1 2007-02-07 Roger Sayle <roger@eyesopen.com>
3 * global.c (compute_regsets): Move declatation of "i" inside of
4 #ifdef ELIMINABLE_REGS to avoid unused variable bootstrap failure.
6 2007-02-07 Jakub Jelinek <jakub@redhat.com>
9 * gimplify.c (gimplify_scan_omp_clauses): Remove special casing
10 of INDIRECT_REF <RESULT_DECL>.
12 * config/i386/i386.c (override_options): Set PTA_SSSE3 for core2.
14 2007-02-06 J"orn Rennecke <joern.rennecke@arc.com>
15 Kaz Kojima <kkojima@gcc.gnu.org>
18 * config/sh/sh.c (expand_cbranchdi4): Use scratch register
20 (sh_initialize_trampoline): Add parentheses to avoid a warning.
22 2007-02-06 Zdenek Dvorak <dvorakz@suse.cz>
24 * doc/loop.texi: Document possibility not to perform disambiguation
25 of loops with multiple latches.
26 * cfgloopmanip.c (alp_enum_p): Removed.
27 (add_loop): Handle subloops. Use get_loop_body_with_size.
28 (create_preheader): Do not allow ENTRY_BLOCK_PTR to be preheader.
29 * cfghooks.c (redirect_edge_and_branch_force): Set dominator for
30 the new forwarder block.
31 (make_forwarder_block): Only call new_bb_cbk if it is not NULL.
32 Handle the case latch is NULL.
33 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Avoid cfg modifications
34 when marking loop exits.
35 * ifcvt.c (if_convert): Ditto. Mark loop exits even if cfg cannot
37 * loop-init.c (loop_optimizer_init): Do not modify cfg. Call
38 disambiguate_loops_with_multiple_latches.
39 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Calculate dominators
40 before fix_loop_structure.
41 * cfgloop.c: Include pointer-set.h and output.h.
42 (canonicalize_loop_headers, HEADER_BLOCK, LATCH_EDGE,
43 update_latch_info, mfb_keep_just, mfb_keep_nonlatch): Removed.
44 (get_loop_latch_edges, find_subloop_latch_edge_by_profile,
45 find_subloop_latch_edge_by_ivs, find_subloop_latch_edge,
46 mfb_redirect_edges_in_set, form_subloop, merge_latch_edges,
47 disambiguate_multiple_latches, get_loop_body_with_size,
48 disambiguate_loops_with_multiple_latches): New functions.
49 (flow_loop_dump): Dump multiple latch edges.
50 (flow_loop_nodes_find): Handle loops with multiple latches.
51 (flow_loops_find): Ditto. Do not call canonicalize_loop_headers.
52 (glb_enum_p): Modified.
53 (get_loop_body): Use get_loop_body_with_size.
54 * cfgloop.h (LOOPS_HAVE_RECORDED_EXITS): New flag.
55 (AVOID_CFG_MODIFICATIONS): New constant.
56 (disambiguate_loops_with_multiple_latches, add_loop,
57 get_loop_body_with_size): Declare.
58 * Makefile.in (cfgloop.o): Add pointer-set.h and output.h.
60 2007-02-06 Seongbae Park <seongbae.park@gmail.com>
63 * global.c (compute_regsets): New function.
64 (global_alloc): Refactored ELIMINABLE_REGSET
65 and NO_GLOBAL_ALLOC_REGS computation out.
66 (rest_of_handle_global_alloc): Call compute_regsets()
67 for non-optimizing case.
69 2007-02-06 Richard Henderson <rth@redhat.com>
71 * config/i386/constraints.md (Y2): Rename from Y.
73 * config/i386/i386.md (movsi_1, movdi_2, pushdf_nointeger,
74 pushdf_integer, movdf_nointeger, movdf_integer, zero_extendsidi2_32,
75 zero_extendsidi2_rex64, truncxfdf2_mixed): Change Y constraints to Y2.
76 (extendsfdf2_mixed, extendsfdf2_sse, truncdfsf_fast_mixed,
77 truncdfsf_fast_sse, truncdfsf_mixed, fix_truncdfdi_sse,
78 fix_truncdfsi_sse, floatsidf2_mixed, floatsidf2_sse,
79 floatdidf2_mixed, floatdidf2_sse, absnegdf2_mixed,
80 absnegdf2_sse, sse_setccdf, fop_df_comm_mixed, fop_df_comm_sse,
81 fop_df_1_mixed, fop_df_1_sse): Change Y constraints to x.
82 * config/i386/mmx.md (mov<MMXMODEI>_internal_rex64,
83 mov<MMXMODEI>_internal, movv2sf_internal_rex64, movv2sf_internal,
84 vec_extractv2si_1): Change Y constraints to Y2.
85 * config/i386/sse.md (vec_setv4sf_0, vec_concatv2df, vec_dupv4si,
86 vec_dupv2di, sse2_concatv2si, vec_concatv4si_1, vec_concatv2di):
87 Change Y constraints to Y2.
88 (sse2_loadld): Change Y constraints to x.
90 2007-02-06 Roger Sayle <roger@eyesopen.com>
92 * config/rs6000/rs6000.md (popcount<mode>2): Rewrite.
93 (parity<mode>2): New define_expand using rs6000_emit_parity.
94 * config/rs6000/rs6000.c (rs6000_emit_popcount,
95 rs6000_emit_parity): New functions.
96 * config/rs6000/rs6000-protos.h (rs6000_emit_popcount,
97 rs6000_emit_parity): Prototype here.
99 2007-02-06 Ian Lance Taylor <iant@google.com>
101 * lower-subreg.c (simple_move_operand): Reject CONST.
102 (resolve_clobber): Call validate_change rather than directly
103 assigning to XEXP (pat, 0).
105 2006-02-06 Paolo Bonzini <bonzini@gnu.org>
107 * Makefile.in (tree-ssa-loop-ivopts.o): Add pointer-set.h dependency.
108 (tree-ssa-reassoc.o): Add pointer-set.h dependency.
109 (tree-cfg.o): Remove hashtab.h dependency.
111 * tree-ssa-loop-ivopts.c: Include pointer-set.h.
112 (struct ivopts_data): Change niters to pointer_map_t.
113 (struct nfe_cache_elt, nfe_hash, nfe_eq): Delete.
114 (niter_for_exit): Create pointer_map on demand. Change for
116 (tree_ssa_iv_optimize_init): Initialize data->niters to NULL.
117 (free_loop_data): Destroy data->niters if created and reset field.
118 (tree_ssa_iv_optimize_finalize): Don't delete data->niters here.
119 (tree_ssa_iv_optimize_loop): Check for presence of stale data.
121 * tree-ssa-reassoc.c: Include pointer-set.h.
122 (bb_rank): Change to long *.
123 (operand_rank): Change to pointer_map_t.
124 (find_operand_rank): Return long, -1 if not found. Declare as inline.
125 (insert_operand_rank): Accept long.
126 (operand_entry_hash, operand_entry_eq): Remove.
127 (get_rank): Return long. Adjust for changes above.
128 (init_reassoc): Change rank type to long. Adjust creation of bb_rank
130 (fini_reassoc): Delete operand_rank with pointer_map_destroy.
132 * tree-ssa-structalias.c (vi_for_tree): Change to pointer_map.
133 (struct tree_vi, tree_vi_t, tree_vi_hash, tree_vi_eq): Delete.
134 (insert_vi_for_tree): Rewrite for pointer_map API. Assert argument
136 (lookup_vi_for_tree): Rewrite for pointer_map API. Return varinfo_t
137 directly since it cannot be NULL.
138 (get_vi_for_tree): Rewrite for pointer_map API.
139 (find_what_p_points_to): Adjust for change to lookup_vi_for_tree.
140 (init_alias_vars): Create vi_for_tree as pointer_map.
141 (delete_points_to_sets): Delete vi_for_tree using pointer_map_destroy.
143 * tree-cfg.c: Don't include hashtab.h.
144 (edge_to_cases): Declare as pointer_map.
145 (struct edge_to_cases_elt, edge_to_cases_hash, edge_to_cases_eq):
147 (edge_to_cases_cleanup): Rewrite as pointer_map_traverse callback.
148 (start_recording_case_labels): Create edge_to_cases as pointer_map.
149 (end_recoding_case_labels): Cleanup edge_to_cases manually before
151 (record_switch_edge): Delete.
152 (get_cases_for_edge): Adjust for pointer_map API, inline
153 record_switch_edge (rewritten for new API), remove goto.
155 2006-02-06 Paolo Bonzini <bonzini@gnu.org>
157 * Makefile.in (tree-nested.o): Add pointer-set.h dependency.
158 * tree-nested.c: Include pointer-set.h.
159 (var_map_elt, var_map_eq, var_map_hash): Delete.
160 (struct nesting_info): Remove GTY marker. Change the two htab_t's
162 (nesting_info_bitmap_obstack): New.
163 (lookup_field_for_decl): Adjust for pointer_map API.
164 (lookup_tramp_for_decl): Adjust for pointer_map API.
165 (get_nonlocal_debug_decl): Adjust for pointer_map API.
166 (get_local_debug_decl): Adjust for pointer_map API.
167 (convert_nl_goto_reference): Adjust for pointer_map API.
168 (convert_nl_goto_receiver): Adjust for pointer_map API.
169 (create_nesting_tree): Create outside GGC space. Create bitmap on
170 the new obstack. Create field_map and var_map as pointer_maps.
171 (free_nesting_tree): Adjust for changes to create_nesting_tree.
173 (lower_nested_functions): Move root here, no need to NULL it.
174 Initialize and release the obstack.
176 2007-02-06 Paolo Bonzini <bonzini@gnu.org>
178 * tree.c (tree_int_map_hash, tree_int_map_eq, tree_int_map_marked_p):
179 Remove prototypes and make them non-static.
180 (struct tree_int_map): Remove.
181 * tree.h (struct tree_int_map): Move here, turning TO into an
183 (tree_int_map_hash, tree_int_map_eq, tree_int_map_marked_p): Declare.
185 * tree.h (TREE_COMPLEXITY): Remove.
186 (struct tree_exp): Remove complexity field.
187 * tree.c (build1_stat): Don't set it.
189 2007-02-06 Dorit Nuzman <dorit@il.ibm.com>
190 Victor Kaplansky <victork@il.ibm.com>
192 * tree-vectorizer.c (vect_is_simple_use): Support induction.
193 (vect_is_simple_reduction): Support reduction with induction as
195 (vect_is_simple_iv_evolution): Fix formatting.
196 * tree-vect-analyze.c (vect_mark_stmts_to_be_vectorized): Fix
197 formatting. Don't mark induction phis for vectorization.
198 (vect_analyze_scalar_cycles): Analyze all inductions, then reductions.
199 * tree-vect-transform.c (get_initial_def_for_induction): New function.
200 (vect_get_vec_def_for_operand): Support induction.
201 (vect_get_vec_def_for_stmt_copy): Fix formatting and add check for
203 (vectorizable_reduction): Support reduction with induction as one of
205 (vectorizable_type_demotion): Use def-type of stmt argument rather
208 * tree-ssa-loop.c (gate_scev_const_prop): Return the value of
209 flag_tree_scev_cprop.
210 * common.opt (tree-scev-cprop): New flag.
212 * tree-vect-transform.c (vect_create_destination_var): Use 'kind' in
213 call to vect_get_new_vect_var.
215 2007-02-06 Ira Rosen <irar@il.ibm.com>
217 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Check that
219 (vect_pattern_recog_1): Likewise.
221 2007-02-05 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
223 * fold-const.c (negate_expr_p): Handle CONJ_EXPR.
224 (fold_negate_expr): Likewise.
226 2007-02-05 Alexandre Oliva <aoliva@redhat.com>
229 * dwarf2out.c (modified_type_die): Follow DECL_ORIGINAL_TYPE
230 even if cv-qualification is the same.
232 2007-02-05 Geoffrey Keating <geoffk@apple.com>
234 * config/rs6000/darwin-tramp.asm (__trampoline_setup): Call
235 __enable_execute_stack on completion.
237 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
239 * config/i386/athlon.md (athlon_fldxf_k8, athlon_fld_k8,
240 athlon_fstxf_k8, athlon_fst_k8, athlon_fist, athlon_fmov,
241 athlon_fadd_load, athlon_fadd_load_k8, athlon_fadd, athlon_fmul,
242 athlon_fmul_load, athlon_fmul_load_k8, athlon_fsgn,
243 athlon_fdiv_load, athlon_fdiv_load_k8, athlon_fdiv_k8,
244 athlon_fpspc_load, athlon_fpspc, athlon_fcmov_load,
245 athlon_fcmov_load_k8, athlon_fcmov_k8, athlon_fcomi_load_k8,
246 athlon_fcomi, athlon_fcom_load_k8, athlon_fcom): Added amdfam10.
248 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
250 * config/i386/i386.md (x86_sahf_1, cmpfp_i_mixed, cmpfp_i_sse,
251 cmpfp_i_i387, cmpfp_iu_mixed, cmpfp_iu_sse, cmpfp_iu_387,
252 swapsi, swaphi_1, swapqi_1, swapdi_rex64, fix_truncsfdi_sse,
253 fix_truncdfdi_sse, fix_truncsfsi_sse, fix_truncdfsi_sse,
254 x86_fldcw_1, floatsisf2_mixed, floatsisf2_sse, floatdisf2_mixed,
255 floatdisf2_sse, floatsidf2_mixed, floatsidf2_sse,
256 floatdidf2_mixed, floatdidf2_sse, muldi3_1_rex64, mulsi3_1,
257 mulsi3_1_zext, mulhi3_1, mulqi3_1, umulqihi3_1, mulqihi3_insn,
258 umulditi3_insn, umulsidi3_insn, mulditi3_insn, mulsidi3_insn,
259 umuldi3_highpart_rex64, umulsi3_highpart_insn,
260 umulsi3_highpart_zext, smuldi3_highpart_rex64,
261 smulsi3_highpart_insn, smulsi3_highpart_zext, x86_64_shld,
262 x86_shld_1, x86_64_shrd, sqrtsf2_mixed, sqrtsf2_sse,
263 sqrtsf2_i387, sqrtdf2_mixed, sqrtdf2_sse, sqrtdf2_i387,
264 sqrtextendsfdf2_i387, sqrtxf2, sqrtextendsfxf2_i387,
265 sqrtextenddfxf2_i387): Added amdfam10_decode.
267 * config/i386/athlon.md (athlon_idirect_amdfam10,
268 athlon_ivector_amdfam10, athlon_idirect_load_amdfam10,
269 athlon_ivector_load_amdfam10, athlon_idirect_both_amdfam10,
270 athlon_ivector_both_amdfam10, athlon_idirect_store_amdfam10,
271 athlon_ivector_store_amdfam10): New define_insn_reservation.
272 (athlon_idirect_loadmov, athlon_idirect_movstore): Added
275 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
277 * config/i386/athlon.md (athlon_call_amdfam10,
278 athlon_pop_amdfam10, athlon_lea_amdfam10): New
279 define_insn_reservation.
280 (athlon_branch, athlon_push, athlon_leave_k8, athlon_imul_k8,
281 athlon_imul_k8_DI, athlon_imul_mem_k8, athlon_imul_mem_k8_DI,
282 athlon_idiv, athlon_idiv_mem, athlon_str): Added amdfam10.
284 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
286 * config/i386/athlon.md (athlon_sseld_amdfam10,
287 athlon_mmxld_amdfam10, athlon_ssest_amdfam10,
288 athlon_mmxssest_short_amdfam10): New define_insn_reservation.
290 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
292 * config/i386/athlon.md (athlon_sseins_amdfam10): New
293 define_insn_reservation.
294 * config/i386/i386.md (sseins): Added sseins to define_attr type
295 and define_attr unit.
296 * config/i386/sse.md: Set type attribute to sseins for insertq
299 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
301 * config/i386/athlon.md (sselog_load_amdfam10, sselog_amdfam10,
302 ssecmpvector_load_amdfam10, ssecmpvector_amdfam10,
303 ssecomi_load_amdfam10, ssecomi_amdfam10,
304 sseaddvector_load_amdfam10, sseaddvector_amdfam10): New
305 define_insn_reservation.
306 (ssecmp_load_k8, ssecmp, sseadd_load_k8, seadd): Added amdfam10.
308 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
310 * config/i386/athlon.md (cvtss2sd_load_amdfam10,
311 cvtss2sd_amdfam10, cvtps2pd_load_amdfam10, cvtps2pd_amdfam10,
312 cvtsi2sd_load_amdfam10, cvtsi2ss_load_amdfam10,
313 cvtsi2sd_amdfam10, cvtsi2ss_amdfam10, cvtsd2ss_load_amdfam10,
314 cvtsd2ss_amdfam10, cvtpd2ps_load_amdfam10, cvtpd2ps_amdfam10,
315 cvtsX2si_load_amdfam10, cvtsX2si_amdfam10): New
316 define_insn_reservation.
318 * config/i386/sse.md (cvtsi2ss, cvtsi2ssq, cvtss2si,
319 cvtss2siq, cvttss2si, cvttss2siq, cvtsi2sd, cvtsi2sdq,
320 cvtsd2si, cvtsd2siq, cvttsd2si, cvttsd2siq,
321 cvtpd2dq, cvttpd2dq, cvtsd2ss, cvtss2sd,
322 cvtpd2ps, cvtps2pd): Added amdfam10_decode attribute.
324 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
326 * config/i386/athlon.md (athlon_ssedivvector_amdfam10,
327 athlon_ssedivvector_load_amdfam10, athlon_ssemulvector_amdfam10,
328 athlon_ssemulvector_load_amdfam10): New define_insn_reservation.
329 (athlon_ssediv, athlon_ssediv_load_k8, athlon_ssemul,
330 athlon_ssemul_load_k8): Added amdfam10.
332 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
334 * config/i386/i386.h (TARGET_SSE_UNALIGNED_MOVE_OPTIMAL): New macro.
335 (x86_sse_unaligned_move_optimal): New variable.
337 * config/i386/i386.c (x86_sse_unaligned_move_optimal): Enable for
339 (ix86_expand_vector_move_misalign): Add code to generate movupd/movups
340 for unaligned vector SSE double/single precision loads for AMDFAM10.
342 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
344 * config/i386/i386.h (TARGET_AMDFAM10): New macro.
345 (TARGET_CPU_CPP_BUILTINS): Add code for amdfam10.
346 Define TARGET_CPU_DEFAULT_amdfam10.
347 (TARGET_CPU_DEFAULT_NAMES): Add amdfam10.
348 (processor_type): Add PROCESSOR_AMDFAM10.
350 * config/i386/i386.md: Add amdfam10 as a new cpu attribute to match
351 processor_type in config/i386/i386.h.
352 Enable imul peepholes for TARGET_AMDFAM10.
354 * config.gcc: Add support for --with-cpu option for amdfam10.
356 * config/i386/i386.c (amdfam10_cost): New variable.
357 (m_AMDFAM10): New macro.
358 (m_ATHLON_K8_AMDFAM10): New macro.
359 (x86_use_leave, x86_push_memory, x86_movx, x86_unroll_strlen,
360 x86_cmove, x86_3dnow_a, x86_deep_branch, x86_use_simode_fiop,
361 x86_promote_QImode, x86_integer_DFmode_moves,
362 x86_partial_reg_dependency, x86_memory_mismatch_stall,
363 x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
364 x86_sse_partial_reg_dependency, x86_sse_typeless_stores,
365 x86_use_ffreep, x86_use_incdec, x86_four_jump_limit,
366 x86_schedule, x86_use_bt, x86_cmpxchg16b, x86_pad_returns):
367 Enable/disable for amdfam10.
368 (override_options): Add amdfam10_cost to processor_target_table.
369 Set up PROCESSOR_AMDFAM10 for amdfam10 entry in
370 processor_alias_table.
371 (ix86_issue_rate): Add PROCESSOR_AMDFAM10.
372 (ix86_adjust_cost): Add code for amdfam10.
374 2007-02-05 Harsha Jagasia <harsha.jagasia@amd.com>
376 * config/i386/i386.opt: Add new Advanced Bit Manipulation (-mabm)
377 instruction set feature flag. Add new (-mpopcnt) flag for popcnt
378 instruction. Add new SSE4A (-msse4a) instruction set feature flag.
379 * config/i386/i386.h: Add builtin definition for SSE4A.
380 * config/i386/i386.md: Add support for ABM instructions
382 * config/i386/sse.md: Add support for SSE4A instructions
383 (movntss, movntsd, extrq, insertq).
384 * config/i386/i386.c: Add support for ABM and SSE4A builtins.
385 Add -march=amdfam10 flag.
386 * config/i386/ammintrin.h: Add support for SSE4A intrinsics.
387 * doc/invoke.texi: Add documentation on flags for sse4a, abm, popcnt
389 * doc/extend.texi: Add documentation for SSE4A builtins.
391 2007-02-05 Bob Wilson <bob.wilson@acm.org>
393 * config/xtensa/xtensa.c (constantpool_mem_p): Skip over SUBREGs.
395 2007-02-05 Richard Guenther <rguenther@suse.de>
397 * tree-vectorizer.h (vectorizable_function): Add argument type
398 argument, change return type.
399 * tree-vect-patterns.c (vect_recog_pow_pattern): Adjust caller.
400 * tree-vect-transform.c (vectorizable_function): Handle extra
401 argument, return vectorized function decl.
402 (build_vectorized_function_call): Remove.
403 (vectorizable_call): Handle calls with result and argument types
404 differing. Handle loop vectorization factor correctly.
405 * targhooks.c (default_builtin_vectorized_function): Adjust for
407 * targhooks.h (default_builtin_vectorized_function): Likewise.
408 * target.h (builtin_vectorized_function): Add argument type
410 * config/i386/i386.c (ix86_builtin_vectorized_function): Handle
411 extra argument, allow vectorizing of lrintf.
412 * doc/tm.texi (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Adjust
413 documentation of target hook.
415 2007-02-05 Hans-Peter Nilsson <hp@axis.com>
418 * config/cris/cris.md ("*andsi_movu", "*andsi_clear", "*andhi_movu")
419 ("*andhi_clear", andu (casesi+45)): For size-changed operand where
420 memory is allowed, require !side_effects_p, not just !MEM_VOLATILE_P.
422 2007-02-05 Roger Sayle <roger@eyesopen.com>
424 * fold-const.c (fold_unary) <REAL_PART>: Test for availability of
425 BUILT_IN_COS before simplifying REAL_PART(CEXPI)) to COS.
426 <IMAG_PART>: Likewise, check for availability of BUILT_IN_SIN.
427 * builtins.c (fold_builtin_sincos): Check for TARGET_C99_FUNCTIONS
428 before canonicalizing sincos to cexpi.
429 (fold_builtin_cexp): Likewise, for canonicalizing cexp to cexpi.
431 2007-02-05 Roger Sayle <roger@eyesopen.com>
433 * config/alpha/alpha.c (alpha_add_builtins): New Helper function.
434 Set TREE_READONLY and TREE_NOTHROW directly, not via attributes.
435 (alpha_init_builtins): Use alpha_add_builtins to process tables.
437 2007-02-05 Roger Sayle <roger@eyesopen.com>
439 * mips-tfile.c (initialize_init_file): Correct endianness test.
441 2007-02-05 Kazu Hirata <kazu@codesourcery.com>
443 * config/m68k/m68k.md (pushdi-1, pushdi, movsi+1): Don't use
446 2007-02-05 Richard Sandiford <richard@codesourcery.com>
448 * dwarf2out.c (dwarf2out_frame_debug_expr): Record the register
449 saves in a PARALLEL before the register assignments.
451 2007-02-05 Richard Sandiford <richard@codesourcery.com>
453 * doc/tm.texi (DWARF_ALT_FRAME_RETURN_COLUMN): Do not require
454 DWARF_FRAME_RETURN_COLUMN to be a general register.
455 * dwarf2out.c (init_return_column_size): New function, split from...
456 (expand_builtin_init_dwarf_reg_sizes): ...here. Allow both
457 DWARF_FRAME_RETURN_COLUMN and DWARF_ALT_FRAME_RETURN_COLUMN
458 to be nongeneral registers.
459 * config/m68k/m68k.h (DWARF_FRAME_REGNUM): Only map FP and
461 (DWARF_FRAME_REGISTERS, DWARF_FRAME_RETURN_COLUMN): Define.
462 (DWARF_ALT_FRAME_RETURN_COLUMN): Define.
464 2007-02-04 Zdenek Dvorak <dvorakz@suse.cz>
466 * cfgcleanup.c (try_optimize_cfg): Avoid removing ENTRY_BLOCK_PTR.
468 2007-02-04 Zdenek Dvorak <dvorakz@suse.cz>
470 * cfgloopmanip.c (loop_delete_branch_edge): Removed.
471 (remove_path): Use can_remove_branch_p and remove_branch instead
472 of loop_delete_branch_edge.
473 * tree-ssa-loop-manip.c (scale_dominated_blocks_in_loop): New function.
474 (tree_transform_and_unroll_loop): Remove dead branches immediately.
475 Update profile using scale_dominated_blocks_in_loop.
476 * cfghooks.c (can_remove_branch_p, remove_branch): New functions.
477 * cfghooks.h (struct cfg_hooks): Add can_remove_branch_p.
478 (can_remove_branch_p, remove_branch): Declare.
479 * tree-cfg.c (tree_can_remove_branch_p): New function.
480 (tree_cfg_hooks): Add tree_can_remove_branch_p.
481 * cfgrtl.c (rtl_can_remove_branch_p): New function.
482 (rtl_cfg_hooks, cfg_layout_rtl_cfg_hook): Add rtl_can_remove_branch_p.
484 2007-02-05 Jan Hubicka <jh@suse.cz>
487 * ipa-inline.c (cgraph_clone_inlined_nodes): When there are unanalyzed
488 nodes in cgraph, don't remove offline copy of the function.
490 2007-02-04 Jan Hubicka <jh@suse.cz>
492 * tree-sra.c (sra_walk_expr): Add linebreaks. BITFIELD_REFs into
493 vectors might cause maybe_lookup_element_for_expr to be called
494 on non-sra-candidate.
496 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
498 * config/bfin/bfin-modes.def, config/bfin/bfin.c,
499 config/bfin/bfin.md, config/bfin/predicates.md: Follow
500 spelling conventions.
502 2007-02-04 Richard Guenther <rguenther@suse.de>
505 * fold-const.c (try_move_mult_to_index): Make sure to not
506 overflow one dimension of a multi-dimensional array access.
508 2007-02-04 Jan Hubicka <jh@suse.cz>
510 * passes.c (init_optimization_passes): Reindent.
512 2007-02-04 Jan Hubicka <jh@suse.cz>
513 Eric Botcazou <ebotcazou@adacore.com>
515 * tree-optimize.c (has_abnormal_outgoing_edge_p): Move to...
516 (execute_fixup_cfg): Break out the abnormal goto code.
517 * tree-inline.c (has_abnormal_outgoing_edge_p): ...here.
518 (make_nonlocal_label_edges): Move here from execute_fixup_cfg.
519 (optimize_inline_calls): Call make_nonlocal_label_edges.
521 2007-02-04 Jan Hubicka <jh@suse.cz>
523 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Return
524 true when something was changed.
525 (rename_ssa_copies): When something was changed, do
526 TODO_remove_unused_locals.
527 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
528 add TODO_remove_unused_locals when instruction was removed.
530 2007-02-04 Jan Hubicka <jh@suse.cz>
532 * ipa-inline.c (try_inline): Improve debug output; work on already
534 (cgraph_decide_inlining_incrementally): Indent; improve debug output;
535 call try_inline for already inlined edges too when flattening;
536 inline also functions that make callee growth but overall unit size
539 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
541 * config/m32c/bitops.md, config/m32c/jump.md,
542 config/m32c/m32c.c, config/m32c/m32c.h, config/m32r/m32r.c,
543 config/m32r/m32r.h, config/m32r/m32r.md,
544 config/m32r/predicates.md, config/m68hc11/larith.asm,
545 config/m68hc11/m68hc11.c, config/m68hc11/m68hc11.h,
546 config/m68k/m68k.h, config/mcore/mcore.md, config/mips/4k.md,
547 config/mips/mips-protos.h, config/mips/mips.c,
548 config/mips/mips.h, config/mips/mips.md, config/mips/mips16.S,
549 config/mn10300/mn10300.h, config/mn10300/predicates.md,
550 config/mt/mt.c, config/mt/mt.h, config/mt/mt.md: Follow
551 spelling conventions.
