * df-core.c (pass_df_initialize_opt): Convert from a global struct to
1314 a subclass of rtl_opt_pass along with...
1315 (pass_data_df_initialize_opt): ...new pass_data instance and...
1316 (make_pass_df_initialize_opt): ...new function.
1317 (pass_df_initialize_no_opt): Convert from a global struct to a
1318 subclass of rtl_opt_pass along with...
1319 (pass_data_df_initialize_no_opt): ...new pass_data instance and...
1320 (make_pass_df_initialize_no_opt): ...new function.
1321 (pass_df_finish): Convert from a global struct to a subclass of
1322 rtl_opt_pass along with...
1323 (pass_data_df_finish): ...new pass_data instance and...
1324 (make_pass_df_finish): ...new function.
1325 * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
1326 rtl_opt_pass along with...
1327 (pass_data_rtl_dse1): ...new pass_data instance and...
1328 (make_pass_rtl_dse1): ...new function.
1329 (pass_rtl_dse2): Convert from a global struct to a subclass of
1330 rtl_opt_pass along with...
1331 (pass_data_rtl_dse2): ...new pass_data instance and...
1332 (make_pass_rtl_dse2): ...new function.
1333 * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
1334 subclass of rtl_opt_pass along with...
1335 (pass_data_dwarf2_frame): ...new pass_data instance and...
1336 (make_pass_dwarf2_frame): ...new function.
1337 * except.c (pass_set_nothrow_function_flags): Convert from a global
1338 struct to a subclass of rtl_opt_pass along with...
1339 (pass_data_set_nothrow_function_flags): ...new pass_data instance
1341 (make_pass_set_nothrow_function_flags): ...new function.
1342 (pass_convert_to_eh_region_ranges): Convert from a global struct to a
1343 subclass of rtl_opt_pass along with...
1344 (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
1346 (make_pass_convert_to_eh_region_ranges): ...new function.
1347 * final.c (pass_compute_alignments): Convert from a global struct to a
1348 subclass of rtl_opt_pass along with...
1349 (pass_data_compute_alignments): ...new pass_data instance and...
1350 (make_pass_compute_alignments): ...new function.
1351 (pass_final): Convert from a global struct to a subclass of
1352 rtl_opt_pass along with...
1353 (pass_data_final): ...new pass_data instance and...
1354 (make_pass_final): ...new function.
1355 (pass_shorten_branches): Convert from a global struct to a subclass of
1356 rtl_opt_pass along with...
1357 (pass_data_shorten_branches): ...new pass_data instance and...
1358 (make_pass_shorten_branches): ...new function.
1359 (pass_clean_state): Convert from a global struct to a subclass of
1360 rtl_opt_pass along with...
1361 (pass_data_clean_state): ...new pass_data instance and...
1362 (make_pass_clean_state): ...new function.
1363 * function.c (pass_instantiate_virtual_regs): Convert from a global
1364 struct to a subclass of rtl_opt_pass along with...
1365 (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
1366 (make_pass_instantiate_virtual_regs): ...new function.
1367 (pass_leaf_regs): Convert from a global struct to a subclass of
1368 rtl_opt_pass along with...
1369 (pass_data_leaf_regs): ...new pass_data instance and...
1370 (make_pass_leaf_regs): ...new function.
1371 (pass_thread_prologue_and_epilogue): Convert from a global struct to a
1372 subclass of rtl_opt_pass along with...
1373 (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
1375 (make_pass_thread_prologue_and_epilogue): ...new function.
1376 (pass_match_asm_constraints): Convert from a global struct to a
1377 subclass of rtl_opt_pass along with...
1378 (pass_data_match_asm_constraints): ...new pass_data instance and...
1379 (make_pass_match_asm_constraints): ...new function.
1380 * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
1381 subclass of rtl_opt_pass along with...
1382 (pass_data_rtl_fwprop): ...new pass_data instance and...
1383 (make_pass_rtl_fwprop): ...new function.
1384 (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
1385 rtl_opt_pass along with...
1386 (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
1387 (make_pass_rtl_fwprop_addr): ...new function.
1388 * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
1389 rtl_opt_pass along with...
1390 (pass_data_rtl_pre): ...new pass_data instance and...
1391 (make_pass_rtl_pre): ...new function.
1392 (pass_rtl_hoist): Convert from a global struct to a subclass of
1393 rtl_opt_pass along with...
1394 (pass_data_rtl_hoist): ...new pass_data instance and...
1395 (make_pass_rtl_hoist): ...new function.
1396 * gimple-low.c (pass_lower_cf): Convert from a global struct to a
1397 subclass of gimple_opt_pass along with...
1398 (pass_data_lower_cf): ...new pass_data instance and...
1399 (make_pass_lower_cf): ...new function.
1400 * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
1401 from a global struct to a subclass of gimple_opt_pass along with...
1402 (pass_data_strength_reduction): ...new pass_data instance and...
1403 (make_pass_strength_reduction): ...new function.
1404 * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
1405 of rtl_opt_pass along with...
1406 (pass_data_rtl_ifcvt): ...new pass_data instance and...
1407 (make_pass_rtl_ifcvt): ...new function.
1408 (pass_if_after_combine): Convert from a global struct to a subclass of
1409 rtl_opt_pass along with...
1410 (pass_data_if_after_combine): ...new pass_data instance and...
1411 (make_pass_if_after_combine): ...new function.
1412 (pass_if_after_reload): Convert from a global struct to a subclass of
1413 rtl_opt_pass along with...
1414 (pass_data_if_after_reload): ...new pass_data instance and...
1415 (make_pass_if_after_reload): ...new function.
1416 * init-regs.c (pass_initialize_regs): Convert from a global struct to
1417 a subclass of rtl_opt_pass along with...
1418 (pass_data_initialize_regs): ...new pass_data instance and...
1419 (make_pass_initialize_regs): ...new function.
1420 * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
1421 of ipa_opt_pass_d along with...
1422 (pass_data_ipa_cp): ...new pass_data instance and...
1423 (make_pass_ipa_cp): ...new function.
1424 * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
1425 global struct to a subclass of gimple_opt_pass along with...
1426 (pass_data_inline_parameters): ...new pass_data instance and...
1427 (make_pass_inline_parameters): ...new function.
1428 * ipa-inline.c (pass_early_inline): Convert from a global struct to a
1429 subclass of gimple_opt_pass along with...
1430 (pass_data_early_inline): ...new pass_data instance and...
1431 (make_pass_early_inline): ...new function.
1432 (pass_ipa_inline): Convert from a global struct to a subclass of
1433 ipa_opt_pass_d along with...
1434 (pass_data_ipa_inline): ...new pass_data instance and...
1435 (make_pass_ipa_inline): ...new function.
1436 * ipa-pure-const.c (pass_local_pure_const): Convert from a global
1437 struct to a subclass of gimple_opt_pass along with...
1438 (pass_data_local_pure_const): ...new pass_data instance and...
1439 (make_pass_local_pure_const): ...new function.
1440 (pass_ipa_pure_const): Convert from a global struct to a subclass of
1441 ipa_opt_pass_d along with...
1442 (pass_data_ipa_pure_const): ...new pass_data instance and...
1443 (make_pass_ipa_pure_const): ...new function.
1444 * ipa-reference.c (pass_ipa_reference): Convert from a global struct
1445 to a subclass of ipa_opt_pass_d along with...
1446 (pass_data_ipa_reference): ...new pass_data instance and...
1447 (make_pass_ipa_reference): ...new function.
1448 * ipa-split.c (pass_split_functions): Convert from a global struct to
1449 a subclass of gimple_opt_pass along with...
1450 (pass_data_split_functions): ...new pass_data instance and...
1451 (make_pass_split_functions): ...new function.
1452 (pass_feedback_split_functions): Convert from a global struct to a
1453 subclass of gimple_opt_pass along with...
1454 (pass_data_feedback_split_functions): ...new pass_data instance and...
1455 (make_pass_feedback_split_functions): ...new function.
1456 * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
1457 global struct to a subclass of simple_ipa_opt_pass along with...
1458 (pass_data_ipa_function_and_variable_visibility): ...new pass_data
1460 (make_pass_ipa_function_and_variable_visibility): ...new function.
1461 (pass_ipa_free_inline_summary): Convert from a global struct to a
1462 subclass of simple_ipa_opt_pass along with...
1463 (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
1464 (make_pass_ipa_free_inline_summary): ...new function.
1465 (pass_ipa_whole_program_visibility): Convert from a global struct to a
1466 subclass of ipa_opt_pass_d along with...
1467 (pass_data_ipa_whole_program_visibility): ...new pass_data instance
1469 (make_pass_ipa_whole_program_visibility): ...new function.
1470 (pass_ipa_profile): Convert from a global struct to a subclass of
1471 ipa_opt_pass_d along with...
1472 (pass_data_ipa_profile): ...new pass_data instance and...
1473 (make_pass_ipa_profile): ...new function.
1474 (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
1475 ipa_opt_pass_d along with...
1476 (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
1477 (make_pass_ipa_cdtor_merge): ...new function.
1478 * ira.c (pass_ira): Convert from a global struct to a subclass of
1479 rtl_opt_pass along with...
1480 (pass_data_ira): ...new pass_data instance and...
1481 (make_pass_ira): ...new function.
1482 (pass_reload): Convert from a global struct to a subclass of
1483 rtl_opt_pass along with...
1484 (pass_data_reload): ...new pass_data instance and...
1485 (make_pass_reload): ...new function.
1486 * jump.c (pass_cleanup_barriers): Convert from a global struct to a
1487 subclass of rtl_opt_pass along with...
1488 (pass_data_cleanup_barriers): ...new pass_data instance and...
1489 (make_pass_cleanup_barriers): ...new function.
1490 * loop-init.c (pass_loop2): Convert from a global struct to a subclass
1491 of rtl_opt_pass along with...
1492 (pass_data_loop2): ...new pass_data instance and...
1493 (make_pass_loop2): ...new function.
1494 (pass_rtl_loop_init): Convert from a global struct to a subclass of
1495 rtl_opt_pass along with...
1496 (pass_data_rtl_loop_init): ...new pass_data instance and...
1497 (make_pass_rtl_loop_init): ...new function.
1498 (pass_rtl_loop_done): Convert from a global struct to a subclass of
1499 rtl_opt_pass along with...
1500 (pass_data_rtl_loop_done): ...new pass_data instance and...
1501 (make_pass_rtl_loop_done): ...new function.
1502 (pass_rtl_move_loop_invariants): Convert from a global struct to a
1503 subclass of rtl_opt_pass along with...
1504 (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
1505 (make_pass_rtl_move_loop_invariants): ...new function.
1506 (pass_rtl_unswitch): Convert from a global struct to a subclass of
1507 rtl_opt_pass along with...
1508 (pass_data_rtl_unswitch): ...new pass_data instance and...
1509 (make_pass_rtl_unswitch): ...new function.
1510 (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
1511 subclass of rtl_opt_pass along with...
1512 (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
1514 (make_pass_rtl_unroll_and_peel_loops): ...new function.
1515 (pass_rtl_doloop): Convert from a global struct to a subclass of
1516 rtl_opt_pass along with...
1517 (pass_data_rtl_doloop): ...new pass_data instance and...
1518 (make_pass_rtl_doloop): ...new function.
1519 * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
1520 a subclass of rtl_opt_pass along with...
1521 (pass_data_lower_subreg): ...new pass_data instance and...
1522 (make_pass_lower_subreg): ...new function.
1523 (pass_lower_subreg2): Convert from a global struct to a subclass of
1524 rtl_opt_pass along with...
1525 (pass_data_lower_subreg2): ...new pass_data instance and...
1526 (make_pass_lower_subreg2): ...new function.
1527 * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
1528 struct to a subclass of ipa_opt_pass_d along with...
1529 (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
1530 (make_pass_ipa_lto_gimple_out): ...new function.
1531 (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
1532 of ipa_opt_pass_d along with...
1533 (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
1534 (make_pass_ipa_lto_finish_out): ...new function.
1535 * mode-switching.c (pass_mode_switching): Convert from a global struct
1536 to a subclass of rtl_opt_pass along with...
1537 (pass_data_mode_switching): ...new pass_data instance and...
1538 (make_pass_mode_switching): ...new function.
1539 * modulo-sched.c (pass_sms): Convert from a global struct to a
1540 subclass of rtl_opt_pass along with...
1541 (pass_data_sms): ...new pass_data instance and...
1542 (make_pass_sms): ...new function.
1543 * omp-low.c (pass_expand_omp): Convert from a global struct to a
1544 subclass of gimple_opt_pass along with...
1545 (pass_data_expand_omp): ...new pass_data instance and...
1546 (make_pass_expand_omp): ...new function.
1547 (pass_lower_omp): Convert from a global struct to a subclass of
1548 gimple_opt_pass along with...
1549 (pass_data_lower_omp): ...new pass_data instance and...
1550 (make_pass_lower_omp): ...new function.
1551 (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
1552 of gimple_opt_pass along with...
1553 (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
1554 (make_pass_diagnose_omp_blocks): ...new function.
1555 * passes.c (pass_early_local_passes): Convert from a global struct to
1556 a subclass of simple_ipa_opt_pass along with...
1557 (pass_data_early_local_passes): ...new pass_data instance and...
1558 (make_pass_early_local_passes): ...new function.
1559 (pass_all_early_optimizations): Convert from a global struct to a
1560 subclass of gimple_opt_pass along with...
1561 (pass_data_all_early_optimizations): ...new pass_data instance and...
1562 (make_pass_all_early_optimizations): ...new function.
1563 (pass_all_optimizations): Convert from a global struct to a subclass
1564 of gimple_opt_pass along with...
1565 (pass_data_all_optimizations): ...new pass_data instance and...
1566 (make_pass_all_optimizations): ...new function.
1567 (pass_all_optimizations_g): Convert from a global struct to a subclass
1568 of gimple_opt_pass along with...
1569 (pass_data_all_optimizations_g): ...new pass_data instance and...
1570 (make_pass_all_optimizations_g): ...new function.
1571 (pass_rest_of_compilation): Convert from a global struct to a subclass
1572 of rtl_opt_pass along with...
1573 (pass_data_rest_of_compilation): ...new pass_data instance and...
1574 (make_pass_rest_of_compilation): ...new function.
1575 (pass_postreload): Convert from a global struct to a subclass of
1576 rtl_opt_pass along with...
1577 (pass_data_postreload): ...new pass_data instance and...
1578 (make_pass_postreload): ...new function.
1579 * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
1580 subclass of rtl_opt_pass along with...
1581 (pass_data_gcse2): ...new pass_data instance and...
1582 (make_pass_gcse2): ...new function.
1583 * postreload.c (pass_postreload_cse): Convert from a global struct to
1584 a subclass of rtl_opt_pass along with...
1585 (pass_data_postreload_cse): ...new pass_data instance and...
1586 (make_pass_postreload_cse): ...new function.
1587 * predict.c (pass_profile): Convert from a global struct to a subclass
1588 of gimple_opt_pass along with...
1589 (pass_data_profile): ...new pass_data instance and...
1590 (make_pass_profile): ...new function.
1591 (pass_strip_predict_hints): Convert from a global struct to a subclass
1592 of gimple_opt_pass along with...
1593 (pass_data_strip_predict_hints): ...new pass_data instance and...
1594 (make_pass_strip_predict_hints): ...new function.
1595 * recog.c (pass_peephole2): Convert from a global struct to a subclass
1596 of rtl_opt_pass along with...
1597 (pass_data_peephole2): ...new pass_data instance and...
1598 (make_pass_peephole2): ...new function.
1599 (pass_split_all_insns): Convert from a global struct to a subclass of
1600 rtl_opt_pass along with...
1601 (pass_data_split_all_insns): ...new pass_data instance and...
1602 (make_pass_split_all_insns): ...new function.
1603 (pass_split_after_reload): Convert from a global struct to a subclass
1604 of rtl_opt_pass along with...
1605 (pass_data_split_after_reload): ...new pass_data instance and...
1606 (make_pass_split_after_reload): ...new function.
1607 (pass_split_before_regstack): Convert from a global struct to a
1608 subclass of rtl_opt_pass along with...
1609 (pass_data_split_before_regstack): ...new pass_data instance and...
1610 (make_pass_split_before_regstack): ...new function.
1611 (pass_split_before_sched2): Convert from a global struct to a subclass
1612 of rtl_opt_pass along with...
1613 (pass_data_split_before_sched2): ...new pass_data instance and...
1614 (make_pass_split_before_sched2): ...new function.
1615 (pass_split_for_shorten_branches): Convert from a global struct to a
1616 subclass of rtl_opt_pass along with...
1617 (pass_data_split_for_shorten_branches): ...new pass_data instance
1619 (make_pass_split_for_shorten_branches): ...new function.
1620 * ree.c (pass_ree): Convert from a global struct to a subclass of
1621 rtl_opt_pass along with...
1622 (pass_data_ree): ...new pass_data instance and...
1623 (make_pass_ree): ...new function.
1624 * reg-stack.c (pass_stack_regs): Convert from a global struct to a
1625 subclass of rtl_opt_pass along with...
1626 (pass_data_stack_regs): ...new pass_data instance and...
1627 (make_pass_stack_regs): ...new function.
1628 (pass_stack_regs_run): Convert from a global struct to a subclass of
1629 rtl_opt_pass along with...
1630 (pass_data_stack_regs_run): ...new pass_data instance and...
1631 (make_pass_stack_regs_run): ...new function.
1632 * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
1633 subclass of rtl_opt_pass along with...
1634 (pass_data_cprop_hardreg): ...new pass_data instance and...
1635 (make_pass_cprop_hardreg): ...new function.
1636 * reginfo.c (pass_reginfo_init): Convert from a global struct to a
1637 subclass of rtl_opt_pass along with...
1638 (pass_data_reginfo_init): ...new pass_data instance and...
1639 (make_pass_reginfo_init): ...new function.
1640 * regmove.c (pass_regmove): Convert from a global struct to a subclass
1641 of rtl_opt_pass along with...
1642 (pass_data_regmove): ...new pass_data instance and...
1643 (make_pass_regmove): ...new function.
1644 * regrename.c (pass_regrename): Convert from a global struct to a
1645 subclass of rtl_opt_pass along with...
1646 (pass_data_regrename): ...new pass_data instance and...
1647 (make_pass_regrename): ...new function.
1648 * reorg.c (pass_delay_slots): Convert from a global struct to a
1649 subclass of rtl_opt_pass along with...
1650 (pass_data_delay_slots): ...new pass_data instance and...
1651 (make_pass_delay_slots): ...new function.
1652 (pass_machine_reorg): Convert from a global struct to a subclass of
1653 rtl_opt_pass along with...
1654 (pass_data_machine_reorg): ...new pass_data instance and...
1655 (make_pass_machine_reorg): ...new function.
1656 * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
1657 of rtl_opt_pass along with...
1658 (pass_data_sched): ...new pass_data instance and...
1659 (make_pass_sched): ...new function.
1660 (pass_sched2): Convert from a global struct to a subclass of
1661 rtl_opt_pass along with...
1662 (pass_data_sched2): ...new pass_data instance and...
1663 (make_pass_sched2): ...new function.
1664 * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
1665 to a subclass of rtl_opt_pass along with...
1666 (pass_data_stack_ptr_mod): ...new pass_data instance and...
1667 (make_pass_stack_ptr_mod): ...new function.
1668 * store-motion.c (pass_rtl_store_motion): Convert from a global struct
1669 to a subclass of rtl_opt_pass along with...
1670 (pass_data_rtl_store_motion): ...new pass_data instance and...
1671 (make_pass_rtl_store_motion): ...new function.
1672 * tracer.c (pass_tracer): Convert from a global struct to a subclass
1673 of gimple_opt_pass along with...
1674 (pass_data_tracer): ...new pass_data instance and...
1675 (make_pass_tracer): ...new function.
1676 * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
1677 to a subclass of gimple_opt_pass along with...
1678 (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
1679 (make_pass_diagnose_tm_blocks): ...new function.
1680 (pass_lower_tm): Convert from a global struct to a subclass of
1681 gimple_opt_pass along with...
1682 (pass_data_lower_tm): ...new pass_data instance and...
1683 (make_pass_lower_tm): ...new function.
1684 (pass_tm_init): Convert from a global struct to a subclass of
1685 gimple_opt_pass along with...
1686 (pass_data_tm_init): ...new pass_data instance and...
1687 (make_pass_tm_init): ...new function.
1688 (pass_tm_mark): Convert from a global struct to a subclass of
1689 gimple_opt_pass along with...
1690 (pass_data_tm_mark): ...new pass_data instance and...
1691 (make_pass_tm_mark): ...new function.
1692 (pass_tm_edges): Convert from a global struct to a subclass of
1693 gimple_opt_pass along with...
1694 (pass_data_tm_edges): ...new pass_data instance and...
1695 (make_pass_tm_edges): ...new function.
1696 (pass_tm_memopt): Convert from a global struct to a subclass of
1697 gimple_opt_pass along with...
1698 (pass_data_tm_memopt): ...new pass_data instance and...
1699 (make_pass_tm_memopt): ...new function.
1700 (pass_ipa_tm): Convert from a global struct to a subclass of
1701 simple_ipa_opt_pass along with...
1702 (pass_data_ipa_tm): ...new pass_data instance and...
1703 (make_pass_ipa_tm): ...new function.
1704 * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
1705 subclass of gimple_opt_pass along with...
1706 (pass_data_call_cdce): ...new pass_data instance and...
1707 (make_pass_call_cdce): ...new function.
1708 * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
1709 subclass of gimple_opt_pass along with...
1710 (pass_data_build_cfg): ...new pass_data instance and...
1711 (make_pass_build_cfg): ...new function.
1712 (pass_split_crit_edges): Convert from a global struct to a subclass of
1713 gimple_opt_pass along with...
1714 (pass_data_split_crit_edges): ...new pass_data instance and...
1715 (make_pass_split_crit_edges): ...new function.
1716 (pass_warn_function_return): Convert from a global struct to a
1717 subclass of gimple_opt_pass along with...
1718 (pass_data_warn_function_return): ...new pass_data instance and...
1719 (make_pass_warn_function_return): ...new function.
1720 (pass_warn_function_noreturn): Convert from a global struct to a
1721 subclass of gimple_opt_pass along with...
1722 (pass_data_warn_function_noreturn): ...new pass_data instance and...
1723 (make_pass_warn_function_noreturn): ...new function.
1724 (pass_warn_unused_result): Convert from a global struct to a subclass
1725 of gimple_opt_pass along with...
1726 (pass_data_warn_unused_result): ...new pass_data instance and...
1727 (make_pass_warn_unused_result): ...new function.
1728 * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
1729 a subclass of gimple_opt_pass along with...
1730 (pass_data_merge_phi): ...new pass_data instance and...
1731 (make_pass_merge_phi): ...new function.
1732 * tree-complex.c (pass_lower_complex): Convert from a global struct to
1733 a subclass of gimple_opt_pass along with...
1734 (pass_data_lower_complex): ...new pass_data instance and...
1735 (make_pass_lower_complex): ...new function.
1736 (pass_lower_complex_O0): Convert from a global struct to a subclass of
1737 gimple_opt_pass along with...
1738 (pass_data_lower_complex_O0): ...new pass_data instance and...
1739 (make_pass_lower_complex_O0): ...new function.
1740 * tree-eh.c (pass_lower_eh): Convert from a global struct to a
1741 subclass of gimple_opt_pass along with...
1742 (pass_data_lower_eh): ...new pass_data instance and...
1743 (make_pass_lower_eh): ...new function.
1744 (pass_refactor_eh): Convert from a global struct to a subclass of
1745 gimple_opt_pass along with...
1746 (pass_data_refactor_eh): ...new pass_data instance and...
1747 (make_pass_refactor_eh): ...new function.
1748 (pass_lower_resx): Convert from a global struct to a subclass of
1749 gimple_opt_pass along with...
1750 (pass_data_lower_resx): ...new pass_data instance and...
1751 (make_pass_lower_resx): ...new function.
1752 (pass_lower_eh_dispatch): Convert from a global struct to a subclass
1753 of gimple_opt_pass along with...
1754 (pass_data_lower_eh_dispatch): ...new pass_data instance and...
1755 (make_pass_lower_eh_dispatch): ...new function.
1756 (pass_cleanup_eh): Convert from a global struct to a subclass of
1757 gimple_opt_pass along with...
1758 (pass_data_cleanup_eh): ...new pass_data instance and...
1759 (make_pass_cleanup_eh): ...new function.
1760 * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
1761 to a subclass of simple_ipa_opt_pass along with...
1762 (pass_data_ipa_lower_emutls): ...new pass_data instance and...
1763 (make_pass_ipa_lower_emutls): ...new function.
1764 * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
1765 a subclass of gimple_opt_pass along with...
1766 (pass_data_if_conversion): ...new pass_data instance and...
1767 (make_pass_if_conversion): ...new function.
1768 * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
1769 subclass of gimple_opt_pass along with...
1770 (pass_data_build_ssa): ...new pass_data instance and...
1771 (make_pass_build_ssa): ...new function.
1772 * tree-loop-distribution.c (pass_loop_distribution): Convert from a
1773 global struct to a subclass of gimple_opt_pass along with...
1774 (pass_data_loop_distribution): ...new pass_data instance and...
1775 (make_pass_loop_distribution): ...new function.
1776 * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
1777 subclass of gimple_opt_pass along with...
1778 (pass_data_mudflap_1): ...new pass_data instance and...
1779 (make_pass_mudflap_1): ...new function.
1780 (pass_mudflap_2): Convert from a global struct to a subclass of
1781 gimple_opt_pass along with...
1782 (pass_data_mudflap_2): ...new pass_data instance and...
1783 (make_pass_mudflap_2): ...new function.
1784 * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
1785 subclass of gimple_opt_pass along with...
1786 (pass_data_mudflap_1): ...new pass_data instance and...
1787 (make_pass_mudflap_1): ...new function.
1788 (pass_mudflap_2): Convert from a global struct to a subclass of
1789 gimple_opt_pass along with...
1790 (pass_data_mudflap_2): ...new pass_data instance and...
1791 (make_pass_mudflap_2): ...new function.
1792 * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
1793 gimple_opt_pass along with...
1794 (pass_data_nrv): ...new pass_data instance and...
1795 (make_pass_nrv): ...new function.
1796 (pass_return_slot): Convert from a global struct to a subclass of
1797 gimple_opt_pass along with...
1798 (pass_data_return_slot): ...new pass_data instance and...
1799 (make_pass_return_slot): ...new function.
1800 * tree-object-size.c (pass_object_sizes): Convert from a global struct
1801 to a subclass of gimple_opt_pass along with...
1802 (pass_data_object_sizes): ...new pass_data instance and...
1803 (make_pass_object_sizes): ...new function.
1804 * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
1805 global struct to a subclass of gimple_opt_pass along with...
1806 (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
1808 (make_pass_cleanup_cfg_post_optimizing): ...new function.
1809 (pass_fixup_cfg): Convert from a global struct to a subclass of
1810 gimple_opt_pass along with...
1811 (pass_data_fixup_cfg): ...new pass_data instance and...
1812 (make_pass_fixup_cfg): ...new function.
1813 * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
1814 (make_pass_mudflap_1): ...new function.
1815 (pass_mudflap_2): Replace declaration with that of...
1816 (make_pass_mudflap_2): ...new function.
1817 (pass_asan): Replace declaration with that of...
1818 (make_pass_asan): ...new function.
1819 (pass_asan_O0): Replace declaration with that of...
1820 (make_pass_asan_O0): ...new function.
1821 (pass_tsan): Replace declaration with that of...
1822 (make_pass_tsan): ...new function.
1823 (pass_tsan_O0): Replace declaration with that of...
1824 (make_pass_tsan_O0): ...new function.
1825 (pass_lower_cf): Replace declaration with that of...
1826 (make_pass_lower_cf): ...new function.
1827 (pass_refactor_eh): Replace declaration with that of...
1828 (make_pass_refactor_eh): ...new function.
1829 (pass_lower_eh): Replace declaration with that of...
1830 (make_pass_lower_eh): ...new function.
1831 (pass_lower_eh_dispatch): Replace declaration with that of...
1832 (make_pass_lower_eh_dispatch): ...new function.
1833 (pass_lower_resx): Replace declaration with that of...
1834 (make_pass_lower_resx): ...new function.
1835 (pass_build_cfg): Replace declaration with that of...
1836 (make_pass_build_cfg): ...new function.
1837 (pass_early_tree_profile): Replace declaration with that of...
1838 (make_pass_early_tree_profile): ...new function.
1839 (pass_cleanup_eh): Replace declaration with that of...
1840 (make_pass_cleanup_eh): ...new function.
1841 (pass_sra): Replace declaration with that of...
1842 (make_pass_sra): ...new function.
1843 (pass_sra_early): Replace declaration with that of...
1844 (make_pass_sra_early): ...new function.
1845 (pass_early_ipa_sra): Replace declaration with that of...
1846 (make_pass_early_ipa_sra): ...new function.
1847 (pass_tail_recursion): Replace declaration with that of...
1848 (make_pass_tail_recursion): ...new function.
1849 (pass_tail_calls): Replace declaration with that of...
1850 (make_pass_tail_calls): ...new function.
1851 (pass_tree_loop): Replace declaration with that of...
1852 (make_pass_tree_loop): ...new function.
1853 (pass_tree_loop_init): Replace declaration with that of...
1854 (make_pass_tree_loop_init): ...new function.
1855 (pass_lim): Replace declaration with that of...
1856 (make_pass_lim): ...new function.
1857 (pass_tree_unswitch): Replace declaration with that of...
1858 (make_pass_tree_unswitch): ...new function.
1859 (pass_predcom): Replace declaration with that of...
1860 (make_pass_predcom): ...new function.
1861 (pass_iv_canon): Replace declaration with that of...
1862 (make_pass_iv_canon): ...new function.
1863 (pass_scev_cprop): Replace declaration with that of...
1864 (make_pass_scev_cprop): ...new function.
1865 (pass_empty_loop): Replace declaration with that of...
1866 (make_pass_empty_loop): ...new function.
1867 (pass_record_bounds): Replace declaration with that of...
1868 (make_pass_record_bounds): ...new function.
1869 (pass_graphite): Replace declaration with that of...
1870 (make_pass_graphite): ...new function.
1871 (pass_graphite_transforms): Replace declaration with that of...
1872 (make_pass_graphite_transforms): ...new function.
1873 (pass_if_conversion): Replace declaration with that of...
1874 (make_pass_if_conversion): ...new function.
1875 (pass_loop_distribution): Replace declaration with that of...
1876 (make_pass_loop_distribution): ...new function.
1877 (pass_vectorize): Replace declaration with that of...
1878 (make_pass_vectorize): ...new function.
1879 (pass_slp_vectorize): Replace declaration with that of...
1880 (make_pass_slp_vectorize): ...new function.
1881 (pass_complete_unroll): Replace declaration with that of...
1882 (make_pass_complete_unroll): ...new function.
1883 (pass_complete_unrolli): Replace declaration with that of...
1884 (make_pass_complete_unrolli): ...new function.
1885 (pass_parallelize_loops): Replace declaration with that of...
1886 (make_pass_parallelize_loops): ...new function.
1887 (pass_loop_prefetch): Replace declaration with that of...
1888 (make_pass_loop_prefetch): ...new function.
1889 (pass_iv_optimize): Replace declaration with that of...
1890 (make_pass_iv_optimize): ...new function.
1891 (pass_tree_loop_done): Replace declaration with that of...
1892 (make_pass_tree_loop_done): ...new function.
1893 (pass_ch): Replace declaration with that of...
1894 (make_pass_ch): ...new function.
1895 (pass_ccp): Replace declaration with that of...
1896 (make_pass_ccp): ...new function.
1897 (pass_phi_only_cprop): Replace declaration with that of...
1898 (make_pass_phi_only_cprop): ...new function.
1899 (pass_build_ssa): Replace declaration with that of...
1900 (make_pass_build_ssa): ...new function.
1901 (pass_build_alias): Replace declaration with that of...
1902 (make_pass_build_alias): ...new function.
1903 (pass_build_ealias): Replace declaration with that of...
1904 (make_pass_build_ealias): ...new function.
1905 (pass_dominator): Replace declaration with that of...
1906 (make_pass_dominator): ...new function.
1907 (pass_dce): Replace declaration with that of...
1908 (make_pass_dce): ...new function.
1909 (pass_dce_loop): Replace declaration with that of...
1910 (make_pass_dce_loop): ...new function.
1911 (pass_cd_dce): Replace declaration with that of...
1912 (make_pass_cd_dce): ...new function.
1913 (pass_call_cdce): Replace declaration with that of...
1914 (make_pass_call_cdce): ...new function.
1915 (pass_merge_phi): Replace declaration with that of...
1916 (make_pass_merge_phi): ...new function.
1917 (pass_split_crit_edges): Replace declaration with that of...
1918 (make_pass_split_crit_edges): ...new function.
1919 (pass_pre): Replace declaration with that of...
1920 (make_pass_pre): ...new function.
1921 (pass_profile): Replace declaration with that of...
1922 (make_pass_profile): ...new function.
1923 (pass_strip_predict_hints): Replace declaration with that of...
1924 (make_pass_strip_predict_hints): ...new function.
1925 (pass_lower_complex_O0): Replace declaration with that of...
1926 (make_pass_lower_complex_O0): ...new function.
1927 (pass_lower_complex): Replace declaration with that of...
1928 (make_pass_lower_complex): ...new function.
1929 (pass_lower_vector): Replace declaration with that of...
1930 (make_pass_lower_vector): ...new function.
1931 (pass_lower_vector_ssa): Replace declaration with that of...
1932 (make_pass_lower_vector_ssa): ...new function.
1933 (pass_lower_omp): Replace declaration with that of...
1934 (make_pass_lower_omp): ...new function.
1935 (pass_diagnose_omp_blocks): Replace declaration with that of...
1936 (make_pass_diagnose_omp_blocks): ...new function.
1937 (pass_expand_omp): Replace declaration with that of...
1938 (make_pass_expand_omp): ...new function.
1939 (pass_expand_omp_ssa): Replace declaration with that of...
1940 (make_pass_expand_omp_ssa): ...new function.
1941 (pass_object_sizes): Replace declaration with that of...
1942 (make_pass_object_sizes): ...new function.
1943 (pass_strlen): Replace declaration with that of...
1944 (make_pass_strlen): ...new function.
1945 (pass_fold_builtins): Replace declaration with that of...
1946 (make_pass_fold_builtins): ...new function.
1947 (pass_stdarg): Replace declaration with that of...
1948 (make_pass_stdarg): ...new function.
1949 (pass_early_warn_uninitialized): Replace declaration with that of...
1950 (make_pass_early_warn_uninitialized): ...new function.
1951 (pass_late_warn_uninitialized): Replace declaration with that of...
1952 (make_pass_late_warn_uninitialized): ...new function.
1953 (pass_cse_reciprocals): Replace declaration with that of...
1954 (make_pass_cse_reciprocals): ...new function.
1955 (pass_cse_sincos): Replace declaration with that of...
1956 (make_pass_cse_sincos): ...new function.
1957 (pass_optimize_bswap): Replace declaration with that of...
1958 (make_pass_optimize_bswap): ...new function.
1959 (pass_optimize_widening_mul): Replace declaration with that of...
1960 (make_pass_optimize_widening_mul): ...new function.
1961 (pass_warn_function_return): Replace declaration with that of...
1962 (make_pass_warn_function_return): ...new function.
1963 (pass_warn_function_noreturn): Replace declaration with that of...
1964 (make_pass_warn_function_noreturn): ...new function.
1965 (pass_cselim): Replace declaration with that of...
1966 (make_pass_cselim): ...new function.
1967 (pass_phiopt): Replace declaration with that of...
1968 (make_pass_phiopt): ...new function.
1969 (pass_forwprop): Replace declaration with that of...
1970 (make_pass_forwprop): ...new function.
1971 (pass_phiprop): Replace declaration with that of...
1972 (make_pass_phiprop): ...new function.
1973 (pass_tree_ifcombine): Replace declaration with that of...
1974 (make_pass_tree_ifcombine): ...new function.
1975 (pass_dse): Replace declaration with that of...
1976 (make_pass_dse): ...new function.
1977 (pass_nrv): Replace declaration with that of...
1978 (make_pass_nrv): ...new function.
1979 (pass_rename_ssa_copies): Replace declaration with that of...
1980 (make_pass_rename_ssa_copies): ...new function.
1981 (pass_sink_code): Replace declaration with that of...
1982 (make_pass_sink_code): ...new function.
1983 (pass_fre): Replace declaration with that of...
1984 (make_pass_fre): ...new function.
1985 (pass_check_data_deps): Replace declaration with that of...
1986 (make_pass_check_data_deps): ...new function.
1987 (pass_copy_prop): Replace declaration with that of...
1988 (make_pass_copy_prop): ...new function.
1989 (pass_vrp): Replace declaration with that of...
1990 (make_pass_vrp): ...new function.
1991 (pass_uncprop): Replace declaration with that of...
1992 (make_pass_uncprop): ...new function.
1993 (pass_return_slot): Replace declaration with that of...
1994 (make_pass_return_slot): ...new function.
1995 (pass_reassoc): Replace declaration with that of...
1996 (make_pass_reassoc): ...new function.
1997 (pass_rebuild_cgraph_edges): Replace declaration with that of...
1998 (make_pass_rebuild_cgraph_edges): ...new function.
1999 (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2000 (make_pass_remove_cgraph_callee_edges): ...new function.
2001 (pass_build_cgraph_edges): Replace declaration with that of...
2002 (make_pass_build_cgraph_edges): ...new function.
2003 (pass_local_pure_const): Replace declaration with that of...
2004 (make_pass_local_pure_const): ...new function.
2005 (pass_tracer): Replace declaration with that of...
2006 (make_pass_tracer): ...new function.
2007 (pass_warn_unused_result): Replace declaration with that of...
2008 (make_pass_warn_unused_result): ...new function.
2009 (pass_diagnose_tm_blocks): Replace declaration with that of...
2010 (make_pass_diagnose_tm_blocks): ...new function.
2011 (pass_lower_tm): Replace declaration with that of...
2012 (make_pass_lower_tm): ...new function.
2013 (pass_tm_init): Replace declaration with that of...
2014 (make_pass_tm_init): ...new function.
2015 (pass_tm_mark): Replace declaration with that of...
2016 (make_pass_tm_mark): ...new function.
2017 (pass_tm_memopt): Replace declaration with that of...
2018 (make_pass_tm_memopt): ...new function.
2019 (pass_tm_edges): Replace declaration with that of...
2020 (make_pass_tm_edges): ...new function.
2021 (pass_split_functions): Replace declaration with that of...
2022 (make_pass_split_functions): ...new function.
2023 (pass_feedback_split_functions): Replace declaration with that of...
2024 (make_pass_feedback_split_functions): ...new function.
2025 (pass_strength_reduction): Replace declaration with that of...
2026 (make_pass_strength_reduction): ...new function.
2027 (pass_ipa_lower_emutls): Replace declaration with that of...
2028 (make_pass_ipa_lower_emutls): ...new function.
2029 (pass_ipa_function_and_variable_visibility): Replace declaration with
2031 (make_pass_ipa_function_and_variable_visibility): ...new function.
2032 (pass_ipa_tree_profile): Replace declaration with that of...
2033 (make_pass_ipa_tree_profile): ...new function.
2034 (pass_early_local_passes): Replace declaration with that of...
2035 (make_pass_early_local_passes): ...new function.
2036 (pass_ipa_whole_program_visibility): Replace declaration with that
2038 (make_pass_ipa_whole_program_visibility): ...new function.
2039 (pass_ipa_lto_gimple_out): Replace declaration with that of...
2040 (make_pass_ipa_lto_gimple_out): ...new function.
2041 (pass_ipa_increase_alignment): Replace declaration with that of...
2042 (make_pass_ipa_increase_alignment): ...new function.
2043 (pass_ipa_inline): Replace declaration with that of...
2044 (make_pass_ipa_inline): ...new function.
2045 (pass_ipa_free_lang_data): Replace declaration with that of...
2046 (make_pass_ipa_free_lang_data): ...new function.
2047 (pass_ipa_free_inline_summary): Replace declaration with that of...
2048 (make_pass_ipa_free_inline_summary): ...new function.
2049 (pass_ipa_cp): Replace declaration with that of...
2050 (make_pass_ipa_cp): ...new function.
2051 (pass_ipa_reference): Replace declaration with that of...
2052 (make_pass_ipa_reference): ...new function.
2053 (pass_ipa_pure_const): Replace declaration with that of...
2054 (make_pass_ipa_pure_const): ...new function.
2055 (pass_ipa_pta): Replace declaration with that of...
2056 (make_pass_ipa_pta): ...new function.
2057 (pass_ipa_lto_finish_out): Replace declaration with that of...
2058 (make_pass_ipa_lto_finish_out): ...new function.
2059 (pass_ipa_tm): Replace declaration with that of...
2060 (make_pass_ipa_tm): ...new function.
2061 (pass_ipa_profile): Replace declaration with that of...
2062 (make_pass_ipa_profile): ...new function.
2063 (pass_ipa_cdtor_merge): Replace declaration with that of...
2064 (make_pass_ipa_cdtor_merge): ...new function.
2065 (pass_cleanup_cfg_post_optimizing): Replace declaration with that
2067 (make_pass_cleanup_cfg_post_optimizing): ...new function.
2068 (pass_init_datastructures): Replace declaration with that of...
2069 (make_pass_init_datastructures): ...new function.
2070 (pass_fixup_cfg): Replace declaration with that of...
2071 (make_pass_fixup_cfg): ...new function.
2072 (pass_expand): Replace declaration with that of...
2073 (make_pass_expand): ...new function.
2074 (pass_instantiate_virtual_regs): Replace declaration with that of...
2075 (make_pass_instantiate_virtual_regs): ...new function.
2076 (pass_rtl_fwprop): Replace declaration with that of...
2077 (make_pass_rtl_fwprop): ...new function.
2078 (pass_rtl_fwprop_addr): Replace declaration with that of...
2079 (make_pass_rtl_fwprop_addr): ...new function.
2080 (pass_jump): Replace declaration with that of...
2081 (make_pass_jump): ...new function.
2082 (pass_jump2): Replace declaration with that of...
2083 (make_pass_jump2): ...new function.
2084 (pass_lower_subreg): Replace declaration with that of...
2085 (make_pass_lower_subreg): ...new function.
2086 (pass_cse): Replace declaration with that of...
2087 (make_pass_cse): ...new function.
2088 (pass_fast_rtl_dce): Replace declaration with that of...
2089 (make_pass_fast_rtl_dce): ...new function.
2090 (pass_ud_rtl_dce): Replace declaration with that of...
2091 (make_pass_ud_rtl_dce): ...new function.
2092 (pass_rtl_dce): Replace declaration with that of...
2093 (make_pass_rtl_dce): ...new function.
2094 (pass_rtl_dse1): Replace declaration with that of...
2095 (make_pass_rtl_dse1): ...new function.
2096 (pass_rtl_dse2): Replace declaration with that of...
2097 (make_pass_rtl_dse2): ...new function.
2098 (pass_rtl_dse3): Replace declaration with that of...
2099 (make_pass_rtl_dse3): ...new function.
2100 (pass_rtl_cprop): Replace declaration with that of...
2101 (make_pass_rtl_cprop): ...new function.
2102 (pass_rtl_pre): Replace declaration with that of...
2103 (make_pass_rtl_pre): ...new function.
2104 (pass_rtl_hoist): Replace declaration with that of...
2105 (make_pass_rtl_hoist): ...new function.
2106 (pass_rtl_store_motion): Replace declaration with that of...
2107 (make_pass_rtl_store_motion): ...new function.
2108 (pass_cse_after_global_opts): Replace declaration with that of...
2109 (make_pass_cse_after_global_opts): ...new function.
2110 (pass_rtl_ifcvt): Replace declaration with that of...
2111 (make_pass_rtl_ifcvt): ...new function.
2112 (pass_into_cfg_layout_mode): Replace declaration with that of...
2113 (make_pass_into_cfg_layout_mode): ...new function.
2114 (pass_outof_cfg_layout_mode): Replace declaration with that of...
2115 (make_pass_outof_cfg_layout_mode): ...new function.
2116 (pass_loop2): Replace declaration with that of...
2117 (make_pass_loop2): ...new function.
2118 (pass_rtl_loop_init): Replace declaration with that of...
2119 (make_pass_rtl_loop_init): ...new function.
2120 (pass_rtl_move_loop_invariants): Replace declaration with that of...
2121 (make_pass_rtl_move_loop_invariants): ...new function.
2122 (pass_rtl_unswitch): Replace declaration with that of...
2123 (make_pass_rtl_unswitch): ...new function.
2124 (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
2125 (make_pass_rtl_unroll_and_peel_loops): ...new function.
2126 (pass_rtl_doloop): Replace declaration with that of...
2127 (make_pass_rtl_doloop): ...new function.
2128 (pass_rtl_loop_done): Replace declaration with that of...
2129 (make_pass_rtl_loop_done): ...new function.
2130 (pass_web): Replace declaration with that of...
2131 (make_pass_web): ...new function.
2132 (pass_cse2): Replace declaration with that of...
2133 (make_pass_cse2): ...new function.
2134 (pass_df_initialize_opt): Replace declaration with that of...
2135 (make_pass_df_initialize_opt): ...new function.
2136 (pass_df_initialize_no_opt): Replace declaration with that of...
2137 (make_pass_df_initialize_no_opt): ...new function.
2138 (pass_reginfo_init): Replace declaration with that of...
2139 (make_pass_reginfo_init): ...new function.
2140 (pass_inc_dec): Replace declaration with that of...
2141 (make_pass_inc_dec): ...new function.
2142 (pass_stack_ptr_mod): Replace declaration with that of...
2143 (make_pass_stack_ptr_mod): ...new function.
2144 (pass_initialize_regs): Replace declaration with that of...
2145 (make_pass_initialize_regs): ...new function.
2146 (pass_combine): Replace declaration with that of...
2147 (make_pass_combine): ...new function.
2148 (pass_if_after_combine): Replace declaration with that of...
2149 (make_pass_if_after_combine): ...new function.
2150 (pass_ree): Replace declaration with that of...
2151 (make_pass_ree): ...new function.
2152 (pass_partition_blocks): Replace declaration with that of...
2153 (make_pass_partition_blocks): ...new function.
2154 (pass_match_asm_constraints): Replace declaration with that of...
2155 (make_pass_match_asm_constraints): ...new function.
2156 (pass_regmove): Replace declaration with that of...
2157 (make_pass_regmove): ...new function.
2158 (pass_split_all_insns): Replace declaration with that of...
2159 (make_pass_split_all_insns): ...new function.
2160 (pass_fast_rtl_byte_dce): Replace declaration with that of...
2161 (make_pass_fast_rtl_byte_dce): ...new function.
2162 (pass_lower_subreg2): Replace declaration with that of...
2163 (make_pass_lower_subreg2): ...new function.
2164 (pass_mode_switching): Replace declaration with that of...
2165 (make_pass_mode_switching): ...new function.
2166 (pass_sms): Replace declaration with that of...
2167 (make_pass_sms): ...new function.
2168 (pass_sched): Replace declaration with that of...
2169 (make_pass_sched): ...new function.
2170 (pass_ira): Replace declaration with that of...
2171 (make_pass_ira): ...new function.
2172 (pass_reload): Replace declaration with that of...
2173 (make_pass_reload): ...new function.
2174 (pass_clean_state): Replace declaration with that of...
2175 (make_pass_clean_state): ...new function.
2176 (pass_branch_prob): Replace declaration with that of...
2177 (make_pass_branch_prob): ...new function.
2178 (pass_value_profile_transformations): Replace declaration with that
2180 (make_pass_value_profile_transformations): ...new function.
2181 (pass_postreload_cse): Replace declaration with that of...
2182 (make_pass_postreload_cse): ...new function.
2183 (pass_gcse2): Replace declaration with that of...
2184 (make_pass_gcse2): ...new function.
2185 (pass_split_after_reload): Replace declaration with that of...
2186 (make_pass_split_after_reload): ...new function.
2187 (pass_branch_target_load_optimize1): Replace declaration with that
2189 (make_pass_branch_target_load_optimize1): ...new function.
2190 (pass_thread_prologue_and_epilogue): Replace declaration with that
2192 (make_pass_thread_prologue_and_epilogue): ...new function.
2193 (pass_stack_adjustments): Replace declaration with that of...
2194 (make_pass_stack_adjustments): ...new function.
2195 (pass_peephole2): Replace declaration with that of...
2196 (make_pass_peephole2): ...new function.
2197 (pass_if_after_reload): Replace declaration with that of...
2198 (make_pass_if_after_reload): ...new function.
2199 (pass_regrename): Replace declaration with that of...
2200 (make_pass_regrename): ...new function.
2201 (pass_cprop_hardreg): Replace declaration with that of...
2202 (make_pass_cprop_hardreg): ...new function.
2203 (pass_reorder_blocks): Replace declaration with that of...
2204 (make_pass_reorder_blocks): ...new function.
2205 (pass_branch_target_load_optimize2): Replace declaration with that
2207 (make_pass_branch_target_load_optimize2): ...new function.
2208 (pass_leaf_regs): Replace declaration with that of...
2209 (make_pass_leaf_regs): ...new function.
2210 (pass_split_before_sched2): Replace declaration with that of...
2211 (make_pass_split_before_sched2): ...new function.
2212 (pass_compare_elim_after_reload): Replace declaration with that of...
2213 (make_pass_compare_elim_after_reload): ...new function.
2214 (pass_sched2): Replace declaration with that of...
2215 (make_pass_sched2): ...new function.
2216 (pass_stack_regs): Replace declaration with that of...
2217 (make_pass_stack_regs): ...new function.
2218 (pass_stack_regs_run): Replace declaration with that of...
2219 (make_pass_stack_regs_run): ...new function.
2220 (pass_df_finish): Replace declaration with that of...
2221 (make_pass_df_finish): ...new function.
2222 (pass_compute_alignments): Replace declaration with that of...
2223 (make_pass_compute_alignments): ...new function.
2224 (pass_duplicate_computed_gotos): Replace declaration with that of...
2225 (make_pass_duplicate_computed_gotos): ...new function.
2226 (pass_variable_tracking): Replace declaration with that of...
2227 (make_pass_variable_tracking): ...new function.
2228 (pass_free_cfg): Replace declaration with that of...
2229 (make_pass_free_cfg): ...new function.
2230 (pass_machine_reorg): Replace declaration with that of...
2231 (make_pass_machine_reorg): ...new function.
2232 (pass_cleanup_barriers): Replace declaration with that of...
2233 (make_pass_cleanup_barriers): ...new function.
2234 (pass_delay_slots): Replace declaration with that of...
2235 (make_pass_delay_slots): ...new function.
2236 (pass_split_for_shorten_branches): Replace declaration with that of...
2237 (make_pass_split_for_shorten_branches): ...new function.
2238 (pass_split_before_regstack): Replace declaration with that of...
2239 (make_pass_split_before_regstack): ...new function.
2240 (pass_convert_to_eh_region_ranges): Replace declaration with that
2242 (make_pass_convert_to_eh_region_ranges): ...new function.
2243 (pass_shorten_branches): Replace declaration with that of...
2244 (make_pass_shorten_branches): ...new function.
2245 (pass_set_nothrow_function_flags): Replace declaration with that of...
2246 (make_pass_set_nothrow_function_flags): ...new function.
2247 (pass_dwarf2_frame): Replace declaration with that of...
2248 (make_pass_dwarf2_frame): ...new function.
2249 (pass_final): Replace declaration with that of...
2250 (make_pass_final): ...new function.
2251 (pass_rtl_seqabstr): Replace declaration with that of...
2252 (make_pass_rtl_seqabstr): ...new function.
2253 (pass_release_ssa_names): Replace declaration with that of...
2254 (make_pass_release_ssa_names): ...new function.
2255 (pass_early_inline): Replace declaration with that of...
2256 (make_pass_early_inline): ...new function.
2257 (pass_inline_parameters): Replace declaration with that of...
2258 (make_pass_inline_parameters): ...new function.
2259 (pass_update_address_taken): Replace declaration with that of...
2260 (make_pass_update_address_taken): ...new function.
2261 (pass_convert_switch): Replace declaration with that of...
2262 (make_pass_convert_switch): ...new function.
2263 * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
2264 to a subclass of simple_ipa_opt_pass along with...
2265 (pass_data_ipa_tree_profile): ...new pass_data instance and...
2266 (make_pass_ipa_tree_profile): ...new function.
2267 * tree-sra.c (pass_sra_early): Convert from a global struct to a
2268 subclass of gimple_opt_pass along with...
2269 (pass_data_sra_early): ...new pass_data instance and...
2270 (make_pass_sra_early): ...new function.
2271 (pass_sra): Convert from a global struct to a subclass of
2272 gimple_opt_pass along with...
2273 (pass_data_sra): ...new pass_data instance and...
2274 (make_pass_sra): ...new function.
2275 (pass_early_ipa_sra): Convert from a global struct to a subclass of
2276 gimple_opt_pass along with...
2277 (pass_data_early_ipa_sra): ...new pass_data instance and...
2278 (make_pass_early_ipa_sra): ...new function.
2279 * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
2280 subclass of gimple_opt_pass along with...
2281 (pass_data_ccp): ...new pass_data instance and...
2282 (make_pass_ccp): ...new function.
2283 (pass_fold_builtins): Convert from a global struct to a subclass of
2284 gimple_opt_pass along with...
2285 (pass_data_fold_builtins): ...new pass_data instance and...
2286 (make_pass_fold_builtins): ...new function.
2287 * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
2288 subclass of gimple_opt_pass along with...
2289 (pass_data_copy_prop): ...new pass_data instance and...
2290 (make_pass_copy_prop): ...new function.
2291 * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
2292 global struct to a subclass of gimple_opt_pass along with...
2293 (pass_data_rename_ssa_copies): ...new pass_data instance and...
2294 (make_pass_rename_ssa_copies): ...new function.
2295 * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
2296 subclass of gimple_opt_pass along with...
2297 (pass_data_dce): ...new pass_data instance and...
2298 (make_pass_dce): ...new function.
2299 (pass_dce_loop): Convert from a global struct to a subclass of
2300 gimple_opt_pass along with...
2301 (pass_data_dce_loop): ...new pass_data instance and...
2302 (make_pass_dce_loop): ...new function.
2303 (pass_cd_dce): Convert from a global struct to a subclass of
2304 gimple_opt_pass along with...
2305 (pass_data_cd_dce): ...new pass_data instance and...
2306 (make_pass_cd_dce): ...new function.
2307 * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
2308 subclass of gimple_opt_pass along with...
2309 (pass_data_dominator): ...new pass_data instance and...
2310 (make_pass_dominator): ...new function.
2311 (pass_phi_only_cprop): Convert from a global struct to a subclass of
2312 gimple_opt_pass along with...
2313 (pass_data_phi_only_cprop): ...new pass_data instance and...
2314 (make_pass_phi_only_cprop): ...new function.
2315 * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
2316 subclass of gimple_opt_pass along with...
2317 (pass_data_dse): ...new pass_data instance and...
2318 (make_pass_dse): ...new function.
2319 * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
2320 a subclass of gimple_opt_pass along with...
2321 (pass_data_forwprop): ...new pass_data instance and...
2322 (make_pass_forwprop): ...new function.
2323 * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
2324 struct to a subclass of gimple_opt_pass along with...
2325 (pass_data_tree_ifcombine): ...new pass_data instance and...
2326 (make_pass_tree_ifcombine): ...new function.
2327 * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
2328 subclass of gimple_opt_pass along with...
2329 (pass_data_ch): ...new pass_data instance and...
2330 (make_pass_ch): ...new function.
2331 * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
2332 subclass of gimple_opt_pass along with...
2333 (pass_data_tree_loop): ...new pass_data instance and...
2334 (make_pass_tree_loop): ...new function.
2335 (pass_tree_loop_init): Convert from a global struct to a subclass of
2336 gimple_opt_pass along with...
2337 (pass_data_tree_loop_init): ...new pass_data instance and...
2338 (make_pass_tree_loop_init): ...new function.
2339 (pass_lim): Convert from a global struct to a subclass of
2340 gimple_opt_pass along with...
2341 (pass_data_lim): ...new pass_data instance and...
2342 (make_pass_lim): ...new function.
2343 (pass_tree_unswitch): Convert from a global struct to a subclass of
2344 gimple_opt_pass along with...
2345 (pass_data_tree_unswitch): ...new pass_data instance and...
2346 (make_pass_tree_unswitch): ...new function.
2347 (pass_predcom): Convert from a global struct to a subclass of
2348 gimple_opt_pass along with...
2349 (pass_data_predcom): ...new pass_data instance and...
2350 (make_pass_predcom): ...new function.
2351 (pass_vectorize): Convert from a global struct to a subclass of
2352 gimple_opt_pass along with...
2353 (pass_data_vectorize): ...new pass_data instance and...
2354 (make_pass_vectorize): ...new function.
2355 (pass_graphite): Convert from a global struct to a subclass of
2356 gimple_opt_pass along with...
2357 (pass_data_graphite): ...new pass_data instance and...
2358 (make_pass_graphite): ...new function.
2359 (pass_graphite_transforms): Convert from a global struct to a subclass
2360 of gimple_opt_pass along with...
2361 (pass_data_graphite_transforms): ...new pass_data instance and...
2362 (make_pass_graphite_transforms): ...new function.
2363 (pass_check_data_deps): Convert from a global struct to a subclass of
2364 gimple_opt_pass along with...
2365 (pass_data_check_data_deps): ...new pass_data instance and...
2366 (make_pass_check_data_deps): ...new function.
2367 (pass_iv_canon): Convert from a global struct to a subclass of
2368 gimple_opt_pass along with...
2369 (pass_data_iv_canon): ...new pass_data instance and...
2370 (make_pass_iv_canon): ...new function.
2371 (pass_scev_cprop): Convert from a global struct to a subclass of
2372 gimple_opt_pass along with...
2373 (pass_data_scev_cprop): ...new pass_data instance and...
2374 (make_pass_scev_cprop): ...new function.
2375 (pass_record_bounds): Convert from a global struct to a subclass of
2376 gimple_opt_pass along with...
2377 (pass_data_record_bounds): ...new pass_data instance and...
2378 (make_pass_record_bounds): ...new function.
2379 (pass_complete_unroll): Convert from a global struct to a subclass of
2380 gimple_opt_pass along with...
2381 (pass_data_complete_unroll): ...new pass_data instance and...
2382 (make_pass_complete_unroll): ...new function.
2383 (pass_complete_unrolli): Convert from a global struct to a subclass of
2384 gimple_opt_pass along with...
2385 (pass_data_complete_unrolli): ...new pass_data instance and...
2386 (make_pass_complete_unrolli): ...new function.
2387 (pass_parallelize_loops): Convert from a global struct to a subclass
2388 of gimple_opt_pass along with...
2389 (pass_data_parallelize_loops): ...new pass_data instance and...
2390 (make_pass_parallelize_loops): ...new function.
2391 (pass_loop_prefetch): Convert from a global struct to a subclass of
2392 gimple_opt_pass along with...
2393 (pass_data_loop_prefetch): ...new pass_data instance and...
2394 (make_pass_loop_prefetch): ...new function.
2395 (pass_iv_optimize): Convert from a global struct to a subclass of
2396 gimple_opt_pass along with...
2397 (pass_data_iv_optimize): ...new pass_data instance and...
2398 (make_pass_iv_optimize): ...new function.
2399 (pass_tree_loop_done): Convert from a global struct to a subclass of
2400 gimple_opt_pass along with...
2401 (pass_data_tree_loop_done): ...new pass_data instance and...
2402 (make_pass_tree_loop_done): ...new function.
2403 * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
2404 struct to a subclass of gimple_opt_pass along with...
2405 (pass_data_cse_reciprocals): ...new pass_data instance and...
2406 (make_pass_cse_reciprocals): ...new function.
2407 (pass_cse_sincos): Convert from a global struct to a subclass of
2408 gimple_opt_pass along with...
2409 (pass_data_cse_sincos): ...new pass_data instance and...
2410 (make_pass_cse_sincos): ...new function.
2411 (pass_optimize_bswap): Convert from a global struct to a subclass of
2412 gimple_opt_pass along with...
2413 (pass_data_optimize_bswap): ...new pass_data instance and...
2414 (make_pass_optimize_bswap): ...new function.
2415 (pass_optimize_widening_mul): Convert from a global struct to a
2416 subclass of gimple_opt_pass along with...
2417 (pass_data_optimize_widening_mul): ...new pass_data instance and...
2418 (make_pass_optimize_widening_mul): ...new function.
2419 * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
2420 subclass of gimple_opt_pass along with...
2421 (pass_data_phiopt): ...new pass_data instance and...
2422 (make_pass_phiopt): ...new function.
2423 (pass_cselim): Convert from a global struct to a subclass of
2424 gimple_opt_pass along with...
2425 (pass_data_cselim): ...new pass_data instance and...
2426 (make_pass_cselim): ...new function.
2427 * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
2428 subclass of gimple_opt_pass along with...
2429 (pass_data_phiprop): ...new pass_data instance and...
2430 (make_pass_phiprop): ...new function.
2431 * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
2432 subclass of gimple_opt_pass along with...
2433 (pass_data_pre): ...new pass_data instance and...
2434 (make_pass_pre): ...new function.
2435 (pass_fre): Convert from a global struct to a subclass of
2436 gimple_opt_pass along with...
2437 (pass_data_fre): ...new pass_data instance and...
2438 (make_pass_fre): ...new function.
2439 * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
2440 subclass of gimple_opt_pass along with...
2441 (pass_data_reassoc): ...new pass_data instance and...
2442 (make_pass_reassoc): ...new function.
2443 * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
2444 subclass of gimple_opt_pass along with...
2445 (pass_data_sink_code): ...new pass_data instance and...
2446 (make_pass_sink_code): ...new function.
2447 * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
2448 subclass of gimple_opt_pass along with...
2449 (pass_data_strlen): ...new pass_data instance and...
2450 (make_pass_strlen): ...new function.
2451 * tree-ssa-structalias.c (pass_build_alias): Convert from a global
2452 struct to a subclass of gimple_opt_pass along with...
2453 (pass_data_build_alias): ...new pass_data instance and...
2454 (make_pass_build_alias): ...new function.
2455 (pass_build_ealias): Convert from a global struct to a subclass of
2456 gimple_opt_pass along with...
2457 (pass_data_build_ealias): ...new pass_data instance and...
2458 (make_pass_build_ealias): ...new function.
2459 (pass_ipa_pta): Convert from a global struct to a subclass of
2460 simple_ipa_opt_pass along with...
2461 (pass_data_ipa_pta): ...new pass_data instance and...
2462 (make_pass_ipa_pta): ...new function.
2463 * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
2464 subclass of gimple_opt_pass along with...
2465 (pass_data_uncprop): ...new pass_data instance and...
2466 (make_pass_uncprop): ...new function.
2467 * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
2468 global struct to a subclass of gimple_opt_pass along with...
2469 (pass_data_late_warn_uninitialized): ...new pass_data instance and...
2470 (make_pass_late_warn_uninitialized): ...new function.
2471 * tree-ssa.c (pass_init_datastructures): Convert from a global struct
2472 to a subclass of gimple_opt_pass along with...
2473 (pass_data_init_datastructures): ...new pass_data instance and...
2474 (make_pass_init_datastructures): ...new function.
2475 (pass_early_warn_uninitialized): Convert from a global struct to a
2476 subclass of gimple_opt_pass along with...
2477 (pass_data_early_warn_uninitialized): ...new pass_data instance and...
2478 (make_pass_early_warn_uninitialized): ...new function.
2479 (pass_update_address_taken): Convert from a global struct to a
2480 subclass of gimple_opt_pass along with...
2481 (pass_data_update_address_taken): ...new pass_data instance and...
2482 (make_pass_update_address_taken): ...new function.
2483 * tree-ssanames.c (pass_release_ssa_names): Convert from a global
2484 struct to a subclass of gimple_opt_pass along with...
2485 (pass_data_release_ssa_names): ...new pass_data instance and...
2486 (make_pass_release_ssa_names): ...new function.
2487 * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
2488 subclass of gimple_opt_pass along with...
2489 (pass_data_stdarg): ...new pass_data instance and...
2490 (make_pass_stdarg): ...new function.
2491 * tree-switch-conversion.c (pass_convert_switch): Convert from a
2492 global struct to a subclass of gimple_opt_pass along with...
2493 (pass_data_convert_switch): ...new pass_data instance and...
2494 (make_pass_convert_switch): ...new function.
2495 * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
2496 to a subclass of gimple_opt_pass along with...
2497 (pass_data_tail_recursion): ...new pass_data instance and...
2498 (make_pass_tail_recursion): ...new function.
2499 (pass_tail_calls): Convert from a global struct to a subclass of
2500 gimple_opt_pass along with...
2501 (pass_data_tail_calls): ...new pass_data instance and...
2502 (make_pass_tail_calls): ...new function.
2503 * tree-vect-generic.c (pass_lower_vector): Convert from a global
2504 struct to a subclass of gimple_opt_pass along with...
2505 (pass_data_lower_vector): ...new pass_data instance and...
2506 (make_pass_lower_vector): ...new function.
2507 (pass_lower_vector_ssa): Convert from a global struct to a subclass of
2508 gimple_opt_pass along with...
2509 (pass_data_lower_vector_ssa): ...new pass_data instance and...
2510 (make_pass_lower_vector_ssa): ...new function.
2511 * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
2512 to a subclass of gimple_opt_pass along with...
2513 (pass_data_slp_vectorize): ...new pass_data instance and...
2514 (make_pass_slp_vectorize): ...new function.
2515 (pass_ipa_increase_alignment): Convert from a global struct to a
2516 subclass of simple_ipa_opt_pass along with...
2517 (pass_data_ipa_increase_alignment): ...new pass_data instance and...
2518 (make_pass_ipa_increase_alignment): ...new function.
2519 * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
2520 gimple_opt_pass along with...
2521 (pass_data_vrp): ...new pass_data instance and...
2522 (make_pass_vrp): ...new function.
2523 * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
2524 subclass of simple_ipa_opt_pass along with...
2525 (pass_data_ipa_free_lang_data): ...new pass_data instance and...
2526 (make_pass_ipa_free_lang_data): ...new function.
2527 * tsan.c (pass_tsan): Convert from a global struct to a subclass of
2528 gimple_opt_pass along with...
2529 (pass_data_tsan): ...new pass_data instance and...
2530 (make_pass_tsan): ...new function.
2531 (pass_tsan_O0): Convert from a global struct to a subclass of
2532 gimple_opt_pass along with...
2533 (pass_data_tsan_O0): ...new pass_data instance and...
2534 (make_pass_tsan_O0): ...new function.
2535 * var-tracking.c (pass_variable_tracking): Convert from a global
2536 struct to a subclass of rtl_opt_pass along with...
2537 (pass_data_variable_tracking): ...new pass_data instance and...
2538 (make_pass_variable_tracking): ...new function.
2539 * web.c (pass_web): Convert from a global struct to a subclass of
2540 rtl_opt_pass along with...
2541 (pass_data_web): ...new pass_data instance and...
2542 (make_pass_web): ...new function.
2543 * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
2544 declaration with that of...
2545 (make_pass_mode_switch_use): ...new function.
2546 (pass_resolve_sw_modes): Replace declaration with that of...
2547 (make_pass_resolve_sw_modes): ...new function.
2548 * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
2549 from a global struct to a subclass of rtl_opt_pass along with...
2550 (pass_data_mode_switch_use): ...new pass_data instance and...
2551 (make_pass_mode_switch_use): ...new function.
2552 * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
2553 from a global struct to a subclass of rtl_opt_pass along with...
2554 (pass_data_resolve_sw_modes): ...new pass_data instance and...
2555 (make_pass_resolve_sw_modes): ...new function.
2556 * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
2557 struct to a subclass of rtl_opt_pass along with...
2558 (pass_data_insert_vzeroupper): ...new pass_data instance and...
2559 (make_pass_insert_vzeroupper): ...new function.
2560 * config/sparc/sparc.c (pass_work_around_errata): Convert from a
2561 global struct to a subclass of rtl_opt_pass along with...
2562 (pass_data_work_around_errata): ...new pass_data instance and...
2563 (make_pass_work_around_errata): ...new function.
2564 * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
2565 struct to a subclass of rtl_opt_pass along with...
2566 (pass_data_mips_machine_reorg2): ...new pass_data instance and...
2567 (make_pass_mips_machine_reorg2): ...new function.
2569 2013-08-05 David Malcolm <dmalcolm@redhat.com>
2571 * passes.c (pass_manager::operator new): New.
2573 2013-08-05 David Malcolm <dmalcolm@redhat.com>
2575 Handwritten part of conversion of passes to C++ classes.
2577 * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
2578 (toplev.o): Add dep on PASS_MANAGER_H.
2579 * cgraphunit.c (cgraph_process_new_functions): Rework invocation
2580 of early local pases to reflect this moving from a global to a
2581 member of gcc::pass_manager.
2582 (cgraph_add_new_function): Likewise.
2583 * lto-cgraph.c (lto_output_node): Update for conversion of
2584 struct ipa_opt_pass_d to a C++ subclass of opt_pass.
2585 * passes.c (opt_pass::clone): New.
2586 (opt_pass::gate): New.
2587 (opt_pass::execute): New.
2588 (opt_pass::opt_pass): New.
2589 (pass_manager::execute_early_local_passes): New.
2590 (pass_manager::execute_pass_mode_switching): new.
2591 (finish_optimization_passes): Convert to...
2592 (pass_manager::finish_optimization_passes): ...this.
2593 (finish_optimization_passes): Update for conversion of passes to
2595 (register_dump_files_1): Use has_gate since we cannot portably
2596 check a vtable entry against NULL.
2597 (dump_one_pass): Likewise.
2598 (ipa_write_summaries_2): Likewise.
2599 (ipa_write_optimization_summaries_1): Likewise.
2600 (ipa_read_summaries_1): Likewise.
2601 (ipa_read_optimization_summaries_1): Likewise.
2602 (execute_ipa_stmt_fixups): Likewise.
2603 (pass_manager::pass_manager): Rewrite pass-creation, invoking
2604 pass-creation functions rather than wiring up globals, and
2605 storing the results in fields of pass_manager generated using
2607 (pass_manager::dump_profile_report): Update for conversion of
2608 passes to C++ classes.
2609 (pass_manager::execute_ipa_summary_passes): Likewise.
2610 (execute_one_ipa_transform_pass): Likewise.
2611 (execute_one_pass): Use has_gate and has_execute since we cannot
2612 portably check a vtable entry against NULL.
2613 * pass_manager.h (pass_manager::finish_optimization_passes): New.
2614 (pass_manager): Use pass-instances.def to add fields for the
2615 various pass instances.
2616 * toplev.c (finalize): Update for move of
2617 finish_optimization_passes to a method of gcc::pass_manager.
2618 * toplev.h (finish_optimization_passes): Move to method of class
2620 * tree-pass.h (struct pass_data): New.
2621 (opt_pass): Convert to C++ class, make it a subclass of pass_data.
2622 (opt_pass::gate): Convert to virtual function.
2623 (opt_pass::~opt_pass): New.
2624 (opt_pass::clone): New.
2625 (opt_pass::execute): Convert to virtual function.
2626 (opt_pass::opt_pass): New.
2627 (opt_pass::ctxt_): new.
2628 (gimple_opt_pass): Convert to subclass of opt_pass.
2629 (gimple_opt_pass::gimple_opt_pass): New.
2630 (rtl_opt_pass): Convert to subclass of opt_pass.
2631 (rtl_opt_pass::rtl_opt_pass): New.
2632 (ipa_opt_pass_d): Convert to subclass of opt_pass.
2633 (ipa_opt_pass_d::ipa_opt_pass_d): New.
2634 (simple_ipa_opt_pass): Convert to subclass of opt_pass.
2635 (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
2636 * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
2637 invocation of pass_mode_switching to reflect this moving from a
2638 global to a member of gcc::pass_manager.
2639 (ix86_option_override): Rework how pass_insert_vzeroupper is
2640 added to the pass_manager to reflect autogenerated changes.
2641 * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
2643 2013-08-05 Richard Earnshaw <rearnsha@arm.com>
2645 PR rtl-optimization/57708
2646 * recog.c (peep2_find_free_register): Validate all regs in a
2649 2013-08-05 Jan Hubicka <jh@suse.cz>
2652 * cgraph.c (verify_cgraph_node): Accept local flags from other
2654 * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
2655 (function_and_variable_visibility): Likewise.
2656 * trans-mem.c (ipa_tm_create_version): TM versions are not local.
2658 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
2660 * graph.c (init_graph_slim_pretty_print): Remove.
2661 (print_graph_cfg): Do not call it. Use local pretty printer.
2662 (start_graph_dump): Likewise.
2664 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
2666 * gimple-pretty-print.c (buffer): Remove.
2667 (initialized): Likewise.
2668 (maybe_init_pretty_print): Likewise.
2669 (print_gimple_stmt): Do not call it. Use non-static local
2670 pretty_printer variable.
2671 (print_gimple_expr): Likewise.
2672 (print_gimple_seq): Likewise.
2673 (gimple_dump_bb): Likewise.
2675 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
2677 * asan.c (asan_pp): Remove.
2678 (asan_pp_initialized): Likewise.
2679 (asan_pp_initialize): Likewise.
2680 (asan_pp_string): Take a pretty_printer parameter. Adjust callers.
2681 (asan_emit_stack_protection): Tidy. Use local pretty printer.
2682 (asan_add_global): Likewise.
2684 2013-08-04 Gabriel Dos Reis <gdr@integrable-solutions.net>
2686 * pretty-print.h (pp_base): Remove. Adjust dependent macros.
2687 * diagnostic.h (diagnostic_flush_buffer): Adjust.
2688 * pretty-print.c (pp_formatted_text_data): Likewise.
2689 (pp_indent): Rename from pp_base_indent.
2690 (pp_format): Rename from pp_base_format.
2691 (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
2692 (pp_format_verbatim): Rename from pp_base_format_verbatim.
2693 (pp_flush): Rename from pp_base_flush.
2694 (pp_set_line_maximum_length): Rename from
2695 pp_base_set_line_maximum_length.
2696 (pp_clear_output_area): Rename from pp_base_clear_output_area.
2697 (pp_set_prefix): Rename from pp_base_set_prefix.
2698 (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
2699 (pp_emit_prefix): Rename from pp_base_emit_prefix.
2700 (pp_append_text): Rename from pp_base_append_text.
2701 (pp_formatted_text): Rename from pp_base_formatted_text.
2702 (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
2703 (pp_remaining_character_count_for_line): Rename from
2704 pp_base_remaining_character_count_for_line.
2705 (pp_newline): Rename from pp_base_newline.
2706 (pp_character): Rename from pp_base_character.
2707 (pp_string): Rename from pp_base_string.
2708 (pp_maybe_space): Rename from pp_base_maybe_space.
2709 * asan.c (asan_pp_string): Adjust.
2710 (asan_emit_stack_protection): Likewise.
2711 (asan_add_global): Likewise.
2712 * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
2713 * tree-mudflap.c (mf_varname_tree): Likewise.
2714 * tree-pretty-print.c (pp_tree_identifier): Rename from
2715 pp_base_tree_identifier.
2716 * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
2717 Declare as function.
2719 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
2721 * pretty-print.h (pp_bar_bar): New.
2722 (pp_ampersand_ampersand): Likewise.
2723 (pp_less_equal): Likewise.
2724 (pp_greater_equal): Likewise.
2725 * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
2726 printer functions instead of pp_string or operators and punctuators.
2727 (dump_gimple_call): Likewise.
2728 (dump_gimple_omp_for): Likewise.
2729 (dump_gimple_transaction): Likewise.
2730 (dump_gimple_phi): Likewise.
2731 (pp_gimple_stmt_1): Likewise.
2732 * sched-vis.c (print_insn): Likewise.
2733 * tree-mudflap.c (mf_varname_tree): Likewise.
2734 * tree-pretty-print.c (dump_block_node): Likewise.
2735 (dump_generic_node): Likewise.
2737 2013-08-02 Jan Hubicka <jh@suse.cz>
2739 * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
2741 * lto-streamer-out.c (tree_is_indexable): Results decls and
2742 parm decls are not indexable.
2743 (DFS_write_tree_body): Do not follow args and results.
2744 (hash_tree): Likewise.
2745 (output_functions): Rearrange so struct function is needed
2746 only when real body is output; be able to also ouptut abstract
2747 functions; output DECL_ARGUMENTS and DECL_RESULT.
2748 (lto_output): When not in WPA, ale store abstract functions.
2749 (write_symbol): Do not care about RESULT_DECL.
2750 (output_symbol_p): Handle correctly sbtract decls.
2751 * lto-streamer-in.c (input_function): Rearrange so struct
2752 function can be NULL at entry; allow streaming of
2753 functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
2754 * ipa.c (symtab_remove_unreachable_nodes): Silence confused
2755 sanity check during LTO.
