1 2009-01-13 Danny Smith <dannysmith@users.sourceforge.net>
4 * gcc.c (process_command): Replace call to execvp with calls
7 2009-01-03 Anatoly Sokolov <aesok@post.ru>
10 * config/avr/t-avr (LIB1ASMFUNCS): Add _tablejump_elpm.
11 * config/avr/libgcc.S (__do_global_ctors, __do_global_dtors): Add
12 variant for devices with 3-byte PC.
13 (__tablejump_elpm__) : New.
15 2009-01-12 Jakub Jelinek <jakub@redhat.com>
18 * c-parser.c (c_parser_postfix_expression): Allow `->' in
19 offsetof member-designator, handle it as `[0].'.
21 2009-01-12 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
23 * pa.c (pa_asm_output_mi_thunk): Use pc-relative branch to thunk
24 function when not using named sections on targets with named sections
25 if branch distance is less than 262132.
27 2009-01-12 Richard Earnshaw <rearnsha@arm.com>
29 * combine.c (combine_instructions): Recompute
30 optimize_this_for_speed_p for each BB in the main combine loop.
32 2009-01-12 Tomas Bily <tbily@suse.cz>
35 * tree-loop-distribution.c (prop_phis): New function.
36 (generate_builtin): Call prop_phis.
37 * testsuite/gcc.dg/tree-ssa/pr38385.c: New file.
39 2009-01-12 Jakub Jelinek <jakub@redhat.com>
41 PR tree-optimization/38807
42 * tree-ssa-reassoc.c (remove_visited_stmt_chain): Don't look at
43 gimple_visited_p unless stmt is GIMPLE_ASSIGN.
45 2009-01-11 Adam Nemet <anemet@caviumnetworks.com>
47 * expmed.c (store_bit_field_1): Properly truncate the paradoxical
48 subreg of op0 to the original op0.
50 2009-01-11 Laurent GUERBY <laurent@guerby.net>
52 * doc/sourcebuild.texi (Source Tree): Move up intl and fixinc.
54 2009-01-11 Markus Schoepflin <markus.schoepflin@comsoft.de>
57 * gcc/mips-tfile.c (parse_def): Fix parsing of def strings
60 2009-01-10 Jakub Jelinek <jakub@redhat.com>
63 * config/arm/arm.c (arm_is_long_call_p): Don't call
64 arm_function_in_section_p if decl isn't a FUNCTION_DECL.
66 2009-01-09 Steven Bosscher <steven@gcc.gnu.org>
68 * regrename.c (regrename_optimize): Fix dumping.
69 (find_oldest_value_reg): Preserve REG_POINTER.
70 (copy_hardreg_forward_1): Likewise.
72 2009-01-09 Diego Novillo <dnovillo@google.com>
74 * gimple.h (struct gimple_statement_base)<uid>: Document
75 the restrictions on its use.
77 (gimple_set_uid): Tidy.
79 2009-01-09 Jakub Jelinek <jakub@redhat.com>
81 * config/i386/i386.c (ix86_expand_movmem, ix86_expand_setmem): Add
82 zero guard even if align_bytes != 0 and count is smaller than
85 2008-01-09 Vladimir Makarov <vmakarov@redhat.com>
87 PR rtl-optimization/38495
88 * ira-emit.c (print_move_list, ira_debug_move_list): New
90 (add_range_and_copies_from_move_list): Print all added ranges.
91 Add ranges to memory optimized destination.
93 2009-01-09 Jakub Jelinek <jakub@redhat.com>
97 * config/i386/i386.c (override_options): Reject
98 -mstringop-strategy=rep_8byte with -m32.
99 (ix86_expand_movmem): For size_needed == 1 set epilogue_size_needed
100 to 1. Do count comparison against epilogue_size_needed at compile
101 time even when count_exp was constant forced into register. For
102 size_needed don't jump to epilogue, instead just avoid aligning
103 and invoke the body algorithm. If need_zero_guard, add zero guard
104 even if count is non-zero, but smaller than size_needed + number of
105 bytes that could be stored for alignment.
106 (ix86_expand_setmem): For size_needed == 1 set epilogue_size_needed
107 to 1. If need_zero_guard, add zero guard even if count is non-zero,
108 but smaller than size_needed + number of bytes that could be stored
109 for alignment. Compare size_needed with epilogue_size_needed instead
110 of desired_align - align, don't adjust size_needed, pass
111 epilogue_size_needed to the epilogue expanders.