553 * config/v850/v850.c, config/v850/v850.h, config/v850/v850.md:
554 Follow spelling conventions.
556 2007-02-03 Douglas Gregor <doug.gregor@gmail.com>
558 * c-opts.c (c_common_post_options): If C++0x mode is enabled, don't
559 warn about C++0x compatibility.
561 2007-02-04 Kazu Hirata <kazu@codesourcery.com>
563 * config/h8300/h8300.c, config/h8300/h8300.h,
564 config/h8300/h8300.md: Follow spelling conventions.
566 2007-02-03 Uros Bizjak <ubizjak@gmail.com>
569 * combine.c (try_combine): Do not substitute source operand
570 with constants wider than 2 * HOST_BITS_PER_WIDE_INT.
572 2007-02-03 Jan Hubicka <jh@suse.cz>
574 PR gcov-profile/30650
575 * value-prof.c (stringop_block_profile): Fix handling of size counter;
576 do not divide by zero for never executed counters.
577 (tree_find_values_to_profile): Fix counters.
578 * gcov-ui.h (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): Fix comments.
580 2007-02-03 Ian Lance Taylor <iant@google.com>
582 * lower-subreg.c (simple_move_operand): New static function,
583 broken out of simple_move. Reject LABEL_REF, SYMBOL_REF, and HIGH
585 (simple_move): Call simple_move_operand.
586 (find_decomposable_subregs): Add special handling of MEMs.
587 (can_decompose_p): Rename from cannot_decompose_p. Reverse
588 meaning of return value. If we see a hard register, test whether
589 it can store a word_mode value. Change all callers.
591 2007-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
593 * pa.md (addvdi3, addvsi3, subvdi3, subvsi3, negvdi2, negvsi2): New
594 ftrapv insns and expanders.
595 (subdi3): Change define_expand operand 1 to arith11_operand, and
596 operand 2 to reg_or_0_operand. Change constraints of 64-bit insn
597 pattern to handle reg_or_0 operands. Revise 32-bit insn pattern to
598 handle 11-bit constants and reg_or_0 operands in operands 1 and 2,
602 * varasm.c (notice_global_symbol): Treat global objects as weak when
605 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
607 * emit-rtl.c (dconstpi): Delete.
609 (init_emit_once): Delete dconstpi and init dconstsqrt2.
610 * real.h (dconstpi): Delete.
612 * builtins.c (fold_builtin_cabs): Use dconstsqrt2.
613 (fold_builtin_hypot): Likewise.
615 2007-02-03 Tom Tromey <tromey@redhat.com>
618 * gcc.c (cpp_unique_options): Any of -ggdb3, -gstabs3,
619 -gcoff3, -gxcoff3, -gvms3 implies -dD.
621 2007-02-03 Kazu Hirata <kazu@codesourcery.com>
623 * c-decl.c, config/avr/avr.c, config/avr/avr.h,
624 config/m68k/m68k.c, config/m68k/netbsd-elf.h,
625 config/mn10300/mn10300.c, config/pdp11/pdp11.h,
626 config/rs6000/cell.md, config/rs6000/darwin.h,
627 config/sh/sh.md, config/sh/sh4-300.md, config/spu/spu.c,
628 config/spu/spu.md, cselib.c, expr.c, haifa-sched.c, hwint.h,
629 jump.c, reload.c, sched-deps.c, sched-int.h, tree-inline.c,
630 tree-profile.c, tree-ssa-live.h, tree-vrp.c: Fix comment
631 typos. Follow spelling conventions.
632 * doc/invoke.texi: Follow spelling conventions.
634 2007-02-03 Roger Sayle <roger@eyesopen.com>
636 * simplify-rtx.c (simplify_relational_operation_1): Implement some
637 canonicalization transformations that attempt to simplify integer
638 constant comparisons to become comparisons against zero.
640 2007-02-03 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
642 * builtins.c (fold_builtin_cabs): Fold cabs(x+xi) into
644 * fold-const.c (fold_binary): Fix comment typos. Fold complex
645 (x,0)-(0,y) into (x,-y). Likewise (0,y)-(x,0) into (-x,y).
647 2007-02-02 Mike Stump <mrs@apple.com>
649 * config/darwin9.h (DARWIN_LINKER_GENERATES_ISLANDS): Add.
650 * config/rs6000/rs6000.c (DARWIN_GENERATE_ISLANDS): Add.
651 (output_call): Use DARWIN_GENERATE_ISLANDS to decide when to
652 generate a branch island.
654 2007-02-02 Bob Wilson <bob.wilson@acm.org>
656 * config/xtensa/xtensa.c (smalloffset_mem_p): Use BASE_REG_P.
657 (xtensa_legitimate_address_p): New.
658 (xtensa_legitimize_address): New.
659 (xtensa_output_addr_const_extra): New.
660 * config/xtensa/xtensa.h (REG_OK_STRICT_FLAG): Define.
662 (REG_OK_FOR_BASE_P): Use BASE_REG_P.
663 (GO_IF_LEGITIMATE_ADDRESS): Move code to xtensa_legitimate_address_p.
664 (LEGITIMIZE_ADDRESS): Move code to xtensa_legitimize_address.
665 (OUTPUT_ADDR_CONST_EXTRA): Move code to xtensa_output_addr_const_extra.
666 * config/xtensa/xtensa-protos.h (xtensa_legitimate_address_p): New.
667 (xtensa_legitimize_address): New.
668 (xtensa_output_addr_const_extra): New.
670 2007-02-02 Steve Ellcey <sje@cup.hp.com>
672 * config/ia64/ia64.c (ia64_print_operand): Fix compare strings.
674 2007-02-02 Ian Lance Taylor <iant@google.com>
676 * expmed.c (expand_divmod): Add comment.
678 2007-02-02 Kazu Hirata <kazu@codesourcery.com>
680 * emit-rtl.c (renumber_insns): Remove.
681 * flags.h: Remove the extern for flag_renumber_insns.
682 * rtl.h: Remove the prototype for renumber_insns.
683 * toplev.c (flag_renumber_insns): Remove.
685 2007-02-02 Hui-May Chang <hm.chang@apple.com>
687 Revert for x86 darwin:
688 2005-06-19 Uros Bizjak <uros@kss-loka.si>
690 * config/i386/i386.c (ix86_function_arg_regno_p): Put back the
691 code before the following patch under TARGET_MACHO.
692 (ix86_function_value_regno_p): Likewise.
694 2007-02-02 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
696 * fold-const.c (negate_expr_p, fold_negate_expr): Handle
699 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
702 * config/ia64/ia64.c (ia64_speculate_insn): Restrict to memory loads to
703 general or fp registers. Add comments.
704 * config/ia64/ia64.md (reg_pred_prefix): Add comment.
706 2007-02-02 Paolo Bonzini <bonzini@gnu.org>
708 * pointer-set.c (insert_aux): Only return insertion slot.
709 (pointer_set_insert): Adjust.
710 (pointer_set_traverse, struct pointer_map_t, pointer_map_create,
711 pointer_map_destroy, pointer_map_insert, pointer_map_contains,
712 pointer_map_traverse): New.
713 * pointer-set.h (pointer_set_traverse, struct pointer_map_t,
714 pointer_map_create, pointer_map_destroy, pointer_map_insert,
715 pointer_map_contains, pointer_map_traverse): Declare.
717 2007-02-02 Jakub Jelinek <jakub@redhat.com>
720 * builtins.c (fold_builtin_sprintf): Do not attempt to optimize
721 sprintf (str, "%s"). Do not optimize sprintf (str, "nopercent", p++).
723 2007-02-02 Maxim Kuvyrkov <mkuvyrkov@ispras.ru>
725 * sched-int.h (ds_to_dk, dk_to_ds): Declare functions.
727 (struct _dep): New type.
728 (dep_t): New typedef.
729 (DEP_PRO, DEP_CON, DEP_KIND): New access macros.
730 (DEP_STATUS): New access macro. The macro with the same name was
731 renamed to DEP_LINK_STATUS.
732 (dep_init): Declare function
734 (struct _dep_link): New type.
735 (dep_link_t): New typedef.
736 (DEP_LINK_NODE, DEP_LINK_NEXT, DEP_LINK_PREV_NEXTP): New access macros.
737 (DEP_LINK_DEP, DEP_LINK_PRO, DEP_LINK_CON, DEP_LINK_KIND): New macros.
738 (DEP_LINK_STATUS): New macro.
739 (debug_dep_links): New debug function.
741 (struct _deps_list): New type.
742 (deps_list_t): New typedef.
743 (DEPS_LIST_FIRST): New access macro.
744 (FOR_EACH_DEP_LINK): New cycle macro.
745 (create_deps_list, free_deps_list, delete_deps_list): Declare
747 (deps_list_empty_p, debug_deps_list, add_back_dep_to_deps_list): Ditto.
748 (find_link_by_pro_in_deps_list, find_link_by_con_in_deps_list): Ditto.
749 (copy_deps_list_change_con): Ditto.
751 (move_dep_link): Declare function.
753 (struct _dep_node): New type.
754 (dep_node_t): New typedef.
755 (DEP_NODE_BACK, DEP_NODE_DEP, DEP_NODE_FORW): New access macros.
757 (struct haifa_insn_data.back_deps): New field to hold backward
758 dependencies of the insn.
759 (struct haifa_insn_data.depend): Rename to forw_deps. Change its type
761 (struct haifa_insn_data.resolved_deps): Rename to resolved_back_deps.
762 Change its type to deps_list_t.
763 (INSN_BACK_DEPS): New access macro to use instead of LOG_LINKS.
764 (INSN_DEPEND): Rename to INSN_FORW_DEPS.
765 (RESOLVED_DEPS): Rename to INSN_RESOLVED_BACK_DEPS.
767 (INSN_COST): Move to haifa-sched.c. Use insn_cost () instead.
769 (DEP_STATUS): Rename to DEP_LINK_STATUS. Fix typo in the comment.
771 (add_forw_dep, delete_back_forw_dep, insn_cost): Update declaration and
775 * sched-deps.c (CHECK): New macro to (en/dis)able sanity checks.
776 (ds_to_dk, dk_to_ds): New functions.
778 (init_dep_1): New static function.
779 (init_dep): New function.
780 (copy_dep): New static function.
782 (dep_link_consistent_p, attach_dep_link, add_to_deps_list): New static
784 (detach_dep_link): New static function.
785 (move_dep_link): New function.
787 (dep_links_consistent_p, dump_dep_links): New static functions.
788 (debug_dep_links): New debugging function.
790 (deps_obstack, dl_obstack, dn_obstack): New static variables.
792 (alloc_deps_list, init_deps_list): New static functions.
793 (create_deps_list): New function.
794 (clear_deps_list): New static function.
795 (free_deps_list, delete_deps_list, deps_list_empty_p): New functions.
796 (deps_list_consistent_p, dump_deps_list): New static functions.
797 (debug_deps_list): New function.
798 (add_back_dep_to_deps_list, find_link_by_pro_in_deps_list): New
800 (find_link_by_con_in_deps_list, copy_deps_list_change_con): Ditto.
802 (maybe_add_or_update_back_dep_1, add_or_update_back_dep_1): Update to
803 use new scheduler dependencies lists.
804 (add_back_dep, delete_all_dependences, fixup_sched_groups): Ditto.
805 (sched_analyze): Ditto. Initialize dependencies lists.
806 (add_forw_dep, compute_forward_dependences): Update to use new
807 scheduler dependencies lists.
809 (init_dependency_caches): Init deps_obstack.
810 (free_dependency_caches): Free deps_obstack.
812 (adjust_add_sorted_back_dep, adjust_back_add_forw_dep): Update to use
813 new scheduler dependencies lists.
814 (delete_forw_dep, add_or_update_back_forw_dep): Ditto.
815 (add_back_forw_dep, delete_back_forw_dep): Ditto.
817 * sched-rgn.c (set_spec_fed, find_conditional_protection, is_pfree):
818 Update to use new scheduler dependencies lists.
819 (is_conditionally_protected, is_prisky, add_branch_dependences): Ditto.
820 (debug_dependencies): Ditto.
821 (schedule_region): Update comments.
823 * sched-ebb.c (earliest_block_with_similiar_load): Update to use new
824 scheduler dependencies lists.
825 (schedule_ebb): Update comments.
827 * rtl.def (DEPS_LIST): Remove.
829 * lists.c (unused_deps_list): Remove.
830 (free_list): Update assertions.
832 (alloc_DEPS_LIST, free_DEPS_LIST_list, free_DEPS_LIST_node): Remove.
833 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
835 * rtl.h (free_DEPS_LIST_list, alloc_DEPS_LIST): Remove declarations.
836 (remove_free_DEPS_LIST_elem, copy_DEPS_LIST_list): Ditto.
838 * haifa-sched.c (comments): Update.
839 (insn_cost1): Remove. Inline the code into insn_cost ().
840 (insn_cost): Update to use new scheduler dependencies lists. Move
841 processing of the dependency cost to dep_cost ().
842 (dep_cost): New function. Use it instead of insn_cost () when
843 evaluating cost of the dependency. Use compatible interface to
844 interact with the target.
845 (priority): Update to use new scheduler dependencies lists.
846 (rank_for_schedule): Ditto. Optimize heuristic that prefers the insn
847 with greater number of insns that depend on the insn.
848 (schedule_insn): Update to use new scheduler dependencies lists. Add
849 code to free backward dependencies lists. Inline and optimize code
850 from resolve_dep () - see PR28071.
851 (ok_for_early_queue_removal): Update to use new scheduler dependencies
852 lists. Update call to targetm.sched.is_costly_dependence hook.
854 (fix_inter_tick, try_ready, fix_tick_ready): Update to use new
855 scheduler dependencies lists.
857 (resolve_dep): Remove. Move the logic to schedule_insn ().
858 (init_h_i_d): Initialize dependencies lists.
860 (process_insn_depend_be_in_spec): Rename to
861 process_insn_forw_deps_be_in_spec. Update to use new scheduler
863 (add_to_speculative_block, create_check_block_twin, fix_recovery_deps):
864 Update to use new scheduler dependencies lists.
865 (clear_priorities, calc_priorities, add_jump_dependencies): Ditto.
867 * ddg.c (create_ddg_dependence, create_ddg_dep_no_link): Update to use
868 new scheduler dependencies lists.
869 (build_intra_loop_deps): Ditto.
871 * target.h (struct _dep): Declare to use in
872 gcc_target.sched.is_costly_dependence.
873 (struct gcc_target.sched.adjust_cost): Fix typo.
874 (struct gcc_target.sched.is_costly_dependence): Change signature to use
875 single dep_t parameter instead of an equivalent triad.
876 (struct gcc_target.sched.adjust_cost_2): Remove.
878 * target-def.h (TARGET_SCHED_ADJUST_COST_2): Remove.
880 * reg-notes.def (DEP_TRUE, DEP_OUTPUT, DEP_ANTI): Update comments.
882 * doc/tm.texi (TARGET_SCHED_IS_COSTLY_DEPENDENCE): Update
884 (TARGET_SCHED_ADJUST_COST_2): Remove documentation.
886 * doc/rtl.texi (LOG_LINKS): Remove part about instruction scheduler.
887 (REG_DEP_TRUE): Document.
889 * config/ia64/ia64.c (ia64_adjust_cost_2): Rename to ia64_adjust_cost.
890 Change signature to correspond to the targetm.sched.adjust_cost hook.
891 Update use in TARGET_SCHED_ADJUST_COST_2.
892 (TARGET_SCHED_ADJUST_COST_2): Rename to TARGET_SCHED_ADJUST_COST.
893 (ia64_dependencies_evaluation_hook, ia64_dfa_new_cycle): Update to use
894 new scheduler dependencies lists.
895 (ia64_gen_check): Ditto.
897 * config/mips/mips.c (vr4130_swap_insns_p): Update to use new scheduler
900 * config/rs6000/rs6000.c (rs6000_is_costly_dependence): Change
901 signature to correspond to the targetm.sched.is_costly_dependence hook.
902 (is_costly_group): Update to use new scheduler dependencies lists.
904 * config/spu/spu.c (spu_sched_adjust_cost): Use insn_cost () function
905 instead of INSN_COST () macro.
907 2007-02-01 Ian Lance Taylor <iant@google.com>
909 * lower-subreg.c (resolve_clobber): Handle a subreg of a concatn.
911 2007-02-01 Guy Martin <gmsoft@gentoo.org>
913 * pa.md (tp_load): Correct mfctl instruction syntax.
915 2007-02-01 Geoffrey Keating <geoffk@apple.com>
917 * config/rs6000/rs6000.c (rs6000_stack_info): Correct
918 altivec_padding_size calculation on AIX. Improve comment, add
919 assert to verify that it's right.
921 * config/rs6000/darwin.h (MD_UNWIND_SUPPORT): Don't define for
924 2007-2-01 Seongbae Park <seongbae.park@gmail.com>
927 * global.c (global_alloc): Add mising initialization of
930 2007-02-01 Roger Sayle <roger@eyesopen.com>
932 * alias.c (init_alias_analysis): Correct whitespace.
933 * bb-reorder.c (fix_edges_for_rarely_executed_code,
934 partition_hot_cold_basic_blocks): Likewise.
935 * builtins.c (expand_builtin_printf, expand_builtin_fprintf,
936 expand_builtin_sprintf, fold_builtin_carg, fold_builtin_sprintf,
937 maybe_emit_sprintf_chk_warning, fold_builtin_sprintf_chk,
938 fold_builtin_snprintf_chk, fold_builtin_printf,
939 fold_builtin_fprintf, do_mpfr_ckconv, do_mpfr_arg1, do_mpfr_arg2,
940 do_mpfr_arg3, do_mpfr_sincos): Likewise.
941 * cfgcleanup.c (cleanup_cfg): Likewise.
942 * cfgexpand.c (tree_expand_cfg): Likewise.
943 * fold-const.c (fold_binary) <RDIV_EXPR>: Likewise.
944 * function.c (get_next_funcdef_no): Likewise.
945 * gengtype.c (main): Likewise.
946 * genmodes.c (main): Likewise.
947 * gcse.c (bypass_conditional_jumps, print_ldst_list): Likewise.
948 * haifa-sched.c (schedule_block, extend_h_i_d): Likewise.
949 * ifcvt.c (noce_emit_move_insn): Likewise.
950 * modulo-sched.c (generate_prolog_epilog, sms_schedule_by_order):
952 * stor-layout.c (get_best_mode): Likewise.
953 * tree-ssa-loop-niter.c (get_val_for): Likewise.
954 * tree-ssa-structalias.c (get_varinfo, get_varinfo_fc,
955 scc_visit, do_ds_constraint, do_complex_constraint, label_visit,
956 perform_var_substitution, solve_graph): Likewise.
957 * tree-vrp.c (vrp_finalize): Likewise.
959 2007-02-01 Ian Lance Taylor <iant@google.com>
961 * lower-subreg.c (simplify_gen_subreg_concatn): If we ask for the
962 high part of a paradoxical subreg, return a constant zero.
964 2007-02-01 Zdenek Dvorak <dvorakz@suse.cz>
966 * toplev.c (lang_dependent_init): Call init_set_costs.
967 * loop-init.c (loop_optimizer_init): Do not call init_set_costs.
969 2007-02-01 Richard Guenther <rguenther@suse.de>
972 * fold-const.c (fold_negate_expr): Allow negating a
973 constant if overflow does not change.
975 2007-02-01 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
977 * doc/c-tree.texi (Expression trees): Improve markup.
978 * doc/tm.texi (Register Classes, Addressing Modes)
979 (Floating Point): Fix spacing after abbreviations. Fix some
982 2007-02-01 Ben Elliston <bje@au.ibm.com>
984 * doc/invoke.texi: Replace "bugfix" with "bug fix" throughout.
985 * doc/contrib.texi: Likewise.
986 * doc/install.texi: Likewise.
988 2007-01-31 Richard Henderson <rth@redhat.com>
989 Ian Lance Taylor <iant@google.com>
991 * lower-subreg.c: New file.
992 * rtl.def (CONCATN): Define.
993 * passes.c (init_optimization_passes): Add pass_lower_subreg and
995 * emit-rtl.c (update_reg_offset): New static function, broken out
996 of gen_rtx_REG_offset.
997 (gen_rtx_REG_offset): Call update_reg_offset.
998 (gen_reg_rtx_offset): New function.
999 * regclass.c: Revert patch of 2006-03-05, restoring
1001 (clear_reg_info_regno): New function.
1002 * dwarf2out.c (concatn_loc_descriptor): New static function.
1003 (loc_descriptor): Handle CONCATN.
1004 * common.opt (fsplit_wide_types): New option.
1005 * opts.c (decode_options): Set flag_split_wide_types when
1007 * timevar.def (TV_LOWER_SUBREG): Define.
1008 * rtl.h (gen_reg_rtx_offset): Declare.
1009 (reg_scan_update): Declare.
1010 * regs.h (clear_reg_info_regno): Declare.
1011 * tree-pass.h (pass_lower_subreg): Declare.
1012 (pass_lower_subreg2): Declare.
1013 * doc/invoke.texi (Option Summary): List -fno-split-wide-types.
1014 (Optimize Options): Add -fsplit-wide-types to -O1 list. Document
1016 * doc/rtl.texi (Regs and Memory): Document concat and concatn.
1017 * Makefile.in (OBJS-common): Add lower-subreg.o.
1018 (lower-subreg.o): New target.
1020 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1022 * config/sh/sh.h (HAVE_SECONDARY_RELOADS): Remove.
1024 2007-01-31 Anatoly Sokolov <aesok@post.ru>
1027 * config/avr/avr.c (DWARF2_ADDR_SIZE): Define.
1029 2007-01-31 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1032 * builtins.c (fold_builtin_sqrt): Use MPFR for constant args.
1034 2007-01-31 Zdenek Dvorak <dvorakz@suse.cz>
1036 * cfgloop.h: Include vec-prim.h.
1037 (enum li_flags): Remove LI_ONLY_OLD.
1038 (loop_iterator): Changed.
1039 (fel_next, fel_init): Iterate over loop tree.
1040 (FOR_EACH_LOOP_BREAK): New macro.
1041 * loop-unswitch.c (unswitch_loops): Do not pass LI_ONLY_OLD to
1043 * tree-ssa-loop-unswitch.c (tree_ssa_unswitch_loops): Ditto.
1044 * modulo-sched.c (sms_schedule): Ditto.
1045 * tree-vectorizer.c (vectorize_loops): Ditto.
1046 * doc/loop.texi: Update information on loop numbering and behavior of
1047 FOR_EACH_LOOP wrto new loops.
1048 * tree-scalar-evolution.c (compute_overall_effect_of_inner_loop,
1049 add_to_evolution_1): Test nestedness of loops instead of comparing
1051 * tree-chrec.c (chrec_fold_plus_poly_poly,
1052 chrec_fold_multiply_poly_poly, chrec_evaluate,
1053 hide_evolution_in_other_loops_than_loop, chrec_component_in_loop_num,
1054 reset_evolution_in_loop): Ditto.
1055 * Makefile.in (CFGLOOP_H): Add vecprim.h dependency.
1057 2007-01-31 Dirk Mueller <dmueller@suse.de>
1059 * c-common.c (warn_about_parentheses): Separate warning about
1060 un-parenthized sequence of comparison operators from the one
1061 which is supposed to warn about x <= y <= z.
1063 2007-01-31 Uros Bizjak <ubizjak@gmail.com>
1065 * optabs.h (enum optab_index): Add new OTI_isinf.
1066 (isinf_optab): Define corresponding macro.
1067 * optabs.c (init_optabs): Initialize isinf_optab.
1068 * genopinit.c (optabs): Implement isinf_optab using isinf?f2
1070 * builtins.c (mathfn_built_in): Handle BUILT_IN_ISINF{,F,L}.
1071 (expand_builtin_interclass_mathfn): Expand BUILT_IN_ISINF{,F,L}
1073 (expand_builtin): Expand BUILT_IN_ISINF{,F,L} using
1074 expand_builtin_interclass_mathfn.
1075 * reg_stack.c (subst_stack_regs_pat): Handle UNSPEC_FXAM.
1076 * config/i386/i386.md (UNSPEC_FXAM): New constant.
1077 (fxam<mode>2_i387): New insn pattern.
1078 (isinf<mode>2) New expander to implement isinf, isinff and isinfl
1079 built-in functions as x87 inline asm.
1081 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1083 * gcc/config/arm/unwind-arm.h (_sleb128_t, _uleb128_t): New.
1085 2007-01-30 Eric Christopher <echristo@apple.com>
1087 * config.gcc: Add geode.
1089 2007-01-31 Kazu Hirata <kazu@codesourcery.com>
1091 * cgraphunit.c, config/arm/arm.c, config/m68k/m68k.c,
1092 ipa-inline.c, tree-profile.c, tree-ssa-live.c,
1093 tree-ssa-math-opts.c, tree-ssanames.c, tree-vect-analyze.c,
1094 value-prof.c: Fix comment typos.
1096 2007-01-30 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1099 * doc/invoke.texi (Wpointer-arith): Document warning.
1101 2007-01-30 Janis Johnson
1103 * doc/extend.texi (Decimal Floating Types): Remove decfloat.h from
1104 the list of discrepancies from the draft TR.
1106 2007-01-30 Dirk Mueller <dmueller@suse.de>
1109 * doc/invoke.texi (-Wreturn-type): Update description to
1112 2007-01-30 Richard Sandiford <richard@codesourcery.com>
1114 * cfgrtl.c (try_redirect_by_replacing_jump): Check only_sets_cc0_p.
1116 2007-01-30 Uros Bizjak <ubizjak@gmail.com>
1118 * builtins.c (expand_builtin_int_interclass_roundingfn): New function
1119 to handle optabs that operate on floating point input argument and
1120 output to integer output.
1121 (expand_builtin_mathfn) [BUILT_IN_ILOGB]: Move from here ...
1122 (expand_builtin_interclass_mathfn) [BUILT_IN_ILOGB]: ... to here.
1123 (expand_builtin): Expand BUILT_IN_ILOGB{,F,L} using
1124 expand_builtin_interclass_mathfn ().
1125 * config/i386/i386.md (fxtractxf3_i387): Rename from *fxtractxf3_i387.
1127 (ilogbxf2, ilogb<mode>2): New expanders to implement ilogb, ilogbf and
1128 ilogbl built-in functions as x87 intrinsics.
1130 2007-01-30 Richard Guenther <rguenther@suse.de>
1133 * dwarf2out.c (reference_to_unused): Query varpool if the
1134 variable was output.
1136 2007-01-30 Richard Guenther <rguenther@suse.de>
1139 * passes.c (execute_one_pass): Reset in_gimple_form to not
1140 confuse non-unit-at-a-time mode.
1142 2007-01-29 Roger Sayle <roger@eyesopen.com>
1143 Richard Guenther <rguenther@suse.de>
1145 * fold-const.c (round_up): Make HIGH an unsigned HOST_WIDE_INT to
1146 avoid undefined behaviour on overflow. Use force_fit_type_double
1147 to construct the constant with the specified TREE_OVERFLOW.
1149 2007-01-29 Janis Johnson <janis187@us.ibm.com>
1151 * config/dfp-bit.c: Add parameterized support for fp exceptions.
1152 * config/dfp-bit.h: Ditto.
1154 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1156 * c-decl.c (pop_scope): Replace warnings with call to
1157 warn_for_unused_label.
1158 * c-common.h (warn_for_unused_label): Declare.
1159 * c-common.c (warn_for_unused_label): Define.
1161 2007-01-29 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1163 * tree-optimize.c (update_inlined_to_pointers): Delete unused
1166 2007-01-29 Janis Johnson <janis187@us.ibm.com>
1168 * Makefile.in (USER_H): Remove decfloat.h.
1169 * ginclude/decfloat.h: Delete, moving contents to ...
1170 * ginclude/float.h: Add support for decimal floating point,
1171 guarded by __STDC_WANT_DEC_FP__.
