1 2009-01-13 Vladimir Makarov <vmakarov@redhat.com>
4 * Makefile.in (ira-lives.o): Add except.h.
6 * ira-lives.c: Include except.h.
7 (process_bb_node_lives): Process can_throw_internal.
9 2009-01-13 Jakub Jelinek <jakub@redhat.com>
11 PR rtl-optimization/38774
12 * combine.c (simplify_set): When undoing cc_use change, don't do
13 PUT_CODE on the newly created comparison, but instead put back the
16 2009-01-13 Joseph Myers <joseph@codesourcery.com>
18 * doc/invoke.texi (ARM Options): Update lists of -mcpu and -march
19 values. Remove duplicate arm8 entry.
21 2009-01-13 Sebastian Pop <sebastian.pop@amd.com>
23 PR tree-optimization/38786
24 * graphite.c (expand_scalar_variables_ssa_name): New, outlined from
25 the SSA_NAME case of expand_scalar_variables_expr.
26 Set the type of an expression to the type of its assign statement.
27 (expand_scalar_variables_expr): Also gather the scalar computation
28 used to index the memory access. Do not pass loop_p.
29 Fix comment. Stop recursion on tcc_constant or tcc_declaration.
30 (expand_scalar_variables_stmt): Pass to expand_scalar_variables_expr
31 the gimple_stmt_iterator where it inserts new code. Do not pass loop_p.
32 (copy_bb_and_scalar_dependences): Do not pass loop_p.
33 (translate_clast): Update call to copy_bb_and_scalar_dependences.
35 2009-01-13 Sebastian Pop <sebastian.pop@amd.com>
37 * graphite.h (debug_value): Removed.
38 * graphite.c (debug_value): Removed.
40 2009-01-13 Richard Earnshaw <rearnsha@arm.com>
42 * arm.c (output_move_double): Don't synthesize thumb-2 ldrd/strd with
43 two 32-bit instructions.
45 2009-01-13 Richard Earnshaw <rearnsha@arm.com>
47 * arm.c (struct processors): Pass for speed down into cost helper
49 (const_ok_for_op): Handle COMPARE and inequality nodes.
50 (arm_rtx_costs_1): Rewrite.
51 (arm_size_rtx_costs): Update prototype.
52 (arm_rtx_costs): Pass speed down to helper functions.
53 (arm_slowmul_rtx_costs): Rework cost calculations.
54 (arm_fastmul_rtx_costs, arm_xscale_rtx_costs): Likewise.
55 (arm_9e_rtx_costs): Likewise.
57 2009-01-13 Uros Bizjak <ubizjak@gmail.com>
59 * config/alpha/alpha.c (alpha_legitimate_address_p): Explicit
60 relocations of local symbols wider than UNITS_PER_WORD are not valid.
61 (alpha_legitimize_address): Do not split local symbols wider than
62 UNITS_PER_WORD into HIGH/LO_SUM parts.
64 2009-01-13 Danny Smith <dannysmith@users.sourceforge.net>
67 * gcc.c (process_command): Replace call to execvp with calls
70 2009-01-03 Anatoly Sokolov <aesok@post.ru>
73 * config/avr/t-avr (LIB1ASMFUNCS): Add _tablejump_elpm.
74 * config/avr/libgcc.S (__do_global_ctors, __do_global_dtors): Add
75 variant for devices with 3-byte PC.
76 (__tablejump_elpm__) : New.
78 2009-01-12 Jakub Jelinek <jakub@redhat.com>
81 * c-parser.c (c_parser_postfix_expression): Allow `->' in
82 offsetof member-designator, handle it as `[0].'.
84 2009-01-12 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
86 * pa.c (pa_asm_output_mi_thunk): Use pc-relative branch to thunk
87 function when not using named sections on targets with named sections
88 if branch distance is less than 262132.
90 2009-01-12 Richard Earnshaw <rearnsha@arm.com>
92 * combine.c (combine_instructions): Recompute
93 optimize_this_for_speed_p for each BB in the main combine loop.
95 2009-01-12 Tomas Bily <tbily@suse.cz>
98 * tree-loop-distribution.c (prop_phis): New function.