114 * c-pretty-print.c (pp_c_expression): Handle GOTO_EXPR like
117 2009-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
119 * pa.c (last_address): Change to unsigned.
120 (update_total_code_bytes): Change argument to unsigned. Don't
121 check if insn addresses are set.
122 (pa_output_function_epilogue): Set last_address to UINT_MAX if insn
123 addresses are not set.
124 (pa_asm_output_mi_thunk): Handle wrap when updating last_address.
126 2009-01-09 Nick Clifton <nickc@redhat.com>
128 * config/sh/symbian.c: Replace uses of DECL_INLINE with
129 DECL_DECLARED_INLINE_P.
131 2009-01-09 Jakub Jelinek <jakub@redhat.com>
134 * dojump.c (do_jump_by_parts_zero_rtx): Use mode instead of
135 GET_MODE (op0) in operand_subword_force calls.
138 * fold-const.c (fold_unary): For COMPOUND_EXPR and COND_EXPR,
139 fold_convert arg0 operands to TREE_TYPE (op0) first.
141 2009-01-08 Vladimir Makarov <vmakarov@redhat.com>
143 * params.def (ira-max-conflict-table-size): Decrease default value
146 2009-01-08 Jakub Jelinek <jakub@redhat.com>
148 PR tree-optimization/37031
149 * lambda-code.c (lambda_collect_parameters): Call pointer_set_destroy
151 (build_access_matrix): Reserve correct size for AM_MATRIX vector,
152 allocate it using gc instead of heap, use VEC_quick_push instead of
154 * graphite.c (build_access_matrix): Allocate AM_MATRIX vector using gc
155 instead of heap, use VEC_quick_push instead of VEC_safe_push.
156 * tree-data-ref.h (struct access_matrix): Change matrix to gc
157 allocated vector from heap allocated.
158 * lambda.h: Add DEF_VEC_ALLOC_P for gc allocated lambda_vector.
159 * tree-loop-linear.c (linear_transform_loops): Allocate nest
160 vector only after perfect_loop_nest_depth call.
162 2009-01-08 Sebastian Pop <sebastian.pop@amd.com>
163 Jan Sjodin <jan.sjodin@amd.com>
165 PR tree-optimization/38559
166 * graphite.c (debug_value, copy_constraint,
167 swap_constraint_variables, scale_constraint_variable, ): New.
168 (get_lower_bound, get_upper_bound): Removed.
169 (graphite_trans_bb_strip_mine): Clean up this code that works
170 only for constant number of iterations. Fully copy upper and
171 lower bound constraints, not only the constant part of them.
172 * graphite.h (debug_value): Declared.
174 2009-01-08 Ira Rosen <irar@il.ibm.com>
176 PR tree-optimization/37194
177 * tree-vect-transform.c (vect_estimate_min_profitable_iters):
178 Don't add the cost of cost model guard in prologue to scalar
179 outside cost in case of known number of iterations.
181 2009-01-07 Nathan Froyd <froydnj@codesourcery.com>
182 Alan Modra <amodra@bigpond.net.au>
184 * config/rs6000/rs6000.c (rs6000_legitimize_address): Check for
185 non-word-aligned REG+CONST addressing.
187 2009-01-07 Uros Bizjak <ubizjak@gmail.com>
190 * config/alpha/alpha.c (alpha_end_function): For TARGET_ABI_OSF, call
191 free_after_compilation when outputting a thunk.
192 (alpha_output_mi_thunk_osf): Assert that we are processing a thunk.
193 Do not call free_after_compilation here.
195 2009-01-07 Uros Bizjak <ubizjak@gmail.com>
197 * config/i386/i386.c (ix86_target_string): Use ARRAY_SIZE.
198 (ix86_valid_target_attribute_inner_p): Ditto.
200 2009-01-07 Jan Sjodin <jan.sjodin@amd.com>
202 PR tree-optimization/38492
203 PR tree-optimization/38498
204 * tree-check.c (operator_is_linear, scev_is_linear_expression): New.
205 * tree-chrec.h (scev_is_linear_expression): Declared.
206 * graphite.c (graphite_cannot_represent_loop_niter): New.
207 (scopdet_basic_block_info): Call graphite_cannot_represent_loop_niter.
