1 2009-01-13 Uros Bizjak <ubizjak@gmail.com>
3 * config/alpha/alpha.c (alpha_legitimate_address_p): Explicit
4 relocations of local symbols wider than UNITS_PER_WORD are not valid.
5 (alpha_legitimize_address): Do not split local symbols wider than
6 UNITS_PER_WORD into HIGH/LO_SUM parts.
8 2009-01-13 Danny Smith <dannysmith@users.sourceforge.net>
11 * gcc.c (process_command): Replace call to execvp with calls
14 2009-01-03 Anatoly Sokolov <aesok@post.ru>
17 * config/avr/t-avr (LIB1ASMFUNCS): Add _tablejump_elpm.
18 * config/avr/libgcc.S (__do_global_ctors, __do_global_dtors): Add
19 variant for devices with 3-byte PC.
20 (__tablejump_elpm__) : New.
22 2009-01-12 Jakub Jelinek <jakub@redhat.com>
25 * c-parser.c (c_parser_postfix_expression): Allow `->' in
26 offsetof member-designator, handle it as `[0].'.
28 2009-01-12 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
30 * pa.c (pa_asm_output_mi_thunk): Use pc-relative branch to thunk
31 function when not using named sections on targets with named sections
32 if branch distance is less than 262132.
34 2009-01-12 Richard Earnshaw <rearnsha@arm.com>
36 * combine.c (combine_instructions): Recompute
37 optimize_this_for_speed_p for each BB in the main combine loop.
39 2009-01-12 Tomas Bily <tbily@suse.cz>
42 * tree-loop-distribution.c (prop_phis): New function.
43 (generate_builtin): Call prop_phis.
44 * testsuite/gcc.dg/tree-ssa/pr38385.c: New file.
46 2009-01-12 Jakub Jelinek <jakub@redhat.com>
48 PR tree-optimization/38807
49 * tree-ssa-reassoc.c (remove_visited_stmt_chain): Don't look at
50 gimple_visited_p unless stmt is GIMPLE_ASSIGN.
52 2009-01-11 Adam Nemet <anemet@caviumnetworks.com>
54 * expmed.c (store_bit_field_1): Properly truncate the paradoxical
55 subreg of op0 to the original op0.
57 2009-01-11 Laurent GUERBY <laurent@guerby.net>
59 * doc/sourcebuild.texi (Source Tree): Move up intl and fixinc.
61 2009-01-11 Markus Schoepflin <markus.schoepflin@comsoft.de>
64 * gcc/mips-tfile.c (parse_def): Fix parsing of def strings
67 2009-01-10 Jakub Jelinek <jakub@redhat.com>
70 * config/arm/arm.c (arm_is_long_call_p): Don't call
71 arm_function_in_section_p if decl isn't a FUNCTION_DECL.
73 2009-01-09 Steven Bosscher <steven@gcc.gnu.org>
75 * regrename.c (regrename_optimize): Fix dumping.
76 (find_oldest_value_reg): Preserve REG_POINTER.
77 (copy_hardreg_forward_1): Likewise.
79 2009-01-09 Diego Novillo <dnovillo@google.com>
81 * gimple.h (struct gimple_statement_base)<uid>: Document
82 the restrictions on its use.
84 (gimple_set_uid): Tidy.
86 2009-01-09 Jakub Jelinek <jakub@redhat.com>
88 * config/i386/i386.c (ix86_expand_movmem, ix86_expand_setmem): Add
89 zero guard even if align_bytes != 0 and count is smaller than
92 2008-01-09 Vladimir Makarov <vmakarov@redhat.com>
94 PR rtl-optimization/38495
95 * ira-emit.c (print_move_list, ira_debug_move_list): New functions.
96 (add_range_and_copies_from_move_list): Print all added ranges.
97 Add ranges to memory optimized destination.
99 2009-01-09 Jakub Jelinek <jakub@redhat.com>
103 * config/i386/i386.c (override_options): Reject
104 -mstringop-strategy=rep_8byte with -m32.
105 (ix86_expand_movmem): For size_needed == 1 set epilogue_size_needed
106 to 1. Do count comparison against epilogue_size_needed at compile
107 time even when count_exp was constant forced into register. For
108 size_needed don't jump to epilogue, instead just avoid aligning
109 and invoke the body algorithm. If need_zero_guard, add zero guard
110 even if count is non-zero, but smaller than size_needed + number of
111 bytes that could be stored for alignment.