2756 * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
2757 RESULT_DECl and DECL_ARGUMENTS.
2758 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
2761 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
2763 * pretty-print.h (pp_underscore): New.
2765 * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
2766 printer functions instead of pp_character.
2767 (dump_binary_rhs): Likewise.
2768 (dump_ternary_rhs): Likewise.
2769 (dump_gimple_call_args): Likewise.
2770 (pp_points_to_solution): Likewise.
2771 (dump_gimple_call): Likewise.
2772 (dump_gimple_switch): Likewise.
2773 (dump_gimple_cond): Likewise.
2774 (dump_gimple_bind): Likewise.
2775 (dump_gimple_try): Likewise.
2776 (dump_gimple_omp_for): Likewise.
2777 (dump_gimple_omp_continue): Likewise.
2778 (dump_gimple_omp_single): Likewise.
2779 (dump_gimple_omp_sections): Likewise.
2780 (dump_gimple_omp_block): Likewise.
2781 (dump_gimple_omp_critical): Likewise.
2782 (dump_gimple_transaction): Likewise.
2783 (dump_gimple_asm): Likewise.
2784 (dump_gimple_phi): Likewise.
2785 (dump_gimple_omp_parallel): Likewise.
2786 (dump_gimple_omp_task): Likewise.
2787 (dump_gimple_omp_atomic_load): Likewise.
2788 (dump_gimple_omp_atomic_store): Likewise.
2789 (dump_gimple_mem_ops): Likewise.
2790 (pp_gimple_stmt_1): Likewise.
2791 (pp_cfg_jump): Likewise.
2792 (dump_implicit_edges): Likewise.
2793 (gimple_dump_bb_for_graph): Likewise.
2794 * graph.c (draw_cfg_node): Likewise.
2795 * langhooks.c (lhd_print_error_function): Likewise.
2796 * sched-vis.c (print_exp): Likewise.
2797 (print_value): Likewise.
2798 (print_pattern): Likewise.
2799 (print_insn): Likewise.
2800 (rtl_dump_bb_for_graph): Likewise.
2801 * tree-pretty-print.c (dump_function_declaration): Likewise.
2802 (dump_array_domain): Likewise.
2803 (dump_omp_clause): Likewise.
2804 (dump_location): Likewise.
2805 (dump_generic_node): Likewise.
2806 (print_struct_decl): Likewise.
2807 * diagnostic.c (diagnostic_show_locus): Use pp_space.
2809 2013-08-03 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2811 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
2812 candidate table when replacing a candidate statement.
2813 (replace_rhs_if_not_dup): Likewise.
2814 (replace_one_candidate): Likewise.
2816 2013-08-02 Jan Hubicka <jh@suse.cz>
2817 Martin Liska <marxin.liska@gmail.com>
2819 * cgraphunit.c (add_new_function): Fix logic when adding from
2821 (assemble_thunk): Rename to ...
2822 (expand_thunk); .. this one; export; get it working with
2823 general functions; make produced gimple valid.
2824 * cgraph.h (expand_thunk): Declare.
2826 2013-08-02 Jan Hubicka <jh@suse.cz>
2828 * ipa-cp.c (gather_context_independent_values): Use
2829 ipa_get_param_move_cost.
2830 (get_replacement_map): Remove PARAM; move parameter folding
2832 (create_specialized_node): Update.
2833 * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
2834 assert that we have gimple body; update move_cost.
2835 (count_formal_params): Assert that we have gimple body.
2836 (ipa_dump_param): New function.
2837 (ipa_alloc_node_params): Break out from ...
2838 (ipa_initialize_node_params): ... here.
2839 (ipa_get_vector_of_formal_parms): ICE when used in WPA.
2840 (ipa_write_node_info): Stream move costs.
2841 (ipa_read_node_info): Read move costs.
2842 (ipa_update_after_lto_read): Do not recompute node params.
2843 * ipa-prop.h (ipa_param_descriptor): Add move_cost.
2844 (ipa_get_param): Check we are not in WPA.
2845 (ipa_get_param_move_cost): New.
2846 * tree-inline.c (tree_function_versioning): Fold replacement as needed.
2847 * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
2848 parm numbers to be present.
2850 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
2852 PR rtl-optimization/58048
2853 * lra-constraints.c (process_alt_operands): Don't check asm
2854 operand on register.
2856 2013-08-02 Eric Botcazou <ebotcazou@adacore.com>
2858 * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
2859 the implied StoreLoad barrier for atomic operations if before.
2861 2013-08-02 Jan Hubicka <jh@suse.cz>
2862 Martin Liska <marxin.liska@gmail.com>
2864 * cgraph.c (cgraph_function_body_availability): Do not check
2866 * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
2867 symtab_node_availability): Declare.
2868 * ipa.c (can_replace_by_local_alias): New.
2869 (function_and_variable_visibility): Use it.
2870 * symtab.c (symtab_for_node_and_aliases,
2871 symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
2873 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
2875 PR rtl-optimization/57963
2876 * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
2877 (lra_constraints): Use them.
2879 2013-08-02 Sofiane Naci <sofiane.naci@arm.com>
2881 * config/arm/types.md (define_attr "type"): Add "load_acq"
2883 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
2885 (cortex_a53_store1): Likewise.
2887 2013-08-01 Jan Hubicka <jh@suse.cz>
2889 * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
2890 partitions are not needed.
2892 2013-08-01 Uros Bizjak <ubizjak@gmail.com>
2894 * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
2895 * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
2896 MAYBE_NON_Q_CLASS_P where appropriate.
2898 2013-08-01 Jan Hubicka <jh@suse.cz>
2900 * cgraph.h (release_function_body): Declare.
2901 * tree.c (free_lang_data_in_decl): Free, parameters and return values
2902 of unused delcarations.
2904 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2906 * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
2907 RTL form when subtracting a constant.
2909 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2911 * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
2912 Generate canonical plus rtx with negated immediate instead of minus
2914 * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
2916 2013-08-01 Jan Hubicka <jh@suse.cz>
2918 * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
2919 (cgraph_release_function_body): Likewise.
2920 (cgraph_can_remove_if_no_direct_calls_p): Likewise.
2921 * cgraph.h (cgrpah_node): Rename abstract_and_needed
2922 to used_as_abstract_origin.
2923 * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
2924 symbols used as abstract origins.
2925 * cgraphunit.c (analyze_functions): Update.
2926 * ipa.c (symtab_remove_unreachable_nodes): Recompute
2927 used_as_abstract_origin.
2928 * tree-inline.c (tree_function_versioning): Update
2929 used_as_abstract_origin; be ready for DECL_RESULT and
2930 DECL_ARGUMENTS to be NULL.
2932 * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
2933 for abstract functions.
2934 * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
2937 2013-08-01 Jan Hubicka <jh@suse.cz>
2939 * profile.c (compute_value_histograms): Fix thinko.
2941 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
2943 * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs. Add
2944 aarch-common-protos.h to extra_headers.
2945 (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2946 * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
2947 * config/aarch64/t-aarch64 (aarch-common.o): Define.
2949 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
2951 * config/aarch64/aarch64.md (define_attr "type"): Delete.
2952 Include "../arm/types.md". Define "type" attribute for all patterns.
2953 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
2956 2013-07-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2958 * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
2959 to support power8 load fusion.
2960 (fusion_gpr_mem_load): Likewise.
2962 * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
2964 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
2965 declarations for power8 load fusion.
2966 (emit_fusion_gpr_load): Likewise.
2968 * config/rs6000/rs6000.c (rs6000_option_override_internal): If
2969 tuning for power8, turn on fusion mode by default. Turn on sign
2970 extending fusion mode if normal fusion mode is on, and we are at
2972 (fusion_gpr_load_p): New function, return true if we can fuse an
2973 addis instruction with a dependent load to a GPR.
2974 (emit_fusion_gpr_load): Emit the instructions for power8 load
2977 * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
2978 (VSX load fusion peepholes): New peepholes to fuse together an
2979 addi instruction with a VSX load instruction.
2981 * config/rs6000/rs6000.md (GPR load fusion peepholes): New
2982 peepholes to fuse an addis instruction with a load to a GPR base
2983 register. If we are supporting sign extending fusions, convert
2984 sign extending loads to zero extending loads and add an explicit
2987 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
2989 * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs. Add
2990 aarch-common-protos.h to extra_headers.
2991 (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
2992 * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
2993 (arm_early_store_addr_dep): Likewise.
2994 (arm_no_early_alu_shift_dep): Likewise.
2995 (arm_no_early_alu_shift_value_dep): Likewise.
2996 (arm_no_early_mul_dep): Likewise.
2997 (arm_no_early_store_addr_dep): Likewise.
2998 (arm_mac_accumulator_is_mul_result): Likewise.
2999 (arm_mac_accumulator_is_result): Likewise.
3000 * config/arm/aarch-common.c: ... here. New file.
3001 * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3003 (arm_early_store_addr_dep): Likewise.
3004 (arm_no_early_alu_shift_dep): Likewise.
3005 (arm_no_early_alu_shift_value_dep): Likewise.
3006 (arm_no_early_mul_dep): Likewise.
3007 (arm_no_early_store_addr_dep): Likewise.
3008 (arm_mac_accumulator_is_mul_result): Likewise.
3009 (arm_mac_accumulator_is_result): Likewise.
3010 * config/arm/aarch-common-protos.h: ... here. New file.
3011 * config/arm/t-arm (aarch-common.o): Define.
3013 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
3015 * config/arm/arm.md: Include new file "types.md".
3016 (define_attr "type"): Move from here to ...
3017 (define_attr "mul32"): Likewise.
3018 (define_attr "mul64"): Likewise.
3019 * config/arm/types.md: ... here. New file.
3021 2013-07-31 Sebastian Huber <sebastian.huber@embedded-brains.de>
3023 * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
3024 * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
3026 2013-07-31 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3028 * gen-pass-instances.awk: Fix offset of substr().
3030 2013-07-31 David Malcolm <dmalcolm@redhat.com>
3032 * Makefile.in (pass-instances.def): New.
3033 (passes.o): Replace dependency on passes.def with one on
3036 * gen-pass-instances.awk: New.
3038 * passes.c (pass_manager::pass_manager): Use pass-instances.def
3039 rather than passes.def, updating local definition of NEXT_PASS
3040 macro to add an extra NUM parameter (currently unused).
3042 2013-07-30 David Malcolm <dmalcolm@redhat.com>
3044 * Makefile.in (PASS_MANAGER_H): New.
3045 (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
3046 (passes.o): Likewise.
3047 (statistics.o): Likewise.
3048 (cgraphunit.o): Likewise.
3049 (context.o): Depend on PASS_MANAGER_H.
3051 * pass_manager.h: New.
3053 * cgraphunit.c (cgraph_add_new_function): Update for moves
3054 of globals to fields of pass_manager.
3055 (analyze_function): Likewise.
3056 (expand_function): Likewise.
3057 (ipa_passes): Likewise.
3058 (compile): Likewise.
3060 * context.c (context::context): New.
3061 * context.h (context::context): New.
3062 (context::get_passes): New.
3063 (context::passes_): New.
3065 * lto-cgraph.c (input_node): Update for moves of globals to
3066 fields of pass_manager.
3068 * passes.c (all_passes): Remove, in favor of a field of the
3069 same name within the new class pass_manager.
3070 (all_small_ipa_passes): Likewise.
3071 (all_lowering_passes): Likewise.
3072 (all_regular_ipa_passes): Likewise.
3073 (all_late_ipa_passes): Likewise.
3074 (all_lto_gen_passes): Likewise.
3075 (passes_by_id): Likewise.
3076 (passes_by_id_size): Likewise.
3077 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3078 the new class pass_manager.
3079 (set_pass_for_id): Convert to...
3080 (pass_manager::set_pass_for_id): ...method.
3081 (get_pass_for_id): Convert to...
3082 (pass_manager::get_pass_for_id): ...method.
3083 (register_one_dump_file): Move body of implementation into...
3084 (pass_manager::register_one_dump_file): ...here.
3085 (register_dump_files_1): Convert to...
3086 (pass_manager::register_dump_files_1): ...method.
3087 (register_dump_files): Convert to...
3088 (pass_manager::register_dump_files): ...method.
3089 (create_pass_tab): Update for moves of globals to fields of
3091 (dump_passes): Move body of implementation into...
3092 (pass_manager::dump_passes): ...here.
3093 (register_pass): Move body of implementation into...
3094 (pass_manager::register_pass): ...here.
3095 (init_optimization_passes): Convert into...
3096 (pass_manager::pass_manager): ...constructor for new
3097 pass_manager class, and initialize the pass_lists array.
3098 (check_profile_consistency): Update for moves of globals to
3099 fields of pass_manager.
3100 (dump_profile_report): Move body of implementation into...
3101 (pass_manager::dump_profile_report): ...here.
3102 (ipa_write_summaries_1): Update for moves of pass lists from
3103 being globals to fields of pass_manager.
3104 (ipa_write_optimization_summaries): Likewise.
3105 (ipa_read_summaries): Likewise.
3106 (ipa_read_optimization_summaries): Likewise.
3107 (execute_all_ipa_stmt_fixups): Likewise.
3109 * statistics.c (statistics_fini): Update for moves of globals to
3110 fields of pass_manager.
3112 * toplev.c (general_init): Replace call to
3113 init_optimization_passes with construction of the pass_manager
3116 * tree-pass.h (all_passes): Remove, in favor of a field of the
3117 same name within the new class pass_manager.
3118 (all_small_ipa_passes): Likewise.
3119 (all_lowering_passes): Likewise.
3120 (all_regular_ipa_passes): Likewise.
3121 (all_lto_gen_passes): Likewise.
3122 (all_late_ipa_passes): Likewise.
3123 (passes_by_id): Likewise.
3124 (passes_by_id_size): Likewise.
3125 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3126 the new class pass_manager.
3127 (get_pass_for_id): Remove.
3129 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
3131 * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
3134 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
3136 * arm.md (mulhi3): New expand pattern.
3138 2013-07-30 Jan Hubicka <jh@suse.cz>
3139 Martin Liska <marxin.liska@gmail.com>
3141 * profile.c (compute_value_histograms): Do not ICE when
3142 there is mismatch only on some counters.
3144 2013-07-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
3146 PR rtl-optimization/57637
3147 * function.c (move_insn_for_shrink_wrap): Also check the
3148 GEN set of the LIVE problem for the liveness analysis
3149 if it exists, otherwise give up.
3151 2013-07-29 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3153 PR tree-optimization/57993
3154 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
3155 replaced statement in the candidate table.
3156 (phi_add_costs): Return infinite cost when the hidden basis does
3157 not dominate all phis on which the candidate is dependent.
3158 (replace_one_candidate): Record replaced statement in the
3161 2013-07-29 Joern Rennecke <joern.rennecke@embecosm.com>
3163 * config/epiphany/epiphany.md (*isub_i+2): New peephole.
3164 (ashlv2si3): New expander.
3165 (*ashlv2si3_i): New define_insn_and_split.
3166 * predicates.md (float_operation): Allow patterns with three
3169 PR rtl-optimization/58021
3170 * mode-switching.c (create_pre_exit): Always split off preceding
3171 insns if we are not at the basic block head.
3173 2013-07-29 Maciej W. Rozycki <macro@codesourcery.com>
3175 * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
3176 (UCLIBC_DYNAMIC_LINKER): New macro.
3177 * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
3179 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
3180 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
3182 (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
3183 (UCLIBC_DYNAMIC_LINKERN32): Likewise.
3184 * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
3185 for SF and DF modes. Use ieee_quad_format for TF mode.
3186 * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
3187 * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
3188 (mips_option_override): Handle `-mnan=legacy'.
3189 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
3190 `-mabs=2008' and `-mnan=2008'.
3191 (OPTION_DEFAULT_SPECS): Add "nan" default.
3192 (ASM_SPEC): Handle `-mnan='.
3193 [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
3194 * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
3195 comment accordingly.
3196 (neg<mode>2): Likewise.
3197 * config/mips/mips.opt (mabs, mnan): New options.
3198 * doc/install.texi (Configuration): Document `--with-nan=' option.
3199 * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
3201 (MIPS Options): Document them.
3202 * config.gcc <mips*-*-*>: Handle `--with-nan='.
3203 * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
3204 * configure: Regenerate.
3205 * config.in: Regenerate.
3207 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
3209 * config/i386/i386.md (float post-reload splitters): Do not check
3210 for subregs of SSE registers.
3212 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
3213 H.J. Lu <hongjiu.lu@intel.com>
3217 * config/i386/i386.md (post-reload splitter
3218 to avoid partial SSE reg dependency stalls): New pattern.
3220 2013-07-29 Dominik Vogt <vogt@linux.vnet.ibm.com>
3222 * config/s390/s390.md ("movcc"): Swap load and store instructions.
3224 2013-07-27 Joern Rennecke <joern.rennecke@embecosm.com>
3226 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
3227 Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
3229 2013-07-26 Cary Coutant <ccoutant@google.com>
3231 * dwarf2out.c (die_checksum_ordered): Don't include template
3232 instantiations in signature.
3233 (is_template_parameter): New function.
3234 (is_template_instantiation): New function.
3235 (generate_skeleton_bottom_up): Don't include template instantiations
3237 (generate_skeleton): Likewise.
3238 (break_out_comdat_types): Move recursive call to break out nested
3240 (prune_unused_types_mark_generic_parms_dies): Call
3241 is_template_parameter.
3243 2013-07-26 Ian Bolton <ian.bolton@arm.com>
3245 * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
3246 uses vector registers.
3247 * config/aarch64/iterators.md: Add attributes rtn and vas.
3249 2013-07-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3250 Richard Earnshaw <richard.earnshaw@arm.com>
3252 * combine.c (simplify_comparison): Re-canonicalize operands
3254 * config/arm/arm.md (movcond_addsi): New splitter.
3256 2013-07-25 Sterling Augustine <saugustine@google.com>
3258 * dwarf2out.c (size_of_pubnames): Move code to...
3259 (include_pubname_in_output): ...here. New.
3260 (want_pubnames): Rearrange.
3261 (output_pubnames): Call include_pubname_in_output. Move assertion.
3263 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
3265 * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
3267 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
3270 * doc/extend.texi: Remove obsolete builtins. Fix
3271 typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
3273 2013-07-25 Jan Hubicka <jh@suse.cz>
3275 * cgraph.c (release_function_body): Break out from ...
3276 (cgraph_release_function_body): ... this one; also release DECL_RESULT
3278 * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
3279 old_tree in the map.
3280 (create_specialized_node): Update.
3281 * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
3283 * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
3284 DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
3285 * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
3286 * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
3289 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3291 * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
3292 addsi3_carryin_alt2_<optab>): Correct output template.
3294 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3296 * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
3297 Adjust for arm_restrict_it.
3298 Remove trailing whitespace.
3300 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
3302 * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
3303 libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
3305 * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
3307 2013-07-25 Vladimir Makarov <vmakarov@redhat.com>
3309 PR rtl-optimization/57960
3310 * lra-constraints.c (process_alt_operands): Use the right mode
3311 when checking strict_low.
3313 2013-07-25 Jan Hubicka <jh@suse.cz>
3315 * lto-symtab.c (lto_cgraph_replace_node): Release function body.
3316 * cgraph.c (cgraph_remove_node): Do not release function body
3317 when in cgraph streaming.
3318 * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
3319 in other partitions are not considered reachable; fix handling of
3322 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3324 * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
3326 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3331 * config/arm/arm.md ("*sibcall_insn): Replace use of
3332 Ss with US. Adjust output for v5 and v4t.
3333 (*sibcall_value_insn): Likewise and loosen predicate on operand0.
3335 * config/arm/constraints.md ("Ss"): Rename to US.
3337 2013-07-25 Terry Guo <terry.guo@arm.com>
3339 * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
3340 shift_add/shift_sub0/shift_sub1 RTXs.
3342 2013-07-24 Bill Schmidt <wschmidt@linux.ibm.com>
3343 Anton Blanchard <anton@au1.ibm.com>
3345 * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
3346 (altivec_vpks<VI_char>ss): Likewise.
3347 (altivec_vpks<VI_char>us): Likewise.
3348 (altivec_vpku<VI_char>us): Likewise.
3349 (altivec_vpku<VI_char>um): Likewise.
3351 2013-07-24 David Malcolm <dmalcolm@redhat.com>
3353 Introduce context class.
3355 * Makefile.in (CONTEXT_H): New.
3356 (OBJS): Add context.o.
3357 (toplev.o): Add CONTEXT_H to dependencies.
3360 * toplev.c (general_init): Create the singleton gcc::context instance.
3366 2013-07-24 Joern Rennecke <joern.rennecke@embecosm.com>
3368 PR rtl-optimization/57968
3369 * mode-switching.c (create_pre_exit): Allow instructions that
3370 don't set a return register to need a non-exit mode.
3372 2013-07-24 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3373 Anton Blanchard <anton@au1.ibm.com>
3375 * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
3376 operands to vperm for little endian.
3377 * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
3378 of lvsl to create the control mask for a vperm for little endian.
3380 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3381 Anton Blanchard <anton@au1.ibm.com>
3383 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
3384 two operands for little-endian.
3386 2013-07-23 Steve Ellcey <sellcey@mips.com>
3388 * config/mips/mips.c (mips_case_values_threshold): New.
3389 (TARGET_CASE_VALUES_THRESHOLD): Define.
3391 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3392 Anton Blanchard <anton@au1.ibm.com>
3394 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
3395 selection of field for vector splat in little endian mode.
3397 2013-07-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3399 * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
3400 expanders to rs6000.md.
3401 (ior<mode>3): Likewise.
3402 (and<mode>3): Likewise.
3403 (one_cmpl<mode>2): Likewise.
3404 (nor<mode>3): Likewise.
3405 (andc<mode>3): Likewise.
3406 (eqv<mode>3): Likewise.
3407 (nand<mode>3): Likewise.
3408 (orc<mode>3): Likewise.
3410 * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
3413 * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
3414 to split multi-word logical operations.
3415 (rs6000_split_logical_di): Likewise.
3416 (rs6000_split_logical): Likewise.
3418 * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
3419 (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
3420 and allow TImode operations in 32-bit.
3421 (vsx_and<mode>3_64bit): Likewise.
3422 (vsx_ior<mode>3_32bit): Likewise.
3423 (vsx_ior<mode>3_64bit): Likewise.
3424 (vsx_xor<mode>3_32bit): Likewise.
3425 (vsx_xor<mode>3_64bit): Likewise.
3426 (vsx_one_cmpl<mode>2_32bit): Likewise.
3427 (vsx_one_cmpl<mode>2_64bit): Likewise.
3428 (vsx_nor<mode>3_32bit): Likewise.
3429 (vsx_nor<mode>3_64bit): Likewise.
3430 (vsx_andc<mode>3_32bit): Likewise.
3431 (vsx_andc<mode>3_64bit): Likewise.
3432 (vsx_eqv<mode>3_32bit): Likewise.
3433 (vsx_eqv<mode>3_64bit): Likewise.
3434 (vsx_nand<mode>3_32bit): Likewise.
3435 (vsx_nand<mode>3_64bit): Likewise.
3436 (vsx_orc<mode>3_32bit): Likewise.
3437 (vsx_orc<mode>3_64bit): Likewise.
3439 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
3440 logical types in GPRs.
3442 * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
3443 logical insns to rs6000.md, and allow TImode operations in
3445 (altivec_ior<mode>3): Likewise.
3446 (altivec_xor<mode>3): Likewise.
3447 (altivec_one_cmpl<mode>2): Likewise.
3448 (altivec_nor<mode>3): Likewise.
3449 (altivec_andc<mode>3): Likewise.
3451 * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
3452 attributes for moving the 128-bit logical operations into
3454 (BOOL_REGS_OUTPUT): Likewise.
3455 (BOOL_REGS_OP1): Likewise.
3456 (BOOL_REGS_OP2): Likewise.
3457 (BOOL_REGS_UNARY): Likewise.
3458 (BOOL_REGS_AND_CR0): Likewise.
3459 (one_cmpl<mode>2): Add support for DI logical operations on
3460 32-bit, splitting the operations to 32-bit.
3464 (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
3465 changes to combine the 32/64-bit code, allow logical operations on
3466 TI mode in 32-bit, and to use similar match_operator patterns like
3467 scalar mode uses. Combine the Altivec and VSX code for logical
3468 operations, and move it here.
3469 (ior<mode>3, 128-bit types): Likewise.
3470 (xor<mode>3, 128-bit types): Likewise.
3471 (one_cmpl<mode>3, 128-bit types): Likewise.
3472 (nor<mode>3, 128-bit types): Likewise.
3473 (andc<mode>3, 128-bit types): Likewise.
3474 (eqv<mode>3, 128-bit types): Likewise.
3475 (nand<mode>3, 128-bit types): Likewise.
3476 (orc<mode>3, 128-bit types): Likewise.
3477 (and<mode>3_internal): Likewise.
3478 (bool<mode>3_internal): Likewise.
3479 (boolc<mode>3_internal1): Likewise.
3480 (boolc<mode>3_internal2): Likewise.
3481 (boolcc<mode>3_internal1): Likewise.
3482 (boolcc<mode>3_internal2): Likewise.
3483 (eqv<mode>3_internal1): Likewise.
3484 (eqv<mode>3_internal2): Likewise.
3485 (one_cmpl1<mode>3_internal): Likewise.
3487 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
3489 * config/microblaze/microblaze.c (microblaze_expand_prologue):
3490 Rename flag_stack_usage to flag_stack_usage_info.
3492 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
3494 * config/microblaze/sync.md: New file.
3495 * config/microblaze/microblaze.md: Include sync.md
3496 * config/microblaze/microblaze.c: Add print_operand 'y'.
3497 * config/microblaze/constraints.md: Add memory_contraint
3498 'Q' which is a single register.
3500 2013-07-23 Eric Botcazou <ebotcazou@adacore.com>
3502 * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
3504 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
3506 * reload.c (find_reloads): Exit loop once we find this operand
3507 cannot be reloaded somehow for this alternative.
3509 * reload.c (find_reloads): Exit loop once we find a hard register.
3511 * rtlanal.c (computed_jump_p): Exit loop once we find label
3514 * i386.c (ix86_pad_returns): Exit loop after setting replace.
3516 * cfgloopmanip.c (remove_path): Exit loop after setting
3519 * gensupport.c (subst_dup): Avoid loop if code is not
3520 MATCH_DUP nor MATCH_OP_DUP.
3522 2013-07-23 Nicklas Bo Jensen <nbjensen@gmail.com>
3524 * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
3526 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3528 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
3529 true for SP_REGNUM if mode == ptr_mode.
3530 * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
3531 with value R0_REGNUM + 31.
3533 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3535 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
3536 pad pointer-typed argument downward.
3538 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3540 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
3541 and __ILP32__ when the ILP32 model is in use.
3543 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3545 * config/aarch64/aarch64.c (POINTER_BYTES): New define.
3546 (aarch64_load_symref_appropriately): In the case of
3547 SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
3548 to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
3549 (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
3550 change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
3551 if 'mode' doesn't equal to 'ptr_mode'.
3552 (aarch64_output_mi_thunk): Add an assertion on the alignment of
3553 'vcall_offset'; change to call aarch64_emit_move differently depending
3554 on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
3555 to calculate the upper bound of 'vcall_offset'.
3556 (aarch64_cannot_force_const_mem): Change to also return true if
3558 (aarch64_legitimize_reload_address): In the case of large
3559 displacements, add new local variable 'xmode' and an assertion
3560 based on it; change to use 'xmode' to generate the new rtx and
3562 (aarch64_asm_trampoline_template): Change to generate the template
3563 differently depending on TARGET_ILP32 or not; change to use
3564 'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
3565 (aarch64_trampoline_size): Removed.
3566 (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
3567 and replace immediate literals with it. Change to use 'ptr_mode'
3568 instead of 'DImode' and call convert_memory_address if the mode
3569 of 'fnaddr' doesn't equal to 'ptr_mode'.
3570 (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
3572 (aarch64_elf_asm_destructor): Likewise.
3573 * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
3574 on TARGET_ILP32 instead of aarch64_trampoline_size.
3575 * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
3576 of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
3577 (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
3578 (loadwb_pair<GPI:mode>_<P:mode>): ... this. Replace PTR with P.
3579 (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
3580 (storewb_pair<GPI:mode>_<P:mode>): ... this.
3581 (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
3582 depending on the value of 'mode'.
3583 (add_losym_<mode>): New.
3584 (ldr_got_small_<mode>): New, based on ldr_got_small.
3585 (ldr_got_small): Remove.
3586 (ldr_got_small_sidi): New.
3587 * config/aarch64/iterators.md (P): New.
3588 (PTR): Change to 'ptr_mode' in the condition.
3590 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3592 * config.gcc (aarch64*-*-*): Support --with-abi.
3593 (aarch64*-*-elf): Support --with-multilib-list.
3594 (aarch64*-*-linux*): Likewise.
3595 (supported_defaults): Add abi to aarch64*-*-*.
3596 * configure.ac: Mention AArch64 for --with-multilib-list.
3597 * configure: Re-generated.
3598 * config/aarch64/biarchilp32.h: New file.
3599 * config/aarch64/biarchlp64.h: New file.
3600 * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
3602 (MULTILIB_DEFAULTS): Ditto.
3603 (DRIVER_SELF_SPECS): Ditto.
3604 (ASM_SPEC): Update to also substitute -mabi.
3605 * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
3606 file whose name depends on -mabi= and -mbig-endian.
3607 * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
3609 (POINTER_SIZE): New define.
3610 (POINTERS_EXTEND_UNSIGNED): Ditto.
3611 (enum aarch64_abi_type): New enumeration tag.
3612 (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
3613 (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
3614 (TARGET_ILP32): New define.
3615 * config/aarch64/aarch64.opt (mabi): New.
3617 (ilp32, lp64): New values for -mabi.
3618 * config/aarch64/t-aarch64 (comma): New define.
3619 (MULTILIB_OPTIONS): Ditto.
3620 (MULTILIB_DIRNAMES): Ditto.
3621 * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
3622 * doc/invoke.texi: Document -mabi for AArch64.
3624 2013-07-23 Georg-Johann Lay <avr@gjlay.de>
3626 * config/avr/avr.md: Explain asm print modifier 'r' for REG.
3628 2013-07-22 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3629 Anton Blanchard <anton@au1.ibm.com>
3631 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
3632 endianness when selecting field to splat.
3634 2013-07-22 Eric Christopher <echristo@gmail.com>
3636 * dwarf2out.c (die_odr_checksum): New function to use
3637 CHECKSUM_ macros and ULEB128 for DIE tag.
3638 (generate_type_signature): Use.
3640 2013-07-22 Eric Botcazou <ebotcazou@adacore.com>
3642 * config.gcc (sparc*-*-*): Accept leon3 processor.
3643 (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
3644 * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
3645 * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
3646 * config/sparc/sparc.opt (enum processor_type): Add leon3.
3647 (mfix-ut699): Adjust comment.
3648 * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
3649 (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
3650 (CPP_CPU_SPEC): Likewise.
3651 (ASM_CPU_SPEC): Likewise.
3652 * config/sparc/sparc.c (leon3_cost): New constant.
3653 (sparc_option_override): Add leon3 support.
3654 (mem_ref): New function.
3655 (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
3656 (sparc_do_work_around_errata): Look into the instruction in the delay
3657 slot and adjust accordingly. Add fix for the data cache nullify issues
3658 of the UT699. Change insertion position for the NOP.
3659 * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
3660 (leon3_load): New reservation.
3661 (leon_store): Bump latency to 2.
3662 (grfpu): New automaton.
3663 (grfpu_alu): New unit.
3664 (grfpu_ds): Likewise.
3665 (leon_fp_alu): Adjust.
3666 (leon_fp_mult): Delete.
3667 (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
3668 (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
3669 * config/sparc/sparc.md (cpu): Add leon3.
3670 * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
3672 (atomic_test_and_set): Likewise.
3675 2013-07-22 Jürgen Urban <JuergenUrban@gmx.de>
3677 * config.gcc (mips*-*-*): Add --with-fpu support. Make single the
3678 default for R5900 targets.
3679 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
3680 (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
3681 * config/mips/mips.c (mips_option_override): Report an error for
3682 -march=r5900 -mhard-float -mdouble-float. Use spu_single_format
3683 for -march=r5900 -mhard-float.
3685 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
3687 * df-problems.c (can_move_insns_across): Exit loop once we
3688 find a non-fixed, non-global register.
3690 * ipa-pure-const.c (propagate_nothrow): Exit loop after
3693 * omega.c (omega_eliminate_red): Break after setting red_found.
3694 (omega_problem_has_red_equations): Similarly after setting found.
3695 (omega_query_variable): Similarly after setting coupled.
3697 2013-07-22 Marek Polacek <polacek@redhat.com>
3699 * gimplify.c: Don't include gimple.h twice.
3701 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3703 * config/arm/constraints.md (Pd): Allow TARGET_THUMB
3704 instead of TARGET_THUMB1.
3705 (Pz): New constraint.
3706 * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
3708 (compare_negsi_si): Likewise.
3709 (compare_addsi2_op0): Likewise.
3710 (compare_addsi2_op1): Likewise.
3711 (addsi3_carryin_<optab>): Likewise.
3712 (addsi3_carryin_alt2_<optab>): Likewise.
3713 (addsi3_carryin_shift_<optab>): Disable cond_exec variant
3714 for arm_restrict_it.
3715 (subsi3_carryin): Likewise.
3716 (arm_subsi3_insn): Add alternatives for 16-bit encoding.
3717 (minmax_arithsi): Disable for arm_restrict_it.
3718 (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
3719 (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
3720 (satsi_<SAT:code>_shift): Likewise.
3721 (arm_shiftsi3): Add alternative for 16-bit encoding.
3722 (arm32_movhf): Disable for arm_restrict_it.
3723 (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
3724 (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
3726 2013-07-22 Sofiane Naci <sofiane.naci@arm.com>
3728 * config/arm/arm.md (attribute "insn"): Delete.
3729 (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
3730 "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
3731 (not_shiftsi): Update for attribute change.
3732 (not_shiftsi_compare0): Likewise.
3733 (not_shiftsi_compare0_scratch): Likewise.
3734 (arm_one_cmplsi2): Likewise.
3735 (thumb1_one_cmplsi2): Likewise.
3736 (notsi_compare0): Likewise.
3737 (notsi_compare0_scratch): Likewise.
3738 (thumb1_movdi_insn): Likewise.
3739 (arm_movsi_insn): Likewise.
3740 (movhi_insn_arch4): Likewise.
3741 (movhi_bytes): Likewise.
3742 (arm_movqi_insn): Likewise.
3743 (thumb1_movqi_insn): Likewise.
3744 (arm32_movhf): Likewise.
3745 (thumb1_movhf): Likewise.
3746 (arm_movsf_soft_insn): Likewise.
3747 (thumb1_movsf_insn): Likewise.
3748 (thumb_movdf_insn): Likewise.
3749 (movsicc_insn): Likewise.
3750 (movsfcc_soft_insn): Likewise.
3751 (and_scc): Likewise.
3752 (cond_move): Likewise.
3753 (if_move_not): Likewise.
3754 (if_not_move): Likewise.
3755 (if_shift_move): Likewise.
3756 (if_move_shift): Likewise.
3757 (if_shift_shift): Likewise.
3758 (if_not_arith): Likewise.
3759 (if_arith_not): Likewise.
3760 (cond_move_not): Likewise.
3761 * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
3762 (neon_mov<mode>): Likewise.
3763 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
3764 (thumb2_movsi_vfp): Likewise.
3765 (movsf_vfp): Likewise.
3766 (thumb2_movsf_vfp): Likewise.
3767 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
3769 (cortexa7_older_only): Likewise.
3770 (cortexa7_younger): Likewise.
3771 * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
3772 (1020alu_shift_op): Likewise.
3773 (1020alu_shift_reg_op): Likewise.
3774 * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
3775 (alu_shift_op): Likewise.
3776 (alu_shift_reg_op): Likewise.
3777 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
3778 (11_alu_shift_op): Likewise.
3779 (11_alu_shift_reg_op): Likewise.
3780 * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
3781 (9_alu_shift_reg_op): Likewise.
3782 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
3784 (cortex_a15_alu_shift): Likewise.
3785 (cortex_a15_alu_shift_reg): Likewise.
3786 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
3787 (cortex_a5_alu_shift): Likewise.
3788 * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
3790 (cortex_a53_alu_shift): Likewise.
3791 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
3793 (cortex_a7_alu_reg): Likewise.
3794 (cortex_a7_alu_shift): Likewise.
3795 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3796 (cortex_a8_alu_shift): Likewise.
3797 (cortex_a8_alu_shift_reg): Likewise.
3798 (cortex_a8_mov): Likewise.
3799 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
3800 (cortex_a9_dp_shift): Likewise.
3801 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
3802 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
3803 (cortex_r4_mov): Likewise.
3804 (cortex_r4_alu_shift): Likewise.
3805 (cortex_r4_alu_shift_reg): Likewise.
3806 * config/arm/fa526.md (526_alu_op): Update for attribute change.
3807 (526_alu_shift_op): Likewise.
3808 * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
3809 * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
3810 (626te_alu_shift_op): Likewise.
3811 * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
3812 (726te_alu_op): Likewise.
3813 (726te_alu_shift_op): Likewise.
3814 (726te_alu_shift_reg_op): Likewise.
3815 * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
3816 (mp626_alu_shift_op): Likewise.
3817 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
3818 (pj4_alu_e1_conds): Likewise.
3819 (pj4_alu): Likewise.
3820 (pj4_alu_conds): Likewise.
3821 (pj4_shift): Likewise.
3822 (pj4_shift_conds): Likewise.
3823 (pj4_alu_shift): Likewise.
3824 (pj4_alu_shift_conds): Likewise.
3826 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3828 * config/arm/predicates.md (shiftable_operator_strict_it):
3830 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
3831 Disable cond_exec version for arm_restrict_it.
3832 (thumb2_smaxsi3): Convert to generate cond_exec.
3833 (thumb2_sminsi3): Likewise.
3834 (thumb32_umaxsi3): Likewise.
3835 (thumb2_uminsi3): Likewise.
3836 (thumb2_abssi2): Adjust constraints for arm_restrict_it.
3837 (thumb2_neg_abssi2): Likewise.
3838 (thumb2_mov_scc): Add alternative for 16-bit encoding.
3839 (thumb2_movsicc_insn): Adjust alternatives.
3840 (thumb2_mov_negscc): Disable for arm_restrict_it.
3841 (thumb2_mov_negscc_strict_it): New pattern.
3842 (thumb2_mov_notscc_strict_it): New pattern.
3843 (thumb2_mov_notscc): Disable for arm_restrict_it.
3844 (thumb2_ior_scc): Likewise.
3845 (thumb2_ior_scc_strict_it): New pattern.
3846 (thumb2_cond_move): Adjust for arm_restrict_it.
3847 (thumb2_cond_arith): Disable for arm_restrict_it.
3848 (thumb2_cond_arith_strict_it): New pattern.
3849 (thumb2_cond_sub): Adjust for arm_restrict_it.
3850 (thumb2_movcond): Likewise.
3851 (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
3852 (thumb2_zero_extendhisi2_v6): Likewise.
3853 (thumb2_zero_extendqisi2_v6): Likewise.
3854 (orsi_notsi_si): Likewise.
3855 (orsi_not_shiftsi_si): Likewise.
3857 2013-07-22 Georg-Johann Lay <avr@gjlay.de>
3859 * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
3860 instruction sequence is 1 byte shorter.
3862 2013-07-22 Uros Bizjak <ubizjak@gmail.com>
3864 * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
3865 it is not needed after split.
3867 2013-07-20 Iain Sandoe <iain@codesourcery.com>
3870 * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
3871 second label for nonlocal goto receivers. Don't output pic base labels
3872 unless we're producing PIC; mark that action unreachable().
3873 (ix86_save_reg): If the function contains a nonlocal label, save the
3875 * config/darwin-protos.h (machopic_should_output_picbase_label): New.
3876 * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
3877 (update_pic_label_number_if_needed): New.
3878 (machopic_output_function_base_name): Adjust for nonlocal receiver
3880 (machopic_should_output_picbase_label): New.
3881 * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
3882 (nonlocal_goto_receiver): New insn and split.
3884 2013-07-20 James Greenhalgh <james.greenhalgh@arm.com>
3886 * config/aarch64/aarch64-builtins.c
3887 (aarch64_fold_builtin): Fold abs in all modes.
3888 * config/aarch64/aarch64-simd-builtins.def
3889 (abs): Enable for all modes.
3890 * config/aarch64/arm_neon.h
3891 (vabs<q>_s<8,16,32,64): Rewrite using builtins.
3892 (vabs_f64): Add missing intrinsic.
3894 2013-07-19 Ian Bolton <ian.bolton@arm.com>
3896 * config/aarch64/arm_neon.h (vabs_s64): New function
3898 2013-07-19 Georg-Johann Lay <avr@gjlay.de>
3901 * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
3902 * config/avr/avr.md (adjust_len): Add `round'.
3903 * config/avr/avr-protos.h (avr_out_round): New prototype.
3904 (avr_out_plus): Add `out_label' argument.
3905 * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
3906 (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
3907 Handle the case where `insn' is just a pattern.
3908 (avr_out_bitop): Handle the case where `insn' is just a pattern.
3909 (avr_out_round): New function.
3910 (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
3912 2013-07-18 David Holsgrove <david.holsgrove@xilinx.com>
3914 * config/microblaze/microblaze.c (microblaze_expand_prologue):
3915 Add check for flag_stack_usage to handle -fstack-usage support
3917 2013-07-18 Pat Haugen <pthaugen@us.ibm.com>
3919 * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
3920 interaction for new Power8 flags and VSX.
3922 2013-07-18 Sriraman Tallam <tmsriram@google.com>
3925 * tree-inline.c (expand_call_inline): Emit errors during
3926 early_inlining only if optimization is not turned on.
3928 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3932 * passes.c (init_optimization_passes): Move the construction of
3933 the pass hierarchy into a new passes.def file.
3935 * Makefile.in (passes.o): Add dependency on passes.def.
3937 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3939 * passes.c (init_optimization_passes): Introduce macros for
3940 constructing the tree of passes (INSERT_PASSES_AFTER,
3941 PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
3942 TERMINATE_PASS_LIST).
3944 2013-07-18 Vladimir Makarov <vmakarov@redhat.com>
3945 Wei Mi <wmi@google.com>
3947 PR rtl-optimization/57878
3948 * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
3950 (reload_pseudo_compare_func): Check nregs first for reload
3953 2013-07-18 David Malcolm <dmalcolm@redhat.com>
3955 * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
3957 2013-07-18 Po-Chun Chang <pchang9@cs.wisc.edu>
3959 * read-rtl.c (validate_const_int): Once an invalid character is
3960 seen, quit the loop.
3962 * gengtype.c (write_roots): Similarly once we find the "deletable"
3963 or "if_marked" option.
3965 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
3967 * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
3968 "xtab" and "sat". Move value "clz" from here to ...
3969 (attriubte "type"): ... here.
3970 (satsi_<SAT:code>): Delete "insn" attribute.
3971 (satsi_<SAT:code>_shift): Likewise.
3972 (arm_zero_extendqisi2addsi): Likewise.
3973 (arm_extendqisi2addsi): Likewise.
3974 (clzsi2): Update for attribute changes.
3975 (rbitsi2): Likewise.
3976 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
3978 (arm_usatsihi): Likewise.
3979 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
3981 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
3983 * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
3984 "arlo_imm". Rename "alu_reg" to "arlo_reg". Rename "simple_alu_shift"
3985 to "extend". Split "alu_shift" into "shift" and "arlo_shift". Split
3986 "alu_shift_reg" into "shift_reg" and "arlo_shift_reg". List types
3987 in alphabetical order.
3988 (attribute "core_cycles"): Update for attribute changes.
3989 (arm_addsi3): Likewise.
3990 (addsi3_compare0): Likewise.
3991 (addsi3_compare0_scratch): Likewise.
3992 (addsi3_compare_op1): Likewise.
3993 (addsi3_compare_op2): Likewise.
3994 (compare_addsi2_op0): Likewise.
3995 (compare_addsi2_op1): Likewise.
3996 (addsi3_carryin_shift_<optab>): Likewise.
3997 (subsi3_carryin_shift): Likewise.
3998 (rsbsi3_carryin_shift): Likewise.
3999 (arm_subsi3_insn): Likewise.
4000 (subsi3_compare0): Likewise.
4001 (subsi3_compare): Likewise.
4002 (arm_andsi3_insn): Likewise.
4003 (thumb1_andsi3_insn): Likewise.
4004 (andsi3_compare0): Likewise.
4005 (andsi3_compare0_scratch): Likewise.
4006 (zeroextractsi_compare0_scratch
4007 (andsi_not_shiftsi_si): Likewise.
4008 (iorsi3_insn): Likewise.
4009 (iorsi3_compare0): Likewise.
4010 (iorsi3_compare0_scratch): Likewise.
4011 (arm_xorsi3): Likewise.
4012 (thumb1_xorsi3_insn): Likewise.
4013 (xorsi3_compare0): Likewise.
4014 (xorsi3_compare0_scratch): Likewise.
4015 (satsi_<SAT:code>_shift): Likewise.
4017 (arm_shiftsi3): Likewise.
4018 (shiftsi3_compare0): Likewise.
4019 (not_shiftsi): Likewise.
4020 (not_shiftsi_compare0): Likewise.
4021 (not_shiftsi_compare0_scratch): Likewise.
4022 (arm_one_cmplsi2): Likewise.
4023 (thumb_one_complsi2): Likewise.
4024 (notsi_compare0): Likewise.
4025 (notsi_compare0_scratch): Likewise.
4026 (thumb1_zero_extendhisi2): Likewise.
4027 (arm_zero_extendhisi2): Likewise.
4028 (arm_zero_extendhisi2_v6): Likewise.
4029 (arm_zero_extendhisi2addsi): Likewise.
4030 (thumb1_zero_extendqisi2): Likewise.
4031 (thumb1_zero_extendqisi2_v6): Likewise.
4032 (arm_zero_extendqisi2): Likewise.
4033 (arm_zero_extendqisi2_v6): Likewise.
4034 (arm_zero_extendqisi2addsi): Likewise.
4035 (thumb1_extendhisi2): Likewise.
4036 (arm_extendhisi2): Likewise.
4037 (arm_extendhisi2_v6): Likewise.
4038 (arm_extendqisi): Likewise.
4039 (arm_extendqisi_v6): Likewise.
4040 (arm_extendqisi2addsi): Likewise.
4041 (thumb1_extendqisi2): Likewise.
4042 (thumb1_movdi_insn): Likewise.
4043 (arm_movsi_insn): Likewise.
4044 (movsi_compare0): Likewise.
4045 (movhi_insn_arch4): Likewise.
4046 (movhi_bytes): Likewise.
4047 (arm_movqi_insn): Likewise.
4048 (thumb1_movqi_insn): Likewise.
4049 (arm32_movhf): Likewise.
4050 (thumb1_movhf): Likewise.
4051 (arm_movsf_soft_insn): Likewise.
4052 (thumb1_movsf_insn): Likewise.
4053 (movdf_soft_insn): Likewise.
4054 (thumb_movdf_insn): Likewise.
4055 (arm_cmpsi_insn): Likewise.
4056 (cmpsi_shiftsi): Likewise.
4057 (cmpsi_shiftsi_swp): Likewise.
4058 (arm_cmpsi_negshiftsi_si): Likewise.
4059 (movsicc_insn): Likewise.
4060 (movsfcc_soft_insn): Likewise.
4061 (arith_shiftsi): Likewise.
4062 (arith_shiftsi_compare0
4063 (arith_shiftsi_compare0_scratch
4064 (sub_shiftsi): Likewise.
4065 (sub_shiftsi_compare0
4066 (sub_shiftsi_compare0_scratch
4067 (and_scc): Likewise.
4068 (cond_move): Likewise.
4069 (if_plus_move): Likewise.
4070 (if_move_plus): Likewise.
4071 (if_move_not): Likewise.
4072 (if_not_move): Likewise.
4073 (if_shift_move): Likewise.
4074 (if_move_shift): Likewise.
4075 (if_shift_shift): Likewise.
4076 (if_not_arith): Likewise.
4077 (if_arith_not): Likewise.
4078 (cond_move_not): Likewise.
4079 (thumb1_ashlsi3): Set type attribute.
4080 (thumb1_ashrsi3): Likewise.
4081 (thumb1_lshrsi3): Likewise.
4082 (thumb1_rotrsi3): Likewise.
4083 (shiftsi3_compare0_scratch): Likewise.
4084 * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
4085 (neon_mov<mode>): Likewise.
4086 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
4088 (thumb2_movsi_insn): Likewise.
4089 (thumb2_cmpsi_neg_shiftsi): Likewise.
4090 (thumb2_extendqisi_v6): Likewise.
4091 (thumb2_zero_extendhisi2_v6): Likewise.
4092 (thumb2_zero_extendqisi2_v6): Likewise.
4093 (thumb2_shiftsi3_short): Likewise.
4094 (thumb2_addsi3_compare0_scratch): Likewise.
4095 (orsi_not_shiftsi_si): Likewise.
4096 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
4097 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
4099 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
4100 (1020alu_shift_op): Likewise.
4101 (1020alu_shift_reg_op): Likewise.
4102 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
4103 (alu_shift_op): Likewise.
4104 (alu_shift_reg_op): Likewise.
4105 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
4106 (11_alu_shift_op): Likewise.
4107 (11_alu_shift_reg_op): Likewise.
4108 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
4109 (9_alu_shift_reg_op): Likewise.
4110 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4112 (cortex_a15_alu_shift): Likewise.
4113 (cortex_a15_alu_shift_reg): Likewise.
4114 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
4116 (cortex_a5_alu_shift): Likewise.
4117 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
4119 (cortex_a53_alu_shift): Likewise.
4120 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4122 (cortex_a7_alu_reg): Likewise.
4123 (cortex_a7_alu_shift): Likewise.
4124 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
4126 (cortex_a8_alu_shift): Likewise.
4127 (cortex_a8_alu_shift_reg): Likewise.
4128 (cortex_a8_mov): Likewise.
4129 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
4130 (cortex_a9_dp_shift): Likewise.
4131 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
4133 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
4135 (cortex_r4_mov): Likewise.
4136 (cortex_r4_alu_shift): Likewise.
4137 (cortex_r4_alu_shift_reg): Likewise.
4138 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
4139 (526_alu_shift_op): Likewise.
4140 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
4141 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
4142 (626te_alu_shift_op): Likewise.
4143 * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
4144 (726te_alu_op): Likewise.
4145 (726te_alu_shift_op): Likewise.
4146 (726te_alu_shift_reg_op): Likewise.
4147 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
4148 (mp626_alu_shift_op): Likewise.
4149 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
4150 (pj4_alu_e1_conds): Likewise.
4151 (pj4_alu): Likewise.
4152 (pj4_alu_conds): Likewise.
4153 (pj4_shift): Likewise.
4154 (pj4_shift_conds): Likewise.
4155 (pj4_alu_shift): Likewise.
4156 (pj4_alu_shift_conds): Likewise.
4157 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4159 (cortexa7_older_only): Likewise.
4160 (cortexa7_younger): Likewise.
4162 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4164 * ipa-pure-const.c (generate_summary): Rename to...
4165 (pure_const_generate_summary): ... this.
4167 2013-07-17 Iain Sandoe <iain@codesourcery.com>
4169 * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
4171 2013-07-17 Yvan Roux <yvan.roux@linaro.org>
4174 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
4177 2013-07-17 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4179 * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
4180 enabled without -march=zEC12.
4181 * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
4184 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
4186 * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
4187 (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
4188 (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
4189 terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
4190 (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
4191 * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
4192 ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
4194 <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
4195 and ISA_HAS_NMADD3_NMSUB3.
4196 * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
4197 (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
4198 (nmsub4<mode>, nmsub3<mode>): Likewise.
4199 (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
4201 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
4203 * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
4204 TARGET_MIPS5400 checking.
4206 2013-07-16 Jakub Jelinek <jakub@redhat.com>
4207 Peter Bergner <bergner@vnet.ibm.com>
4209 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
4210 registers in the comment.
4211 (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
4212 (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
4213 rather than FIRST_PSEUDO_REGISTERS.
4215 2013-07-16 Peter Bergner <bergner@vnet.ibm.com>
4217 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
4218 enable extra ISA flags with TARGET_HTM.
4220 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
4222 * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
4225 2013-07-15 Cong Hou <congh@google.com>
4227 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
4228 in compare function for sorting.
4230 2013-07-15 Peter Bergner <bergner@vnet.ibm.com>
4232 * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
4233 * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
4234 * config/rs6000/rs6000.opt: Add -mhtm option.
4235 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
4236 (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
4237 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
4238 __HTM__ if the HTM instructions are available.
4239 * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
4240 htm_spr_reg_operand): New define_predicates.
4241 * config/rs6000/rs6000.md (define_attr "type"): Add htm.
4242 (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
4244 * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
4245 BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
4246 HTM builtin functions.
4247 * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
4248 (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
4249 (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
4250 (rs6000_builtin_mask_calculate): Likewise.
4251 (rs6000_option_override_internal): Likewise.
4252 (bdesc_htm): Add new HTM builtin support.
4253 (htm_spr_num): New function.
4254 (htm_spr_regno): Likewise.
4255 (rs6000_htm_spr_icode): Likewise.
4256 (htm_expand_builtin): Likewise.
4257 (htm_init_builtins): Likewise.
4258 (rs6000_expand_builtin): Add support for HTM builtin functions.
4259 (rs6000_init_builtins): Likewise.
4260 (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
4262 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
4263 (TARGET_HTM, MASK_HTM): Define macros.
4264 (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
4265 (FIXED_REGISTERS): Likewise.
4266 (CALL_USED_REGISTERS): Likewise.
4267 (CALL_REALLY_USED_REGISTERS): Likewise.
4268 (REG_ALLOC_ORDER): Likewise.
4269 (enum reg_class): Likewise.
4270 (REG_CLASS_NAMES): Likewise.
4271 (REG_CLASS_CONTENTS): Likewise.
4272 (REGISTER_NAMES): Likewise.
4273 (ADDITIONAL_REGISTER_NAMES): Likewise.
4274 (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
4275 RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
4276 (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
4277 * config/rs6000/htm.md: New file.
4278 * config/rs6000/htmintrin.h: New file.
4279 * config/rs6000/htmxlintrin.h: New file.
4281 2013-07-15 Marcus Shawcroft <marcus.shawcroft@arm.com>
4283 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4284 Define SYMBOL_TINY_GOT, update comment.
4285 * config/aarch64/aarch64.c
4286 (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
4287 (aarch64_expand_mov_immediate): Likewise.
4288 (aarch64_print_operand): Likewise.
4289 (aarch64_classify_symbol): Likewise.
4290 * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
4291 (ldr_got_tiny): Define.
4293 2013-07-13 Tobias Grosser <tobias@grosser.es>
4295 PR tree-optimization/54094
4296 * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
4297 scheduling dimension for the parallelism check from the polyhedral
4298 information in the AST.
4299 * graphite-dependences.c (carries_deps): Do not assume the schedule is
4302 2013-07-13 Jason Merrill <jason@redhat.com>
4304 * print-tree.c (debug_vec_tree): Use debug_raw.
4305 (debug_raw (vec<tree, va_gc> &)): New.
4306 (debug_raw (vec<tree, va_gc> *)): New.
4307 * tree.h: Declare them.
4309 2013-07-13 Bin Cheng <bin.cheng@arm.com>
4311 * ifcvt.c (ifcvt_after_combine): New static variable.
4312 (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
4314 (if_convert): New parameter after_combine. Set ifcvt_after_combine.
4315 (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
4316 rest_of_handle_if_after_reload): Pass new argument for if_convert.
4318 2013-07-12 Maciej W. Rozycki <macro@codesourcery.com>
4320 * config/mips/mips.c (mips_expand_call): Remove empty statement.
4322 2013-07-12 Michael Matz <matz@suse.de>
4325 * convert.c (convert_to_real): Reject non-float inner types.
4327 2013-07-12 Tejas Belagod <tejas.belagod@arm.com>
4329 * config/aarch64/aarch64-protos.h
4330 (aarch64_simd_immediate_valid_for_move): Remove.
4331 * config/aarch64/aarch64.c (simd_immediate_info): New member.
4332 (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
4334 (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
4336 2013-07-11 Steve Ellcey <sellcey@mips.com>
4338 * config/mips/mips.c (mips_conditional_register_usage): Do not
4339 use t[0-7] registers in MIPS16 mode when optimizing for size.
4341 2013-07-11 Sriraman Tallam <tmsriram@google.com>
4343 * config/i386/i386.c (dispatch_function_versions): Fix array
4344 indexing of function_version_info to match actual_versions.
4346 2013-07-11 Teresa Johnson <tejohnson@google.com>
4348 * vec.h (struct va_gc): Move release out-of-line.
4349 (va_gc::release): Call ggc_free on released vec.
4351 2013-07-11 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
4353 * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
4354 Require GOT register as additional operand in UNSPEC.
4355 ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
4356 ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4357 ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4358 ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
4359 ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
4360 ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
4361 ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
4363 2013-07-11 Georg-Johann Lay <avr@gjlay.de>
4366 * config/avr/avr.c (avr_set_current_function): Sanity-check signal
4367 name seen by assembler/linker rather if available.
4369 2013-07-11 Andreas Schwab <schwab@suse.de>
4371 * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
4373 2013-07-10 Vladimir Makarov <vmakarov@redhat.com>
4375 * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
4377 2013-07-10 Joseph Myers <joseph@codesourcery.com>
4379 * doc/tm.texi.in: Move hook documentation to ....
4380 * target.def: ... here.
4382 * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
4384 * doc/tm.texi: Regenerate.
4386 2013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
4389 * doc/invoke.texi: Document -Wconditionally-supported.
4391 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
4394 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
4397 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
4400 * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
4401 (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
4402 (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
4403 Remove duplicate devices.
4404 * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
4405 * config/avr/t-multilib: Regenerate.
4406 * config/avr/avr-tables.opt: Regenerate.
4407 * doc/avr-mmcu.texi: Regenerate.
4409 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
4412 * config/avr/avr.opt (Waddr-space-convert): Fix typo.
4414 2013-07-10 Graham Stott <graham.stott@btinternet.com>
4416 * config/mips/mips.c (mips_rtx_costs): Very slightly increase
4417 the cost of MULT when optimizing for size.
4419 2013-07-10 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4421 * config/cr16/cr16-protos.h: Don't include target.h.
4423 2013-07-09 Joseph Myers <joseph@codesourcery.com>
4425 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
4426 adjust register size for TDmode and TFmode for VSX registers.
4428 2013-07-08 Kai Tietz <ktietz@redhat.com>
4431 * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
4432 hook_bool_const_tree_true.
4434 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4436 * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
4437 * config/s390/s390.h: Remove F*_REGNUM macro definitions.
4438 * config/s390/s390.md: Define FPR*_REGNUM constants.
4439 Fix FPR2_REGNUM constant (18 -> 17).
4440 ("*trunc<BFP:mode><DFP_ALL:mode>2")
4441 ("*trunc<DFP_ALL:mode><BFP:mode>2")
4442 ("trunc<BFP:mode><DFP_ALL:mode>2")
4443 ("trunc<DFP_ALL:mode><BFP:mode>2")
4444 ("*extend<BFP:mode><DFP_ALL:mode>2")
4445 ("*extend<DFP_ALL:mode><BFP:mode>2")
4446 ("extend<BFP:mode><DFP_ALL:mode>2")
4447 ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
4450 2013-07-08 Graham Stott <graham.stott@btinternet.com>
4452 * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
4454 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4456 * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
4457 and cfun_fpr_bit_p to cfun_fpr_save_p.
4458 (s390_frame_area, s390_register_info, s390_frame_info)
4459 (s390_emit_prologue, s390_emit_epilogue)
4460 (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
4462 * config/s390/s390.h: Define *_REGNUM macros for floating point
4465 2013-07-08 Eric Botcazou <ebotcazou@adacore.com>
4467 * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
4469 2013-07-08 Po-Chun Chang <pchang9@cs.wisc.edu>
4471 PR rtl-optimization/57786
4472 * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
4473 and break out of the loop when it is set to false.
4475 2013-07-08 Jakub Jelinek <jakub@redhat.com>
4478 * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
4479 Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
4480 (const_int 63)) 0)).
4481 * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
4482 using simplify_gen_unary instead of gen_rtx_*_EXTEND.
4483 * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
4485 PR rtl-optimization/57829
4486 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
4487 mask bits outside of mode are just sign-extension from mode to HWI.
4489 2013-07-08 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
4491 * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
4492 * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
4493 adjust_address instead of change_address to keep info about alignment.
4494 (emit_strmov): Remove.
4495 (emit_memmov): New function.
4496 (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
4497 (expand_movmem_epilogue): Likewise and return updated rtx for
4499 (expand_constant_movmem_prologue): Likewise and return updated rtx for
4500 destination and source.
4501 (decide_alignment): Refactor, handle vector_loop.
4502 (ix86_expand_movmem): Likewise.
4503 (ix86_expand_setmem): Likewise.
4504 * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
4506 2013-07-07 Uros Bizjak <ubizjak@gmail.com>
4508 * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
4509 signature_TM2_ebx, it interferes with signature_INTEL_ebx.
4511 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
4513 * config/i386/sse.md (sse_movlhps): Change alternative 3
4514 of operand 2 to "m".
4516 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
4519 * config/i386/sse.md (iptr): New mode attribute.
4520 (sse2_movq128): Add pointer size overrides for Intel asm dialect.
4521 (<sse>_vm<plusminus_insn><mode>3): Ditto.
4522 (<sse>_vmmul<mode>3): Ditto.
4523 (<sse>_vmdiv<mode>3): Ditto.
4524 (sse_vmrcpv4sf2): Ditto.
4525 (<sse>_vmsqrt<mode>2): Ditto.
4526 (sse_vmrsqrtv4sf2): Ditto.
4527 (<sse>_vm<code><mode>3): Ditto.
4528 (avx_vmcmp<mode>3): Ditto.
4529 (<sse>_vmmaskcmp<mode>3): Ditto.
4530 (<sse>_comi): Ditto.
4531 (<sse>_ucomi): Ditto.
4532 (*xop_vmfrcz_<mode>): Ditto.
4533 (*fmai_fmadd_<mode>): Ditto.
4534 (*fmai_fmsub_<mode>): Ditto.
4535 (*fmai_fnmadd_<mode>): Ditto.
4536 (*fmai_fnmsub_<mode>): Ditto.
4537 (*fma4i_vmfmadd_<mode>): Ditto.
4538 (*fma4i_vmfmsub_<mode>): Ditto.
4539 (*fma4i_vmfnmadd_<mode>): Ditto.
4540 (*fma4i_vmfnmsub_<mode>): Ditto.
4541 (*xop_vmfrcz_<mode>): Ditto.
4542 (sse_cvtps2pi): Ditto.
4543 (sse_cvttps2pi): Ditto.
4544 (sse_cvtss2si): Ditto.
4545 (sse_cvtss2si_2): Ditto.
4546 (sse_cvtss2siq_2): Ditto.
4547 (sse_cvttss2si): Ditto.
4548 (sse_cvttss2siq): Ditto.
4549 (sse_cvtsd2si): Ditto.
4550 (sse_cvtsd2si_2): Ditto.
4551 (sse_cvtsd2siq_2): Ditto.
4552 (sse_cvttsd2si): Ditto.
4553 (sse_cvttsd2siq): Ditto.
4554 (sse_cvtsd2ss): Ditto.
4555 (sse_cvtss2sd): Ditto.
4556 (avx2_pbroadcast<mode>): Ditto.
4557 (avx2_pbroadcast<mode>_1): Ditto.
4558 (*avx_vperm_broadcast_v4sf): Ditto.
4560 (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
4561 (sse_movlhps): Ditto.
4562 (sse_storehps): Ditto.
4563 (sse_loadhps): Ditto.
4564 (sse_storelps): Ditto.
4565 (sse_loadlps): Ditto.
4566 (*vec_concatv4sf): Ditto.
4567 (*vec_interleave_highv2df): Ditto.
4568 (*vec_interleave_lowv2df): Ditto.
4569 (*vec_extractv2df_1_sse): Ditto.
4570 (*vec_extractv2df_0_sse): Ditto.
4571 (sse2_storelpd): Ditto.
4572 (sse2_loadlpd): Ditto.
4573 (sse2_movsd): Ditto.
4574 (*vec_concatv4si): Ditto.
4575 (vec_concatv2di): Ditto.
4577 * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
4578 for Intel asm dialect.
4579 (mmx_punpcklwd): Ditto.
4580 (mmx_punpckldq): Ditto.
4582 * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
4583 for intel assembler dialect.
4585 2013-07-06 Jakub Jelinek <jakub@redhat.com>
4588 * fold-const.c (tree_call_nonnegative_warnv_p): Return true
4589 for BUILT_IN_C{LZ,LRSB}*.
4590 * tree.h (CASE_INT_FN): Add FN##IMAX case.
4591 * tree-vrp.c (extract_range_basic): Handle
4592 BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*. For
4593 BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
4594 fall thru to code calling set_value*.
4595 * builtins.c (expand_builtin): Remove *IMAX cases.
4596 (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
4597 if width is bigger than 2*HWI.
4599 2013-07-05 Vladimir Makarov <vmakarov@redhat.com>
4601 PR rtl-optimization/55342
4602 * lra-int.h (lra_subreg_reload_pseudos): New.
4603 * lra.c: Add undoing optional reloads to the block diagram.
4604 (lra_subreg_reload_pseudos): New.
4605 (lra_optional_reload_pseudos): Change comments.
4606 (lra): Init and clear lra_subreg_reload_pseudos. Clear
4607 lra_optional_reload_pseudos after undo transformations.
4608 * lra-assigns.c (pseudo_prefix_title): New.
4609 (lra_setup_reg_renumber): Use it.
4610 (spill_for): Ditto. Check subreg reload pseudos too.
4611 (assign_by_spills): Consider subreg reload pseudos too.
4612 * lra-constraints.c (simplify_operand_subreg): Use
4613 lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
4614 (curr_insn_transform): Recognize and do optional reloads.
4615 (undo_optional_reloads): New.
4616 (lra_undo_inheritance): Call undo_optional_reloads.
4618 2013-07-05 Thomas Quinot <quinot@adacore.com>
4620 * tree-complex.c (expand_complex_operations_1): Fix typo.
4622 2013-07-04 Tejas Belagod <tejas.belagod@arm.com>
4624 * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
4625 (tune_params): New member 'const vec_costs'.
4626 * config/aarch64/aarch64.c (generic_vector_cost): New.
4627 (generic_tunings): New member 'generic_vector_cost'.
4628 (aarch64_builtin_vectorization_cost): New.
4629 (aarch64_add_stmt_cost): New.
4630 (TARGET_VECTORIZE_ADD_STMT_COST): New.
4631 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
4633 2013-07-03 Jakub Jelinek <jakub@redhat.com>
4636 * config/i386/predicates.md (vsib_address_operand): Disallow
4637 SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
4639 2013-07-03 Hans-Peter Nilsson <hp@bitrange.com>
4642 * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
4643 expand_builtin_setjmp_receiver.
4644 (expand_label): Adjust, call expand_builtin_setjmp_receiver
4645 with NULL for the label parameter.
4646 * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
4647 the frame-pointer. Adjust comments.
4648 [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
4649 only if LABEL is non-NULL.
4651 2013-07-03 Yufeng Zhang <yufeng.zhang@arm.com>
4653 * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
4654 (ARM_ABI_AAPCS64): Ditto.
4656 (ARM_DEFAULT_ABI): Ditto.
4658 2013-07-03 James Greenhalgh <james.greenhalgh@arm.com>
4660 * config/aarch64/aarch64-builtins.c
4661 (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
4662 * config/aarch64/aarch64-simd-builtins.def (ld1): New.
4664 * config/aarch64/aarch64-simd.md
4665 (aarch64_ld1<VALL:mode>): New.
4666 (aarch64_st1<VALL:mode>): Likewise.
4667 * config/aarch64/arm_neon.h
4668 (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
4670 2013-07-02 Sriraman Tallam <tmsriram@google.com>
4672 * config/i386/i386.c (gate_insert_vzeroupper): Check if
4674 (ix86_option_override_internal):Turn on all -mavx target flags by
4675 default as they are dependent on AVX anyway.
4677 2013-07-02 Cary Coutant <ccoutant@google.com>
4679 * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
4681 (loc_checksum_ordered): Likewise.
4682 (hash_loc_operands): Remove inline keyword.
4684 2013-07-02 Jakub Jelinek <jakub@redhat.com>
4686 PR tree-optimization/57741
4687 * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
4688 non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
4689 or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
4690 Allow REAL_CST step_exprs if flag_associative_math.
4691 (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
4693 2013-07-02 Ian Bolton <ian.bolton@arm.com>
4695 * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
4697 2013-07-02 Ian Bolton <ian.bolton@arm.com>
4699 * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
4701 2013-07-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4703 * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
4705 (iorsi3_insn): Likewise.
4706 (arm_xorsi3): Likewise.
4708 2013-07-01 Sofiane Naci <sofiane.naci@arm.com>
4710 * arm.md (attribute "wtype"): Delete. Move attribute values from here
4712 (attribute "type"): ... here, and prefix with "wmmx_".
4713 (attribute "core_cycles"): Update for attribute changes.
4714 * iwmmxt.md (tbcstv8qi): Update for attribute changes.
4715 (tbcstv4hi): Likewise.
4716 (tbcstv2si): Likewise.
4717 (iwmmxt_iordi3): Likewise.
4718 (iwmmxt_xordi3): Likewise.
4719 (iwmmxt_anddi3): Likewise.
4720 (iwmmxt_nanddi3): Likewise.
4721 (iwmmxt_arm_movdi): Likewise.
4722 (iwmmxt_movsi_insn): Likewise.
4723 (mov<mode>_internal): Likewise.
4724 (and<mode>3_iwmmxt): Likewise.
4725 (ior<mode>3_iwmmxt): Likewise.
4726 (xor<mode>3_iwmmxt): Likewise.
4727 (add<mode>3_iwmmxt): Likewise.
4728 (ssaddv8qi3): Likewise.
4729 (ssaddv4hi3): Likewise.
4730 (ssaddv2si3): Likewise.
4731 (usaddv8qi3): Likewise.
4732 (usaddv4hi3): Likewise.
4733 (usaddv2si3): Likewise.
4734 (sub<mode>3_iwmmxt): Likewise.
4735 (sssubv8qi3): Likewise.
4736 (sssubv4hi3): Likewise.
4737 (sssubv2si3): Likewise.
4738 (ussubv8qi3): Likewise.
4739 (ussubv4hi3): Likewise.
4740 (ussubv2si3): Likewise.
4741 (mulv4hi3_iwmmxt): Likewise.
4742 (smulv4hi3_highpart): Likewise.
4743 (umulv4hi3_highpart): Likewise.
4744 (iwmmxt_wmacs): Likewise.
4745 (iwmmxt_wmacsz): Likewise.
4746 (iwmmxt_wmacu): Likewise.
4747 (iwmmxt_wmacuz): Likewise.
4748 (iwmmxt_clrdi): Likewise.
4749 (iwmmxt_clrv8qi): Likewise.
4750 (iwmmxt_clr4hi): Likewise.
4751 (iwmmxt_clr2si): Likewise.
4752 (iwmmxt_uavgrndv8qi3): Likewise.
4753 (iwmmxt_uavgrndv4hi3): Likewise.
4754 (iwmmxt_uavgv8qi3): Likewise.
4755 (iwmmxt_uavgv4hi3): Likewise.
4756 (iwmmxt_tinsrb): Likewise.
4757 (iwmmxt_tinsrh): Likewise.
4758 (iwmmxt_tinsrw): Likewise.
4759 (iwmmxt_textrmub): Likewise.
4760 (iwmmxt_textrmsb): Likewise.
4761 (iwmmxt_textrmuh): Likewise.
4762 (iwmmxt_textrmsh): Likewise.
4763 (iwmmxt_textrmw): Likewise.
4764 (iwmxxt_wshufh): Likewise.
4765 (eqv8qi3): Likewise.
4766 (eqv4hi3): Likewise.
4767 (eqv2si3): Likewise.
4768 (gtuv8qi3): Likewise.
4769 (gtuv4hi3): Likewise.
4770 (gtuv2si3): Likewise.
4771 (gtv8qi3): Likewise.
4772 (gtv4hi3): Likewise.
4773 (gtv2si3): Likewise.
4774 (smax<mode>3_iwmmxt): Likewise.
4775 (umax<mode>3_iwmmxt): Likewise.
4776 (smin<mode>3_iwmmxt): Likewise.
4777 (umin<mode>3_iwmmxt): Likewise.
4778 (iwmmxt_wpackhss): Likewise.
4779 (iwmmxt_wpackwss): Likewise.
4780 (iwmmxt_wpackdss): Likewise.
4781 (iwmmxt_wpackhus): Likewise.
4782 (iwmmxt_wpackwus): Likewise.
4783 (iwmmxt_wpackdus): Likewise.
4784 (iwmmxt_wunpckihb): Likewise.
4785 (iwmmxt_wunpckihh): Likewise.
4786 (iwmmxt_wunpckihw): Likewise.
4787 (iwmmxt_wunpckilb): Likewise.
4788 (iwmmxt_wunpckilh): Likewise.
4789 (iwmmxt_wunpckilw): Likewise.
4790 (iwmmxt_wunpckehub): Likewise.
4791 (iwmmxt_wunpckehuh): Likewise.
4792 (iwmmxt_wunpckehuw): Likewise.
4793 (iwmmxt_wunpckehsb): Likewise.
4794 (iwmmxt_wunpckehsh): Likewise.
4795 (iwmmxt_wunpckehsw): Likewise.
4796 (iwmmxt_wunpckelub): Likewise.
4797 (iwmmxt_wunpckeluh): Likewise.
4798 (iwmmxt_wunpckeluw): Likewise.
4799 (iwmmxt_wunpckelsb): Likewise.
4800 (iwmmxt_wunpckelsh): Likewise.
4801 (iwmmxt_wunpckelsw): Likewise.
4802 (ror<mode>3): Likewise.
4803 (ashr<mode>3_iwmmxt): Likewise.
4804 (lshr<mode>3_iwmmxt): Likewise.
4805 (ashl<mode>3_iwmmxt): Likewise.
4806 (ror<mode>3_di): Likewise.
4807 (ashr<mode>3_di): Likewise.
4808 (lshr<mode>3_di): Likewise.
4809 (ashl<mode>3_di): Likewise.
4810 (iwmmxt_wmadds): Likewise.
4811 (iwmmxt_wmaddu): Likewise.
4812 (iwmmxt_tmia): Likewise.
4813 (iwmmxt_tmiaph): Likewise.
4814 (iwmmxt_tmiabb): Likewise.
4815 (iwmmxt_tmiatb): Likewise.
4816 (iwmmxt_tmiabt): Likewise.
4817 (iwmmxt_tmiatt): Likewise.
4818 (iwmmxt_tmovmskb): Likewise.
4819 (iwmmxt_tmovmskh): Likewise.
4820 (iwmmxt_tmovmskw): Likewise.
4821 (iwmmxt_waccb): Likewise.
4822 (iwmmxt_wacch): Likewise.
4823 (iwmmxt_waccw): Likewise.
4824 (iwmmxt_waligni): Likewise.
4825 (iwmmxt_walignr): Likewise.
4826 (iwmmxt_walignr0): Likewise.
4827 (iwmmxt_walignr1): Likewise.
4828 (iwmmxt_walignr2): Likewise.
4829 (iwmmxt_walignr3): Likewise.
4830 (iwmmxt_wsadb): Likewise.
4831 (iwmmxt_wsadh): Likewise.
4832 (iwmmxt_wsadbz): Likewise.
4833 (iwmmxt_wsadhz): Likewise.
4834 * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
4835 (iwmmxt_wabsdiffb): Likewise.
4836 (iwmmxt_wabsdiffh): Likewise.
4837 (iwmmxt_wabsdiffw): Likewise.
4838 (iwmmxt_waddsubhx): Likewise
4839 (iwmmxt_wsubaddhx): Likewise.
4840 (addc<mode>3): Likewise.
4841 (iwmmxt_avg4): Likewise.
4842 (iwmmxt_avg4r): Likewise.
4843 (iwmmxt_wmaddsx): Likewise.
4844 (iwmmxt_wmaddux): Likewise.
4845 (iwmmxt_wmaddsn): Likewise.
4846 (iwmmxt_wmaddun): Likewise.
4847 (iwmmxt_wmulwsm): Likewise.
4848 (iwmmxt_wmulwum): Likewise.
4849 (iwmmxt_wmulsmr): Likewise.
4850 (iwmmxt_wmulumr): Likewise.
4851 (iwmmxt_wmulwsmr): Likewise.
4852 (iwmmxt_wmulwumr): Likewise.
4853 (iwmmxt_wmulwl): Likewise.
4854 (iwmmxt_wqmulm): Likewise.
4855 (iwmmxt_wqmulwm): Likewise.