99 (generate_builtin): Call prop_phis.
100 * testsuite/gcc.dg/tree-ssa/pr38385.c: New file.
102 2009-01-12 Jakub Jelinek <jakub@redhat.com>
104 PR tree-optimization/38807
105 * tree-ssa-reassoc.c (remove_visited_stmt_chain): Don't look at
106 gimple_visited_p unless stmt is GIMPLE_ASSIGN.
108 2009-01-11 Adam Nemet <anemet@caviumnetworks.com>
110 * expmed.c (store_bit_field_1): Properly truncate the paradoxical
111 subreg of op0 to the original op0.
113 2009-01-11 Laurent GUERBY <laurent@guerby.net>
115 * doc/sourcebuild.texi (Source Tree): Move up intl and fixinc.
117 2009-01-11 Markus Schoepflin <markus.schoepflin@comsoft.de>
120 * gcc/mips-tfile.c (parse_def): Fix parsing of def strings
121 starting with digits.
123 2009-01-10 Jakub Jelinek <jakub@redhat.com>
126 * config/arm/arm.c (arm_is_long_call_p): Don't call
127 arm_function_in_section_p if decl isn't a FUNCTION_DECL.
129 2009-01-09 Steven Bosscher <steven@gcc.gnu.org>
131 * regrename.c (regrename_optimize): Fix dumping.
132 (find_oldest_value_reg): Preserve REG_POINTER.
133 (copy_hardreg_forward_1): Likewise.
135 2009-01-09 Diego Novillo <dnovillo@google.com>
137 * gimple.h (struct gimple_statement_base)<uid>: Document
138 the restrictions on its use.
140 (gimple_set_uid): Tidy.
142 2009-01-09 Jakub Jelinek <jakub@redhat.com>
144 * config/i386/i386.c (ix86_expand_movmem, ix86_expand_setmem): Add
145 zero guard even if align_bytes != 0 and count is smaller than
148 2008-01-09 Vladimir Makarov <vmakarov@redhat.com>
150 PR rtl-optimization/38495
151 * ira-emit.c (print_move_list, ira_debug_move_list): New functions.
152 (add_range_and_copies_from_move_list): Print all added ranges.
153 Add ranges to memory optimized destination.
155 2009-01-09 Jakub Jelinek <jakub@redhat.com>
159 * config/i386/i386.c (override_options): Reject
160 -mstringop-strategy=rep_8byte with -m32.
161 (ix86_expand_movmem): For size_needed == 1 set epilogue_size_needed
162 to 1. Do count comparison against epilogue_size_needed at compile
163 time even when count_exp was constant forced into register. For
164 size_needed don't jump to epilogue, instead just avoid aligning
165 and invoke the body algorithm. If need_zero_guard, add zero guard
166 even if count is non-zero, but smaller than size_needed + number of
167 bytes that could be stored for alignment.
168 (ix86_expand_setmem): For size_needed == 1 set epilogue_size_needed
169 to 1. If need_zero_guard, add zero guard even if count is non-zero,
170 but smaller than size_needed + number of bytes that could be stored
171 for alignment. Compare size_needed with epilogue_size_needed instead
172 of desired_align - align, don't adjust size_needed, pass
173 epilogue_size_needed to the epilogue expanders.
176 * c-pretty-print.c (pp_c_expression): Handle GOTO_EXPR like BIND_EXPR.
178 2009-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
180 * pa.c (last_address): Change to unsigned.
181 (update_total_code_bytes): Change argument to unsigned. Don't
182 check if insn addresses are set.
183 (pa_output_function_epilogue): Set last_address to UINT_MAX if insn
184 addresses are not set.
185 (pa_asm_output_mi_thunk): Handle wrap when updating last_address.
187 2009-01-09 Nick Clifton <nickc@redhat.com>
189 * config/sh/symbian.c: Replace uses of DECL_INLINE with
190 DECL_DECLARED_INLINE_P.
192 2009-01-09 Jakub Jelinek <jakub@redhat.com>
195 * dojump.c (do_jump_by_parts_zero_rtx): Use mode instead of
196 GET_MODE (op0) in operand_subword_force calls.
199 * fold-const.c (fold_unary): For COMPOUND_EXPR and COND_EXPR,
200 fold_convert arg0 operands to TREE_TYPE (op0) first.