112 (ix86_expand_setmem): For size_needed == 1 set epilogue_size_needed
113 to 1. If need_zero_guard, add zero guard even if count is non-zero,
114 but smaller than size_needed + number of bytes that could be stored
115 for alignment. Compare size_needed with epilogue_size_needed instead
116 of desired_align - align, don't adjust size_needed, pass
117 epilogue_size_needed to the epilogue expanders.
120 * c-pretty-print.c (pp_c_expression): Handle GOTO_EXPR like BIND_EXPR.
122 2009-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
124 * pa.c (last_address): Change to unsigned.
125 (update_total_code_bytes): Change argument to unsigned. Don't
126 check if insn addresses are set.
127 (pa_output_function_epilogue): Set last_address to UINT_MAX if insn
128 addresses are not set.
129 (pa_asm_output_mi_thunk): Handle wrap when updating last_address.
131 2009-01-09 Nick Clifton <nickc@redhat.com>
133 * config/sh/symbian.c: Replace uses of DECL_INLINE with
134 DECL_DECLARED_INLINE_P.
136 2009-01-09 Jakub Jelinek <jakub@redhat.com>
139 * dojump.c (do_jump_by_parts_zero_rtx): Use mode instead of
140 GET_MODE (op0) in operand_subword_force calls.
143 * fold-const.c (fold_unary): For COMPOUND_EXPR and COND_EXPR,
144 fold_convert arg0 operands to TREE_TYPE (op0) first.
146 2009-01-08 Vladimir Makarov <vmakarov@redhat.com>
148 * params.def (ira-max-conflict-table-size): Decrease default value
151 2009-01-08 Jakub Jelinek <jakub@redhat.com>
153 PR tree-optimization/37031
154 * lambda-code.c (lambda_collect_parameters): Call pointer_set_destroy
156 (build_access_matrix): Reserve correct size for AM_MATRIX vector,
157 allocate it using gc instead of heap, use VEC_quick_push instead of
159 * graphite.c (build_access_matrix): Allocate AM_MATRIX vector using gc
160 instead of heap, use VEC_quick_push instead of VEC_safe_push.
161 * tree-data-ref.h (struct access_matrix): Change matrix to gc
162 allocated vector from heap allocated.
163 * lambda.h: Add DEF_VEC_ALLOC_P for gc allocated lambda_vector.
164 * tree-loop-linear.c (linear_transform_loops): Allocate nest
165 vector only after perfect_loop_nest_depth call.
167 2009-01-08 Sebastian Pop <sebastian.pop@amd.com>
168 Jan Sjodin <jan.sjodin@amd.com>
170 PR tree-optimization/38559
171 * graphite.c (debug_value, copy_constraint,
172 swap_constraint_variables, scale_constraint_variable, ): New.
173 (get_lower_bound, get_upper_bound): Removed.
174 (graphite_trans_bb_strip_mine): Clean up this code that works
175 only for constant number of iterations. Fully copy upper and
176 lower bound constraints, not only the constant part of them.
177 * graphite.h (debug_value): Declared.
179 2009-01-08 Ira Rosen <irar@il.ibm.com>
181 PR tree-optimization/37194
182 * tree-vect-transform.c (vect_estimate_min_profitable_iters):
183 Don't add the cost of cost model guard in prologue to scalar
184 outside cost in case of known number of iterations.
186 2009-01-07 Nathan Froyd <froydnj@codesourcery.com>
187 Alan Modra <amodra@bigpond.net.au>
189 * config/rs6000/rs6000.c (rs6000_legitimize_address): Check for
190 non-word-aligned REG+CONST addressing.
192 2009-01-07 Uros Bizjak <ubizjak@gmail.com>
195 * config/alpha/alpha.c (alpha_end_function): For TARGET_ABI_OSF, call
196 free_after_compilation when outputting a thunk.
197 (alpha_output_mi_thunk_osf): Assert that we are processing a thunk.
198 Do not call free_after_compilation here.
200 2009-01-07 Uros Bizjak <ubizjak@gmail.com>
202 * config/i386/i386.c (ix86_target_string): Use ARRAY_SIZE.
203 (ix86_valid_target_attribute_inner_p): Ditto.
205 2009-01-07 Jan Sjodin <jan.sjodin@amd.com>
207 PR tree-optimization/38492
208 PR tree-optimization/38498
209 * tree-check.c (operator_is_linear, scev_is_linear_expression): New.
