1 2005-01-25 Aldy Hernandez <aldyh@redhat.com>
3 * config/frv/frv.c (frv_legitimize_tls_address): New.
4 (TARGET_HAVE_TLS): Define.
5 (FRV_SYMBOL_REF_TLS_P): Define.
6 (frv_override_options): Handle new register classes.
7 (frv_legitimate_address_p): Reject tls addresses.
8 (frv_legitimize_address): Handle TLS addresses.
9 (gen_inlined_tls_plt): New.
11 (frv_legitimize_tls_address): New.
12 (unspec_got_name): Add TLS entries.
13 (got12_operand): Add R_FRV_TLSMOFF12 case.
14 (frv_emit_move): Fixup TLS addresses.
15 (frv_emit_movsi): Legitimize TLS addresses.
17 * config/frv/frv.h (MASK_BIG_TLS): New.
18 (TARGET_BIG_TLS): New.
19 (HAVE_AS_TLS): Define.
20 (TARGET_SWITCHES): Add -mTLS and -mtls options.
21 (enum reg_class): Add GR8_REGS, GR9_REGS, GR89_REGS.
22 (REG_CLASS_NAMES): Same.
23 (REG_CLASS_CONTENTS): Same.
24 (CONSTRAINT_LEN): New.
25 (REG_CLASS_FROM_CONSTRAINT): New.
26 (PREDICATE_CODES): Add symbolic_operand.
28 * config/frv/frv.md (define_constants): Add UNSPEC_GETTLSOFF,
29 UNSPEC_TLS_LOAD_GOTTLSOFF12, UNSPEC_TLS_INDIRECT_CALL,
30 UNSPEC_TLS_TLSDESC_LDD, UNSPEC_TLS_TLSDESC_LDD_AUX,
31 UNSPEC_TLS_TLSOFF_LD, UNSPEC_TLS_LDDI, UNSPEC_TLSOFF_HILO,
32 R_FRV_GOTTLSOFF_HI, R_FRV_GOTTLSOFF_LO, R_FRV_TLSMOFFHI,
33 R_FRV_TLSMOFFLO, R_FRV_TLSMOFF12, R_FRV_TLSDESCHI,
34 R_FRV_TLSDESCLO, R_FRV_GOTTLSDESCHI, R_FRV_GOTTLSDESCLO, GR8_REG,
35 GR9_REG, GR14_REG, LRREG.
36 (type): Add load_or_call attribute.
37 ("load_or_call"): New reservation.
38 ("call_gettlsoff"): New.
39 ("tls_indirect_call"): New.
40 ("tls_load_gottlsoff12"): New.
42 ("tls_tlsdesc_ldd"): New.
43 ("tls_tlsoff_ld"): New.
46 * config/frv/frv-protos.h (symbolic_operand): Protoize.
48 2005-01-25 Alexandre Oliva <aoliva@redhat.com>
50 * config/frv/frv.h (ASM_OUTPUT_DWARF_DTPREL): Define.
51 * config/frv/frv-protos.h (frv_output_dwarf_dtprel): Declare.
52 * config/frv/frv.c (TLS_BIAS): Define.
53 (frv_output_dwarf_dtprel): New.
55 2005-01-26 Jakub Jelinek <jakub@redhat.com>
57 PR rtl-optimization/19579
58 * ifcvt.c (noce_try_cmove_arith): If emitting instructions to set up
59 both A and B, see if they don't clobber registers the other expr uses.
61 2005-01-25 J"orn Rennecke <joern.rennecke@st.com>
63 * real.c (do_add): Initialize signalling and canonical members.
65 * real.c (real_from_integer): Zero out destination.
67 2005-01-18 Jan Hubicka <jh@suse.cz>
69 PR tree-optimize/19337
70 * tree-inline.c (initialize_inlined_parameters): Copy
71 saved_static_chain_decl only.
73 2005-01-25 Richard Henderson <rth@redhat.com>
77 * config/i386/i386.c (x86_inter_unit_moves): Zero.
78 (ix86_preferred_reload_class): Rewrite fp-constant section, with
79 80387 enabled, to return a proper subclass. Return the subset that
80 overlaps with GENERAL_REGS for PLUS.
81 * config/i386/i386.md (movsi_1, movdi_2): Set type to "mmx" for pxor.
83 2005-01-25 Richard Henderson <rth@redhat.com>
86 * tree-complex.c (expand_complex_div_wide): Use the correct formulae.
88 2005-01-25 Roger Sayle <roger@eyesopen.com>
90 * expmed.c (choose_multiplier): Change interface to return the
91 multiplier lowpart as an rtx instead of an unsigned HOST_WIDE_INT.
92 (expand_divmod): Update calls to choose_multiplier.
94 2005-01-24 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
97 * expmed.c (extract_bit_field): Use adjust_address instead of
98 gen_lowpart when op0 is a MEM.
100 2005-01-24 Steven Bosscher <stevenb@suse.de>
102 * cfgexpand.c (expand_gimple_tailcall): Fix typo.
104 2005-01-24 Richard Sandiford <rsandifo@redhat.com>
106 * config/mips/irix-csr.c: New file.
107 * config/mips/t-iris6 (irix-csr.o): New rule to build it.
108 (EXTRA_MULTILIB_PARTS): Add irix-csr.o.
109 * config/mips/iris6.h (ENDFILE_SPEC): Include it in n32 and n64
112 2005-01-24 Eric Botcazou <ebotcazou@libertysurf.fr>
115 * config.gcc (sparc-*-elf*): Remove sol2.h, sparc/sol2.h and
116 sparc/elf.h, add sparc/sp-elf.h.
117 (sparc-*-rtems*): Likewise.
118 (sparclite-*-elf*): Remove sol2.h, sparc/sol2.h, sparc/elf.h and
119 tm-dwarf2.h, add sparc/sp-elf.h.
120 (sparc86x-*-elf): Likewise.
121 (sparc64-*-elf*): Remove sol2.h, sparc/sol2.h and tm-dwarf2.h.
122 * config/sparc/liteelf.h (TARGET_SUB_OS_CPP_BUILTINS): Rename into
123 TARGET_OS_CPP_BUILTINS.
124 * config/sparc/sp86x-elf (TARGET_SUB_OS_CPP_BUILTINS): Likewise.
125 * config/sparc/rtemself.h (TARGET_SUB_OS_CPP_BUILTINS): Likewise.
127 * config/sparc/openbsd64.h (NO_IMPLICIT_EXTERN_C): Undefine.
128 * config/sparc/sp64-elf.h (NO_IMPLICIT_EXTERN_C): New macro.
129 (SWITCH_TAKES_ARG): Likewise.
130 (LOCAL_LABEL_PREFIX): Likewise.
131 (ASM_GENERATE_INTERNAL_LABEL): Likewise.
132 (TARGET_N_FORMAT_TYPES): Delete.
133 (TARGET_FORMAT_TYPES): Likewise.
134 (ASM_DECLARE_FUNCTION_SIZE): Likewise.
135 * config/sparc/elf.h: Delete.
136 * config/sparc/sp-elf.h: New file.
138 2005-01-24 Kazu Hirata <kazu@cs.umass.edu>
140 * tree-flow-inline.h (phi_arg_from_edge): Remove.
141 * tree-flow.h: Remove the corresponding prototype.
142 * tree-ssa-operands.h, tree-ssa-threadupdate.c, tree-ssa.c,
143 tree-vectorizer.c: Use dest_idx instead of phi_arg_from_edge.
145 2005-01-24 Jakub Jelinek <jakub@redhat.com>
148 * config/rs6000/rs6000.c (rs6000_file_start): Create toc section
149 for AIX ABI or ELF -fPIC.
150 (rs6000_emit_load_toc_table): Don't create toc_section here.
151 (rs6000_xcoff_file_start): Nor here.
153 2005-01-24 J"orn Rennecke <joern.rennecke@st.com>
155 * sh.c (ra.h): Don't #include.
156 (hard_regs_intersect_p): New function, resurrected from ra.c.
158 * sh.c: Fix 1996 Copyright.
160 2005-01-24 Richard Henderson <rth@redhat.com>
161 Aldy Hernandez <aldyh@redhat.com>
163 * regrename.c (note_sets): Handle subregs.
165 2005-01-24 Joseph S. Myers <joseph@codesourcery.com>
168 * genconditions.c (write_header, write_conditions): Elide file if
171 2005-01-24 Kazu Hirata <kazu@cs.umass.edu>
173 * c-opts.c, expmed.c, gengtype.c, tree-complex.c,
174 config/arm/iwmmxt.md, config/avr/avr.c, config/s390/s390.h:
177 2005-01-24 Jakub Jelinek <jakub@redhat.com>
179 * builtins.c (get_pointer_alignment, gimplify_va_arg_expr,
180 expand_builtin_printf, expand_builtin_fprintf,
181 expand_builtin_sprintf): Use POINTER_TYPE_P instead TREE_CODE
182 checking against POINTER_TYPE.
183 (validate_arglist): Handle POINTER_TYPE code by checking
186 2005-01-24 Paolo Bonzini <bonzini@gnu.org>
188 * rtlanal.c (non_rtx_starting_operands, for_each_rtx_1,
190 (for_each_rtx): Call for_each_rtx_1.
191 * rtl.h (init_rtlanal): Declare.
192 * toplev.c (backend_init): Call init_rtlanal.
194 2005-01-24 Jakub Jelinek <jakub@redhat.com>
196 * flow.c (propagate_one_insn): Formatting.
199 * flow.c (libcall_dead_p): Be more conservative if unsure.
200 If there are any instructions between insn and call, see if they are
201 all dead before saying the libcall is dead.
203 2005-01-24 Uros Bizjak <uros@kss-loka.si>
205 * config/i386/i386.md (*extendsfdf2_sse): Use "nonimmediate_operand"
206 operand constraint for operand 0.
208 2005-01-23 Richard Henderson <rth@redhat.com>
211 * tree-complex.c (expand_complex_div_wide): Set EDGE_FALLTHRU on
212 edges leading to join block.
214 2005-01-23 Roger Sayle <roger@eyesopen.com>
216 * expmed.c (expand_mult_highpart): Make static. Change type of
217 constant multiplier argument from unsigned HOST_WIDE_INT to rtx.
218 (expand_divmod): Updates calls to expand_mult_highpart by using
219 gen_int_mode to pass a CONST_INT rtx.
220 * rtl.h (expand_mult_highpart): Remove prototype.
222 2005-01-23 Kazu Hirata <kazu@cs.umass.edu>
224 * flow.c (regset_bytes, regset_size): Remove.
226 * cfganal.c: Fix a reference to Harvey's paper.
228 2005-01-23 Steven Bosscher <stevenb@suse.de>
230 PR rtl-optimization/19464
231 * tree-optimize.c (init_tree_optimization_passes): Add one more
232 copyrename pass just before out-of-ssa.
234 2005-01-23 Joseph S. Myers <joseph@codesourcery.com>
237 * recog.c (recog_memoized): Don't define if GENERATOR_FILE.
238 * ggc-none.c (ggc_free): Define.
240 2005-01-23 Kazu Hirata <kazu@cs.umass.edu>
242 * cse.c (max_reg, max_insn_uid): Remove.
243 (cse_main): Don't access max_reg or max_insn_uid.
245 * c-common.c (explicit_flag_signed_bitfields,
246 lang_statement_code_p, lang_expand_function_end): Remove.
247 * c-common.h: Remove the corresponding declarations.
248 * c-opts.c (c_common_handle_option): Remove all write access
249 to explicit_flag_signed_bitfields.
251 * cgraph.c (cgraph_varpool_n_nodes): Remove.
252 (cgraph_varpool_node): Don't access cgraph_varpool_n_nodes.
253 * cgraph.h: Remove the corresponding declaration.
255 * gcse.c (null_pointer_info): Remove.
257 2005-01-23 Roger Sayle <roger@eyesopen.com>
258 Eric Botcazou <ebotcazou@libertysurf.fr>
260 * combine.c (simplify_logical): Only simplify logical expressions
261 of the form ior(and(x,y),z) by the inverse distributive law if the
262 result is cheaper than the original.