208 (graphite_loop_normal_form): Use gcc_assert.
209 (scan_tree_for_params): Use CASE_CONVERT.
210 (phi_node_is_iv, bb_contains_non_iv_scalar_phi_nodes): New.
211 (build_scop_conditions_1): Call bb_contains_non_iv_scalar_phi_nodes.
212 Use gcc_assert. Discard scops that contain unhandled cases.
213 (build_scop_conditions): Return a boolean status for unhandled cases.
214 (strip_mine_profitable_p): Print the loop number, not its depth.
215 (is_interchange_valid): Pass the depth of the loop nest, don't
216 recompute it wrongly.
217 (graphite_trans_bb_block): Same.
218 (graphite_trans_bb_block): Print tentative of loop blocking.
219 (graphite_trans_scop_block): Do not print that the loop has been
221 (graphite_transform_loops): Do not handle scops that contain condition
224 2009-01-07 H.J. Lu <hongjiu.lu@intel.com>
226 AVX Programming Reference (December, 2008)
227 * config/i386/avxintrin.h (_mm256_stream_si256): New.
228 (_mm256_stream_pd): Likewise.
229 (_mm256_stream_ps): Likewise.
231 * config/i386/i386.c (ix86_builtins): Add IX86_BUILTIN_MOVNTDQ256,
232 IX86_BUILTIN_MOVNTPD256 and IX86_BUILTIN_MOVNTPS256.
233 (ix86_special_builtin_type): Add VOID_FTYPE_PV4DI_V4DI.
234 (bdesc_special_args): Add __builtin_ia32_movntdq256,
235 __builtin_ia32_movntpd256 and __builtin_ia32_movntps256.
236 (ix86_init_mmx_sse_builtins): Handle VOID_FTYPE_PV4DI_V4DI.
237 (ix86_expand_special_args_builtin): Likewise.
239 * config/i386/sse.md (AVXMODEDI): New.
240 (avx_movnt<mode>): Likewise.
241 (avx_movnt<mode>): Likewise.
242 (<sse>_movnt<mode>): Remove AVX support.
243 (sse2_movntv2di): Likewise.
245 2009-01-07 Richard Guenther <rguenther@suse.de>
248 * fold-const.c (extract_muldiv): Remove obsolete comment.
249 (fold_plusminus_mult_expr): Undo MINUS_EXPR
250 to PLUS_EXPR canonicalization for the canonicalization.
252 2009-01-07 Gerald Pfeifer <gerald@pfeifer.com>
254 * doc/install.texi (alpha*-dec-osf*): Remove note on 32-bit
255 hosted cross-compilers generating less efficient code.
257 2009-01-06 Richard Sandiford <rdsandiford@googlemail.com>
259 * function.h (rtl_data): Add a dbr_scheduled_p field.
260 * reorg.c (dbr_schedule): Set it.
261 (gate_handle_delay_slots): Check it.
262 * config/mips/mips.c (mips_base_delayed_branch): Delete.
263 (mips_reorg): Check flag_delayed_branch instead of
264 mips_base_delayed_branch.
265 (mips_override_options): Don't set mips_base_delayed_branch
266 or flag_delayed_branch.
268 2009-01-06 Richard Sandiford <rdsandiford@googlemail.com>
270 PR rtl-optimization/38426.
271 * ira.c (ira): Set current_function_is_leaf earlier.
273 2009-01-06 Jakub Jelinek <jakub@redhat.com>
275 PR rtl-optimization/38722
276 * combine.c (try_combine): Don't modify PATTERN (i3) and notes
277 too early, only set a flag and modify after last possible
280 2009-01-06 Janis Johnson <janis187@us.ibm.com>
283 * ginclude/float.h: Rename DECnn_DEN to DECnn_SUBNORMAL_MIN.
284 * real.c (decimal_single_format): Correct values of emin and emax.
285 (decimal_double_format): Ditto.
286 (decimal_quad_format): Ditto.
287 * c-cppbuiltin.c (builtin_define_decimal_float_constants): Adjust
288 computation of DECnn_MIN and DECnn_MAX for corrected values of
289 emin and emax. Define __DECnn_SUBNORMAL_MIN__ instead of
290 __DECnn_MIN__, and adjust its computation for the corrected value
293 2009-01-06 Jan Hubicka <jh@suse.cz>
296 * i386.c (ix86_expand_call): Use ARRAY_SIZE.