4856 (iwmmxt_wqmulmr): Likewise.
4857 (iwmmxt_wqmulwmr): Likewise.
4858 (iwmmxt_waddbhusm): Likewise.
4859 (iwmmxt_waddbhusl): Likewise.
4860 (iwmmxt_wqmiabb): Likewise.
4861 (iwmmxt_wqmiabt): Likewise.
4862 (iwmmxt_wqmiatb): Likewise.
4863 (iwmmxt_wqmiatt): Likewise.
4864 (iwmmxt_wqmiabbn): Likewise.
4865 (iwmmxt_wqmiabtn): Likewise.
4866 (iwmmxt_wqmiatbn): Likewise.
4867 (iwmmxt_wqmiattn): Likewise.
4868 (iwmmxt_wmiabb): Likewise.
4869 (iwmmxt_wmiabt): Likewise.
4870 (iwmmxt_wmiatb): Likewise.
4871 (iwmmxt_wmiatt): Likewise.
4872 (iwmmxt_wmiabbn): Likewise.
4873 (iwmmxt_wmiabtn): Likewise.
4874 (iwmmxt_wmiatbn): Likewise.
4875 (iwmmxt_wmiattn): Likewise.
4876 (iwmmxt_wmiawbb): Likewise.
4877 (iwmmxt_wmiawbt): Likewise.
4878 (iwmmxt_wmiawtb): Likewise.
4879 (iwmmxt_wmiawtt): Likewise.
4880 (iwmmxt_wmiawbbn): Likewise.
4881 (iwmmxt_wmiawbtn): Likewise.
4882 (iwmmxt_wmiawtbn): Likewise.
4883 (iwmmxt_wmiawttn): Likewise.
4884 (iwmmxt_wmerge): Likewise.
4885 (iwmmxt_tandc<mode>3): Likewise.
4886 (iwmmxt_torc<mode>3): Likewise.
4887 (iwmmxt_torvsc<mode>3): Likewise.
4888 (iwmmxt_textrc<mode>3): Likewise.
4889 * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
4890 (wmmxt_pack): Likewise.
4891 (wmmxt_mult_c1): Likewise.
4892 (wmmxt_mult_c2): Likewise.
4893 (wmmxt_alu_c1): Likewise.
4894 (wmmxt_alu_c2): Likewise.
4895 (wmmxt_alu_c3): Likewise.
4896 (wmmxt_transfer_c1): Likewise.
4897 (wmmxt_transfer_c2): Likewise.
4898 (wmmxt_transfer_c3): Likewise.
4899 (marvell_f_iwmmxt_wstr): Likewise.
4900 (marvell_f_iwmmxt_wldr): Likewise.
4902 2013-06-29 Yufeng Zhang <yufeng.zhang@arm.com>
4904 * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
4906 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
4909 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
4910 * lra-constraints.c (need_for_split_p): Check call used hard regs
4911 living through calls.
4913 * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
4914 call used regs for call insn.
4916 2013-06-28 Jakub Jelinek <jakub@redhat.com>
4919 * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
4920 mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
4922 2013-06-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
4924 * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
4926 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
4928 * lra-constraints.c (need_for_split_p): Check call used hard regs
4929 living through calls.
4931 2013-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4934 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
4935 to tie with any other modes. Eliminate Altivec vector mode tests,
4936 since these are a subset of ALTIVEC or VSX vector modes. Simplify
4937 code, to return 0 if testing MODE2 for a condition, if we've
4938 already tested MODE1 for the same condition.
4940 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
4942 * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
4945 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
4947 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4948 Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
4950 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
4952 * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
4954 (aarch64_symbolic_constant_p): Remove.
4955 * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
4956 static. Fix line length and white space.
4957 (aarch64_symbolic_constant_p): Remove.
4958 * config/aarch64/predicates.md (aarch64_valid_symref):
4959 Use aarch64_classify_symbol_expression.
4961 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4963 * config/arm/constraints.md (Ts): New constraint.
4964 * config/arm/arm.md (arm_movqi_insn): Add alternatives for
4966 (compare_scc): Use "Ts" constraint for operand 0.
4967 (ior_scc_scc): Likewise.
4968 (and_scc_scc): Likewise.
4969 (and_scc_scc_nodom): Likewise.
4970 (ior_scc_scc_cmp): Likewise for operand 7.
4971 (and_scc_scc_cmp): Likewise.
4972 * config/arm/thumb2.md (thumb2_movsi_insn):
4973 Add alternatives for 16-bit encodings.
4974 (thumb2_movhi_insn): Likewise.
4975 (thumb2_movsicc_insn): Likewise.
4976 (thumb2_and_scc): Take 'and' outside cond_exec. Use "Ts" constraint.
4977 (thumb2_negscc): Use "Ts" constraint.
4978 Move mvn instruction outside cond_exec block.
4979 * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
4980 for 16-bit encodings.
4982 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4984 * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
4986 (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
4987 (mulsi3subsi): Likewise.
4988 (mulsidi3adddi): Likewise.
4989 (mulsidi3_v6): Likewise.
4990 (umulsidi3_v6): Likewise.
4991 (umulsidi3adddi_v6): Likewise.
4992 (smulsi3_highpart_v6): Likewise.
4993 (umulsi3_highpart_v6): Likewise.
4994 (mulhisi3tb): Likewise.
4995 (mulhisi3bt): Likewise.
4996 (mulhisi3tt): Likewise.
4997 (maddhisi4): Likewise.
4998 (maddhisi4tb): Likewise.
4999 (maddhisi4tt): Likewise.
5000 (maddhidi4): Likewise.
5001 (maddhidi4tb): Likewise.
5002 (maddhidi4tt): Likewise.
5003 (zeroextractsi_compare0_scratch): Likewise.
5004 (insv_zero): Likewise.
5005 (insv_t2): Likewise.
5006 (anddi_notzesidi_di): Likewise.
5007 (anddi_notsesidi_di): Likewise.
5008 (andsi_notsi_si): Likewise.
5009 (iordi_zesidi_di): Likewise.
5010 (xordi_zesidi_di): Likewise.
5011 (andsi_iorsi3_notsi): Likewise.
5013 (smax_m1): Likewise.
5015 (not_shiftsi): Likewise.
5016 (unaligned_loadsi): Likewise.
5017 (unaligned_loadhis): Likewise.
5018 (unaligned_loadhiu): Likewise.
5019 (unaligned_storesi): Likewise.
5020 (unaligned_storehi): Likewise.
5021 (extv_reg): Likewise.
5022 (extzv_t2): Likewise.
5024 (udivsi3): Likewise.
5025 (arm_zero_extendhisi2addsi): Likewise.
5026 (arm_zero_extendqisi2addsi): Likewise.
5027 (compareqi_eq0): Likewise.
5028 (arm_extendhisi2_v6): Likewise.
5029 (arm_extendqisi2addsi): Likewise.
5030 (arm_movt): Likewise.
5031 (thumb2_ldrd): Likewise.
5032 (thumb2_ldrd_base): Likewise.
5033 (thumb2_ldrd_base_neg): Likewise.
5034 (thumb2_strd): Likewise.
5035 (thumb2_strd_base): Likewise.
5036 (thumb2_strd_base_neg): Likewise.
5037 (arm_negsi2): Add alternative for 16-bit encoding.
5038 (arm_one_cmplsi2): Likewise.
5040 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5042 * config/arm/predicates.md (arm_cond_move_operator): New predicate.
5043 * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
5044 (movdfcc): Likewise.
5045 * config/arm/vfp.md (*thumb2_movsf_vfp):
5046 Disable predication for arm_restrict_it.
5047 (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
5048 (*thumb2_movdfcc_vfp): Likewise.
5049 (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
5050 *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
5051 *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
5052 *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
5053 *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
5054 *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
5055 *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
5056 *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
5057 *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
5058 *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
5059 *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
5060 *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
5061 Disable predication for arm_restrict_it.
5063 2013-06-28 Kirill Yukhin <kirill.yukhin@intel.com>
5065 * config/i386/bmiintrin.h (_bextr_u32): New.
5066 (_bextr_u64): Ditto.
5068 2013-06-27 Richard Sandiford <rdsandiford@googlemail.com>
5070 * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
5071 (mips64r5900el-*-elf*): Include mips/n32-elf.h.
5072 * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
5073 (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
5074 * config/mips/n32-elf.h: ...this new file.
5076 2013-06-27 Marc Glisse <marc.glisse@inria.fr>
5079 * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
5080 IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
5082 2013-06-27 Catherine Moore <clm@codesourcery.com>
5084 * config/mips/mips-tables.opt: Regenerate.
5085 * config/mips/mips-cpus.def: Add m14ke and m14kec.
5086 * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
5087 * doc/invoke.texi: Add -m14kc.
5089 2013-06-27 Jakub Jelinek <jakub@redhat.com>
5092 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
5093 constraints of operand 1 and 2.
5096 * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
5097 to match RTL canonicalization. Swap predicates and
5098 constraints of operand 1 and 2.
5100 2013-06-27 Vladimir Makarov <vmakarov@redhat.com>
5102 * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
5103 Process OP_INOUT regs for splitting too.
5105 2013-06-27 Jakub Jelinek <jakub@redhat.com>
5107 * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
5108 decl before the loop, initialize to NULL.
5109 (vectorizable_load): Initialize ptr_incr to NULL.
5111 2013-06-27 Martin Jambor <mjambor@suse.cz>
5114 * ipa-ref.h (ipa_maybe_record_reference): Declare.
5115 * ipa-ref.c (ipa_maybe_record_reference): New function.
5116 * cgraphclones.c (cgraph_create_virtual_clone): Use it.
5117 * ipa-cp.c (create_specialized_node): Record potential references from
5119 * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
5121 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
5123 * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
5124 parameter 'mode' of type 'enum machine_mode mode'; change to pass
5125 'mode' to force_reg.
5126 (aarch64_add_offset): Update calls to aarch64_force_temporary.
5127 (aarch64_expand_mov_immediate): Likewise.
5129 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
5131 * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
5132 'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
5134 2013-06-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5136 * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
5137 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5138 (struct machine_function): Add tbegin_p.
5139 (s390_canonicalize_comparison): Fold CC mode compares to
5140 conditional jump if possible.
5141 (s390_emit_jump): Return the emitted jump.
5142 (s390_branch_condition_mask, s390_branch_condition_mnemonic):
5143 Handle CCRAWmode compares.
5144 (s390_option_override): Default to -mhtm if available.
5145 (s390_reg_clobbered_rtx): Handle floating point regs as well.
5146 (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
5147 FPRs instead of df_regs_ever_live_p.
5148 (s390_optimize_nonescaping_tx): New function.
5149 (s390_init_frame_layout): Extend clobbered_regs array to cover
5151 (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
5152 (s390_expand_tbegin): New function.
5153 (enum s390_builtin): New enum definition.
5154 (code_for_builtin): New array definition.
5155 (s390_init_builtins): New function.
5156 (s390_expand_builtin): New function.
5157 (TARGET_INIT_BUILTINS): Define.
5158 (TARGET_EXPAND_BUILTIN): Define.
5159 * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
5160 * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
5161 (s390_alc_comparison): Likewise.
5162 * config/s390/s390-modes.def: Add CCRAWmode.
5163 * config/s390/s390.h (processor_flags): Add PF_TX.
5164 (TARGET_CPU_HTM): Define macro.
5165 (TARGET_HTM): Define macro.
5166 (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
5167 * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
5168 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5169 (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
5170 (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
5171 (TBEGIN_MASK, TBEGINC_MASK): New constants.
5172 ("*cc_to_int"): Move up.
5173 ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
5174 constants other than 0.
5175 ("*ccraw_to_int"): New insn and splitter definition.
5176 ("tbegin", "tbegin_nofloat", "tbegin_retry")
5177 ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
5178 ("tx_assist"): New expander.
5179 ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
5180 ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
5181 * config/s390/s390.opt: Add -mhtm option.
5182 * config/s390/s390-protos.h (s390_emit_jump): Add return type.
5183 * config/s390/htmxlintrin.h: New file.
5184 * config/s390/htmintrin.h: New file.
5185 * config/s390/s390intrin.h: New file.
5186 * doc/extend.texi: Document htm builtins.
5187 * config.gcc: Add the new header files to extra_headers.
5189 2013-06-26 Thomas Schwinge <thomas@codesourcery.com>
5191 * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
5192 (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
5194 2013-06-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5195 Pat Haugen <pthaugen@us.ibm.com>
5196 Peter Bergner <bergner@vnet.ibm.com>
5198 * config/rs6000/power8.md: New.
5199 * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
5200 setting for power8 entry.
5201 * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
5202 * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
5203 test for Power4/Power5 only.
5204 (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
5206 (force_new_group): Adjust comment.
5207 * config/rs6000/rs6000.md: Include power8.md.
5209 2013-06-26 Greta Yorsh <Greta.Yorsh@arm.com>
5211 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
5212 * config/arm/arm-protos.h (arm_max_conditional_execute): New
5214 (tune_params): Update comment.
5215 * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
5216 (arm_max_conditional_execute): New function.
5217 (thumb2_final_prescan_insn): Use max_insn_skipped and
5218 MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
5220 2013-06-25 Jakub Jelinek <jakub@redhat.com>
5222 PR tree-optimization/57705
5223 * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
5224 SSA_NAME step, provided that it is not defined inside the loop.
5225 (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
5226 (get_initial_def_for_induction): Handle SSA_NAME IV step.
5228 2013-06-25 Martin Jambor <mjambor@suse.cz>
5231 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
5232 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
5234 (ipa_note_param_call): Initialize member_ptr flag.
5235 (ipa_analyze_indirect_call_uses): Set member_ptr flag.
5236 (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
5237 (ipa_write_indirect_edge_info): Stream member_ptr flag.
5238 (ipa_read_indirect_edge_info): Likewise.
5240 2013-06-25 Richard Biener <rguenther@suse.de>
5243 * passes.c (init_optimization_passes): Move pass_fold_builtins
5244 and pass_dce earlier with -Og.
5246 2013-06-25 Eric Botcazou <ebotcazou@adacore.com>
5248 * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
5249 <BIT_FIELD_REF>: Remove trailing TAB.
5250 * varasm.c (output_constructor_bitfield): Fix formatting glitch and
5253 2013-06-24 Martin Jambor <mjambor@suse.cz>
5255 PR tree-optimization/57358
5256 * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
5257 (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
5258 (ipa_analyze_params_uses): Generate pessimistic info when true.
5260 2013-06-24 Martin Jambor <mjambor@suse.cz>
5262 PR tree-optimization/57539
5263 * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
5264 global.inlined_to of the new node to it. All callers changed.
5265 * ipa-inline-transform.c (clone_inlined_nodes): New variable
5266 inlining_into, pass it to cgraph_clone_node.
5267 * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
5268 ipa_free_edge_args_substructures.
5269 (ipa_edge_duplication_hook): Only add edges from inlined nodes to
5270 rdesc linked list. Do not assert rdesc edges have inlined caller.
5271 Assert we have found an rdesc in the rdesc list.
5273 2013-06-24 Richard Biener <rguenther@suse.de>
5275 * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
5276 (pointer_set_lookup): Declare.
5277 (class pointer_map): New template class implementing a
5278 generic pointer to T map.
5279 (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
5280 pointer_map<T>::contains, pointer_map<T>::insert,
5281 pointer_map<T>::traverse): New functions.
5282 * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
5283 (pointer_set_lookup): New function.
5284 (pointer_set_contains): Use pointer_set_lookup.
5285 (pointer_set_insert): Likewise.
5286 (insert_aux): Remove.
5287 (struct pointer_map_t): Embed a pointer_set_t.
5288 (pointer_map_create): Adjust.
5289 (pointer_map_destroy): Likewise.
5290 (pointer_map_contains): Likewise.
5291 (pointer_map_insert): Likewise.
5292 (pointer_map_traverse): Likewise.
5293 * tree-streamer.h (struct streamer_tree_cache_d): Use a
5294 pointer_map<unsigned> instead of a pointer_map_t.
5295 * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
5296 (streamer_tree_cache_lookup): Likewise.
5297 (streamer_tree_cache_create): Likewise.
5298 (streamer_tree_cache_delete): Likewise.
5299 * lto-streamer.h (struct lto_tree_ref_encoder): Use a
5300 pointer_map<unsigned> instead of a pointer_map_t.
5301 (lto_init_tree_ref_encoder): Adjust.
5302 (lto_destroy_tree_ref_encoder): Likewise.
5303 * lto-section-out.c (lto_output_decl_index): Likewise.
5304 (lto_record_function_out_decl_state): Likewise.
5305 * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
5307 2013-06-24 Richard Biener <rguenther@suse.de>
5309 PR tree-optimization/57488
5310 * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
5312 2013-06-24 Alan Modra <amodra@gmail.com>
5314 * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
5315 (gen_easy_altivec_constant): Likewise.
5316 * config/rs6000/predicates.md (easy_vector_constant_add_self,
5317 easy_vector_constant_msb): Likewise.
5319 2013-06-23 Jakub Jelinek <jakub@redhat.com>
5322 * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
5323 add missing return true.
5325 2013-06-23 Oleg Endo <olegendo@gcc.gnu.org>
5328 * config/sh/predicates.md (general_extend_operand): Invoke
5329 general_movsrc_operand for memory operands.
5330 (general_movsrc_operand): Allow reg+reg addressing, do not use
5331 general_operand for memory operands.
5333 2013-06-23 Sriraman Tallam <tmsriram@google.com>
5335 * config/i386/i386.c (ix86_pragma_target_parse): Restore target
5336 when current target options does not apply.
5337 * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
5338 * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
5339 * config/i386/bmiintrin.h: Pass appropriate target
5340 attributes to header.
5341 * config/i386/mmintrin.h: Ditto.
5342 * config/i386/nmmintrin.h: Ditto.
5343 * config/i386/avx2intrin.h: Ditto.
5344 * config/i386/fxsrintrin.h: Ditto.
5345 * config/i386/tbmintrin.h: Ditto.
5346 * config/i386/xsaveintrin.h: Ditto.
5347 * config/i386/f16cintrin.h: Ditto.
5348 * config/i386/xtestintrin.h: Ditto.
5349 * config/i386/xsaveoptintrin.h: Ditto.
5350 * config/i386/bmi2intrin.h: Ditto.
5351 * config/i386/lzcntintrin.h: Ditto.
5352 * config/i386/smmintrin.h: Ditto.
5353 * config/i386/wmmintrin.h: Ditto.
5354 * config/i386/x86intrin.h: Remove all header include guards.
5355 * config/i386/prfchwintrin.h: Ditto.
5356 * config/i386/pmmintrin.h: Ditto.
5357 * config/i386/tmmintrin.h: Ditto.
5358 * config/i386/xmmintrin.h: Ditto.
5359 * config/i386/popcntintrin.h: Ditto.
5360 * config/i386/rdseedintrin.h: Ditto.
5361 * config/i386/ammintrin.h: Ditto.
5362 * config/i386/emmintrin.h: Ditto.
5363 * config/i386/immintrin.h: Remove all header include guards.
5364 * config/i386/fma4intrin.h: Ditto.
5365 * config/i386/lwpintrin.h: Ditto.
5366 * config/i386/xopintrin.h: Ditto.
5367 * config/i386/ia32intrin.h: Ditto.
5368 * config/i386/avxintrin.h: Ditto.
5369 * config/i386/rtmintrin.h: Ditto.
5370 * config/i386/fmaintrin.h: Ditto.
5371 * config/i386/mm3dnow.h: Ditto.
5373 2013-06-22 Sriraman Tallam <tmsriram@google.com>
5375 * common/config/i386/i386-common.c: Handle LZCNT.
5377 2013-06-22 Andi Kleen <ak@linux.intel.com>
5379 * doc/extend.texi: Use __atomic_store_n instead of
5380 __atomic_store in HLE example.
5382 2013-06-22 Oleg Endo <olegendo@gcc.gnu.org>
5384 * config/sh/sh.c: Remove <cstdlib> workaround.
5386 2013-06-21 Andi Kleen <ak@linux.intel.com>
5388 * doc/extend.texi: Dont use __atomic_clear in HLE example. Fix typo.
5390 2013-06-21 Andi Kleen <ak@linux.intel.com>
5392 * doc/extend.texi: Document that __atomic_clear and
5393 __atomic_test_and_set should only be used with bool.
5395 2013-06-20 Jan Hubicka <jh@suse.cz>
5397 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
5399 * tree.c (decls_same_for_odr): New function.
5400 (same_for_edr): New function.
5401 (types_same_for_odr): New function.
5402 (get_binfo_at_offset): Use it.
5403 * tree.h (types_same_for_odr): Declare.
5405 2013-06-20 Oleg Endo <olegendo@gcc.gnu.org>
5406 Jason Merrill <jason@redhat.com>
5408 * system.h: Include <cstdlib> as well as <stdlib.h>.
5410 2013-06-20 Uros Bizjak <ubizjak@gmail.com>
5413 * config/i386/i386.c (construct_container): Report error if
5414 long double is used with disabled x87 float returns.
5416 2013-06-20 Jan Hubicka <jh@suse.cz>
5418 * lto-cgraph.c (input_symtab): Do not set cgraph state.
5420 2013-06-20 Joern Rennecke <joern.rennecke@embecosm.com>
5422 PR rtl-optimization/57425
5423 PR rtl-optimization/57569
5424 * alias.c (write_dependence_p): Remove parameters mem_mode and
5425 canon_mem_addr. Add parameters x_mode, x_addr and x_canonicalized.
5426 Changed all callers.
5427 (canon_anti_dependence): Get comments and semantics in sync.
5428 Add parameter mem_canonicalized. Changed all callers.
5429 * rtl.h (canon_anti_dependence): Update prototype.
5431 2013-06-20 Richard Biener <rguenther@suse.de>
5433 * data-streamer-in.c (streamer_read_uhwi): Optimize single
5434 byte case, inline streamer_read_uchar and defer section
5437 2013-06-20 Richard Biener <rguenther@suse.de>
5439 PR tree-optimization/57584
5440 * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
5441 SSA names into the expanded expression that take part in
5442 abnormal coalescing.
5444 2013-06-19 Sharad Singhai <singhai@google.com>
5446 * gcov.c (print_usage): Handle new option.
5447 (process_args): Ditto.
5448 (get_gcov_intermediate_filename): New function.
5449 (output_intermediate_file): New function.
5450 (output_gcov_file): New function
5451 (generate_results): Handle new option.
5452 (release_function): Relase demangled name.
5453 (read_graph_file): Handle demangled name.
5454 (output_lines): Ditto.
5455 * doc/gcov.texi: Document gcov intermediate format.
5457 2013-06-19 Vladimir Makarov <vmakarov@redhat.com>
5460 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
5461 (lra_emit_add): Use the functions. Add comment about Y as an
5464 2013-06-19 David Edelsohn <dje.gcc@gmail.com>
5467 * collect2.c (collect_atexit): New.
5468 (collect_exit): Delete.
5469 (main): Register collect_atexit with atexit.
5470 (collect_wait): Change collect_exit to exit.
5472 * collect2.h (collect_exit): Delete.
5473 * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
5475 2013-06-19 Wei Mi <wmi@google.com>
5477 PR rtl-optimization/57518
5478 * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
5479 if regno is used in paradoxical subreg.
5480 (update_equiv_regs): Check pdx_subregs[regno] before
5481 set a reg to be equivalent with a mem.
5483 2013-06-19 Matthias Klose <doko@ubuntu.com>
5486 * file-find.h (find_a_file): Add a mode parameter.
5487 * file-find.c (find_a_file): Likewise.
5488 * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
5489 with X_OK for the executables.
5490 * collect2.c (main): Call find_a_file with X_OK.
5492 2013-06-19 Steve Ellcey <sellcey@mips.com>
5495 * config/mips/mips.md (casesi_internal_mips16_<mode>):
5496 Use NEXT_INSN instead of next_real_insn.
5498 2013-06-19 Jan Hubicka <jh@suse.cz>
5500 * cgraph.h (const_value_known_p): Replace by ...
5501 (ctor_for_folding): .. this one.
5502 * cgraphunit.c (process_function_and_variable_attributes): Use it.
5503 * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
5504 * expr.c (expand_expr_real_1): Likewise.
5505 (string_constant): Likewise.
5506 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
5507 * ipa.c (process_references): Likewise.
5508 (symtab_remove_unreachable_nodes): Likewise.
5509 * ipa-inline-analysis.c (param_change_prob): Likewise.
5510 * gimple-fold.c (canonicalize_constructor_val): Likewise.
5511 (get_base_constructor): Likwise.
5512 * varpool.c (varpool_remove_node): Likewise.
5513 (varpool_remove_initializer): LIkewise.
5514 (dump_varpool_node): LIkwise.
5515 (const_value_known_p): Rewrite to ...
5516 (ctor_for_folding): ... this one.
5518 2013-06-19 Jakub Jelinek <jakub@redhat.com>
5521 * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
5522 PERSONALITY in $PATH derived prefixes.
5524 2013-06-19 Jeff Law <law@redhat.com>
5526 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
5529 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
5530 (simplify_bitwise_binary): Use it to simpify certain binary ops on
5533 2013-06-19 Sofiane Naci <sofiane.naci@arm.com>
5535 * config/arm/vfp.md: Move VFP instruction classification documentation
5537 * config/arm/arm.md: ... here. Update instruction classification
5540 2013-06-19 Richard Earnshaw <rearnsha@arm.com>
5542 arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
5543 (peepholes for eq(reg, not-0)): Ensure condition register is dead after
5544 pattern. Use more efficient sequences on ARMv5 and Thumb2.
5546 2013-06-19 Steven Bosscher <steven@gcc.gnu.org>
5549 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5550 with NEXT_INSN. Use tablejump_p to check for jump table data
5553 2013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
5556 * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
5557 that now in C++ the value is correct per the C++ standards.
5559 2013-06-19 Richard Biener <rguenther@suse.de>
5561 * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
5564 2013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5567 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5570 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5571 with next_active_insn.
5573 2013-06-18 Sriraman Tallam <tmsriram@google.com>
5575 * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
5576 functions are inlined during failures to flag an error.
5577 * tree-inline.c (expand_call_inline): Allow the error to be flagged
5578 in early inline pass.
5580 2013-06-18 H.J. Lu <hongjiu.lu@intel.com>
5582 * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
5585 2013-06-18 Julian Brown <julian@codesourcery.com>
5587 * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
5588 Permit virtual register pre-reload if !strict.
5589 (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
5591 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
5593 * config/arm/neon.md (movmisalign<mode>): Use
5594 neon_perm_struct_or_reg_operand instead of
5595 neon_struct_or_register_operand.
5596 (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
5597 neon_permissive_struct_operand instead of neon_struct_operand.
5598 * config/arm/constraints.md (Un, Um, Us): Adjust calls to
5599 neon_vector_mem_operand.
5600 * config/arm/predicates.md (neon_struct_operand): Adjust call to
5601 neon_vector_mem_operand.
5602 (neon_permissive_struct_operand): New.
5603 (neon_struct_or_register_operand): Rename to...
5604 (neon_perm_struct_or_reg_operand): This. Adjust call to
5605 neon_vector_mem_operand.
5607 2013-06-18 Richard Biener <rguenther@suse.de>
5609 * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
5610 * lto-streamer.h: Include pointer-set.h.
5611 (struct lto_decl_slot): Remove.
5612 (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
5613 Remove next_index entry.
5614 (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5615 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5616 (lto_init_tree_ref_encoder): Adjust.
5617 (lto_destroy_tree_ref_encoder): Likewise.
5618 * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
5619 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
5620 (lto_output_decl_index): Adjust.
5621 (lto_new_out_decl_state): Likewise.
5622 (lto_record_function_out_decl_state): Likewise.
5623 * lto-streamer-out.c (copy_function): Likewise.
5625 2013-06-18 Richard Biener <rguenther@suse.de>
5627 * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
5628 * cgraphunit.c: Include cfgloop.h.
5629 (init_lowered_empty_function): Initialize the loop tree.
5630 (assemble_thunk): Insert new BBs into loops.
5632 2013-06-18 Richard Biener <rguenther@suse.de>
5634 * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
5635 * tree-streamer.c (streamer_tree_cache_create): Make maintaining
5636 the map from cache entry to cache index optional.
5637 (streamer_tree_cache_replace_tree): Adjust accordingly.
5638 (streamer_tree_cache_append): Likewise.
5639 (streamer_tree_cache_delete): Likewise.
5640 * lto-streamer-in.c (lto_data_in_create): Do not maintain the
5641 streamer cache map from cache entry to cache index.
5642 * lto-streamer-out.c (create_output_block): Adjust.
5644 2013-06-18 Sofiane Naci <sofiane.naci@arm.com>
5646 * config/arm/arm.md (attribute "insn"): Move multiplication and
5647 division attributes to...
5648 (attribute "type"): ... here. Remove mult.
5649 (attribute "mul32"): New attribute.
5650 (attribute "mul64"): Add umaal.
5651 (*arm_mulsi3): Update attributes.
5652 (*arm_mulsi3_v6): Likewise.
5653 (*thumb_mulsi3): Likewise.
5654 (*thumb_mulsi3_v6): Likewise.
5655 (*mulsi3_compare0): Likewise.
5656 (*mulsi3_compare0_v6): Likewise.
5657 (*mulsi_compare0_scratch): Likewise.
5658 (*mulsi_compare0_scratch_v6): Likewise.
5659 (*mulsi3addsi): Likewise.
5660 (*mulsi3addsi_v6): Likewise.
5661 (*mulsi3addsi_compare0): Likewise.
5662 (*mulsi3addsi_compare0_v6): Likewise.
5663 (*mulsi3addsi_compare0_scratch): Likewise.
5664 (*mulsi3addsi_compare0_scratch_v6): Likewise.
5665 (*mulsi3subsi): Likewise.
5666 (*mulsidi3adddi): Likewise.
5667 (*mulsi3addsi_v6): Likewise.
5668 (*mulsidi3adddi_v6): Likewise.
5669 (*mulsidi3_nov6): Likewise.
5670 (*mulsidi3_v6): Likewise.
5671 (*umulsidi3_nov6): Likewise.
5672 (*umulsidi3_v6): Likewise.
5673 (*umulsidi3adddi): Likewise.
5674 (*umulsidi3adddi_v6): Likewise.
5675 (*smulsi3_highpart_nov6): Likewise.
5676 (*smulsi3_highpart_v6): Likewise.
5677 (*umulsi3_highpart_nov6): Likewise.
5678 (*umulsi3_highpart_v6): Likewise.
5679 (mulhisi3): Likewise.
5680 (*mulhisi3tb): Likewise.
5681 (*mulhisi3bt): Likewise.
5682 (*mulhisi3tt): Likewise.
5683 (maddhisi4): Likewise.
5684 (*maddhisi4tb): Likewise.
5685 (*maddhisi4tt): Likewise.
5686 (maddhidi4): Likewise.
5687 (*maddhidi4tb): Likewise.
5688 (*maddhidi4tt): Likewise.
5690 (udivsi3): Likewise.
5691 * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
5692 (thumb2_mulsi_short_compare0): Likewise.
5693 (thumb2_mulsi_short_compare0_scratch): Likewise.
5694 * config/arm/arm1020e.md (1020mult1): Update attribute change.
5695 (1020mult2): Likewise.
5696 (1020mult3): Likewise.
5697 (1020mult4): Likewise.
5698 (1020mult5): Likewise.
5699 (1020mult6): Likewise.
5700 * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
5702 (cortex_a15_mult64): Likewise.
5703 (cortex_a15_sdiv): Likewise.
5704 (cortex_a15_udiv): Likewise.
5705 * config/arm/arm1026ejs.md (mult1): Update attribute change.
5711 * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
5712 (pj4_ir_div): Likewise.
5713 * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
5714 (11_mult2): Likewise.
5715 (11_mult3): Likewise.
5716 (11_mult4): Likewise.
5717 (11_mult5): Likewise.
5718 (11_mult6): Likewise.
5719 (11_mult7): Likewise.
5720 * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
5721 (cortex_a8_mla): Likewise.
5722 (cortex_a8_mull): Likewise.
5723 (cortex_a8_smulwy): Likewise.
5724 (cortex_a8_smlald): Likewise.
5725 * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
5726 * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
5727 (cortex_r4_mul_3): Likewise.
5728 (cortex_r4_mla_4): Likewise.
5729 (cortex_r4_mla_3): Likewise.
5730 (cortex_r4_smlald): Likewise.
5731 (cortex_r4_mull): Likewise.
5732 (cortex_r4_sdiv): Likewise.
5733 (cortex_r4_udiv): Likewise.
5734 * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
5735 (cortex_a7_idiv): Likewise.
5736 * config/arm/arm926ejs.md (9_mult1): Update attribute change.
5737 (9_mult2): Likewise.
5738 (9_mult3): Likewise.
5739 (9_mult4): Likewise.
5740 (9_mult5): Likewise.
5741 (9_mult6): Likewise.
5742 * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
5743 (cortex_a53_sdiv): Likewise.
5744 (cortex_a53_udiv): Likewise.
5745 * config/arm/fa726te.md (726te_mult_op): Update attribute change.
5746 * config/arm/fmp626.md (mp626_mult1): Update attribute change.
5747 (mp626_mult2): Likewise.
5748 (mp626_mult3): Likewise.
5749 (mp626_mult4): Likewise.
5750 * config/arm/fa526.md (526_mult1): Update attribute change.
5751 (526_mult2): Likewise.
5752 * config/arm/arm-generic.md (mult): Update attribute change.
5753 (mult_ldsched_strongarm): Likewise.
5754 (mult_ldsched): Likewise.
5755 (multi_cycle): Likewise.
5756 * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
5757 * config/arm/fa606te.md (606te_mult1): Update attribute change.
5758 (606te_mult2): Likewise.
5759 (606te_mult3): Likewise.
5760 (606te_mult4): Likewise.
5761 * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
5762 (cortex_a9_mac16): Likewise.
5763 (cortex_a9_multiply): Likewise.
5764 (cortex_a9_mac): Likewise.
5765 (cortex_a9_multiply_long): Likewise.
5766 * config/arm/fa626te.md (626te_mult1): Update attribute change.
5767 (626te_mult2): Likewise.
5768 (626te_mult3): Likewise.
5769 (626te_mult4): Likewise.
5771 2013-06-18 Richard Biener <rguenther@suse.de>
5774 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
5776 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5779 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
5780 with next_active_insn.
5782 2013-06-18 Alan Modra <amodra@gmail.com>
5784 * config/rs6000/rs6000.h (enum data_align): New.
5785 (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
5786 (DATA_ABI_ALIGNMENT): Define.
5787 (CONSTANT_ALIGNMENT): Correct comment.
5788 * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
5789 * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
5791 2013-06-17 David Malcolm <dmalcolm@redhat.com>
5793 * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
5794 ATTRIBUTE_UNUSED marking.
5796 2013-06-17 Sofiane Naci <sofiane.naci@arm.com>
5798 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
5799 alternative and update.
5800 (aarch64_dup_lanedi): Delete.
5801 * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
5802 * config/aarch64/aarch64-simd-builtins.def: Update.
5804 2013-06-17 Richard Biener <rguenther@suse.de>
5806 * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
5807 (lto_input_scc): Declare.
5808 (lto_input_tree_1): Likewise.
5809 (struct lto_stats_d): Add num_tree_bodies_output and
5810 num_pickle_refs_output.
5811 * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
5812 (lto_read_tree_1): Split out from ...
5813 (lto_read_tree): ... this.
5814 (lto_input_scc): New function.
5815 (lto_input_tree_1): Split out from ...
5816 (lto_input_tree): ... this. Handle LTO_tree_scc.
5817 (lto_data_in_create): Create the streamer cache without hashes.
5818 * lto-streamer-out.c (create_output_block): Create the streamer
5819 cache with hashes when not doing WPA.
5820 (lto_write_tree_1): Split out from ...
5821 (lto_write_tree): ... this.
5822 (get_symbol_initial_value): New function.
5823 (lto_output_tree_1): Split out from ...
5824 (lto_output_tree): ... this. Write trees as series of SCCs
5825 using a DFS walk via DFS_write_tree.
5826 (struct sccs, struct scc_entry): New types.
5827 (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
5828 (DFS_write_tree_body): New function.
5829 (DFS_write_tree): Likewise.
5830 (hash_tree): Likewise.
5831 (scc_entry_compare): Likewise.
5832 (hash_scc): Likewise.
5833 (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
5834 * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
5835 TREE_CHAIN as regular reference.
5836 (streamer_read_integer_cst): Remove.
5837 (streamer_get_pickled_tree): Adjust.
5838 * tree-streamer-out.c (streamer_write_chain): Disable streaming
5839 of DECL_EXTERNALs in BLOCK_VARS for now.
5840 (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
5842 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
5843 Add hash value argument and record that if hashes are recorded
5845 (streamer_tree_cache_insert_1): Adjust.
5846 (streamer_tree_cache_insert): Likewise.
5847 (streamer_tree_cache_insert_at): Rename to ...
5848 (streamer_tree_cache_replace_tree): ... this and adjust.
5849 (streamer_tree_cache_append): Adjust.
5850 (record_common_node): Likewise.
5851 (streamer_tree_cache_create): Add argument whether to
5852 record hash values together with trees.
5853 (streamer_tree_cache_delete): Adjust.
5854 * tree-streamer.h (struct streamer_tree_cache_d): Add
5856 (streamer_read_integer_cst): Remove.
5857 (streamer_tree_cache_insert): Adjust.
5858 (streamer_tree_cache_append): Likewise.
5859 (streamer_tree_cache_insert_at): Rename to ...
5860 (streamer_tree_cache_replace_tree): ... this and adjust.
5861 (streamer_tree_cache_create): Add argument whether to record hashes.
5862 (streamer_tree_cache_get): Rename to ...
5863 (streamer_tree_cache_get_tree): ... this.
5864 (streamer_tree_cache_get_hash): New function.
5865 * tree.c (cache_integer_cst): New function.
5866 * tree.h (cache_integer_cst): Declare.
5867 (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
5868 * lto-symtab.c (lto_varpool_replace_node): Only release
5869 DECL_INITIAL of non-prevailing decls.
5870 * varpool.c (varpool_remove_initializer): Do not release
5871 DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
5873 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
5875 * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
5876 (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
5877 * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
5878 (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
5879 (mul<mode>3_mul3): Handle TARGET_MIPS5900.
5880 (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
5881 (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
5882 (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
5883 instead of TARGET_64BIT.
5884 (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
5885 Require ISA_HAS_<D>DIV.
5887 2013-06-16 Richard Sandiford <rdsandiford@googlemail.com>
5889 * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
5890 (mips*-*-linux*): Move default with_llsc setting to where other
5892 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
5894 (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
5895 Likewise. Remove default with_tune setting. Move default float
5896 setting to its own block. Handle with_llsc in the same block as above.
5898 2013-06-16 Joern Rennecke <joern.rennecke@embecosm.com>
5900 PR rtl-optimization/57425
5901 PR rtl-optimization/57569
5902 * alias.c (write_dependence_p): Add new parameters mem_mode,
5903 canon_mem_addr and mem_canonicalized. Change type of writep to bool.
5904 Changed all callers.
5905 (canon_anti_dependence): New function.
5906 * cse.c (check_dependence): Use canon_anti_dependence.
5907 * cselib.c (cselib_invalidate_mem): Likewise.
5908 * rtl.h (canon_anti_dependence): Declare.
5910 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
5912 * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
5913 * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
5914 ".set mips3" for 64-bit targets.
5916 2013-06-15 Dehao Chen <dehao@google.com>
5918 * tree-flow.h (gimple_check_call_matching_types): Add new argument.
5919 * gimple-low.c (gimple_check_call_matching_types): Likewise.
5920 (gimple_check_call_args): Likewise.
5921 * value-prof.c (check_ic_target): Likewise.
5922 * ipa-inline.c (early_inliner): Likewise.
5923 * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
5924 * cgraph.c (cgraph_create_edge_1): Likewise.
5925 (cgraph_make_edge_direct): Likewise.
5927 2013-06-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5930 * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
5931 rs6000_output_move_128bit to handle emitting quad memory
5932 operations. Set attribute length to 8 bytes.
5934 2013-06-14 Vidya Praveen <vidyapraveen@arm.com>
5936 * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
5938 (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
5939 (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
5940 (aarch64_<su>mlsl<mode>): Likewise.
5942 2013-06-14 Mike Stump <mikestump@comcast.net>
5944 * Makefile.in (TARGET_H): Add insn-codes.h.
5946 2013-06-14 Alan Modra <amodra@gmail.com>
5950 * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
5951 EXPAND_MEMORY and EXPAND_WRITE to recursive call. Don't use
5952 bitfield expansion when EXPAND_MEMORY.
5953 (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
5955 2013-06-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5957 * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
5958 test for clearing quad memory on 32-bit later.
5960 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
5962 * fold-const.c (negate_expr_p): Handle VECTOR_CST.
5963 (fold_negate_expr): Likewise.
5964 (fold_real_zero_addition_p): Handle vectors.
5965 (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
5967 2013-06-14 Alan Modra <amodra@gmail.com>
5969 * varasm.c (force_const_mem): Revert 2013-06-07 change.
5971 2013-06-13 Jan Hubicka <jh@suse.cz>
5973 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
5974 Local comdats are not externally visible.
5975 * symtab.c (dump_symtab_base): Dump externally visible.
5976 (verify_symtab_base): Verify back links in the symtab hash.
5978 2013-06-13 Bin Cheng <bin.cheng@arm.com>
5980 * fold-const.c (operand_equal_p): Consider NOP_EXPR and
5981 CONVERT_EXPR as equal nodes.
5983 2013-06-13 Bin Cheng <bin.cheng@arm.com>
5985 * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
5987 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
5989 * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
5990 Generalize to complex and vector.
5991 * tree.c (build_all_ones_cst): New function.
5992 * tree.h (build_all_ones_cst): Declare it.
5994 2013-06-13 Alan Modra <amodra@gmail.com>
5996 * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
5997 * config/rs6000/rs6000.md (signbittf2): New insn.
5998 (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
5999 (abstf2_internal, cmptf_internal2): Likewise.
6000 * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6002 2013-06-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6003 Pat Haugen <pthaugen@us.ibm.com>
6004 Peter Bergner <bergner@vnet.ibm.com>
6006 * config/rs6000/rs6000.c (emit_load_locked): Add support for
6007 power8 byte, half-word, and quad-word atomic instructions.
6008 (emit_store_conditional): Likewise.
6009 (rs6000_expand_atomic_compare_and_swap): Likewise.
6010 (rs6000_expand_atomic_op): Likewise.
6012 * config/rs6000/sync.md (larx): Add new modes for power8.
6014 (AINT): New mode iterator to include TImode as well as normal
6015 integer modes on power8.
6016 (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
6017 that VSX registers are not considered. Use AINT mode iterator
6018 instead of INT1 to allow inclusion of quad word atomic operations
6020 (load_locked<mode>): Likewise.
6021 (store_conditional<mode>): Likewise.
6022 (atomic_compare_and_swap<mode>): Likewise.
6023 (atomic_exchange<mode>): Likewise.
6024 (atomic_nand<mode>): Likewise.
6025 (atomic_fetch_<fetchop_name><mode>): Likewise.
6026 (atomic_nand_fetch<mode>): Likewise.
6027 (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
6029 (ATOMIC): On power8, add QImode, HImode modes.
6030 (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
6031 modes that promote to SImode.
6032 (load_lockedti): Convert TImode arguments to PTImode, so that we
6033 get a guaranteed even/odd register pair.
6034 (load_lockedpti): Likewise.
6035 (store_conditionalti): Likewise.
6036 (store_conditionalpti): Likewise.
6038 * config/rs6000/rs6000.md (QHI): New mode iterator for power8
6039 atomic load/store instructions.
6042 2013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
6044 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
6046 (insn_count): New attribute, with most cases extracted from...
6047 (length): ...here. Redefine most cases in terms of insn_count.
6048 (single_insn): Delete.
6049 (can_delay): Use insn_count to check for single instructions.
6050 (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
6051 (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
6052 (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
6053 (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
6054 (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
6055 (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
6056 (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
6057 (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
6058 (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
6059 (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
6060 rather than "length".
6061 (tls_get_tp_<mode>): Likewise. Remove redundant "no_delay" attribute.
6062 * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
6063 Use "insn_count" rather than "length".
6064 * config/mips/mips-dsp.md
6065 (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
6066 (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
6069 2013-06-12 Marc Glisse <marc.glisse@inria.fr>
6071 PR tree-optimization/57361
6072 * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
6074 2013-06-12 Sofiane Naci <sofiane.naci@arm.com>
6076 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
6078 (aarch64_simd_combine<mode>): New instruction expansion.
6079 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
6081 * config/aarch64/aarch64.c (aarch64_split_combine): New function.
6082 * config/aarch64/iterators.md (Vdbl): Add entry for DF.
6084 2013-06-12 Jan Hubicka <jh@suse.cz>
6086 * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
6087 decl has when in streaming stage.
6088 * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
6089 * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
6091 2013-06-12 Roland Stigge <stigge@antcom.de>
6094 * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
6096 2013-06-12 Jakub Jelinek <jakub@redhat.com>
6098 PR tree-optimization/57537
6099 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
6100 vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
6102 2013-06-12 Richard Biener <rguenther@suse.de>
6104 * data-streamer.h (streamer_write_char_stream): CSE
6105 obs->current_pointer.
6106 * data-streamer-out.c (streamer_write_uhwi_stream): Inline
6107 streamer_write_char_stream manually and optimize the resulting loop.
6108 (streamer_write_hwi_stream): Likewise.
6110 2013-06-12 Jan Hubicka <jh@suse.cz>
6112 * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
6113 * cgraph.h (varpool_create_empty_node): Declare.
6114 * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
6116 * symtab.c (symtab_unregister_node): Be lax about missin entries
6118 (symtab_get_node): Update comment.
6119 * varpool.c (varpool_create_empty_node): Break out from ...
6120 (varpool_node_for_decl): ... here.
6121 * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
6123 2013-06-12 Eric Botcazou <ebotcazou@adacore.com>
6125 * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
6126 <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
6127 part. Use straight-line flow at the end.
6128 <COMPONENT_REF>: Remove superfluous else.
6129 <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
6131 2013-06-12 Jakub Jelinek <jakub@redhat.com>
6134 * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
6135 target hook even for !TREE_PUBLIC decls. If no resolution info
6136 is available, return false for common and external decls.
6138 2013-06-12 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
6140 * config/rl78/constraints.md (U): New constraint.
6141 * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
6144 2013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
6147 * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
6148 to allow returning address to AT_PLATFORM name.
6150 2013-06-11 Jan Hubicka <jh@suse.cz>
6152 * cgraph.c (cgraph_create_function_alias): Set weakref flag.
6153 * cgraph.h (symtab_node_base): Add weakref flag.
6154 * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
6155 (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
6156 (output_weakrefs): Use weakref flag.
6157 * fold-const.c (simple_operand_p): Handle WEAK.
6158 * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
6159 * ipa.c (varpool_externally_visible_p): Drop weakref.
6160 (function_and_variable_visibility): Update comment; fix weakref
6161 sanity checks; do not clear DECL_WEAK on them.
6162 * lto-cgraph.c (lto_output_node): update.
6163 (lto_output_varpool_node): Update.
6164 (input_overwrite_node): Update.
6165 (input_node): Update.
6166 (input_varpool_node): Update.
6167 * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
6168 (lto_symtab_merge_symbols): Add sanity check.
6169 (lto_symtab_prevailing_decl): Do not special case weakrefs.
6170 * passes.c (rest_of_decl_compilation): Set static flag, too.
6171 * symtab.c (dump_symtab_base): Dump weakref.
6172 (verify_symtab_base): Sanity check weakrefs.
6173 (symtab_make_decl_local): Remove duplicated code.
6174 (symtab_alias_ultimate_target): Simplify.
6175 * varpool.c (varpool_create_variable_alias): Set weakref flag.
6177 2013-06-11 DJ Delorie <dj@redhat.com>
6179 * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
6180 (rl78_unwind_word_mode): New.
6182 2013-06-11 David Malcolm <dmalcolm@redhat.com>
6184 * final.c (debug_prefix_maps): Make static.
6186 2013-06-11 David Malcolm <dmalcolm@redhat.com>
6188 * function.c (initial_trampoline): Remove stray copy.
6190 2013-06-11 Sofiane Naci <sofiane.naci@arm.com>
6192 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
6194 2013-06-11 Martin Jambor <mjambor@suse.cz>
6196 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
6197 within bounds at the beginning of the function.
6199 2013-06-11 Alan Modra <amodra@gmail.com>
6201 * varasm.c (get_section): Don't die on !DECL_P decl. Tidy error
6203 (get_named_section): Don't NULL !DECL_P decl.
6205 2013-06-11 Igor Zamyatin <igor.zamyatin@intel.com>
6207 * doc/invoke.texi (core-avx2): Document.
6209 (atom): Updated with MOVBE.
6211 2013-06-11 Richard Biener <rguenther@suse.de>
6213 * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
6215 2013-06-11 Anton Blanchard <anton@samba.org>
6217 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
6218 correct shift value in little-endian mode.
6220 2013-06-11 Jakub Jelinek <jakub@redhat.com>
6223 * varasm.c (get_variable_align): Move #endif to the right place.
6225 2013-06-10 Cary Coutant <ccoutant@google.com>
6227 * dwarf2out.c (hash_external_ref): Use die_symbol or signature
6228 for hash so that hash table traversal order is deterministic.
6230 2013-06-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6231 Pat Haugen <pthaugen@us.ibm.com>
6232 Peter Bergner <bergner@vnet.ibm.com>
6234 * config/rs6000/vector.md (GPR move splitter): Do not split moves
6235 of vectors in GPRS if they are direct moves or quad word load or
6238 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
6240 (direct_move_p): Likewise.
6241 (quad_load_store_p): Likewise.
6243 * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
6244 classes into bins based on the physical register type.
6245 (reg_class_to_reg_type): Likewise.
6246 (IS_STD_REG_TYPE): Likewise.
6247 (IS_FP_VECT_REG_TYPE): Likewise.
6248 (reload_fpr_gpr): Arrays to determine what insn to use if we can
6249 use direct move instructions.
6250 (reload_gpr_vsx): Likewise.
6251 (reload_vsx_gpr): Likewise.
6252 (rs6000_init_hard_regno_mode_ok): Precalculate the register type
6253 information that is a simplification of register classes. Also
6254 precalculate direct move reload helpers.
6255 (direct_move_p): New function to return true if the operation can
6256 be done as a direct move instruciton.
6257 (quad_load_store_p): New function to return true if the operation
6258 is a quad memory operation.
6259 (rs6000_legitimize_address): If quad memory, only allow register
6260 indirect for TImode addresses.
6261 (rs6000_legitimate_address_p): Likewise.
6262 (enum reload_reg_type): Delete, replace with rs6000_reg_type.
6263 (rs6000_reload_register_type): Likewise.
6264 (register_to_reg_type): Return register type.
6265 (rs6000_secondary_reload_simple_move): New helper function for
6266 secondary reload and secondary memory needed to identify anything
6267 that is a simple move, and does not need reloading.
6268 (rs6000_secondary_reload_direct_move): New helper function for
6269 secondary reload to identify cases that can be done with several
6270 instructions via the direct move instructions.
6271 (rs6000_secondary_reload_move): New helper function for secondary
6272 reload to identify moves between register types that can be done.
6273 (rs6000_secondary_reload): Add support for quad memory operations
6274 and for direct move.
6275 (rs6000_secondary_memory_needed): Likewise.
6276 (rs6000_debug_secondary_memory_needed): Change argument names.
6277 (rs6000_output_move_128bit): New function to return the move to
6278 use for 128-bit moves, including knowing about the various
6279 limitations of quad memory operations.
6281 * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
6282 memory operations. call rs6000_output_move_128bit for the actual
6283 instruciton(s) to generate.
6284 (vsx_movti_64bit): Likewise.
6286 * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
6287 (UNSPEC_P8V_MTVSRWZ): Likewise.
6288 (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
6289 (UNSPEC_P8V_MTVSRD): Likewise.
6290 (UNSPEC_P8V_XXPERMDI): Likewise.
6291 (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
6292 (UNSPEC_FUSION_GPR): Likewise.
6293 (FMOVE128_GPR): New iterator for direct move.
6294 (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
6297 (zero_extend<mode>di2_internal1): Add support for power8 32-bit
6298 loads and direct move instructions.
6299 (zero_extendsidi2_lfiwzx): Likewise.
6300 (extendsidi2_lfiwax): Likewise.
6301 (extendsidi2_nocell): Likewise.
6302 (floatsi<mode>2_lfiwax): Likewise.
6304 (floatunssi<mode>2_lfiwzx): Likewise.
6306 (fix_trunc<mode>_stfiwx): Likewise.
6307 (fixuns_trunc<mode>_stfiwx): Likewise.
6308 (mov<mode>_hardfloat, 32-bit floating point): Likewise.
6309 (mov<move>_hardfloat64, 64-bit floating point): Likewise.
6310 (parity<mode>2_cmpb): Set length/type attr.
6311 (unnamed shift right patterns, mov<mode>_internal2): Change type attr
6312 for 'mr.' to fast_compare.
6313 (bpermd_<mode>): Change type attr to popcnt.
6314 (p8_fmrgow_<mode>): New insns for power8 direct move support.
6315 (p8_mtvsrwz_1): Likewise.
6316 (p8_mtvsrwz_2): Likewise.
6317 (reload_fpr_from_gpr<mode>): Likewise.
6318 (p8_mtvsrd_1): Likewise.
6319 (p8_mtvsrd_2): Likewise.
6320 (p8_xxpermdi_<mode>): Likewise.
6321 (reload_vsx_from_gpr<mode>): Likewise.
6322 (reload_vsx_from_gprsf): Likewise.
6323 (p8_mfvsrd_3_<mode>): LIkewise.
6324 (reload_gpr_from_vsx<mode>): Likewise.
6325 (reload_gpr_from_vsxsf): Likewise.
6326 (p8_mfvsrd_4_disf): Likewise.
6327 (multi-word GPR splits): Do not split direct moves or quad memory
6330 2013-06-10 David Malcolm <dmalcolm@redhat.com>
6332 * tree-into-ssa.c (interesting_blocks): Make static.
6334 2013-06-10 Jakub Jelinek <jakub@redhat.com>
6337 * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
6338 CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
6339 Use DATA_ABI_ALIGNMENT for that case instead if defined.
6340 (get_variable_align): New function.
6341 (get_variable_section, emit_bss, emit_common,
6342 assemble_variable_contents, place_block_symbol): Use
6343 get_variable_align instead of DECL_ALIGN.
6344 (assemble_noswitch_variable): Add align argument, use it
6345 instead of DECL_ALIGN.
6346 (assemble_variable): Adjust caller. Use get_variable_align
6347 instead of DECL_ALIGN.
6348 * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
6350 (DATA_ABI_ALIGNMENT): Define.
6351 * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
6352 * config/i386/i386.c (x86_data_alignment): Add opt argument. If
6353 opt is false, only return the psABI mandated alignment increase.
6354 * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
6355 (DATA_ABI_ALIGNMENT): ... this.
6356 * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
6357 (DATA_ABI_ALIGNMENT): ... this.
6358 * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
6359 * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
6360 (DATA_ABI_ALIGNMENT): ... this.
6361 * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
6362 * doc/tm.texi: Regenerated.
6364 2013-06-10 Uros Bizjak <ubizjak@gmail.com>
6366 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
6367 cmp_code to construct REG_EQUAL note.
6369 2013-06-09 Jakub Jelinek <jakub@redhat.com>
6372 * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
6373 that operands[2] doesn't overlap with operands[0].
6375 2013-06-09 David Edelsohn <dje.gcc@gmail.com>
6376 Jan Hubicka <jh@suse.cz>
6378 * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
6379 hack to mark symbols as used.
6381 2013-06-08 Vladimir Makarov <vmakarov@redhat.com>
6383 PR rtl-optimization/57559
6384 * lra-constraints.c (process_alt_operands): Don't discourage
6385 memory with known offset for offsetable memory constraint.
6386 * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
6388 2013-06-08 Eric Botcazou <ebotcazou@adacore.com>
6390 * varasm.c (struct oc_local_state): Reorder fields.
6391 (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
6392 and adjust accordingly.
6393 (output_constructor): Reorder initialization code and adjust call to
6394 output_constructor_bitfield.
6396 2013-06-07 Jan Hubicka <jh@suse.cz>
6398 * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
6400 2013-06-07 David Malcolm <dmalcolm@redhat.com>
6402 * tree-object-size.c (unknown): Make const.
6404 2013-06-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6406 * config/s390/s390.md (cpu_facility): Add cpu_zarch.
6407 ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
6408 for last alternative in the cpu_facility attribute.
6410 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6413 * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
6414 (xordi3): Change operand 2 constraint to arm_xordi_operand.
6415 * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
6416 * config/arm/constraints.md (Dg): New constraint.
6417 * config/arm/neon.md (xordi3_neon): Remove.
6418 (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
6419 * config/arm/predicates.md (arm_xordi_operand): New predicate.
6421 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6423 * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
6424 Clean up alternatives.
6426 2013-06-07 Alan Modra <amodra@gmail.com>
6428 * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
6431 2013-06-07 Alan Modra <amodra@gmail.com>
6433 * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
6435 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6437 * config/arm/constraints.md (Df): New constraint.
6438 * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
6439 Correct length attribute for last two alternatives.
6441 2013-06-07 Alan Modra <amodra@gmail.com>
6443 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6444 override user -mfp-in-toc.
6445 (offsettable_ok_by_alignment): Consider just the current access
6446 rather than the whole object, unless BLKmode. Handle
6447 CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
6448 (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
6449 for -mcmodel=medium.
6450 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
6451 override user -mfp-in-toc or -msum-in-toc. Default to
6452 -mno-fp-in-toc for -mcmodel=medium.
6454 2013-06-06 DJ Delorie <dj@redhat.com>
6456 * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
6457 TARGET_VALID_POINTER_MODE.
6459 2013-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6460 Pat Haugen <pthaugen@us.ibm.com>
6461 Peter Bergner <bergner@vnet.ibm.com>
6463 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6464 Document new power8 builtins.
6466 * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
6467 condition code register, to allow 128-bit logical operations to be
6468 done in the VSX or GPR registers.
6469 (nor<mode>3): Use the canonical form for nor.
6470 (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
6471 vclz*, and vpopcnt* vector instructions.
6472 (nand<mode>3): Likewise.
6473 (orc<mode>3): Likewise.
6474 (clz<mode>2): LIkewise.
6475 (popcount<mode>2): Likewise.
6477 * config/rs6000/predicates.md (int_reg_operand): Rework tests so
6478 that only the GPRs are recognized.
6480 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6481 support for new power8 builtins.
6483 * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
6485 (xscvdpspn): Likewise.
6491 (vpopcnt): Likewise.
6492 (vpopcntb): Likewise.
6493 (vpopcnth): Likewise.
6494 (vpopcntw): Likewise.
6495 (vpopcntd): Likewise.
6500 (eqv_v16qi3): Likewise.
6501 (eqv_v8hi3): Likewise.
6502 (eqv_v4si3): Likewise.
6503 (eqv_v2di3): Likewise.
6504 (eqv_v4sf3): Likewise.
6505 (eqv_v2df3): Likewise.
6507 (nand_v16qi3): Likewise.
6508 (nand_v8hi3): Likewise.
6509 (nand_v4si3): Likewise.
6510 (nand_v2di3): Likewise.
6511 (nand_v4sf3): Likewise.
6512 (nand_v2df3): Likewise.
6514 (orc_v16qi3): Likewise.
6515 (orc_v8hi3): Likewise.
6516 (orc_v4si3): Likewise.
6517 (orc_v2di3): Likewise.
6518 (orc_v4sf3): Likewise.
6519 (orc_v2df3): Likewise.
6521 * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
6522 allow power8 quad mode in 64-bit.
6523 (rs6000_builtin_vectorized_function): Add support to vectorize
6524 ISA 2.07 count leading zeros, population count builtins.
6525 (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
6526 V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
6527 (builtin_function_type): Add vgbbd builtin function which takes an
6529 (altivec_expand_vec_perm_const): Add support for new power8 merge
6532 * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
6533 that does not include TImdoe for use with 32-bit.
6534 (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
6536 (UNSPEC_VSX_CVDPSPN): Likewise.
6537 (vsx_xscvdpspn): Likewise.
6538 (vsx_xscvspdpn): Likewise.
6539 (vsx_xscvdpspn_scalar): Likewise.
6540 (vsx_xscvspdpn_directmove): Likewise.
6541 (vsx_and<mode>3): Split logical operations into 32-bit and
6542 64-bit. Add support to do logical operations on TImode as well as
6543 VSX vector types. Allow logical operations to be done in either
6544 VSX registers or in general purpose registers in 64-bit mode. Add
6545 splitters if GPRs were used. For AND, add clobber of CCmode to
6546 allow use of ANDI on GPRs. Rewrite nor to use the canonical RTL
6548 (vsx_and<mode>3_32bit): Likewise.
6549 (vsx_and<mode>3_64bit): Likewise.
6550 (vsx_ior<mode>3): Likewise.
6551 (vsx_ior<mode>3_32bit): Likewise.
6552 (vsx_ior<mode>3_64bit): Likewise.
6553 (vsx_xor<mode>3): Likewise.
6554 (vsx_xor<mode>3_32bit): Likewise.
6555 (vsx_xor<mode>3_64bit): Likewise.
6556 (vsx_one_cmpl<mode>2): Likewise.
6557 (vsx_one_cmpl<mode>2_32bit): Likewise.
6558 (vsx_one_cmpl<mode>2_64bit): Likewise.
6559 (vsx_nor<mode>3): Likewise.
6560 (vsx_nor<mode>3_32bit): Likewise.
6561 (vsx_nor<mode>3_64bit): Likewise.
6562 (vsx_andc<mode>3): Likewise.
6563 (vsx_andc<mode>3_32bit): Likewise.
6564 (vsx_andc<mode>3_64bit): Likewise.
6565 (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
6566 and xxlorc instructions.
6567 (vsx_eqv<mode>3_64bit): Likewise.
6568 (vsx_nand<mode>3_32bit): Likewise.
6569 (vsx_nand<mode>3_64bit): Likewise.
6570 (vsx_orc<mode>3_32bit): Likewise.
6571 (vsx_orc<mode>3_64bit): Likewise.
6573 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
6575 * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
6577 (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
6578 (p8_vmrgow): Likewise.
6579 (altivec_and<mode>3): Add clobber of CCmode to allow AND using
6580 GPRs to be split under VSX.
6581 (p8v_clz<mode>2): Add power8 count leading zero support.
6582 (p8v_popcount<mode>2): Add power8 population count support.
6583 (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
6586 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
6589 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
6591 (vec_nand): Likewise.
6592 (vec_vclz): Likewise.
6593 (vec_vclzb): Likewise.
6594 (vec_vclzd): Likewise.
6595 (vec_vclzh): Likewise.
6596 (vec_vclzw): Likewise.
6597 (vec_vgbbd): Likewise.
6598 (vec_vmrgew): Likewise.
6599 (vec_vmrgow): Likewise.
6600 (vec_vpopcnt): Likewise.
6601 (vec_vpopcntb): Likewise.
6602 (vec_vpopcntd): Likewise.
6603 (vec_vpopcnth): Likewise.
6604 (vec_vpopcntw): Likewise.
6606 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
6608 PR rtl-optimization/57468
6609 * config/i386/i386.c (inline_secondary_memory_needed): Ignore
6612 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
6614 PR rtl-optimization/57459
6615 * lra-constraints.c (update_ebb_live_info): Fix typo for operand
6616 type when setting live regs.
6618 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
6620 * config/s390/s390.opt (mlra): New option.
6621 * config/s390/s390.c (s390_decompose_address): Check displacement
6622 for all registers for LRA.
6623 (s390_secondary_reload): Don't used secondary reloads for LRA.
6624 (s390_lra_p): New function.
6625 (TARGET_LRA_P): Define.
6626 * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
6627 of attribute cpu_facility to zarch for the last alternative.
6628 (*cmpmem_short): Ditto.
6630 2013-06-06 Eric Botcazou <ebotcazou@adacore.com>
6632 * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
6633 (arm_compute_static_chain_stack_bytes): Use it. Tidy up.
6634 (arm_expand_prologue): Likewise.
6636 2013-06-06 Teresa Johnson <tejohnson@google.com>
6639 * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
6640 as this is now done by redirect_edge_and_branch_force.
6641 * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
6642 barriers, and fix interaction with splitting.
6643 * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
6644 * cfgcleanup.c (try_forward_edges): Fix early return value to properly
6645 reflect changes made in the routine.
6646 * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
6647 (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
6648 since this is called in cfglayout mode, and replace partition fixup
6649 with assert as that is now done by force_nonfallthru_and_redirect.
6650 (add_reg_crossing_jump_notes): Handle the fact that some jumps may
6651 already be marked with region crossing note.
6652 (insert_section_boundary_note): Make non-static, gate on flag
6653 has_bb_partition, rewrite to also check for multiple partitions.
6654 (rest_of_handle_reorder_blocks): Remove call to
6655 insert_section_boundary_note, now done later during free_cfg.
6656 (duplicate_computed_gotos): Don't duplicate partition crossing edge.
6657 * bb-reorder.h (insert_section_boundary_note): Declare.
6658 * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
6659 * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
6660 invoke insert_section_boundary_note.
6661 (try_redirect_by_replacing_jump): Remove unnecessary
6662 check for region crossing note.
6663 (fixup_partition_crossing): New function.
6664 (rtl_redirect_edge_and_branch): Fixup partition boundaries.
6665 (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
6666 in non-cfglayout mode.
6667 (force_nonfallthru_and_redirect): Fixup partition boundaries,
6668 remove old code that tried to do this. Emit barrier correctly
6669 when we are in cfglayout mode.
6670 (last_bb_in_partition): New function.
6671 (rtl_split_edge): Correctly fixup partition boundaries.
6672 (commit_one_edge_insertion): Remove old code that tried to
6673 fixup region crossing edge since this is now handled in
6674 split_block, and set up insertion point correctly since
6675 block may now end in a jump.
6676 (verify_hot_cold_block_grouping): Guard against checking when not in
6677 linearized RTL mode.
6678 (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
6680 (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
6681 rtl_verify_flow_info, so not called in cfglayout mode.
6682 (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
6683 (fixup_reorder_chain): Remove old code that attempted to fixup region
6684 crossing note as this is now handled in force_nonfallthru_and_redirect.
6685 (duplicate_insn_chain): Don't duplicate switch section notes.
6686 (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
6688 * basic-block.h (emit_barrier_after_bb): Declare.
6690 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6692 * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
6693 sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
6694 arm_usatsihi): Adjust alternatives for arm_restrict_it.
6696 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6698 * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
6700 * config/arm/ldmstm.md: Regenerate.
6702 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6704 * config/arm/sync.md (atomic_loaddi_1):
6705 Disable predication for arm_restrict_it.
6706 (arm_load_exclusive<mode>): Likewise.
6707 (arm_load_exclusivesi): Likewise.
6708 (arm_load_exclusivedi): Likewise.
6709 (arm_load_acquire_exclusive<mode>): Likewise.
6710 (arm_load_acquire_exclusivesi): Likewise.
6711 (arm_load_acquire_exclusivedi): Likewise.
6712 (arm_store_exclusive<mode>): Likewise.
6713 (arm_store_exclusive<mode>): Likewise.
6714 (arm_store_release_exclusivedi): Likewise.
6715 (arm_store_release_exclusive<mode>): Likewise.
6717 2013-06-06 Richard Biener <rguenther@suse.de>
6719 * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
6721 (lto_tag_is_tree_code_p): Adjust.
6722 (lto_tag_is_gimple_code_p): Likewise.
6723 (lto_gimple_code_to_tag): Likewise.
6724 (lto_tag_to_gimple_code): Likewise.
6725 (lto_tree_code_to_tag): Likewise.
6726 (lto_tag_to_tree_code): Likewise.
6727 * data-streamer.h (streamer_write_hwi_in_range): Use
6728 uhwi streaming to stream the normalized range.
6729 (streamer_read_hwi_in_range): Likewise.
6731 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6733 * config/arm/arm.md (enabled_for_depr_it): New attribute.
6734 (predicable_short_it): Likewise.
6735 (predicated): Likewise.
6736 (enabled): Handle above.
6737 (define_cond_exec): Set predicated attribute to yes.
6739 2013-06-05 Mike Stump <mikestump@comcast.net>
6741 * gdbinit.in (__FUNCTION__): Add.
6743 2013-06-05 Uros Bizjak <ubizjak@gmail.com>
6745 * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
6746 GE, GT, GEU and GTU compares, modulo DImode compares with zero.
6748 2013-06-05 Jan Hubicka <jh@suse.cz>
6750 * varasm.c (mark_decl_referenced): Revert the removal until targets
6753 2013-06-05 David Edelsohn <dje.gcc@gmail.com>
6755 * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
6756 instead of mark_decl_referenced.
6758 2013-06-05 Jan Hubicka <jh@suse.cz>
6760 * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
6761 (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
6762 and symtab_used_from_object_file_p.
6763 (cgraph_make_node_local_1): Clear forced_by_abi.
6764 (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
6765 * cgraph.h (symtab_node_base): Add forced_by_abi.
6766 (decide_is_variable_needed): Remove.
6767 (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
6768 * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
6769 (decide_is_symbol_needed): ... this one; handle symbols in general;
6770 always analyze virtuals; honnor forced_by_abi.
6771 (cgraph_finalize_function): Update.
6772 (varpool_finalize_decl): Update.
6773 (symbol_defined_and_needed): Remove.
6774 (analyze_functions): Update.
6775 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
6776 output_refs, input_overwrite_node): Handle forced_by_abi.
6777 * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
6778 (address_taken_from_non_vtable_p): ... this one.
6779 (comdat_can_be_unshared_p_1): New function.
6780 (cgraph_comdat_can_be_unshared_p): Rename to ...
6781 (comdat_can_be_unshared_p): ... this one; handle symbols in general.
6782 (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
6783 (function_and_variable_visibility): Clear forced_by_abi as needed.
6784 * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
6785 (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
6786 * symtab.c (dump_symtab_base): Dump forced_by_abi.
6787 * varpool.c (decide_is_variable_needed): Remove.
6789 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6791 * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
6792 (arm_option_override): Override arm_restrict_it where appropriate.
6793 (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
6794 * config/arm/arm.opt (mrestrict-it): New command-line option.
6795 * doc/invoke.texi: Document -mrestrict-it.
6797 2013-06-05 David Malcolm <dmalcolm@redhat.com>
6799 * tsan.c (tsan_atomic_table): Make const.
6801 2013-06-05 Richard Biener <rguenther@suse.de>
6803 * tree-streamer.c (streamer_tree_cache_insert_1): Update the
6804 index associated with the tree we are supposed to replace.
6805 * tree-streamer-out.c (pack_ts_base_value_fields): Output
6806 TREE_ASM_WRITTEN as zero for everything but SSA names.
6808 2013-06-05 David Malcolm <dmalcolm@redhat.com>
6810 * tree-ssa-structalias.c (call_stmt_vars): Make static.
6812 2013-06-04 Jan Hubicka <jh@suse.cz>
6814 * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
6815 (input_node, input_varpool_node): Handle correctly external same
6817 * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
6818 nodes at ltrans stage.
6820 2013-06-04 Jan Hubicka <jh@suse.cz>
6822 * ipa-inline.c (update_caller_keys): Fix availability test.
6823 (update_callee_keys): Likewise.
6824 * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
6825 to follow ELF standard.
6827 2013-06-04 Jürgen Urban <JuergenUrban@gmx.de>
6829 * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
6830 (mips64r5900el-*-elf*): New configurations.
6831 * config/mips/mips-cpus.def (r5900): New processor.
6832 * config/mips/mips-tables.opt: Regenerate.
6833 * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
6834 (mips_issue_rate): Handle PROCESSOR_R5900.
6835 (mips_reorg_process_insns): Force reorder mode for the R5900.
6836 * config/mips/mips.h (TARGET_MIPS5900): Define.
6837 (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
6839 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
6841 * config/mips/mips.md (processor): Add r5900.
6842 (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
6844 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6846 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
6847 into function to generate MOVI instruction.
6848 * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
6849 (aarch64_preferred_simd_mode): Turn into wrapper.
6850 (aarch64_output_scalar_simd_mov_immediate): New function.
6851 * config/aarch64/aarch64-protos.h: Add prototype for above.
6853 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6855 * config/aarch64/aarch64.c (simd_immediate_info): Remove
6856 element_char member.
6857 (sizetochar): Return signed char.
6858 (aarch64_simd_valid_immediate): Remove elchar and other
6859 unnecessary variables.
6860 (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
6861 Calculate element_char as required.
6862 * config/aarch64/aarch64-protos.h: Update and move prototype
6863 for aarch64_output_simd_mov_immediate.
6864 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6867 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6869 * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
6870 information completed by aarch64_simd_valid_immediate.
6871 (aarch64_legitimate_constant_p): Update arguments.
6872 (aarch64_simd_valid_immediate): Work with struct rather than many
6874 (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
6875 (aarch64_simd_make_constant): Update arguments.
6876 (aarch64_output_simd_mov_immediate): Work with struct rather than
6877 many pointers. Output immediate directly rather than as operand.
6878 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6880 * config/aarch64/constraints.md (Dn): Update arguments.
6882 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6884 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
6886 (aarch64_simd_immediate_valid_for_move): Remove.
6887 (aarch64_simd_scalar_immediate_valid_for_move): Update call.
6888 (aarch64_simd_make_constant): Update call.
6889 (aarch64_output_simd_mov_immediate): Update call.
6890 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
6892 * config/aarch64/constraints.md (Dn): Update call.
6894 2013-06-04 Ian Bolton <ian.bolton@arm.com>
6896 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
6897 return type to bool for prototype.
6898 (aarch64_legitimate_constant_p): Check for true instead of not -1.
6899 (aarch64_simd_valid_immediate): Fix up each return to return a bool.
6900 (aarch64_simd_immediate_valid_for_move): Update retval for bool.
6902 2013-06-04 Catherine Moore <clm@codesourcery.com>
6904 * config/mips/mips.opt (meva): New.
6905 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
6906 (ASM_SPEC): Handle -meva.
6907 * doc/invoke.texi (meva): Document.
6909 2013-06-04 Alan Modra <amodra@gmail.com>
6911 * config/rs6000/rs6000.c (output_toc): Correct little-endian float
6914 2013-06-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6916 * rtl.def: Add extra fourth optional field to define_cond_exec.
6917 * gensupport.c (process_one_cond_exec): Process attributes from
6919 * doc/md.texi: Document fourth field in define_cond_exec.
6921 2013-06-04 Eric Botcazou <ebotcazou@adacore.com>
6923 * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
6924 out the processing order as in store_bit_field_1.
6926 2013-06-04 Jan Hubicka <jh@suse.cz>
6929 * cgraphunit.c (cgraph_process_same_body_aliases): Create
6930 non-VAR_DECL node if it does not exist yet.
6932 2013-06-03 Richard Sandiford <rdsandiford@googlemail.com>
6934 * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
6935 (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
6936 (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
6937 target_cpu_default setting.
6939 2013-06-03 Teresa Johnson <tejohnson@google.com>
6941 * dumpfile.c (opt_info_switch_p): Change -fopt-info
6942 default to -fopt-info=optimized instead of all.
6943 * doc/invoke.texi: Ditto.
6944 * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
6945 success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
6946 (execute_vect_slp): Emit BB vectorization success under
6947 MSG_OPTIMIZED_LOCATIONS.
6948 * tree-vect-slp.c (vect_slp_transform_bb): Change
6949 MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
6950 * tree-vect-loop.c (vect_transform_loop): Ditto.
6952 2013-06-03 Jason Merrill <jason@redhat.com>
6955 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6956 Use TARGET_EXPR for C++.
6958 2013-06-03 Jakub Jelinek <jakub@redhat.com>
6960 PR rtl-optimization/57268
6961 * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
6962 if DEBUG_INSN_P (insn).
6965 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
6967 PR rtl-optimization/57268
6968 * sched-deps.c (sched_analyze_2): Flush dependence lists if
6969 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
6971 2013-06-03 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
6973 * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
6974 (ix86_avoid_lea_for_addr): Likewise.
6975 (exact_dependency_1): Likewise.
6976 (ix86_adjust_cost): Likewise.
6977 (swap_top_of_ready_list): Fix formatting and !reload_completed check
6979 (do_reorder_for_imul): Fix typo, formatting and
6980 !reload_completed check removed.
6981 (ix86_sched_reorder): Fix typo and formatting.
6982 (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
6985 2013-06-03 Sofiane Naci <sofiane.naci@arm.com>
6987 * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
6989 2013-06-03 Eric Botcazou <ebotcazou@adacore.com>
6991 * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
6992 <STRING_CST>: Likewise.
6993 <VECTOR_CST>: Likewise.
6995 2013-06-01 Janus Weil <janus@gcc.gnu.org>
6996 Mikael Morin <mikael@gcc.gnu.org>
6998 * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
6999 * config.in: Regenerated.