1173 2007-01-29 Mike Stump <mrs@apple.com>
1175 * doc/gccint.texi (Top): Rename Loop Representation to Loop
1176 Analysis and Representation to resolve case insensitive conflict.
1177 * doc/loop.texi (Loop Analysis and Representation): Likewise.
1179 2007-01-28 Daniel Berlin <dberlin@dberlin.org>
1181 * tree.h (struct tree_memory_tag): Add aliases member.
1182 (MTAG_ALIASES): New macro.
1183 * tree-ssa-alias.c (alias_bitmap_obstack): New variable.
1184 (add_may_alias): Remove pointer-set. Update for may_aliases being
1186 (mark_aliases_call_clobbered): Update for may_aliases being a
1188 (compute_tag_properties): Ditto.
1189 (create_partition_for): Ditto.
1190 (compute_memory_partitions): Ditto.
1191 (dump_may_aliases_for): Ditto.
1192 (is_aliased_with): Ditto.
1193 (add_may_alias_for_new_tag): Ditto.
1194 (rewrite_alias_set_for): Rewrite for may_aliases being a bitmap.
1195 (compute_is_aliased): New function.
1196 (compute_may_aliases): Call compute_is_aliased).
1197 (init_alias_info): Initialize alias_bitmap_obstack.
1198 (union_alias_set_into): New function.
1199 (compute_flow_sensitive_aliasing): Use union_aliases_into.
1200 (have_common_aliases_p): Rewrite to take two bitmaps and use
1202 (compute_flow_insensitive_aliasing): Stop using pointer-sets.
1204 (finalize_ref_all_pointers): Update for add_may_alias changes.
1205 (new_type_alias): Ditto.
1206 * tree-flow-inline.h (may_aliases): Return a bitmap.
1207 * tree-dfa.c (dump_variable): Check for MTAG_P'ness.
1208 * tree-ssa.c (verify_flow_insensitive_alias_info): Update for
1209 may_aliases being a bitmap.
1210 * tree-flow.h (struct var_ann_d): Remove may_aliases member.
1211 may_aliases now returns a bitmap.
1212 * tree-ssa-structalias.c (merge_smts_into): Update for may_aliases
1214 * tree-ssa-operands.c (add_virtual_operand): Update for
1215 may_aliases being a bitmap.
1217 2007-01-29 Daniel Berlin <dberlin@dberlin.org>
1219 PR tree-optimization/30630
1220 * tree-ssa-structalias.c (do_complex_constraint): Mark correct
1221 variable as changed.
1223 2007-01-29 Simon Martin <simartin@users.sourceforge.net>
1226 * gimplify.c (gimplify_target_expr): Make sure that the TARGET_EXPR is
1227 expanded only once even if an error occurs.
1229 2007-01-29 Ben Elliston <bje@au.ibm.com>
1231 * gcov-io.h (__gcov_indirect_call_profiler): Declare.
1232 (__gcov_average_profiler): Likewise.
1233 (__gcov_ior_profiler): Likewise.
1234 (__gcov_merge_ior): Likewise.
1236 2007-01-28 Jan Hubicka <jh@suse.cz>
1238 * builtins.c (expand_builtin_memset): Fix typo in my last patch.
1239 * value-prof.c (stringop_block_profile): Likewise.
1241 2007-01-28 Jan Hubicka <jh@suse.cz>
1243 * expr.c (emit_block_move_via_movmem, emit_block_move_via_libcall): Add
1244 variant handling histograms; add wrapper.
1245 (clear_storage_via_libcall): Export.
1246 (emit_block_move_hints): Break out from ...; add histograms.
1247 (emit_block_move): ... this one.
1248 (clear_storage_hints): Break out from ...; add histograms.
1249 (clear_storage): ... this one.
1250 (set_storage_via_memset): Handle histogram.
1251 * expr.h (emit_block_move_via_libcall, emit_block_move_hints): Declare.
1252 (clear_storage_hints, clear_storage_via_libcall): Declare.
1253 (set_storage_via_setmem): Update prototype.
1254 * doc/md.texi (movmem, setmem): Document new arguments.
1256 * value-prof.c (dump_histogram_value, tree_find_values_to_profile): Add
1258 (stringop_block_profile): New global function.
1259 (tree_stringops_values_to_profile): Profile block size and alignment.
1260 * value-prof.h (enum hist_type): add HIST_TYPE_AVERAGE and
1262 (struct profile_hooks): Add gen_average_profiler and gen_ior_profiler.
1263 (stringop_block_profile): Declare.
1264 * builtins.c: Include value-prof.h.
1265 (expand_builtin_memcpy, expand_builtin_memset): Pass block profile.
1266 * gcov-ui.h (GCOV_COUNTER_NAMES): Add new counter.
1267 (GCOV_COUNTER_AVERAGE, GCOV_COUNTER_IOR): New constants.
1268 (GCOV_COUNTERS, GCOV_LAST_VALUE_COUNTER): Update.
1269 * profile.c (instrument_values): Add new counters.
1270 * cfgexpand.c (expand_gimple_basic_block): Propagate histograms to
1272 * tree-profile.c (tree_average_profiler_fn, tree_ior_profiler_fn): New.
1273 (tree_init_edge_profiler): Build new profilers.
1274 (tree_gen_average_profiler, tree_gen_ior_profiler): New.
1275 (pass_tree_profile): Add dump.
1276 (tree_profile_hooks): Update.
1277 * Makefile.in (LIBGCOV): Add new constants.
1278 * libgcov.c (__gcov_merge_ior, __gcov_average_profiler,
1279 __gcov_ior_profiler): New.
1280 * i386.md (movmem/setmem expanders): Add new optional arguments.
1282 2007-01-28 David Edelsohn <edelsohn@gnu.org>
1284 * doc/md.texi (Standard Pattern Names): Document blockage pattern.
1286 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
1288 * tree-ssa-loop-unswitch.c: Include tree-inline.h.
1289 (tree_unswitch_single_loop): Pass eni_size_weights to
1290 tree_num_loop_insns.
1291 * tree-ssa-loop-manip.c: Include tree-inline.h.
1292 (can_unroll_loop_p): Pass eni_size_weights to
1293 tree_num_loop_insns.
1294 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p):
1295 Pass eni_size_weights to estimate_num_insns.
1296 * tree.h (init_inline_once): Export.
1297 * toplev.c (backend_init): Call init_inline_once.
1298 * cgraphunit.c (cgraph_process_new_functions,
1299 cgraph_analyze_function): Pass eni_inlining_weights to
1301 * ipa-inline.c (compute_inline_parameters): Ditto.
1302 * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Pass weights
1303 to estimate_num_insns.
1304 (try_unroll_loop_completely): Pass eni_size_weights to
1305 tree_num_loop_insns.
1306 * tree-eh.c (decide_copy_try_finally): Pass eni_size_weights
1307 ot estimate_num_insns.
1308 * tree-ssa-loop-prefetch.c: Include tree-inline.h.
1309 (loop_prefetch_arrays): Pass eni_time_weights to tree_num_loop_insns.
1310 * tree-inline.c (eni_inlining_weights, eni_size_weights,
1311 eni_time_weights): New variables.
1312 (init_inline_once): Initialize them.
1313 (struct eni_data): Mew.
1314 (estimate_num_insns_1, estimate_num_insns): Use weights.
1315 * tree-inline.h (struct eni_weights_d): New.
1316 (eni_inlining_weights, eni_size_weights, eni_time_weights): Declare.
1317 (estimate_num_insns): Declaration changed.
1318 * cfgloop.h (tree_num_loop_insns): Declaration changed.
1319 * Makefile.in (tree-ssa-loop-unswitch.o, tree-ssa-loop-prefetch.o,
1320 tree-ssa-loop-manip.o): Add TREE_INLINE_H dependency.
1322 2007-01-28 Zdenek Dvorak <dvorakz@suse.cz>
1324 * tree-data-ref.c (conflict_fn): Assert that the number of affine
1325 relations in the conflict function is valid.
1327 2007-01-27 Ian Lance Taylor <iant@google.com>
1329 * common.opt: Add fstrict-overflow.
1330 * opts.c (decode_options): Set flag_strict_overflow if -O2.
1331 * flags.h (TYPE_OVERFLOW_WRAPS): Define.
1332 (TYPE_OVERFLOW_UNDEFINED): Define.
1333 (TYPE_OVERFLOW_TRAPS): Define. This replaces TYPE_TRAP_SIGNED.
1335 * tree.h (TYPE_TRAP_SIGNED): Don't define.
1336 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_UNDEFINED.
1337 (fold_negate_expr): Likewise.
1338 (make_range): Likewise.
1339 (extract_muldiv_1): Likewise.
1340 (maybe_canonicalize_comparison): Likewise.
1341 (fold_comparison): Likewise.
1342 (fold_binary): Likewise.
1343 (tree_expr_nonnegative_p): Likewise.
1344 (tree_expr_nonzero_p): Likewise.
1345 * tree-vrp.c (compare_values): Likewise.
1346 (extract_range_from_binary_expr): Likewise.
1347 (extract_range_from_unary_expr): Likewise.
1348 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness):
1350 (nowrap_type_p): Likewise.
1351 * tree-scalar-evolution.c (simple_iv): Likewise.
1352 * fold-const.c (negate_expr_p): Use TYPE_OVERFLOW_WRAPS.
1353 (build_range_check): Likewise.
1354 (extract_muldiv_1): Likewise.
1355 (fold_comparison): Likewise.
1356 * tree-vrp.c (vrp_int_const_binop): Likewise.
1357 (extract_range_from_unary_expr): Likewise.
1358 * convert.c (convert_to_integer): Likewise.
1359 * fold-const.c (fold_negate_expr): Use TYPE_OVERFLOW_TRAPS.
1360 (fold_comparison): Likewise.
1361 (fold_binary): Likewise.
1362 * optabs.c (optab_for_tree_code): Likewise.
1363 * tree-vectorizer.c (vect_is_simple_reduction): Likewise.
1364 * simplify-rtx.c (simplify_const_relational_operation): Check
1365 flag_strict_overflow and flag_trapv.
1366 (simplify_const_relational_operation): Likewise.
1367 * doc/invoke.texi (Option Summary): Mention -fstrict-overflow.
1368 (Optimize Options): Add -fstrict-overflow to -O2 list. Document
1371 2007-01-27 Roger Sayle <roger@eyesopen.com>
1373 * tree.c (tree_fold_gcd): Delete.
1374 * tree.h (tree_fold_gcd): Remove prototype.
1375 * tree-data-ref.c (tree_fold_divides_p): Don't use tree_fold_gcd to
1376 test whether one constant integer is a multiple of another. Instead
1377 call int_const_binop with TRUNC_MOD_EXPR and test for a zero result.
1378 * fold-const.c (multiple_of_p): We've determined both TOP and
1379 BOTTOM are integer constants so we can call int_const_binop directly
1380 instead of the more generic const_binop.
1382 2007-01-27 Roger Sayle <roger@eyesopen.com>
1384 * fold-const.c (size_binop): In the fast-paths for X+0, 0+X, X-0 and
1385 1*X check that the constant hasn't overflowed, to preserve the
1387 (round_up): Provide an efficient implementation when rouding-up an
1388 INTEGER_CST to a power-of-two.
1390 2007-01-28 Ralf Wildenhues <Ralf.Wildenhues@gmx.de>
1392 * doc/sourcebuild.texi: Add comma for clarity.
1393 * doc/extend.texi: Fix some typos.
1394 * doc/passes.texi: Likewise.
1395 * doc/cppinternals.texi: Likewise.
1396 * doc/c-tree.texi: Likewise.
1397 * doc/tree-ssa.texi: Likewise.
1398 * doc/install.texi: Likewise.
1400 2007-01-27 Jan Hubicka <jh@suse.cz>
1402 * tree-sra.c (sra_walk_function): Don't rely on aliases being build.
1403 (pass_sra): Do not require alias information.
1404 * passes.c (init_optimization_passes): Add SRA
1406 2007-01-27 Steven Bosscher <steven@gcc.gnu.org>
1408 * tracer.c (rest_of_handle_tracer): We already cleaned
1409 up the CFG in tracer() so don't do it here again.
1410 * cfgcleanup.c (rest_of_handle_jump2): Don't repeat
1411 cleanup_cfg here, either. And don't call renumber_insns.
1413 * cfgrtl.c (rtl_verify_flow_info_1): Don't verify that BB_END
1414 and BB_HEAD are in the insn stream here. Instead make sure
1415 that BB_INSN is valid on all insns. Also, do check here that
1416 there are no pending branch predictions...
1417 (rtl_verify_flow_info): ...instead of doing it here. Checks
1418 for BB_END and BB_HEAD moved from rtl_verify_flow_info_1 to
1421 2007-01-26 Roger Sayle <roger@eyesopen.com>
1423 * config/i386/i386.c (ix86_swap_binary_operands_p): New helper
1424 function to simplify/factorize operand order canonicalization.
1425 (ix86_fixup_binary_operands): Reorganize using the above function.
1426 (ix86_binary_operator_ok): Likewise.
1428 2007-01-27 Jakub Jelinek <jakub@redhat.com>
1430 * genattrtab.c (struct attr_value_list, insn_code_values): Move to
1431 file scope from optimize_attrs.
1432 (simplify_test_exp): If insn_code_values is not NULL, use it to speed
1434 (optimize_attrs): Clear insn_code_values after freeing it.
1436 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
1438 * tree-ssa-address.c (create_mem_ref): Remove ", bsi" from
1439 a parts.base assignment.
1441 2007-01-26 Zdenek Dvorak <dvorakz@suse.cz>
1443 * tree-data-ref.c (dump_subscript): Use dump_conflict_function.
1444 (compute_subscript_distance, initialize_data_dependence_relation,
1445 finalize_ddr_dependent, analyze_ziv_subscript,
1446 analyze_siv_subscript_cst_affine,
1447 compute_overlap_steps_for_affine_univar,
1448 compute_overlap_steps_for_affine_1_2, analyze_subscript_affine_affine,
1449 analyze_siv_subscript, analyze_miv_subscript,
1450 analyze_overlapping_iterations, subscript_dependence_tester_1,
1451 compute_self_dependence, free_dependence_relation): Work
1452 with affine_fn instead of chrecs.
1453 (dump_affine_function, dump_conflict_function, affine_function_equal_p,
1454 common_affine_function, affine_function_base,
1455 affine_function_constant_p, affine_fn_op, affine_fn_plus,
1456 affine_fn_minus, affine_fn_free, conflict_fn_not_known,
1457 conflict_fn_no_dependence, free_conflict_function, free_subscripts,
1458 conflict_fn, affine_fn_cst, affine_fn_univar): New functions.
1459 (all_chrecs_equal_p): Removed.
1460 * tree-data-ref.h (affine_fn, conflict_function): New types.
1461 (struct subscript): Change type of conflicting_iterations_in_a
1462 and conflicting_iterations_in_b.
1464 2007-01-26 Steve Ellcey <sje@cup.hp.com>
1467 * config/pa/pa.h (TARGET_HPUX_11): New.
1468 * config/pa/pa-hpux11.h (TARGET_HPUX_11): New.
1469 * config/pa/pa.c (pa_init_builtins): Use TARGET_HPUX_11.
1471 2007-01-26 Daniel Berlin <dberlin@dberlin.org>
1472 Richard Guenther <rguenther@suse.de>
1474 * tree-ssa-structalias.c (solve_graph): Handle case
1475 we merged the variable to another.
1477 2007-01-25 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1479 * builtins.c (fold_builtin_1): Treat ccos and ccosh as 'even'
1482 * fold-const.c (negate_mathfn_p): Treat casin, casinh, catan,
1483 catanh, cproj, csin, csinh, ctan and ctanh as 'odd' functions.
1485 2007-01-25 DJ Delorie <dj@redhat.com>
1487 * config/m32c/m32c.c (m32c_cannot_change_mode_class): We don't
1488 allow changes to modes which don't fit in those registers.
1490 * reload1.c (choose_reload_regs): Check for invalid subregs before
1491 computing their locations, not after.
1493 2007-01-25 Geoffrey Keating <geoffk@apple.com>
1496 * config/rs6000/rs6000.c (first_altivec_reg_to_save): On Darwin,
1497 save Altivec registers in an eh_return function.
1498 (compute_vrsave_mask): Likewise.
1499 (rs6000_stack_info): Correct AIX/Darwin stack alignment computation
1500 for saving Altivec registers.
1501 (rs6000_emit_prologue): Don't allocate stack twice in
1502 eh_return function. Correct expected value of altivec_save_offset
1503 when using save_world. Describe save of R0 to stack when using
1504 save_world. Describe stack pointer adjustment when using
1505 save_world. Remove duplicated eh_return parameter register saving.
1506 Update sp_offset variable after save_world.
1507 * config/rs6000/t-darwin (LIB2FUNCS_STATIC_EXTRA): Remove
1509 (LIB2FUNCS_EXTRA): Add darwin-world.asm.
1510 * config/rs6000/darwin.h (SUBTARGET_OVERRIDE_OPTIONS): -m64
1513 2007-01-25 Steve Ellcey <sje@cup.hp.com>
1515 * config.gcc (ia64*-*-hpux*): Make posix threads the default.
1517 2007-01-25 Steve Ellcey <sje@cup.hp.com>
1520 * config/pa/pa.c (pa_init_builtins): Set asm names for finite routines.
1521 * config/ia64/ia64.c (ia64_init_builtins): Ditto.
1523 2007-01-25 Richard Guenther <rguenther@suse.de>
1525 * doc/invoke.texi (-Wcoverage-mismatch): Document.
1526 * common.opt (-Wcoverage-mismatch): New warning option.
1527 * coverage.c (get_coverage_counts): Ignore coverage mismatch
1528 if -Wcoverage-mismatch is given.
1530 2007-01-25 Razya Ladelsky <razya@il.ibm.com>
1532 * ipa-cp.c (ipcp_insert_stage, ipcp_driver): Support for SSA.
1533 (ipcp_driver): Change to static definition.
1534 Add dumping of the ifunctions.
1535 (constant_val_insert): Remove unused parameter. Support for SSA.
1536 (ipcp_propagate_const): Support for SSA.
1537 (ipcp_profile_bb_print): Print only analyzed nodes.
1538 (ipcp_replace_map_create): Remove support for Fortran constant
1540 * ipa-prop.c (ipa_method_modify_stmt,
1541 ipa_callsite_compute_param): Support for SSA.
1542 * ipa-prop.h (ipcp_driver): Remove declaration.
1543 (IS_VALID_TREE_MAP_INDEX): Add define.
1545 2007-01-24 Geoffrey Keating <geoffk@apple.com>
1547 * unwind-dw2.c (execute_stack_op): Handle DW_OP_swap.
1549 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1552 * builtins.c (fold_builtin_cabs): Use MPFR to evaluate a
1553 constant argument to cabs and do it without checking for
1554 -funsafe-math-optimizations.
1556 2007-01-24 Douglas Gregor <dgregor@osl.iu.edu>
1558 * c-common.h (RID_FIRST_CXX0X): New.
1559 (RID_LAST_CXX0X): New.
1560 * c-opts.c (c_common_handle_option): -Wc++0x-compat is triggered
1562 * c.opt (Wc++0x-compat): New.
1563 * doc/invoke.texi (-Wc++0x-compat): Document.
1565 2007-01-24 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1567 * builtins.c (fold_builtin_carg): New.
1568 (fold_builtin_1): Use it.
1570 2007-01-24 Jan Hubicka <jh@suse.cz>
1572 * ipa-inline.c (cgraph_decide_inlining): Initialize initial_insns.
1574 * ipa-inline.c (initial_insns, max_insns): Delete.
1575 (compute_max_insns): New function.
1576 (cgraph_decide_inlining_of_small_function): Use it; take minimal amount
1577 of insns as base for code growth.
1578 (cgraph_decide_inlining): Make initial_insns local; do not compute
1580 * params.def (PARAM_INLINE_UNIT_GROWTH): Set to 60.
1581 * doc/invoke.texi (inline-unit-growth): Update docs.
1583 2007-01-24 Jakub Jelinek <jakub@redhat.com>
1585 * config/i386/i386.h (x86_cmpxchg16b): Remove const.
1586 (TARGET_CMPXCHG16B): Define to x86_cmpxchg16b.
1587 * config/i386/i386.c (x86_cmpxchg16b): Remove const.
1588 (override_options): Add PTA_CX16 flag. Set x86_cmpxchg16b
1589 for CPUs that have PTA_CX16 set.
1592 * gimplify.c (omp_check_private): New function.
1593 (gimplify_scan_omp_clauses): Use it for
1594 firstprivate/lastprivate/reduction.
1597 * gimplify.c (omp_add_variable): Don't call omp_notice_variable
1598 on TYPE_SIZE_UNIT for GOVD_LOCAL VLAs.
1601 * omp-low.c (lower_omp_for_lastprivate): Add dlist argument.
1602 If lower_lastprivate_clauses emits some statements, append them
1603 to dlist rather than body_p and to body_p append an initializer.
1604 (lower_omp_for): Adjust caller.
1606 2007-01-24 Steve Ellcey <sje@cup.hp.com>
1608 * target.h (globalize_decl_name): New.
1609 * target-def.h (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
1610 * output.h (default_globalize_decl_name): New.
1611 * varasm.c (asm_output_bss): Use globalize_decl_name instead of
1613 (globalize_decl): Ditto.
1614 (default_globalize_decl_name): New.
1615 * config/ia64/ia64.c (ia64_globalize_decl_name): New.
1616 (ia64_handle_version_id_attribute): New.
1617 (TARGET_ASM_GLOBALIZE_DECL_NAME): New.
1618 (ia64_asm_output_external): Use globalize_decl_name instead
1620 * doc/extend.texi (version_id): New pragma.
1621 * doc/tm.texi (ARGET_ASM_GLOBALIZE_DECL_NAME): New target hook.
1623 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
1625 * unwind-dw2-fde.c (get_cie_encoding): Replaced _Unwind_Word with
1626 _uleb128_t and _Unwind_SWord with _sleb128_t.
1627 * unwind-dw2.c (extract_cie_info, execute_stack_op, execute_cfa_program,
1628 uw_frame_state_for, uw_update_context_1): Likewise.
1629 * unwind-c.c (parse_lsda_header, PERSONALITY_FUNCTION): Likewise.
1630 * unwind-pe.h (read_uleb128, read_sleb128,
1631 read_encoded_value_with_base): Likewise.
1632 * unwind-generic.h: Define _sleb128_t and _uleb128_t types.
1634 2007-01-24 Richard Guenther <rguenther@suse.de>
1636 * builtins.c (expand_builtin_cexpi): Get the fndecl
1637 for cexp in the correct way.
1639 2007-01-24 Jan Hubicka <jh@suse.cz>
1641 * tree-ssa-dce.c (eliminate_unnecesary_stmts): Remove dead LHS of calls.
1643 2007-01-24 Andreas Krebbel <krebbel1@de.ibm.com>
1645 * c-cppbuiltin.c (builtin_define_type_sizeof): New function.
1646 (c_cpp_builtins): New builtin macros: __SIZEOF_INT__, __SIZEOF_LONG__,
1647 __SIZEOF_LONG_LONG__, __SIZEOF_SHORT__, __SIZEOF_POINTER__,
1648 __SIZEOF_FLOAT__, __SIZEOF_DOUBLE__, __SIZEOF_LONG_DOUBLE__,
1649 __SIZEOF_SIZE_T__, __SIZEOF_WCHAR_T__, __SIZEOF_WINT_T__ and
1650 __SIZEOF_PTRDIFF_T__.
1651 * doc/cpp.texi: Documentation for the new builtin macros added.
1653 2007-01-24 Uros Bizjak <ubizjak@gmail.com>
1655 * config/i386/i386.md (tanxf2, tan<mode>2, atan<mode>2, log<mode>2,
1656 log10<mode>2, log2<mode>2, expxf2, exp10xf2, exp2xf2): Use op2
1657 instead of operands[2] to avoid access past the end of array.
1659 2007-01-24 Richard Sandiford <richard@codesourcery.com>
1661 * reload1.c (emit_reload_insns): Pass the reload register
1662 for a non-spill output reload through forget_old_reloads_1.
1664 2007-01-23 Joseph Myers <joseph@codesourcery.com>
1666 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Define
1667 _SOFT_DOUBLE if doubles use software floating-point.
1668 * config/rs6000/libgcc-ppc-glibc.ver: Export additional long
1669 double functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1670 * config/rs6000/darwin-ldouble.c: Also compile functions for
1671 hard-float without FPRs. Use fmsub function for all __NO_FPRS__
1672 cases. Compile extra functions if _SOFT_DOUBLE, not _SOFT_FLOAT.
1673 * config/rs6000/linuxspe.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Remove
1674 commented-out long double override.
1675 (CPP_LONGDOUBLE_DEFAULT_SPEC): Likewise.
1676 * config/rs6000/eabispe.h: Likewise.
1677 * config/rs6000/rs6000.c (rs6000_override_options): Don't override
1678 long double for non-SPE.
1679 (rs6000_handle_option): Likewise.
1680 (invalid_e500_subreg): Disallow more subregs involding DImode,
1681 DFmode, TImode or TFmode.
1682 (rs6000_legitimate_offset_address_p): Check TFmode offsets for
1684 (legitimate_lo_sum_address_p): Also check for TFmode for E500
1686 (rs6000_legitimize_address): Also handle TFmode for E500 double.
1687 (rs6000_legitimize_reload_address): Also handle TFmode for E500
1689 (rs6000_legitimate_address): Also check for TFmode for E500
1691 (rs6000_emit_move): Use DFmode subregs of TFmode for E500 double.
1692 (spe_build_register_parallel): Handle TFmode and TCmode.
1693 (rs6000_spe_function_arg): Handle TFmode and TCmode for E500
1695 (function_arg): Handle TFmode and TCmode for E500 double.
1696 (rs6000_init_libfuncs): Initialize extra libfuncs for soft double
1698 (print_operand): Handle TFmode and TImode for %y.
1699 (rs6000_generate_compare): Handle TFmode comparisons for E500
1701 (spe_func_has_64bit_regs_p): Check for TFmode for E500 double.
1702 (rs6000_function_value): Handle TFmode and TCmode for E500 double.
1703 (rs6000_libcall_value): Handle TFmode and TCmode for E500 double.
1704 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Check for
1705 TFmode for E500 double.
1706 * config/rs6000/rs6000.md (FP): Allow TF for E500 double.
1707 (floatsidf2): Enable for E500 double.
1708 (movtf_softfloat): Use rs6000_nonimmediate_operand.
1709 (extenddftf2): Change to extenddftf2_fprs.
1710 (extenddftf2): Call gen_spe_extenddftf2 or gen_extenddftf2_fprs
1711 depending on TARGET_E500_DOUBLE.
1712 (extendsftf2): Enable for E500 double.
1713 (trunctfdf2): Enable for E500 double.
1714 (trunctfsf2): Change to trunctfsf2_fprs.
1715 (trunctfsf2): Call gen_spe_trunctfsf2 or gen_trunctfsf2_fprs
1716 depending on TARGET_E500_DOUBLE.
1717 (floatsitf2): Enable for E500 double.
1718 (fix_trunctfsi2): Change to fix_trunctfsi2_fprs.
1719 (fix_trunctfsi2): Call gen_spe_fix_trunctfsi2 or
1720 gen_fix_trunctfsi2_fprs depending on TARGET_E500_DOUBLE.
1721 (negtf2): Change to negtf2_internal.
1722 (negtf2): New expander.
1723 (abstf2): Enable for E500 double. Call gen_spe_abstf2_tst,
1724 gen_spe_abstf2_cmp or gen_abstf2_internal depending on
1725 TARGET_E500_DOUBLE and flag_unsafe_math_optimizations.
1726 (movdi_internal32): Use rs6000_nonimmediate_operand.
1727 (unnamed splitter): Likewise.
1728 * config/rs6000/spe.md (CMPTFEQ_GPR, TSTTFEQ_GPR, CMPTFGT_GPR,
1729 TSTTFGT_GPR, CMPTFLT_GPR, TSTTFLT_GPR): New unspecs.
1730 (SPE64TF, DITI): New mode macros.
1731 (frob_df_di): Change to frob_<SPE64:mode>_<DITI:mode>; allow more
1734 (frob_<mode>_di_2): New.
1735 (frob_tf_di_8_2): New.
