1 2011-12-02 Sameera Deshpande <sameera.deshpande@arm.com>
3 * gimple-pretty-print.c (dump_gimple_call): Dereference fn only if
6 2011-12-02 Martin Jambor <mjambor@suse.cz>
8 PR tree-optimization/50622
9 * tree-sra.c (load_assign_lhs_subreplacements): Force gimple operand
10 if both lacc and racc are grp_partial_lhs.
12 2011-12-01 Kaz Kojima <kkojima@gcc.gnu.org>
15 * config/sh/sh.c (expand_ashiftrt): Handle TARGET_SH2A same as
17 (shl_sext_kind): Likewise.
18 * config/sh/sh.h (SH_DYNAMIC_SHIFT_COST): Likewise.
19 * config/sh/sh.md (ashlsi3_sh2a, ashrsi3_sh2a, lshrsi3_sh2a):
21 (ashlsi3_std): Handle TARGET_SH2A same as TARGET_SH3.
23 (ashrsi3_d): Likewise.
24 (lshrsi3_d): Likewise.
27 2011-12-01 Diego Novillo <dnovillo@google.com>
30 * ipa-inline.c (can_inline_edge_p): If the edge E has a
31 statement, use the statement's inline indicator instead
33 Remove consistency check.
35 2011-12-01 Diego Novillo <dnovillo@google.com>
40 2011-11-29 Diego Novillo <dnovillo@google.com>
42 * gimple.c (gimple_call_set_cannot_inline): Move from gimple.h.
43 Update field call_stmt_cannot_inline_p from call
44 graph edge, if needed.
45 * gimple.h (gimple_call_set_cannot_inline): Move to gimple.c.
47 2011-12-01 Nathan Sidwell <nathan@acm.org>
50 * coverage.c (build_var): Keep coverage variables static.
52 2011-12-01 Jakub Jelinek <jakub@redhat.com>
54 PR tree-optimization/51356
55 * tree-vect-patterns.c (vect_recog_bool_pattern): Give up if
56 vectype doesn't have VECTOR_MODE_P.
59 * tree-ssa-dce.c (remove_dead_stmt): Add a debug stmt when removing
60 as unnecessary a store to a variable with gimple reg type.
61 * tree-ssa-live.c (remove_unused_locals): Clear TREE_ADDRESSABLE bit
62 on local unreferenced variables.
63 * cfgexpand.c (expand_gimple_basic_block): Don't emit DEBUG_INSNs
64 for !target_for_debug_bind variables.
66 2011-12-01 Patrick Marlier <patrick.marlier@gmail.com>
69 * cgraph.h (cgraph_call_node_duplication_hooks): Declare.
70 * cgraph.c (cgraph_call_node_duplication_hooks): Make global.
71 * cgraphunit.c (cgraph_copy_node_for_versioning): Call it.
73 2011-12-01 Andrew Pinski <apinski@cavium.com>
76 * tree.c (free_lang_data_in_decl): Clear FIELD_DECL's
79 2011-12-01 Jakub Jelinek <jakub@redhat.com>
81 PR tree-optimization/51246
82 * tree-predcom.c (replace_ref_with): Handle also clobber on the
85 PR rtl-optimization/51014
86 * loop-unroll.c (apply_opt_in_copies): Ignore label DEBUG_INSNs
87 both from bb and orig_bb.
89 2011-12-01 Joern Rennecke <joern.rennecke@embecosm.com>
91 PR tree-optimization/50802
92 * tree-vrp.c (simplify_conversion_using_ranges): Rewrite test
93 considering what happens to ranges during sign changes and/or
94 intermediate narrowing conversions.
96 2011-11-30 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
99 * config/pa/pa.md (in_branch_delay): Disallow frame related insns.
100 (in_nullified_branch_delay): Likewise.
101 (in_call_delay): Likewise.
103 2011-11-30 Richard Henderson <rth@redhat.com>
105 * predict.c (expr_expected_value_1): Assume compare-and-swap builtin
106 boolean return is true.
108 2011-11-30 Jakub Jelinek <jakub@redhat.com>
110 PR rtl-optimization/51044
111 * dse.c (emit_inc_dec_insn_before): Ensure new_insn is an instruction
112 rather than just insn body.
115 * function.c (thread_prologue_and_epilogue_insns): If
116 stack_realign_drap, add drap_reg to SET_UP_BY_PROLOGUE
120 * tree-eh.c (cleanup_empty_eh_merge_phis): Add check to
121 avoid creating duplicate edges here.
122 (cleanup_empty_eh_unsplit): And remove it in the caller.
124 2011-11-30 Andrew Pinski <apinski@cavium.com>
127 * c-parser.c (c_parser_postfix_expression): Check groktypename results
128 before looking at the main variant.
130 2011-11-30 Jakub Jelinek <jakub@redhat.com>
132 PR rtl-optimization/48721
133 * sched-deps.c (sched_analyze_insn): For SIBLING_CALL_P set
134 reg_pending_barrier to TRUE_BARRIER.
136 PR rtl-optimization/48721
137 * sched-ebb.c (begin_move_insn): Insert empty unreachable
138 block after BARRIER if insn is followed by it.
140 2011-11-30 Richard Henderson <rth@redhat.com>
142 * config/sparc/sparc-opts.h (enum sparc_memory_model_type): New.
143 * config/sparc/sparc.opt (mmemory-model=): New option.
144 * doc/invoke.texi (Sparc Options): Document it.
145 * config/sparc/sparc.c (sparc_option_override): Provide default
146 for sparc_memory_model.
147 (sparc_emit_membar_for_model): Omit barrier combinations that are
148 implied by the memory model.
150 2011-11-30 Richard Henderson <rth@redhat.com>
152 * config/sparc/sync.md (I124MODE): New mode iterator.
153 (memory_barrier): Remove.
154 (atomic_exchangesi): New.
155 (swapsi): Rename from *swapsi.
156 (atomic_test_and_set<I124MODE>): New.
157 (ldstubqi): Rename from *ldstubqi, merge with expander.
158 (ldstub<I24MODE>): Rename from *ldstub<I24MODE>, merge with expander.
159 (sync_lock_test_and_setsi): Remove.
160 (sync_lock_test_and_set<I12MODE>): Remove.
162 2011-11-30 Richard Henderson <rth@redhat.com>
164 * config/sparc/constraints.md ("w"): New.
165 * config/sparc/predicates.md (mem_noofs_operand): New.
166 * config/sparc/sparc.c (sparc_expand_compare_and_swap_12): Make static.
167 Generate the boolean result of the atomic_compare_exchange.
168 (sparc_expand_compare_and_swap): New.
169 * config/sparc/sync.md (atomic_compare_and_swap<I>): New.
170 (atomic_compare_and_swap<I48MODE>_1): Rename from
171 *sync_compare_and_swap<I48MODE>, use mem_noofs_operand.
172 (*atomic_compare_and_swapdi_v8plus): Rename from
173 *sync_compare_and_swapdi_v8plus, use mem_noofs_operand.
174 (sync_compare_and_swap<I12MODE>): Remove.
175 (sync_compare_and_swap<I48MODE>): Remove.
177 2011-11-30 Richard Henderson <rth@redhat.com>
179 * config/sparc/predicates.md (register_or_v9_zero_operand): New.
180 * config/sparc/sparc.md (UNSPEC_ATOMIC): New.
181 * config/sparc/sync.md (atomic_load<I>): New.
182 (atomic_loaddi_1, atomic_store<I>, atomic_storedi_1): New.
184 2011-11-30 Richard Henderson <rth@redhat.com>
186 * config/sparc/predicates.md (zero_or_v7_operand): New.
187 * config/sparc/sparc.c (sparc_emit_membar_for_model): New.
188 * config/sparc/sparc-protos.h: Update.
189 * config/sparc/sync.md (mem_thread_fence): New.
190 (memory_barrier): Use sparc_emit_membar_for_model.
191 (membar, *membar_empty, *membar_storestore, *membar_storeload): New.
192 (*membar_v8): Accept and ignore the membar mask.
193 (*membar): Accept and print the membar mask.
195 2011-11-30 Matthew Gretton-Dann <matthew.gretton-dann@arm.com>
197 * config/arm/arm.c (arm_issue_rate): Cortex-A15 can triple issue.
198 * config/arm/arm.md (mul64): New attribute.
199 (generic_sched): Cortex-A15 is not scheduled generically.
200 (cortex-a15.md): Include.
201 * config/arm/cortex-a15.md: New machine description.
202 * config/arm/t-arm (MD_INCLUDES): Add cortex-a15.md.
204 2011-11-30 Matthew Gretton-Dann <matthew.gretton-dann@arm.com>
206 * config/arm/t-arm (MD_INCLUDES): Ensure all md files are listed.
208 2011-11-30 Iain Sandoe <iains@gcc.gnu.org>
210 * doc/tm.texi.in (NEXT_OBJC_RUNTIME): Update macro description.
211 * doc/tm.texi: Regenerate.
212 * flags.h (flag_next_runtime): Remove references.
213 * toplev.c: Likewise.
214 * defaults.h (NEXT_OBJC_RUNTIME): Provide default.
215 * gccspec.c (lang_specific_driver): Adjust use of NEXT_OBJC_RUNTIME.
216 * config/darwin.c (darwin_override_options): Provide default
217 Objective-C abi settings and target conflict checks.
218 * config/darwin.h (SUBTARGET_C_COMMON_OVERRIDE_OPTIONS): Remove unused
219 code. Adjust indenting.
221 2011-11-29 Joseph Myers <joseph@codesourcery.com>
223 * config/sh/sh.h (enum reg_class, REG_CLASS_NAMES)
224 (REG_CLASS_CONTENTS): Add NON_SP_REGS.
225 (REGCLASS_HAS_GENERAL_REG): Handle NON_SP_REGS.
226 * config/sh/constraints.md (u): New constraint.
228 2011-11-29 Oleg Endo <oleg.endo@t-online.de>
231 * config/sh/sh.c (sh_secondary_reload): Add case when FPUL
232 register is being loaded from a pseudo in memory.
234 2011-11-29 DJ Delorie <dj@redhat.com>
236 * config.gcc (rl78-*-elf): New case.
237 * doc/extend.texi: Add RL78 documentation.
238 * doc/invoke.texi: Likewise.
239 * doc/md.texi: Likewise.
240 * doc/contrib.texi: Add RL78.
241 * doc/install.texi: Add rl78-*-elf.
242 * config/rl78: New directory for the Renesas RL78.
244 2011-11-29 Jakub Jelinek <jakub@redhat.com>
246 PR tree-optimization/51247
247 * tree-vrp.c (extract_range_from_assert): For signed 1-bit precision
248 types instead of adding 1 subtract -1 and instead of subtracting 1
249 add -1 to avoid overflows.
251 2011-11-29 Andrew MacLeod <amacleod@redhat.com>
254 * optabs.c (maybe_optimize_fetch_op): New. Look for more optimal
255 instructions for a FECTH_OP or OP_FECTH sequence.
256 (expand_atomic_fetch_op): Call maybe_optimize_fetch_op.
258 2011-11-29 Uros Bizjak <ubizjak@gmail.com>
260 * config/i386/sync.md (UNSPEC_LDA, UNSPEC_STA): New unspecs.
261 (movdi_via_fpu): Remove.
262 (loaddi_via_fpu): New insn pattern.
263 (storedi_via_fpu): Ditto.
264 (atomic_loaddi_fpu): Use loaddi_via_fpu and storedi_via_fpu.
265 (atomic_storedi_fpu): Ditto.
266 * reg-stack.c (get_true_reg): Handle UNSPEC_LDA.
267 (subst_stack_regs_pat): Handle UNSPEC_STA.
269 2011-11-29 Uros Bizjak <ubizjak@gmail.com>
271 * config/i386/i386.md (*floathi<mode>2_i387_with_temp): Do not
272 allocate scratch memory for alternative 0.
274 2011-11-29 Michael Meissner <meissner@linux.vnet.ibm.com>
276 * config/rs6000/rs6000-builtins.def: Completely rewrite builtin
277 handling to centralize more of the builtins in this file. Change
278 some builtin enumerations to be more consistant. Use a new mask
279 to hold the current builtins, including SPE and PAIRED builtins
280 which no longer are set via target_flags masks. Add
281 -mdebug=builtin debug support. For power machines, define all
282 Altivec and VSX buitins when the compiler starts, but don't allow
283 the use of a builtin unless the appropriate switch is used, or
284 #pragma GCC target is used to change the options. If the user
285 uses #pragma GCC target, update the appropriate hardware macros.
286 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Ditto.
287 * config/rs6000/rs6000.opt (rs6000_builtin_mask): Ditto.
288 * config/rs6000/rs6000-c.c (rs6000_macro_to_expand): Ditto.
289 (rs6000_define_or_undefine_macro): Ditto.
290 (rs6000_target_modify_macros): Ditto.
291 (rs6000_cpu_cpp_builtins): Ditto.
292 (altivec_overloaded_builtins): Ditto.
293 (altivec_build_resolved_builtin): Ditto.
294 * config/rs6000/rs6000.c (rs6000_target_modify_macros_ptr): Ditto.
295 (rs6000_builtin_info): Ditto.
296 (def_builtin): Ditto.
297 (enable_mask_for_builtins): Ditto.
298 (DEBUG_FMT_X): Ditto.
299 (rs6000_debug_reg_global): Ditto.
300 (rs6000_builtin_mask_calculate): Ditto.
301 (rs6000_option_override_internal): Ditto.
302 (rs6000_builtin_conversion): Ditto.
303 (rs6000_builtin_vectorized_function): Ditto.
307 (builtin_description_predicates): Ditto.
308 (bdesc_altivec_preds): Ditto.
309 (bdesc_spe_predicates): Ditto.
310 (bdesc_spe_evsel): Ditto.
311 (bdesc_paired_preds): Ditto.
314 (rs6000_overloaded_builtin_p): Ditto.
315 (rs6000_expand_unop_builtin): Ditto.
316 (bdesc_2arg_spe): Ditto.
317 (spe_expand_builtin): Ditto.
318 (rs6000_invalid_builtin): Ditto.
319 (rs6000_expand_builtin): Ditto.
320 (rs6000_init_builtins): Ditto.
321 (spe_init_builtins): Ditto.
322 (paired_init_builtins): Ditto.
323 (altivec_init_builtins): Ditto.
324 (builtin_function_type): Ditto.
325 (rs6000_common_init_builtins): Ditto.
326 (rs6000_builtin_reciprocal): Ditto.
327 (rs6000_builtin_mask_names): Ditto.
328 (rs6000_pragma_target_parse): Ditto.
329 (rs6000_function_specific_print): Ditto.
330 * config/rs6000/rs6000.h (MASK_DEBUG_BUILTIN): Ditto.
331 (MASK_DEBUG_ALL): Ditto.
332 (TARGET_DEBUG_BUILTIN): Ditto.
333 (TARGET_EXTRA_BUILTINS): Ditto.
334 (REGISTER_TARGET_PRAGMAS): Ditto.
335 (enum rs6000_btc): Ditto.
336 (RS6000_BTC_*): Ditto.
337 (RS6000_BTM_*): Ditto.
338 (enum rs6000_builtins): Ditto.
339 * config/rs6000/rs6000-protos.h (rs6000_overloaded_builtin_p): Ditto.
340 (rs6000_target_modify_macros): Ditto.
341 (rs6000_target_modify_macros_ptr): Ditto.
343 * config/rs6000/vector.md (unsfloat<VEC_int><mode>2): Use the
344 standard name for converting vector unsigned values to floating
345 point. Delete the TARGET_VECTORIZE_BUILTIN_CONVERSION hook, which
346 is handled by machine independent code.
347 * config/rs6000/rs6000.c (TARGET_VECTORIZE_BUILTIN_CONVERSION): Ditto.
348 (rs6000_builtin_conversion): Ditto.
350 2011-11-29 Dodji Seketeli <dodji@redhat.com>
352 * tree-diagnostic (struct loc_t): Rename into struct loc_map_pair.
353 (maybe_unwind_expanded_macro_loc): Adjust.
355 2011-11-29 David Edelsohn <dje.gcc@gmail.com>
356 Aldy Hernandez <aldyh@redhat.com>
358 * trans-mem.c (ipa_tm_create_version_alias): Mangle new_decl
360 (ipa_tm_create_version): Same.
362 2011-11-29 Bernd Schmidt <bernds@codesourcery.com>
364 * haifa-sched.c (recompute_todo_spec): Simplify and correct the
365 code checking for a clobber of a condition register when deciding
366 whether to predicate.
368 2011-11-29 Diego Novillo <dnovillo@google.com>
370 * gimple.c (gimple_call_set_cannot_inline): Move from gimple.h.
371 Update field call_stmt_cannot_inline_p from call graph edge, if needed.
372 * gimple.h (gimple_call_set_cannot_inline): Move to gimple.c.
374 2011-11-29 Sameera Deshpande <sameera.deshpande@arm.com>
376 * config/arm/arm.md (arm_movdi): Update NEG_POOL_RANGE.
377 (movdf_soft_insn): Likewise.
378 * config/arm/fpa.md (thumb2_movdf_fpa): Likewise.
379 * config/arm/neon.md (neon_mov<mode>): Likewise.
380 * config/arm/vfp.md (movdi_vfp): Likewise.
381 (movdi_vfp_cortexa8): Likewise.
382 (movdf_vfp): Likewise.
384 2011-11-29 Jakub Jelinek <jakub@redhat.com>
387 * expr.c (expand_expr_addr_expr_1): Don't call force_operand for
390 2011-11-29 Ira Rosen <ira.rosen@linaro.org>
392 PR tree-optimization/51301
393 * tree-vect-patterns.c (vect_recog_over_widening_pattern): Check that
394 the last statement doesn't convert to a bigger type than the original
395 type of the computation.
397 2011-11-28 Richard Henderson <rth@redhat.com>
399 * config/rs6000/rs6000.c (rs6000_expand_atomic_compare_and_swap):
400 Handle overlap between retval and oldval. Always compute boolval
403 2011-11-28 Jakub Jelinek <jakub@redhat.com>
406 * tree-ssa.c (execute_update_addresses_taken): Remove
407 var ={v} {CLOBBER} stmts instead of rewriting them into
408 var_N ={v} {CLOBBER}.
411 * function.c (convert_jumps_to_returns): When redirecting an edge
412 succ to EXIT_BLOCK_PTR, clear EDGE_CROSSING flag.
414 PR tree-optimization/50078
415 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Copy over
416 TREE_THIS_VOLATILE also from the old to new lhs resp. rhs.
418 PR tree-optimization/50682
419 * tree-eh.c (maybe_remove_unreachable_handlers): New function.
420 * tree-flow.h (maybe_remove_unreachable_handlers): New prototype.
421 * tree-optimize.c (execute_cleanup_cfg_post_optimizing): Call it.
423 2011-11-28 Georg-Johann Lay <avr@gjlay.de>
425 * doc/extend.texi (AVR Built-in Functions): Add documentation for
426 __builtin_avr_map8 and __builtin_avr_map16.
428 * config/avr/avr.md: Document new %t and %T asm output codes.
429 (define_c_enum "unspec"): Add UNSPEC_MAP_BITS.
430 (adjust_len): Add map_bits.
431 (map_bitsqi, map_bitshi): New insns.
432 * config/avr/avr-protos.h (avr_out_map_bits): New.
433 * config/avr/avr-protos.c (print_operand): Implement %t and %T.
434 (adjust_insn_length): Handle ADJUST_LEN_MAP_BITS.
435 (avr_double_int_push_digit): New function.
436 (avr_map, avr_revert_map, avr_swap_map, avr_id_map): New functions.
437 (avr_sig_map, avr_map_hamming_byte): New functions.
438 (avr_out_swap_bits, avr_out_revert_bits, avr_move_bits,
439 avr_out_map_bits): New functions.
440 (enum avr_builtin_id): Add AVR_BUILTIN_MAP8, AVR_BUILTIN_MAP16.
441 (avr_init_builtins): Populate __builtin_avr_map8, __builtin_avr_map16.
442 (bdesc_2arg): Add __builtin_avr_map8, __builtin_avr_map16 ...
443 (avr_expand_builtin): ...and expand them.
444 * config/avr/avr-c.c (avr_cpu_cpp_builtins): New built-in defines:
445 __BUILTIN_AVR_MAP8, __BUILTIN_AVR_MAP16.
447 2011-11-27 Richard Sandiford <rdsandiford@googlemail.com>
450 * config/mips/mips.c (mips_rtx_costs): Handle baddu.
452 2011-11-27 Richard Sandiford <rdsandiford@googlemail.com>
454 * config/mips/mips.c (mips16_unextended_reference_p): Test for BLKmode.
456 2011-11-27 Richard Sandiford <rdsandiford@googlemail.com>
458 * config/mips/mips-protos.h (mips_emit_binary): Declare.
459 * config/mips/mips.c (mips_emit_binary): Make global.
460 (mips_set_mips16_mode): Turn off -mfix-r4000 in MIPS16 mode.
461 (mips_conditional_register_usage): Don't treat LO and HI as
462 register operands in MIPS16 mode.
463 (mips_mulsidi3_gen_fn): Use {u,}mulsidi3_{32,64}bit_mips16
465 * config/mips/predicates.md (muldiv_target_operand): New predicate.
466 (move_operand): Allow hilo_operand.
467 * config/mips/mips.md (mul<mode>3): Explicitly specify LO as the
468 target of MIPS16 multiplies, then move it into the target register.
469 (mul<mode>3_internal, *macc2, *msac2): Use muldiv_target_operand.
470 (<u>mulsidi3_32bit_mips16): New expander.
471 (<u>mulsidi3_32bit): Use muldiv_target_operand.
472 (<u>mulsidi3_32bit_r4000): Disable for ISA_HAS_DSP.
473 (<u>mulsidi3_64bit): Require !TARGET_MIPS16. Split into
474 <u>mulsidi3_64bit_split.
475 (<u>mulsidi3_64bit_mips16): New expander.
476 (<u>mulsidi3_64bit_split): Likewise, using expansions from
477 two previous define_splits.
478 (<u>mulsidi3_64bit_hilo, *muls<u>_di, <u>msubsidi4): Use
479 muldiv_target_operand.
480 (<su>mulsi3_highpart): Use <su>mulsi3_highpart_split for MIPS16 code.
481 (<su>mulsi3_highpart_internal): Require !TARGET_MIPS16.
482 Split into <su>mulsi3_highpart_split.
483 (<su>mulsi3_highpart_split): New expander.
484 (<su>muldi3_highpart): Turn into a define_expand.
485 Use <su>muldi3_highpart_split for MIPS16 code.
486 (<su>muldi3_highpart_internal): Renamed from <su>muldi3_highpart.
487 Require !TARGET_MIPS16. Split into <su>muldi3_highpart_split.
488 (<su>muldi3_highpart_split): New expander.
489 (<u>mulditi3): Explicitly specify LO as the target of MIPS16
490 multiplies, then move it into the target register.
491 (<u>mulditi3_internal, <u>maddsidi4): Use muldiv_target_operand.
492 (divmod<mode>4, udivmod<mode>4): Turn into define_expands.
493 Use <u>divmod<mode>4_split for MIPS16 code, then explicitly
494 move LO into operand 0.
495 (divmod<mode>4_internal, udivmod<mode>4_internal): Renamed
496 from <u>divmod<mode>4. Use muldiv_target_operand.
497 Require !TARGET_MIPS16. Split into <u>divmod<mode>4_split.
498 (<u>divmod<mode>4_split): New expander.
499 (<u>divmod<GPR:mode>4_hilo_<HILO:mode>): Use muldiv_target_operand.
500 (mfhi<GPR:mode>_<HILO:mode>): Use hilo_operand.
502 2011-11-27 Richard Sandiford <rdsandiford@googlemail.com>
504 * hard-reg-set.h (target_hard_regs): Add x_accessible_reg_set
505 and x_operand_reg_set.
506 (accessible_reg_set, operand_reg_set): New macros.
507 * reginfo.c (init_reg_sets): Initialize accessible_reg_set and
509 (saved_accessible_reg_set, saved_operand_reg_set): New variables.
510 (save_register_info): Save them.
511 (restore_register_info): Restore them.
512 (init_reg_sets_1): Limit operand_reg_set to accessible_reg_set.
513 Remove NO_REGS registers from operand_reg_set. Treat members
514 of operand_reg_set as fixed.
515 * recog.c (general_operand): Check operand_reg_set rather than NO_REGS.
516 (register_operand, nonmemory_operand): Likewise.
517 * varasm.c (make_decl_rtl): Always use DECL_MODE as the mode of
518 register variables. Check accessible_reg_set and operand_reg_set.
519 * config/mips/mips.c (mips_conditional_register_usage): Remove
520 inaccessible register from accessible_reg_set, rather than just
523 2011-11-27 Gerald Pfeifer <gerald@pfeifer.com>
525 * config/pa/pa-linux.h (TARGET_GAS): Remove comment.
527 2011-11-26 Richard Henderson <rth@redhat.com>
529 * config/arm/arm.c (arm_init_libfuncs): Call init_sync_libfuncs
532 2011-11-26 Richard Henderson <rth@redhat.com>
534 * optabs.c (expand_atomic_fetch_op): Always return result.
536 * optabs.c (init_sync_libfuncs_1): Include max in iteration.
538 * omp-low.c (expand_omp_atomic): Assume anything aligned to
539 BIGGEST_ALIGNMENT is aligned.
541 2011-11-26 Richard Henderson <rth@redhat.com>
543 * config/m68k/m68k.md (UNSPECV_CAS_1, UNSPECV_CAS_2): New.
544 (UNSPECV_TAS_1, UNSPECV_TAS_2): New.
545 (I): New mode iterator.
546 (xz): New mode attribute.
547 * config/m68k/sync.md: New file.
549 * config/m68k/linux.h (TARGET_INIT_LIBFUNCS): New.
550 * config/m68k/m68k.c (m68k_init_sync_libfuncs): New.
552 2011-11-26 Nathan Sidwell <nathan@acm.org>
554 PR gcov-profile/51297
555 * gcov.c (main): Allocate initial names and sources arrays.
556 (find_source): Don't check for null name or source arrays here.
558 2011-11-26 Jakub Jelinek <jakub@redhat.com>
560 PR rtl-optimization/49912
561 * cfgrtl.c (rtl_verify_flow_info_1): Ignore also EDGE_PRESERVE bit
562 when counting n_branch.
564 2011-11-26 Uros Bizjak <ubizjak@gmail.com>
566 * config/i386/sync.md (movdi_via_fpu): Add %Z insn suffixes.
568 2011-11-26 Joern Rennecke <joern.rennecke@embecosm.com>
571 * calls.c (mem_overlaps_already_clobbered_arg_p):
572 Return false if no outgoing arguments have been stored so far.
574 2011-11-25 Georg-Johann Lay <avr@gjlay.de>
576 * config/avr/avr.c (print_operand): Support code = 'i' for CONST_INT.
578 2011-11-24 Andrew MacLeod <amacleod@redhat.com>
581 * tree.h (is_builtin_name): No longer external.
582 * builtins.c (is_builtin_name): Make static.
583 * varasm.c (incorporeal_function_p): __sync and __atomic external calls
584 are not incorporeal and may need asm label processing.
586 2011-11-25 Georg-Johann Lay <avr@gjlay.de>
589 * config/avr/avr-protos.h (avr_log_t): Add field .builtin.