7000 * configure: Regenerated.
7002 2013-06-01 Jan Hubicka <jh@suse.cz>
7005 * cgraphunit.c (compile): When weakref is not supported,
7006 set up transparent aliases before final output pass.
7007 * varasm.c (assemble_alias): Do not try to do it here.
7009 2013-06-01 Jan Hubicka <jh@suse.cz>
7012 * passes.c (for_per_function): Skip unanalyzed functions.
7014 2013-06-01 Jan Hubicka <jh@suse.cz>
7016 * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
7017 (lto_symtab_merge_symbols_1): ... this one.
7018 (lto_symtab_merge_cgraph_nodes): Rename to ...
7019 (lto_symtab_merge_symbols): ... this one; simplify.
7020 * cgraph.c (same_body_aliases_done): Rename to ...
7021 (cpp_implicit_aliases_done): ... this one.
7022 (cgraph_create_function_alias): Update.
7023 (cgraph_same_body_alias): Update.
7024 (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
7025 (verify_edge_corresponds_to_fndecl): Simplify.
7026 * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
7027 (cgraph_node): Remove same_body_alias.
7028 (varpool_node): Remove alias_of and extra_name_alias.
7029 (same_body_aliases_done): Rename to ..
7030 (cpp_implicit_aliases_done): ... this one.
7031 (symtab_alias_ultimate_target): Add default parameter.
7032 (symtab_resolve_alias): New function.
7033 (fixup_same_cpp_alias_visibility): Declare.
7034 (cgraph_function_node): Add default parameter.
7035 (cgraph_node_asm_name): Likewise.
7036 (cgraph_function_or_thunk_node): Add default parameter; do
7037 not ICE when it is NULL.
7038 (varpool_variable_node): Likewise.
7039 * tree-emutls.c (create_emultls_var): Update.
7040 (ipa_lower_emutls): Update.
7041 * cgraphunit.c (cgraph_decide_is_function_needed): Update.
7042 (cgraph_reset_node): Reset alias info.
7043 (cgraph_finalize_function): Update.
7044 (fixup_same_cpp_alias_visibility): Move to symtab.c.
7045 (analyze_function): Simplify.
7046 (cgraph_process_same_body_aliases): Simplify.
7047 (analyze_functions): Fixup same body aliases.
7048 (handle_alias_pairs): Simplify.
7049 (assemble_thunk): Update.
7050 (assemble_thunks_and_aliases): Update.
7051 (output_weakrefs): Rewrite.
7052 * lto-cgraph.c (lto_output_node): Rewrite alias handling.
7053 (lto_output_varpool_node): Likewise.
7054 (compute_ltrans_boundary): Remve assert.
7055 (get_alias_symbol): New functoin.
7056 (input_node): Rewrite alias handling.
7057 (input_varpool_node): Likewise.
7058 * ipa-pure-const.c (propagate_pure_const): Fix formating.
7059 * ipa.c (process_references): Handle weakrefs correctly.
7060 (symtab_remove_unreachable_nodes): Likewise.
7061 * trans-mem.c (get_cg_data): Update.
7062 (ipa_tm_create_version_alias): Update.
7063 (ipa_tm_execute): Update.
7064 * symtab.c (dump_symtab_base): Dump aliases.
7065 (verify_symtab_base): Verify aliases.
7066 (symtab_node_availability): New function.
7067 (symtab_alias_ultimate_target): Simplify.
7068 (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
7069 handle all the fixup cases.
7070 (symtab_resolve_alias): New function.
7071 * passes.c (ipa_write_summaries): Handle weakrefs.
7072 * varpool.c (varpool_analyze_node): Simplify.
7073 (assemble_aliases): Update.
7074 (varpool_create_variable_alias): Simplify.
7075 (varpool_extra_name_alias): Simplify.
7076 * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
7077 (lto_symtab_merge_symbols): ... this one.
7079 2013-06-01 Dinar Temirbulatov <dinar@kugelworks.com>
7082 PR rtl-optimization/57268
7083 * sched-deps.c (sched_analyze_2): Flush dependence lists if
7084 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7086 2013-06-01 Tobias Burnus <burnus@net-b.de>
7089 2013-05-31 Tobias Burnus <burnus@net-b.de>
7092 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
7095 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
7097 PR rtl-optimization/57268
7098 * sched-deps.c (sched_analyze_2): Flush dependence lists if
7099 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7101 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
7103 * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
7104 unordered comparison operators when -fno-trapping-math is in effect
7106 * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
7107 and implement unordered comparison operators properly on the e500.
7109 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
7111 * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
7112 for constant scalar integers.
7113 (simplify_relational_operation_1): Likewise.
7115 2013-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7117 * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
7118 * config/rs6000/rs6000.md (cpu): Reorder. Split long line.
7121 2013-05-31 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
7122 Igor Zamyatin <igor.zamyatin@intel.com>
7124 Silvermont (SLM) architecture performance tuning.
7125 * config/i386/i386.h (enum ix86_tune_indices): Add
7126 X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
7127 (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
7129 * config/i386/i386.c (initial_ix86_tune_features)
7130 <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
7131 (ix86_lea_outperforms): Handle Silvermont tuning.
7132 (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
7134 (ix86_use_lea_for_mov): Likewise.
7135 (ix86_avoid_lea_for_addr): Likewise.
7136 (ix86_lea_for_add_ok): Likewise.
7137 (exact_dependency_1): New function.
7138 (exact_store_load_dependency): Likewise.
7139 (ix86_adjust_cost): Handle Silvermont tuning.
7140 (do_reoder_for_imul): Likewise.
7141 (swap_top_of_ready_list): New function.
7142 (ix86_sched_reorder): Changed to handle Silvermont tuning.
7144 * config/i386/i386.md (peepholes that split memory operand in fp
7147 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
7149 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7150 Remove un-necessary braces.
7152 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
7154 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7155 Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
7157 2013-05-31 Tobias Burnus <burnus@net-b.de>
7160 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
7162 2013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7165 * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
7166 * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
7167 * config/arm/neon.md (iordi3_neon): Remove.
7168 (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
7169 * config/arm/predicates.md (imm_for_neon_logic_operand):
7170 Move to earlier in the file.
7171 (neon_logic_op2): Likewise.
7172 (arm_iordi_operand_neon): New predicate.
7174 2013-05-31 Richard Biener <rguenther@suse.de>
7176 PR tree-optimization/57478
7177 PR tree-optimization/57453
7178 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
7181 2013-05-31 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
7183 * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
7184 (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
7186 2013-05-30 Tobias Burnus <burnus@net-b.de>
7187 Thomas Koenig <tkoenig@gcc.gnu.org>
7190 * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
7191 powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
7193 2013-05-30 Steven Bosscher <steven@gcc.gnu.org>
7195 * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
7197 2013-05-30 Vladimir Makarov <vmakarov@redhat.com>
7199 * target.def (register_usage_leveling_p): New hook.
7200 * targhooks.c (default_register_usage_leveling_p): New.
7201 * targhooks.h (default_register_usage_leveling_p): New prototype.
7202 * lra-assigns.c (register_usage_leveling_p): Use the hook.
7203 * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
7204 * doc/tm.texi: Update.
7205 * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
7207 2013-05-30 Ian Bolton <ian.bolton@arm.com>
7209 * config/aarch64/aarch64.md (insv<mode>): New define_expand.
7210 (*insv_reg<mode>): New define_insn.
7212 2013-05-30 Joern Rennecke <joern.rennecke@embecosm.com>
7214 PR rtl-optimization/57439
7215 * postreload.c (move2add_valid_value_p): Check that we have
7216 a zero subreg_regno_offset when accessing the register in
7219 2013-05-30 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
7220 Igor Zamyatin <igor.zamyatin@intel.com>
7222 Silvermont (SLM) architecture pipeline model, tuning and
7224 * config.gcc: Add slm config options and target.
7226 * config/i386/slm.md: New.
7228 * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
7230 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): New case
7232 (ix86_target_macros_internal): Likewise.
7234 * gcc/config/i386/i386.c (slm_cost): New cost.
7235 (m_SLM): New macro flag.
7236 (initial_ix86_tune_features): Set m_SLM.
7237 (x86_accumulate_outgoing_args): Likewise.
7238 (x86_arch_always_fancy_math_387): Likewise.
7239 (processor_target_table): Add slm cost.
7240 (cpu_names): Add slm cpu name.
7241 (x86_option_override_internal): Set SLM ISA.
7242 (ix86_issue_rate): New case PROCESSOR_SLM.
7243 (ia32_multipass_dfa_lookahead): Likewise.
7244 (fold_builtin_cpu): Add slm.
7246 * config/i386/i386.h (TARGET_SLM): New target macro.
7247 (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
7248 (processor_type): Add PROCESSOR_SLM.
7250 * config/i386/i386.md (cpu): Add new value "slm".
7251 (slm.md): Include slm.md.
7253 2013-05-30 Bernd Schmidt <bernds@codesourcery.com>
7254 Zhenqiang Chen <zhenqiang.chen@linaro.org>
7256 * config/arm/arm-protos.h: Add and update function protos.
7257 * config/arm/arm.c (use_simple_return_p): New added.
7258 (thumb2_expand_return): Check simple_return flag.
7259 * config/arm/arm.md: Add simple_return and conditional simple_return.
7260 * config/arm/iterators.md: Add iterator for return and simple_return.
7262 2013-05-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
7264 * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
7265 (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
7266 (arm_emit_vfp_multi_reg_pop): Likewise.
7267 (thumb2_emit_ldrd_pop): Likewise.
7268 (arm_expand_epilogue): Add misc REG_CFA notes.
7269 (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
7271 2013-05-29 Lawrence Crowl <crowl@google.com>
7273 * config/arm/t-arm: Update for below.
7275 * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
7276 Change type to hash_table. Update dependent calls and types.
7278 * config/i386/t-cygming: Update for below.
7280 * config/i386/t-interix: Update for below.
7282 * config/i386/winnt.c (i386_pe_section_type_flags::htab):
7283 Change type to hash_table. Update dependent calls and types.
7284 (i386_find_on_wrapper_list::wrappers): Likewise.
7286 * config/ia64/t-ia64: Update for below.
7288 * config/ia64/ia64.c (bundle_state_table):
7289 Change type to hash_table. Update dependent calls and types.
7291 * config/mips/mips.c (mips_reorg_process_insns::htab):
7292 Change type to hash_table. Update dependent calls and types.
7294 * config/sol2.c (solaris_comdat_htab):
7295 Change type to hash_table. Update dependent calls and types.
7297 * config/t-sol2: Update for above.
7299 2013-05-29 Teresa Johnson <tejohnson@google.com>
7301 * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
7302 functions are not yet marked as defined.
7304 2013-05-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7305 Pat Haugen <pthaugen@us.ibm.com>
7306 Peter Bergner <bergner@vnet.ibm.com>
7308 * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
7312 (vrotl<mode>3): Likewise.
7313 (vashl<mode>3): Likewise.
7314 (vlshr<mode>3): Likewise.
7315 (vashr<mode>3): Likewise.
7317 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7318 support for power8 V2DI builtins.
7320 * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
7321 power8 V2DI builtins.
7322 (vupkhsw): Likewise.
7323 (vupklsw): Likewise.
7324 (vaddudm): Likewise.
7329 (vpkudum): Likewise.
7330 (vpksdss): Likewise.
7331 (vpkudus): Likewise.
7332 (vpksdus): Likewise.
7337 (vsubudm): Likewise.
7338 (vcmpequd): Likewise.
7339 (vcmpgtsd): Likewise.
7340 (vcmpgtud): Likewise.
7341 (vcmpequd_p): Likewise.
7342 (vcmpgtsd_p): Likewise.
7343 (vcmpgtud_p): Likewise.
7344 (vupkhsw): Likewise.
7345 (vupklsw): Likewise.
7346 (vaddudm): Likewise.
7351 (vpksdss): Likewise.
7352 (vpksdus): Likewise.
7353 (vpkudum): Likewise.
7354 (vpkudus): Likewise.
7359 (vsubudm): Likewise.
7361 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
7362 support for power8 V2DI instructions.
7364 * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
7365 power8 V2DI instructions. Combine pack and unpack insns to use an
7366 iterator for each mode. Check whether a particular mode supports
7367 Altivec instructions instead of just checking TARGET_ALTIVEC.
7368 (UNSPEC_VPKUWUM): Likewise.
7369 (UNSPEC_VPKSHSS): Likewise.
7370 (UNSPEC_VPKSWSS): Likewise.
7371 (UNSPEC_VPKUHUS): Likewise.
7372 (UNSPEC_VPKSHUS): Likewise.
7373 (UNSPEC_VPKUWUS): Likewise.
7374 (UNSPEC_VPKSWUS): Likewise.
7375 (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
7376 (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
7377 (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
7378 (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
7379 (UNSPEC_VUPKHSB): Likewise.
7380 (UNSPEC_VUNPACK_HI_SIGN): Likewise.
7381 (UNSPEC_VUNPACK_LO_SIGN): Likewise.
7382 (UNSPEC_VUPKHSH): Likewise.
7383 (UNSPEC_VUPKLSB): Likewise.
7384 (UNSPEC_VUPKLSH): Likewise.
7386 (VI_char): Likewise.
7387 (VI_scalar): Likewise.
7388 (VI_unit): Likewise.
7390 (VP_small): Likewise.
7391 (VP_small_lc): Likewise.
7392 (VU_char): Likewise.
7393 (add<mode>3): Likewise.
7394 (altivec_vaddcuw): Likewise.
7395 (altivec_vaddu<VI_char>s): Likewise.
7396 (altivec_vadds<VI_char>s): Likewise.
7397 (sub<mode>3): Likewise.
7398 (altivec_vsubcuw): Likewise.
7399 (altivec_vsubu<VI_char>s): Likewise.
7400 (altivec_vsubs<VI_char>s): Likewise.
7401 (altivec_vavgs<VI_char>): Likewise.
7402 (altivec_vcmpbfp): Likewise.
7403 (altivec_eq<mode>): Likewise.
7404 (altivec_gt<mode>): Likewise.
7405 (altivec_gtu<mode>): Likewise.
7406 (umax<mode>3): Likewise.
7407 (smax<mode>3): Likewise.
7408 (umin<mode>3): Likewise.
7409 (smin<mode>3): Likewise.
7410 (altivec_vpkuhum): Likewise.
7411 (altivec_vpkuwum): Likewise.
7412 (altivec_vpkshss): Likewise.
7413 (altivec_vpkswss): Likewise.
7414 (altivec_vpkuhus): Likewise.
7415 (altivec_vpkshus): Likewise.
7416 (altivec_vpkuwus): Likewise.
7417 (altivec_vpkswus): Likewise.
7418 (altivec_vpks<VI_char>ss): Likewise.
7419 (altivec_vpks<VI_char>us): Likewise.
7420 (altivec_vpku<VI_char>us): Likewise.
7421 (altivec_vpku<VI_char>um): Likewise.
7422 (altivec_vrl<VI_char>): Likewise.
7423 (altivec_vsl<VI_char>): Likewise.
7424 (altivec_vsr<VI_char>): Likewise.
7425 (altivec_vsra<VI_char>): Likewise.
7426 (altivec_vsldoi_<mode>): Likewise.
7427 (altivec_vupkhsb): Likewise.
7428 (altivec_vupkhs<VU_char>): Likewise.
7429 (altivec_vupkls<VU_char>): Likewise.
7430 (altivec_vupkhsh): Likewise.
7431 (altivec_vupklsb): Likewise.
7432 (altivec_vupklsh): Likewise.
7433 (altivec_vcmpequ<VI_char>_p): Likewise.
7434 (altivec_vcmpgts<VI_char>_p): Likewise.
7435 (altivec_vcmpgtu<VI_char>_p): Likewise.
7436 (abs<mode>2): Likewise.
7437 (vec_unpacks_hi_v16qi): Likewise.
7438 (vec_unpacks_hi_v8hi): Likewise.
7439 (vec_unpacks_lo_v16qi): Likewise.
7440 (vec_unpacks_hi_<VP_small_lc>): Likewise.
7441 (vec_unpacks_lo_v8hi): Likewise.
7442 (vec_unpacks_lo_<VP_small_lc>): Likewise.
7443 (vec_pack_trunc_v8h): Likewise.
7444 (vec_pack_trunc_v4si): Likewise.
7445 (vec_pack_trunc_<mode>): Likewise.
7447 * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
7449 (vec_vmaxsd): Likewise.
7450 (vec_vmaxud): Likewise.
7451 (vec_vminsd): Likewise.
7452 (vec_vminud): Likewise.
7453 (vec_vpksdss): Likewise.
7454 (vec_vpksdus): Likewise.
7455 (vec_vpkudum): Likewise.
7456 (vec_vpkudus): Likewise.
7457 (vec_vrld): Likewise.
7458 (vec_vsld): Likewise.
7459 (vec_vsrad): Likewise.
7460 (vec_vsrd): Likewise.
7461 (vec_vsubudm): Likewise.
7462 (vec_vupkhsw): Likewise.
7463 (vec_vupklsw): Likewise.
7465 2013-05-29 Jan Hubicka <jh@suse.cz>
7467 * cgraph.h (symtab_node_base): Add definition, alias and analyzed
7468 flags; reorder rest of fields in more consistent way.
7469 (varpool_node): Remove analyzed, finalized and alias.
7470 (cgraph_ndoe): Likewise.
7471 (symtab_alias_ultimate_target): New function.
7472 (cgraph_function_node): Move offline.
7473 (cgraph_reset_node): Declare.
7474 (cgraph_comdat_can_be_unshared_p): Remove.
7475 (varpool_remove_initializer): Declare.
7476 (varpool_first_defined_variable, varpool_next_defined_variable
7477 cgraph_first_defined_function, cgraph_next_defined_function): Update.
7478 (cgraph_function_with_gimple_body_p): Update.
7479 (varpool_all_refs_explicit_p): Update.
7480 (symtab_alias_target): New function.
7481 (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
7482 (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
7483 (cgraph_function_or_thunk_node): Simplify using
7484 symtab_alias_ultimate_target.
7485 (varpool_variable_node): Likewise.
7486 * cgraph.c (cgraph_create_function_alias): Update.
7487 (cgraph_add_thunk): Update.
7488 (cgraph_remove_node): Update.
7489 (dump_cgraph_node): Do not dump removed flags.
7490 (cgraph_function_body_availability): Update.
7491 (cgraph_propagate_frequency): Update.
7492 (verify_cgraph_node): Check sanity of local flag.
7493 (cgraph_function_node): Move here from cgraph.h; revamp for
7494 cgraph_function_or_thunk_node.
7495 * lto-symtab.c (lto_varpool_replace_node): Update.
7496 (lto_symtab_resolve_can_prevail_p): Update.
7497 (lto_symtab_merge_cgraph_nodes): Update.
7498 * ipa-cp.c (determine_versionability, initialize_node_lattices,
7499 propagate_constants_accross_call, devirtualization_time_bonus,
7500 ipcp_propagate_stage): Update.
7501 * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
7502 * ipa-inline-transform.c (clone_inlined_nodes,
7503 preserve_function_body_p): Update.
7504 * ipa-reference.c (propagate): Update.
7505 (write_node_summary_p): Update.
7506 * toplev.c (wrapup_global_declaration_2): Update.
7507 * cgraphunit.c (cgraph_analyze_function): Rename to ...
7508 (analyze_function) ... this one.
7509 (cgraph_process_new_functions): Update.
7510 (cgraph_reset_node): Export.
7511 (cgraph_finalize_function): Update.
7512 (cgraph_add_new_function): Update.
7513 (process_function_and_variable_attributes): Update.
7514 (varpool_finalize_decl): Update.
7515 (symbol_finalized): Remove.
7516 (symbol_finalized_and_needed): Rename to ...
7517 (symbol_defined_and_needed): ... update.
7518 (cgraph_analyze_functions): Update.
7519 (handle_alias_pairs): Update.
7520 (mark_functions_to_output): Update.
7521 (assemble_thunk): Update.
7522 (output_in_order): Update.
7523 (output_weakrefs): Update.
7524 (finalize_compilation_unit): Update.
7525 * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
7526 lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
7527 input_node, input_varpool_node): Update.
7528 * dbxout.c (dbxout_expand_expr): Update.
7529 * cgraphclones.c (cgraph_clone_node): Update.
7530 (cgraph_copy_node_for_versioning): Update.
7531 (cgraph_materialize_clone): Update.
7532 (cgraph_materialize_all_clones): Update.
7533 * ipa-pure-const.c (analyze_function, pure_const_write_summary,
7534 propagate_pure_const, propagate_nothrow): Update.
7535 * lto-streamer-out.c (lto_output, write_symbol): Update.
7536 * ipa-utils.c (ipa_reverse_postorder): Update.
7537 * ipa-inline.c (can_inline_edge_p): Update.
7538 (update_caller_keys, ipa_inline): Update.
7539 * dwarf2out.c (reference_to_unused,
7540 premark_types_used_by_global_vars_helper): Update.
7541 * tree-eh.c (tree_could_trap_p): Update.
7542 * ipa-split.c (consider_split, execute_split_functions): Update.
7543 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
7544 has_addr_references_p): Update; move ahead in file for better
7546 (process_references): Simplify.
7547 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
7549 (cgraph_comdat_can_be_unshared_p): Make static.
7550 (cgraph_externally_visible_p): Update.
7551 (varpool_externally_visible_p): Update.
7552 (function_and_variable_visibility): Update.
7553 * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
7554 ipa_tm_mark_force_output_node): Update.
7555 * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
7556 estimate_edge_devirt_benefit, inline_generate_summary,
7557 inline_write_summary): Update.
7558 * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
7559 * ipa-prop.c (ipa_compute_jump_functions): Update.
7560 (ipa_print_node_params, ipa_prop_read_section,
7561 ipa_update_after_lto_read, read_replacements_section): Update.
7562 * varasm.c (mark_decl_referenced): Update.
7563 (assemble_alias, dump_tm_clone_pairs): Update.
7564 * tree-inline.c (copy_bb): Update.
7565 (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
7567 * symtab.c (dump_symtab_base): Print new flags.
7568 (verify_symtab_base): Verify new flags.
7569 (symtab_alias_ultimate_target): New function.
7570 * tree-ssa-structalias.c (get_constraint_for_ssa_var,
7571 create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
7573 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
7575 * i386.c (ix86_get_function_versions_dispatcher,
7576 ix86_generate_version_dispatcher_body): Update.
7577 (fold_builtin_cpu): Use varpool_add_new_variable.
7578 * varpool.c (varpool_remove_initializer): Break out from ...
7579 (varpool_remove_node): ... this one.
7580 (dump_varpool_node, varpool_node_for_asm,
7581 cgraph_variable_initializer_availability, varpool_analyze_node,
7582 varpool_assemble_decl, varpool_remove_unreferenced_decls,
7583 varpool_finalize_named_section_flags, varpool_create_variable_alias):
7586 2013-05-29 Jan Hubicka <jh@suse.cz>
7588 * passes.c (init_optimization_passes): Move OMP expansion into lowering.
7590 2013-05-29 Easwaran Raman <eraman@google.com>
7592 PR tree-optimization/57442
7593 * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
7594 when control exits the main loop.
7596 2013-05-29 Sandeep Kumar Singh <Sandeep.Singh2@kpitcummins.com>
7598 * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
7600 * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
7601 * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
7602 * rx/t-rx: Add rx100 under multi library matches option for nofpu
7605 2013-05-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7607 PR tree-optimization/57441
7608 * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
7609 Don't limit size of incr_vec to number of candidates.
7611 2013-05-29 Steve Ellcey <sellcey@imgtec.com>
7613 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
7614 and mips16 directories.
7615 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
7616 (MULTILIB_DIRNAMES): Ditto.
7617 (MULTILIB_EXCEPTIONS): Add new exceptions.
7618 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
7619 (MULTILIB_DIRNAMES): Ditto.
7620 (MULTILIB_EXCEPTIONS): Add new exceptions.
7622 2012-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7623 Marcus Shawcroft <marcus.shawcroft@arm.com>
7625 * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
7626 SYMBOL_TINY_ABSOLUTE.
7627 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
7628 SYMBOL_TINY_ABSOLUTE.
7629 (aarch64_expand_mov_immediate): Likewise.
7630 (aarch64_classify_symbol): Likewise.
7631 (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
7632 Permit SYMBOL_TINY_ABSOLUTE.
7633 * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
7635 2013-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7636 Marcus Shawcroft <marcus.shawcroft@arm.com>
7638 * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
7639 Refactor if/switch. Replace gcc_assert with if.
7641 2013-05-29 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
7643 * config/i386/i386.c (initial_ix86_tune_features): Enable
7644 FP Reassociation for AMD bdver1 and bdver2.
7646 2013-05-29 Martin Jambor <mjambor@suse.cz>
7648 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
7649 and IMAGPART_EXPR do not occur within other handled_components.
7651 2013-05-29 Richard Biener <rguenther@suse.de>
7653 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
7654 access on whether the use is in the BB we currently try to
7656 (vect_bb_vectorization_profitable_p): Pass the BB we currently
7657 vectorize to vect_bb_slp_scalar_cost.
7659 2013-05-29 Richard Biener <rguenther@suse.de>
7661 * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
7662 computing scalar cost offsetted by stmts that are kept live
7664 (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
7665 for computation of scalar cost.
7667 2013-05-28 Steve Ellcey <sellcey@mips.com>
7669 * config/mips/mips-cpus.def (mips32r2): Change processor type.
7671 2013-05-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
7673 * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
7674 array notation built-in reduction functions.
7675 * doc/passes.texi (Passes): Added documentation about changes done
7677 * doc/invoke.texi (C Dialect Options): Added documentation about
7678 the -fcilkplus flag.
7679 * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
7680 (BUILTINS_DEF): Depend on cilkplus.def.
7681 * builtins.def: Include cilkplus.def. Define DEF_CILKPLUS_BUILTIN.
7682 * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
7683 * cilkplus.def: New file.
7685 2013-05-28 Joern Rennecke <joern.rennecke@embecosm.com>
7687 PR rtl-optimization/57439
7688 * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
7690 2013-05-28 Easwaran Raman <eraman@google.com>
7692 PR tree-optimization/57337
7693 * tree-ssa-reassoc.c (appears_later_in_bb): New function.
7694 (find_insert_point): Correctly identify the insertion point
7695 when two statements with the same UID is compared.
7697 2013-05-28 Richard Biener <rguenther@suse.de>
7699 PR tree-optimization/56787
7700 * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
7701 from the list of data references.
7702 * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
7704 (vect_analyze_loop_operations): Likewise.
7705 (vect_transform_loop): Remove clobbers.
7707 2013-05-28 Martin Jambor <mjambor@suse.cz>
7709 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
7710 and REALPART_EXPRs have scalar type.
7712 2013-05-28 Richard Biener <rguenther@suse.de>
7714 PR tree-optimization/57411
7715 * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
7717 * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
7718 virtual operand propagation.
7720 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
7722 * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
7723 destination register for bmasksi_vis.
7724 (vector_init_bshuffle): Likewise.
7725 * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
7727 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
7729 * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
7730 * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
7731 mode if the instruction isn't available in the original mode.
7732 * config/sparc/sparc.opt (mfix-ut699): New option.
7733 * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
7734 (divdf3): Turn into expander.
7735 (divdf3_nofix): New insn.
7736 (divdf3_fix): Likewise.
7737 (divsf3): Disable if -mfix-ut699.
7738 (sqrtdf2): Turn into expander.
7739 (sqrtdf2_nofix): New insn.
7740 (sqrtdf2_fix): Likewise.
7741 (sqrtsf2): Disable if -mfix-ut699.
7743 2013-05-27 Richard Biener <rguenther@suse.de>
7746 * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
7747 block for the new loop.
7749 2013-05-27 Richard Biener <rguenther@suse.de>
7751 PR tree-optimization/57343
7752 * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
7753 use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
7754 (number_of_iterations_cond): Do not build the folded tree.
7756 2013-05-27 Richard Biener <rguenther@suse.de>
7760 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7761 OEP_CONSTANT_ADDRESS_OF retained.
7763 PR tree-optimization/57417
7764 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
7766 (set_ssa_val_to): Compare addresses using
7767 get_addr_base_and_unit_offset.
7769 2013-05-27 Joern Rennecke <joern.rennecke@embecosm.com>
7771 PR rtl-optimization/56833
7772 * postreload.c (move2add_record_mode): New function.
7773 (move2add_record_sym_value, move2add_valid_value_p): Likewise.
7774 (move2add_use_add2_insn): Use move2add_record_sym_value.
7775 (move2add_use_add3_insn): Likewise.
7776 (reload_cse_move2add): Use move2add_valid_value_p and
7777 move2add_record_mode. Invalidate call-clobbered and REG_INC
7778 affected regs by setting reg_mode to VOIDmode.
7779 (move2add_note_store): Don't pretend the inside of a SUBREG is
7780 the actual destination. Invalidate single/leading registers by
7781 setting reg_mode to VOIDmode.
7782 Use move2add_record_sym_value, move2add_valid_value_p and
7783 move2add_record_mode.
7785 2013-05-27 Richard Biener <rguenther@suse.de>
7787 PR tree-optimization/57396
7788 * tree-affine.c (double_int_constant_multiple_p): Properly
7789 return false for val == 0 and div != 0.
7791 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
7793 * config/mips/mips.h: Use #elif in preprocessor conditions.
7795 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
7798 * config/mips/constraints.md (kl): New constraint.
7799 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
7800 (divmod<mode>4_internal): Rename to divmod<mode>4. Use "kl" as the
7801 constraint for operand 0. Split after CSE for MIPS16. Emit a move
7803 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
7805 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
7808 * config/mips/mips.c (mips_can_inline_p): New function.
7809 (TARGET_CAN_INLINE_P): Define.
7811 2013-05-25 Steven Bosscher <steven@gcc.gnu.org>
7813 * sched-int.h (ds_t, dw_t): Make unsigned int.
7814 Fix documentation that describes how all the ds_t bits are used.
7815 Reserve the last bit for delayed-branch scheduling.
7816 (BITS_PER_DEP_STATUS): Move to ds_t typedef.
7817 (BITS_PER_DEP_WEAK): Fix definition and documentation.
7818 (gen_dep_weak_1): Remove prototype.
7819 * sched-deps.c (get_dep_weak_1): Make static.
7820 * target.def (speculate_insn, needs_block_p, gen_spec_check,
7821 get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
7822 * doc/tm.texi: Regenerate.
7823 * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
7825 2013-05-24 Steven Bosscher <steven@gcc.gnu.org>
7828 * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
7830 2013-05-24 Nathan Sidwell <nathan@codesourcery.com>
7831 Sandra Loosemore <sandra@codesourcery.com>
7833 * config.gcc (powerpc-*): Allow native for with-cpu.
7835 2013-05-24 Jeff Law <law@redhat.com>
7837 PR tree-optimization/57124
7838 * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
7839 conversion feeding a condition if the range has an overflow
7840 if -fstrict-overflow. Add warnings for when we do make the
7843 2013-05-24 Dehao Chen <dehao@google.com>
7845 * tree-cfg.c (locus_discrim_map): Fix the typo.
7846 (locus_discrim_hasher): Likewise.
7847 (locus_discrim_hasher::hash): Likewise.
7848 (locus_discrim_hasher::equal): Likewise.
7850 2013-05-24 Martin Jambor <mjambor@suse.cz>
7852 PR tree-optimization/57294
7853 * cgraph.h (ipa_record_stmt_references): Declare.
7854 * cgraphbuild.c (ipa_record_stmt_references): New function.
7855 (build_cgraph_edges): Use ipa_record_stmt_references.
7856 (rebuild_cgraph_edges): Likewise.
7857 (cgraph_rebuild_references): Likewise.
7858 * ipa-prop.c (ipa_modify_call_arguments): Discard references
7859 associated with the old statement and build references from the
7860 newly built statements.
7861 * ipa-ref.c (ipa_remove_stmt_references): New function.
7862 * ipa-ref.h (ipa_remove_stmt_references): Declare.
7864 2013-05-24 Vladimir Makarov <vmakarov@redhat.com>
7866 * lra-constraints.c (emit_spill_move): Use smaller mode for
7868 (check_and_process_move): Consider mem-reg moves for secondary
7870 (curr_insn_transform): Don't lose insns emitted before for
7871 secondary memory moves.
7872 (inherit_in_ebb): Mark defined reg. Add usage only if it is not a
7873 reg set up in the current insn.
7875 2013-05-24 Dehao Chen <dehao@google.com>
7877 * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
7879 (locus_descrim_hasher::equal): Likewise.
7880 (build_gimple_cfg): New discriminator assignment algorithm.
7881 (make_edges): Likewise.
7882 (next_discriminator_for_locus): Likewise.
7883 (same_line_p): Likewise.
7884 (assign_discriminators): Likewise.
7885 (make_cond_expr_edges): Likewise.
7886 (make_gimple_switch_edges): Likewise.
7887 (make_goto_expr_edges): Likewise.
7888 (make_gimple_asm_edges): Likewise.
7890 2013-05-24 Ian Bolton <ian.bolton@arm.com>
7892 * config/aarch64/aarch64.c (aarch64_print_operand): Change the
7893 X format specifier to only display bottom 16 bits.
7894 * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
7895 immediate to match for operand 2, since it will be masked.
7897 2013-05-24 Richard Biener <rguenther@suse.de>
7899 PR tree-optimization/57287
7900 * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
7901 all SSA names that occur in abnormal PHIs.
7903 2013-05-24 Alexander Ivchenko <alexander.ivchenko@intel.com>
7906 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
7907 that index is not negative.
7909 2013-05-24 Eric Botcazou <ebotcazou@adacore.com>
7911 PR rtl-optimization/55177
7912 * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
7913 (simplify_byte_swapping_operation): New.
7914 (simplify_binary_operation_1): Call it for AND, IOR and XOR.
7915 (simplify_relational_operation_1): Deal with BSWAP.
7917 2013-05-23 Richard Henderson <rth@redhat.com>
7920 * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
7921 (ix86_reorg): Call it.
7923 2013-05-23 Uros Bizjak <ubizjak@gmail.com>
7926 * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
7927 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
7928 REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
7930 2013-05-23 Christian Bruel <christian.bruel@st.com>
7933 * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
7935 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7936 Marcus Shawcroft <marcus.shawcroft@arm.com>
7938 * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
7939 * config/aarch64/constraints.md (Usa): Remove.
7940 * doc/md.texi (AArch64 Usa): Remove.
7942 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
7943 Marcus Shawcroft <marcus.shawcroft@arm.com>
7945 * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
7946 * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
7947 * config/aarch64/predicates.md (aarch64_const_address): Remove.
7948 (aarch64_mov_operand): Use aarch64_mov_operand_p.
7950 2013-05-23 Vidya Praveen <vidyapraveen@arm.com>
7952 * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
7953 instruction (AdvSIMD).
7954 * config/aarch64/aarch64-builtins.c
7955 (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
7956 * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
7958 2013-05-23 Martin Jambor <mjambor@suse.cz>
7961 * tree.h (contains_bitfld_component_ref_p): Declare.
7962 * tree-sra.c (contains_bitfld_comp_ref_p): Move...
7963 * tree.c (contains_bitfld_component_ref_p): ...here. Adjust its
7965 * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
7966 not access a bit-field. Assert all final offsets are byte-aligned.
7968 2013-05-23 Richard Biener <rguenther@suse.de>
7970 PR tree-optimization/57380
7971 * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
7972 least one invariant or re-used load.
7973 * passes.c (init_optimization_passes): Move pass_phiprop before
7976 2013-05-23 James Greenhalgh <james.greenhalgh@arm.com>
7978 * config/aarch64/aarch64-simd.md
7979 (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
7981 2013-05-23 Richard Biener <rguenther@suse.de>
7984 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
7985 OEP_CONSTANT_ADDRESS_OF retained.
7987 2013-05-23 Jakub Jelinek <jakub@redhat.com>
7990 * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
7991 don't lower unit. Handle unit not being always BITS_PER_WORD.
7993 2013-05-23 Richard Biener <rguenther@suse.de>
7995 PR rtl-optimization/57341
7996 * ira.c (validate_equiv_mem_from_store): Use anti_dependence
7997 instead of true_dependence.
7999 2013-05-22 David Malcolm <dmalcolm@redhat.com>
8001 * bb-reorder.c (branch_threshold): Make const.
8002 (exec_threshold): Ditto.
8004 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
8005 Pat Haugen <pthaugen@us.ibm.com>
8006 Peter Bergner <bergner@vnet.ibm.com>
8008 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8009 documentation for the power8 crypto builtins.
8011 * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
8013 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
8014 macros for defining power8 builtin functions.
8015 (BU_P8V_AV_2): Likewise.
8016 (BU_P8V_AV_P): Likewise.
8017 (BU_P8V_VSX_1): Likewise.
8018 (BU_P8V_OVERLOAD_1): Likewise.
8019 (BU_P8V_OVERLOAD_2): Likewise.
8020 (BU_CRYPTO_1): Likewise.
8021 (BU_CRYPTO_2): Likewise.
8022 (BU_CRYPTO_3): Likewise.
8023 (BU_CRYPTO_OVERLOAD_1): Likewise.
8024 (BU_CRYPTO_OVERLOAD_2): Likewise.
8025 (XSCVSPDP): Fix typo, point to the correct instruction.
8026 (VCIPHER): Add power8 crypto builtins.
8027 (VCIPHERLAST): Likewise.
8028 (VNCIPHER): Likewise.
8029 (VNCIPHERLAST): Likewise.
8030 (VPMSUMB): Likewise.
8031 (VPMSUMH): Likewise.
8032 (VPMSUMW): Likewise.
8033 (VPERMXOR_V2DI): Likewise.
8034 (VPERMXOR_V4SI: Likewise.
8035 (VPERMXOR_V8HI: Likewise.
8036 (VPERMXOR_V16QI: Likewise.
8037 (VSHASIGMAW): Likewise.
8038 (VSHASIGMAD): Likewise.
8040 (VPERMXOR): Likewise.
8041 (VSHASIGMA): Likewise.
8043 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8044 __CRYPTO__ if the crypto instructions are available.
8045 (altivec_overloaded_builtins): Add support for overloaded power8
8048 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
8049 support for power8 crypto builtins.
8050 (builtin_function_type): Likewise.
8051 (altivec_init_builtins): Add support for builtins that take vector
8052 long long (V2DI) arguments.
8054 * config/rs6000/crypto.md: New file, define power8 crypto
8057 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
8058 Pat Haugen <pthaugen@us.ibm.com>
8059 Peter Bergner <bergner@vnet.ibm.com>
8061 * doc/invoke.texi (Option Summary): Add power8 options.
8062 (RS/6000 and PowerPC Options): Likewise.
8064 * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
8065 constraints.md instead of rs6000.h. Reorder w* constraints. Add
8066 wm, wn, wr documentation.
8068 * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
8069 registers if direct move instructions are enabled.
8070 (wn): New constraint for no registers.
8071 (wq): New constraint for quad word even GPR registers.
8072 (wr): New constraint if 64-bit instructions are enabled.
8073 (wv): New constraint if power8 vector instructions are enabled.
8074 (wQ): New constraint for quad word memory locations.
8076 * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
8077 constraint for 0..15 for crypto instructions.
8078 (gpc_reg_operand): If VSX allow registers in VSX registers as well
8079 as GPR and floating point registers.
8080 (int_reg_operand): New predicate to match only GPR registers.
8081 (base_reg_operand): New predicate to match base registers.
8082 (quad_int_reg_operand): New predicate to match even GPR registers
8083 for quad memory operations.
8084 (vsx_reg_or_cint_operand): New predicate to allow vector logical
8085 operations in both GPR and VSX registers.
8086 (quad_memory_operand): New predicate for quad memory operations.
8087 (reg_or_indexed_operand): New predicate for direct move support.
8089 * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
8090 Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
8091 (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
8092 (POWERPC_MASKS): Add power8 options.
8093 (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
8096 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8097 Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
8099 * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
8100 (-mpower8-fusion): New power8 options.
8101 (-mpower8-fusion-sign): Likewise.
8102 (-mpower8-vector): Likewise.
8103 (-mcrypto): Likewise.
8104 (-mdirect-move): Likewise.
8105 (-mquad-memory): Likewise.
8107 * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
8109 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
8111 (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
8112 (rs6000_debug_vector_unit): Add p8_vector.
8113 (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
8114 definitions. Also print fusion state.
8115 (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
8116 (rs6000_builtin_mask_calculate): Add power8 builtin support.
8117 (rs6000_option_override_internal): Add support for power8.
8118 (rs6000_common_init_builtins): Add debugging for skipped builtins
8120 (rs6000_adjust_cost): Add power8 support.
8121 (rs6000_issue_rate): Likewise.
8122 (insn_must_be_first_in_group): Likewise.
8123 (insn_must_be_last_in_group): Likewise.
8124 (force_new_group): Likewise.
8125 (rs6000_register_move_cost): Likewise.
8126 (rs6000_opt_masks): Likewise.
8128 * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
8129 power8 capable assembler, default to power7 options.
8130 (TARGET_DIRECT_MOVE): Likewise.
8131 (TARGET_CRYPTO): Likewise.
8132 (TARGET_P8_VECTOR): Likewise.
8133 (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
8134 (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
8135 (VECTOR_MEM_P8_VECTOR_P): Likewise.
8136 (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
8137 (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
8138 (TARGET_XSCVDPSPN): Likewise.
8139 (TARGET_XSCVSPDPN): Likewsie.
8140 (TARGET_SYNC_HI_QI): Likewise.
8141 (TARGET_SYNC_TI): Likewise.
8142 (MASK_CRYPTO): Likewise.
8143 (MASK_DIRECT_MOVE): Likewise.
8144 (MASK_P8_FUSION): Likewise.
8145 (MASK_P8_VECTOR): Likewise.
8146 (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
8147 temporary used by some of the direct move instructions to get two FP
8148 temporary registers does not force creation of a stack frame.
8149 (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
8150 (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
8151 that any VSX registers are tieable, even if they are also an
8152 Altivec vector mode.
8153 (r6000_reg_class_enum): Add wm, wr, wv constraints.
8154 (RS6000_BTM_P8_VECTOR): Power8 builtin support.
8155 (RS6000_BTM_CRYPTO): Likewise.
8156 (RS6000_BTM_COMMON): Likewise.
8158 * config/rs6000/rs6000.md (cpu attribute): Add power8.
8159 * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
8160 (enum rs6000_vector): Add power8 vector support.
8162 2013-05-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8166 * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
8167 (any_sibcall_could_use_r3): this and handle indirect calls.
8168 (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
8170 2013-05-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8172 * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
8174 2013-05-22 Richard Biener <rguenther@suse.de>
8177 * profile.c (branch_prob): Do not split blocks that are
8178 abnormally receiving from ECF_RETURNS_TWICE functions.
8180 2013-05-22 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
8182 * recog.c (offsettable_address_addr_space_p): Fix calculation of
8183 address mode. Move pointer mode initialization to the same place.
8185 2013-05-22 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
8187 * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
8188 while it has any effect.
8190 2013-05-21 Easwaran Raman <eraman@google.com>
8192 PR tree-optimization/57322
8193 * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
8194 UID of the statement added to the BB to be 1.
8196 2013-05-21 Jakub Jelinek <jakub@redhat.com>
8198 PR tree-optimization/57331
8199 * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
8200 of conversion from pointer type to integral type with integer.
8202 2013-05-21 Martin Jambor <mjambor@suse.cz>
8205 * ipa-prop.c (ipa_read_node_info): Process param_used and
8206 controlled_uses in the same order as when writing.
8208 2013-05-21 Magnus Granberg <baldrick@free.fr>
8211 * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
8213 2013-05-21 Richard Biener <rguenther@suse.de>
8215 PR tree-optimization/57318
8216 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
8217 estimate stmts with side-effects as likely eliminated.
8219 2013-05-21 Richard Biener <rguenther@suse.de>
8221 PR tree-optimization/57330
8222 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
8223 preserve the call stmts fntype.
8225 2013-05-21 Richard Biener <rguenther@suse.de>
8227 PR tree-optimization/57303
8228 * tree-ssa-sink.c (statement_sink_location): Improve killing
8229 stmt detection and properly handle self-assignments.
8231 2013-05-21 Christian Bruel <christian.bruel@st.com>
8233 * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
8234 spanning registers. LEAF_REG_REMAP is supported only for contiguous
8235 registers. Set register size out of the PARALLEL loop.
8237 2013-05-20 Oleg Endo <olegendo@gcc.gnu.org>
8240 * config/sh/sh.md (fmasf4): Remove empty constraints strings.
8241 (*fmasf4, *fmasf4_media): New insns.
8243 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
8245 * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
8246 * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
8247 (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
8248 (mips_idiv_insns): Update the comments to say that the returned
8249 instruction counts are in units of BASE_INSN_LENGTH.
8250 (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
8251 by BASE_INSN_LENGTH rather than 4. Add the jump separately,
8252 using 2 rather than 4 as the length of indirect MIPS16 and
8253 microMIPS jumps. Use NOP_INSN_LENGTH rather than 4 as the
8254 length of a NOP. Don't divide MIPS16 lengths by 2.
8255 (mips16_split_long_branches): Assume a branch is long if the
8256 length is greater than 4 rather than 8.
8257 * config/mips/mips.md (length): Give MIPS16 lengths directly,
8258 rather than multiplying them by 2. Multiply instruction counts
8259 by BASE_INSN_LENGTH rather than 4.
8260 (*jump_mips16, tls_get_tp_mips16_<mode>)
8261 (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
8263 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
8265 * config/mips/mips.md (extended_mips16): Remove branch case.
8266 (length): Remove duplicated extended_mips16 test.
8268 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
8270 * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
8272 2013-05-18 Richard Sandiford <rdsandiford@googlemail.com>
8274 * recog.h (Recog_data): Rename to...
8275 (recog_data_d): ...this.
8276 (recog_data): Update accordingly.
8277 * recog.c (recog_data): Likewise.
8278 * reload.c (save_recog_data): Likewise.
8279 * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
8280 (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
8282 2013-05-17 Julian Brown <julian@codesourcery.com>
8284 * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
8285 found in a REG_EQUAL note, invalidate it.
8287 2013-05-17 Easwaran Raman <eraman@google.com>
8289 * tree-ssa-reassoc.c (find_insert_point): New function.
8290 (insert_stmt_after): Likewise.
8291 (get_def_stmt): Likewise.
8292 (ensure_ops_are_available): Likewise.
8293 (not_dominated_by): Likewise.
8294 (rewrite_expr_tree): Do not move statements beyond what is
8295 necessary. Remove call to swap_ops_for_binary_stmt...
8296 (reassociate_bb): ... and move it here.
8297 (build_and_add_sum): Assign UIDs for new statements.
8298 (linearize_expr): Likewise.
8299 (do_reassoc): Renumber gimple statement UIDs.
8301 2013-05-17 Jan Hubicka <jh@suse.cz>
8303 * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
8305 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
8306 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
8308 (output_weakrefs): Update.
8310 2013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
8311 Martin Jambor <mjambor@suse.cz>
8314 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
8315 value that corresponds to the given aggval is found in values vector.
8317 2013-05-17 Uros Bizjak <ubizjak@gmail.com>
8319 * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
8320 sse, sse2, sse3, ssse3 and sse4a flags to options.
8322 2013-05-17 David Malcolm <dmalcolm@redhat.com>
8324 * gengtype-state.c: (s_expr_writer): New class, to handle
8325 prettifying of output layout of s-expressions.
8326 (state_writer): New class, to write out gtype.state.
8327 (state_written_type_count): Move this variable into member data of
8329 (s_expr_writer::s_expr_writer): New code: constructor for new class
8330 (state_writer::state_writer(): ditto
8331 (s_expr_writer::write_new_line): New function
8332 (s_expr_writer::write_any_indent): ditto
8333 (s_expr_writer::begin_s_expr): ditto
8334 (s_expr_writer::end_s_expr): ditto
8335 (write_state_fileloc): convert to method of state_writer...
8336 (state_writer:: write_state_fileloc): ...and use methods of
8337 s_expr_writer to write indentation into the gtype.state output file
8338 to visually represent the hierarchical structure of the list
8340 (write_state_fields): ditto, renaming to...
8341 (state_writer::write_state_fields)
8342 (write_state_a_string): ditto, renaming to...
8343 (state_writer::write_state_a_string)
8344 (write_state_string_option): ditto, renaming to...
8345 (state_writer::write_state_string_option)
8346 (write_state_type_option): ditto, renaming to...
8347 (state_writer::write_state_type_option)
8348 (write_state_nested_option): ditto, renaming to...
8349 (state_writer::write_state_nested_option)
8350 (write_state_option): ditto, renaming to...
8351 (state_writer::write_state_option)
8352 (write_state_options): ditto, renaming to...
8353 (state_writer::write_state_options)
8354 (write_state_lang_bitmap): ditto, renaming to...
8355 (state_writer::write_state_lang_bitmap)
8356 (write_state_version): ditto, renaming to...
8357 (state_writer::write_state_version)
8358 (write_state_scalar_type): ditto, renaming to...
8359 (state_writer::write_state_scalar_type)
8360 (write_state_string_type): ditto, renaming to...
8361 (state_writer::write_state_string_type)
8362 (write_state_undefined_type): ditto, renaming to...
8363 (state_writer::write_state_undefined_type)
8364 (write_state_struct_union_type): ditto, renaming to...
8365 (state_writer::write_state_struct_union_type)
8366 (write_state_struct_type): ditto, renaming to...
8367 (state_writer::write_state_struct_type)
8368 (write_state_user_struct_type): ditto, renaming to...
8369 (state_writer::write_state_user_struct_type)
8370 (write_state_lang_struct_type): ditto, renaming to...
8371 (state_writer::write_state_lang_struct_type)
8372 (write_state_param_struct_type): ditto, renaming to...
8373 (state_writer::write_state_param_struct_type)
8374 (write_state_pointer_type): ditto, renaming to...
8375 (state_writer::write_state_pointer_type)
8376 (write_state_array_type): ditto, renaming to...
8377 (state_writer::write_state_array_type)
8378 (write_state_gc_used): ditto, renaming to...
8379 (state_writer::write_state_gc_used)
8380 (write_state_common_type_content): ditto, renaming to...
8381 (state_writer::write_state_common_type_content)
8382 (write_state_type): ditto, renaming to...
8383 (state_writer::write_state_type)
8384 (write_state_pair_list): ditto, renaming to...
8385 (state_writer::write_state_pair_list)
8386 (write_state_pair): ditto, renaming to...
8387 (state_writer::write_state_pair)
8388 (write_state_typedefs): ditto, renaming to...
8389 (state_writer::write_state_typedefs)
8390 (write_state_structures): ditto, renaming to...
8391 (state_writer::write_state_structures)
8392 (write_state_param_structs): ditto, renaming to...
8393 (state_writer::write_state_param_structs)
8394 (write_state_variables): ditto, renaming to...
8395 (state_writer::write_state_variables)
8396 (write_state_srcdir): ditto, renaming to...
8397 (state_writer::write_state_srcdir)
8398 (write_state_files_list): ditto, renaming to...
8399 (state_writer::write_state_files_list)
8400 (write_state_languages): ditto, renaming to...
8401 (state_writer::write_state_languages)
8402 (write_state): create a state_writer instance and use it when
8403 writing out the state file
8405 2013-05-17 Mike Stump <mikestump@comcast.net>
8407 PR rtl-optimization/57304
8408 * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
8409 accessing DF_REF_REAL_LOC.
8411 2013-05-17 Jakub Jelinek <jakub@redhat.com>
8413 PR rtl-optimization/57281
8414 PR rtl-optimization/57300
8415 * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
8416 (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
8417 what the other splitter did if the registers are dead.
8419 2013-05-17 Richard Biener <rguenther@suse.de>
8421 * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
8424 2013-05-17 Jakub Jelinek <jakub@redhat.com>
8426 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
8429 2013-05-17 Marek Polacek <polacek@redhat.com>
8431 * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
8432 length when doing non-zero store of storing '\0' to '\0'.
8434 2013-05-17 Jakub Jelinek <jakub@redhat.com>
8436 * tree-vect-patterns.c (vect_recog_rotate_pattern): For
8437 vect_external_def oprnd1 with loop_vinfo, try to emit
8438 optional cast, negation and and stmts on the loop preheader
8439 edge instead of into the pattern def seq.
8441 PR tree-optimization/57051
8442 * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
8443 case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
8445 2013-05-16 Nick Clifton <nickc@redhat.com>
8447 * config/rl78/rl78.c (rl78_attribute_table): Add naked.
8448 (rl78_is_naked_func): New function.
8449 (rl78_expand_prologue): Skip prologue generation for naked functions.
8450 (rl78_expand_epilogue): Skip epilogue generation for naked functions.
8451 * doc/extend.texi (naked): Add RL78 to the list of processors
8452 that supports this attribute.
8454 2013-05-16 Jeff Law <law@redhat.com>
8456 * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
8458 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
8460 * config/i386/driver-i386.c (host_detect_local_cpu): Determine
8461 cache parameters using detect_caches_amd also for CYRIX,
8462 NSC and TM2 signatures.
8464 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
8465 Dzianis Kahanovich <mahatma@eu.by>
8469 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8470 VIA/Centaur processors and determine their cache parameters
8471 using detect_caches_amd.
8473 2013-05-16 Teresa Johnson <tejohnson@google.com>
8475 * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
8476 (rtl_verify_edges): New function.
8477 (rtl_verify_bb_insns): Ditto.
8478 (rtl_verify_bb_pointers): Ditto.
8479 (rtl_verify_bb_insn_chain): Ditto.
8480 (rtl_verify_fallthru): Ditto.
8481 (rtl_verify_bb_layout): Ditto.
8482 (rtl_verify_flow_info_1): Outline checks into new functions.
8483 (rtl_verify_flow_info): Ditto.
8485 2013-05-16 Steve Ellcey <sellcey@imgtec.com>
8487 * cfghooks.c (copy_bbs): Add update_dominance argument.
8488 * cfghooks.h (copy_bbs): Update prototype.
8489 * tree-cfg.c (gimple_duplicate_sese_region):
8490 Add update_dominance argument.
8491 * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
8492 * tree-ssa-loop-ch.c (copy_loop_headers): Update
8493 gimple_duplicate_sese_region call.
8494 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
8495 Update copy_bbs call.
8496 * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
8497 * trans-mem.c (ipa_uninstrument_transaction): Ditto.
8499 2013-05-16 Jakub Jelinek <jakub@redhat.com>
8501 * tree-vectorizer.h (NUM_PATTERNS): Increment.
8502 * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
8503 vect_recog_rotate_pattern.
8504 (vect_recog_rotate_pattern): New function.
8506 2013-05-16 Jason Merrill <jason@redhat.com>
8508 * Makefile.in (LLINKER): New variable.
8509 (mostlyclean): Remove link mutex.
8510 * configure.ac: Handle --enable-link-mutex.
8511 * lock-and-run.sh: New script.
8513 2013-05-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8516 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
8519 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8521 * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
8523 2013-05-16 Greta Yorsh <Greta.Yorsh@arm.com>
8525 * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
8526 * config/arm/arm.c (next_consecutive_mem): New function.
8527 (gen_movmem_ldrd_strd): Likewise.
8528 * config/arm/arm.md (movmemqi): Update condition and code.
8529 (unaligned_loaddi, unaligned_storedi): New patterns.
8531 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8533 * config.gcc: Obsolete *-*-solaris2.9*.
8534 * doc/install.texi (Specific, *-*-solaris2*): Document it.
8536 2013-05-16 Richard Biener <rguenther@suse.de>
8538 * passes.c (init_optimization_passes): Move pass_parallelize_loops
8539 earlier, after GRAPHITE transforms and IV canonicalization.
8541 2013-05-16 Jakub Jelinek <jakub@redhat.com>
8543 * omp-low.c (extract_omp_for_data): For collapsed loops,
8544 if at least one of the loops is known at compile time to
8545 iterate zero times, set count to 0.
8546 (expand_omp_regimplify_p): New function.
8547 (expand_omp_for_generic): For collapsed loops, if at least
8548 one of the loops isn't known to iterate at least once,
8549 add runtime check with setting count to 0.
8550 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
8551 For unsigned types if it isn't known at compile time that
8552 the loop will iterate at least once, add runtime check to bypass
8553 the whole loop if initial condition isn't true.
8555 2013-05-16 Nathan Sidwell <nathan@codesourcery.com>
8557 * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
8559 2013-05-16 Marc Glisse <marc.glisse@inria.fr>
8562 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
8563 transformations to avoid an infinite loop.
8565 2013-05-16 Marek Polacek <polacek@redhat.com>
8567 * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
8569 2013-05-15 Leif Ekblad <leif@rdos.net>
8571 * config/i386/i386.c (ix86_decompose_address): Use
8572 DEFAULT_TLS_SEG_REG to access TLS segment register.
8573 * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
8574 * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
8575 (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
8577 2013-05-15 Richard Sandiford <rdsandiford@googlemail.com>
8580 * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
8581 sibling calls to functions that would normally be lazily bound,
8582 unless $gp is call-clobbered.
8584 2013-05-15 Uros Bizjak <ubizjak@gmail.com>
8586 * config/i386/i386.c (ix86_option_override_internal): Update
8587 processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags. Add
8588 PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
8589 non-SSE 3dNow! targets. Enable TARGET_PRFCHW for TARGET_3DNOW targets.
8590 * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
8592 (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
8594 2013-05-15 Andreas Schwab <schwab@suse.de>
8596 * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
8597 for rotlhi3+1 and rotlqi3+1, resp. Fix reference to non-existing
8600 2013-05-15 Teresa Johnson <tejohnson@google.com>
8602 * loop-unroll.c (report_unroll_peel): Check decision before
8603 emitting unroll/peel message.
8605 2013-05-15 Teresa Johnson <tejohnson@google.com>
8607 * function.h (has_bb_partition): New rtl_data flag.
8608 (bb_reorder_complete): Ditto.
8609 * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
8610 instead of flag_reorder_blocks_and_partition.
8611 * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
8612 with some enhancements.
8613 (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
8614 * bb-reorder.c (connect_traces): Check for has_bb_partition
8615 instead of flag_reorder_blocks_and_partition.
8616 (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
8617 (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
8618 verify_hot_cold_block_grouping.
8619 (partition_hot_cold_basic_blocks): Set has_bb_partition.
8621 2013-05-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8624 * config/arm/predicates.md (call_insn_operand): New predicate.
8625 * config/arm/constraints.md ("Cs", "Ss"): New constraints.
8626 * config/arm/arm.md (*call_insn, *call_value_insn): Match only
8627 if insn is not a tail call.
8628 (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
8630 * config/arm/arm.h (enum reg_class): New caller save register class.
8631 (REG_CLASS_NAMES): Likewise.
8632 (REG_CLASS_CONTENTS): Likewise.
8633 * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
8636 2013-05-15 Richard Biener <rguenther@suse.de>
8638 * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
8639 of MSG_OPTIMIZED_LOCATIONS.
8640 * tree-vect-slp.c (vect_make_slp_decision): Likewise.
8641 (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
8643 * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
8644 of MSG_OPTIMIZED_LOCATIONS.
8645 (execute_vect_slp): Likewise.
8646 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
8647 (vect_create_cond_for_alias_checks): Likewise.
8648 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8649 (vect_recog_widen_mult_pattern): Likewise.
8650 (vect_recog_widen_sum_pattern): Likewise.
8651 (vect_recog_over_widening_pattern): Likewise.
8652 (vect_recog_widen_shift_pattern): Likewise.
8653 (vect_recog_vector_vector_shift_pattern): Likewise.
8654 (vect_recog_divmod_pattern): Likewise.
8655 (vect_recog_mixed_size_cond_pattern): Likewise.
8656 (vect_recog_bool_pattern): Likewise.
8657 (vect_pattern_recog_1): Likewise.
8659 2013-05-15 Martin Jambor <mjambor@suse.cz>
8661 * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
8662 non-functions to builtin_unreachable.
8663 * ipa-inline-transform.c (inline_call): Do not assert estimates were
8664 correct when new direct edges were discovered.
8666 2013-05-15 Martin Jambor <mjambor@suse.cz>
8668 * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
8669 header, print symbol order instead of node uid, print more information
8670 about indirect edge targets.
8671 (ipa_make_edge_direct_to_target): Print symbol order instead of node
8673 (ipa_make_edge_direct_to_target): Likewise.
8674 (remove_described_reference): Likewise.
8675 (propagate_controlled_uses): Likewise.
8676 (ipa_print_node_params): Also print symbol order.
8677 (ipcp_transform_function): Print symbol order instead of node uids.
8678 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
8679 (cgraph_get_create_real_symbol_node): Likewise.
8680 * ipa-cp.c (print_lattice): Likewise.
8681 (print_all_lattices): Likewise.
8682 (determine_versionability): Likewise.
8683 (initialize_node_lattices): Likewise.
8684 (estimate_local_effects): Likewise.
8685 (update_profiling_info): Likewise.
8686 (create_specialized_node): Likewise.
8687 (perhaps_add_new_callers): Likewise.
8688 (decide_about_value): Likewise.
8689 (decide_whether_version_node): Likewise.
8690 (identify_dead_nodes): Likewise.
8691 * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
8692 (dump_inline_summary): Likewise.
8693 (estimate_node_size_and_time): Likewise.
8694 (inline_analyze_function): Likewise.
8695 * ipa-inline.c (report_inline_failed_reason): Likewise.
8696 (want_early_inline_function_p): Likewise.
8697 (edge_badness): Likewise.
8698 (update_edge_key): Likewise.
8699 (inline_small_functions): Likewise. Add dumping of order to two other
8701 * ipa-pure-const.c (pure_const_read_summary): Print symbol order
8702 instead of node uids.
8703 (propagate_pure_const): Likewise.
8704 (propagate_pure_const): Likewise.
8705 * ipa-utils.c (dump_cgraph_node_set): Likewise.
8706 * lto-cgraph.c (input_node): Explicitly specify we dump uid.
8707 * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
8709 * tree-pretty-print.c (dump_function_header): Likewise.
8710 * tree-sra.c (convert_callers_for_node): Dump in traditional format.
8711 Print symbol order instead of node uids.
8713 2013-05-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
8715 * config/s390/s390.c (s390_register_move_cost): Don't impose the
8716 FPR<->GPR move cost penalty if ldgr/lgdr can be used.
8718 2013-05-15 Richard Biener <rguenther@suse.de>
8720 PR tree-optimization/57275
8721 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
8722 return value for fail to do runtime alias checks for gather loads.
8724 2013-05-15 Jan Hubicka <jh@suse.cz>
8728 * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
8729 weakrefs are not external.
8730 (lto_symtab_merge_decls): Fix thinko when dealing with
8731 non-lto_symtab decls.
8732 (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
8733 (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
8734 * varpool.c (dump_varpool_node): Dump more flags.
8736 2013-05-15 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
8738 * config/i386/i386.c (processor_alias_table): Add instruction
8739 FSGSBASE for AMD bdver3 architecture.
8741 2013-05-14 Jakub Jelinek <jakub@redhat.com>
8743 * tree.c (warn_deprecated_use): Print file:line using locus color.
8744 * diagnostic.c (diagnostic_report_current_module): Print file:line
8745 and file:line:column using locus color.
8747 2013-05-14 Mike Stump <mikestump@comcast.net>
8749 * gdbinit.in: Add __null.
8751 2013-05-14 Mike Stump <mikestump@comcast.net>
8753 * recog.h: Rename struct recog_data to Recog_data.
8754 * recog.c: Likewise.
8755 * reload.c (can_reload_into): Likewise.
8756 * config/picochip/picochip.c: Likewise.
8758 2013-05-14 Mike Stump <mikestump@comcast.net>
8760 * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
8762 2013-05-14 Steven Bosscher <steven@gcc.gnu.org>
8764 * resource.h (struct resources): Remove unch_memory member.
8765 (CLEAR_RESOURCE): Don't clear unch_memory.
8766 * resource.c (mark_referenced_resources): Don't set it.
8767 (mark_set_resources): Likewise.
8768 (mark_target_live_regs): Don't clear it.
8769 (init_resource_info): Likewise.
8770 * reorg.c (resource_conflicts_p): Don't compare it.
8771 (redundant_insn): Don't set it.
8773 * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
8775 * emit-rtl.c (next_label): Remove unused function.
8776 (skip_consecutive_labels, link_cc0_insns): Move to ...
8777 * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
8778 only place where these functions are used, and make them static.
8780 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
8782 * fold-const.c (fold_negate_expr): Handle vectors.
8783 (fold_truth_not_expr): Make it static.
8784 (fold_invert_truthvalue): New static function.
8785 (invert_truthvalue_loc): Handle vectors. Do not call
8786 fold_truth_not_expr directly.
8787 (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
8788 <TRUTH_NOT_EXPR>: Do not cast to boolean.
8789 (fold_comparison): Handle vector constants.
8790 (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
8791 (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
8792 * tree.h (fold_truth_not_expr): Remove declaration.
8794 2013-05-14 James Greenhalgh <james.greenhalgh@arm.com>
8796 * config/aarch64/aarch64-simd.md
8797 (aarch64_vcond_internal<mode>): Rename to...
8798 (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
8799 (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
8800 float modes. Clarify all iterator modes.
8801 (vcond<mode><mode>): Use new name for vcond expanders.
8802 (vcond<v_cmp_result><mode>): Likewise.
8803 (vcondu<mode><mode>: Likewise.
8804 * config/aarch64/iterators.md (VDQF_COND): New.
8806 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
8809 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
8810 variable for the shift amount. Check that we shift by non-negative
8813 2013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
8816 * config/arm/arm.h (EPILOGUE_USES): Only return true
8817 for LR_REGNUM after epilogue_completed.
8819 2013-05-14 Joern Rennecke <joern.rennecke@embecosm.com>
8821 * config/avr/avr.c (avr_encode_section_info): Bail out if the type
8824 2013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8827 * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
8828 and Solaris 11+/x86 with gld.
8829 * configure: Regenerate.
8831 2013-05-14 Jakub Jelinek <jakub@redhat.com>
8833 * expmed.c (expand_shift_1): Canonicalize rotates by
8834 constant bitsize / 2 to bitsize - 1.
8835 * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
8836 case ROTATERT>: Likewise.
8839 2013-05-10 Jakub Jelinek <jakub@redhat.com>
8841 * config/i386/i386.md (rotateinv): New code attr.
8842 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
8843 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
8844 roll $31, %eax, etc.
8846 2013-05-14 Richard Biener <rguenther@suse.de>
8849 * tree-eh.c (sink_clobbers): Give up for successors with
8850 multiple predecessors and no virtual uses.
8852 2013-05-14 Eric Botcazou <ebotcazou@adacore.com>
8854 * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
8855 * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
8857 2013-05-14 Jakub Jelinek <jakub@redhat.com>
8860 * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
8861 the case when both op0 and op1 have VOIDmode.
8863 2013-05-14 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
8865 * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
8866 in multiply-accumulate mode.
8868 2013-05-13 Guozhi Wei <carrot@google.com>
8870 * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
8872 2013-05-13 Kai Tietz <ktietz@redhat.com>
8875 * config/i386/cygming.h (TARGET_PECOFF): Define as true.
8876 * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
8877 (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
8878 * config/i386/i386.c (ix86_option_override_internal): Likewise.
8879 (ix86_expand_prologue): Likewise.
8880 (ix86_expand_split_stack_prologue): Likewise.
8881 (legitimate_pic_address_disp_p): Likewise.
8882 (legitimize_pic_address): Likewise.
8883 (legitimize_tls_address): Likewise.
8884 (legitimize_pe_coff_symbol): Likewise.
8885 (output_pic_addr_const): Likewise.
8886 (construct_plt_address): Likewise.
8887 (ix86_expand_call): Likewise.
8888 (x86_output_mi_thunk): Likewise.
8889 (x86_function_profiler): Likewise.
8891 2013-05-13 Sofiane Naci <sofiane.naci@arm.com>
8893 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
8894 similar switch cases.
8895 (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
8896 (aarch64_simd_mov_to_<mode>low): Delete.
8897 (aarch64_simd_mov_to_<mode>high): Delete.
8898 (move_lo_quad_<mode>): Add w<-r alternative.
8899 (aarch64_simd_move_hi_quad_<mode>): Likewise.
8900 (aarch64_simd_mov_from_*): Update type attribute.
8901 * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
8904 2013-05-13 Jan Hubicka <jh@suse.cz>
8906 * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
8907 * config/i386/i386.c (ix86_compute_frame_layout,
8908 ix86_expand_epilogue, emit_i387_cw_initialization,
8909 ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
8910 ix86_local_alignment): Fix use of size/speed predicates.
8912 2013-05-13 Jakub Jelinek <jakub@redhat.com>
8914 PR tree-optimization/45216
8915 PR tree-optimization/57157
8916 * tree-ssa-forwprop.c (simplify_rotate): Only recognize
8917 the (-Y) & (B - 1) variant if OP is |.
8918 * expmed.c (expand_shift_1): For rotations by const0_rtx just
8919 return shifted. Use (-op1) & (prec - 1) as other_amount
8920 instead of prec - op1.
8922 2013-05-13 Martin Jambor <mjambor@suse.cz>
8925 * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
8926 (ipa_constant_data): New type.
8927 (ipa_jump_func): Use ipa_constant_data to hold information about
8928 constant jump functions.
8929 (ipa_get_jf_constant): Adjust to jump function type changes.
8930 (ipa_get_jf_constant_rdesc): New function.
8931 (ipa_param_descriptor): New field controlled_uses.
8932 (ipa_get_controlled_uses): New function.
8933 (ipa_set_controlled_uses): Likewise.
8934 * ipa-ref.h (ipa_find_reference): Declare.
8935 * ipa-prop.c (ipa_cst_ref_desc): New type.
8936 (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
8938 (ipa_set_jf_constant): Likewise. Also create reference descriptions.
8939 New parameter cs. Adjust all callers.
8940 (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
8941 (remove_described_reference): New function.
8942 (jfunc_rdesc_usable): Likewise.
8943 (try_make_edge_direct_simple_call): Decrement controlled use count,
8944 attempt to remove reference if it hits zero.
8945 (combine_controlled_uses_counters): New function.
8946 (propagate_controlled_uses): Likewise.
8947 (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
8948 (ipa_edge_duplication_hook): Duplicate reference descriptions.
8949 (ipa_print_node_params): Print described use counter.
8950 (ipa_write_jump_function): Adjust to jump function type changes.
8951 (ipa_read_jump_function): New parameter CS, pass it to
8952 ipa_set_jf_constant. Adjust caller.
8953 (ipa_write_node_info): Stream controlled use count
8954 (ipa_read_node_info): Likewise.
8955 * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
8957 * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
8958 count. Remove cloning-added reference if it reaches zero.
8959 * ipa-ref.c (ipa_find_reference): New function.
8961 2013-05-13 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
8963 * config/i386/i386.c (processor_target_table): Modified default
8964 alignment values for AMD BD and BT architectures.
8966 2013-05-13 Marc Glisse <marc.glisse@inria.fr>
8968 * tree-vect-generic.c (uniform_vector_p): Move ...
8969 * tree.c (uniform_vector_p): ... here.
8970 * tree.h (uniform_vector_p): Declare it.
8971 * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
8974 2013-05-13 Jakub Jelinek <jakub@redhat.com>
8976 PR tree-optimization/57230
8977 * tree-ssa-strlen.c (handle_char_store): Record length for
8978 array store from STRING_CST.
8980 PR tree-optimization/57230
8981 * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
8984 2013-05-12 Joern Rennecke <joern.rennecke@embecosm.com>
8986 * config/epiphany/epiphany.c (epiphany_init): Check size of
8987 NUM_MODES_FOR_MODE_SWITCHING.
8988 (epiphany_expand_prologue):
8989 Remove CONFIG_REGNUM initial value handling code.
8990 (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
8991 (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
8992 (emit_set_fp_mode, epiphany_mode_after): Likewise.
8993 (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
8994 Don't return 1 for FP_MODE_NONE.
8995 * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
8996 Add value for EPIPHANY_MSW_ENTITY_CONFIG.
8997 (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
8998 * config/epiphany/epiphany.md (save_config): New pattern.
9000 2013-05-12 Uros Bizjak <ubizjak@gmail.com>
9002 * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9004 2013-05-10 Uros Bizjak <ubizjak@gmail.com>
9006 * config/i386/i386.md (memory): Handle sseishft1.
9007 * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9008 (*vec_extractv2di_1): Ditto.
9010 2013-05-10 Vladimir Makarov <vmakarov@redhat.com>
9012 * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
9015 2013-05-10 Sebastian Huber <sebastian.huber@embedded-brains.de>
9017 * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
9018 Add mthumb/march=armv7-a multilib.
9019 Add mthumb/march=armv7-r multilib.
9020 Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
9022 2013-05-10 Ralf Corsépius <ralf.corsepius@rtems.org>
9024 * config/v850/t-rtems: Add more multilibs.
9026 2013-05-10 Richard Biener <rguenther@suse.de>
9028 PR tree-optimization/57214
9029 * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
9030 not propagate from SSA names that occur in abnormal PHI nodes.
9032 2013-05-10 Marc Glisse <marc.glisse@inria.fr>
9034 * stor-layout.c (element_precision): New function.
9035 * machmode.h (element_precision): Declare it.
9036 * tree.c (build_minus_one_cst): New function.
9037 (element_precision): Likewise.
9038 * tree.h (build_minus_one_cst): Declare new function.
9039 (element_precision): Likewise.
9040 * fold-const.c (operand_equal_p): Use element_precision.
9041 (fold_binary_loc): Handle vector types.
9042 * convert.c (convert_to_integer): Use element_precision.
9043 * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
9046 2013-05-10 Richard Sandiford <rdsandiford@googlemail.com>
9048 * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
9049 (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
9050 (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
9051 (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9052 * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
9053 (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
9054 (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
9055 (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9056 * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
9057 (Uuw8): New constraints.
9058 (Usb4): Move into alphabetical order.
9059 * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
9060 (sd8_operand, ub8_operand, uw8_operand): New predicates.
9061 * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
9062 previously unnamed patterns.
9063 (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
9064 (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
9065 (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
9066 (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
9067 of set_attr_alternative/if_then_else. Use extended_mips16 instead
9068 of specific lengths.
9070 2013-05-10 Jakub Jelinek <jakub@redhat.com>
9072 * config/i386/i386.md (rotateinv): New code attr.
9073 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9074 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9075 roll $31, %eax, etc.
9077 PR tree-optimization/45216
9078 PR tree-optimization/57157
9079 * tree-ssa-forwprop.c (simplify_rotate): New function.
9080 (ssa_forward_propagate_and_combine): Call it.
9082 2013-05-10 Richard Biener <rguenther@suse.de>
9084 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
9085 disable peeling when we version for aliasing.
9086 (vector_alignment_reachable_p): Honor explicit user alignment.
9087 (vect_supportable_dr_alignment): Likewise.
9088 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
9089 STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
9090 * tree-vect-loop.c (vect_transform_loop): First apply versioning,
9091 then peeling to arrange for the cost-model check to come first.
9093 2013-05-10 Alan Modra <amodra@gmail.com>
9095 * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
9096 (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
9097 * configure: Regenerate.
9099 2013-05-10 Alan Modra <amodra@gmail.com>
9102 * varasm.c (default_elf_select_section): Move !DECL_P check..
9103 (get_named_section): ..to here before calling get_section_name.
9105 (default_section_type_flags): Add DECL_P check.
9106 * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
9107 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
9109 2013-05-09 Joern Rennecke <joern.rennecke@embecosm.com>
9111 * config/epiphany/epiphany.c (epiphany_expand_prologue):
9112 When using gen_stack_adjust_str with a register offset, add a
9113 REG_FRAME_RELATED_EXPR note.
9115 2013-05-09 Uros Bizjak <ubizjak@gmail.com>
9117 * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
9118 (*vec_extractv4si_zext_mem): Ditto.
9119 (*vec_extractv2di): Add 0->x and x->x alternatives.
9120 * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
9121 * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
9123 2013-05-09 Jason Merrill <jason@redhat.com>
9125 N3639 C++1y VLA support
9126 * gimplify.c (gimplify_vla_decl): Don't touch an existing
9129 * tree.c (build_constructor_va): New.
9130 * tree.h: Declare it.
9132 2013-05-09 Martin Jambor <mjambor@suse.cz>
9135 * gimple-fold.c (canonicalize_constructor_val): Call
9136 cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
9138 2013-05-09 Jan Hubicka <jh@suse.cz>
9139 Richard Biener <rguenther@suse.de>
9142 * symtab.c (symtab_make_decl_local): Do not add private names.
9144 2013-05-09 Jan Hubicka <jh@suse.cz>
9147 * symtab.c (insert_to_assembler_name_hash): Handle clones.
9148 (unlink_from_assembler_name_hash): Likewise.
9149 (symtab_prevail_in_asm_name_hash, symtab_register_node,
9150 symtab_unregister_node, symtab_initialize_asm_name_hash,
9151 change_decl_assembler_name): Update.
9153 2013-05-09 Sofiane Naci <sofiane.naci@arm.com>
9155 * config/aarch64/aarch64.md: New movtf split.