210 * tree-chrec.h (scev_is_linear_expression): Declared.
211 * graphite.c (graphite_cannot_represent_loop_niter): New.
212 (scopdet_basic_block_info): Call graphite_cannot_represent_loop_niter.
213 (graphite_loop_normal_form): Use gcc_assert.
214 (scan_tree_for_params): Use CASE_CONVERT.
215 (phi_node_is_iv, bb_contains_non_iv_scalar_phi_nodes): New.
216 (build_scop_conditions_1): Call bb_contains_non_iv_scalar_phi_nodes.
217 Use gcc_assert. Discard scops that contain unhandled cases.
218 (build_scop_conditions): Return a boolean status for unhandled cases.
219 (strip_mine_profitable_p): Print the loop number, not its depth.
220 (is_interchange_valid): Pass the depth of the loop nest, don't
221 recompute it wrongly.
222 (graphite_trans_bb_block): Same.
223 (graphite_trans_bb_block): Print tentative of loop blocking.
224 (graphite_trans_scop_block): Do not print that the loop has been
226 (graphite_transform_loops): Do not handle scops that contain condition
229 2009-01-07 H.J. Lu <hongjiu.lu@intel.com>
231 AVX Programming Reference (December, 2008)
232 * config/i386/avxintrin.h (_mm256_stream_si256): New.
233 (_mm256_stream_pd): Likewise.
234 (_mm256_stream_ps): Likewise.
236 * config/i386/i386.c (ix86_builtins): Add IX86_BUILTIN_MOVNTDQ256,
237 IX86_BUILTIN_MOVNTPD256 and IX86_BUILTIN_MOVNTPS256.
238 (ix86_special_builtin_type): Add VOID_FTYPE_PV4DI_V4DI.
239 (bdesc_special_args): Add __builtin_ia32_movntdq256,
240 __builtin_ia32_movntpd256 and __builtin_ia32_movntps256.
241 (ix86_init_mmx_sse_builtins): Handle VOID_FTYPE_PV4DI_V4DI.
242 (ix86_expand_special_args_builtin): Likewise.
244 * config/i386/sse.md (AVXMODEDI): New.
245 (avx_movnt<mode>): Likewise.
246 (avx_movnt<mode>): Likewise.
247 (<sse>_movnt<mode>): Remove AVX support.
248 (sse2_movntv2di): Likewise.
250 2009-01-07 Richard Guenther <rguenther@suse.de>
253 * fold-const.c (extract_muldiv): Remove obsolete comment.
254 (fold_plusminus_mult_expr): Undo MINUS_EXPR
255 to PLUS_EXPR canonicalization for the canonicalization.
257 2009-01-07 Gerald Pfeifer <gerald@pfeifer.com>
259 * doc/install.texi (alpha*-dec-osf*): Remove note on 32-bit
260 hosted cross-compilers generating less efficient code.
262 2009-01-06 Richard Sandiford <rdsandiford@googlemail.com>
264 * function.h (rtl_data): Add a dbr_scheduled_p field.
265 * reorg.c (dbr_schedule): Set it.
266 (gate_handle_delay_slots): Check it.
267 * config/mips/mips.c (mips_base_delayed_branch): Delete.
268 (mips_reorg): Check flag_delayed_branch instead of
269 mips_base_delayed_branch.
270 (mips_override_options): Don't set mips_base_delayed_branch
271 or flag_delayed_branch.
273 2009-01-06 Richard Sandiford <rdsandiford@googlemail.com>
275 PR rtl-optimization/38426.
276 * ira.c (ira): Set current_function_is_leaf earlier.
278 2009-01-06 Jakub Jelinek <jakub@redhat.com>
280 PR rtl-optimization/38722
281 * combine.c (try_combine): Don't modify PATTERN (i3) and notes
282 too early, only set a flag and modify after last possible
285 2009-01-06 Janis Johnson <janis187@us.ibm.com>
288 * ginclude/float.h: Rename DECnn_DEN to DECnn_SUBNORMAL_MIN.
289 * real.c (decimal_single_format): Correct values of emin and emax.
290 (decimal_double_format): Ditto.
291 (decimal_quad_format): Ditto.