202 2009-01-08 Vladimir Makarov <vmakarov@redhat.com>
204 * params.def (ira-max-conflict-table-size): Decrease default value
207 2009-01-08 Jakub Jelinek <jakub@redhat.com>
209 PR tree-optimization/37031
210 * lambda-code.c (lambda_collect_parameters): Call pointer_set_destroy
212 (build_access_matrix): Reserve correct size for AM_MATRIX vector,
213 allocate it using gc instead of heap, use VEC_quick_push instead of
215 * graphite.c (build_access_matrix): Allocate AM_MATRIX vector using gc
216 instead of heap, use VEC_quick_push instead of VEC_safe_push.
217 * tree-data-ref.h (struct access_matrix): Change matrix to gc
218 allocated vector from heap allocated.
219 * lambda.h: Add DEF_VEC_ALLOC_P for gc allocated lambda_vector.
220 * tree-loop-linear.c (linear_transform_loops): Allocate nest
221 vector only after perfect_loop_nest_depth call.
223 2009-01-08 Sebastian Pop <sebastian.pop@amd.com>
224 Jan Sjodin <jan.sjodin@amd.com>
226 PR tree-optimization/38559
227 * graphite.c (debug_value, copy_constraint,
228 swap_constraint_variables, scale_constraint_variable, ): New.
229 (get_lower_bound, get_upper_bound): Removed.
230 (graphite_trans_bb_strip_mine): Clean up this code that works
231 only for constant number of iterations. Fully copy upper and
232 lower bound constraints, not only the constant part of them.
233 * graphite.h (debug_value): Declared.
235 2009-01-08 Ira Rosen <irar@il.ibm.com>
237 PR tree-optimization/37194
238 * tree-vect-transform.c (vect_estimate_min_profitable_iters):
239 Don't add the cost of cost model guard in prologue to scalar
240 outside cost in case of known number of iterations.
242 2009-01-07 Nathan Froyd <froydnj@codesourcery.com>
243 Alan Modra <amodra@bigpond.net.au>
245 * config/rs6000/rs6000.c (rs6000_legitimize_address): Check for
246 non-word-aligned REG+CONST addressing.
248 2009-01-07 Uros Bizjak <ubizjak@gmail.com>
251 * config/alpha/alpha.c (alpha_end_function): For TARGET_ABI_OSF, call
252 free_after_compilation when outputting a thunk.
253 (alpha_output_mi_thunk_osf): Assert that we are processing a thunk.
254 Do not call free_after_compilation here.
256 2009-01-07 Uros Bizjak <ubizjak@gmail.com>
258 * config/i386/i386.c (ix86_target_string): Use ARRAY_SIZE.
259 (ix86_valid_target_attribute_inner_p): Ditto.
261 2009-01-07 Jan Sjodin <jan.sjodin@amd.com>
263 PR tree-optimization/38492
264 PR tree-optimization/38498
265 * tree-check.c (operator_is_linear, scev_is_linear_expression): New.
266 * tree-chrec.h (scev_is_linear_expression): Declared.
267 * graphite.c (graphite_cannot_represent_loop_niter): New.
268 (scopdet_basic_block_info): Call graphite_cannot_represent_loop_niter.
269 (graphite_loop_normal_form): Use gcc_assert.
270 (scan_tree_for_params): Use CASE_CONVERT.
271 (phi_node_is_iv, bb_contains_non_iv_scalar_phi_nodes): New.
272 (build_scop_conditions_1): Call bb_contains_non_iv_scalar_phi_nodes.
273 Use gcc_assert. Discard scops that contain unhandled cases.
274 (build_scop_conditions): Return a boolean status for unhandled cases.
275 (strip_mine_profitable_p): Print the loop number, not its depth.
276 (is_interchange_valid): Pass the depth of the loop nest, don't
277 recompute it wrongly.
278 (graphite_trans_bb_block): Same.
279 (graphite_trans_bb_block): Print tentative of loop blocking.