1736 (frob_di_df): Change to frob_di_<mode>; allow more modes.
1738 (frob_di_df_2): Change to frob_<DITI:mode>_<SPE64:mode>_2; allow
1740 (frob_ti_<mode>_8_2): New.
1741 (frob_ti_tf_2): New.
1742 (mov_si<mode>_e500_subreg0, mov_si<mode>_e500_subreg0_2,
1743 mov_si<mode>_e500_subreg4, mov_si<mode>_e500_subreg4_2): Allow
1745 (mov_sitf_e500_subreg8, mov_sitf_e500_subreg8_2,
1746 mov_sitf_e500_subreg12, mov_sitf_e500_subreg12_2): New.
1747 (spe_trunctfdf2_internal1, spe_trunctfsf2, spe_extenddftf2,
1748 spe_fix_trunctfsi2, spe_fix_trunctfsi2_internal,
1749 spe_negtf2_internal, spe_abstf2_cmp, spe_abstf2_tst): New.
1750 (cmptfeq_gpr, tsttfeq_gpr, cmptfgt_gpr, tsttfgt_gpr, cmptflt_gpr,
1753 2007-01-23 Ian Lance Taylor <iant@google.com>
1755 * Makefile.in (OBJS-common): Reformat, alphabetize, but put
1757 (OBJS-archive): Reformat, alphabetize.
1758 (OBJS): Change out_object_file to OBJS-md.
1760 2007-01-23 Uros Bizjak <ubizjak@gmail.com>
1762 * config/i386/i386.md: Use REG_P, MEM_P, CONST_INT_P, LABEL_P,
1763 JUMP_P and CALL_P predicates where applicable.
1764 * config/i386/i386.c: Ditto.
1765 * config/i386/i386.md: Ditto.
1766 * config/i386/mmx.md: Ditto.
1767 * config/i386/predicates.md: Ditto.
1769 2007-01-22 Andreas Schwab <schwab@suse.de>
1771 * config/m68k/m68k.h: Fix comment.
1773 2007-01-22 Jan Hubicka <jh@suse.cz>
1775 * passes.c (init_optimization_passes): Do not rerun
1776 pass_early_warn_uninitialized.
1778 2007-01-22 Richard Guenther <rguenther@suse.de>
1780 PR tree-optimization/30038
1781 * tree-ssa-math-opts.c (maybe_record_sincos): New static helper
1783 (execute_cse_sincos_1): Likewise.
1784 (execute_cse_sincos): Likewise.
1785 (gate_cse_sincos): Likewise.
1786 (pass_cse_sincos): New pass CSEing sin() and cos() calls using
1787 the cexpi() canonicalization of sincos().
1788 * tree-pass.h (pass_cse_sincos): Declare.
1789 * passes.c (init_optimization_passes): New pass pas_cse_sincos.
1791 2007-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
1793 PR rtl-optimization/29329
1794 * combine.c (replaced_rhs_insn): Rename to i2mod.
1795 (replaced_rhs_value): Rename to i2mod_new_rhs.
1796 (i2mod_old_rhs): New global variable.
1797 (combine_instructions): Adjust for above change. Save a copy of
1798 the old RHS into i2mod_old_rhs when the contents of a REG_EQUAL
1799 note are substituted in the second instruction.
1800 (distribute_notes) <REG_DEAD>: Adjust for above change. Do not
1801 ditch the note if it pertains to the second eliminated register
1802 and this register is mentioned in i2mod_old_rhs.
1805 2006-09-12 Eric Botcazou <ebotcazou@libertysurf.fr>
1807 * combine.c (distribute_notes) <REG_DEAD>: Do not consider SETs past
1808 the insn to which the note was originally attached.
1810 2007-01-21 Jan Hubicka <jh@suse.cz>
1812 * ipa-inline.c (inlining_mode): Comment, move up.
1813 (cgraph_decide_inlining_incrementally): Do not perform inlining
1814 itself; fix handling of flattening of self recursive functions.
1815 (cgraph_find_cycles): Remove.
1816 (cgraph_flatten_node): Remove.
1817 (cgraph_decide_inlining): Use incremental inliner to handle flattening.
1818 (try_inline): New function.
1819 (cgraph_early_inlining): Update call of
1820 cgraph_decide_inlining_incrementally. Apply inlining here.
1821 (apply_inline): Update call of cgraph_decide_inlining_incrementally.
1823 2007-01-21 Dirk Mueller <dmueller@suse.de>
1826 * tree-vrp.c (check_array_bounds): do not warn
1827 about ADDR_EXPR's of ARRAY_REF's which are immediately
1828 used in binary expressions.
1830 2007-01-21 Ira Rosen <irar@il.ibm.com>
1832 * tree-vectorizer.h (struct _stmt_vec_info): Add new field
1833 read_write_dep and macros for its access.
1834 * tree-vectorizer.c (new_stmt_vec_info): Initialize the new field.
1835 * tree-vect-analyze.c (vect_analyze_data_ref_dependence): Remove
1836 argument, call vect_check_interleaving for every independent pair of
1837 data-refs. Mark loads that access the same memory location as a store
1839 (vect_check_dependences): Remove.
1840 (vect_analyze_data_ref_dependences): Remove vect_check_dependences
1841 call, fix the call to vect_analyze_data_ref_dependence.
1842 (vect_analyze_data_ref_access): For statements that access the same
1843 data-ref, check that they are not stores; for loads, check that there
1844 is no store that access the same location.
1846 2007-01-20 Roger Sayle <roger@eyesopen.com>
1847 Joseph Myers <joseph@codesourcery.com>
1849 * doc/invoke.texi (-fdump-rtl-combine): Fix under/overfull hbox.
1850 (-fdump-rtl-stack): Likewise.
1851 (-fno-signed-zeros): Use @minus{} for a minus sign. Correct typo.
1852 (-mcheck-zero-division, -mcpu): Fix under/overfull hbox.
1853 (-mpt-fixed): Use @minus{} for minus sign.
1854 (Using Precompiled Headers): Fix under/overfull hbox.
1856 2007-01-20 Jan Hubicka <jh@suse.cz>
1858 * tree-flow.h (struct stmt_ann_d): Move references_memory to proper
1859 place within annotation.
1861 2007-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1863 * pa.c (output_move_double): Change array size of xoperands to 4.
1865 2007-01-20 Jan Hubicka <jh@suse.cz>
1867 * tree-tailcall.c (adjust_return_value): Do not use RESULT_DECL
1870 2007-01-19 Ian Lance Taylor <iant@google.com>
1872 * expmed.c (expand_divmod) [TRUNC_MOD_EXPR, TRUNC_DIV_EXPR]: Cast
1873 constant to unsigned HOST_WIDE_INT before negating.
1875 2007-01-19 Ian Lance Taylor <iant@google.com>
1877 * tree-ssa-operands.h (struct vuse_vec_d): Change num_vuse field
1879 (VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Use unsigned comparison.
1880 (VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1881 (SET_VUSE_VECT_ELEMENT) [ENABLE_CHECKING]: Likewise.
1882 (SET_VUSE_ELEMENT_VAR) [ENABLE_CHECKING]: Likewise.
1883 (SET_VUSE_ELEMENT_PTR) [ENABLE_CHECKING]: Likewise.
1884 (realloc_vdef, realloc_vuse): Change second parameter to
1886 (ssa_operand_iterator_d): Change vuse_index and mayuse_index
1888 * tree-ssa-operands.c (realloc_vop): Change num_elem parameter to
1889 unsigned. Change x and lim locals to unsigned.
1890 (realloc_vdef, realloc_vuse): Change num_elem parameter to
1892 (finalize_ssa_vuse_ops): Change old_i local to unsigned.
1893 (copy_virtual_operands): Change i and n locals to unsigned.
1895 2007-01-19 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1898 * builtins.c (fold_builtin_1): Handle builtin fdim.
1900 2007-01-20 Jan Hubicka <jh@suse.cz>
1902 * tree-ssa.c (init_tree_ssa): Do not call init_alias_heapvars.
1903 * tree-ssa-structalias.c (compute_points_to_sets): Do call
1904 init_alias_heapvars.
1905 (init_alias_heapvars): Initialize only when not already initialized.
1906 (delete_alias_heapvars): Set heapvar_for_stmt to NULL.
1908 2007-01-19 Roger Sayle <roger@eyesopen.com>
1910 * common.opt (fsigned-zeros): New command line option.
1911 * flags.h (HONOR_SIGNED_ZEROS): Control via flag_signed_zeros instead
1912 of flag_unsafe_math_optimizations.
1913 * opts.c (set_fast_math_flags): The -ffast-math command line option
1914 implies -fno-signed-zeros.
1915 (fast_math_flags_set_p): Likewise.
1917 * doc/invoke.texi: Document new -fno-signed-zeros option, and update
1918 the documentation of -ffast-math appropriately. Wrap long lines.
1920 2007-01-19 Steve Ellcey <sje@cup.hp.com>
1922 * system.h (ASM_MAKE_LABEL_LINKONCE): Poison.
1923 * varasm.c (globalize_decl): Remove ASM_MAKE_LABEL_LINKONCE ifdef.
1925 2007-01-19 Tomas Bily <tbily@suse.cz>
1927 * cgraphunit.c (cgraph_finalize_function): Updating of pid
1929 (tree_init_ic_make_global_vars): New function
1930 (tree_init_edge_profiler): call of tree_init_ic_make_global_vars
1931 (tree_gen_ic_profiler): New function
1932 (tree_gen_ic_func_profiler): New function
1933 (tree_profiling): Added calling of tree_gen_ic_func_profiler
1934 (tree_profile_hooks): Added hook for indirec/virtual calls
1935 * value-prof.c (tree_find_values_to_profile): New case for
1937 (tree_values_to_profile): Call for determining indirect/virtual
1939 (tree_indirect_call_to_profile): New function
1940 (tree_ic_transform): New function
1941 (tree_ic): New function
1942 (find_func_by_pid): New function
1943 (init_pid_map): New function
1944 (tree_value_profile_transformations): Added check for
1945 indirect/virtual call transformation
1946 * value-prof.h (enum hist_type): New counter type for
1947 indirect/virtual calls
1948 (profile_hooks): Added new hook for profiling indirect/virtual
1950 * profile.c (instrument_values): New case for indirect/virtual
1952 * gcov-io.h (GCOV_LAST_VALUE_COUNTER): Changed to 6
1953 (GCOV_COUNTER_V_INDIR): New counter type
1954 (GCOV_COUNTER_NAMES): New name of counter "indirect" added
1955 (GCOV_MERGE_FUNCTIONS): New merge function for indirect/virtual
1957 * cgraph.c: Definition of cgraph_max_pid
1958 (cgraph_create_node): Default init of pid attribute
1959 * cgraph.h: Declaration of cgraph_max_pid
1960 (struct cgraph_node): Added pid attribute
1961 * libgcov.c (__gcov_indirect_call_profiler): New function
1962 (__gcov_one_value_profiler_body): New function
1963 (__gcov_one_value_profiler): Body was moved to
1964 __gcov_one_value_profiler_body and calls it
1966 2007-01-19 Basile Starynkevitch <basile@starynkevitch.net>
1968 * doc/gty.texi (Options): Document the mark_hook option to GTY.
1969 * gengtype.c (write_types_data, write_func_for_structure,
1970 write_types, ggc_wtd, pch_wtd): Add skip_hooks to
1971 write_types_data, ggc_wtd, pch_wtd for processing mark_hook.
1972 (walk_type, write_func_for_structure): Generate the mark_hook if
1975 2007-01-19 Jan Hubicka <jh@suse.cz>
1977 * ipa-inline.c (cgraph_decide_inlining_incrementally): Instead of
1978 'early' argument take inlining mode argument specifying whether to
1979 inline for size/speeed or all functions; add support for flattening;
1981 (cgraph_early_inlining): Update call of decide_inlining_incrementally.
1983 2007-01-19 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1986 * toplev.c (warn_deprecated_use): Use %qD instead of %qs to print
1987 the name of the declared identifier.
1989 2007-01-19 Dirk Mueller <dmueller@suse.de>
1991 * config/i386.h (CONDITIONAL_REGISTER_USAGE): Store
1992 result of PIC_OFFSET_TABLE_REGNUM in temporary variable to avoid
1993 duplicate evaluation.
1995 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
1997 * config/i386/i386.md (acos<mode>2): Rename from acossf2 and acosdf2.
1998 Macroize expander using X87MODEF12 mode macro. Extend operand 1
1999 to XFMode, use acosxf2 and truncate result to requested mode.
2000 Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
2001 (asin<mode>2): Similarly, with asin expanders.
2002 (*fscalexf4_i387): Rename from *fscalexf4.
2003 (expNcorexf3): New expander.
2004 (expxf2, exp10xf2, exp2xf2): Use expNcorexf3 expander.
2005 (exp<mode>2): Rename from expsf2 and expdf2. Macroize expander using
2006 X87MODEF12 mode macro. Extend operand 1 to XFMode, use expxf2 and
2007 truncate result to requested mode. Use SSE_FLOAT_MODE_P to disable
2008 patterns for SSE math.
2009 (exp10<mode>2): Similarly, with exp10 expanders.
2010 (exp2<mode>2): Similarly, with exp2 expanders.
2011 (expm1<mode>2): Similarly, with expm1 expanders.
2012 (ldexp<mode>3): Similarly, with ldexp expanders.
2013 (log<mode>2, log10<mode>2, log2<mode>2, log1p<mode>2, logb<mode>2):
2014 Use gen_truncxf<mode>2_i387_noop to truncate result.
2016 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2018 * config/m68k/m68k.h (PREFERRED_STACK_BOUNDARY): Define to 32
2019 for ColdFire targets.
2021 2007-01-19 Nathan Sidwell <nathan@codesourcery.com>
2022 Richard Sandiford <richard@codesourcery.com>
2024 * config/m68k/m68k.h (M68K_STATIC_CHAIN_REG_NAME): New macro.
2025 (INITIALIZE_TRAMPOLINE): Use STATIC_CHAIN_REGNUM.
2026 (__transfer_from_trampoline): Use M68K_STATIC_CHAIN_REG_NAME.
2027 * config/m68k/m68kelf.h (STATIC_CHAIN_REGNUM): Override.
2028 (M68K_STATIC_CHAIN_REG_NAME): Likewise.
2029 * config/m68k/netbsd-elf.h (M68K_STATIC_CHAIN_REG_NAME): Likewise.
2031 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2033 * config/m68k/m68k.md (adddi_dilshr32): Rename to...
2034 (*adddi_dilshr32): ...this. Fix formatting. Remove commented-out
2035 non-canonical pattern. Restrict to !TARGET_COLDFIRE.
2036 (*adddi_dilshr32_cf): New pattern.
2037 (adddi3, subdi3): Remove first alternatives.
2039 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2041 * config/m68k/m68k.c (notice_update_cc): If an SFmode move is
2042 implemented using move.l, do not use its cc result for floating-point
2045 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2047 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Stop the 'T' constraint
2048 from accepting 's' constraints if flag_pic.
2050 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2052 * config/m68k/m68k.md (bordered, bunordered, buneq, bunge, bungt)
2053 (bunle, bunlt, bltgt, bordered_rev, bunordered_rev, buneq_rev)
2054 (bunge_rev, bungt_rev, bunle_rev, bunlt_rev, bltgt_rev): Change
2055 condition from TARGET_68881 to TARGET_HARD_FLOAT.
2057 2007-01-19 Sandra Loosemore <sandra@codesourcery.com>
2059 * longlong.h (count_leading_zeros, COUNT_LEADING_ZEROS_0): Add
2060 ColdFire alternatives.
2061 * config/m68k/m68k.h (CLZ_DEFINED_VALUE_AT_ZERO): New macro.
2062 * config/m68k/m68k.md (clzsi2): Define for ColdFire
2063 architectures that support ff1 instruction.
2065 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2066 Julian Brown <julian@codesourcery.com>
2068 * config/m68k/m68k.h (CONST_OK_FOR_LETTER_P): Add an 'R' case.
2069 * config/m68k/m68k.md (*movsi_cfv4): Fold into...
2070 (*movsi_cf): ...here. Remove unnecessary 'R' from 'Rg'.
2073 2007-01-19 Richard Sandiford <richard@codesourcery.com>
2075 * config/m68k/m68k-protos.h (valid_mov3q_const): Take a HOST_WIDE_INT
2077 (output_move_const_into_data_reg, output_move_simode_const): Delete.
2078 * config/m68k/m68k.c (const_method, const_int_cost): Take a
2079 HOST_WIDE_INT instead of an rtx.
2080 (m68k_rtx_costs): Update call accordingly.
2081 (output_move_const_into_data_reg): Likewise. Fix formatting.
2082 (valid_mov3q_const): Take a HOST_WIDE_INT instead of an rtx.
2084 (output_move_simode_const): Update calls after above changes.
2085 Rework to use automatic variables and predicates like MEM_P.
2086 * config/m68k/m68k.md (pushexthisi_const): Update call to
2089 2007-01-19 Dirk Mueller <dmueller@suse.de>
2091 * tree-ssa-alias.c (perform_var_substitution): Fix typo
2094 2007-01-19 Richard Guenther <rguenther@suse.de>
2096 * builtins.c (expand_builtin_cexpi): Fall back to expanding
2097 via cexp in case sincos is not available.
2099 2007-01-19 Richard Guenther <rguenther@suse.de>
2101 * doc/tm.texi (TARGET_HAS_SINCOS): Document new target macro.
2102 * defaults.h (TARGET_HAS_SINCOS): Default to off.
2103 * config/linux.h (TARGET_HAS_SINCOS): Set to on if we have glibc.
2104 * config/alpha/linux.h (TARGET_HAS_SINCOS): Likewise.
2105 * config/sparc/linux.h (TARGET_HAS_SINCOS): Likewise.
2106 * config/sparc/linux64.h (TARGET_HAS_SINCOS): Likewise.
2107 * config/rs6000/linux.h (TARGET_HAS_SINCOS): Likewise.
2108 * config/rs6000/linux64.h (TARGET_HAS_SINCOS): Likewise.
2110 2007-01-19 Uros Bizjak <ubizjak@gmail.com>
2112 * config/i386/i386.md (*fpatanxf3_i387, fpatan_extend<mode>xf3_i387):
2114 (atan2sf3_1, atan2df3_1, atan2xf3_1): Remove insn patterns.
2115 (atan2xf3): Directly generate RTL pattern.
2116 (atan2<mode>3): Rename from atan2sf3 and atan2df3 and macroize insn
2117 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
2118 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
2119 disable patterns for SSE math.
2120 (atan<mode>2): Rename from atansf2 and atandf2 and macroize insn
2121 patterns using X87MODEF12 mode macro. Use fpatan_extend<mode>xf3_i387
2122 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
2123 disable patterns for SSE math.
2125 2007-01-19 Alexandre Oliva <aoliva@redhat.com>
2127 * libgcc-std.ver: Fix typo in %inherit for GCC_4.3.0.
2129 2007-01-18 Roger Sayle <roger@eyesopen.com>
2131 * fold-const.c (fold_unary) <VIEW_CONVERT_EXPR>: Optimize away a
2132 VIEW_CONVERT_EXPR to the same type as it's operand.
2134 2007-01-18 David Edelsohn <edelsohn@gnu.org>
2136 * config/rs6000/darwin-ldouble.c: Only build _SOFT_FLOAT if
2137 configured for long double 128.
2139 2007-01-18 Mike Stump <mrs@apple.com>
2141 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Fix build
2144 2007-01-18 Michael Meissner <michael.meissner@amd.com>
2146 * i386.c (ix86_compute_frame_layout): Make fprintf's in #if 0 code
2149 2007-01-18 Jan Hubicka <jh@suse.cz>
2151 * tree-ssa-operands.c (vop_free_bucket_size): Never return value
2152 greater than NUM_VOP_FREE_BUCKETS.
2154 2007-01-18 Daniel Berlin <dberlin@dberlin.org>
2156 * tree-ssa-structalias.c: Update comments.
2157 (ptabitmap_obstack): Removed.
2159 (oldpta_obstack): Ditto.
2160 (stats): Add a few members.
2161 (struct variable_info): Remove node, complex, address_taken, and
2162 indirect_target members. Add oldsolution member.
2163 (new_var_info): Do not initialize removed members.
2164 (constraint_expr_type): Remove INCLUDES.
2165 (constraint_graph): Add size, implicit_preds, rep,
2166 indirect_cycles, eq_rep, label, direct_nodes, and complex members.
2167 (FIRST_REF_NODE): New macro.
2168 (LAST_REF_NODE): Ditto.
2169 (FIRST_ADDR_NODE): Ditto.
2170 (find): New function.
2172 (dump_constraint): Do not handle INCLUDES.
2173 (insert_into_complex): Do not insert duplicate constraints.
2174 (condense_varmap_nodes): Renamed and rewritten into ...
2175 (merge_node_constraints): This. Also fix bug in handling of
2176 offseted copy constraints.
2177 (clear_edges_for_node): No longer need to deal with preds at all,
2178 or removing associated preds/succs.
2179 (merge_graph_nodes): Deal with indirect_cycles.
2180 Don't deal with predecessors.
2181 (add_implicit_graph_edge): New function.
2182 (add_pred_graph_edge): Ditto.
2183 (add_graph_edge): Don't deal with predecessors.
2184 (build_constraint_graph): Removed.
2185 (build_pred_graph): New function.
2186 (build_succ_graph): Ditto.
2187 (struct scc_info): Removed in_component. Added roots, dfs, and
2188 node_mapping. Remove visited_index, unification_queue.
2189 (scc_visit): Deal with union-find we do now.
2190 Deal with cycles with REF nodes.
2191 (collapse_nodes): Renamed and rewritten to ...
2192 (unify_nodes): This.
2193 (process_unification_queue): Removed.
2194 (topo_visit): Cleanup
2195 (do_da_constraint): Use find.
2196 (do_sd_constraint): Ditto.
2197 (do_ds_constraint): Ditto.
2198 (do_complex_constraint): Ditto.
2199 (init_scc_info): Update for removed and added members.
2200 (find_and_collapse_graph_cycles): Renamed and rewritten into ...
2201 (find_indirect_cycles): This.
2202 (equivalence_class): New variable.
2203 (label_visit): New function.
2204 (perform_variable_substitution): Rewritten.
2205 (free_var_substitution_info): New function.
2206 (find_equivalent_node): Ditto.
2207 (move_complex_constraints): Ditto.
2208 (eliminate_indirect_cycles): Ditto.
2209 (solve_graph): Only propagate changed bits.
2210 Use indirect cycle elimination.
2212 (tree_id_t): Rename to tree_vi_t, delete id member, add vi member.
2213 (tree_id_eq): Renamed to ...
2214 (tree_vi_eq): This. Update for member change
2215 (insert_id_for_tree): Renamed and rewritten to ...
2216 (insert_vi_for_tree): This.
2217 (lookup_id_for_tree): Renamed and rewritten to ...
2218 (lookup_vi_for_tree): This.
2219 (get_id_for_tree): Renamed and rewritten to ...
2220 (get_vi_for_tree): Ditto.
2221 (get_constraint_exp_from_ssa_var): Update to use get_vi_for_tree.
2222 (process_constraint): Don't handle INCLUDES.
2223 Remove special ADDRESSOF case.
2224 (find_func_aliases): Rewrite to use vi functions instead of id
2226 (create_function_info_for): Ditto.
2227 (create_variable_info_for): Ditto.
2228 (intra_create_variable_infos): Ditto.
2229 (merge_smts_into): Ditto.
2230 (find_what_p_points_to): Ditto.
2231 (init_base_vars): Ditto.
2232 (init_alias_vars): Ditto.
2233 (remove_preds_and_fake_succs): New function.
2234 (dump_sa_points_to_info): Dump new stats.
2235 (dump_solution_for_var): Use find.
2236 (set_used_smts): Fix formatting.
2237 (compute_points_to_sets): Updated for new functions.
2238 (ipa_pta_execute): Ditto.
2240 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2241 Richard Sandiford <richard@codesourcery.com>
2243 * doc/tm.texi (TARGET_FUNCTION_VALUE): Expand documentation of
2245 * calls.c (expand_call): If the return value is a PARALLEL,
2246 extract its first member.
2247 * config/m68k/linux.h (FUNCTION_EXTRA_EPILOGUE): Remove.
2248 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
2249 use FUNCTION_EXTRA_EPILOGUE.
2250 (m68k_function_value): Return a PARALLEL if the return value
2251 is of a pointer type.
2252 * config/m68k/netbsd-elf.h (current_function_returns_pointer)
2253 (FUNCTION_EXTRA_EPILOGUE): Remove.
2254 * config/m68k/m68k.md (D0_REG): New constant.
2256 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2258 * config/m68k/m68k.c (m68k_output_function_epilogue): Don't
2259 output a NOP for empty epilogues.
2261 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2263 * config/m68k/m68k.c (m68k_use_return_insn): Update comments
2264 before function. Extend register save check to include all
2265 registers, not just integer ones.
2267 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2269 * config/m68k/m68k-protos.h (use_return_insn): Rename to...
2270 (m68k_use_return_insn): ...this.
2271 * config/m68k/m68k.h (USE_RETURN_INSN): Delete.
2272 * config/m68k/m68k.c (use_return_insn): Rename to...
2273 (m68k_use_return_insn): ...this.
2274 * config/m68k/m68k.md (return): Use m68k_use_return_insn instead
2277 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
2279 * config/m68k/fpgnulib.c (__truncdfsf2): Implement round to
2280 nearest even, fix denormal rounding overflow.
2282 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2284 * config/m68k/m68k.md (movsf_cf_hard): Use fsmove instead of
2285 f%$smove and f%$move.
2286 (movdf_cf_hard): Use fdmove for cases 0 and 3 and fmove for case 1.
2287 (extendsfdf2_cf): Use fdmove instead of f%&move.
2288 (truncdfsf2_cf): Use fsmove instead of f%$smove.
2289 (add<mode>3_cf, sub<mode>3_cf): Use <FP:prec> instead of <FP:round>.
2291 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2293 * config/m68k/m68k.md (movdf_cf_hard): Use output_move_double for
2296 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2298 * real.h (real_format): Add a canonical_nan_lsbs_set field.
2299 (coldfire_single_format): Declare.
2300 (coldfire_double_format): Likewise.
2301 * real.c (encode_ieee_single): Use canonical_nan_lsbs_set instead
2302 of qnan_msb_set to determine the lower bits of a canonical
2304 (encode_ieee_double): Likewise.
2305 (encode_ieee_quad): Likewise.
2306 (ieee_single_format): Initialize canonical_nan_lsbs_set.
2307 (mips_single_format): Likewise.
2308 (ieee_double_format): Likewise.
2309 (mips_double_format): Likewise.
2310 (ieee_extended_motorola_format): Likewise.
2311 (ieee_extended_intel_96_format): Likewise.
2312 (ieee_extended_intel_128_format): Likewise.
2313 (ieee_extended_intel_96_round_53_format): Likewise.
2314 (ibm_extended_format): Likewise.
2315 (mips_extended_format): Likewise.
2316 (ieee_quad_format): Likewise.
2317 (mips_quad_format): Likewise.
2318 (vax_f_format): Likewise.
2319 (vax_d_format): Likewise.
2320 (vax_g_format): Likewise.
2321 (i370_single_format): Likewise.
2322 (i370_double_format): Likewise.
2323 (decimal_single_format): Likewise.
2324 (decimal_double_format): Likewise.
2325 (decimal_quad_format): Likewise.
2326 (c4x_single_format): Likewise.
2327 (c4x_extended_format): Likewise.
2328 (real_internal_format): Likewise.
2329 (coldfire_single_format): New real_format.
2330 (coldfire_double_format): Likewise.
2331 * config/pdp11/pdp11.c (pdp11_f_format): Initialize
2332 canonical_nan_lsbs_set.
2333 (pdp11_d_format): Likewise.
2334 * config/m68k/m68k.c (override_options): Override REAL_FORMAT_MODE
2335 if TARGET_COLDFIRE_CPU.
2337 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2339 * config/m68k/m68k-protos.h (m68k_output_pic_call): Delete.
2340 (output_call, m68k_legitimize_call_address): Declare.
2341 * config/m68k/m68k.h (EXTRA_CONSTRAINT): Remove unnecessary
2342 parenthesees. Add support for a 'W' constraint.