264 2005-01-23 Kazu Hirata <kazu@cs.umass.edu>
266 * alias.c, c-common.h, c-incpath.c, c-incpath.h, expr.c,
267 fold-const.c, gimplify.c, params.h, tree-data-ref.c,
268 tree-if-conv.c, tree-nested.c, tree-outof-ssa.c,
269 tree-ssa-dom.c, tree-vectorizer.c, tree.def, config/darwin.c,
270 config/freebsd-spec.h, config/arm/arm.h,
271 config/h8300/h8300.md, config/i386/i386.md,
272 config/i386/predicates.md, config/i386/sse.md,
273 config/ia64/ia64.c, config/ip2k/ip2k.c, config/s390/s390.c,
274 config/vax/vax.md: Fix comment typos. Follow spelling
277 2005-01-23 Steven Bosscher <stevenb@suse.de>
279 * gengtype.c (walk_type): Produce `not equal to' compares for loop
280 tests, instead of `less than'.
282 2005-01-23 Richard Sandiford <rsandifo@redhat.com>
284 * function.c (assign_parm_setup_block): When creating a new stack slot
285 for a parameter, get its alignment from the parameter's DECL_ALIGN
286 rather than the type's TYPE_ALIGN. Make sure that the parameter
287 is at least word aligned.
289 2005-01-22 Roger Sayle <roger@eyesopen.com>
292 * config/avr/avr.c (avr_hard_regno_mode_ok): Rewrite.
294 2005-01-22 Richard Henderson <rth@redhat.com>
297 * config/i386/i386.md (movsfcc_1_sse_max): Use nonimmediate_operand
298 in both compare operands.
299 (movdfcc_1_sse_max): Likewise.
300 (movsfcc_1_sse): Likewise. Add earlyclobber for scratch.
301 (movdfcc_1_sse): Likewise.
302 * config/i386/i386.c (ix86_split_sse_movcc): Emit copies into the
303 scratch register as needed.
305 2005-01-22 Richard Henderson <rth@redhat.com>
307 * config/i386/i386.md (smaxsf3): Fix mnemonic typo.
309 * config/i386/i386.c (ix86_prepare_fp_compare_args): Fix is_sse test.
311 2005-01-22 Richard Henderson <rth@redhat.com>
313 * genopinit.c (optabs): Use smin/smax for floating point too.
314 * doc/md.texi: Update to match. Clarify that floating point
315 results are undefined for +0/-0 and NaN.
316 * doc/rtl.texi: Likewise.
317 * rtl.def (SMIN, SMAX): Likewise
318 * tree.def (MIN_EXPR, MAX_EXPR): Likewise.
320 * config/alpha/alpha.md (smaxdf3, smindf3, smaxsf3, sminsf3): Add
321 leading 's' to the name.
322 * config/ia64/ia64.md (smaxsf3, sminsf3, smaxdf3, smindf3,
323 smaxxf3, sminxf3): Likewise.
324 * config/rs6000/rs6000.md (smaxdf3, smindf3, smaxsf3, sminsf3):
327 2005-01-22 Paul Brook <paul@codesourcery.com>
329 * config/arm/arm.md: Use "Uy" constraint for wcgr load/stores.
331 2005-01-21 Mark Dettinger <dettinge@de.ibm.com>
333 * config/s390/s390.c (struct processor_costs): New fields
334 dlgr, dlr, dr, dsgfr, dsgr.
335 (z900_cost, z990_cost): Values for new fields.
336 (s390_rtx_costs): New cases MEM und COMPARE in switch
337 statement. Modified handling of SIGN_EXTEND, ZERO_EXTEND,
338 DIV, MOD, UDIV, UMOD.
340 2005-01-21 Ulrich Weigand <uweigand@de.ibm.com>
342 * config/s390/s390.md ("doloop_si64"): Reload input value directly
343 into the register being decremented.
344 ("doloop_si31", "doloop_di"): Likewise.
345 ("*doloop_si_long"): Adapt pattern.
347 2005-01-21 Ulrich Weigand <uweigand@de.ibm.com>
349 * config/s390/s390.h (HARD_REGNO_NREGS): Fix computation for
351 (CLASS_MAX_NREGS): Likewise.
353 2005-01-21 Daniel Berlin <dberlin@dberlin.org>
355 * doc/tree-ssa.texi (Statement Operands): Add example for new
356 must-def macro. Note deprecation of old operands interface.
358 2005-01-22 Richard Sandiford <rsandifo@redhat.com>
360 PR tree-optimization/19484
361 * tree-cfg.c (remove_fallthru_edge): New function.
362 (cleanup_control_flow): Remove fallthru edges from calls that are
363 now known not to return.
365 2005-01-20 Daniel Berlin <dberlin@dberlin.org>
367 Fix PR tree-optimization/19038
368 * tree-ssa-dom.c (cprop_operand): Don't replace loop invaeriant
369 copies with loop variant ones.
371 2005-01-22 Kazu Hirata <kazu@cs.umass.edu>
373 * cfganal.c, real.h, reorg.c, timevar.def, tree-ssa-ccp.c,
374 config/alpha/alpha-protos.h, config/alpha/alpha.h,
375 config/alpha/alpha.md, config/alpha/predicates.md,
376 config/sparc/freebsd.h, config/sparc/netbsd-elf.h,
377 config/sparc/sol2.h: Update copyright.
379 * tree-cfg.c (remove_forwarder_block_with_phi): Look at the
380 first label to see if it is a nonlocal label.
382 2005-01-22 David Edelsohn <edelsohn@gnu.org>
383 Andrew Pinski <pinskia@physics.uc.edu>
386 * config/rs6000/rs6000.c (rs6000_va_start): Saturate n_gpr at
387 maximum number of GPRs. Saturate n_fpr at maximum number of FPRs.
389 2005-01-22 Ralf Corsepius <ralf.corsepius@rtems.org>
391 * config/rs6000/rtems.h: Update copyright.
393 2005-01-22 Ralf Corsepius <ralf.corsepius@rtems.org>
396 * config/rs6000/rtems.h: Resurrect cpp_os_rtems_spec from gcc < 3.4.
397 (CPP_OS_RTEMS_SPEC): New (From gcc-3.3's config/rs6000/sys4.h).
398 (SUBSUBTARGET_EXTRA_SPECS): Use CPP_OS_RTEMS_SPEC.
400 2005-01-22 Volker Reichelt <reichelt@igpm.rwth-aachen.de>
403 * c-typeck.c (convert_arguments): Check for error_mark_node.
405 2005-01-21 Ian Lance Taylor <ian@airs.com>
407 PR tree-optimization/13000
408 * tree-inline.c: Include "tree-flow.h".
409 (expand_call_inline): If warn_return_type, warn if non-void inline
410 function falls through.
411 * tree-cfg.c (execute_warn_function_return): Don't warn about
412 control reaching end if TREE_NO_WARNING is set. Set
414 * gimple-low.c (block_may_fallthru): Don't assume that SWITCH_EXPR
416 * gimplify.c (shortcut_cond_expr): Don't emit a jump over the else
417 branch if we don't need one.
418 * c-typeck.c: Include "tree-flow.h"
419 (c_finish_bc_stmt): Don't add a goto if the current statement
420 list doesn't fall through to the current point.
422 2005-01-21 Roger Sayle <roger@eyesopen.com>
424 PR rtl-optimization/576
425 * real.c (real_arithmetic): Change return type from void to bool
426 to return an indication that the result may be inexact.
427 * real.h (real_arithmeric): Update prototype.
428 * fold-const.c (const_binop): Don't constant fold floating
429 point expressions when the user specifies -frounding-math and
430 the result may depend upon the run-time rounding mode.
431 (fold_convert_const_real_from_real): Clean-up.
432 (fold_initializer): Ignore flag_rounding_math for initializers.
433 * simplify-rtx.c (simplify_binary_operation): Likewise, don't
434 constant fold FP operations with flag_rounding_math if the
435 result may depend upon the run-time rounding mode.
437 2005-01-21 Tom Tromey <tromey@redhat.com>
439 * c-cppbuiltin.c (define__GNUC__): Correct assertion.
441 2005-01-21 Kazu Hirata <kazu@cs.umass.edu>
443 * tree-cfg.c: Fix comment typos.
445 2005-01-21 J"orn Rennecke <joern.rennecke@st.com>
447 * passes.c (rest_of_handle_flow2): Close / open the flow2 dump file
448 around the call to rest_of_handle_branch_target_load_optimize.
450 2005-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
452 * config.gcc (sparc-*-netbsdelf*): Include dbxelf.h
453 (sparc64-*-openbsd*): Likewise.
454 (sparclite-*-elf*): Include tm-dwarf2.h.
455 (sparc86x-*-elf*): Likewise.
456 (sparc64-*-elf*): Likewise.
457 * config/sparc/linux64.h (DWARF2_DEBUGGING_INFO): Delete.
458 (DBX_DEBUGGING_INFO): Likewise.
459 * config/sparc/liteelf.h (DWARF2_DEBUGGING_INFO): Likewise.
460 (PREFERRED_DEBUGGING_TYPE): Likewise.
461 * config/sparc/netbsd-elf.h (PREFERRED_DEBUGGING_TYPE): Likewise.
462 * config/sparc/sp64-elf.h (DBX_DEBUGGING_INFO): Likewise.
463 (PREFERRED_DEBUGGING_TYPE): Likewise.
464 * config/sparc/sp86x-elf.h (DWARF2_DEBUGGING_INFO): Likewise.
465 (PREFERRED_DEBUGGING_TYPE): Likewise.
466 * config/sparc/sparc.h (DBX_DEBUGGING_INFO): Likewise.
468 2005-01-21 Hans-Peter Nilsson <hp@bitrange.com>
471 * combine.c (combine_simplify_rtx): Revert change of 2004-12-31.
473 2005-01-20 Paul Brook <paul@codesourcery.com>
475 * doc/rtl.texi: Document value extension requirements for CONST_INT.
477 2005-01-20 Kazu Hirata <kazu@cs.umass.edu>
479 * tree-cfg.c (tree_verify_flow_info): Check that a nonlocal
480 label is first in a sequence of labels.
482 * tree-cfg.c (tree_verify_flow_info): Fix a typo.
484 2005-01-20 Janis Johnson <janis187@us.ibm.com>
485 Giovanni Bajo <giovannibajo@gcc.gnu.org>
487 * doc/sourcebuild.texi (Test Directives): New.
489 2005-01-20 Roger Sayle <roger@eyesopen.com>
491 * real.c (real_floor): Don't leave the result uninitialized when
492 mode is VOIDMode, but return the unrounded intermediate value.
493 (real_ceil): Likewise.
495 2005-01-20 Paul Brook <paul@codesourcery.com>
497 * config/arm/arm.md (insv): Use gen_int_mode.
499 2005-01-20 Eric Botcazou <ebotcazou@libertysurf.fr>
501 * config/sparc/freebsd.h (ASM_OUTPUT_INTERNAL_LABELREF): Delete.
502 * config/sparc/linux.h (ASM_OUTPUT_INTERNAL_LABELREF): Likewise.
503 * config/sparc/linux64.h (ASM_OUTPUT_INTERNAL_LABELREF): Likewise.
504 * config/sparc/netbsd-elf.h (ASM_OUTPUT_INTERNAL_LABELREF): Likewise.
505 * config/sparc/sol2.h (ASM_OUTPUT_INTERNAL_LABELREF): Likewise.
507 2005-01-20 Daniel Berlin <dberlin@dberlin.org>
510 * dwarf2out.c (concat_loc_descriptor): We don't know we can use
511 fbreg, so we have to assume we can't.
512 (loc_descriptor_from_tree_1): Ditto.
513 (containing_function_has_frame_base): New function.
514 (add_location_or_const_value_attribute): Use it.
515 Also try to generate a frame_base from a single element location
518 2005-01-20 Kazu Hirata <kazu@cs.umass.edu>
520 PR tree-optimization/15349
521 * timevar.def (TV_TREE_MERGE_PHI): New.