280 (graphite_trans_scop_block): Do not print that the loop has been
282 (graphite_transform_loops): Do not handle scops that contain condition
285 2009-01-07 H.J. Lu <hongjiu.lu@intel.com>
287 AVX Programming Reference (December, 2008)
288 * config/i386/avxintrin.h (_mm256_stream_si256): New.
289 (_mm256_stream_pd): Likewise.
290 (_mm256_stream_ps): Likewise.
292 * config/i386/i386.c (ix86_builtins): Add IX86_BUILTIN_MOVNTDQ256,
293 IX86_BUILTIN_MOVNTPD256 and IX86_BUILTIN_MOVNTPS256.
294 (ix86_special_builtin_type): Add VOID_FTYPE_PV4DI_V4DI.
295 (bdesc_special_args): Add __builtin_ia32_movntdq256,
296 __builtin_ia32_movntpd256 and __builtin_ia32_movntps256.
297 (ix86_init_mmx_sse_builtins): Handle VOID_FTYPE_PV4DI_V4DI.
298 (ix86_expand_special_args_builtin): Likewise.
300 * config/i386/sse.md (AVXMODEDI): New.
301 (avx_movnt<mode>): Likewise.
302 (avx_movnt<mode>): Likewise.
303 (<sse>_movnt<mode>): Remove AVX support.
304 (sse2_movntv2di): Likewise.
306 2009-01-07 Richard Guenther <rguenther@suse.de>
309 * fold-const.c (extract_muldiv): Remove obsolete comment.
310 (fold_plusminus_mult_expr): Undo MINUS_EXPR
311 to PLUS_EXPR canonicalization for the canonicalization.
313 2009-01-07 Gerald Pfeifer <gerald@pfeifer.com>
315 * doc/install.texi (alpha*-dec-osf*): Remove note on 32-bit
316 hosted cross-compilers generating less efficient code.
318 2009-01-06 Richard Sandiford <rdsandiford@googlemail.com>
320 * function.h (rtl_data): Add a dbr_scheduled_p field.
321 * reorg.c (dbr_schedule): Set it.
322 (gate_handle_delay_slots): Check it.
323 * config/mips/mips.c (mips_base_delayed_branch): Delete.
324 (mips_reorg): Check flag_delayed_branch instead of
325 mips_base_delayed_branch.
326 (mips_override_options): Don't set mips_base_delayed_branch
327 or flag_delayed_branch.
329 2009-01-06 Richard Sandiford <rdsandiford@googlemail.com>
331 PR rtl-optimization/38426.
332 * ira.c (ira): Set current_function_is_leaf earlier.
334 2009-01-06 Jakub Jelinek <jakub@redhat.com>
336 PR rtl-optimization/38722
337 * combine.c (try_combine): Don't modify PATTERN (i3) and notes
338 too early, only set a flag and modify after last possible
341 2009-01-06 Janis Johnson <janis187@us.ibm.com>
344 * ginclude/float.h: Rename DECnn_DEN to DECnn_SUBNORMAL_MIN.
345 * real.c (decimal_single_format): Correct values of emin and emax.
346 (decimal_double_format): Ditto.
347 (decimal_quad_format): Ditto.
348 * c-cppbuiltin.c (builtin_define_decimal_float_constants): Adjust
349 computation of DECnn_MIN and DECnn_MAX for corrected values of
350 emin and emax. Define __DECnn_SUBNORMAL_MIN__ instead of
351 __DECnn_MIN__, and adjust its computation for the corrected value
354 2009-01-06 Jan Hubicka <jh@suse.cz>
357 * i386.c (ix86_expand_call): Use ARRAY_SIZE.
359 2009-01-06 Gerald Pfeifer <gerald@pfeifer.com>
361 * doc/contrib.texi (Contributors): Slightly adjust the end note.
362 Add Robert Clark to the list of testers.
364 2009-01-06 Jan Hubicka <jh@suse.cz>
365 Kai Tietz <kai.tietz@onevision.com>
367 * i386.md (*msabi_syvabi): Add SSE regs clobbers.
368 * i386.c (ix86_expand_call): Add clobbers.
370 2009-01-06 Jan Hubicka <jh@suse.cz>
371 Kai Tietz <kai.tietz@onevision.com>
373 * i386.h (CONDITIONAL_CALL_USAGE): SSE regs are not used for w64 ABI.