590 * config/avr/avr-log.c (avr_log_set_avr_log): Initialize it.
591 Don't bypass TARGET_ALL_DEBUG. Print self-info with ?.
592 (avr_log_vadump): Support %D and %X to print double_int.
593 (avr_double_int_pop_digit): New static function.
594 (avr_dump_double_int_hex): New static function.
596 2011-11-24 Enkovich Ilya <ilya.enkovich@intel.com>
599 * config/i386/i386.c (distance_non_agu_define_in_bb): Fix insn
602 2011-11-24 Andrew MacLeod <amacleod@redhat.com>
604 * optab.c (maybe_emit_atomic_exchange): New. Try to emit an
605 atomic_exchange pattern.
606 (maybe_emit_sync_lock_test_and_set): New. Try to emit an exchange
607 using __sync_lock_test_and_set.
608 (maybe_emit_compare_and_swap_exchange_loop): New. Try to emit an
609 exchange using a compare_and_swap loop.
610 (expand_sync_lock_test_and_set): New. Expand sync_lock_test_and_set.
611 (expand_atomic_test_and_set): New. Expand test_and_set operation.
612 (expand_atomic_exchange): Use new maybe_emit_* functions.
613 (expand_atomic_store): Use new maybe_emit_* functions.
614 * builtins.c (expand_builtin_sync_lock_test_and_set): Call
615 expand_sync_lock_test_and_set routine.
616 (expand_builtin_atomic_exchange): Remove parameter from call.
617 (expand_builtin_atomic_clear): Use atomic_clear pattern if present.
618 (expand_builtin_atomic_test_and_set): Add target and simply call
619 expand_atomic_test_and_set.
620 (expand_builtin): Add target to expand_builtin_atomic_test_and_set.
621 * expr.h (expand_atomic_exchange): Add parameter.
622 (expand_sync_lock_test_and_set): New prototype.
623 (expand_atomic_test_and_set, expand_atomic_clear): New prototypes.
625 2011-11-24 H.J. Lu <hongjiu.lu@intel.com>
628 * config/i386/i386.h (processor_costs): Revert revision 181357.
629 * config/i386/i386.c (cost models): Likewise.
630 (core_cost): Likewise.
631 (promote_duplicated_reg): Likewise.
632 (promote_duplicated_reg_to_size): Likewise.
633 (processor_target): Likewise.
634 (expand_set_or_movmem_via_loop_with_iter): Likewise.
635 (expand_set_or_movmem_via_loop): Likewise.
636 (emit_strset): Likewise.
637 (expand_movmem_epilogue): Likewise.
638 (expand_setmem_epilogue): Likewise.
639 (expand_movmem_prologue): Likewise.
640 (expand_setmem_prologue): Likewise.
641 (expand_constant_movmem_prologue): Likewise.
642 (expand_constant_setmem_prologue): Likewise.
643 (decide_alg): Likewise.
644 (decide_alignment): Likewise.
645 (ix86_expand_movmem): Likewise.
646 (ix86_expand_setmem): Likewise.
647 (ix86_slow_unaligned_access): Likewise.
648 * config/i386/i386.md (strset): Likewise.
649 * config/i386/sse.md (vec_dupv4si): Likewise.
650 (vec_dupv2di): Likewise.
652 2011-11-24 Georg-Johann Lay <avr@gjlay.de>
654 * config/avr/avr.md (*rotlhi2.15): Set length to 4.
656 2011-11-24 Razya Ladelsky <razya@il.ibm.com>
658 * tree-data-ref.c (initialize_data_dependence_relation): Update
659 comment for the self dependence case.
660 (compute_self_dependence): Remove.
661 * tree-vect-data-refs.c (vect_analyze_data_refs): Remove call to
662 compute_self_dependenc.
664 2011-11-24 Alan Modra <amodra@gmail.com>
666 * c-decl.c (warn_if_shadowing): Don't warn if shadowed
667 identifier is from system header.
669 2011-11-23 Jakub Jelinek <jakub@redhat.com>
672 * config/i386/i386.c (decide_alg): Initialize *dynamic_check
675 2011-11-23 Vladimir Makarov <vmakarov@redhat.com>
677 PR rtl-optimization/48455
678 * doc/invoke.texi (-fira-region): Document default values.
680 * flags-types.h (enum ira_region): Add new value
681 IRA_REGION_AUTODETECT.
683 * common.opt (fira-region): Set up initial value to
684 IRA_REGION_AUTODETECT.
686 * toplev.c (process_options): Set up flag_ira_region depending on
689 * ira.c (ira.c): Remove optimize guard for ira_build.
691 2011-11-23 Chung-Lin Tang <cltang@codesourcery.com>
693 PR rtl-optimization/50496
694 * jump.c (redirect_jump): Assert fail on nlabel == NULL_RTX
695 only after epilogue is created. Add comments.
697 2011-11-22 Richard Henderson <rth@redhat.com>
699 * config/ia64/ia64.c (ia64_expand_atomic_op): Add model parameter.
700 Generate the barrier required for the memory model.
701 (rtx_needs_barrier): Handle UNSPEC_FETCHADD_REL, UNSPEC_CMPXCHG_REL.
702 * config/ia64/ia64-protos.h: Update.
703 * config/ia64/ia64.md (UNSPEC_FETCHADD_REL): New.
704 (UNSPEC_CMPXCHG_REL): New.
705 * config/ia64/sync.md (mem_thread_fence): New.
706 (atomic_load<IMODE>, atomic_store<IMODE>): New.
707 (atomic_compare_and_swap<IMODE>): New.
708 (cmpxchg_acq_<I124MODE>, cmpxchg_acq_di): New.
709 (atomic_exchange<IMODE>): New.
710 (xchg_acq_<IMODE>): Rename from sync_lock_test_and_set<IMODE>.
711 (atomic_<FETCHOP><IMODE>, atomic_nand<IMODE>): New.
712 (atomic_fetch_<FETCHOP><IMODE>, atomic_fetch_nand<IMODE>): New.
713 (atomic_<FETCHOP>_fetch<IMODE>, atomic_nand_fetch<IMODE>): New.
714 (fetchadd_rel_<I48MODE>): New.
715 (sync_<FETCHOP><IMODE>, sync_nand<IMODE>): Remove.
716 (sync_old_<FETCHOP><IMODE>, sync_old_nand<IMODE>): Remove.
717 (sync_new_<FETCHOP><IMODE>, sync_new_nand<IMODE>): Remove.
718 (sync_compare_and_swap<IMODE>): Remove.
719 (sync_lock_release<IMODE>): Remove.
721 * config/ia64/ia64.md: Use define_c_enum for UNSPEC_*
722 and UNSPECV_* constants.
724 2011-11-22 Tom de Vries <tom@codesourcery.com>
726 PR rtl-optimization/50764
727 * sched-deps.c (sched_analyze_insn): Make sure frame-related insns are
728 not moved past preceding jump.
730 2011-11-22 Jeff Law <law@redhat.com>
732 * doc/contrib.texi: Add entry for David Binderman.
734 2011-11-22 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
736 * postreload.c (reload_combine): Mark reg_state as invalid at
737 volatile insns if there has been a use already.
739 2011-11-21 Aldy Hernandez <aldyh@redhat.com>
741 * opts.c (finish_options): Do not fail for -fgnu-tm.
742 * gimple-streamer-out.c (output_gimple_stmt): Handle
744 * gimple-streamer-in.c (input_gimple_stmt): Same.
745 * lto-cgraph.c (input_overwrite_node): Read tm_clone bit.
746 (lto_output_node): Write tm_clone bit.
748 2011-11-22 Ian Lance Taylor <iant@google.com>
750 * doc/install.texi (Configuration): Correct doc of
751 --enable-build-poststage1-with-cxx: it is not experimental.
753 2011-11-22 Jakub Jelinek <jakub@redhat.com>
755 PR tree-optimization/51074
756 * fold-const.c (fold_binary_loc): Fix up VEC_INTERLEAVE_*_EXPR
757 handling for BYTES_BIG_ENDIAN.
758 * optabs.c (can_vec_perm_for_code_p): Likewise.
760 2011-11-22 Michael Matz <matz@suse.de>
763 * trans-mem.c (expand_block_tm): Ignore clobbers.
765 2011-11-22 Iain Sandoe <iains@gcc.gnu.org>
767 * config/darwin.h (ENDFILE_SPEC): New.
768 (DARWIN_CRT1_SPEC): Add crttms.o.
769 (STARTFILE_SPEC): Likewise for dynamic libs and bundles.
770 * config/i386.h (ENDFILE_SPEC): Amend to include crttme.o.
772 2011-11-22 Michael Matz <matz@suse.de>
775 * tree.c (iterative_hash_expr): Remove break after return.
777 2011-11-22 Iain Sandoe <iains@gcc.gnu.org>
779 * target.def (tm_clone_table_section): New hook.
780 * doc/tm.texi.in (TARGET_ASM_TM_CLONE_TABLE_SECTION): Define.
781 * doc/tm.texi: Regenerate.
782 * varasm.c (dump_tm_clone_pairs): Use target tm_clone_table_section.
783 * output.h (default_clone_table_section): New prototype.
784 * config/darwin.h (TARGET_ASM_TM_CLONE_TABLE_SECTION): New.
785 * config/darwin-protos.h (darwin_tm_clone_table_section): New prototype.
786 * config/darwin.c (darwin_tm_clone_table_section): New.
788 2011-11-21 Torvald Riegel <triegel@redhat.com>
790 * trans-mem.c (diagnose_tm_1): Print an expression instead of a
791 declaration in error messages for indirect calls.
793 2011-11-21 David S. Miller <davem@davemloft.net>
795 * config/sparc/sparc.c (sparc_regmode_natural_size): New function
796 implementing REGMODE_NATURAL_SIZE taking into consideration vector
798 (sparc_modes_tieable_p): Similarly for MODES_TIEABLE_P.
799 * config/sparc/sparc-protos.h (sparc_regmode_natural_size,
800 sparc_modes_tieable_p): Declare.
801 * gcc/config/sparc/sparc.h (REGMODE_NATURAL_SIZE,
802 MODES_TIEABLE_P): Use new helper functions.
805 2011-11-16 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
808 * expmed.c (store_bit_field_1): Use extract_bit_field on big
809 endian targets if the source cannot be exactly covered by word
812 2011-11-21 Jakub Jelinek <jakub@redhat.com>
815 * var-tracking.c (loc_cmp): For ENTRY_VALUEs use ENTRY_VALUE_EXP
816 macro instead of XEXP.
818 2011-11-21 Andreas Tobler <andreast@fgznet.ch>
820 * configure: Regenerate.
822 2011-11-21 Georg-Johann Lay <avr@gjlay.de>
824 * config/avr/avr.c (output_reload_in_const): Loading a byte with 0
826 * config/avr/avr.md (*movhi, *movpsi, *movsi, *movsf): Zero to any
827 register does not change cc0. Same for any constant to ld-register.
829 2011-11-21 Uros Bizjak <ubizjak@gmail.com>
831 * config/i386/mmx.md (unspec) <UNSPEC_MOVNTQ, UNSPEC_PFRCP,
832 UNSPEC_PFRCPIT1, UNSPEC_PFRCPIT2, UNSPEC_PFRSQRT, UNSPEC_PFRSQIT1>:
833 Move from config/i386/i386.md
834 (unspecv) <UNSPECV_EMMS, UNSPECV_FEMMS>: Ditto.
835 * config/i386/sse.md (unspec) <UNSPEC_MOVNT,UNSPEC_MOVU, UNSPEC_LDDQU,
836 UNSPEC_PSHUFB, UNSPEC_PSIGN, UNSPEC_PALIGNR, UNSPEC_EXTRQI,
837 UNSPEC_EXTRQ, UNSPEC_INSERTQI, UNSPEC_INSERTQ, UNSPEC_BLENDV,
838 UNSPEC_INSERTPS, UNSPEC_DP, UNSPEC_MOVNTDQA, UNSPEC_MPSADBW,
839 UNSPEC_PHMINPOSUW, UNSPEC_PTEST, UNSPEC_PCMPESTR, UNSPEC_PCMPISTR,
840 UNSPEC_FMADDSUB, UNSPEC_XOP_UNSIGNED_CMP, UNSPEC_XOP_TRUEFALSE,
841 UNSPEC_XOP_PERMUTE, UNSPEC_FRCZ, UNSPEC_AESENC, UNSPEC_AESENCLAST,
842 UNSPEC_AESDEC, UNSPEC_AESDECLAST, UNSPEC_AESIMC,
843 UNSPEC_AESKEYGENASSIST, UNSPEC_PCLMUL, UNSPEC_PCMP, UNSPEC_VPERMIL,
844 UNSPEC_VPERMIL2, UNSPEC_VPERMIL2F128, UNSPEC_CAST, UNSPEC_VTESTP,
845 UNSPEC_VCVTPH2PS, UNSPEC_VCVTPS2PH, UNSPEC_VPERMSI, UNSPEC_VPERMDF,
846 UNSPEC_VPERMSF, UNSPEC_VPERMTI, UNSPEC_GATHER, UNSPEC_VSIBADDR>: Ditto.
847 (unspecv) <UNSPECV_LDMXCSR, UNSPECV_STMXCSR, UNSPECV_CLFLUSH,
848 UNSPECV_MONITOR, UNSPECV_MWAIT, UNSPECV_VZEROALL, UNSPECV_VZEROUPPER>:
850 * config/i386/sync.md (unspec) <UNSPEC_LFENCE, UNSPEC_SFENCE,
851 UNSPEC_MFENCE, UNSPEC_MOVA>: Ditto.
852 (unspecv) <UNSPECV_CMPXCHG_1, UNSPECV_CMPXCHG_2, UNSPECV_CMPXCHG_3,
853 UNSPECV_CMPXCHG_4, UNSPECV_XCHG, UNSPECV_LOCK>: Ditto.
854 (sse2_lfence): Move from config/i386/sse.md.
855 (*sse2_lfence): Ditto.
857 (*sse_sfence): Ditto.
858 (sse2_mfence): Ditto.
859 (mfence_sse2): Ditto. Rename from *sse2_mfence. Enable also
861 (mem_thread_fence): Use mfence_sse2.
863 2011-11-21 Georg-Johann Lay <avr@gjlay.de>
865 * config/avr/avr.h (struct base_arch_s): Add field sfr_offset.
866 * config/avr/avr-devices.c: Ditto. And initialize it.
867 * config/avr/avr-c.c (avr_cpu_cpp_builtins): New built-in define
869 * config/avr/avr-protos.h (out_movqi_r_mr, out_movqi_mr_r): Remove.
870 (out_movhi_r_mr, out_movhi_mr_r): Remove.
871 (out_movsi_r_mr, out_movsi_mr_r): Remove.
872 * config/avr/avr.md (*cbi, *sbi): Use %i instead of %m-0x20.
873 (*insv.io, *insv.not.io): Ditto.
874 * config/avr/avr.c (out_movsi_r_mr, out_movsi_mr_r): Make static.
875 (print_operand): Implement "%i" to print address as I/O address.
876 (output_movqi): Clean up call of out_movqi_mr_r.
877 (output_movhi): Clean up call of out_movhi_mr_r.
878 (avr_file_start): Use avr_current_arch->sfr_offset instead of
879 magic -0x20. Use TMP_REGNO, ZERO_REGNO instead of 0, 1.
880 (avr_out_sbxx_branch): Use %i instead of %m-0x20.
881 (out_movqi_r_mr, out_movqi_mr_r): Ditto. And make static.
882 (out_movhi_r_mr, out_movhi_mr_r): Ditto. And use avr_asm_len.
883 (out_shift_with_cnt): Clean up code: Use avr_asm_len.
884 (output_movsisf): Use output_reload_insisf for all CONSTANT_P sources.
885 (avr_out_movpsi): USE avr_out_reload_inpsi for all CONSTANT_P sources.
886 Clean up call of avr_out_store_psi.
887 (output_reload_in_const): Don't cut symbols longer than 2 bytes.
888 (output_reload_insisf): Filter CONST_INT_P or CONST_DOUBLE_P to
889 try if setting pre-cleared register is advantageous.
890 (avr_out_plus_1): Use gen_int_mode instead of GEN_INT.
892 2011-11-20 Joey Ye <joey.ye@arm.com>
894 * expr.c (expand_expr_real_1): Correctly handle strict volatile
895 bitfield loads smaller than mode size.
897 2011-11-20 Richard Henderson <rth@redhat.com>
899 * config/mips/mips.c (mips_init_libfuncs): Call init_sync_libfuncs.
901 2011-11-20 Uros Bizjak <ubizjak@gmail.com>
904 * config/i386/i386.c (ix86_expand_vcond): Generate TARGET_XOP
905 patterns for supported mode only.
908 * config/i386/i386.c (ix86_expand_builtin)
909 <IX86_BUILTIN_GATHERALTSIV4DI>: Use CODE_FOR_avx2_gathersiv4di.
911 2011-11-20 Uros Bizjak <ubizjak@gmail.com>
913 * config/i386/i386.md (UNSPEC_MOVNTI): Remove.
914 (UNSPEC_MOVNTQ): New unspec.
915 * config/i386/mmx.md (sse_movntq): Rename from sse_movntdi.
916 Use UNSPEC_MOVNTQ instead of UNSPEC_MOVNT.
917 * config/i386/sse.md (sse2_movnti<mode>): Use UNSPEC_MOVNT instead of
919 (STORENT_MODE): Add DI and V4DI modes.
921 2011-11-20 Nathan Sidwell <nathan@acm.org>
923 PR gcov-profile/51113
924 * coverage.c (build_var): Propagate visibility for public
925 decls. Use make_decl_one_only for PIC correctness.
928 * gcc.misc-tests/gcov-13.c: Check gcovpart-13b coverage
929 * gcc.misc-tests/gcov-16.c: New.
930 * gcc.misc-tests/gcov-17.c: New.
931 * g++.dg/gcov/gcov-8.C: New.
932 * g++.dg/gcov/gcov-9.C: New.
933 * g++.dg/gcov/gcov-10.C: New.
935 2011-11-19 Eric Botcazou <ebotcazou@adacore.com>
937 PR rtl-optimization/51187
938 * reorg.c (relax_delay_slots): Do not consider a jump useless if there
939 is a barrier between the jump and its target label.
941 2011-11-19 Patrick Marlier <patrick.marlier@gmail.com>
944 * tracer.c (ignore_bb_p): Don't copy GIMPLE_TRANSACTION.
946 2011-11-19 Eric Botcazou <ebotcazou@adacore.com>
948 * expmed.c (store_bit_field_1): Revert bogus formatting change.
950 2011-11-19 Richard Earnshaw <rearnsha@arm.com>
953 * config/arm/arm.c (neon_disambiguate_copy): Correctly handle partial
954 overlap of src and dest operands.
956 2011-11-19 Iain Sandoe <iains@gcc.gnu.org>
958 * config/darwin.h (ASM_DEBUG_SPEC): New.
959 * config/darwin9.h (ASM_DEBUG_SPEC): New.
961 2011-11-18 Jan Hubicka <jh@suse.cz>
963 * config/i386/i386.c (atom_cost): Disable SSE loop until alignment
966 2011-11-18 Anatoly Sokolov <aesok@post.ru>
968 * config/cris/cris.c (cris_function_value_regno_p): Make static.
969 (TARGET_FUNCTION_VALUE_REGNO_P): Define.
970 * config/cris/cris.h (FUNCTION_VALUE_REGNO_P): Remove.
971 * config/cris/cris-protos.h (cris_function_value_regno_p): Remove.
973 2011-11-18 H.J. Lu <hongjiu.lu@intel.com>
976 * doc/extend.texi: Document __builtin_ia32_movnti64.
978 * config/i386/emmintrin.h (_mm_stream_si64): New.
980 * config/i386/i386-builtin-types.def: Add VOID_FTYPE_PLONGLONG_LONGLONG.
982 * config/i386/i386.c (ix86_builtins): Add IX86_BUILTIN_MOVNTI64.
983 (bdesc_special_args): Update __builtin_ia32_movnti. Add
984 __builtin_ia32_movnti64.
985 (ix86_expand_special_args_builtin): Handle
986 VOID_FTYPE_PLONGLONG_LONGLONG.
988 * config/i386/i386.md (UNSPEC_MOVNTI): New.
990 * config/i386/sse.md (sse2_movntsi): Renamed to ...
991 (sse2_movnti<mode>): This.
993 2011-11-18 Georg-Johann Lay <avr@gjlay.de>
996 * config/avr/avr.h (base_arch_s): Add field n_segments.
997 (ADDR_SPACE_PGM1, ADDR_SPACE_PGM2, ADDR_SPACE_PGM3,
998 ADDR_SPACE_PGM4, ADDR_SPACE_PGM5, ADDR_SPACE_PGMX): New address spaces.
999 (AVR_HAVE_ELPM, AVR_HAVE_ELPMX): New defines.
1000 (INIT_EXPANDERS): New define.
1001 * config/avr/avr-protos.h (avr_mem_pgmx_p): New.
1002 (avr_init_expanders): New.
1003 (avr_emit_movmemhi, avr_out_movmem): New.
1004 (avr_xload_libgcc_p): New.
1005 * config/avr/avr-c.c (avr_register_target_pragmas): Register
1006 address spaces __pgm1, __pgm2, __pgm3, __pgm4 __pgm5, __pgmx.
1007 (avr_cpu_cpp_builtins): Add built-in defines __PGM1,
1008 __PGM2, __PGM3, __PGM4, __PGM5, __PGMX.
1009 * config/avr/avr-devices.c (avr_arch_types): Set field n_segments.
1011 * config/avr/avr.c (AVR_SECTION_PROGMEM): Change define to cover
1012 3 bits instead of just 1.
1013 (xstring_empty, xstring_e, rampz_rtx): New static GTYed variables.
1014 (progmem_section): Change from section to array of sections.
1015 (progmem_section_prefix): New static variable.
1016 (avr_file_start): Print set for __RAMPZ__
1017 (avr_option_override): Move initialization of RTXes from here...
1018 (avr_init_expanders): ...to this new function.
1019 (avr_pgm_segment): New static function.
1020 (avr_decl_pgm_p): Handle error_mark_node.
1021 (avr_mem_pgmx_p, avr_decl_pgmx_p): New static functions.
1022 (avr_out_xload, avr_find_unused_d_reg): New static functions.
1023 (expand_prologue, expand_epilogue): Use rampz_rtx.
1024 (print_operand): Hande CONST_STRING.
1025 (avr_xload_libgcc_p): New static function.
1026 (avr_out_lpm_no_lpmx, avr_out_lpm): Handle ELPM.
1027 (avr_progmem_p): Return 2 for 24-bit flash address space.
1028 (avr_out_sbxx_branch): Clean-up code from ASn macros.
1029 (out_movqi_r_mr, out_movqi_mr_r): Ditto. And recognize RAMPZ's
1030 address and print symbolically.
1031 (avr_asm_named_section, avr_section_type_flags,
1032 avr_encode_section_info, avr_asm_select_section,
1033 avr_addr_space_address_mode, avr_addr_space_pointer_mode,
1034 avr_addr_space_legitimate_address_p, avr_addr_space_convert,
1035 avr_addr_space_legitimize_address): Handle new address spaces.
1036 (avr_output_progmem_section_asm_op): New static function.
1037 (avr_asm_init_sections): Initialize progmem_section[].
1038 (adjust_insn_length): Handle ADJUST_LEN_XLOAD, ADJUST_LEN_MOVMEM.
1039 (avr_const_address_lo16): New static function.
1040 (avr_assemble_integer): Use it to handle 3-byte integers.
1041 (avr_emit_movmemhi, avr_out_movmem): New functions.
1043 * config/avr/predicates.md (nox_general_operand): Handle new
1045 * config/avr/avr.md (unspec): Add UNSPEC_MOVMEM.
1046 (adjust_len): Add xload, movmem.
1047 (SP_ADDR): New define_constants.
1048 (isa): Add "lpm", "lpmx", "elpm", "elpmx".
1049 (enabled): Handle them.
1050 (load<mode>_libgcc): New expander.
1051 (*load.<mode>.libgcc): Rename to load_<mode>_libgcc.
1052 (xload8_A, xload<mode>_A): New insn-and-splits.
1053 (xload_8, xload_<mode>_libgcc, xload_<mode>, loadmem_elpm): New insns.
1054 (mov<mode>): Handle new address spaces.
1055 (movmemhi): Rewrite using avr_emit_movmemhi.
1056 (MOVMEM_r_d): New mode attribute.
1057 (movmem_<mode>, movmem_qi_elpm): New insns.
1058 (setmemhi, *clrmemqi, *clrmemhi, strlenhi, *strlenhi): Unquote
1059 C-code. Use label instead of hard-coded instrunction lengths.
1061 2011-11-18 Martin Jambor <mjambor@suse.cz>
1063 PR tree-optimization/50605
1064 * gimple.c (is_gimple_ip_invariant_address): Also handle MEM_REFs
1065 of IPA invariant decls.
1067 2011-11-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1069 * tree-outof-ssa.c (insert_back_edge_copies): Add call to
1070 mark_dfs_back_edges.
1072 2011-11-18 Iain Sandoe <iains@gcc.gnu.org>
1075 * configure.ac: Remove ranlib special-casing for Darwin.
1076 * configure: Regenerate.
1078 2011-11-18 Uros Bizjak <ubizjak@gmail.com>
1080 PR tree-optimization/51118
1081 * fold-const.c (fold_checksum_tree): Check for TS_TYPED structure
1082 before using TREE_TYPE accessor on expr.
1084 2011-11-17 Jan Hubicka <jh@suse.cz>
1087 * i386.c (atom_cost): Fix 32bit memset description.
1088 (expand_set_or_movmem_via_loop_with_iter): Output proper bounds check
1090 (expand_movmem_epilogue): Handle epilogues up to size 15 w/o producing
1092 (decide_alg): sse_loop is not useable wthen SSE2 is disabled;
1093 when not optimizing always use rep movsb or lincall; do not produce
1094 word sized loops when optimizing memset for size (to avoid need
1095 for large constants).
1096 (ix86_expand_movmem): Get into sync with ix86_expand_setmem;
1097 choose unroll factors better; always do 128bit moves when producing
1098 SSE loops; do not produce loopy epilogue when size is too small.
1099 (promote_duplicated_reg_to_size): Do not look into desired alignments
1100 when doing vector expansion.
1101 (ix86_expand_setmem): Track better when promoted value is available;
1102 choose unroll factors more sanely. Output loopy epilogue only
1105 2011-11-17 Steve Ellcey <sje@cup.hp.com>
1108 * output.h (fprint_w): Remove.
1109 * final.c (fprint_w): Remove.
1110 (output_addr_const): Change fprint_w back to fprintf.
1112 2011-11-17 Andrew Pinski <apinski@cavium.com>
1114 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for MIPS.
1116 2011-11-17 Andrew MacLeod <amacleod@redhat.com>
1118 * builtins.c (expand_builtin): Remove 4th parameter representing
1119 weak/strong mode when __atomic_compare_exchange becomes a library call.
1121 2011-11-17 Richard Henderson <rth@redhat.com>
1123 * config/rs6000/rs6000.c (rs6000_expand_atomic_compare_and_swap): Get
1124 new pseudo for target after convert_modes.
1125 (rs6000_expand_atomic_exchange, rs6000_expand_atomic_op): Likewise.