2343 (LEGITIMATE_PIC_OPERAND_P): Remove SYMBOL_REF_FLAG handling.
2344 (PRINT_OPERAND_PUNCT_VALID_P): Remove comment about 'o'.
2345 (m68k_symbolic_call, m68k_symbolic_jump): Declare.
2346 * config/m68k/m68k.c (m68k_symbolic_call, m68k_symbolic_jump): New
2348 (override_options): Initialize them. Do not set flag_no_function_cse
2349 for TARGET_ID_SHARED_LIBRARY.
2350 (m68k_output_pic_call): Delete.
2351 (m68k_legitimize_call_address): New function.
2352 (print_operand): Remove the %o prefix. Handle the %p prefix.
2353 (output_call): New function.
2354 (m68k_output_mi_thunk): Use m68k_symbolic_jump. Always load the
2355 target address from the GOT if symbolic jumps are not allowed.
2356 * config/m68k/m68k.md (call, general_operand): Do not set
2357 SYMBOL_REF_FLAG. Use m68k_legitimize_call_address instead.
2358 Merge separate flag_pic and !flag_pic define_insns into...
2359 (*call, *call_value): ...these new patterns. Match the address
2360 rather than the containing MEM and require it to be a call_operand.
2361 Use output_call to generate the asm template.
2362 * config/m68k/predicates.md (const_call_operand): New predicate.
2363 (call_operand): Likewise.
2365 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
2367 * config/m68k/m68k.h (REGISTER_MOVE_COST): Simplify definition.
2368 (STACK_GROWS_DOWNWARD): Define to 1.
2369 (FUNCTION_VALUE, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P): Equivocate
2370 comments, emphasizing that these values are only defaults.
2371 * config/m68k/linux.h (LINK_SPEC): Fix formatting in #undef.
2372 * config/m68k/m68k.c (const_method): Remove trailing whitespace.
2374 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2376 * config/m68k/m68k.md (cmpsi): Remove outdated flag_pic handling.
2378 2007-01-18 Kazu Hirata <kazu@codesourcery.com>
2379 Richard Sandiford <richard@codesourcery.com>
2381 * config/m68k/m68k.h (DATA_REGNO_P, ADDRESS_REGNO_P, INT_REGNO_P)
2382 (FP_REGNO_P): New macros.
2383 (REGNO_OK_FOR_INDEX_P, REGNO_OK_FOR_BASE_P, REGNO_OK_FOR_DATA_P)
2384 (REGNO_OK_FOR_FP_P, REG_OK_FOR_INDEX_P, REG_OK_FOR_BASE_P): Use them.
2385 (EH_RETURN_STACKADJ_RTX): Use A0_REG.
2386 * config/m68k/m68k.c (m68k_regno_mode_ok): Use the new REGNO macros.
2388 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
2390 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
2391 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-rtems*): Add t-floatlib
2393 * config/m68k/t-floatlib: New file, extracting common code from...
2394 * config/m68k/t-m68kbare, config/m68k/t-m68kelf,
2395 * config/m68k/t-uclinux: Here.
2396 * config/m68k/fpgnulib.c: Do not compile extendeed precision
2397 routines on ColdFire targets.
2399 2007-01-18 Nathan Sidwell <nathan@codesourcery.com>
2401 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
2402 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*)
2403 (m68k-*-uclinuxoldabi, m68k-*-uclinux*, m68k-*-linux*)
2404 (m68k-*-rtems*): Use tm_file rather than m68k/m68k.h and
2405 explicitly set MOTOROLA to 1.
2406 * config/m68k/m68k.h (MOTOROLA): Simplify definition accordingly.
2408 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2409 Nathan Sidwell <nathan@codesourcery.com>
2411 * config/m68k/m68k.h (PCC_STATIC_STRUCT_RETURN): Delete.
2412 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Add MOTOROLA cases.
2413 * config/m68k/coff.h (REGISTER_PREFIX_MD): Delete.
2414 * config/m68k/m68020-elf.h (LIB_SPEC): Delete.
2415 * config/m68k/m68k-none.h (CC1_SPEC, CPP_SUBTARGET_SPEC): Delete.
2416 * config/m68k/m68kelf.h (IMMEDIATE_PREFIX, REGISTER_PREFIX_MD)
2417 (ASM_OUTPUT_REG_PUSH, ASM_OUTPUT_REG_POP): Delete.
2418 (NO_DOLLAR_IN_LABEL, PCC_STATIC_STRUCT_RETURN): Don't undefine.
2419 (BSS_ASM_OP, ASM_OUTPUT_SKIP, ASM_OUTPUT_ASCII): Delete.
2420 * config/m68k/m68kemb.h (PCC_STATIC_STRUCT_RETURN): Don't undefine.
2421 * config/m68k/linux.h (SIZE_TYPE, PTRDIFF_TYPE, WCHAR_TYPE)
2422 (WCHAR_TYPE_SIZE, TARGET_OBJFMT_CPP_BUILTINS): Delete.
2423 (TARGET_OS_CPP_BUILTINS): Don't define mc68000 and mc68020 here.
2424 (DBX_REGISTER_NUMBER): Delete.
2425 * config/m68k/netbsd-elf.h (IMMEDIATE_PREFIX): Delete.
2426 (PCC_STATIC_STRUCT_RETURN): Don't undefine.
2427 * config/m68k/openbsd.h (PCC_STATIC_STRUCT_RETURN): Define.
2429 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2431 * config.gcc (m68k-*-uclinux*): Add flat.h to $tm_file.
2432 * config/flat.h: New file.
2433 * crtstuff.c (USE_PT_GNU_EH_FRAME): Don't define if
2435 * config/m68k/m68k.h (ASM_PREFERRED_EH_DATA_FORMAT): Do not use
2436 indirect references for -msep-data or -mid-shared-library.
2437 Do not use PC-relative code addresses either.
2438 * config/m68k/m68k.c (override_options): Restrict -fPIC error
2440 * config/m68k/uclinux.h (STARTFILE_SPEC): Define. Use Scrt1.o
2441 for shared libraries and crt1.o for executables. Use crti.o and
2443 (ENDFILE_SPEC): Use crtend.o and crtn.o.
2444 (LIB_SPEC): Suppress -Rlibc.gdb if -static-libc is given.
2445 Do not add -elf2flt or -shared-lib-id options here.
2446 (LINK_SPEC): Define. Pass -elf2flt if no -elf2flt option is given.
2447 Pass -shared-lib-id if -mid-shared-library, taking the library
2448 identifier from -mshared-library-id if given, otherwise
2450 (EH_FRAME_IN_DATA_SECTION): Do not undefine.
2451 (INIT_SECTION_ASM_OP, FINI_SECTION_ASM_OP): Likewise.
2452 (TARGET_OS_CPP_BUILTINS): Define __GXX_MERGED_TYPEINFO_NAMES=0
2453 and __GXX_TYPEINFO_EQUALITY_INLINE=0 if -mid-shared-library.
2454 (DRIVER_SELF_SPECS): Map unadorned PIC options to -msep-data.
2455 * config/m68k/t-uclinux (EXTRA_MULTILIB_PARTS): Add crtbegin.o
2457 * config/m68k/lb1sf68.asm (PICCALL): Use an lea and pc-relative
2458 jump sequence for ISA A and ISA A+.
2459 (PICJUMP): Likewise.
2461 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2463 * config.gcc (m68k-*-uclinux*): Base the port on the common
2464 and m68k GNU/Linux files rather than on the generic ELF ones.
2465 * config/m68k/uclinux.h (TARGET_VERSION): Override.
2466 (TARGET_OS_CPP_BUILTINS): Use LINUX_TARGET_OS_CPP_BUILTINS.
2468 2007-01-18 Julian Brown <julian@codesourcery.com>
2469 Richard Sandiford <richard@codesourcery.com>
2471 * config/m68k/m68k.h (LONG_DOUBLE_TYPE_SIZE): Make 64-bit on ColdFire.
2472 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2473 * config/m68k/netbsd-elf.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Undefine
2475 * config/m68k/uclinux-oldabi.h (LONG_DOUBLE_TYPE_SIZE): Redefine to
2477 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2479 2007-01-18 Richard Sandiford <richard@codesourcery.com>
2481 * doc/install.texi: Document m68k-uclinuxoldabi.
2482 * config.gcc (m68k-*-uclinuxoldabi): New configuration.
2483 * config/m68k/uclinux-oldabi.h: New file, copied from
2484 config/m68k/uclinux.h.
2486 2007-01-18 Kaz Kojima <kkojima@rr.iij4u.or.jp>
2488 * config/m32r/m32r-protos.h (m32r_expand_epilogue): Declare it.
2489 * config/m32r/m32r.c (m32r_setup_incoming_varargs): Use gen_frame_mem.
2490 (m32r_compute_frame_size): Use unsigned for regno.
2491 (m32r_reload_lr): Use gen_frame_mem.
2493 (m32r_output_function_epilogue): Don't output the function epilogue
2495 (m32r_expand_epilogue): New.
2496 (direct_return): Return false if the function has the interrupt
2498 (m32r_hard_regno_rename_ok): Remove code for the textual epilogue.
2499 * config/m32r/m32r.md (epilogue): New expander.
2500 (return_lr, return_rte): New insns.
2501 (return): Make it expander.
2502 (return_normal): New expander.
2504 2007-01-18 Josh Conner <jconner@apple.com>
2507 * config/rs6000/rs6000.c (rs6000_emit_vector_compare): Add
2508 support for UNLE, UNLT, UNGE, and UNGT.
2510 2007-01-18 Jan Hubicka <jh@suse.cz>
2512 * tree-vrp.c (finalize_jump_threads): Do not call cleanup_cfg by hand.
2513 * tree-tailcall (add_virtual_phis): Likewise.
2514 (optimize_tail_call): Return TODOs.
2515 (execute_tail_calls): Return TODOs.
2516 * tree-ssa-ccp (execute_fold_all_builtins): Do cleanup_cfg via TODO.
2517 * tree-cfgcleanup.c (cleanup_tree_cfg_loop): Return if something
2519 * tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_value):
2520 Cleanup cfg using TODO.
2521 * tree-flow.h (cleanup_tree_cfg_loop): Update prototype.
2522 * passes.c (execute_function_todo): When cleanup did something, remove
2524 * tree-cfg.c (pass_build_cfg): Add cleanup_cfg TODO.
2525 (make_edges): Don't cleanup_cfg.
2527 2007-01-18 Uros Bizjak <ubizjak@gmail.com>
2529 * reg-stack.c (subst_stack_regs_pat) [UNSPEC_SINCOS_COS,
2530 UNSPEC_XTRACT_FRACT]: Use generic code for instructions that
2531 operate on the top of stack.
2532 [UNSPEC_SINCOS_SIN, UNSPEC_XTRACT_EXP, UNSPEC_TAN]: Rewrite
2533 register handling of instructions that output to the second
2535 [UNSPEC_TAN_ONE, UNSPEC_TAN_TAN]: Remove.
2536 (move_for_stack_reg): Special-case check for dead destination
2537 stack slot for constant load of 1.0 inside UNSPEC_TAN.
2539 * config/i386/i386.md (UNSPEC_TAN): New constant.
2540 (UNSPEC_TAN_ONE, UNSPEC_TAN_TAN): Remove.
2541 (fptanxf4_i387, fptan_extend<mode>xf4_i387): New patterns
2542 to correctly model move of constant 1.0 to top stack slot.
2543 (*tandf3_1, *tansf3_1, *tanxf3_1): Remove insn patterns.
2544 (unnamed peephole2 pattern): Remove corresponding peephole2
2545 pattern that optimizes tan insn and loading of constant 1.0.
2546 (tanxf2): Use fptanxf4_i387.
2547 (tan<mode>2): Rename from tansf2 and tandf2 and macroize insn
2548 patterns using X87MODEF12 mode macro. Use fptan_extend<mode>xf4_i387
2549 and truncate result to requested mode. Use SSE_FLOAT_MODE_P to
2550 disable patterns for SSE math.
2551 (sincos<mode>3): Use truncxf<mode>2_i387_noop for truncation.
2552 (fyl2x_extend<mode>xf3_i387): Use X87MODEF12 for operand 1.
2554 2007-01-18 Dirk Mueller <dmueller@suse.de>
2555 Richard Guenther <rguenther@suse.de>
2558 * doc/invoke.texi (Warray-bounds): Document -Warray-bounds.
2559 * common.opt (Warray-bounds): Add new warning option.
2560 * c-opts.c (c_common_handle_option): Define -Warray-bounds
2562 * Makefile.in: make tree-vrp.o depend on toplev.h
2563 * tree-vrp.c (vrp_finalize): Call check_array_refs if -Warray-bounds
2565 (check_array_refs, check_array_bounds, check_array_ref): New.
2567 2007-01-18 Jan Hubicka <jh@suse.cz>
2569 * tree-ssa-ccp.c (ccp_finalize): Return if something changed.
2570 (execute_ssa_ccp): Return flags conditionally.
2571 * tree-ssa-propagate.c (substitue_and_fold): Return if something was
2573 * tree-ssa-propagate.h (substitute_and_fold): Update prototype.
2575 2007-01-18 Steven Bosscher <steven@gcc.gnu.org>
2577 * cfgcleanup.c (cleanup_cfg): Detect cfglayout mode and set
2578 the CLEANUP_CFGLAYOUT flag when in cfglayout mode.
2580 * Makefile.c (GTFILES): Add cfglayout.h.
2581 * gengtype.c (open_base_files): Likewise.
2582 * cfglayout.c (cfg_layout_function_footer,
2583 cfg_layout_function_header) Reindent to make gengtype happy.
2584 * cfglayout.h (cfg_layout_function_footer,
2585 cfg_layout_function_header): Add GTY(()) marker.
2587 * ifcvt.c (noce_try_sign_mask): Make sure INSN_B is non-null.
2589 2007-01-18 Ben Elliston <bje@au.ibm.com>
2591 * genautomata.c (write_automata): Include xstrerror output in the
2592 error message if writing the DFA description file fails.
2594 2007-01-17 H.J. Lu <hongjiu.lu@intel.com>
2596 * config/mips/mips-protos.h (mips_output_external): Make it
2598 * config/mips/iris.h (TARGET_ASM_EXTERNAL_LIBCALL): Removed.
2599 * config/mips/mips.c (irix_output_external_libcall): Likewise.
2600 (extern_list): Likewise.
2601 (extern_head): Likewise.
2602 (TARGET_ASM_FILE_END): Likewise.
2603 (mips_file_end): Likewise.
2604 (mips_output_external): Rewritten.
2606 2007-01-18 Ben Elliston <bje@au.ibm.com>
2608 * genpreds.c (write_insn_preds_c): Only write out the function
2609 body for regclass_for_constraint if we have register constraints.
2611 2007-01-17 Tom Tromey <tromey@redhat.com>
2613 * doc/sourcebuild.texi (libgcj Tests): Use sourceware.org.
2614 * doc/install.texi (Testing): Use sourceware.org.
2615 (Binaries): Likewise.
2616 (Specific): Likewise.
2617 * doc/contrib.texi (Contributors): Use sourceware.org.
2619 2007-01-17 Anatoly Sokolov <aesok@post.ru>
2621 * config/avr/avr.h (AVR_HAVE_LPMX): New macro.
2622 (AVR_ENHANCED): Rename to ...
2623 (AVR_HAVE_MUL): ... new.
2624 (avr_enhanced_p): Rename to ...
2625 (avr_have_mul_p): ... new.
2626 (TARGET_CPU_CPP_BUILTINS): Use 'avr_have_mul_p' instead of
2627 'avr_enhanced_p' for "__AVR_ENHANCED__". Define "__AVR_HAVE_MUL__".
2628 * config/avr/avr.c (avr_enhanced_p): Rename to ...
2629 (avr_have_mul_p): ... new.
2630 (base_arch_s): Rename 'enhanced' to 'have_mul'.
2631 (avr_override_options): Use 'avr_have_mul_p' and 'have_mul' instead of
2632 'avr_enhanced_p' and 'enhanced'.
2633 (ashlhi3_out, ashrhi3_out, lshrhi3_out, avr_rtx_costs): Use
2634 AVR_HAVE_MUL instead of AVR_ENHANCED.
2635 * avr.md (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
2636 (mulqi3, *mulqi3_enh, *mulqi3_call, mulqihi3, umulqihi3, mulhi3,
2637 *mulhi3_enh, *mulhi3_call, mulsi3, *mulsi3_call): Use AVR_HAVE_MUL
2638 instead of AVR_ENHANCED.
2639 (*tablejump_enh): Use AVR_HAVE_LPMX instead of AVR_ENHANCED.
2640 * libgcc.S: Use __AVR_HAVE_MUL__ instead of __AVR_ENHANCED__.
2641 (__tablejump__): Use __AVR_HAVE_LPMX__ instead of __AVR_ENHANCED__.
2643 2007-01-17 Ian Lance Taylor <iant@google.com>
2645 * vec.h (VEC_reserve_exact): Define.
2646 (vec_gc_p_reserve_exact): Declare.
2647 (vec_gc_o_reserve_exact): Declare.
2648 (vec_heap_p_reserve_exact): Declare.
2649 (vec_heap_o_reserve_exact): Declare.
2650 (VEC_OP (T,A,reserve_exact)): New static inline function, three
2652 (VEC_OP (T,A,reserve)) [all versions]: Remove handling of
2654 (VEC_OP (T,A,alloc)) [all versions]: Call ...reserve_exact.
2655 (VEC_OP (T,A,copy)) [all versions]: Likewise.
2656 (VEC_OP (T,a,safe_grow)) [all versions]: Likewise.
2657 * vec.c (calculate_allocation): Add exact parameter. Change all
2659 (vec_gc_o_reserve_1): New static function, from vec_gc_o_reserve.
2660 (vec_gc_p_reserve, vec_gc_o_reserve): Call vec_gc_o_reserve_1.
2661 (vec_gc_p_reserve_exact, vec_gc_o_reserve_exact): New functions.
2662 (vec_heap_o_reserve_1): New static function, from vec_heap_o_reserve.
2663 (vec_heap_p_reserve, vec_heap_o_reserve): Call vec_heap_o_reserve_1.
2664 (vec_heap_p_reserve_exact): New function.
2665 (vec_heap_o_reserve_exact): New function.
2667 2007-01-17 Jan Hubicka <jh@suse.cz>
2669 * ipa-type-escape.c (look_for_casts): Revamp using handled_component_p.
2671 2007-01-17 Eric Christopher <echristo@apple.com>
2673 * config.gcc: Support core2 processor.
2675 2007-01-16 Jan Hubicka <jh@suse.cz>
2677 * tree-ssanames.c (release_dead_ssa_names): Instead of ggc_freeing
2678 the names, just unlink the chain so we don't crash on dangling pointers
2681 2007-01-16 Jan Hubicka <jh@suse.cz>
2683 * cgraph.h (cgraph_decide_inlining_incrementally): Kill.
2684 * tree-pass.h: Reorder to make IPA passes appear toegher.
2685 (pass_early_inline, pass_inline_parameters, pass_apply_inline): Declare.
2686 * cgraphunit.c (cgraph_finalize_function): Do not compute inling
2687 parameters, do not call early inliner.
2688 * ipa-inline.c: Update comments. Include tree-flow.h
2689 (cgraph_decide_inlining): Do not compute inlining parameters.
2690 (cgraph_decide_inlining_incrementally): Return TODOs; assume to
2691 be called with function context set up.
2692 (pass_ipa_inline): Remove unreachable functions before pass.
2693 (cgraph_early_inlining): Simplify assuming to be called from the
2695 (pass_early_inline): New pass.
2696 (cgraph_gate_ipa_early_inlining): New gate.
2697 (pass_ipa_early_inline): Turn into simple wrapper.
2698 (compute_inline_parameters): New function.
2699 (gate_inline_passes): New gate.
2700 (pass_inline_parameters): New pass.
2701 (apply_inline): Move here from tree-optimize.c
2702 (pass_apply_inline): New pass.
2703 * ipa.c (cgraph_remove_unreachable_nodes): Verify cgraph after
2705 * tree-inline.c (optimize_inline_calls): Return TODOs rather than
2707 (tree_function_versioning): Do not allocate dummy struct function.
2708 * tree-inline.h (optimize_inline_calls): Update prototype.
2709 * tree-optimize.c (execute_fixup_cfg): Export.
2710 (pass_fixup_cfg): Remove
2711 (tree_rest_of_compilation): Do not apply inlines.
2712 * tree-flow.h (execute_fixup_cfg): Declare.
2713 * Makefile.in (gt-passes.c): New.
2714 * passes.c: Include gt-passes.h
2715 (init_optimization_passes): New passes.
2716 (nnodes, order): New static vars.
2717 (do_per_function_toporder): New function.
2718 (execute_one_pass): Dump current pass here.
2719 (execute_ipa_pass_list): Don't dump current pass here.
2721 2007-01-16 Janis Johnson <janis187@us.ibm.com>
2723 * config/dfp-bit.c (dfp_compare_op): Return separate value for NaN.
2724 (DFP_NE, DFP_LE, DFP_GE): Return false for NaN.
2726 2007-01-16 David Edelsohn <edelsohn@gnu.org>
2728 * config/rs6000/darwin-ldouble.c: Build file for SOFT_FLOAT.
2729 (strong_alias): Define.
2730 (__gcc_qmul): Provide non-FMA for soft-float.
2736 (__gcc_qunord): New.
2746 * config/rs6000/rs6000.c (rs6000_init_libfuncs): Initialize
2747 soft-float functions.
2748 * config/rs6000/libgcc-ppc-glibc.ver: Version soft-float symbols.
2749 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Do not warn
2750 about long double soft float.
2752 2007-01-16 Dorit Nuzman <dorit@il.ibm.com>
2753 Tehila Meyzels <tehila@il.ibm.com>
2755 * tree-vectorizer.h (is_pattern_stmt_p): New.
2756 * tree-vect-analyze.c (vect_determine_vectorization_factor): Fix
2757 formatting (tabs instead of spaces). Cleanup and clarify setting
2758 of STMT_VINFO_VECTYPE. Call is_pattern_stmt_p.
2759 * tree-vect-transform.c (vect_get_vec_def_for_operand): Fix typo.
2760 (vectorizable_type_demotion): Check that types are integral.
2761 (vectorizable_type_promotion): Likewise.
2762 (vectorizable_store): Fix typo. Eliminate new-line at end of
2765 2007-01-16 Jan Hubicka <jh@suse.cz>
2767 * tree-ssanames.c (release_dead_ssa_names): Remove invalidated
2770 2007-01-15 Eric Christopher <echristo@apple.com>
2772 * ifcvt.c: Include vec.h, vecprim.h.
2773 (check_cond_move_block): New argument regs.
2774 Reorganize. Add registers used to regs.
2775 (cond_move_process_if_block): Use regs set above as
2778 2007-01-15 Eric Christopher <echristo@apple.com>
2780 * config/darwin.h: Update copyright.
2781 (TARGET_OPTION_TRANSLATE_TABLE): Add umbrella.
2782 (LINK_COMMAND_SPEC): Add -u.
2783 (LINK_SPEC): Fix umbrella for above.
2785 2007-01-15 Joseph S. Myers <joseph@codesourcery.com>
2787 * config/soft-fp/op-common.h, config/soft-fp/op-4.h: Update from
2790 2007-01-15 Tom Tromey <tromey@redhat.com>
2792 * doc/sourcebuild.texi (libgcj Tests): Don't mention jacks.
2793 * doc/install.texi (Testing): Don't mention jacks.
2794 (Configuration): Document --enable-java-maintainer-mode. Move
2795 --with-java-home to libgcj-specific section. Document
2797 (Prerequisites): Mention --enable-java-maintainer-mode, ecj1.
2799 2007-01-15 Jan Hubicka <jh@suse.cz>
2801 * tree-ssa-dce.c (DCE_TODOs): New.
2802 (propagate_necessity): Return if something changed.
2803 (eliminate_unnecessary_stmts): Likewise.
2804 (perform_tree_ssa_dce): Return TODO flags when needed.
2805 (pass_dce, pass_dce_loop, pass_cd_dce): Remove TODO flags.
2807 2007-01-15 Uros Bizjak <ubizjak@gmail.com>
2809 * config/i386/i386.md (fyl2xxf3_i387): Rename from fyl2x_xf3.
2810 (fyl2x_extend<mode>xf3_i387): New insn pattern.
2811 (log<mode>2): Rename from logsf2 and logdf2 and macroize insn
2812 patterns using X87MODEF12 mode macro. Extend operand 1
2813 to XFmode. Use SSE_FLOAT_MODE_P to disable patterns for SSE math.
2814 (log10<mode>2): Ditto.
2815 (log2<mode>2): Ditto.
2816 (log1p<mode>2): Ditto.
2817 (logb<mode>2): Ditto.
2818 (fyl2xp1xf3_i387): Rename from fyl2xp1_xf3.
2819 (fyl2xp1_extend<mode>xf3_i387): New insn pattern.
2820 (*fxtractxf3_i387): Rename from *fxtractxf3.
2821 (fxtract_extend<mode>xf3_i387): New insn pattern.
2822 (ilogbsi2): Use match_dup 3, not match_operand:XF 3.
2823 * config/i386/i386.c (ix86_emit_i387_log1p): Use gen_fyl2xp1xf3_i387()
2824 and gen_fyl2xxf3_i387().
2826 2007-01-14 Zdenek Dvorak <dvorakz@suse.cz>
2828 * loop-unswitch.c (unswitch_loop): Do not call fix_loop_placement.
2829 * cfgloopmanip.c (fix_loop_placement): Made static. Use
2830 get_loop_exit_edges. Changed return type to bool.
2831 * cfgloop.h (fix_loop_placement): Declaration removed.
2833 2007-01-14 Dorit Nuzman <dorit@il.ibm.com>
2835 * param.h (MIN_VECT_LOOP_BOUND): New.
2836 * params.def (MIN_VECT_LOOP_BOUND): New.
2837 * tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Takes another
2838 argument - minimum threshold for number of iterations.
2839 * tree-vectorizer.h (slpeel_tree_peel_loop_to_edge): Add another
2840 argument to declaration.
2841 * tree-vect-analyze.c (vect_analyze_operations): Check value of
2842 MIN_VECT_LOOP_BOUND.
2843 * tree-vect-transform.c (vect_do_peeling_for_loop_bound): Call
2844 slpeel_tree_peel_loop_to_edge with additional argument.
2845 (vect_do_peeling_for_alignment): Likewise.
2846 * doc/invoke.texi (min-vect-loop-bound): Document new param option.
2848 2007-01-14 Uros Bizjak <ubizjak@gmail.com>
2851 * config/i386/i386.c (print_operand) ['z']: Output 'b' for
2854 2007-01-14 Jan Hubicka <jh@suse.cz>
2856 * tree-dfa.c (remove_referenced_var): New function.
2857 * tree-ssa-live.c (remove_unused_locals): Walk referenced vars and
2858 prune referenced vars list too.
2859 * tree-flow.h (remove_referenced_var): Declare.
2861 2007-01-14 Jan Hubicka <jh@suse.cz>
2863 * tree-eh.c (add_stmt_to_eh_region_fn): Do not add call_exprs
2865 (remove_stmt_from_eh_region_fn): Do not remove call_exprs.
2866 (verify_eh_throw_stmt_node, verify_eh_throw_table_statements): Kill.
2867 * except.h (verify_eh_throw_table_statements): Kill prototype.
2868 * cfgexpand.c (expand_gimple_basic_block): Propagate Eh regions
2869 into call exrepssions.
2870 * tree-optimize.c (execute_free_cfg_annotatiosn): Do not call
2872 * tree-cfg.c: Include pointer-set.h.
2873 (verify_node_sharing): Work on pointer set.
2874 (verify_eh_throw_stmt_node): New.
2875 (verify_stmts): Use pointers sets, verify throw_stmt.
2877 2007-01-13 Zdenek Dvorak <dvorakz@suse.cz>
2879 * ipa-reference.c (analyze_function): Consider also addresses taken
2882 2007-01-12 Roger Sayle <roger@eyesopen.com>
2884 * c-typeck.c (null_pointer_constant_p): Replace use of
2885 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2886 (build_c_cast): Likewise.
