1 2007-02-07 Ian Lance Taylor <iant@google.com>
3 * lower-subreg.c (simple_move): Reject PARTIAL_INT modes.
5 2007-02-07 Roger Sayle <roger@eyesopen.com>
7 * config/rs6000/rs6000.md (ctz<mode>2, ffs<mode>2, popcount<mode>2,
8 parity<mode>2, smulsi3_highpart, abstf2_internal, allocate_stack,
9 tablejumpdi, movsi_to_cr_one): Remove constraints from
10 define_expand's match_operands.
12 2007-02-07 Roger Sayle <roger@eyesopen.com>
14 * global.c (compute_regsets): Move declatation of "i" inside of
15 #ifdef ELIMINABLE_REGS to avoid unused variable bootstrap failure.
17 2007-02-07 Jakub Jelinek <jakub@redhat.com>
20 * gimplify.c (gimplify_scan_omp_clauses): Remove special casing
21 of INDIRECT_REF <RESULT_DECL>.
23 * config/i386/i386.c (override_options): Set PTA_SSSE3 for core2.
25 2007-02-06 J"orn Rennecke <joern.rennecke@arc.com>
26 Kaz Kojima <kkojima@gcc.gnu.org>
29 * config/sh/sh.c (expand_cbranchdi4): Use scratch register
31 (sh_initialize_trampoline): Add parentheses to avoid a warning.
33 2007-02-06 Zdenek Dvorak <dvorakz@suse.cz>
35 * doc/loop.texi: Document possibility not to perform disambiguation
36 of loops with multiple latches.
37 * cfgloopmanip.c (alp_enum_p): Removed.
38 (add_loop): Handle subloops. Use get_loop_body_with_size.
39 (create_preheader): Do not allow ENTRY_BLOCK_PTR to be preheader.
40 * cfghooks.c (redirect_edge_and_branch_force): Set dominator for
41 the new forwarder block.
42 (make_forwarder_block): Only call new_bb_cbk if it is not NULL.
43 Handle the case latch is NULL.
44 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Avoid cfg modifications
45 when marking loop exits.
46 * ifcvt.c (if_convert): Ditto. Mark loop exits even if cfg cannot
48 * loop-init.c (loop_optimizer_init): Do not modify cfg. Call
49 disambiguate_loops_with_multiple_latches.
50 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Calculate dominators
51 before fix_loop_structure.
52 * cfgloop.c: Include pointer-set.h and output.h.
53 (canonicalize_loop_headers, HEADER_BLOCK, LATCH_EDGE,
54 update_latch_info, mfb_keep_just, mfb_keep_nonlatch): Removed.
55 (get_loop_latch_edges, find_subloop_latch_edge_by_profile,
56 find_subloop_latch_edge_by_ivs, find_subloop_latch_edge,
57 mfb_redirect_edges_in_set, form_subloop, merge_latch_edges,
58 disambiguate_multiple_latches, get_loop_body_with_size,
59 disambiguate_loops_with_multiple_latches): New functions.
60 (flow_loop_dump): Dump multiple latch edges.
61 (flow_loop_nodes_find): Handle loops with multiple latches.
62 (flow_loops_find): Ditto. Do not call canonicalize_loop_headers.
63 (glb_enum_p): Modified.
64 (get_loop_body): Use get_loop_body_with_size.
65 * cfgloop.h (LOOPS_HAVE_RECORDED_EXITS): New flag.
66 (AVOID_CFG_MODIFICATIONS): New constant.
67 (disambiguate_loops_with_multiple_latches, add_loop,
68 get_loop_body_with_size): Declare.
69 * Makefile.in (cfgloop.o): Add pointer-set.h and output.h.
71 2007-02-06 Seongbae Park <seongbae.park@gmail.com>
74 * global.c (compute_regsets): New function.
75 (global_alloc): Refactored ELIMINABLE_REGSET
76 and NO_GLOBAL_ALLOC_REGS computation out.
77 (rest_of_handle_global_alloc): Call compute_regsets()
78 for non-optimizing case.
80 2007-02-06 Richard Henderson <rth@redhat.com>
82 * config/i386/constraints.md (Y2): Rename from Y.
84 * config/i386/i386.md (movsi_1, movdi_2, pushdf_nointeger,
85 pushdf_integer, movdf_nointeger, movdf_integer, zero_extendsidi2_32,
86 zero_extendsidi2_rex64, truncxfdf2_mixed): Change Y constraints to Y2.
87 (extendsfdf2_mixed, extendsfdf2_sse, truncdfsf_fast_mixed,
88 truncdfsf_fast_sse, truncdfsf_mixed, fix_truncdfdi_sse,
89 fix_truncdfsi_sse, floatsidf2_mixed, floatsidf2_sse,
90 floatdidf2_mixed, floatdidf2_sse, absnegdf2_mixed,
91 absnegdf2_sse, sse_setccdf, fop_df_comm_mixed, fop_df_comm_sse,
92 fop_df_1_mixed, fop_df_1_sse): Change Y constraints to x.
93 * config/i386/mmx.md (mov<MMXMODEI>_internal_rex64,
94 mov<MMXMODEI>_internal, movv2sf_internal_rex64, movv2sf_internal,
95 vec_extractv2si_1): Change Y constraints to Y2.
96 * config/i386/sse.md (vec_setv4sf_0, vec_concatv2df, vec_dupv4si,
97 vec_dupv2di, sse2_concatv2si, vec_concatv4si_1, vec_concatv2di):
98 Change Y constraints to Y2.
99 (sse2_loadld): Change Y constraints to x.
101 2007-02-06 Roger Sayle <roger@eyesopen.com>
103 * config/rs6000/rs6000.md (popcount<mode>2): Rewrite.
104 (parity<mode>2): New define_expand using rs6000_emit_parity.
105 * config/rs6000/rs6000.c (rs6000_emit_popcount,
106 rs6000_emit_parity): New functions.
107 * config/rs6000/rs6000-protos.h (rs6000_emit_popcount,
108 rs6000_emit_parity): Prototype here.
110 2007-02-06 Ian Lance Taylor <iant@google.com>
112 * lower-subreg.c (simple_move_operand): Reject CONST.
113 (resolve_clobber): Call validate_change rather than directly
114 assigning to XEXP (pat, 0).
116 2006-02-06 Paolo Bonzini <bonzini@gnu.org>
118 * Makefile.in (tree-ssa-loop-ivopts.o): Add pointer-set.h dependency.
119 (tree-ssa-reassoc.o): Add pointer-set.h dependency.
120 (tree-cfg.o): Remove hashtab.h dependency.
122 * tree-ssa-loop-ivopts.c: Include pointer-set.h.
123 (struct ivopts_data): Change niters to pointer_map_t.
124 (struct nfe_cache_elt, nfe_hash, nfe_eq): Delete.
125 (niter_for_exit): Create pointer_map on demand. Change for
127 (tree_ssa_iv_optimize_init): Initialize data->niters to NULL.
128 (free_loop_data): Destroy data->niters if created and reset field.
129 (tree_ssa_iv_optimize_finalize): Don't delete data->niters here.
130 (tree_ssa_iv_optimize_loop): Check for presence of stale data.
132 * tree-ssa-reassoc.c: Include pointer-set.h.
133 (bb_rank): Change to long *.
134 (operand_rank): Change to pointer_map_t.
135 (find_operand_rank): Return long, -1 if not found. Declare as inline.
136 (insert_operand_rank): Accept long.
137 (operand_entry_hash, operand_entry_eq): Remove.
138 (get_rank): Return long. Adjust for changes above.
139 (init_reassoc): Change rank type to long. Adjust creation of bb_rank
141 (fini_reassoc): Delete operand_rank with pointer_map_destroy.
143 * tree-ssa-structalias.c (vi_for_tree): Change to pointer_map.
144 (struct tree_vi, tree_vi_t, tree_vi_hash, tree_vi_eq): Delete.
145 (insert_vi_for_tree): Rewrite for pointer_map API. Assert argument
147 (lookup_vi_for_tree): Rewrite for pointer_map API. Return varinfo_t
148 directly since it cannot be NULL.
149 (get_vi_for_tree): Rewrite for pointer_map API.
150 (find_what_p_points_to): Adjust for change to lookup_vi_for_tree.
151 (init_alias_vars): Create vi_for_tree as pointer_map.
152 (delete_points_to_sets): Delete vi_for_tree using pointer_map_destroy.
154 * tree-cfg.c: Don't include hashtab.h.
155 (edge_to_cases): Declare as pointer_map.
156 (struct edge_to_cases_elt, edge_to_cases_hash, edge_to_cases_eq):
158 (edge_to_cases_cleanup): Rewrite as pointer_map_traverse callback.
159 (start_recording_case_labels): Create edge_to_cases as pointer_map.
160 (end_recoding_case_labels): Cleanup edge_to_cases manually before
162 (record_switch_edge): Delete.
163 (get_cases_for_edge): Adjust for pointer_map API, inline
164 record_switch_edge (rewritten for new API), remove goto.
166 2006-02-06 Paolo Bonzini <bonzini@gnu.org>
168 * Makefile.in (tree-nested.o): Add pointer-set.h dependency.
169 * tree-nested.c: Include pointer-set.h.
170 (var_map_elt, var_map_eq, var_map_hash): Delete.
171 (struct nesting_info): Remove GTY marker. Change the two htab_t's
173 (nesting_info_bitmap_obstack): New.
174 (lookup_field_for_decl): Adjust for pointer_map API.
175 (lookup_tramp_for_decl): Adjust for pointer_map API.
176 (get_nonlocal_debug_decl): Adjust for pointer_map API.
177 (get_local_debug_decl): Adjust for pointer_map API.
178 (convert_nl_goto_reference): Adjust for pointer_map API.
179 (convert_nl_goto_receiver): Adjust for pointer_map API.
180 (create_nesting_tree): Create outside GGC space. Create bitmap on
181 the new obstack. Create field_map and var_map as pointer_maps.
182 (free_nesting_tree): Adjust for changes to create_nesting_tree.
184 (lower_nested_functions): Move root here, no need to NULL it.
185 Initialize and release the obstack.
187 2007-02-06 Paolo Bonzini <bonzini@gnu.org>
189 * tree.c (tree_int_map_hash, tree_int_map_eq, tree_int_map_marked_p):
190 Remove prototypes and make them non-static.
191 (struct tree_int_map): Remove.
192 * tree.h (struct tree_int_map): Move here, turning TO into an
194 (tree_int_map_hash, tree_int_map_eq, tree_int_map_marked_p): Declare.
196 * tree.h (TREE_COMPLEXITY): Remove.
197 (struct tree_exp): Remove complexity field.
198 * tree.c (build1_stat): Don't set it.
200 2007-02-06 Dorit Nuzman <dorit@il.ibm.com>
201 Victor Kaplansky <victork@il.ibm.com>
203 * tree-vectorizer.c (vect_is_simple_use): Support induction.
204 (vect_is_simple_reduction): Support reduction with induction as
206 (vect_is_simple_iv_evolution): Fix formatting.
207 * tree-vect-analyze.c (vect_mark_stmts_to_be_vectorized): Fix
208 formatting. Don't mark induction phis for vectorization.
209 (vect_analyze_scalar_cycles): Analyze all inductions, then reductions.
210 * tree-vect-transform.c (get_initial_def_for_induction): New function.
211 (vect_get_vec_def_for_operand): Support induction.
212 (vect_get_vec_def_for_stmt_copy): Fix formatting and add check for
214 (vectorizable_reduction): Support reduction with induction as one of
216 (vectorizable_type_demotion): Use def-type of stmt argument rather
219 * tree-ssa-loop.c (gate_scev_const_prop): Return the value of
220 flag_tree_scev_cprop.
221 * common.opt (tree-scev-cprop): New flag.
223 * tree-vect-transform.c (vect_create_destination_var): Use 'kind' in
224 call to vect_get_new_vect_var.
226 2007-02-06 Ira Rosen <irar@il.ibm.com>
228 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Check that
230 (vect_pattern_recog_1): Likewise.
232 2007-02-05 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
234 * fold-const.c (negate_expr_p): Handle CONJ_EXPR.
235 (fold_negate_expr): Likewise.
237 2007-02-05 Alexandre Oliva <aoliva@redhat.com>
240 * dwarf2out.c (modified_type_die): Follow DECL_ORIGINAL_TYPE
241 even if cv-qualification is the same.
243 2007-02-05 Geoffrey Keating <geoffk@apple.com>
245 * config/rs6000/darwin-tramp.asm (__trampoline_setup): Call
246 __enable_execute_stack on completion.
248 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
250 * config/i386/athlon.md (athlon_fldxf_k8, athlon_fld_k8,
251 athlon_fstxf_k8, athlon_fst_k8, athlon_fist, athlon_fmov,
252 athlon_fadd_load, athlon_fadd_load_k8, athlon_fadd, athlon_fmul,
253 athlon_fmul_load, athlon_fmul_load_k8, athlon_fsgn,
254 athlon_fdiv_load, athlon_fdiv_load_k8, athlon_fdiv_k8,
255 athlon_fpspc_load, athlon_fpspc, athlon_fcmov_load,
256 athlon_fcmov_load_k8, athlon_fcmov_k8, athlon_fcomi_load_k8,
257 athlon_fcomi, athlon_fcom_load_k8, athlon_fcom): Added amdfam10.
259 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
261 * config/i386/i386.md (x86_sahf_1, cmpfp_i_mixed, cmpfp_i_sse,
262 cmpfp_i_i387, cmpfp_iu_mixed, cmpfp_iu_sse, cmpfp_iu_387,
263 swapsi, swaphi_1, swapqi_1, swapdi_rex64, fix_truncsfdi_sse,
264 fix_truncdfdi_sse, fix_truncsfsi_sse, fix_truncdfsi_sse,
265 x86_fldcw_1, floatsisf2_mixed, floatsisf2_sse, floatdisf2_mixed,
266 floatdisf2_sse, floatsidf2_mixed, floatsidf2_sse,
267 floatdidf2_mixed, floatdidf2_sse, muldi3_1_rex64, mulsi3_1,
268 mulsi3_1_zext, mulhi3_1, mulqi3_1, umulqihi3_1, mulqihi3_insn,
269 umulditi3_insn, umulsidi3_insn, mulditi3_insn, mulsidi3_insn,
270 umuldi3_highpart_rex64, umulsi3_highpart_insn,
271 umulsi3_highpart_zext, smuldi3_highpart_rex64,
272 smulsi3_highpart_insn, smulsi3_highpart_zext, x86_64_shld,
273 x86_shld_1, x86_64_shrd, sqrtsf2_mixed, sqrtsf2_sse,
274 sqrtsf2_i387, sqrtdf2_mixed, sqrtdf2_sse, sqrtdf2_i387,
275 sqrtextendsfdf2_i387, sqrtxf2, sqrtextendsfxf2_i387,
276 sqrtextenddfxf2_i387): Added amdfam10_decode.
278 * config/i386/athlon.md (athlon_idirect_amdfam10,
279 athlon_ivector_amdfam10, athlon_idirect_load_amdfam10,
280 athlon_ivector_load_amdfam10, athlon_idirect_both_amdfam10,
281 athlon_ivector_both_amdfam10, athlon_idirect_store_amdfam10,
282 athlon_ivector_store_amdfam10): New define_insn_reservation.
283 (athlon_idirect_loadmov, athlon_idirect_movstore): Added
286 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
288 * config/i386/athlon.md (athlon_call_amdfam10,
289 athlon_pop_amdfam10, athlon_lea_amdfam10): New
290 define_insn_reservation.
291 (athlon_branch, athlon_push, athlon_leave_k8, athlon_imul_k8,
292 athlon_imul_k8_DI, athlon_imul_mem_k8, athlon_imul_mem_k8_DI,
293 athlon_idiv, athlon_idiv_mem, athlon_str): Added amdfam10.
295 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
297 * config/i386/athlon.md (athlon_sseld_amdfam10,
298 athlon_mmxld_amdfam10, athlon_ssest_amdfam10,
299 athlon_mmxssest_short_amdfam10): New define_insn_reservation.
301 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
303 * config/i386/athlon.md (athlon_sseins_amdfam10): New
304 define_insn_reservation.
305 * config/i386/i386.md (sseins): Added sseins to define_attr type
306 and define_attr unit.
307 * config/i386/sse.md: Set type attribute to sseins for insertq
310 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
312 * config/i386/athlon.md (sselog_load_amdfam10, sselog_amdfam10,
313 ssecmpvector_load_amdfam10, ssecmpvector_amdfam10,
314 ssecomi_load_amdfam10, ssecomi_amdfam10,
315 sseaddvector_load_amdfam10, sseaddvector_amdfam10): New
316 define_insn_reservation.
317 (ssecmp_load_k8, ssecmp, sseadd_load_k8, seadd): Added amdfam10.
319 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
321 * config/i386/athlon.md (cvtss2sd_load_amdfam10,
322 cvtss2sd_amdfam10, cvtps2pd_load_amdfam10, cvtps2pd_amdfam10,
323 cvtsi2sd_load_amdfam10, cvtsi2ss_load_amdfam10,
324 cvtsi2sd_amdfam10, cvtsi2ss_amdfam10, cvtsd2ss_load_amdfam10,
325 cvtsd2ss_amdfam10, cvtpd2ps_load_amdfam10, cvtpd2ps_amdfam10,
326 cvtsX2si_load_amdfam10, cvtsX2si_amdfam10): New
327 define_insn_reservation.
329 * config/i386/sse.md (cvtsi2ss, cvtsi2ssq, cvtss2si,
330 cvtss2siq, cvttss2si, cvttss2siq, cvtsi2sd, cvtsi2sdq,
331 cvtsd2si, cvtsd2siq, cvttsd2si, cvttsd2siq,
332 cvtpd2dq, cvttpd2dq, cvtsd2ss, cvtss2sd,
333 cvtpd2ps, cvtps2pd): Added amdfam10_decode attribute.
335 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
337 * config/i386/athlon.md (athlon_ssedivvector_amdfam10,
338 athlon_ssedivvector_load_amdfam10, athlon_ssemulvector_amdfam10,
339 athlon_ssemulvector_load_amdfam10): New define_insn_reservation.
340 (athlon_ssediv, athlon_ssediv_load_k8, athlon_ssemul,
341 athlon_ssemul_load_k8): Added amdfam10.
343 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
345 * config/i386/i386.h (TARGET_SSE_UNALIGNED_MOVE_OPTIMAL): New macro.
346 (x86_sse_unaligned_move_optimal): New variable.
348 * config/i386/i386.c (x86_sse_unaligned_move_optimal): Enable for
350 (ix86_expand_vector_move_misalign): Add code to generate movupd/movups
351 for unaligned vector SSE double/single precision loads for AMDFAM10.
353 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
355 * config/i386/i386.h (TARGET_AMDFAM10): New macro.
356 (TARGET_CPU_CPP_BUILTINS): Add code for amdfam10.
357 Define TARGET_CPU_DEFAULT_amdfam10.
358 (TARGET_CPU_DEFAULT_NAMES): Add amdfam10.
359 (processor_type): Add PROCESSOR_AMDFAM10.
361 * config/i386/i386.md: Add amdfam10 as a new cpu attribute to match
362 processor_type in config/i386/i386.h.
363 Enable imul peepholes for TARGET_AMDFAM10.
365 * config.gcc: Add support for --with-cpu option for amdfam10.
367 * config/i386/i386.c (amdfam10_cost): New variable.
368 (m_AMDFAM10): New macro.
369 (m_ATHLON_K8_AMDFAM10): New macro.
370 (x86_use_leave, x86_push_memory, x86_movx, x86_unroll_strlen,
371 x86_cmove, x86_3dnow_a, x86_deep_branch, x86_use_simode_fiop,
372 x86_promote_QImode, x86_integer_DFmode_moves,
373 x86_partial_reg_dependency, x86_memory_mismatch_stall,
374 x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
375 x86_sse_partial_reg_dependency, x86_sse_typeless_stores,
376 x86_use_ffreep, x86_use_incdec, x86_four_jump_limit,
377 x86_schedule, x86_use_bt, x86_cmpxchg16b, x86_pad_returns):
378 Enable/disable for amdfam10.
379 (override_options): Add amdfam10_cost to processor_target_table.