298 2009-01-06 Gerald Pfeifer <gerald@pfeifer.com>
300 * doc/contrib.texi (Contributors): Slightly adjust the end note.
301 Add Robert Clark to the list of testers.
303 2009-01-06 Jan Hubicka <jh@suse.cz>
304 Kai Tietz <kai.tietz@onevision.com>
306 * i386.md (*msabi_syvabi): Add SSE regs clobbers.
307 * i386.c (ix86_expand_call): Add clobbers.
309 2009-01-06 Jan Hubicka <jh@suse.cz>
310 Kai Tietz <kai.tietz@onevision.com>
312 * i386.h (CONDITIONAL_CALL_USAGE): SSE regs are not used for w64 ABI.
313 * i386.c (struct ix86_frame): Add padding0 and nsseregs.
314 (ix86_nsaved_regs): Count only general purpose regs.
315 (ix86_nsaved_sseregs): New.
316 (ix86_compute_frame_layout): Update nsseregs; set preferred alignment
317 to 16 for w64; compute padding and size of sse reg save area.
318 (ix86_emit_save_regs, ix86_emit_save_regs_using_mov): Save only
319 general purpose regs.
320 (ix86_emit_save_sse_regs_using_mov): New.
321 (ix86_expand_prologue): Save SSE regs if needed.
322 (ix86_emit_restore_regs_using_mov): Use only general purpose regs.
323 (ix86_emit_restore_sse_regs_using_mov): New.
324 (ix86_expand_epilogue): Save SSE regs if needed.
326 2009-01-06 Jan Hubicka <jh@suse.cz>
327 Kai Tietz <kai.tietz@onevision.com>
329 * i386.h (ACCUMULATE_OUTGOING_ARGS): Enable for MSABI
330 * i386.c (init_cumulative_args): Disallow calls of MSABI functions
331 when accumulate outgoing args is off.
333 2009-01-06 H.J. Lu <hongjiu.lu@intel.com>
336 * ira-color.c (ira_reuse_stack_slot): Check ENABLE_IRA_CHECKING
337 before using pseudos_have_intersected_live_ranges_p.
339 * ira-int.h (ira_assert): Always define.
341 2009-01-06 H.J. Lu <hongjiu.lu@intel.com>
343 AVX Programming Reference (December, 2008)
344 * config/i386/avxintrin.h (_mm_permute2_pd): Removed.
345 (_mm256_permute2_pd): Likewise.
346 (_mm_permute2_ps): Likewise.
347 (_mm256_permute2_ps): Likewise.
348 * config/i386/i386.md (UNSPEC_VPERMIL2): Likewise.
349 * config/i386/sse.md (avx_vpermil2<mode>3): Likewise.
351 * config/i386/i386.c (ix86_builtins): Remove
352 IX86_BUILTIN_VPERMIL2PD, IX86_BUILTIN_VPERMIL2PS,
353 IX86_BUILTIN_VPERMIL2PD256 and IX86_BUILTIN_VPERMIL2PS256.
354 (ix86_builtin_type): Remove V8SF_FTYPE_V8SF_V8SF_V8SI_INT,
355 V4DF_FTYPE_V4DF_V4DF_V4DI_INT, V4SF_FTYPE_V4SF_V4SF_V4SI_INT
356 and V2DF_FTYPE_V2DF_V2DF_V2DI_INT.
357 (bdesc_args): Remove __builtin_ia32_vpermil2pd,
358 __builtin_ia32_vpermil2ps, __builtin_ia32_vpermil2pd256 and
359 __builtin_ia32_vpermil2ps256.
360 (ix86_init_mmx_sse_builtins): Updated.
361 (ix86_expand_args_builtin): Likewise.
363 2009-01-05 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
365 * pa.c (output_call): Relocate non-jump insns in the delay slot of
366 long absolute calls when generating PA 2.0 code.
368 2009-01-05 Vladimir Makarov <vmakarov@redhat.com>
370 PR rtl-optimization/38583
371 * params.h (IRA_MAX_CONFLICT_TABLE_SIZE): New macro.
373 * params.def (ira-max-conflict-table-size): New.
375 * doc/invoke.texi (ira-max-conflict-table-size): Decribe.
377 * ira.h (ira_conflicts_p): New external definition.