374 * i386.c (struct ix86_frame): Add padding0 and nsseregs.
375 (ix86_nsaved_regs): Count only general purpose regs.
376 (ix86_nsaved_sseregs): New.
377 (ix86_compute_frame_layout): Update nsseregs; set preferred alignment
378 to 16 for w64; compute padding and size of sse reg save area.
379 (ix86_emit_save_regs, ix86_emit_save_regs_using_mov): Save only
380 general purpose regs.
381 (ix86_emit_save_sse_regs_using_mov): New.
382 (ix86_expand_prologue): Save SSE regs if needed.
383 (ix86_emit_restore_regs_using_mov): Use only general purpose regs.
384 (ix86_emit_restore_sse_regs_using_mov): New.
385 (ix86_expand_epilogue): Save SSE regs if needed.
387 2009-01-06 Jan Hubicka <jh@suse.cz>
388 Kai Tietz <kai.tietz@onevision.com>
390 * i386.h (ACCUMULATE_OUTGOING_ARGS): Enable for MSABI
391 * i386.c (init_cumulative_args): Disallow calls of MSABI functions
392 when accumulate outgoing args is off.
394 2009-01-06 H.J. Lu <hongjiu.lu@intel.com>
397 * ira-color.c (ira_reuse_stack_slot): Check ENABLE_IRA_CHECKING
398 before using pseudos_have_intersected_live_ranges_p.
400 * ira-int.h (ira_assert): Always define.
402 2009-01-06 H.J. Lu <hongjiu.lu@intel.com>
404 AVX Programming Reference (December, 2008)
405 * config/i386/avxintrin.h (_mm_permute2_pd): Removed.
406 (_mm256_permute2_pd): Likewise.
407 (_mm_permute2_ps): Likewise.
408 (_mm256_permute2_ps): Likewise.
409 * config/i386/i386.md (UNSPEC_VPERMIL2): Likewise.
410 * config/i386/sse.md (avx_vpermil2<mode>3): Likewise.
412 * config/i386/i386.c (ix86_builtins): Remove
413 IX86_BUILTIN_VPERMIL2PD, IX86_BUILTIN_VPERMIL2PS,
414 IX86_BUILTIN_VPERMIL2PD256 and IX86_BUILTIN_VPERMIL2PS256.
415 (ix86_builtin_type): Remove V8SF_FTYPE_V8SF_V8SF_V8SI_INT,
416 V4DF_FTYPE_V4DF_V4DF_V4DI_INT, V4SF_FTYPE_V4SF_V4SF_V4SI_INT
417 and V2DF_FTYPE_V2DF_V2DF_V2DI_INT.
418 (bdesc_args): Remove __builtin_ia32_vpermil2pd,
419 __builtin_ia32_vpermil2ps, __builtin_ia32_vpermil2pd256 and
420 __builtin_ia32_vpermil2ps256.
421 (ix86_init_mmx_sse_builtins): Updated.
422 (ix86_expand_args_builtin): Likewise.
424 2009-01-05 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
426 * pa.c (output_call): Relocate non-jump insns in the delay slot of
427 long absolute calls when generating PA 2.0 code.
429 2009-01-05 Vladimir Makarov <vmakarov@redhat.com>
431 PR rtl-optimization/38583
432 * params.h (IRA_MAX_CONFLICT_TABLE_SIZE): New macro.
434 * params.def (ira-max-conflict-table-size): New.
436 * doc/invoke.texi (ira-max-conflict-table-size): Decribe.
438 * ira.h (ira_conflicts_p): New external definition.
440 * ira-conflicts.c (build_conflict_bit_table): Do not build too big
441 table. Report this. Return result of building.
442 (ira_build_conflicts): Use ira_conflicts_p. Check result of
443 building conflict table.
445 * ira-color.c (fast_allocation): Use num instead of ira_allocnos_num.
446 (ira_color): Use ira_conflicts_p.
448 * global.c: Include ira.h.
449 (pseudo_for_reload_consideration_p, build_insn_chain): Use
452 * Makefile.in (global.o): Add ira.h.