292 * c-cppbuiltin.c (builtin_define_decimal_float_constants): Adjust
293 computation of DECnn_MIN and DECnn_MAX for corrected values of
294 emin and emax. Define __DECnn_SUBNORMAL_MIN__ instead of
295 __DECnn_MIN__, and adjust its computation for the corrected value
298 2009-01-06 Jan Hubicka <jh@suse.cz>
301 * i386.c (ix86_expand_call): Use ARRAY_SIZE.
303 2009-01-06 Gerald Pfeifer <gerald@pfeifer.com>
305 * doc/contrib.texi (Contributors): Slightly adjust the end note.
306 Add Robert Clark to the list of testers.
308 2009-01-06 Jan Hubicka <jh@suse.cz>
309 Kai Tietz <kai.tietz@onevision.com>
311 * i386.md (*msabi_syvabi): Add SSE regs clobbers.
312 * i386.c (ix86_expand_call): Add clobbers.
314 2009-01-06 Jan Hubicka <jh@suse.cz>
315 Kai Tietz <kai.tietz@onevision.com>
317 * i386.h (CONDITIONAL_CALL_USAGE): SSE regs are not used for w64 ABI.
318 * i386.c (struct ix86_frame): Add padding0 and nsseregs.
319 (ix86_nsaved_regs): Count only general purpose regs.
320 (ix86_nsaved_sseregs): New.
321 (ix86_compute_frame_layout): Update nsseregs; set preferred alignment
322 to 16 for w64; compute padding and size of sse reg save area.
323 (ix86_emit_save_regs, ix86_emit_save_regs_using_mov): Save only
324 general purpose regs.
325 (ix86_emit_save_sse_regs_using_mov): New.
326 (ix86_expand_prologue): Save SSE regs if needed.
327 (ix86_emit_restore_regs_using_mov): Use only general purpose regs.
328 (ix86_emit_restore_sse_regs_using_mov): New.
329 (ix86_expand_epilogue): Save SSE regs if needed.
331 2009-01-06 Jan Hubicka <jh@suse.cz>
332 Kai Tietz <kai.tietz@onevision.com>
334 * i386.h (ACCUMULATE_OUTGOING_ARGS): Enable for MSABI
335 * i386.c (init_cumulative_args): Disallow calls of MSABI functions
336 when accumulate outgoing args is off.
338 2009-01-06 H.J. Lu <hongjiu.lu@intel.com>
341 * ira-color.c (ira_reuse_stack_slot): Check ENABLE_IRA_CHECKING
342 before using pseudos_have_intersected_live_ranges_p.
344 * ira-int.h (ira_assert): Always define.
346 2009-01-06 H.J. Lu <hongjiu.lu@intel.com>
348 AVX Programming Reference (December, 2008)
349 * config/i386/avxintrin.h (_mm_permute2_pd): Removed.
350 (_mm256_permute2_pd): Likewise.
351 (_mm_permute2_ps): Likewise.
352 (_mm256_permute2_ps): Likewise.
353 * config/i386/i386.md (UNSPEC_VPERMIL2): Likewise.
354 * config/i386/sse.md (avx_vpermil2<mode>3): Likewise.
356 * config/i386/i386.c (ix86_builtins): Remove
357 IX86_BUILTIN_VPERMIL2PD, IX86_BUILTIN_VPERMIL2PS,
358 IX86_BUILTIN_VPERMIL2PD256 and IX86_BUILTIN_VPERMIL2PS256.
359 (ix86_builtin_type): Remove V8SF_FTYPE_V8SF_V8SF_V8SI_INT,
360 V4DF_FTYPE_V4DF_V4DF_V4DI_INT, V4SF_FTYPE_V4SF_V4SF_V4SI_INT
361 and V2DF_FTYPE_V2DF_V2DF_V2DI_INT.
362 (bdesc_args): Remove __builtin_ia32_vpermil2pd,
363 __builtin_ia32_vpermil2ps, __builtin_ia32_vpermil2pd256 and
364 __builtin_ia32_vpermil2ps256.
365 (ix86_init_mmx_sse_builtins): Updated.
366 (ix86_expand_args_builtin): Likewise.
368 2009-01-05 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
370 * pa.c (output_call): Relocate non-jump insns in the delay slot of
371 long absolute calls when generating PA 2.0 code.
373 2009-01-05 Vladimir Makarov <vmakarov@redhat.com>
375 PR rtl-optimization/38583
376 * params.h (IRA_MAX_CONFLICT_TABLE_SIZE): New macro.