379 * ira-conflicts.c (build_conflict_bit_table): Do not build too big
380 table. Report this. Return result of building.
381 (ira_build_conflicts): Use ira_conflicts_p. Check result of
382 building conflict table.
384 * ira-color.c (fast_allocation): Use num instead of ira_allocnos_num.
385 (ira_color): Use ira_conflicts_p.
387 * global.c: Include ira.h.
388 (pseudo_for_reload_consideration_p, build_insn_chain): Use
391 * Makefile.in (global.o): Add ira.h.
393 * ira-build.c (mark_all_loops_for_removal,
394 propagate_some_info_from_allocno): New.
395 (remove_unnecessary_allocnos): Call
396 propagate_some_info_from_allocno.
397 (remove_low_level_allocnos): New.
398 (remove_unnecessary_regions): Add parameter. Call
399 mark_all_loops_for_removal and remove_low_level_allocnos. Pass
400 parameter to remove_unnecessary_regions.
401 (ira_build): Remove all regions but root if the conflict table was
402 not built. Update conflict hard regs for allocnos crossing calls.
404 * ira.c (ira_conflicts_p): New global.
405 (ira): Define and use ira_conflicts_p.
407 * reload1.c (compute_use_by_pseudos, reload, count_pseudo,
408 count_spilled_pseudo, find_reg, alter_reg, finish_spills,
409 emit_input_reload_insns, delete_output_reload): Use ira_conflicts_p.
411 2009-01-06 Ben Elliston <bje@au.ibm.com>
413 * gengtype-lex.l (YY_NO_INPUT): Define.
415 2009-01-05 Andrew Pinski <andrew_pinski@playstation.sony.com>
418 * c-common.c (handle_vector_size_attribute): Also reject
421 2009-01-05 Sebastian Pop <sebastian.pop@amd.com>
423 PR tree-optimization/38492
424 * graphite.c (rename_map_elt, debug_rename_elt,
425 debug_rename_map_1, debug_rename_map, new_rename_map_elt,
426 rename_map_elt_info, eq_rename_map_elts,
427 get_new_name_from_old_name, bb_in_sese_p): Moved around.
428 (sese_find_uses_to_rename_use): Renamed sese_build_livein_liveouts_use.
429 (sese_find_uses_to_rename_bb): Renamed sese_build_livein_liveouts_bb.
430 (sese_build_livein_liveouts): New.
431 (new_sese, free_sese): New.
432 (new_scop): Call new_sese.
433 (free_scop): Call free_sese.
434 (rename_variables_from_edge, rename_phis_end_scop): Removed.
435 (register_old_new_names): Renamed register_old_and_new_names.
436 (register_scop_liveout_renames, add_loop_exit_phis,
437 insert_loop_close_phis, struct igp,
438 default_liveout_before_guard, add_guard_exit_phis,
439 insert_guard_phis, copy_renames): New.
440 (translate_clast): Call insert_loop_close_phis and insert_guard_phis.
441 (sese_add_exit_phis_edge): Renamed scop_add_exit_phis_edge.
442 (rewrite_into_sese_closed_ssa): Renamed scop_insert_phis_for_liveouts.
443 (scop_adjust_phis_for_liveouts): New.
444 (gloog): Call scop_adjust_phis_for_liveouts.
446 * graphite.h (struct sese): Documented. Added fields liveout,
448 (SESE_LIVEOUT, SESE_LIVEIN, SESE_LIVEIN_VER, SESE_NUM_VER): New.
449 (new_sese, free_sese, sese_build_livein_liveouts): Declared.
450 (struct scop): Added field liveout_renames.
451 (SCOP_LIVEOUT_RENAMES): New.
453 2009-01-05 Harsha Jagasia <harsha.jagasia@amd.com>
455 PR tree-optimization/38510
456 * graphite.c (recompute_all_dominators): Call mark_irreducible_loops.
457 (translate_clast): Call recompute_all_dominators before
459 (gloog): Call recompute_all_dominators before graphite_verify.
461 2009-01-05 Harsha Jagasia <harsha.jagasia@amd.com>
462 Jan Sjodin <jan.sjodin@amd.com>
464 PR tree-optimization/38500
465 * graphite.c (create_sese_edges): Call fix_loop_structure after
468 2009-01-05 Joel Sherrill <joel.sherrill@oarcorp.com>
470 * config.gcc: Add m32r*-*-rtems*.