380 Set up PROCESSOR_AMDFAM10 for amdfam10 entry in
381 processor_alias_table.
382 (ix86_issue_rate): Add PROCESSOR_AMDFAM10.
383 (ix86_adjust_cost): Add code for amdfam10.
385 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
387 * config/i386/i386.opt: Add new Advanced Bit Manipulation (-mabm)
388 instruction set feature flag. Add new (-mpopcnt) flag for popcnt
389 instruction. Add new SSE4A (-msse4a) instruction set feature flag.
390 * config/i386/i386.h: Add builtin definition for SSE4A.
391 * config/i386/i386.md: Add support for ABM instructions
393 * config/i386/sse.md: Add support for SSE4A instructions
394 (movntss, movntsd, extrq, insertq).
395 * config/i386/i386.c: Add support for ABM and SSE4A builtins.
396 Add -march=amdfam10 flag.
397 * config/i386/ammintrin.h: Add support for SSE4A intrinsics.
398 * doc/invoke.texi: Add documentation on flags for sse4a, abm, popcnt
400 * doc/extend.texi: Add documentation for SSE4A builtins.
402 2007-02-05 Bob Wilson <bob.wilson@acm.org>
404 * config/xtensa/xtensa.c (constantpool_mem_p): Skip over SUBREGs.
406 2007-02-05 Richard Guenther <rguenther@suse.de>
408 * tree-vectorizer.h (vectorizable_function): Add argument type
409 argument, change return type.
410 * tree-vect-patterns.c (vect_recog_pow_pattern): Adjust caller.
411 * tree-vect-transform.c (vectorizable_function): Handle extra
412 argument, return vectorized function decl.
413 (build_vectorized_function_call): Remove.
414 (vectorizable_call): Handle calls with result and argument types
415 differing. Handle loop vectorization factor correctly.
416 * targhooks.c (default_builtin_vectorized_function): Adjust for
418 * targhooks.h (default_builtin_vectorized_function): Likewise.
419 * target.h (builtin_vectorized_function): Add argument type
421 * config/i386/i386.c (ix86_builtin_vectorized_function): Handle
422 extra argument, allow vectorizing of lrintf.
423 * doc/tm.texi (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Adjust
424 documentation of target hook.
426 2007-02-05 Hans-Peter Nilsson <hp@axis.com>
429 * config/cris/cris.md ("*andsi_movu", "*andsi_clear", "*andhi_movu")
430 ("*andhi_clear", andu (casesi+45)): For size-changed operand where
431 memory is allowed, require !side_effects_p, not just !MEM_VOLATILE_P.
433 2007-02-05 Roger Sayle <roger@eyesopen.com>
435 * fold-const.c (fold_unary) <REAL_PART>: Test for availability of
436 BUILT_IN_COS before simplifying REAL_PART(CEXPI)) to COS.
437 <IMAG_PART>: Likewise, check for availability of BUILT_IN_SIN.
438 * builtins.c (fold_builtin_sincos): Check for TARGET_C99_FUNCTIONS
439 before canonicalizing sincos to cexpi.
440 (fold_builtin_cexp): Likewise, for canonicalizing cexp to cexpi.
442 2007-02-05 Roger Sayle <roger@eyesopen.com>
444 * config/alpha/alpha.c (alpha_add_builtins): New Helper function.
445 Set TREE_READONLY and TREE_NOTHROW directly, not via attributes.
446 (alpha_init_builtins): Use alpha_add_builtins to process tables.
448 2007-02-05 Roger Sayle <roger@eyesopen.com>
450 * mips-tfile.c (initialize_init_file): Correct endianness test.
452 2007-02-05 Kazu Hirata <kazu@codesourcery.com>
454 * config/m68k/m68k.md (pushdi-1, pushdi, movsi+1): Don't use
457 2007-02-05 Richard Sandiford <richard@codesourcery.com>
459 * dwarf2out.c (dwarf2out_frame_debug_expr): Record the register
460 saves in a PARALLEL before the register assignments.
462 2007-02-05 Richard Sandiford <richard@codesourcery.com>
464 * doc/tm.texi (DWARF_ALT_FRAME_RETURN_COLUMN): Do not require
465 DWARF_FRAME_RETURN_COLUMN to be a general register.
466 * dwarf2out.c (init_return_column_size): New function, split from...
467 (expand_builtin_init_dwarf_reg_sizes): ...here. Allow both
468 DWARF_FRAME_RETURN_COLUMN and DWARF_ALT_FRAME_RETURN_COLUMN
469 to be nongeneral registers.
470 * config/m68k/m68k.h (DWARF_FRAME_REGNUM): Only map FP and
472 (DWARF_FRAME_REGISTERS, DWARF_FRAME_RETURN_COLUMN): Define.
473 (DWARF_ALT_FRAME_RETURN_COLUMN): Define.
475 2007-02-04 Zdenek Dvorak <dvorakz@suse.cz>
477 * cfgcleanup.c (try_optimize_cfg): Avoid removing ENTRY_BLOCK_PTR.
479 2007-02-04 Zdenek Dvorak <dvorakz@suse.cz>
481 * cfgloopmanip.c (loop_delete_branch_edge): Removed.
482 (remove_path): Use can_remove_branch_p and remove_branch instead
483 of loop_delete_branch_edge.
484 * tree-ssa-loop-manip.c (scale_dominated_blocks_in_loop): New function.
485 (tree_transform_and_unroll_loop): Remove dead branches immediately.
486 Update profile using scale_dominated_blocks_in_loop.
487 * cfghooks.c (can_remove_branch_p, remove_branch): New functions.
488 * cfghooks.h (struct cfg_hooks): Add can_remove_branch_p.
489 (can_remove_branch_p, remove_branch): Declare.
490 * tree-cfg.c (tree_can_remove_branch_p): New function.
491 (tree_cfg_hooks): Add tree_can_remove_branch_p.
492 * cfgrtl.c (rtl_can_remove_branch_p): New function.
493 (rtl_cfg_hooks, cfg_layout_rtl_cfg_hook): Add rtl_can_remove_branch_p.
495 2007-02-05 Jan Hubicka <jh@suse.cz>
498 * ipa-inline.c (cgraph_clone_inlined_nodes): When there are unanalyzed
499 nodes in cgraph, don't remove offline copy of the function.
501 2007-02-04 Jan Hubicka <jh@suse.cz>
503 * tree-sra.c (sra_walk_expr): Add linebreaks. BITFIELD_REFs into
504 vectors might cause maybe_lookup_element_for_expr to be called
505 on non-sra-candidate.
507 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
509 * config/bfin/bfin-modes.def, config/bfin/bfin.c,
510 config/bfin/bfin.md, config/bfin/predicates.md: Follow
511 spelling conventions.
513 2007-02-04 Richard Guenther <rguenther@suse.de>
516 * fold-const.c (try_move_mult_to_index): Make sure to not
517 overflow one dimension of a multi-dimensional array access.
519 2007-02-04 Jan Hubicka <jh@suse.cz>
521 * passes.c (init_optimization_passes): Reindent.
523 2007-02-04 Jan Hubicka <jh@suse.cz>
524 Eric Botcazou <ebotcazou@adacore.com>
526 * tree-optimize.c (has_abnormal_outgoing_edge_p): Move to...
527 (execute_fixup_cfg): Break out the abnormal goto code.
528 * tree-inline.c (has_abnormal_outgoing_edge_p): ...here.
529 (make_nonlocal_label_edges): Move here from execute_fixup_cfg.
530 (optimize_inline_calls): Call make_nonlocal_label_edges.
532 2007-02-04 Jan Hubicka <jh@suse.cz>
534 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Return
535 true when something was changed.
536 (rename_ssa_copies): When something was changed, do
537 TODO_remove_unused_locals.
538 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
539 add TODO_remove_unused_locals when instruction was removed.
541 2007-02-04 Jan Hubicka <jh@suse.cz>
543 * ipa-inline.c (try_inline): Improve debug output; work on already
545 (cgraph_decide_inlining_incrementally): Indent; improve debug output;
546 call try_inline for already inlined edges too when flattening;
547 inline also functions that make callee growth but overall unit size
550 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
552 * config/m32c/bitops.md, config/m32c/jump.md,
553 config/m32c/m32c.c, config/m32c/m32c.h, config/m32r/m32r.c,
554 config/m32r/m32r.h, config/m32r/m32r.md,
555 config/m32r/predicates.md, config/m68hc11/larith.asm,
556 config/m68hc11/m68hc11.c, config/m68hc11/m68hc11.h,
557 config/m68k/m68k.h, config/mcore/mcore.md, config/mips/4k.md,
558 config/mips/mips-protos.h, config/mips/mips.c,
559 config/mips/mips.h, config/mips/mips.md, config/mips/mips16.S,
560 config/mn10300/mn10300.h, config/mn10300/predicates.md,
561 config/mt/mt.c, config/mt/mt.h, config/mt/mt.md: Follow
562 spelling conventions.
564 * config/v850/v850.c, config/v850/v850.h, config/v850/v850.md:
565 Follow spelling conventions.
567 2007-02-03 Douglas Gregor <doug.gregor@gmail.com>
569 * c-opts.c (c_common_post_options): If C++0x mode is enabled, don't
570 warn about C++0x compatibility.
572 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
574 * config/h8300/h8300.c, config/h8300/h8300.h,
575 config/h8300/h8300.md: Follow spelling conventions.
577 2007-02-03 Uros Bizjak <ubizjak@gmail.com>
580 * combine.c (try_combine): Do not substitute source operand
581 with constants wider than 2 * HOST_BITS_PER_WIDE_INT.
583 2007-02-03 Jan Hubicka <jh@suse.cz>
585 PR gcov-profile/30650
586 * value-prof.c (stringop_block_profile): Fix handling of size counter;
587 do not divide by zero for never executed counters.
588 (tree_find_values_to_profile): Fix counters.
589 * gcov-ui.h (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): Fix comments.
591 2007-02-03 Ian Lance Taylor <iant@google.com>
593 * lower-subreg.c (simple_move_operand): New static function,
594 broken out of simple_move. Reject LABEL_REF, SYMBOL_REF, and HIGH
596 (simple_move): Call simple_move_operand.
597 (find_decomposable_subregs): Add special handling of MEMs.
598 (can_decompose_p): Rename from cannot_decompose_p. Reverse
599 meaning of return value. If we see a hard register, test whether
600 it can store a word_mode value. Change all callers.
602 2007-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
604 * pa.md (addvdi3, addvsi3, subvdi3, subvsi3, negvdi2, negvsi2): New
605 ftrapv insns and expanders.
606 (subdi3): Change define_expand operand 1 to arith11_operand, and
607 operand 2 to reg_or_0_operand. Change constraints of 64-bit insn
608 pattern to handle reg_or_0 operands. Revise 32-bit insn pattern to
609 handle 11-bit constants and reg_or_0 operands in operands 1 and 2,
613 * varasm.c (notice_global_symbol): Treat global objects as weak when
616 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
618 * emit-rtl.c (dconstpi): Delete.
620 (init_emit_once): Delete dconstpi and init dconstsqrt2.
621 * real.h (dconstpi): Delete.
623 * builtins.c (fold_builtin_cabs): Use dconstsqrt2.
624 (fold_builtin_hypot): Likewise.
626 2007-02-03 Tom Tromey <tromey@redhat.com>
629 * gcc.c (cpp_unique_options): Any of -ggdb3, -gstabs3,
630 -gcoff3, -gxcoff3, -gvms3 implies -dD.
632 2007-02-03 Kazu Hirata <kazu@codesourcery.com>
634 * c-decl.c, config/avr/avr.c, config/avr/avr.h,
635 config/m68k/m68k.c, config/m68k/netbsd-elf.h,
636 config/mn10300/mn10300.c, config/pdp11/pdp11.h,
637 config/rs6000/cell.md, config/rs6000/darwin.h,
638 config/sh/sh.md, config/sh/sh4-300.md, config/spu/spu.c,
639 config/spu/spu.md, cselib.c, expr.c, haifa-sched.c, hwint.h,
640 jump.c, reload.c, sched-deps.c, sched-int.h, tree-inline.c,
641 tree-profile.c, tree-ssa-live.h, tree-vrp.c: Fix comment
642 typos. Follow spelling conventions.
643 * doc/invoke.texi: Follow spelling conventions.
645 2007-02-03 Roger Sayle <roger@eyesopen.com>
647 * simplify-rtx.c (simplify_relational_operation_1): Implement some
648 canonicalization transformations that attempt to simplify integer
649 constant comparisons to become comparisons against zero.
651 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
653 * builtins.c (fold_builtin_cabs): Fold cabs(x+xi) into
655 * fold-const.c (fold_binary): Fix comment typos. Fold complex
656 (x,0)-(0,y) into (x,-y). Likewise (0,y)-(x,0) into (-x,y).
658 2007-02-02 Mike Stump <mrs@apple.com>
660 * config/darwin9.h (DARWIN_LINKER_GENERATES_ISLANDS): Add.
661 * config/rs6000/rs6000.c (DARWIN_GENERATE_ISLANDS): Add.
662 (output_call): Use DARWIN_GENERATE_ISLANDS to decide when to
663 generate a branch island.
665 2007-02-02 Bob Wilson <bob.wilson@acm.org>
667 * config/xtensa/xtensa.c (smalloffset_mem_p): Use BASE_REG_P.
668 (xtensa_legitimate_address_p): New.
669 (xtensa_legitimize_address): New.
670 (xtensa_output_addr_const_extra): New.
671 * config/xtensa/xtensa.h (REG_OK_STRICT_FLAG): Define.
673 (REG_OK_FOR_BASE_P): Use BASE_REG_P.
674 (GO_IF_LEGITIMATE_ADDRESS): Move code to xtensa_legitimate_address_p.
675 (LEGITIMIZE_ADDRESS): Move code to xtensa_legitimize_address.
676 (OUTPUT_ADDR_CONST_EXTRA): Move code to xtensa_output_addr_const_extra.
677 * config/xtensa/xtensa-protos.h (xtensa_legitimate_address_p): New.
678 (xtensa_legitimize_address): New.
679 (xtensa_output_addr_const_extra): New.
681 2007-02-02 Steve Ellcey <sje@cup.hp.com>
683 * config/ia64/ia64.c (ia64_print_operand): Fix compare strings.
685 2007-02-02 Ian Lance Taylor <iant@google.com>
687 * expmed.c (expand_divmod): Add comment.
689 2007-02-02 Kazu Hirata <kazu@codesourcery.com>
691 * emit-rtl.c (renumber_insns): Remove.
692 * flags.h: Remove the extern for flag_renumber_insns.
693 * rtl.h: Remove the prototype for renumber_insns.
694 * toplev.c (flag_renumber_insns): Remove.
696 2007-02-02 Hui-May Chang <hm.chang@apple.com>
698 Revert for x86 darwin:
699 2005-06-19 Uros Bizjak <uros@kss-loka.si>
701 * config/i386/i386.c (ix86_function_arg_regno_p): Put back the
702 code before the following patch under TARGET_MACHO.
703 (ix86_function_value_regno_p): Likewise.
705 2007-02-02 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
707 * fold-const.c (negate_expr_p, fold_negate_expr): Handle
710 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
713 * config/ia64/ia64.c (ia64_speculate_insn): Restrict to memory loads to
714 general or fp registers. Add comments.
715 * config/ia64/ia64.md (reg_pred_prefix): Add comment.
717 2007-02-02 Paolo Bonzini <bonzini@gnu.org>
719 * pointer-set.c (insert_aux): Only return insertion slot.
720 (pointer_set_insert): Adjust.
721 (pointer_set_traverse, struct pointer_map_t, pointer_map_create,
722 pointer_map_destroy, pointer_map_insert, pointer_map_contains,
723 pointer_map_traverse): New.
724 * pointer-set.h (pointer_set_traverse, struct pointer_map_t,
725 pointer_map_create, pointer_map_destroy, pointer_map_insert,
726 pointer_map_contains, pointer_map_traverse): Declare.
728 2007-02-02 Jakub Jelinek <jakub@redhat.com>
731 * builtins.c (fold_builtin_sprintf): Do not attempt to optimize
732 sprintf (str, "%s"). Do not optimize sprintf (str, "nopercent", p++).
734 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
736 * sched-int.h (ds_to_dk, dk_to_ds): Declare functions.
738 (struct _dep): New type.
739 (dep_t): New typedef.
740 (DEP_PRO, DEP_CON, DEP_KIND): New access macros.
741 (DEP_STATUS): New access macro. The macro with the same name was
742 renamed to DEP_LINK_STATUS.
743 (dep_init): Declare function
745 (struct _dep_link): New type.
746 (dep_link_t): New typedef.
747 (DEP_LINK_NODE, DEP_LINK_NEXT, DEP_LINK_PREV_NEXTP): New access macros.
748 (DEP_LINK_DEP, DEP_LINK_PRO, DEP_LINK_CON, DEP_LINK_KIND): New macros.
749 (DEP_LINK_STATUS): New macro.
750 (debug_dep_links): New debug function.
752 (struct _deps_list): New type.
753 (deps_list_t): New typedef.
754 (DEPS_LIST_FIRST): New access macro.
755 (FOR_EACH_DEP_LINK): New cycle macro.
756 (create_deps_list, free_deps_list, delete_deps_list): Declare
758 (deps_list_empty_p, debug_deps_list, add_back_dep_to_deps_list): Ditto.
759 (find_link_by_pro_in_deps_list, find_link_by_con_in_deps_list): Ditto.
760 (copy_deps_list_change_con): Ditto.
762 (move_dep_link): Declare function.
764 (struct _dep_node): New type.
765 (dep_node_t): New typedef.
766 (DEP_NODE_BACK, DEP_NODE_DEP, DEP_NODE_FORW): New access macros.
768 (struct haifa_insn_data.back_deps): New field to hold backward
769 dependencies of the insn.
770 (struct haifa_insn_data.depend): Rename to forw_deps. Change its type
772 (struct haifa_insn_data.resolved_deps): Rename to resolved_back_deps.
773 Change its type to deps_list_t.
774 (INSN_BACK_DEPS): New access macro to use instead of LOG_LINKS.
775 (INSN_DEPEND): Rename to INSN_FORW_DEPS.
776 (RESOLVED_DEPS): Rename to INSN_RESOLVED_BACK_DEPS.
778 (INSN_COST): Move to haifa-sched.c. Use insn_cost () instead.
780 (DEP_STATUS): Rename to DEP_LINK_STATUS. Fix typo in the comment.
782 (add_forw_dep, delete_back_forw_dep, insn_cost): Update declaration and
786 * sched-deps.c (CHECK): New macro to (en/dis)able sanity checks.
787 (ds_to_dk, dk_to_ds): New functions.
789 (init_dep_1): New static function.
790 (init_dep): New function.
791 (copy_dep): New static function.
793 (dep_link_consistent_p, attach_dep_link, add_to_deps_list): New static
795 (detach_dep_link): New static function.
796 (move_dep_link): New function.
798 (dep_links_consistent_p, dump_dep_links): New static functions.
799 (debug_dep_links): New debugging function.
801 (deps_obstack, dl_obstack, dn_obstack): New static variables.
803 (alloc_deps_list, init_deps_list): New static functions.
804 (create_deps_list): New function.
805 (clear_deps_list): New static function.
806 (free_deps_list, delete_deps_list, deps_list_empty_p): New functions.
807 (deps_list_consistent_p, dump_deps_list): New static functions.
808 (debug_deps_list): New function.
809 (add_back_dep_to_deps_list, find_link_by_pro_in_deps_list): New
811 (find_link_by_con_in_deps_list, copy_deps_list_change_con): Ditto.