522 * tree-cfg.c (tree_forwarder_block_p): Add a new argument
524 (remove_forwarder_block, cleanup_forwarder_blocks): Adjust the
525 calls to tree_forwarder_block_p.
526 (remove_forwarder_block_with_phi, merge_phi_nodes,
527 gate_merge_phi, pass_merge_phi): New.
528 * tree-optimize.c (init_tree_optimization_passes): Add
530 * tree-pass.h: Add an extern for pass_merge_phi;
532 2005-01-20 Richard Henderson <rth@redhat.com>
535 * config/i386/emmintrin.h (_mm_castpd_ps, _mm_castpd_si128): New.
536 (_mm_castps_pd, _mm_castps_si128): New.
537 (_mm_castsi128_ps, _mm_castsi128_pd): New.
539 2005-01-20 Richard Henderson <rth@redhat.com>
542 * config/i386/mmintrin.h (_mm_cvtsi32_si64): Use
543 __builtin_ia32_vec_init_v2si.
544 (_mm_cvtsi64_si32): Use __builtin_ia32_vec_ext_v2si.
545 * config/i386/i386.c (IX86_BUILTIN_VEC_EXT_V2SI): New.
546 (ix86_init_mmx_sse_builtins): Create it.
547 (ix86_expand_builtin): Expand it.
548 (ix86_expand_vector_set): Handle V2SFmode and V2SImode.
549 * config/i386/mmx.md (vec_extractv2sf_0, vec_extractv2sf_1): New.
550 (vec_extractv2si_0, vec_extractv2si_1): New.
552 2005-01-20 Richard Henderson <rth@redhat.com>
555 * config/i386/i386.c (ix86_expand_push): New.
556 * config/i386/mmx.md (push<MMXMODE>1): New.
557 * config/i386/sse.md (push<SSEMODE>1): New.
558 * config/i386/i386-protos.h: Update.
560 2005-01-20 Kazu Hirata <kazu@cs.umass.edu>
562 * params.def, params.h, predict.c, tree-ssa-loop-im.c: Update
565 2005-01-20 Richard Henderson <rth@redhat.com>
567 * rtl.def (CONST_VECTOR): Use RTX_CONST_OBJ.
568 * rtl.h (CONSTANT_P): Don't special case CONST_VECTOR.
570 2005-01-19 Richard Henderson <rth@redhat.com>
573 * config/i386/i386.c (ix86_expand_vector_move_misalign): Convert
574 to V4SFmode in SSE1 fallback load path.
576 2005-01-19 Richard Henderson <rth@redhat.com>
578 * config/i386/i386.c (ix86_expand_vector_init_one_var): Fix typo
579 in QImode expansion to ix86_expand_vector_set.
580 (ix86_expand_vector_init_general): Fix typo in V8HImode recursive call.
582 2005-01-19 Richard Henderson <rth@redhat.com>
585 * config/i386/i386.c (ix86_preferred_reload_class): Return a proper
586 subclass of the input class.
587 (ix86_secondary_memory_needed): Always true for cross-MMX classes.
588 Always true for cross-SSE1 classes. Rationalize conditionals.
589 * config/i386/i386.h (SSE_CLASS_P, MMX_CLASS_P): Use straight equality.
590 * config/i386/i386.md (movsi_1): Add MMX/SSE zeros. Fix alternatives
591 for SSE1. Don't check TARGET_INTER_UNIT_MOVES.
592 (movdi_2): Add MMX/SSE zeros.
593 (movdi_1_rex64): Likewise. Don't check TARGET_INTER_UNIT_MOVES.
594 (movsf_1): Don't check TARGET_INTER_UNIT_MOVES.
595 (zero_extendsidi2_32, zero_extendsidi2_rex64): Likewise.
596 (movsi_1_nointernunit, movdi_1_rex64_nointerunit): Remove.
597 (movsf_1_nointerunit, zero_extendsidi2_32_1): Remove.
598 (zero_extendsidi2_rex64_1): Remove.
599 (MOV0 peephole): Check GENERAL_REG_P.
601 2005-01-19 Richard Henderson <rth@redhat.com>
604 * config/i386/i386.c (ix86_expand_vector_set): Fix third and fourth
606 (ix86_expand_vector_extract): Likewise.
608 2005-01-19 Richard Henderson <rth@redhat.com>
611 * expr.c (emit_move_change_mode): New argument force; use
612 simplify_gen_subreg if true.
613 (emit_move_via_alt_mode): Merge into ...
614 (emit_move_via_integer): ... here.
615 (emit_move_ccmode): Use emit_move_change_mode directly.
617 2005-01-19 Richard Henderson <rth@redhat.com>
620 * config/alpha/alpha.c (alpha_rtx_costs): Handle HIGH.
621 (alpha_preferred_reload_class): Handle CONST_VECTOR.
622 (alpha_emit_set_const_1): Add no_output parameter; don't emit
624 (alpha_emit_set_const): Likewise. Make static.
625 (alpha_emit_set_long_const): Make static.
626 (alpha_extract_integer): Split out from alpha_expand_mov.
627 (alpha_split_const_mov): Likewise.
628 (alpha_expand_mov): Use them. Handle CONST_VECTOR.
629 (alpha_legitimate_constant_p): New.
630 * config/alpha/alpha-protos.h: Update.
631 * config/alpha/alpha.h (REGISTER_MOVE_COST): Correct fp<->gp cost.
632 (LEGITIMATE_CONSTANT_P): Re-implement with a function.
633 * config/alpha/alpha.md (movsi): Add n alternative.
634 (movsi_nt_vms, movdi_er_nofix, movdi_er_fix, movdi_fix): Likewise.
635 (mov<VEC>_fix, mov<VEC>_nofix): Add i alternative.
636 (splitters for all of the above): Use alpha_split_const_mov.
637 * config/alpha/predicates.md (non_add_const_operand): New.
638 (non_zero_const_operand): New.
639 (input_operand): Use alpha_legitimate_constant_p after reload.
641 2005-01-19 Zdenek Dvorak <dvorakz@suse.cz>
643 PR tree-optimization/19038
644 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): New function.
645 (add_candidate): Add ivs with increment in latch only if
646 allow_ip_end_pos_p is true.
647 (determine_iv_cost): Use empty_block_p.
649 2005-01-19 Daniel Berlin <dberlin@dberlin.org>
651 * cfganal.c (compute_dominance_frontiers_1): Replace with new algorithm
652 (compute_dominance_frontiers): Ditto.
654 2005-01-19 Ralf Corsepius <ralf.corsepius@rtems.org>
657 * config/sh/t-rtems: New.
658 * config.gcc (sh-*-rtems*): Reflect having added config/sh/t-rtems.
660 2005-01-19 Zdenek Dvorak <dvorakz@suse.cz>
662 * tree-ssa-ccp.c (maybe_fold_offset_to_component_ref): Always subtract
663 the offset of the selected field.
665 2005-01-19 Kazu Hirata <kazu@cs.umass.edu>
667 * tree-cfg.c (remove_forwarder_block): Fix the check to
668 prevent a nonlocal label from appearing in the middle of a
671 2005-01-19 Hans-Peter Nilsson <hp@axis.com>
673 PR rtl-optimization/19462
674 * reorg.c (find_end_label): Create return insn only if
675 current_function_epilogue_delay_list is empty.
677 2005-01-19 Paolo Bonzini <bonzini@gnu.org>
679 * doc/passes.texi: Remove paragraph mentioning new-ra.
681 2005-01-19 Jakub Jelinek <jakub@redhat.com>
683 PR rtl-optimization/15139
684 * combine.c: Include params.h.
685 (count_rtxs): New function.
686 (record_value_for_reg): If replace_rtx would replace at least
687 2 occurrences of REG in VALUE and TEM is really large, replace REG with
688 (clobber (const_int 0)) instead of TEM.
689 * params.def (PARAM_MAX_LAST_VALUE_RTL): New.
690 * params.h (MAX_LAST_VALUE_RTL): New.
691 * Makefile.in (combine.o): Depend on $(PARAMS_H).
692 * doc/invoke.texi (--param max-last-value-rtl=N): Document.
695 * c-typeck.c (digest_init): Only call build_vector if all constructor
696 elements are *_CST nodes.
697 * gimplify.c (gimplify_init_constructor): Likewise.
700 * c-typeck.c (digest_init): Only call build_vector if inside_init
703 2005-01-18 Tobias Schl"uter <tobias.schlueter@physik.uni-muenchen.de>
705 * toplev.c (init_asm_output): Remove dead #ifdef.
707 2005-01-18 Kaz Kojima <kkojima@gcc.gnu.org>
709 * config/sh/linux.h (TARGET_C99_FUNCTIONS): Define.
711 2005-01-18 Eric Botcazou <ebotcazou@libertysurf.fr>
713 * calls.c (expand_call): Check DECL_BUILT_IN_CLASS before
714 accessing DECL_FUNCTION_CODE.
715 * dojump.c (do_jump): Likewise.
716 * gimplify.c (gimplify_call_expr): Likewise.
717 * predict.c (expr_expected_value): Likewise.
718 (strip_builtin_expect): Likewise.
719 * tree-inline.c (estimate_num_insns_1): Likewise.
720 * tree-ssa-loop-im.c (stmt_cost): Likewise
721 * fold-const.c (fold): Test for BUILT_IN_NORMAL.
722 (tree_expr_nonnegative_p): Likewise.
724 2005-01-18 Eric Botcazou <ebotcazou@libertysurf.fr>
726 * config/sparc/sparc.c (load_pic_register): Emit the appropriate
727 variant of the load_pcrel_sym pattern.
728 * config/sparc/sparc.md (P macro): Move to the top.
729 (load_pcrel_sym): Macroize using P.
730 (save_register_window): Likewise.
732 2005-01-18 Aldy Hernandez <aldyh@redhat.com>
734 * config/rs6000/rs6000.md ("sunordered"): Disable for e500.
737 2005-01-18 Andrew Pinski <pinskia@physics.uc.edu>
740 * c-typeck.c (build_asm_expr): Strip nops off
741 input memory operands.
743 2005-01-18 David Edelsohn <edelsohn@gnu.org>
745 * config/rs6000/aix43.h (CPLUSPLUS_CPP_SPEC): Delete
746 _XOPEN_SOURCE, _XOPEN_SOURCE_EXTENDED, _LARGE_FILE_API.
747 * config/rs6000/aix51.h (CPLUSPLUS_CPP_SPEC): Same.
748 * config/rs6000/aix52.h (CPLUSPLUS_CPP_SPEC): Same.
750 2005-01-18 Joel Sherrill <joel.sherrill@oarcorp.com>
753 * config/i386/i386.c (override_options): If the 80387 is disabled,
754 then do not return FP values using FP registers.
756 2005-01-18 Kazu Hirata <kazu@cs.umass.edu>
758 * basic-block.h, c-common.c, c-cppbuiltin.c, c-lang.c,
759 c-tree.h, cfgbuild.c, cgraph.c, cgraph.h, collect2.c,
760 combine.c, config.gcc, coverage.h, cse.c, cselib.c,
761 defaults.h, df.c, dwarf2asm.c, dwarf2out.c, explow.c, expr.c,
762 flow.c, fold-const.c, gcse.c, ggc-page.c, gimple-low.c,
763 gimplify.c, ifcvt.c, langhooks-def.h, lcm.c, optabs.h,
764 output.h, postreload-gcse.c, postreload.c, recog.c,
765 resource.c, rtl.def, rtlanal.c, sched-deps.c, sched-rgn.c,
766 targhooks.h, toplev.c, tree-data-ref.c, tree-eh.c,
767 tree-flow.h, tree-loop-linear.c, tree-mudflap.h, tree-nrv.c,
768 tree-optimize.c, tree-outof-ssa.c, tree-pass.h,
769 tree-scalar-evolution.c, tree-ssa-copy.c, tree-ssa-dce.c,
770 tree-ssa-dse.c, tree-ssa-loop-ivopts.c, tree-ssa-operands.c,
771 tree-ssa-pre.c, tree-ssa.c, tree-vectorizer.c, tree.def,
772 unwind-dw2-fde-darwin.c, var-tracking.c: Update copyright.