1127 2011-11-17 Richard Henderson <rth@redhat.com>
1129 * config/rs6000/rs6000.c (rs6000_expand_atomic_compare_and_swap): Get
1130 new pseudo for target after convert_modes.
1131 (rs6000_expand_atomic_exchange, rs6000_expand_atomic_op): Likewise.
1133 2011-11-17 Richard Henderson <rth@redhat.com>
1135 * builtins.c (expand_builtin_mem_thread_fence): Remove.
1136 (expand_builtin_mem_signal_fence): Remove.
1137 (expand_builtin_atomic_thread_fence): Use expand_mem_thread_fence.
1138 (expand_builtin_sync_synchronize): Likewise.
1139 (expand_builtin_atomic_signal_fence): Use expand_mem_signal_fence.
1140 * optabs.c (expand_asm_memory_barrier): Split out from
1141 expand_builtin_mem_signal_fence.
1142 (expand_mem_thread_fence): New, a combination of code from
1143 expand_builtin_mem_thread_fence and expand_builtin_sync_synchronize.
1144 (expand_mem_signal_fence): Moved and renamed from
1145 expand_builtin_mem_signal_fence.
1146 (expand_atomic_exchange): Use expand_mem_thread_fence.
1147 (expand_atomic_load, expand_atomic_store): Likewise.
1148 * expr.h, optabs.h: Update decls.
1150 2011-11-17 Bin Cheng <bin.cheng@arm.com>
1152 PR rtl-optimization/50663
1153 * cprop.c (implicit_set_indexes): New global variable.
1154 (insert_set_in_table): Add additional parameter and record implicit
1156 (hash_scan_set): Add additional parameter and pass it to above.
1157 (hash_scan_insn): Pass false to hash_scan_set.
1158 (compute_hash_table_work): Pass true to hash_scan_set.
1159 (compute_cprop_data): Add implicit set to AVIN of block which the
1160 implicit set is recorded for.
1161 (one_cprop_pass): Handle implicit_set_indexes array.
1163 2011-11-17 Michael Matz <matz@suse.de>
1167 * tree-ssa-live.c (mark_all_vars_used_1): Recurse only for decls of
1169 (remove_unused_locals): Ditto.
1171 2011-11-16 Richard Henderson <rth@redhat.com>
1173 * config/alpha/alpha.c (alpha_split_compare_and_swap_12): Copy
1174 ALIAS_SET_MEMORY_BARRIER when creating a new memory.
1175 (alpha_split_atomic_exchange_12): Likewise.
1176 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Likewise.
1178 2011-11-16 Joseph Myers <joseph@codesourcery.com>
1180 * config/i386/i386elf.h (PREFERRED_DEBUGGING_TYPE): Remove.
1182 2011-11-16 David Edelsohn <dje.gcc@gmail.com>
1184 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags):
1185 Default to minimum alignment if decl is NULL.
1187 2011-11-16 Uros Bizjak <ubizjak@gmail.com>
1189 * config/i386/sse.md (round<mode>2_vec_pack_sfix): Optimize V2DFmode
1191 (<sse4_1>_round<ssemodesuffix>_vec_pack_sfix<avxsizesuffix>): Ditto.
1193 2011-11-16 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
1195 * doc/invoke.texi: Document AMD bdver1 and btver1.
1197 2011-11-16 Richard Earnshaw <rearnsha@arm.com>
1198 Bernd Schmidt <bernds@coudesourcery.com>
1199 Sebastian Huber <sebastian.huber@embedded-brains.de>
1202 * config/arm/arm.c (store_multiple_sequence): Avoid cases where
1203 the base reg is stored iff compiling for Thumb1.
1205 2011-11-16 Razya Ladelsky <razya@il.ibm.com>
1207 PR tree-optimization/49960
1208 * tree-data-ref.c (initialize_data_dependence_relation): Add
1210 Remove call to compute_self_dependence.
1211 (compute_affine_dependence): Remove the !DDR_SELF_REFERENCE condition.
1212 (compute_self_dependence): Remove old code. Add call to
1213 compute_affine_dependence.
1214 (compute_all_dependences): Remove call to compute_self_dependence.
1215 Add call to compute_affine_dependence.
1217 2011-11-16 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1220 * expmed.c (store_bit_field_1): Use extract_bit_field on big
1221 endian targets if the source cannot be exactly covered by word
1224 2011-11-15 Joseph Myers <joseph@codesourcery.com>
1226 * config/i386/i386elf.h (ASM_OUTPUT_ASCII): Change STRING_LIMIT to
1229 2011-11-15 Richard Henderson <rth@redhat.com>
1231 * config/alpha/alpha.c (alpha_pre_atomic_barrier): New.
1232 (alpha_post_atomic_barrier): New.
1233 (alpha_split_atomic_op): New memmodel argument; honor it.
1234 (alpha_split_compare_and_swap): Take array of operands. Honor
1235 memmodel; always set bool output
1236 (alpha_expand_compare_and_swap_12): Similarly.
1237 (alpha_split_compare_and_swap_12): Similarly.
1238 (alpha_split_atomic_exchange): Similarly. Rename from
1239 alpha_split_lock_test_and_set.
1240 (alpha_expand_atomic_exchange_12): Similarly. Rename from
1241 alpha_expand_lock_test_and_set_12.
1242 (alpha_split_atomic_exchange_12): Similarly. Rename from
1243 alpha_split_lock_test_and_set_12.
1244 * config/alpha/alpha-protos.h: Update.
1245 * config/alpha/alpha.md (UNSPECV_CMPXCHG): New.
1246 * config/alpha/constraints.md ("w"): New.
1247 * config/alpha/predicates.md (mem_noofs_operand): New.
1248 * config/alpha/sync.md (atomic_compare_and_swap<mode>): Rename from
1249 sync_compare_and_swap<mode>; add the new parameters.
1250 (atomic_exchange<mode>): Update from sync_test_and_set<mode>.
1251 (atomic_fetch_<op><mode>): Update from sync_old_<op><mode>.
1252 (atomic_<op>_fetch<mode>): Update from sync_new_<op><mode>.
1253 (atomic_<op><mode>): Update from sync_<op><mode>.
1255 2011-11-16 Tom de Vries <tom@codesourcery.com>
1257 * tree-ssa-tail-merge.c (replace_block_by): Add frequency of bb2 to bb1.
1259 2011-11-16 Alan Modra <amodra@gmail.com>
1261 PR rtl-optimization/51051
1263 * function.c (thread_prologue_and_epilogue_insns): Guard
1264 emitting return with single_succ_p test.
1266 2011-11-15 Joern Rennecke <joern.rennecke@embecosm.com>
1268 * config/epiphany/epiphany.md (mov<mode>cc): Fix code to
1269 get mode from CMP_OP1 if CMP_OP0 is VOIDmode.
1271 2011-11-15 Uros Bizjak <ubizjak@gmail.com>
1273 * config/i386/sse.md (vec_pack_trunc_v2df): Optimize sequence for AVX.
1274 (vec_pack_sfix_trunc_v2df): Ditto.
1275 (vec_pack_sfix_v2df): Ditto.
1276 (vec_pack_sfix_trunc_v4df): Generate fix_truncv4dfv4si2 and
1277 avx_vec_concatv8si patterns.
1278 (vec_pack_sfix_v4df): Generate avx_cvtpd2dq256 and
1279 avx_vec_concatv8si patterns.
1281 2011-11-15 Torvald Riegel <triegel@redhat.com>
1283 * c-parser.c (c_parser_transaction_expression): Require parentheses
1284 when parsing transaction expressions.
1286 2011-11-15 Tristan Gingold <gingold@adacore.com>
1288 * incpath.c (get_added_cpp_dirs): New function.
1289 * incpath.h (get_added_cpp_dirs): Declare.
1290 * config/vms/vms-c.c (vms_c_register_includes): New function.
1291 (vms_std_modules): New variable.
1292 * config/vms/vms.h (TARGET_EXTRA_INCLUDES): Define.
1293 (vms_c_register_includes): Declare.
1295 2011-11-15 Tristan Gingold <gingold@adacore.com>
1297 * c-family/c-pragma.h (pragma_extern_prefix): Declare.
1298 * c-family/c-pragma.c (pragma_extern_prefix): Make it public.
1299 * config/vms/vms-c.c (saved_extern_prefix): New variable.
1300 (vms_pragma_extern_prefix): New function.
1301 (vms_c_register_pragma): Register vms_pragma_extern_prefix.
1303 2011-11-15 Georg-Johann Lay <avr@gjlay.de>
1306 * config/avr/avr.h (ADDR_SPACE_PGM): New address spaces.
1307 (REGISTER_TARGET_PRAGMAS): New define.
1308 * config/avr/avr-protos.h (avr_mem_pgm_p): New.
1309 (avr_load_libgcc_p): New.
1310 (avr_register_target_pragmas): New.
1311 (asm_output_external_libcall): Remove.
1312 (avr_log_t): Add field "progmem". Order alphabetically.
1313 * config/avr/avr-log.c (avr_log_set_avr_log): Set avr_log.progmem.
1314 * config/avr/avr-c.c (langhooks.h): New include.
1315 (avr_register_target_pragmas): New function. Register address
1317 (avr_cpu_cpp_builtins): Add built-in define __PGM.
1319 * config/avr/avr.c: Include "c-family/c-common.h".
1320 (TARGET_LEGITIMATE_ADDRESS_P): Remove define.
1321 (TARGET_LEGITIMIZE_ADDRESS): Remove define.
1322 (TARGET_ADDR_SPACE_SUBSET_P): Define to...
1323 (avr_addr_space_subset_p): ...this new static function.
1324 (TARGET_ADDR_SPACE_CONVERT): Define to...
1325 (avr_addr_space_convert): ...this new static function.
1326 (TARGET_ADDR_SPACE_ADDRESS_MODE): Define to...
1327 (avr_addr_space_address_mode): ...this new static function.
1328 (TARGET_ADDR_SPACE_POINTER_MODE): Define to...
1329 (avr_addr_space_pointer_mode): ...this new static function.
1330 (TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P): Define to...
1331 (avr_addr_space_legitimate_address_p): ...this new static function.
1332 (TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS): Define to...
1333 (avr_addr_space_legitimize_address): ...this new static function.
1334 (avr_mode_code_base_reg_class): Handle address spaces.
1335 (avr_regno_mode_code_ok_for_base_p): Ditto.
1336 (lpm_addr_reg_rtx, lpm_reg_rtx, all_regs_rtx): New static variables.
1337 (avr_option_override): Initialize them.
1338 (output_reload_in_const): Use all_regs_rtx. Fix signedness of loop
1340 (avr_pgm_segment): New static function.
1341 (avr_decl_pgm_p, avr_mem_pgm_p): New static functions.
1342 (avr_out_lpm, avr_out_lpm_no_lpmx): New static functions.
1343 (output_movqi, output_movhi, output_movsisf, avr_out_movpsi): Call
1344 avr_out_lpm to handle loads from progmem.
1345 (avr_load_libgcc_p): New static function.
1346 (avr_progmem_p): Test if decl is in flash.
1347 (avr_pgm_pointer_const_p): New static function.
1348 (avr_nonconst_pointer_addrspace): New static function.
1349 (avr_pgm_check_var_decl): New static function.
1350 (avr_insert_attributes): Use it. Change error message to report
1351 cause (progmem or address space) when code wants to write to flash.
1352 (avr_section_type_flags): Unset section flag SECTION_BSS for
1355 * config/avr/predicates.md (nop_general_operand): New predicate.
1356 (nox_general_operand): New predicate.
1357 * config/avr/avr.md (LPM_REGNO): New define_constant.
1358 (load<mode>_libgcc): New expander.
1359 (*load.<mode>.libgcc): New insn.
1360 (mov<mode>): Handle loads from non-generic AS.
1361 (movmemhi): Ditto. Propagate address space information to newly
1363 (movqi_insn, *movhi, *movpsi, *movsi, *movsf): Change predicate #1
1364 to nox_general_operand.
1365 (ashrqi3, ashrhi3, ashrsi3): Change predicate #1 to
1366 nop_general_operand.
1367 (ashlqi3, *ashlqi3, ashlhi3, ashlsi3): Ditto.
1368 (lshrqi3, *lshrqi3, lshrhi3, lshrsi3): Ditto.
1369 (split-lpmx): New split.
1370 (*ashlhi3_const, *ashlsi3_const, *ashrhi3_const, *ashrsi3_const,
1371 *lshrhi3_const, *lshrsi3_const): Indent, unquote C.
1373 2011-11-15 Maxim Kuvyrkov <maxim@codesourcery.com>
1375 * ipa-cp.c (ipa_value_from_jfunc): Make global.
1376 (ipa_cst_from_jfunc): Remove, use ipa_value_from_jfunc instead.
1377 (get_indirect_edge_target): Rename, make global.
1378 (devirtualization_time_bonus, estimate_local_effects,)
1379 (ipcp_discover_new_direct_edges): Update.
1380 * ipa-inline-analysis.c (evaluate_conditions_for_edge):
1381 Generalize to also handle types. Rename to ...
1382 (evaluate_properties_for_edge): Use instead of
1383 evaluate_conditions_for_edge.
1384 (estimate_edge_devirt_benefit): New function.
1385 (estimate_calls_size_and_time): Use it.
1386 (estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,)
1387 (inline_merge_summary): Update.
1388 (do_estimate_edge_time, do_estimate_edge_growth): Update. Calculate
1389 parameter information at the call site and pass it on to subroutines.
1390 * tree-inline.c (estimate_num_insns): Distinguish between direct and
1392 (init_inline_once): Set size and time costs or indirect calls.
1393 * tree-inline.h (eni_weights): Add indirect_call_cost.
1395 2011-11-15 Tom de Vries <tom@codesourcery.com>
1397 PR tree-optimization/51005
1398 * tree-ssa-tail-merge.c (delete_basic_block_same_succ): Rename to
1399 mark_basic_block_deleted.
1400 (update_worklist): Inline purge_bbs.
1401 (purge_bbs, unlink_virtual_phi, update_vuses, vop_at_entry)
1402 (delete_block_update_dominator_info): Remove.
1403 (replace_block_by): Remove update_vops parameter. Partially evaluate
1404 for update_vops == false.
1405 (apply_clusters): Remove update_vops parameter. Remove update_vops
1406 argument in replace_block_by call.
1407 (update_debug_stmts): Remove MAY_HAVE_DEBUG_STMTS test.
1408 (tail_merge_optimize): Remove update_vops argument to apply_clusters.
1409 Remove call to purge_bbs. Add calls to calculate_dominance_info and
1410 free_dominance_info. Add MAY_HAVE_DEBUG_STMTS before calling
1411 update_debug_stmts. Mark vop var for renaming, if necessary.
1413 2011-11-15 Bernd Schmidt <bernds@codesourcery.com>
1415 PR rtl-optimization/51051
1416 * cfgrtl.c (cfg_layout_can_merge_blocks_p): Return FALSE if the
1417 move would cause fallthrough into the exit block.
1419 2011-11-14 Richard Henderson <rth@redhat.com>
1421 * config/rs6000/rs6000.c (emit_load_locked): Assert the mode is
1423 (emit_store_conditional): Likewise.
1424 (rs6000_pre_atomic_barrier, rs6000_post_atomic_barrier): New.
1425 (rs6000_adjust_atomic_subword): New.
1426 (rs6000_mask_atomic_subword, rs6000_finish_atomic_subword): New.
1427 (rs6000_expand_atomic_op): Rename from rs6000_emit_sync; rewrite
1428 for pre-reload expansion.
1429 (rs6000_split_compare_and_swap, rs6000_split_compare_and_swapqhi,
1430 rs6000_expand_compare_and_swapqhi): Merge into ...
1431 (rs6000_expand_atomic_compare_and_swap): ... here. New function.
1432 rs6000_split_lock_test_and_set; expand immediately. Handle
1434 * config/rs6000/rs6000.md (UNSPEC_LWSYNC): Move and rename
1435 from UNSPECV_LWSYNC.
1436 * config/rs6000/sync.md (fetchopsi_constr, fetchopdi_constr): Remove.
1437 (mem_thread_fence): New.
1438 (hwsync): Rename from memory_barrier.
1439 (*hwsync): Rename from *sync_internal.
1440 (lwsync, *lwsync): Mirror hwsync implementation.
1441 (isync): Don't reference memory.
1443 (atomic_load<INT>, atomic_store<INT>): New.
1444 (ATOMIC): New mode iterator.
1445 (load_locked<ATOMIC>): Rename from load_locked_<GPR>.
1446 (store_conditional<ATOMIC>): Rename from store_conditional_<GPR>.
1447 (sync_compare_and_swap<GPR>): Remove.
1448 (sync_compare_and_swaphi, sync_compare_and_swapqi): Remove.
1449 (sync_compare_and_swapqhi_internal): Remove.
1450 (sync_lock_test_and_set<GPR>): Remove.
1451 (sync_<FETCHOP><INT1>): Remove.
1452 (*sync_<FETCHOP>si_internal, *sync_<FETCHOP>di_internal): Remove.
1453 (sync_nand<INT1>, *sync_nand<GPR>_internal): Remove.
1454 (sync_old_<FETCHOP><GPR>): Remove.
1455 (*sync_old_<FETCHOP>si_internal): Remove.
1456 (*sync_old_<FETCHOP>di_internal): Remove.
1457 (sync_old_nand<INT1>): Remove.
1458 (*sync_old_nand<GPR>_internal): Remove.
1459 (sync_new_<FETCHOP><GPR>): Remove.
1460 (*sync_new_<FETCHOP>si_internal): Remove.
1461 (*sync_new_<FETCHOP>di_internal): Remove.
1462 (sync_new_nand<INT1>): Remove.
1463 (*sync_new_nand<GPR>_internal): Remove.
1464 (*atomic_andsi, *atomic_anddi): Remove.
1465 (*sync_addshort_internal, *sync_subshort_internal): Remove.
1466 (*sync_andsi_internal, *sync_boolsi_internal): Remove.
1467 (*sync_boolcshort_internal): Remove.
1468 (sync_lock_release<INT1>): Remove.
1469 (atomic_compare_and_swap<INT1>): New.
1470 (atomic_exchange<INT1>): New.
1471 (atomic_<FETCHOP><INT1>, atomic_nand<INT1>): New.
1472 (atomic_fetch_<FETCHOP><INT1>, atomic_fetch_nand<INT1>): New.
1473 (atomic_<FETCHOP>_fetch<INT1>, atomic_nand_fetch<INT1>): New.
1475 2011-11-14 Uros Bizjak <ubizjak@gmail.com>
1477 * config/i386/sse.md (round<mode>2): Use register_operand for
1478 operand 1 predicate.
1479 (round<mode>2_sfix): Ditto.
1480 (round<mode>2_vec_pack_sfix): Ditto for operand 1 and operand 2.
1482 2011-11-14 Nathan Sidwell <nathan@acm.org>
1484 * gcov.c (canonicalize_name): Protect use of S_ISLNK.
1486 2011-11-14 Jan Hubicka <jh@suse.cz>
1489 * config/i386/i386.c (core cost model): Correct pasto.
1491 2011-11-14 Uros Bizjak <ubizjak@gmail.com>
1493 * config/i386/sse.md (round<mode>2_sfix): New expander.
1494 (round<mode>2_vec_pack_sfix): Ditto.
1495 (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
1496 (<sse4_1>_round<ssemodesuffix>_vec_pack_sfix<avxsizesuffix>): Ditto.
1497 * config/i386/builtin-types.def (V4SI_FTYPE_V4SF_ROUND,
1498 V8SI_FTYPE_V8SF_ROUND, V4SI_FTYPE_V2DF_V2DF_ROUND,
1499 V8SI_FTYPE_V4DF_V4DF_ROUND): New builtin types.
1500 * config/i386/i386.c (ix86_builtins): Add
1501 IX86_BUILTIN_{FLOORPD,CEILPD,ROUNDPD_AZ}_VEC_PACK_SFIX{,256} and
1502 IX86_BUILTIN_{FLOORPS,CEILPS,ROUNDPS_AZ}_SFIX{,256} defines.
1503 (bdesc_args): Add __builtin_ia32_{floorpd,ceilpd}_vec_pack_sfix{,256},
1504 __builtin_ia32_roundpd_az_vec_pack_sfix{,256},
1505 __builtin_ia32_{floorps,ceilps}_sfix{,256}and
1506 __builtin_ia32_roundps_az_sfix{,256} descriptions.
1507 (ix86_expand_sse_round_vec_pack_sfix): New.
1508 (ix86_expand_args_builtin): Handle V4SI_FTYPE_V4SF_ROUND,
1509 V8SI_FTYPE_V8SF_ROUND, V4SI_FTYPE_V2DF_V2DF_ROUND and
1510 V8SI_FTYPE_V4DF_V4DF_ROUND types. Check last argument of
1511 CODE_FOR_sse4_1_roundpd_vec_pack_sfix, CODE_FOR_sse4_1_roundps_sfix,
1512 CODE_FOR_avx_roundpd_vec_pack_sfix256 and CODE_FOR_avx_roundps_sfix256.
1513 (ix86_builtin_vectorized_function): Handle
1514 BUILT_IN_{I,L,LL}FLOOR{,F}, BUILT_IN_{I,L,LL}CEIL{,F} and
1515 BUILT_IN_{I,L,LL}ROUND{,F}
1517 2011-11-14 Jan Hubicka <jh@suse.cz>
1520 * cgraph.h (varpool_externally_visible_p): Declare.
1521 (varpool_add_new_variable): Declare.
1522 * tree-emultls.c (get_emutls_init_templ_addr,
1523 new_emutls_decl): Use varpool_add_new_variable.
1524 * ipa.c (varpool_externally_visible_p): Export.
1525 * varpool.c (varpool_add_new_variable): New function.
1527 2011-11-14 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
1528 Jan Hubicka <jh@suse.cz>
1530 * config/i386/i386.h (processor_costs): Add second dimension to
1531 stringop_algs array.
1532 * config/i386/i386.c (cost models): Initialize second dimension of
1533 stringop_algs arrays.
1534 (core_cost): New costs based on generic64 costs with updated stringop
1536 (promote_duplicated_reg): Add support for vector modes, add
1538 (promote_duplicated_reg_to_size): Likewise.
1539 (processor_target): Set core costs for core variants.
1540 (expand_set_or_movmem_via_loop_with_iter): New function.
1541 (expand_set_or_movmem_via_loop): Enable reuse of the same iters in
1542 different loops, produced by this function.
1543 (emit_strset): New function.
1544 (expand_movmem_epilogue): Add epilogue generation for bigger sizes,
1545 use SSE-moves where possible.
1546 (expand_setmem_epilogue): Likewise.
1547 (expand_movmem_prologue): Likewise for prologue.
1548 (expand_setmem_prologue): Likewise.
1549 (expand_constant_movmem_prologue): Likewise.
1550 (expand_constant_setmem_prologue): Likewise.
1551 (decide_alg): Add new argument align_unknown. Fix algorithm of
1552 strategy selection if TARGET_INLINE_ALL_STRINGOPS is set; Skip sse_loop
1553 (decide_alignment): Update desired alignment according to chosen move
1555 (ix86_expand_movmem): Change unrolled_loop strategy to use SSE-moves.
1556 (ix86_expand_setmem): Likewise.
1557 (ix86_slow_unaligned_access): Implementation of new hook
1558 slow_unaligned_access.
1559 * config/i386/i386.md (strset): Enable half-SSE moves.
1560 * config/i386/sse.md (vec_dupv4si): Add expand for vec_dupv4si.
1561 (vec_dupv2di): Add expand for vec_dupv2di.
1563 2011-11-14 Dimitrios Apostolou <jimis@gmx.net>
1567 * config/elfos.h (TARGET_ASM_INTERNAL_LABEL): Remove.
1568 * varasm.c (default_elf_internal_label): Remove.
1570 2011-11-14 Kai Tietz <ktietz@redhat.com>
1572 * gcov.c (generate_results): Add missing semicolon and correct indent.
1574 2011-11-14 Ira Rosen <ira.rosen@linaro.org>
1577 * tree-vect-stmts.c (vectorizable_condition): Initialize comp_vectype.
1579 2011-11-14 Mingjie Xing <mingjie.xing@gmail.com>
1581 * doc/invoke.texi (Wunused-result): Change @pxref{Variable Attributes}
1582 to @pxref{Function Attributes}.
1584 2011-11-13 Oleg Endo <oleg.endo@t-online.de>
1587 * config/sh/sh.h (IS_LITTLE_ENDIAN_OPTION, UNSUPPORTED_SH2A):
1589 (DRIVER_SELF_SPECS): Use new macros to filter out unsupported options
1590 taking the default configuration into account.
1592 2011-11-13 Jonathan Wakely <jwakely.gcc@gmail.com>
1594 * doc/install.texi (Specific): Fix anchor for
1595 x86_64-*-solaris2.1[0-9]*.
1597 2011-11-13 Nathan Sidwell <nathan@acm.org>
1599 * gcov.c (source_prefix, source_length): New globals.
1600 (flag_relative_only): Likewise.
1601 (print_usage, options, process_args): Update.
1602 (generate_results): Use coverage.name, check
1603 flag_relative_only. Adjust messages.
1604 (find_source): Check source_prefix.
1605 (output_lines): Use coverage.name, adjust messages.
1606 * doc/gcov.texi (Invoking Gcov): Document new options.
1608 2011-11-12 Jason Merrill <jason@redhat.com>
1611 * gimplify.c (gimplify_target_expr): Add a clobber to the cleanup.
1612 (gimplify_modify_expr): Don't try to simplify it.
1614 2011-11-12 Dimitrios Apostolou <jimis@gmx.net>
1617 * config/elfos.h (STRING_ASM_OP): Define.
1618 * varasm.c (default_elf_asm_output_limited_string): Use it.
1621 * gcc/configure.ac: Check for stpcpy declaration.
1622 * system.h: Declare it if not.
1623 * config.in, configure: Regenerate.
1625 2011-11-12 Richard Henderson <rth@redhat.com>
1627 * config/rs6000/rs6000.md (fix_trunc<SFDF>si2_stfiwx): Use
1628 nonimmediate_operand for the destination.
1629 (fixuns_trunc<SFDF>si2_stfiwx): Likewise.
1631 * builtins.c (expand_builtin_compare_and_swap): If target is const0,
1632 don't pass the target to expand_atomic_compare_and_swap.
1633 (expand_builtin_atomic_compare_exchange): Likewise.
1635 * config/sh/linux.h (TARGET_INIT_LIBFUNCS): New.
1636 * config/sh/sh.c (sh_init_sync_libfuncs): New.
1638 * config/pa/pa-linux.h (TARGET_SYNC_LIBCALL): New.
1639 * config/pa/pa.h (TARGET_SYNC_LIBCALL): New default.
1640 * config/pa/pa.c (pa_init_libfuncs): Use init_sync_libfuncs.
1642 2011-11-12 Iain Sandoe <iains@gcc.gnu.org>
1645 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address):
1646 Only expand a symbol ref. into an access when the entity is defined
1649 2011-11-12 Jan Hubicka <jh@suse.cz>
1652 * cgraphunit.c (output_weakrefs): Output really only weakrefs.
1654 2011-11-12 Nathan Sidwell <nathan@acm.org>
1656 * gcov.c (struct name_map): New.
1657 (names, n_names, a_names): New global vars.
1658 (print_usage): Adjust usage.
1659 (generate_results): Canonicalize main file name.
1660 (release_structures): Adjust.