378 * params.def (ira-max-conflict-table-size): New.
380 * doc/invoke.texi (ira-max-conflict-table-size): Decribe.
382 * ira.h (ira_conflicts_p): New external definition.
384 * ira-conflicts.c (build_conflict_bit_table): Do not build too big
385 table. Report this. Return result of building.
386 (ira_build_conflicts): Use ira_conflicts_p. Check result of
387 building conflict table.
389 * ira-color.c (fast_allocation): Use num instead of ira_allocnos_num.
390 (ira_color): Use ira_conflicts_p.
392 * global.c: Include ira.h.
393 (pseudo_for_reload_consideration_p, build_insn_chain): Use
396 * Makefile.in (global.o): Add ira.h.
398 * ira-build.c (mark_all_loops_for_removal,
399 propagate_some_info_from_allocno): New.
400 (remove_unnecessary_allocnos): Call
401 propagate_some_info_from_allocno.
402 (remove_low_level_allocnos): New.
403 (remove_unnecessary_regions): Add parameter. Call
404 mark_all_loops_for_removal and remove_low_level_allocnos. Pass
405 parameter to remove_unnecessary_regions.
406 (ira_build): Remove all regions but root if the conflict table was
407 not built. Update conflict hard regs for allocnos crossing calls.
409 * ira.c (ira_conflicts_p): New global.
410 (ira): Define and use ira_conflicts_p.
412 * reload1.c (compute_use_by_pseudos, reload, count_pseudo,
413 count_spilled_pseudo, find_reg, alter_reg, finish_spills,
414 emit_input_reload_insns, delete_output_reload): Use ira_conflicts_p.
416 2009-01-06 Ben Elliston <bje@au.ibm.com>
418 * gengtype-lex.l (YY_NO_INPUT): Define.
420 2009-01-05 Andrew Pinski <andrew_pinski@playstation.sony.com>
423 * c-common.c (handle_vector_size_attribute): Also reject
426 2009-01-05 Sebastian Pop <sebastian.pop@amd.com>
428 PR tree-optimization/38492
429 * graphite.c (rename_map_elt, debug_rename_elt,
430 debug_rename_map_1, debug_rename_map, new_rename_map_elt,
431 rename_map_elt_info, eq_rename_map_elts,
432 get_new_name_from_old_name, bb_in_sese_p): Moved around.
433 (sese_find_uses_to_rename_use): Renamed sese_build_livein_liveouts_use.
434 (sese_find_uses_to_rename_bb): Renamed sese_build_livein_liveouts_bb.
435 (sese_build_livein_liveouts): New.
436 (new_sese, free_sese): New.
437 (new_scop): Call new_sese.
438 (free_scop): Call free_sese.
439 (rename_variables_from_edge, rename_phis_end_scop): Removed.
440 (register_old_new_names): Renamed register_old_and_new_names.
441 (register_scop_liveout_renames, add_loop_exit_phis,
442 insert_loop_close_phis, struct igp,
443 default_liveout_before_guard, add_guard_exit_phis,
444 insert_guard_phis, copy_renames): New.
445 (translate_clast): Call insert_loop_close_phis and insert_guard_phis.
446 (sese_add_exit_phis_edge): Renamed scop_add_exit_phis_edge.
447 (rewrite_into_sese_closed_ssa): Renamed scop_insert_phis_for_liveouts.
448 (scop_adjust_phis_for_liveouts): New.
449 (gloog): Call scop_adjust_phis_for_liveouts.
451 * graphite.h (struct sese): Documented. Added fields liveout,
453 (SESE_LIVEOUT, SESE_LIVEIN, SESE_LIVEIN_VER, SESE_NUM_VER): New.
454 (new_sese, free_sese, sese_build_livein_liveouts): Declared.
455 (struct scop): Added field liveout_renames.
456 (SCOP_LIVEOUT_RENAMES): New.
458 2009-01-05 Harsha Jagasia <harsha.jagasia@amd.com>
460 PR tree-optimization/38510
461 * graphite.c (recompute_all_dominators): Call mark_irreducible_loops.
462 (translate_clast): Call recompute_all_dominators before
464 (gloog): Call recompute_all_dominators before graphite_verify.
466 2009-01-05 Harsha Jagasia <harsha.jagasia@amd.com>
467 Jan Sjodin <jan.sjodin@amd.com>
469 PR tree-optimization/38500
470 * graphite.c (create_sese_edges): Call fix_loop_structure after
473 2009-01-05 Joel Sherrill <joel.sherrill@oarcorp.com>
475 * config.gcc: Add m32r*-*-rtems*.