774 * config/darwin.c, config/darwin.h, config/freebsd-spec.h,
775 config/alpha/alpha.c, config/arm/vfp.md,
776 config/i386/emmintrin.h, config/i386/i386-modes.def,
777 config/i386/i386-protos.h, config/i386/i386.h,
778 config/i386/pmmintrin.h, config/i386/ppro.md,
779 config/i386/predicates.md, config/i386/xmmintrin.h,
780 config/i860/i860.c, config/ia64/ia64-modes.def,
781 config/ia64/ia64-protos.h, config/ia64/ia64.h,
782 config/ia64/ia64.md, config/ia64/itanium1.md,
783 config/ia64/itanium2.md, config/ia64/predicates.md,
784 config/iq2000/iq2000.h, config/mips/linux64.h,
785 config/rs6000/aix41.h, config/rs6000/aix43.h,
786 config/rs6000/aix52.h, config/rs6000/darwin-fallback.c,
787 config/rs6000/rs6000.c, config/rs6000/rs6000.h,
788 config/rs6000/spe.md, config/sh/sh.md, config/sparc/linux.h,
789 config/sparc/linux64.h, config/sparc/litecoff.h,
790 config/sparc/sp64-elf.h, config/sparc/sparc.c,
791 config/sparc/sparc.h, config/sparc/sparc.md,
792 config/sparc/sysv4-only.h: Update copyright.
794 2005-01-18 Richard Henderson <rth@redhat.com>
797 * config/i386/i386.c (ix86_expand_fp_movcc): Fail for LTGT and UNEQ.
799 2005-01-18 Marc Espie <espie@openbsd.org>
800 * config/i386/openbsdelf.h: Typo.
802 2005-01-18 Richard Henderson <rth@redhat.com>
804 * config/i386/i386.c (ix86_expand_fp_movcc): Remove TARGET_IEEE_FP
805 special case for sse.
806 * config/i386/predicates.md (sse_comparison_operator): Likewise.
808 2005-01-18 Dorit Naishlos <dorit@il.ibm.com>
810 * tree-inline.c (estimate_num_insns_1): Added cases for
811 ALIGN_INDIRECT_REF, MISALIGNED_INDIRECT_REF, and REALIGN_LOAD_EXPR.
813 2005-01-18 Eric Botcazou <ebotcazou@libertysurf.fr>
815 PR rtl-optimization/19296
816 * combine.c (simplify_comparison): Rewrite the condition under
817 which a non-paradoxical SUBREG of a PLUS can be lifted when
818 compared against a constant.
820 2005-01-18 Andi Kleen <ak@muc.de>
822 * c-typeck.c: (convert_for_assignment): Check warn_pointer_sign.
823 * c.opt (-Wpointer-sign): Add.
824 * doc/invoke.texi: (-Wpointer-sign): Add.
826 2005-01-18 Uros Bizjak <uros@kss-loka.si>
828 * config/i386/i386.c (override_options): Revert 2004-11-24 change.
829 * config/i386/i386.md (fmodsf3, dremsf3, *sinsf2, *cossf2, sincossf3,
830 *tansf3_1, tansf2, atan2sf3_1, atan2sf3, atansf2, asinsf2, acossf2,
831 logsf2, log10sf2, log2sf2, log1psf2, logbsf2, ilogbsf2, expsf2,
832 exp10sf2, exp2sf2, expm1sf2, rintsf2, floorsf2, ceilsf2, btruncsf2,
833 nearbyintsf2): Disable for TARGET_SSE_MATH. Leave patterns enabled
834 for TARGET_MIX_SSE_I387.
835 (fmoddf3, dremdf3, *sindf2, *sinextendsfdf2, *cosdf2, *cosextendsfdf2,
836 sincosdf3, *sincosextendsfdf3, *tandf3_1, tandf2, atan2df3_1,
837 atan2df3, atandf2, asindf2, acosdf2, logdf2, log10df2, log2df2,
838 log1pdf2, logbdf2, expdf2, exp10df2, exp2df2, expm1df2, rintdf2,
839 floordf2, ceildf2, btruncdf2, nearbyintdf2): Disable for
840 (TARGET_SSE2 && TARGET_SSE_MATH). Leave patterns enabled for
842 (atan2sf3, atan2df3, atan2xf3): Remove register constraints
845 2005-01-18 Uros Bizjak <uros@kss-loka.si>
848 * config/i386/mmx.md (*movv2sf_internal_rex64, *movv2sf_internal):
849 Add movaps alternative for xmm reg->reg move.
851 2005-01-17 Jeff Law <law@redhat.com>
853 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Clear SSA_NAME_VALUE
854 at the end of the main DOM loop rather than just before DOM exits.
856 2005-01-17 Diego Novillo <dnovillo@redhat.com>
858 PR tree-optimization/19121
859 * tree-ssa-alias.c (compute_flow_sensitive_aliasing): When
860 adding aliases to a name tag, also add them to the pointer's
862 * tree-ssa-copy.c (merge_alias_info): Do not merge flow
863 sensitive alias info at all. Only check that the two pointers
864 have compatible pointed-to sets.
865 * tree-ssa.c (verify_name_tags): Verify that the alias set of
866 a pointer's type tag is a superset of the alias set of the
869 2005-01-17 James E Wilson <wilson@specifixinc.com>
872 * config/ia64/ia64.md (movxf): Handle general register source. Adjust
873 comment to document why.
875 2005-01-17 Richard Henderson <rth@redhat.com>
877 * config/i386/sse.md (smaxv4sf3_finite, sse_vmsmaxv4sf3_finite,
878 sminv4sf3_finite, sse_vmsminv4sf3_finite, smaxv2df3_finite,
879 sse2_vmsmaxv2df3_finite, sminv2df3_finite,
880 sse2_vmsminv2df3_finite): New.
881 (smaxv4sf3, sse_vmsmaxv4sf3, sminv4sf3, sse_vmsminv4sf3, smaxv2df3,
882 sse2_vmsmaxv2df3, sminv2df3, sse2_vmsminv2df3): Remove commutative.
883 Force op1 into register.
885 2005-01-17 Kelley Cook <kcook@gcc.gnu.org>
887 * Makefile.in (STAGEMOVESTUFF): Stage all the stamp files to prevent
888 bubblestrap from always rebuilding build directory.
890 2005-01-17 Ulrich Weigand <uweigand@de.ibm.com>
892 * config/s390/s390.md ("*clc"): Remove incorrect '='.
893 ("*cmpmem_short"): Likewise.
895 2005-01-17 Steven Bosscher <stevenb@suse.de>
897 * basic-block.h: Document BB_* flags.
898 * regrename.c (copyprop_hardreg_forward): Don't use BB_VISITED,
899 use an sbitmap instead.
900 * sched-rgn.c (compute_trg_info): Likewise.
902 2005-01-17 Richard Sandiford <rsandifo@redhat.com>
904 * config.gcc (mips64*-*-linux*): Set the default abi to n32. Remove
905 redundant target_cpu_default and extra_parts lines. Use the default
907 * config/mips/linux64.h (DRIVER_DEFAULT_ABI_SELF_SPEC): Delete.
908 (SUBTARGET_EXTRA_SPECS): Delete.
909 (DRIVER_SELF_SPECS): Remove architecture lines. Set the ABI directly.
910 (SUBTARGET_ASM_SPEC): Remove -mabi=n32 and -mabi=64 mappings.
911 * config/mips/t-linux64 (MULTILIB_DIRNAMES): Use "32" for -mabi=32
912 and "n32" for -mabi=n32.
914 2005-01-17 Ian Lance Taylor <ian@airs.com>
917 * c-typeck.c (build_c_cast): Revert patch of 2003-02-16: don't
918 fold constant variables into initial values.
920 2005-01-17 Kazu Hirata <kazu@cs.umass.edu>
922 * tree-cfg.c (tree_can_merge_blocks_p): Reorder two checks.
924 * tree-cfg.c (tree_forwarder_block_p): Speed up by walking
925 through the statements backward.
927 2005-01-17 Ian Lance Taylor <ian@airs.com>
930 * tree-inline.c (expand_call_inline): Set TREE_NO_WARNING on
931 a variable set to the return value of the inlined function.
933 2005-01-17 Mark Dettinger <dettinge@de.ibm.com>
935 * rtlanal.c (rtx_cost): Assign cost of 0 to a SUBREG
936 when modes are tieable.
938 2005-01-17 Ranjit Mathew <rmathew@hotmail.com>
940 * gthr-posix.h (__gthread_active_p): Use pthread_cancel instead
941 of pthread_create to find out if threads are enabled.
942 * gthr-posix95.h (__gthread_active_p): Likewise.
944 2005-01-17 Paolo Bonzini <bonzini@gnu.org>
946 * common.opt (-fnew-ra): Remove.
948 * toplev.h (flag_new_regalloc): Remove.
949 * Makefile.in (ra*.*): Don't mention.
950 * passes.c (rest_of_handle_new_regalloc): Remove.
951 (rest_of_handle_combine, rest_of_compilation): Always consider
952 flag_new_regalloc as false.
953 * doc/invoke.texi: Don't document -fnew-ra.
955 2005-01-17 Paolo Bonzini <bonzini@gnu.org>
957 * bb-reorder.c (fix_edges_for_rarely_executed_code): Remove
958 last parameter to reg_scan.
959 * loop.c (loop_optimize): Likewise.
960 * passes.c (rest_of_handle_tracer, rest_of_handle_if_conversion,
961 rest_of_handle_web, rest_of_handle_cfg, rest_of_handle_jump_bypass,
962 rest_of_handle_life, rest_of_handle_cse, rest_of_handle_cse2,
963 rest_of_handle_gcse, rest_of_handle_loop_optimize,
964 rest_of_handle_loop2, rest_of_handle_jump2): Likewise.
965 * regclass.c (reg_scan): Likewise, for the declaration.
966 * rtl.h (reg_scan): Likewise, for the prototype.
968 2005-01-17 Kazu Hirata <kazu@cs.umass.edu>
970 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Don't call
973 2005-01-16 Daniel Berlin <dberlin@dberlin.org>
975 * tree-ssa-pre.c (add_to_sets): s1 may be NULL.
976 (compute_avail): Uses don't go in tmp_gen.
978 2005-01-16 Steven Bosscher <stevenb@suse.de>
980 * ggc-page.c (ggc_alloc_stat): Use __builtin_ctzl instead of a
981 loop to look for a free slot in a page entry.
983 2005-01-16 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
986 * defaults.h (TARGET_DEFERRED_OUTPUT_DEFS): Provide default.
987 * toplev.c (compile_file): Call process_pending_assemble_output_defs
988 just before targetm.asm_out.file_end.
989 * tree.h (process_pending_assemble_output_defs): Declare.
990 * varasm.c (assemble_output_def, process_pending_assemble_output_defs):
992 (assemble_alias): Defer generation of assembly code for defines when
993 TARGET_DEFERRED_OUTPUT_DEFS is true.
994 * config/rs6000/aix41.h (TARGET_DEFERRED_OUTPUT_DEFS): Define.
995 * config/rs6000/aix43.h (TARGET_DEFERRED_OUTPUT_DEFS): Define.
996 * doc/tm.texi (TARGET_DEFERRED_OUTPUT_DEFS): document.
998 2005-01-15 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1001 * pa.c (pa_scalar_mode_supported_p): New function.
1002 (TARGET_SCALAR_MODE_SUPPORTED_P): Define.
1003 * pa.h (MIN_UNITS_PER_WORD): Add comment.
1005 * pa-protos.h (prefetch_operand): Delete.
1006 (prefetch_cc_operand, prefetch_nocc_operand): New declations.
1007 * pa.c (prefetch_operand): Delete.
1008 (prefetch_cc_operand, prefetch_nocc_operand): New functions.
1009 * pa.h (EXTRA_CONSTRAINT): Add `W' constraint.