2888 2007-01-12 Roger Sayle <roger@eyesopen.com>
2890 * tree.h (force_fit_type_double): Remove unused final argument.
2891 * c-common.c (constant_expression_warning): Replace use of
2892 TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
2893 (convert_and_check): Likewise.
2894 (shorten_compare): Update call to force_fit_type_double.
2895 (c_common_truthvalue_conversion) <INTEGER_CST>: Use integer_zerop.
2896 * convert.c (convert_to_pointer): Update call to
2897 force_fit_type_double.
2898 * fold-const.c (force_fit_type_double): Remove overflowed_const
2900 (int_const_binop, fold_convert_const_int_from_int,
2901 fold_convert_const_int_from_real, fold_div_compare,
2902 fold_sign_changed_comparison, fold_unary, fold_negate_const,
2903 fold_abs_const, fold_not_const): Remove the final argument from
2904 calls to force_fit_type_double.
2906 2007-01-12 Andrew Pinski <andrew_pinski@playstation.sony.com>
2908 * configure.ac: Set insn to "nop" for spu-*-* also.
2909 * configure: Regenerate.
2911 2007-01-12 Olga Golovanevsky <olga@il.ibm.com>
2913 * builtins.def : Add BUILT_IN_FREE.
2915 2007-01-12 Jan Hubicka <jh@suse.cz>
2917 PR tree-optimization/30443
2918 * tree-inline.c (tree_function_versioning): Do not optimize when
2919 cloning for inlining.
2921 2007-01-12 Zdenek Dvorak <dvorakz@suse.cz>
2923 * doc/loop.texi: Document recording of loop exits.
2924 * cfgloopmanip.c (loopify, duplicate_loop): Use alloc_loop.
2925 (update_single_exits_after_duplication,
2926 update_single_exit_for_duplicated_loop,
2927 update_single_exit_for_duplicated_loops): Removed.
2928 (duplicate_loop_to_header_edge): Do not call
2929 update_single_exits_after_duplication and
2930 update_single_exit_for_duplicated_loops.
2931 (loop_version): Do not update single_exit information.
2932 (fix_loop_structure): Use record_loop_exits instead of
2933 mark_single_exit_loops.
2934 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update
2935 the lists of loop exits.
2936 * cfghooks.c (redirect_edge_and_branch, redirect_edge_and_branch_force,
2937 split_edge, merge_blocks): Update the lists of loop exits.
2938 * modulo-sched.c (sms_schedule): Pass LOOPS_HAVE_RECORDED_EXITS to
2939 loop_optimizer_init.
2940 * loop-init.c (loop_optimizer_init): Call record_loop_exits instead
2941 of mark_single_exit_loops.
2942 (loop_optimizer_finalize): Call release_recorded_exits.
2943 * tree-ssa-loop.c (tree_loop_optimizer_init): Pass
2944 LOOPS_HAVE_RECORDED_EXITS to loop_optimizer_init.
2945 * tree-vectorizer.c (slpeel_tree_duplicate_loop_to_edge_cfg): Do not
2946 update single exit information.
2947 * lambda-code.c (perfect_nestify): Ditto.
2948 * cfgloop.c (flow_loop_free): Destroy the list of exits of the loop.
2949 (mark_single_exit_loops): Removed.
2950 (alloc_loop, loop_exit_hash, loop_exit_eq, loop_exit_free,
2951 get_exit_descriptions, rescan_loop_exit, record_loop_exits,
2952 dump_recorded_exit, dump_recorded_exits, release_recorded_exits): New
2954 (get_loop_exit_edges, single_exit): Use recorded exit lists.
2955 (add_bb_to_loop, remove_bb_from_loops): Update the lists of loop exits.
2956 (verify_loop_structure): Verify consistency of the exit lists.
2957 (flow_loops_find): Use alloc_loop. Initialize exits hash.
2958 (set_single_exit): Removed.
2959 * cfgloop.h (struct loop_exit): New function.
2960 (struct loop): single_exit_ field replaced by exits field.
2961 (LOOPS_HAVE_MARKED_SINGLE_EXITS): Replaced by LOOPS_HAVE_RECORDED_EXITS.
2962 (struct loops): Added exits hash.
2963 (mark_single_exit_loops, set_single_exit): Declaration removed.
2964 (release_recorded_exits, record_loop_exits, rescan_loop_exit): Declare.
2966 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2968 * doc/invoke.texi: Avoid use of @headitem.
2970 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2972 * cse.c (cse_insn): Move HAVE_CC0 code after declarations.
2974 2007-01-12 Richard Sandiford <richard@codesourcery.com>
2976 * doc/install.texi: Fix m68k-*-* anchor and add m68k-*-* to the
2979 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2980 Richard Sandiford <richard@codesourcery.com>
2982 * doc/invoke.texi: Document -mno-bitfield, -mno-rtd and -mno-short.
2983 * config/m68k/m68k.opt: Resort options.
2984 (mbitfield, mrtd, mshort): Remove RejectNegative properties.
2986 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
2987 Richard Sandiford <richard@codesourcery.com>
2989 * doc/invoke.texi: Document the macros that are defined by
2990 m68k's -mtune and -mhard-float options.
2991 * config/m68k/m68k-protos.h (m68k_cpp_cpu_ident) Declare.
2992 (m68k_cpp_cpu_family): Likewise.
2993 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Add a full set
2994 of __ucfv*__ macros. Define __mcffpu__ if generating code for
2995 ColdFire FPUs. Define __mcf_cpu_* and __mcf_family_* macros.
2996 * config/m68k/m68k.c (m68k_cpp_cpu_ident): New function.
2997 (m68k_cpp_cpu_family): Likewise.
2999 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3001 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Treat all mc68*
3002 macros besides mc68000 as tuning macros. Use a switch statement
3003 to set them and mcpu32.
3005 2007-01-12 Julian Brown <julian@codesourcery.com>
3007 * config/m68k/m68k.h: Use TARGET_68040 instead of TARGET_68040_ONLY.
3008 (TARGET_68040_ONLY): Rename to...
3009 (TARGET_68040): ...this.
3010 * config/m68k/m68k.c: Use TARGET_68040 instead of TARGET_68040_ONLY.
3011 * config/m68k/m68k.md: Likewise.
3013 2007-01-12 Julian Brown <julian@codesourcery.com>
3014 Nathan Sidwell <nathan@codesourcery.com>
3015 Richard Sandiford <richard@codesourcery.com>
3017 * config.gcc (m680[012]0-*-*, m68k*-*-*): Set m68k_cpu_ident to
3018 the -mcpu= argument associated with the --with-cpu setting.
3019 Define M68K_DEFAULT_TUNE to the default -mtune= option,
3020 if different from the one implied by the -mcpu setting.
3021 Accept --with-cpu=FOO if FOO is listed in m68k-devices.def,
3022 using mcpu=FOO as the default CPU option. Set target_cpu_default2.
3023 * doc/invoke.texi: Mention ColdFire in the introduction to the
3024 m68k options. Document the new -march, -mcpu, -mtune, -mdiv,
3025 -mno-div and -mhard-float options. Make -m68881 a synonym for
3026 -mhard-float. Document the previously-undocumented -m5206e,
3027 -m528x, -m5307 and -m5407 options. Tweak the existing option
3028 documentation for consistency.
3029 * doc/install.texi: Mention new --with-cpu arguments.
3030 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Only use the
3031 default CPU if neither -mcpu nor -march are specified.
3032 (ASM_CPU_SPEC): Pass down -mcpu and -march options.
3033 (TARGET_CPU_CPP_BUILTINS): Set __mcfisa*__ macros from
3034 TARGET_ISA*. Set the legacy __mcf*__ cpu macros in the same way,
3035 using m68k_tune to decide between families that implement the
3036 same ISA. Use m68k_tune to set __mcfv4e__.
3037 (FL_BITFIELD, FL_68881, FL_COLDFIRE, FL_CF_HWDIV, FL_CF_MAC)
3038 (FL_CF_EMAC, FL_CF_EMAC_B, FL_CF_USP, FL_CF_FPU, FL_ISA_68000)
3039 (FL_ISA_68010, FL_ISA_68020, FL_ISA_68040, FL_ISA_A, FL_ISA_B)
3040 (FL_ISA_C, FL_ISA_MMU): New macros.
3041 (MASK_COLDFIRE): Delete.
3042 (TARGET_68010, TARGET_68020, TARGET_68040_ONLY, TARGET_COLDFIRE)
3043 (TARGET_ISAB): Redefine in terms of m68k_cpu_flags.
3044 (TARGET_68881, TARGET_COLDFIRE_FPU): Redefine in terms of m68k_fpu.
3045 (TARGET_HARD_FLOAT): Do not define here.
3046 (TARGET_ISAAPLUS, TARGET_ISAC): New macros.
3047 (TUNE_68000): New macro.
3048 (TUNE_68000_10): Redefine in terms of TUNE_68000 and TUNE_68010.
3049 (TUNE_68010, TUNE_68030, TUNE_68040, TUNE_68060, TUNE_CPU32)
3050 (TUNE_CFV2): Redefine in terms of m68k_tune.
3051 (uarch_type, target_device, fpu_type): New enums.
3052 (m68k_cpu, m68k_tune, m68k_fpu, m68k_cpu_flags): Declare.
3053 * config/m68k/m68k.c (TARGET_DEFAULT): Remove MASK_68881.
3054 (FL_FOR_isa_00, FL_FOR_isa_10, FL_FOR_isa_20, FL_FOR_isa_40)
3055 (FL_FOR_isa_cpu32, FL_FOR_isa_a, FL_FOR_isa_aplus, FL_FOR_isa_b)
3056 (FL_FOR_isa_c): New macros.
3057 (m68k_isa): New enum.
3058 (m68k_target_selection): New structure.
3059 (all_devices, all_isas, all_microarchs): New tables.
3060 (m68k_cpu_entry, m68k_arch_entry, m68k_tune_entry, m68k_cpu)
3061 (m68k_tune, m68k_fpu, m68k_cpu_flags): New variables.
3062 (MASK_ALL_CPU_BITS): Delete.
3063 (m68k_find_selection): New function.
3064 (m68k_handle_option): Handle -mcpu=, -march= and -mtune=.
3065 Map the legacy target options to a combination of the new ones.
3066 (override_options): Set m68k_cpu, m68k_tune, m68k_fpu and
3067 m68k_cpu_flags. Handle M68K_DEFAULT_TUNE. Use m68k_cpu_flags
3068 to derive default MASK_BITFIELD, MASK_CF_HWDIV and MASK_HARD_FLOAT
3070 * config/m68k/m68k.opt (m5200, m5206e, m528x, m5307, m5407, mcfv4e)
3071 (m68010, m68020, m68020-40, m68020-60, m68030, m68040): Remove Mask
3073 (m68881, msoft-float): Change mask from 68881 to HARD_FLOAT.
3074 (march=, mcpu=, mdiv, mhard-float, mtune=): New options.
3075 * config/m68k/m68k-devices.def: New file.
3077 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3078 Nathan Sidwell <nathan@codesourcery.com>
3080 * config/m68k/m68k.h (ASM_CPU_SPEC, ASM_SPEC, EXTRA_SPECS)
3081 (SUBTARGET_EXTRA_SPECS): New macros.
3082 * config/m68k/linux.h (ASM_SPEC): Remove CPU flags;
3083 use %(asm_cpu_spec) instead.
3084 * config/m68k/m68k-none.h (ASM_SPEC): Likewise.
3085 * config/m68k/openbsd.h (ASM_SPEC): Likewise.
3086 * config/m68k/netbsd-elf.h (ASM_SPEC): Likewise.
3087 (EXTRA_SPECS): Rename to...
3088 (SUBTARGET_EXTRA_SPECS): ...this.
3090 2007-01-12 Nathan Sidwell <nathan@codesourcery.com>
3091 Richard Sandiford <richard@codesourcery.com>
3092 Julian Brown <julian@codesourcery.com>
3094 * config.gcc (m68k-*-aout*, m68k-*-coff*, m68020-*-elf*, m68k-*-elf*)
3095 (m68k-*-uclinux*, m68k-*-linux*, m68k-*-rtems*): Set default_m68k_cpu
3096 to the configuration's default CPU.
3097 (m68010-*-netbsdelf*, m68k*-*-netbsdelf*, m68k*-*-openbsd*): Likewise.
3098 Remove default masks.
3099 (m680[012]0-*-*): Set the default with_cpu to the first part of
3101 (m68k*-*-*): Set the default with_cpu to m$default_m68k_cpu.
3102 (m68k*-*-linux): Extend the --with-cpu handling to...
3103 (m680[012]0-*-*, m68k*-*-*): ...these configurations. Allow m68000
3104 and m68010. Don't set target_cpu_default2.
3105 * doc/install.texi: Document --with-cpu for m68k.
3106 * config/m68k/m68k.h (OPTION_DEFAULT_SPECS): Define.
3107 * config/m68k/m68k-none.h (TARGET_CPU_DEFAULT, M68K_CPU_m68k)
3108 (M68K_CPU_m68000, M68K_CPU_m68010, M68K_CPU_m68020, M68K_CPU_m68030)
3109 (M68K_CPU_m68040, M68K_CPU_m68302, M68K_CPU_m68332, TARGET_DEFAULT)
3110 (ASM_CPU_DEFAULT_SPEC, CC1_CPU_DEFAULT_SPEC): Delete.
3111 (ASM_SPEC): Remove use of %(asm_cpu_default).
3112 (EXTRA_SPECS, SUBTARGET_EXTRA_SPECS, MULTILIB_DEFAULTS): Delete.
3113 * config/m68k/linux.h (TARGET_DEFAULT): Delete.
3114 (CPP_SPEC): Merge definitions. Do not handle __HAVE_68881__ here.
3115 * config/m68k/netbsd-elf.h (TARGET_OS_CPP_BUILTINS): Define
3116 __HAVE_FPU__ if TARGET_HARD_FLOAT.
3117 (TARGET_DEFAULT): Delete.
3118 (EXTRA_SPECS): Delete cpp_cpu_default_spec, cpp_cpu_spec,
3119 cpp_fpu_spec, asm_default_spec and netbsd_cpp_spec.
3120 (CPP_CPU_SPEC): Delete.
3121 (TARGET_VERSION): Merge definitions, using TARGET_68010 to pick
3122 the appropriate string.
3123 (CPP_CPU_DEFAULT_SPEC, ASM_DEFAULT_SPEC, CPP_FPU_SPEC): Delete.
3124 (CPP_SPEC): Define to NETBSD_CPP_SPEC.
3125 (ASM_SPEC): Don't use %(asm_default_spec).
3126 * config/m68k/m68k.c (TARGET_DEFAULT_TARGET_FLAGS): Remove
3127 TARGET_DEFAULT and add MASK_68881.
3128 * config/m68k/m68k.md: Remove mention of TARGET_DEFAULT from comments.
3130 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3132 * config.gcc (m68010-*-netbsdelf*): Add MASK_68010.
3133 (m68k*-*-netbsdelf*, m68k*-*-openbsd*, m68k*-linux*): Add
3134 MASK_68010 alongside MASK_68020.
3135 * doc/invoke.texi: Document -m68010.
3136 * config/m68k/m68k.opt (m68010): New.
3137 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Define mc68010
3139 (TUNE_68010): New macro.
3140 * config/m68k/m68k-none.h (M68K_CPU_m68k, M68K_CPU_m68010)
3141 (M68K_CPU_m68020, M68K_CPU_m68030, M68K_CPU_m68040)
3142 (M68K_CPU_m68332): Add MASK_68010.
3143 * config/m68k/linux.h (TARGET_DEFAULT): Add MASK_68010 to
3144 fallback definition.
3145 * config/m68k/netbsd-elf.h (CPP_CPU_SPEC): Remove now-redundant
3147 * config/m68k/m68k.c (MASK_ALL_CPU_BITS): Add MASK_68010.
3148 (m68k_handle_option): Handle OPT_m68010. Add MASK_68010
3149 to all entries that use MASK_68020.
3150 (output_move_simode_const, output_move_himode, output_move_qimode)
3151 (output_move_stricthi, output_move_strictqi): Use TARGET_68010
3152 instead of TARGET_68020 to select clr behavior. Remove comment
3153 about there being no TARGET_68010.
3154 * config/m68k/m68k.md: Likewise throughout.
3156 2007-01-12 Julian Brown <julian@codesourcery.com>
3158 * config/m68k/m68k.h (TARGET_ISAB): New macro.
3159 * config/m68k/m68k.c: Use TARGET_ISAB rather than TARGET_CFV4.
3160 * config/m68k/m68k.md: Likewise.
3162 2007-01-12 Julian Brown <julian@codesourcery.com>
3164 * config/m68k/m68k.h (LEGITIMATE_INDEX_P, LEGITIMIZE_ADDRESS): Use
3165 TARGET_COLDFIRE_FPU instead of TARGET_CFV4E.
3167 2007-01-12 Julian Brown <julian@codesourcery.com>
3169 * config/m68k/m68k.h (TUNE_68040_60): New macro.
3170 * config/m68k/m68k.c (standard_68881_constant_p): Use it.
3171 * config/m68k/m68k.md: Likewise.
3173 2007-01-12 Julian Brown <julian@codesourcery.com>
3174 Richard Sandiford <richard@codesourcery.com>
3176 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Use TUNE_68030
3177 instead of TARGET_68030, TUNE_68040 instead of TARGET_68040,
3178 TUNE_68060 instead of TARGET_68060 and TUNE_CPU32 instead of
3180 (TARGET_CPU32): Rename to...
3181 (TUNE_CPU32): ...this.
3182 (TUNE_68000_10, TUNE_68030, TUNE_68040, TUNE_68060)
3183 (TUNE_CFV2): New macros.
3184 * config/m68k/netbsd-elf.h (LONG_DOUBLE_TYPE_SIZE): Simplify;
3185 remove conditions that are implied by TARGET_68020.
3186 * config/m68k/m68k.c (m68k_output_function_prologue): Use TUNE_68040
3187 instead of TARGET_68040 and TUNE_CPU32 instead of TARGET_CPU32.
3188 (m68k_output_function_epilogue): Likewise.
3189 (m68k_rtx_costs): Likewise. Use TUNE_68060 instead of TARGET_68060
3190 and TUNE_CFV2 instead of TARGET_5200. Use TUNE_68000_10 instead of
3191 "!TARGET_68020 && !TARGET_COLDFIRE" to choose between 68000 and
3192 non-68000 timings. Refactor multiplication and division costs.
3193 (output_addsi3): Use TUNE_68040 instead of TARGET_68040 and
3194 TUNE_CPU32 instead of TARGET_CPU32.
3195 (standard_68881_constant_p): Use TUNE_68040 instead of TARGET_68040
3196 and TUNE_68060 instead of TARGET_68060.
3197 * config/m68k/m68k.md: Use TUNE_68040 instead of TARGET_68040,
3198 TUNE_68060 instead of TARGET_68060, and TUNE_CPU32 instead of
3200 (movsi_const0): Use TUNE_68000_10 rather than "!TARGET_68020
3201 && !TARGET_COLDFIRE" to choose between moveq and clr.
3202 Likewise in the unnamed movsf pattern.
3203 (ashlsi_17_24, lshrsi_17_24): Guard with TUNE_68000_10 rather than
3204 "!TARGET_68020 && !TARGET_COLDFIRE". Likewise the unnamed
3207 2007-01-12 Richard Sandiford <richard@codesourcery.com>
3209 * config/m68k/m68k.h (TARGET_CPU_CPP_BUILTINS): Increase amount
3210 of tabbing before backslashes.
3212 2007-01-11 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
3214 * pa-linux.h (ASM_OUTPUT_INTERNAL_LABEL): Undefine.
3215 * pa.h (ASM_OUTPUT_LABEL): Output colon when using GAS.
3216 (ASM_OUTPUT_INTERNAL_LABEL): Define.
3218 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
3220 * tree-ssa-loop-ivopts.c (extract_cond_operands): Split from
3221 find_interesting_uses_cond.
3222 (find_interesting_uses_cond): Use extract_cond_operands.
3223 (rewrite_use_compare): Use extract_cond_operands and
3224 force_gimple_operand_bsi. Do not call update_stmt.
3225 (determine_use_iv_cost_condition): Use extract_cond_operands.
3226 Return cheaper of using original bound and changing the exit bound.
3228 2007-01-11 Zdenek Dvorak <dvorakz@suse.cz>
3230 PR tree-optimization/29516
3231 * tree-ssa-address.c (tree_mem_ref_addr, add_to_parts,
3232 most_expensive_mult_to_index, addr_to_parts,
3233 create_mem_ref, maybe_fold_tmr): Make the type of
3234 fields of TARGET_MEM_REF sizetype.
3235 (move_fixed_address_to_symbol, move_pointer_to_base):
3237 * tree.def (TARGET_MEM_REF): Add comment on types of
3240 2007-01-11 Joseph Myers <joseph@codesourcery.com>
3242 * c-common.c (vector_types_convertible_p): Treat opaque types as
3243 always convertible if they have the same size, but not otherwise.
3245 2007-01-11 Steven Bosscher <steven@gcc.gnu.org>
3247 * ifcvt.c (struct noce_if_info): Add comments to the fields.
3248 Remove the b_unconditional field.
3249 (noce_try_sign_mask): Do not look at b_unconditional.
3250 (noce_process_if_block): Do not use merge_if_blocks. Update
3251 the CFG here. Do not set b_unconditional.
3252 (cond_move_process_if_block): Likewise.
3253 (find_cond_trap): Likewise.
3254 (check_cond_move_block): Require simple jump insns at the end
3257 2007-01-11 Jan Hubicka <jh@suse.cz>
3259 PR tree-optimization/1046
3260 * tree-tailcall.c (suitable_for_tail_call_opt_p): Use TREE_ADDRESSABLE
3261 when alias info is not ready.
3262 (pass_tail_recursion): Do not require aliasing.
3263 * tree-ssa-copyrename.c (pass_rename_ssa_cop): Likewise.
3264 * tree-ssa-ccp.c (pass_ccp, pass_fold_builtins): Likewise.
3265 * tree-ssa-copy.c (pass_copy_prop): Likewise.
3266 * tree-ssa-forwprop.c (pass_forwprop): Likewise.
3267 * tree-ssa-dce.c (pass_dce, pass_dce_loop, pass_cd_dce): Likewise.
3268 * passes.c (init_optimization_passes): Execute rename_ssa_copies,
3269 ccp, forwprop, copy_prop, merge_phi, copy_prop, dce and tail recursion
3271 * tree-ssa-operands.c (add_virtual_operand, get_indirect_ref_operand):
3272 When aliasing is not build, mark statement as volatile.
3274 2007-01-11 Tom Tromey <tromey@redhat.com>
3276 PR preprocessor/15185, PR preprocessor/20989:
3277 * doc/cppopts.texi <-MT>: Update description of algorithm for
3278 computing default target.
3279 <-M, -MD>: Reword "basename" text.
3281 2007-01-11 Roger Sayle <roger@eyesopen.com>
3283 * builtins.c (expand_builtin_pow, expand_builtin_powi,
3284 fold_builtin_cabs, fold_builtin_sqrt, fold_builtin_trunc,
3285 fold_builtin_floor, fold_builtin_ceil, fold_builtin_round,
3286 fold_builtin_int_int_roundingfn, fold_builtin_bitop,
3287 fold_builtin_bswap, real_constp, fold_builtin_pow,
3288 fold_builtin_powi, fold_builtin_signbit, fold_builtin_copysign,
3289 do_mpfr_arg1, do_mpfr_arg2, do_mpfr_arg3, do_mpfr_sincos): Replace
3290 uses of the macro TREE_CONSTANT_OVERFLOW with TREE_OVERFLOW.
3291 * convert.c (convert_to_pointer): Likewise.
3292 * expr.c (highest_pow2_factor, expand_expr_real_1): Likewise.
3293 * fold-const.c (force_fit_type, fold_negate_expr, int_const_binop,
3294 const_binop, fold_convert_const_int_from_int,
3295 fold_convert_const_int_from_real,
3296 fold_convert_const_real_from_real, sign_bit_p,
3297 optimize_minmax_comparison, extract_muldiv_1, fold_div_compare,
3298 fold_sign_changed_comparison, fold_unary, fold_comparison,
3299 fold_binary, multiple_of_p, tree_Expr_non_zero_p,
3300 fold_negate_const, fold_abs_const, fold_not_const): Likewise.
3301 * print-tree.c (print_node_brief, print_node): Likewise.
3302 * stor-layout.c (place_field, layout_type): Likewise.
3303 * tree-chrec.c (keep_cast): Likewise.
3304 * tree.c (build_vector, build_real, build_real_from_int_cst,
3305 build_complex): Likewise.
3307 2007-01-11 Roger Sayle <roger@eyesopen.com>
3309 * tree.h (TREE_CONSTANT_OVERFLOW): Obsolete. For the time being,
3310 treat TREE_CONSTANT_OVERFLOW as a synonym of TREE_OVERFLOW.
3312 2007-01-11 Paolo Bonzini <bonzini@gnu.org>
3314 * configure.ac (strict1_warn): Rename to strict_warn.
3315 (WERROR, --enable-werror, symlink hacks, stage1_cflags,
3316 cc_set_by_configure, quoted_cc_set_by_configure,
3317 stage_prefix_set_by_configure, quoted_stage_prefix_set_by_configure,
3318 all_boot_languages, all_stagestuff): Remove.
3319 (target_list): Remove bootstrap targets.
3320 * Makefile.in (quickstrap): Unconditionally make a synonym of all.
3321 (BOOT_LANGUAGES, STAGE1_CFLAGS, STAGE1_CHECKING,
3322 REMAKEFLAGS, FLAGS_TO_PASS, PREPEND_DOTDOT_TO_RELATIVE_PATHS,
3323 SUBDIR_FLAGS_TO_PASS, WERROR_FLAGS, STRICT2_WARN, LANG_STAGESTUFF,
3324 VOL_FILES, POSTSTAGE1_FLAGS_TO_PASS, STAGE2_FLAGS_TO_PASS,
3325 STAGEPROFILE_FLAGS_TO_PASS, STAGEFEEDBACK_FLAGS_TO_PASS, stage1_build,
3326 stage1_copy, stage2_build, stage2_copy, stageprofile_build,
3327 stageprofile_copy, stage3_build, stage3_copy, stagefeedback_build,
3328 stagefeedback_copy, stage4_build, clean_s1, clean_sw, bootstrap,
3329 bootstrap-lean, bootstrap2, bootstrap2-lean, bootstrap3,
3330 bootstrap3-lean, bootstrap4, bootstrap4-lean, unstage1, unstage2,
3331 unstage3, unstage4, unstageprofile, unstagefeedback, restage, restage2,
3332 restage3, restage4, restageprofile, restagefeedback, bubbleestrap,
3333 cleanstrap, unstrap, restrap, *compare, *compare3, *compare4,
3334 *compare-lean, *compare3-lean, *compare4-lean, stage1-start, stage1,
3335 stage2-start, stage2, stage3-start, stage3, stage4-start, stage4,
3336 stageprofile-start, stageprofile, stagefeedback-start, stagefeedback,
3337 risky-stage1, risky-stage2, risky-stage3, risky-stage4): Remove.
3338 (ORDINARY_FLAGS_TO_PASS): Rename to FLAGS_TO_PASS.
3339 (STAGECOPYSTUFF, STAGEMOVESTUFF): Consolidate into MOSTLYCLEANFILES.
3340 (mostlyclean): Adjust.
3341 (clean, distclean): Don't mention bootstrap stuff.
3342 * configure: Regenerate.
3343 * ada/config-lang.in, cp/config-lang.in, forttran/config-lang.in,
3344 java/config-lang.in, objc/config-lang.in, objcp/config-lang.in,
3345 treelang/config-lang.in (stagestuff): Remove.
3346 * doc/sourcebuild.texi (stage1, stage2, stage3, stage4,
3347 stageprofile, stagefeedback, stagestuff): Remove mention.
3349 2007-01-11 Nick Clifton <nickc@redhat.com>
3351 * config/mcore/predicates.md (mcore_general_movesrc_operand):
3353 (mcore_general_movdst_operand): Do not accept CONST_INTs.
3354 (mcore_arith_K_S_operand): Run the test for the S constraint not
3355 the test for the M constraint.