813 (maybe_add_or_update_back_dep_1, add_or_update_back_dep_1): Update to
814 use new scheduler dependencies lists.
815 (add_back_dep, delete_all_dependences, fixup_sched_groups): Ditto.
816 (sched_analyze): Ditto. Initialize dependencies lists.
817 (add_forw_dep, compute_forward_dependences): Update to use new
818 scheduler dependencies lists.
820 (init_dependency_caches): Init deps_obstack.
821 (free_dependency_caches): Free deps_obstack.
823 (adjust_add_sorted_back_dep, adjust_back_add_forw_dep): Update to use
824 new scheduler dependencies lists.
825 (delete_forw_dep, add_or_update_back_forw_dep): Ditto.
826 (add_back_forw_dep, delete_back_forw_dep): Ditto.
828 * sched-rgn.c (set_spec_fed, find_conditional_protection, is_pfree):
829 Update to use new scheduler dependencies lists.
830 (is_conditionally_protected, is_prisky, add_branch_dependences): Ditto.
831 (debug_dependencies): Ditto.
832 (schedule_region): Update comments.
834 * sched-ebb.c (earliest_block_with_similiar_load): Update to use new
835 scheduler dependencies lists.
836 (schedule_ebb): Update comments.
838 * rtl.def (DEPS_LIST): Remove.
840 * lists.c (unused_deps_list): Remove.
841 (free_list): Update assertions.
843 (alloc_DEPS_LIST, free_DEPS_LIST_list, free_DEPS_LIST_node): Remove.
844 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
846 * rtl.h (free_DEPS_LIST_list, alloc_DEPS_LIST): Remove declarations.
847 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
849 * haifa-sched.c (comments): Update.
850 (insn_cost1): Remove. Inline the code into insn_cost ().
851 (insn_cost): Update to use new scheduler dependencies lists. Move
852 processing of the dependency cost to dep_cost ().
853 (dep_cost): New function. Use it instead of insn_cost () when
854 evaluating cost of the dependency. Use compatible interface to
855 interact with the target.
856 (priority): Update to use new scheduler dependencies lists.
857 (rank_for_schedule): Ditto. Optimize heuristic that prefers the insn
858 with greater number of insns that depend on the insn.
859 (schedule_insn): Update to use new scheduler dependencies lists. Add
860 code to free backward dependencies lists. Inline and optimize code
861 from resolve_dep () - see PR28071.
862 (ok_for_early_queue_removal): Update to use new scheduler dependencies
863 lists. Update call to targetm.sched.is_costly_dependence hook.
865 (fix_inter_tick, try_ready, fix_tick_ready): Update to use new
866 scheduler dependencies lists.
868 (resolve_dep): Remove. Move the logic to schedule_insn ().
869 (init_h_i_d): Initialize dependencies lists.
871 (process_insn_depend_be_in_spec): Rename to
872 process_insn_forw_deps_be_in_spec. Update to use new scheduler
874 (add_to_speculative_block, create_check_block_twin, fix_recovery_deps):
875 Update to use new scheduler dependencies lists.
876 (clear_priorities, calc_priorities, add_jump_dependencies): Ditto.
878 * ddg.c (create_ddg_dependence, create_ddg_dep_no_link): Update to use
879 new scheduler dependencies lists.
880 (build_intra_loop_deps): Ditto.
882 * target.h (struct _dep): Declare to use in
883 gcc_target.sched.is_costly_dependence.
884 (struct gcc_target.sched.adjust_cost): Fix typo.
885 (struct gcc_target.sched.is_costly_dependence): Change signature to use
886 single dep_t parameter instead of an equivalent triad.
887 (struct gcc_target.sched.adjust_cost_2): Remove.
889 * target-def.h (TARGET_SCHED_ADJUST_COST_2): Remove.
891 * reg-notes.def (DEP_TRUE, DEP_OUTPUT, DEP_ANTI): Update comments.
893 * doc/tm.texi (TARGET_SCHED_IS_COSTLY_DEPENDENCE): Update
895 (TARGET_SCHED_ADJUST_COST_2): Remove documentation.
897 * doc/rtl.texi (LOG_LINKS): Remove part about instruction scheduler.
898 (REG_DEP_TRUE): Document.
900 * config/ia64/ia64.c (ia64_adjust_cost_2): Rename to ia64_adjust_cost.
901 Change signature to correspond to the targetm.sched.adjust_cost hook.
902 Update use in TARGET_SCHED_ADJUST_COST_2.
903 (TARGET_SCHED_ADJUST_COST_2): Rename to TARGET_SCHED_ADJUST_COST.
904 (ia64_dependencies_evaluation_hook, ia64_dfa_new_cycle): Update to use
905 new scheduler dependencies lists.
906 (ia64_gen_check): Ditto.
908 * config/mips/mips.c (vr4130_swap_insns_p): Update to use new scheduler
911 * config/rs6000/rs6000.c (rs6000_is_costly_dependence): Change
912 signature to correspond to the targetm.sched.is_costly_dependence hook.
913 (is_costly_group): Update to use new scheduler dependencies lists.
915 * config/spu/spu.c (spu_sched_adjust_cost): Use insn_cost () function
916 instead of INSN_COST () macro.
918 2007-02-01 Ian Lance Taylor <iant@google.com>
920 * lower-subreg.c (resolve_clobber): Handle a subreg of a concatn.
922 2007-02-01 Guy Martin <gmsoft@gentoo.org>
924 * pa.md (tp_load): Correct mfctl instruction syntax.
926 2007-02-01 Geoffrey Keating <geoffk@apple.com>
928 * config/rs6000/rs6000.c (rs6000_stack_info): Correct
929 altivec_padding_size calculation on AIX. Improve comment, add
930 assert to verify that it's right.
932 * config/rs6000/darwin.h (MD_UNWIND_SUPPORT): Don't define for
935 2007-2-01 Seongbae Park <seongbae.park@gmail.com>
938 * global.c (global_alloc): Add mising initialization of
941 2007-02-01 Roger Sayle <roger@eyesopen.com>
943 * alias.c (init_alias_analysis): Correct whitespace.
944 * bb-reorder.c (fix_edges_for_rarely_executed_code,
945 partition_hot_cold_basic_blocks): Likewise.
946 * builtins.c (expand_builtin_printf, expand_builtin_fprintf,
947 expand_builtin_sprintf, fold_builtin_carg, fold_builtin_sprintf,
948 maybe_emit_sprintf_chk_warning, fold_builtin_sprintf_chk,
949 fold_builtin_snprintf_chk, fold_builtin_printf,
950 fold_builtin_fprintf, do_mpfr_ckconv, do_mpfr_arg1, do_mpfr_arg2,
951 do_mpfr_arg3, do_mpfr_sincos): Likewise.
952 * cfgcleanup.c (cleanup_cfg): Likewise.
953 * cfgexpand.c (tree_expand_cfg): Likewise.
954 * fold-const.c (fold_binary) <RDIV_EXPR>: Likewise.
955 * function.c (get_next_funcdef_no): Likewise.
956 * gengtype.c (main): Likewise.
957 * genmodes.c (main): Likewise.
958 * gcse.c (bypass_conditional_jumps, print_ldst_list): Likewise.
959 * haifa-sched.c (schedule_block, extend_h_i_d): Likewise.
960 * ifcvt.c (noce_emit_move_insn): Likewise.
961 * modulo-sched.c (generate_prolog_epilog, sms_schedule_by_order):
963 * stor-layout.c (get_best_mode): Likewise.
964 * tree-ssa-loop-niter.c (get_val_for): Likewise.
965 * tree-ssa-structalias.c (get_varinfo, get_varinfo_fc,
966 scc_visit, do_ds_constraint, do_complex_constraint, label_visit,
967 perform_var_substitution, solve_graph): Likewise.
968 * tree-vrp.c (vrp_finalize): Likewise.
970 2007-02-01 Ian Lance Taylor <iant@google.com>
972 * lower-subreg.c (simplify_gen_subreg_concatn): If we ask for the
973 high part of a paradoxical subreg, return a constant zero.
975 2007-02-01 Zdenek Dvorak <dvorakz@suse.cz>
977 * toplev.c (lang_dependent_init): Call init_set_costs.
978 * loop-init.c (loop_optimizer_init): Do not call init_set_costs.
980 2007-02-01 Richard Guenther <rguenther@suse.de>
983 * fold-const.c (fold_negate_expr): Allow negating a
984 constant if overflow does not change.
986 2007-02-01 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
988 * doc/c-tree.texi (Expression trees): Improve markup.
989 * doc/tm.texi (Register Classes, Addressing Modes)
990 (Floating Point): Fix spacing after abbreviations. Fix some
993 2007-02-01 Ben Elliston <bje@au.ibm.com>
995 * doc/invoke.texi: Replace "bugfix" with "bug fix" throughout.
996 * doc/contrib.texi: Likewise.
997 * doc/install.texi: Likewise.
999 2007-01-31 Richard Henderson <rth@redhat.com>
1000 Ian Lance Taylor <iant@google.com>
1002 * lower-subreg.c: New file.
1003 * rtl.def (CONCATN): Define.
1004 * passes.c (init_optimization_passes): Add pass_lower_subreg and
1006 * emit-rtl.c (update_reg_offset): New static function, broken out
1007 of gen_rtx_REG_offset.
1008 (gen_rtx_REG_offset): Call update_reg_offset.
1009 (gen_reg_rtx_offset): New function.
1010 * regclass.c: Revert patch of 2006-03-05, restoring
1012 (clear_reg_info_regno): New function.
1013 * dwarf2out.c (concatn_loc_descriptor): New static function.
1014 (loc_descriptor): Handle CONCATN.
1015 * common.opt (fsplit_wide_types): New option.
1016 * opts.c (decode_options): Set flag_split_wide_types when
1018 * timevar.def (TV_LOWER_SUBREG): Define.
1019 * rtl.h (gen_reg_rtx_offset): Declare.
1020 (reg_scan_update): Declare.
1021 * regs.h (clear_reg_info_regno): Declare.
1022 * tree-pass.h (pass_lower_subreg): Declare.
1023 (pass_lower_subreg2): Declare.
1024 * doc/invoke.texi (Option Summary): List -fno-split-wide-types.
1025 (Optimize Options): Add -fsplit-wide-types to -O1 list. Document
1027 * doc/rtl.texi (Regs and Memory): Document concat and concatn.
1028 * Makefile.in (OBJS-common): Add lower-subreg.o.
1029 (lower-subreg.o): New target.
1031 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1033 * config/sh/sh.h (HAVE_SECONDARY_RELOADS): Remove.
1035 2007-01-31 Anatoly Sokolov <aesok@post.ru>
1038 * config/avr/avr.c (DWARF2_ADDR_SIZE): Define.
1040 2007-01-31 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1043 * builtins.c (fold_builtin_sqrt): Use MPFR for constant args.
1045 2007-01-31 Zdenek Dvorak <dvorakz@suse.cz>
1047 * cfgloop.h: Include vec-prim.h.
1048 (enum li_flags): Remove LI_ONLY_OLD.
1049 (loop_iterator): Changed.
1050 (fel_next, fel_init): Iterate over loop tree.
1051 (FOR_EACH_LOOP_BREAK): New macro.
1052 * loop-unswitch.c (unswitch_loops): Do not pass LI_ONLY_OLD to
1054 * tree-ssa-loop-unswitch.c (tree_ssa_unswitch_loops): Ditto.
1055 * modulo-sched.c (sms_schedule): Ditto.
1056 * tree-vectorizer.c (vectorize_loops): Ditto.
1057 * doc/loop.texi: Update information on loop numbering and behavior of
1058 FOR_EACH_LOOP wrto new loops.
1059 * tree-scalar-evolution.c (compute_overall_effect_of_inner_loop,
1060 add_to_evolution_1): Test nestedness of loops instead of comparing
1062 * tree-chrec.c (chrec_fold_plus_poly_poly,
1063 chrec_fold_multiply_poly_poly, chrec_evaluate,
1064 hide_evolution_in_other_loops_than_loop, chrec_component_in_loop_num,
1065 reset_evolution_in_loop): Ditto.
1066 * Makefile.in (CFGLOOP_H): Add vecprim.h dependency.
1068 2007-01-31 Dirk Mueller <dmueller@suse.de>
1070 * c-common.c (warn_about_parentheses): Separate warning about
1071 un-parenthized sequence of comparison operators from the one
1072 which is supposed to warn about x <= y <= z.
1074 2007-01-31 Uros Bizjak <ubizjak@gmail.com>
1076 * optabs.h (enum optab_index): Add new OTI_isinf.
1077 (isinf_optab): Define corresponding macro.
1078 * optabs.c (init_optabs): Initialize isinf_optab.
1079 * genopinit.c (optabs): Implement isinf_optab using isinf?f2
1081 * builtins.c (mathfn_built_in): Handle BUILT_IN_ISINF{,F,L}.
1082 (expand_builtin_interclass_mathfn): Expand BUILT_IN_ISINF{,F,L}
1084 (expand_builtin): Expand BUILT_IN_ISINF{,F,L} using
1085 expand_builtin_interclass_mathfn.
1086 * reg_stack.c (subst_stack_regs_pat): Handle UNSPEC_FXAM.
1087 * config/i386/i386.md (UNSPEC_FXAM): New constant.
1088 (fxam<mode>2_i387): New insn pattern.
1089 (isinf<mode>2) New expander to implement isinf, isinff and isinfl
1090 built-in functions as x87 inline asm.
1092 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1094 * gcc/config/arm/unwind-arm.h (_sleb128_t, _uleb128_t): New.
1096 2007-01-30 Eric Christopher <echristo@apple.com>
1098 * config.gcc: Add geode.
1100 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1102 * cgraphunit.c, config/arm/arm.c, config/m68k/m68k.c,
1103 ipa-inline.c, tree-profile.c, tree-ssa-live.c,
1104 tree-ssa-math-opts.c, tree-ssanames.c, tree-vect-analyze.c,
1105 value-prof.c: Fix comment typos.
1107 2007-01-30 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1110 * doc/invoke.texi (Wpointer-arith): Document warning.
1112 2007-01-30 Janis Johnson
1114 * doc/extend.texi (Decimal Floating Types): Remove decfloat.h from
1115 the list of discrepancies from the draft TR.
1117 2007-01-30 Dirk Mueller <dmueller@suse.de>
1120 * doc/invoke.texi (-Wreturn-type): Update description to
1123 2007-01-30 Richard Sandiford <richard@codesourcery.com>
1125 * cfgrtl.c (try_redirect_by_replacing_jump): Check only_sets_cc0_p.
1127 2007-01-30 Uros Bizjak <ubizjak@gmail.com>
1129 * builtins.c (expand_builtin_int_interclass_roundingfn): New function
1130 to handle optabs that operate on floating point input argument and
1131 output to integer output.
1132 (expand_builtin_mathfn) [BUILT_IN_ILOGB]: Move from here ...
1133 (expand_builtin_interclass_mathfn) [BUILT_IN_ILOGB]: ... to here.
1134 (expand_builtin): Expand BUILT_IN_ILOGB{,F,L} using
1135 expand_builtin_interclass_mathfn ().
1136 * config/i386/i386.md (fxtractxf3_i387): Rename from *fxtractxf3_i387.
1138 (ilogbxf2, ilogb<mode>2): New expanders to implement ilogb, ilogbf and
1139 ilogbl built-in functions as x87 intrinsics.
1141 2007-01-30 Richard Guenther <rguenther@suse.de>
1144 * dwarf2out.c (reference_to_unused): Query varpool if the
1145 variable was output.
1147 2007-01-30 Richard Guenther <rguenther@suse.de>
1150 * passes.c (execute_one_pass): Reset in_gimple_form to not
1151 confuse non-unit-at-a-time mode.
1153 2007-01-29 Roger Sayle <roger@eyesopen.com>
1154 Richard Guenther <rguenther@suse.de>
1156 * fold-const.c (round_up): Make HIGH an unsigned HOST_WIDE_INT to
1157 avoid undefined behaviour on overflow. Use force_fit_type_double
1158 to construct the constant with the specified TREE_OVERFLOW.
1160 2007-01-29 Janis Johnson <janis187@us.ibm.com>
1162 * config/dfp-bit.c: Add parameterized support for fp exceptions.
1163 * config/dfp-bit.h: Ditto.
1165 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1167 * c-decl.c (pop_scope): Replace warnings with call to
1168 warn_for_unused_label.
1169 * c-common.h (warn_for_unused_label): Declare.
1170 * c-common.c (warn_for_unused_label): Define.
1172 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1174 * tree-optimize.c (update_inlined_to_pointers): Delete unused
1177 2007-01-29 Janis Johnson <janis187@us.ibm.com>
1179 * Makefile.in (USER_H): Remove decfloat.h.
1180 * ginclude/decfloat.h: Delete, moving contents to ...
1181 * ginclude/float.h: Add support for decimal floating point,
1182 guarded by __STDC_WANT_DEC_FP__.
1184 2007-01-29 Mike Stump <mrs@apple.com>
1186 * doc/gccint.texi (Top): Rename Loop Representation to Loop
1187 Analysis and Representation to resolve case insensitive conflict.
1188 * doc/loop.texi (Loop Analysis and Representation): Likewise.
1190 2007-01-28 Daniel Berlin <dberlin@dberlin.org>
1192 * tree.h (struct tree_memory_tag): Add aliases member.
1193 (MTAG_ALIASES): New macro.
1194 * tree-ssa-alias.c (alias_bitmap_obstack): New variable.
1195 (add_may_alias): Remove pointer-set. Update for may_aliases being
1197 (mark_aliases_call_clobbered): Update for may_aliases being a
1199 (compute_tag_properties): Ditto.
1200 (create_partition_for): Ditto.
1201 (compute_memory_partitions): Ditto.
1202 (dump_may_aliases_for): Ditto.
1203 (is_aliased_with): Ditto.
1204 (add_may_alias_for_new_tag): Ditto.
1205 (rewrite_alias_set_for): Rewrite for may_aliases being a bitmap.
1206 (compute_is_aliased): New function.
1207 (compute_may_aliases): Call compute_is_aliased).
1208 (init_alias_info): Initialize alias_bitmap_obstack.
1209 (union_alias_set_into): New function.
1210 (compute_flow_sensitive_aliasing): Use union_aliases_into.
1211 (have_common_aliases_p): Rewrite to take two bitmaps and use
1213 (compute_flow_insensitive_aliasing): Stop using pointer-sets.
1215 (finalize_ref_all_pointers): Update for add_may_alias changes.
1216 (new_type_alias): Ditto.
1217 * tree-flow-inline.h (may_aliases): Return a bitmap.
1218 * tree-dfa.c (dump_variable): Check for MTAG_P'ness.
1219 * tree-ssa.c (verify_flow_insensitive_alias_info): Update for
1220 may_aliases being a bitmap.
1221 * tree-flow.h (struct var_ann_d): Remove may_aliases member.
1222 may_aliases now returns a bitmap.
1223 * tree-ssa-structalias.c (merge_smts_into): Update for may_aliases
1225 * tree-ssa-operands.c (add_virtual_operand): Update for
1226 may_aliases being a bitmap.
1228 2007-01-29 Daniel Berlin <dberlin@dberlin.org>
1230 PR tree-optimization/30630
1231 * tree-ssa-structalias.c (do_complex_constraint): Mark correct
1232 variable as changed.
1234 2007-01-29 Simon Martin <simartin@users.sourceforge.net>
1237 * gimplify.c (gimplify_target_expr): Make sure that the TARGET_EXPR is
1238 expanded only once even if an error occurs.
1240 2007-01-29 Ben Elliston <bje@au.ibm.com>
1242 * gcov-io.h (__gcov_indirect_call_profiler): Declare.
1243 (__gcov_average_profiler): Likewise.
1244 (__gcov_ior_profiler): Likewise.
1245 (__gcov_merge_ior): Likewise.