1010 (PREDICATE_CODES): Delete prefetch_operand. Add prefetch_cc_operand
1011 and prefetch_nocc_operand.
1012 * pa.md (prefetch): Rework to avoid reload problems handling short
1013 displacements when a cache control completer needs to be provided.
1014 (prefetch_32, prefetch_64): Delete.
1015 (prefetch_cc, prefetch_nocc): New patterns.
1017 2005-01-15 David Edelsohn <edelsohn@gnu.org>
1019 * config/rs6000/aix52.h (CPLUSPLUS_CPP_SPEC): Revert previous change.
1021 2005-01-15 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1024 * pa.c (pa_asm_output_aligned_common): Warn if specified alignment
1025 exceeds maximum alignment for global common data.
1026 * pa64-hpux.h (MAX_OFILE_ALIGNMENT): Define to 32768.
1027 * som.h (MAX_OFILE_ALIGNMENT): Likewise.
1029 2004-01-15 Roger Sayle <roger@eyesopen.com>
1031 * tree-ssa-dom.c (extract_range_from_cond): Correct condition.
1033 2004-01-15 Roger Sayle <roger@eyesopen.com>
1035 * harg-reg-set.h (reg_class_names): Prototype global array.
1036 * regclass.c (reg_class_names): Declare here and initialize to
1038 (dump_regclass): Remove local declaration of reg_class_names.
1039 (regclass): Likewise.
1040 * cfg.c (dump_flow_info): Likewise.
1041 * ra-debug.c (reg_class_names): Likewise.
1042 * regrename.c (reg_class_names): Likewise.
1043 * reload.c (reg_class_names): Likewise.
1044 * reload1.c (spill_failure): Likewise.
1045 * config/m68hc11/m68hc11.c (reg_class_names): Likewise.
1047 2005-01-15 Ulrich Weigand <uweigand@de.ibm.com>
1049 * config/s390/s390.md ("reload_outti"): Remove predicate for
1050 output operand. Abort if operand is not a MEM.
1051 ("reload_outdi", "reload_outdf"): Likewise.
1053 2005-01-15 Marc Espie <espie@openbsd.org>
1055 * config.gcc (*-*-openbsd*): Set HAS_LIBC_R for OpenBSD <= 3.2.
1056 (i[34567]86-*-openbsd*): Switch to ELF for OpenBSD >= 3.4.
1057 * config/openbsd.h: Fix C++ includes for native configurations.
1058 Add proper OS_CPP_BUILTINS. Fix libspec for recent OpenBSD.
1059 Add trampoline support.
1060 * config/i386/openbsdelf.h: New.
1062 2005-01-15 Marc Espie <espie@openbsd.org>
1064 * collect2.c (main): Explicitly parse -dynamic-linker option.
1066 2005-01-15 Jakub Jelinek <jakub@redhat.com>
1068 PR tree-optimization/19060
1069 * tree-ssa-dom.c (extract_range_from_cond) <case LT_EXPR, GT_EXPR>:
1070 Return 0 if op1 <= TYPE_MIN_VALUE () resp. op1 >= TYPE_MAX_VALUE ().
1071 (simplify_cond_and_lookup_avail_expr): Add assert for dummy == 0
1072 and handle extract_range_from_cond returning false.
1073 * fold-const.c (fold): Optimize comparisons with min/max even for
1074 width > HOST_BITS_PER_WIDE_INT.
1076 2005-01-15 Ralf Corsepius <ralf.corsepius@rtems.org>
1078 * config/mips/rtems.h (MIPS_DEFAULT_GVALUE): Set to 0.
1079 * config/mips/t-rtems (MULTILIBS_DIRNAMES,MULTILIB_OPTIONS):
1080 Remove little endian multilib variants.
1081 Add mips32 multilib variant.
1083 2005-01-15 Kazu Hirata <kazu@cs.umass.edu>
1085 * explow.c (copy_all_reg, stabilize): Remove.
1086 * expr.h: Remove the corresponding prototypes.
1088 * rtlanal.c (regs_set_between_p): Remove.
1089 * rtl.h: Remove the corresponding prototype.
1091 2005-01-14 David Edelsohn <edelsohn@gnu.org>
1093 * config/rs6000/rs6000.c (rs6k_nonimmediate_operand): Rename
1094 to rs6000_nonimmediate_operand.
1095 * config/rs6000/rs6000.h (PREDICATE_CODES): Rename
1096 rs6k_nonimmediate_operand to rs6000_nonimmediate_operand.
1097 * config/rs6000/rs6000.md (movsi_internal1): Rename predicate to
1098 rs6000_nonimmediate_operand.
1099 * config/rs6000/spe.md (movdf_e500_double): Rename predicate to
1100 rs6000_nonimmediate_operand.
1102 2005-01-14 Andrew Pinski <pinskia@physics.uc.edu>
1104 * tree-ssa-alias.c (pass_may_alias): Add TODO_verify_stmts.
1106 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
1108 * config/rs6000/rs6000.h (CLASS_MAX_NREGS): DF goes in 1 register
1110 (CANNOT_CHANGE_MODE_CLASS): Restrict DI mode changes on e500v2.
1111 (PREDICATE_CODES): Add rs6k_nonimmediate_operand.
1113 * config/rs6000/rs6000.c (invalid_e500_subreg): New.
1114 (rs6k_nonimmediate_operand): New.
1115 (rs6000_legitimate_offset_address_p): Handle DI modes on e500v2
1117 (legitimate_lo_sum_address_p): Same.
1118 (rs6000_legitimize_address): Same.
1119 (rs6000_legitimize_reload_address): Same.
1120 (rs6000_legitimate_address): Same.
1121 (spe_build_register_parallel): Pass DF and DC modes in a DI
1124 * config/rs6000/rs6000.md ("*movsi_internal1"): Change predicate
1125 to rs6k_nonimmediate_operand.
1127 * config/rs6000/spe.md ("*frob_df_di"): New.
1128 ("*frob_di_df"): New.
1129 ("*frob_di_df_2"): New.
1130 ("*mov_sidf_e500_subreg0"): New.
1131 ("*mov_sidf_e500_subreg4"): New.
1132 ("*movdf_e500_double"): Change predicate to
1133 rs6k_nonimmediate_operand.
1135 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
1137 * postreload.c (move2add_note_store): Only call
1138 trunc_int_for_mode on scalar integers.
1140 2005-01-14 Steven Bosscher <stevenb@suse.de>
1142 * tree-ssa-dce.c (visited_control_parents): New sbitmap to
1143 replace BB_VISITED uses.
1144 (find_obviously_necessary_stmts): Don't clear BB_VISITED.
1145 (propagate_necessity): Check the bitmap instead of BB_VISITED.
1146 (tree_dce_done): Free visited_control_parents.
1147 (perform_tree_ssa_dce): Allocate and clear it.
1148 * tree-ssa-pre.c (compute_antic_aux): Make non-recursive.
1149 (compute_antic): Iterate from here using a DFS. Use an sbitmap
1150 instead of BB_VISITED.
1152 2005-01-14 Kazu Hirata <kazu@cs.umass.edu>
1154 * c-tree.h, coverage.h, langhooks-def.h, optabs.h, output.h,
1155 rtl.h, targhooks.h, tree-flow.h, tree-mudflap.h, tree.h:
1156 Remove unused prototypes.
1158 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
1160 * config/rs6000/rs6000.c (rs6000_stack_info): Fix caching of SPE
1161 64-bit register usage.
1162 (spe_func_has_64bit_regs_p): Add FIXME note.
1164 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
1166 * config/rs6000/rs6000.c (rs6000_generate_compare): Replace
1167 flag_finite_math_only with flag_unsafe_math_optimizations.
1169 2005-01-14 J. D. Johnston <jjohnst@us.ibm.com>
1171 * config/s390/tpf-unwind.h (s390_fallback_frame_state): Correct
1173 (__tpf_eh_return): Copy TPF private stack area from the module
1174 boundary stack frame. Add check for when module addresses are
1175 equal, but no stub address is found.
1177 2005-01-14 Richard Earnshaw <rearnsha@arm.com>
1180 * arm.h (struct machine_function): Add call_via field.
1181 (thumb_call_via_label): Declare.
1182 * arm.c (thumb_call_via_label): New variable.
1183 (thumb_call_reg_needed): New variable.
1184 (arm_output_function_epilogue): For Thumb code, output any per-function
1185 call-indirect trampolines.
1186 (thumb_call_via_reg): New function.
1187 (arm_file_end): New function.
1188 (TARGET_ASM_FILE_END): Call arm_file_end.
1189 (aof_file_end): Likewise.
1190 * arm-protos.h (thumb_call_via_reg): Declare.
1191 * arm.md (call_reg_thumb, call_value_reg_thumb): Call
1192 thumb_call_via_reg in normal case.
1194 2005-01-14 Jakub Jelinek <jakub@redhat.com>
1197 PR rtl-optimization/19348
1198 * recog.c (peephole2_optimize): Do global life update if some peephole
1199 decides it doesn't need at least one of its inputs and that change
1200 influences liveness at the start of the basic block.
1202 * basic-block.h (EXECUTE_IF_AND_COMPL_IN_REG_SET): Needs 2 REGSET
1203 arguments instead of 1.
1205 2005-01-14 Eric Botcazou <ebotcazou@libertysurf.fr>
1208 * varasm.c (initializer_constant_valid_p) <ADDR_EXPR>: Return
1209 zero for nested functions needing a static chain or functions
1210 with a non-constant address.
1212 2005-01-13 Roger Sayle <roger@eyesopen.com>
1214 * simplify-rtx.c (simplify_binary_operation) <AND>: Optimize
1215 (and (sign_extend X) C) into (zero_extend (and X C)).
1217 2005-01-13 David O'Brien <obrien@FreeBSD.org>
1219 * config/freebsd-spec.h: Make KSE pthread lib logic the default.
1221 2005-01-13 Richard Henderson <rth@redhat.com>
1226 * config/i386/i386.md (cmpdf, cmpsf, bunordered, bordered, buneq,
1227 bunge, bungt, bunle, bunlt, bltgt): Enable for TARGET_SSE_MATH,
1228 not just TARGET_SSE.
1229 (cmpfp_i_387): Rename from cmpfp_i. Move after sse patterns.
1230 (cmpfp_i_mixed): Rename from cmpfp_i_sse; use for TARGET_MIX_SSE_I387.
1231 (cmpfp_i_sse): Rename from cmpfp_i_sse_only; use for TARGET_SSE_MATH.
1232 (cmpfp_iu_mixed, cmpfp_iu_sse, cmpfp_iu_387): Similarly.
1233 (fp_jcc_1_mixed, fp_jcc_1_sse, fp_jcc_1_387): Similarly.
1234 (fp_jcc_2_mixed, fp_jcc_2_sse, fp_jcc_2_387): Similarly.
1235 (fp_jcc_3_387, fp_jcc_4_387, fp_jcc_5_387, fp_jcc_6_387,
1236 fp_jcc_7_387, fp_jcc_8_387): Rename from fp_jcc_N.
1237 (movdicc_c_rex64): Rename with '*'.
1238 (movsfcc, movdfcc): Add checks for 387 and sse math to condition.
1239 (movsfcc_1_sse_min, movsfcc_1_sse_max, movsfcc_1_sse): New.
1240 (movsfcc_1_387): Rename from movsfcc_1.
1241 (movdfcc_1_sse_min, movdfcc_1_sse_max, movdfcc_1_sse): New.
1242 (movdfcc_1, movdfcc_1_rex64): Add check for 387.
1243 (sminsf3, smaxsf3, smindf3, smaxdf3): New.
1244 (minsf3, minsf, minsf_nonieee, minsf_sse, mindf3, mindf,
1245 mindf_nonieee, mindf_sse, maxsf3, maxsf, maxsf_nonieee, maxsf_sse,
1246 maxdf3, maxdf, maxdf_nonieee, maxdf_sse, sse_movsfcc, sse_movsfcc_eq,
1247 sse_movdfcc, sse_movdfcc_eq, sse_movsfcc_const0_1,
1248 sse_movsfcc_const0_2, sse_movsfcc_const0_3, sse_movsfcc_const0_4,
1249 sse_movdfcc_const0_1, sse_movdfcc_const0_2, sse_movdfcc_const0_3,
1250 sse_movdfcc_const0_4): Remove.