454 * ira-build.c (mark_all_loops_for_removal,
455 propagate_some_info_from_allocno): New.
456 (remove_unnecessary_allocnos): Call
457 propagate_some_info_from_allocno.
458 (remove_low_level_allocnos): New.
459 (remove_unnecessary_regions): Add parameter. Call
460 mark_all_loops_for_removal and remove_low_level_allocnos. Pass
461 parameter to remove_unnecessary_regions.
462 (ira_build): Remove all regions but root if the conflict table was
463 not built. Update conflict hard regs for allocnos crossing calls.
465 * ira.c (ira_conflicts_p): New global.
466 (ira): Define and use ira_conflicts_p.
468 * reload1.c (compute_use_by_pseudos, reload, count_pseudo,
469 count_spilled_pseudo, find_reg, alter_reg, finish_spills,
470 emit_input_reload_insns, delete_output_reload): Use ira_conflicts_p.
472 2009-01-06 Ben Elliston <bje@au.ibm.com>
474 * gengtype-lex.l (YY_NO_INPUT): Define.
476 2009-01-05 Andrew Pinski <andrew_pinski@playstation.sony.com>
479 * c-common.c (handle_vector_size_attribute): Also reject
482 2009-01-05 Sebastian Pop <sebastian.pop@amd.com>
484 PR tree-optimization/38492
485 * graphite.c (rename_map_elt, debug_rename_elt,
486 debug_rename_map_1, debug_rename_map, new_rename_map_elt,
487 rename_map_elt_info, eq_rename_map_elts,
488 get_new_name_from_old_name, bb_in_sese_p): Moved around.
489 (sese_find_uses_to_rename_use): Renamed sese_build_livein_liveouts_use.
490 (sese_find_uses_to_rename_bb): Renamed sese_build_livein_liveouts_bb.
491 (sese_build_livein_liveouts): New.
492 (new_sese, free_sese): New.
493 (new_scop): Call new_sese.
494 (free_scop): Call free_sese.
495 (rename_variables_from_edge, rename_phis_end_scop): Removed.
496 (register_old_new_names): Renamed register_old_and_new_names.
497 (register_scop_liveout_renames, add_loop_exit_phis,
498 insert_loop_close_phis, struct igp,
499 default_liveout_before_guard, add_guard_exit_phis,
500 insert_guard_phis, copy_renames): New.
501 (translate_clast): Call insert_loop_close_phis and insert_guard_phis.
502 (sese_add_exit_phis_edge): Renamed scop_add_exit_phis_edge.
503 (rewrite_into_sese_closed_ssa): Renamed scop_insert_phis_for_liveouts.
504 (scop_adjust_phis_for_liveouts): New.
505 (gloog): Call scop_adjust_phis_for_liveouts.
507 * graphite.h (struct sese): Documented. Added fields liveout,
509 (SESE_LIVEOUT, SESE_LIVEIN, SESE_LIVEIN_VER, SESE_NUM_VER): New.
510 (new_sese, free_sese, sese_build_livein_liveouts): Declared.
511 (struct scop): Added field liveout_renames.
512 (SCOP_LIVEOUT_RENAMES): New.
514 2009-01-05 Harsha Jagasia <harsha.jagasia@amd.com>
516 PR tree-optimization/38510
517 * graphite.c (recompute_all_dominators): Call mark_irreducible_loops.
518 (translate_clast): Call recompute_all_dominators before
520 (gloog): Call recompute_all_dominators before graphite_verify.
522 2009-01-05 Harsha Jagasia <harsha.jagasia@amd.com>
523 Jan Sjodin <jan.sjodin@amd.com>
525 PR tree-optimization/38500
526 * graphite.c (create_sese_edges): Call fix_loop_structure after
529 2009-01-05 Joel Sherrill <joel.sherrill@oarcorp.com>
531 * config.gcc: Add m32r*-*-rtems*.
532 * config/m32r/rtems.h: New file.
534 2009-01-05 Ben Elliston <bje@au.ibm.com>
536 * Makefile.in (.po.gmo): Use mkinstalldirs, not test -d || mkdir.
538 (po/gcc.pot): Likewise.