9156 (*movtf_aarch64): Update.
9157 (aarch64_movdi_tilow): Handle TF modes and rename to
9158 aarch64_movdi_<mode>low.
9159 (aarch64_movdi_tihigh): Handle TF modes and rename to
9160 aarch64_movdi_<mode>high
9161 (aarch64_movtihigh_di): Handle TF modes and rename to
9162 aarch64_mov<mode>high_di
9163 (aarch64_movtilow_di): Handle TF modes and rename to
9164 aarch64_mov<mode>low_di
9165 (aarch64_movtilow_tilow): Remove spurious whitespace.
9166 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
9168 (aarch64_print_operand): Update.
9170 2013-05-09 Alan Modra <amodra@gmail.com>
9172 * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
9174 * configure: Regenerate.
9176 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
9178 * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
9179 splitter preparation statements.
9180 * config/i386/sse.md (*vec_extract* splitters): Ditto.
9181 (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
9184 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9186 * gimple-ssa-strength-reduction.c (count_candidates): Change
9187 return value to int.
9188 (analyze_candidates_and_replace): Change type of length to int.
9190 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
9192 * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
9193 (*vec_extract<mode>): Use VI12_128 mode iterator.
9194 (*vec_extract<mode>_mem): Ditto.
9195 (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
9198 2013-05-08 Diego Novillo <dnovillo@google.com>
9203 2012-08-17 Diego Novillo <dnovillo@google.com>
9206 * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
9207 * config.in: Regenerate.
9208 * configure: Regenerate.
9209 * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
9211 2013-05-08 Jan Hubicka <jh@suse.cz>
9214 * cgraph.c (cgraph_make_node_local_1): Se unique_name.
9215 * cgraph.h (symtab_node_base): Add unique_name.
9216 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9217 input_overwrite_node, input_varpool_node): Stream unique_name.
9218 * cgraphclones.c (cgraph_create_virtual_clone,
9219 cgraph_function_versioning): Set unique_name.
9220 * ipa.c (function_and_variable_visibility): Set unique_name.
9222 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9224 * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
9225 (alloc_cand_and_find_basis): Restrict conditional candidate
9226 processing to CAND_MULTs.
9228 2013-05-08 Jan Hubicka <jh@suse.cz>
9231 lto-symtab.c (lto_symtab_symbol_p): New function.
9232 (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
9233 lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
9234 lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
9235 Skip static symbols.
9237 2013-05-08 Paolo Carlini <paolo.carlini@oracle.com>
9239 PR tree-optimization/57200
9240 * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
9241 Only call inform if the preceding warning_at returns true.
9243 2013-05-07 Han Shen <shenhan@google.com>
9245 * cfgexpand.c (record_or_union_type_has_array_p): New function.
9246 (expand_used_vars): Add logic handling '-fstack-protector-strong'.
9247 * common.opt (fstack-protector-strong): New option.
9248 * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
9249 * doc/invoke.texi (Optimization Options): Document
9250 "-fstack-protector-strong".
9251 * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
9253 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
9255 * config/mips/mips.c (mips_machine_reorg2): Return 0.
9257 2013-05-07 Vladimir Makarov <vmakarov@redhat.com>
9259 * ira.c (update_equiv_regs): Add insn having equiv memory even if
9260 it is not lhs of the insn.
9261 (setup_reg_equiv): Remove insn having equiv memory which it is not
9263 * lra-constraints.c (process_address): Try to improve generation
9264 code for address base + disp.
9265 (lra_constraints): Make correct the code for checking insn setting
9266 up backward equivalence. Remove insn only if it is in the init
9268 * lra-eliminations.c (update_reg_eliminate): Change return value.
9269 (lra_eliminate): Use the result.
9271 2013-05-07 Uros Bizjak <ubizjak@gmail.com>
9273 * config/i386/sse.md (ssescalarnummask): New mode attribute.
9274 (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
9275 (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
9276 *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
9277 register target operands.
9278 (*vec_extractv8hi_sse2): New pattern.
9279 (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
9280 (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
9281 (*vec_extract<mode>_mem): New insn and split pattern.
9283 2013-05-07 Christophe Lyon <christophe.lyon@linaro.org>
9285 * config/arm/arm.c (arm_asan_shadow_offset): New function.
9286 (TARGET_ASAN_SHADOW_OFFSET): Define.
9287 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
9288 (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
9290 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9292 * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
9293 (incr_vec_index): Return -1 if increment not found.
9294 (create_add_on_incoming_edge): Assert if increment not found.
9295 (record_increment): Limit number of increments recorded.
9296 (all_phi_incrs_profitable): Return false if an increment not found.
9297 (replace_profitable_candidates): Don't process increments that were
9299 (analyze_candidates_and_replace): Limit size of incr_vec.
9301 2013-05-07 Richard Biener <rguenther@suse.de>
9303 * calls.c (special_function_p): setjmp-like functions are leaf.
9304 * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
9305 * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
9307 2013-05-07 Sofiane Naci <sofiane.naci@arm.com>
9309 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
9310 (aarch64_simd_mov<mode>): New expander.
9311 (aarch64_simd_mov_to_<mode>low): New instruction pattern.
9312 (aarch64_simd_mov_to_<mode>high): Likewise.
9313 (aarch64_simd_mov_from_<mode>low): Likewise.
9314 (aarch64_simd_mov_from_<mode>high): Likewise.
9315 (aarch64_dup_lane<mode>): Update.
9316 (aarch64_dup_lanedi): New instruction pattern.
9317 * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
9318 * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
9320 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9322 * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
9323 (replace_mult_candidate): Remove unnecessary argument; remove
9324 unnecessary parameter from call to introduce_cast_before_cand.
9325 (replace_unconditional_candidate): Remove unnecessary parameter
9326 from call to replace_mult_candidate.
9327 (replace_conditional_candidate): Likewise.
9328 (insert_initializers): Use make_temp_ssa_name.
9329 (introduce_cast_before_cand): Remove unnecessary argument; use
9331 (replace_one_candidate): Remove unnecessary argument; remove
9332 unnecessary parameter from calls to introduce_cast_before_cand.
9333 (replace_profitable_candidates): Remove unnecessary parameters
9334 from calls to replace_one_candidate.
9336 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9338 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9339 phi def as possibly hiding a basis for a CAND_ADD whose operands
9340 have been commuted in the analysis.
9341 (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
9343 2013-05-07 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
9345 * config/aarch64/aarch64.md
9346 (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
9347 shift value between 0-4.
9349 2013-05-07 Richard Biener <rguenther@suse.de>
9351 * double-int.h (rshift): New overload.
9352 * double-int.c (rshift): New function.
9353 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
9354 (create_reference_ops_from_ref): Remove.
9355 (vn_reference_insert): Use shared ops for constructing the
9356 reference and copy it.
9358 2013-05-07 Richard Biener <rguenther@suse.de>
9361 * tree-eh.c (sink_clobbers): Properly propagate
9362 SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
9364 2013-05-07 Jakub Jelinek <jakub@redhat.com>
9366 PR tree-optimization/57149
9367 * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
9368 (can_skip_redundant_opnd, compute_uninit_opnds_pos,
9369 collect_phi_def_edges, execute_late_warn_uninitialized): Use
9370 uninit_undefined_value_p instead of ssa_undefined_value_p.
9373 * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
9374 for modifier == EXPAND_INITIALIZER.
9376 2013-05-07 Anton Blanchard <anton@samba.org>
9378 * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
9379 for powerpc64 little endian.
9380 * configure: Regenerate.
9382 2013-05-06 Graham Stott <grahams@btinternet.com>
9384 * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
9385 mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
9386 wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
9387 shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
9389 2013-05-06 Graham Stott <grahams@btinternet.com>
9391 * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
9392 codes which allow non-lvalues.
9394 2013-05-06 Marc Glisse <marc.glisse@inria.fr>
9396 * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
9397 components are all 1s.
9398 (integer_minus_onep): New function.
9399 * tree.h (integer_minus_onep): Declare it.
9400 * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
9401 integer_minus_onep instead of integer_all_onesp.
9403 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
9406 * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
9407 variations of these patterns.
9409 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
9411 * config/i386/i386.md (isa): Add x64_sse4 member.
9412 (enabled): Handle x64_sse4.
9413 (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
9414 instruction for 64bit SSE4_1 targets. Update insn attributes.
9415 (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
9416 instruction for SSE4_1 targets. Update insn attributes.
9417 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9418 with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
9419 (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
9421 (*vec_extractv4si): Rename from *sse4_1_pextrd.
9422 (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
9423 (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
9425 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
9428 * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
9430 2013-05-06 Maxim Kuznetsov <maks.kuznetsov@gmail.com>
9432 * final.c (do_assembler_dialects): Don't handle curly braces and
9433 vertical bar escaped by % as dialect delimiters.
9434 (output_asm_insn): Print curly braces and vertical bar if escaped
9435 by % and ASSEMBLER_DIALECT defined.
9436 * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
9437 * doc/tm.texi: Regenerated.
9439 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
9441 * config/mips/mips.c: Include tree-pass.h.
9442 (mips_reorg): Split in pre- and post-dbr_schedule parts.
9443 (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
9444 (pass_mips_machine_reorg2): New machine specific pass.
9445 (insert_pass_mips_machine_reorg2): New pass plugin definition.
9446 (mips_option_override): Register the new pass.
9447 * rtl.h (cleanup_barriers): Remove prototype.
9448 (dbr_schedule): Likewise.
9449 * jump.c (cleanup_barriers): Make static.
9450 * reorg.c (dbr_schedule): Likewise.
9452 2013-05-06 Richard Biener <rguenther@suse.de>
9454 PR tree-optimization/57185
9455 * tree-parloops.c (add_field_for_reduction): Handle anonymous
9458 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
9461 * config/i386/i386.c (add_parameter_dependencies): Add dependence
9462 between "first_arg" and "insn", not "last" and "insn".
9464 2013-05-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9466 * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
9467 (find_candidates_in_block): Re-enable slsr_process_phi.
9468 (create_phi_basis): Fix double counting of candidate adjustment.
9470 2013-05-06 Richard Biener <rguenther@suse.de>
9473 * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
9474 the edge is also fallthru, preserve it and just clear the
9476 * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
9477 also complex, preserve that and just clear the fallthru flag.
9478 * tree-inline.c (update_ssa_across_abnormal_edges): Also
9479 update virtual operands.
9481 2013-05-06 Alan Modra <amodra@gmail.com>
9483 * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
9484 (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
9485 * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
9486 * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
9487 (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9488 * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
9490 2013-05-06 Alan Modra <amodra@gmail.com>
9492 * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
9493 (ASM_SPEC): ..here. Emit DEFAULT_ASM_ENDIAN too.
9494 (DEFAULT_ASM_ENDIAN): Define.
9495 (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9496 * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
9497 Update -K PIC clause from sysv4.h.
9498 (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
9499 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
9501 2013-05-06 Alan Modra <amodra@gmail.com>
9503 * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
9504 twice for little-endian.
9505 (ashrdi3_no_power, ashrdi3): Support little-endian.
9507 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
9510 * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
9511 * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
9513 * config/sh/iterators.md (SMIN_SMAX): New code iterator.
9514 * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
9515 clips_min_const_int, clips_max_const_int, clipu_max_const_int):
9518 2013-05-05 Steven Bosscher <steven@gcc.gnu.org>
9519 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
9521 * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
9522 * config/pa/pa.opt: Make mbig-switch a no-op.
9523 * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
9524 (CASE_VECTOR_MODE): Always return SImode.
9525 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
9526 for the !TARGET_BIG_SWITCH case.
9527 * config/pa/pa-linux.h: Likewise.
9528 * config/pa/pa-openbsd.h: Likewise.
9529 * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
9530 * config/pa/pa.md (short_jump): Remove define_insn.
9531 (casesi): Remove code for the !TARGET_BIG_SWITCH case.
9532 (casesi0): Remove define_insn.
9533 (type): Remove btable_branch.
9534 (pa_combine_type): Likewise.
9535 (in_nullified_branch_delay): Likewise.
9536 (in_call_delay): Likewise.
9537 (define_delay): Likewise.
9538 (define_insn_reservation "Z3"): Likewise.
9539 (define_insn_reservation "Z4"): Likewise.
9540 * config/pa/pa.c (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
9541 (pa_adjust_insn_length): Remove adjustment for btable branches.
9542 * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
9545 2013-05-05 Uros Bizjak <ubizjak@gmail.com>
9547 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9548 from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
9549 Add m->r,x alternatives.
9550 (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
9551 splitters using SWI48x mode iterator.
9552 (*vec_extract_v2di_0_sse): Rename from *sse2_storeq. Disable for
9553 TARGET_64BIT. Add m->x alternative.
9554 (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
9555 Add o->x alternative. Enable for TARGET_SSE.
9556 (sse_storeq): Remove expander.
9557 (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
9558 with memory input operand.
9559 (*vec_extractv2di_1 splitter): New.
9560 (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
9561 * config/i386/i386.md (ssevecmodelower): New mode attribute.
9563 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
9565 * config/rs6000/rs6000.c (INT_P): Reformat. Delete obsolete comment.
9566 (INT_LOWPART): Delete.
9567 (extract_MB): Adjust.
9568 (extract_ME): Adjust.
9569 (print_operand): Adjust.
9571 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
9573 * config/rs6000/predicates.md (reg_or_add_cint_operand,
9574 reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9575 (reg_or_logical_cint_operand, easy_fp_constant,
9576 logical_const_operand): Delete "CONST_DOUBLE" case.
9577 * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
9578 "HOST_BITS_PER_WIDE_INT == 64" test.
9579 (num_insns_constant): Ditto. Delete CONST_DOUBLE DImode/VOIDmode case.
9580 (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
9581 (rs6000_emit_set_const): Delete CONST_DOUBLE case.
9582 (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
9584 (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
9585 CONST_DOUBLE DImode/VOIDmode case.
9586 (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
9587 (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case. Delete
9588 CONST_DOUBLE VOIDmode case.
9589 (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
9590 (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
9591 * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
9592 Delete CONST_DOUBLE case.
9593 (splitters for mov FMOVE64 const_double): Delete
9594 "HOST_BITS_PER_WIDE_INT == 32" case. Delete
9595 "HOST_BITS_PER_WIDE_INT >= 64" test.
9596 (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
9598 (mov DI const_double): Delete.
9600 2013-05-04 Jakub Jelinek <jakub@redhat.com>
9602 * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
9603 on op shows all bits zero in mode of a lowpart subreg, return zero.
9605 2013-05-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9608 * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
9609 to save TFmode registers and DImode to save TImode registers for
9610 caller save operations.
9611 (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
9612 mark being partially clobbered since they only use the first
9615 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
9616 and TDmode only use the upper 64-bits of each VSX register.
9618 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9620 * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
9621 (find_candidates_in_block): Disable slsr_process_phi.
9623 2013-05-03 Guozhi Wei <carrot@google.com>
9625 * coverage.c (coverage_obj_init): Move the construction of gcov
9627 (build_init_ctor): ... here.
9629 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9631 * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
9632 (slsr_cand_d): Redefine def_phi.
9633 (stride_status, phi_adjust_status, count_phis_status): New enums.
9634 (find_phi_def): New.
9635 (find_basis_for_base_expr): New.
9636 (find_basis_for_candidate): Handle hidden bases.
9637 (alloc_cand_and_find_basis): Handle phi candidates.
9638 (slsr_process_phi): New.
9639 (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
9640 (create_mul_imm_cand): Likewise.
9641 (create_add_ssa_cand): Exclude phi base candidates.
9642 (create_add_imm_cand): Likewise.
9643 (slsr_process_cast): Likewise.
9644 (slsr_process_copy): Likewise.
9645 (find_candidates_in_block): Handle phi candidates.
9646 (dump_candidate): Likewise.
9647 (unconditional_cands): Delete.
9648 (unconditional_cands_with_known_stride_p): Delete.
9649 (phi_dependent_cand_p): New.
9650 (cand_increment): Handle phi-dependent candidates.
9651 (replace_dependent): Delete.
9652 (replace_mult_candidate): New.
9653 (replace_unconditional_candidate): New.
9654 (incr_vec_index): Move to avoid forward reference.
9655 (create_add_on_incoming_edge): New.
9656 (create_phi_basis): New.
9657 (replace_dependents): Delete.
9658 (replace_conditional_candidate): New.
9659 (phi_add_costs): New.
9660 (replace_uncond_cands_and_profitable_phis): New.
9661 (record_increment): Handle phi adjustments.
9662 (record_phi_increments): New.
9663 (record_increments): Handle phi adjustments.
9664 (phi_incr_cost): New.
9665 (lowest_cost_path): Handle phis.
9666 (total_savings): Likewise.
9667 (analyze_increments): Likewise.
9668 (ncd_with_phi): New.
9669 (ncd_of_cand_and_phis): New.
9670 (nearest_common_dominator_for_cands): Handle phi increments.
9671 (all_phi_incrs_profitable): New.
9672 (replace_profitable_candidates): Handle phi-dependent candidates.
9673 (analyze_candidates_and_replace): Likewise.
9675 2013-05-03 Teresa Johnson <tejohnson@google.com>
9678 * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
9679 do not exceed REG_BR_PROB_BASE.
9681 2013-05-03 Jeff Law <law@redhat.com>
9683 PR tree-optimization/57144
9684 * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
9685 operand of the condition will bit into the new type when eliminating
9686 a cast feeding a condition.
9688 2013-05-03 Jakub Jelinek <jakub@redhat.com>
9690 PR rtl-optimization/57130
9691 * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
9692 of COMPARE as in_code to the recursive call if needed.
9694 2013-05-03 Uros Bizjak <ubizjak@gmail.com>
9696 * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
9697 (enabled): Handle new members.
9698 * config/i386/sse.md (*vec_concatv2si): Merge from
9699 *vec_concatv2si_sse2 and vec_concatv2si_sse.
9700 (vec_concatv2di): Merge with *vec_concatv2di_rex64.
9702 2013-05-03 Joern Rennecke <joern.rennecke@embecosm.com>
9704 PR tree-optimization/57027
9705 * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
9706 for fnms opportunity, check we got the prerequisite kind
9707 of tree / gimple before using accessor functions.
9709 2013-05-03 Richard Biener <rguenther@suse.de>
9711 * double-int.h (lshift): New overload without precision
9713 (operator *=, operator +=, operator -=): Move ...
9714 * double-int.c (operator *=, operator +=, operator -=): ... here
9715 and implement more efficiently.
9716 (mul_double_with_sign): Remove.
9717 (lshift_double): Adjust to take unsinged shift argument, push
9718 dispatching code to callers.
9719 (mul_double_wide_with_sign): Add early out for callers that
9720 are not interested in high parts or overflow.
9721 (lshift): New function.
9722 (lshift, rshift, alshift, arshift, llshift, lrshift): Add
9724 (lrotate, rrotate): Use logical shifts.
9725 * expr.c (get_inner_reference): Use lshift.
9726 * fixed-value.c (do_fixed_divide): Likewise.
9727 * tree-dfa.c (get_ref_base_and_extent): Likewise.
9728 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
9729 (indirect_refs_may_alias_p): Likewise.
9730 (stmt_kills_ref_p_1): Likewise.
9732 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
9734 * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
9736 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
9738 * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
9739 scalar form of FABD instruction.
9741 2013-05-02 Vladimir Makarov <vmakarov@redhat.com>
9743 * lra-constraints.c (process_alt_operands): Add checking alt
9744 number to choose the best alternative.
9746 2013-05-02 Richard Biener <rguenther@suse.de>
9748 * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
9749 bitmap and its handling.
9750 (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
9752 2013-05-02 Richard Biener <rguenther@suse.de>
9755 * tree-inline.c (copy_loops): Properly handle removed loops.
9756 (copy_cfg_body): Mark destination loops for fixup if source
9759 2013-05-02 Greta Yorsh <Greta.Yorsh@arm.com>
9762 * config/arm/arm.c (arm_expand_epilogue): Check really_return before
9763 generating simple_return for naked functions.
9765 2013-05-02 Martin Jambor <mjambor@suse.cz>
9768 * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
9769 * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
9771 (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
9772 ipa_agg_replacement_value structures.
9773 (known_aggs_to_agg_replacement_list): Likewise.
9774 * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
9775 (read_agg_replacement_chain): Likewise.
9776 (ipcp_transform_function): Also check that by_ref flags match.
9778 2013-05-02 Richard Biener <rguenther@suse.de>
9780 * graphds.h (struct graph): Add obstack member.
9781 * graphds.c (new_graph): Initialize obstack and allocate
9783 (add_edge): Allocate edge from the obstack.
9784 (free_graph): Free the obstack instead of all edges and vertices.
9786 2013-05-02 Teresa Johnson <tejohnson@google.com>
9788 * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
9790 * cfg.c (update_bb_profile_for_threading): Ditto.
9791 * tree-inline.c (copy_bb): Ditto.
9792 (copy_edges_for_bb): Ditto.
9793 (initialize_cfun): Ditto.
9794 (copy_cfg_body): Ditto.
9795 (expand_call_inline): Ditto.
9796 * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
9797 (estimate_node_size_and_time): Ditto.
9798 (inline_merge_summary): Ditto.
9799 * cgraphclones.c (cgraph_clone_edge): Ditto.
9800 (cgraph_clone_node): Ditto.
9801 * sched-rgn.c (compute_dom_prob_ps): Ditto.
9802 (compute_trg_info): Ditto.
9804 2013-05-02 Ian Bolton <ian.bolton@arm.com>
9806 * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
9807 S reg when fp attribute set.
9808 (movdi_aarch64): Only allow to/from D reg when fp attribute set.
9810 2013-05-02 Ian Bolton <ian.bolton@arm.com>
9812 * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
9814 (*and_one_cmplsi3_compare0_uxtw): Likewise.
9815 (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
9816 (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
9818 2013-05-02 Richard Biener <rguenther@suse.de>
9820 * tree-scalar-evolution.c (scev_info_hasher): Remove.
9821 (struct instantiate_cache_entry): New type.
9822 (struct instantiate_cache_entry_hasher): New hashtable descriptor.
9823 (struct instantiate_cache_type): New type.
9824 (set_instantiated_value, get_instantiated_value): Remove.
9825 (get_instantiated_value_entry): New function.
9826 (instantiate_scev_name): Use the new cache and adjust.
9827 (instantiate_scev_poly): Adjust.
9828 (instantiate_scev_binary): Likewise.
9829 (instantiate_array_ref): Likewise.
9830 (instantiate_scev_convert): Likewise.
9831 (instantiate_scev_not): Likewise.
9832 (instantiate_scev_3): Likewise.
9833 (instantiate_scev_2): Likewise.
9834 (instantiate_scev_r): Likewise.
9835 (instantiate_scev): Likewise.
9836 (resolve_mixers): Likewise.
9838 2013-05-01 Vladimir Makarov <vmakarov@redhat.com>
9841 * lra-constraints.c (best_small_class_operands_num): Remove.
9842 (process_alt_operands): Remove small_class_operands_num. Take
9843 small classes operands into losers and only if the operand is not
9844 matched. Modify debugging output.
9845 (curr_insn_transform): Remove best_small_class_operands_num.
9848 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9850 * config/aarch64/aarch64-builtins.c
9851 (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
9852 * config/aarch64/aarch64-simd-builtins.def
9853 (reduc_splus_): Add new modes.
9854 (reduc_uplus_): New.
9855 * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
9856 (reduc_uplus_v4sf): Likewise.
9857 (reduc_splus_v4sf): Likewise.
9858 (aarch64_addv<mode>): Likewise.
9859 (reduc_uplus_<mode>): Likewise.
9860 (reduc_splus_<mode>): Likewise.
9861 (aarch64_addvv2di): Likewise.
9862 (reduc_uplus_v2di): Likewise.
9863 (reduc_splus_v2di): Likewise.
9864 (aarch64_addvv2si): Likewise.
9865 (reduc_uplus_v2si): Likewise.
9866 (reduc_splus_v2si): Likewise.
9867 (reduc_<sur>plus_<mode>): New.
9868 (reduc_<sur>plus_v2di): Likewise.
9869 (reduc_<sur>plus_v2si): Likewise.
9870 (reduc_<sur>plus_v4sf): Likewise.
9871 (aarch64_addpv4sf): Likewise.
9872 * config/aarch64/arm_neon.h
9873 (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
9874 * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
9875 add UNSPEC_SADDV, UNSPEC_UADDV.
9877 (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
9879 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9881 * config/aarch64/arm_neon.h
9882 (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
9884 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9886 * config/aarch64/aarch64-builtins
9887 (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
9889 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9891 * config/aarch64/aarch64-simd-builtins.def
9893 (reduc_smin_): Likewise.
9894 (reduc_umax_): Likewise.
9895 (reduc_umin_): Likewise.
9896 (reduc_smax_nan_): Likewise.
9897 (reduc_smin_nan_): Likewise.
9900 (smax): Update for V2SF, V4SF and V2DF modes.
9903 (smin_nan): Likewise.
9904 * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
9905 (<su><maxmin><mode>3): ...This, refactor.
9906 (s<maxmin><mode>3): New.
9907 (<maxmin_uns><mode>3): Likewise.
9908 (reduc_<maxmin_uns>_<mode>): Refactor.
9909 (reduc_<maxmin_uns>_v4sf): Likewise.
9910 (reduc_<maxmin_uns>_v2si): Likewise.
9911 (aarch64_<fmaxmin><mode>: Remove.
9912 * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
9914 (vmin<q>_f<32,64>): Likewise.
9915 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9917 (su): Add mappings for smax, smin, umax, umin.
9919 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
9920 (FMAXMIN): Rename as...
9921 (FMAXMIN_UNS): ...This.
9923 (fmaxminv): Likewise.
9924 (fmaxmin): Likewise.
9926 (maxmin_uns_op): Likewise.
9928 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9930 * config/aarch64/arm_neon.h
9931 (vac<ge, gt><sd>_f<32, 64>): Rename to...
9932 (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
9933 (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
9935 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9937 * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
9938 * config/aarch64/iterators.md (FAC_COMPARISONS): New.
9940 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9942 * config/aarch64/aarch64-simd.md
9943 (vcond<mode>_internal): Handle special cases for constant masks.
9944 (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
9945 (vcondu<mode><mode>): Likewise.
9946 (vcond<v_cmp_result><mode>): New.
9948 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9950 * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
9951 (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
9952 * config/aarch64/aarch64-simd-builtins.def
9953 (cmeq): Update to BUILTIN_VALLDI.
9958 * config/aarch64/arm_neon.h
9959 (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
9960 to builtins or C as appropriate.
9962 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
9964 * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
9966 (cmhi): Rename to...
9968 * config/aarch64/aarch64-simd.md
9969 (simd_mode): Add SF.
9970 (aarch64_vcond_internal): Use new names for unsigned comparison insns.
9971 (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
9972 * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
9973 (cstore<mode>_neg): ...This.
9974 * config/aarch64/iterators.md
9976 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
9978 (UCOMPARISONS): Likewise.
9979 (optab): Add missing comparisons.
9987 (V_cmp_result): Add DF, SF modes.
9988 (v_cmp_result): Likewise.
9991 * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
9993 2013-05-01 Greta Yorsh <Greta.Yorsh@arm.com>
9995 * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
9996 define_insn to define_insn_and_split.
9997 (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
9998 (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
9999 (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10000 (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10001 (thumb2_negscc): Likewise.
10003 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
10005 * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10007 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
10009 * config/arm/thumb2.md: Remove trailing whitespaces.
10011 2013-04-30 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
10013 * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
10014 Use gen_int_mode rather than GEN_INT.
10016 2013-04-30 H.J. Lu <hongjiu.lu@intel.com>
10018 * value-prof.c (stream_in_histogram_value): Remove the strayed
10021 2013-04-30 Richard Biener <rguenther@suse.de>
10023 PR middle-end/57122
10024 * cfghooks.c (split_edge): Properly check for the loop latch edge.
10026 2013-04-30 Richard Biener <rguenther@suse.de>
10028 PR middle-end/57107
10029 * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
10031 2013-04-30 Andrey Belevantsev <abel@ispras.ru>
10033 PR rtl-optimization/56957
10034 PR rtl-optimization/57105
10035 * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
10036 variable. Use just INSN_UID for determining whether an insn
10037 should be only disconnected from the insn stream.
10038 * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
10040 2013-04-30 Jakub Jelinek <jakub@redhat.com>
10042 PR tree-optimization/57104
10043 * tsan.c (instrument_expr): Don't instrument accesses to
10044 DECL_HARD_REGISTER VAR_DECLs.
10046 2013-04-30 Richard Biener <rguenther@suse.de>
10048 * function.h (loops_for_fn): New inline function.
10049 (set_loops_for_fn): Likewise.
10050 * cfgloop.h (place_new_loop): Add struct function parameter.
10051 (get_loop): Likewise.
10052 (get_loops): Likewise.
10053 (number_of_loops): Likewise.
10054 (fel_next): Adjust.
10055 (fel_init): Likewise.
10056 * cfg.c (get_loop_copy): Adjust.
10057 * cfgloop.c (flow_loops_dump): Likewise.
10058 (record_loop_exits): Likewise.
10059 (verify_loop_structure): Likewise.
10060 * cfgloopanal.c (mark_irreducible_loops): Likewise.
10061 (estimate_reg_pressure_cost): Likewise.
10062 (mark_loop_exit_edges): Likewise.
10063 * cfgloopmanip.c (place_new_loop): Likewise.
10064 (add_loop): Likewise.
10065 (duplicate_loop): Likewise.
10066 * graph.c (draw_cfg_nodes): Likewise.
10067 * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
10068 * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
10069 (extract_affine_chrec): Likewise.
10070 (build_scop_iteration_domain): Likewise.
10071 * graphite.c (graphite_initialize): Likewise.
10072 * ira-build.c (create_loop_tree_nodes): Likewise.
10073 (more_one_region_p): Likewise.
10074 (rebuild_regno_allocno_maps): Likewise.
10075 (mark_loops_for_removal): Likewise.
10076 (mark_all_loops_for_removal): Likewise.
10077 (remove_unnecessary_regions): Likewise.
10078 (ira_build): Likewise.
10079 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
10080 * loop-init.c (fix_loop_structure): Likewise.
10081 (gate_rtl_move_loop_invariants): Likewise.
10082 (gate_rtl_unswitch): Likewise.
10083 (gate_rtl_unroll_and_peel_loops): Likewise.
10084 (rtl_doloop): Likewise.
10085 * lto-streamer-in.c (input_cfg): Likewise.
10086 * lto-streamer-out.c (output_cfg): Likewise.
10087 * modulo-sched.c (sms_schedule): Likewise.
10088 * predict.c (tree_estimate_probability): Likewise.
10089 (tree_estimate_probability_driver): Likewise.
10090 (estimate_loops): Likewise.
10091 * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
10092 (move_sese_region_to_fn): Likewise.
10093 (debug_loop_num): Likewise.
10094 * tree-chrec.c (chrec_evaluate): Likewise.
10095 (hide_evolution_in_other_loops_than_loop): Likewise.
10096 (chrec_component_in_loop_num): Likewise.
10097 (reset_evolution_in_loop): Likewise.
10098 (evolution_function_is_invariant_rec_p): Likewise.
10099 * tree-if-conv.c (main_tree_if_conversion): Likewise.
10100 * tree-inline.c (copy_loops): Likewise.
10101 (copy_cfg_body): Likewise.
10102 (tree_function_versioning): Likewise.
10103 * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
10104 * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
10106 (add_to_evolution_1): Likewise.
10107 (scev_const_prop): Likewise.
10108 * tree-scalar-evolution.h (get_chrec_loop): Likewise.
10109 * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
10110 * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
10111 (tree_ssa_lim_initialize): Likewise.
10112 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
10113 (verify_loop_closed_ssa): Likewise.
10114 * tree-ssa-loop.c (tree_ssa_loop_init): Likewise.
10115 (tree_ssa_loop_im): Likewise.
10116 (tree_ssa_loop_unswitch): Likewise.
10117 (tree_vectorize): Likewise.
10118 (check_data_deps): Likewise.
10119 (tree_ssa_loop_ivcanon): Likewise.
10120 (tree_ssa_loop_bounds): Likewise.
10121 (tree_complete_unroll): Likewise.
10122 (tree_complete_unroll_inner): Likewise.
10123 (tree_parallelize_loops): Likewise.
10124 (tree_ssa_loop_prefetch): Likewise.
10125 (tree_ssa_loop_ivopts): Likewise.
10126 * tree-ssa.c (execute_update_addresses_taken): Liekwise.
10127 * tree-vectorizer.c (vectorize_loops): Likewise.
10129 2013-04-29 Mike Frysinger <vapier@gentoo.org>
10131 * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
10132 (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
10133 * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
10134 with EABI_LINK_SPEC.
10136 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
10139 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
10142 2013-04-29 Vladimir Makarov <vmakarov@redhat.com>
10145 * lra-constraints.c (process_alt_operands): Discourage a bit more
10146 using memory for pseudos. Print cost dump for alternatives.
10147 Modify cost values for conflicts with early clobbers.
10148 (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
10150 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
10153 * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
10155 2013-04-29 Ian Bolton <ian.bolton@arm.com>
10157 * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
10158 from/to S register.
10159 (movdi_aarch64): Support LDR/STR from/to D register.
10161 2013-04-29 Ian Bolton <ian.bolton@arm.com>
10163 * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
10164 or higher by default.
10166 2013-04-29 Richard Biener <rguenther@suse.de>
10168 PR middle-end/57075
10169 * tree-inline.c (copy_edges_for_bb): Still split the bbs,
10170 even if not adding abnormal edges for calls that can make
10173 2013-04-29 Richard Biener <rguenther@suse.de>
10175 PR middle-end/57103
10176 * tree-cfg.c (move_stmt_op): Fix condition under which to update
10178 (move_stmt_r): Remove redundant checking.
10180 2013-04-29 Teresa Johnson <tejohnson@google.com>
10183 * basic-block.h (apply_scale): New function.
10184 (apply_probability): Use apply_scale.
10185 * gimple-streamer-in.c (input_bb): Ditto.
10186 * lto-streamer-in.c (input_cfg): Ditto.
10187 * lto-cgraph.c (merge_profile_summaries): Ditto.
10188 * tree-optimize.c (execute_fixup_cfg): Ditto.
10189 * tree-inline.c (copy_bb): Update comment to use apply_scale.
10190 (copy_edges_for_bb): Ditto.
10191 (copy_cfg_body): Ditto.
10193 2013-04-29 Tom de Vries <tom@codesourcery.com>
10195 * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
10196 (replace_block_by): Don't set LOOPS_NEED_FIXUP.
10197 (tail_merge_optimize): Handle current_loops == NULL.
10199 2013-04-26 Jeff Law <law@redhat.com>
10201 * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
10202 (simplify_cond_using_ranges): Generalize code to simplify
10203 COND_EXPRs where one argument is a constant and the other
10204 is an SSA_NAME created by an integral type conversion.
10206 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10208 * config/arm/arm.md (store_minmaxsi): Use only when
10209 optimize_insn_for_size_p.
10211 2013-04-29 Christian Bruel <christian.bruel@st.com>
10214 * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
10216 2013-04-29 Richard Biener <rguenther@suse.de>
10218 PR middle-end/57089
10219 * omp-low.c (expand_omp_taskreg): If the parent function had a broken
10220 loop tree make sure to schedule a fixup for the child as well.
10221 (expand_omp_for_generic): Properly add loops.
10222 (expand_omp_for_static_nochunk): Likewise.
10223 (expand_omp_for_static_chunk): Likewise.
10224 (expand_omp_for): For the degenerate case fixup loops.
10225 (expand_omp_sections): Fix default bb placement in loops.
10226 (expand_omp_atomic_pipeline): Properly add loops.
10228 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10230 * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
10232 2013-04-29 Tom de Vries <tom@codesourcery.com>
10234 * tree-ssa-tail-merge.c: Update header comment.
10236 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10238 * config/aarch64/arm_neon.h
10239 (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
10240 (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
10241 (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
10242 (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
10243 (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
10244 (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
10245 (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
10246 (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
10248 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10250 * config/aarch64/aarch64-simd.md
10251 (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
10252 (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
10253 fix_trunc, fixuns_trunc.
10254 (ftrunc<VDQF:mode>2): New.
10255 * config/aarch64/iterators.md (optab): Add fix, fixuns.
10256 (fix_trunc_optab): New.
10258 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10260 * config/aarch64/aarch64-builtins.c
10261 (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
10262 iceilf, lround, iroundf.
10264 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
10267 * config/i386/i386.h (enum ix86_tune_indices)
10268 <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
10269 New, split from X86_TUNE_INTER_UNIT_MOVES.
10270 <X86_TUNE_INTER_UNIT_MOVES>: Remove.
10271 (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
10272 (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
10273 (TARGET_INTER_UNIT_MOVES): Remove.
10274 * config/i386/i386.c (initial_ix86_tune_features): Update.
10275 Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
10276 (ix86_expand_convert_uns_didf_sse): Use
10277 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10278 (ix86_expand_vector_init_one_nonzero): Ditto.
10279 (ix86_expand_vector_init_interleave): Ditto.
10280 (inline_secondary_memory_needed): Return true for moves from SSE class
10281 registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
10282 to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
10283 * config/i386/constraints.md (Yi, Ym): Depend on
10284 TARGET_INTER_UNIT_MOVES_TO_VEC.
10285 (Yj, Yn): New constraints.
10286 * config/i386/i386.md (*movdi_internal): Change constraints of
10287 operand 1 from Yi to Yj and from Ym to Yn.
10288 (*movsi_internal): Ditto.
10289 (*movdf_internal): Ditto.
10290 (*movsf_internal): Ditto.
10291 (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
10292 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10293 (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
10294 (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
10295 (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
10296 * config/i386/sse.md (movdi_to_sse): Ditto.
10297 (sse2_stored): Change constraint of operand 1 from Yi to Yj.
10298 Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
10299 TARGET_INTER_UNIT_MOVES.
10300 (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
10301 (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
10302 instead of TARGET_INTER_UNIT_MOVES.
10303 * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
10304 operand 1 from Yi to Yj and from Ym to Yn.
10306 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10308 * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
10309 (float_truncate_hi_): Likewise.
10310 (float_extend_lo_): Likewise.
10311 (float_truncate_lo_): Likewise.
10312 * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
10313 (aarch64_float_extend_lo_v2df): Likewise.
10314 (vec_unpacks_hi_v4sf): Likewise.
10315 (aarch64_float_truncate_lo_v2sf): Likewise.
10316 (aarch64_float_truncate_hi_v4sf): Likewise.
10317 (vec_pack_trunc_v2df): Likewise.
10318 (vec_pack_trunc_df): Likewise.
10320 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10322 * config/aarch64/aarch64-builtins.c
10323 (aarch64_fold_builtin): Fold float conversions.
10324 * config/aarch64/aarch64-simd-builtins.def
10325 (floatv2si, floatv4si, floatv2di): New.
10326 (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
10327 * config/aarch64/aarch64-simd.md
10328 (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
10329 * config/aarch64/iterators.md (FLOATUORS): New.
10330 (optab): Add float, floatuns.
10331 (su_optab): Likewise.
10333 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10335 * config/aarch64/aarch64-builtins.c
10336 (aarch64_builtin_vectorized_function): Use new names for
10338 * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
10339 (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
10340 (fcvtzu): Split as...
10341 (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
10342 (fcvtas): Split as...
10343 (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
10344 (fcvtau): Split as...
10345 (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
10346 (fcvtps): Split as...
10347 (lceilv2sf, lceilv4sf, lceilv2df): ...This.
10348 (fcvtpu): Split as...
10349 (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
10350 (fcvtms): Split as...
10351 (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
10352 (fcvtmu): Split as...
10353 (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
10354 (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
10355 (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
10356 (lfrintnusf, lfrintnudf): Likewise.
10357 * config/aarch64/aarch64-simd.md
10358 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
10360 (aarch64_fcvt<frint_suffix><su><mode>): Remove.
10361 * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
10362 (fcvt_pattern): Likewise.
10364 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10366 * config/aarch64/aarch64-simd.md
10367 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
10368 (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
10370 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10372 * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
10373 (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
10374 (vrnd<a,m,n,p>_f32): Implement using builtins.
10375 (vrnd<i,x><q>_f<32, 64>): New.
10377 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10379 * config/aarch64/aarch64-builtins.c
10380 (aarch64_builtin_vectorized_function): Fold to standard pattern names.
10381 * config/aarch64/aarch64-simd-builtins.def (frintn): New.
10382 (frintz): Rename to...
10384 (frintp): Rename to...
10386 (frintm): Rename to...
10388 (frinti): Rename to...
10389 (nearbyint): ...this.
10390 (frintx): Rename to...
10392 (frinta): Rename to...
10394 * config/aarch64/aarch64-simd.md
10395 (aarch64_frint<frint_suffix><mode>): Delete.
10396 (<frint_pattern><mode>2): Convert to insn.
10397 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
10398 * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
10399 (frint_pattern): Likewise.
10400 (frint_suffix): Likewise.
10402 2013-04-29 Richard Biener <rguenther@suse.de>
10404 PR tree-optimization/57081
10405 * loop-init.c: Include tree-flow.h.
10406 (loop_optimizer_finalize): Free number of iteration estimates.
10407 * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
10409 2013-04-29 Jakub Jelinek <jakub@redhat.com>
10411 PR tree-optimization/57083
10412 * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
10413 non-singleton shift count range, zero extend low_bound for uns case.
10415 * config/i386/predicates.md (general_vector_operand): New predicate.
10416 * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
10417 (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
10418 if they aren't nonimmediate operands. If their original values
10419 satisfy const_vector_equal_evenodd_p, don't shift them.
10420 * config/i386/sse.md (mul<mode>3): Use general_vector_operand
10421 predicates. For the SSE4.1 case force operands[{1,2}] into registers
10422 if not nonimmediate_operand.
10423 (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
10424 instead of register_operand.
10425 (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
10427 2013-04-28 Eric Botcazou <ebotcazou@adacore.com>
10429 * stor-layout.c (finalize_size_functions): Allocate a structure and
10430 reset cfun before dumping the functions.
10432 2013-04-27 Jakub Jelinek <jakub@redhat.com>
10434 * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
10437 * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
10438 use xop_pmacsdqh if uns_p.
10439 * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
10440 the immediate rotate count.
10442 2013-04-26 Vladimir Makarov <vmakarov@redhat.com>
10444 * rtl.h (struct rtx_def): Add comment for field jump.
10445 (LRA_SUBREG_P): New macro.
10446 * recog.c (register_operand): Check LRA_SUBREG_P.
10447 * lra.c (lra): Add note at the end of RTL code. Align non-empty
10449 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10450 (lra_final_code_change): Skip subreg change for operators.
10451 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10452 if there are no operand changes.
10453 * lra-constraints.c (curr_insn_set): New.
10454 (match_reload): Set LRA_SUBREG_P.
10455 (emit_spill_move): Ditto.
10456 (check_and_process_move): Use curr_insn_set. Process only single
10457 set insns. Don't initialize sec_mem_p and change_p.
10458 (simplify_operand_subreg): Use LRA_SUBREG_P.
10459 (reg_in_class_p): New function.
10460 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
10461 of #ifdef. Add code to remove cycling.
10462 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10463 non-null disp. Reload inner instead of disp when base and index
10464 are null. Try to put lo_sum into register.
10465 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10466 (check_and_process_move): Move code for move cost check to
10467 simple_move_p. Remove equiv_substitution.
10468 (simple_move_p): New function.
10469 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
10470 curr_insn_set. Call check_and_process_move only for single set
10471 insns. Use the new function. Move call of check_and_process_move
10472 after operand equiv substitution and address process.
10474 2013-04-26 Jakub Jelinek <jakub@redhat.com>
10477 * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
10478 with nonlocal goto receivers or returns twice calls, ignore
10479 unininitialized values from abnormal edges to nl goto receiver
10480 or returns twice call.
10482 2013-04-26 Jakub Jelinek <jakub@redhat.com>
10484 PR tree-optimization/57051
10485 * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
10486 and VEC_RSHIFT_EXPR if shift count is a multiple of element
10489 2013-04-26 Richard Biener <rguenther@suse.de>
10491 * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
10492 (expand_omp_taskreg): Likewise. Mark loops for fixup.
10493 * tree-cfg.c (move_block_to_fn): Remap loop fathers.
10494 (fixup_loop_arrays_after_move): New function.
10495 (move_sese_region_to_fn): Properly outline the loop tree parts
10496 of the SESE region.
10498 2013-04-26 Uros Bizjak <ubizjak@gmail.com>
10500 * config/i386/i386.md (type, unit): Fix long lines.
10502 2013-04-26 Richard Biener <rguenther@suse.de>
10504 * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
10505 (lto-streamer-out.o): Likewise.
10506 * cfgloop.c (init_loops_structure): Export, add struct function
10507 argument and adjust.
10508 (flow_loops_find): Adjust.
10509 * cfgloop.h (enum loop_estimation): Add EST_LAST.
10510 (init_loops_structure): Declare.
10511 * lto-streamer-in.c: Include cfgloop.h.
10512 (input_cfg): Input the loop tree.
10513 * lto-streamer-out.c: Include cfgloop.h.
10514 (output_cfg): Output the loop tree.
10515 (output_struct_function_base): Do not drop PROP_loops.
10517 2013-03-26 Richard Biener <rguenther@suse.de>
10519 * tree-cfg.c (execute_build_cfg): Build the loop tree.
10520 (pass_build_cfg): Provide PROP_loops.
10521 (move_sese_region_to_fn): Remove loops that are outlined into fn
10523 * tree-inline.c: Include cfgloop.h.
10524 (initialize_cfun): Do not drop PROP_loops.
10525 (copy_loops): New function.
10526 (copy_cfg_body): Copy loop structure.
10527 (tree_function_versioning): Initialize destination loop tree.
10528 * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
10529 (pass_parallelize_loops): Do IL verification.
10530 * loop-init.c (loop_optimizer_init): Fixup loops if required.
10531 * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
10532 the CFG make sure we fixup loops as well.
10533 * tree-ssa-tail-merge.c: Include cfgloop.h.
10534 (replace_block_by): When merging loop latches mark loops for fixup.
10535 * lto-streamer-out.c (output_struct_function_base): Drop
10536 PROP_loops for now.
10537 * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
10538 (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
10539 * ipa-split.c: Include cfgloop.h.
10540 (split_function): Add the new return block to the loop tree root.
10541 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
10542 whether we have removed the forwarder block.
10543 (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
10544 * cfgloop.h (place_new_loop): Declare.
10545 * cfgloopmanip.c (place_new_loop): Export.
10546 * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
10547 (tree-switch-conversion.o): Likewise.
10548 (tree-complex.o): Likewise.
10549 (tree-inline.o): Likewise.
10550 (tree-ssa-tailmerge.o): Likewise.
10551 (ipa-split.o): Likewise.
10552 (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
10553 (tree-ssa-copy.o): Likewise.
10554 * tree-switch-conversion.c: Include cfgloop.h
10555 (process_switch): If we emit a bit-test cascade, schedule loops
10557 * tree-complex.c: Include cfgloop.h.
10558 (expand_complex_div_wide): Properly add new basic-blocks to loops.
10559 * asan.c: Include cfgloop.h.
10560 (create_cond_insert_point): Properly add new basic-blocks to
10561 loops, schedule loop fixup.
10562 * cfgloop.c (verify_loop_structure): Check that looks are not
10564 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
10566 (expand_omp_for_generic): Likewise.
10567 (expand_omp_sections): Likewise.
10568 (expand_omp_atomic_pipeline): Schedule loops for fixup.
10569 * tree-ssa-copy.c: Include tree-scalar-evolution.h.
10570 (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
10571 is initialized, not when loops are present.
10572 * tree-parloops.c (parallelize_loops): Remove checking here.
10573 * passes.c (init_optimization_passes): Schedule a copy-propagation
10574 pass before complete unrolling of inner loops.
10576 2013-04-26 Jakub Jelinek <jakub@redhat.com>
10578 * Makefile.in (toplev.o): Depend on diagnostic-color.h.
10579 * diagnostic-color.c (should_colorize): Remove _WIN32 version.
10580 (colorize_init): Add argument to _WIN32 version.
10581 * toplev.c: Include diagnostic-color.h.
10582 (process_options): Default to -fdiagnostics-color=auto if
10583 GCC_COLORS env var is in the environment.
10584 * common.opt (fdiagnostics-color=): Add Var and Init.
10585 * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
10586 env var is in the environment, the default is auto rather than never.
10588 * diagnostic.h (file_name_as_prefix): Add context argument.
10589 * diagnostic.c (file_name_as_prefix): Likewise. Colorize
10590 the string as locus.
10591 * langhooks.c (lhd_print_error_function): Adjust caller.
10593 2013-04-25 Lawrence Crowl <crowl@google.com>
10595 * var-tracking.c (shared_hash_def::htab):
10596 Change type to hash_table. Update dependent calls and types.
10598 2013-04-25 Lawrence Crowl <crowl@google.com>
10600 * Makefile.in: Update as needed below.
10602 * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
10603 Move declaration to after the type's method definitons.
10605 * attribs.c (htab_t scoped_attributes::attribute_hash):
10606 Change type to hash_table. Update dependent calls and types.
10608 * bitmap.c (htab_t bitmap_desc_hash):
10609 Change type to hash_table. Update dependent calls and types.
10611 * cselib.c (htab_t cselib_hash_table):
10612 Change type to hash_table. Update dependent calls and types.
10614 * data-streamer.h (struct string_slot): Move to lto-streamer.h.
10615 (hash_string_slot_node): Move implementation into lto-streamer.h
10616 struct string_slot_hasher.
10617 (eq_string_slot_node): Likewise.
10619 * data-streamer-out.c: Update output_block::string_hash_table
10620 dependent calls and types.
10622 * dwarf2cfi.c (htab_t trace_index):
10623 Change type to hash_table. Update dependent calls and types.
10625 * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
10626 Change type to hash_table. Update dependent calls and types.
10627 (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
10628 (htab_t optimize_external_refs::map): Likewise.
10629 (htab_t output_comp_unit::extern_map): Likewise.
10630 (htab_t output_comdat_type_unit::extern_map): Likewise.
10631 (htab_t output_macinfo::macinfo_htab): Likewise.
10632 (htab_t optimize_location_lists::htab): Likewise.
10633 (htab_t dwarf2out_finish::comdat_type_table): Likewise.
10635 * except.c (htab_t ehspec_hash_type):
10636 Change type to hash_table. Update dependent calls and types.
10637 (assign_filter_values::ttypes): Likewise.
10638 (assign_filter_values::ehspec): Likewise.
10639 (sjlj_assign_call_site_values::ar_hash): Likewise.
10640 (convert_to_eh_region_ranges::ar_hash): Likewise.
10642 * gcse.c (htab_t pre_ldst_table):
10643 Change type to hash_table. Update dependent calls and types.
10645 * ggc-common.c (htab_t saving_htab):
10646 Change type to hash_table. Update dependent calls and types.
10647 (htab_t loc_hash): Likewise.
10648 (htab_t ptr_hash): Likewise.
10649 (call_count): Rename ggc_call_count.
10650 (call_alloc): Rename ggc_call_alloc.
10651 (loc_descriptor): Rename make_loc_descriptor.
10652 (add_statistics): Rename ggc_add_statistics.
10654 * ggc-common.c (saving_htab):
10655 Change type to hash_table. Update dependent calls and types.
10657 * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
10658 (push_gimplify_context): Likewise.
10659 (pop_gimplify_context): Likewise.
10660 (struct gimple_temp_hash_elt): Added.
10661 (struct gimplify_hasher): Likewise.
10662 (struct gimplify_ctx.temp_htab):
10663 Change type to hash_table. Update dependent calls and types.
10665 * gimple-fold.c: Include gimplify-ctx.h.
10667 * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
10668 Change type to hash_table. Update dependent calls and types.
10669 (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
10670 avoid potential global name collision.
10672 * gimplify.c: Include gimplify-ctx.h.
10673 (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
10674 (htab_t gimplify_ctx::temp_htab):
10675 Update dependent calls and types for new type hash_table.
10676 (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
10677 (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
10679 * gimplify-ctx.h: New.
10680 (struct gimple_temp_hash_elt): Move from gimplify.c.
10681 (class gimplify_hasher): New.
10682 (struct gimplify_ctx): Move from gimple.h.
10683 (htab_t gimplify_ctx::temp_htab):
10684 Change type to hash_table. Update dependent calls and types.
10686 * graphite-clast-to-gimple.c: Include graphite-htab.h.
10687 (htab_t ivs_params::newivs_index):
10688 Change type to hash_table. Update dependent calls and types.
10689 (htab_t ivs_params::params_index): Likewise.
10690 (htab_t print_generated_program::params_index): Likewise.
10691 (htab_t gloog::newivs_index): Likewise.
10692 (htab_t gloog::params_index): Likewise.
10694 * graphite.c: Include graphite-htab.h.
10695 4htab_t graphite_transform_loops::bb_pbb_mapping):
10696 Change type to hash_table. Update dependent calls and types.
10698 * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
10699 (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
10700 (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
10702 * graphite-dependences.c: Include graphite-htab.h.
10703 (loop_is_parallel_p): Change hash table type of parameter.
10705 * graphite-htab.h: New.
10706 (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
10707 (extern find_pbb_via_hash): Move from graphite-poly.h.
10708 (extern loop_is_parallel_p): Move from graphite-poly.h.
10709 (extern get_loop_body_pbbs): Move from graphite-poly.h.
10711 * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
10712 (extern loop_is_parallel_p): Move to graphite-htab.h.
10713 (extern get_loop_body_pbbs): Move to graphite-htab.h.
10715 * haifa-sched.c (htab_t delay_htab):
10716 Change type to hash_table. Update dependent calls and types.
10717 (htab_t delay_htab_i2): Likewise.
10719 * ira-color.c (htab_t allocno_hard_regs_htab):
10720 Change type to hash_table. Update dependent calls and types.
10722 * ira-costs.c (htab_t cost_classes_htab):
10723 Change type to hash_table. Update dependent calls and types.
10725 * loop-invariant.c (htab_t merge_identical_invariants::eq):
10726 Change type to hash_table. Update dependent calls and types.
10728 * loop-iv.c (htab_t bivs):
10729 Change type to hash_table. Update dependent calls and types.
10731 * loop-unroll.c (htab_t opt_info::insns_to_split):
10732 Change type to hash_table. Update dependent calls and types.
10733 (htab_t opt_info::insns_with_var_to_expand): Likewise.
10735 * lto-streamer.h (struct string_slot): Move from data-streamer.h
10736 (struct string_slot_hasher): New.
10737 (htab_t output_block::string_hash_table):
10738 Change type to hash_table. Update dependent calls and types.
10740 * lto-streamer-in.c (freeing_string_slot_hasher): New.
10741 (htab_t file_name_hash_table):
10742 Change type to hash_table. Update dependent calls and types.
10744 * lto-streamer-out.c: Update output_block::string_hash_table dependent
10747 * lto-streamer.c (htab_t tree_htab):
10748 Change type to hash_table. Update dependent calls and types.
10750 * omp-low.c: Include gimplify-ctx.h.
10752 * passes.c (htab_t name_to_pass_map):
10753 Change type to hash_table. Update dependent calls and types.
10754 (pass_traverse): Rename to passes_pass_traverse.
10756 * plugin.c (htab_t event_tab):
10757 Change type to hash_table. Update dependent calls and types.
10759 * postreload-gcse.c (htab_t expr_table):
10760 Change type to hash_table. Update dependent calls and types.
10761 (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
10763 * sese.c (debug_rename_map_1): Make extern.
10764 (htab_t copy_bb_and_scalar_dependences::rename_map):
10765 Change type to hash_table. Update dependent calls and types.
10767 * sese.h (extern debug_rename_map): Move to .c file.
10769 * store-motion.c (htab_t store_motion_mems_table):
10770 Change type to hash_table. Update dependent calls and types.
10772 * trans-mem.c (htab_t tm_new_mem_hash):
10773 Change type to hash_table. Update dependent calls and types.
10775 * tree-browser.c (htab_t TB_up_ht):
10776 Change type to hash_table. Update dependent calls and types.
10778 * tree-cfg.c (htab_t discriminator_per_locus):
10779 Change type to hash_table. Update dependent calls and types.
10781 * tree-complex.c: Include tree-hasher.h
10782 (htab_t complex_variable_components):
10783 Change type to hash_table. Update dependent calls and types.
10785 * tree-eh.c (htab_t finally_tree):
10786 Change type to hash_table. Update dependent calls and types.
10788 * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
10789 struct int_tree_hasher.
10790 (extern int_tree_map_eq): Likewise.
10791 (uid_decl_map_hash): Removed.
10792 (extern decl_tree_map_eq): Likewise.
10794 * tree-hasher.h: New.
10795 (struct int_tree_hasher): New.
10796 (typedef int_tree_htab_type): New.
10798 * tree-inline.c: Include gimplify-ctx.h.
10800 * tree-mudflap.c: Include gimplify-ctx.h.
10802 * tree-parloops.c: Include tree-hasher.h.
10803 (htab_t eliminate_local_variables_stmt::decl_address):
10804 Change type to hash_table. Update dependent calls and types.
10805 (htab_t separate_decls_in_region::decl_copies): Likewise.
10807 * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
10808 Change type to hash_table. Update dependent calls and types.
10810 * tree-sra.c (candidates):
10811 Change type to hash_table. Update dependent calls and types.
10813 * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
10815 (int_tree_map_hash): Likewise.
10817 * tree-ssa-dom.c (htab_t avail_exprs):
10818 Change type to hash_table. Update dependent calls and types.
10820 * tree-ssa-live.c (var_map_base_init::tree_to_index):
10821 Change type to hash_table. Update dependent calls and types.
10823 * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
10824 Change type to hash_table. Update dependent calls and types.
10826 * tree-ssa-phiopt.c (seen_ssa_names):
10827 Change type to hash_table. Update dependent calls and types.
10829 * tree-ssa-strlen.c (decl_to_stridxlist_htab):
10830 Change type to hash_table. Update dependent calls and types.
10832 * tree-ssa-uncprop.c (equiv):
10833 Change type to hash_table. Update dependent calls and types.
10835 2013-04-25 Jakub Jelinek <jakub@redhat.com>
10837 PR rtl-optimization/57003
10838 * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
10839 call note_stores with kill_clobbered_value callback again after
10840 killing regs_invalidated_by_call.
10842 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10844 * config/aarch64/aarch64-simd.md
10845 (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
10846 (aarch64_simd_bsl<mode>): Likewise.
10847 * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
10849 2013-04-25 Marek Polacek <polacek@redhat.com>
10851 PR tree-optimization/57066
10852 * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
10854 2013-04-25 James Greenhalgh <jame.greenhalgh@arm.com>
10856 * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
10858 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10860 * config/aarch64/aarch64-builtins.c
10861 (aarch64_fold_builtin): New.
10862 * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
10863 * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
10864 * config/aarch64/aarch64-simd-builtins.def (abs): New.
10865 * config/aarch64/arm_neon.h
10866 (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
10868 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10869 Tejas Belagod <tejas.belagod@arm.com>
10871 * config/aarch64/aarch64-builtins.c
10872 (aarch64_gimple_fold_builtin): New.
10873 * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
10874 * config/aarch64/aarch64-simd-builtins.def (addv): New.
10875 * config/aarch64/aarch64-simd.md (addpv4sf): New.
10876 (addvv4sf): Update.
10877 * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
10879 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10881 * config/aarch64/aarch64.md
10882 (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
10884 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10886 * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
10887 (*ngcsi_uxtw): New pattern.
10889 2013-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10890 Julian Brown <julian@codesourcery.com>
10892 * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
10893 (TB_DREG): Add T_V4HF.
10894 (v4hf_UP): New macro.
10895 (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10896 (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10897 Handle initialisation of V4HF. Adjust initialisation of reinterpret
10899 (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
10900 (arm_vector_mode_supported_p): Handle V4HF.
10901 (arm_mangle_map): Handle V4HFmode.
10902 * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
10903 * config/arm/arm_neon_builtins.def: Add entries for
10904 vcvtv4hfv4sf, vcvtv4sfv4hf.
10905 * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
10906 (neon_vcvtv4hfv4sf): Likewise.
10907 * config/arm/neon-gen.ml: Handle half-precision floating point
10909 * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
10910 * config/arm/arm_neon.h: Regenerate.
10911 * config/arm/neon.ml (type elts): Add F16.
10912 (type vectype): Add T_float16x4, T_floatHF.
10913 (type vecmode): Add V4HF.
10914 (type features): Add Requires_FP_bit feature.
10915 (elt_width): Handle F16.
10916 (elt_class): Likewise.
10917 (elt_of_class_width): Likewise.
10918 (mode_of_elt): Refactor.
10919 (type_for_elt): Handle F16, fix error messages.
10920 (vectype_size): Handle T_float16x4.
10921 (vcvt_sh): New function.
10922 (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
10923 (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
10924 (string_of_mode): Handle V4HF.
10925 * doc/arm-neon-intrinsics.texi: Regenerate.
10927 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
10929 * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
10930 format specifier in 'X' case.
10932 2013-04-25 Alan Modra <amodra@gmail.com>
10935 * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
10936 rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
10937 (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
10938 Repeat for many other rotate/shift and mask patterns using subregs.
10939 Name lshiftrt insns.
10940 (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
10941 on WORDS_BIG_ENDIAN.
10943 2013-04-25 Alan Modra <amodra@gmail.com>
10945 * config.gcc: Support little-endian powerpc-linux targets.
10946 * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
10947 (LINK_OS_LINUX_SPEC): Define.
10948 * config/rs6000/linuxspe.h (TARGET_DEFAULT):
10949 Preserve MASK_LITTLE_ENDIAN.
10950 * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
10951 * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
10952 * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
10953 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
10954 (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
10955 * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
10956 Correct fp word order for little-endian. Don't shift toc entries
10957 smaller than a word for little-endian.
10958 * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
10959 (bswapdi2 splits): Correct low-part subreg for little-endian.
10960 Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
10961 low/high where such is correct only for be.
10962 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
10963 little-endian for -mcall-aixdesc.
10965 2013-04-25 Alan Modra <amodra@gmail.com>
10967 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
10968 replace_equiv_address_nv.
10970 2013-04-25 Alan Modra <amodra@gmail.com>
10972 * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
10974 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10977 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10978 * rtl.h (struct rtx_def): ...
10980 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
10982 PR rtl-optimizations/57046
10983 * lra-constraints (split_reg): Set up lra_risky_transformations_p
10984 for multi-reg splits.
10986 2013-04-24 H.J. Lu <hongjiu.lu@intel.com>
10988 * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
10990 2013-04-24 Sterling Augustine <saugustine@google.com>
10992 * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
10993 (comp_dir_string, debug_str_dwo_section): New.
10994 (DEBUG_STR_DWO_SECTION): Rename to ...
10995 (DEBUG_DWO_STR_SECTION): ... this.
10996 (DEBUG_NORM_STR_SECTION): Delete.
10997 (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
10998 (DEBUG_STR_DWO_SECTION_FLAGS): New.
10999 (find_AT_string): Move most logic to ...
11000 (find_AT_string_in_table): ... here. New.
11001 (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11002 add_skeleton_AT_string. Delete logic.
11003 (output_skeleton_debug_sections): Remove call to
11004 add_top_level_skeleton_die_attrs.
11005 (add_comp_dir_attribute): Move logic to comp_dir_string.
11006 (dwarf2out_init): Initialize debug_str_dwo_section.
11007 (output_indirect_string): Call find_string_form.
11008 (output_indirect_strings): Rewrite.
11009 (prune_unused_types): Empty skeleton_debug_str_hash.
11010 Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
11011 (dwarf2out_finish): Call output_indirect_strings.
11013 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
11015 * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
11017 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11019 * rtl.h (struct rtx_def): Add comment for field jump.
11020 (LRA_SUBREG_P): New macro.
11021 * recog.c (register_operand): Check LRA_SUBREG_P.
11022 * lra.c (lra): Add note at the end of RTL code. Align non-empty
11024 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11025 (lra_final_code_change): Skip subreg change for operators.
11026 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11027 if there are no operand changes.
11028 * lra-constraints.c (curr_insn_set): New.
11029 (match_reload): Set LRA_SUBREG_P.
11030 (emit_spill_move): Ditto.
11031 (check_and_process_move): Use curr_insn_set. Process only single
11032 set insns. Don't initialize sec_mem_p and change_p.
11033 (simplify_operand_subreg): Use LRA_SUBREG_P.
11034 (reg_in_class_p): New function.
11035 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
11036 of #ifdef. Add code to remove cycling.
11037 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11038 non-null disp. Reload inner instead of disp when base and index
11039 are null. Try to put lo_sum into register.
11040 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11041 (check_and_process_move): Move code for move cost check to
11042 simple_move_p. Remove equiv_substitution.
11043 (simple_move_p): New function.
11044 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
11045 curr_insn_set. Call check_and_process_move only for single set
11046 insns. Use the new function. Move call of check_and_process_move
11047 after operand equiv substitution and address process.
11049 2013-04-24 James Greenhalgh <james.greenhalgh@arm.com>
11051 * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
11052 (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
11053 (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
11055 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
11057 * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
11059 2013-04-24 Marek Polacek <polacek@redhat.com>
11061 * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
11062 * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
11063 (select_loops_exit_conditions): Likewise.
11064 (number_of_iterations_for_all_loops): Likewise.
11065 (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
11066 (scev_analysis): Likewise.
11068 2013-04-02 Catherine Moore <clm@codesourcery.com>
11069 Chao-ying Fu <fu@mips.com>
11071 * config/mips/micromips.md (jraddiusp): New pattern.
11072 * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
11073 instruction if possible.
11075 2013-04-24 Alan Modra <amodra@gmail.com>
11077 * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
11079 2013-04-24 Julian Brown <julian@codesourcery.com>
11080 Chung-Lin Tang <cltang@codesourcery.com>
11082 * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
11083 dependency behavior in enumeration type DIE generation. Add TODO note
11084 to comments about future DW_FORM_sdata/udata re-work of related code.
11086 2013-04-23 Lawrence Crowl <crowl@google.com>
11088 * Makefile.in: Update as needed below.
11090 * hash-table.h (class hash_table):
11091 Correct many methods with parameter types compare_type to the correct
11092 value_type. (Correct code was unlikely to notice the change.)
11093 (hash_table::elements_with_deleted) New.
11094 (class hashtable::iterator): New.
11095 (hashtable::begin()): New.
11096 (hashtable::end()): New.
11097 (FOR_EACH_HASH_TABLE_ELEMENT): New.
11099 * statistics.c (statistics_hashes):
11100 Change type to hash_table. Update dependent calls and types.
11102 * tree-into-ssa.c (var_infos):
11103 Change type to hash_table. Update dependent calls and types.
11105 * tree-ssa-coalesce.c (struct coalesce_list_d.list):
11106 Change type to hash_table. Update dependent calls and types.
11108 * tree-ssa-loop-im.c (struct mem_ref.refs):
11109 Change type to hash_table. Update dependent calls and types.
11111 * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
11112 Change type to hash_table. Update dependent calls and types.
11114 * tree-ssa-sccvn.c (vn_tables_s::nary):
11115 Change type to hash_table. Update dependent calls and types.
11116 (vn_tables_s::phis): Likewise.
11117 (vn_tables_s::references): Likewise.
11119 * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
11120 (vn_reference_eq): Update parameter and return types.
11122 * tree-ssa-structalias.c (pointer_equiv_class_table):
11123 Change type to hash_table. Update dependent calls and types.
11124 (location_equiv_class_table): Likewise.
11126 * tree-vect-data-refs.c: Consequential changes for making
11127 peeling a hash_table.
11129 * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
11130 (destroy_loop_vec_info): Dependent hash_table update.
11132 * tree-vectorizer.h (peeling_htab):
11133 Change type to hash_table. Update dependent calls and types.
11135 2013-04-23 Shiva Chen <shiva0217@gmail.com>
11137 * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
11138 to check the register content is equal or not.
11139 * lra-constraints.c (match_reload): Use lra_assign_reg_val
11140 to assign register content record.
11141 * lra-eliminations.c (update_reg_eliminate): Use
11142 lra_update_reg_val_offset to update register content offset.
11143 * lra-int.h (struct lra_reg): Add offset member.
11144 (lra_reg_val_equal_p): New static inline function.
11145 (lra_update_reg_val_offset): New static inline function.
11146 (lra_assign_reg_val): New static inline function.
11147 * lra.c (lra_create_new_reg): Use lra_assign_reg_val
11148 to assign register content record.
11149 (initialize_lra_reg_info_element): Initial offset to zero.
11151 2013-04-23 Catherine Moore <clm@codesourcery.com>
11153 * config/mips/mips.md (*movhi_internal, *movqi_internal): New
11154 operands. Record compression.
11156 2013-04-23 Xinliang David Li <davidxl@google.com>
11158 * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
11160 2013-04-23 Richard Biener <rguenther@suse.de>
11162 PR middle-end/57036
11163 * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
11164 parameter, only add abnormal goto edges from the copied body
11165 if the call could perform abnormal gotos.
11166 (copy_cfg_body): Adjust.
11168 2013-04-23 Sofiane Naci <sofiane.naci@arm.com>
11170 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
11172 2013-04-23 Andreas Schwab <schwab@linux-m68k.org>
11174 * coretypes.h (gimple_stmt_iterator): Add struct to make
11177 2013-04-23 Richard Biener <rguenther@suse.de>
11179 PR tree-optimization/57026
11180 * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
11181 from SSA names occuring in abnormal PHI nodes.
11183 2013-04-22 Andi Kleen <ak@linux.intel.com>
11185 * lto/lto.c (print_lto_report_1): Fix LTO report names.
11187 2013-04-22 Andi Kleen <ak@linux.intel.com>
11189 * lto/lto.c (print_lto_report_1): Declare early.
11190 (read_cgraph_and_symbols): Call print_lto_report_1 early.
11192 2013-04-22 Andi Kleen <ak@linux.intel.com>
11194 * common.opt (-flto-report-wpa): Add.
11195 * doc/invoke.texi (-flto-report-wpa): Add.
11196 * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
11199 2013-04-22 Xinliang David Li <davidxl@google.com>
11201 * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
11202 * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
11203 * Makefile.in: New dependency
11205 David Daney <ddaney.cavm@gmail.com>
11207 * configure.ac (gcc_cv_as_micromips_support): Use the
11208 --fatal-warnings option.
11209 * configure: Regenerate.
11211 2013-04-22 Marek Polacek <polacek@redhat.com>
11214 * tsan.c (instrument_expr): Don't instrument expression
11215 in case its size is zero.
11217 2013-04-22 Uros Bizjak <ubizjak@gmail.com>
11221 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
11223 * config/alpha/alpha.c (TARGET_LRA_P): New define.
11225 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
11227 * coretypes.h (gimple_stmt_iterator_d): Forward declare.
11228 (gimple_stmt_iterator): New typedef.
11229 * gimple.h (gimple_stmt_iterator): Rename to...
11230 (gimple_stmt_iterator_d): ... This.
11231 * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
11232 trees be valid for GIMPLE and GENERIC.
11233 (TARGET_GIMPLE_FOLD_BUILTIN): New.
11234 * gimple-fold.c (gimple_fold_call): Call target hook
11235 gimple_fold_builtin.
11236 * hooks.c (hook_bool_gsiptr_false): New.
11237 * hooks.h (hook_bool_gsiptr_false): New.
11238 * target.def (fold_stmt): New.
11239 * doc/tm.texi: Regenerate.
11241 2013-04-22 Vladimir Makarov <vmakarov@redhat.com>
11244 * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
11245 a set sp if no stack realignment.
11247 2013-04-22 Nick Clifton <nickc@redhat.com>
11249 * config.gcc (tilegx-linux): Extend extra_objs rather than
11251 (tilepro-linux): Likewise.
11253 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
11255 * config/aarch64/aarch64-builtins.c
11257 (CF0, CF1, CF2, CF3, CF4, CF10): New.
11258 (VAR<1-12>): Add MAP parameter.
11259 (BUILTIN_*): Likewise.
11260 * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
11261 * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
11262 (aarch64_ushl_n<mode>): Likewise.
11263 (aarch64_sshr_n<mode>): Likewise.
11264 (aarch64_ushr_n<mode>): Likewise.
11265 (aarch64_<maxmin><mode>): Likewise.
11266 (aarch64_sqrt<mode>): Likewise.
11267 * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
11268 (vshr<q>_n_*): Likewise.
11270 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
11272 * config/aarch64/aarch64-builtins.c
11273 (aarch64_simd_builtin_type_mode): Handle SF types.
11275 (BUILTIN_GPF): Define.
11276 (aarch64_init_simd_builtins): Handle SF types.
11277 * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
11278 (frecps): Likewise.
11279 (frecpx): Likewise.
11280 * config/aarch64/aarch64-simd.md
11281 (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
11282 (aarch64_frecpe<mode>): New.
11283 (aarch64_frecps<mode>): Likewise.
11284 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
11285 (v8type): Add frecp<esx>.
11286 (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
11287 (aarch64_frecps<mode>): Likewise.
11288 * config/aarch64/iterators.md (FRECP): New.
11289 (frecp_suffix): Likewise.
11290 * config/aarch64/arm_neon.h
11291 (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
11293 2013-04-22 Christian Bruel <christian.bruel@st.com>
11296 * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
11297 (REG_CLASS_NAMES): Idem.
11298 (REG_CLASS_CONTENTS): Idem.
11299 (REGCLASS_HAS_FP_REG): Idem.
11300 * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
11301 (sh_conditional_register_usage): Idem.
11303 2013-04-21 Jeff Law <law@redhat.com>
11305 * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
11306 (ssa_forward_propagate_and_combine): Use it.
11308 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
11310 * lra.c: Update the flow chart diagram.
11312 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
11314 PR rtl-optimization/56847
11315 * lra-constraints.c (process_alt_operands): Discourage alternative
11316 with non-matche doffsettable memory constraint fro memory with
11319 2013-04-19 Richard Biener <rguenther@suse.de>
11321 PR tree-optimization/56982
11322 * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
11324 * gimplify.c (gimplify_call_expr): Notice special calls.
11325 (gimplify_modify_expr): Likewise.
11326 * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
11327 abnormal control flow receivers.
11328 (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
11329 in the same way as cfun->has_nonlocal_labels.
11330 (gimple_purge_dead_abnormal_call_edges): Likewise.
11331 (stmt_starts_bb_p): Make setjmp-like abnormal control flow
11332 receivers start a basic-block.
11334 2013-04-19 Richard Biener <rguenther@suse.de>
11336 * tree-vectorizer.h (struct _slp_instance): Move load_permutation
11338 (struct _slp_tree): ... here. Make it a vector of unsigned ints.
11339 (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
11340 (SLP_TREE_LOAD_PERMUTATION): Add.
11341 (vect_transform_slp_perm_load): Adjust prototype.
11342 * tree-vect-slp.c (vect_free_slp_tree): Adjust.
11343 (vect_free_slp_instance): Likewise.
11344 (vect_create_new_slp_node): Likewise.
11345 (vect_supported_slp_permutation_p): Remove.
11346 (vect_slp_rearrange_stmts): Adjust.
11347 (vect_supported_load_permutation_p): Likewise. Inline
11348 vect_supported_slp_permutation_p here.
11349 (vect_analyze_slp_instance): Compute load permutations per
11350 slp node instead of per instance.
11351 (vect_get_slp_defs): Adjust.
11352 (vect_transform_slp_perm_load): Likewise.
11353 (vect_schedule_slp_instance): Remove redundant code.
11354 (vect_schedule_slp): Remove hack for PR56270, add it ...
11355 * tree-vect-stmts.c (vectorizable_load): ... here, do not
11356 CSE loads for SLP. Adjust.
11358 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
11360 * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
11361 spelling in two comments.
11363 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
11366 * config/arm/arm.c (load_multiple_sequence): Require SP
11367 as base register for loads if SP is in the register list.
11369 2013-04-19 Martin Jambor <mjambor@suse.cz>
11371 PR tree-optimization/56718
11372 * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
11373 * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
11374 and made public. Adjusted all callers.
11375 (ipa_intraprocedural_devirtualization): New function.
11376 * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
11377 (ipa_intraprocedural_devirtualization): Likewise.
11378 * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
11380 2013-04-19 Richard Biener <rguenther@suse.de>
11382 PR tree-optimization/57000
11383 * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
11385 2013-04-19 Terry Guo <terry.guo@arm.com>
11387 * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
11389 (cortex_m4_v_a, cortex_m4_v_b): ... new cpu units.
11390 (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
11391 (cortex_m4_fmacs): Use new reservations.
11392 (cortex_m4_f_load, cortex_m4_f_store): Likewise.
11394 2013-04-18 Vladimir Makarov <vmakarov@redhat.com>
11396 PR rtl-optimization/56999
11397 * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
11399 (lra_coalesce): Remove split_origin_bitmap and related code.
11400 * lra.c (lra): Coalesce after undoing inheritance. Recreate live
11401 ranges if necessary.