1251 * config/i386/i386.c (ix86_expand_fp_movcc): For TARGET_SSE_MATH,
1252 recognize min/max early. Update for changed sse cmove patterns.
1253 (ix86_split_sse_movcc): New.
1254 * config/i386/i386-protos.h: Update.
1256 2005-01-13 Steven Bosscher <stevenb@suse.de>
1258 * tree-ssa-dse.c (fix_phi_uses): Use SSA operand iterators.
1259 (fix_stmt_v_may_defs): Likewise.
1261 2005-01-13 Richard Henderson <rth@redhat.com>
1263 * config/i386/i386.c (ix86_expand_fp_absneg_operator): Use elt_mode
1264 for converting the mask.
1266 2005-01-13 David Edelsohn <edelsohn@gnu.org>
1268 * config/rs6000/aix52.h (CPLUSPLUS_CPP_SPEC): Change _XOPEN_SOURCE
1271 2005-01-13 Richard Henderson <rth@redhat.com>
1273 * config/i386/i386.c (IX86_BUILTIN_MOVQ, IX86_BUILTIN_LOADD,
1274 IX86_BUILTIN_STORED, IX86_BUILTIN_MOVQ2DQ,
1275 IX86_BUILTIN_MOVDQ2Q): Remove.
1276 (IX86_BUILTIN_VEC_EXT_V4SI): New.
1277 (ix86_init_mmx_sse_builtins, ix86_expand_builtin): Update to match.
1278 (ix86_expand_vector_extract): For V4S[FI], extract element 0 after
1280 * config/i386/sse.md (sse_concatv2sf): Accept zero operand 2.
1281 (sse2_pextrw): Fix immediate constraint.
1282 (sse2_loadq, sse2_loadq_rex64): Remove.
1283 * config/i386/emmintrin.h (_mm_cvtsi128_si32, _mm_cvtsi128_si64x):
1284 Use __builtin_ia32_vec_ext_<size>.
1285 (_mm_cvtsi32_si128, _mm_cvtsi64x_si128): Use _mm_set_epi<size>.
1287 2005-01-13 Aldy Hernandez <aldyh@redhat.com>
1289 * function.c (assign_parm_setup_block): Look inside original
1290 entry_parm when inspecting PARALLEL.
1292 2005-01-13 Ralf Corsepius <ralf.corsepius@rtems.org>
1293 Joel Sherrill <joel@oarcorp.com>
1296 * gthr-rtems.h (__gthread_recursive_mutex_t): New type.
1297 (__GTHREAD_RECURSIVE_MUTEX_INIT_FUNCTION): Define to
1298 rtems_gxx_recursive_mutex_init.
1299 (__gthread_recursive_mutex_lock): New function.
1300 (__gthread_recursive_mutex_trylock): Likewise.
1301 (__gthread_recursive_mutex_unlock): Likewise.
1303 2005-01-13 Ralf Corsepius <ralf.corsepius@rtems.org>
1305 * config/i386/t-rtems-i386: Multilib on -mtune instead of -mcpu.
1307 2005-01-13 Jan Beulich <jbeulich@novell.com>
1309 * config/ia64/ia64.md (ashlti3, ashlti3_internal): New.
1310 (ashrti3_internal): Indicate output is early clobber. Generate result
1311 into output rather than first input. Use move for low word of output
1312 if shift count is exactly 64.
1313 (lshrti3_internal): Likewise.
1315 2005-01-13 Hans-Peter Nilsson <hp@bitrange.com>
1319 * reload1.c (delete_output_reload): Don't delete an output reload
1320 if the pseudo lives longer than a single basic block. Adjust and
1321 improve wording of comment.
1323 2005-01-12 Andrew Pinski <pinskia@physics.uc.edu>
1326 * config/darwin.c (machopic_select_section): Use TYPE_SIZE_UNIT instead
1327 of TYPE_SIZE where we mean the number of bytes.
1329 2005-01-12 Aldy Hernandez <aldyh@redhat.com>
1331 * function.c (assign_parm_setup_block): Relax condition on
1332 multi-register optimization.
1334 2005-01-12 Nick Clifton <nickc@redhat.com>
1336 * config/sh/sh.md (udivsi3_sh2a, divsi3_sh2a): Give these patterns
1337 an "in_delay_slot" attribute of "no" to prevent them being used in
1338 delay slots. This is forbidden because they might generate
1341 2005-01-12 Alan Modra <amodra@bigpond.net.au>
1344 * config/rs6000/rs6000.md (movtf_internal): Replace r->o and m->r
1347 2005-01-12 Nick Clifton <nickc@redhat.com>
1349 * config/iq2000/iq2000.h (ASM_SPEC): Undefine (to stop -Qy being
1350 passed on to GAS) but do not define, as GAS no longer supports or
1351 needs the -m2000 option.
1353 2005-01-11 Roger Sayle <roger@eyesopen.com>
1355 * builtins.c (fold_builtin_fabs): Convert argument to the appropriate
1356 type as args of unprototyped builtins aren't automatically promoted.
1357 (fold_builtin_abs): Likewise.
1359 2005-01-11 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1361 * Makefile.in: Set a `build-warn' variable.
1363 2005-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1365 PR tree-optimization/17949
1366 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): New function.
1367 (find_interesting_uses_address): Use it.
1369 2005-01-11 Aldy Hernandez <aldyh@redhat.com>
1371 * regrename.c (kill_value): Handle subreg's that won't simplify.
1373 2005-01-11 Richard Henderson <rth@redhat.com>
1376 * config/i386/i386.h (enum ix86_builtins): Move ...
1377 * config/i386/i386.c: ... here.
1378 (IX86_BUILTIN_MOVDDUP, IX86_BUILTIN_MMX_ZERO, IX86_BUILTIN_PEXTRW,
1379 IX86_BUILTIN_PINSRW, IX86_BUILTIN_LOADAPS, IX86_BUILTIN_LOADSS,
1380 IX86_BUILTIN_STORESS, IX86_BUILTIN_SSE_ZERO, IX86_BUILTIN_PEXTRW128,
1381 IX86_BUILTIN_PINSRW128, IX86_BUILTIN_LOADAPD, IX86_BUILTIN_LOADSD,
1382 IX86_BUILTIN_STOREAPD, IX86_BUILTIN_STORESD, IX86_BUILTIN_STOREHPD,
1383 IX86_BUILTIN_STORELPD, IX86_BUILTIN_SETPD1, IX86_BUILTIN_SETPD,
1384 IX86_BUILTIN_CLRPD, IX86_BUILTIN_LOADPD1, IX86_BUILTIN_LOADRPD,
1385 IX86_BUILTIN_STOREPD1, IX86_BUILTIN_STORERPD, IX86_BUILTIN_LOADDQA,
1386 IX86_BUILTIN_STOREDQA, IX86_BUILTIN_CLRTI,
1387 IX86_BUILTIN_LOADDDUP): Remove.
1388 (IX86_BUILTIN_VEC_INIT_V2SI, IX86_BUILTIN_VEC_INIT_V4HI,
1389 IX86_BUILTIN_VEC_INIT_V8QI, IX86_BUILTIN_VEC_EXT_V2DF,
1390 IX86_BUILTIN_VEC_EXT_V2DI, IX86_BUILTIN_VEC_EXT_V4SF,
1391 IX86_BUILTIN_VEC_EXT_V8HI, IX86_BUILTIN_VEC_EXT_V4HI,
1392 IX86_BUILTIN_VEC_SET_V8HI, IX86_BUILTIN_VEC_SET_V4HI): New.
1393 (ix86_init_builtins): Make static.
1394 (ix86_init_mmx_sse_builtins): Update for changed builtins.
1395 (ix86_expand_binop_builtin): Only use ix86_fixup_binary_operands
1396 if all the modes match. Otherwise, fake it.
1397 (get_element_number, ix86_expand_vec_init_builtin,
1398 ix86_expand_vec_ext_builtin, ix86_expand_vec_set_builtin): New.
1399 (ix86_expand_builtin): Make static. Update for changed builtins.
1400 (ix86_expand_vector_move_misalign): Use sse2_loadlpd with zero
1401 operand instead of sse2_loadsd. Cast sse1 fallback to V4SFmode.
1402 (ix86_expand_vector_init_duplicate): New.
1403 (ix86_expand_vector_init_low_nonzero): New.
1404 (ix86_expand_vector_init_one_var, ix86_expand_vector_init_general):
1405 Split out from ix86_expand_vector_init; handle integer modes.
1406 (ix86_expand_vector_init): Use them.
1407 (ix86_expand_vector_set, ix86_expand_vector_extract): New.
1408 * config/i386/i386-protos.h: Update.
1409 * config/i386/predicates.md (reg_or_0_operand): New.
1410 * config/i386/mmx.md (mov<MMXMODEI>_internal): Add 'r' variants.
1411 (movv2sf_internal): Likewise. And a splitter to match them all.
1412 (vec_dupv2sf, mmx_concatv2sf, vec_setv2sf, vec_extractv2sf,
1413 vec_initv2sf, vec_dupv4hi, vec_dupv2si, mmx_concatv2si, vec_setv2si,
1414 vec_extractv2si, vec_initv2si, vec_setv4hi, vec_extractv4hi,
1415 vec_initv4hi, vec_setv8qi, vec_extractv8qi, vec_initv8qi): New.
1416 (mmx_pinsrw): Fix operand ordering.
1417 * config/i386/sse.md (movv4sf splitter): Use direct pattern,
1418 rather than sse_loadss expander.
1419 (movv2df splitter): Similarly.
1420 (sse_loadss, sse_loadlss): Remove.
1421 (vec_dupv4sf, sse_concatv2sf, sse_concatv4sf, vec_extractv4sf_0): New.
1422 (vec_setv4sf, vec_setv2df): Use ix86_expand_vector_set.
1423 (vec_extractv4sf, vec_extractv2df): Use ix86_expand_vector_extract.
1424 (sse3_movddup): Rename with '*'.
1425 (sse3_movddup splitter): Use gen_rtx_REG instead of gen_lowpart.
1426 (sse2_loadsd): Remove.
1427 (vec_dupv2df_sse3): Rename from sse3_loadddup.
1428 (vec_dupv2df, vec_concatv2df_sse3, vec_concatv2df): New.
1429 (sse2_pinsrw): Fix argument ordering.
1430 (sse2_loadld, sse2_loadq): Add sse1 alternatives.
1431 (sse2_stored): Remove 'r' destination.
1432 (vec_dupv4si, vec_dupv2di, sse2_concatv2si, sse1_concatv2si,
1433 vec_concatv4si_1, vec_concatv2di, vec_setv2di, vec_extractv2di,
1434 vec_initv2di, vec_setv4si, vec_extractv4si, vec_initv4si,
1435 vec_setv8hi, vec_extractv8hi, vec_initv8hi, vec_setv16qi,
1436 vec_extractv16qi, vec_initv16qi): New.
1438 * config/i386/emmintrin.h (__m128i, __m128d): Use typedef, not define.
1439 (_mm_set_sd, _mm_set1_pd, _mm_setzero_pd, _mm_set_epi64x,
1440 _mm_set_epi32, _mm_set_epi16, _mm_set_epi8, _mm_setzero_si128): Use
1442 (_mm_load_pd, _mm_store_pd): Use plain dereference.
1443 (_mm_load_si128, _mm_store_si128): Likewise.
1444 (_mm_load1_pd): Use _mm_set1_pd.
1445 (_mm_load_sd): Use _mm_set_sd.
1446 (_mm_store_sd, _mm_storeh_pd): Use __builtin_ia32_vec_ext_v2df.
1447 (_mm_store1_pd, _mm_storer_pd): Use _mm_store_pd.
1448 (_mm_set_epi64): Use _mm_set_epi64x.