1247 2007-01-28 Jan Hubicka <jh@suse.cz>
1249 * builtins.c (expand_builtin_memset): Fix typo in my last patch.
1250 * value-prof.c (stringop_block_profile): Likewise.
1252 2007-01-28 Jan Hubicka <jh@suse.cz>
1254 * expr.c (emit_block_move_via_movmem, emit_block_move_via_libcall): Add
1255 variant handling histograms; add wrapper.
1256 (clear_storage_via_libcall): Export.
1257 (emit_block_move_hints): Break out from ...; add histograms.
1258 (emit_block_move): ... this one.
1259 (clear_storage_hints): Break out from ...; add histograms.
1260 (clear_storage): ... this one.
1261 (set_storage_via_memset): Handle histogram.
1262 * expr.h (emit_block_move_via_libcall, emit_block_move_hints): Declare.
1263 (clear_storage_hints, clear_storage_via_libcall): Declare.
1264 (set_storage_via_setmem): Update prototype.
1265 * doc/md.texi (movmem, setmem): Document new arguments.
1267 * value-prof.c (dump_histogram_value, tree_find_values_to_profile): Add
1269 (stringop_block_profile): New global function.
1270 (tree_stringops_values_to_profile): Profile block size and alignment.
1271 * value-prof.h (enum hist_type): add HIST_TYPE_AVERAGE and
1273 (struct profile_hooks): Add gen_average_profiler and gen_ior_profiler.
1274 (stringop_block_profile): Declare.
1275 * builtins.c: Include value-prof.h.
1276 (expand_builtin_memcpy, expand_builtin_memset): Pass block profile.
1277 * gcov-ui.h (GCOV_COUNTER_NAMES): Add new counter.
1278 (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): New constants.
1279 (GCOV_COUNTERS, GCOV_LAST_VALUE_COUNTER): Update.
1280 * profile.c (instrument_values): Add new counters.
1281 * cfgexpand.c (expand_gimple_basic_block): Propagate histograms to
1283 * tree-profile.c (tree_average_profiler_fn, tree_ior_profiler_fn): New.
1284 (tree_init_edge_profiler): Build new profilers.
1285 (tree_gen_average_profiler, tree_gen_ior_profiler): New.
1286 (pass_tree_profile): Add dump.
1287 (tree_profile_hooks): Update.
1288 * Makefile.in (LIBGCOV): Add new constants.
1289 * libgcov.c (__gcov_merge_ior, __gcov_average_profiler,
1290 __gcov_ior_profiler): New.
1291 * i386.md (movmem/setmem expanders): Add new optional arguments.
1293 2007-01-28 David Edelsohn <edelsohn@gnu.org>
1295 * doc/md.texi (Standard Pattern Names): Document blockage pattern.
1297 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
1299 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
1300 (tree_unswitch_single_loop): Pass eni_size_weights to
1301 tree_num_loop_insns.
1302 * tree-ssa-loop-manip.c: Include tree-inline.h.
1303 (can_unroll_loop_p): Pass eni_size_weights to
1304 tree_num_loop_insns.
1305 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
1306 Pass eni_size_weights to estimate_num_insns.
1307 * tree.h (init_inline_once): Export.
1308 * toplev.c (backend_init): Call init_inline_once.
1309 * cgraphunit.c (cgraph_process_new_functions,
1310 cgraph_analyze_function): Pass eni_inlining_weights to
1312 * ipa-inline.c (compute_inline_parameters): Ditto.
1313 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
1314 to estimate_num_insns.
1315 (try_unroll_loop_completely): Pass eni_size_weights to
1316 tree_num_loop_insns.
1317 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
1318 ot estimate_num_insns.
1319 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
1320 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
1321 * tree-inline.c (eni_inlining_weights, eni_size_weights,
1322 eni_time_weights): New variables.
1323 (init_inline_once): Initialize them.
1324 (struct eni_data): Mew.
1325 (estimate_num_insns_1, estimate_num_insns): Use weights.
1326 * tree-inline.h (struct eni_weights_d): New.
1327 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
1328 (estimate_num_insns): Declaration changed.
1329 * cfgloop.h (tree_num_loop_insns): Declaration changed.
1330 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
1331 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
1333 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
1335 * tree-data-ref.c (conflict_fn): Assert that the number of affine
1336 relations in the conflict function is valid.
1338 2007-01-27 Ian Lance Taylor <iant@google.com>
1340 * common.opt: Add fstrict-overflow.
1341 * opts.c (decode_options): Set flag_strict_overflow if -O2.
1342 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
1343 (TYPE_OVERFLOW_UNDEFINED): Define.
1344 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
1346 * tree.h (TYPE_TRAP_SIGNED): Don't define.
1347 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
1348 (fold_negate_expr): Likewise.
1349 (make_range): Likewise.
1350 (extract_muldiv_1): Likewise.
1351 (maybe_canonicalize_comparison): Likewise.
1352 (fold_comparison): Likewise.
1353 (fold_binary): Likewise.
1354 (tree_expr_nonnegative_p): Likewise.
1355 (tree_expr_nonzero_p): Likewise.
1356 * tree-vrp.c (compare_values): Likewise.
1357 (extract_range_from_binary_expr): Likewise.
1358 (extract_range_from_unary_expr): Likewise.
1359 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
1361 (nowrap_type_p): Likewise.
1362 * tree-scalar-evolution.c (simple_iv): Likewise.
1363 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
1364 (build_range_check): Likewise.
1365 (extract_muldiv_1): Likewise.
1366 (fold_comparison): Likewise.
1367 * tree-vrp.c (vrp_int_const_binop): Likewise.
1368 (extract_range_from_unary_expr): Likewise.
1369 * convert.c (convert_to_integer): Likewise.
1370 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
1371 (fold_comparison): Likewise.
1372 (fold_binary): Likewise.
1373 * optabs.c (optab_for_tree_code): Likewise.
1374 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
1375 * simplify-rtx.c (simplify_const_relational_operation): Check
1376 flag_strict_overflow and flag_trapv.
1377 (simplify_const_relational_operation): Likewise.
1378 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
1379 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
1382 2007-01-27 Roger Sayle <roger@eyesopen.com>
1384 * tree.c (tree_fold_gcd): Delete.
1385 * tree.h (tree_fold_gcd): Remove prototype.
1386 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
1387 test whether one constant integer is a multiple of another. Instead
1388 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
1389 * fold-const.c (multiple_of_p): We've determined both TOP and
1390 BOTTOM are integer constants so we can call int_const_binop directly
1391 instead of the more generic const_binop.
1393 2007-01-27 Roger Sayle <roger@eyesopen.com>
1395 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
1396 1*X check that the constant hasn't overflowed, to preserve the
1398 (round_up): Provide an efficient implementation when rouding-up an
1399 INTEGER_CST to a power-of-two.
1401 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
1403 * doc/sourcebuild.texi: Add comma for clarity.
1404 * doc/extend.texi: Fix some typos.
1405 * doc/passes.texi: Likewise.
1406 * doc/cppinternals.texi: Likewise.
1407 * doc/c-tree.texi: Likewise.
1408 * doc/tree-ssa.texi: Likewise.
1409 * doc/install.texi: Likewise.
1411 2007-01-27 Jan Hubicka <jh@suse.cz>
1413 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
1414 (pass_sra): Do not require alias information.
1415 * passes.c (init_optimization_passes): Add SRA
1417 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
1419 * tracer.c (rest_of_handle_tracer): We already cleaned
1420 up the CFG in tracer() so don't do it here again.
1421 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
1422 cleanup_cfg here, either. And don't call renumber_insns.
1424 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
1425 and BB_HEAD are in the insn stream here. Instead make sure
1426 that BB_INSN is valid on all insns. Also, do check here that
1427 there are no pending branch predictions...
1428 (rtl_verify_flow_info): ...instead of doing it here. Checks
1429 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
1432 2007-01-26 Roger Sayle <roger@eyesopen.com>
1434 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
1435 function to simplify/factorize operand order canonicalization.
1436 (ix86_fixup_binary_operands): Reorganize using the above function.
1437 (ix86_binary_operator_ok): Likewise.
1439 2007-01-27 Jakub Jelinek <jakub@redhat.com>
1441 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
1442 file scope from optimize_attrs.
1443 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
1445 (optimize_attrs): Clear insn_code_values after freeing it.
1447 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
1449 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
1450 a parts.base assignment.
1452 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
1454 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
1455 (compute_subscript_distance, initialize_data_dependence_relation,
1456 finalize_ddr_dependent, analyze_ziv_subscript,
1457 analyze_siv_subscript_cst_affine,
1458 compute_overlap_steps_for_affine_univar,
1459 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
1460 analyze_siv_subscript, analyze_miv_subscript,
1461 analyze_overlapping_iterations, subscript_dependence_tester_1,
1462 compute_self_dependence, free_dependence_relation): Work
1463 with affine_fn instead of chrecs.
1464 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
1465 common_affine_function, affine_function_base,
1466 affine_function_constant_p, affine_fn_op, affine_fn_plus,
1467 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
1468 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
1469 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
1470 (all_chrecs_equal_p): Removed.
1471 * tree-data-ref.h (affine_fn, conflict_function): New types.
1472 (struct subscript): Change type of conflicting_iterations_in_a
1473 and conflicting_iterations_in_b.
1475 2007-01-26 Steve Ellcey <sje@cup.hp.com>
1478 * config/pa/pa.h (TARGET_HPUX_11): New.
1479 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
1480 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
1482 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
1483 Richard Guenther <rguenther@suse.de>
1485 * tree-ssa-structalias.c (solve_graph): Handle case
1486 we merged the variable to another.
1488 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1490 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
1493 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
1494 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
1496 2007-01-25 DJ Delorie <dj@redhat.com>
1498 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
1499 allow changes to modes which don't fit in those registers.
1501 * reload1.c (choose_reload_regs): Check for invalid subregs before
1502 computing their locations, not after.
1504 2007-01-25 Geoffrey Keating <geoffk@apple.com>
1507 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
1508 save Altivec registers in an eh_return function.
1509 (compute_vrsave_mask): Likewise.
1510 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
1511 for saving Altivec registers.
1512 (rs6000_emit_prologue): Don't allocate stack twice in
1513 eh_return function. Correct expected value of altivec_save_offset
1514 when using save_world. Describe save of R0 to stack when using
1515 save_world. Describe stack pointer adjustment when using
1516 save_world. Remove duplicated eh_return parameter register saving.
1517 Update sp_offset variable after save_world.
1518 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
1520 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
1521 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
1524 2007-01-25 Steve Ellcey <sje@cup.hp.com>
1526 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
1528 2007-01-25 Steve Ellcey <sje@cup.hp.com>
1531 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
1532 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
1534 2007-01-25 Richard Guenther <rguenther@suse.de>
1536 * doc/invoke.texi (-Wcoverage-mismatch): Document.
1537 * common.opt (-Wcoverage-mismatch): New warning option.
1538 * coverage.c (get_coverage_counts): Ignore coverage mismatch
1539 if -Wcoverage-mismatch is given.
1541 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
1543 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
1544 (ipcp_driver): Change to static definition.
1545 Add dumping of the ifunctions.
1546 (constant_val_insert): Remove unused parameter. Support for SSA.
1547 (ipcp_propagate_const): Support for SSA.
1548 (ipcp_profile_bb_print): Print only analyzed nodes.
1549 (ipcp_replace_map_create): Remove support for Fortran constant
1551 * ipa-prop.c (ipa_method_modify_stmt,
1552 ipa_callsite_compute_param): Support for SSA.
1553 * ipa-prop.h (ipcp_driver): Remove declaration.
1554 (IS_VALID_TREE_MAP_INDEX): Add define.
1556 2007-01-24 Geoffrey Keating <geoffk@apple.com>
1558 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
1560 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1563 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
1564 constant argument to cabs and do it without checking for
1565 -funsafe-math-optimizations.
1567 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
1569 * c-common.h (RID_FIRST_CXX0X): New.
1570 (RID_LAST_CXX0X): New.
1571 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
1573 * c.opt (Wc++0x-compat): New.
1574 * doc/invoke.texi (-Wc++0x-compat): Document.
1576 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1578 * builtins.c (fold_builtin_carg): New.
1579 (fold_builtin_1): Use it.
1581 2007-01-24 Jan Hubicka <jh@suse.cz>
1583 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
1585 * ipa-inline.c (initial_insns, max_insns): Delete.
1586 (compute_max_insns): New function.
1587 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
1588 of insns as base for code growth.
1589 (cgraph_decide_inlining): Make initial_insns local; do not compute
1591 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
1592 * doc/invoke.texi (inline-unit-growth): Update docs.
1594 2007-01-24 Jakub Jelinek <jakub@redhat.com>
1596 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
1597 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
1598 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
1599 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
1600 for CPUs that have PTA_CX16 set.
1603 * gimplify.c (omp_check_private): New function.
1604 (gimplify_scan_omp_clauses): Use it for
1605 firstprivate/lastprivate/reduction.
1608 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
1609 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
1612 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
1613 If lower_lastprivate_clauses emits some statements, append them
1614 to dlist rather than body_p and to body_p append an initializer.
1615 (lower_omp_for): Adjust caller.
1617 2007-01-24 Steve Ellcey <sje@cup.hp.com>
1619 * target.h (globalize_decl_name): New.
1620 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
1621 * output.h (default_globalize_decl_name): New.
1622 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
1624 (globalize_decl): Ditto.
1625 (default_globalize_decl_name): New.
1626 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
1627 (ia64_handle_version_id_attribute): New.
1628 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
1629 (ia64_asm_output_external): Use globalize_decl_name instead
1631 * doc/extend.texi (version_id): New pragma.
1632 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
1634 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
1636 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
1637 _uleb128_t and _Unwind_SWord with _sleb128_t.
1638 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
1639 uw_frame_state_for, uw_update_context_1): Likewise.
1640 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
1641 * unwind-pe.h (read_uleb128, read_sleb128,
1642 read_encoded_value_with_base): Likewise.
1643 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
1645 2007-01-24 Richard Guenther <rguenther@suse.de>
1647 * builtins.c (expand_builtin_cexpi): Get the fndecl
1648 for cexp in the correct way.
1650 2007-01-24 Jan Hubicka <jh@suse.cz>
1652 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
1654 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
1656 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
1657 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
1658 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
1659 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
1660 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
1661 __SIZEOF_PTRDIFF_T__.
1662 * doc/cpp.texi: Documentation for the new builtin macros added.
1664 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
1666 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
1667 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
1668 instead of operands[2] to avoid access past the end of array.
1670 2007-01-24 Richard Sandiford <richard@codesourcery.com>
1672 * reload1.c (emit_reload_insns): Pass the reload register
1673 for a non-spill output reload through forget_old_reloads_1.
1675 2007-01-23 Joseph Myers <joseph@codesourcery.com>
1677 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
1678 _SOFT_DOUBLE if doubles use software floating-point.
1679 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
1680 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1681 * config/rs6000/darwin-ldouble.c: Also compile functions for
1682 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
1683 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1684 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
1685 commented-out long double override.
1686 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
1687 * config/rs6000/eabispe.h: Likewise.
1688 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
1689 long double for non-SPE.
1690 (rs6000_handle_option): Likewise.
1691 (invalid_e500_subreg): Disallow more subregs involding DImode,
1692 DFmode, TImode or TFmode.
1693 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
1695 (legitimate_lo_sum_address_p): Also check for TFmode for E500
1697 (rs6000_legitimize_address): Also handle TFmode for E500 double.
1698 (rs6000_legitimize_reload_address): Also handle TFmode for E500
1700 (rs6000_legitimate_address): Also check for TFmode for E500
1702 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
1703 (spe_build_register_parallel): Handle TFmode and TCmode.
1704 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
1706 (function_arg): Handle TFmode and TCmode for E500 double.
1707 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
1709 (print_operand): Handle TFmode and TImode for %y.
1710 (rs6000_generate_compare): Handle TFmode comparisons for E500
1712 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
1713 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
1714 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
1715 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
1716 TFmode for E500 double.
1717 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
1718 (floatsidf2): Enable for E500 double.
1719 (movtf_softfloat): Use rs6000_nonimmediate_operand.
1720 (extenddftf2): Change to extenddftf2_fprs.
1721 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
1722 depending on TARGET_E500_DOUBLE.
1723 (extendsftf2): Enable for E500 double.
1724 (trunctfdf2): Enable for E500 double.
1725 (trunctfsf2): Change to trunctfsf2_fprs.
1726 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
1727 depending on TARGET_E500_DOUBLE.
1728 (floatsitf2): Enable for E500 double.
1729 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
1730 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
1731 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
1732 (negtf2): Change to negtf2_internal.
1733 (negtf2): New expander.
1734 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
1735 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
1736 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
1737 (movdi_internal32): Use rs6000_nonimmediate_operand.
1738 (unnamed splitter): Likewise.
1739 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
1740 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
1741 (SPE64TF, DITI): New mode macros.
1742 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
1745 (frob_<mode>_di_2): New.
1746 (frob_tf_di_8_2): New.
1747 (frob_di_df): Change to frob_di_<mode>; allow more modes.
1749 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
1751 (frob_ti_<mode>_8_2): New.
1752 (frob_ti_tf_2): New.
1753 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
1754 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
1756 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
1757 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
1758 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
1759 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
1760 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
1761 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
1764 2007-01-23 Ian Lance Taylor <iant@google.com>
1766 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
1768 (OBJS-archive): Reformat, alphabetize.
1769 (OBJS): Change out_object_file to OBJS-md.
1771 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
1773 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
1774 JUMP_P and CALL_P predicates where applicable.
1775 * config/i386/i386.c: Ditto.
1776 * config/i386/i386.md: Ditto.
1777 * config/i386/mmx.md: Ditto.
1778 * config/i386/predicates.md: Ditto.
1780 2007-01-22 Andreas Schwab <schwab@suse.de>
1782 * config/m68k/m68k.h: Fix comment.
1784 2007-01-22 Jan Hubicka <jh@suse.cz>
1786 * passes.c (init_optimization_passes): Do not rerun
1787 pass_early_warn_uninitialized.
1789 2007-01-22 Richard Guenther <rguenther@suse.de>
1791 PR tree-optimization/30038
1792 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
1794 (execute_cse_sincos_1): Likewise.
1795 (execute_cse_sincos): Likewise.
1796 (gate_cse_sincos): Likewise.
1797 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
1798 the cexpi() canonicalization of sincos().
1799 * tree-pass.h (pass_cse_sincos): Declare.
1800 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
1802 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
1804 PR rtl-optimization/29329
1805 * combine.c (replaced_rhs_insn): Rename to i2mod.
1806 (replaced_rhs_value): Rename to i2mod_new_rhs.
1807 (i2mod_old_rhs): New global variable.
1808 (combine_instructions): Adjust for above change. Save a copy of
1809 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
1810 note are substituted in the second instruction.
1811 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
1812 ditch the note if it pertains to the second eliminated register
1813 and this register is mentioned in i2mod_old_rhs.
1816 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
1818 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
1819 the insn to which the note was originally attached.
1821 2007-01-21 Jan Hubicka <jh@suse.cz>
1823 * ipa-inline.c (inlining_mode): Comment, move up.
1824 (cgraph_decide_inlining_incrementally): Do not perform inlining
1825 itself; fix handling of flattening of self recursive functions.
1826 (cgraph_find_cycles): Remove.
1827 (cgraph_flatten_node): Remove.
1828 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
1829 (try_inline): New function.
1830 (cgraph_early_inlining): Update call of
1831 cgraph_decide_inlining_incrementally. Apply inlining here.
1832 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
1834 2007-01-21 Dirk Mueller <dmueller@suse.de>
1837 * tree-vrp.c (check_array_bounds): do not warn
1838 about ADDR_EXPR's of ARRAY_REF's which are immediately
1839 used in binary expressions.
1841 2007-01-21 Ira Rosen <irar@il.ibm.com>
1843 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
1844 read_write_dep and macros for its access.