1661 (name_search, name_sort): New callbacks.
1662 (find_source): Look for and create a canonical name.
1663 (canonicalize_name): New.
1664 (make_gcov_file_name): Reimplement and fix mangling.
1666 * doc/gcov.texi: Update documentation about path preservation.
1668 2011-11-11 David S. Miller <davem@davemloft.net>
1670 * config/sparc/linux.h (ASM_GENERATE_INTERNAL_LABEL): Delete.
1671 * config/sparc/linux64.h (ASM_GENERATE_INTERNAL_LABEL): Delete.
1673 2011-11-11 Jakub Jelinek <jakub@redhat.com>
1675 * config/i386/i386-protos.h (ix86_maybe_emit_epilogue_vzeroupper):
1677 * config/i386/i386.c (ix86_maybe_emit_epilogue_vzeroupper): New
1679 (ix86_expand_epilogue): Use it.
1680 * config/i386/i386.md (return, simple_return): Call it in the
1683 PR tree-optimization/51091
1684 * tree-stdarg.c (execute_optimize_stdarg): Ignore TREE_CLOBBER_P
1685 rhs also in the va_list_simple_ptr case.
1687 PR tree-optimization/51058
1688 * tree-vect-slp.c (vect_remove_slp_scalar_calls): New function.
1689 (vect_schedule_slp): Call it.
1690 * tree-vect-stmts.c (vectorizable_call): If slp_node != NULL,
1691 don't replace scalar calls with setting of their lhs to zero here.
1693 PR tree-optimization/51074
1694 * fold-const.c (vec_cst_ctor_to_array, fold_vec_perm): New functions.
1695 (fold_binary_loc): Handle VEC_EXTRACT_EVEN_EXPR,
1696 VEC_EXTRACT_ODD_EXPR, VEC_INTERLEAVE_HIGH_EXPR and
1697 VEC_INTERLEAVE_LOW_EXPR with VECTOR_CST or CONSTRUCTOR operands.
1698 (fold_ternary_loc): Handle VEC_PERM_EXPR with VECTOR_CST or
1699 CONSTRUCTOR operands.
1700 * tree-ssa-propagate.c (valid_gimple_rhs_p): Handle ternary
1702 * tree-vect-generic.c (lower_vec_perm): Mask sel_int elements
1703 to 0 .. 2 * elements - 1.
1705 2011-11-11 Richard Henderson <rth@redhat.com>
1707 * gimple-pretty-print.c (dump_gimple_omp_atomic_load): Dump needed.
1708 (dump_gimple_omp_atomic_store): Likewise.
1709 * optabs.c (can_atomic_exchange_p): New.
1710 * optabs.h (can_atomic_exchange_p): Declare.
1711 * omp-low.c (expand_omp_atomic_load): Implement.
1712 (expand_omp_atomic_store): Likewise.
1713 (expand_omp_atomic): Update for new arguments to load/store.
1715 2011-11-11 Eric Botcazou <ebotcazou@adacore.com>
1717 * cprop.c: Adjust outdated comments throughout.
1718 (hash_scan_set): Rename PAT parameter into SET.
1719 (cprop_pavloc): Rename into...
1720 (cprop_avloc): ...this.
1721 (cprop_absaltered): Rename into...
1722 (cprop_kill): ...this.
1723 (alloc_cprop_mem): Adjust for above renaming.
1724 (free_cprop_mem): Likewise.
1725 (compute_cprop_data): Likewise.
1726 (compute_local_properties): Rename TRANSP parameter into KILL and
1727 adjust throughout. Rework comments.
1728 (try_replace_reg): Fix long line.
1729 (cprop_jump): Likewise.
1730 (constprop_register): Fix prototype and take INSN last.
1731 (cprop_insn): Adjust calls to above function. Fix long lines.
1732 (bypass_block): Likewise.
1733 (one_cprop_pass): Likewise.
1735 2011-11-11 Uros Bizjak <ubizjak@gmail.com>
1737 * config/alpha/elf.h (ELF_ASCII_ESCAPES): Rename from ESCAPES.
1738 (ELF_STRING_LIMIT): Rename from STRING_LIMIT.
1740 2011-11-11 Jason Merrill <jason@redhat.com>
1742 * varasm.c (default_elf_asm_output_limited_string): #ifdef
1744 (default_elf_asm_output_ascii): Likewise.
1746 2011-11-11 Uros Bizjak <ubizjak@gmail.com>
1749 * config/i386/constraints.md (j): New address constraint.
1750 * config/i386/predicates.md (lea_address_operand): Redefine as
1752 * config/i386/i386.md (*lea_3_zext): Use "j" constraint for operand 1.
1753 (*lea_4_zext): Ditto.
1755 2011-11-11 David S. Miller <davem@davemloft.net>
1758 2011-11-05 David S. Miller <davem@davemloft.net>
1760 * config/sparc/sparc.md (UNSPEC_SHORT_LOAD): New unspec.
1761 (zero-extend_v8qi_vis, zero_extend_v4hi_vis): New expanders.
1762 (*zero_extend_v8qi_<P:mode>_insn,
1763 *zero_extend_v4hi_<P:mode>_insn): New insns.
1764 * config/sparc/sparc.c (vector_init_move_words,
1765 vector_init_prepare_elts, sparc_expand_vector_init_vis2,
1766 sparc_expand_vector_init_vis1): New functions.
1767 (vector_init_bshuffle): Rewrite to handle more cases and make use
1768 of locs[] array prepared by vector_init_prepare_elts.
1769 (vector_init_fpmerge, vector_init_faligndata): Delete.
1770 (sparc_expand_vector_init): Rewrite using new infrastructure.
1772 2011-11-11 Jakub Jelinek <jakub@redhat.com>
1774 * opts-common.c (generate_canonical_option): Free opt_text
1775 it it has been allocated here and not stored anywhere.
1777 * tree-vect-slp.c (vect_free_slp_tree): Also free SLP_TREE_CHILDREN
1779 (vect_create_new_slp_node): Don't allocate node before checking stmt
1781 (vect_free_oprnd_info): Remove FREE_DEF_STMTS argument, always
1782 free def_stmts vectors and additionally free oprnd_info.
1783 (vect_build_slp_tree): Adjust callers. Call it even if
1784 stop_recursion. If vect_create_new_slp_node or
1785 vect_build_slp_tree fails, properly handle freeing memory.
1786 If it succeeded, clear def_stmts in oprnd_info.
1788 2011-11-11 Dimitrios Apostolou <jimis@gmx.net>
1790 * defaults.h (ASM_OUTPUT_LABEL, ASM_OUTPUT_INTERNAL_LABEL): Expand on
1791 multiple lines for readability.
1792 * output.h (fprint_whex, fprint_w, fprint_ul, sprint_ul): Declare.
1793 (ASM_OUTPUT_LABELREF): Call fputs instead of asm_fprintf.
1794 * dwarf2asm.c (dw2_assemble_integer, dw2_asm_output_data,
1795 dw2_asm_data_uleb128, dw2_asm_delta_uleb128,
1796 dw2_asm_delta_sleb128): Convert fprintf calls to the new functions.
1797 * dwarf2out.c (dwarf2out_source_line): Call fprint_ul instead of
1799 * final.c (fprint_whex, fprint_w, fprint_ul, sprint_ul): New functions
1800 serving as fast replacements for fprintf integer to string conversions.
1801 (sprint_ul_rev): Internal helper for the above.
1802 (output_addr_const) <CONST_INT>: Don't use fprintf.
1803 * varasm.c: Fix some whitespace inconsistencies.
1804 (default_elf_asm_output_limited_string): New.
1805 (default_elf_asm_output_ascii): Likewise.
1806 (default_elf_internal_label): Likewise.
1807 * config/elfos.h (ASM_GENERATE_INTERNAL_LABEL): Use sprint_ul and
1808 stpcpy instead of sprintf.
1809 (TARGET_ASM_INTERNAL_LABEL): Define as default_elf_internal_label.
1810 (ELF_ASCII_ESCAPES, ELF_STRING_LIMIT): Rename from old ESCAPES and
1811 STRING_LIMIT macros.
1812 (ASM_OUTPUT_LIMITED_STRING, ASM_OUTPUT_ASCII): Call functions that
1813 provide the same functionality.
1814 * config/i386/i386.c (print_reg): Call fprint_ul instead of fprintf.
1816 2011-11-10 Andrew MacLeod <amacleod@redhat.com>
1818 * doc/extend.texi: Document __atomic_test_and_set and __atomic_clear.
1820 2011-11-10 Andrew MacLeod <amacleod@redhat.com>
1823 * builtins.c (expand_builtin_atomic_clear): New. Expand atomic_clear.
1824 (expand_builtin_atomic_test_and_set): New. Expand atomic test_and_set.
1825 (expand_builtin): Add cases for test_and_set and clear.
1826 * sync-builtins.def (BUILT_IN_ATOMIC_TEST_AND_SET): New.
1827 (BUILT_IN_ATOMIC_CLEAR): New.
1829 2011-11-10 Roberto Agostino Vitillo <ravitillo@lbl.gov>
1832 * dwarf2out.c (set_cur_line_info_table): Restore the last is_stmt
1833 value in the current line table.
1835 2011-11-10 Nathan Sidwell <nathan@acm.org>
1837 * gcov.c (struct function_info): Make src an index, not a pointer.
1838 (struct source_info): Remove index and next source fields.
1839 (fn_end): New static var.
1840 (sources_index): Remove.
1841 (sources): Now a pointer to an array, not a list.
1842 (n_sources, a_sources): New.
1843 (process_file): Adjust for changes to read_graph_file. Insert
1844 functions into source lists and check line numbers here.
1845 (generate_results): Only allocate lines for sources with
1846 contents. Adjust for source array.
1847 (release_structures): Likewise.
1848 (find_source): Return source index, adjust for source array.
1849 (read_graph_file): Return function list. Don't insert into source
1851 (read_count_file): Take list of functions.
1852 (solve_flow_graph): Reverse the arc lists here.
1853 (add_line_counts): Adjust for source array.
1855 2011-11-10 Jakub Jelinek <jakub@redhat.com>
1858 * tree-object-size.c (addr_object_size): Check TREE_CODE of
1859 MEM_REF's operand rather than code of the MEM_REF itself.
1861 2011-11-10 Andrew MacLeod <amacleod@redhat.com>
1863 PR rtl-optimization/51040
1864 * optabs.c (expand_atomic_fetch_op): Patchup code for NAND should
1865 be AND followed by NOT.
1866 * builtins.c (expand_builtin_atomic_fetch_op): Patchup code for NAND
1867 should be AND followed by NOT.
1869 2011-11-10 Jakub Jelinek <jakub@redhat.com>
1871 * vec.h (VEC_BASE): If base is at offset 0 in the structure,
1872 use &(P)->base even if P is NULL.
1874 PR rtl-optimization/51023
1875 * combine.c (simplify_comparison) <case SIGN_EXTEND>: Don't use
1876 val_signbit_known_clear_p for signed comparison narrowing
1877 optimization. Don't check for non-VOIDmode, use
1878 HWI_COMPUTABLE_MODE_P macro.
1879 <case ZERO_EXTEND>: Don't check for non-VOIDmode.
1880 Optimize even when const_op is equal to GET_MODE_MASK (mode),
1881 don't optimize if const_op is negative.
1883 2011-11-10 Richard Guenther <rguenther@suse.de>
1885 PR tree-optimization/51042
1886 * tree-ssa-pre.c (phi_translate_1): Avoid recursing on
1887 self-referential expressions. Refactor code to avoid duplication.
1889 2011-11-10 Richard Guenther <rguenther@suse.de>
1891 PR tree-optimization/51070
1892 * tree-loop-distribution.c (generate_builtin): Do not replace
1893 the loop with a builtin if the partition contains statements which
1894 results are used outside of the loop.
1895 (pass_loop_distribution): Verify and collect.
1897 2011-11-10 Richard Guenther <rguenther@suse.de>
1899 PR tree-optimization/51030
1900 * tree-ssa-phiopt.c (jump_function_from_stmt): New function.
1901 (value_replacement): Use it to handle trivial non-empty
1902 intermediate blocks.
1904 2011-11-10 Richard Guenther <rguenther@suse.de>
1907 * gimple.c (gimple_has_side_effects): Remove checking code
1908 that doesn't belong here.
1910 2011-11-10 Ira Rosen <ira.rosen@linaro.org>
1912 PR tree-optimization/51058
1913 * tree-vect-slp.c (vect_get_constant_vectors): Handle CALL_EXPR.
1915 2011-11-10 Jakub Jelinek <jakub@redhat.com>
1917 PR tree-optimization/51000
1918 * tree-vect-patterns.c (vect_recog_bool_pattern): If adding
1919 a pattern stmt for a bool store, adjust DR_STMT too.
1920 Don't handle bool conversions to single bit precision lhs.
1921 * tree-vect-stmts.c (vect_remove_stores): If next is a pattern
1922 stmt, remove its related stmt and free its stmt_vinfo.
1923 (free_stmt_vec_info): Free also pattern stmt's vinfo and
1924 pattern def stmt's vinfo.
1925 * tree-vect-loop.c (destroy_loop_vec_info): Don't try to
1926 free pattern stmt's vinfo here.
1927 (vect_transform_loop): When calling vect_remove_stores,
1928 do gsi_next first and don't call gsi_remove. If not strided
1929 store, free stmt vinfo for gsi_stmt (si) rather than stmt.
1931 2011-11-09 Jakub Jelinek <jakub@redhat.com>
1934 * config/i386/i386.c (expand_vec_perm_interleave2): If d->vmode is
1935 V4SImode, !TARGET_SSE2 and punpck[lh]* is needed, change dremap.vmode
1938 * function.h (requires_stack_frame_p): New prototype.
1939 * function.c (requires_stack_frame_p): No longer static.
1940 * config/i386/i386.c (ix86_finalize_stack_realign_flags): If
1941 stack_realign_fp was just a conservative guess for a function
1942 which doesn't use sp/fp/argp at all, clear frame_pointer_needed
1943 and stack realignment.
1945 2011-11-09 Paolo Carlini <paolo.carlini@oracle.com>
1947 PR preprocessor/51061
1948 * incpath.c (merge_include_chains): Make sure to not pass null
1949 pointers to free_path.
1951 2011-11-09 Paolo Carlini <paolo.carlini@oracle.com>
1953 * doc/invoke.texi ([Option Summary, C++ Language Options]):
1954 Add -Wzero-as-null-pointer-constant.
1956 2011-11-07 Matthew Gretton-Dann <matthew.gretton-dann@arm.com>
1958 * config/arm/arm-cores.def: Add -mcpu=cortex-a7.
1959 * config/arm/arm-tables.opt: Regenerate.
1960 * config/arm/arm-tune.md: Likewise.
1961 * config/arm/bpabi.h (BE8_LINK_SPEC): Add Cortex A-7.
1962 * doc/invoke.texi: Document -mcpu=cortex-a7.
1964 2011-11-09 Richard Guenther <rguenther@suse.de>
1967 2011-11-09 Richard Guenther <rguenther@suse.de>
1969 * tree-cfg.c (verify_gimple_call): Verify that
1970 gimple_call_cannot_inline_p is returning a conservative
1971 correct result according to gimple_check_call_matching_types.
1973 2011-11-09 Richard Guenther <rguenther@suse.de>
1975 PR tree-optimization/51039
1976 * tree-cfg.c (verify_gimple_call): Verify that
1977 gimple_call_cannot_inline_p is returning a conservative
1978 correct result according to gimple_check_call_matching_types.
1979 * ipa-inline-analysis.c (estimate_function_body_sizes): Remove
1980 code dealing with un-inlinablility.
1981 * gimple-streamer-in.c (input_gimple_stmt): Update the
1984 2011-11-09 Richard Guenther <rguenther@suse.de>
1986 PR tree-optimization/51039
1987 * tree-inline.c (setup_one_parameter): Always perform a
1988 valid gimple type change.
1989 (declare_return_variable): Likewise.
1991 2011-11-09 Jakub Jelinek <jakub@redhat.com>
1993 * config/rs6000/vector.md (vcondv4sfv4si, vcondv4siv4sf,
1994 vconduv4sfv4si): New patterns.
1995 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Handle
1996 different dest_mode from comparison mode.
1998 2011-11-09 Richard Guenther <rguenther@suse.de>
2000 * gimple-fold.c (canonicalize_constructor_val): Mark
2001 address-taken variables addressable.
2003 2011-11-09 Ira Rosen <ira.rosen@linaro.org>
2005 PR tree-optimization/51015
2006 * tree-vect-loop.c (vect_determine_vectorization_factor): Expect
2007 vectype to be set for pattern def stmts.
2009 2011-11-09 Alan Modra <amodra@gmail.com>
2011 * function.c (bb_active_p): Delete.
2012 (dup_block_and_redirect, active_insn_between): New functions.
2013 (convert_jumps_to_returns, emit_return_for_exit): New functions,
2015 (thread_prologue_and_epilogue_insns): ..here. Delete
2016 shadowing variables. Don't do prologue register clobber tests
2017 when shrink wrapping already failed. Delete all last_bb_active
2018 code. Instead compute tail block candidates for duplicating
2019 exit path. Remove these from antic set. Duplicate tails when
2020 reached from both blocks needing a prologue/epilogue and
2021 blocks not needing such.
2022 * ifcvt.c (dead_or_predicable): Test both flag_shrink_wrap and
2024 * bb-reorder.c (get_uncond_jump_length): Make global.
2025 * bb-reorder.h (get_uncond_jump_length): Declare.
2026 * cfgrtl.c (rtl_create_basic_block): Comment typo fix.
2027 (rtl_split_edge): Likewise. Warning fix.
2028 (rtl_duplicate_bb): New function.
2029 (rtl_cfg_hooks): Enable can_duplicate_block_p and duplicate_block.
2030 * Makefile.in (function.o): Update dependencies.
2032 2011-11-08 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2033 Georg-Johann Lay <avr@gjlay.de>
2035 * doc/tm.texi.in (MODE_CODE_BASE_REG_CLASS): Add address space
2037 (REGNO_MODE_CODE_OK_FOR_BASE_P): Likewise.
2038 * doc/tm.texi: Regenerate.
2040 * config/cris/cris.h (MODE_CODE_BASE_REG_CLASS): Add address
2042 (REGNO_MODE_CODE_OK_FOR_BASE_P): Likewise.
2043 * config/bfin/bfin.h (MODE_CODE_BASE_REG_CLASS): Likewise.
2044 (REGNO_MODE_CODE_OK_FOR_BASE_P): Likewise.
2045 * config/avr/avr.h (MODE_CODE_BASE_REG_CLASS): Add address space
2047 (REGNO_MODE_CODE_OK_FOR_BASE_P): Ditto.
2048 * config/avr/avr-protos.h (avr_mode_code_base_reg_class): Ditto.
2049 (avr_regno_mode_code_ok_for_base_p): Ditto.
2050 * config/avr/avr.c (avr_mode_code_base_reg_class): Ditto.
2051 (avr_regno_mode_code_ok_for_base_p): Ditto.
2052 (avr_reg_ok_for_addr_p): Pass AS down to
2053 avr_regno_mode_code_ok_for_base_p.
2055 * addresses.h (base_reg_class): Add address space argument.
2056 Pass to MODE_CODE_BASE_REG_CLASS.
2057 (ok_for_base_p_1): Add address space argument. Pass to
2058 REGNO_MODE_CODE_OK_FOR_BASE_P.
2059 (regno_ok_for_base_p): Add address space argument. Pass to
2062 * regrename.c (scan_rtx_address): Add address space argument.
2063 Pass address space to regno_ok_for_base_p and base_reg_class.
2064 Update recursive calls.
2065 (scan_rtx): Pass address space to scan_rtx_address.
2066 (build_def_use): Likewise.
2067 * regcprop.c (replace_oldest_value_addr): Add address space
2068 argument. Pass to regno_ok_for_base_p and base_reg_class.
2069 Update recursive calls.
2070 (replace_oldest_value_mem): Pass address space to
2071 replace_oldest_value_addr.
2072 (copyprop_hardreg_forward_1): Likewise.
2074 * reload.c (find_reloads_address_1): Add address space argument.
2075 Pass address space to base_reg_class and regno_ok_for_base_p.
2076 Update recursive calls.
2077 (find_reloads_address): Pass address space to base_reg_class,
2078 regno_ok_for_base_p, and find_reloads_address_1.
2079 (find_reloads): Pass address space to base_reg_class.
2080 (find_reloads_subreg_address): Likewise.
2082 * ira-costs.c (record_reg_classes): Update calls to base_reg_class.
2083 (ok_for_base_p_nonstrict): Add address space argument. Pass to
2085 (record_address_regs): Add address space argument. Pass to
2086 base_reg_class and ok_for_base_p_nonstrict. Update recursive calls.
2087 (record_operand_costs): Pass address space to record_address_regs.
2088 (scan_one_insn): Likewise.
2090 * caller-save.c (init_caller_save): Update call to base_reg_class.
2091 * ira-conflicts.c (ira_build_conflicts): Likewise.
2092 * reload1.c (maybe_fix_stack_asms): Likewise.
2094 2011-11-08 Michael Matz <matz@suse.de>
2096 * gengtype.c (write_field_root): Avoid out-of-scope access of newv.
2098 * tree-stdarg.c (execute_optimize_stdarg): Accept clobbers.
2100 * tree.h (TREE_CLOBBER_P): New macro.
2101 * gimple.h (gimple_clobber_p): New inline function.
2102 * gimplify.c (gimplify_bind_expr): Add clobbers for all variables
2103 that go out of scope and live in memory.
2104 * tree-ssa-operands.c (get_expr_operands): Transfer volatility also
2106 * cfgexpand.c (decl_to_stack_part): New static variable.
2107 (add_stack_var): Allocate it, and remember mapping.
2108 (fini_vars_expansion): Deallocate it.
2109 (stack_var_conflict_p): Add early outs.
2110 (visit_op, visit_conflict, add_scope_conflicts_1,
2111 add_scope_conflicts): New static functions.
2112 (expand_used_vars_for_block): Don't call add_stack_var_conflict, tidy.
2113 (expand_used_vars): Add scope conflicts.
2114 (expand_gimple_stmt_1): Expand clobbers to nothing.
2115 (expand_debug_expr): Ditto.
2117 * tree-pretty-print.c (dump_generic_node): Dump clobbers nicely.
2118 * tree-ssa-live.c (remove_unused_locals): Remove clobbers that
2119 refer to otherwise unused locals.
2120 * tree-sra.c (build_accesses_from_assign): Ignore clobbers.
2121 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Clobbers of
2122 SSA names aren't necessary.
2123 (propagate_necessity): Accept and ignore constructors on the rhs, tidy.
2124 * gimple.c (walk_gimple_op): Accept constructors like mem_rhs.
2125 * tree-ssa-structalias.c (find_func_aliases): Clobbers don't store
2127 * tree-ssa-sccvn.c (vn_reference_lookup_3): Ditto, in particular they
2128 don't zero-initialize something.
2129 * tree-ssa-phiopt.c (cond_if_else_store_replacement_1): Ignore
2130 clobber RHS, we don't want PHI nodes with those.
2132 2011-11-08 Jakub Jelinek <jakub@redhat.com>
2134 * config/i386/i386.c (ix86_preferred_simd_mode): Even for TARGET_AVX
2135 if not TARGET_PREFER_AVX128 return 32-byte integer vectors.
2137 2011-11-08 Richard Guenther <rguenther@suse.de>
2139 PR tree-optimization/51012
2140 * ipa-prop.c (update_indirect_edges_after_inlining): Fixup
2141 non-inlinable state.
2142 * cgraph.c (cgraph_make_edge_direct): Likewise.
2144 2011-11-08 Eric Botcazou <ebotcazou@adacore.com>
2146 PR rtl-optimization/47698
2147 * ifcvt.c (noce_operand_ok): Move around comment.
2149 2011-11-08 Richard Guenther <rguenther@suse.de>
2152 * lto-opts.c (append_to_collect_gcc_options): Split out from...
2153 (lto_write_options): ... here. Prepend frontend specific flags.
2155 2011-11-08 Jakub Jelinek <jakub@redhat.com>
2157 * config/i386/sse.md (*avx_vperm2f128<mode>3_nozero): Emit mask
2158 0x12 and 0x20 as vinsert[fi]128 instead of vperm2[fi]128.
2160 * tree-vect-stmts.c (vectorizable_call): Add SLP_NODE argument.
2161 Handle vectorization of SLP calls.
2162 (vect_analyze_stmt): Adjust caller, add call to it for SLP too.
2163 (vect_transform_stmt): Adjust vectorizable_call caller, remove
2165 * tree-vect-slp.c (vect_get_and_check_slp_defs): For calls start
2167 (vect_build_slp_tree): Allow CALL_EXPR.
2169 2011-11-08 Richard Guenther <rguenther@suse.de>
2171 * gimple-fold.c (canonicalize_constructor_val): Make sure
2172 we have referenced vars setup before adding to them.
2174 2011-11-07 Richard Henderson <rth@redhat.com>
2175 Aldy Hernandez <aldyh@redhat.com>
2176 Andrew MacLeod <amacleod@redhat.com>
2177 Torvald Riegel <triegel@redhat.com>
2179 Merged from transactional-memory.
2181 * gtm-builtins.def: New file.
2182 * trans-mem.c: New file.
2183 * trans-mem.h: New file.
2185 * opts.c (finish_options): Error out when using -flto and -fgnu-tm.
2187 * config/i386/i386.c: Define TARGET_VECTORIZE* transactional variants.
2188 (ix86_handle_tm_regparm_attribute, struct bdesc_tm,
2189 ix86_builtin_tm_load, ix86_builtin_tm_store,
2190 ix86_init_tm_builtins): New.
2191 (ix86_init_builtins): Initialize TM builtins.
2192 (struct ix86_attribute_table): Add "*tm regparm".
2193 * config/i386/i386-builtin-types.def (PV2SI): Define.
2195 Define V2SI_FTYPE_PCV2SI.
2196 Define V4SF_FTYPE_PCV4SF.
2197 Define V8SF_FTYPE_PCV8SF.
2198 Define VOID_PV2SI_V2SI.
2200 * doc/invoke.texi (C Dialect Options): Document -fgnu-tm and
2201 tm-max-aggregate-size.
2202 * doc/tm.texi.in: Add TARGET_VECTORIZE_BUILTIN_TM_LOAD and
2203 TARGET_VECTORIZE_BUILTIN_TM_STORE hooks.
2204 * doc/tm.texi: Regenerate.
2206 * attribs.c (apply_tm_attr): New.
2207 (init_attributes): Allow '*' prefix for overrides.
2208 (register_attribute): Likewise.
2209 * builtin-attrs.def (ATTR_TM_TMPURE, ATTR_TM_REGPARM): New.
2210 (ATTR_TM_NOTHROW_LIST, ATTR_TM_TMPURE_NOTHROW_LIST,
2211 ATTR_TM_PURE_TMPURE_NOTHROW_LIST, ATTR_TM_NORETURN_NOTHROW_LIST,
2212 ATTR_TM_CONST_NOTHROW_LIST, ATTR_TMPURE_MALLOC_NOTHROW_LIST,
2213 ATTR_TMPURE_NOTHROW_LIST): New.
2214 * builtin-types.def (BT_FN_I[1248]_VPTR, BT_FN_FLOAT_VPTR,
2215 BT_FN_DOUBLE_VPTR, BT_FN_LDOUBLE_VPTR, BT_FN_VOID_VPTR_I[1248],
2216 BT_FN_VOID_VPTR_FLOAT, BT_FN_VOID_VPTR_DOUBLE,
2217 BT_FN_VOID_VPTR_LDOUBLE, BT_FN_VOID_VPTR_SIZE): New.
2218 * builtins.def: Include gtm-builtins.def. Add comments regarding
2219 transactional memory synchronization.