471 * config/m32r/rtems.h: New file.
473 2009-01-05 Ben Elliston <bje@au.ibm.com>
475 * Makefile.in (.po.gmo): Use mkinstalldirs, not test -d || mkdir.
477 (po/gcc.pot): Likewise.
479 2009-01-04 David S. Miller <davem@davemloft.net>
481 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
482 (STARTING_FRAME_OFFSET): Always set to zero.
484 2009-01-04 Richard Sandiford <rdsandiford@googlemail.com>
486 * tree.def (LSHIFT_EXPR, RSHIFT_EXPR): Add commentary.
487 * tree-cfg.c (verify_gimple_assign_binary): Allow shifts of
488 fixed-point types, and vectors of the same.
490 2009-01-04 Richard Sandiford <rdsandiford@googlemail.com>
492 * config/mips/sync.md (*mb_barrier): Rename to...
493 (*memory_barrier): ...this.
495 2009-01-04 Jonathan Wakely <jwakely.gcc@gmail.com>
497 * doc/extend.texi (Function Attributes): Move @cindex after @item
498 for 'artificial' and 'flatten'. Fix grammar for 'externally_visible'
499 and put in alphabetical order. Fix 'target' name and put in order.
500 * doc/invoke.texi (-Wstrict-null-sentinel, -fipa-matrix-reorg): Fix
503 2009-01-04 Uros Bizjak <ubizjak@gmail.com>
505 * config/s390/s390.md (UNSPEC_MB): Rename from UNSPECV_MB.
506 (memory_barrier): Expand as unspec instead of unspec_volatile.
507 Remove mem:BLK from insn operands. Use Pmode scratch register.
508 (*memory_barrier): Define as unspec instead of unspec_volatile.
509 Use (match_dup 0) as input operand.
511 * config/sparc/sparc.md (UNSPEC_MEMBAR): Rename from UNSPECV_MEMBAR.
512 * config/sparc/sync.md (memory_barrier): Expand as unspec instead of
513 unspec_volatile. Remove mem:BLK from insn operands. Use Pmode
514 scratch register. Remove operand 1.
515 (*stbar): Define as unspec instead of unspec_volatile.
516 Use (match_dup 0) as input operand, remove (const_int 8).
517 (*membar): Define as unspec instead of unspec_volatile.
518 Use (match_dup 0) as input operand, remove input operand 2.
520 * config/xtensa/xtensa.md (UNSPEC_MEMW): Rename from UNSPECV_MEMW.
521 (memory_barrier): Expand as unspec instead of unspec_volatile.
522 Remove mem:BLK from insn operands. Use Pmode scratch register.
523 (*memory_barrier): Define as unspec instead of unspec_volatile.
524 Use (match_dup 0) as input operand.
526 * config/ia64/sync.md (memory_barrier): Redefine as expander pattern.
527 Remove mem:BLK from insn operands. Use Pmode scratch register.
528 Set volatile flag on operand 0.
529 (*memory_barrier): New insn pattern.
531 * config/rs6000/sync.md (memory_barrier): Remove mem:BLK from
533 (*memory_barrier): Use (match_dup 0) as input operand.
535 * config/mips/sync.md (memory_barrier): Redefine as expander pattern.
536 Remove mem:BLK from insn operands. Use Pmode scratch register.
537 Set volatile flag on operand 0.
538 (*mb_internal): New insn pattern.
540 * config/alpha/sync.md (*memory_barrier): Rename from *mb_internal.
542 2009-01-04 Steven Bosscher <steven@gcc.gnu.org>
545 * function.c (struct temp_slot): Move to the section of the file
546 that deals with temp slots. Remove field 'address'.
547 (temp_slot_address_table): New hash table of address -> temp slot.
548 (struct temp_slot_address_entry): New struct, items for the table.
549 (temp_slot_address_compute_hash, temp_slot_address_hash,
550 temp_slot_address_eq, insert_temp_slot_address): Support functions
552 (find_temp_slot_from_address): Rewrite to use the new hash table.
553 (remove_unused_temp_slot_addresses): Remove addresses of temp
554 slots that have been made available.
555 (remove_unused_temp_slot_addresses_1): Call-back for htab_traverse,
556 worker function for remove_unused_temp_slot_addresses.