1845 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
1846 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
1847 argument, call vect_check_interleaving for every independent pair of
1848 data-refs. Mark loads that access the same memory location as a store
1850 (vect_check_dependences): Remove.
1851 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
1852 call, fix the call to vect_analyze_data_ref_dependence.
1853 (vect_analyze_data_ref_access): For statements that access the same
1854 data-ref, check that they are not stores; for loads, check that there
1855 is no store that access the same location.
1857 2007-01-20 Roger Sayle <roger@eyesopen.com>
1858 Joseph Myers <joseph@codesourcery.com>
1860 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
1861 (-fdump-rtl-stack): Likewise.
1862 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
1863 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
1864 (-mpt-fixed): Use @minus{} for minus sign.
1865 (Using Precompiled Headers): Fix under/overfull hbox.
1867 2007-01-20 Jan Hubicka <jh@suse.cz>
1869 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
1870 place within annotation.
1872 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1874 * pa.c (output_move_double): Change array size of xoperands to 4.
1876 2007-01-20 Jan Hubicka <jh@suse.cz>
1878 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
1881 2007-01-19 Ian Lance Taylor <iant@google.com>
1883 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
1884 constant to unsigned HOST_WIDE_INT before negating.
1886 2007-01-19 Ian Lance Taylor <iant@google.com>
1888 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
1890 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
1891 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1892 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
1893 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
1894 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1895 (realloc_vdef, realloc_vuse): Change second parameter to
1897 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
1899 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
1900 unsigned. Change x and lim locals to unsigned.
1901 (realloc_vdef, realloc_vuse): Change num_elem parameter to
1903 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
1904 (copy_virtual_operands): Change i and n locals to unsigned.
1906 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1909 * builtins.c (fold_builtin_1): Handle builtin fdim.
1911 2007-01-20 Jan Hubicka <jh@suse.cz>
1913 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
1914 * tree-ssa-structalias.c (compute_points_to_sets): Do call
1915 init_alias_heapvars.
1916 (init_alias_heapvars): Initialize only when not already initialized.
1917 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
1919 2007-01-19 Roger Sayle <roger@eyesopen.com>
1921 * common.opt (fsigned-zeros): New command line option.
1922 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
1923 of flag_unsafe_math_optimizations.
1924 * opts.c (set_fast_math_flags): The -ffast-math command line option
1925 implies -fno-signed-zeros.
1926 (fast_math_flags_set_p): Likewise.
1928 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
1929 the documentation of -ffast-math appropriately. Wrap long lines.
1931 2007-01-19 Steve Ellcey <sje@cup.hp.com>
1933 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
1934 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
1936 2007-01-19 Tomas Bily <tbily@suse.cz>
1938 * cgraphunit.c (cgraph_finalize_function): Updating of pid
1940 (tree_init_ic_make_global_vars): New function
1941 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
1942 (tree_gen_ic_profiler): New function
1943 (tree_gen_ic_func_profiler): New function
1944 (tree_profiling): Added calling of tree_gen_ic_func_profiler
1945 (tree_profile_hooks): Added hook for indirec/virtual calls
1946 * value-prof.c (tree_find_values_to_profile): New case for
1948 (tree_values_to_profile): Call for determining indirect/virtual
1950 (tree_indirect_call_to_profile): New function
1951 (tree_ic_transform): New function
1952 (tree_ic): New function
1953 (find_func_by_pid): New function
1954 (init_pid_map): New function
1955 (tree_value_profile_transformations): Added check for
1956 indirect/virtual call transformation
1957 * value-prof.h (enum hist_type): New counter type for
1958 indirect/virtual calls
1959 (profile_hooks): Added new hook for profiling indirect/virtual
1961 * profile.c (instrument_values): New case for indirect/virtual
1963 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
1964 (GCOV_COUNTER_V_INDIR): New counter type
1965 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
1966 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
1968 * cgraph.c: Definition of cgraph_max_pid
1969 (cgraph_create_node): Default init of pid attribute
1970 * cgraph.h: Declaration of cgraph_max_pid
1971 (struct cgraph_node): Added pid attribute
1972 * libgcov.c (__gcov_indirect_call_profiler): New function
1973 (__gcov_one_value_profiler_body): New function
1974 (__gcov_one_value_profiler): Body was moved to
1975 __gcov_one_value_profiler_body and calls it
1977 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
1979 * doc/gty.texi (Options): Document the mark_hook option to GTY.
1980 * gengtype.c (write_types_data, write_func_for_structure,
1981 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
1982 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
1983 (walk_type, write_func_for_structure): Generate the mark_hook if
1986 2007-01-19 Jan Hubicka <jh@suse.cz>
1988 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
1989 'early' argument take inlining mode argument specifying whether to
1990 inline for size/speeed or all functions; add support for flattening;
1992 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
1994 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1997 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
1998 the name of the declared identifier.
2000 2007-01-19 Dirk Mueller <dmueller@suse.de>
2002 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
2003 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
2004 duplicate evaluation.
2006 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
2008 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
2009 Macroize expander using X87MODEF12 mode macro. Extend operand 1
2010 to XFMode, use acosxf2 and truncate result to requested mode.
2011 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
2012 (asin<mode>2): Similarly, with asin expanders.
2013 (*fscalexf4_i387): Rename from *fscalexf4.
2014 (expNcorexf3): New expander.
2015 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
2016 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
2017 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
2018 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
2019 patterns for SSE math.
2020 (exp10<mode>2): Similarly, with exp10 expanders.
2021 (exp2<mode>2): Similarly, with exp2 expanders.
2022 (expm1<mode>2): Similarly, with expm1 expanders.
2023 (ldexp<mode>3): Similarly, with ldexp expanders.
2024 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
2025 Use gen_truncxf<mode>2_i387_noop to truncate result.
2027 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2029 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
2030 for ColdFire targets.
2032 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
2033 Richard Sandiford <richard@codesourcery.com>
2035 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
2036 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
2037 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
2038 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
2039 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
2040 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
2042 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2044 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
2045 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
2046 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
2047 (*adddi_dilshr32_cf): New pattern.
2048 (adddi3, subdi3): Remove first alternatives.
2050 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2052 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
2053 implemented using move.l, do not use its cc result for floating-point
2056 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2058 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
2059 from accepting 's' constraints if flag_pic.
2061 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2063 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
2064 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
2065 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
2066 condition from TARGET_68881 to TARGET_HARD_FLOAT.
2068 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
2070 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
2071 ColdFire alternatives.
2072 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
2073 * config/m68k/m68k.md (clzsi2): Define for ColdFire
2074 architectures that support ff1 instruction.
2076 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2077 Julian Brown <julian@codesourcery.com>
2079 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
2080 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
2081 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
2084 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2086 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
2088 (output_move_const_into_data_reg, output_move_simode_const): Delete.
2089 * config/m68k/m68k.c (const_method, const_int_cost): Take a
2090 HOST_WIDE_INT instead of an rtx.
2091 (m68k_rtx_costs): Update call accordingly.
2092 (output_move_const_into_data_reg): Likewise. Fix formatting.
2093 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
2095 (output_move_simode_const): Update calls after above changes.
2096 Rework to use automatic variables and predicates like MEM_P.
2097 * config/m68k/m68k.md (pushexthisi_const): Update call to
2100 2007-01-19 Dirk Mueller <dmueller@suse.de>
2102 * tree-ssa-alias.c (perform_var_substitution): Fix typo
2105 2007-01-19 Richard Guenther <rguenther@suse.de>
2107 * builtins.c (expand_builtin_cexpi): Fall back to expanding
2108 via cexp in case sincos is not available.
2110 2007-01-19 Richard Guenther <rguenther@suse.de>
2112 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
2113 * defaults.h (TARGET_HAS_SINCOS): Default to off.
2114 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
2115 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
2116 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
2117 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
2118 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
2119 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
2121 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
2123 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
2125 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
2126 (atan2xf3): Directly generate RTL pattern.
2127 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
2128 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
2129 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
2130 disable patterns for SSE math.
2131 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
2132 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
2133 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
2134 disable patterns for SSE math.
2136 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
2138 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
2140 2007-01-18 Roger Sayle <roger@eyesopen.com>
2142 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
2143 VIEW_CONVERT_EXPR to the same type as it's operand.
2145 2007-01-18 David Edelsohn <edelsohn@gnu.org>
2147 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
2148 configured for long double 128.
2150 2007-01-18 Mike Stump <mrs@apple.com>
2152 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
2155 2007-01-18 Michael Meissner <michael.meissner@amd.com>
2157 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
2160 2007-01-18 Jan Hubicka <jh@suse.cz>
2162 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
2163 greater than NUM_VOP_FREE_BUCKETS.
2165 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
2167 * tree-ssa-structalias.c: Update comments.
2168 (ptabitmap_obstack): Removed.
2170 (oldpta_obstack): Ditto.
2171 (stats): Add a few members.
2172 (struct variable_info): Remove node, complex, address_taken, and
2173 indirect_target members. Add oldsolution member.
2174 (new_var_info): Do not initialize removed members.
2175 (constraint_expr_type): Remove INCLUDES.
2176 (constraint_graph): Add size, implicit_preds, rep,
2177 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
2178 (FIRST_REF_NODE): New macro.
2179 (LAST_REF_NODE): Ditto.
2180 (FIRST_ADDR_NODE): Ditto.
2181 (find): New function.
2183 (dump_constraint): Do not handle INCLUDES.
2184 (insert_into_complex): Do not insert duplicate constraints.
2185 (condense_varmap_nodes): Renamed and rewritten into ...
2186 (merge_node_constraints): This. Also fix bug in handling of
2187 offseted copy constraints.
2188 (clear_edges_for_node): No longer need to deal with preds at all,
2189 or removing associated preds/succs.
2190 (merge_graph_nodes): Deal with indirect_cycles.
2191 Don't deal with predecessors.
2192 (add_implicit_graph_edge): New function.
2193 (add_pred_graph_edge): Ditto.
2194 (add_graph_edge): Don't deal with predecessors.
2195 (build_constraint_graph): Removed.
2196 (build_pred_graph): New function.
2197 (build_succ_graph): Ditto.
2198 (struct scc_info): Removed in_component. Added roots, dfs, and
2199 node_mapping. Remove visited_index, unification_queue.
2200 (scc_visit): Deal with union-find we do now.
2201 Deal with cycles with REF nodes.
2202 (collapse_nodes): Renamed and rewritten to ...
2203 (unify_nodes): This.
2204 (process_unification_queue): Removed.
2205 (topo_visit): Cleanup
2206 (do_da_constraint): Use find.
2207 (do_sd_constraint): Ditto.
2208 (do_ds_constraint): Ditto.
2209 (do_complex_constraint): Ditto.
2210 (init_scc_info): Update for removed and added members.
2211 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
2212 (find_indirect_cycles): This.
2213 (equivalence_class): New variable.
2214 (label_visit): New function.
2215 (perform_variable_substitution): Rewritten.
2216 (free_var_substitution_info): New function.
2217 (find_equivalent_node): Ditto.
2218 (move_complex_constraints): Ditto.
2219 (eliminate_indirect_cycles): Ditto.
2220 (solve_graph): Only propagate changed bits.
2221 Use indirect cycle elimination.
2223 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
2224 (tree_id_eq): Renamed to ...
2225 (tree_vi_eq): This. Update for member change
2226 (insert_id_for_tree): Renamed and rewritten to ...
2227 (insert_vi_for_tree): This.
2228 (lookup_id_for_tree): Renamed and rewritten to ...
2229 (lookup_vi_for_tree): This.
2230 (get_id_for_tree): Renamed and rewritten to ...
2231 (get_vi_for_tree): Ditto.
2232 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
2233 (process_constraint): Don't handle INCLUDES.
2234 Remove special ADDRESSOF case.
2235 (find_func_aliases): Rewrite to use vi functions instead of id
2237 (create_function_info_for): Ditto.
2238 (create_variable_info_for): Ditto.
2239 (intra_create_variable_infos): Ditto.
2240 (merge_smts_into): Ditto.
2241 (find_what_p_points_to): Ditto.
2242 (init_base_vars): Ditto.
2243 (init_alias_vars): Ditto.
2244 (remove_preds_and_fake_succs): New function.
2245 (dump_sa_points_to_info): Dump new stats.
2246 (dump_solution_for_var): Use find.
2247 (set_used_smts): Fix formatting.
2248 (compute_points_to_sets): Updated for new functions.
2249 (ipa_pta_execute): Ditto.
2251 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2252 Richard Sandiford <richard@codesourcery.com>
2254 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
2256 * calls.c (expand_call): If the return value is a PARALLEL,
2257 extract its first member.
2258 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
2259 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
2260 use FUNCTION_EXTRA_EPILOGUE.
2261 (m68k_function_value): Return a PARALLEL if the return value
2262 is of a pointer type.
2263 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
2264 (FUNCTION_EXTRA_EPILOGUE): Remove.
2265 * config/m68k/m68k.md (D0_REG): New constant.
2267 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2269 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
2270 output a NOP for empty epilogues.
2272 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2274 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
2275 before function. Extend register save check to include all
2276 registers, not just integer ones.
2278 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2280 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
2281 (m68k_use_return_insn): ...this.
2282 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
2283 * config/m68k/m68k.c (use_return_insn): Rename to...
2284 (m68k_use_return_insn): ...this.
2285 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
2288 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
2290 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
2291 nearest even, fix denormal rounding overflow.
2293 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2295 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
2296 f%$smove and f%$move.
2297 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
2298 (extendsfdf2_cf): Use fdmove instead of f%&move.
2299 (truncdfsf2_cf): Use fsmove instead of f%$smove.
2300 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
2302 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2304 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
2307 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2309 * real.h (real_format): Add a canonical_nan_lsbs_set field.
2310 (coldfire_single_format): Declare.
2311 (coldfire_double_format): Likewise.
2312 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
2313 of qnan_msb_set to determine the lower bits of a canonical
2315 (encode_ieee_double): Likewise.
2316 (encode_ieee_quad): Likewise.
2317 (ieee_single_format): Initialize canonical_nan_lsbs_set.
2318 (mips_single_format): Likewise.
2319 (ieee_double_format): Likewise.
2320 (mips_double_format): Likewise.
2321 (ieee_extended_motorola_format): Likewise.
2322 (ieee_extended_intel_96_format): Likewise.
2323 (ieee_extended_intel_128_format): Likewise.
2324 (ieee_extended_intel_96_round_53_format): Likewise.
2325 (ibm_extended_format): Likewise.
2326 (mips_extended_format): Likewise.
2327 (ieee_quad_format): Likewise.
2328 (mips_quad_format): Likewise.
2329 (vax_f_format): Likewise.
2330 (vax_d_format): Likewise.
2331 (vax_g_format): Likewise.
2332 (i370_single_format): Likewise.
2333 (i370_double_format): Likewise.
2334 (decimal_single_format): Likewise.
2335 (decimal_double_format): Likewise.
2336 (decimal_quad_format): Likewise.
2337 (c4x_single_format): Likewise.
2338 (c4x_extended_format): Likewise.
2339 (real_internal_format): Likewise.
2340 (coldfire_single_format): New real_format.
2341 (coldfire_double_format): Likewise.
2342 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
2343 canonical_nan_lsbs_set.
2344 (pdp11_d_format): Likewise.
2345 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
2346 if TARGET_COLDFIRE_CPU.
2348 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2350 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
2351 (output_call, m68k_legitimize_call_address): Declare.
2352 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
2353 parenthesees. Add support for a 'W' constraint.
2354 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
2355 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
2356 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
2357 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
2359 (override_options): Initialize them. Do not set flag_no_function_cse
2360 for TARGET_ID_SHARED_LIBRARY.
2361 (m68k_output_pic_call): Delete.
2362 (m68k_legitimize_call_address): New function.
2363 (print_operand): Remove the %o prefix. Handle the %p prefix.
2364 (output_call): New function.
2365 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
2366 target address from the GOT if symbolic jumps are not allowed.
2367 * config/m68k/m68k.md (call, general_operand): Do not set
2368 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
2369 Merge separate flag_pic and !flag_pic define_insns into...
2370 (*call, *call_value): ...these new patterns. Match the address
2371 rather than the containing MEM and require it to be a call_operand.
2372 Use output_call to generate the asm template.
2373 * config/m68k/predicates.md (const_call_operand): New predicate.
2374 (call_operand): Likewise.
2376 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
2378 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
2379 (STACK_GROWS_DOWNWARD): Define to 1.
2380 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
2381 comments, emphasizing that these values are only defaults.
2382 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
2383 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
2385 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2387 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
2389 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2390 Richard Sandiford <richard@codesourcery.com>
2392 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
2393 (FP_REGNO_P): New macros.
2394 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
2395 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
2396 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
2397 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
2399 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
2401 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
2402 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
2404 * config/m68k/t-floatlib: New file, extracting common code from...
2405 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
2406 * config/m68k/t-uclinux: Here.
2407 * config/m68k/fpgnulib.c: Do not compile extendeed precision
2408 routines on ColdFire targets.
2410 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
2412 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
2413 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
2414 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
2415 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
2416 explicitly set MOTOROLA to 1.
2417 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
2419 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2420 Nathan Sidwell <nathan@codesourcery.com>
2422 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
2423 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
2424 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
2425 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
2426 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
2427 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
2428 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
2429 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
2430 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
2431 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
2432 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
2433 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
2434 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
2435 (DBX_REGISTER_NUMBER): Delete.
2436 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
2437 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
2438 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
2440 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2442 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
2443 * config/flat.h: New file.
2444 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
2446 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
2447 indirect references for -msep-data or -mid-shared-library.
2448 Do not use PC-relative code addresses either.
2449 * config/m68k/m68k.c (override_options): Restrict -fPIC error
2451 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
2452 for shared libraries and crt1.o for executables. Use crti.o and
2454 (ENDFILE_SPEC): Use crtend.o and crtn.o.
2455 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
2456 Do not add -elf2flt or -shared-lib-id options here.
2457 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
2458 Pass -shared-lib-id if -mid-shared-library, taking the library
2459 identifier from -mshared-library-id if given, otherwise
2461 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
2462 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
2463 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
2464 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
2465 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
2466 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
2468 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
2469 jump sequence for ISA A and ISA A+.
2470 (PICJUMP): Likewise.
2472 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2474 * config.gcc (m68k-*-uclinux*): Base the port on the common
2475 and m68k GNU/Linux files rather than on the generic ELF ones.
2476 * config/m68k/uclinux.h (TARGET_VERSION): Override.
2477 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
2479 2007-01-18 Julian Brown <julian@codesourcery.com>
2480 Richard Sandiford <richard@codesourcery.com>
2482 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
2483 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2484 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
2486 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
2488 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2490 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2492 * doc/install.texi: Document m68k-uclinuxoldabi.
2493 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
2494 * config/m68k/uclinux-oldabi.h: New file, copied from
2495 config/m68k/uclinux.h.
2497 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
2499 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
2500 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
2501 (m32r_compute_frame_size): Use unsigned for regno.
2502 (m32r_reload_lr): Use gen_frame_mem.
2504 (m32r_output_function_epilogue): Don't output the function epilogue
2506 (m32r_expand_epilogue): New.
2507 (direct_return): Return false if the function has the interrupt
2509 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
2510 * config/m32r/m32r.md (epilogue): New expander.
2511 (return_lr, return_rte): New insns.
2512 (return): Make it expander.
2513 (return_normal): New expander.
2515 2007-01-18 Josh Conner <jconner@apple.com>
2518 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
2519 support for UNLE, UNLT, UNGE, and UNGT.
2521 2007-01-18 Jan Hubicka <jh@suse.cz>
2523 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
2524 * tree-tailcall (add_virtual_phis): Likewise.
2525 (optimize_tail_call): Return TODOs.
2526 (execute_tail_calls): Return TODOs.