2220 (DEF_TM_BUILTIN): New.
2221 * c-parser.c (struct c_parser): Add in_transaction.
2222 (c_parser_transaction, c_parser_transaction_expression,
2223 c_parser_transaction_cancel, c_parser_transaction_attributes): New.
2224 (c_parser_attribute_any_word): Split out from c_parser_attributes.
2225 (c_parser_statement_after_labels): Handle RID_TRANSACTION*.
2226 (c_parser_unary_expression): Same.
2227 * c-tree.h (c_finish_transaction): Declare.
2228 * c-typeck.c (c_finish_transaction): New.
2229 (build_function_call_vec): Call tm_malloc_replacement.
2230 * calls.c (is_tm_builtin): New.
2231 (flags_from_decl_or_type): Add ECF_TM_BUILTIN and ECF_TM_PURE.
2232 * cfgbuild.c (make_edges): Add edges for REG_TM notes.
2233 * cfgexpand.c (expand_call_stmt): Call
2234 mark_transaction_restart_calls.
2235 (gimple_expand_cfg): Free the tm_restart map.
2236 (mark_transaction_restart_calls): New.
2237 * cfgrtl.c (purge_dead_edges): Look for REG_TM notes.
2238 * cgraph.c (dump_cgraph_node): Handle tm_clone.
2239 * cgraph.h (struct cgraph_node): Add tm_clone field.
2240 (decl_is_tm_clone): New.
2241 (struct cgraph_local_info): Add tm_may_enter_irr.
2242 (cgraph_copy_node_for_versioning): Declare.
2243 * cgraphunit.c (cgraph_copy_node_for_versioning): Export;
2244 copy analyzed from old version.
2245 * combine.c (distribute_notes): Handle REG_TM notes.
2246 * common.opt: Add -fgnu-tm.
2247 * crtstuff.c (__TMC_LIST__, __TMC_END__): New.
2248 (__do_global_dtors_aux): Deregister clone table.
2249 (frame_dummy): Register clone table.
2250 * emit-rtl.c (try_split): Handle REG_TM. Early return if no function
2252 * gimple-low.c (lower_stmt): Handle GIMPLE_EH_ELSE and
2254 (gimple_stmt_may_fallthru): Handle GIMPLE_EH_ELSE.
2255 * gimple-pretty-print.c: Include trans-mem.h.
2256 (dump_gimple_fmt): Add %x.
2257 (dump_gimple_call): Dump arguments for calls to _ITM_beginTransaction.
2258 (dump_gimple_eh_else, dump_gimple_transaction): New.
2259 (dump_gimple_stmt): Handle GIMPLE_EH_ELSE and GIMPLE_TRANSACTION.
2260 * gimple.c (gimple_build_eh_else, gimple_build_transaction): New.
2261 (walk_gimple_seq): Honor removed_stmt. Document usage of removed_stmt
2263 (walk_gimple_op): Handle GIMPLE_TRANSACTION.
2264 (walk_gimple_stmt): Initialize and honor removed_stmt.
2265 Handle GIMPLE_EH_ELSE and GIMPLE_TRANSACTION.
2266 (gimple_copy): Handle GIMPLE_EH_ELSE and GIMPLE_TRANSACTION.
2267 * gimple.def (GIMPLE_TRANSACTION, GIMPLE_EH_ELSE): New.
2268 * gimple.h (struct gimple_statement_eh_else,
2269 gimple_statement_transaction, GTMA_*): New.
2270 (gimple_statement_d): Add gimple_statement_eh_else and
2272 (gimple_build_eh_else, gimple_build_transaction,
2273 gimple_fold_call, diagnose_tm_safe_errors): Declare.
2274 (get_call_expr_in): Remove prototype.
2275 (gimple_has_substatements): Add GIMPLE_EH_ELSE and GIMPLE_TRANSACTION.
2276 (gimple_eh_else_n_body, gimple_eh_else_e_body,
2277 gimple_eh_else_set_n_body, gimple_eh_else_set_e_body,
2278 gimple_transaction_body, gimple_transaction_label,
2279 gimple_transaction_label_ptr, gimple_transaction_subcode,
2280 gimple_transaction_set_body, gimple_transaction_set_label,
2281 gimple_transaction_set_subcode): New.
2282 (struct walk_stmt_info): Use BOOL_BITFIELD; add removed_stmt.
2283 * gimplify.c (create_tmp_var_name): Use clean_symbol_name.
2284 (voidify_wrapper_expr): Handle TRANSACTION_EXPR.
2285 (gimplify_transaction): New.
2286 (gimplify_expr): Handle TRANSACTION_EXPR.
2287 * gsstruct.def (GSS_EH_ELSE, GSS_TRANSACTION): New.
2288 * ipa-inline.c (can_inline_edge_p): Do not inline TM safe calling
2290 * Makefile.in: Add trans-mem.o and dependencies.
2291 (BUILTINS_DEF): Add gtm-builtins.def.
2292 (gimple-pretty-print.o): Depend on TRANS_MEM_H.
2293 (GTFILES): Add trans-mem.c.
2294 * omp-low.c (WALK_SUBSTMTS): Add GIMPLE_TRANSACTION.
2295 * output.h (record_tm_clone_pair, finish_tm_clone_pairs,
2296 get_tm_clone_pair): Declare.
2297 * params.def (PARAM_TM_MAX_AGGREGATE_SIZE): New.
2298 * passes.c (init_optimization_passes): Place transactional memory
2300 * print-tree.c (print_node): Dump tm-clone.
2301 * recog.c (peep2_attempt): Handle REG_TM.
2302 * reg-notes.def (TM): New.
2303 * rtlanal.c (alloc_reg_note): Handle REG_TM.
2304 * target.def (builtin_tm_load, builtin_tm_store): New.
2305 * targhooks.c (default_builtin_tm_load_store): New.
2306 * targhooks.h (default_builtin_tm_load_store): Declare.
2307 * timevar.def (TV_TRANS_MEM): New.
2308 * toplev.c (compile_file): Call finish_tm_clone_pairs.
2309 * tree-cfg.c (make_edges): Handle GIMPLE_TRANSACTION.
2310 (cleanup_dead_labels): Handle GIMPLE_TRANSACTION. Avoid unnecessary
2311 writes into the statements to update labels.
2312 (is_ctrl_altering_stmt): Add TM ending statements. Handle
2314 (verify_gimple_transaction): New.
2315 (verify_gimple_stmt): Handle GIMPLE_TRANSACTION.
2316 (verify_gimple_in_seq_2): Handle GIMPLE_EH_ELSE and GIMPLE_TRANSACTION.
2317 (gimple_redirect_edge_and_branch): Handle TM_TRANSACTION.
2318 (dump_function_to_file): Display [tm-clone] if applicable.
2319 * tree-eh.c (struct_ptr_eq): Make inline and move to tree.h.
2320 (struct_ptr_hash): Same.
2321 (collect_finally_tree): Handle GIMPLE_EH_ELSE.
2322 (replace_goto_queue_1): Likewise.
2324 (honor_protect_cleanup_actions): Handle GIMPLE_EH_ELSE.
2325 (lower_try_finally_nofallthru): Likewise.
2326 (lower_try_finally_onedest): Likewise.
2327 (lower_try_finally_copy): Likewise.
2328 (lower_try_finally_switch): Likewise.
2329 (lower_try_finally): Likewise.
2330 (decide_copy_try_finally): Likewise.
2331 (lower_eh_constructs_2): Likewise.
2332 (refactor_eh_r): Likewise.
2333 * tree-flow.h (struct gimple_df): Add tm_restart field.
2334 Define tm_restart_node.
2335 * tree-inline.c (remap_gimple_stmt): Handle GIMPLE_TRANSACTION.
2336 (estimate_num_insns): Likewise.
2337 (init_inline_once): Init tm_cost.
2338 * tree-inline.h (struct eni_weights_d): Add tm_cost.
2339 * tree-pass.h (pass_diagnose_tm_blocks, pass_lower_tm, pass_tm_init,
2340 pass_tm_mark, pass_tm_memopt, pass_tm_edges, pass_ipa_tm): Declare.
2341 * tree-pretty-print.c (dump_generic_node): Handle TRANSACTION_EXPR.
2342 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Handle
2343 BUILT_IN_TM_MEMSET, BUILT_IN_TM_MEMCPY, BUILT_IN_TM_MEMMOVE.
2344 Add support for TM vector loads. Add support for TM logging builtins.
2345 (call_may_clobber_ref_p_1): Add support for vector stores.
2346 * tree-ssa-structalias.c (find_func_aliases): Add support for TM
2347 vector stores and loads. Handle BUILT_IN_TM_MEMSET,
2348 BUILT_IN_TM_MEMCPY, BUILT_IN_TM_MEMMOVE.
2349 * tree.c (strip_invariant_refs): Moved from gimple.c to here.
2350 (local_define_builtin): Handle ECF_TM_PURE.
2351 (build_common_builtin_nodes): Set __builtin_eh_pointer to ECF_TM_PURE.
2352 * tree.def (TRANSACTION_EXPR): New.
2353 * tree.h (strip_invariant_refs): Moved from gimple.h to here.
2354 (TRANSACTION_EXPR_BODY, TRANSACTION_EXPR_CHECK,
2355 TRANSACTION_EXPR_OUTER, TRANSACTION_EXPR_RELAXED,
2356 BUILTIN_TM_LOAD_STORE_P, BUILTIN_TM_LOAD_P, BUILTIN_TM_STORE_P,
2357 CASE_BUILT_IN_TM_LOAD, CASE_BUILT_IN_TM_STORE): New.
2358 (ECF_TM_PURE, ECF_TM_BUILTIN): New.
2359 (struct tree_function_decl): Add tm_clone_flag.
2360 (struct_ptr_eq, struct_ptr_hash): New.
2361 (apply_tm_attr): Declare.
2362 (is_tm_safe_or_pure): New.
2363 (build_tm_abort_call, is_tm_safe, is_tm_pure,
2364 is_tm_may_cancel_outer, is_tm_ending_fndecl, record_tm_replacement,
2365 tm_malloc_replacement): Declare.
2366 * varasm.c (tm_clone_hash): New.
2367 (record_tm_clone_pair, finish_tm_clone_pairs, get_tm_clone_pair,
2368 dump_tm_clone_to_vec, dump_tm_clone_pairs, tm_alias_pair_cmp): New.
2369 (struct tm_alias_pair): New. Declare VEC types for object.
2371 2011-11-07 Richard Henderson <rth@redhat.com>
2373 * optabs.h (OTI_sync_compare_and_swap, OTI_sync_lock_test_and_set,
2374 OTI_sync_old_add, OTI_sync_old_sub, OTI_sync_old_ior,
2375 OTI_sync_old_and, OTI_sync_old_xor, OTI_sync_old_nand,
2376 OTI_sync_new_add, OTI_sync_new_sub, OTI_sync_new_ior,
2377 OTI_sync_new_and, OTI_sync_new_xor, OTI_sync_new_nand): Move and
2378 rename from the direct_optab_index enum.
2379 (sync_compare_and_swap_optab, sync_lock_test_and_set_optab,
2380 sync_old_add_optab, sync_old_sub_optab, sync_old_ior_optab,
2381 sync_old_and_optab, sync_old_xor_optab, sync_old_nand_optab,
2382 sync_new_add_optab, sync_new_sub_optab, sync_new_ior_optab,
2383 sync_new_and_optab, sync_new_xor_optab, sync_new_nand_optab): Read
2384 from the optab_table, not the direct_optab_table.
2385 (init_sync_libfuncs): Declare.
2386 (can_compare_and_swap_p): Update parameters.
2387 * optabs.c (init_sync_libfuncs_1, init_sync_libfuncs): New.
2388 (can_compare_and_swap_p): Add allow_libcall parameter; if true,
2389 test for the legacy compare-and-swap libcall.
2390 (expand_atomic_exchange): Use the legacy test-and-set libcall.
2391 (expand_atomic_compare_and_swap): Use the legacy CAS libcall.
2392 (struct atomic_op_functions): Update for optab type changes.
2393 (maybe_emit_op): Likewise.
2394 (expand_atomic_fetch_op): Use the legacy fetch-op libcalls.
2395 * builtins.c (fold_builtin_atomic_always_lock_free): Update call
2396 to can_compare_and_swap_p.
2397 * omp-low.c (expand_omp_atomic_fetch_op): Likewise.
2398 (expand_omp_atomic_pipeline): Likewise.
2399 * genopinit.c (optabs): Make sync_old_*_optab, sync_new_*_optab,
2400 sync_compare_and_swap_optab, sync_lock_test_and_set_optab regular
2402 * doc/md.texi (sync_compare_and_swap): Update docs for libcalls.
2404 2011-11-07 Jakub Jelinek <jakub@redhat.com>
2406 * config/i386/i386-bultin-types.def (V8SI_FTYPE_V4DF_V4DF): Add.
2407 * config/i386/i386.c (enum ix86_builtins): Add
2408 IX86_BUILTIN_VEC_PACK_SFIX256.
2409 (bdesc_args): Add __builtin_ia32_vec_pack_sfix256.
2410 (ix86_expand_args_builtin): Handle V8SI_FTYPE_V4DF_V4DF.
2411 (ix86_builtin_vectorized_function): Also vectorize lrint using
2412 256-bit vectors for -mavx.
2414 2011-11-07 Anatoly Sokolov <aesok@post.ru>
2416 * config/cris/constraints.md: New file.
2417 * config/cris/cris.h (REG_CLASS_FROM_LETTER, CONSTRAINT_LEN,
2418 CRIS_CONST_OK_FOR_LETTER_P, CONST_OK_FOR_CONSTRAINT_P,
2419 CONST_DOUBLE_OK_FOR_LETTER_P, EXTRA_MEMORY_CONSTRAINT,
2420 EXTRA_CONSTRAINT, EXTRA_CONSTRAINT_Q, EXTRA_CONSTRAINT_R,
2421 EXTRA_CONSTRAINT_T, EXTRA_CONSTRAINT_S, EXTRA_CONSTRAINT_U): Remove.
2422 * config/cris/cris.c: Incule "tm-constrs.h".
2423 (cris_print_operand): Use satisfies_constraint_O.
2424 (cris_normal_notice_update_cc, cris_rtx_costs): Use
2425 satisfies_constraint_I.
2426 (cris_address_cost): Use satisfies_constraint_L.
2427 * config/cris/cris.md: Include "constraints.md".
2428 (*mov_side<mode>, *mov_sidesisf, *mov_side<mode>_mem,
2429 *mov_sidesisf_mem, *clear_side<mode>, *ext_sideqihi,
2430 *ext_side<mode>si, *op_side<mode>, *op_swap_side<mode>,
2431 *extopqihi_side, *extop<mode>si_side, *extopqihi_swap_side,
2432 *extop<mode>si_swap_side): Use satisfies_constraint_N and
2433 satisfies_constraint_J.
2434 (moversideqi movemsideqi mover2side peephole2): Use
2435 satisfies_constraint_N and satisfies_constraint_J.
2436 (andu peephole2): Use satisfies_constraint_I and
2437 satisfies_constraint_O.
2439 2011-11-07 Uros Bizjak <ubizjak@gmail.com>
2441 * config/i386/i386.c (ix86_builtin_vectorized_function): Handle
2442 BUILT_IN_IRINT, BUILT_IN_IRINTF, BUILT_IN_LLRINT and BUILT_IN_LLRINTF.
2444 2011-11-07 Andrew MacLeod <amacleod@redhat.com>
2446 * optabs.c (get_atomic_op_for_code): Fill in optab table at runtime so
2447 SWITCHABLE_TARGET can change the values during compilation.
2448 (expand_atomic_fetch_op): Handle parameter change ripples for
2449 get_atomic_op_for_code call.
2451 2011-11-07 Andrew MacLeod <amacleod@redhat.com>
2453 * doc/extend.texi: Docuemnt behaviour change for __atomic_exchange and
2455 * optabs.c (expand_atomic_exchange): Expand to __sync_lock_test_and_set
2456 only when originated from that builtin.
2457 (expand_atomic_store): Expand to __sync_lock_release when originated
2459 * builtins.c (expand_builtin_sync_lock_test_and_set): Add flag that
2460 expand_atomic_exchange call originated from here.
2461 (expand_builtin_sync_lock_release): Add flag that expand_atomic_store
2462 call originated from here.
2463 (expand_builtin_atomic_exchange): Add origination flag.
2464 (expand_builtin_atomic_store): Add origination flag.
2465 * expr.h (expand_atomic_exchange, expand_atomic_store): Add boolean
2466 parameters to indicate implementation fall back options.
2468 2011-11-07 Georg-Johann Lay <avr@gjlay.de>
2470 * config/avr/avr.c (output_reload_in_const): Can handle CONSTANT_P
2471 now, not only CONST_INT and CONST_DOUBLE.
2472 (output_movqi): Use output_reload_in_const.
2473 (output_reload_inhi): Ditto.
2474 (output_reload_insisf): Move assertion to output_reload_in_const.
2475 (avr_out_reload_inpsi): Ditto.
2477 2011-11-07 Nathan Sidwell <nathan@acm.org>
2479 * gcov.c (object_summary): Replace with ...
2480 (object_runs): ... this.
2481 (process_file): Remove functions with no data.
2482 (generate_results): Ignore files with no lines.
2483 (release_function): New helper, broken out of ...
2484 (release_structures): ... here. Use it.
2485 (read_count_file): Adjust for new data file format.
2486 (output_lines): Use object_runs.
2487 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Obsolete.
2488 (struct gcov_ctr_info): Move definition.
2489 (struct gcov_fn_info): Add key field, use gcov_ctr_info for
2491 (struct gcov_info): Add merge function array, remove mask and
2492 counts. Trailing array is array of pointers to function info.
2493 * coverage.c (struct function_list): Replace counter numbers with
2494 counter arrays. Add fndecl field. GTYify.
2495 (counts_entry): Remove chain workspace.
2496 (functions_head): GTYify.
2497 (prg_n_ctrs): Remove.
2499 (tree_ctr_tables): Remove.
2500 (read_counts_file): Cope with blank entries and expect program
2501 summaries before functions. Don't warn on missing entries.
2502 (coverage_counter_alloc): Allocate individual function arrays.
2503 (tree_coverage_counter_ref, tree_coverage_counter_addr): Adjust
2504 for individual function arrays.
2505 (coverage_end_function): GTYify function list object. Finalize
2506 function's counter arrays.
2507 (build_var): New. Create a counter-related variable with
2508 appropriate linkage.
2509 (build_fn_info_type): Adjust for new runtime structure.
2510 (build_fn_info_value): Rename to ...
2511 (build_fn_info): ... here. Build new format data.
2512 (build_ctr_info_type, build_ctr_info_value): Remove.
2513 (build_info_type): New. Build new format data structure.
2514 (build_info): Adjust for new format data.
2515 (create_coverage): Likewise.
2516 * gcov-dump.c (tag_function): Recognize placeholders.
2518 2011-11-07 Georg-Johann Lay <avr@gjlay.de>
2520 * config/avr/constraints.md (Cm2): New constraint for int -2.
2521 * config/avr/avr.md (addqi3): Use it. New alternatives for +/-2.
2522 (*negqihi2): New insn.
2524 2011-11-07 H.J. Lu <hongjiu.lu@intel.com>
2526 * dwarf2cfi.c (dwarf2out_frame_debug_expr): Check
2527 HARD_FRAME_POINTER_REGNUM instead of hard_frame_pointer_rtx in Rule 18.
2529 2011-11-07 Jakub Jelinek <jakub@redhat.com>
2531 * config/i386/i386.c (ix86_expand_builtin): If gather mask
2532 argument is known to have all high bits set, pass pc_rtx as
2533 second argument to the expander instead of op0.
2534 * config/i386/sse.md (*avx2_gathersi<mode>_2,
2535 *avx2_gatherdi<mode>_2): New patterns.
2536 * config/i386/avx2intrin.h (_mm256_i32gather_pd,
2537 _mm256_i64gather_pd, _mm256_i32gather_ps): Set mask using
2538 _mm256_cmp_pd with zero vector arguments and _CMP_EQ_OQ instead of
2541 PR tree-optimization/50789
2542 * tree-vect-stmts.c (process_use): Add force argument, avoid
2543 exist_non_indexing_operands_for_use_p check if true.
2544 (vect_mark_stmts_to_be_vectorized): Adjust callers. Handle
2545 STMT_VINFO_GATHER_P.
2546 (gen_perm_mask): New function.
2547 (perm_mask_for_reverse): Use it.
2548 (reverse_vec_element): Rename to...
2549 (permute_vec_elements): ... this. Add Y and MASK_VEC arguments,
2550 generalize for any permutations.
2551 (vectorizable_load): Adjust caller. Handle STMT_VINFO_GATHER_P.
2552 * target.def (TARGET_VECTORIZE_BUILTIN_GATHER): New hook.
2553 * doc/tm.texi.in (TARGET_VECTORIZE_BUILTIN_GATHER): Document it.
2554 * doc/tm.texi: Regenerate.
2555 * tree-data-ref.c (initialize_data_dependence_relation,
2556 compute_self_dependence): No longer static.
2557 * tree-data-ref.h (initialize_data_dependence_relation,
2558 compute_self_dependence): New prototypes.
2559 * tree-vect-data-refs.c (vect_check_gather): New function.
2560 (vect_analyze_data_refs): Detect possible gather load data refs.
2561 * tree-vectorizer.h (struct _stmt_vec_info): Add gather_p field.
2562 (STMT_VINFO_GATHER_P): Define.
2563 (vect_check_gather): New prototype.
2564 * config/i386/i386-builtin-types.def: Add types for alternate
2566 * config/i386/sse.md (AVXMODE48P_DI): Remove.
2567 (VEC_GATHER_MODE): Rename mode_attr to...
2568 (VEC_GATHER_IDXSI): ... this.
2569 (VEC_GATHER_IDXDI, VEC_GATHER_SRCDI): New mode_attrs.
2570 (avx2_gathersi<mode>, *avx2_gathersi<mode>): Use <VEC_GATHER_IDXSI>
2571 instead of <VEC_GATHER_MODE>.
2572 (avx2_gatherdi<mode>): Use <VEC_GATHER_IDXDI> instead of
2573 <AVXMODE48P_DI> and <VEC_GATHER_SRCDI> instead of VEC_GATHER_MODE
2574 on src and mask operands.
2575 (*avx2_gatherdi<mode>): Likewise. Use VEC_GATHER_MODE iterator
2576 instead of AVXMODE48P_DI.
2577 (avx2_gatherdi<mode>256, *avx2_gatherdi<mode>256): Removed.
2578 * config/i386/i386.c (enum ix86_builtins): Add
2579 IX86_BUILTIN_GATHERALTSIV4DF, IX86_BUILTIN_GATHERALTDIV8SF,
2580 IX86_BUILTIN_GATHERALTSIV4DI and IX86_BUILTIN_GATHERALTDIV8SI.
2581 (ix86_init_mmx_sse_builtins): Create those builtins.
2582 (ix86_expand_builtin): Handle those builtins and adjust expansions
2583 of other gather builtins.
2584 (ix86_vectorize_builtin_gather): New function.
2585 (TARGET_VECTORIZE_BUILTIN_GATHER): Define.
2587 2011-11-07 Uros Bizjak <ubizjak@gmail.com>
2589 * config/i386/f16cintrin.h: Remove extra _X86INTRIN_H_INCLUDED check.
2591 2011-11-07 Tristan Gingold <gingold@adacore.com>
2593 * config.gcc (*-*-*vms*): Set c_target_objs, cxx_target_objs.
2594 * config/vms/t-vms: Add vms-c.o rule.
2595 * config/vms/vms-c.c: New file.
2596 * config/vms/vms-protos.h (vms_c_register_pragma): New prototype.
2597 * config/vms/vms.h (REGISTER_TARGET_PRAGMAS): Define.
2599 2011-11-07 Tristan Gingold <gingold@adacore.com>
2601 * config/alpha/vms.h (TARGET_OBJECT_SUFFIX,
2602 TARGET_EXECUTABLE_SUFFIX, TARGET_OS_CPP_BUILTINS,
2603 TARGET_ABI_OPEN_VMS, LONG_TYPE_SIZE, ADA_LONG_TYPE_SIZE,
2604 POINTER_SIZE, POINTERS_EXTEND_UNSIGNED): Move to config/vms/vms.h
2605 (SUBTARGET_OS_CPP_BUILTINS): Define.
2606 (TARGET_DEFAULT): Tune according to POINTER_SIZE.
2607 (MASK_RETURN_ADDR): Define in 64 bit mode.
2608 * config/ia64/vms.h: Likewise.
2609 * config/vms/vms.h: New file.
2610 * config/vms/vms64.h: New file.
2611 * config/alpha/vms64.h: Removed.
2612 * config/ia64/vms64.h: Removed.
2613 * config.gcc (alpha64-dec-*vms*, alpha*-dec-*vms*, ia64-hp-*vms*):
2614 Adjust for above change.
2616 2011-11-07 Enkovich Ilya <ilya.enkovich@intel.com>
2619 * config/i386/i386-protos.h (ix86_use_lea_for_mov): New.
2620 * config/i386/i386.c (ix86_use_lea_for_mov): Likewise.
2621 * config/i386/i386.md (movsi_internal): Emit lea if profitable.
2622 (movdi_internal_rex64): Likewise.
2624 2011-11-07 Sergey Ostanevich <sergos.gnu@gmail.com>
2626 PR rtl-optimization/47698
2627 * ifcvt.c (noce_operand_ok): Return false for mems with side effects.
2629 2011-11-07 Tristan Gingold <gingold@adacore.com>
2631 * common/config/alpha/alpha-common.c (alpha_option_init_struct):
2633 (TARGET_OPTION_INIT_STRUCT): Define.
2634 * config/alpha/alpha.c (TARGET_MIN_ANCHOR_OFFSET)
2635 (TARGET_MAX_ANCHOR_OFFSET)
2636 (TARGET_USE_BLOCKS_FOR_CONSTANT_P): Redefine.
2638 2011-11-06 Quentin Neill <quentin.neill@amd.com>
2640 Fix r180999, update ChangeLog
2641 * config.gcc: Add f16cintrin.h.
2642 * config/i386/f16cintrin.h: Add missing endif.
2644 2011-11-06 Sebastian Huber <sebastian.huber@embedded-brains.de>
2646 * config.gcc (arm*-*-rtemseabi*): New target.
2647 * config/arm/rtems-eabi.h: New.
2648 * config/arm/t-rtems-eabi: New.
2650 2011-11-06 David S. Miller <davem@davemloft.net>
2652 * config/sparc/sparc.md (UNSPEC_SHORT_LOAD): Delete.
2653 (zero_extend_v8qi_vis, zero_extend_v4hi_vis,
2654 *zero_extend_v8qi_<P:mode>_insn,
2655 *zero_extend_v4hi_<P:mode>_insn): Express using vec_merge
2656 and vec_duplicate instead of using an UNSPEC.