540 2009-01-04 David S. Miller <davem@davemloft.net>
542 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
543 (STARTING_FRAME_OFFSET): Always set to zero.
545 2009-01-04 Richard Sandiford <rdsandiford@googlemail.com>
547 * tree.def (LSHIFT_EXPR, RSHIFT_EXPR): Add commentary.
548 * tree-cfg.c (verify_gimple_assign_binary): Allow shifts of
549 fixed-point types, and vectors of the same.
551 2009-01-04 Richard Sandiford <rdsandiford@googlemail.com>
553 * config/mips/sync.md (*mb_barrier): Rename to...
554 (*memory_barrier): ...this.
556 2009-01-04 Jonathan Wakely <jwakely.gcc@gmail.com>
558 * doc/extend.texi (Function Attributes): Move @cindex after @item
559 for 'artificial' and 'flatten'. Fix grammar for 'externally_visible'
560 and put in alphabetical order. Fix 'target' name and put in order.
561 * doc/invoke.texi (-Wstrict-null-sentinel, -fipa-matrix-reorg): Fix
564 2009-01-04 Uros Bizjak <ubizjak@gmail.com>
566 * config/s390/s390.md (UNSPEC_MB): Rename from UNSPECV_MB.
567 (memory_barrier): Expand as unspec instead of unspec_volatile.
568 Remove mem:BLK from insn operands. Use Pmode scratch register.
569 (*memory_barrier): Define as unspec instead of unspec_volatile.
570 Use (match_dup 0) as input operand.
572 * config/sparc/sparc.md (UNSPEC_MEMBAR): Rename from UNSPECV_MEMBAR.
573 * config/sparc/sync.md (memory_barrier): Expand as unspec instead of
574 unspec_volatile. Remove mem:BLK from insn operands. Use Pmode
575 scratch register. Remove operand 1.
576 (*stbar): Define as unspec instead of unspec_volatile.
577 Use (match_dup 0) as input operand, remove (const_int 8).
578 (*membar): Define as unspec instead of unspec_volatile.
579 Use (match_dup 0) as input operand, remove input operand 2.
581 * config/xtensa/xtensa.md (UNSPEC_MEMW): Rename from UNSPECV_MEMW.
582 (memory_barrier): Expand as unspec instead of unspec_volatile.
583 Remove mem:BLK from insn operands. Use Pmode scratch register.
584 (*memory_barrier): Define as unspec instead of unspec_volatile.
585 Use (match_dup 0) as input operand.
587 * config/ia64/sync.md (memory_barrier): Redefine as expander pattern.
588 Remove mem:BLK from insn operands. Use Pmode scratch register.
589 Set volatile flag on operand 0.
590 (*memory_barrier): New insn pattern.
592 * config/rs6000/sync.md (memory_barrier): Remove mem:BLK from
594 (*memory_barrier): Use (match_dup 0) as input operand.
596 * config/mips/sync.md (memory_barrier): Redefine as expander pattern.
597 Remove mem:BLK from insn operands. Use Pmode scratch register.
598 Set volatile flag on operand 0.
599 (*mb_internal): New insn pattern.
601 * config/alpha/sync.md (*memory_barrier): Rename from *mb_internal.
603 2009-01-04 Steven Bosscher <steven@gcc.gnu.org>
606 * function.c (struct temp_slot): Move to the section of the file
607 that deals with temp slots. Remove field 'address'.
608 (temp_slot_address_table): New hash table of address -> temp slot.
609 (struct temp_slot_address_entry): New struct, items for the table.
610 (temp_slot_address_compute_hash, temp_slot_address_hash,
611 temp_slot_address_eq, insert_temp_slot_address): Support functions
613 (find_temp_slot_from_address): Rewrite to use the new hash table.
614 (remove_unused_temp_slot_addresses): Remove addresses of temp
615 slots that have been made available.
616 (remove_unused_temp_slot_addresses_1): Call-back for htab_traverse,
617 worker function for remove_unused_temp_slot_addresses.
618 (assign_stack_temp_for_type): Don't clear the temp slot address list.
619 Add the temp slot address to the address -> temp slot map.
620 (update_temp_slot_address): Update via insert_temp_slot_address.