2527 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
2528 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
2530 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
2531 Cleanup cfg using TODO.
2532 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
2533 * passes.c (execute_function_todo): When cleanup did something, remove
2535 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
2536 (make_edges): Don't cleanup_cfg.
2538 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
2540 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
2541 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
2542 operate on the top of stack.
2543 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
2544 register handling of instructions that output to the second
2546 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
2547 (move_for_stack_reg): Special-case check for dead destination
2548 stack slot for constant load of 1.0 inside UNSPEC_TAN.
2550 * config/i386/i386.md (UNSPEC_TAN): New constant.
2551 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
2552 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
2553 to correctly model move of constant 1.0 to top stack slot.
2554 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
2555 (unnamed peephole2 pattern): Remove corresponding peephole2
2556 pattern that optimizes tan insn and loading of constant 1.0.
2557 (tanxf2): Use fptanxf4_i387.
2558 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
2559 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
2560 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
2561 disable patterns for SSE math.
2562 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
2563 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
2565 2007-01-18 Dirk Mueller <dmueller@suse.de>
2566 Richard Guenther <rguenther@suse.de>
2569 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
2570 * common.opt (Warray-bounds): Add new warning option.
2571 * c-opts.c (c_common_handle_option): Define -Warray-bounds
2573 * Makefile.in: make tree-vrp.o depend on toplev.h
2574 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
2576 (check_array_refs, check_array_bounds, check_array_ref): New.
2578 2007-01-18 Jan Hubicka <jh@suse.cz>
2580 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
2581 (execute_ssa_ccp): Return flags conditionally.
2582 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
2584 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
2586 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
2588 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
2589 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
2591 * Makefile.c (GTFILES): Add cfglayout.h.
2592 * gengtype.c (open_base_files): Likewise.
2593 * cfglayout.c (cfg_layout_function_footer,
2594 cfg_layout_function_header) Reindent to make gengtype happy.
2595 * cfglayout.h (cfg_layout_function_footer,
2596 cfg_layout_function_header): Add GTY(()) marker.
2598 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
2600 2007-01-18 Ben Elliston <bje@au.ibm.com>
2602 * genautomata.c (write_automata): Include xstrerror output in the
2603 error message if writing the DFA description file fails.
2605 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
2607 * config/mips/mips-protos.h (mips_output_external): Make it
2609 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
2610 * config/mips/mips.c (irix_output_external_libcall): Likewise.
2611 (extern_list): Likewise.
2612 (extern_head): Likewise.
2613 (TARGET_ASM_FILE_END): Likewise.
2614 (mips_file_end): Likewise.
2615 (mips_output_external): Rewritten.
2617 2007-01-18 Ben Elliston <bje@au.ibm.com>
2619 * genpreds.c (write_insn_preds_c): Only write out the function
2620 body for regclass_for_constraint if we have register constraints.
2622 2007-01-17 Tom Tromey <tromey@redhat.com>
2624 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
2625 * doc/install.texi (Testing): Use sourceware.org.
2626 (Binaries): Likewise.
2627 (Specific): Likewise.
2628 * doc/contrib.texi (Contributors): Use sourceware.org.
2630 2007-01-17 Anatoly Sokolov <aesok@post.ru>
2632 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
2633 (AVR_ENHANCED): Rename to ...
2634 (AVR_HAVE_MUL): ... new.
2635 (avr_enhanced_p): Rename to ...
2636 (avr_have_mul_p): ... new.
2637 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
2638 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
2639 * config/avr/avr.c (avr_enhanced_p): Rename to ...
2640 (avr_have_mul_p): ... new.
2641 (base_arch_s): Rename 'enhanced' to 'have_mul'.
2642 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
2643 'avr_enhanced_p' and 'enhanced'.
2644 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
2645 AVR_HAVE_MUL instead of AVR_ENHANCED.
2646 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
2647 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
2648 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
2649 instead of AVR_ENHANCED.
2650 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
2651 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
2652 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
2654 2007-01-17 Ian Lance Taylor <iant@google.com>
2656 * vec.h (VEC_reserve_exact): Define.
2657 (vec_gc_p_reserve_exact): Declare.
2658 (vec_gc_o_reserve_exact): Declare.
2659 (vec_heap_p_reserve_exact): Declare.
2660 (vec_heap_o_reserve_exact): Declare.
2661 (VEC_OP (T,A,reserve_exact)): New static inline function, three
2663 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
2665 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
2666 (VEC_OP (T,A,copy)) [all versions]: Likewise.
2667 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
2668 * vec.c (calculate_allocation): Add exact parameter. Change all
2670 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
2671 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
2672 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
2673 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
2674 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
2675 (vec_heap_p_reserve_exact): New function.
2676 (vec_heap_o_reserve_exact): New function.
2678 2007-01-17 Jan Hubicka <jh@suse.cz>
2680 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
2682 2007-01-17 Eric Christopher <echristo@apple.com>
2684 * config.gcc: Support core2 processor.
2686 2007-01-16 Jan Hubicka <jh@suse.cz>
2688 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
2689 the names, just unlink the chain so we don't crash on dangling pointers
2692 2007-01-16 Jan Hubicka <jh@suse.cz>
2694 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
2695 * tree-pass.h: Reorder to make IPA passes appear toegher.
2696 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
2697 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
2698 parameters, do not call early inliner.
2699 * ipa-inline.c: Update comments. Include tree-flow.h
2700 (cgraph_decide_inlining): Do not compute inlining parameters.
2701 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
2702 be called with function context set up.
2703 (pass_ipa_inline): Remove unreachable functions before pass.
2704 (cgraph_early_inlining): Simplify assuming to be called from the
2706 (pass_early_inline): New pass.
2707 (cgraph_gate_ipa_early_inlining): New gate.
2708 (pass_ipa_early_inline): Turn into simple wrapper.
2709 (compute_inline_parameters): New function.
2710 (gate_inline_passes): New gate.
2711 (pass_inline_parameters): New pass.
2712 (apply_inline): Move here from tree-optimize.c
2713 (pass_apply_inline): New pass.
2714 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
2716 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
2718 (tree_function_versioning): Do not allocate dummy struct function.
2719 * tree-inline.h (optimize_inline_calls): Update prototype.
2720 * tree-optimize.c (execute_fixup_cfg): Export.
2721 (pass_fixup_cfg): Remove
2722 (tree_rest_of_compilation): Do not apply inlines.
2723 * tree-flow.h (execute_fixup_cfg): Declare.
2724 * Makefile.in (gt-passes.c): New.
2725 * passes.c: Include gt-passes.h
2726 (init_optimization_passes): New passes.
2727 (nnodes, order): New static vars.
2728 (do_per_function_toporder): New function.
2729 (execute_one_pass): Dump current pass here.
2730 (execute_ipa_pass_list): Don't dump current pass here.
2732 2007-01-16 Janis Johnson <janis187@us.ibm.com>
2734 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
2735 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
2737 2007-01-16 David Edelsohn <edelsohn@gnu.org>
2739 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
2740 (strong_alias): Define.
2741 (__gcc_qmul): Provide non-FMA for soft-float.
2747 (__gcc_qunord): New.
2757 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
2758 soft-float functions.
2759 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
2760 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
2761 about long double soft float.
2763 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
2764 Tehila Meyzels <tehila@il.ibm.com>
2766 * tree-vectorizer.h (is_pattern_stmt_p): New.
2767 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
2768 formatting (tabs instead of spaces). Cleanup and clarify setting
2769 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
2770 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
2771 (vectorizable_type_demotion): Check that types are integral.
2772 (vectorizable_type_promotion): Likewise.
2773 (vectorizable_store): Fix typo. Eliminate new-line at end of
2776 2007-01-16 Jan Hubicka <jh@suse.cz>
2778 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
2781 2007-01-15 Eric Christopher <echristo@apple.com>
2783 * ifcvt.c: Include vec.h, vecprim.h.
2784 (check_cond_move_block): New argument regs.
2785 Reorganize. Add registers used to regs.
2786 (cond_move_process_if_block): Use regs set above as
2789 2007-01-15 Eric Christopher <echristo@apple.com>
2791 * config/darwin.h: Update copyright.
2792 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
2793 (LINK_COMMAND_SPEC): Add -u.
2794 (LINK_SPEC): Fix umbrella for above.
2796 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
2798 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
2801 2007-01-15 Tom Tromey <tromey@redhat.com>
2803 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
2804 * doc/install.texi (Testing): Don't mention jacks.
2805 (Configuration): Document --enable-java-maintainer-mode. Move
2806 --with-java-home to libgcj-specific section. Document
2808 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
2810 2007-01-15 Jan Hubicka <jh@suse.cz>
2812 * tree-ssa-dce.c (DCE_TODOs): New.
2813 (propagate_necessity): Return if something changed.
2814 (eliminate_unnecessary_stmts): Likewise.
2815 (perform_tree_ssa_dce): Return TODO flags when needed.
2816 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
2818 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
2820 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
2821 (fyl2x_extend<mode>xf3_i387): New insn pattern.
2822 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
2823 patterns using X87MODEF12 mode macro. Extend operand 1
2824 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
2825 (log10<mode>2): Ditto.
2826 (log2<mode>2): Ditto.
2827 (log1p<mode>2): Ditto.
2828 (logb<mode>2): Ditto.
2829 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
2830 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
2831 (*fxtractxf3_i387): Rename from *fxtractxf3.
2832 (fxtract_extend<mode>xf3_i387): New insn pattern.
2833 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
2834 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
2835 and gen_fyl2xxf3_i387().
2837 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
2839 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
2840 * cfgloopmanip.c (fix_loop_placement): Made static. Use
2841 get_loop_exit_edges. Changed return type to bool.
2842 * cfgloop.h (fix_loop_placement): Declaration removed.
2844 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
2846 * param.h (MIN_VECT_LOOP_BOUND): New.
2847 * params.def (MIN_VECT_LOOP_BOUND): New.
2848 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
2849 argument - minimum threshold for number of iterations.
2850 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
2851 argument to declaration.
2852 * tree-vect-analyze.c (vect_analyze_operations): Check value of
2853 MIN_VECT_LOOP_BOUND.
2854 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
2855 slpeel_tree_peel_loop_to_edge with additional argument.
2856 (vect_do_peeling_for_alignment): Likewise.
2857 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
2859 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
2862 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
2865 2007-01-14 Jan Hubicka <jh@suse.cz>
2867 * tree-dfa.c (remove_referenced_var): New function.
2868 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
2869 prune referenced vars list too.
2870 * tree-flow.h (remove_referenced_var): Declare.
2872 2007-01-14 Jan Hubicka <jh@suse.cz>
2874 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
2876 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
2877 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
2878 * except.h (verify_eh_throw_table_statements): Kill prototype.
2879 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
2880 into call exrepssions.
2881 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
2883 * tree-cfg.c: Include pointer-set.h.
2884 (verify_node_sharing): Work on pointer set.
2885 (verify_eh_throw_stmt_node): New.
2886 (verify_stmts): Use pointers sets, verify throw_stmt.
2888 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
2890 * ipa-reference.c (analyze_function): Consider also addresses taken
2893 2007-01-12 Roger Sayle <roger@eyesopen.com>
2895 * c-typeck.c (null_pointer_constant_p): Replace use of
2896 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2897 (build_c_cast): Likewise.
2899 2007-01-12 Roger Sayle <roger@eyesopen.com>
2901 * tree.h (force_fit_type_double): Remove unused final argument.
2902 * c-common.c (constant_expression_warning): Replace use of
2903 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2904 (convert_and_check): Likewise.
2905 (shorten_compare): Update call to force_fit_type_double.
2906 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
2907 * convert.c (convert_to_pointer): Update call to
2908 force_fit_type_double.
2909 * fold-const.c (force_fit_type_double): Remove overflowed_const
2911 (int_const_binop, fold_convert_const_int_from_int,
2912 fold_convert_const_int_from_real, fold_div_compare,
2913 fold_sign_changed_comparison, fold_unary, fold_negate_const,
2914 fold_abs_const, fold_not_const): Remove the final argument from
2915 calls to force_fit_type_double.
2917 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
2919 * configure.ac: Set insn to "nop" for spu-*-* also.
2920 * configure: Regenerate.
2922 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
2924 * builtins.def : Add BUILT_IN_FREE.
2926 2007-01-12 Jan Hubicka <jh@suse.cz>
2928 PR tree-optimization/30443
2929 * tree-inline.c (tree_function_versioning): Do not optimize when
2930 cloning for inlining.
2932 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
2934 * doc/loop.texi: Document recording of loop exits.
2935 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
2936 (update_single_exits_after_duplication,
2937 update_single_exit_for_duplicated_loop,
2938 update_single_exit_for_duplicated_loops): Removed.
2939 (duplicate_loop_to_header_edge): Do not call
2940 update_single_exits_after_duplication and
2941 update_single_exit_for_duplicated_loops.
2942 (loop_version): Do not update single_exit information.
2943 (fix_loop_structure): Use record_loop_exits instead of
2944 mark_single_exit_loops.
2945 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
2946 the lists of loop exits.
2947 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
2948 split_edge, merge_blocks): Update the lists of loop exits.
2949 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
2950 loop_optimizer_init.
2951 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
2952 of mark_single_exit_loops.
2953 (loop_optimizer_finalize): Call release_recorded_exits.
2954 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
2955 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
2956 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
2957 update single exit information.
2958 * lambda-code.c (perfect_nestify): Ditto.
2959 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
2960 (mark_single_exit_loops): Removed.
2961 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
2962 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
2963 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
2965 (get_loop_exit_edges, single_exit): Use recorded exit lists.
2966 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
2967 (verify_loop_structure): Verify consistency of the exit lists.
2968 (flow_loops_find): Use alloc_loop. Initialize exits hash.
2969 (set_single_exit): Removed.
2970 * cfgloop.h (struct loop_exit): New function.
2971 (struct loop): single_exit_ field replaced by exits field.
2972 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
2973 (struct loops): Added exits hash.
2974 (mark_single_exit_loops, set_single_exit): Declaration removed.
2975 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
2977 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2979 * doc/invoke.texi: Avoid use of @headitem.
2981 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2983 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
2985 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2987 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
2990 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2991 Richard Sandiford <richard@codesourcery.com>
2993 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
2994 * config/m68k/m68k.opt: Resort options.
2995 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
2997 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2998 Richard Sandiford <richard@codesourcery.com>
3000 * doc/invoke.texi: Document the macros that are defined by
3001 m68k's -mtune and -mhard-float options.
3002 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
3003 (m68k_cpp_cpu_family): Likewise.
3004 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
3005 of __ucfv*__ macros. Define __mcffpu__ if generating code for
3006 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
3007 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
3008 (m68k_cpp_cpu_family): Likewise.
3010 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3012 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
3013 macros besides mc68000 as tuning macros. Use a switch statement
3014 to set them and mcpu32.
3016 2007-01-12 Julian Brown <julian@codesourcery.com>
3018 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
3019 (TARGET_68040_ONLY): Rename to...
3020 (TARGET_68040): ...this.
3021 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
3022 * config/m68k/m68k.md: Likewise.
3024 2007-01-12 Julian Brown <julian@codesourcery.com>
3025 Nathan Sidwell <nathan@codesourcery.com>
3026 Richard Sandiford <richard@codesourcery.com>
3028 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
3029 the -mcpu= argument associated with the --with-cpu setting.
3030 Define M68K_DEFAULT_TUNE to the default -mtune= option,
3031 if different from the one implied by the -mcpu setting.
3032 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
3033 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
3034 * doc/invoke.texi: Mention ColdFire in the introduction to the
3035 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
3036 -mno-div and -mhard-float options. Make -m68881 a synonym for
3037 -mhard-float. Document the previously-undocumented -m5206e,
3038 -m528x, -m5307 and -m5407 options. Tweak the existing option
3039 documentation for consistency.
3040 * doc/install.texi: Mention new --with-cpu arguments.
3041 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
3042 default CPU if neither -mcpu nor -march are specified.
3043 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
3044 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
3045 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
3046 using m68k_tune to decide between families that implement the
3047 same ISA. Use m68k_tune to set __mcfv4e__.
3048 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
3049 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
3050 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
3051 (FL_ISA_C, FL_ISA_MMU): New macros.
3052 (MASK_COLDFIRE): Delete.
3053 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
3054 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
3055 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
3056 (TARGET_HARD_FLOAT): Do not define here.
3057 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
3058 (TUNE_68000): New macro.
3059 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
3060 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
3061 (TUNE_CFV2): Redefine in terms of m68k_tune.
3062 (uarch_type, target_device, fpu_type): New enums.
3063 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
3064 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
3065 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
3066 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
3067 (FL_FOR_isa_c): New macros.
3068 (m68k_isa): New enum.
3069 (m68k_target_selection): New structure.
3070 (all_devices, all_isas, all_microarchs): New tables.
3071 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
3072 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
3073 (MASK_ALL_CPU_BITS): Delete.
3074 (m68k_find_selection): New function.
3075 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
3076 Map the legacy target options to a combination of the new ones.
3077 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
3078 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
3079 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
3081 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
3082 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
3084 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
3085 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
3086 * config/m68k/m68k-devices.def: New file.
3088 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3089 Nathan Sidwell <nathan@codesourcery.com>
3091 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
3092 (SUBTARGET_EXTRA_SPECS): New macros.
3093 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
3094 use %(asm_cpu_spec) instead.
3095 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
3096 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
3097 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
3098 (EXTRA_SPECS): Rename to...
3099 (SUBTARGET_EXTRA_SPECS): ...this.
3101 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
3102 Richard Sandiford <richard@codesourcery.com>
3103 Julian Brown <julian@codesourcery.com>
3105 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
3106 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
3107 to the configuration's default CPU.
3108 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
3109 Remove default masks.
3110 (m680[012]0-*-*): Set the default with_cpu to the first part of
3112 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
3113 (m68k*-*-linux): Extend the --with-cpu handling to...
3114 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
3115 and m68010. Don't set target_cpu_default2.
3116 * doc/install.texi: Document --with-cpu for m68k.
3117 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
3118 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
3119 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
3120 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
3121 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
3122 (ASM_SPEC): Remove use of %(asm_cpu_default).
3123 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
3124 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
3125 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
3126 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
3127 __HAVE_FPU__ if TARGET_HARD_FLOAT.
3128 (TARGET_DEFAULT): Delete.
3129 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
3130 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
3131 (CPP_CPU_SPEC): Delete.
3132 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
3133 the appropriate string.
3134 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
3135 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
3136 (ASM_SPEC): Don't use %(asm_default_spec).
3137 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
3138 TARGET_DEFAULT and add MASK_68881.
3139 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
3141 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3143 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
3144 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
3145 MASK_68010 alongside MASK_68020.
3146 * doc/invoke.texi: Document -m68010.
3147 * config/m68k/m68k.opt (m68010): New.
3148 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
3150 (TUNE_68010): New macro.
3151 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
3152 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
3153 (M68K_CPU_m68332): Add MASK_68010.
3154 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
3155 fallback definition.
3156 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
3158 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
3159 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
3160 to all entries that use MASK_68020.
3161 (output_move_simode_const, output_move_himode, output_move_qimode)
3162 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
3163 instead of TARGET_68020 to select clr behavior. Remove comment
3164 about there being no TARGET_68010.
3165 * config/m68k/m68k.md: Likewise throughout.
3167 2007-01-12 Julian Brown <julian@codesourcery.com>
3169 * config/m68k/m68k.h (TARGET_ISAB): New macro.
3170 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
3171 * config/m68k/m68k.md: Likewise.
3173 2007-01-12 Julian Brown <julian@codesourcery.com>
3175 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
3176 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
3178 2007-01-12 Julian Brown <julian@codesourcery.com>
3180 * config/m68k/m68k.h (TUNE_68040_60): New macro.
3181 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
3182 * config/m68k/m68k.md: Likewise.
3184 2007-01-12 Julian Brown <julian@codesourcery.com>
3185 Richard Sandiford <richard@codesourcery.com>
3187 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
3188 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
3189 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
3191 (TARGET_CPU32): Rename to...