476 * config/m32r/rtems.h: New file.
478 2009-01-05 Ben Elliston <bje@au.ibm.com>
480 * Makefile.in (.po.gmo): Use mkinstalldirs, not test -d || mkdir.
482 (po/gcc.pot): Likewise.
484 2009-01-04 David S. Miller <davem@davemloft.net>
486 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
487 (STARTING_FRAME_OFFSET): Always set to zero.
489 2009-01-04 Richard Sandiford <rdsandiford@googlemail.com>
491 * tree.def (LSHIFT_EXPR, RSHIFT_EXPR): Add commentary.
492 * tree-cfg.c (verify_gimple_assign_binary): Allow shifts of
493 fixed-point types, and vectors of the same.
495 2009-01-04 Richard Sandiford <rdsandiford@googlemail.com>
497 * config/mips/sync.md (*mb_barrier): Rename to...
498 (*memory_barrier): ...this.
500 2009-01-04 Jonathan Wakely <jwakely.gcc@gmail.com>
502 * doc/extend.texi (Function Attributes): Move @cindex after @item
503 for 'artificial' and 'flatten'. Fix grammar for 'externally_visible'
504 and put in alphabetical order. Fix 'target' name and put in order.
505 * doc/invoke.texi (-Wstrict-null-sentinel, -fipa-matrix-reorg): Fix
508 2009-01-04 Uros Bizjak <ubizjak@gmail.com>
510 * config/s390/s390.md (UNSPEC_MB): Rename from UNSPECV_MB.
511 (memory_barrier): Expand as unspec instead of unspec_volatile.
512 Remove mem:BLK from insn operands. Use Pmode scratch register.
513 (*memory_barrier): Define as unspec instead of unspec_volatile.
514 Use (match_dup 0) as input operand.
516 * config/sparc/sparc.md (UNSPEC_MEMBAR): Rename from UNSPECV_MEMBAR.
517 * config/sparc/sync.md (memory_barrier): Expand as unspec instead of
518 unspec_volatile. Remove mem:BLK from insn operands. Use Pmode
519 scratch register. Remove operand 1.
520 (*stbar): Define as unspec instead of unspec_volatile.
521 Use (match_dup 0) as input operand, remove (const_int 8).
522 (*membar): Define as unspec instead of unspec_volatile.
523 Use (match_dup 0) as input operand, remove input operand 2.
525 * config/xtensa/xtensa.md (UNSPEC_MEMW): Rename from UNSPECV_MEMW.
526 (memory_barrier): Expand as unspec instead of unspec_volatile.
527 Remove mem:BLK from insn operands. Use Pmode scratch register.
528 (*memory_barrier): Define as unspec instead of unspec_volatile.
529 Use (match_dup 0) as input operand.
531 * config/ia64/sync.md (memory_barrier): Redefine as expander pattern.
532 Remove mem:BLK from insn operands. Use Pmode scratch register.
533 Set volatile flag on operand 0.
534 (*memory_barrier): New insn pattern.
536 * config/rs6000/sync.md (memory_barrier): Remove mem:BLK from
538 (*memory_barrier): Use (match_dup 0) as input operand.
540 * config/mips/sync.md (memory_barrier): Redefine as expander pattern.
541 Remove mem:BLK from insn operands. Use Pmode scratch register.
542 Set volatile flag on operand 0.
543 (*mb_internal): New insn pattern.
545 * config/alpha/sync.md (*memory_barrier): Rename from *mb_internal.
547 2009-01-04 Steven Bosscher <steven@gcc.gnu.org>
550 * function.c (struct temp_slot): Move to the section of the file
551 that deals with temp slots. Remove field 'address'.
552 (temp_slot_address_table): New hash table of address -> temp slot.
553 (struct temp_slot_address_entry): New struct, items for the table.
554 (temp_slot_address_compute_hash, temp_slot_address_hash,
555 temp_slot_address_eq, insert_temp_slot_address): Support functions
557 (find_temp_slot_from_address): Rewrite to use the new hash table.
558 (remove_unused_temp_slot_addresses): Remove addresses of temp
559 slots that have been made available.
560 (remove_unused_temp_slot_addresses_1): Call-back for htab_traverse,
561 worker function for remove_unused_temp_slot_addresses.