11403 2013-04-18 Uros Bizjak <ubizjak@gmail.com>
11405 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
11407 (ix86_expand_call): Remove clobbered_registers array and use
11408 x86_64_ms_sysv_extra_clobbered_registers instead.
11409 * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
11411 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
11413 * config/i386/i386.md (*call_rex64_ms_sysv): Use
11414 call_rex64_ms_sysv_operation predicate. Remove explicit clobbers.
11415 (*call_value_rex64_ms_sysv): Ditto.
11417 2013-04-18 Cary Coutant <ccoutant@google.com>
11419 * dwarf2out.c (output_pubnames): Check die_perennial_p of
11420 parent instead of die_mark.
11422 2013-04-18 Diego Novillo <dnovillo@google.com>
11424 * gimple.c (create_gimple_tmp): New.
11425 (get_expr_type): New.
11426 (build_assign): New.
11427 (build_type_cast): New.
11428 * gimple.h (enum ssa_mode): Define.
11429 (gimple_seq_set_location): New.
11430 * asan.c (build_check_stmt): Change some gimple_build_* calls
11431 to use build_assign and build_type_cast.
11433 2013-04-18 Richard Biener <rguenther@suse.de>
11435 * tree-vect-data-refs.c (vect_analyze_group_access): Properly
11436 handle negative step. Remove redundant checks.
11437 (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
11438 * tree-vect-stmts.c (vectorizable_load): Instead of asserting
11439 for negative step and grouped loads fail to vectorize.
11441 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
11443 * emit-rtl.c (reset_insn_used_flags): New function.
11444 (reset_all_used_flags): Use it.
11445 (verify_insn_sharing): New function.
11446 (verify_rtl_sharing): Fix verification for SEQUENCEs.
11448 2013-04-18 Jakub Jelinek <jakub@redhat.com>
11450 PR tree-optimization/56984
11451 * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
11452 and (x >> M) >= N don't register any assertion if N << M is the
11455 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
11457 * lower-subreg.c (resolve_simple_move): If called self-recursive,
11458 do not delete_insn insns that have not yet been emitted, only
11459 unlink them with remove_insn.
11460 * df-scan.c (df_insn_delete): Revert r197492.
11462 2013-04-17 Steven Bosscher <steven@gcc.gnu.org>
11464 * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
11465 * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
11467 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
11469 * config/arm/arm.md (movsicc_insn): Convert define_insn into
11470 define_insn_and_split.
11471 (and_scc,ior_scc,negscc): Likewise.
11472 (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
11474 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
11476 * config/arm/arm.c (use_return_insn): Return 0 for targets that
11477 can benefit from using a sequence of LDRD instructions in epilogue
11478 instead of a single LDM instruction.
11480 2013-04-17 Manuel López-Ibáñez <manu@gcc.gnu.org>
11483 * doc/extend.texi: Fix typo.
11485 2013-04-17 Richard Biener <rguenther@suse.de>
11487 * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
11488 (vect_build_slp_tree): ... here.
11489 (vect_build_slp_tree_1): Compute which stmts of the SLP group
11490 match. Remove special-casing of mismatched complex loads.
11491 (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
11492 re-try the match with swapped commutative operands.
11493 (vect_supported_load_permutation_p): Remove special-casing of
11494 mismatched complex loads.
11495 (vect_analyze_slp_instance): Adjust.
11497 2013-04-17 Richard Biener <rguenther@suse.de>
11499 PR rtl-optimization/56921
11500 * cfgloop.h (struct loop): Add simple_loop_desc member.
11501 (struct niter_desc): Mark with GTY(()).
11502 (simple_loop_desc): Do not use aux field but simple_loop_desc.
11503 * loop-iv.c (get_simple_loop_desc): Likewise.
11504 (free_simple_loop_desc): Likewise.
11507 2013-04-16 Richard Biener <rguenther@suse.de>
11509 PR rtl-optimization/56921
11510 * loop-init.c (pass_rtl_move_loop_invariants): Add
11511 TODO_do_not_ggc_collect to todo_flags_finish.
11512 (pass_rtl_unswitch): Same.
11513 (pass_rtl_unroll_and_peel_loops): Same.
11514 (pass_rtl_doloop): Same.
11516 2013-04-17 Eric Botcazou <ebotcazou@adacore.com>
11518 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
11519 (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
11520 Use nonoverlapping_component_refs_of_decl_p to disambiguate component
11522 (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
11523 * tree-streamer.c (record_common_node): Adjust reference in comment.
11525 2013-04-17 Terry Guo <terry.guo@arm.com>
11527 * config/arm/cortex-m4.md: Add a new bypass.
11529 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11531 * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
11533 (*subs_<optab><mode>_multp2): New pattern.
11534 (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11535 (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
11537 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11539 * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
11540 (*subs_mul_imm_<mode>): New pattern.
11542 2013-04-16 David Edelsohn <dje.gcc@gmail.com>
11545 * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
11546 (vsx_movti_64bit): Change j->wa to O->wa. Add n->r alternative.
11547 (vsx_movti_32bit): Change j->wa to O->wa.
11549 2013-04-16 Richard Biener <rguenther@suse.de>
11551 PR rtl-optimization/56921
11552 * loop-init.c (pass_rtl_move_loop_invariants): Add
11553 TODO_do_not_ggc_collect to todo_flags_finish.
11554 (pass_rtl_unswitch): Same.
11555 (pass_rtl_unroll_and_peel_loops): Same.
11556 (pass_rtl_doloop): Same.
11558 2013-04-16 Greta Yorsh <Greta.Yorsh@arm.com>
11560 * config/arm/arm.c (emit_multi_reg_push): New declaration
11561 for an existing function.
11562 (arm_emit_strd_push): New function.
11563 (arm_expand_prologue): Used here.
11564 (arm_emit_ldrd_pop): New function.
11565 (arm_expand_epilogue): Used here.
11566 (arm_get_frame_offsets): Update condition.
11567 (arm_emit_multi_reg_pop): Add a special case for load of a single
11568 register with writeback.
11570 2013-04-16 Uros Bizjak <ubizjak@gmail.com>
11572 * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
11575 2013-04-16 Richard Biener <rguenther@suse.de>
11577 PR tree-optimization/56756
11578 * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
11579 (first_mem_ref_loc): New.
11580 (execute_sm): Place the load temporarily before a previous
11581 access instead of in the latch edge to ensure its SSA dependencies
11582 are defined at points dominating the load.
11584 2013-04-16 Steven Bosscher <steven@gcc.gnu.org>
11586 * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
11587 correct fix by moving header and footer insn to the footer of
11588 the merged basic block. Clear BB_END of the merged-away block.
11590 PR middle-end/43631
11591 * emit-rtl.c (make_note_raw): New function.
11592 (link_insn_into_chain): New static inline function.
11593 (add_insn): Use it.
11594 (add_insn_before, add_insn_after): Factor insn chain linking code...
11595 (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
11596 using link_insn_into_chain.
11597 (note_outside_basic_block_p): New helper function for emit_note_after
11598 and emit_note_before.
11599 (emit_note_after): Use nobb variant of add_insn_after if the note
11600 should not be contained in a basic block.
11601 (emit_note_before): Use nobb variant of add_insn_before if the note
11602 should not be contained in a basic block.
11603 (emit_note_copy): Use make_note_raw.
11604 (emit_note): Likewise.
11605 * bb-reorder.c (insert_section_boundary_note): Remove hack to set
11606 BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
11607 * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
11608 the moved barrier the tail of the basic block it follows.
11609 * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
11611 2013-04-15 Jakub Jelinek <jakub@redhat.com>
11613 PR tree-optimization/56962
11614 * gimple-ssa-strength-reduction.c (record_increment): Only set
11615 initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
11616 either rhs1 or rhs2 is equal to c->base_expr.
11618 2013-04-15 Richard Biener <rguenther@suse.de>
11620 PR tree-optimization/56933
11621 * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
11623 (GROUP_READ_WRITE_DEPENDENCE): Remove.
11624 (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
11625 * tree-vect-data-refs.c (vect_analyze_group_access): Move
11626 dependence check ...
11627 vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
11629 * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
11630 GROUP_READ_WRITE_DEPENDENCE.
11632 2013-04-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
11634 * emit-rtl.c (reset_all_used_flags): New function.
11635 (verify_rtl_sharing): Call reset_all_used_flags before and after
11636 performing the checks.
11638 2013-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11640 * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
11641 * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
11642 * config/arm/constraints.md (De): New constraint.
11643 * config/arm/neon.md (anddi3_neon): Delete.
11644 (neon_vand<mode>): Expand to standard anddi3 pattern.
11645 * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
11646 Move earlier in the file.
11647 (neon_inv_logic_op2): Likewise.
11648 (arm_anddi_operand_neon): New predicate.
11650 2013-04-15 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11652 * configure.ac (gcc_cv_ld_as_needed): Set
11653 gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
11654 Use -z ignore, -z record on *-*-solaris2*.
11655 (HAVE_LD_AS_NEEDED): Update comment.
11656 (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
11657 * configure: Regenerate.
11658 * config.in: Regenerate.
11659 * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
11660 LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
11661 * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
11662 * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
11663 equivalents. Fix markup.
11664 * doc/tm.texi: Regenerate.
11666 2013-04-15 Andrew Hsieh <andrewhsieh.google.com>
11668 * config/i386/i386.opt: New option mstack-protector-guard=.
11669 * config/i386/i386-opts.h: Add enum stack_protector_guard.
11670 * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
11671 TARGET_SSP_TLS_GUARD.
11672 * config/i386/i386.c (ix86_option_override_internal): Set
11673 ix86_stack_protector_guard.
11674 * config/i386/i386.md (stack_protect_set): Enable for
11675 TARGET_SSP_TLS_GUARD only.
11676 (stack_protect_set_<mode>): Ditto.
11677 (stack_protect_test): Ditto.
11678 (stack_protect_test_<mode>): Ditto.
11679 * doc/invoke.texi (i386 Option): Document.
11681 2013-04-15 Eric Botcazou <ebotcazou@adacore.com>
11684 * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
11685 (S_MODES): Set H_MODE bit.
11686 (SF_MODES): Set only S_MODE and SF_MODE bits.
11687 (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
11688 (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
11689 <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
11690 <MODE_FLOAT>: Likewise.
11692 2013-04-15 Joey Ye <joey.ye@arm.com>
11694 * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
11696 2013-04-15 Joey Ye <joey.ye@arm.com>
11698 * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
11700 (thumb_far_jump_used_p): Count instruction size and set
11703 2013-04-14 Eric Botcazou <ebotcazou@adacore.com>
11705 * reorg.c (fill_simple_delay_slots): Reindent block of code.
11706 * resource.c (mark_target_live_regs): Reformat conditional block.
11708 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
11710 * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
11711 notes, they are emitted only just before final.
11712 * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
11714 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
11716 * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
11717 * cfgrtl.c (delete_insn): Call it here instead.
11718 * lra-spills.c (lra_final_code_change): Use delete_insn.
11719 * haifa-sched.c (sched_remove_insn): Likewise.
11720 * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
11721 returning to the nop pool.
11722 (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
11723 use delete_insn for definitive removal. Clear BLOCK_FOR_INSN.
11725 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
11727 * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
11728 * doc/tm.texi: Regenerated.
11730 2013-04-12 Uros Bizjak <ubizjak@gmail.com>
11732 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
11735 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
11737 * df-core.c (df_find_def): Compare register numbers.
11738 (df_find_use): Likewise.
11740 2013-04-12 Vladimir Makarov <vmakarov@redhat.com>
11743 * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
11744 lra_in_progress for return.
11746 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
11748 * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
11749 define_insn into define_insn_and_split and emit movsicc patterns.
11751 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
11753 * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
11755 2013-04-12 Richard Biener <rguenther@suse.de>
11757 * tree-pass.h (TODO_do_not_ggc_collect): New.
11758 * passes.c (execute_one_ipa_transform_pass): Honor
11759 TODO_do_not_ggc_collect.
11760 (execute_one_pass): Likewise.
11763 2013-04-10 Richard Biener <rguenther@suse.de>
11765 * passes.c (init_optimization_passes): Remove reload pass.
11766 * ira.c (do_reload): Merge into ...
11768 (rest_of_handle_reload): Remove.
11769 (pass_reload): Likewise.
11770 * config/i386/i386.c (ix86_option_override): Refer to ira instead
11771 of reload for vzeroupper pass placement.
11773 2013-04-12 Jakub Jelinek <jakub@redhat.com>
11775 PR tree-optimization/56918
11776 PR tree-optimization/56920
11777 * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
11778 instead of op1 - op2. Pass 2 * TYPE_PRECISION (type) as second
11779 argument to rshift method. For 2 * HOST_BITS_PER_WIDE_INT precision
11780 use wide_mul_with_sign method.
11782 2013-04-12 Richard Biener <rguenther@suse.de>
11784 * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
11785 not be considered a gimple constant.
11787 2013-04-12 Marc Glisse <marc.glisse@inria.fr>
11789 * fold-const.c (const_binop): Handle vector shifts by a scalar.
11790 (fold_binary_loc): Call const_binop also for mixed vector-scalar
11793 2013-04-12 Manuel López-Ibáñez <manu@gcc.gnu.org>
11794 Jakub Jelinek <jakub@redhat.com>
11796 * opts.c: Include diagnostic-color.h.
11797 (common_handle_option): Handle OPT_fdiagnostics_color_.
11798 * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
11799 (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
11800 (diagnostic-color.o): New.
11801 * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
11802 (diagnostic_color_rule): New enum.
11803 * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
11804 * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
11805 the location string.
11806 * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
11807 either NULL, or color kind.
11808 * diagnostic-color.c: New file.
11809 * diagnostic-color.h: New file.
11810 * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
11812 * doc/invoke.texi (-fdiagnostics-color): Document.
11813 * pretty-print.h (pp_show_color): Define.
11814 (struct pretty_print_info): Add show_color field.
11815 * diagnostic.c: Include diagnostic-color.h.
11816 (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
11817 macros. Colorize error:, warning: etc. strings and also the location
11819 (diagnostic_show_locus): Colorize the caret line.
11820 * pretty-print.c: Include diagnostic-color.h.
11821 (pp_base_format): Handle %r and %R format specifiers. Colorize strings
11822 inside of %< %> quotes or quoted through q format modifier.
11824 2013-04-12 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
11826 * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
11828 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11830 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
11831 code in CC_NZ mode.
11832 * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
11835 2013-04-11 Marek Polacek <polacek@redhat.com>
11837 PR tree-optimization/48184
11838 * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
11840 2013-04-11 Eric Botcazou <ebotcazou@adacore.com>
11842 * stor-layout.c (skip_simple_constant_arithmetic): Move to...
11843 * tree.c (skip_simple_constant_arithmetic): ...here and make public.
11844 (skip_simple_arithmetic): Tidy up.
11845 * tree.h (skip_simple_constant_arithmetic): Declare.
11847 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11849 * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
11851 2013-04-11 Richard Biener <rguenther@suse.de>
11853 * tree-vect-loop.c (get_initial_def_for_induction): Properly
11854 generate vector constants.
11856 2013-04-11 Richard Biener <rguenther@suse.de>
11858 PR tree-optimization/56878
11859 * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
11860 * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
11862 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
11863 Prefer to align the DR with the most invariant base address.
11865 2013-04-11 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
11867 * opts.c (common_handle_option): Fix formatting and add FALLTHRU
11870 2013-04-11 James Greenhalgh <james.greenhalgh@arm.com>
11872 * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
11873 floating-point vector comparisons against 0.
11875 2013-04-11 Jakub Jelinek <jakub@redhat.com>
11877 PR tree-optimization/56899
11878 * fold-const.c (extract_muldiv_1): Apply distributive law
11879 only if TYPE_OVERFLOW_WRAPS (ctype).
11881 2013-04-11 Bin Cheng <bin.cheng@arm.com>
11884 * ira-costs.c (scan_one_insn): Check whether the source rtx of
11885 loading has side effect.
11887 2013-04-10 Steven Bosscher <steven@gcc.gnu.org>
11889 * config/sparc/sparc.c: Include tree-pass.h.
11890 (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
11891 (sparc_reorg): Rename to sparc_do_work_around_errata. Move to
11892 head of file. Change return type. Split off gate function.
11893 (sparc_gate_work_around_errata): New function.
11894 (pass_work_around_errata): New pass definition.
11895 (insert_pass_work_around_errata) New pass insert definition to
11896 insert pass_work_around_errata just after delayed-branch scheduling.
11897 (sparc_option_override): Insert the pass.
11898 * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
11900 2013-04-10 David S. Miller <davem@davemloft.net>
11902 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
11903 or -mcpu=hypersparc.
11905 * target.def (cstore_mode): New hook.
11906 * target.h: Include insn-codes.h
11907 * targhooks.c: Likewise.
11908 (default_cstore_mode): New function.
11909 * targhooks.h: Declare it.
11910 * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
11911 * doc/tm.texi: Rebuild.
11912 * expmed.c (emit_cstore): Obtain cstore boolean result mode using
11913 target hook, rather than inspecting the insn_data.
11914 * config/sparc/sparc.c (sparc_cstore_mode): New function.
11915 (TARGET_CSTORE_MODE): Redefine.
11916 (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
11918 * config/sparc/predicates.md (cstore_result_operand): New special
11920 * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
11921 Use it for operand 0.
11922 (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
11923 (*snesi_special): Likewise.
11924 (*snesi_zero): Likewise.
11925 (*seqsi_zero): Likewise.
11926 (*sltu_insn): Likewise.
11927 (*sgeu_insn): Likewise.
11928 (*seqdi_special): Make operand 0 and comparison operation be of
11930 (*snedi_special): Likewise.
11931 (*snedi_special_vis3): Likewise.
11932 (*neg_snesi_zero): Rename to *neg_snesisi_zero.
11933 (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
11934 (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
11935 (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
11936 (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
11937 (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
11938 (*sltu_extend_sp64): Likewise.
11939 (*neg_sltu_insn): Rename to *neg_sltusi_insn.
11940 (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
11941 (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
11942 (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
11943 (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
11945 2013-04-10 Yufeng Zhang <yufeng.zhang@arm.com>
11947 * config/aarch64/aarch64.c (aarch64_print_extension): New function.
11948 (aarch64_start_file): Use the new function.
11950 2013-04-10 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
11951 Jason Merrill <jason@redhat.com>
11953 * common.opt: Add -gdwarf.
11954 * opts.c (common_handle_option): Handle it.
11955 * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
11957 2013-04-10 Richard Biener <rguenther@suse.de>
11959 * passes.c (execute_todo): Do not call ggc_collect conditional here.
11960 (execute_one_ipa_transform_pass): But unconditionally here.
11961 (execute_one_pass): And here.
11962 (init_optimization_passes): Remove reload pass.
11963 * tree-pass.h (TODO_ggc_collect): Remove.
11964 (pass_reload): Likewise.
11965 * ira.c (do_reload): Merge into ...
11967 (rest_of_handle_reload): Remove.
11968 (pass_reload): Likewise.
11969 * config/i386/i386.c (ix86_option_override): Refer to ira instead
11970 of reload for vzeroupper pass placement.
11971 * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
11972 and todo_flags_finish of all passes.
11974 2013-04-10 Richard Biener <rguenther@suse.de>
11976 * tree-vectorizer.h (struct _slp_oprnd_info): Remove
11977 first_const_oprnd field, rename first_def_type to first_op_type.
11978 * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
11979 (vect_get_and_check_slp_defs): Always use the type of the
11980 operand. Allow mixed vect_external_def, vect_constant_def types.
11981 (vect_get_constant_vectors): Handle mixed vect_external_def,
11982 vect_constant_def types.
11984 2013-04-10 Joern Rennecke <joern.rennecke@embecosm.com>
11986 PR tree-optimization/55524
11987 * tree-ssa-math-opts.c
11988 (convert_mult_to_fma): Don't use an fms construct
11989 when we don't have an fms operation, but fnma, and it looks
11990 likely that we'll be able to use the latter.
11992 2013-04-10 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
11994 * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
11996 * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
11997 inline fail caused by overwritable functions.
11999 2013-04-10 Chung-Ju Wu <jasonwucj@gmail.com>
12001 * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12002 unnecessary bits in the constant power of two case.
12004 2013-04-10 Richard Biener <rguenther@suse.de>
12006 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12007 broken code swapping operands.
12008 (vect_build_slp_tree): Do not compute load permutations here.
12009 (vect_analyze_slp_instance): Compute load permutations here,
12010 after building the SLP tree.
12012 2013-04-09 Christian Bruel <christian.bruel@st.com>
12014 * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
12015 of next/prev_real_insn.
12017 2013-04-09 Jan Hubicka <jh@suse.cz>
12019 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
12020 Drop aliased parameter.
12021 (function_and_variable_visibility): Do not handle alias pairs.
12022 * cgraph.c (varpool_externally_visible_p): Update prototype.
12023 * varpool.c (varpool_add_new_variable): Update.
12025 2013-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12027 * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
12029 2013-04-09 Steven Bosscher <steven@gcc.gnu.org>
12031 * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
12033 * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
12035 2013-04-09 Marek Polacek <polacek@redhat.com>
12037 PR tree-optimization/48762
12038 * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
12040 2013-04-09 Richard Biener <rguenther@suse.de>
12042 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
12044 (vect_build_slp_tree): Likewise.
12045 (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
12046 calculating the cost of a SLP instance.
12047 (vect_analyze_slp_instance): Use it from here, after building
12050 2013-04-09 Jakub Jelinek <jakub@redhat.com>
12052 PR middle-end/56883
12053 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
12054 expand_omp_for_static_chunk): Use simple_p = true in
12055 force_gimple_operand_gsi calls when assigning to addressable decls.
12057 2013-04-09 Jeff Law <law@redhat.com>
12059 * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
12060 when the boolean was created by converting a wider object which
12061 had a boolean range.
12063 2013-04-09 Richard Biener <rguenther@suse.de>
12065 * tree-vectorizer.h (slp_void_p): Remove.
12066 (slp_tree): Typedef before _slp_tree declaration.
12067 (struct _slp_tree): Use a vector of slp_tree as children.
12068 (vect_get_place_in_interleaving_chain): Remove.
12069 * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
12071 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
12073 (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
12074 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
12075 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
12076 vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
12077 Use slp_node instead of slp_void_p and adjust.
12079 2013-04-09 Richard Biener <rguenther@suse.de>
12081 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
12082 work that is not necessary.
12084 2013-04-09 Jakub Jelinek <jakub@redhat.com>
12086 PR tree-optimization/56854
12087 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
12088 forward into clobber stmts if it would change MEM_REF lhs into
12091 2013-04-09 Maxim Kuvyrkov <maxim@kugelworks.com>
12093 * tree.c (type_hash_lookup, type_hash_add): Make static.
12094 * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
12096 2013-04-09 Richard Biener <rguenther@suse.de>
12098 * tree.h (unsave_expr_now): Remove.
12099 * tree-inline.c (mark_local_for_remap_r): Remove.
12100 (unsave_expr_1): Likewise.
12101 (unsave_r): Likewise.
12102 (unsave_expr_now): Likewise.
12103 * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
12104 (propagate_tree_value): Likewise.
12106 2013-04-08 Steven Bosscher <steven@gcc.gnu.org>
12108 * doc/rtl.texi (sequence): Rewrite documentation to match the
12109 current use of SEQUENCE rtl objects.
12110 * rtl.def (SEQUENCE): Likewise.
12112 * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
12113 Update documentation.
12114 (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
12115 NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
12117 * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
12119 2013-04-08 Teresa Johnson <tejohnson@google.com>
12121 * basic-block.h (GCOV_COMPUTE_SCALE): Define.
12122 * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
12124 (estimate_edge_size_and_time): Add comment to suggest using rounding
12126 (estimate_node_size_and_time): Ditto.
12127 (remap_edge_change_prob): Use helper rounding divide methods.
12128 * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
12129 (gimple_mod_pow2_value_transform): Ditto.
12130 (gimple_mod_subtract_transform): Ditto.
12131 (gimple_ic_transform): Ditto.
12132 (gimple_stringops_transform): Ditto.
12133 * stmt.c (conditional_probability): Ditto.
12134 (emit_case_dispatch_table): Ditto.
12135 * lto-cgraph.c (merge_profile_summaries): Ditto.
12136 * tree-optimize.c (execute_fixup_cfg): Ditto.
12137 * cfgcleanup.c (try_forward_edges): Ditto.
12138 * cfgloopmanip.c (scale_loop_profile): Ditto.
12140 (duplicate_loop_to_header_edge): Ditto.
12141 (lv_adjust_loop_entry_edge): Ditto.
12142 * tree-vect-loop.c (vect_transform_loop): Ditto.
12143 * profile.c (compute_branch_probabilities): Ditto.
12144 * cfgbuild.c (compute_outgoing_frequencies): Ditto.
12145 * lto-streamer-in.c (input_cfg): Ditto.
12146 * gimple-streamer-in.c (input_bb): Ditto.
12147 * ipa-cp.c (update_profiling_info): Ditto.
12148 (update_specialized_profile): Ditto.
12149 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
12150 * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
12152 * sched-rgn.c (compute_dom_prob_ps): Ditto.
12153 (compute_trg_info): Ditto.
12154 * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
12155 (purge_dead_edges): Ditto.
12156 * loop-unswitch.c (unswitch_loop): Ditto.
12157 * cgraphclones.c (cgraph_clone_edge): Ditto.
12158 (cgraph_clone_node): Ditto.
12159 * tree-inline.c (copy_bb): Ditto.
12160 (copy_edges_for_bb): Ditto.
12161 (initialize_cfun): Ditto.
12162 (copy_cfg_body): Ditto.
12163 (expand_call_inline): Ditto.
12165 2013-04-08 Kai Tietz <ktietz@redhat.com>
12167 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
12168 TARGET_CYGWIN64 by TARGET_64BIT.
12170 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
12172 * config/epiphany/epiphany.md (GPR_1): New constant.
12173 (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
12174 * config/epiphany/epiphany.c (gen_compare_reg):
12175 For flag_finite_math_only, avoid swapping operands when r0 and/or r1
12176 is already in place.
12177 Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
12178 Don't require being called during rtl expansion; If y operlaps r0,
12180 (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
12181 (epiphany_expand_epilogue): Likewise.
12183 * config/epiphany/epiphany.c (epiphany_select_cc_mode):
12184 Don't use CC_FPmode for ORDERED / UNORDERED.
12185 * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
12187 * config/epiphany/constraints.md (CnL): New constraint.
12188 * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
12189 * config/epiphany/predicates.md (add_operand): Allow 1024.
12191 * config/epiphany/epiphany.md (logical_op): New code iterator.
12192 (op_mnc): New code attribute.
12193 (<op_mnc>_f, mov_f, cstoresi4): New patterns.
12194 (mov_f+1, mov_f+2): New peephole2 patterns.
12196 * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
12197 (cstoresi4): Also allow re-use of zero result when doing a NE
12198 comparison to a non-zero operand.
12199 Use (clobber (scratch)) for first insn if the gpr output is not needed.
12201 * config/epiphany/epiphany.md (<insn_opname>v2si3):
12202 Use gen_addsi3_i / gen_subsi3_i.
12204 2013-04-08 Jakub Jelinek <jakub@redhat.com>
12208 * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
12209 contain anything but clobbers, at most one __builtin_stack_restore,
12210 optionally debug stmts and final resx, and if it has at least one
12211 incoming EH edge. Don't check for SSA_NAME on LHS of a clobber.
12212 (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
12213 Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
12214 which isn't defaut definition, remove them.
12215 (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
12216 instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
12217 * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
12218 with MEM_REF LHS with SSA_NAME address.
12220 2013-04-08 Jeff Law <law@redhat.com>
12222 * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
12224 2013-04-08 Richard Biener <rguenther@suse.de>
12226 * gimple-pretty-print.c (debug_gimple_stmt): Do not print
12228 * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
12229 determined vector type.
12230 (vect_analyze_data_refs): Likewise.
12231 (vect_get_new_vect_var): Adjust.
12232 (vect_create_destination_var): Preserve SSA name versions.
12233 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
12234 not dump anything here.
12236 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
12238 * config/epiphany/epiphany.h (struct GTY (()) machine_function):
12239 Add member lr_slot_known.
12240 * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
12242 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
12243 Remove code that sets lr_slot_offset according to what a previous
12244 version of epiphany_emit_save_restore used to do.
12245 (epiphany_emit_save_restore): When doing an lr save or restore,
12246 set/verify lr_slot_known and lr_slot_offset.
12248 2013-04-08 Xinyu Qi <xyqi@marvell.com>
12251 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
12254 2013-04-08 Richard Biener <rguenther@suse.de>
12256 * alias.c (find_base_term): Fix thinko in previous change.
12258 2013-04-08 Jakub Jelinek <jakub@redhat.com>
12260 * tree-loop-distribution.c (const_with_all_bytes_same): New function.
12261 (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
12262 TYPE_PRECISION is equal to mode bitsize. Use const_with_all_bytes_same
12263 if possible to compute val.
12264 (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
12265 For QImode integers don't require anything about precision. Use
12266 const_with_all_bytes_same to find out if the constant doesn't have
12267 repeated bytes in it.
12269 2013-04-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12271 * config/s390/s390.c (s390_expand_insv): Only accept insertions
12274 2013-04-08 Marek Polacek <polacek@redhat.com>
12276 PR rtl-optimization/48182
12277 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
12280 2013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
12283 * config/pa/pa.c (legitimize_pic_address): Before incrementing label
12284 nuses, make sure we have a label.
12286 2013-04-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12289 * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
12290 (rs6000_emit_swdiv_low_precision): Remove.
12291 (rs6000_emit_swdiv): Rewrite to handle between one and four
12292 iterations of Newton-Raphson generally; modify required number of
12293 iterations for some cases.
12294 * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
12296 2013-04-05 Steven Bosscher <steven@gcc.gnu.org>
12298 * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
12299 set-but-unused variable.
12301 * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
12302 basic blocks of released function bodies garbage-collectable.
12304 * ree.c (find_and_remove_re): Do not call df_finish_pass here.
12305 (struct rtl_opt_pass): Add TODO_df_finish.
12307 * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
12309 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12311 * config/arm/constraints.md (q): New constraint.
12312 * config/arm/ldrdstrd.md: New file.
12313 * config/arm/arm.md (ldrdstrd.md) New include.
12314 (arm_movdi): Use "q" instead of "r" constraint
12315 for double-word memory access.
12316 (movdf_soft_insn): Likewise.
12317 * config/arm/vfp.md (movdi_vfp): Likewise.
12318 * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
12319 * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
12320 * config/arm/arm.c (gen_operands_ldrd_strd): New function.
12321 (mem_ok_for_ldrd_strd): Likewise.
12322 (output_move_double): Update assertion.
12324 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12326 * config/arm/arm.md: Comment on splitting Thumb1 patterns.
12328 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12330 * config/arm/arm.md (arm_smax_insn): Convert define_insn into
12331 define_insn_and_split.
12332 (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
12334 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12336 * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
12337 define_insn_and_split.
12338 (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
12339 (shiftsi3_compare): New pattern.
12340 (rrx): New pattern.
12341 * config/arm/unspecs.md (UNSPEC_RRX): New.
12343 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12345 * config/arm/arm.md (negdi_extendsidi): New pattern.
12346 (negdi_zero_extendsidi): Likewise.
12348 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12350 * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
12351 define_insn_and_split.
12352 (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
12353 (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
12355 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12357 * config/arm/arm.md (arm_subdi3): Convert define_insn into
12358 define_insn_and_split.
12359 (subdi_di_zesidi,subdi_di_sesidi): Likewise.
12360 (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
12362 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12364 * config/arm/arm.md (subsi3_carryin): New pattern.
12365 (subsi3_carryin_const): Likewise.
12366 (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
12367 (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
12369 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12371 * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
12373 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12375 * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
12376 (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
12378 2013-04-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12380 * config/arm/arm.c (arm_expand_builtin): Change fcode
12381 type to unsigned int.
12383 2013-04-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12385 * doc/invoke.texi (ARM Options): Document cortex-a53 support.
12387 2013-04-04 Ian Lance Taylor <iant@google.com>
12389 * doc/standards.texi (Standards): The Go frontend supports the Go 1
12392 2013-04-04 Steven Bosscher <steven@gcc.gnu.org>
12394 PR middle-end/56729
12395 * df-scan.c (df_insn_delete): Disable failing assert.
12397 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12399 * config/arm/arm-protos.h (arm_builtin_vectorized_function):
12400 New function prototype.
12401 * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
12402 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
12403 (arm_builtin_vectorized_function): New function.
12405 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12407 * config/arm/arm_neon_builtins.def: New file.
12408 * config/arm/arm.c (neon_builtin_data): Move contents to
12409 arm_neon_builtins.def.
12410 (enum arm_builtins): Include neon builtin definitions.
12411 (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
12412 * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
12414 2013-04-04 Marek Polacek <polacek@redhat.com>
12416 PR tree-optimization/48186
12417 * predict.c (maybe_hot_frequency_p): Return false if
12418 HOT_BB_FREQUENCY_FRACTION is 0.
12419 (cgraph_maybe_hot_edge_p): Likewise.
12421 2013-04-04 Richard Biener <rguenther@suse.de>
12423 PR tree-optimization/56826
12424 * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
12427 2013-04-04 Richard Biener <rguenther@suse.de>
12429 PR tree-optimization/56213
12430 * tree-vect-data-refs.c (vect_check_strided_load): Remove.
12431 (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
12433 2013-04-04 Richard Biener <rguenther@suse.de>
12435 PR tree-optimization/56837
12436 * tree-loop-distribution.c (classify_partition): For non-zero
12437 values require that the value has the same precision as its
12438 mode to be useful as memset value.
12440 2013-04-03 Nick Clifton <nickc@redhat.com>
12442 * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
12443 (fmssf4): Use fmsf.s on E3V5 architectures.
12444 (fnmasf4): Use fnmaf.s on E3V5 architectures.
12445 (fnmssf4): Use fnmsf.s on E3V5 architectures.
12447 2013-04-03 Jeff Law <law@redhat.com>
12449 * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
12450 (lra-eliminations.o): Likewise.
12452 2013-04-03 Teresa Johnson <tejohnson@google.com>
12454 * gcov-io.c (compute_working_sets): Moved most of body of old
12455 compute_working_sets here from profile.c.
12456 * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
12457 (gcov_working_set_t): Moved typedef here from basic-block.h
12458 (compute_working_set): Declare.
12459 * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
12460 (get_working_sets): Renamed from compute_working_set,
12461 replace most of body with call to new compute_working_sets.
12462 (get_exec_counts): Replace call to compute_working_sets
12463 to get_working_sets.
12464 * profile.h (get_working_sets): Renamed from compute_working_set.
12465 * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
12466 to get_working_sets.
12467 * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
12468 * gcov-dump.c (dump_working_sets): New function.
12470 2013-04-03 Kenneth Zadeck <zadeck@naturalbridge.com>
12472 * hwint.c (sext_hwi, zext_hwi): New functions.
12473 * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
12474 HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
12475 HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
12476 HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
12477 HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
12478 (sext_hwi, zext_hwi): New functions.
12480 2013-04-03 Jeff Law <law@redhat.com>
12482 PR tree-optimization/56799
12483 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
12484 back test for widening conversion erroneously dropped in prior change.
12486 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12489 * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
12490 instead of next_real_insn.
12492 2013-04-03 Marek Polacek <polacek@redhat.com>
12495 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
12497 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12500 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
12502 (thumb1_output_casesi): Likewise.
12503 (thumb2_output_casesi): Likewise.
12505 2013-04-03 Richard Biener <rguenther@suse.de>
12507 PR tree-optimization/56817
12508 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
12510 (tree_unroll_loops_completely_1): ... new function to manually
12511 walk the loop tree, properly defering outer loops of unrolled
12512 loops to later iterations.
12514 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
12516 * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
12517 (vectorizable_load): Likewise.
12518 * tree-vect-slp.c (vect_build_slp_tree): Likewise.
12519 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
12521 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
12523 * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
12526 2013-04-03 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
12528 * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
12530 2013-04-03 Bin Cheng <bin.cheng@arm.com>
12532 * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
12534 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
12536 PR tree-optimization/56790
12537 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
12540 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
12542 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
12544 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
12545 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
12548 2013-04-03 Jakub Jelinek <jakub@redhat.com>
12551 * tree.h (force_folding_builtin_constant_p): New decl.
12552 * builtins.c (force_folding_builtin_constant_p): New variable.
12553 (fold_builtin_constant_p): Fold immediately also if
12554 force_folding_builtin_constant_p.
12556 2013-04-03 Richard Biener <rguenther@suse.de>
12558 PR tree-optimization/56812
12559 * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
12560 DRs of the same interleaving chain are independent.
12562 2013-04-02 Jason Merrill <jason@redhat.com>
12564 * gdbinit.in (pbb): Use debug fn.
12566 2013-04-02 Lawrence Crowl <crowl@google.com>
12568 * sese.h (struct ivtype_map_elt_s): Remove unused.
12569 (extern debug_ivtype_map): Remove unused.
12570 (extern eq_ivtype_map_elts): Remove unused.
12571 * sese.c (debug_ivtype_map): Removed unused.
12572 (debug_ivtype_map_1): Removed unused.
12573 (debug_ivtype_elt): Remove unused.
12574 (eq_ivtype_map_elts): Remove unused.
12577 2013-04-02 Kai Tietz <ktietz@redhat.com>
12580 * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
12581 * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
12582 * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
12584 (legitimize_pe_coff_symbol): Likewise.
12585 (is_imported_p): New helper-function.
12586 (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
12587 for Windows x64 targets.
12588 (ix86_expand_prologue): Optimize for pe-coff targets.
12589 (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
12590 (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
12591 medium/large code-model.
12592 (legitimize_pic_address): Likewise.
12593 (legitimize_tls_address): Likewise.
12594 (ix86_expand_call): Likewise.
12595 (x86_output_mi_thunk): Likewise.
12596 (get_dllimport_decl): Add new beimport argument.
12597 (construct_plt_address): Don't assert for x64 pe-coff targets.
12598 * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
12600 (SYMBOL_FLAG_STUBVAR): New macro.
12601 (SYMBOL_REF_STUBVAR_P): Likewise.
12602 * config/i386/winnt.c (stub_list): New structure.
12603 (stub_head): New local variable.
12604 (i386_pe_record_stub): New function.
12605 (i386_pe_file_end): Emit refptr-stubs.
12607 2013-04-02 Jakub Jelinek <jakub@redhat.com>
12609 PR rtl-optimization/56745
12610 * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
12611 if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
12614 * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
12615 and both of them are MEM_REFs, just compare first argument for
12616 equality and attempt to deal even with differing offsets.
12619 * tree-cfg.c (verify_gimple_assign_single): Allow lhs
12620 of gimple_clobber_p to be MEM_REF.
12621 * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
12622 an assignment from TREE_CLOBBER_P. Allow it to be MEM_REF
12623 after gimplification.
12624 * asan.c (get_mem_ref_of_assignment): Don't instrument
12625 gimple_clobber_p stmts.
12626 * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
12627 gimple_clobber_p stmt if they have MEM_REF lhs and
12628 are dead because of another gimple_clobber_p stmt.
12629 * tree-ssa-live.c (clear_unused_block_pointer): Treat
12630 gimple_clobber_p stmts like debug stmts.
12631 (remove_unused_locals): Remove clobbers with MEM_REF lhs
12632 that refer to unused VAR_DECLs or uninitialized values.
12633 * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
12634 gimple_clobber_p stmts if they refer to removed parameters.
12635 (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
12638 2013-04-02 Uros Bizjak <ubizjak@gmail.com>
12640 * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
12641 using SWI48 mode attribute.
12643 2013-04-02 Wei Mi <wmi@google.com>
12645 * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
12646 ashl<mode>3_mask, *<shift_insn><mode>3_mask and
12647 *<rotate_insn><mode>3_mask in i386.md.
12649 2013-04-02 Alexander Ivchenko <alexander.ivchenko@intel.com>
12651 * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
12653 2013-04-02 Richard Biener <rguenther@suse.de>
12655 PR tree-optimization/56778
12656 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
12657 Runtime alias tests are not supported for gather loads.
12658 * tree-vect-loop-manip.c (vect_loop_versioning): Insert
12659 stmts referenced from SSA operands before updating SSA form.
12661 2013-04-02 Ian Caulfield <ian.caulfield@arm.com>
12662 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12664 * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
12665 * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
12666 * config/arm/cortex-a53.md: New file.
12667 * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
12668 * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
12669 * config/arm/arm.c (arm_issue_rate): Likewise.
12670 * config/arm/arm-tune.md: Regenerate
12671 * config/arm/arm-tables.opt: Regenerate.
12672 * config/arm/arm-cores.def: Add cortex-a53.
12674 2013-04-02 Zhenqiang Chen <zhenqiang.chen@arm.com>
12676 * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
12679 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
12681 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
12682 scalar load/store operations using B/H registers.
12683 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
12685 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
12687 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
12689 * config/aarch64/aarch64.c
12690 (aarch64_simd_scalar_immediate_valid_for_move): New.
12691 * config/aarch64/aarch64-protos.h
12692 (aarch64_simd_scalar_immediate_valid_for_move): New.
12693 * config/aarch64/constraints.md (Dh, Dq): New.
12694 * config/aarch64/iterators.md (hq): New.
12696 2013-04-02 Eric Botcazou <ebotcazou@adacore.com>
12698 * reorg.c (get_branch_condition): Deal with conditional returns.
12699 (fill_simple_delay_slots): Remove dead code dealing with jumps.
12701 2013-04-01 Wei Mi <wmi@google.com>
12703 * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
12704 Truncate operand 2 using %b asm operand modifier.
12705 (*<shift_insn><mode>3_mask): Ditto.
12706 (*<rotate_insn><mode>3_mask): Ditto.
12708 2013-04-01 Steven Bosscher <steven@gcc.gnu.org>
12710 PR middle-end/56798
12711 * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
12713 2013-03-31 Kaz Kojima <kkojima@gcc.gnu.org>
12715 * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
12717 (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
12719 2013-03-30 Lawrence Crowl <crowl@google.com>
12721 * dse.c (clear_alias_sets): Remove never set.
12722 (disqualified_clear_alias_sets): Remove never set.
12723 (clear_alias_mode_pool): Remove never set.
12724 (dse_step0): Remove condition that is never true.
12725 (canon_address): Remove condition that is never true.
12726 (dse_step7): Remove condition that is never true.
12727 (rest_of_handle_dse): Remove condition that is never true.
12728 (rest_of_handle_dse::did_global): Remove never read from above.
12729 (dse_step2_spill): Remove never called from above.
12730 (dse_step5_spill): Remove never called from above.
12732 2013-03-30 Steven Bosscher <steven@gcc.gnu.org>
12734 * doc/md.texi (Standard Names) <casesi>: Update documentation for
12735 JUMP_TABLE_DATA changes.
12736 * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
12737 * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
12738 (Insns) <jump_table_data>: New entry.
12739 * doc/tm.texi: Regenerate.
12741 * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
12743 * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
12744 for table jump at the end of a basic block using tablejump_p.
12745 * targhooks.c (default_invalid_within_doloop): Likewise.
12746 * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
12747 target hook implementation that is identical to the default hook.
12748 (rs6000_invalid_within_doloop): Remove.
12750 * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
12751 unused variable from tablejump_p call.
12753 * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
12754 * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
12755 (INSN_DELETED_P): Likewise.
12756 (emit_jump_table_data): New prototype.
12757 * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
12758 after 4th as unused.
12759 * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
12760 * sched-vis.c (print_insn): Likewise.
12761 * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
12762 insn for compatibility with back ends that use next_active_insn to
12763 identify jump table data.
12764 (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
12765 (remove_insn): Likewise.
12766 (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
12768 (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
12769 (emit_jump_table_data): New function.
12771 * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
12772 basic block, a JUMP_TABLE_DATA never is.
12773 (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
12774 * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
12775 off from code handling real insns.
12776 * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
12777 * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
12778 test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
12779 * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
12780 is not a NONDEBUG_INSN_P.
12781 * ira-costs.c (scan_one_insn): Likewise.
12782 * jump.c (mark_all_labels): Likewise.
12783 (mark_jump_label_1): Likewise.
12784 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12785 * lra.c (get_insn_freq): Expect all insns reaching here to be in
12787 (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
12788 * predict.c (expensive_function_p): Use FOR_BB_INSNS.
12789 * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
12790 JUMP_TABLE_DATA_P insns.
12791 (calculate_elim_costs_all_insns): Likewise.
12792 (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
12793 (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
12794 (delete_output_reload): Code style fixups.
12795 * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
12796 insn flags on this non-insn.
12797 * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
12798 as scheduling barriers, for pre-change compatibility.
12799 * stmt.c (emit_case_dispatch_table): Emit jump table data not as
12800 JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
12802 * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
12803 redundant JUMP_TABLE_DATA_P test.
12804 * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
12805 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12806 (frv_for_each_packet): Likewise.
12807 * config/i386/i386.c (min_insn_size): Likewise.
12808 (ix86_avoid_jump_mispredicts): Likewise.
12809 * config/m32r/m32r.c (m32r_is_insn): Likewise.
12810 * config/mep/mep.c (mep_reorg_erepeat): Likewise.
12811 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12812 (mips16_insn_length): Robustify.
12813 (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
12814 (mips16_split_long_branches): Likewise.
12815 * config/pa/pa.c (pa_combine_instructions): Likewise.
12816 * config/rs6000/rs6000.c (get_next_active_insn): Treat
12817 JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
12818 * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
12819 as contributing to pool range lengths.
12820 * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
12821 Remove redundant JUMP_TABLE_DATA_P test.
12822 (sh_loop_align): Likewise.
12823 (split_branches): Likewise.
12824 (sh_insn_length_adjustment): Likewise.
12825 * config/spu/spu.c (get_branch_target): Likewise.
12827 2013-03-29 Jan Hubicka <jh@suse.cz>
12829 * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
12830 gcov streaming; stream hot bb threshold to ltrans.
12831 * predict.c (get_hot_bb_threshold): Break out from ....
12832 (maybe_hot_count_p): ... here.
12833 (set_hot_bb_threshold): New function.
12834 * lto-section-in.c (lto_section_name): Add profile.
12835 * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
12836 * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
12837 and data-streamer.h
12838 (histogram_entry): New structure.
12839 (histogram, histogram_pool): New global vars.
12840 (histogram_hash): New structure.
12841 (histogram_hash::hash): New method.
12842 (histogram_hash::equal): Likewise.
12843 (account_time_size): New function.
12844 (cmp_counts): New function.
12845 (dump_histogram): New function.
12846 (ipa_profile_generate_summary): New function.
12847 (ipa_profile_write_summary): New function.
12848 (ipa_profile_read_summary): New function.
12849 (ipa_profile): Decide on threshold.
12850 (pass_ipa_profile): Add ipa_profile_write_summary and
12851 ipa_profile_read_summary.
12852 * Makefile.in (ipa.o): Update dependencies.
12853 * lto-streamer.h (LTO_section_ipa_profile): New section.
12855 2013-03-29 Gabriel Dos Reis <gdr@integrable-solutions.net>
12857 * tree.h (VAR_P): New.
12859 2013-03-29 Paolo Carlini <paolo.carlini@oracle.com>
12862 * doc/invoke.texi ([-fwhole-program]): Fix typo.
12864 2013-03-29 Steven Bosscher <steven@gcc.gnu.org>
12866 * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
12867 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12868 (control_flow_insn_p): Likewise.
12869 * cfgrtl.c (duplicate_insn_chain): Likewise.
12870 * final.c (get_attr_length_1): Likewise.
12871 (shorten_branches): Likewise.
12872 (final_scan_insn): Likewise.
12873 * function.c (instantiate_virtual_regs): Likewise.
12874 * gcse.c (insert_insn_end_basic_block): Likewise.
12875 * ira-costs.c (scan_one_insn): Likewise.
12876 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
12877 * lra.c (check_rtl): Likewise.
12878 * reload1.c (elimination_costs_in_insn): Likewise.
12879 * reorg.c (follow_jumps): Likewise.
12881 * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
12882 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
12883 (thumb_far_jump_used_p): Likewise.
12884 * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
12885 (workaround_speculation): Likewise.
12886 (add_sched_insns_for_speculation): Likewise.
12887 * config/c6x/c6x.c (reorg_emit_nops): Likewise.
12888 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
12889 (frv_for_each_packet): Likewise.
12890 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
12891 * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
12892 (final_emit_insn_group_barriers): Likewise.
12893 * config/m32r/m32r.c (m32r_is_insn): Likewise.
12894 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
12895 (mips16_insn_length): Likewise.
12896 * config/pa/pa.c (pa_reorg): Likewise.
12897 (pa_combine_instructions): Likewise.
12898 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
12899 * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
12900 (sh_reorg): Likewise.
12901 (split_branches): Likewise.
12902 * config/spu/spu.c (get_branch_target): Likewise.
12904 * config/s390/s390.c (s390_chunkify_start): Simplify logic using
12907 2013-03-29 Kirill Yukhin <kirill.yukhin@intel.com>
12909 * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
12910 Fix declaration name.
12912 2013-03-28 Lawrence Crowl <crowl@google.com>
12914 * graphds.h (struct graph.indicies): Remove unused.
12915 * graphite-poly.h (struct graph.original_pddrs): Remove unused.
12916 (SCOP_ORIGINAL_PDDRS): Remove unused.
12917 * sese.h (extern insert_loop_close_phis): Removed unused.
12918 (extern insert_guard_phis): Removed unused.
12919 (extern ivtype_map_elt_info): Removed unused.
12920 (new_ivtype_map_elt): Removed unused.
12921 * sese.c (ivtype_map_elt_info): Removed unused.
12923 2013-03-28 Lawrence Crowl <crowl@google.com>
12925 * Makefile.in: Add several missing include dependences.
12927 (test-dump.o): New. This object is not added to any executable,
12928 but is present for ad-hoc testing.
12930 (debug (const bitmap_head_def &)): New.
12931 (debug (const bitmap_head_def *)): New.
12933 (extern debug (const bitmap_head_def &)): New.
12934 (extern debug (const bitmap_head_def *)): New.
12936 (debug (edge_def &)): New.
12937 (debug (edge_def *)): New.
12939 (debug (basic_block_def &)): New.
12940 (debug (basic_block_def *)): New.
12942 (dump_node (const_tree, int, FILE *)): Correct source file.
12944 (debug (die_struct &)): New.
12945 (debug (die_struct *)): New.
12947 (extern debug (die_struct &)): New.
12948 (extern debug (die_struct *)): New.
12949 * gimple-pretty-print.c
12950 (debug (gimple_statement_d &)): New.
12951 (debug (gimple_statement_d *)): New.
12952 * gimple-pretty-print.h
12953 (extern debug (gimple_statement_d &)): New.
12954 (extern debug (gimple_statement_d *)): New.
12956 (debug (ira_allocno_copy &)): New.
12957 (debug (ira_allocno_copy *)): New.
12958 (debug (ira_allocno &)): New.
12959 (debug (ira_allocno *)): New.
12961 (extern debug (ira_allocno_copy &)): New.
12962 (extern debug (ira_allocno_copy *)): New.
12963 (extern debug (ira_allocno &)): New.
12964 (extern debug (ira_allocno *)): New.
12966 (debug (live_range &)): New.
12967 (debug (live_range *)): New.
12969 (debug (lra_live_range &)): New.
12970 (debug (lra_live_range *)): New.
12972 (debug (lra_live_range &)): New.
12973 (debug (lra_live_range *)): New.
12975 (debug (omega_pb_d &)): New.
12976 (debug (omega_pb_d *)): New.
12978 (extern debug (omega_pb_d &)): New.
12979 (extern debug (omega_pb_d *)): New.
12981 (debug (const rtx_def &)): New.
12982 (debug (const rtx_def *)): New.
12984 (debug_tree (tree): Move within file.
12985 (debug_raw (const tree_node &)): New.
12986 (debug_raw (const tree_node *)): New.
12987 (dump_tree_via_hooks (const tree_node *, int)): New.
12988 (debug (const tree_node &)): New.
12989 (debug (const tree_node *)): New.
12990 (debug_verbose (const tree_node &)): New.
12991 (debug_verbose (const tree_node *)): New.
12992 (debug_head (const tree_node &)): New.
12993 (debug_head (const tree_node *)): New.
12994 (debug_body (const tree_node &)): New.
12995 (debug_body (const tree_node *)): New.
12996 (debug_vec_tree (tree): Move and reimplement in terms of dump.
12997 (debug (vec<tree, va_gc> &)): New.
12998 (debug (vec<tree, va_gc> *)): New.
13000 (extern debug (const rtx_def &)): New.
13001 (extern debug (const rtx_def *)): New.
13003 (debug_raw (simple_bitmap_def &)): New.
13004 (debug_raw (simple_bitmap_def *)): New.
13005 (debug (simple_bitmap_def &)): New.
13006 (debug (simple_bitmap_def *)): New.
13008 (extern debug (simple_bitmap_def &)): New.
13009 (extern debug (simple_bitmap_def *)): New.
13010 (extern debug_raw (simple_bitmap_def &)): New.
13011 (extern debug_raw (simple_bitmap_def *)): New.
13013 (debug (vinsn_def &)): New.
13014 (debug (vinsn_def *)): New.
13015 (debug_verbose (vinsn_def &)): New.
13016 (debug_verbose (vinsn_def *)): New.
13017 (debug (expr_def &)): New.
13018 (debug (expr_def *)): New.
13019 (debug_verbose (expr_def &)): New.
13020 (debug_verbose (expr_def *)): New.
13021 (debug (vec<rtx> &)): New.
13022 (debug (vec<rtx> *)): New.
13024 (extern debug (vinsn_def &)): New.
13025 (extern debug (vinsn_def *)): New.
13026 (extern debug_verbose (vinsn_def &)): New.
13027 (extern debug_verbose (vinsn_def *)): New.
13028 (extern debug (expr_def &)): New.
13029 (extern debug (expr_def *)): New.
13030 (extern debug_verbose (expr_def &)): New.
13031 (extern debug_verbose (expr_def *)): New.
13032 (extern debug (vec<rtx> &)): New.
13033 (extern debug (vec<rtx> *)): New.
13035 (_list_iter_cond_expr): Make inline instead of static.
13037 (debug (sreal &)): New.
13038 (debug (sreal *)): New.
13040 (extern debug (sreal &)): New.
13041 (extern debug (sreal *)): New.
13043 (extern debug_raw (const tree_node &)): New.
13044 (extern debug_raw (const tree_node *)): New.
13045 (extern debug (const tree_node &)): New.
13046 (extern debug (const tree_node *)): New.
13047 (extern debug_verbose (const tree_node &)): New.
13048 (extern debug_verbose (const tree_node *)): New.
13049 (extern debug_head (const tree_node &)): New.
13050 (extern debug_head (const tree_node *)): New.
13051 (extern debug_body (const tree_node &)): New.
13052 (extern debug_body (const tree_node *)): New.
13053 (extern debug (vec<tree, va_gc> &)): New.
13054 (extern debug (vec<tree, va_gc> *)): New.
13056 (debug (struct loop &)): New.
13057 (debug (struct loop *)): New.
13058 (debug_verbose (struct loop &)): New.
13059 (debug_verbose (struct loop *)): New.
13060 * tree-dump.c: Add header dependence.
13062 (extern debug (struct loop &)): New.
13063 (extern debug (struct loop *)): New.
13064 (extern debug_verbose (struct loop &)): New.
13065 (extern debug_verbose (struct loop *)): New.
13067 (debug (data_reference &)): New.
13068 (debug (data_reference *)): New.
13069 (debug (vec<data_reference_p> &)): New.
13070 (debug (vec<data_reference_p> *)): New.
13071 (debug (vec<ddr_p> &)): New.
13072 (debug (vec<ddr_p> *)): New.
13074 (extern debug (data_reference &)): New.
13075 (extern debug (data_reference *)): New.
13076 (extern debug (vec<data_reference_p> &)): New.
13077 (extern debug (vec<data_reference_p> *)): New.
13078 (extern debug (vec<ddr_p> &)): New.
13079 (extern debug (vec<ddr_p> *)): New.
13081 (debug (pt_solution &)): New.
13082 (debug (pt_solution *)): New.
13084 (extern debug (pt_solution &)): New.
13085 (extern debug (pt_solution *)): New.
13087 (debug (_var_map &)): New.
13088 (debug (_var_map *)): New.
13089 (debug (tree_live_info_d &)): New.
13090 (debug (tree_live_info_d *)): New.
13092 (extern debug (_var_map &)): New.
13093 (extern debug (_var_map *)): New.
13094 (extern debug (tree_live_info_d &)): New.
13095 (extern debug (tree_live_info_d *)): New.
13097 2013-03-28 Jan Hubicka <jh@suse.cz>
13099 * lto-cgraph.c (merge_profile_summaries): Fix overflows.
13101 2013-03-28 Ian Bolton <ian.bolton@arm.com>
13103 * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
13104 record only when desired or required.
13106 2013-03-28 Uros Bizjak <ubizjak@gmail.com>
13108 * config/i386/i386.md (*vec_extract2vdi_1): Merge with
13109 *vec_extractv2di_1_rex64. Use x64 isa attribute.
13111 2013-03-28 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
13113 * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
13114 (*andsi3_compare0_uxtw): New pattern.
13115 (*and_<SHIFT:optab><mode>3_compare0): New pattern.
13116 (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
13118 2013-03-28 Jan Hubicka <jh@suse.cz>
13120 * data-streamer-in.c (streamer_read_gcov_count): New function.
13121 * gimple-streamer-out.c: Include value-prof.h.
13122 (output_gimple_stmt): Output histogram.
13123 (output_bb): Use streamer_write_gcov_count.
13124 * value-prof.c: Include data-streamer.h
13125 (dump_histogram_value): Add HIST_TYPE_MAX.
13126 (stream_out_histogram_value): New function.
13127 (stream_in_histogram_value): New function.
13128 * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
13129 (stream_out_histogram_value, stream_in_histogram_value): Declare.
13130 * data-streamer-out.c (streamer_write_gcov_count): New function.
13131 (streamer_write_gcov_count_stream): New function.
13132 * lto-cgraph.c (lto_output_edge): Update counter streaming.
13133 (lto_output_node): Likewise.
13134 (input_node, input_edge): Likewise.
13135 * lto-streamer-out.c (output_cfg): Update streaming.
13136 * lto-streamer-in.c (input_cfg): Likewise.
13137 * data-streamer.h (streamer_write_gcov_count,
13138 streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
13139 * gimple-streamer-in.c: Include value-prof.h
13140 (input_gimple_stmt): Input histograms.
13141 (input_bb): Update profile streaming.
13143 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
13145 * genmodes.c (emit_max_int): New function.
13146 (emit_insn_modes_h): Added call to emit_max_function.
13147 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
13149 * machmode.def: Fixed comment.
13151 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
13153 * combine.c (try_combine): Removed useless assert.
13154 * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
13156 2013-03-28 Marek Polacek <polacek@redhat.com>
13157 Richard Biener <rguenther@suse.de>
13159 PR tree-optimization/56695
13160 * tree-vect-stmts.c (vectorizable_condition): Unconditionally
13161 build signed result of a vector comparison.
13162 * tree-cfg.c (verify_gimple_comparison): Check that a result
13163 of a vector comparison has signed type.
13165 2013-03-28 Richard Biener <rguenther@suse.de>
13167 PR tree-optimization/37021
13168 * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
13169 do not restrict gaps between groups.
13170 * tree-vect-stmts.c (vectorizable_load): Properly account for
13171 a gap between groups.
13173 2013-03-28 Eric Botcazou <ebotcazou@adacore.com>
13175 * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
13176 general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
13179 2013-03-27 Gerald Pfeifer <gerald@pfeifer.com>
13181 * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
13182 * doc/extend.texi (Named Address Spaces): Ditto.
13183 (Variable Attributes): Ditto.
13185 2013-03-27 Kai Tietz <ktietz@redhat.com>
13187 * config.build: Add support for cygwin x64 target.
13188 * config.gcc: Likewise.
13189 * config.host: Likewise.
13190 * configure.ac: Likewise
13191 * configure: Regenerated.
13193 2013-03-27 Kai Tietz <ktietz@redhat.com>
13195 * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
13196 * config/i386/t-cygwin-w64: New file.
13197 * config/i386/cygwin-w64.h: New file.
13198 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
13199 and add support for x64-cygwin target.
13200 (CPP_SPEC): Likewise.
13201 (CXX_WRAP_SPEC_LIST): Undefine before define.
13202 (LIBGCJ_SONAME): Use 15 as version.
13204 2013-03-27 Richard Biener <rguenther@suse.de>
13206 PR tree-optimization/56716
13207 * tree-ssa-structalias.c (perform_var_substitution): Adjust
13208 dumping for ref nodes.
13210 2013-03-27 Martin Jambor <mjambor@suse.cz>
13212 PR tree-optimization/55334
13213 * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
13214 restricted pointers to arrays.
13216 2013-03-27 Gabriel Dos Reis <gdr@integrable-solutions.net>
13218 * Makefile.in (.SUFFIXES): Add .cc.
13219 (.c.o): Apply same recipe for implicit rule .cc.o.
13221 2013-03-27 Richard Biener <rguenther@suse.de>
13223 PR tree-optimization/37021
13224 * tree-vect-data-refs.c (vect_check_strided_load): Allow
13225 REALPART/IMAGPART_EXPRs around the supported refs.
13226 * tree-ssa-structalias.c (find_func_aliases): Assume that
13227 floating-point values are not used to transfer pointers.
13229 2013-03-27 Alexander Ivchenko <alexander.ivchenko@intel.com>
13231 * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
13232 * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
13233 * gcc/doc/tm.texi: Regenerate.
13234 * gcc/targhooks.h (default_has_ifunc_p): New.
13235 * gcc/targhooks.c (default_has_ifunc_p): Ditto.
13236 * gcc/config/linux-protos.h: New file.
13237 * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
13238 this hook for linux which disables support of indirect functions in
13240 * gcc/config/linux-android.c: New file.
13241 * gcc/config/t-linux-android.c: Ditto.
13242 * gcc/config.gcc: Added new object file linux-android.o.
13243 * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
13244 Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
13245 * gcc/varasm.c (do_assemble_alias): Likewise.
13246 * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
13247 doesn't support indirect functions.
13248 * configure: Regenerate.
13250 2013-03-27 Bin Cheng <bin.cheng@arm.com>
13253 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
13254 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
13257 2013-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
13259 * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
13261 2013-03-27 Terry Guo <terry.guo@arm.com>
13263 * config/arm/arm-cores.def: Added core cortex-r7.
13264 * config/arm/arm-tune.md: Regenerated.
13265 * config/arm/arm-tables.opt: Regenerated.
13266 * doc/invoke.texi: Added entry for core cortex-r7.
13268 2013-03-27 Walter Lee <walt@tilera.com>
13270 * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
13271 double-decrement of next_scratch_regno.
13273 2013-03-27 Walter Lee <walt@tilera.com>
13275 * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
13277 (insn_v1mulus): Ditto.
13278 (insn_v2muls): Ditto.
13280 2013-03-27 Walter Lee <walt@tilera.com>
13282 * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
13283 (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
13285 2013-03-27 Walter Lee <walt@tilera.com>
13287 * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
13288 (*sibcall_value): Ditto.
13290 2013-03-27 Walter Lee <walt@tilera.com>
13292 * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
13293 (insn_mnz_v8qi): ... this ...
13294 (insn_mnz_v4hi): ... and this. Replace (const_int 0) with the
13296 (insn_v<n>mnz): Replaced by ...
13297 (insn_v1mnz): ... this ...
13298 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
13300 (insn_mz_<mode>): Replaced by ...
13301 (insn_mz_v8qi): ... this ...
13302 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
13304 (insn_v<n>mz): Replaced by ...
13305 (insn_v1mz): ... this ...
13306 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
13309 2013-03-26 Eric Botcazou <ebotcazou@adacore.com>
13311 * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
13313 2013-03-26 Roland McGrath <mcgrathr@google.com>
13315 * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
13316 than fprintf with a non-constant, non-format string.
13318 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
13320 * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
13321 using nox64 isa attribute. Use nonimmediate_x86nomem_operand as
13322 operand 0 predicate.
13323 (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
13324 attribute. Use general_x64nomem_operand as operand 1 predicate.
13325 (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
13326 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
13327 (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
13328 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
13329 (mov<mode>_insv_1): Remove expander. Merge insn with
13330 movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
13331 Use general_x64nomem_operand as operand 1 predicate.
13332 (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
13333 (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
13334 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
13335 (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
13336 attribute. Use nonimmediate_x64nomem_operand as operand 2 predicate.
13337 (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
13338 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
13339 (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
13340 isa attribute. Use general_x64nomem_operand as operand 2 predicate.
13341 * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
13342 (general_x64nomem_operand): Ditto.
13344 2013-03-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13346 * config/rtems.opt: Add -pthread option.
13348 2013-03-26 Richard Biener <rguenther@suse.de>
13350 * alias.c (find_base_term): Avoid redundant and not used recursion.
13351 (base_alias_check): Get the initial base term from the caller.
13352 (true_dependence_1): Compute and pass base terms to base_alias_check.
13353 (write_dependence_p): Likewise.
13354 (may_alias_p): Likewise.
13356 2013-03-26 Sofiane Naci <sofiane.naci@arm.com>
13358 * config/aarch64/aarch64.c (aarch64_classify_address): Support
13359 PC-relative load in SI modes and above only.
13361 2013-03-26 Xinyu Qi <xyqi@marvell.com>
13363 * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
13364 * config/arm/iwmmxt.md (WCGR0): Update.
13365 (WCGR1, WCGR2, WCGR3): Likewise.
13367 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
13369 * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
13370 Use x64 and nox64 isa attributes.
13372 2013-03-26 Richard Biener <rguenther@suse.de>
13374 * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
13375 alignment computations and rely on get_object_alignment_1
13376 for the !TYPE_P case.
13377 Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
13379 2013-03-26 Walter Lee <walt@tilera.com>
13381 * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
13382 * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
13384 2013-03-25 Jeff Law <law@redhat.com>
13386 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
13387 check for INTEGRAL_TYPE_P that was missing due to checking in
13388 wrong version of prior patch.
13390 2013-03-25 Walter Lee <walt@tilera.com>
13392 * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
13393 TILEGX_INSN_SHUFFLEBYTES1.
13394 * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
13396 (tilegx_builtins): Ditto.
13397 * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
13399 2013-03-25 Walter Lee <walt@tilera.com>
13401 * config/tilegx/tilegx.md (floatsisf2): New pattern.
13402 (floatunssisf2): New pattern.
13403 (floatsidf2): New pattern.
13404 (floatunssidf2): New pattern.
13406 2013-03-25 Walter Lee <walt@tilera.com>
13408 * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
13409 tests for constraint J, K, N, P.
13411 2013-03-25 Walter Lee <walt@tilera.com>
13413 * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
13414 Use indirect/pcrel encoding.
13415 * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
13418 2013-03-25 Steve Ellcey <sellcey@mips.com>
13420 * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
13421 74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
13422 * config/mips/mips.c (mips_option_override): Set IMADD default.
13423 * config/mips/mips.h (PTF_AVOID_IMADD): New.
13424 (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
13425 (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
13426 * config/mips/mips.md (mimadd): New flag for integer madd/msub.
13427 * doc/invoke.texi (-mimadd/-mno-imadd): New.
13429 2013-03-25 Jeff Law <law@redhat.com>
13431 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
13432 slightly to avoid creating and folding useless trees. Simplify
13433 slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
13435 2013-03-25 Uros Bizjak <ubizjak@gmail.com>
13437 * config/i386/i386.md (*zero_extendsidi2): Merge with
13438 *zero_extendsidi2_rex64. Use x64 and nox64 isa attributes.
13439 * config/i386/predicates.md (x86_64_zext_operand): Rename from
13440 x86_64_zext_general_operand. Use nonimmediate_operand on 32bit
13441 targets. Clarify comment.
13443 2013-03-25 Martin Jambor <mjambor@suse.cz>
13445 * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
13446 pass-through jump functions differently.
13447 (ipa_read_jump_function): Likewise. Also use setter functions to set
13450 2013-03-25 Martin Jambor <mjambor@suse.cz>
13452 * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
13453 ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
13455 (ipa_get_indirect_edge_target): New function.
13456 (devirtualization_time_bonus): New parameter known_aggs, pass it to
13457 ipa_get_indirect_edge_target. Update all callers.
13458 (ipcp_discover_new_direct_edges): New parameter aggvals. Pass it to
13459 ipa_get_indirect_edge_target_1 instead of calling
13460 ipa_get_indirect_edge_target.
13461 (create_specialized_node): Pass aggvlas to
13462 ipcp_discover_new_direct_edges.
13464 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13466 * config/arm/arm.md (f_sels, f_seld): New types.
13467 (*cmov<mode>): New pattern.
13468 * config/arm/predicates.md (arm_vsel_comparison_operator): New
13471 2013-03-25 Kai Tietz <ktietz@redhat.com>
13473 * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
13474 POSIX-printf for mingw-hosted builds.
13476 2013-03-25 Richard Biener <rguenther@suse.de>
13478 PR middle-end/56694
13479 * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
13480 must-not-throw stmt location.
13482 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13484 * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
13485 Emit load-acquire versions when acq is true.
13486 (arm_emit_store_exclusive): Add rel parameter.
13487 Emit store-release versions when rel is true.
13488 (arm_split_compare_and_swap): Use acquire-release instructions
13490 of barriers when appropriate.
13491 (arm_split_atomic_op): Likewise.
13492 * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
13493 * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
13494 (VUNSPEC_SLX): Likewise.
13495 (VUNSPEC_LDA): Likewise.
13496 (VUNSPEC_STL): Likewise.
13497 * config/arm/sync.md (atomic_load<mode>): New pattern.
13498 (atomic_store<mode>): Likewise.
13499 (arm_load_acquire_exclusive<mode>): Likewise.
13500 (arm_load_acquire_exclusivesi): Likewise.
13501 (arm_load_acquire_exclusivedi): Likewise.
13502 (arm_store_release_exclusive<mode>): Likewise.
13504 2013-03-25 Catherine Moore <clm@codesourcery.com>
13506 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
13507 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
13508 * config/mip/predicates.md (lwsp_swsp_operand,
13509 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
13510 sb16_operand, db4_operand, db7_operand, ib3_operand,
13511 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
13512 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
13513 andi16_operand): New predicates.
13514 * config/mips/mips.md (compression): New attribute.
13515 (enabled): New attribute.
13516 (length): Consider compression in computing length.
13517 (shift_compression): New code attribute.
13518 (*add<mode>3): New operands. Record compression.
13519 (sub<mode>3): Likewise.
13520 (one_cmpl<mode>2): Likewise.
13521 (*and<mode>3): Likewise.
13522 (*ior<mode>3): Likewise.
13523 (unnamed pattern for xor): Likewise.
13524 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
13525 (*<optab><mode>3): Likewise.
13526 (*mov<mode>_internal: Likewise.
13527 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
13528 (mips_unsigned_immediate_p): New.
13529 (umips_lwsp_swsp_address_p): New.
13530 (m16_based_address_p): New.
13531 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
13532 (mips_unsigned_immediate_p): New prototype.
13533 (lwsp_swsp_address_p): New prototype.
13534 (m16_based_address_p): New prototype.
13535 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
13536 (mips_signed_immediate_p): New function.
13537 (m16_based_address_p): New function.
13538 (lwsp_swsp_address_p): New function.
13539 (mips_print_operand_punctuation): Recognize short delay slot insns
13540 for microMIPS.add<mode>3"
13542 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13545 * config/arm/iterators.md (v_cmp_result): New mode attribute.
13546 * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
13548 2013-03-25 Richard Biener <rguenther@suse.de>
13550 PR tree-optimization/56689
13551 * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
13554 2013-03-25 Richard Biener <rguenther@suse.de>
13556 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
13558 (memory_references): Likewise.
13559 (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
13560 gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
13561 ref_indep_loop_p_2, find_refs_for_sm): Adjust.
13562 (gather_mem_refs_in_loops): Fold into ...
13563 (analyze_memory_references): ... this. Move initialization
13564 to tree_ssa_lim_initialize.
13565 (fill_always_executed_in): Rename to ...
13566 (fill_always_executed_in_1): ... this.
13567 (fill_always_executed_in): Move contains_call computation to
13568 this new function from ...
13569 (tree_ssa_lim_initialize): ... here.
13570 (tree_ssa_lim): Call fill_always_executed_in.
13572 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
13574 * postreload.c (reload_combine): Fix code detecting returns.
13576 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
13578 * function.c (emit_use_return_register_into_block): On cc0 targets,
13579 do not emit the sequence between cc0 setter and user.
13581 2013-03-25 Kai Tietz <ktietz@redhat.com>
13583 * config/i386/predicates.md (local_symbolic_operand): Interpret
13584 dll-imported symbols as none-local.
13586 2013-03-25 Richard Biener <rguenther@suse.de>
13588 * tree-ssa-loop-im.c (struct depend): Remove.
13589 (struct lim_aux_data): Make depends a vec of gimples.
13590 (free_lim_aux_data): Adjust.
13591 (add_dependency): Likewise.
13592 (set_level): Likewise.
13594 2013-03-25 Richard Biener <rguenther@suse.de>
13596 PR middle-end/56434
13597 * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
13598 the pointer returned by calls with ECF_MALLOC set.
13600 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
13602 * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
13604 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
13606 * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
13607 using MMXMODE mode iterator.
13608 (*move<mode>_internal): Merge with *movv2sf_internal and
13609 *movv2sf_internal_rex64 using MMXMODE mode iterator.
13611 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
13613 * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
13614 (record_last_mem_set_info): Likewise.
13616 * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
13617 of XNEWVEC followed by memset.
13618 (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
13620 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
13622 * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
13623 config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
13624 config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
13625 config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
13626 config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
13627 config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
13628 dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
13629 the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
13630 BARRIER_P instead of GET_CODE.
13632 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
13634 * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
13635 inaccuracy in the probing code.
13637 * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
13638 (ctrapdi4): Likewise.
13640 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
13642 * calls.c (expand_call): Add missing guard to code handling return
13643 of non-BLKmode structures in MSB.
13644 * function.c (expand_function_end): Likewise.
13646 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
13648 * combine.c (try_combine): Adjust comment. Do not add the set of
13649 insn #0 if the destination indirectly is set or dies in insn #2.
13650 Tidy up code to distribute a new note.
13652 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
13654 * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
13655 also for alternatives 16 and 17.
13657 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
13659 * config/i386/sse.md (*mov<mode>_internal): Merge with
13660 *mov<mode>_internal_rex64. Use x64 and nox64 isa attributes.
13661 Emit insn template depending on type attribute. Use
13662 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13663 movd instead of movq mnemonic for interunit moves. Rewrite mode
13664 attribute calculation. Remove unit attribute calculation.
13665 Set prefix attribute to maybe_vex for sselog1 and ssemov types.
13666 Set prefix_data16 attribute for DImode ssemov types.
13667 Use Ym instead of y for SSE-MMX conversion alternatives.
13668 Reorder operand constraints.
13670 2013-03-22 Steven Bosscher <steven@gcc.gnu.org>
13672 * df.h (df_insn_delete): Adjust prototype.
13673 * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
13674 and let it decide whether mark the basic block dirty.
13675 (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
13676 * df-scan.c (df_insn_info_delete): New helper function, split
13677 off from df_insn_delete.
13678 (df_scan_free_bb_info): Use it.
13679 (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
13681 (df_insn_delete): Likewise. Take insn rtx as argument. Verify
13682 that the insn is actually an insn and it has a non-NULL basic block.
13683 Do not mark basic block dirty if only deleting a DEBUG_INSN.
13685 2013-03-22 Richard Biener <rguenther@suse.de>
13687 * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
13689 (mem_ref_alloc): Do not allocate them.
13690 (refs_independent_p): Do not query or maintain a cache.
13692 2013-03-22 Richard Biener <rguenther@suse.de>
13694 * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
13695 (gather_mem_refs_in_loops): Do not compute it.
13696 (analyze_memory_references): Do not allocate it.
13697 (tree_ssa_lim_finalize): Do not free it.
13698 (for_all_locs_in_loop): Do not query all_refs_in_loop.
13700 2013-03-22 Richard Biener <rguenther@suse.de>
13702 * is-a.h (as_a): Use gcc_checking_assert.
13704 2013-03-22 Ian Bolton <ian.bolton@arm.com>
13706 * config/aarch64/aarch64.c (aarch64_print_operand): New
13707 format specifier for printing a constant in hex.
13708 * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
13709 format specifier for printing second operand.
13711 2013-03-22 Richard Biener <rguenther@suse.de>
13713 * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
13715 (gather_mem_refs_in_loops): Perform store accumulation here.
13716 (create_vop_ref_mapping_loop): Remove.
13717 (create_vop_ref_mapping): Likewise.
13718 (analyze_memory_references): Initialize refs_stored_in_loop.
13719 (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
13720 (record_indep_loop): Remove.
13721 (record_dep_loop): New function.
13722 (ref_indep_loop_p_1): Adjust to only walk over references
13723 in the loop, not its subloops.