621 (free_temp_slots): Call remove_unused_temp_slot_addresses.
622 (pop_temp_slots): Likewise.
623 (init_temp_slots): Allocate the address -> temp slot map, or empty
624 the map if it is already allocated.
625 (prepare_function_start): Initialize temp slot processing.
627 2009-01-04 Steven Bosscher <steven@gcc.gnu.org>
630 * cfgexpand.c (estimate_stack_frame_size): Simplify the estimate:
631 Calculate the size of all stack vars assuming no packing of stack
632 vars will happen, replacing a quadratic algorithm with a linear one.
634 2009-01-03 Jakub Jelinek <jakub@redhat.com>
637 * expmed.c (store_bit_field_1): Don't modify op0 if movstrict insn
640 2009-01-03 Diego Novillo <dnovillo@google.com>
642 * doc/contrib.texi: Update contributions.
644 2009-01-03 Jakub Jelinek <jakub@redhat.com>
647 * builtins.c (fold_builtin_memory_op): Give up if either operand
648 is volatile. Set srctype or desttype to non-qualified version
652 * builtins.c (fold_builtin_expect): Only check DECL_WEAK for VAR_DECLs
655 2009-01-02 Kenneth Zadeck <zadeck@naturalbridge.com>
657 PR rtl-optimization/35805
658 * df-problems.c (df_lr_finalize): Add recursive call to resolve lr
659 problem if fast dce is able to remove any instructions.
660 * dce.c (dce_process_block): Fix dump message.
662 2009-01-02 Mark Mitchell <mark@codesourcery.com>
665 * tree-ssa-pre.c (compute_antic): Correct loop bounds.
667 2009-01-02 Jakub Jelinek <jakub@redhat.com>
670 * tree-flow.h (op_code_prio, op_prio): New prototypes.
671 * tree-pretty-print.c (op_code_prio): New function.
672 (op_prio): No longer static. Use op_code_prio.
673 * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs):
674 Use op_prio and op_code_prio to determine if () should be
675 printed around operand(s) or not.
677 * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
678 dump_gimple_call, dump_gimple_switch, dump_gimple_cond,
679 dump_gimple_label, dump_gimple_try, dump_symbols, dump_gimple_phi,
680 dump_gimple_mem_ops, dump_bb_header, dump_bb_end, pp_cfg_jump): Use
681 pp_character instead of pp_string for single letter printing.
683 2009-01-02 Richard Sandiford <rdsandiford@googlemail.com>
685 * doc/extend.texi: Fix '#pragma GCC option' typo.
687 2009-01-02 Richard Guenther <rguenther@suse.de>
689 * doc/install.texi (--enable-checking): Mention different
691 (--enable-stage1-checking): Document.
693 2009-01-01 Andrew Pinski <pinskia@gmail.com>
696 * tree-cfg.c (verify_expr): Add INDIRECT_REF case. Change MODIFY_EXPR
699 2009-01-02 Ben Elliston <bje@au.ibm.com>
701 * config/fp-bit.h (pack_d): Constify argument.
702 * config/fp-bit.c (makenan): Constify return type. Remove casts.
703 (isnan): Constify argument.
707 (_fpadd_parts): Constify return type.
708 (_fpmul_parts): Likewise.
709 (_fpdiv_parts): Likewise.
711 2009-01-01 Jakub Jelinek <jakub@redhat.com>
714 * c-typeck.c (add_pending_init): Add IMPLICIT argument. Only
715 warn about overwriting initializer with side-effects or
716 -Woverride-init if !IMPLICIT.
717 (output_init_element): Likewise. Pass IMPLICIT down to
719 (process_init_element): Add IMPLICIT argument. Pass it down
720 to output_init_element.
721 (push_init_element, pop_init_level, set_designator): Adjust
722 process_init_element callers.
723 (set_nonincremental_init, set_nonincremental_init_from_string):
724 Adjust add_pending_init callers.
725 (output_pending_init_elements): Adjust output_init_element callers.
726 * c-tree.h (process_init_element): Adjust prototype.
727 * c-parser.c (c_parser_initelt, c_parser_initval): Adjust
728 process_init_element callers.