2658 2011-11-07 Alan Modra <amodra@gmail.com>
2661 * config/rs6000/rs6000.c (rs6000_emit_stack_reset): Always emit
2662 blockage for ABI_V4.
2664 2011-11-06 Dave Korn <dave.korn.cygwin@gmail.com>
2666 * config/i386/cygwin.h (LIBGCJ_SONAME): Updated to match recent
2667 libgcj version bump.
2668 * config/i386/mingw32.h (LIBGCJ_SONAME): Likewise.
2670 2011-11-06 Eric Botcazou <ebotcazou@adacore.com>
2672 * gcse.c: Adjust outdated comments throughout.
2673 (struct mem_conflict_info): New structure.
2674 (mems_conflict_for_gcse_p): Use it to communicate with caller.
2675 (load_killed_in_block_p): Pass it to note_stores.
2676 (hash_expr): Remove superfluous line break.
2677 (hash_scan_set): Rename PAT parameter into SET.
2678 (hash_scan_insn): Reorder cases.
2679 (canon_list_insert): Fix long line.
2680 (edge_list): Delete.
2681 (prune_expressions): Rename E local variable into EXPR.
2682 (compute_pre_data): Return struct edge_list * object.
2683 (pre_expr_reaches_here_p_work): Fix formatting.
2684 (process_insert_insn): Move around comment.
2685 (pre_edge_insert): Fix long line.
2686 (pre_insert_copies): Likewise.
2687 (gcse_emit_move_after): Swap SRC and DEST parameters.
2688 (pre_delete): Adjust call to gcse_emit_move_after.
2689 (pre_gcse): Take struct edge_list * parameter. Fix long line.
2690 (one_pre_gcse_pass): Use flag_gcse_lm condition for all routines.
2691 Use a local list of edges.
2692 (hoist_code): Fix long line. Adjust call to gcse_emit_move_after.
2693 (pre_ldst_expr_hash): Fix long line.
2694 (free_ldst_mems): Rename into...
2695 (free_ld_motion_mems): ...this.
2696 (first_ls_expr): Delete.
2697 (next_ls_expr): Likewise.
2698 (print_ldst_list): Do not use above two functions.
2699 (simple_mem): Adjust interface.
2700 (compute_ld_motion_mems): Fix formatting.
2701 (update_ld_motion_stores): Reuse local variable.
2703 2011-11-06 Joseph Myers <joseph@codesourcery.com>
2705 * c-decl.c (shadow_tag_warned, grokdeclarator): Handle _Alignas
2707 (build_null_declspecs): Initialize align_log and alignas_p fields.
2708 (declspecs_add_alignas): New.
2709 * c-parser.c (c_token_starts_declspecs): Handle RID_ALIGNAS.
2710 (c_parser_declspecs): Handle _Alignas specifiers.
2711 (c_parser_alignas_specifier): New.
2712 (c_parser_alignof_expression): Diagnose alignof use for non-C1X.
2713 Diagnose _Alignof (expression).
2714 * c-tree.h (struct c_declspecs): Add align_log and alignas_p fields.
2715 (declspecs_add_alignas): Declare.
2716 * ginclude/stddef.h (max_align_t): Define for C1X and C++11.
2717 * ginclude/stdalign.h: New.
2718 * Makefile.in (USER_H): Add stdalign.h.
2720 2011-11-06 Joern Rennecke <joern.rennecke@embecosm.com>
2722 * regset.h (fixed_reg_set_regset): Declare.
2723 * dse.c: Include regset.h .
2724 (struct insn_info): Add member fixed_regs_live.
2725 (note_add_store_info): New typedef.
2726 (note_add_store): New function.
2727 (emit_inc_dec_insn_before): Expect arg to be of type insn_info_t.
2728 Use gen_add3_insn / gen_move_insn.
2729 Check new insn for unwanted clobbers before emitting it.
2730 (check_for_inc_dec): Rename to...
2731 (check_for_inc_dec_1:) ... this. Return bool. Take insn_info
2732 parameter. Changed all callers in file.
2733 (check_for_inc_dec, copy_fixed_regs): New functions.
2734 (scan_insn): Set fixed_regs_live field of insn_info.
2735 * rtl.h (check_for_inc_dec): Update prototype.
2736 * postreload.c (reload_cse_simplify): Take new signature of
2737 check_ind_dec into account.
2738 * reginfo.c (fixed_reg_set_regset): New variable.
2739 (init_reg_sets_1): Initialize it.
2741 2011-11-06 Jakub Jelinek <jakub@redhat.com>
2743 * tree-cfg.c (gimple_can_merge_blocks_p): For -O0 don't remove
2746 2011-11-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
2748 * config/pa/pa.c (pa_hpux_init_libfuncs): Rename to pa_init_libfuncs.
2749 Remove dependence of declaration and target define on definition of
2750 HPUX_LONG_DOUBLE_LIBRARY. Update implementation.
2752 2011-11-06 Andrew Macleod <amacleod@redhat.com>
2753 Richard Henderson <rth@redhat.com>
2754 Aldy Hernandez <aldyh@redhat.com>
2756 Merged from cxx-mem-model.
2758 * cppbuiltin.c (define__GNUC__): Define __ATOMIC memory models
2759 * coretypes.h (enum memmodel): New. enumerated memory model type.
2760 * Makefile.in (cppbuiltin.o) Add missing dependency on $(TREE_H)
2761 * genopinit,c (optabs): Add atomic direct optab handlers.
2762 * sync-builtins.def (BUILT_IN_ATOMIC_*): New builtins.
2763 * builtin-types.def (BT_CONST_VOLATILE_PTR,
2764 BT_FN_I{1,2,4,8,16}_CONST_VPTR_INT, BT_FN_VOID_VPTR_INT,
2765 BT_FN_BOOL_VPTR_INT, BT_FN_BOOL_SIZE_CONST_VPTR,
2766 BT_FN_I{1,2,4,8,16}_VPTR_I{1,2,4,8,16}_INT,
2767 BT_FN_VOID_VPTR_I{1,2,4,8,16}_INT, BT_FN_VOID_SIZE_VPTR_PTR_INT,
2768 BT_FN_VOID_SIZE_CONST_VPTR_PTR_INT, BT_FN_VOID_SIZE_VPTR_PTR_PTR_INT,
2769 BT_FN_BOOL_VPTR_PTR_I{1,2,4,8,16}_BOOL_INT_INT): New builtin types.
2770 * expr.h (expand_atomic_*): Add prototypes.
2771 (expand_{bool,val}_compare_and_swap): Remove prototypes.
2772 * c-typeck.c (build_function_call_vec): Don't reprocess __atomic
2774 * common.opt (Winvalid-memory-model): New warning flag.
2775 (finline-atomics): New. Flag to disable atomic inlining.
2776 * params.h (ALLOW_LOAD_DATA_RACES): New.
2777 (ALLOW_PACKED_LOAD_DATA_RACES): New.
2778 (ALLOW_PACKED_STORE_DATA_RACES): New.
2779 * params.def (PARAM_ALLOW_LOAD_DATA_RACES): New.
2780 (PARAM_ALLOW_PACKED_LOAD_DATA_RACES): New.
2781 (PARAM_ALLOW_PACKED_STORE_DATA_RACES): New.
2782 * builtins.c (is_builtin_name): Handle __atomic.
2783 (get_memmodel): New. Extract memory model.
2784 (expand_expr_force_mode): New. Factor out common code for ensuring an
2785 integer argument is in the proper mode.
2786 (expand_builtin_sync_operation): Remove ignore param. Always call
2787 expand_atomic_fetch_op instead of the old expanders.
2788 (expand_builtin_compare_and_swap,
2789 expand_builtin_sync_lock_test_and_set): Use expand_expr_force_mode,
2790 call atomic expanders instead of sync expanders.
2791 (expand_builtin_sync_lock_release): Call atomic_store expander.
2792 (expand_builtin_atomic_compare_exchange, expand_builtin_atomic_load,
2793 expand_builtin_atomic_store, expand_builtin_atomic_fetch_op): New.
2794 (expand_builtin_atomic_exchange): New.
2795 (fold_builtin_atomic_always_lock_free,
2796 expand_builtin_atomic_always_lock_free,
2797 fold_builtin_atomic_is_lock_free, expand_builtin_atomic_is_lock_free):
2799 (expand_builtin_mem_thread_fence, expand_builtin_atomic_thread_fence,
2800 expand_builtin_atomic_signal_fence): New.
2801 (expand_builtin_mem_signal_fence): New.
2802 (expand_builtin): Add cases for BUILT_IN_ATOMIC_*.
2803 (fold_builtin_2): Add cases for BUILT_IN_ATOMIC_{IS,ALWAYS}_LOCK_FREE.
2804 * optabs.h (DOI_atomic_*): Define new atomics.
2805 (atomic_*_optab): Define.
2806 (can_compare_and_swap_p, expand_atomic_compare_and_swap): New
2808 * optabs.c (expand_sync_operation, expand_sync_fetch_operation): Remove.
2809 (expand_sync_lock_test_and_set): Remove.
2810 (expand_atomic_load, expand_atomic_store): New.
2811 (expand_atomic_exchange): New.
2812 (expand_atomic_compare_and_swap): New. Implements
2813 atomic_compare_exchange via compare and swap.
2814 (struct atomic_op_functions): Opcode table struct for fetch ops.
2815 (get_atomic_op_for_code): New. Return an opcode table entry.
2816 (maybe_emit_op): New. Try to emit a fetch op.
2817 (expand_atomic_fetch_op): New.
2818 (expand_val_compare_and_swap_1): Remove.
2819 (expand_val_compare_and_swap, expand_bool_compare_and_swap): Remove.
2820 (expand_atomic_compare_and_swap): Rename from
2821 expand_atomic_compare_exchange. Rewrite to return both success and
2822 oldval return values; expand via both atomic and sync optabs.
2823 (can_compare_and_swap_p): New.
2824 (expand_compare_and_swap_loop): Use expand_atomic_compare_and_swap.
2825 (maybe_gen_insn): Handle 7 and 8 operands.
2826 * omp-low.c (expand_omp_atomic_fetch_op): Don't test individual
2827 fetch_op optabs, only test can_compare_and_swap_p. Use __atomic
2828 builtins instead of __sync builtins.
2829 (expand_omp_atomic_pipeline): Use can_compare_and_swap_p.
2830 * doc/extend.texi: Document __atomic built-in functions.
2831 * doc/invoke.texi: Document data race parameters.
2832 * doc/md.texi: Document atomic patterns.
2833 * config/i386/i386.md (UNSPEC_MOVA): New.
2834 (UNSPECV_CMPXCHG): Split into ...
2835 (UNSPECV_CMPXCHG_1, UNSPECV_CMPXCHG_2,
2836 UNSPECV_CMPXCHG_3, UNSPECV_CMPXCHG_4): New.
2837 * config/i386/sync.md (ATOMIC): New mode iterator.
2838 (atomic_load<ATOMIC>, atomic_store<ATOMIC>): New.
2839 (atomic_loaddi_fpu, atomic_storedi_fpu, movdi_via_fpu): New.
2840 (mem_thread_fence): Rename from memory_barrier.
2841 Handle the added memory model parameter.
2842 (mfence_nosse): Rename from memory_barrier_nosse.
2843 (sync_compare_and_swap<CASMODE>): Split into ...
2844 (atomic_compare_and_swap<SWI124>): this and ...
2845 (atomic_compare_and_swap<CASMODE>): this. Handle the new parameters.
2846 (atomic_compare_and_swap_single<SWI>): Rename from
2847 sync_compare_and_swap<SWI>; rewrite to use split unspecs.
2848 (atomic_compare_and_swap_double<DCASMODE>): Rename from
2849 sync_double_compare_and_swap<DCASMODE>; rewrite to use split unspecs.
2850 (*atomic_compare_and_swap_doubledi_pic): Rename from
2851 sync_double_compare_and_swapdi_pic; rewrite to use split unspecs.
2852 (atomic_fetch_add<SWI>): Rename from sync_old_add<SWI>; add memory
2854 (*atomic_fetch_add_cmp<SWI>): Similarly.
2855 (atomic_add<SWI>, atomic<any_logic><SWI>): Similarly.
2856 (atomic_sub<SWI>): Similarly. Use x86_maybe_negate_const_int.
2857 (sync_lock_test_and_set<SWI>): Merge with ...
2858 (atomic_exchange<SWI>): ... this.
2860 2011-11-6 Richard Guenther <rguenther@suse.de>
2862 * ipa-prop.c (ipa_modify_call_arguments): Re-compute inlinable flag.
2864 2011-11-06 Ira Rosen <ira.rosen@linaro.org>
2866 * tree-vectorizer.h (vectorizable_condition): Add argument.
2867 * tree-vect-loop.c (vectorizable_reduction): Fail for condition
2868 in SLP. Update calls to vectorizable_condition.
2869 * tree-vect-stmts.c (vect_is_simple_cond): Add basic block info to
2870 the arguments. Pass it to vect_is_simple_use_1.
2871 (vectorizable_condition): Add slp_node to the arguments. Support
2872 vectorization of basic blocks. Fail for reduction in SLP. Update
2873 calls to vect_is_simple_cond and vect_is_simple_use. Support SLP:
2874 call vect_get_slp_defs to get vector operands.
2875 (vect_analyze_stmt): Update calls to vectorizable_condition.
2876 (vect_transform_stmt): Likewise.
2877 * tree-vect-slp.c (vect_create_new_slp_node): Handle COND_EXPR.
2878 (vect_get_and_check_slp_defs): Handle COND_EXPR. Allow pattern
2880 (vect_build_slp_tree): Handle COND_EXPR.
2881 (vect_analyze_slp_instance): Push pattern statements to root node.
2882 (vect_get_constant_vectors): Fix comments. Handle COND_EXPR.
2884 2011-11-05 David S. Miller <davem@davemloft.net>
2886 * config/sparc/sparc.md (UNSPEC_SHORT_LOAD): New unspec.
2887 (zero-extend_v8qi_vis, zero_extend_v4hi_vis): New expanders.
2888 (*zero_extend_v8qi_<P:mode>_insn,
2889 *zero_extend_v4hi_<P:mode>_insn): New insns.
2890 * config/sparc/sparc.c (vector_init_move_words,
2891 vector_init_prepare_elts, sparc_expand_vector_init_vis2,
2892 sparc_expand_vector_init_vis1): New functions.
2893 (vector_init_bshuffle): Rewrite to handle more cases and make use
2894 of locs[] array prepared by vector_init_prepare_elts.
2895 (vector_init_fpmerge, vector_init_faligndata): Delete.
2896 (sparc_expand_vector_init): Rewrite using new infrastructure.
2898 2011-11-05 Joern Rennecke <joern.rennecke@embecosm.com>
2900 * config.gcc (epiphany-*-*): New architecture.
2901 (epiphany-*-elf): New configuration.
2902 * config/epiphany, common/config/epiphany : New directories.
2903 * doc/extend.texi (disinterrupt attribute): Add Epiphany.
2904 (interrupt attribute): Add Epiphany.
2905 (long_call, short_call attribute): Add Epiphany.
2906 * doc/invoke.texi (Options): Add Epiphany options.
2907 * doc/md.texi (Machine Constraints): Add Epiphany constraints.
2908 * doc/install.texi (Options specification):
2909 Add --with-stack-offset=@var{num} description.
2910 (host/target specific issues): Add epiphany-*-elf.
2911 * doc/contrib.texi (Contributors): Mention Epiphany port.
2913 2011-11-05 Jakub Jelinek <jakub@redhat.com>
2915 PR tree-optimization/50693
2916 * tree-cfg.c (gimple_can_merge_blocks_p): Allow merging with
2917 non-forced user labels.
2918 (gimple_merge_blocks): Turn non-forced user labels into
2919 debug bind stmt with the label as first operand and reset value.
2920 (gimple_duplicate_bb): Don't duplicate label debug stmts.
2921 * dwarf2out.c (gen_label_die): Handle NOTE_INSN_DELETED_DEBUG_LABEL.
2922 * final.c (final_scan_insn): Likewise.
2923 (rest_of_clean_state): Don't dump NOTE_INSN_DELETED_DEBUG_LABEL.
2924 * var-tracking.c (debug_label_num): New variable.
2925 (delete_debug_insns): Don't delete DEBUG_INSNs for LABEL_DECLs,
2926 instead turn them into NOTE_INSN_DELETED_DEBUG_LABEL notes.
2927 * cfglayout.c (skip_insns_after_block, duplicate_insn_chain): Handle
2928 NOTE_INSN_DELETED_DEBUG_LABEL.
2929 (duplicate_insn_chain): Don't duplicate LABEL_DECL DEBUG_INSNs.
2930 * insn-notes.def (DELETED_DEBUG_LABEL): New note kind.
2931 * print-rtl.c (print_rtx): Handle NOTE_INSN_DELETED_DEBUG_LABEL.
2932 * gengtype.c (adjust_field_rtx_def): Likewise.
2933 * config/i386/i386.c (ix86_output_function_epilogue): For MachO
2934 clear CODE_LABEL_NUMBER of NOTE_INSN_DELETED_DEBUG_LABEL
2935 if their are at the end of function and nop hasn't been emitted.
2936 * config/rs6000/rs6000.c (rs6000_output_function_epilogue): Likewise.
2938 2011-11-05 Georg-Johann Lay <avr@gjlay.de>
2940 PR rtl-optimization/50448
2941 * cprop.c (try_replace_reg): Also try to replace uses of FROM that
2944 2011-11-05 Peter Dufault <dufault@hda.com>,
2945 Sebastian Huber <sebastian.huber@embedded-brains.de>
2947 * config/rs6000/rtems.h (SUBSUBTARGET_OVERRIDE_OPTIONS):
2948 Extend rs6000_spe handling.
2950 2011-11-05 Ralf Corsépius <ralf.corsepius@rtems.org>
2952 * config/rs6000/t-rtems: Add -mcpu=8540/-mfloat-gprs=double multilib.
2953 Remove -mcpu=601 multilib.
2954 Remove -Dmpc8260 multilib.
2955 * config/rs6000/rtems.h: Allow --float-gprs=... to override grps
2958 2011-11-05 Quentin Neill <quentin.neill@amd.com>
2960 Piledriver f16cintrin.h fix.
2961 * config/i386/f16cintrin.h: Contents moved from immintrin.h.
2962 * config/i386/immintrin.h: Include f16cintrin.h.
2963 * config.gcc (i[34567]86-*-*, x86_64-*-*): Add f16cintrin.h.
2966 2011-11-04 Eric Botcazou <ebotcazou@adacore.com>
2969 * c-parser.c (c_parser_postfix_expression) <RID_OFFSETOF>: Adjust call
2971 * c-typeck.c (build_unary_op) <ADDR_EXPR>: Call fold_offsetof_1.
2973 2011-11-04 Alan Modra <amodra@gmail.com>
2975 * reload1.c (gen_reload): Don't use REGNO on SUBREGs.
2976 * print-rtl.c (print_rtx): Don't segfault on negative regno.
2978 2011-11-04 David S. Miller <davem@davemloft.net>
2981 * config/sparc/sparc.c (sparc_expand_conditional_move): Handle the
2982 fact that sparc_emit_float_lib_cmp modifies the comparison in
2985 2011-11-04 Ralf Corsépius <ralf.corsepius@rtems.org>
2987 * config/lm32/t-rtems: New.
2988 * config.gcc (lm32-*-rtems*): Add t-rtems.
2990 2011-11-04 Eric Botcazou <ebotcazou@adacore.com>
2993 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=v8.
2995 2011-11-04 Jiangning Liu <jiangning.liu@arm.com>
2997 PR rtl-optimization/38644
2998 * config/arm/arm.c (thumb1_expand_epilogue): Add memory barrier
2999 for epilogue having stack adjustment.
3001 2011-11-04 Georg-Johann Lay <avr@gjlay.de>
3004 * config/avr/avr-modes.def: New file defining PSImode.
3005 * config/avr/avr-c.c (__INT24_MAX__, __INT24_MIN__,
3006 __UINT24_MAX__): New built-in defines.
3007 * config/avr/avr.md (adjust_len): Add tstpsi, mov24, reload_in24,
3008 ashlpsi, ashrpsi, lshrpsi.
3009 (QISI, QIDI, HISI, HIDI, MPUSH, rotx, rotsmode): Add PSI.
3010 (MOVMODE): New mode iterator.
3011 (movpsi): New expander.
3012 (movqi, movhi, movsi, movsf, movpsi): Write as one using MOVMODE.
3013 (*reload_inpsi, *movpsi): New insns.
3014 (*reload_inpsi): New RTL peephole.
3015 (addpsi3, *addpsi3_zero_extend.qi, *addpsi3_zero_extend.hi,
3016 *addpsi3_sign_extend.hi): New insns.
3017 (subpsi3, *subpsi3_zero_extend.qi, *subpsi3_zero_extend.hi,
3018 *subpsi3_sign_extend.hi): New insns.
3019 (divmodpsi4, udivmodpsi4): New define insn-and-split.
3020 (*divmodpsi4_call, *udivmodpsi4_call): New insns.
3021 (andpsi3, iorpsi3, xorpsi3): New insns.
3022 (*rotlpsi2.1, *rotlpsi2.23): New insns.
3023 (*rotw<mode>): Insn condition only allow even-sized modes.
3024 (*rotb<mode>): Insn condition allows odd-sized modes.
3025 (ashlpsi3, ashrpsi3, lshrpsi3, *addpsi3.lt0): New insns.
3026 (negpsi2, one_cmplpsi2): New insns.
3027 (extendqipsi2, extendhipsi2, extendpsisi2): New insns.
3028 (zero_extendqipsi2, zero_extendhipsi2, zero_extendpsisi2): New
3030 (*cmppsi, *negated_tstpsi, *reversed_tstpsi): New insns.
3031 (cbranchpsi4): New expander.
3032 * config/avr/constraints.md (Ca3, Co3, Cx3): New constraints.
3033 * config/avr/avr-protos.h (avr_out_tstpsi, avr_out_movpsi,
3034 avr_out_ashlpsi3, avr_out_ashrpsi3, avr_out_lshrpsi3,
3035 avr_out_reload_inpsi): New prototypes.
3037 * config/avr/avr.c (TARGET_SCALAR_MODE_SUPPORTED_P): Define to...
3038 (avr_scalar_mode_supported_p): ...this new static function.
3039 (avr_asm_len): Always return "".
3040 (avr_out_load_psi, avr_out_store_psi): New static functions.
3041 (avr_out_movpsi, avr_out_reload_inpsi): New functions.
3042 (avr_out_tstpsi): New function.
3043 (avr_out_ashlpsi3, avr_out_ashrpsi3, avr_out_lshrpsi3): New functions.
3044 (avr_out_plus_1, output_reload_in_const): Handle 3-byte types.
3045 (avr_simplify_comparison_p): Ditto.
3046 (adjust_insn_length): Handle ADJUST_LEN_RELOAD_IN24,
3047 ADJUST_LEN_MOV24, ADJUST_LEN_TSTPSI, ADJUST_LEN_ASHLPSI,
3048 ADJUST_LEN_ASHRPSI, ADJUST_LEN_LSHRPSI.
3049 (avr_rtx_costs_1): Report PSI costs.
3050 (avr_libcall_value): Handle odd-sized parameters.
3051 (avr_init_builtin_int24): New static function to define built-in
3052 24-bit types __int24 and __uint24.
3053 (avr_init_builtins): Use it.
3055 2011-11-04 Thomas Doerfler <thomas.doerfler@embedded-brains.de>
3058 * config/arm/rtems-elf.h, config/arm/t-rtems: Add optional
3059 support for VFP floating point model.
3061 2011-11-04 Tristan Gingold <gingold@adacore.com>
3063 * config/alpha/vms.h (ASM_OUTPUT_DEF): Do not switch section.
3065 2011-11-04 Ira Rosen <ira.rosen@linaro.org>
3068 2011-10-24 Ira Rosen <ira.rosen@linaro.org>
3070 PR tree-optimization/50730
3071 * tree-vect-data-refs.c (vect_analyze_data_refs): Stop basic block
3072 analysis if encountered unsupported data-ref.
3074 2011-11-04 Jakub Jelinek <jakub@redhat.com>
3076 * config/i386/i386.c (ix86_expand_vector_convert_uns_vsivsf): New
3078 * config/i386/i386-protos.h (ix86_expand_vector_convert_uns_vsivsf):
3080 * config/i386/sse.md (floatuns<sseintvecmodelower><mode>2): Use it.
3081 For floatunsv8siv8sf2 require TARGET_AVX2.
3083 * config/i386/i386.c (ix86_expand_adjust_ufix_to_sfix_si): Add
3084 XORP argument. Subtract 0x1p31 instead of 0x1p32. Use normal
3085 signalling comparison instead of non-signalling. Store into
3086 *XORP pseudo holding 0x80000000 integers if 0x1p31 has been
3087 subtracted and 0 otherwise.
3088 * config/i386/i386-protos.h (ix86_expand_adjust_ufix_to_sfix_si):
3090 * config/i386/sse.md (fixuns_trunc<mode><sseintvecmodelower>2): Enable
3091 already for TARGET_SSE2. Xor in vector initialized by
3092 ix86_expand_adjust_ufix_to_sfix_si at the end.
3093 (vec_pack_ufix_trunc_<mode>): Likewise.
3095 * tree-vect-stmts.c (vectorizable_conversion): Rewritten to handle
3096 not just FLOAT_EXPR and FIX_TRUNC_EXPR, but also CONVERT_EXPR_CODE_P,
3097 WIDEN_MULT_EXPR and WIDEN_LSHIFT_EXPR to handle what
3098 vectorizable_type_demotion and vectorizable_type_promotion did.
3099 Additionally handle FLOAT_EXPR and FIX_TRUNC_EXPR where the integer
3100 is {,un}signed {char,short}.
3101 (vect_create_vectorized_demotion_stmts): Fix comment typo. For
3102 recursive calls unconditionally use VEC_PACK_TRUNC_EXPR.
3103 Push vec_dest back to the vec_dsts vector at the end.
3104 (vect_create_vectorized_promotion_stmts): Don't recurse, do just
3105 one step. Removed multi_step_cvt, vec_dsts, slp_node and
3106 prev_stmt_info arguments, add vec_dest argument. Push always
3107 into vec_tmp, not just when multi_step_cvt != 0, replace *vec_oprdn0
3108 with vec_tmp at the end after freeing old *vec_oprnd0 vector.
3109 (vectorizable_type_demotion, vectorizable_type_promotion): Removed.
3110 (vect_analyze_stmt): Don't call vectorizable_type_demotion and
3111 vectorizable_type_promotion. Call vectorizable_conversion even
3112 for SLP bb vectorization.
3113 (vect_transform_stmt): Call vectorizable_conversion instead of
3114 vectorizable_type_demotion and vectorizable_type_promotion.
3115 (supportable_widening_operation): Clear *multi_step_cvt first,
3116 simplify c1/c2 computation, free *interm_types vector on failure.
3117 (supportable_narrowing_operation): Clear *multi_step_cvt first,
3118 free *interm_types vector on failure, handle multi-step
3121 2011-11-04 Tristan Gingold <gingold@adacore.com>
3123 * config/alpha/alpha.c (alpha_write_linkage): Remove fundecl
3124 argument. Conditionally generate crash debug info. Adjust
3125 for alpha_funcs_tree removal.
3126 (machine_function): Add links field.