13724 (ref_indep_loop_p): Rename to ...
13725 (ref_indep_loop_p_2): ... this and recurse over the loop tree,
13726 maintaining a more fine-grained cache.
13727 (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
13728 (tree_ssa_lim_finalize): Free refs_stored_in_loop.
13730 2013-03-22 Richard Biener <rguenther@suse.de>
13732 * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
13733 (struct mem_ref): Make accesses_in_loop a vec of a vec of
13734 aggregate mem_ref_loc.
13735 (free_mem_ref_locs): Inline into ...
13736 (memref_free): ... this and adjust.
13737 (mem_ref_alloc): Adjust.
13738 (mem_ref_locs_alloc): Remove.
13739 (record_mem_ref_loc): Adjust.
13740 (get_all_locs_in_loop): Rewrite into ...
13741 (for_all_locs_in_loop): ... this iterator.
13742 (rewrite_mem_ref_loc): New functor.
13743 (rewrite_mem_refs): Use for_all_locs_in_loop.
13744 (sm_set_flag_if_changed): New functor.
13745 (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
13746 (ref_always_accessed): New functor.
13747 (ref_always_accessed_p): Use for_all_locs_in_loop.
13749 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
13751 * tree-pass.h (PROP_gimple_lvec): New.
13752 * passes.c (dump_properties): Handle PROP_gimple_lvec.
13753 (init_optimization_passes): Move pass_lower_vector.
13754 * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
13756 (pass_lower_vector): Provide PROP_gimple_lvec.
13757 (pass_lower_vector_ssa): Likewise.
13758 * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
13760 2013-03-21 Mark Wielaard <mjw@redhat.com>
13762 * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
13764 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
13766 * config/i386/i386.md (*movdi_internal): Disparage slightly
13767 all MMX moves to/from memory. Use Yi instead of x for SSE-MMX
13768 conversion alternatives.
13770 2013-03-21 Jakub Jelinek <jakub@redhat.com>
13772 PR middle-end/48087
13773 * diagnostic.def (DK_WERROR): New kind.
13774 * diagnostic.h (werrorcount): Define.
13775 * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
13776 promoted to DK_ERROR, increment DK_WERROR counter instead of
13778 * toplev.c (toplev_main): Call print_ignored_options even if
13779 just werrorcount is non-zero. Exit with FATAL_EXIT_CODE
13780 even if just werrorcount is non-zero.
13783 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
13785 (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
13786 (string_cst_pool_decl): New function.
13787 (optimize_one_addr_into_implicit_ptr): New function.
13788 (resolve_addr_in_expr): Optimize DWARF location expression
13789 DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
13790 which doesn't live in memory, but has DW_AT_location or
13791 DW_AT_const_value, or refers to a string literal, into
13792 DW_OP_GNU_implicit_pointer.
13793 (optimize_location_into_implicit_ptr): New function.
13794 (resolve_addr): If removing DW_AT_location of a variable because
13795 it was DW_OP_addr of address of the variable, but the variable doesn't
13796 live in memory, try to emit const value attribute for the initializer.
13798 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
13800 * tree.h (VECTOR_TYPE_P): New macro.
13801 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
13802 TYPE_MODE): Use it.
13803 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
13804 VEC_COND_EXPR cannot be lvalues.
13805 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
13807 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
13809 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
13810 Restrict the transformation to equal modes.
13812 2013-03-21 Richard Biener <rguenther@suse.de>
13814 PR tree-optimization/39326
13815 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
13816 (MEM_ANALYZABLE): Adjust.
13817 (record_mem_ref_loc): Move bitmap ops ...
13818 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
13819 unanalyzable refs, do not record locations for it.
13820 (analyze_memory_references): Allocate ref zero as shared
13822 (refs_independent_p): Do not test for unanalyzed mems here.
13823 (ref_indep_loop_p_1): Special-case disambiguation against
13824 the unanalyzed ref.
13825 (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
13827 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
13829 * config/arm/arm-protos.h (tune_params): Add
13830 prefer_neon_for_64bits field.
13831 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
13832 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
13833 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
13834 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
13835 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
13836 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
13837 (arm_option_override): Handle -mneon-for-64bits new option.
13838 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
13839 (prefer_neon_for_64bits): Declare new variable.
13840 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
13841 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
13842 (arch_enabled): Handle new arch types. Remove support for onlya8
13844 (one_cmpldi2): Use new arch names.
13845 (zero_extend<mode>di2, extend<mode>di2): Ditto.
13846 * config/arm/arm.opt (mneon-for-64bits): Add option.
13847 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
13848 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
13849 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
13851 * doc/invoke.texi (-mneon-for-64bits): Document.
13853 2013-03-21 Richard Biener <rguenther@suse.de>
13855 PR tree-optimization/39326
13856 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
13857 (sort_bbs_in_loop_postorder_cmp): New function.
13858 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
13860 2013-03-21 Richard Biener <rguenther@suse.de>
13862 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
13863 (vect_insert_into_interleaving_chain): Likewise.
13864 (vect_drs_dependent_in_basic_block): Inline ...
13865 (vect_slp_analyze_data_ref_dependence): ... here. New function,
13867 (vect_analyze_data_ref_dependence): ... here. Simplify.
13868 (vect_check_interleaving): Simplify.
13869 (vect_analyze_data_ref_dependences): Likewise. Split out ...
13870 (vect_slp_analyze_data_ref_dependences): ... this new function.
13871 (dr_group_sort_cmp): New function.
13872 (vect_analyze_data_ref_accesses): Compute data-reference groups
13873 here instead of in vect_analyze_data_ref_dependence. Use
13874 a more efficient algorithm.
13875 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
13876 vect_slp_analyze_data_ref_dependences. Call
13877 vect_analyze_data_ref_accesses earlier.
13878 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13879 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
13880 (vect_slp_analyze_data_ref_dependences): New prototype.
13882 2013-03-21 Richard Biener <rguenther@suse.de>
13884 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
13885 ref is stored in the loop.
13886 (find_refs_for_sm): Walk only over all stores.
13887 (store_motion_loop): Allocate from lim_bitmap_obstack.
13888 (store_motion): Likewise.
13890 2013-03-21 Richard Biener <rguenther@suse.de>
13892 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
13893 Update virtual SSA form.
13895 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13897 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
13898 * configure: Regenerate.
13899 * config.in: Regenerate.
13900 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
13901 if !HAVE_LD_EH_FRAME_CIEV3.
13903 2013-03-21 Richard Biener <rguenther@suse.de>
13905 * tree-cfg.c (verify_expr_no_block): New function.
13906 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
13907 nor DECL_VALUE_EXPR have locations with associated blocks.
13908 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
13909 (clear_unused_block_pointer): Remove code dealing with
13910 blocks in DECL_DEBUG_EXPR locations.
13912 2013-03-21 Richard Biener <rguenther@suse.de>
13914 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
13915 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
13916 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
13917 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
13918 instead of DECL_DEBUG_EXPR_IS_FROM.
13919 * gimplify.c (gimplify_modify_expr): Likewise.
13920 * tree-cfg.c (verify_expr_location_1): Likewise.
13921 * tree-complex.c (create_one_component_var): Likewise.
13922 * tree-sra.c (create_access_replacement): Likewise.
13923 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
13924 (clear_unused_block_pointer): Likewise.
13925 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
13926 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
13927 * var-tracking.c (var_debug_decl): Likewise.
13928 (track_expr_p): Likewise.
13929 * tree-inline.c (add_local_variables): Likewise. Set
13930 DECL_HAS_DEBUG_EXPR_P after copying it.
13931 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
13932 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
13934 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
13937 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
13938 * configure: Regenerate.
13939 * config.in: Regenerate.
13940 * config/i386/i386.md (*movdf_internal): Use
13941 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
13942 movd instead of movq mnemonic for interunit moves.
13943 (*movdi_internal): Ditto.
13945 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
13947 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
13948 (abd<mode>_3): New pattern.
13949 (aba<mode>_3): New pattern.
13950 (fabd<mode>_3): New pattern.
13952 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
13954 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
13955 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
13956 occurrence of REGISTER_PREFIX as its empty string.
13958 2013-03-20 Jeff Law <law@redhat.com>
13960 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
13961 addititional equivalences for equality comparisons between an SSA_NAME
13962 and a constant where the SSA_NAME was set from a widening conversion.
13964 2013-03-20 Walter Lee <walt@tilera.com>
13966 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
13968 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
13970 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
13971 depending on type attribute.
13972 (*movti_internal): Ditto.
13973 (*movtf_internal): Ditto.
13974 (*movxf_internal): Ditto.
13975 (*movdf_internal): Ditto.
13976 (*movsf_internal): Ditto.
13978 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
13980 * config/i386/i386.md (*movti_internal): Set prefix attribute to
13981 maybe_vex for sselog1 and ssemov types.
13982 (*movdi_internal): Reorder operand constraints.
13983 (*movsi_internal): Ditto. Set prefix attribute to
13984 maybe_vex for sselog1 and ssemov types.
13985 (*movtf_internal): Set prefix attribute to maybe_vex
13986 for sselog1 and ssemov types.
13987 (*movdf_internal): Ditto. Set prefix_data16 attribute for
13988 DImode ssemov types. Reorder operand constraints.
13989 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
13990 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
13991 attribute for SImode ssemov types. Reorder operand constraints.
13993 2013-03-20 Martin Jambor <mjambor@suse.cz>
13995 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
13996 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
13998 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
14000 * config/rs6000/predicates.md (indexed_address, update_address_mem
14001 update_indexed_address_mem): New predicates.
14002 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14003 attribute for load/store instructions.
14004 * config/rs6000/dfp.md (movsd_store): Likewise.
14005 (movsd_load): Likewise.
14006 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14007 (unnamed HI->DI extend define_insn): Likewise.
14008 (unnamed SI->DI extend define_insn): Likewise.
14009 (unnamed QI->SI extend define_insn): Likewise.
14010 (unnamed QI->HI extend define_insn): Likewise.
14011 (unnamed HI->SI extend define_insn): Likewise.
14012 (unnamed HI->SI extend define_insn): Likewise.
14013 (extendsfdf2_fpr): Likewise.
14014 (movsi_internal1): Likewise.
14015 (movsi_internal1_single): Likewise.
14016 (movhi_internal): Likewise.
14017 (movqi_internal): Likewise.
14018 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
14019 attribute for load/store instructions.
14020 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
14022 (mov<mode>_softfloat): Likewise.
14023 (mov<mode>_hardfloat32): Likewise.
14024 (mov<mode>_hardfloat64): Likewise.
14025 (mov<mode>_softfloat64): Likewise.
14026 (movdi_internal32): Likewise.
14027 (movdi_internal64): Likewise.
14028 (probe_stack_<mode>): Likewise.
14030 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14032 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
14033 floating point, and decimal floating point to reload iterator.
14035 * config/rs6000/constraints.md (wl constraint): New constraints to
14036 return FLOAT_REGS if certain options are used to reduce the number
14037 of separate patterns that exist in the file.
14038 (wx constraint): Likewise.
14039 (wz constraint): Likewise.
14041 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14042 -mdebug=reg, print wg, wl, wx, and wz constraints.
14043 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14044 Initialize the reload functions for 64-bit binary/decimal floating
14046 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14047 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14048 create the buffer on the stack to overcome not having a 32-bit
14050 (rs6000_emit_move): Likewise.
14051 (rs6000_secondary_memory_needed_rtx): Likewise.
14052 (rs6000_alloc_sdmode_stack_slot): Likewise.
14053 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14054 via xxlxor, just like DFmode 0.0.
14056 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
14057 define as 1 if we are running on a power7 or newer.
14058 (enum r6000_reg_class_enum): Add new constraints.
14060 * config/rs6000/dfp.md (movsd): Delete, combine with binary
14061 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
14062 with other moves by using conditional constraits (wg). Use LFIWZX
14063 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
14064 (movsd splitter): Likewise.
14065 (movsd_hardfloat): Likewise.
14066 (movsd_softfloat): Likewise.
14068 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
14069 binary and decimal floating point moves.
14070 (fmove_ok): New attributes to combine binary and decimal floating
14071 point moves, and to combine power6x (mfpgpr) moves along normal
14073 (real_value_to_target): Likewise.
14074 (f32_lr): Likewise.
14075 (f32_lm): Likewise.
14076 (f32_li): Likewise.
14077 (f32_sr): Likewise.
14078 (f32_sm): Likewise.
14079 (f32_si): Likewise.
14080 (movsf): Combine binary and decimal floating point moves. Combine
14081 power6x (mfpgpr) moves with other moves by using conditional
14082 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
14083 (mov<mode> for SFmode/SDmode); Likewise.
14084 (SFmode/SDmode splitters): Likewise.
14085 (movsf_hardfloat): Likewise.
14086 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
14087 (movsf_softfloat): Likewise.
14088 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
14090 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
14091 wx and wz constraints.
14093 * config/rs6000/constraints.md (wg constraint): New constraint to
14094 return FLOAT_REGS if -mmfpgpr (power6x) was used.
14096 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
14099 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14100 -mdebug=reg, print wg, wl, wx, and wz constraints.
14101 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14102 Initialize the reload functions for 64-bit binary/decimal floating
14104 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14105 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14106 create the buffer on the stack to overcome not having a 32-bit
14108 (rs6000_emit_move): Likewise.
14109 (rs6000_secondary_memory_needed_rtx): Likewise.
14110 (rs6000_alloc_sdmode_stack_slot): Likewise.
14111 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14112 via xxlxor, just like DFmode 0.0.
14114 * config/rs6000/dfp.md (movdd): Delete, combine with binary
14115 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
14116 with other moves by using conditional constraits (wg). Use LFIWZX
14117 and STFIWX for loading SDmode on power7.
14118 (movdd splitters): Likewise.
14119 (movdd_hardfloat32): Likewise.
14120 (movdd_softfloat32): Likewise.
14121 (movdd_hardfloat64_mfpgpr): Likewise.
14122 (movdd_hardfloat64): Likewise.
14123 (movdd_softfloat64): Likewise.
14125 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
14126 64-bit binary and decimal floating point moves.
14127 (FMOVE64X): Likewise.
14128 (movdf): Combine 64-bit binary and decimal floating point moves.
14129 Combine power6x (mfpgpr) moves with other moves by using
14130 conditional constraits (wg).
14131 (mov<mode> for DFmode/DDmode): Likewise.
14132 (DFmode/DDmode splitters): Likewise.
14133 (movdf_hardfloat32): Likewise.
14134 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
14135 (movdf_softfloat32): Likewise.
14136 (movdf_hardfloat64_mfpgpr): Likewise.
14137 (movdf_hardfloat64): Likewise.
14138 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
14139 (movdf_softfloat64): Likewise.
14140 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
14141 (reload_<mode>_load): Move to later in the file so they aren't in
14142 the middle of the floating point move insns.
14143 (reload_<mode>_store): Likewise.
14145 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
14148 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
14149 constraint if -mdebug=reg.
14150 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
14151 Enable using dd reload support if needed.
14153 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
14154 binary and decimal floating point moves in rs6000.md.
14155 (movtd_internal): Likewise.
14157 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
14158 decimal floating point moves.
14160 (movtf_internal): Likewise.
14161 (mov<mode>_internal, TDmode/TFmode): Likewise.
14162 (movtf_softfloat): Likewise.
14163 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
14165 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
14166 movdi_internal64, using wg constraint for move direct operations.
14167 (movdi_internal64): Likewise.
14169 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
14170 MODES_TIEABLE_P for selected modes. Print the numerical value of
14171 the various virtual registers. Use GPR/FPR first/last values,
14172 instead of hard coding the register numbers. Print which modes
14173 have reload functions registered.
14174 (rs6000_option_override_internal): If -mdebug=reg, trace the options
14175 settings before/after setting cpu, target and subtarget settings.
14176 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
14177 and for secondary reload failures in rs6000_secondary_reload_inner.
14178 (rs6000_secondary_reload_fail): Likewise.
14179 (rs6000_secondary_reload_inner): Likewise.
14181 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
14182 macros for first/last GPR and FPR registers.
14183 (LAST_GPR_REGNO): Likewise.
14184 (FIRST_FPR_REGNO): Likewise.
14185 (LAST_FPR_REGNO): Likewise.
14187 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
14188 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
14189 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
14190 (vcond<mode><mode>): Likewise.
14191 (vcondu<mode><mode>): Likewise.
14192 (vector_gtu<mode>): Likewise.
14193 (vector_gte<mode>): Likewise.
14194 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
14195 to prevent the compiler from converting DImode operations to TImode.
14196 (ior<mode>3): Likewise.
14197 (and<mode>3): Likewise.
14198 (one_cmpl<mode>2): Likewise.
14199 (nor<mode>3): Likewise.
14200 (andc<mode>3): Likewise.
14202 * config/rs6000/constraints.md (wt constraint): New constraint
14203 that returns VSX_REGS if TImode is allowed in VSX registers.
14205 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
14206 constant under VSX.
14208 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
14209 similar to TImode, but it is restricted to being in the GPRs.
14211 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
14212 TImode to occupy a single VSX register.
14214 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
14215 -mvsx-timode for power7/power8.
14216 (power7 cpu): Likewise.
14217 (power8 cpu): Likewise.
14219 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
14220 sure that TFmode/TDmode take up two registers if they are ever
14221 allowed in the upper VSX registers.
14222 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
14224 (rs6000_init_hard_regno_mode_ok): Likewise.
14225 (rs6000_debug_reg_global): Add debugging for PTImode and wt
14226 constraint. Print if LRA is turned on.
14227 (rs6000_option_override_internal): Give an error if -mvsx-timode
14228 and VSX is not enabled.
14229 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
14230 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
14231 to reg+offset addressing. Use PTImode when checking offset
14232 addresses for validity.
14233 (reg_offset_addressing_ok_p): Likewise.
14234 (rs6000_legitimate_offset_address_p): Likewise.
14235 (rs6000_legitimize_address): Likewise.
14236 (rs6000_legitimize_reload_address): Likewise.
14237 (rs6000_legitimate_address_p): Likewise.
14238 (rs6000_eliminate_indexed_memrefs): Likewise.
14239 (rs6000_emit_move): Likewise.
14240 (rs6000_secondary_reload): Likewise.
14241 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
14242 reloads to fpr registers to continue to use reg+offset addressing,
14243 but 64-bit reloads to altivec registers need reg+reg addressing.
14244 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
14245 it. Treat LO_SUM like a PLUS operation.
14246 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
14247 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
14248 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
14249 registers to share a register with a smaller sized type, since VSX
14250 puts scalars in the upper 64-bits.
14251 (print_operand): Add support for PTImode.
14252 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
14253 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
14254 registers, but don't have arithmetic support.
14255 (rs6000_memory_move_cost): Add test for VSX.
14256 (rs6000_opt_masks): Add -mvsx-timode.
14258 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
14261 (VSr): Use wt constraint for TImode.
14262 (VSv): Drop TImode support.
14263 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
14264 (vsx_movti_64bit): Likewise.
14265 (vsx_movti_32bit): Likewise.
14266 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
14267 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
14268 one '?' on the appropriate output constraint. Do not allow TImode
14269 logical operations on 32-bit systems.
14270 (vsx_ior<mode>3): Likewise.
14271 (vsx_xor<mode>3): Likewise.
14272 (vsx_one_cmpl<mode>2): Likewise.
14273 (vsx_nor<mode>3): Likewise.
14274 (vsx_andc<mode>3): Likewise.
14275 (vsx_concat_<mode>): Likewise.
14276 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
14278 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
14279 OPTION_MASK_VSX_TIMODE.
14280 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
14281 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
14283 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
14284 (TI2 iterator): New iterator for TImode, PTImode.
14285 (wd mode attribute): Add values for vector types.
14286 (movti_string): Replace TI move operations with operations for TImode
14287 and PTImode. Add support for TImode being allowed in VSX registers.
14288 (mov<mode>_string, TImode/PTImode): Likewise.
14289 (movti_ppc64): Likewise.
14290 (mov<mode>_ppc64, TImode/PTImode): Likewise.
14291 (TI mode splitters): Likewise.
14293 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
14296 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
14298 PR tree-optimization/56355
14299 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
14300 Also handle integers with undefined overflow.
14302 2013-03-20 Catherine Moore <clm@codesourcery.com>
14303 Maciej W. Rozycki <macro@codesourcery.com>
14304 Tom de Vries <tom@codesourcery.com>
14305 Nathan Sidwell <nathan@codesourcery.com>
14306 Iain Sandoe <iain@codesourcery.com>
14307 Nathan Froyd <froydnj@codesourcery.com>
14308 Chao-ying Fu <fu@mips.com>
14310 * doc/extend.texi: (micromips, nomicromips, nocompression):
14311 Document new function attributes.
14312 * doc/invoke.texi (minterlink-compressed, mmicromips,
14313 m14k, m14ke, m14kec): Document new options.
14314 (minterlink-mips16): Update documentation.
14315 * doc/md.texi (ZC, ZD): Document new constraints.
14316 * configure.ac (gcc_cv_as_micromips): Check if linker
14317 supports the .set micromips directive.
14318 * configure: Regenerate.
14319 * config.in: Regenerate.
14320 * config/mips/mips-tables.opt: Regenerate.
14321 * config/mips/micromips.md: New file.
14322 * constraints.md (ZC, ZD): New constraints.
14323 * config/mips/predicates.md (movep_src_register): New predicate.
14324 (movep_src_operand): New predicate.
14325 (non_volatile_mem_operand): New predicate.
14326 * config/mips/mips.md (multimem): New type.
14327 (length): Differentiate between 17-bit and 18-bit branch offsets.
14328 (MOVEP1, MOVEP2): New mode iterator.
14329 (mov_<load>l): Use ZC constraint.
14330 (mov_<load>r): Likewise.
14331 (mov_<store>l): Likewise.
14332 (mov_<store>r): Likewise.
14333 (*branch_equality<mode>_inverted): Add microMIPS support.
14334 (*branch_equality<mode>): Likewise.
14335 (*jump_absolute): Likewise.
14336 (indirect_jump_<mode>): Likewise.
14337 (tablejump_<mode>): Likewise.
14338 (<optab>_internal): Likewise.
14339 (sibcall_internal): Likewise.
14340 (sibcall_value_internal): Likewise.
14341 (prefetch): Use constraint ZD.
14342 * config/mips/mips.opt (minterlink-compressed): New option.
14343 (minterlink-mips16): Now an alias for minterlink-compressed.
14344 (mmicromips): New option.
14345 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14346 (compare_and_swap_12): Likewise.
14347 (sync_add<mode>): Likewise.
14348 (sync_<optab>_12): Likewise.
14349 (sync_old_<optab>_12): Likewise.
14350 (sync_new_<optab>_12): Likewise.
14351 (sync_nand_12): Likewise.
14352 (sync_old_nand_12): Likewise.
14353 (sync_new_nand_12): Likewise.
14354 (sync_sub<mode>): Likewise.
14355 (sync_old_add<mode>): Likewise.
14356 (sync_old_sub<mode>): Likewise.
14357 (sync_new_add<mode>): Likewise.
14358 (sync_new_sub<mode>): Likewise.
14359 (sync_<optab><mode>): Likewise.
14360 (sync_old_<optab><mode>): Likewise.
14361 (sync_new_<optab><mode>): Likewise.
14362 (sync_nand<mode>): Likewise.
14363 (sync_old_nand<mode>): Likewise.
14364 (sync_new_nand<mode>): Likewise.
14365 (sync_lock_test_and_set<mode>): Likewise.
14366 (test_and_set_12): Likewise.
14367 (atomic_compare_and_swap<mode>): Likewise.
14368 (atomic_exchange<mode>_llsc): Likewise.
14369 (atomic_fetch_add<mode>_llsc): Likewise.
14370 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14371 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14372 (umips_save_restore_pattern_p): Likewise.
14373 (umips_load_store_pair_p): Likewise.
14374 (umips_output_load_store_pair): Likewise.
14375 (umips_movep_target_p): Likewise.
14376 (umips_12bit_offset_address_p): Likewise.
14377 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14378 (mips_base_mips16): Rename this...
14379 (mips_base_compression_flags): ...to this. Update all uses.
14380 (mips_attribute_table): Add micromips, nomicromips and nocompression.
14381 (mips_mips16_decl_p): Delete.
14382 (mips_nomips16_decl_p): Delete.
14383 (mips_get_compress_on_flags): New function.
14384 (mips_get_compress_off_flags): New function.
14385 (mips_get_compress_mode): New function.
14386 (mips_get_compress_on_name): New function.
14387 (mips_get_compress_off_name): New function.
14388 (mips_insert_attributes): Support multiple compression types.
14389 (mips_merge_decl_attributes): Likewise.
14390 (umips_12bit_offset_address_p): New function.
14391 (mips_start_function_definition): Emit .set micromips directive.
14392 (mips_call_may_need_jalx_p): New function.
14393 (mips_function_ok_for_sibcall): Add microMIPS support.
14394 (mips_print_operand_punctuation): Support short delay slots and
14396 (umips_swm_mask, umips_swm_encoding): New.
14397 (umips_build_save_restore): New function.
14398 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14399 (was_mips16_p): Remove.
14400 (old_compression_mode): New.
14401 (mips_set_compression_mode): New function.
14402 (mips_set_current_function): Add microMIPS support.
14403 (mips_option_override): Likewise.
14404 (umips_save_restore_pattern_p): New function.
14405 (umips_output_save_restore): New function.
14406 (umips_load_store_pair_p_1): New function.
14407 (umips_load_store_pair_p): New function.
14408 (umips_output_load_store_pair_1): New function.
14409 (umips_output_load_store_pair): New function.
14410 (umips_movep_target_p) New function.
14411 (mips_prepare_pch_save): Add microMIPS support.
14412 * config/mips/mips.h (TARGET_COMPRESSION): New.
14413 (TARGET_CPU_CPP_BUILTINS): Update macro
14414 to use new compression flags and to support microMIPS.
14415 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14416 (MIPS_ARCH_FLOAT_SPEC): Likewise.
14417 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14418 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14419 (ASM_SPEC): Support mmicromips and mno-micromips.
14420 (M16STORE_REG_P): New macro.
14421 (MIPS_CALL): Support TARGET_MICROMIPS.
14422 (MICROMIPS_J): New macro.
14423 (mips_base_mips16): Rename this...
14424 (mips_base_compression_flags): ...to this.
14425 (UMIPS_12BIT_OFFSET_P): New macro.
14426 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14427 (MULTILIB_DIRNAMES): Likewise.
14428 2013-03-20 Richard Biener <rguenther@suse.de>
14430 PR tree-optimization/56661
14431 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
14432 the result does not have to be distinct.
14434 2013-03-20 Richard Biener <rguenther@suse.de>
14436 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
14439 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14440 Steven Bosscher <steven@gcc.gnu.org>
14442 PR rtl-optimization/56605
14443 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
14445 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
14448 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
14449 that require movd instead of movq.
14451 2013-03-20 Richard Biener <rguenther@suse.de>
14453 * tree-ssa-structalias.c (struct variable_info): Add pointer
14454 to the first field of an aggregate with sub-vars. Make
14455 this and the pointer to the next subfield its ID.
14456 (vi_next): New function.
14457 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
14458 storedanything_id, integer_id): Increment by one.
14459 (new_var_info, get_call_vi, lookup_call_clobber_vi,
14460 get_call_clobber_vi): Adjust.
14461 (solution_set_expand): Simplify and speedup.
14462 (solution_set_add): Inline into ...
14463 (set_union_with_increment): ... this. Adjust accordingly.
14464 (do_sd_constraint): Likewise.
14465 (do_ds_constraint): Likewise.
14466 (do_complex_constraint): Simplify.
14467 (build_pred_graph): Adjust.
14468 (solve_graph): Likewise. Simplify and speedup.
14469 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
14470 get_constraint_for_component_ref, get_constraint_for_1,
14471 first_vi_for_offset, first_or_preceding_vi_for_offset,
14472 create_function_info_for, create_variable_info_for_1,
14473 create_variable_info_for, intra_create_variable_infos): Adjust.
14474 (init_base_vars): Push NULL for ID zero.
14475 (compute_points_to_sets): Adjust.
14477 2013-03-20 Richard Biener <rguenther@suse.de>
14479 * cfgloop.c (verify_loop_structure): Streamline and avoid
14480 ICEing on corrupt loop tree.
14481 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
14484 2013-03-20 Richard Biener <rguenther@suse.de>
14486 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
14487 check whether an SSA update is needed.
14489 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
14491 * config/mips/constraints.md (T): Rename to...
14495 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
14496 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
14498 2013-03-19 Ian Bolton <ian.bolton@arm.com>
14500 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
14501 (*subsi3_carryin_uxtw): Likewise.
14503 2013-03-19 Ian Bolton <ian.bolton@arm.com>
14505 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
14506 (*rorsi3_insn_uxtw): Likewise.
14508 2013-03-19 Ian Bolton <ian.bolton@arm.com>
14510 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
14511 (*extrsi5_insn_uxtw): Likewise.
14513 2013-03-19 Richard Biener <rguenther@suse.de>
14515 PR tree-optimization/56273
14516 * passes.c (init_optimization_passes): Move second VRP after DOM.
14518 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
14520 * config/i386/i386.md (*movti_internal): Merge from
14521 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
14522 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
14523 nox64 isa attributes.
14525 2013-03-18 Richard Biener <rguenther@suse.de>
14527 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
14529 (merge_node_constraints): Likewise.
14530 (build_succ_graph): Likewise.
14531 (valid_graph_edge): Inline into single caller.
14532 (unify_nodes): Likewise. Use bitmap_set_bit return value
14534 (scc_visit): Fix formatting and variable use.
14535 (do_sd_constraint): Use gcc_checking_assert.
14536 (do_ds_constraint): Likewise.
14537 (do_complex_constraint): Likewise.
14538 (condense_visit): Likewise. Cleanup.
14539 (dump_pred_graph): New function.
14540 (perform_var_substitution): Dump the pred-graph before
14541 variable substitution.
14542 (find_equivalent_node): Use gcc_checking_assert.
14543 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
14545 2013-03-18 Richard Biener <rguenther@suse.de>
14547 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
14548 Remove cond_expr_stmt_list argument and do not gimplify the
14550 (vect_loop_versioning): Adjust.
14551 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
14552 Cleanup to use less temporaries.
14553 (vect_create_data_ref_ptr): Cleanup.
14555 2013-03-18 Jakub Jelinek <jakub@redhat.com>
14557 PR tree-optimization/56635
14558 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
14559 require types_compatible_p types.
14561 2013-03-18 Nick Clifton <nickc@redhat.com>
14563 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
14564 spurious backslash.
14566 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
14567 Add missing line to comment describing function.
14569 2013-03-18 Richard Biener <rguenther@suse.de>
14571 PR tree-optimization/56210
14572 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
14573 Handle string / character search functions.
14574 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
14576 2013-03-18 Richard Biener <rguenther@suse.de>
14578 PR middle-end/56483
14579 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
14580 and implement properly.
14581 * gimple.h (gimple_cond_single_var_p): Remove.
14583 2013-03-18 Richard Biener <rguenther@suse.de>
14585 * tree-data-ref.h (find_data_references_in_loop): Declare.
14586 * tree-data-ref.c (get_references_in_stmt): Use a stack
14587 vector pre-allocated in the callers.
14588 (find_data_references_in_stmt): Adjust.
14589 (graphite_find_data_references_in_stmt): Likewise.
14590 (create_rdg_vertices): Likewise.
14591 (find_data_references_in_loop): Export.
14592 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14593 Compute dependences here...
14594 (vect_analyze_data_refs): ...not here. When we encounter
14595 a non-vectorizable data reference in basic-block vectorization
14596 truncate the data reference vector. Do not bother to
14597 fixup data-dependence information for gather loads.
14598 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
14599 of data references, as reported.
14601 2013-03-18 Richard Biener <rguenther@suse.de>
14603 PR tree-optimization/3713
14604 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
14605 has_constants and expr.
14606 (stmt_has_constants): Properly valueize SSA names when deciding
14607 whether the stmt has constants.
14609 2013-03-18 Richard Biener <rguenther@suse.de>
14611 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
14612 whole function when there is nothing to do.
14613 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
14614 * tree-vectorizer.c (vectorize_loops): Update virtual and
14615 loop-closed SSA once.
14616 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
14618 2013-03-18 Richard Biener <rguenther@suse.de>
14620 PR middle-end/56113
14621 * domwalk.c (bb_postorder): New global static.
14622 (cmp_bb_postorder): New function.
14623 (walk_dominator_tree): Replace scheme imposing an order for
14624 visiting dominator sons by one sorting them at the time they
14625 are pushed on the stack.
14627 2013-03-18 Richard Biener <rguenther@suse.de>
14629 PR tree-optimization/39326
14630 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
14631 (struct mem_ref): Replace mem member with ao_ref typed member.
14632 (MEM_ANALYZABLE): Adjust.
14633 (memref_eq): Likewise.
14634 (mem_ref_alloc): Likewise.
14635 (gather_mem_refs_stmt): Likewise.
14636 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
14637 (execute_sm_if_changed_flag_set): Adjust.
14638 (execute_sm): Likewise.
14639 (ref_always_accessed_p): Likewise.
14640 (refs_independent_p): Likewise.
14641 (can_sm_ref_p): Likewise.
14643 2013-03-18 Jakub Jelinek <jakub@redhat.com>
14646 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
14647 return 1 even for !unsignedp.
14649 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
14651 * config/i386/i386.md (isa): Add x64 and nox64.
14652 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
14653 (*pushtf): Enable *roF alternative for x64 isa only.
14654 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
14655 mode attribute of integer alternatives to DImode for TARGET_64BIT.
14656 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
14657 (*movtf_internal): Merge from *movtf_internal_rex64 and
14658 *movtf_internal_sse. Use x64 and nox64 isa attributes.
14659 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
14660 nox64 isa attributes.
14661 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
14662 nox64 isa attributes.
14663 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
14665 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
14667 * config/alpha/alpha.c (TARGET_LRA_P): New define.
14669 2013-03-17 Jakub Jelinek <jakub@redhat.com>
14672 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
14673 class names. Remove trailing comma after "ALL_REGS".
14675 2013-03-16 Jan Hubicka <jh@suse.cz>
14677 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
14678 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
14679 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
14680 of cgraph_get_create_node.
14681 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14683 2013-03-16 Jason Merrill <jason@redhat.com>
14686 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
14687 with DW_AT_default_value.
14689 2013-03-16 Jakub Jelinek <jakub@redhat.com>
14691 * BASE-VER: Set to 4.9.0.
14693 2013-03-14 Andi Kleen <ak@linux.intel.com>
14696 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
14697 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
14698 Document _x* TSX intrinsics.
14700 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
14701 David Holsgrove <david.holsgrove@xilinx.com>
14703 * configure.ac: Add MicroBlaze TLS support detection.
14704 * configure: Regenerate.
14705 * config/microblaze/microblaze-protos.h
14706 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
14707 symbol_mentioned_p, label_mentioned_p): Add prototypes.
14708 * config/microblaze/microblaze.c (microblaze_address_type): Add
14709 ADDRESS_TLS and tls_reloc address types.
14710 (microblaze_address_info): Add tls_reloc.
14711 (TARGET_HAVE_TLS): Define.
14712 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
14713 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
14714 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
14715 load_tls_operand, microblaze_call_tls_get_addr,
14716 microblaze_legitimize_tls_address): New functions.
14717 (microblaze_classify_unspec): Handle UNSPEC_TLS.
14718 (get_base_reg): Use microblaze_tls_symbol_p.
14719 (microblaze_classify_address): Handle TLS.
14720 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
14721 label_mentioned_p and microblaze_tls_referenced_p.
14722 (microblaze_legitimize_address): Handle TLS.
14723 (microblaze_address_insns): Handle ADDRESS_TLS.
14724 (pic_address_needs_scratch): Handle TLS.
14725 (print_operand_address): Handle TLS.
14726 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
14727 (microblaze_expand_move): Handle TLS.
14728 (microblaze_legitimate_constant_p): Check
14729 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
14730 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14731 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
14732 (PIC_OFFSET_TABLE_REGNUM): Set.
14733 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
14734 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
14735 (addsi3, movsi_internal2, movdf_internal): Update constraints
14736 * config/microblaze/predicates.md (arith_plus_operand): Define
14737 (move_operand): Redefine as move_src_operand,
14738 check microblaze_tls_referenced_p.
14740 2013-03-14 Ian Bolton <ian.bolton@arm.com>
14742 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
14743 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
14745 2013-03-14 Ian Bolton <ian.bolton@arm.com>
14747 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
14750 2013-03-14 Jakub Jelinek <jakub@redhat.com>
14752 PR tree-optimization/53265
14753 * common.opt (Waggressive-loop-optimizations): New option.
14754 * tree-ssa-loop-niter.c: Include tree-pass.h.
14755 (do_warn_aggressive_loop_optimizations): New function.
14756 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
14757 if number_of_latch_executions returned constant.
14758 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
14759 early. If number_of_latch_executions returned constant, set
14760 nb_iterations_upper_bound back to it.
14761 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
14763 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
14764 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
14766 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
14767 (MULTILIB_OSDIRNAMES): Set.
14768 * genmultilib: If defaultosdirname doesn't start with :: , set
14769 defaultosdirname2 instead, clear it and emit two . multilib_raw
14770 entries instead of just one.
14772 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
14774 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14775 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
14776 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
14777 (SUBTARGET_OVERRIDE_OPTIONS): New.
14779 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
14782 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
14783 (musermode): Convert to Var(TARGET_USERMODE).
14784 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
14785 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
14786 * config/sh/sh.c (sh_option_override): Use
14787 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
14788 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
14790 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
14792 (udivsi3_i4_single, divsi3_i4_single): Use
14793 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
14795 2013-03-13 Dave Korn <dave.korn.cygwin@....>
14797 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
14800 2013-03-13 Richard Biener <rguenther@suse.de>
14802 PR tree-optimization/56608
14803 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
14804 calls when vectorizing basic-blocks.
14806 2013-03-13 Jakub Jelinek <jakub@redhat.com>
14809 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
14812 2013-03-12 Jakub Jelinek <jakub@redhat.com>
14814 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
14816 2013-03-11 Jan Hubicka <jh@suse.cz>
14819 * lto-streamer-out.c (output_symbol_p): Skip references from
14820 constructors of external variables.
14822 2013-03-11 Jan Hubicka <jh@suse.cz>
14824 PR middle-end/56571
14825 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
14827 * emit-rtl.c (verify_rtx_sharing): Likewise.
14828 (copy_insn_1): Likewise.
14829 * rtl.c (copy_rtx): Likewise.
14831 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
14834 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
14835 output_operand_lossage message.
14837 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
14840 * arm.c (shift_op): Validate RTL pattern on the fly.
14841 (arm_print_operand, case 'S'): Don't use shift_operator to validate
14844 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
14847 * config/pa/pa.md (call_value): Check for calls to powf and direct to
14848 new call patterns that clobber %fr12.
14849 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
14850 split and postreload patterns.
14851 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
14852 registers %fr12 and %fr12R as call used.
14854 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
14856 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
14857 (canon_address, record_store, replace_read, check_mem_read_rtx,
14858 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
14859 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
14860 rest_of_handle_dse): Likewise.
14862 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
14864 PR middle-end/56524
14865 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
14867 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
14868 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
14869 (save_optabs_if_changed): Replace with...
14870 (init_tree_optimization_optabs): ...this.
14871 * optabs.c (save_optabs_if_changed): Rename to...
14872 (init_tree_optimization_optabs): ...this. Take the optimization node
14873 as argument. Do nothing if the base optabs are already correct.
14874 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
14875 to recompute optabs.
14876 * function.h (function): Remove optabs field.
14877 * function.c (invoke_set_current_function_hook): Call
14878 init_tree_optimization_optabs. Use the result to initialize
14881 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
14883 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
14884 if GTMA_HAS_NO_INSTRUMENTATION.
14885 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
14886 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
14887 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
14888 * gimple-pretty-print.c (dump_gimple_transaction): Handle
14889 GTMA_HAS_NO_INSTRUMENTATION.
14891 2013-03-08 Jakub Jelinek <jakub@redhat.com>
14893 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
14896 2013-03-08 Marek Polacek <polacek@redhat.com>
14897 Jakub Jelinek <jakub@redhat.com>
14899 PR tree-optimization/56478
14900 * predict.c (is_comparison_with_loop_invariant_p): Change the
14901 type of loop_step to tree.
14902 (predict_loops): Adjust.
14903 (predict_iv_comparison): Perform the computations on double_ints.
14905 2013-03-08 Richard Biener <rguenther@suse.de>
14907 PR tree-optimization/56570
14908 * tree-cfg.c (verify_expr_location_1): Verify locations for
14910 * tree-sra.c (create_access_replacement): Strip locations
14911 from DECL_DEBUG_EXPRs.
14913 2013-03-08 Richard Biener <rguenther@suse.de>
14915 * tree-inline.c (expand_call_inline): Do not associate
14916 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
14917 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
14919 2013-03-08 Richard Biener <rguenther@suse.de>
14921 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
14922 or block changes with -Og. Fix for location / block encoding
14923 changes and PHI arguments with locations.
14925 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
14927 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
14929 (struct output_info): Likewise.
14930 (register_overhead): Remove bad gcc_assert.
14931 (bitmap_find_bit): If there is only a single bitmap element, do not
14932 count a miss as a search.
14933 (print_statistics): Update for counter type changes.
14934 (dump_bitmap_statistics): Likewise. Print headers such that they
14935 are properly lined up with the printed counters.
14937 2013-03-07 Jakub Jelinek <jakub@redhat.com>
14939 PR tree-optimization/56559
14940 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
14941 check that it has only a single use.
14943 2013-03-07 Richard Biener <rguenther@suse.de>
14945 * doc/invoke.texi (fwhole-program): Discourage use in combination
14948 2013-03-06 Jakub Jelinek <jakub@redhat.com>
14950 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
14952 PR tree-optimization/56539
14953 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
14954 instead of GSI_CONTINUE_LINKING as last argument to
14955 force_gimple_operand_gsi. Adjust function comment.
14957 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
14960 PR middle-end/56548
14961 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
14962 promoted mode, convert the result back to the original mode.
14964 2013-03-06 Richard Biener <rguenther@suse.de>
14966 PR middle-end/56294
14967 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
14968 (insert_updated_phi_nodes_compare_uids): New function.
14969 (update_ssa): Sort symbols_to_rename after UID before
14970 traversing it to insert PHI nodes.
14972 2013-03-06 Richard Biener <rguenther@suse.de>
14974 PR middle-end/50494
14975 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
14976 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
14979 2013-02-13 Richard Biener <rguenther@suse.de>
14982 * varasm.c (output_constant_def_1): Get the decl representing
14983 the constant as argument.
14984 (output_constant_def): Wrap output_constant_def_1.
14985 (make_decl_rtl): Use output_constant_def_1 with the decl
14986 representing the constant.
14987 (build_constant_desc): Optionally re-use a decl already
14988 representing the constant.
14989 (tree_output_constant_def): Adjust.
14991 2013-03-06 Joey Ye <joey.ye@arm.com>
14994 * gcc.c (convert_white_space): New function.
14995 (main): Handles white space in function name.
14997 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
15000 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15001 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
15002 to SH_DIV_CALL_TABLE for TARGET_SH2.
15003 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15005 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15006 call-table options.
15008 2013-03-05 Sterling Augustine <saugustine@google.com>
15009 Cary Coutant <ccoutant@google.com>
15012 * dwarf2out.c (resolve_addr): Don't call
15013 remove_loc_list_addr_table_entries a second time for the same
15016 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15019 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
15020 (avoid_complex_debug_insns): New function.
15021 (expand_debug_locations): Call it.
15023 PR rtl-optimization/56484
15024 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
15025 lifetimes of hard registers on small register class machines.
15027 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
15029 * config/microblaze/microblaze-protos.h: Rename
15030 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
15031 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
15033 (microblaze_fast_interrupt_function_p): New function.
15034 (microblaze_is_interrupt_handler): Rename to
15035 microblaze_is_interrupt_variant and add fast_interrupt check.
15036 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
15037 (save_restore_insns): Likewise.
15038 (compute_frame_size): Likewise.
15039 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
15040 (microblaze_globalize_label): Likewise.
15041 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
15042 * config/microblaze/microblaze.md: Use wrapper
15043 microblaze_is_interrupt_variant.
15045 2013-03-05 Kai Tietz <ktietz@redhat.com>
15047 * sdbout.c (sdbout_one_type): Switch to current function's section
15048 supporting cold/hot.
15050 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
15052 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
15055 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15057 PR middle-end/56461
15058 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
15059 if VALGRIND_GET_VBITS is defined, temporarily make object
15060 memory all defined, and restore previous valgrind addressability
15061 and definability afterwards. Free this_object at the end.
15063 PR middle-end/56461
15064 * lra.c (lra): Call lra_clear_live_ranges if live_p,
15065 right before calling lra_create_live_ranges, also call it
15066 when clearing live_p. Only call lra_clear_live_ranges
15067 at the end if live_p.
15069 PR middle-end/56461
15070 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
15072 2013-03-05 Richard Biener <rguenther@suse.de>
15074 PR tree-optimization/56521
15075 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
15078 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
15081 * except.h (remove_unreachable_eh_regions): New prototype.
15082 * except.c (remove_eh_handler_splicer): New function, split out
15083 of remove_eh_handler.
15084 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
15085 warning about running it on many EH regions one at a time.
15086 (remove_unreachable_eh_regions_worker): New function, walk the
15087 EH tree in depth-first order and remove non-marked regions.
15088 (remove_unreachable_eh_regions): New function.
15089 * tree-eh.c (mark_reachable_handlers): New function, split out
15090 from remove_unreachable_handlers.
15091 (remove_unreachable_handlers): Use mark_reachable_handlers and
15092 remove_unreachable_eh_regions.
15093 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
15094 and remove_unreachable_eh_regions.
15096 2013-03-05 Richard Biener <rguenther@suse.de>
15098 PR middle-end/56525
15099 * loop-init.c (fix_loop_structure): Remove loops in two stages,
15100 not freeing them until the end.
15102 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
15104 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
15106 2013-03-05 Richard Biener <rguenther@suse.de>
15108 PR tree-optimization/56270
15109 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
15110 of loads after scheduling an SLP instance.
15112 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15114 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
15116 (check_gcc_parallelize): Run guality.exp as a separate job from
15117 vect.exp with unsorted.exp and $(dg_target_exps) separately from
15118 struct-layout-1.exp with stackalign.exp.
15120 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
15122 PR middle-end/56461
15123 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
15124 load_index sbitmap even if some bit in it isn't set.
15126 PR middle-end/56461
15127 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
15128 (discover_iteration_bound_by_body_walk): Change queues to
15129 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
15130 spelling in comment. Call safe_push on queues[bound_index] directly.
15131 Release queues[queue_index] in every iteration unconditionally.
15132 Release bounds vector.
15134 PR middle-end/56461
15135 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
15136 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
15137 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
15140 2013-03-05 Richard Biener <rguenther@suse.de>
15143 * tree-inline.c (remap_blocks_to_null): New function.
15144 (expand_call_inline): When expanding a call stmt without
15145 an associated block inline remap all callee blocks to NULL.
15147 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15149 PR rtl-optimization/56494
15150 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
15151 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
15152 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
15154 PR middle-end/56461
15155 * sel-sched-ir.c (free_sched_pools): Release
15156 succs_info_pool.stack[succs_info_pool.max_top] vectors too
15157 if succs_info_pool.max_top isn't -1.
15160 * opts.c (opts_obstack, opts_concat): Moved to...
15161 * opts-common.c (opts_obstack, opts_concat): ... here.
15163 2013-03-04 Jakub Jelinek <jakub@redhat.com>
15165 PR middle-end/56461
15166 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
15168 2013-03-04 Martin Jambor <mjambor@suse.cz>
15170 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
15171 all appropriate places.
15173 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
15175 PR tree-optimization/56424
15176 * ipa-split.c (split_function): Do not set the RSO flag if result is
15177 not by reference and its type is a register type.
15179 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
15181 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
15182 (microblaze_legitimate_pic_operand): Likewise
15183 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
15184 new function microblaze_legitimate_pic_operand
15185 * config/microblaze/microblaze-protos.h
15186 (microblaze_legitimate_pic_operand): Declare.
15188 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15190 * config/microblaze/predicates.md (call_insn_simple_operand):
15191 New predicate for supported rtx code types.
15192 * config/microblaze/microblaze.md (call_internal1): Use
15193 call_insn_simple_operand predicate.
15195 2013-03-04 Jakub Jelinek <jakub@redhat.com>
15197 PR middle-end/56461
15198 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
15199 partitions.ordered_remove.
15201 PR middle-end/56461
15202 * tree-vect-stmts.c (vectorizable_conversion): Don't call
15203 vec_oprnds0.create (1) for modifier == NONE.
15205 PR middle-end/56461
15206 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
15207 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
15208 vec_oprnds1 right before pushing anything to it for
15211 PR middle-end/56461
15212 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
15213 set nbbs to 0 instead of having separate code path.
15214 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
15215 instead of false as last argument if returning NULL.
15217 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
15219 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
15220 the attribute is now called "target" instead of "option".
15221 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
15222 * doc/tm.texi.in (Target Attributes): Likewise document the correct
15223 attribute/pragma name for TARGET_OPTION_VALID_P and
15224 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
15225 * doc/tm.texi: Regenerated.
15227 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
15229 * config/microblaze/microblaze.c:
15230 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
15231 * config/microblaze/microblaze.h: Add -mxl-reorder to
15233 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
15234 instructions emitted if TARGET_REORDER.
15235 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
15236 or 0 for -m/-mno case, but initialises as 2 to detect default use case
15239 2013-03-01 Xinliang David Li <davidxl@google.com>
15241 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
15244 2013-03-01 Jakub Jelinek <jakub@redhat.com>
15246 PR middle-end/56461
15247 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
15248 vector even when returning true. Fix up function comment formatting.
15250 PR middle-end/56461
15251 * ira-build.c (ira_loop_nodes_count): New variable.
15252 (create_loop_tree_nodes): Initialize it.
15253 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
15255 PR middle-end/56461
15256 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
15257 method on dr_chain and result_chain.
15258 * tree-vect-stmts.c (vectorizable_store): Only call
15259 result_chain.create if j == 0.
15261 PR middle-end/56461
15262 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
15263 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
15264 before overwriting it.
15266 2013-03-01 Tobias Burnus <burnus@net-b.de>
15268 * doc/extended.texi (C Extensions): Change order in @menu
15270 (Other MIPS Built-in Functions): Move last MIPS entry before
15271 "picoChip Built-in Functions".
15272 (SH Built-in Functions): Move after RX Built-in Functions.
15273 * doc/gcc.texi (Introduction): Change order in @menu
15275 * doc/md.texi (Constraints): Ditto.
15276 * gty.texi (Type Information): Ditto.
15277 (User-provided marking routines for template types): Make
15279 * doc/invoke.texi (AArch64 Options): Move before
15280 "Adapteva Epiphany Options".
15282 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
15283 Jakub Jelinek <jakub@redhat.com>
15286 * asan.c (gate_asan): Lookup no_sanitize_address instead of
15287 no_address_safety_analysis attribute.
15288 * doc/extend.texi (no_address_safety_attribute): Rename to
15289 no_sanitize_address attribute, mention no_address_safety_analysis
15290 attribute as deprecated alias.
15292 2013-02-28 Jakub Jelinek <jakub@redhat.com>
15294 PR middle-end/56461
15295 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
15296 type to vec<vec<tree> > *.
15297 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
15298 to be vec<tree> instead of vec<tree> *, set vec_defs
15299 to vNULL and call vec_defs.create (number_of_vects), adjust other
15301 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
15302 vectorizable_condition): Adjust vect_get_slp_defs callers.
15304 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
15306 * config/aarch64/aarch64.c
15307 (aarch64_float_const_representable): Remove unused variable.
15309 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
15311 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
15313 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
15315 * config/aarch64/aarch64-builtins.c
15316 (aarch64_init_simd_builtins): Make static.
15318 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
15320 * config/aarch64/aarch64.c
15321 (aarch64_simd_make_constant): Make static.
15323 2013-02-28 Martin Jambor <mjambor@suse.cz>
15325 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
15326 with no initialization to the RHS of debug statements.
15328 2013-02-28 Martin Jambor <mjambor@suse.cz>
15330 PR tree-optimization/56294
15331 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
15333 (get_access_replacement): Do not call create_access_replacement.
15334 Assert a replacement exists.
15335 (get_repl_default_def_ssa_name): Create the replacement declaration
15338 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15340 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
15341 final_end_function.
15343 2013-02-28 Marek Polacek <polacek@redhat.com>
15345 PR rtl-optimization/56466
15346 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
15347 if we're changing a loop.
15348 (peel_loops_completely): Likewise.
15350 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
15353 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
15355 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
15358 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
15359 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
15360 INTX_FTYPE_FX, FX_FTYPE_INTX.
15361 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
15363 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
15365 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
15366 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
15367 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
15368 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
15369 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
15370 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
15371 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
15372 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
15373 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
15374 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
15375 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
15376 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15377 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
15378 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
15379 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
15380 (avrxmega6): Increase max flash segments from 5 to 6.
15381 * config/avr/t-multilib: Regenerate.
15382 * config/avr/avr-tables.opt: Regenerate.
15383 * doc/avr-mmcu.texi: Regenerate.
15385 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
15387 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
15388 (avr_device_to_arch): Rename to avr_device_to_ld.
15389 (avr_device_to_as): New prototype.
15390 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
15391 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
15392 * config/avr/driver-avr.c (avr_device_to_as): New.
15393 (avr_device_to_arch): Rename to avr_device_to_ld.
15395 2013-02-27 Jakub Jelinek <jakub@redhat.com>
15397 PR middle-end/56461
15398 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
15399 method on dr_chain and result_chain.
15401 PR middle-end/56461
15402 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
15403 pointer_set_destroy on not_executed_last_iteration.
15405 PR middle-end/56461
15406 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
15408 PR middle-end/56461
15409 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
15410 FOR_EACH_DEFINED_FUNCTION when freeing state.
15412 PR middle-end/56461
15413 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
15415 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
15418 PR middle-end/56461
15419 * ipa-cp.c (decide_whether_version_node): Call vec_free on
15420 known_aggs[i].items and release known_aggs vector.
15422 PR middle-end/56461
15423 * ipa-reference.c (propagate): Free node_info even for alias nodes.
15425 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15427 * config/microblaze/microblaze.c (microblaze_emit_compare):
15428 Use xor for EQ/NE comparisions.
15429 * config/microblaze/microblaze.md (cstoresf4): Add constraints
15430 (cbranchsf4): Adjust operator to comparison_operator.
15432 2013-02-27 Jakub Jelinek <jakub@redhat.com>
15434 PR middle-end/56461
15435 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
15437 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
15438 vec_safe_push, always update *slot.
15439 (redirect_edge_var_map_clear): Use vec_free.
15440 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
15441 (free_var_map_entry): Use vec_free.
15442 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
15443 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
15445 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
15447 PR middle-end/45472
15448 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
15449 when the may_trap_p bit of the exprs being merged differs.
15450 Reorder tests for speculativeness in the logical and operator.
15452 2013-02-27 Jakub Jelinek <jakub@redhat.com>
15454 * incpath.c (add_standard_paths): Use reconcat instead of concat
15455 where appropriate and avoid leaking memory.
15457 * opts.h: Include obstack.h.
15458 (opts_concat): New prototype.
15459 (opts_obstack): New declaration.
15460 * opts.c (opts_concat): New function.
15461 (opts_obstack): New variable.
15462 (init_options_struct): Call gcc_init_obstack on opts_obstack.
15463 (finish_options): Use opts_concat instead of concat
15464 and XOBNEWVEC instead of XNEWVEC.
15465 * opts-common.c (generate_canonical_option, decode_cmdline_option,
15466 generate_option): Likewise.
15467 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
15468 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
15471 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
15472 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
15474 2013-02-26 Jakub Jelinek <jakub@redhat.com>
15476 PR middle-end/56461
15477 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
15479 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
15481 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
15482 (arm_block_move_unaligned_straight): Likewise.
15483 (arm_adjust_block_mem): Likewise.
15485 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
15488 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
15489 temp, cond and label.
15490 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
15493 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
15494 * config/c6x/c6x.h (dbx_register_map): Update declaration.
15497 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
15498 of prologue/epilogue functions.
15501 * config/tilegx/tilegx.c (tilegx_expand_prologue):
15502 Remove unused variable cfa_offset.
15503 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
15506 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
15507 type promotion to unsigned.
15510 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
15511 for HOST_WIDE_INT of 32 bit / same size as int.
15512 (arm_block_move_unaligned_straight): Likewise.
15513 (arm_adjust_block_mem): Likewise.
15516 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
15519 2013-02-26 Marek Polacek <polacek@redhat.com>
15521 PR tree-optimization/56426
15522 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
15524 2013-02-26 Richard Biener <rguenther@suse.de>
15527 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
15528 unused variable loops.
15530 2013-02-26 Jakub Jelinek <jakub@redhat.com>
15532 PR tree-optimization/56448
15533 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
15534 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
15535 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
15536 later operands of the references, or even first operand for
15537 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
15539 PR tree-optimization/56443
15540 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
15541 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
15542 to type_for_mode langhook.
15544 2013-02-25 Matt Turner <mattst88@gmail.com>
15546 * doc/invoke.texi: Document r4700.
15548 2013-02-25 Richard Biener <rguenther@suse.de>
15550 PR tree-optimization/56175
15551 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
15553 (simplify_bitwise_binary): ... here. Also guard the conversion
15554 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
15556 2013-02-25 Catherine Moore <clm@codesourcery.com>
15559 2013-02-24 Catherine Moore <clm@codesourcery.com>
15560 Maciej W. Rozycki <macro@codesourcery.com>
15561 Tom de Vries <tom@codesourcery.com>
15562 Nathan Sidwell <nathan@codesourcery.com>
15563 Iain Sandoe <iain@codesourcery.com>
15564 Nathan Froyd <froydnj@codesourcery.com>
15565 Chao-ying Fu <fu@mips.com>
15567 * doc/extend.texi: (micromips, nomicromips, nocompression):
15568 Document new function attributes.
15569 * doc/invoke.texi (minterlink-compressed, mmicromips,
15570 m14k, m14ke, m14kec): Document new options.
15571 (minterlink-mips16): Update documentation.
15572 * doc/md.texi (ZC, ZD): Document new constraints.
15573 * configure.ac (gcc_cv_as_micromips): Check if linker
15574 supports the .set micromips directive.
15575 * configure: Regenerate.
15576 * config.in: Regenerate.
15577 * config/mips/mips-tables.opt: Regenerate.
15578 * config/mips/micromips.md: New file.
15579 * constraints.md (ZC, AD): New constraints.
15580 * config/mips/predicates.md (movep_src_register): New predicate.
15581 (movep_src_operand): New predicate.
15582 (non_volatile_mem_operand): New predicate.
15583 * config/mips/mips.md (multimem): New type.
15584 (length): Differentiate between 17-bit and 18-bit branch offsets.
15585 (MOVEP1, MOVEP2): New mode iterator.
15586 (mov_<load>l): Use ZC constraint.
15587 (mov_<load>r): Likewise.
15588 (mov_<store>l): Likewise.
15589 (mov_<store>r): Likewise.
15590 (*branch_equality<mode>_inverted): Add microMIPS support.
15591 (*branch_equality<mode>): Likewise.
15592 (*jump_absolute): Likewise.
15593 (indirect_jump_<mode>): Likewise.
15594 (tablejump_<mode>): Likewise.
15595 (<optab>_internal): Likewise.
15596 (sibcall_internal): Likewise.
15597 (sibcall_value_internal): Likewise.
15598 (prefetch): Use constraint ZD.
15599 * config/mips/mips.opt (minterlink-compressed): New option.
15600 (minterlink-mips16): Now an alias for minterlink-compressed.
15601 (mmicromips): New option.
15602 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15603 (compare_and_swap_12): Likewise.
15604 (sync_add<mode>): Likewise.
15605 (sync_<optab>_12): Likewise.
15606 (sync_old_<optab>_12): Likewise.
15607 (sync_new_<optab>_12): Likewise.
15608 (sync_nand_12): Likewise.
15609 (sync_old_nand_12): Likewise.
15610 (sync_new_nand_12): Likewise.
15611 (sync_sub<mode>): Likewise.
15612 (sync_old_add<mode>): Likewise.
15613 (sync_old_sub<mode>): Likewise.
15614 (sync_new_add<mode>): Likewise.
15615 (sync_new_sub<mode>): Likewise.
15616 (sync_<optab><mode>): Likewise.
15617 (sync_old_<optab><mode>): Likewise.
15618 (sync_new_<optab><mode>): Likewise.
15619 (sync_nand<mode>): Likewise.
15620 (sync_old_nand<mode>): Likewise.
15621 (sync_new_nand<mode>): Likewise.
15622 (sync_lock_test_and_set<mode>): Likewise.
15623 (test_and_set_12): Likewise.
15624 (atomic_compare_and_swap<mode>): Likewise.
15625 (atomic_exchange<mode>_llsc): Likewise.
15626 (atomic_fetch_add<mode>_llsc): Likewise.
15627 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15628 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15629 (umips_save_restore_pattern_p): Likewise.
15630 (umips_load_store_pair_p): Likewise.
15631 (umips_output_load_store_pair): Likewise.
15632 (umips_movep_target_p): Likewise.
15633 (umips_12bit_offset_address_p): Likewise.
15634 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15635 (mips_base_mips16): Rename this...
15636 (mips_base_compression_flags): ...to this. Update all uses.
15637 (mips_attribute_table): Add micromips, nomicromips and nocompression.
15638 (mips_mips16_decl_p): Delete.
15639 (mips_nomips16_decl_p): Delete.
15640 (mips_get_compress_on_flags): New function.
15641 (mips_get_compress_off_flags): New function.
15642 (mips_get_compress_mode): New function.
15643 (mips_get_compress_on_name): New function.
15644 (mips_get_compress_off_name): New function.
15645 (mips_insert_attributes): Support multiple compression types.
15646 (mips_merge_decl_attributes): Likewise.
15647 (umips_12bit_offset_address_p): New function.
15648 (mips_start_function_definition): Emit .set micromips directive.
15649 (mips_call_may_need_jalx_p): New function.
15650 (mips_function_ok_for_sibcall): Add microMIPS support.
15651 (mips_print_operand_punctuation): Support short delay slots and
15653 (umips_swm_mask, umips_swm_encoding): New.
15654 (umips_build_save_restore): New function.
15655 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15656 (was_mips16_p): Remove.
15657 (old_compression_mode): New.
15658 (mips_set_compression_mode): New function.
15659 (mips_set_current_function): Add microMIPS support.
15660 (mips_option_override): Likewise.
15661 (umips_save_restore_pattern_p): New function.
15662 (umips_output_save_restore): New function.
15663 (umips_load_store_pair_p_1): New function.
15664 (umips_load_store_pair_p): New function.
15665 (umips_output_load_store_pair_1): New function.
15666 (umips_output_load_store_pair): New function.
15667 (umips_movep_target_p) New function.
15668 (mips_prepare_pch_save): Add microMIPS support.
15669 * config/mips/mips.h (TARGET_COMPRESSION): New.
15670 (TARGET_CPU_CPP_BUILTINS): Update macro
15671 to use new compression flags and to support microMIPS.
15672 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15673 (MIPS_ARCH_FLOAT_SPEC): Likewise.
15674 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15675 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15676 (ASM_SPEC): Support mmicromips and mno-micromips.
15677 (M16STORE_REG_P): New macro.
15678 (MIPS_CALL): Support TARGET_MICROMIPS.
15679 (MICROMIPS_J): New macro.
15680 (mips_base_mips16): Rename this...
15681 (mips_base_compression_flags): ...to this.
15682 (UMIPS_12BIT_OFFSET_P): New macro.
15683 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15684 (MULTILIB_DIRNAMES): Likewise.
15686 2013-02-25 Tom de Vries <tom@codesourcery.com>
15688 PR rtl-optimization/56131
15689 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
15690 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
15691 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
15693 2013-02-25 Tobias Burnus <burnus@net-b.de>
15695 * doc/invoke.texi (-fsanitize=): Move from optimization
15696 to debugging options.
15698 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
15700 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
15702 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
15703 Alexander Monakov <amonakov@ispras.ru>
15705 PR middle-end/56077
15706 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
15707 flush pending lists also on non-jumps. Adjust comment.
15709 2013-02-24 Catherine Moore <clm@codesourcery.com>
15710 Maciej W. Rozycki <macro@codesourcery.com>
15711 Tom de Vries <tom@codesourcery.com>
15712 Nathan Sidwell <nathan@codesourcery.com>
15713 Iain Sandoe <iain@codesourcery.com>
15714 Nathan Froyd <froydnj@codesourcery.com>
15715 Chao-ying Fu <fu@mips.com>
15717 * doc/extend.texi: (micromips, nomicromips, nocompression):
15718 Document new function attributes.
15719 * doc/invoke.texi (minterlink-compressed, mmicromips,
15720 m14k, m14ke, m14kec): Document new options.
15721 (minterlink-mips16): Update documentation.
15722 * doc/md.texi (ZC, ZD): Document new constraints.
15723 * configure.ac (gcc_cv_as_micromips): Check if linker
15724 supports the .set micromips directive.
15725 * configure: Regenerate.
15726 * config.in: Regenerate.
15727 * config/mips/mips-tables.opt: Regenerate.
15728 * config/mips/micromips.md: New file.
15729 * constraints.md (ZC, AD): New constraints.
15730 * config/mips/predicates.md (movep_src_register): New predicate.
15731 (movep_src_operand): New predicate.
15732 (non_volatile_mem_operand): New predicate.
15733 * config/mips/mips.md (multimem): New type.
15734 (length): Differentiate between 17-bit and 18-bit branch offsets.
15735 (MOVEP1, MOVEP2): New mode iterator.
15736 (mov_<load>l): Use ZC constraint.
15737 (mov_<load>r): Likewise.
15738 (mov_<store>l): Likewise.
15739 (mov_<store>r): Likewise.
15740 (*branch_equality<mode>_inverted): Add microMIPS support.
15741 (*branch_equality<mode>): Likewise.
15742 (*jump_absolute): Likewise.
15743 (indirect_jump_<mode>): Likewise.
15744 (tablejump_<mode>): Likewise.
15745 (<optab>_internal): Likewise.
15746 (sibcall_internal): Likewise.
15747 (sibcall_value_internal): Likewise.
15748 (prefetch): Use constraint ZD.
15749 * config/mips/mips.opt (minterlink-compressed): New option.
15750 (minterlink-mips16): Now an alias for minterlink-compressed.
15751 (mmicromips): New option.
15752 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15753 (compare_and_swap_12): Likewise.
15754 (sync_add<mode>): Likewise.
15755 (sync_<optab>_12): Likewise.
15756 (sync_old_<optab>_12): Likewise.
15757 (sync_new_<optab>_12): Likewise.
15758 (sync_nand_12): Likewise.
15759 (sync_old_nand_12): Likewise.
15760 (sync_new_nand_12): Likewise.
15761 (sync_sub<mode>): Likewise.
15762 (sync_old_add<mode>): Likewise.
15763 (sync_old_sub<mode>): Likewise.
15764 (sync_new_add<mode>): Likewise.
15765 (sync_new_sub<mode>): Likewise.
15766 (sync_<optab><mode>): Likewise.
15767 (sync_old_<optab><mode>): Likewise.
15768 (sync_new_<optab><mode>): Likewise.
15769 (sync_nand<mode>): Likewise.
15770 (sync_old_nand<mode>): Likewise.
15771 (sync_new_nand<mode>): Likewise.
15772 (sync_lock_test_and_set<mode>): Likewise.
15773 (test_and_set_12): Likewise.
15774 (atomic_compare_and_swap<mode>): Likewise.
15775 (atomic_exchange<mode>_llsc): Likewise.
15776 (atomic_fetch_add<mode>_llsc): Likewise.
15777 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15778 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15779 (umips_save_restore_pattern_p): Likewise.
15780 (umips_load_store_pair_p): Likewise.
15781 (umips_output_load_store_pair): Likewise.
15782 (umips_movep_target_p): Likewise.
15783 (umips_12bit_offset_address_p): Likewise.
15784 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15785 (mips_base_mips16): Rename this...
15786 (mips_base_compression_flags): ...to this. Update all uses.
15787 (mips_attribute_table): Add micromips, nomicromips and nocompression.
15788 (mips_mips16_decl_p): Delete.
15789 (mips_nomips16_decl_p): Delete.
15790 (mips_get_compress_on_flags): New function.
15791 (mips_get_compress_off_flags): New function.
15792 (mips_get_compress_mode): New function.
15793 (mips_get_compress_on_name): New function.
15794 (mips_get_compress_off_name): New function.
15795 (mips_insert_attributes): Support multiple compression types.
15796 (mips_merge_decl_attributes): Likewise.
15797 (umips_12bit_offset_address_p): New function.
15798 (mips_start_function_definition): Emit .set micromips directive.
15799 (mips_call_may_need_jalx_p): New function.
15800 (mips_function_ok_for_sibcall): Add microMIPS support.
15801 (mips_print_operand_punctuation): Support short delay slots and
15803 (umips_swm_mask, umips_swm_encoding): New.
15804 (umips_build_save_restore): New function.
15805 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15806 (was_mips16_p): Remove.
15807 (old_compression_mode): New.
15808 (mips_set_compression_mode): New function.
15809 (mips_set_current_function): Add microMIPS support.
15810 (mips_option_override): Likewise.
15811 (umips_save_restore_pattern_p): New function.
15812 (umips_output_save_restore): New function.
15813 (umips_load_store_pair_p_1): New function.
15814 (umips_load_store_pair_p): New function.
15815 (umips_output_load_store_pair_1): New function.
15816 (umips_output_load_store_pair): New function.
15817 (umips_movep_target_p) New function.
15818 (mips_prepare_pch_save): Add microMIPS support.
15819 * config/mips/mips.h (TARGET_COMPRESSION): New.
15820 (TARGET_CPU_CPP_BUILTINS): Update macro
15821 to use new compression flags and to support microMIPS.
15822 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15823 (MIPS_ARCH_FLOAT_SPEC): Likewise.
15824 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15825 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15826 (ASM_SPEC): Support mmicromips and mno-micromips.
15827 (M16STORE_REG_P): New macro.
15828 (MIPS_CALL): Support TARGET_MICROMIPS.
15829 (MICROMIPS_J): New macro.
15830 (mips_base_mips16): Rename this...
15831 (mips_base_compression_flags): ...to this.
15832 (UMIPS_12BIT_OFFSET_P): New macro.
15833 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15834 (MULTILIB_DIRNAMES): Likewise.
15836 2013-02-24 Jakub Jelinek <jakub@redhat.com>
15839 * target-globals.c (save_target_globals): For init_reg_sets and
15840 target_reinit remporarily set this_fn_optabs to this_target_optabs.
15842 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
15844 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
15845 * config/aarch64/t-aarch64
15846 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
15848 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
15850 PR inline-asm/56148
15851 * lra-constraints.c (process_alt_operands): Reload operand
15852 conflicting with earlier clobber only if no more other conflicting
15855 2013-02-22 Jakub Jelinek <jakub@redhat.com>
15858 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
15859 if not linking a shared library.
15861 2013-02-22 Seth LaForge <sethml@google.com>
15863 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
15865 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
15867 * config/arm/arm.md (split for extendsidi): Update condition.
15868 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
15869 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
15870 (qhs_zextenddi_cstr): Likewise.
15872 2013-02-21 Jakub Jelinek <jakub@redhat.com>
15874 PR middle-end/56420
15875 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
15876 avoid signed wrapping.
15877 (expand_mult): Handle properly multiplication by
15878 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
15879 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
15880 in the compiler if coeff is HOST_WIDE_INT_MIN.
15881 (expand_divmod): Don't make ext_op1 static, change it's type to
15882 uhwi. Avoid undefined behavior in -INTVAL (op1).
15884 PR rtl-optimization/50339
15885 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
15887 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
15888 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
15889 into splitting_ashiftrt field.
15890 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
15892 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
15895 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
15897 PR middle-end/56108
15898 * trans-mem.c (execute_tm_mark): Do not expand transactions that
15899 are sure to go irrevocable.
15901 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
15903 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
15904 scalars are valid operands.
15906 2013-02-21 Martin Jambor <mjambor@suse.cz>
15908 PR tree-optimization/56310
15909 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
15910 only matching indices and non-negative final offsets.
15911 (intersect_aggregates_with_edge): Pass src_idx to
15912 agg_replacements_to_vector. Pass src_idx insstead of index to
15913 intersect_with_agg_replacements.
15915 2013-02-21 Martin Jambor <mjambor@suse.cz>
15917 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
15918 instead of hard-wired defaults.
15920 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
15922 * doc/invoke.texi (MIPS Options): Update documentation of the
15923 floating-point multiply-accumulate instruction restrictions.
15925 2013-02-21 Kostya Serebryany <kcc@google.com>
15927 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
15928 asan_shadow_offset on x86_64 linux.
15930 2013-02-21 Richard Biener <rguenther@suse.de>
15932 PR tree-optimization/56415
15934 2013-02-11 Richard Biener <rguenther@suse.de>
15936 PR tree-optimization/56273
15937 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
15940 2013-02-21 Jakub Jelinek <jakub@redhat.com>
15943 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
15946 PR inline-asm/56405
15947 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
15948 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
15950 2013-02-20 Jan Hubicka <jh@suse.cz>
15952 PR tree-optimization/56265
15953 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
15954 when target is referenced for first time.
15956 2013-02-20 Richard Biener <rguenther@suse.de>
15958 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
15959 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
15960 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
15961 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
15962 not return anything.
15963 (rename_ssa_copies): Do not remove unused locals.
15964 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
15965 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
15966 * passes.c (execute_function_todo): Do not schedule unused locals
15967 removal if cleanup_tree_cfg did something.
15968 * tree-ssa-live.c (remove_unused_locals): Dump statistics
15969 about the number of removed locals.
15971 2013-02-20 Richard Biener <rguenther@suse.de>
15973 PR tree-optimization/56398
15974 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
15976 2013-02-20 Martin Jambor <mjambor@suse.cz>
15978 PR tree-optimization/55334
15979 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
15980 restricted pointers to arrays.
15982 2013-02-20 Richard Biener <rguenther@suse.de>
15983 Jakub Jelinek <jakub@redhat.com>
15985 PR tree-optimization/56396
15986 * tree-ssa-ccp.c (n_const_val): New static variable.
15987 (get_value): Return NULL for SSA names we don't have a lattice
15989 (ccp_initialize): Initialize n_const_val.
15990 * tree-ssa-copy.c (n_copy_of): New static variable.
15991 (init_copy_prop): Initialize n_copy_of.
15992 (get_value): Return NULL_TREE for SSA names we don't have a
15995 2013-02-20 Martin Jambor <mjambor@suse.cz>
15997 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
15999 2013-02-20 Richard Biener <rguenther@suse.de>
16001 * genpreds.c (write_lookup_constraint): Do not compare first
16002 letter of the constraint again.
16004 2013-02-20 Richard Biener <rguenther@suse.de>
16006 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16008 (get_use_iv_cost): Terminate hashtable walk when coming across
16011 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
16013 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
16014 reassociation for avx2 targets.
16016 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16018 * config/microblaze/microblaze.c: microblaze_has_clz = 0
16019 Add version check for v8.10.a to enable microblaze_has_clz
16020 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
16021 version and TARGET_PATTERN_COMPARE check
16022 * config/microblaze/microblaze.md: New clzsi2 instruction
16024 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16026 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
16027 function before branching.
16029 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
16031 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
16033 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
16035 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
16037 PR middle-end/55889
16038 * sel-sched.c: Include ira.h.
16039 (implicit_clobber_conflict_p): New function.
16040 (moveup_expr): Use it.
16041 * Makefile.in (sel-sched.o): Depend on ira.h.
16043 2013-02-19 Richard Biener <rguenther@suse.de>
16045 PR tree-optimization/56384
16046 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
16047 (vn_hash_type): Split out from ...
16048 (vn_hash_constant_with_type): ... here.
16049 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
16050 (vn_phi_eq): Compare types from vn_phi_s structure.
16051 (vn_phi_lookup): Populate vn_phi_s type.
16052 (vn_phi_insert): Likewise.
16054 2013-02-19 Jakub Jelinek <jakub@redhat.com>
16056 PR tree-optimization/56350
16057 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
16058 if haven't found reduction or nested cycle operand, rather than
16059 asserting we must find it.
16061 PR tree-optimization/56381
16062 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
16065 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
16066 Jakub Jelinek <jakub@redhat.com>
16069 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
16070 (swap_optab_enable): Same.
16071 (init_all_optabs): Use argument instead of global.
16072 * tree.h (struct tree_optimization_option): New field target_optabs.
16073 * expr.h (init_all_optabs): Add argument to prototype.
16074 (TREE_OPTIMIZATION_OPTABS): New.
16075 (save_optabs_if_changed): Protoize.
16076 * optabs.h: Declare this_fn_optabs.