3192 (TUNE_CPU32): ...this.
3193 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
3194 (TUNE_CFV2): New macros.
3195 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
3196 remove conditions that are implied by TARGET_68020.
3197 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
3198 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
3199 (m68k_output_function_epilogue): Likewise.
3200 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
3201 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
3202 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
3203 non-68000 timings. Refactor multiplication and division costs.
3204 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
3205 TUNE_CPU32 instead of TARGET_CPU32.
3206 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
3207 and TUNE_68060 instead of TARGET_68060.
3208 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
3209 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
3211 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
3212 && !TARGET_COLDFIRE" to choose between moveq and clr.
3213 Likewise in the unnamed movsf pattern.
3214 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
3215 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
3218 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3220 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
3221 of tabbing before backslashes.
3223 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3225 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
3226 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
3227 (ASM_OUTPUT_INTERNAL_LABEL): Define.
3229 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
3231 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
3232 find_interesting_uses_cond.
3233 (find_interesting_uses_cond): Use extract_cond_operands.
3234 (rewrite_use_compare): Use extract_cond_operands and
3235 force_gimple_operand_bsi. Do not call update_stmt.
3236 (determine_use_iv_cost_condition): Use extract_cond_operands.
3237 Return cheaper of using original bound and changing the exit bound.
3239 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
3241 PR tree-optimization/29516
3242 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
3243 most_expensive_mult_to_index, addr_to_parts,
3244 create_mem_ref, maybe_fold_tmr): Make the type of
3245 fields of TARGET_MEM_REF sizetype.
3246 (move_fixed_address_to_symbol, move_pointer_to_base):
3248 * tree.def (TARGET_MEM_REF): Add comment on types of
3251 2007-01-11 Joseph Myers <joseph@codesourcery.com>
3253 * c-common.c (vector_types_convertible_p): Treat opaque types as
3254 always convertible if they have the same size, but not otherwise.
3256 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
3258 * ifcvt.c (struct noce_if_info): Add comments to the fields.
3259 Remove the b_unconditional field.
3260 (noce_try_sign_mask): Do not look at b_unconditional.
3261 (noce_process_if_block): Do not use merge_if_blocks. Update
3262 the CFG here. Do not set b_unconditional.
3263 (cond_move_process_if_block): Likewise.
3264 (find_cond_trap): Likewise.
3265 (check_cond_move_block): Require simple jump insns at the end
3268 2007-01-11 Jan Hubicka <jh@suse.cz>
3270 PR tree-optimization/1046
3271 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
3272 when alias info is not ready.
3273 (pass_tail_recursion): Do not require aliasing.
3274 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
3275 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
3276 * tree-ssa-copy.c (pass_copy_prop): Likewise.
3277 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
3278 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
3279 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
3280 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
3282 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
3283 When aliasing is not build, mark statement as volatile.
3285 2007-01-11 Tom Tromey <tromey@redhat.com>
3287 PR preprocessor/15185, PR preprocessor/20989:
3288 * doc/cppopts.texi <-MT>: Update description of algorithm for
3289 computing default target.
3290 <-M, -MD>: Reword "basename" text.
3292 2007-01-11 Roger Sayle <roger@eyesopen.com>
3294 * builtins.c (expand_builtin_pow, expand_builtin_powi,
3295 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
3296 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
3297 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
3298 fold_builtin_bswap, real_constp, fold_builtin_pow,
3299 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
3300 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
3301 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
3302 * convert.c (convert_to_pointer): Likewise.
3303 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
3304 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
3305 const_binop, fold_convert_const_int_from_int,
3306 fold_convert_const_int_from_real,
3307 fold_convert_const_real_from_real, sign_bit_p,
3308 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
3309 fold_sign_changed_comparison, fold_unary, fold_comparison,
3310 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
3311 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
3312 * print-tree.c (print_node_brief, print_node): Likewise.
3313 * stor-layout.c (place_field, layout_type): Likewise.
3314 * tree-chrec.c (keep_cast): Likewise.
3315 * tree.c (build_vector, build_real, build_real_from_int_cst,
3316 build_complex): Likewise.
3318 2007-01-11 Roger Sayle <roger@eyesopen.com>
3320 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
3321 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
3323 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
3325 * configure.ac (strict1_warn): Rename to strict_warn.
3326 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
3327 cc_set_by_configure, quoted_cc_set_by_configure,
3328 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
3329 all_boot_languages, all_stagestuff): Remove.
3330 (target_list): Remove bootstrap targets.
3331 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
3332 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
3333 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
3334 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
3335 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
3336 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
3337 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
3338 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
3339 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
3340 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
3341 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
3342 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
3343 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
3344 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
3345 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
3346 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
3347 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
3348 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
3349 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
3350 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
3351 (mostlyclean): Adjust.
3352 (clean, distclean): Don't mention bootstrap stuff.
3353 * configure: Regenerate.
3354 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
3355 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
3356 treelang/config-lang.in (stagestuff): Remove.
3357 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
3358 stageprofile, stagefeedback, stagestuff): Remove mention.
3360 2007-01-11 Nick Clifton <nickc@redhat.com>
3362 * config/mcore/predicates.md (mcore_general_movesrc_operand):
3364 (mcore_general_movdst_operand): Do not accept CONST_INTs.
3365 (mcore_arith_K_S_operand): Run the test for the S constraint not
3366 the test for the M constraint.
3367 (mcore_addsub_operand): Do not accept integer values that are
3368 larger than 32 bits.
3369 * config/mcore/mcore.md: Remove unused constraints from split.
3370 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
3372 (allocate_stack): Likewise.
3373 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
3374 operands to 32 bits.
3375 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
3377 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
3378 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
3379 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
3380 mcore_output_bclri, mcore_output_andn, output_inline_const,
3381 mcore_output_move, mcore_output_movedouble): Likewise.
3382 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
3383 (output_inline_const): Likewise.
3384 (output_inline_const): Fix format strings used in sprintf
3386 * config/mcore/mcore-protos.h: Update prototypes for changed
3387 functions in mcore.c.
3388 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
3389 HOST_WIDE_INT and not int.
3390 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
3391 CONST_OK_FOR_N): Likewise.
3392 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
3393 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
3396 2007-01-10 Jan Hubicka <jh@suse.cz>
3398 * tree-vrp.c (remove_range_assertions): Release defs.
3399 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
3400 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
3402 2007-01-10 Paul Brook <paul@codesourcery.com>
3404 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
3405 * config/arm/arm.md (mulsi3subsi): New insn.
3407 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
3409 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
3410 (tree_transform_and_unroll_loop): New.
3411 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
3414 2007-01-10 Robert Kennedy <jimbob@google.com>
3416 * fold-const.c (fold_comparison): Fold comparisons like (x *
3417 1000 < 0) to (x < 0).
3419 2007-01-10 Ian Lance Taylor <iant@google.com>
3421 * tree-pretty-print.c (dump_generic_node): Print parentheses when
3422 operands have the same priority.
3424 2007-01-10 Tom Tromey <tromey@redhat.com>
3426 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
3427 (fold_binary): Likewise.
3428 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
3429 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
3430 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
3431 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
3432 * langhooks.h (struct lang_hooks): Removed field
3433 'can_use_bit_fields_p'.
3435 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
3437 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
3439 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
3441 * function.c (get_last_funcdef_no): New function.
3442 * function.h (get_last_funcdef_no): Declare.
3443 * tree-inline.c (initialize_cfun): Add initialization.
3444 (tree_function_versioning): Cleanup.
3446 2007-01-10 Jan Hubicka <jh@suse.cz>
3448 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
3451 2007-01-10 Sa Liu <saliu@de.ibm.com>
3452 Ben Elliston <bje@au.ibm.com>
3454 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
3455 for SAVE_NONLOCAL and Pmode for any other save level.
3456 * spu-protos.h (spu_restore_stack_block): Declare.
3457 * spu.md (save_stack_block): Remove.
3458 (restore_stack_block): Call spu_restore_stack_block.
3459 * spu.c (spu_restore_stack_block): New function.
3460 (spu_expand_epilogue): Remove old comment.
3462 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
3464 PR tree-optimization/30322
3465 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
3466 (cand_value_at): Return the value as aff_tree.
3467 (may_eliminate_iv): Convert the bound from aff_tree to tree.
3468 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
3469 aff_combination_mult): New functions.
3470 (aff_combination_add): Use aff_combination_add_cst.
3471 (aff_combination_convert): Allow conversions to a wider type.
3472 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
3473 * tree-affine.h (aff_combination_mult): Declare.
3475 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
3477 * doc/tm.texi: Update documentation to reflect reality of exec
3478 and start file search behaviours. Update copyright year.
3479 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
3480 header file directories.
3482 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
3484 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
3485 (*cosxf2): Rename to cosxf2_i387.
3486 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
3487 using X87MODEF12 mode macro. Rename patterns to
3488 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
3490 (*cosdf2, *cossf2): Ditto.
3491 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
3492 to match extended input operands.
3493 (sincos<mode>3): New expander.
3494 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
3495 insn patterns and corresponding splitters.
3497 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
3499 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
3500 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
3502 2007-01-09 Nicolas Pitre <nico@cam.org>
3505 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
3507 2007-01-08 Geoffrey Keating <geoffk@apple.com>
3509 * target.h (struct gcc_target): New field library_rtti_comdat.
3510 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
3511 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
3512 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
3513 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
3515 2007-01-08 Geoffrey Keating <geoffk@apple.com>
3517 * doc/invoke.texi (Optimize Options): Correct description of -O0.
3519 2007-01-08 Richard Guenther <rguenther@suse.de>
3521 * tree.h (force_fit_type_double): Export.
3522 (force_fit_type): Remove.
3523 * fold-const.c (force_fit_type_double): New function.
3524 (force_fit_type): Remove.
3525 (int_const_binop): Use it.
3526 (fold_convert_const_int_from_int): Likewise.
3527 (fold_convert_const_int_from_real): Likewise.
3528 (fold_div_compare): Likewise.
3529 (fold_sign_changed_comparison): Likewise.
3530 (fold_unary): Likewise.
3531 (fold_negate_const): Likewise.
3532 (fold_abs_const): Likewise.
3533 (fold_not_const): Likewise.
3534 * c-common.c (shorten_compare): Use force_fit_type_double.
3535 * convert.c (convert_to_pointer): Likewise.
3537 2007-01-08 Richard Guenther <rguenther@suse.de>
3539 * tree.h (build_int_cst_wide_type): Export.
3540 * tree.c (build_int_cst_wide_type): New function.
3541 (build_int_cst_wide): Fix comment.
3542 * builtins.c (fold_builtin_object_size): Use build_int_cst
3543 to build -1 or 0 of the correct type. Use fit_double_type
3544 to check for overflow.
3545 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
3547 (decode_field_reference): Likewise.
3548 (all_ones_mask_p): Likewise.
3549 (native_interpret_int): Use build_int_cst_wide_type.
3550 (fold_binary): Use build_int_cst_type to build an all-ones
3552 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
3554 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
3556 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
3558 2007-01-09 Ben Elliston <bje@au.ibm.com>
3560 * genautomata.c (STATS_OPTION): New option.
3561 (stats_flag): New flag.
3562 (gen_automata_option): Handle it.
3563 (initiate_automaton_gen): Ditto.
3564 (write_automata): Output statistics only if stats_flag is
3565 set. Likewise, output time statistics only if time_flag is set.
3566 * doc/md.texi (Processor pipeline description): Document new flag.
3568 2007-01-08 Richard Guenther <rguenther@suse.de>
3570 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
3571 * tree.c (build_int_cst_type): Likewise.
3572 (size_in_bytes): Don't call force_fit_type on the result.
3573 (int_fits_type_p): Use fit_double_type.
3574 * fold-const.c (fit_double_type): New function.
3575 (force_fit_type): Use it.
3576 * tree.h (fit_double_type): Export.
3578 2007-01-08 Jan Hubicka <jh@suse.cz>
3580 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
3581 * ipa.c (function_and_variable_visibility): Fix return type.
3583 2007-01-08 Richard Guenther <rguenther@suse.de>
3585 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
3586 of offset to build the index.
3587 * tree-pretty-print.c (dump_generic_node): Don't build negated
3588 const just for printing.
3589 * c-pretty-print.c (pp_c_integer_constant): Likewise.
3590 * builtins.c (fold_builtin_int_roundingfn): Check if result
3591 fits the type by using force_fit_type and comparing the result.
3592 * predict.c (predict_loops): Use compare_tree_int for comparison.
3593 * tree.c (build_int_cst): Fall back to integer_type_node for
3595 (build_int_cst_wide): Assert type is non-null.
3597 2007-01-08 Roberto Costa <roberto.costa@st.com>
3599 * tree-vrp.c (extract_range_from_cond_expr): New.
3600 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
3601 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
3603 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
3604 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
3605 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
3606 used as expressions.
3607 * tree-object-size.c (cond_expr_object_size): New.
3608 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
3610 2007-01-08 Jan Hubicka <jh@suse.cz>
3612 * tree-ssa-forwprop.c (forward_propagate_into_cond,
3613 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
3616 2007-01-08 Richard Guenther <rguenther@suse.de>
3618 PR tree-optimization/23603
3619 * tree-vrp.c (set_value_range_to_truthvalue): New function.
3620 (extract_range_from_binary): Fall back to truthvalue instead of
3621 varying for TRUTH_*_EXPR.
3622 (extract_range_from_comparison): Fall back to truthvalue instead of
3624 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
3625 if all visited PHI values were constant.
3627 2007-01-08 Jan Hubicka <jh@suse.cz>
3629 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
3630 (cgraph_analyze_function): break out from ...
3631 (cgraph_finalize_compilation_unit): ... here.
3632 (cgraph_expand_function): Remove forgoten commented out line.
3633 (cgraph_optimize): Analyze functions.
3635 2007-01-08 Jan Hubicka <jh@suse.cz>
3637 * tree-pas.h (TODO_remove_function): New flag.
3638 (TODO_update*): Renumber.
3639 (pass_ipa_increase_alignment,
3640 pass_ipa_function_and_variable_visibility): New passes.
3641 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
3642 (cgraph_function_and_variable_visibility): Move to ipa.c
3643 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
3644 cgraph_increase_alignment.
3645 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
3646 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
3647 calling tree-inline.
3648 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
3649 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
3650 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
3651 (gate_increase_alignment): New function.
3652 (pass_ipa_increase_alignment): New pass.
3653 * ipa.c: Inline tree-pass.h and timevar.h
3654 (function_and_variable_visibility): Move here from cgraphunit.c
3655 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
3656 * passes.c (init_optimization_passes): Add the two new passes.
3657 (execute_todo): Handle cgraph_remove_functions.
3659 2007-01-08 Nick Clifton <nickc@redhat.com>
3661 * config/frv/predicates.md (reg_or_0_operand): Accept
3664 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
3666 * config/bfin/rtems.h, config/bfin/t-rtems: New.
3667 * config.gcc: Add bfin*-rtems*.
3669 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3671 * c.opt: Add -flax-vector-conversions.
3672 * c-typeck.c (convert_for_assignment): Pass flag to
3673 vector_types_convertible_p to allow emission of note.
3674 (digest_init): Likewise.
3675 * c-opts.c: Handle -flax-vector-conversions.
3676 * c-common.c (flag_lax_vector_conversions): New.
3677 (vector_types_convertible_p): Unless -flax-vector conversions
3678 has been passed, disallow conversions between vectors with
3679 differing numbers of subparts and/or element types. If such
3680 a conversion is disallowed, possibly emit a note on the first
3681 occasion only to inform the user of -flax-vector-conversions.
3682 The new last argument specifies this.
3683 * c-common.h (flag_lax_vector_conversions): New.
3684 (vector_types_convertible_p): Add extra argument.
3685 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
3686 char_type_node for V*QI type vectors.
3687 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3688 Update to satisfy new typechecking rules.
3689 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
3691 * doc/invoke.texi (C Dialect Options): Document
3692 -flax-vector-conversions.
3694 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3696 PR tree-optimization/29877
3697 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
3698 a register variable on the RHS to not be replaceable.
3700 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
3701 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
3702 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
3703 const_simm15, const_pow2, const_npow2): Added.
3704 * config/score/misc.md (insv, extv, extzv, movmemsi,
3705 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
3706 * config/score/score.c (score_address_cost, score_select_cc_mode):
3708 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
3709 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
3710 Update score_print_operand.
3711 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
3712 Adjust register allocate order and update some macro define.
3713 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
3714 mdx_block_move_straight, mdx_block_move_loop_head,
3715 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
3716 mdx_block_move): Added.
3717 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
3718 mdp_limm): Updated and fix some bug and typo.
3719 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
3720 (movsf, movdf, doloop_end): Added.
3722 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
3724 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
3725 config/arm/thumb2.md: Fix comment typos.
3726 * doc/extend.texi: Fix a typo.
3728 2007-01-07 Eric Christopher <echristo@apple.com>
3730 * configure.ac: Check for __stack_chk_fail for darwin.
3731 * configure: Regenerate.
3733 2007-01-07 Richard Guenther <rguenther@suse.de>
3735 * tree-vrp.c (extract_range_from_assert): CSE calls to
3736 compare_values where possible.
3737 (extract_range_from_unary_expr): Likewise.
3739 2007-01-07 Anatoly Sokolov <aesok@post.ru>
3741 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
3742 * config/avr/avr.c (call_insn_operand): Delete function.
3743 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
3744 instead of register number. Use predicates.
3745 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
3746 call_insn_operand): Add.
3748 2007-01-06 Jan Hubicka <jh@suse.cz>
3750 * tree-pass.h (pass_build_cgraph_edges): Declare.
3751 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
3752 (visited_nodes): Remove.
3753 (cgraph_create_edges): Move to cgraphbuild.c; rename to
3754 build_cgrpah_edges; make visited_nodes local.
3755 (cgraph_process_new_functions): DO not call initialize_inline_failed.
3756 (record_references_in_initializer): Move to cgraphbuild.c
3757 (initialize_inline_failed, rebuild_cgraph_edges,
3758 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
3759 (verify_cgraph_node): Make visited_nodes local.
3760 (cgraph_analyze_function): Do not call cgraph_create_edges and
3761 initialize_inline_failed.
3762 (cgraph_expand_function): Do not call cgraph_lower_function;
3763 assert that function is already lowered.
3764 * Makefile.in (cgraphbuild.o): New.
3765 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
3766 at the end of lowering passes.
3768 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
3770 * ifcvt.c (cond_move_convert_if_block): New function, code
3771 factored out from...
3772 (cond_move_process_if_block): ...here. Call the new function
3773 on the THEN and ELSE blocks.
3774 (merge_if_block): Do not copy global_live_at_end, merge_blocks
3775 already takes care of this.
3777 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3780 * tree.h (TREE_OVERFLOW_P): New.
3781 * c-typeck.c (parser_build_unary_op): Warn only if result
3782 overflowed and operands did not.
3783 (parser_build_binary_op): Likewise.
3784 (convert_for_assignment): Remove redundant overflow_warning.
3785 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
3787 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3789 * c-typeck.c (store_init_value): Split over two lines to follow
3790 the GNU coding style.
3792 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
3794 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
3795 __GXX_EXPERIMENTAL_CXX0X__.
3796 * doc/cpp.texi: Same.
3798 2007-01-05 Richard Guenther <rguenther@suse.de>
3801 * tree.c (get_narrower): Do not construct COMPONENT_REFs
3802 with mismatched types. Instead explicitly build a
3803 conversion NOP_EXPR.
3805 2007-01-05 Ian Lance Taylor <iant@google.com>
3807 * c-common.c (decl_with_nonnull_addr_p): New function.
3808 (c_common_truthvalue_conversion): Call it.
3809 * c-typeck.c (build_binary_op): Likewise.
3810 * c-common.h (decl_with_nonnull_addr_p): Declare.