3127 (alpha_start_function): Conditionally generate crash debug info.
3128 (alpha_end_function): Adjust call to alpha_write_linkage.
3129 (alpha_funcs): Remove.
3130 (links_kind): Remove.
3131 (alpha_links): Remove num, target and lkind field. Add func field.
3132 (alpha_links_tree): Remove.
3133 (alpha_funcs_tree): Remove.
3134 (alpha_need_linkage): Remove.
3135 (alpha_use_linkage): Change prototype. Adjust.
3136 (alpha_write_one_linkage): Use ASM_OUTPUT_INTERNAL_LABEL.
3137 Use SYMBOL_REF_EXTERNAL_P and SYMBOL_REF_LOCAL_P macro.
3138 * config/alpha/alpha-protos.h (alpha_use_linkage): Update.
3139 (alpha_need_linkage): Remove.
3140 * config/alpha/alpha.md: Update calls to alpha_use_linkage.
3141 Adjust calls to alpha_need_linkage.
3143 2011-11-03 Uros Bizjak <ubizjak@gmail.com>
3145 * sched-vis.c (print_value): Handle STRICT_LOW_PART.
3147 2011-11-03 Uros Bizjak <ubizjak@gmail.com>
3149 * config/i386/i386.md (lround<X87MODEF:mode><SWI248x:mode>2,
3150 rint<mode>2, floor<mode>2, lfloor<MODEF:mode><SWI48:mode>2,
3151 btrunc<mode>2, lwp_lwpval<mode>3): Use operands[N] instead of operandN.
3153 2011-11-03 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3156 * config/arm/t-bpabi: New file.
3157 * config.gcc (arm*-*-linux*): Add arm/t-bpabi to tmake_file for
3159 (arm*-*-uclinux*): Add arm/t-bpabi to tmake_file for
3160 arm*-*-uclinux*eabi.
3161 (arm*-*-eabi*, arm*-*-symbianelf*): Add arm/t-bpabi to tmake_file
3164 2011-11-03 Michael Matz <matz@suse.de>
3167 * configure.ac: Check for -fno-exceptions -fno-rtti.
3168 * configure: Regenerate.
3169 * Makefile.in (NOEXCEPTION_FLAGS): New flag.
3170 (ALL_CXXFLAGS): Use it.
3172 2011-11-03 Uros Bizjak <ubizjak@gmail.com>
3174 * config/i386/i386.md: Use {} for multi-line preparation statements.
3176 2011-11-03 Eric Botcazou <ebotcazou@adacore.com>
3178 * config/sparc/sparc.md (movtf_insn_sp32_no_fpu): Consolidate into...
3179 (movtf_insn_sp32): ...this.
3180 (movtf_insn_sp64_no_fpu): Consolidate into...
3181 (movtf_insn_sp64): ...this.
3182 (movtf_insn_sp64_hq): Do not test TARGET_FPU.
3183 * config/sparc/sparc.c (sparc_legitimate_address_p): Likewise.
3185 2011-11-03 Tristan Gingold <gingold@adacore.com>
3187 * config/vms/vms.c (vms_patch_builtins): Fix typo.
3189 2011-11-03 Richard Guenther <rguenther@suse.de>
3192 * lto-opts.c: Re-implement.
3193 * lto-streamer.h (lto_register_user_option): Remove.
3194 (lto_read_file_options): Likewise.
3195 (lto_reissue_options): Likewise.
3196 (lto_clear_user_options): Likewise.
3197 (lto_clear_file_options): Likewise.
3198 * opts-global.c (post_handling_callback): Remove.
3199 (set_default_handlers): Do not set post_handling_callback.
3200 (decode_options): Remove LTO specific code.
3201 * lto-wrapper.c (merge_and_complain): New function.
3202 (run_gcc): Read all input file options and
3203 prepend a merged set before the linker driver options.
3204 * gcc.c (driver_post_handling_callback): Remove.
3205 (set_option_handlers): Do not set post_handling_callback.
3206 * opts-common.c (handle_option): Do not call post_handling_callback.
3207 * opts.h (struct cl_option_handlers): Remove post_handling_callback.
3209 2011-11-03 Richard Guenther <rguenther@suse.de>
3211 * collect2.c (main): Guard object_nbr variable with TARGET_AIX_VERSION.
3213 2011-11-03 Martin Jambor <mjambor@suse.cz>
3215 * ipa-prop.c (type_change_info): New fields offset, object,
3216 known_current_type and multiple_types_encountered.
3217 (extr_type_from_vtbl_ptr_store): New function.
3218 (check_stmt_for_type_change): Use it, set multiple_types_encountered if
3219 the result is different from the previous one.
3220 (detect_type_change): Renamed to detect_type_change_1. New parameter
3221 comp_type. Set up new fields in tci, build known type jump
3222 functions if the new type can be identified.
3223 (detect_type_change): New function.
3224 * tree.h (DECL_CONTEXT): Comment new use.
3226 2011-11-03 Richard Guenther <rguenther@suse.de>
3229 * lto-wrapper.c (get_options_from_collect_gcc_options): Properly
3230 decode an encoded literal '.
3232 2011-11-03 Tristan Gingold <gingold@adacore.com>
3234 * collect2.c (main): Add support of -f (response file) on AIX.
3236 2011-11-03 Ira Rosen <ira.rosen@linaro.org>
3238 PR tree-optimization/50912
3239 * tree-vectorizer.h (slp_void_p): New.
3240 (struct _slp_tree): Replace left and right with children. Update
3242 (struct _slp_oprnd_info): New.
3243 (vect_get_vec_defs): Declare.
3244 (vect_get_slp_defs): Update arguments.
3245 * tree-vect-loop.c (vect_create_epilog_for_reduction): Call
3246 vect_get_vec_defs instead of vect_get_slp_defs.
3247 (vectorizable_reduction): Likewise.
3248 * tree-vect-stmts.c (vect_get_vec_defs): Remove static, add argument.
3249 Update call to vect_get_slp_defs.
3250 (vectorizable_conversion): Update call to vect_get_vec_defs.
3251 (vectorizable_assignment, vectorizable_shift,
3252 vectorizable_operation): Likewise.
3253 (vectorizable_type_demotion): Call vect_get_vec_defs instead of
3255 (vectorizable_type_promotion, vectorizable_store): Likewise.
3256 (vect_analyze_stmt): Fix typo.
3257 * tree-vect-slp.c (vect_free_slp_tree): Update SLP tree traversal.
3258 (vect_print_slp_tree, vect_mark_slp_stmts,
3259 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
3260 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
3261 vect_schedule_slp_instance): Likewise.
3262 (vect_create_new_slp_node): New.
3263 (vect_create_oprnd_info, vect_free_oprnd_info): Likewise.
3264 (vect_get_and_check_slp_defs): Pass information about defs using
3265 oprnds_info, allow any number of operands.
3266 (vect_build_slp_tree): Likewise. Update calls to
3267 vect_get_and_check_slp_defs. Fix comments.
3268 (vect_analyze_slp_instance): Move node creation to
3269 vect_create_new_slp_node.
3270 (vect_get_slp_defs): Allow any number of operands.
3272 2011-11-02 Peter Bergner <bergner@vnet.ibm.com>
3273 Iain Sandoe <iains@gcc.gnu.org>
3275 * config/rs6000/rs6000.c (USE_HIDDEN_LINKONCE): New define.
3276 (get_ppc476_thunk_name): Use it.
3277 (rs6000_code_end): Likewise.
3278 (macho_branch_islands): Fix typo.
3280 2011-11-02 Paolo Carlini <paolo.carlini@oracle.com>
3281 Jason Merrill <jason@redhat.com>
3284 * configure.ac: Add -Wno-narrowing to warning options.
3285 * doc/invoke.texi ([-Wnarrowing], [-Wc++0x-compat]): Update.
3287 2011-11-02 Eric Botcazou <ebotcazou@adacore.com>
3290 * config/sparc/sparc.md (movsf_insn): Reindent constraints.
3291 (movdf_insn_sp32): Likewise. Remove redundant G constraint.
3292 (movdf_insn_sp64): Likewise.
3293 (DFmode splitter): Do not test TARGET_FPU.
3294 (movtf_insn_sp32): Reindent constraints.
3295 (movtf_insn_sp32_no_fpu): Likewise.
3296 (movtf_insn_sp64): Likewise.
3297 (movtf_insn_sp64_hq): Likewise.
3298 (movtf_insn_sp64_no_fpu): Likewise.
3300 2011-11-02 Paolo Carlini <paolo.carlini@oracle.com>
3303 * builtins.c (fold_builtin_memchr): Fix cast.
3305 2011-11-02 Teresa Johnson <tejohnson@google.com>
3307 * config/i386/predicates.md (promotable_binary_operator): Add minus
3308 to the list of promotable operators.
3310 2011-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3312 * gthr-single.h, gthr.h: Move to ../libgcc.
3313 * gthr-aix.h: Move to ../libgcc/config/rs6000.
3314 * gthr-dce.h: Move to ../libgcc/config/pa.
3315 * gthr-lynx.h: Move to ../libgcc/config.
3316 * gthr-mipssde.h: Move to ../libgcc/config/mips.
3317 * gthr-posix.h: Move to ../libgcc/config.
3318 * gthr-rtems.h: Likewise.
3319 * gthr-tpf.h: Move to ../libgcc/config/s390.
3320 * gthr-vxworks.h: Move to ../libgcc/config.
3321 * gthr-win32.h: Move to ../libgcc/config/i386.
3322 * configure.ac (gthread_flags): Remove
3323 (gthr-default.h): Don't create.
3324 (thread_file): Don't substitute.
3325 * configure: Regenerate.
3326 * Makefile.in (GCC_THREAD_FILE): Remove.
3327 (GTHREAD_FLAGS): Remove.
3328 (libgcc.mvars): Remove GTHREAD_FLAGS.
3329 * config/t-vxworks (EXTRA_HEADERS): Remove.
3331 2011-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3332 Paolo Bonzini <bonzini@gnu.org>
3334 * configure.ac (libgcc_tm_file_list, libgcc_tm_include_list): Remove.
3335 * configure: Regenerate.
3336 * Makefile.in (libgcc_tm_file_list, libgcc_tm_include_list): Remove.
3337 (TM_H): Remove libgcc_tm.h, $(libgcc_tm_file_list).
3338 (libgcc_tm.h, cs-libgcc_tm.h): Remove.
3339 (clean): Remove libgcc_tm.h
3340 * mkconfig.sh: Don't include libgcc_tm.h in tm.h.
3341 * config.gcc (libgcc_tm_file): Remove.
3342 (arm*-*-linux*): Remove libgcc_tm_file for arm*-*-linux-*eabi.
3343 (arm*-*-uclinux*): Remove libgcc_tm_file for arm*-*-uclinux*eabi.
3344 (arm*-*-eabi*, arm*-*-symbianelf*): Remove libgcc_tm_file.
3345 (avr-*-rtems*): Likewise.
3346 (avr-*-*): Likewise.
3347 (frv-*-elf): Likewise.
3348 (frv-*-*linux*): Likewise.
3349 (h8300-*-rtems*): Likewise.
3350 (h8300-*-elf*): Likewise.
3351 (i[34567]86-*-darwin*): Likewise.
3352 (x86_64-*-darwin*): Likewise.
3353 (rx-*-elf*): Likewise.
3354 (tic6x-*-elf): Likewise.
3355 (tic6x-*-uclinux): Likewise.
3356 (i[34567]86-*-linux*, x86_64-*-linux*): Likewise.
3358 2011-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3360 * Makefile.in (LIBGCC2_DEBUG_CFLAGS LIBGCC2_CFLAGS)
3361 (LIBGCC2_INCLUDES, TARGET_LIBGCC2_CFLAGS, LIB2FUNCS_EXTRA)
3362 (LIB2FUNCS_STATIC_EXTRA, LIB2FUNCS_EXCLUDE, T, T_TARGET)
3363 (INCLUDES_FOR_TARGET): Remove.
3364 (LIBGCC2_CFLAGS): Don't export.
3365 (LIB2FUNCS_ST, LIB2_DIVMOD_FUNCS, LIB2ADD, LIB2ADD_ST, srcdirify):
3367 (libgcc-support): Remove $(LIB2ADD), $(LIB2ADD_ST) dependencies.
3368 (libgcc.mvars): Likewise.
3369 Don't emit LIB2FUNCS_ST, LIB2FUNCS_EXCLUDE, LIB2ADD, LIB2ADD_ST,
3370 LIB2_SIDITI_CONV_FUNCS, LIB2_DIVMOD_FUNCS, LIBGCC2_CFLAGS,
3371 TARGET_LIBGCC2_CFLAGS.
3373 * libgcc2.c, libgcc2.h, gbl-ctors.h, longlong.h: Move to ../libgcc.
3374 * config/darwin-64.c: Move to ../libgcc/config.
3375 * config/divmod.c, config/floatunsidf.c, config/floatunsisf.c,
3376 config/floatunsitf.c, config/floatunsixf.c, config/udivmod.c,
3377 config/udivmodsi4.c: Move to ../libgcc/config.
3378 * config/gthr-posix.c: Move to ../libgcc/config/alpha.
3379 * config/memcmp.c, config/memcpy.c, config/memmove.c,
3380 config/memset.c: Move to ../libgcc/config.
3381 * config/t-darwin (TARGET_LIBGCC2_CFLAGS): Remove.
3382 * config/t-freebsd: Remove.
3383 * config/t-freebsd-thread: Move to ../libgcc/config.
3384 * config/t-libgcc-pic: Move to ../libgcc/config.
3385 * config/t-libunwind (TARGET_LIBGCC2_CFLAGS): Remove.
3386 * config/t-linux: Remove.
3387 * config/t-lynx (TARGET_LIBGCC2_CFLAGS, LIBGCC, INSTALL_LIBGCC):
3389 * config/t-openbsd-thread: Move to ../libgcc/config.
3390 * config/t-rtems (LIBGCC2_INCLUDES): Remove.
3391 * config/t-sol2 (TARGET_LIBGCC2_CFLAGS): Remove.
3392 * config/t-svr4: Remove.
3393 * config/t-vxworks (LIBGCC, INSTALL_LIBGCC, TARGET_LIBGCC2_CFLAGS)
3394 (LIBGCC2_DEBUG_CFLAGS, LIB2FUNCS_EXTRA, LIBGCC2_INCLUDES): Remove.
3395 * config/vxlib.c, config/vxlib-tls.c: Move to ../libgcc/config.
3396 * config/alpha/qrnnd.asm: Move to ../libgcc/config/alpha/qrnnd.S.
3397 * config/alpha/t-alpha, config/alpha/t-ieee: Remove.
3398 * config/alpha/t-vms (LIB2FUNCS_EXTRA, LIBGCC, INSTALL_LIBGCC): Remove.
3399 * config/alpha/vms-gcc_shell_handler.c: Move to ../libgcc/config/alpha.
3400 * config/arm/bpabi.c, config/arm/unaligned-funcs.c,
3401 config/arm/fp16.c, config/arm/linux-atomic.c,
3402 config/arm/linux-atomic-64bit.c: Move to ../libgcc/config/arm.
3403 * config/arm/t-arm-elf (LIBGCC, INSTALL_LIBGCC)
3404 (TARGET_LIBGCC2_CFLAGS): Remove.
3405 * config/arm/t-bpabi, config/arm/t-linux: Remove.
3406 * config/arm/t-linux-eabi (TARGET_LIBGCC2_CFLAGS)
3407 (LIB2FUNCS_STATIC_EXTRA): Remove.
3408 * config/arm/t-netbsd: Remove.
3409 * config/arm/t-strongarm-elf (LIBGCC, INSTALL_LIBGCC)
3410 (TARGET_LIBGCC2_CFLAGS): Remove.
3411 * config/arm/t-symbian (LIB2FUNCS_STATIC_EXTRA): Remove.
3412 * config/arm/t-wince-pe (LIBGCC, INSTALL_LIBGCC)
3413 (TARGET_LIBGCC2_CFLAGS): Remove.
3414 * config/avr/t-avr (LIB2FUNCS_EXCLUDE, TARGET_LIBGCC2_CFLAGS)
3415 (LIBGCC, INSTALL_LIBGCC): Remove.
3416 * config/bfin/t-bfin-elf (TARGET_LIBGCC2_CFLAGS): Remove.
3417 * config/bfin/t-bfin-linux: Likewise.
3418 * config/bfin/t-bfin-uclinux: Likewise.
3419 * config/c6x/eqd.c, config/c6x/eqf.c, config/c6x/ged.c,
3420 config/c6x/gef.c, config/c6x/gtd.c, config/c6x/gtf.c,
3421 config/c6x/led.c, config/c6x/lef.c, config/c6x/ltd.c,
3422 config/c6x/ltf.c: Move to ../libgcc/config/c6x.
3423 * config/c6x/t-c6x-elf (LIB2FUNCS_EXCLUDE, LIB2FUNCS_EXTRA): Remove.
3424 * config/c6x/t-c6x-uclinux (TARGET_LIBGCC2_CFLAGS): Remove.
3425 * config/cris/arit.c: Move to ../libgcc/config/cris.
3426 * config/cris/cris_abi_symbol.c: Remove.
3427 * config/cris/cris.h: Remove obsolete comment.
3428 * config/cris/mulsi3.asm: Move to ../libgcc/config/cris/mulsi3.S.
3429 * config/cris/t-cris (LIB2FUNCS_EXTRA, CRIS_LIB1CSRC)
3430 ($(LIB2FUNCS_EXTRA)): Remove.
3431 * config/cris/t-elfmulti (LIB2FUNCS_STATIC_EXTRA, INSTALL_LIBGCC)
3433 * config/cris/t-linux (TARGET_LIBGCC2_CFLAGS): Remove.
3434 * config/fr30/t-fr30: Remove.
3435 * config/frv/cmovd.c, config/frv/cmovh.c, config/frv/cmovw.c,
3436 config/frv/modi.c, config/frv/uitod.c, config/frv/uitof.c,
3437 config/frv/ulltod.c, config/frv/ulltof.c, config/frv/umodi.c: Move
3438 to ../libgcc/config/frv.
3439 * config/frv/t-frv (LIB2FUNCS_EXTRA, TARGET_LIBGCC2_CFLAGS)
3440 (cmovh.c, cmovw.c, cmovd.c, modi.c, umodi.c, uitof.c, uitod.c)
3441 (ulltof.c, LIBGCC, INSTALL_LIBGCC): Remove.
3442 * config/frv/t-linux (TARGET_LIBGCC2_CFLAGS): Remove.
3443 * config/h8300/clzhi2.c, config/h8300/ctzhi2.c,
3444 config/h8300/fixunssfsi.c, config/h8300/parityhi2.c,
3445 config/h8300/popcounthi2.c: Move to ../libgcc/config/h8300.
3446 * config/h8300/t-h8300 (LIB2FUNCS_EXTRA, TARGET_LIBGCC2_CFLAGS)
3447 (LIBGCC, INSTALL_LIBGCC): Remove.
3448 * config/i386/gthr-win32.c: Move to ../libgcc/config/i386.
3449 * config/i386/t-cygming (LIBGCC2_INCLUDES): Remove.
3450 * config/i386/t-cygwin: Remove.
3451 * config/i386/t-darwin (LIB2_SIDITI_CONV_FUNCS, LIB2FUNCS_EXTRA)
3452 (LIB2FUNCS_EXCLUDE): Remove.
3453 * config/i386/t-darwin64 (LIB2_SIDITI_CONV_FUNCS, LIB2FUNCS_EXTRA)
3454 (LIBGCC, INSTALL_LIBGCC): Remove.
3455 * config/i386/t-gthr-win32: Move to ../libgcc/config/i386.
3456 * config/i386/t-linux64 (LIBGCC, INSTALL_LIBGCC): Remove.
3457 * config/i386/t-mingw-w32: Likewise.
3458 * config/i386/t-mingw-w64: Likewise.
3459 * config/i386/t-openbsd: Likewise.
3460 * config/i386/t-nto: Remove.
3461 * config/ia64/quadlib.c: Move to ../libgcc/config/ia64.
3462 * config/ia64/t-hpux (LIBGCC, INSTALL_LIBGCC, LIB2FUNCS_EXTRA)
3463 (quadlib.c): Remove.
3464 * config/ia64/t-ia64: Remove comment.
3465 * config/iq2000/lib2extra-funcs.c: Move to
3466 ../libgcc/config/iq2000/lib2funcs.c.
3467 * config/iq2000/t-iq2000: Remove.
3468 * config/m32c/m32c-lib2.c: Move to ../libgcc/config/m32c/lib2funcs.c.
3469 * config/m32c/m32c-lib2-trapv.c: Move to ../libgcc/config/m32c/trapv.c.
3470 * config/m32r/t-linux (TARGET_LIBGCC2_CFLAGS): Remove.
3471 * config/m32c/t-m32c (LIB2FUNCS_EXTRA): Remove.
3472 * config/m32r/t-m32r (TARGET_LIBGCC2_CFLAGS, LIBGCC)
3473 (INSTALL_LIBGCC): Remove.
3474 * config/m68k/fpgnulib.c: Move to ../libgcc/config/m68k.
3475 * config/m68k/t-floatlib: Remove.
3476 * config/m68k/t-mlibs (LIBGCC, INSTALL_LIBGCC): Remove.
3477 * config/mcore/t-mcore (TARGET_LIBGCC2_CFLAGS): Remove.
3479 (LIBGCC, INSTALL_LIBGCC): Remove.
3480 * config/mep/mep-lib2.c: Move to ../libgcc/config/mep/lib2funcs.c.
3481 * config/mep/mep-tramp.c: Move to ../libgcc/config/mep/tramp.c.
3482 * config/mep/t-mep (LIB2FUNCS_EXTRA): Remove.
3483 * config/mips/t-elf (TARGET_LIBGCC2_CFLAGS, LIBGCC)
3484 (INSTALL_LIBGCC): Remove.
3485 * config/mips/t-isa3264: Likewise.
3486 * config/mips/t-mips (LIB2_SIDITI_CONV_FUNCS): Remove.
3487 * config/mips/t-r3900 (TARGET_LIBGCC2_CFLAGS, LIBGCC)
3488 (INSTALL_LIBGCC): Remove.
3489 * config/mips/t-sde (LIBGCC, INSTALL_LIBGCC): Remove.
3490 * config/mips/t-sr71k (TARGET_LIBGCC2_CFLAGS, LIBGCC)
3491 (INSTALL_LIBGCC): Remove.
3492 * config/mips/t-vr (TARGET_LIBGCC2_CFLAGS)
3493 (LIB2FUNCS_STATIC_EXTRA): Remove.
3494 * config/mips/vr4120-div.S: Move to ../libgcc/config/mips.
3495 * config/mmix/t-mmix (TARGET_LIBGCC2_CFLAGS): Remove.
3496 * config/mn10300/t-mn10300 (LIBGCC, INSTALL_LIBGCC): Remove.
3497 * config/pa/fptr.c, config/pa/linux-atomic.c: Move to
3498 ../libgcc/config/pa.
3499 * config/pa/lib2funcs.asm: Move to ../libgcc/config/pa/lib2funcs.S.
3500 * config/pa/quadlib.c: Move to ../libgcc/config/pa.
3501 * config/pa/t-dce-thr (LIBGCC, INSTALL_LIBGCC): Remove.
3502 * config/pa/t-linux, config/pa/t-linux64: Remove.
3503 * config/pa/t-pa-hpux, config/pa/t-pa-hpux10,
3504 config/pa/t-pa-hpux11, config/pa/t-pa64: Remove.
3505 * config/pdp11/t-pdp11 (TARGET_LIBGCC2_CFLAGS, LIB2FUNCS_EXTRA):
3507 * config/picochip/libgccExtras: Move to ../libgcc/config/picochip.
3508 * config/picochip/t-picochip (LIB2FUNCS_EXTRA, RANLIB_FOR_TARGET)
3509 (TARGET_LIBGCC2_CFLAGS, LIBGCC2_DEBUG_CFLAGS): Remove.
3510 * config/rs6000/crtresfpr.asm: Move to
3511 ../libgcc/config/rs6000/crtresfpr.S.
3512 * config/rs6000/crtresgpr.asm: Move to
3513 ../libgcc/config/rs6000/crtresgpr.S.
3514 * config/rs6000/crtresxfpr.asm: Move to
3515 ../libgcc/config/rs6000/crtresxfpr.S.
3516 * config/rs6000/crtresxgpr.asm: Move to
3517 ../libgcc/config/rs6000/crtresxgpr.S.
3518 * config/rs6000/crtsavfpr.asm: Move to
3519 ../libgcc/config/rs6000/crtsavfpr.S.
3520 * config/rs6000/crtsavgpr.asm: Move to
3521 ../libgcc/config/rs6000/crtsavgpr.S.
3522 * config/rs6000/darwin-asm.h: Move to ../libgcc/config/rs6000.
3523 * config/rs6000/darwin-fpsave.asm: Move to
3524 ../libgcc/config/rs6000/darwin-fpsave.S.
3525 * config/rs6000/darwin-gpsave.asm: Move to
3526 ../libgcc/config/rs6000/darwin-gpsave.S.
3527 * config/rs6000/darwin-tramp.asm: Move to
3528 ../libgcc/config/rs6000/darwin-tramp.S.
3529 * config/rs6000/darwin-vecsave.asm: Move to
3530 ../libgcc/config/rs6000/darwin-vecsave.S.
3531 * config/rs6000/darwin-world.asm: Move to
3532 ../libgcc/config/rs6000/darwin-world.S.
3533 * config/rs6000/e500crtres32gpr.asm: Move to
3534 ../libgcc/config/rs6000/e500crtres32gpr.S.
3535 * config/rs6000/e500crtres64gpr.asm: Move to
3536 ../libgcc/config/rs6000/e500crtres64gpr.S.
3537 * config/rs6000/e500crtres64gprctr.asm: Move to
3538 ../libgcc/config/rs6000/e500crtres64gprctr.S.
3539 * config/rs6000/e500crtrest32gpr.asm: Move to
3540 ../libgcc/config/rs6000/e500crtrest32gpr.S.
3541 * config/rs6000/e500crtrest64gpr.asm: Move to
3542 ../libgcc/config/rs6000/e500crtrest64gpr.S.
3543 * config/rs6000/e500crtresx32gpr.asm: Move to
3544 ../libgcc/config/rs6000/e500crtresx32gpr.S.
3545 * config/rs6000/e500crtresx64gpr.asm: Move to
3546 ../libgcc/config/rs6000/e500crtresx64gpr.S.
3547 * config/rs6000/e500crtsav32gpr.asm: Move to
3548 ../libgcc/config/rs6000/e500crtsav32gpr.S.
3549 * config/rs6000/e500crtsav64gpr.asm: Move to
3550 ../libgcc/config/rs6000/e500crtsav64gpr.S.
3551 * config/rs6000/e500crtsav64gprctr.asm: Move to
3552 ../libgcc/config/rs6000/e500crtsav64gprctr.S.
3553 * config/rs6000/e500crtsavg32gpr.asm: Move to
3554 ../libgcc/config/rs6000/e500crtsavg32gpr.S.
3555 * config/rs6000/e500crtsavg64gpr.asm: Move to
3556 ../libgcc/config/rs6000/e500crtsavg64gpr.S.
3557 * config/rs6000/e500crtsavg64gprctr.asm: Move to
3558 ../libgcc/config/rs6000/e500crtsavg64gprctr.S.