557 (assign_stack_temp_for_type): Don't clear the temp slot address list.
558 Add the temp slot address to the address -> temp slot map.
559 (update_temp_slot_address): Update via insert_temp_slot_address.
560 (free_temp_slots): Call remove_unused_temp_slot_addresses.
561 (pop_temp_slots): Likewise.
562 (init_temp_slots): Allocate the address -> temp slot map, or empty
563 the map if it is already allocated.
564 (prepare_function_start): Initialize temp slot processing.
566 2009-01-04 Steven Bosscher <steven@gcc.gnu.org>
569 * cfgexpand.c (estimate_stack_frame_size): Simplify the estimate:
570 Calculate the size of all stack vars assuming no packing of stack
571 vars will happen, replacing a quadratic algorithm with a linear one.
573 2009-01-03 Jakub Jelinek <jakub@redhat.com>
576 * expmed.c (store_bit_field_1): Don't modify op0 if movstrict insn
579 2009-01-03 Diego Novillo <dnovillo@google.com>
581 * doc/contrib.texi: Update contributions.
583 2009-01-03 Jakub Jelinek <jakub@redhat.com>
586 * builtins.c (fold_builtin_memory_op): Give up if either operand
587 is volatile. Set srctype or desttype to non-qualified version
591 * builtins.c (fold_builtin_expect): Only check DECL_WEAK for VAR_DECLs
594 2009-01-02 Kenneth Zadeck <zadeck@naturalbridge.com>
596 PR rtl-optimization/35805
597 * df-problems.c (df_lr_finalize): Add recursive call to resolve lr
598 problem if fast dce is able to remove any instructions.
599 * dce.c (dce_process_block): Fix dump message.
601 2009-01-02 Mark Mitchell <mark@codesourcery.com>
604 * tree-ssa-pre.c (compute_antic): Correct loop bounds.
606 2009-01-02 Jakub Jelinek <jakub@redhat.com>
609 * tree-flow.h (op_code_prio, op_prio): New prototypes.
610 * tree-pretty-print.c (op_code_prio): New function.
611 (op_prio): No longer static. Use op_code_prio.
612 * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs):
613 Use op_prio and op_code_prio to determine if () should be
614 printed around operand(s) or not.
616 * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
617 dump_gimple_call, dump_gimple_switch, dump_gimple_cond,
618 dump_gimple_label, dump_gimple_try, dump_symbols, dump_gimple_phi,
619 dump_gimple_mem_ops, dump_bb_header, dump_bb_end, pp_cfg_jump): Use
620 pp_character instead of pp_string for single letter printing.
622 2009-01-02 Richard Sandiford <rdsandiford@googlemail.com>
624 * doc/extend.texi: Fix '#pragma GCC option' typo.
626 2009-01-02 Richard Guenther <rguenther@suse.de>
628 * doc/install.texi (--enable-checking): Mention different
630 (--enable-stage1-checking): Document.
632 2009-01-01 Andrew Pinski <pinskia@gmail.com>
635 * tree-cfg.c (verify_expr): Add INDIRECT_REF case. Change MODIFY_EXPR
638 2009-01-02 Ben Elliston <bje@au.ibm.com>
640 * config/fp-bit.h (pack_d): Constify argument.
641 * config/fp-bit.c (makenan): Constify return type. Remove casts.
642 (isnan): Constify argument.
646 (_fpadd_parts): Constify return type.
647 (_fpmul_parts): Likewise.
648 (_fpdiv_parts): Likewise.
650 2009-01-01 Jakub Jelinek <jakub@redhat.com>
653 * c-typeck.c (add_pending_init): Add IMPLICIT argument. Only
654 warn about overwriting initializer with side-effects or
655 -Woverride-init if !IMPLICIT.
656 (output_init_element): Likewise. Pass IMPLICIT down to
658 (process_init_element): Add IMPLICIT argument. Pass it down
659 to output_init_element.
660 (push_init_element, pop_init_level, set_designator): Adjust
661 process_init_element callers.
662 (set_nonincremental_init, set_nonincremental_init_from_string):
663 Adjust add_pending_init callers.
664 (output_pending_init_elements): Adjust output_init_element callers.
665 * c-tree.h (process_init_element): Adjust prototype.
666 * c-parser.c (c_parser_initelt, c_parser_initval): Adjust
667 process_init_element callers.