3356 (mcore_addsub_operand): Do not accept integer values that are
3357 larger than 32 bits.
3358 * config/mcore/mcore.md: Remove unused constraints from split.
3359 (andsi3): Use HOST_WIDE_INT instead of int to hold an INTVAL.
3361 (allocate_stack): Likewise.
3362 * config/mcore/mcore.c (mcore_print_operand): Restrict output of P
3363 operands to 32 bits.
3364 (mcore_const_costs): Use HOST_WIDE_INT instead of int to hold an
3366 (mcore_and_cost, mcore_modify_comparison, const_ok_for_mcore,
3367 mcore_const_ok_for_inline, mcore_const_trick_uses_not,
3368 try_constant_tricks, mcore_num_ones, mcore_num_zeros,
3369 mcore_output_bclri, mcore_output_andn, output_inline_const,
3370 mcore_output_move, mcore_output_movedouble): Likewise.
3371 (mcore_output_cmov): Use CONST_OK_FOR_M and CONST_OK_FOR_N.
3372 (output_inline_const): Likewise.
3373 (output_inline_const): Fix format strings used in sprintf
3375 * config/mcore/mcore-protos.h: Update prototypes for changed
3376 functions in mcore.c.
3377 * config/mcore/mcore.h (CONST_OK_FOR_I): Cast values to
3378 HOST_WIDE_INT and not int.
3379 (CONST_OK_FOR_J, CONST_OK_FOR_K, CONST_OK_FOR_L, CONST_OK_FOR_M,
3380 CONST_OK_FOR_N): Likewise.
3381 (LEGITIMATE_CONSTANT_P): Also check CONSTANT_P.
3382 (GO_IF_LEGITIMATE_INDEX): Use HOST_WIDE_INT instead of int to hold
3385 2007-01-10 Jan Hubicka <jh@suse.cz>
3387 * tree-vrp.c (remove_range_assertions): Release defs.
3388 * tree-ssa-loop-ivopts.c (rmeove_statement): Likewise.
3389 * tree-ssa-dom.c (remove_stmt_or_phi): Likewise.
3391 2007-01-10 Paul Brook <paul@codesourcery.com>
3393 * config/arm/arm.c (arm_rtx_costs_1): Handle mutiply-subtract.
3394 * config/arm/arm.md (mulsi3subsi): New insn.
3396 2007-01-10 Zdenek Dvorak <dvorakz@suse.cz>
3398 * tree-ssa-loop-manip.c (tree_unroll_loop): Make it a wrapper over ...
3399 (tree_transform_and_unroll_loop): New.
3400 * tree-flow.h (transform_callback, tree_transform_and_unroll_loop):
3403 2007-01-10 Robert Kennedy <jimbob@google.com>
3405 * fold-const.c (fold_comparison): Fold comparisons like (x *
3406 1000 < 0) to (x < 0).
3408 2007-01-10 Ian Lance Taylor <iant@google.com>
3410 * tree-pretty-print.c (dump_generic_node): Print parentheses when
3411 operands have the same priority.
3413 2007-01-10 Tom Tromey <tromey@redhat.com>
3415 * fold-const.c (fold_truthop): Don't check can_use_bit_fields_p.
3416 (fold_binary): Likewise.
3417 * langhooks.c (lhd_can_use_bit_fields_p): Removed.
3418 * langhooks-def.h (lhd_can_use_bit_fields_p): Removed.
3419 (LANG_HOOKS_CAN_USE_BIT_FIELDS_P): Removed.
3420 (LANG_HOOKS_INITIALIZER): Remove LANG_HOOKS_CAN_USE_BIT_FIELDS_P.
3421 * langhooks.h (struct lang_hooks): Removed field
3422 'can_use_bit_fields_p'.
3424 2007-01-10 Ralf Corsépius <ralf.corsepius@rtems.org>
3426 * config/bfin/t-bfin, config/bfin/t-bfin-elf: Remove GCC_CFLAGS.
3428 2007-01-10 Razya Ladelsky <razya@il.ibm.com>
3430 * function.c (get_last_funcdef_no): New function.
3431 * function.h (get_last_funcdef_no): Declare.
3432 * tree-inline.c (initialize_cfun): Add initialization.
3433 (tree_function_versioning): Cleanup.
3435 2007-01-10 Jan Hubicka <jh@suse.cz>
3437 * tree-inline.c (setup_one_parameter): Do not propagate into abnormal
3440 2007-01-10 Sa Liu <saliu@de.ibm.com>
3441 Ben Elliston <bje@au.ibm.com>
3443 * spu.h (STACK_SAVE_AREA): Use VOIDmode for SAVE_FUNCTION, SImode
3444 for SAVE_NONLOCAL and Pmode for any other save level.
3445 * spu-protos.h (spu_restore_stack_block): Declare.
3446 * spu.md (save_stack_block): Remove.
3447 (restore_stack_block): Call spu_restore_stack_block.
3448 * spu.c (spu_restore_stack_block): New function.
3449 (spu_expand_epilogue): Remove old comment.
3451 2007-01-09 Zdenek Dvorak <dvorakz@suse.cz>
3453 PR tree-optimization/30322
3454 * tree-ssa-loop-ivopts.c (fold_affine_expr, iv_value): Removed.
3455 (cand_value_at): Return the value as aff_tree.
3456 (may_eliminate_iv): Convert the bound from aff_tree to tree.
3457 * tree-affine.c (aff_combination_add_cst, aff_combination_add_product,
3458 aff_combination_mult): New functions.
3459 (aff_combination_add): Use aff_combination_add_cst.
3460 (aff_combination_convert): Allow conversions to a wider type.
3461 (tree_to_aff_combination): Handle BIT_NOT_EXPR.
3462 * tree-affine.h (aff_combination_mult): Declare.
3464 2007-01-09 Carlos O'Donell <carlos@codesourcery.com>
3466 * doc/tm.texi: Update documentation to reflect reality of exec
3467 and start file search behaviours. Update copyright year.
3468 * doc/invoke.texi: Explain how GCC_EXEC_PREFIX is used to find
3469 header file directories.
3471 2007-01-09 Uros Bizjak <ubizjak@gmail.com>
3473 * config/i386/i386.md (*sinxf2): Rename to *sinxf2_i387.
3474 (*cosxf2): Rename to cosxf2_i387.
3475 (*sindf2, *sinsf2): Extend operand 1 to XFmode. Macroize patterns
3476 using X87MODEF12 mode macro. Rename patterns to
3477 *sin_extend<mode>xf2_i387. Use SSE_FLOAT_MODE_P to disable patterns
3479 (*cosdf2, *cossf2): Ditto.
3480 (sincosdf3, sincossf3): Ditto. Rewrite corresponding splitters
3481 to match extended input operands.
3482 (sincos<mode>3): New expander.
3483 (*sinextendsfdf2, *cosextendsfdf2, *sincosextendsfdf3): Remove
3484 insn patterns and corresponding splitters.
3486 2007-01-09 Kaz Kojima <kkojima@gcc.gnu.org>
3488 * config/sh/t-linux (TARGET_LIBGCC2_CFLAGS): Delete.
3489 (SHLIB_MAPFILES, SHLIB_LINK, SHLIB_INSTALL): Likewise.
3491 2007-01-09 Nicolas Pitre <nico@cam.org>
3494 * arm/ieee754-df.S (Lad_s): Also test the low word of X for zero.
3496 2007-01-08 Geoffrey Keating <geoffk@apple.com>
3498 * target.h (struct gcc_target): New field library_rtti_comdat.
3499 * target-def.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): New.
3500 (TARGET_CXX): Add TARGET_CXX_LIBRARY_RTTI_COMDAT.
3501 * doc/tm.texi (C++ ABI): Document TARGET_CXX_LIBRARY_RTTI_COMDAT.
3502 * config/darwin.h (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
3504 2007-01-08 Geoffrey Keating <geoffk@apple.com>
3506 * doc/invoke.texi (Optimize Options): Correct description of -O0.
3508 2007-01-08 Richard Guenther <rguenther@suse.de>
3510 * tree.h (force_fit_type_double): Export.
3511 (force_fit_type): Remove.
3512 * fold-const.c (force_fit_type_double): New function.
3513 (force_fit_type): Remove.
3514 (int_const_binop): Use it.
3515 (fold_convert_const_int_from_int): Likewise.
3516 (fold_convert_const_int_from_real): Likewise.
3517 (fold_div_compare): Likewise.
3518 (fold_sign_changed_comparison): Likewise.
3519 (fold_unary): Likewise.
3520 (fold_negate_const): Likewise.
3521 (fold_abs_const): Likewise.
3522 (fold_not_const): Likewise.
3523 * c-common.c (shorten_compare): Use force_fit_type_double.
3524 * convert.c (convert_to_pointer): Likewise.
3526 2007-01-08 Richard Guenther <rguenther@suse.de>
3528 * tree.h (build_int_cst_wide_type): Export.
3529 * tree.c (build_int_cst_wide_type): New function.
3530 (build_int_cst_wide): Fix comment.
3531 * builtins.c (fold_builtin_object_size): Use build_int_cst
3532 to build -1 or 0 of the correct type. Use fit_double_type
3533 to check for overflow.
3534 * fold-const.c (optimize_bit_field_compare): Use build_int_cst_type
3536 (decode_field_reference): Likewise.
3537 (all_ones_mask_p): Likewise.
3538 (native_interpret_int): Use build_int_cst_wide_type.
3539 (fold_binary): Use build_int_cst_type to build an all-ones
3541 * stor-layout.c (set_sizetype): Use build_int_cst_wide_type.
3543 2007-01-08 Daniel Jacobowitz <dan@codesourcery.com>
3545 * config/pa/t-pa64 (libgcc_stub.a): Use $(T).
3547 2007-01-09 Ben Elliston <bje@au.ibm.com>
3549 * genautomata.c (STATS_OPTION): New option.
3550 (stats_flag): New flag.
3551 (gen_automata_option): Handle it.
3552 (initiate_automaton_gen): Ditto.
3553 (write_automata): Output statistics only if stats_flag is
3554 set. Likewise, output time statistics only if time_flag is set.
3555 * doc/md.texi (Processor pipeline description): Document new flag.
3557 2007-01-08 Richard Guenther <rguenther@suse.de>
3559 * builtins.c (fold_builtin_int_roundingfn): Use fit_double_type.
3560 * tree.c (build_int_cst_type): Likewise.
3561 (size_in_bytes): Don't call force_fit_type on the result.
3562 (int_fits_type_p): Use fit_double_type.
3563 * fold-const.c (fit_double_type): New function.
3564 (force_fit_type): Use it.
3565 * tree.h (fit_double_type): Export.
3567 2007-01-08 Jan Hubicka <jh@suse.cz>
3569 * tree-vectorizer.c (gate_increase_alignment): Fix return type.
3570 * ipa.c (function_and_variable_visibility): Fix return type.
3572 2007-01-08 Richard Guenther <rguenther@suse.de>
3574 * tree-ssa-ccp.c (maybe_fold_offset_to_array_ref): Use type
3575 of offset to build the index.
3576 * tree-pretty-print.c (dump_generic_node): Don't build negated
3577 const just for printing.
3578 * c-pretty-print.c (pp_c_integer_constant): Likewise.
3579 * builtins.c (fold_builtin_int_roundingfn): Check if result
3580 fits the type by using force_fit_type and comparing the result.
3581 * predict.c (predict_loops): Use compare_tree_int for comparison.
3582 * tree.c (build_int_cst): Fall back to integer_type_node for
3584 (build_int_cst_wide): Assert type is non-null.
3586 2007-01-08 Roberto Costa <roberto.costa@st.com>
3588 * tree-vrp.c (extract_range_from_cond_expr): New.
3589 (extract_range_from_expr): Handle COND_EXPR nodes used as expressions.
3590 * tree-ssa-ccp.c (get_maxval_strlen): Handle COND_EXPR nodes used
3592 (fold_stmt): Bug fix, avoid infinite recursion when folding COND_EXPRs.
3593 * tree-ssa-forwprop.c (simplify_cond, forward_propagate_into_cond,
3594 tree_ssa_forward_propagate_single_use_vars): Handle COND_EXPR nodes
3595 used as expressions.
3596 * tree-object-size.c (cond_expr_object_size): New.
3597 (collect_object_sizes_for): Handle COND_EXPR nodes used as expressions.
3599 2007-01-08 Jan Hubicka <jh@suse.cz>
3601 * tree-ssa-forwprop.c (forward_propagate_into_cond,
3602 tree_ssa_forward_propagate_single_use_va): Release defs of propagated
3605 2007-01-08 Richard Guenther <rguenther@suse.de>
3607 PR tree-optimization/23603
3608 * tree-vrp.c (set_value_range_to_truthvalue): New function.
3609 (extract_range_from_binary): Fall back to truthvalue instead of
3610 varying for TRUTH_*_EXPR.
3611 (extract_range_from_comparison): Fall back to truthvalue instead of
3613 (vrp_visit_phi_node): Don't adjust new range bounds to +INF/-INF
3614 if all visited PHI values were constant.
3616 2007-01-08 Jan Hubicka <jh@suse.cz>
3618 * cgraphunit.c (cgraph_process_new_functions): Reset reachable flag.
3619 (cgraph_analyze_function): break out from ...
3620 (cgraph_finalize_compilation_unit): ... here.
3621 (cgraph_expand_function): Remove forgoten commented out line.
3622 (cgraph_optimize): Analyze functions.
3624 2007-01-08 Jan Hubicka <jh@suse.cz>
3626 * tree-pas.h (TODO_remove_function): New flag.
3627 (TODO_update*): Renumber.
3628 (pass_ipa_increase_alignment,
3629 pass_ipa_function_and_variable_visibility): New passes.
3630 * cgraphunit.c (cgraph_increase_alignment): Move to tree-vectorizer.c
3631 (cgraph_function_and_variable_visibility): Move to ipa.c
3632 (cgraph_optimize): Don't call cgraph_function_and_variable_visibility,
3633 cgraph_increase_alignment.
3634 * ipa-inline.c (cgraph_decide_inlining): Don't push timevar.
3635 (cgraph_decide_inlining_incrementally): Push TV_INTEGRATION before
3636 calling tree-inline.
3637 (cgraph_early_inlining): Do not call cgraph_remove_unreachable_nodes.
3638 (pass_ipa_inline, pass_early_ipa_inlining): Set TODO_remove_functions
3639 * tree-vectorizer.c (increase_alignment): Move here from cgraphunit.c
3640 (gate_increase_alignment): New function.
3641 (pass_ipa_increase_alignment): New pass.
3642 * ipa.c: Inline tree-pass.h and timevar.h
3643 (function_and_variable_visibility): Move here from cgraphunit.c
3644 * tree-optimize.c (pass_early_local_passes): Add TODO_remove_functions.
3645 * passes.c (init_optimization_passes): Add the two new passes.
3646 (execute_todo): Handle cgraph_remove_functions.
3648 2007-01-08 Nick Clifton <nickc@redhat.com>
3650 * config/frv/predicates.md (reg_or_0_operand): Accept
3653 2007-01-08 Ralf Corsépius <ralf.corsepius@rtems.org>
3655 * config/bfin/rtems.h, config/bfin/t-rtems: New.
3656 * config.gcc: Add bfin*-rtems*.
3658 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3660 * c.opt: Add -flax-vector-conversions.
3661 * c-typeck.c (convert_for_assignment): Pass flag to
3662 vector_types_convertible_p to allow emission of note.
3663 (digest_init): Likewise.
3664 * c-opts.c: Handle -flax-vector-conversions.
3665 * c-common.c (flag_lax_vector_conversions): New.
3666 (vector_types_convertible_p): Unless -flax-vector conversions
3667 has been passed, disallow conversions between vectors with
3668 differing numbers of subparts and/or element types. If such
3669 a conversion is disallowed, possibly emit a note on the first
3670 occasion only to inform the user of -flax-vector-conversions.
3671 The new last argument specifies this.
3672 * c-common.h (flag_lax_vector_conversions): New.
3673 (vector_types_convertible_p): Add extra argument.
3674 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use
3675 char_type_node for V*QI type vectors.
3676 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3677 Update to satisfy new typechecking rules.
3678 * config/rs6000/altivec.h (vec_cmple): Use vec_cmpge, for both
3680 * doc/invoke.texi (C Dialect Options): Document
3681 -flax-vector-conversions.
3683 2007-01-08 Mark Shinwell <shinwell@codesourcery.com>
3685 PR tree-optimization/29877
3686 * tree-ssa-ter.c (is_replaceable_p): Deem assignments with
3687 a register variable on the RHS to not be replaceable.
3689 2007-01-08 Chen Liqin <liqin@sunnorth.com.cn>
3690 * config/score/t-score-elf (MULTILIB_OPTIONS): Change.
3691 * config/score/predicates.md (const_uimm5, sr0_operand, const_simm12,
3692 const_simm15, const_pow2, const_npow2): Added.
3693 * config/score/misc.md (insv, extv, extzv, movmemsi,
3694 move_lbu_a/b, mov_lhu_a/b etc): Added and fix some bug.
3695 * config/score/score.c (score_address_cost, score_select_cc_mode):
3697 Change CONST_OK_FOR_LETTER_P/EXTRA_CONSTRAINT define.
3698 Update score_rtx_costs for MACRO TARGET_RTX_COSTS.
3699 Update score_print_operand.
3700 * config/score/score.h (DATA_ALIGNMENT, SELECT_CC_MODE): Added.
3701 Adjust register allocate order and update some macro define.
3702 * config/score/score-mdaux.c (mdx_unaligned_load, mdx_unsigned_store,
3703 mdx_block_move_straight, mdx_block_move_loop_head,
3704 mdx_block_move_loop_body, mdx_block_move_loop_foot, mdx_block_move_loop,
3705 mdx_block_move): Added.
3706 (mdx_movsicc, mdp_select_add_imm, mdp_select, mds_zero_extract_andi,
3707 mdp_limm): Updated and fix some bug and typo.
3708 * config/score/score.md (movqi/hi/si, add/sub/zero/ext): Updated.
3709 (movsf, movdf, doloop_end): Added.
3711 2007-01-08 Kazu Hirata <kazu@codesourcery.com>
3713 * config/arm/arm.c, config/arm/arm.h, config/arm/arm.md,
3714 config/arm/thumb2.md: Fix comment typos.
3715 * doc/extend.texi: Fix a typo.
3717 2007-01-07 Eric Christopher <echristo@apple.com>
3719 * configure.ac: Check for __stack_chk_fail for darwin.
3720 * configure: Regenerate.
3722 2007-01-07 Richard Guenther <rguenther@suse.de>
3724 * tree-vrp.c (extract_range_from_assert): CSE calls to
3725 compare_values where possible.
3726 (extract_range_from_unary_expr): Likewise.
3728 2007-01-07 Anatoly Sokolov <aesok@post.ru>
3730 * config/avr/avr-protos.h (call_insn_operand): Delete prototype.
3731 * config/avr/avr.c (call_insn_operand): Delete function.
3732 * config/avr/avr.md (*pushqi, *pushhi, *pushsi, *pushsf): Use REG_SP
3733 instead of register number. Use predicates.
3734 * config/avr/predicates.md (const0_operand, reg_or_0_operand,
3735 call_insn_operand): Add.
3737 2007-01-06 Jan Hubicka <jh@suse.cz>
3739 * tree-pass.h (pass_build_cgraph_edges): Declare.
3740 * cgraphunit.c (record_refernece): Move to cgraphbuild.c
3741 (visited_nodes): Remove.
3742 (cgraph_create_edges): Move to cgraphbuild.c; rename to
3743 build_cgrpah_edges; make visited_nodes local.
3744 (cgraph_process_new_functions): DO not call initialize_inline_failed.
3745 (record_references_in_initializer): Move to cgraphbuild.c
3746 (initialize_inline_failed, rebuild_cgraph_edges,
3747 pass_rebuild_cgraph_edges): Move to cgraphbuild.c.
3748 (verify_cgraph_node): Make visited_nodes local.
3749 (cgraph_analyze_function): Do not call cgraph_create_edges and
3750 initialize_inline_failed.
3751 (cgraph_expand_function): Do not call cgraph_lower_function;
3752 assert that function is already lowered.
3753 * Makefile.in (cgraphbuild.o): New.
3754 * passes.c (init_optimization_passes): Add pass_build_cgraph_edges
3755 at the end of lowering passes.
3757 2007-01-06 Steven Bosscher <steven@gcc.gnu.org>
3759 * ifcvt.c (cond_move_convert_if_block): New function, code
3760 factored out from...
3761 (cond_move_process_if_block): ...here. Call the new function
3762 on the THEN and ELSE blocks.
3763 (merge_if_block): Do not copy global_live_at_end, merge_blocks
3764 already takes care of this.
3766 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3769 * tree.h (TREE_OVERFLOW_P): New.
3770 * c-typeck.c (parser_build_unary_op): Warn only if result
3771 overflowed and operands did not.
3772 (parser_build_binary_op): Likewise.
3773 (convert_for_assignment): Remove redundant overflow_warning.
3774 * c-common.c (overflow_warning): Don't check or set TREE_OVERFLOW.
3776 2007-01-05 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3778 * c-typeck.c (store_init_value): Split over two lines to follow
3779 the GNU coding style.
3781 2007-01-05 Benjamin Kosnik <bkoz@redhat.com>
3783 * c-cppbuiltin.c (c_cpp_builtins): __GXX_EXPERIMENTAL_CPP0X__ to
3784 __GXX_EXPERIMENTAL_CXX0X__.
3785 * doc/cpp.texi: Same.
3787 2007-01-05 Richard Guenther <rguenther@suse.de>
3790 * tree.c (get_narrower): Do not construct COMPONENT_REFs
3791 with mismatched types. Instead explicitly build a
3792 conversion NOP_EXPR.
3794 2007-01-05 Ian Lance Taylor <iant@google.com>
3796 * c-common.c (decl_with_nonnull_addr_p): New function.
3797 (c_common_truthvalue_conversion): Call it.
3798 * c-typeck.c (build_binary_op): Likewise.
3799 * c-common.h (decl_with_nonnull_addr_p): Declare.
3801 2007-01-05 Jakub Jelinek <jakub@redhat.com>
3804 * libgcc2.c (__divdc3): Compare c and d against 0.0 instead of
3807 2007-01-05 Joel Brobecker <brobecker@adacore.com>
3809 * doc/install.texi (Final install): Document the fact that
3810 the GNAT runtime should not be stripped.
3812 2007-01-04 Jan Hubicka <jh@suse.cz>
3814 * tree-inline.c (fold_marked_statements): Update operand caches
3815 and EH after folding
3817 2007-01-04 Ian Lance Taylor <iant@google.com>
3819 * c-common.c (check_function_nonnull): Whitespace fix.
3821 2007-01-04 Jan Hubicka <jh@suse.cz>
3823 * tree-optimize.c (execute_fixup_cfg): Correct previously mistakely
3824 comitted older version of patch.
3825 (pass_fixup_cfg): Add TODOs to verify flow and statements, dump
3826 function, celanup cfg and collect garbage.
3828 2007-01-04 Mike Stump <mrs@apple.com>
3830 * Makefile.in (mostlyclean): Don't remove libgcc anymore.
3833 2007-01-04 Eric Christopher <echristo@apple.com>
3835 * libgcc2.c (__bswapsi2): Use SItype.
3836 (__bswapdi2): Use DItype.
3837 * libgcc2.h: Update for above.
3839 2007-01-04 Paul Brook <paul@codesourcery.com>
3841 * config/arm/arm.md (arm_mulsi3, thumb_mulsi3, mulsi3_compare0,
3842 mulsi_compare0_scratch, mulsi3addsi, mulsi3addsi_compare0,
3843 mulsi3addsi_compare0_scratch, mulsidi3adddi, mulsidi3,
3844 umulsidi3, umulsidi3adddi, smulsi3_highpart,
3845 umulsi3_highpart): Make conditional on !arm_arch6.
3846 (arm_mulsi3_v6, thumb_mulsi3_v6, mulsi3_compare0_v6,
3847 mulsi_compare0_scratch_v6, mulsi3addsi_v6, mulsi3addsi_compare0_v6,
3848 mulsi3addsi_compare0_scratch_v6, mulsidi3adddi_v6, mulsidi3_v6,
3849 umulsidi3_v6, umulsidi3adddi_v6, smulsi3_highpart_v6,
3850 umulsi3_highpart_v6): New insns.
3852 2007-01-04 Roger Sayle <roger@eyesopen.com>
3854 * fold-const.c (fold_convert): When casting an expression to void,
3855 fold_ignored_result may discover a GIMPLE_MODIFY_STMT which doesn't
3856 have a type. Instead of attempting to build a NOP_EXPR, return
3857 these "special" trees directly.
3859 2007-01-04 Joseph Myers <joseph@codesourcery.com>
3861 * config/rs6000/rs6000.c (rs6000_rtx_costs): Make adjustment for
3862 MULT inside MINUS as either argument. Use rs6000_cost->dmul -
3863 rs6000_cost->fp not 0 as adjustment for outer NEG.
3865 2007-01-04 Jan Hubicka <jh@suse.cz>
3867 * cgraph.c (cgraph_release_function_body): New function.
3868 (cgraph_remove_node): Use it.
3869 * cgraph.h (cgraph_release_function_body): Declare.
3870 * cgraphunit.c (cgraph_expand_function): Use it.
3871 * ipa.c (cgraph_remove_unreahchable_nodes): Use it.
3872 * tree-ssa.c (delete_tree_ssa): Allow to be called before aliasing
3873 is initialized and while compilation of other function is running.
3874 * tree-optimize.c (execute_free_cfg_annotations): Move code to clear
3875 statement CFG annotations from here to ...
3876 * tree-cfg.c (delete_tree_cfg_annotations): ... here.
3878 2007-01-04 Zdenek Dvorak <dvorakz@suse.cz>
3880 * cfgloop.h (enum li_flags): Make the constants powers of two.
3882 2007-01-04 Jan Hubicka <jh@suse.cz>
3884 * tree-inline.c (copy_bb): Insert new statements to statements_to_fold
3886 (fold_marked_statements): New function.
3887 (optimize_inline_calls, tree_function_versioning): Fold new statements.
3888 * tree-inline.h (copy_body_data): Add statements_to_fold.
3890 2007-01-03 Daniel Jacobowitz <dan@codesourcery.com>
3892 * config.gcc: Mention libgcc/config.host.
3893 * Makefile.in: Update comments mentioning libgcc.
3894 (LIBGCC, INSTALL_LIBGCC, GCC_PARTS, mklibgcc): Delete.
3895 (all.cross, start.encap, rest.encap, rest.cross): Update
3896 dependencies for libgcc move.
3897 (libgcc.mk, LIBGCC_DEPS, libgcov.a, libgcc.a, stmp-multilib)
3898 (clean-target, clean-target-libgcc): Delete.
3899 (srcdirify, GCC_EXTRA_PARTS): New macros.
3900 (libgcc-support, libgcc.mvars): New rules.
3901 (distclean): Remove mention of mklibgcc.
3902 (install): Don't reference INSTALL_LIBGCC.
3903 (install-common): Don't reference EXTRA_PARTS.
3904 (install-libgcc, install-multilib): Delete rules.
3905 * mklibgcc.in: Delete file.
3906 * doc/configfiles.texi: Don't mention mklibgcc.
3908 * config/i386/t-darwin (SHLIB_VERPFX): Delete (moved to libgcc).
3909 * config/i386/t-darwin64 (SHLIB_VERPFX): Likewise.
3910 * config/rs6000/t-darwin (SHLIB_VERPFX): Likewise.
3911 * config/rs6000/t-ppccomm (TARGET_LIBGCC2_CFLAGS, SHLIB_MAPFILES)
3912 (mklibgcc, ldblspecs): Likewise.
3914 * config/i386/t-nwld (libgcc.def, libc.def, libpcre.def)
3915 (posixpre.def): Use $(T).
3916 (SHLIB_EXT, SHLIB_NAME, SHLIB_SLIBDIR_QUAL, SHLIB_DEF, SHLIB_MAP)
3917 (SHLIB_SRC, SHLIB_INSTALL): Delete.
3918 (SHLIB_LINK): Make dummy.
3919 * config/t-slibgcc-darwin: Delete contents except for dummy SHLIB_LINK.