16077 * optabs.c (save_optabs_if_changed): New.
16078 Declare this_fn_optabs.
16079 (init_optabs): Add argument to init_all_optabs() call.
16080 * function.c (invoke_set_current_function_hook): Handle per
16082 * function.h (struct function): New field optabs.
16083 * config/mips/mips.c (mips_set_mips16_mode): Handle when
16084 optimization_current_node has changed.
16085 * target-globals.h (save_target_globals_default_opts): Protoize.
16086 * target-globals.c (save_target_globals_default_opts): New.
16088 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
16091 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
16092 registers %fr12 and %fr12R as call used.
16095 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
16096 and HImode, require all displacements to be an integer multiple of
16098 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
16099 only allow QImode and HImode when reload is in progress and strict is
16100 true. Likewise for symbolic addresses. Use base14_operand to check
16101 displacements in REG+BASE addresses.
16103 2013-02-18 Richard Biener <rguenther@suse.de>
16105 PR tree-optimization/56366
16106 * tree-vect-loop.c (get_initial_def_for_induction): Properly
16107 handle sign-conversion of outer-loop initial induction value.
16109 2013-02-18 Richard Biener <rguenther@suse.de>
16111 PR middle-end/56349
16112 * cfghooks.c (merge_blocks): If we merge a latch into another
16113 block adjust references to it.
16114 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
16115 (verify_loop_structure): Verify that a recorded latch is in fact
16118 2013-02-18 Richard Biener <rguenther@suse.de>
16120 PR tree-optimization/56321
16121 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
16122 order SSA name release and virtual operand unlinking.
16124 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16126 * config/microblaze/microblaze.md (save_stack_block): Define.
16127 (restore_stack_block): Likewise.
16129 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16131 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
16132 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
16133 * config/microblaze/microblaze.c (microblaze_option_override):
16134 Bail out early for PIC modes when target does not support PIC.
16136 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16138 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
16139 Replace with a microblaze version.
16140 (microblaze_trampoline_init): Adapt for microblaze.
16141 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
16144 2013-02-16 Jakub Jelinek <jakub@redhat.com>
16145 Dodji Seketeli <dodji@redhat.com>
16148 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
16149 (instrument_mem_region_access): Do not forget to always put
16150 instrumentation of the of 'base' and 'base + len' in a "if (len !=
16151 0) statement, even for cases where either 'base' or 'base + len'
16152 are not instrumented -- because they have been previously
16153 instrumented. Simplify the logic by putting all the statements
16154 instrument 'base + len' inside a sequence, and then insert that
16155 sequence right before the current insertion point. Then, to
16156 instrument 'base + len', just get an iterator on that statement.
16157 And do not forget to update the pointer to iterator the function
16158 received as argument.
16160 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
16162 PR rtl-optimization/56348
16163 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
16165 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
16167 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
16168 (clean_graph_dump_file): Pass base to start_graph_dump.
16170 2013-02-14 Richard Henderson <rth@redhat.com>
16173 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
16175 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
16177 * collect2-aix.h: Define F_LOADONLY.
16179 2013-02-14 Richard Biener <rguenther@suse.de>
16182 * varasm.c (output_constant_def_1): Get the decl representing
16183 the constant as argument.
16184 (output_constant_def): Wrap output_constant_def_1.
16185 (make_decl_rtl): Use output_constant_def_1 with the decl
16186 representing the constant.
16187 (build_constant_desc): Optionally re-use a decl already
16188 representing the constant.
16189 (tree_output_constant_def): Adjust.
16191 2013-02-14 Dodji Seketeli <dodji@redhat.com>
16194 * asan.c (instrument_builtin_call): Really put the length of the
16195 second source argument into src1_len.
16197 2013-02-13 Jakub Jelinek <jakub@redhat.com>
16199 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
16200 argument. If it is false, don't create edge from then_bb to
16202 (insert_if_then_before_iter): Pass true to it.
16203 (build_check_stmt): Pass false to it.
16204 (transform_statements): Flush hash table only on extended basic
16205 block boundaries, rather than at the beginning of every bb.
16206 Don't flush hash table on nonfreeing_call_p calls.
16207 * tree-flow.h (nonfreeing_call_p): New prototype.
16208 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
16210 2013-02-13 David S. Miller <davem@davemloft.net>
16212 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
16214 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
16217 * ira.c (max_regno_before_ira): Move from ...
16219 (fix_reg_equiv_init): Use max_regno_before_ira instead of
16222 2013-02-13 Jakub Jelinek <jakub@redhat.com>
16224 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
16226 2013-02-13 Richard Biener <rguenther@suse.de>
16229 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
16230 globals in MEM_REFs.
16232 2013-02-13 Richard Biener <rguenther@suse.de>
16234 * loop-init.c (loop_optimizer_init): Clear loop state when
16235 re-initializing preserved loops.
16236 * loop-unswitch.c (unswitch_single_loop): Return whether
16237 we unswitched the loop. Do not verify loop state here.
16238 (unswitch_loops): When we unswitched a loop discover new loops.
16240 2013-02-13 Kostya Serebryany <kcc@google.com>
16242 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
16244 * sanitizer.def: Rename __asan_init to __asan_init_v1.
16246 2013-02-12 Dodji Seketeli <dodji@redhat.com>
16248 Avoid instrumenting duplicated memory access in the same basic block
16249 * Makefile.in (asan.o): Add new dependency on hash-table.h
16250 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
16251 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
16252 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
16253 (free_mem_ref_resources, has_mem_ref_been_instrumented)
16254 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
16255 (get_mem_ref_of_assignment): New functions.
16256 (get_mem_refs_of_builtin_call): Extract from
16257 instrument_builtin_call and tweak a little bit to make it fit with
16259 (instrument_builtin_call): Use the new
16260 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
16261 of is_gimple_builtin_call.
16262 (instrument_derefs, instrument_mem_region_access): Insert the
16263 instrumented memory reference into the hash table.
16264 (maybe_instrument_assignment): Renamed instrument_assignment into
16265 this, and change it to advance the iterator when instrumentation
16266 actually happened and return true in that case. This makes it
16267 homogeneous with maybe_instrument_assignment, and thus give a
16268 chance to callers to be more 'regular'.
16269 (transform_statements): Clear the memory reference hash table
16270 whenever we enter a new BB, when we cross a function call, or when
16271 we are done transforming statements. Use
16272 maybe_instrument_assignment instead of instrumentation. No more
16273 need to special case maybe_instrument_assignment and advance the
16274 iterator after calling it; it's now handled just like
16275 maybe_instrument_call. Update comment.
16277 2013-02-13 Richard Biener <rguenther@suse.de>
16279 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
16280 Fix loop discovery code.
16282 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
16284 PR inline-asm/56148
16285 * lra-constraints.c (process_alt_operands): Match early clobber
16286 operand with itself. Check conflicts with earlyclobber only if
16287 the operand is not reloaded. Prefer to reload conflicting operand
16288 if earlyclobber and matching operands are the same.
16290 2013-02-12 Richard Biener <rguenther@suse.de>
16293 * lto-streamer-out.c (write_symbol): Do not output symbols
16294 for hard register variables.
16296 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
16299 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
16300 (umulsidi3_insn, mulsidi3_insn): New insns.
16302 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
16304 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
16305 (struct tune_params): Add vec_costs field.
16306 * config/arm/arm.c (arm_builtin_vectorization_cost)
16307 (arm_add_stmt_cost): New functions.
16308 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
16309 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
16310 (arm_default_vec_cost): New struct of type cpu_vec_costs.
16311 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
16312 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
16313 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
16314 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
16316 2013-02-12 Richard Biener <rguenther@suse.de>
16319 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
16320 decls again if possible.
16322 2013-02-12 Richard Biener <rguenther@suse.de>
16324 PR middle-end/56288
16325 * tree-ssa.c (verify_ssa_name): Fix check, move
16326 SSA_NAME_IN_FREE_LIST check up.
16328 2013-02-12 Jakub Jelinek <jakub@redhat.com>
16329 Steven Bosscher <steven@gcc.gnu.org>
16331 PR rtl-optimization/56151
16332 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
16333 equal to op0 or op1, and last_insn pattern is CODE operation
16334 with MEM dest and one of the operands matches that MEM.
16336 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
16338 * doc/extend.texi: Document Function Multiversioning and "default"
16339 parameter string to target attribute.
16340 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
16341 target attribute parameter is "default".
16342 (ix86_compare_version_priority): Remove checks for target attribute.
16343 (ix86_mangle_function_version_assembler_name): Change error to sorry.
16344 Remove check for target attribute equal to NULL. Add assert.
16345 (ix86_generate_version_dispatcher_body): Change error to sorry.
16347 2013-02-11 Iain Sandoe <iain@codesourcery.com>
16348 Jack Howarth <howarth@bromo.med.uc.edu>
16349 Patrick Marlier <patrick.marlier@gmail.com>
16352 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
16353 define ENDFILE_SPEC as TM_DESTRUCTOR.
16354 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
16356 2013-02-11 Alexander Potapenko <glider@google.com>
16357 Jack Howarth <howarth@bromo.med.uc.edu>
16358 Jakub Jelinek <jakub@redhat.com>
16361 * config/darwin.c (cdtor_record): Rename ctor_record.
16362 (sort_cdtor_records): Rename sort_ctor_records.
16363 (finalize_dtors): New routine to sort destructors by
16364 priority before use in assemble_integer.
16365 (machopic_asm_out_destructor): Use finalize_dtors if needed.
16367 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
16369 PR rtl-optimization/56275
16370 * simplify-rtx.c (avoid_constant_pool_reference): Check that
16371 offset is non-negative and less than cmode size before
16372 calling simplify_subreg.
16374 2013-02-11 Richard Biener <rguenther@suse.de>
16376 PR tree-optimization/56264
16377 * cfgloop.h (fix_loop_structure): Adjust prototype.
16378 * loop-init.c (fix_loop_structure): Return the number of
16379 newly discovered loops.
16380 * tree-cfgcleanup.c (repair_loop_structures): When new loops
16381 are discovered, do a full loop-closed SSA rewrite.
16383 2013-02-11 Richard Biener <rguenther@suse.de>
16385 PR tree-optimization/56273
16386 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16388 (check_array_ref): Fix missing newline in dumps.
16389 (search_for_addr_array): Likewise.
16391 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
16393 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
16395 2013-02-09 Jakub Jelinek <jakub@redhat.com>
16398 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
16400 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
16402 PR rtl-optimization/56246
16403 * lra-constraints.c (simplify_operand_subreg): Try to reuse
16405 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
16406 constraints are satisfied.
16408 2013-02-08 Jeff Law <law@redhat.com>
16411 * emit-rtl.c (reg_is_parm_p): New function.
16412 * regs.h (reg_is_parm_p): New prototype.
16413 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
16414 callee-clobbered registers.
16416 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
16419 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
16420 If there is no implicit builtin declaration, just return NULL.
16422 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
16424 * config/i386/sse.md (FMAMODEM): New mode iterator.
16425 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
16426 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
16428 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
16430 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
16431 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
16432 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
16434 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16436 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
16437 (microblaze*-*-elf): Likewise.
16438 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
16440 * config/microblaze/microblaze-c.c: Add builtin defines for
16441 _LITTLE_ENDIAN and _BIG_ENDIAN.
16442 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
16443 add to TARGET_DEFAULT flags.
16444 Expand ASM_SPEC and LINK_SPEC.
16445 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
16446 * config/microblaze/microblaze.md: Update extendsidi2 and
16447 movdi_internal instructions to use low-order / high-order reg
16449 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
16450 options and inversemask / mask of LITTLE_ENDIAN.
16451 * config/microblaze/t-microblaze: Expand multilib options to
16452 include mlittle-endian (le) and update exceptions patterns.
16454 2013-02-08 Jakub Jelinek <jakub@redhat.com>
16456 PR rtl-optimization/56195
16457 * lra-constraints.c (get_reload_reg): Don't reuse regs
16458 if they have smaller mode than requested, if they have
16459 wider mode than requested, try to return a SUBREG.
16461 PR tree-optimization/56250
16462 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
16463 if type is unsigned and code isn't MULT_EXPR.
16465 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
16467 PR tree-optimization/56064
16468 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
16469 bits according to mode.
16470 * fixed-value.h (fixed_from_double_int)
16471 (const_fixed_from_double_int): Adjust comments.
16473 2013-02-08 Richard Biener <rguenther@suse.de>
16476 * lto-streamer.h (struct data_in): Remove current_file, current_line
16477 and current_col members.
16478 * lto-streamer-out.c (lto_output_location): Stream changed bits
16479 en-block for efficiency.
16480 * lto-streamer-in.c (clear_line_info): Remove.
16481 (lto_input_location): Cache current file, line and column
16482 globally via local statics. Read changed bits en-block.
16483 (input_function): Do not call clear_line_info.
16484 (lto_read_body): Likewise.
16485 (lto_input_toplevel_asms): Likewise.
16487 2013-02-08 Michael Matz <matz@suse.de>
16489 PR tree-optimization/52448
16490 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
16491 (nt_call_phase): New static.
16492 (add_or_mark_expr): Only mark accesses with newer phase than any
16494 (nonfreeing_call_p): New.
16495 (nt_init_block): Update nt_call_phase, mark blocks as visited.
16496 (nt_fini_block): Keep blocks marked as visited.
16497 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
16499 2013-02-08 Richard Biener <rguenther@suse.de>
16501 * ira.c (ira): Free broken dominator information.
16503 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
16505 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
16507 2013-02-08 Marek Polacek <polacek@redhat.com>
16509 * cfgloop.c (verify_loop_structure): Add more checking of headers.
16511 2013-02-08 Richard Biener <rguenther@suse.de>
16513 PR middle-end/56181
16514 * cfgloop.h (flow_loops_find): Adjust.
16515 (bb_loop_header_p): Declare.
16516 * cfgloop.c (bb_loop_header_p): New function split out from ...
16517 (flow_loops_find): ... here. Adjust function signature,
16518 support incremental loop structure update.
16519 (verify_loop_structure): Cleanup. Verify a loop is a loop.
16520 * cfgloopmanip.c (fix_loop_structure): Move ...
16521 * loop-init.c (fix_loop_structure): ... here.
16522 (apply_loop_flags): Split out from ...
16523 (loop_optimizer_init): ... here.
16524 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
16525 in incremental mode, only remove dead loops here.
16527 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
16530 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
16531 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
16532 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
16533 (*round<mode>3.libgcc): New insns for fixed-modes.
16534 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
16535 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
16536 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
16537 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
16538 implementations. Define to __builtin_avr_absFX,
16539 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
16540 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
16541 __builtin_avr_countlsFX, respectively.
16542 * config/avr/avr-c.c (target.h): Include it.
16543 (enum avr_builtin_id): New enum.
16544 (avr_resolve_overloaded_builtin): New static function.
16545 (avr_register_target_pragmas): Use it to set
16546 targetm.resolve_overloaded_builtin.
16547 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
16548 tree nodes used by DEF_BUILTIN.
16549 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
16550 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
16551 <AVR_BUILTIN_xxBITS>: Same.
16553 2013-02-08 Richard Biener <rguenther@suse.de>
16555 * cfgloop.c (verify_loop_structure): Properly handle
16556 a loop exiting to another loop header.
16557 * ira-int.h (ira_loops): Remove.
16558 * ira.c (ira_loops): Remove.
16559 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
16560 (do_reload): Use loop_optimizer_finalize.
16561 * ira-build.c (create_loop_tree_nodes): Use get_loops and
16562 number_of_loops to access the loop tree.
16563 (more_one_region_p): Likewise.
16564 (finish_loop_tree_nodes): Likewise.
16565 (rebuild_regno_allocno_maps): Likewise.
16566 (mark_loops_for_removal): Likewise.
16567 (mark_all_loops_for_removal): Likewise.
16568 (remove_unnecessary_regions): Likewise.
16569 (ira_build): Likewise.
16570 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
16572 2013-02-08 Richard Biener <rguenther@suse.de>
16574 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
16575 * ipa-pure-const.c (analyze_function): Avoid calling
16576 mark_irreducible_loops twice.
16577 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
16579 2013-02-07 David S. Miller <davem@davemloft.net>
16581 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
16583 * var-tracking.c (vt_add_function_parameter): Test the presence of
16584 HAVE_window_save properly and do not remap argument registers when
16585 we have a leaf function.
16587 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
16590 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
16592 * config/i386/i386.c (ix86_print_operand): Ditto.
16594 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
16596 * lra-constraints.c (process_alt_operands): Fix recently added comment.
16598 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
16600 PR rtl-optimization/56225
16601 * lra-constraints.c (process_alt_operands): Check that reload hard
16602 reg can hold value for strict_low_part.
16604 2013-02-07 Jakub Jelinek <jakub@redhat.com>
16607 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
16608 dwarf2out_end_function.
16609 (in_first_function_p, maybe_at_text_label_p,
16610 first_loclabel_num_not_at_text_label): New variables.
16611 (dwarf2out_var_location): In the first function find out
16612 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
16613 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
16616 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
16618 PR rtl-optimization/56178
16619 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
16620 SUBREG of a register. Tidy up related block of code.
16621 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
16622 note if the source is a register or a SUBREG of a register.
16624 2013-02-07 Jakub Jelinek <jakub@redhat.com>
16627 * config/rs6000/rs6000.md (ptrm): New mode attr.
16628 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
16629 call_value_indirect_aix<pttrsize>,
16630 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
16633 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
16635 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
16636 if -bnortl. Convert to strcmp and strncmp.
16638 2013-02-07 Alan Modra <amodra@gmail.com>
16641 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
16642 addresses won't wrap when offsetting.
16643 (rs6000_secondary_reload): Provide secondary reloads needed for
16644 wrapping LO_SUM addresses.
16646 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
16648 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
16649 MACH, just __MACH__.
16651 2013-02-06 Richard Biener <rguenther@suse.de>
16653 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
16654 instead of calling fix_loop_structure.
16656 2013-02-06 Jakub Jelinek <jakub@redhat.com>
16658 PR middle-end/56217
16659 * omp-low.c (use_pointer_for_field): Return false if
16660 lower_send_shared_vars doesn't generate any copy-out code.
16662 2013-02-06 Tom de Vries <tom@codesourcery.com>
16664 PR rtl-optimization/56131
16665 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
16666 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
16667 of the label is NULL. Add comment.
16669 2013-02-05 Jakub Jelinek <jakub@redhat.com>
16671 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
16674 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
16675 (STATIC_LIBTSAN_LIBS): Likewise.
16676 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
16677 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
16678 is defined, don't add anything else beyond that.
16679 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
16680 (LINK_COMMAND_SPEC): Use them.
16682 PR tree-optimization/56205
16683 * tree-stdarg.c (check_all_va_list_escapes): Return true if
16684 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
16685 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
16687 2013-02-05 Richard Biener <rguenther@suse.de>
16689 PR tree-optimization/53342
16690 PR tree-optimization/53185
16691 * tree-vectorizer.h (vect_check_strided_load): Remove.
16692 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
16693 not disallow peeling for vectorized strided loads.
16694 (vect_check_strided_load): Make static and simplify.
16695 (vect_analyze_data_refs): Adjust.
16696 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
16697 correctly when vectorizing strided loads.
16699 2013-02-05 Richard Biener <rguenther@suse.de>
16701 * doc/install.texi: Refer to ISL, not PPL.
16703 2013-02-05 Jan Hubicka <jh@suse.cz>
16705 PR tree-optimization/55789
16706 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
16708 2013-02-05 Jan Hubicka <jh@suse.cz>
16710 PR tree-optimization/55789
16711 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
16712 the dead call anyway.
16714 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
16717 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
16719 2013-02-04 Alexander Potapenko <glider@google.com>
16720 Jack Howarth <howarth@bromo.med.uc.edu>
16721 Jakub Jelinek <jakub@redhat.com>
16724 * config/darwin.c (sort_ctor_records): Stabilized qsort
16725 on constructor priority by using original position.
16726 (finalize_ctors): New routine to sort constructors by
16727 priority before use in assemble_integer.
16728 (machopic_asm_out_constructor): Use finalize_ctors if needed.
16730 2013-02-04 Jakub Jelinek <jakub@redhat.com>
16733 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
16734 about visibility on artificial decls.
16735 * config/sol2.c (solaris_assemble_visibility): Likewise.
16737 2013-02-04 Kai Tietz <ktietz@redhat.com>
16740 * config/i386/i386.c (function_value_ms_64): Add additional valtype
16741 argument and improve checking of return-argument types for 16-byte
16743 (ix86_function_value_1): Add additional valtype argument on call
16744 of function_value_64.
16745 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
16746 handling infunction_value_64 function.
16748 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
16750 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
16752 2013-02-04 Richard Biener <rguenther@suse.de>
16754 PR tree-optimization/56188
16755 * tree-ssa-structalias.c (label_visit): Consider case with
16756 initially non-empty points-to set.
16757 (perform_var_substitution): Dump node mapping and clean up.
16759 2013-02-04 Richard Guenther <rguenther@suse.de>
16762 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
16763 node prevail as last resort.
16764 (lto_symtab_merge_decls): Remove guard on LTRANS here.
16765 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
16767 2013-02-04 Richard Biener <rguenther@suse.de>
16769 PR tree-optimization/56113
16770 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
16772 (equiv_class_lookup_or_add): ... this.
16773 (label_visit): Adjust and fix error in previous patch.
16774 (perform_var_substitution): Adjust.
16776 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
16778 * config/sh/divtab.c: Fix formatting and comments throughout the file.
16779 * config/sh/sh4-300.md: Likewise.
16780 * config/sh/sh4a.md: Likewise.
16781 * config/sh/constraints.md: Likewise.
16782 * config/sh/sh.md: Likewise.
16783 * config/sh/netbsd-elf.h: Likewise.
16784 * config/sh/predicates.md: Likewise.
16785 * config/sh/sh-protos.h: Likewise.
16786 * config/sh/ushmedia.h: Likewise.
16787 * config/sh/linux.h: Likewise.
16788 * config/sh/sh.c: Likewise.
16789 * config/sh/superh.h: Likewise.
16790 * config/sh/elf.h: Likewise.
16791 * config/sh/sh4.md: Likewise.
16792 * config/sh/sh.h: Likewise.
16794 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
16796 * config/pa/constraints.md: Adjust unused letters. Change "T"
16797 constraint to match_test floating_point_store_memory_operand().
16798 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
16799 (base14_operand): New.
16800 (floating_point_store_memory_operand): New.
16801 (integer_store_memory_operand): Revise to use base14_operand and
16802 reg_plus_base_memory_operand.
16803 (move_dest_operand): Allow symbolic_memory_operands.
16804 (symbolic_memory_operand): Check for LO_SOM.
16805 (symbolic_operand): Change default case to break.
16806 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
16807 CONST_DOUBLE values to be reloaded by putting them into memory when
16808 the destination is a floating point register.
16809 (movdf): Remove code to handle CONST_DOUBLE.
16811 (reload_indf_r1): New.
16812 (reload_insf_r1): New.
16813 Consistently use "Q" and "T" constraints with integer and floating
16814 point move instructions, respectively.
16815 (movdi): Remove FAIL.
16816 Change predicate for source operand unamed DImode move from
16817 general_operand to move_src_operand.
16818 (umulsidi3): Change predicate for destination operand to
16820 Likewise for similar unamed patterns.
16821 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
16822 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
16823 (hppa_legitimize_address): Simplify mask calculation.
16824 (pa_emit_move_sequence): Revised handling of secondary reloads from
16825 REG+D addresses for floating point loads and stores. Directly handle
16826 loading CONST0_RTX (mode) to a floating point register.
16827 (pa_secondary_reload): Handle reloading DF and SFmode constant values
16828 to floating point registers. Don't restrict secondary reloads to
16829 floating point registers to integer modes. Revise some comments and
16831 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16832 (pa_legitimate_address_p): New.
16833 (pa_legitimize_reload_address): New.
16834 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
16835 (STRICT_REG_OK_FOR_BASE_P): New.
16836 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
16837 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
16839 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
16840 Andrew Dixie <andrewd@gentrack.com>
16842 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
16845 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
16847 * expmed.c (extract_bit_field_1): Pass the full width of the
16848 structure to get_best_reg_extraction_insn.
16850 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
16853 * configure.ac (use_cxa_atexit): Add AIX.
16854 * configure: Regenerate.
16856 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
16858 2013-02-01 Jakub Jelinek <jakub@redhat.com>
16861 * final.c (need_profile_function): New variable.
16862 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
16863 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
16864 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
16865 notes, targetm.asm_out.function_prologue doesn't emit anything,
16866 HAVE_prologue and profiler should be emitted before prologue,
16867 set need_profile_function instead of emitting it.
16868 (final_scan_insn): If need_profile_function, emit
16869 profile_function on the first NOTE_INSN_BASIC_BLOCK or
16870 NOTE_INSN_FUNCTION_BEG note.
16872 2013-02-01 Richard Henderson <rth@redhat.com>
16874 * config/rs6000/rs6000.md (smulditi3): New.
16877 * config/alpha/alpha.md (umulditi3): New.
16879 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
16881 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
16882 (ASM_OUTPUT_ALIGNED_LOCAL): New.
16884 2013-02-01 Richard Biener <rguenther@suse.de>
16886 PR tree-optimization/56113
16887 * tree-ssa-structalias.c (label_visit): Reduce work for
16888 single-predecessor nodes.
16890 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
16892 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
16893 range isn't testing for zero.
16895 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
16897 PR middle-end/56113
16898 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
16900 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
16901 Nick Clifton <nickc@redhat.com>
16903 * config/v850/constraints.md (Q): Define as a memory constraint.
16904 * config/v850/predicates.md (label_ref_operand): New predicate.
16905 (e3v5_shift_operand): New predicate.
16906 (ior_operator): New predicate.
16907 * config/v850/t-v850: Add e3v5 multilib.
16908 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
16909 (v850_gen_movdi): Prototype.
16910 * config/v850/v850.c: Add support for e3v5 architecture.
16911 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
16913 (construct_save_jarl): Add e3v5 long JARL support.
16914 (v850_adjust_insn_length): New function. Adjust length of call
16915 insns when using e3v5 instructions.
16916 (v850_gen_movdi): New function: Generate instructions to move a
16918 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
16919 (CPP_SPEC): Define __v850e3v5__ as appropriate.
16920 (TARGET_USE_FPU): Enable for e3v5.
16921 (CONST_OK_FOR_W): New macro.
16922 (ADJUST_INSN_LENGTH): Define.
16923 * config/v850/v850.md (UNSPEC_LOOP): Define.
16924 (attr cpu): Add v850e3v5.
16925 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
16926 (movdi): New pattern.
16927 (movdi_internal): New pattern.
16928 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
16929 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
16930 (cstoresf4): Likewise.
16931 (cstoredf4): Likewise.
16932 (insv): New pattern.
16933 (rotlso3_a): New pattern.
16934 (rotlsi3_b): New pattern
16935 (rotlsi3_v850e3v5): New pattern.
16936 (doloop_begin): New pattern.
16937 (fix_loop_counter): New pattern.
16938 (doloop_end): New pattern.
16939 (branch_normal): Add e3v5 long branch support.
16940 (branch_invert): Likewise.
16941 (branch_z_normal): Likewise.
16942 (branch_z_invert): Likewise.
16943 (branch_nz_normal): Likewise.
16944 (branch_nz_invert): Likewise.
16945 (call_internal_short): Add e3v5 register-indirect JARL support.
16946 (call_internal_long): Likewise.
16947 (call_value_internal_short): Likewise.
16948 (call_value_internal_long): Likewise.
16949 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
16950 (mloop): New option.
16951 * config.gcc: Add support for configuring v840e3v5 target.
16952 * doc/invoke.texi: Document new v850 specific command line options.
16954 2013-01-31 Paul Koning <ni1d@arrl.net>
16958 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
16959 children if parent is a class.
16960 (prune_unused_types_prune): Don't add DW_AT_declaration.
16962 2013-01-31 Richard Biener <rguenther@suse.de>
16964 PR tree-optimization/56157
16965 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
16966 match up operand with SLP child.
16968 2013-01-31 Jason Merrill <jason@redhat.com>
16971 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
16972 parameters the first time.
16973 (gen_scheduled_generic_parms_dies): Check completeness here.
16975 2013-01-31 Richard Biener <rguenther@suse.de>
16977 PR middle-end/53073
16978 * common.opt (faggressive-loop-optimizations): New flag,
16979 enabled by default.
16980 * doc/invoke.texi (faggressive-loop-optimizations): Document.
16981 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
16982 infer_loop_bounds_from_undefined by it.
16984 2013-01-31 Richard Biener <rguenther@suse.de>
16986 PR tree-optimization/56150
16987 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
16988 visit virtual operands.
16989 (find_uses_to_rename_bb): Likewise.
16991 2013-01-31 Richard Biener <rguenther@suse.de>
16993 PR tree-optimization/56150
16994 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
16995 mixed store non-store stmts.
16997 2013-01-30 Jakub Jelinek <jakub@redhat.com>
17000 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17001 LIBASAN_EARLY_SPEC is defined.
17002 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17003 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17005 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17008 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17009 invalid args instead of ICEing on it.
17010 (ix86_valid_target_attribute_tree): Return error_mark_node if
17011 ix86_valid_target_attribute_inner_p failed.
17012 (ix86_valid_target_attribute_p): Return false only if
17013 ix86_valid_target_attribute_tree returned error_mark_node. Allow
17014 target("default") attribute.
17015 (sorted_attr_string): Change argument from const char * to tree,
17016 merge in all target attribute arguments rather than just one.
17017 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
17018 instead of free. Avoid using strcat.
17019 (ix86_mangle_function_version_assembler_name): Mangle
17020 target("default") as if no target attribute is present. Adjust
17021 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
17022 instead of xmalloc and XDELETEVEC instead of free.
17023 (ix86_function_versions): Don't return true if one of the decls
17024 doesn't have target attribute. If they don't and one of the decls
17025 is DECL_FUNCTION_VERSIONED, report an error. Adjust
17026 sorted_attr_string caller. Use XDELETEVEC instead of free.
17027 (ix86_supports_function_versions): Remove.
17028 (make_name): Fix up formatting.
17029 (make_dispatcher_decl): Remove resolver_name and its initialization.
17030 Avoid leaking memory.
17031 (is_function_default_version): Return true if there is
17032 target("default") attribute rather than no target attribute at all.
17033 (make_resolver_func): Avoid leaking memory.
17034 (ix86_generate_version_dispatcher_body): Likewise.
17035 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
17036 * target.def (supports_function_versions): Remove.
17037 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
17038 * doc/tm.texi: Regenerated.
17040 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
17042 PR rtl-optimization/56144
17043 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
17044 for values with side effects.
17046 2013-01-30 Richard Biener <rguenther@suse.de>
17048 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
17049 (sparseset_pop): Likewise.
17050 * cfganal.c (compute_idf): Likewise. Increase work-stack size
17051 to be able to use quick_push in the worker loop.
17053 2013-01-30 Marek Polacek <polacek@redhat.com>
17055 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
17057 2013-01-30 Richard Biener <rguenther@suse.de>
17060 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
17062 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
17064 PR tree-optimization/56064
17065 * fixed-value.c (fixed_from_double_int): New function.
17066 * fixed-value.h (fixed_from_double_int): New prototype.
17067 (const_fixed_from_double_int): New static inline function.
17068 * fold-const.c (native_interpret_fixed): New static function.
17069 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
17070 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
17071 (native_encode_fixed): New static function.
17072 (native_encode_expr) <FIXED_CST>: Use it.
17073 (native_interpret_int): Move double_int worker code to...
17074 * double-int.c (double_int::from_buffer): ...this new static method.
17075 * double-int.h (double_int::from_buffer): Prototype it.
17077 2013-01-30 Richard Biener <rguenther@suse.de>
17079 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
17080 New pointer-map and obstack.
17081 (init_alias_vars): Allocate pointer-map and obstack.
17082 (delete_points_to_sets): Free them.
17083 (find_what_var_points_to): Cache result.
17084 (find_what_p_points_to): Adjust for changed interface of
17085 find_what_var_points_to.
17086 (compute_points_to_sets): Likewise.
17087 (ipa_pta_execute): Likewise.
17089 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17091 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
17092 * configure: Regenerate.
17093 * config.in: Regenerate.
17094 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
17095 #nobits/#progbits if supported.
17097 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
17100 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
17101 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
17102 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
17104 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17106 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
17107 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
17109 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17111 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
17112 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
17114 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17116 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
17118 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
17119 * config/arm/cortex-a7.md: New bypasses using
17120 arm_mac_accumulator_is_result.
17122 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17124 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
17125 (cortex_a7_neon_mla): Likewise.
17126 (cortex_a7_fpfmad): New reservation.
17127 (cortex_a7_fpmacs): Use ffmas and update required units.
17128 (cortex_a7_fpmuld): Update required units and latency.
17129 (cortex_a7_fpmacd): Likewise.
17130 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
17131 (cortex_a7_neon). Likewise.
17132 (bypass) Update participating units.
17134 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17136 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
17137 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
17139 * config/arm/vfp11.md (vfp_farith): Use ffmas.
17140 (vfp_fmul): Use ffmad.
17141 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
17142 (cortex_r4_fmacd): Use ffmad.
17143 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
17144 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
17145 (cortex_a9_fmacd): Use ffmad.
17146 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
17147 (cortex_a8_vfp_macd): Use ffmad.
17148 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
17149 (cortex_a5_fpmacd): Use ffmad.
17150 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
17151 (cortex_a15_vfp_macd): Use ffmad.
17152 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
17154 2013-01-29 Jason Merrill <jason@redhat.com>
17157 * varasm.c (default_assemble_visibility): Don't warn about
17158 visibility on artificial decls.
17160 2013-01-29 Richard Biener <rguenther@suse.de>
17162 PR tree-optimization/56113
17163 * tree-ssa-structalias.c (equiv_class_lookup): Also return
17165 (label_visit): Free duplicate bitmaps and record the leader instead.
17166 (perform_var_substitution): Adjust.
17168 2013-01-29 Richard Biener <rguenther@suse.de>
17170 PR tree-optimization/55270
17171 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
17172 the CFG, schedule loops for fixup.
17174 2013-01-29 Nick Clifton <nickc@redhat.com>
17176 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
17179 2013-01-28 Leif Ekblad <leif@rdos.net>
17181 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
17182 * config/i386/i386.h (TARGET_RDOS): New macro.
17183 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
17184 * config/i386/i386.c (ix86_option_override_internal): For 64bit
17185 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
17186 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
17187 DEFAULT_LARGE_SECTION_THRESHOLD.
17188 * config/i386/i386.md (R14_REG, R15_REG): New constants.
17189 * config/i386/rdos.h: New file.
17190 * config/i386/rdos64.h: New file.
17192 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
17195 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
17198 2013-01-28 Jakub Jelinek <jakub@redhat.com>
17200 PR rtl-optimization/56117
17201 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
17202 call cselib_lookup_from_insn on the MEM before calling
17203 add_insn_mem_dependence.
17205 2013-01-28 Richard Biener <rguenther@suse.de>
17207 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
17208 to a stmt that didn't have one.
17209 (copy_phis_for_bb): Likewise for PHI arguments.
17210 (copy_debug_stmt): Likewise for debug stmts.
17212 2013-01-28 Richard Biener <rguenther@suse.de>
17214 PR tree-optimization/56034
17215 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
17216 (partition_builtin_p): Adjust.
17217 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
17218 it is the last partition.
17219 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
17220 up the vertex for the definition.
17221 (classify_partition): Classify whether a partition is a
17222 PKIND_REDUCTION, thus has uses outside of the loop.
17223 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
17224 Merge all PKIND_REDUCTION partitions into the last partition.
17225 (tree_loop_distribution): Seed partitions from reductions as well.
17227 2013-01-28 Jakub Jelinek <jakub@redhat.com>
17229 PR tree-optimization/56125
17230 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
17231 pow(x,c) into sqrt(x) * powi(x, n/2) or
17232 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
17233 optimizing for size.
17234 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
17235 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
17238 PR tree-optimization/56094
17239 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
17240 to UNKNOWN_LOCATION while gimplifying expr.
17242 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
17245 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
17246 operand 0 in movabs insn template for -masm=intel asm alternative.
17247 (*movabs<mode>_2): Ditto for operand 1.
17249 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
17252 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
17255 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17257 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
17260 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
17262 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
17263 Undef to avoid warning.
17265 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
17267 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
17268 * configure: Regenerate.
17270 2013-01-25 Jakub Jelinek <jakub@redhat.com>
17272 PR tree-optimization/56098
17273 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
17274 for stmts with volatile ops.
17275 (cond_store_replacement): Don't optimize if assign has volatile ops.
17276 (cond_if_else_store_replacement_1): Don't optimize if either
17277 then_assign or else_assign have volatile ops.
17278 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
17281 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
17283 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
17285 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
17287 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
17288 missing ':' in asm example.
17290 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
17292 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
17293 entries into lane and laneq entries.
17294 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
17295 Remove AdvSIMD scalar modes.
17296 (aarch64_sq<r>dmulh_laneq<mode>): New.
17297 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
17299 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
17300 builtin implementations to relfect changes in RTL in aarch64-simd.md.
17301 * config/aarch64/iterators.md (VCOND): New.
17304 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
17307 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
17308 Add NULL LIBNAME argument to existing definitions.
17309 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
17310 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
17311 * config/avr/avr.c (DEF_BUILTIN): Same.
17312 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
17313 (avr_expand_builtin): Expand to a vanilla call if a libgcc
17314 implementation is available (DECL_ASSEMBLER_NAME is set).
17315 (avr_fold_absfx): New static function.
17316 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
17317 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
17318 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
17319 AVR_BUILTIN_ABSLLK.
17320 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
17321 (abshk, absk, abslk, absllk): Provide as static inline functions.
17323 2013-01-25 Marek Polacek <polacek@redhat.com>
17325 PR tree-optimization/56035
17326 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
17328 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
17330 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
17331 (*movtf_internal_rex64): Add (!o,C) alternative
17332 (*movxf_internal_rex64): Ditto.
17333 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
17335 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
17337 * doc/invoke.texi: fix typo.
17338 * doc/objc.texi: fix typo.
17340 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
17342 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
17343 for the first two alternatives.
17345 2013-01-24 Diego Novillo <dnovillo@google.com>
17347 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
17348 (ggc-zone.o): Remove.
17349 * configure.ac: Remove option --with-gc.
17350 * configure: Re-generate.
17351 * doc/install.texi: Remove documentation for --with-gc.
17352 * gengtype.c (write_enum_defn): Remove. Update all users.
17353 (write_Types_process_field): Remove generation of gt_e_* argument.
17354 (output_type_enum): Remove. Update all users.
17355 (write_enum_defn): Remove. Update all users.
17356 (enum alloc_zone): Remove. Update all users.
17357 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
17358 * ggc-common.c (ggc_splay_alloc): Remove first argument.
17359 Update all callers.
17360 (struct ptr_data): Remove field TYPE. Update all users.
17361 (gt_pch_note_object): Remove argument TYPE. Update all users.
17362 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
17364 * ggc-none.c (ggc_alloc_typed_stat): Remove.
17365 (struct alloc_zone): Remove.
17366 (ggc_internal_alloc_zone_stat): Remove.
17367 (ggc_internal_cleared_alloc_zone_stat): Remove.
17368 * ggc-page.c (ggc_alloc_typed_stat): Remove.
17369 (ggc_pch_count_object): Remove last argument. Update all users.
17370 (ggc_pch_alloc_object): Remove last argument. Update all users.
17371 (struct alloc_zone): Remove.
17372 * ggc-zone.c: Remove.
17373 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
17374 (struct alloc_zone): Remove.
17375 (ggc_alloc_typed_stat): Remove.
17376 (ggc_alloc_typed): Remove.
17377 (ggc_splay_alloc): Remove first argument.
17378 (rtl_zone): Remove. Update all users.
17379 (tree_zone): Remove. Update all users.
17380 (tree_id_zone): Remove. Update all users.
17381 (ggc_internal_zone_alloc_stat): Remove. Update all users.
17382 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
17383 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
17384 * tree-ssanames.c: Remove references to zone allocator in comments.
17386 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
17388 * config/avr/avr.c (avr_out_fract): Make register numbers that
17389 might be outside of source operand signed.
17391 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
17393 * config/i386/constraints.md (Yf): New constraint.
17394 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
17395 of f constraint to conditionaly disable x87 register preferences.
17396 (*movdf_internal): Ditto.
17397 (*movsf_internal): Ditto.
17399 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
17401 PR inline-asm/55934
17402 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
17403 that have operands with impossible constraints.
17404 Add a FIXME for a speed-up opportunity.
17405 * lra-constraints.c (process_alt_operands): Verify that a class
17406 selected from constraints on asms is valid for the operand mode.
17407 (curr_insn_transform): Remove incorrect comment.
17409 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
17411 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
17412 TOC operand is a valid symbol ref in the constant pool.
17414 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17416 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
17418 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
17421 * config/avr/stdfix.h: New file.
17422 * t-avr (stdfix-gcc.h): New rule to build it.
17423 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
17425 2013-01-23 Kostya Serebryany <kcc@google.com>
17427 * config/darwin.h: remove dependency on
17428 CoreFoundation (asan on Mac OS).
17430 2013-01-23 Jakub Jelinek <jakub@redhat.com>
17433 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
17434 instead of cmpdi_operand for first comparison operand.
17435 Don't assert that comparison operands aren't both constants.
17437 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
17439 * doc/install.texi (Downloading the Source): Update references to
17440 downloading separate components.
17442 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
17444 * doc/extend.texi (__int128): Improve grammar.
17446 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
17449 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
17450 alternative to (o,r).
17451 (*movdi_internal_rex64): Remove (!o,n) alternative.
17452 (DImode immediate->memory splitter): Remove.
17453 (DImode immediate->memory peephole2): Remove.
17454 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
17455 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
17456 alternative to (!o,*r).
17457 (*movtf_internal_sse): New pattern.
17458 (*movxf_internal_rex64): New pattern.
17459 (*movxf_internal): Disable for TARGET_64BIT.
17460 (*movdf_internal_rex64): Remove (!o,F) alternative.
17462 2013-01-22 Jakub Jelinek <jakub@redhat.com>
17464 PR middle-end/56074
17465 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
17466 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
17467 * tree-vect-loop-manip.c (find_loop_location): Also ignore
17468 stmt locations where LOCATION_LOCUS of the stmt location is
17469 UNKNOWN_LOCATION or BUILTINS_LOCATION.
17472 * config/i386/i386.md (UNSPEC_STOS): New.
17473 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
17474 *strsetqi_1): Add UNSPEC_STOS.
17476 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
17479 * doc/invoke.texi: Remove left over -Wsynth example.
17481 2013-01-21 Jakub Jelinek <jakub@redhat.com>
17483 PR tree-optimization/56051
17484 * fold-const.c (fold_binary_loc): Don't fold
17485 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
17486 a narrowing conversion, or widening conversion from signed
17489 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
17491 PR rtl-optimization/56023
17492 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
17493 dependent on debug instruction.
17495 2013-01-21 Martin Jambor <mjambor@suse.cz>
17497 PR middle-end/56022
17498 * function.c (allocate_struct_function): Call
17499 invoke_set_current_function_hook earlier.
17501 2013-01-21 Jakub Jelinek <jakub@redhat.com>
17503 * reload1.c (init_reload): Only initialize reload_obstack
17504 during the first call.
17506 2013-01-21 Marek Polacek <polacek@redhat.com>
17508 * cfgloop.c (verify_loop_structure): Fix up grammar.
17510 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
17512 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
17513 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
17515 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17518 * config/arm/marvell-pj4.md: Update copyright year.
17519 Fix up use of alu to alu_reg and simple_alu_imm.
17521 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
17523 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
17525 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
17528 * lra-constraints.c (curr_insn_transform): Don't reuse original
17529 insn for secondary memory move when memory mode should be different.
17531 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17533 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
17534 atomic_storedi_1): New patterns.
17536 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
17538 btver2 pipeline descriptions.
17539 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
17541 * config/i386/i386.md (btver2_decode): New type attributes.
17542 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
17544 * config/i386/btver2.md: New file describing btver2 pipelines.
17546 2013-01-19 Andrew Pinski <apinski@cavium.com>
17548 PR tree-optimization/52631
17549 * tree-ssa-sccvn (visit_use): Before looking up the original
17550 statement, try looking up the simplified expression.
17552 2013-01-19 Anthony Green <green@moxielogic.com>
17554 * config/moxie/moxie.c (moxie_expand_prologue): Set
17555 current_function_static_stack_size.
17557 2013-01-18 Jakub Jelinek <jakub@redhat.com>
17559 PR tree-optimization/56029
17560 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
17561 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
17563 2013-01-18 Sharad Singhai <singhai@google.com>
17565 PR tree-optimization/55995
17566 * dumpfile.c (dump_loc): Print location only if available.
17567 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
17569 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
17572 * lra-constraints.c (curr_insn_transform): Reuse original insn for
17573 secondary memory move.
17574 (inherit_reload_reg): Use rclass instead of cl for
17575 check_secondary_memory_needed_p.
17577 2013-01-18 Jakub Jelinek <jakub@redhat.com>
17579 PR middle-end/56015
17580 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
17581 the case where writing real complex part of target modifies op1.
17583 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
17585 * config/aarch64/aarch64-simd.md
17586 (aarch64_vcond_internal<mode>): Handle unordered cases.
17587 * config/aarch64/iterators.md (v_cmp_result): New.
17589 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
17590 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17592 * config/arm/marvell-pj4.md: New file.
17593 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
17594 * config/arm/arm.md (generic_sched): Add marvell_pj4.
17595 (generic_vfp): Likewise.
17596 * config/arm/arm-cores.def: Add marvell-pj4.
17597 * config/arm/arm-tune.md: Regenerate.
17598 * config/arm/arm-tables.opt: Regenerate.
17599 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
17600 * doc/invoke.texi: Document marvell-pj4.
17602 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
17604 * config/aarch64/arm_neon.h: Map scalar types to standard types.
17606 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
17611 * var-tracking.c (negative_power_of_two_p): New.
17612 (global_get_addr_cache, local_get_addr_cache): New.
17613 (get_addr_from_global_cache, get_addr_from_local_cache): New.
17614 (vt_canonicalize_addr): Rewrite using the above. Adjust the
17616 (vt_stack_offset_p): Remove.
17617 (vt_canon_true_dep): Always canonicalize loc's address.
17618 (clobber_overlapping_mems): Make sure we have a MEM.
17619 (local_get_addr_clear_given_value): New.
17620 (val_reset): Clear local cached entries.
17621 (compute_bb_dataflow): Create and release the local cache.
17622 Disable duplicate MEMs clobbering.
17623 (emit_notes_in_bb): Clobber MEMs likewise.
17624 (vt_emit_notes): Create and release the local cache.
17625 (vt_initialize, vt_finalize): Create and release the global
17626 cache, respectively.
17627 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
17629 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
17631 PR libmudflap/53359
17632 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
17633 not found in the symtab.
17635 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
17638 PR rtl-optimization/55547
17639 PR rtl-optimization/53827
17642 * alias.c (offset_overlap_p): New, factored out of...
17643 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
17644 the conservative special case for symbolic constants. Don't
17645 adjust zero sizes on alignment.
17647 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
17649 PR rtl-optimization/52573
17650 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
17651 REG_UNUSED for the same register.
17653 2013-01-17 Richard Biener <rguenther@suse.de>
17654 Marek Polacek <polacek@redhat.com>
17656 PR rtl-optimization/55833
17657 * loop-unswitch.c (unswitch_loops): Move loop verification...
17658 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
17659 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
17660 Set it to true when we're removing a loop from hierarchy tree in
17661 an irreducible region.
17662 (fix_bb_placements): Adjust caller.
17663 (fix_loop_placements): Likewise.
17665 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
17667 * config/avr/builtins.def (DEF_BUILTIN): Factor out
17668 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
17669 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
17670 Remove ID. Adjust comments.
17671 * config/avr/avr-c.c (avr_builtin_name): Remove.
17672 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
17673 * config/avr/avr.c (avr_tolower): New static function.
17674 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
17675 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
17676 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
17679 2013-01-17 Jan Hubicka <jh@suse.cz>
17681 PR tree-optimization/55273
17682 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
17684 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
17687 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
17688 store through atomic_store<mode>_1.
17689 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
17691 2013-01-17 Martin Jambor <mjambor@suse.cz>
17693 PR tree-optimizations/55264
17694 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
17695 for virtual methods.
17696 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
17697 virtual methods before inlining is over.
17698 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
17700 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
17703 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
17705 PR rtl-optimization/56005
17706 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
17707 pending reads for prefetch.
17709 2013-01-16 Ian Bolton <ian.bolton@arm.com>
17711 * config/aarch64/aarch64.md
17712 (*cstoresi_neg_uxtw): New pattern.
17713 (*cmovsi_insn_uxtw): New pattern.
17714 (*<optab>si3_uxtw): New pattern.
17715 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
17716 (*<optab>si3_insn_uxtw): New pattern.
17717 (*bswapsi2_uxtw): New pattern.
17719 2013-01-16 Richard Biener <rguenther@suse.de>
17721 * tree-inline.c (tree_function_versioning): Remove set but
17722 never used variable.
17724 2013-01-16 Richard Biener <rguenther@suse.de>
17726 PR tree-optimization/55964
17727 * tree-flow.h (rename_variables_in_loop): Remove.
17728 (rename_variables_in_bb): Likewise.
17729 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
17730 (copy_loop_before): Adjust and delete update-ssa status.
17731 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
17732 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
17733 (rename_variables_in_loop): Remove.
17734 (slpeel_update_phis_for_duplicate_loop): Likewise.
17735 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
17736 use available cfg machinery instead of duplicating it.
17737 Update PHI nodes and perform poor-mans SSA update here.
17738 (slpeel_tree_peel_loop_to_edge): Adjust.
17740 2013-01-16 Richard Biener <rguenther@suse.de>
17742 PR tree-optimization/54767
17743 PR tree-optimization/53465
17744 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
17745 (vrp_visit_phi_node): For PHI arguments coming via backedges
17746 drop all symbolical range information.
17747 (execute_vrp): Compute backedges.
17749 2013-01-16 Richard Biener <rguenther@suse.de>
17751 * doc/install.texi: Update CLooG and ISL requirements to
17754 2013-01-16 Christian Bruel <christian.bruel@st.com>
17757 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
17758 (broken_move): Handle UNSPECV_SP_SWITCH_B.
17759 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
17761 2013-01-16 DJ Delorie <dj@redhat.com>
17763 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
17764 (UNSPECV_SP_SWITCH_E): New.
17765 (sp_switch_1): Change to an unspec.
17766 (sp_switch_2): Change to an unspec. Don't use post-inc when we
17769 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
17771 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
17772 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
17773 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
17774 (expand_mem_thread_fence): Ditto.
17775 (expand_mem_signal_fence): Ditto.
17776 (expand_atomic_load): Ditto.
17777 (expand_atomic_store): Ditto.
17779 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
17781 PR rtl-optimization/55547
17782 PR rtl-optimization/53827
17785 * alias.c (memrefs_conflict_p): Set sizes to negative after
17788 2013-01-15 Jakub Jelinek <jakub@redhat.com>
17791 * function.c (thread_prologue_and_epilogue_insns): Always
17792 add crtl->drap_reg to set_up_by_prologue.set, even if
17793 stack_realign_drap is false.
17795 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
17797 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
17798 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
17799 *call): Fix indention.
17801 2013-01-15 Tom de Vries <tom@codesourcery.com>
17804 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
17807 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
17809 PR rtl-optimization/55153
17810 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
17812 2013-01-15 Martin Jambor <mjambor@suse.cz>
17814 PR tree-optimization/55920
17815 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
17816 accesses as grp_to_be_debug_replaced.
17818 2013-01-15 Jakub Jelinek <jakub@redhat.com>
17820 PR tree-optimization/55920
17821 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
17822 there is non-useless type conversion needed from debug rhs to lhs,
17823 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
17825 2013-01-15 Joseph Myers <joseph@codesourcery.com>
17826 Mikael Pettersson <mikpe@it.uu.se>
17829 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
17831 (ASM_OUTPUT_CASE_LABEL): Remove.
17832 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
17833 * final.c (shorten_branches): Update alignment of labels before
17834 jump tables if CASE_VECTOR_SHORTEN_MODE.
17836 2013-01-15 Richard Biener <rguenther@suse.de>
17839 * system.h: Do not include gmp.h for building host tools.
17841 2013-01-15 Richard Biener <rguenther@suse.de>
17843 PR middle-end/55882
17844 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
17845 account for bitpos when computing alignment.
17847 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
17849 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
17850 (ix86_target_macros_internal): Likewise.
17852 * config/i386/i386.c (m_CORE2I7): Removed.
17853 (m_CORE_HASWELL): New macro.
17854 (m_CORE_ALL): Likewise.
17855 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
17856 (initial_ix86_arch_features): Likewise.
17857 (processor_target_table): Initializations for Core avx2.
17858 (cpu_names): New names "core-avx2".
17859 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
17860 PROCESSOR_CORE_HASWELL.
17861 (ix86_issue_rate): New case.
17862 (ia32_multipass_dfa_lookahead): Likewise.
17863 (ix86_sched_init_global): Likewise.
17865 * config/i386/i386.h (TARGET_HASWELL): New macro.
17866 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
17867 (processor_type): New PROCESSOR_HASWELL.
17869 2013-01-15 Jakub Jelinek <jakub@redhat.com>
17871 PR tree-optimization/55955
17872 * tree-vect-loop.c (vectorizable_reduction): Give up early on
17873 *SHIFT_EXPR and *ROTATE_EXPR codes.
17875 PR tree-optimization/48766
17876 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
17877 -ftrapv disable -fwrapv.
17879 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17882 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
17883 etc. to 1 and not to __flash.
17884 Use LL suffix for __INT24_MAX__ with -mint8.
17885 Use ULL suffix for __UINT24_MAX__ with -mint8.
17887 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17889 * config/avr/avr-arch.h
17890 (struct base_arch_s): Use typedef avr_arch_t instead.
17891 (struct arch_info_s): Use typedef avr_arch_info_t instead.
17892 (struct mcu_type_s): Use typedef avr_mcu_t instead.
17893 * config/avr/avr.c: Same.
17894 * config/avr/avr-devices.c: Same.
17895 * config/avr/driver-avr.c: Same.
17896 * config/avr/gen-avr-mmcu-texi.c: Same.
17897 * config/avr/avr-mcus.def: Adjust comment.
17899 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
17901 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
17902 * config/aarch64/iterators.md (VALLDI): New.
17904 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
17905 Andi Kleen <ak@linux.intel.com>
17908 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
17909 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
17912 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17914 * config/avr/avr-stdint.h: Remove trailing blanks.
17915 * config/avr/avr-log.h: Same.
17916 * config/avr/avr-arch.h: Same.
17917 * config/avr/avr-devices.c: Same.
17918 * config/avr/avr-dimode.md: Same.
17919 * config/avr/predicates.md: Same.
17920 * config/avr/avr-c.c: Same. And fix typo.
17922 * config/avr/avr-protos.h: Same. And:
17923 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
17924 (init_cumulative_args): Rename to avr_init_cumulative_args.
17925 (expand_prologue): Rename to avr_expand_prologue.
17926 (expand_epilogue): Rename to avr_expand_epilogue.
17927 (adjust_insn_length): Rename to avr_adjust_insn_length.
17928 (notice_update_cc): Rename to avr_notice_update_cc.
17929 (final_prescan_insn): Rename to avr_final_prescan_insn.
17930 * config/avr/avr.c: Same.
17931 * config/avr/avr.h: Same.
17932 * config/avr/avr.md: Remove trailing blanks.
17933 (prologue): Use avr_expand_prologue.
17934 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
17936 2013-01-14 Richard Biener <rguenther@suse.de>
17938 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
17939 verify_location, collect_subblocks): New functions.
17940 (verify_gimple_in_cfg): Verify that locations only reference
17941 BLOCKs in the functions BLOCK tree.
17943 2013-01-14 Richard Biener <rguenther@suse.de>
17945 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
17947 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
17949 (insert_out_of_ssa_copy_on_edge): Likewise.
17950 (rewrite_close_phi_out_of_ssa): Likewise.
17951 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
17953 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
17954 propagated constants.
17955 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
17958 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
17960 * config/avr/avr-modes.def: Add GPL copyright notice.
17962 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
17964 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
17965 MEMMODEL_MASK to determine memory model.
17966 (atomic_store<mode>): Ditto from operands[2].
17967 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
17969 2013-01-13 Jakub Jelinek <jakub@redhat.com>
17972 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
17973 (fold_gimple_assign): Don't call unshare_expr here.
17974 (fold_ctor_reference): Call unshare_expr.
17976 2013-01-13 Terry Guo <terry.guo@arm.com>
17978 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
17979 * doc/fragments.texi: Document MULTILIB_REUSE.
17980 * gcc.c (multilib_reuse): New internal spec.
17981 (set_multilib_dir): Also search multilib from multilib_reuse.
17982 * genmultilib (tmpmultilib3): Refactor code.
17983 (tmpmultilib4): Ditto.
17984 (multilib_reuse): New multilib argument.
17986 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
17988 * Makefile.in: Update copyright.
17990 2013-01-12 Tom de Vries <tom@codesourcery.com>
17992 PR middle-end/55890
17993 * calls.c (expand_call): Check if arg_nr is valid.
17995 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17997 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
17998 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
17999 documentation. Add missing '__' in front of
18000 __builtin_ia32_packssdw256.
18002 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
18005 * config/s390/s390.c (s390_preferred_reload_class): Do not return
18006 NO_REGS for larl operands.
18007 (s390_reload_larl_operand): Use s390_load_address instead of
18010 2013-01-11 Richard Biener <rguenther@suse.de>
18012 * tree-cfg.c (verify_node_sharing_1): Split out from ...
18013 (verify_node_sharing): ... here.
18014 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
18016 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
18018 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
18019 Substitute TREECHECKING.
18020 * configure: Regenerate.
18021 * Makefile.in (TREECHECKING): New.
18023 2013-01-11 Richard Guenther <rguenther@suse.de>
18025 PR tree-optimization/44061
18026 * tree-vrp.c (extract_range_basic): Compute zero as
18027 value-range for __builtin_constant_p of function parameters.
18029 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
18031 Update copyright years.
18033 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
18035 PR rtl-optimization/55672
18036 * lra-eliminations.c (mark_not_eliminable): Permit addition with
18037 const to be eliminable.
18039 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
18041 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
18042 * configure: Regenerate.
18044 2013-01-10 Richard Biener <rguenther@suse.de>
18046 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
18048 2013-01-10 Richard Biener <rguenther@suse.de>
18051 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
18052 locations for virtual PHI arguments.
18053 (rewrite_update_phi_arguments): Likewise.
18055 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
18057 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
18060 2013-01-10 Jakub Jelinek <jakub@redhat.com>
18062 PR tree-optimization/55921
18063 * tree-complex.c (expand_complex_asm): New function.
18064 (expand_complex_operations_1): Call it for GIMPLE_ASM.
18066 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
18069 * config/s390/s390.c (s390_symref_operand_p)
18070 (s390_loadrelative_operand_p): Merge the two functions.
18071 (s390_check_qrst_address, print_operand_address): Add parameters
18072 to s390_loadrelative_operand_p invokation.
18073 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
18074 (s390_reload_larl_operand, s390_secondary_reload): Use
18075 s390_loadrelative_operand_p instead of s390_symref_operand_p.
18076 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
18078 2013-01-09 Mike Stump <mikestump@comcast.net>
18080 * dse.c (record_store): Remove unnecessary assert.
18082 2013-01-09 Jan Hubicka <jh@suse.cz>
18084 PR tree-optimization/55569
18085 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
18086 * cfgloop.h (scale_loop_profile): Likewise.
18088 2013-01-09 Jan Hubicka <jh@suse.cz>
18091 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
18093 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
18095 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
18097 PR middle-end/55114
18098 * expr.h (maybe_emit_group_store): Declare.
18099 * expr.c (maybe_emit_group_store): New function.
18100 * builtins.c (expand_builtin_int_roundingfn): Call it.
18101 (expand_builtin_int_roundingfn_2): Likewise.
18103 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
18105 PR rtl-optimization/55829
18106 * lra-constraints.c (match_reload): Add code for absent output.
18107 (curr_insn_transform): Add code for reloads of matched inputs
18110 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
18112 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
18113 attribute of movddup insn to DF.
18114 (*vec_interleave_lowv2df): Ditto.
18115 (vec_dupv2df): Ditto.
18117 2013-01-09 Jan Hubicka <jh@suse.cz>
18119 PR tree-optimiation/55875
18120 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
18121 EVERY_ITERATION parameter.
18122 (number_of_iterations_exit): Check if exit is executed every iteration.
18123 (idx_infer_loop_bounds): Similarly here.
18124 (n_of_executions_at_most): Simplify
18125 to only test for cases where statement is dominated by the
18126 particular bound; handle correctly the "postdominance" test.
18127 (scev_probably_wraps_p): Use max loop iterations info
18128 as a global bound first.
18130 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
18131 Nick Clifton <nickc@redhat.com>
18133 * config/v850/v850.md (cbranchsf4): New pattern.
18134 (cstoresf4): New pattern.
18135 (cbranchdf4): New pattern.
18136 (cstoredf4): New pattern.
18137 (movsicc): Disallow floating point comparisons.
18138 (cmpsf_le_insn): Fix order of operators.
18139 (cmpsf_lt_insn): Likewise.
18140 (cmpsf_eq_insn): Likewise.
18141 (cmpdf_le_insn): Likewise.
18142 (cmpdf_lt_insn): Likewise.
18143 (cmpdf_eq_insn): Likewise.
18144 (cmpsf_ge_insn): Use LE comparison.
18145 (cmpdf_ge_insn): Likewise.
18146 (cmpsf_gt_insn): Use LT comparison.
18147 (cmpdf_gt_insn): Likewise.
18148 (cmpsf_ne_insn): Delete pattern.
18149 (cmpdf_ne_insn): Delete pattern.
18150 * config/v850/v850.c (v850_gen_float_compare): Use
18151 gen_cmpdf_eq_insn for NE comparison.
18152 (v850_float_z_comparison_operator)
18153 (v850_float_nz_comparison_operator): Move from here ...
18154 * config/v850/predicates.md: ... to here. Move GT and GE
18155 comparisons into v850_float_z_comparison_operator.
18156 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
18158 (v850_float_nz_comparison_operator): Likewise.
18160 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18162 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
18163 with calls to gen_insvsi/gen_insvdi.
18165 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
18167 * config/i386/i386.c (initial_ix86_tune_features): Set up
18168 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
18170 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
18171 Jakub Jelinek <jakub@redhat.com>
18173 PR tree-optimization/48189
18174 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
18175 If nitercst is 0, don't predict the exit edge.
18177 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
18179 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
18180 in asm_fprintf with reg_names.
18181 (aarch64_print_operand_address): Likewise.
18182 (aarch64_return_addr): Likewise.
18183 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
18185 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18187 * config/pa/pa.h (VAL_U6_BITS_P): Define.
18188 (INT_U6_BITS): Likewise.
18189 * config/pa/predicates.md (uint6_operand): New predicate.
18190 (shift5_operand, shift6_operand): Likewise.
18191 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
18193 (lshrdi3): Use shift6_operand.
18194 (shrpsi4, shrpdi4): New insn patterns.
18195 (extzv): Delete expander.
18196 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
18197 predicates in unamed zero extract patterns. Tighten common constraint.
18198 (extv): Delete expander.
18199 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
18200 predicates in unamed sign extract patterns. Tighten common constraint.
18201 (insv): Delete expander.
18202 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
18203 predicates in unamed insert patterns. Tighten common constraint.
18204 Change uint32_operand predicate to uint6_operand predicate in unamed
18205 DImode pattern to insert constant values of type 1...1xxxx.
18207 2013-01-04 Jan Hubicka <jh@suse.cz>
18209 PR tree-optimization/55823
18210 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
18213 2013-01-08 Jakub Jelinek <jakub@redhat.com>
18214 Uros Bizjak <ubizjak@gmail.com>
18216 PR rtl-optimization/55845
18217 * df-problems.c (can_move_insns_across): Stop scanning at
18218 volatile_insn_p source instruction or give up if
18219 across_from .. across_to range contains any volatile_insn_p
18222 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
18224 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
18225 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
18227 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
18228 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
18230 2013-01-08 Jakub Jelinek <jakub@redhat.com>
18233 * asan.c (asan_clear_shadow): New function.
18234 (asan_emit_stack_protection): Use it.
18236 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
18238 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
18239 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
18240 with tab instead of space.
18242 2013-01-08 Nick Clifton <nickc@redhat.com>
18244 * config/rl78/rl78.c (rl78_expand_prologue): Always select
18245 register bank 0 at the start of an interrupt handler.
18246 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
18249 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
18251 * config/aarch64/aarch64-simd.md
18252 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
18253 (aarch64_simd_bsl): Likewise.
18254 (aarch64_vcond_internal<mode>): Likewise.
18255 (vcond<mode><mode>): Likewise.
18256 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
18257 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
18259 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
18261 * config/aarch64/aarch64-builtins.c
18262 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
18264 2013-01-08 Martin Jambor <mjambor@suse.cz>
18267 * tree-sra.c (analyze_access_subtree): Return true also after
18268 potentially creating a debug-only replacement.
18270 2013-01-08 Jakub Jelinek <jakub@redhat.com>
18272 PR middle-end/55890
18273 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
18275 PR tree-optimization/54120
18276 * tree-vrp.c (range_fits_type_p): Don't allow
18277 src_precision < precision from signed vr to unsigned_p
18278 if vr->min or vr->max is negative.
18279 (simplify_float_conversion_using_ranges): Test can_float_p
18280 against CODE_FOR_nothing.
18282 2013-01-08 Jakub Jelinek <jakub@redhat.com>
18283 Richard Biener <rguenther@suse.de>
18285 PR middle-end/55851
18286 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
18287 types instead of just INTEGER_TYPE types.
18289 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
18291 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
18294 2013-01-07 Steve Ellcey <sellcey@mips.com>
18297 * config/mips/mips.opt: Change mad to mmad to match documentation.
18299 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18302 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
18303 .progmemx.data now.
18305 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18308 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
18309 (avr_addrspace_t): Add .section_name field.
18310 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
18312 (avr_addrspace): Same. Initialize .section_name. Remove last
18313 NULL entry. Put __memx into .progmemx.data.
18314 (progmem_section_prefix): Remove.
18315 (avr_asm_init_sections): No need to initialize progmem_section.
18316 (avr_asm_named_section): Use avr_addrspace[].section_name to get
18317 section name prefix.
18318 (avr_asm_select_section): Ditto. And use get_unnamed_section to
18319 retrieve the progmem section.
18320 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
18321 boundary to run over avr_addrspace[].
18322 (avr_register_target_pragmas): Ditto.
18324 2013-01-06 Jakub Jelinek <jakub@redhat.com>
18326 * varasm.c (output_constant_def_contents): For asan_protect_global
18327 protected strings, adjust DECL_ALIGN if needed, before testing for
18329 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
18330 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
18332 (output_object_block): For asan protected decls, emit asan padding
18333 after their contents.
18334 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
18335 (asan_finish_file): Test it here instead.
18337 2013-01-07 Nick Clifton <nickc@redhat.com>
18338 Matthias Klose <doko@debian.org>
18339 Doug Kwan <dougkwan@google.com>
18340 H.J. Lu <hongjiu.lu@intel.com>
18343 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
18345 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
18347 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
18349 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
18351 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
18353 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18356 * doc/install.texi (Cross-Compiler-Specific Options): Document
18359 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
18361 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
18362 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
18363 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
18364 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
18365 vqmovun_high_s64): Fix source operand number and update copyright.
18367 2013-01-07 Richard Biener <rguenther@suse.de>
18369 PR middle-end/55890
18370 * gimple.h (gimple_call_builtin_p): New overload.
18371 * gimple.c (validate_call): New function.
18372 (gimple_call_builtin_p): Likewise.
18373 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18374 Use gimple_call_builtin_p.
18375 (find_func_clobbers): Likewise.
18376 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
18377 (strlen_optimize_stmt): Likewise.
18379 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
18381 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
18382 (vld1q_dup_*): Likewise.
18383 (vld1_*): Likewise.
18384 (vld1q_*): Likewise.
18385 (vld1_lane_*): Likewise.
18386 (vld1q_lane_*): Likewise.
18388 2013-01-07 Richard Biener <rguenther@suse.de>
18390 * lto-streamer.h (LTO_minor_version): Bump to 2.
18392 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
18394 * config/aarch64/aarch64-protos.h
18395 (aarch64_const_double_zero_rtx_p): Rename to...
18396 (aarch64_float_const_zero_rtx_p): ...this.
18397 (aarch64_float_const_representable_p): New.
18398 (aarch64_output_simd_mov_immediate): Likewise.
18399 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
18400 move immediate case.
18401 * config/aarch64/aarch64.c
18402 (aarch64_const_double_zero_rtx_p): Rename to...
18403 (aarch64_float_const_zero_rtx_p): ...this.
18404 (aarch64_print_operand): Allow printing of new constants.
18405 (aarch64_valid_floating_const): New.
18406 (aarch64_legitimate_constant_p): Check for valid floating-point
18408 (aarch64_simd_valid_immediate): Likewise.
18409 (aarch64_vect_float_const_representable_p): New.
18410 (aarch64_float_const_representable_p): Likewise.
18411 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
18412 (aarch64_output_simd_mov_immediate): New.
18413 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
18414 (*movdf_aarch64): Likewise.
18415 * config/aarch64/constraints.md (Ufc): New.
18416 (Y): call aarch64_float_const_zero_rtx.
18417 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
18419 2013-01-07 Richard Biener <rguenther@suse.de>
18421 PR tree-optimization/55888
18422 PR tree-optimization/55862
18423 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
18424 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
18425 not if it is contained therein.
18427 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18429 * config/avr/t-avr: Typo.
18431 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18434 * config/avr/t-avr: Don't automatically rebuild
18435 $(srcdir)/config/avr/t-multilib
18436 $(srcdir)/config/avr/avr-tables.opt
18437 $(srcdir)/doc/avr-mmcu.texi
18438 (avr-mcus): New phony target to build them on request.
18439 (s-avr-mlib, s-avr-mmcu-texi): Remove.
18440 * avr/avr-mcus.def: Adjust comments.
18442 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
18444 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
18446 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
18448 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
18450 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
18452 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
18454 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
18456 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
18457 to generate profiling.
18458 * config/rs6000/aix64.h (LIB_SPEC): Same.
18460 2013-01-04 Andrew Pinski <apinski@cavium.com>
18462 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
18464 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
18466 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
18468 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
18470 (ix86_expand_move): Ditto.
18471 (ix86_zero_extend_to_Pmode): Ditto.
18472 (ix86_expand_call): Ditto.
18473 (ix86_expand_special_args_builtin): Ditto.
18474 (ix86_expand_builtin): Ditto.
18476 2013-01-04 Richard Biener <rguenther@suse.de>
18478 PR tree-optimization/55862
18479 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
18480 translating them through PHI nodes.
18482 2013-01-04 Martin Jambor <mjambor@suse.cz>
18484 PR tree-optimization/55755
18485 * tree-sra.c (sra_modify_assign): Do not check that an access has no
18486 children when trying to avoid producing a VIEW_CONVERT_EXPR.
18488 2013-01-04 Marek Polacek <polacek@redhat.com>
18490 PR middle-end/55859
18491 * opts.c (default_options_optimization): Clarify error message.
18493 2013-01-04 Richard Biener <rguenther@suse.de>
18495 PR middle-end/55863
18496 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
18499 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18502 * config/pa/pa.md (movsi): Revert previous change.
18503 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
18506 2013-01-03 Richard Henderson <rth@redhat.com>
18508 * config/i386/i386.c (ix86_expand_move): Always assign to op1
18509 after eliminating TLS symbols.
18511 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
18514 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
18515 * graphite-poly.c (debug_gmp_value): Likewise.
18517 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
18520 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
18521 selected code model, define __code_mode_small__, __code_model_medium__,
18522 __code_model_large__, __code_model_32__ or __code_model_kernel__.
18523 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
18524 xchg temporary register with %k. Declare temporary register as
18526 [__x86_64__]: For medium and large code models, preserve %rbx register.
18528 2013-01-03 Richard Biener <rguenther@suse.de>
18530 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
18531 (dump_subscript): Adjust.
18532 (finalize_ddr_dependent): Do not dump redundant info.
18533 (analyze_siv_subscript): Adjust.
18534 (subscript_dependence_tester): Likewise.
18535 (compute_affine_dependence): Likewise.
18537 2013-01-03 Richard Biener <rguenther@suse.de>
18540 2013-01-03 Richard Biener <rguenther@suse.de>
18542 PR tree-optimization/55857
18543 * tree-vect-stmts.c (vectorizable_load): Do not setup
18544 re-alignment for invariant loads.
18546 2013-01-02 Richard Biener <rguenther@suse.de>
18548 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18549 invariant load do not generate a vector load from the scalar location.
18551 2013-01-03 Richard Biener <rguenther@suse.de>
18553 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
18554 for not vectorizing.
18555 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
18556 not build INDIRECT_REFs, call get_name once only.
18557 (vect_create_data_ref_ptr): Likewise. Dump base object kind
18558 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
18560 2013-01-03 Richard Biener <rguenther@suse.de>
18562 PR tree-optimization/55857
18563 * tree-vect-stmts.c (vectorizable_load): Do not setup
18564 re-alignment for invariant loads.
18566 2013-01-03 Richard Biener <rguenther@suse.de>
18569 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
18570 prefer a built-in decl.
18572 2013-01-03 Jakub Jelinek <jakub@redhat.com>
18574 * gcc.c (process_command): Update copyright notice dates.
18575 * gcov.c (print_version): Likewise.
18576 * gcov-dump.c (print_version): Likewise.
18578 PR rtl-optimization/55838
18579 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
18580 iv0.step, iv1.step and step.
18582 2013-01-03 Jakub Jelinek <jakub@redhat.com>
18583 Marc Glisse <marc.glisse@inria.fr>
18585 PR tree-optimization/55832
18586 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
18587 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
18588 integer_{one,zero}_node.
18590 2013-01-03 Jakub Jelinek <jakub@redhat.com>
18593 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
18594 * var-tracking.c (reverse_op): Don't add reverse ops to
18595 VALUEs that have already
18596 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
18598 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
18600 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
18602 2013-01-02 Teresa Johnson <tejohnson@google.com>
18604 * dumpfile.c (dump_loc): Print filename with location.
18605 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
18606 new location_t parameter to emit complete unroll message with
18607 new dump framework.
18608 (canonicalize_loop_induction_variables): Compute loops location
18609 and pass to try_unroll_loop_completely.
18610 * loop-unroll.c (report_unroll_peel): New function.
18611 (peel_loops_completely): Use new dump format with location
18612 for main dumpfile message, and invoke report_unroll_peel on success.
18613 (decide_unrolling_and_peeling): Ditto.
18614 (decide_peel_once_rolling): Remove old dumpfile message subsumed
18615 by report_unroll_peel.
18616 (decide_peel_completely): Ditto.
18617 (decide_unroll_constant_iterations): Ditto.
18618 (decide_unroll_runtime_iterations): Ditto.
18619 (decide_peel_simple): Ditto.
18620 (decide_unroll_stupid): Ditto.
18621 * cfgloop.c (get_loop_location): New function.
18622 * cfgloop.h (get_loop_location): Declare.
18624 2013-01-02 Sriraman Tallam <tmsriram@google.com>
18626 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
18629 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18631 PR middle-end/55198
18632 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
18633 BLKmode objects when EXPAND_MEMORY is specified.
18635 2013-01-02 Sriraman Tallam <tmsriram@google.com>
18637 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
18639 (fold_builtin_cpu): Do not share cpu model decls across statements.
18641 2013-01-02 Jason Merrill <jason@redhat.com>
18644 * tree.c (build_array_type_1): Revert earlier change.
18646 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
18648 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
18650 * config/aarch64/aarch64-tune.md: Re-generate.
18652 2013-01-02 Richard Biener <rguenther@suse.de>
18654 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
18655 invariant load do not generate a vector load from the scalar location.
18657 2013-01-02 Richard Biener <rguenther@suse.de>
18660 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
18661 * configure: Regenerate.
18663 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
18665 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
18666 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
18667 (expand_builtin_int_roundingfn_2): Keep the original target around
18668 for the fallback case.
18670 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
18672 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
18673 to be clear for sign changes.
18675 2013-01-01 Jan Hubicka <jh@suse.cz>
18677 * ipa-inline-analysis.c: Fix formatting.
18679 2013-01-01 Jakub Jelinek <jakub@redhat.com>
18681 PR tree-optimization/55831
18682 * tree-vect-loop.c (get_initial_def_for_induction): Use
18683 gsi_after_labels instead of gsi_start_bb.
18685 Copyright (C) 2013 Free Software Foundation, Inc.
18687 Copying and distribution of this file, with or without modification,
18688 are permitted in any medium without royalty provided the copyright
18689 notice and this notice are preserved.