3559 * config/rs6000/eabi.asm: Move to ../libgcc/config/rs6000/eabi.S.
3560 * config/rs6000/t-aix43 (LIBGCC, INSTALL_LIBGCC, LIB2FUNCS_EXTRA)
3561 (TARGET_LIBGCC2_CFLAGS): Remove.
3562 * config/rs6000/t-aix52: Likewise.
3563 * config/rs6000/t-darwin: Remove.
3564 * config/rs6000/t-darwin64 (LIB2_SIDITI_CONV_FUNCS)
3565 (LIB2FUNCS_EXTRA): Remove.
3566 * config/rs6000/t-fprules (LIBGCC, INSTALL_LIBGCC): Remove.
3567 * config/rs6000/t-linux64 (TARGET_LIBGCC2_CFLAGS): Remove.
3568 * config/rs6000/t-lynx (LIB2FUNCS_EXTRA, tramp.S, LIBGCC)
3569 (INSTALL_LIBGCC): Remove.
3570 * config/rs6000/t-netbsd (LIB2FUNCS_EXTRA)
3571 (LIB2FUNCS_STATIC_EXTRA, tramp.S, crtsavfpr.S, crtresfpr.S)
3572 (crtsavgpr.S, crtresgpr.S, crtresxfpr.S, crtresxgpr.S, LIBGCC)
3573 (INSTALL_LIBGCC, $(T)crtsavfpr$(objext), $(T)crtresfpr$(objext))
3574 ($(T)crtsavgpr$(objext), $(T)crtresgpr$(objext))
3575 ($(T)crtresxfpr$(objext), $(T)crtresxgpr$(objext)): Remove.
3576 * config/rs6000/t-ppccomm (LIB2FUNCS_EXTRA)
3577 (LIB2FUNCS_STATIC_EXTRA, eabi.S, tramp.S): Remove.
3578 * config/rs6000/t-spe (LIBGCC, INSTALL_LIBGCC): Remove.
3579 * config/rs6000/t-vxworks: Remove comment.
3580 * config/rs6000/tramp.asm: Move to ../libgcc/config/rs6000/tramp.S.
3581 * config/rx/t-rx (LIBGCC, INSTALL_LIBGCC): Remove.
3582 * config/sh/linux-atomic.asm: Move to
3583 ../libgcc/config/sh/linux-atomic.S.
3584 * config/sh/t-linux (LIB2FUNCS_EXTRA): Remove.
3585 * config/sh/t-netbsd: Remove.
3586 * config/sh/t-sh (TARGET_LIBGCC2_CFLAGS, LIBGCC, INSTALL_LIBGCC):
3588 * config/sparc/t-elf (LIBGCC, INSTALL_LIBGCC): Remove.
3589 * config/sparc/t-leon: Likewise.
3590 * config/sparc/t-leon3: Likewise.
3591 * config/sparc/t-linux64: Likewise.
3592 * config/sparc/t-netbsd64: Fix typo.
3594 * config/spu/divmodti4.c, config/spu/divv2df3.c,
3595 config/spu/float_disf.c, config/spu/float_unsdidf.c,
3596 config/spu/float_unsdisf.c, config/spu/float_unssidf.c,
3597 config/spu/mfc_multi_tag_release.c,
3598 config/spu/mfc_multi_tag_reserve.c, config/spu/mfc_tag_release.c,
3599 config/spu/mfc_tag_reserve.c, config/spu/mfc_tag_table.c,
3600 config/spu/multi3.c: Move to ../libgcc/config/spu.
3601 * config/spu/t-spu-elf (TARGET_LIBGCC2_CFLAGS, LIB2FUNCS_EXCLUDE)
3602 (LIB2FUNCS_STATIC_EXTRA, LIB2_SIDITI_CONV_FUNCS, LIBGCC)
3603 (INSTALL_LIBGCC): Remove.
3604 * config/stormy16/stormy16-lib2.c: Move to
3605 ../libgcc/config/stormy16/lib2.c.
3606 * config/stormy16/stormy16-lib2-ashlsi3.c: Move to
3607 ../libgcc/config/stormy16/ashlsi3.c.
3608 * config/stormy16/stormy16-lib2-ashrsi3.c: Move to
3609 ../libgcc/config/stormy16/ashrsi3.c.
3610 * config/stormy16/stormy16-lib2-clzhi2.c: Move to
3611 ../libgcc/config/stormy16/clzhi2.c.
3612 * config/stormy16/stormy16-lib2-cmpsi2.c: Move to
3613 ../libgcc/config/stormy16/cmpsi2.c.
3614 * config/stormy16/stormy16-lib2-ctzhi2.c: Move to
3615 ../libgcc/config/stormy16/ctzhi2.c.
3616 * config/stormy16/stormy16-lib2-divsi3.c: Move to
3617 ../libgcc/config/stormy16/divsi3.c.
3618 * config/stormy16/stormy16-lib2-ffshi2.c: Move to
3619 ../libgcc/config/stormy16/ffshi2.c.
3620 * config/stormy16/stormy16-lib2-lshrsi3.c: Move to
3621 ../libgcc/config/stormy16/lshrsi3.c.
3622 * config/stormy16/stormy16-lib2-modsi3.c: Move to
3623 ../libgcc/config/stormy16/modsi3.c.
3624 * config/stormy16/stormy16-lib2-parityhi2.c: Move to
3625 ../libgcc/config/stormy16/parityhi2.c.
3626 * config/stormy16/stormy16-lib2-popcounthi2.c: Move to
3627 ../libgcc/config/stormy16/popcounthi2.c.
3628 * config/stormy16/stormy16-lib2-ucmpsi2.c: Move to
3629 ../libgcc/config/stormy16/ucmpsi2.c.
3630 * config/stormy16/stormy16-lib2-udivmodsi4.c: Move to
3631 ../libgcc/config/stormy16/udivmodsi4.c.
3632 * config/stormy16/stormy16-lib2-udivsi3.c: Move to
3633 ../libgcc/config/stormy16/udivsi3.c.
3634 * config/stormy16/stormy16-lib2-umodsi3.c: Move to
3635 ../libgcc/config/stormy16/umodsi3.c.
3636 * config/stormy16/t-stormy16: Move to ../libgcc/config/t-stormy16.
3637 * config/v850/t-v850 (INSTALL_LIBGCC): Remove.
3638 * config/xtensa/lib2funcs.S: Move to ../libgcc/config/xtensa.
3639 * config/xtensa/t-elf: Remove.
3640 * config/xtensa/t-xtensa (LIB2FUNCS_EXTRA): Remove.
3641 * config.gcc (*-*-freebsd*): Remove t-freebsd, t-freebsd-thread
3643 (*-*-linux*, frv-*-*linux*, *-*-kfreebsd*-gnu, *-*-knetbsd*-gnu,
3644 *-*-gnu*, *-*-kopensolaris*-gnu): Remove t-linux from tmake_file.
3645 (*-*-netbsd*): Remove t-libgcc-pic from tmake_file.
3646 (*-*-openbsd*): Likewise.
3647 Remove t-openbsd-thread for posix threads.
3648 (alpha*-*-linux*): Remove alpha/t-alpha, alpha/t-ieee from tmake_file.
3649 (alpha*-*-freebsd*): Likewise.
3650 (alpha*-*-netbsd*): Likewise.
3651 (alpha*-*-openbsd*): Likewise.
3652 (alpha64-dec-*vms*): Likewise.
3653 (alpha*-dec-*vms*): Likewise.
3654 (arm*-*-netbsdelf*): Remove arm/t-netbsd from tmake_file.
3655 (arm*-*-linux*): Remove t-linux from tmake_file.
3656 Remove arm/t-bpabi from tmake_file for arm*-*-linux-*eabi.
3657 (arm*-*-uclinux*): Remove arm/t-bpabi from tmake_file for
3658 arm*-*-uclinux*eabi.
3659 (arm*-*-eabi*, arm*-*-symbianelf* ): Remove arm/t-bpabi from
3660 tmake_file for arm*-*-eabi*.
3661 (fr30-*-elf): Remove tmake_file.
3662 (hppa*64*-*-linux*): Remove tmake_file.
3663 (hppa*-*-linux*): Likewise.
3664 (hppa[12]*-*-hpux10*): Remove pa/t-pa-hpux10, pa/t-pa-hpux from
3666 (hppa*64*-*-hpux11*): Remove pa/t-pa64, pa/t-pa-hpux from tmake_file.
3667 (hppa[12]*-*-hpux11*): Remove pa/t-pa-hpux11, pa/t-pa-hpux from
3669 (i[34567]86-*-elf*): Remove tmake_file.
3670 (x86_64-*-elf*): Likewise.
3671 (i[34567]86-*-nto-qnx*): Likewise.
3672 (i[34567]86-*-cygwin*): Remove i386/t-cygwin from tmake_file.
3673 (i[34567]86-*-mingw*, x86_64-*-mingw*): Remove i386/t-gthr-win32
3674 from tmake_file if using win32 threads.
3675 (iq2000*-*-elf*): Remove tmake-file.
3676 (microblaze*-linux*): Likewise.
3677 (sh-*-elf*, sh[12346l]*-*-elf*, sh-*-linux*)
3678 (sh[2346lbe]*-*-linux*, sh-*-netbsdelf*, shl*-*-netbsdelf*)
3679 (sh5-*-netbsd*, sh5l*-*-netbsd*, sh64-*-netbsd*)
3680 (sh64l*-*-netbsd*): Remove sh/t-netbsd from tmake_file for
3681 sh5*-*-netbsd*, sh64*-netbsd*, *-*-netbsd.
3682 (xtensa*-*-elf*): Remove tmake_file.
3684 2011-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3686 * Makefile.in (LIB1ASMSRC): Don't export.
3687 (libgcc.mvars): Don't emit LIB1ASMFUNCS, LIB1ASMSRC.
3688 * config/arm/arm.c: Update lib1funcs.asm filename.
3689 * config/arm/linux-eabi.h: Likewise.
3690 * config/arm/bpabi-v6m.S, config/arm/bpabi.S,
3691 config/arm/ieee754-df.S, config/arm/ieee754-sf.S: Move to
3692 ../libgcc/config/arm.
3693 * config/arm/lib1funcs.asm: Move to ../libgcc/config/arm/lib1funcs.S.
3694 * config/arm/t-arm (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3695 * config/arm/t-arm-elf (LIB1ASMFUNCS): Remove.
3696 * config/arm/t-bpabi: Likewise.
3697 * config/arm/t-linux (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3698 * config/arm/t-linux-eabi (LIB1ASMFUNCS): Remove.
3699 * config/arm/t-strongarm-elf: Likewise.
3700 * config/arm/t-symbian: Likewise.
3701 * config/arm/t-vxworks: Likewise.
3702 * config/arm/t-wince-pe: Likewise.
3703 * config/avr/libgcc.S: Move to ../libgcc/config/avr.
3704 * config/avr/t-avr (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3705 * config/bfin/lib1funcs.asm: Move to ../libgcc/config/bfin/lib1funcs.S.
3706 * config/bfin/t-bfin: Remove.
3707 * config/bfin/t-bfin-elf (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3708 * config/bfin/t-bfin-linux: Likewise.
3709 * config/bfin/t-bfin-uclinux: Likewise.
3710 * config/c6x/lib1funcs.asm: Move to ../libgcc/config/c6x/lib1funcs.S.
3711 * config/c6x/t-c6x-elf (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3712 * config/fr30/lib1funcs.asm: Move to ../libgcc/config/fr30/lib1funcs.S.
3713 * config/fr30/t-fr30 (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3714 * config/frv/lib1funcs.asm: Move to ../libgcc/config/frv/lib1funcs.S.
3715 * config/frv/t-frv (CROSS_LIBGCC1, LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3716 * config/h8300/fixunssfsi.c: Update lib1funcs.asm filename.
3717 * config/h8300/lib1funcs.asm: Move to
3718 ../libgcc/config/h8300/lib1funcs.S.
3719 * config/h8300/t-h8300 (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3720 * config/i386/cygwin.asm: Move to ../libgcc/config/i386/cygwin.S.
3721 * config/i386/t-cygming (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3722 * config/i386/t-interix: Likewise.
3723 * config/ia64/lib1funcs.asm: Move to ../libgcc/config/ia64/lib1funcs.S.
3724 * config/ia64/t-hpux (LIB1ASMFUNCS, LIBGCC1_TEST): Remove.
3725 * config/ia64/t-ia64 (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3726 * config/iq2000/t-iq2000 (LIBGCC1, CROSS_LIBGCC1): Remove.
3727 * config/m32c/m32c.c: Update m32c-lib1.S filename.
3728 * config/m32c/m32c-lib1.S: Move to ../libgcc/config/m32c/lib1funcs.S.
3729 * config/m32c/t-m32c (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3730 * config/m32r/t-linux (CROSS_LIBGCC1, LIBGCC1, LIBGCC1_TEST): Remove.
3731 * config/m68k/lb1sf68.asm: Move to ../libgcc/config/m68k/lb1sf68.S.
3732 * config/m68k/t-floatlib (LIB1ASMSRC, LIB1ASMFUNCS): New file.
3733 * config/mcore/lib1.asm: Move to ../libgcc/config/mcore/lib1funcs.S.
3734 * config/mcore/t-mcore (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3735 * config/mep/mep-lib1.asm: Move to ../libgcc/config/mep/lib1funcs.S.
3736 * config/mep/t-mep (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3737 * config/mips/mips16.S: Move to ../libgcc/config/mips.
3738 * config/mips/t-libgcc-mips16: Remove.
3739 * config/mips/t-sr71k (LIBGCC1, CROSS_LIBGCC1): Remove.
3740 * config/pa/milli64.S: Move to ../libgcc/config/pa.
3741 * config/pa/t-linux (LIB1ASMFUNCS, LIB1ASMSRC): Remove.
3742 * config/pa/t-linux64: Likewise.
3743 * config/picochip/libgccExtras/fake_libgcc.asm: Move to
3744 ../libgcc/config/picochip/lib1funcs.S.
3745 * config/picochip/t-picochip (LIB1ASMFUNCS, LIB1ASMSRC): Remove.
3746 * config/sh/lib1funcs.asm: Move to ../libgcc/config/sh/lib1funcs.S.
3747 * config/sh/lib1funcs.h: Move to ../libgcc/config/sh.
3748 * config/sh/sh.h: Update lib1funcs.asm filename.
3749 * config/sh/t-linux (LIB1ASMFUNCS_CACHE): Remove.
3750 * config/sh/t-netbsd: Likewise.
3751 * config/sh/t-sh (LIB1ASMSRC, LIB1ASMFUNCS, LIB1ASMFUNCS_CACHE):
3753 * config/sh/t-sh64 (LIB1ASMFUNCS): Remove.
3754 * config/sparc/lb1spc.asm: Move to ../libgcc/config/sparc/lb1spc.S.
3755 * config/sparc/lb1spl.asm: Remove.
3756 * config/sparc/t-elf (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3757 * config/sparc/t-leon: Likewise.
3758 * config/spu/t-spu-elf (LIBGCC1, CROSS_LIBGCC1): Remove.
3759 * config/v850/lib1funcs.asm: Move to ../libgcc/config/v850/lib1funcs.S.
3760 * config/v850/t-v850 (LIB1ASMSRC, LIB1ASMFUNCS): Remove
3761 * config/vax/lib1funcs.asm: Move to ../libgcc/config/vax/lib1funcs.S.
3762 * config/vax/t-linux: Remove.
3763 * config/xtensa/ieee754-df.S, config/xtensa/ieee754-sf.S: Move to
3764 ../libgcc/config/xtensa.
3765 * config/xtensa/lib1funcs.asm: Move to
3766 ../libgcc/config/xtensa/lib1funcs.S.
3767 * config/xtensa/t-xtensa (LIB1ASMSRC, LIB1ASMFUNCS): Remove.
3768 * config.gcc (bfin*-rtems*): Remove bfin/t-bfin from tmake_file.
3769 (bfin*-*): Likewise.
3770 (mips64*-*-linux*, mipsisa64*-*-linux*): Remove
3771 mips/t-libgcc-mips16 from tmake_file.
3772 (mips*-*-linux*): Likewise.
3773 (mips*-sde-elf*): Likewise.
3774 (mipsisa32-*-elf*, mipsisa32el-*-elf*, mipsisa32r2-*-elf*)
3775 (mipsisa32r2el-*-elf*, mipsisa64-*-elf*, mipsisa64el-*-elf*)
3776 (mipsisa64r2-*-elf*, mipsisa64r2el-*-elf*): Likewise.
3777 (mipsisa64sb1-*-elf*, mipsisa64sb1el-*-elf*): Likewise.
3778 (mips-*-elf*, mipsel-*-elf*): Likewise.
3779 (mips64-*-elf*, mips64el-*-elf*): Likewise.
3780 (mips64orion-*-elf*, mips64orionel-*-elf*): Likewise.
3781 (mips*-*-rtems*): Likewise.
3782 (mipstx39-*-elf*, mipstx39el-*-elf*): Likewise.
3783 (vax-*-linux*): Remove vax/t-linux from tmake_file.
3785 2011-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3787 * config.gcc (extra_parts): Remove.
3788 (*-*-freebsd*): Remove extra_parts.
3789 (*-*-linux*, frv-*-*linux*, *-*-kfreebsd*-gnu, *-*-knetbsd*-gnu,
3790 *-*-gnu*, *-*-kopensolaris*-gnu): Likewise.
3791 (*-*-netbsd*): Remove t-libc-ok, t-netbsd from tmake_file.
3792 Remove extra_parts for *-*-netbsd*1.[7-9]*, *-*-netbsd[2-9]*,
3793 *-*-netbsdelf[2-9]*.
3794 (*-*-openbsd*): Remove t-libc-ok from tmake_file.
3795 (alpha*-*-linux*): Remove extra_parts.
3796 (alpha*-*-freebsd*): Likewise.
3797 (bfin*-linux-uclibc*): Likewise.
3798 (fr30-*-elf): Likewise.
3799 (moxie-*-elf): Likewise.
3800 (moxie-*-uclinux*): Likewise.
3801 (h8300-*-rtems*): Remove h8300/t-elf from tmake_file.
3802 (h8300-*-elf*): Likewise.
3803 (hppa*64*-*-hpux11*): Remove extra_parts.
3804 (i[34567]86-*-elf*): Remove i386/t-i386elf, i386/t-crtstuff from
3806 (x86_64-*-elf*): Likewise.
3807 (i[34567]86-*-freebsd*): Remove tmake_file.
3808 (x86_64-*-freebsd*): Likewise.
3809 (x86_64-*-netbsd*): Likewise.
3810 (i[34567]86-*-openbsd2.*, i[34567]86-*openbsd3.[0123]): Remove
3811 t-libc-ok from tmake_file.
3812 (i[34567]86-*-linux*, i[34567]86-*-kfreebsd*-gnu,
3813 i[34567]86-*-knetbsd*-gnu, i[34567]86-*-gnu*,
3814 i[34567]86-*-kopensolaris*-gnu): Remove i386/t-crtstuff from
3817 (x86_64-*-linux*, x86_64-*-kfreebsd*-gnu, x86_64-*-knetbsd*-gnu):
3818 Remove i386/t-crtstuff from tmake_file.
3819 (i[34567]86-*-lynxos*): Likewise.
3821 (ia64*-*-elf*): Remove extra_parts.
3822 (ia64*-*-freebsd*): Likewise.
3823 (ia64*-*-linux*): Likewise.
3824 (ia64-hp-*vms*): Remove ia64/t-vms from tmake_file.
3825 (m32r-*-elf*): Remove extra_parts.
3826 (m32rle-*-elf*): Likewise.
3827 (m32r-*-rtems*): Likewise.
3828 (m68k-*-elf*, fido-*-elf*): Likewise.
3829 (m68k*-*-openbsd*): Remove t-libc-ok from tmake_file.
3830 (m68k-*-rtems*): Remove extra_parts.
3831 (mep-*-*): Likewise.
3832 (microblaze*-linux*): Likewise.
3833 (mips64*-*-linux*, mipsisa64*-*-linux*): Likewise.
3834 (mips*-*-linux*): Likewise.
3835 (powerpc-*-lynxos*): Likewise.
3836 (s390x-ibm-tpf*): Likewise.
3837 (score-*-elf): Likewise.
3839 (sh-*-elf*, sh[12346l]*-*-elf*, sh-*-linux*, sh[2346lbe]*-*-linux*,
3840 sh-*-netbsdelf*, shl*-*-netbsdelf*, sh5-*-netbsd*, sh5l*-*-netbsd*,
3841 sh64-*-netbsd*, sh64l*-*-netbsd*): Remove sh/t-elf from tmake_file.
3842 Remove sh/t-superh from tmake_file for sh*-superh-elf.
3843 Remove sh/t-linux64 from tmake_file for sh64*-*-linux*.
3844 (sh-*-rtems*): Remove sh/t-elf from tmake_file.
3845 (sh-wrs-vxworks): Likewise.
3846 (sparc-*-linux*): Remove extra_parts.
3847 (sparc64-*-linux*): Likewise.
3848 (sparc64-*-freebsd*, ultrasparc-*-freebsd*): Likewise.
3849 (xstormy16-*-elf): Likewise.
3850 (xtensa*-*-linux*): Remove xtensa/t-linux from tmake_file.
3851 (am33_2.0-*-linux*): Remove extra_parts.
3852 * configure.ac (extra_parts): Don't substitute.
3853 * configure: Regenerate.
3854 * crtstuff.c: Move to ../libgcc.
3855 * Makefile.in (CRTSTUFF_CFLAGS): Remove.
3856 (EXTRA_PARTS): Remove.
3857 (CRTSTUFF_T_CFLAGS): Remove.
3858 (MOSTLYCLEANFILES): Remove $(EXTRA_PARTS).
3859 (GCC_EXTRA_PARTS): Remove.
3860 (libgcc.mvars): Remove GCC_EXTRA_PARTS, CRTSTUFF_CFLAGS,
3861 CRTSTUFF_T_CFLAGS, CRTSTUFF_T_CFLAGS_S.
3862 Emit GCC_CFLAGS, INHIBIT_LIBC_CFLAGS.
3863 ($(T)crtbegin.o, $(T)crtend.o, $(T)crtbeginS.o, $(T)crtendS.o)
3864 ($(T)crtbeginT.o): Remove.
3865 * config/alpha/t-vms (EXTRA_PARTS): Remove.
3866 ($(T)vms-dwarf2.o, $(T)vms-dwarf2eh.o): Remove.
3867 * config/alpha/vms-dwarf2.asm: Move to
3868 ../libgcc/config/alpha/vms-dwarf2.S.
3869 * config/alpha/vms-dwarf2eh.asm: Move to
3870 ../libgcc/config/alpha/vms-dwarf2eh.S.
3871 * config/arm/crti.asm: Move to ../libgcc/config/arm/crti.S.
3872 * config/arm/crtn.asm: Move to ../libgcc/config/arm/crtn.S.
3873 * config/arm/t-arm-elf (EXTRA_MULTILIB_PARTS): Remove.
3874 ($(T)crti.o, $(T)crtn.o): Remove.
3875 * config/arm/t-linux: Remove comment.
3876 * config/arm/t-linux-eabi (EXTRA_MULTILIB_PARTS): Remove.
3877 * config/arm/t-strongarm-elf (EXTRA_MULTILIB_PARTS): Remove.
3878 ($(T)crti.o, $(T)crtn.o): Remove.
3879 * config/arm/t-symbian (EXTRA_MULTILIB_PARTS): Remove.
3880 * config/bfin/crti.s: Move to ../libgcc/config/bfin/crti.S.
3881 * config/bfin/crtn.s: Move to ../libgcc/config/bfin/crtn.S.
3882 * config/bfin/crtlibid.s: Move to ../libgcc/config/bfin/crtlibid.S.
3883 * config/bfin/t-bfin (EXTRA_PARTS): Remove.
3884 ($(T)crti.o, $(T)crtn.o): Remove.
3885 * config/bfin/t-bfin-elf (CRTSTUFF_T_CFLAGS): Remove.
3886 ($(T)crti.o, $(T)crtn.o, $(T)crtlibid.o): Remove
3887 (EXTRA_MULTILIB_PARTS): Remove.
3888 * config/bfin/t-bfin-linux (CRTSTUFF_T_CFLAGS,
3889 EXTRA_MULTILIB_PARTS): Remove.
3890 * config/bfin/t-bfin-uclinux (CRTSTUFF_T_CFLAGS): Remove.
3891 ($(T)crtlibid.o): Remove.
3892 (EXTRA_MULTILIB_PARTS): Remove.
3893 * config/c6x/crti.s: Move to ../libgcc/config/c6x/crti.S.
3894 * config/c6x/crtn.s: Move to ../libgcc/config/c6x/crtn.S.
3895 * config/c6x/t-c6x-elf ($(T)crti.o, $(T)crtn.o): Remove.
3896 (EXTRA_MULTILIB_PARTS): Remove.
3897 (CRTSTUFF_T_CFLAGS, CRTSTUFF_T_CFLAGS_S): Remove.
3898 * config/c6x/t-c6x-uclinux (CRTSTUFF_T_CFLAGS,
3899 CRTSTUFF_T_CFLAGS_S): Remove.
3900 * config/cris/t-elfmulti (CRTSTUFF_T_CFLAGS): Remove.
3901 * config/cris/t-linux (CRTSTUFF_T_CFLAGS_S): Remove.
3902 * config/fr30/crti.asm: Move to ../libgcc/config/fr30/crti.S.
3903 * config/fr30/crtn.asm: Move to ../libgcc/config/fr30/crtn.S.
3904 * config/fr30/t-fr30 ($(T)crti.o, $(T)crtn.o): Remove.
3905 * config/frv/frvbegin.c, config/frv/frvend.c: Move to
3906 ../libgcc/config/frv.
3907 * config/frv/t-frv (EXTRA_MULTILIB_PARTS): Remove.
3908 (FRVSTUFF_CFLAGS, $(T)frvbegin$(objext), $(T)frvend$(objext)): Remove.
3909 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Remove.
3910 (CRTSTUFF_T_CFLAGS): Remove.
3911 * config/h8300/crti.asm: Move to ../libgcc/config/h8300/crti.S.
3912 * config/h8300/crtn.asm: Move to ../libgcc/config/h8300/crtn.S.
3913 * config/h8300/t-elf: Remove.
3914 * config/i386/cygming-crtbegin.c, config/i386/cygming-crtend.c:
3915 Move to ../libgcc/config/i386.
3916 * config/i386/t-crtstuff: Remove.
3917 * config/i386/t-i386elf: Remove.
3918 * config/i386/t-linux64 (EXTRA_MULTILIB_PARTS): Remove.
3919 * config/i386/t-nto (CRTSTUFF_T_CFLAGS, EXTRA_PARTS): Remove.
3920 * config/ia64/crtbegin.asm: Move to ../libgcc/config/ia64/crtbegin.S.
3921 * config/ia64/crtend.asm: Move to ../libgcc/config/ia64/crtend.S.
3922 * config/ia64/crti.asm: Move to ../libgcc/config/ia64/crti.S.
3923 * config/ia64/crtn.asm: Move to ../libgcc/config/ia64/crtn.S.
3924 * config/ia64/t-vms: Remove.
3925 * config/ia64/vms-crtinit.asm: Move to
3926 ../libgcc/config/ia64/vms-crtinit.S.
OSDN Git Service