3921 * config/frv/t-linux (EXTRA_MULTILIB_PARTS): Clear.
3923 * config/alpha/t-crtfm: Use $(T) in rules for EXTRA_PARTS.
3924 * config/alpha/t-vms, config/alpha/t-vms64, config/fr30/t-fr30,
3925 config/i386/t-rtems-i386, config/ia64/t-ia64, config/rs6000/t-beos,
3926 config/rs6000/t-newas, config/sparc/t-elf: Likewise.
3928 * configure.ac (all_outputs): Remove mklibgcc.
3929 * configure: Regenerated.
3931 2007-01-03 Josh Conner <jconner@apple.com>
3934 * calls.c (compute_argument_addresses): Set stack and stack_slot
3935 for partial args, too.
3936 (store_one_arg): Use locate.size.constant for the size when
3937 generating a save_area.
3939 2007-01-03 Robert Kennedy <jimbob@google.com>
3941 * tree-cfg.c (tree_merge_blocks): Release SSA_NAME phi results
3942 whose definitions are deleted due to basic block merging.
3944 2007-01-03 Paul Brook <paul@codesourcery.com>
3947 * config/arm/arm.c (output_return_instruction): Pop PC in interrupt
3949 (use_return_insn): Return 0 for Thumb interrupt functions.
3950 (print_multi_reg): Add rfe argument for IRQ returns.
3951 (arm_output_epilogue): Pop interrupt return address directly into PC.
3952 (arm_expand_prologue): Only adjust IRQ return address in Arm mode.
3954 2007-01-03 Paul Brook <paul@codesourcery.com>
3956 Merge from sourcerygxx-4_1.
3957 * config/arm/thumb2.md: New file.
3958 * config/arm/elf.h (JUMP_TABLES_IN_TEXT_SECTION): Return True for
3960 * config/arm/coff.h (JUMP_TABLES_IN_TEXT_SECTION): Ditto.
3961 * config/arm/aout.h (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3962 (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump tables.
3963 * config/arm/aof.h (ASM_OUTPUT_ADDR_DIFF_ELT): Output Thumb-2 jump
3965 (ASM_OUTPUT_ADDR_VEC_ELT): Add !Thumb-2 assertion.
3966 * config/arm/ieee754-df.S: Use macros for Thumb-2/Unified asm
3968 * config/arm/ieee754-sf.S: Ditto.
3969 * config/arm/arm.c (thumb_base_register_rtx_p): Rename...
3970 (thumb1_base_register_rtx_p): ... to this.
3971 (thumb_index_register_rtx_p): Rename...
3972 (thumb1_index_register_rtx_p): ... to this.
3973 (thumb_output_function_prologue): Rename...
3974 (thumb1_output_function_prologue): ... to this.
3975 (thumb_legitimate_address_p): Rename...
3976 (thumb1_legitimate_address_p): ... to this.
3977 (thumb_rtx_costs): Rename...
3978 (thumb1_rtx_costs): ... to this.
3979 (thumb_compute_save_reg_mask): Rename...
3980 (thumb1_compute_save_reg_mask): ... to this.
3981 (thumb_final_prescan_insn): Rename...
3982 (thumb1_final_prescan_insn): ... to this.
3983 (thumb_expand_epilogue): Rename...
3984 (thumb1_expand_epilogue): ... to this.
3985 (arm_unwind_emit_stm): Rename...
3986 (arm_unwind_emit_sequence): ... to this.
3987 (thumb2_legitimate_index_p, thumb2_legitimate_address_p,
3988 thumb1_compute_save_reg_mask, arm_dwarf_handle_frame_unspec,
3989 thumb2_index_mul_operand, output_move_vfp, arm_shift_nmem,
3990 arm_save_coproc_regs, thumb_set_frame_pointer, arm_print_condition,
3991 thumb2_final_prescan_insn, thumb2_asm_output_opcode, arm_output_shift,
3992 thumb2_output_casesi): New functions.
3993 (TARGET_DWARF_HANDLE_FRAME_UNSPEC): Define.
3994 (FL_THUMB2, FL_NOTM, FL_DIV, FL_FOR_ARCH6T2, FL_FOR_ARCH7,
3995 FL_FOR_ARCH7A, FL_FOR_ARCH7R, FL_FOR_ARCH7M, ARM_LSL_NAME,
3996 THUMB2_WORK_REGS): Define.
3997 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv, arm_condexec_count,
3998 arm_condexec_mask, arm_condexec_masklen)): New variables.
3999 (all_architectures): Add armv6t2, armv7, armv7a, armv7r and armv7m.
4000 (arm_override_options): Check new CPU capabilities.
4001 Set new architecture flag variables.
4002 (arm_isr_value): Handle v7m interrupt functions.
4003 (user_return_insn): Return 0 for v7m interrupt functions. Handle
4005 (const_ok_for_arm): Handle Thumb-2 constants.
4006 (arm_gen_constant): Ditto. Use movw when available.
4007 (arm_function_ok_for_sibcall): Return false for v7m interrupt
4009 (legitimize_pic_address, arm_call_tls_get_addr): Handle Thumb-2.
4010 (thumb_find_work_register, arm_load_pic_register,
4011 legitimize_tls_address, arm_address_cost, load_multiple_sequence,
4012 emit_ldm_seq, emit_stm_seq, arm_select_cc_mode, get_jump_table_size,
4013 print_multi_reg, output_mov_long_double_fpa_from_arm,
4014 output_mov_long_double_arm_from_fpa, output_mov_double_fpa_from_arm,
4015 output_mov_double_fpa_from_arm, output_move_double,
4016 arm_compute_save_reg_mask, arm_compute_save_reg0_reg12_mask,
4017 output_return_instruction, arm_output_function_prologue,
4018 arm_output_epilogue, arm_get_frame_offsets, arm_regno_class,
4019 arm_output_mi_thunk, thumb_set_return_address): Ditto.
4020 (arm_expand_prologue): Handle Thumb-2. Use arm_save_coproc_regs.
4021 (arm_coproc_mem_operand): Allow POST_INC/PRE_DEC.
4022 (arithmetic_instr, shift_op): Use arm_shift_nmem.
4023 (arm_print_operand): Use arm_print_condition. Handle '(', ')', '.',
4025 (arm_final_prescan_insn): Use extract_constrain_insn_cached.
4026 (thumb_expand_prologue): Use thumb_set_frame_pointer.
4027 (arm_file_start): Output directive for unified syntax.
4028 (arm_unwind_emit_set): Handle stack alignment instruction.
4029 * config/arm/lib1funcs.asm: Remove default for __ARM_ARCH__.
4030 Add v6t2, v7, v7a, v7r and v7m.
4031 (RETLDM): Add Thumb-2 code.
4032 (do_it, shift1, do_push, do_pop, COND, THUMB_SYNTAX): New macros.
4033 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Define __thumb2__.
4034 (TARGET_THUMB1, TARGET_32BIT, TARGET_THUMB2, TARGET_DSP_MULTIPLY,
4035 TARGET_INT_SIMD, TARGET_UNIFIED_ASM, ARM_FT_STACKALIGN, IS_STACKALIGN,
4036 THUMB2_TRAMPOLINE_TEMPLATE, TRAMPOLINE_ADJUST_ADDRESS,
4037 ASM_OUTPUT_OPCODE, THUMB2_GO_IF_LEGITIMATE_ADDRESS,
4038 THUMB2_LEGITIMIZE_ADDRESS, CASE_VECTOR_PC_RELATIVE,
4039 CASE_VECTOR_SHORTEN_MODE, ADDR_VEC_ALIGN, ASM_OUTPUT_CASE_END,
4040 ADJUST_INSN_LENGTH): Define.
4041 (TARGET_REALLY_IWMMXT, TARGET_IWMMXT_ABI, CONDITIONAL_REGISTER_USAGE,
4042 STATIC_CHAIN_REGNUM, HARD_REGNO_NREGS, INDEX_REG_CLASS,
4043 BASE_REG_CLASS, MODE_BASE_REG_CLASS, SMALL_REGISTER_CLASSES,
4044 PREFERRED_RELOAD_CLASS, SECONDARY_OUTPUT_RELOAD_CLASS,
4045 SECONDARY_INPUT_RELOAD_CLASS, LIBCALL_VALUE, FUNCTION_VALUE_REGNO_P,
4046 TRAMPOLINE_SIZE, INITIALIZE_TRAMPOLINE, HAVE_PRE_INCREMENT,
4047 HAVE_POST_DECREMENT, HAVE_PRE_DECREMENT, HAVE_PRE_MODIFY_DISP,
4048 HAVE_POST_MODIFY_DISP, HAVE_PRE_MODIFY_REG, HAVE_POST_MODIFY_REG,
4049 REGNO_MODE_OK_FOR_BASE_P, LEGITIMATE_CONSTANT_P,
4050 REG_MODE_OK_FOR_BASE_P, REG_OK_FOR_INDEX_P, GO_IF_LEGITIMATE_ADDRESS,
4051 LEGITIMIZE_ADDRESS, THUMB2_LEGITIMIZE_ADDRESS,
4052 GO_IF_MODE_DEPENDENT_ADDRESS, MEMORY_MOVE_COST, BRANCH_COST,
4053 ASM_APP_OFF, ASM_OUTPUT_CASE_LABEL, ARM_DECLARE_FUNCTION_NAME,
4054 FINAL_PRESCAN_INSN, PRINT_OPERAND_PUNCT_VALID_P,
4055 PRINT_OPERAND_ADDRESS): Adjust for Thumb-2.
4056 (arm_arch_notm, arm_arch_thumb2, arm_arch_hwdiv): New declarations.
4057 * config/arm/arm-cores.def: Add arm1156t2-s, cortex-a8, cortex-r4 and
4059 * config/arm/arm-tune.md: Regenerate.
4060 * config/arm/arm-protos.h: Update prototypes.
4061 * config/arm/vfp.md: Enable patterns for Thumb-2.
4062 (arm_movsi_vfp): Add movw alternative. Use output_move_vfp.
4063 (arm_movdi_vfp, movsf_vfp, movdf_vfp): Use output_move_vfp.
4064 (thumb2_movsi_vfp, thumb2_movdi_vfp, thumb2_movsf_vfp,
4065 thumb2_movdf_vfp, thumb2_movsfcc_vfp, thumb2_movdfcc_vfp): New.
4066 * config/arm/libunwind.S: Add Thumb-2 code.
4067 * config/arm/constraints.md: Update include Thumb-2.
4068 * config/arm/ieee754-sf.S: Add Thumb-2/Unified asm support.
4069 * config/arm/ieee754-df.S: Ditto.
4070 * config/arm/bpabi.S: Ditto.
4071 * config/arm/t-arm (MD_INCLUDES): Add thumb2.md.
4072 * config/arm/predicates.md (low_register_operand,
4073 low_reg_or_int_operand, thumb_16bit_operator): New.
4074 (thumb_cmp_operand, thumb_cmpneg_operand): Rename...
4075 (thumb1_cmp_operand, thumb1_cmpneg_operand): ... to this.
4076 * config/arm/t-arm-elf: Add armv7 multilib.
4077 * config/arm/arm.md: Update patterns for Thumb-2 and Unified asm.
4079 (UNSPEC_STACK_ALIGN, ce_count): New.
4080 (arm_incscc, arm_decscc, arm_umaxsi3, arm_uminsi3,
4081 arm_zero_extendsidi2, arm_zero_extendqidi2): New
4083 * config/arm/fpa.md: Update patterns for Thumb-2 and Unified asm.
4084 (thumb2_movsf_fpa, thumb2_movdf_fpa, thumb2_movxf_fpa,
4085 thumb2_movsfcc_fpa, thumb2_movdfcc_fpa): New insns.
4086 * config/arm/cirrus.md: Update patterns for Thumb-2 and Unified asm.
4087 (cirrus_thumb2_movdi, cirrus_thumb2_movsi_insn,
4088 thumb2_cirrus_movsf_hard_insn, thumb2_cirrus_movdf_hard_insn): New
4090 * doc/extend.texi: Document ARMv7-M interrupt functions.
4091 * doc/invoke.texi: Document Thumb-2 new cores+architectures.
4093 2007-01-03 Jakub Jelinek <jakub@redhat.com>
4095 * unwind-dw2.c (SIGNAL_FRAME_BIT, EXTENDED_CONTEXT_BIT): Define.
4096 (struct _Unwind_Context): Rename args_size to flags, remove
4097 signal_frame field, add a new args_size field and version field.
4098 (_Unwind_IsSignalFrame, _Unwind_SetSignalFrame,
4099 _Unwind_IsExtendedContext): New inline functions.
4100 (_Unwind_GetGR, _Unwind_SetGR, _Unwind_GetGRPtr, _Unwind_SetGRPtr):
4101 Assume by_value array is only present if _Unwind_IsExtendedContext.
4102 (_Unwind_GetIPInfo, execute_cfa_program, uw_frame_state_for): Use
4103 _Unwind_IsSignalFrame.
4104 (__frame_state_for): Initialize context.flags to EXTENDED_CONTEXT_BIT.
4105 (uw_update_context_1): Use _Unwind_SetSignalFrame.
4106 (uw_init_context_1): Initialize context->flags to
4107 EXTENDED_CONTEXT_BIT.
4108 * config/rs6000/linux-unwind.h (frob_update_context): Use
4109 _Unwind_SetSignalFrame.
4111 2007-01-03 Andrew Pinski <andrew_pinski@playstation.sony.com>
4114 * gimplify.c (gimplify_modify_expr_complex_part): Move below
4115 tree_to_gimple_tuple. Call tree_to_gimple_tuple when we need
4118 2007-01-03 Kazu Hirata <kazu@codesourcery.com>
4120 * config/i386/i386.h (NON_STACK_REG_P, REGNO_OK_FOR_SIREG_P,
4121 REGNO_OK_FOR_DIREG_P, REWRITE_ADDRESS, ASM_OPERAND_LETTER,
4122 RET, AT_SP): Remove.
4123 * config/i386/i386.md (*sse_prologue_save_insn): Use return
4126 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4127 * cfgbuild.c (find_basic_blocks): Likewise.
4128 * cfgrtl.c (rtl_create_basic_block): Likewise.
4129 * function.c (temp_slots_at_level): Likewise.
4130 * reg-stack.c (stack_regs_mentioned): Likewise.
4131 * regclass.c (allocate_reg_info): Likewise.
4132 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4133 set_bb_for_stmt, move_block_to_fn): Likewise.
4134 * tree-complex.c (tree_lower_complex): Likewise.
4135 * vec.h (VEC_safe_grow_cleared): New.
4137 * cgraphunit.c, tree-ssa-alias.c: Fix comment typos.
4139 2007-01-03 Zdenek Dvorak <dvorakz@suse.cz>
4141 * loop-unswitch.c (unswitch_loop): Pass probabilities to loopify.
4142 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Pass probabilities
4144 * cfgloopmanip.c (scale_loop_frequencies): Export.
4145 (loopify): Scale the frequencies by prescribed coefficients.
4146 (set_zero_probability): New function.
4147 (duplicate_loop_to_header_edge): Improve updating of frequencies.
4148 (lv_adjust_loop_entry_edge, loop_version): Set probabilities
4149 and frequencies according to arguments.
4150 * tree-ssa-loop-manip.c (tree_unroll_loop): Set probabilities
4152 * cfg.c (scale_bbs_frequencies_int): Allow scaling the frequencies up.
4153 * modulo-sched.c (sms_schedule): Set probabilities for entering
4154 versioned loop correctly.
4155 * tree-vect-transform.c (vect_transform_loop): Ditto.
4156 * cfgloop.h (loopify, loop_version): Declaration changed.
4157 (scale_loop_frequencies): Declared.
4159 2007-01-02 Jan Hubicka <jh@suse.cz>
4161 * cgraph.c: Include tree-flow.h
4162 (cgraph_add_new-function): Handle IPA_SSA mode; execute
4164 * cgraph.h (enum cgraph_state): Add CGRAPH_STATE_IPA_SSA.
4165 * tree-pass.h (pass_all_early_optimizations): Declare.
4166 * cgraphunit.c (cgraph_process_new_functions): Add IPA_SSA; execute
4168 (cgraph_analyze_function): Do early_local_passes.
4169 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
4170 Do not add referenced vars.
4171 * tree-optimize.c (gate_all_optimizations): Do not execute when not in
4173 (gate_all_early_local_passes): New gate.
4174 (pass_early_local_passes): Use new gate.
4175 (execute_early_local_optimizations): New functions.
4176 (gate_all_early_optimizations): New gate.
4177 (pass_all_early_optimizations): New pass.
4178 (execute_free_datastructures): Free SSA only when initialized.
4179 (gate_init_datastructures): Init only when optimizing.
4180 (tree_lowering_passes): Do early local passes when called late.
4181 * tree-profile.c (do_tree_profiling): Don't profile functions added
4183 (do_early_tree_profiling, pass_early_tree_profile): Kill.
4184 * tree-cfg.c (update_modified_stmts): Do not update when operands are
4186 * passes.c (init_optimizations_passes): Reorder so we go into SSA
4187 during early_local_passes.
4188 * Makefile.in (cgraph.o): Add dependency on tree-flow.h.
4191 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
4193 * Makefile.in: Update copyright year.
4195 2007-01-02 Carlos O'Donell <carlos@codesourcery.com>
4197 * Makefile.in: Export GCC_EXEC_PREFIX before calling $(RUNTEST)
4198 in $(lang_checks) and check-consistency targets.
4200 2007-01-02 Jan Hubicka <jh@suse.cz>
4202 * tree-mudflap.c (mf_decl_cache_locals, mf_build_check_statement_for):
4203 Do not add referenced vars.
4204 * tree-cfg.c (update_modified_stmts): Do not update when SSA operands
4206 * passes.c (init_optimization_passes): Put mudflap_2 after
4207 free_datastructures.
4209 2007-01-02 Jan Hubicka <jh@suse.cz>
4211 * tree-optimize (execute_fixup_cfg): Set after_inlining flag.
4212 Set NOTHROW flag on call statements proved to be nothrow.
4213 Update statement of local calls so new pure/const functions are
4214 updated. Update_ssa when in ssa form. Mark PHI nodes of nonlocal
4216 (tree_rest_of_compilation): Register hooks and initialize bitmap
4217 early. Do not set after_inlining flag.
4219 2007-01-02 Steve Ellcey <sje@cup.hp.com>
4221 * sbitmap.c (HOST_BITS_PER_LONG_LONG): Change to
4222 HOST_BITS_PER_LONGLONG
4224 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4227 * c-typeck.c (store_init_value): Don't emit pedantic overflow
4228 warning for non-static initializers.
4230 2007-01-02 Steven Bosscher <steven@gcc.gnu.org>
4232 * config/alpha/alpha.md, arm/arm.c, darwin.c, frv/frv.md,
4233 m32r/m32r.c, m32r/m32r.c, mn10300/mn10300.md, pa/pa.c,
4234 rs6000/rs6000.c, s390/s390.md, sh/sh.md, sparc/sparc.c:
4235 Always use set_unique_reg_note to add REG_EQUAL notes.
4237 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4240 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4242 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4243 * cfgbuild.c (find_basic_blocks): Likewise.
4244 * cfgrtl.c (rtl_create_basic_block): Likewise.
4245 * function.c (temp_slots_at_level): Likewise.
4246 * reg-stack.c (stack_regs_mentioned): Likewise.
4247 * regclass.c (allocate_reg_info): Likewise.
4248 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4249 set_bb_for_stmt, move_block_to_fn): Likewise.
4250 * tree-complex.c (tree_lower_complex): Likewise.
4251 * vec.h (VEC_safe_grow_cleared): New.
4253 2007-01-02 Ian Lance Taylor <iant@google.com>
4255 * c-common.c (c_common_truthvalue_conversion): When warning about
4256 using an assignment as a truth value, set TREE_NO_WARNING.
4258 2007-01-02 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4261 * c.opt (Wold-style-declaration): New.
4262 * doc/invoke.texi (C-only Warning Options): New.
4263 (Wold-style-declaration): Document it.
4264 (Wextra): Enabled by -Wextra.
4265 * c-opts.c (c_common_post_options): Enabled by -Wextra.
4266 * c-decl.c (declspecs_add_scspec): Replace -Wextra with
4267 -Wold-style-declaration.
4269 2007-01-02 Kazu Hirata <kazu@codesourcery.com>
4271 * alias.c (init_alias_analysis): Use VEC_safe_grow_cleared.
4272 * cfgbuild.c (find_basic_blocks): Likewise.
4273 * cfgrtl.c (rtl_create_basic_block): Likewise.
4274 * function.c (temp_slots_at_level): Likewise.
4275 * reg-stack.c (stack_regs_mentioned): Likewise.
4276 * regclass.c (allocate_reg_info): Likewise.
4277 * tree-cfg.c (init_empty_tree_cfg, build_tree_cfg, create_bb,
4278 set_bb_for_stmt, move_block_to_fn): Likewise.
4279 * tree-complex.c (tree_lower_complex): Likewise.
4280 * vec.h (VEC_safe_grow_cleared): New.
4282 2007-01-02 Douglas Gregor <doug.gregor@gmail.com>
4284 * c-common.c (c_common_nodes_and_builtins): Since variants of
4285 void_type_node get built before it is given a name, we need to
4286 give those variants the name, too.
4287 (complete_array_type): We need to work with the canonical main
4288 type of the array, from which we will build the qualified version.
4289 * params.def (PARAM_VERIFY_CANONICAL_TYPES): New.
4290 * print-tree.c (print_node): Display canonical type information
4292 * stor-layout.c (layout_type): When we don't know the
4293 alignment of a type for which we're building an array, we end up
4294 guessing wrong, so make the type require structural equality.
4295 * tree.c (make_node_stat): When we build a new type, it is its
4297 (build_type_attribute_qual_variant): When building an attribute
4298 variant, its canonical type is the non-attribute variant. However,
4299 if the attributes are target-dependent and they differ, we need to
4300 use structural equality checks for this type.
4301 (build_qualified_type): A qualified type is not equivalent to its
4302 unqualified variant; set the canonical type appropriately.
4303 (build_distinct_type_copy): When building a distinct type from
4304 another type, the new type is its own canonical type.
4305 (build_variant_type_copy): When building a new type variant, we
4306 assume that it is equivalent to the original type.
4307 (build_pointer_type_for_mode): When building a pointer type, also
4308 build a canonical type pointer.
4309 (build_reference_type_for_mode): When building a reference type,
4310 also build a canonical type reference.
4311 (build_index_type): When we can't hash an index type (e.g.,
4312 because its maximum value is negative), the index type requires
4313 structural equality tests.
4314 (build_array_type): Build the canonical form of an array type.
4315 (build_function_type): Function types require structural equality,
4316 because they contain default arguments, attributes, etc.
4317 (build_method_type_directly): Ditto for method types.
4318 (build_offset_type): Build the canonical offset type.
4319 (build_complex_type): Build the canonical vector type.
4320 (make_vector_type): Build the canonical vector type.
4321 * tree.h (TYPE_CANONICAL): New.
4322 (TYPE_STRUCTURAL_EQUALITY_P): New.
4323 (SET_TYPE_STRUCTURAL_EQUALITY): New.
4324 (struct tree_type): Added "canonical" field.
4325 * params.h (VERIFY_CANONICAL_TYPES): New.
4326 * doc/c-tree.texi (TYPE_CANONICAL): Document.
4327 (TYPE_STRUCTURAL_EQUALITY_P): Document.
4328 (SET_TYPE_STRUCTURAL_EQUALITY): Document.
4329 * doc/invoke.texi (verify-canonical-types): Document --param
4330 parameter for verifying canonical types.
4332 2007-01-02 Joseph Myers <joseph@codesourcery.com>
4334 * config.gcc (powerpc-*-eabispe*, powerpc-*-eabisimaltivec*,
4335 powerpc-*-eabisim*, powerpc-*-eabialtivec*, powerpc-*-eabi*,
4336 powerpc-*-rtems*, powerpc-wrs-vxworks, powerpc-wrs-vxworksae,
4337 powerpcle-*-eabisim*, powerpcle-*-eabi*): Add rs6000/e500.h to
4339 * config/rs6000/e500.h: New.
4340 * config/rs6000/eabi.h (TARGET_SPE_ABI, TARGET_SPE, TARGET_E500,
4341 TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE, TARGET_E500_DOUBLE):
4343 * config/rs6000/linuxspe.h (TARGET_SPE_ABI, TARGET_SPE,
4344 TARGET_E500, TARGET_ISEL, TARGET_FPRS, TARGET_E500_SINGLE,
4345 TARGET_E500_DOUBLE): Remove.
4346 * config/rs6000/vxworks.h (TARGET_SPE_ABI, TARGET_SPE,
4347 TARGET_E500, TARGET_ISEL, TARGET_FPRS): Remove.
4348 * config/rs6000/rs6000.h (CHECK_E500_OPTIONS): Define.
4349 * config/rs6000/rs6000.c (rs6000_override_options): Use
4352 2007-01-02 Joseph Myers <joseph@codesourcery.com>
4354 * config/rs6000/rs6000.c (print_operand): Check (TARGET_SPE ||
4355 TARGET_E500_DOUBLE), not TARGET_E500, for %y.
4356 (rs6000_generate_compare, rs6000_emit_sCOND, output_cbranch,
4357 rs6000_emit_cmove): Don't check TARGET_E500.
4358 * config/rs6000/rs6000.md (bunordered, bordered, sunordered,
4359 sordered): Don't check TARGET_E500.
4361 2007-01-01 Eric Christopher <echristo@apple.com>
4363 * config/mips/mips.c (mips_regno_mode_ok_for_base_p): Use
4364 HARD_REGISTER_NUM_P.
4366 2007-01-01 Roger Sayle <roger@eyesopen.com>
4368 * fold-const.c (fold_binary) <EQ_EXPR>: Fold "(X^C1) eq/ne C2" into
4369 "X eq/ne (C1^C2)". Fold "(X^Z) eq/ne (Y^Z)" as "X eq/ne Y" when Z
4370 has no side-effects. Fold "(X^C1) eq/ne (Y^C2)" as "(X^(C1^C2))
4373 2007-01-01 Mike Stump <mrs@apple.com>
4375 * configure.ac: Remove support for building with Apple's gcc-3.1.
4377 2007-01-02 Joseph Myers <joseph@codesourcery.com>
4380 * caller-save.c (add_stored_regs): Only handle SUBREGs if inner
4381 REG is a hard register. Do not modify REG before calling
4383 * rtlanal.c (subreg_get_info): Don't assert size of XMODE is a
4384 multiple of the size of YMODE for certain lowpart cases.
4386 2007-01-01 Andrew Pinski <pinskia@gmail.com>
4389 * gimplify (voidify_wrapper_expr): Update for
4392 2007-01-01 Andreas Schwab <schwab@suse.de>
4395 * config/ia64/ia64.c (ia64_compute_frame_size): Account space for
4396 save of BR0 in extra_spill_size instead of spill_size.
4397 (ia64_expand_prologue): Save BR0 outside of the gr/br/fr spill
4399 (ia64_expand_epilogue): Restore BR0 from its new location.
4401 2007-01-01 Andrew Pinski <pinskia@gmail.com>
4403 * gimplify.c (gimplify_init_constructor <case VECTOR_TYPE>):
4404 Use a temporary variable if the left hand side is not a gimple
4407 2007-01-01 Andrew Pinski <pinskia@gmail.com>
4409 * gimplify.c (gimplify_return_expr): Make the temporary variable
4410 for the return expression, a gimple register variable.
4412 2007-01-01 Jan Hubicka <jh@suse.cz>
4414 * emit-rtl.c (emit_copy_of_insn_after): Do not call copy_insn_1 for
4417 2007-01-01 Mike Stump <mrs@apple.com>
4419 * configure.ac (HAVE_GAS_LITERAL16): Add autoconf check for
4421 * config/darwin.c (machopic_select_rtx_section): Use
4423 (darwin_mergeable_constant_section): Likewise.
4424 * configure: Regenerate.
4425 * config.in: Regenerate.
4427 2007-01-01 Jan Hubicka <jh@suse.cz>
4428 Andrew Pinski <pinskia@gmail.com>
4430 * cgraphunit.c (cgraph_optimize): Call cgraph_add_new_functions
4431 before starting IPA passes.