1449 (_mm_set1_epi64x, _mm_set1_epi64, _mm_set1_epi32, _mm_set_epi16,
1450 _mm_set1_epi8, _mm_setr_epi64, _mm_setr_epi32, _mm_setr_epi16,
1451 _mm_setr_epi8): Use _mm_set_foo form.
1452 (_mm_loadl_epi64, _mm_movpi64_epi64, _mm_move_epi64): Use _mm_set_epi64.
1453 (_mm_storel_epi64, _mm_movepi64_pi64): Use __builtin_ia32_vec_ext_v2di.
1454 (_mm_extract_epi16): Use __builtin_ia32_vec_ext_v8hi.
1455 (_mm_insert_epi16): Use __builtin_ia32_vec_set_v8hi.
1456 * config/i386/mmintrin.h (_mm_setzero_si64): Use plain cast.
1457 (_mm_set_pi32): Use __builtin_ia32_vec_init_v2si.
1458 (_mm_set_pi16): Use __builtin_ia32_vec_init_v4hi.
1459 (_mm_set_pi8): Use __builtin_ia32_vec_init_v8qi.
1460 (_mm_set1_pi16, _mm_set1_pi8): Use _mm_set_piN variant.
1461 * config/i386/pmmintrin.h (_mm_loaddup_pd): Use _mm_load1_pd.
1462 (_mm_movedup_pd): Use _mm_shuffle_pd.
1463 * config/i386/xmmintrin.h (_mm_setzero_ps, _mm_set_ss,
1464 _mm_set1_ps, _mm_set_ps, _mm_setr_ps): Use constructor form.
1465 (_mm_cvtpi16_ps, _mm_cvtpu16_ps, _mm_cvtpi8_ps, _mm_cvtpu8_ps,
1466 _mm_cvtps_pi8, _mm_cvtpi32x2_ps): Avoid __builtin_ia32_mmx_zero;
1468 (_mm_load_ss, _mm_load1_ps): Use _mm_set* form.
1469 (_mm_load_ps, _mm_loadr_ps): Use raw dereference.
1470 (_mm_store_ss): Use __builtin_ia32_vec_ext_v4sf.
1471 (_mm_store_ps): Use raw dereference.
1472 (_mm_store1_ps): Use _mm_storeu_ps.
1473 (_mm_storer_ps): Use _mm_store_ps.
1474 (_mm_extract_pi16): Use __builtin_ia32_vec_ext_v4hi.
1475 (_mm_insert_pi16): Use __builtin_ia32_vec_set_v4hi.
1477 2005-01-11 Stan Shebs <shebs@apple.com>
1479 * config/rs6000/rs6000.c (machopic_output_stub): Issue
1480 ldu instead of lwzu if 64-bit -mdynamic-no-pic.
1482 2005-01-11 Andrew Pinski <pinskia@physics.uc.edu>
1485 * config/rs6000/rs6000.c (rs6000_special_round_type_align):
1486 Skip all DECLs except for FIELD_DECLs.
1488 2005-01-11 Andreas Krebbel <krebbel1@de.ibm.com>
1490 * config/s390/s390.c (override_options): Return error if
1491 -mbackchain, -mpacked-stack and -mhard-float are used together.
1492 (s390_va_start): Remove the backchain && packed-stack special case.
1493 (s390_gimplify_va_arg): Likewise.
1494 * doc/invoke.texi: Remove the ABI incompatibility note.
1496 2005-01-11 Andreas Krebbel <krebbel1@de.ibm.com>
1498 * config/s390/s390.c (struct s390_frame_layout): Remove
1500 (s390_frame_info, s390_emit_prologue): Replace occurrences of
1501 save_backchain_p with TARGET_BACKCHAIN.
1503 2005-01-11 Alan Modra <amodra@bigpond.net.au>
1506 * builtins.c (std_gimplify_va_arg_expr): Adjust alignment of *ap.
1507 * expr.h (struct locate_and_pad_arg_data): Add "boundary".
1508 * function.c (locate_and_pad_parm): Set new field.
1509 (assign_parm_find_stack_rtl): Use it instead of FUNCTION_ARG_BOUNDARY.
1510 Tweak where_pad test to include "none". Always set mem align for
1512 (assign_parm_adjust_stack_rtl): Discard stack_parm if alignment
1513 not sufficient for type.
1514 (assign_parm_setup_block): If stack_parm is zero on entry, always
1515 make a new stack local. Block move old stack parm if necessary
1516 to new aligned stack local.
1517 (assign_parm_setup_stack): Use a block move to handle
1518 potentially misaligned entry_parm.
1519 (assign_parms_unsplit_complex): Specify required alignment when
1520 creating stack local.
1521 * calls.c (compute_argument_addresses): Override alignment of stack
1522 arg calculated from its type with the alignment given by
1523 FUNCTION_ARG_BOUNDARY.
1524 (store_one_arg): Likewise.
1526 2005-01-11 Jan Beulich <jbeulich@novell.com>
1528 * config/ia64/ia64.md (zero_extendsidi2): Replace zxt4 by addp4.
1529 Change respective itanium_class attribute to ialu.
1530 (shladdp4_internal): New.
1531 * config/ia64/predicates.md (shladd_log2_operand): New.
1533 2005-01-11 Richard Henderson <rth@redhat.com>
1535 * expr.c (store_constructor): Use rtvec_alloc instead of
1536 alloca+gen_rtvec_v, and an incorrect number passed to alloca.
1538 2005-01-11 Kazu Hirata <kazu@cs.umass.edu>
1540 * config/alpha/alpha.c, config/i386/mmx.md: Fix comment typos.
1542 2005-01-11 Alan Modra <amodra@bigpond.net.au>
1544 * varasm.c (default_section_type_flags_1): Don't set SECTION_SMALL.
1545 * config/ia64/ia64.c (TARGET_SECTION_TYPE_FLAGS): Define.
1546 (TARGET_RWRELOC): Define.
1547 (ia64_rwreloc_section_type_flags): Delete.
1548 (ia64_section_type_flags): New function.
1549 * config/ia64/hpux.h (TARGET_SECTION_TYPE_FLAGS): Don't define.
1550 (TARGET_RWRELOC): Define.
1552 2005-01-10 David Mosberger <davidm@hpl.hp.com>
1555 * config/ia64/ia64.c (process_set): For alloc insn, only call
1556 process_epilogue is !frame_pointer_needed.
1558 2005-01-10 Roger Sayle <roger@eyesopen.com>
1561 * c-common.c (c_common_truthvalue_conversion): TRUTH_NOT_EXPR is a
1562 unary operator and can't be treated as a binary/comparison operator.
1564 2005-01-10 Richard Henderson <rth@redhat.com>
1566 * config/i386/i386.c (ix86_function_value): Use type_natural_mode.
1567 (ix86_return_in_memory): Likewise.
1568 (function_arg_advance): Likewise. Mirror structure in function_arg
1569 for choosing register to advance.
1571 2005-01-10 Kazu Hirata <kazu@cs.umass.edu>
1573 * tree-vectorizer.c, tree.def: Fix comment typos.
1575 2005-01-10 Mark Dettinger <dettinge@de.ibm.com>
1577 * config/s390/s390.c (struct processor_costs): 4 new fields:
1578 ddbr, ddr, debr, der.
1579 (s390_rtx_costs): More precise handling of divide instructions.
1581 2005-01-10 David Edelsohn <edelsohn@gnu.org>
1583 * config/rs6000/t-aix43 (BOOT_LDFLAGS): Define.
1585 2005-01-10 Jan Beulich <jbeulich@novell.com>
1587 * config/ia64/ia64.c (ia64_in_small_data_p): Also handle the section
1588 names resulting from -ffunction-sections/-fdata-sections and linkonce
1590 * varasm.c (default_section_type_flags_1): Also set SECTION_SMALL
1591 based on the section name. Rearrange the section name comparison logic
1592 slightly so that each section name is compared against at most once.
1594 2005-01-10 Ben Elliston <bje@au.ibm.com>
1596 * doc/invoke.texi (Code Gen Options): Add PowerPC to the list of
1597 targets for which -fPIC reduces limitations on the GOT size.
1599 2005-01-09 Falk Hueffner <falk@debian.org>
1601 * fold-const.c (fold): Also handle EXACT_DIV_EXPR when folding
1604 2005-01-09 David Edelsohn <edelsohn@gnu.org>
1607 * collect2.c (main): Set aixrtl_flag for -brtl option.
1608 (resolve_lib_name): Search for .so file extension before .a
1611 2005-01-09 Dorit Naishlos <dorit@il.ibm.com>
1613 * tree-vectorizer.c (vect_enhance_data_refs_alignment): Add dump prints.
1614 (vect_analyze_data_refs_alignment): Add dump prints.
1616 2005-01-09 Ira Rosen <irar@il.ibm.com>
1618 * tree-vectorizer.c (vect_analyze_offset_expr): Use
1619 expr_invariant_in_loop_p.
1620 Initialize outputs first thing in the function.
1621 (vect_update_ivs_after_vectorizer): Call initial_condition_in_loop_num.
1622 (vect_is_simple_iv_evolution): Call initial_condition_in_loop_num.
1623 (vect_analyze_pointer_ref_access): Check that the initial condition of
1624 the access function is loop invariant.
1626 2005-01-09 Richard Henderson <rth@redhat.com>
1628 * config/i386/i386.c (bdesc_2arg): Update names for mmx_ prefixes.
1629 (ix86_expand_builtin): Likewise. Frob MASKMOVQ wrt the input mem
1630 just like MASKMOVDQU. Return plain zero for MMX_ZERO.
1631 * config/i386/i386.md (MMXMODEI, mov<MMXMODEI>,
1632 mov<MMXMODEI>_internal_rex64, mov<MMXMODEI>_internal, movv2sf,
1633 movv2sf_internal_rex64, movv2sf_internal, MMXMODE,
1634 movmisalign<MMXMODE>, mmx_pmovmskb, mmx_maskmovq, mmx_maskmovq_rex,
1635 sse_movntdi, addv8qi3, addv4hi3, addv2si3, mmx_adddi3, ssaddv8qi3,
1636 ssaddv4hi3, usaddv8qi3, usaddv4hi3, subv8qi3, subv4hi3, subv2si3,
1637 mmx_subdi3, sssubv8qi3, sssubv4hi3, ussubv8qi3, ussubv4hi3,
1638 mulv4hi3, smulv4hi3_highpart, umulv4hi3_highpart, mmx_pmaddwd,
1639 sse2_umulsidi3, mmx_iordi3, mmx_xordi3, mmx_anddi3, mmx_nanddi3,
1640 mmx_uavgv8qi3, mmx_uavgv4hi3, mmx_psadbw, mmx_pinsrw, mmx_pinsrw,
1641 mmx_pextrw, mmx_pshufw, eqv8qi3, eqv4hi3, eqv2si3, gtv8qi3, gtv4hi3,
1642 gtv2si3, umaxv8qi3, smaxv4hi3, uminv8qi3, sminv4hi3, ashrv4hi3,
1643 ashrv2si3, lshrv4hi3, lshrv2si3, mmx_lshrdi3, ashlv4hi3, ashlv2si3,
1644 mmx_ashldi3, mmx_packsswb, mmx_packssdw, mmx_packuswb, mmx_punpckhbw,
1645 mmx_punpckhwd, mmx_punpckhdq, mmx_punpcklbw, mmx_punpcklwd,
1646 mmx_punpckldq, emms, addv2sf3, subv2sf3, subrv2sf3, gtv2sf3, gev2sf3,
1647 eqv2sf3, pfmaxv2sf3, pfminv2sf3, mulv2sf3, femms, pf2id, pf2iw,
1648 pfacc, pfnacc, pfpnacc, pi2fw, floatv2si2, pfrcpv2sf2, pfrcpit1v2sf3,
1649 pfrcpit2v2sf3, pfrsqrtv2sf2, pfrsqit1v2sf3, pmulhrwv4hi3, pswapdv2si2,
1650 pswapdv2sf2): Move to mmx.md; rename as necessary with leading
1652 (mmx_clrdi, pavgusb): Remove.