3812 2007-01-05 Jakub Jelinek <jakub@redhat.com>
3815 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
3818 2007-01-05 Joel Brobecker <brobecker@adacore.com>
3820 * doc/install.texi (Final install): Document the fact that
3821 the GNAT runtime should not be stripped.
3823 2007-01-04 Jan Hubicka <jh@suse.cz>
3825 * tree-inline.c (fold_marked_statements): Update operand caches
3826 and EH after folding
3828 2007-01-04 Ian Lance Taylor <iant@google.com>
3830 * c-common.c (check_function_nonnull): Whitespace fix.
3832 2007-01-04 Jan Hubicka <jh@suse.cz>
3834 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
3835 comitted older version of patch.
3836 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
3837 function, celanup cfg and collect garbage.
3839 2007-01-04 Mike Stump <mrs@apple.com>
3841 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
3844 2007-01-04 Eric Christopher <echristo@apple.com>
3846 * libgcc2.c (__bswapsi2): Use SItype.
3847 (__bswapdi2): Use DItype.
3848 * libgcc2.h: Update for above.
3850 2007-01-04 Paul Brook <paul@codesourcery.com>
3852 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
3853 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
3854 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
3855 umulsidi3, umulsidi3adddi, smulsi3_highpart,
3856 umulsi3_highpart): Make conditional on !arm_arch6.
3857 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
3858 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
3859 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
3860 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
3861 umulsi3_highpart_v6): New insns.
3863 2007-01-04 Roger Sayle <roger@eyesopen.com>
3865 * fold-const.c (fold_convert): When casting an expression to void,
3866 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
3867 have a type. Instead of attempting to build a NOP_EXPR, return
3868 these "special" trees directly.
3870 2007-01-04 Joseph Myers <joseph@codesourcery.com>
3872 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
3873 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
3874 rs6000_cost->fp not 0 as adjustment for outer NEG.
3876 2007-01-04 Jan Hubicka <jh@suse.cz>
3878 * cgraph.c (cgraph_release_function_body): New function.
3879 (cgraph_remove_node): Use it.
3880 * cgraph.h (cgraph_release_function_body): Declare.
3881 * cgraphunit.c (cgraph_expand_function): Use it.
3882 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
3883 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
3884 is initialized and while compilation of other function is running.
3885 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
3886 statement CFG annotations from here to ...
3887 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
3889 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
3891 * cfgloop.h (enum li_flags): Make the constants powers of two.
3893 2007-01-04 Jan Hubicka <jh@suse.cz>
3895 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
3897 (fold_marked_statements): New function.
3898 (optimize_inline_calls, tree_function_versioning): Fold new statements.
3899 * tree-inline.h (copy_body_data): Add statements_to_fold.
3901 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
3903 * config.gcc: Mention libgcc/config.host.
3904 * Makefile.in: Update comments mentioning libgcc.
3905 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
3906 (all.cross, start.encap, rest.encap, rest.cross): Update
3907 dependencies for libgcc move.
3908 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
3909 (clean-target, clean-target-libgcc): Delete.
3910 (srcdirify, GCC_EXTRA_PARTS): New macros.
3911 (libgcc-support, libgcc.mvars): New rules.
3912 (distclean): Remove mention of mklibgcc.
3913 (install): Don't reference INSTALL_LIBGCC.
3914 (install-common): Don't reference EXTRA_PARTS.
3915 (install-libgcc, install-multilib): Delete rules.
3916 * mklibgcc.in: Delete file.
3917 * doc/configfiles.texi: Don't mention mklibgcc.
3919 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
3920 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
3921 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
3922 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
3923 (mklibgcc, ldblspecs): Likewise.
3925 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
3926 (posixpre.def): Use $(T).
3927 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
3928 (SHLIB_SRC, SHLIB_INSTALL): Delete.
3929 (SHLIB_LINK): Make dummy.
3930 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
3932 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
3934 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
3935 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
3936 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
3937 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
3939 * configure.ac (all_outputs): Remove mklibgcc.
3940 * configure: Regenerated.
3942 2007-01-03 Josh Conner <jconner@apple.com>
3945 * calls.c (compute_argument_addresses): Set stack and stack_slot
3946 for partial args, too.
3947 (store_one_arg): Use locate.size.constant for the size when
3948 generating a save_area.
3950 2007-01-03 Robert Kennedy <jimbob@google.com>
3952 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
3953 whose definitions are deleted due to basic block merging.
3955 2007-01-03 Paul Brook <paul@codesourcery.com>
3958 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
3960 (use_return_insn): Return 0 for Thumb interrupt functions.
3961 (print_multi_reg): Add rfe argument for IRQ returns.
3962 (arm_output_epilogue): Pop interrupt return address directly into PC.
3963 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
3965 2007-01-03 Paul Brook <paul@codesourcery.com>
3967 Merge from sourcerygxx-4_1.
3968 * config/arm/thumb2.md: New file.
3969 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
3971 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
3972 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3973 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
3974 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
3976 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3977 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
3979 * config/arm/ieee754-sf.S: Ditto.
3980 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
3981 (thumb1_base_register_rtx_p): ... to this.
3982 (thumb_index_register_rtx_p): Rename...
3983 (thumb1_index_register_rtx_p): ... to this.
3984 (thumb_output_function_prologue): Rename...
3985 (thumb1_output_function_prologue): ... to this.
3986 (thumb_legitimate_address_p): Rename...
3987 (thumb1_legitimate_address_p): ... to this.
3988 (thumb_rtx_costs): Rename...
3989 (thumb1_rtx_costs): ... to this.
3990 (thumb_compute_save_reg_mask): Rename...
3991 (thumb1_compute_save_reg_mask): ... to this.
3992 (thumb_final_prescan_insn): Rename...
3993 (thumb1_final_prescan_insn): ... to this.
3994 (thumb_expand_epilogue): Rename...
3995 (thumb1_expand_epilogue): ... to this.
3996 (arm_unwind_emit_stm): Rename...
3997 (arm_unwind_emit_sequence): ... to this.
3998 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
3999 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
4000 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
4001 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
4002 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
4003 thumb2_output_casesi): New functions.
4004 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
4005 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
4006 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
4007 THUMB2_WORK_REGS): Define.
4008 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
4009 arm_condexec_mask, arm_condexec_masklen)): New variables.
4010 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
4011 (arm_override_options): Check new CPU capabilities.
4012 Set new architecture flag variables.
4013 (arm_isr_value): Handle v7m interrupt functions.
4014 (user_return_insn): Return 0 for v7m interrupt functions. Handle
4016 (const_ok_for_arm): Handle Thumb-2 constants.
4017 (arm_gen_constant): Ditto. Use movw when available.
4018 (arm_function_ok_for_sibcall): Return false for v7m interrupt
4020 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
4021 (thumb_find_work_register, arm_load_pic_register,
4022 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
4023 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
4024 print_multi_reg, output_mov_long_double_fpa_from_arm,
4025 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
4026 output_mov_double_fpa_from_arm, output_move_double,
4027 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
4028 output_return_instruction, arm_output_function_prologue,
4029 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
4030 arm_output_mi_thunk, thumb_set_return_address): Ditto.
4031 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
4032 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
4033 (arithmetic_instr, shift_op): Use arm_shift_nmem.
4034 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
4036 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
4037 (thumb_expand_prologue): Use thumb_set_frame_pointer.
4038 (arm_file_start): Output directive for unified syntax.
4039 (arm_unwind_emit_set): Handle stack alignment instruction.
4040 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
4041 Add v6t2, v7, v7a, v7r and v7m.
4042 (RETLDM): Add Thumb-2 code.
4043 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
4044 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
4045 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
4046 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
4047 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
4048 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
4049 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
4050 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
4051 ADJUST_INSN_LENGTH): Define.
4052 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
4053 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
4054 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
4055 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
4056 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
4057 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
4058 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
4059 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
4060 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
4061 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
4062 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
4063 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
4064 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
4065 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
4066 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
4067 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
4068 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
4070 * config/arm/arm-tune.md: Regenerate.
4071 * config/arm/arm-protos.h: Update prototypes.
4072 * config/arm/vfp.md: Enable patterns for Thumb-2.
4073 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
4074 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
4075 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
4076 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
4077 * config/arm/libunwind.S: Add Thumb-2 code.
4078 * config/arm/constraints.md: Update include Thumb-2.
4079 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
4080 * config/arm/ieee754-df.S: Ditto.
4081 * config/arm/bpabi.S: Ditto.
4082 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
4083 * config/arm/predicates.md (low_register_operand,
4084 low_reg_or_int_operand, thumb_16bit_operator): New.
4085 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
4086 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
4087 * config/arm/t-arm-elf: Add armv7 multilib.
4088 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
4090 (UNSPEC_STACK_ALIGN, ce_count): New.
4091 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
4092 arm_zero_extendsidi2, arm_zero_extendqidi2): New
4094 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
4095 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
4096 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
4097 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
4098 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
4099 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
4101 * doc/extend.texi: Document ARMv7-M interrupt functions.
4102 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
4104 2007-01-03 Jakub Jelinek <jakub@redhat.com>
4106 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
4107 (struct _Unwind_Context): Rename args_size to flags, remove
4108 signal_frame field, add a new args_size field and version field.
4109 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
4110 _Unwind_IsExtendedContext): New inline functions.
4111 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
4112 Assume by_value array is only present if _Unwind_IsExtendedContext.
4113 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
4114 _Unwind_IsSignalFrame.
4115 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
4116 (uw_update_context_1): Use _Unwind_SetSignalFrame.
4117 (uw_init_context_1): Initialize context->flags to
4118 EXTENDED_CONTEXT_BIT.
4119 * config/rs6000/linux-unwind.h (frob_update_context): Use
4120 _Unwind_SetSignalFrame.
4122 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
4125 * gimplify.c (gimplify_modify_expr_complex_part): Move below
4126 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
4129 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
4131 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
4132 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
4133 RET, AT_SP): Remove.
4134 * config/i386/i386.md (*sse_prologue_save_insn): Use return
4137 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4138 * cfgbuild.c (find_basic_blocks): Likewise.
4139 * cfgrtl.c (rtl_create_basic_block): Likewise.
4140 * function.c (temp_slots_at_level): Likewise.
4141 * reg-stack.c (stack_regs_mentioned): Likewise.
4142 * regclass.c (allocate_reg_info): Likewise.
4143 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4144 set_bb_for_stmt, move_block_to_fn): Likewise.
4145 * tree-complex.c (tree_lower_complex): Likewise.
4146 * vec.h (VEC_safe_grow_cleared): New.
4148 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
4150 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
4152 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
4153 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
4155 * cfgloopmanip.c (scale_loop_frequencies): Export.
4156 (loopify): Scale the frequencies by prescribed coefficients.
4157 (set_zero_probability): New function.
4158 (duplicate_loop_to_header_edge): Improve updating of frequencies.
4159 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
4160 and frequencies according to arguments.
4161 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
4163 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
4164 * modulo-sched.c (sms_schedule): Set probabilities for entering
4165 versioned loop correctly.
4166 * tree-vect-transform.c (vect_transform_loop): Ditto.
4167 * cfgloop.h (loopify, loop_version): Declaration changed.
4168 (scale_loop_frequencies): Declared.
4170 2007-01-02 Jan Hubicka <jh@suse.cz>
4172 * cgraph.c: Include tree-flow.h
4173 (cgraph_add_new-function): Handle IPA_SSA mode; execute
4175 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
4176 * tree-pass.h (pass_all_early_optimizations): Declare.
4177 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
4179 (cgraph_analyze_function): Do early_local_passes.
4180 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
4181 Do not add referenced vars.
4182 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
4184 (gate_all_early_local_passes): New gate.
4185 (pass_early_local_passes): Use new gate.
4186 (execute_early_local_optimizations): New functions.
4187 (gate_all_early_optimizations): New gate.
4188 (pass_all_early_optimizations): New pass.
4189 (execute_free_datastructures): Free SSA only when initialized.
4190 (gate_init_datastructures): Init only when optimizing.
4191 (tree_lowering_passes): Do early local passes when called late.
4192 * tree-profile.c (do_tree_profiling): Don't profile functions added
4194 (do_early_tree_profiling, pass_early_tree_profile): Kill.
4195 * tree-cfg.c (update_modified_stmts): Do not update when operands are
4197 * passes.c (init_optimizations_passes): Reorder so we go into SSA
4198 during early_local_passes.
4199 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
4202 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
4204 * Makefile.in: Update copyright year.
4206 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
4208 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
4209 in $(lang_checks) and check-consistency targets.
4211 2007-01-02 Jan Hubicka <jh@suse.cz>
4213 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
4214 Do not add referenced vars.
4215 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
4217 * passes.c (init_optimization_passes): Put mudflap_2 after
4218 free_datastructures.
4220 2007-01-02 Jan Hubicka <jh@suse.cz>
4222 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
4223 Set NOTHROW flag on call statements proved to be nothrow.
4224 Update statement of local calls so new pure/const functions are
4225 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
4227 (tree_rest_of_compilation): Register hooks and initialize bitmap
4228 early. Do not set after_inlining flag.
4230 2007-01-02 Steve Ellcey <sje@cup.hp.com>
4232 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
4233 HOST_BITS_PER_LONGLONG
4235 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4238 * c-typeck.c (store_init_value): Don't emit pedantic overflow
4239 warning for non-static initializers.
4241 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
4243 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
4244 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
4245 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
4246 Always use set_unique_reg_note to add REG_EQUAL notes.
4248 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4251 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4253 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4254 * cfgbuild.c (find_basic_blocks): Likewise.
4255 * cfgrtl.c (rtl_create_basic_block): Likewise.
4256 * function.c (temp_slots_at_level): Likewise.
4257 * reg-stack.c (stack_regs_mentioned): Likewise.
4258 * regclass.c (allocate_reg_info): Likewise.
4259 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4260 set_bb_for_stmt, move_block_to_fn): Likewise.
4261 * tree-complex.c (tree_lower_complex): Likewise.
4262 * vec.h (VEC_safe_grow_cleared): New.
4264 2007-01-02 Ian Lance Taylor <iant@google.com>
4266 * c-common.c (c_common_truthvalue_conversion): When warning about
4267 using an assignment as a truth value, set TREE_NO_WARNING.
4269 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4272 * c.opt (Wold-style-declaration): New.
4273 * doc/invoke.texi (C-only Warning Options): New.
4274 (Wold-style-declaration): Document it.
4275 (Wextra): Enabled by -Wextra.
4276 * c-opts.c (c_common_post_options): Enabled by -Wextra.
4277 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
4278 -Wold-style-declaration.
4280 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4282 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4283 * cfgbuild.c (find_basic_blocks): Likewise.
4284 * cfgrtl.c (rtl_create_basic_block): Likewise.
4285 * function.c (temp_slots_at_level): Likewise.
4286 * reg-stack.c (stack_regs_mentioned): Likewise.
4287 * regclass.c (allocate_reg_info): Likewise.
4288 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4289 set_bb_for_stmt, move_block_to_fn): Likewise.
4290 * tree-complex.c (tree_lower_complex): Likewise.
4291 * vec.h (VEC_safe_grow_cleared): New.
4293 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
4295 * c-common.c (c_common_nodes_and_builtins): Since variants of
4296 void_type_node get built before it is given a name, we need to
4297 give those variants the name, too.
4298 (complete_array_type): We need to work with the canonical main
4299 type of the array, from which we will build the qualified version.
4300 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
4301 * print-tree.c (print_node): Display canonical type information
4303 * stor-layout.c (layout_type): When we don't know the
4304 alignment of a type for which we're building an array, we end up
4305 guessing wrong, so make the type require structural equality.
4306 * tree.c (make_node_stat): When we build a new type, it is its
4308 (build_type_attribute_qual_variant): When building an attribute
4309 variant, its canonical type is the non-attribute variant. However,
4310 if the attributes are target-dependent and they differ, we need to
4311 use structural equality checks for this type.
4312 (build_qualified_type): A qualified type is not equivalent to its
4313 unqualified variant; set the canonical type appropriately.
4314 (build_distinct_type_copy): When building a distinct type from
4315 another type, the new type is its own canonical type.
4316 (build_variant_type_copy): When building a new type variant, we
4317 assume that it is equivalent to the original type.
4318 (build_pointer_type_for_mode): When building a pointer type, also
4319 build a canonical type pointer.
4320 (build_reference_type_for_mode): When building a reference type,
4321 also build a canonical type reference.
4322 (build_index_type): When we can't hash an index type (e.g.,
4323 because its maximum value is negative), the index type requires
4324 structural equality tests.
4325 (build_array_type): Build the canonical form of an array type.
4326 (build_function_type): Function types require structural equality,
4327 because they contain default arguments, attributes, etc.
4328 (build_method_type_directly): Ditto for method types.
4329 (build_offset_type): Build the canonical offset type.
4330 (build_complex_type): Build the canonical vector type.
4331 (make_vector_type): Build the canonical vector type.
4332 * tree.h (TYPE_CANONICAL): New.
4333 (TYPE_STRUCTURAL_EQUALITY_P): New.
4334 (SET_TYPE_STRUCTURAL_EQUALITY): New.
4335 (struct tree_type): Added "canonical" field.
4336 * params.h (VERIFY_CANONICAL_TYPES): New.
4337 * doc/c-tree.texi (TYPE_CANONICAL): Document.
4338 (TYPE_STRUCTURAL_EQUALITY_P): Document.
4339 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
4340 * doc/invoke.texi (verify-canonical-types): Document --param
4341 parameter for verifying canonical types.
4343 2007-01-02 Joseph Myers <joseph@codesourcery.com>
4345 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
4346 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
4347 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
4348 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
4350 * config/rs6000/e500.h: New.
4351 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
4352 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
4354 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
4355 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
4356 TARGET_E500_DOUBLE): Remove.
4357 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
4358 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
4359 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
4360 * config/rs6000/rs6000.c (rs6000_override_options): Use
4363 2007-01-02 Joseph Myers <joseph@codesourcery.com>
4365 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
4366 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
4367 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
4368 rs6000_emit_cmove): Don't check TARGET_E500.
4369 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
4370 sordered): Don't check TARGET_E500.
4372 2007-01-01 Eric Christopher <echristo@apple.com>
4374 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
4375 HARD_REGISTER_NUM_P.
4377 2007-01-01 Roger Sayle <roger@eyesopen.com>
4379 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
4380 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
4381 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
4384 2007-01-01 Mike Stump <mrs@apple.com>
4386 * configure.ac: Remove support for building with Apple's gcc-3.1.
4388 2007-01-02 Joseph Myers <joseph@codesourcery.com>
4391 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
4392 REG is a hard register. Do not modify REG before calling
4394 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
4395 multiple of the size of YMODE for certain lowpart cases.
4397 2007-01-01 Andrew Pinski <pinskia@gmail.com>
4400 * gimplify (voidify_wrapper_expr): Update for
4403 2007-01-01 Andreas Schwab <schwab@suse.de>
4406 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
4407 save of BR0 in extra_spill_size instead of spill_size.
4408 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
4410 (ia64_expand_epilogue): Restore BR0 from its new location.
4412 2007-01-01 Andrew Pinski <pinskia@gmail.com>
4414 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
4415 Use a temporary variable if the left hand side is not a gimple
4418 2007-01-01 Andrew Pinski <pinskia@gmail.com>
4420 * gimplify.c (gimplify_return_expr): Make the temporary variable
4421 for the return expression, a gimple register variable.
4423 2007-01-01 Jan Hubicka <jh@suse.cz>
4425 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
4428 2007-01-01 Mike Stump <mrs@apple.com>
4430 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
4432 * config/darwin.c (machopic_select_rtx_section): Use
4434 (darwin_mergeable_constant_section): Likewise.
4435 * configure: Regenerate.
4436 * config.in: Regenerate.
4438 2007-01-01 Jan Hubicka <jh@suse.cz>
4439 Andrew Pinski <pinskia@gmail.com>
4441 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
4442 before starting IPA passes.