562 (assign_stack_temp_for_type): Don't clear the temp slot address list.
563 Add the temp slot address to the address -> temp slot map.
564 (update_temp_slot_address): Update via insert_temp_slot_address.
565 (free_temp_slots): Call remove_unused_temp_slot_addresses.
566 (pop_temp_slots): Likewise.
567 (init_temp_slots): Allocate the address -> temp slot map, or empty
568 the map if it is already allocated.
569 (prepare_function_start): Initialize temp slot processing.
571 2009-01-04 Steven Bosscher <steven@gcc.gnu.org>
574 * cfgexpand.c (estimate_stack_frame_size): Simplify the estimate:
575 Calculate the size of all stack vars assuming no packing of stack
576 vars will happen, replacing a quadratic algorithm with a linear one.
578 2009-01-03 Jakub Jelinek <jakub@redhat.com>
581 * expmed.c (store_bit_field_1): Don't modify op0 if movstrict insn
584 2009-01-03 Diego Novillo <dnovillo@google.com>
586 * doc/contrib.texi: Update contributions.
588 2009-01-03 Jakub Jelinek <jakub@redhat.com>
591 * builtins.c (fold_builtin_memory_op): Give up if either operand
592 is volatile. Set srctype or desttype to non-qualified version
596 * builtins.c (fold_builtin_expect): Only check DECL_WEAK for VAR_DECLs
599 2009-01-02 Kenneth Zadeck <zadeck@naturalbridge.com>
601 PR rtl-optimization/35805
602 * df-problems.c (df_lr_finalize): Add recursive call to resolve lr
603 problem if fast dce is able to remove any instructions.
604 * dce.c (dce_process_block): Fix dump message.
606 2009-01-02 Mark Mitchell <mark@codesourcery.com>
609 * tree-ssa-pre.c (compute_antic): Correct loop bounds.
611 2009-01-02 Jakub Jelinek <jakub@redhat.com>
614 * tree-flow.h (op_code_prio, op_prio): New prototypes.
615 * tree-pretty-print.c (op_code_prio): New function.
616 (op_prio): No longer static. Use op_code_prio.
617 * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs):
618 Use op_prio and op_code_prio to determine if () should be
619 printed around operand(s) or not.
621 * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
622 dump_gimple_call, dump_gimple_switch, dump_gimple_cond,
623 dump_gimple_label, dump_gimple_try, dump_symbols, dump_gimple_phi,
624 dump_gimple_mem_ops, dump_bb_header, dump_bb_end, pp_cfg_jump): Use
625 pp_character instead of pp_string for single letter printing.
627 2009-01-02 Richard Sandiford <rdsandiford@googlemail.com>
629 * doc/extend.texi: Fix '#pragma GCC option' typo.
631 2009-01-02 Richard Guenther <rguenther@suse.de>
633 * doc/install.texi (--enable-checking): Mention different
635 (--enable-stage1-checking): Document.
637 2009-01-01 Andrew Pinski <pinskia@gmail.com>
640 * tree-cfg.c (verify_expr): Add INDIRECT_REF case. Change MODIFY_EXPR
643 2009-01-02 Ben Elliston <bje@au.ibm.com>
645 * config/fp-bit.h (pack_d): Constify argument.
646 * config/fp-bit.c (makenan): Constify return type. Remove casts.
647 (isnan): Constify argument.
651 (_fpadd_parts): Constify return type.
652 (_fpmul_parts): Likewise.
653 (_fpdiv_parts): Likewise.
655 2009-01-01 Jakub Jelinek <jakub@redhat.com>
658 * c-typeck.c (add_pending_init): Add IMPLICIT argument. Only
659 warn about overwriting initializer with side-effects or
660 -Woverride-init if !IMPLICIT.
661 (output_init_element): Likewise. Pass IMPLICIT down to
663 (process_init_element): Add IMPLICIT argument. Pass it down
664 to output_init_element.
665 (push_init_element, pop_init_level, set_designator): Adjust
666 process_init_element callers.
667 (set_nonincremental_init, set_nonincremental_init_from_string):
668 Adjust add_pending_init callers.
669 (output_pending_init_elements): Adjust output_init_element callers.
670 * c-tree.h (process_init_element): Adjust prototype.
671 * c-parser.c (c_parser_initelt, c_parser_initval): Adjust
672 process_init_element callers.