1653 (ldmxcsr, stmxcsr, sfence, sfence_insn): Move to sse.md; rename
1654 with leading sse_ prefix.
1655 * config/i386/sse.md: Receive them.
1656 * config/i386/mmx.md: New file.
1657 (MMXMODE12, MMXMODE24, mmxvecsize): New.
1658 (subrv2sf3): Turn into expander for normal subtraction.
1659 (mmx_addv2sf3, mmx_mulv2sf3, mmx_smaxv2sf3, mmx_sminv2sf3,
1660 mmx_eqv2sf3, mmx_mulv4hi3, mmx_smulv4hi3_highpart,
1661 mmx_umulv4hi3_highpart, mmx_pmaddwd, mmx_pmulhrwv4hi3, sse2_umulsidi3,
1662 mmx_umaxv8qi3, mmx_smaxv4hi3, mmx_uminv8qi3, mmx_sminv4hi3): Mark
1663 commutative; use ix86_binary_operator_ok.
1664 (mmx_add<MMXMODEI>3, mmx_ssadd<MMXMODE12>3, mmx_usadd<MMXMODE12>3,
1665 mmx_sub<MMXMODEI>3, mmx_sssub<MMXMODE12>3, mmx_ussub<MMXMODE12>3
1666 mmx_ashr<MMXMODE24>3, mmx_lshr<MMXMODE23>3, mmx_ashl<MMXMODE24>3
1667 mmx_eq<MMXMODEI>3, mmx_gt<MMXMODEI>3, mmx_and<MMXMODEI>3,
1668 mmx_nand<MMXMODEI>3, mmx_ior<MMXMODEI>3, mmx_xor<MMXMODEI>3):
1669 Macroize from existing patterns; use ix86_binary_operator_ok.
1670 (mmx_packsswb, mmx_packssdw, mmx_packuswb): Add memory alternative.
1671 (mmx_punpckhbw, mmx_punpcklbw, mmx_punpckhwd, mmx_punpcklwd,
1672 mmx_punpckhdq, mmx_punpckhdq, mmx_punpckldq): Likewise. Model
1673 with vec_select+vec_concat.
1674 (mmx_pshufw, mmx_pshufw_1): Likewise.
1675 (mmx_uavgv8qi3): Merge pavgusb. Model correcty.
1676 (mmx_uavgv4hi3): Model correctly.
1677 * config/i386/mmintrin.h (_mm_and_si64, _mm_andnot_si64, _mm_or_si64,
1678 _mm_xor_si64): Remove casts.
1680 2005-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1682 PR tree-optimization/19224
1683 * tree-scalar-evolution.c (get_instantiated_value,
1684 set_instantiated_value): New functions.
1685 (instantiate_parameters_1): Cache the results.
1686 (instantiate_parameters, resolve_mixers): Initialize and free
1689 2005-01-08 David Edelsohn <edelsohn@gnu.org>
1691 * config/i386/i386.md (addhi_4): Correct reference in comment.
1694 2005-01-08 Richard Henderson <rth@redhat.com>
1696 * config/i386/emmintrin.h (_mm_cvtsi128_si32): Move earlier.
1697 (_mm_cvtsi128_si64x): Likewise.
1698 (_mm_srl_epi64, _mm_srl_epi32, _mm_srl_epi16, _mm_sra_epi32,
1699 _mm_sra_epi16, _mm_sll_epi64, _mm_sll_epi32, _mm_sll_epi16): Use
1700 the _mm_{srl,sll}i_foo counterpart, and _mm_cvtsi128_si32.
1701 * config/i386/i386-modes.def: Add V16HI, V32QI, V4DF, V8SF.
1702 * config/i386/i386-protos.h: Update.
1703 * config/i386/i386.c (print_operand): Add 'H'.
1704 (ix86_fixup_binary_operands): Split out from ...
1705 (ix86_expand_binary_operator): ... here.
1706 (ix86_fixup_binary_operands_no_copy): New.
1707 (ix86_expand_fp_absneg_operator): Handle vector mode results.
1708 (bdesc_2arg): Update names for sse{,2,3}_ prefixes.
1709 (ix86_init_mmx_sse_builtins): Remove *maskncmp* special cases.
1710 (safe_vector_operand): Use CONST0_RTX.
1711 (ix86_expand_binop_builtin): Use ix86_fixup_binary_operands.
1712 (ix86_expand_builtin): Merge CODE_FOR_sse2_maskmovdqu_rex64 and
1713 CODE_FOR_sse2_maskmovdqu. Special case SSE version of MASKMOVDQU
1714 expansion. Update names for sse{,2,3}_ prefixes. Remove *maskncmp*
1716 * config/i386/i386.h (IX86_BUILTIN_CMPNGTSS): New.
1717 (IX86_BUILTIN_CMPNGESS): New.
1718 * config/i386/i386.md (UNSPEC_FIX_NOTRUNC): New.
1719 (attr type): Add sselog1.
1720 (attr unit, attr memory): Handle it.
1721 (movti, movti_internal, movti_rex64): Move near other integer moves.
1722 (movtf, movtf_internal): Move near other fp moves.
1723 (SSEMODE, SSEMODEI, vec_setv2df, vec_extractv2df, vec_initv2df,
1724 vec_setv4sf, vec_extractv4sf, vec_initv4sf, movv4sf, movv4sf_internal,
1725 movv2df, movv2df_internal, mov<SSEMODEI>, mov<SSEMODEI>_internal,
1726 movmisalign<SSEMODE>, sse_movups_1, sse_movmskps, sse_movntv4sf,
1727 sse_movhlps, sse_movlhps, sse_storehps, sse_loadhps, sse_storelps,
1728 sse_loadlps, sse_loadss, sse_loadss_1, sse_movss, sse_storess,
1729 sse_shufps, addv4sf3, vmaddv4sf3, subv4sf3, vmsubv4sf3, negv4sf2,
1730 mulv4sf3, vmmulv4sf3, divv4sf3, vmdivv4sf3, rcpv4sf2, vmrcpv4sf2,
1731 rsqrtv4sf2, vmrsqrtv4sf2, sqrtv4sf2, vmsqrtv4sf2, sse_andv4sf3,
1732 sse_nandv4sf3, sse_iorv4sf3, sse_xorv4sf3, sse2_andv2df3,
1733 sse2_nandv2df3, sse2_iorv2df3, sse2_xorv2df3, sse2_andv2di3,
1734 sse2_nandv2di3, sse2_iorv2di3, sse2_xorv2di3, maskcmpv4sf3,
1735 vmmaskcmpv4sf3, sse_comi, sse_ucomi, sse_unpckhps, sse_unpcklps,
1736 smaxv4sf3, vmsmaxv4sf3, sminv4sf3, vmsminv4sf3, cvtpi2ps, cvtps2pi,
1737 cvttps2pi, cvtsi2ss, cvtsi2ssq, cvtss2si, cvtss2siq, cvttss2si,
1738 cvttss2siq, addv2df3, vmaddv2df3, subv2df3, vmsubv2df3, mulv2df3,
1739 vmmulv2df3, divv2df3, vmdivv2df3, smaxv2df3, vmsmaxv2df3, sminv2df3,
1740 vmsminv2df3, sqrtv2df2, vmsqrtv2df2, maskcmpv2df3, vmmaskcmpv2df3,
1741 sse2_comi, sse2_ucomi, sse2_movmskpd, sse2_pmovmskb, sse2_maskmovdqu,
1742 sse2_maskmovdqu_rex64, sse2_movntv2df, sse2_movntv2di, sse2_movntsi,
1743 cvtdq2ps, cvtps2dq, cvttps2dq, cvtdq2pd, cvtpd2dq, cvttpd2dq,
1744 cvtpd2pi, cvttpd2pi, cvtpi2pd, cvtsd2si, cvtsd2siq, cvttsd2si,
1745 cvttsd2siq, cvtsi2sd, cvtsi2sdq, cvtsd2ss, cvtss2sd, cvtpd2ps,
1746 cvtps2pd, addv16qi3, addv8hi3, addv4si3, addv2di3, ssaddv16qi3,
1747 ssaddv8hi3, usaddv16qi3, usaddv8hi3, subv16qi3, subv8hi3, subv4si3,
1748 subv2di3, sssubv16qi3, sssubv8hi3, ussubv16qi3, ussubv8hi3, mulv8hi3,
1749 smulv8hi3_highpart, umulv8hi3_highpart, sse2_umulsidi3,
1750 sse2_umulv2siv2di3, sse2_pmaddwd, sse2_uavgv16qi3, sse2_uavgv8hi3,
1751 sse2_psadbw, sse2_pinsrw, sse2_pextrw, sse2_pshufd, sse2_pshuflw,
1752 sse2_pshufhw, eqv16qi3, eqv8hi3, eqv4si3, gtv16qi3, gtv8hi3,
1753 gtv4si3, umaxv16qi3, smaxv8hi3, uminv16qi3, sminv8hi3, ashrv8hi3,
1754 ashrv4si3, lshrv8hi3, lshrv4si3, lshrv2di3, ashlv8hi3, ashlv4si3,
1755 ashlv2di3, sse2_ashlti3, sse2_lshrti3, sse2_unpckhpd, sse2_unpcklpd,
1756 sse2_packsswb, sse2_packssdw, sse2_packuswb, sse2_punpckhbw,
1757 sse2_punpckhwd, sse2_punpckhdq, sse2_punpcklbw, sse2_punpcklwd,
1758 sse2_punpckldq, sse2_punpcklqdq, sse2_punpckhqdq, sse2_movupd,
1759 sse2_movdqu, sse2_movdq2q, sse2_movdq2q_rex64, sse2_movq2dq,
1760 sse2_movq2dq_rex64, sse2_loadd, sse2_stored, sse2_storehpd,
1761 sse2_loadhpd, sse2_storelpd, sse2_loadlpd, sse2_movsd, sse2_loadsd,
1762 sse2_loadsd_1, sse2_storesd, sse2_shufpd, sse2_clflush, sse2_mfence,
1763 mfence_insn, sse2_lfence, lfence_insn, mwait, monitor, addsubv4sf3,
1764 addsubv2df3, haddv4sf3, haddv2df3, hsubv4sf3, hsubv2df3, movshdup,
1765 movsldup, lddqu, loadddup, movddup): Move to sse.md. Any with
1766 non-optabs meanings renamed with an "sse{,2,3}_" prefix at the
1768 (SSEPUSH, push<SSEPUSH>): Remove.
1769 (MMXPUSH, push<MMXPUSH>): Remove.
1770 (sse_movaps, sse_movaps_1, sse_movups): Remove.
1771 (sse2_movapd, sse2_movdqa, sse2_movq): Remove.
1772 (sse2_andti3, sse2_nandti3, sse2_iorti3, sse2_xorti3): Remove.
1773 (sse_clrv4sf, sse_clrv2df, sse2_clrti): Remove.
1774 (maskncmpv4sf3, vmmaskncmpv4sf3): Remove.
1775 (maskncmpv2df3, vmmaskncmpv2df3): Remove.
1776 (ashrv8hi3_ti, ashrv4si3_ti, lshrv8hi3_ti, lshrv4si3_ti): Remove.
1777 (lshrv2di3_ti, ashlv8hi3_ti, ashlv4si3_ti, ashlv2di3_ti): Remove.
1778 * config/i386/athlon.md (athlon_sselog_load): Handle sselog1.
1779 (athlon_sselog_load_k8, athlon_sselog, athlon_sselog_k8): Likewise.
1780 * config/i386/ppro.md (ppro_sse_div_V4SF_load): Fix memory attr.
1781 (ppro_sse_log_V4SF_load): Similarly. Handle sselog1.
1782 (ppro_sse_log_V4SF): Handle sselog1.
1783 * config/i386/predicates.md (const_0_to_1_operand): New.
1784 (const_0_to_255_mul_8_operand): New.
1785 (const_1_to_31_operand): Rename from const_int_1_31_operand.
1786 (const_2_to_3_operand, const_4_to_7_operand): New.
1787 * config/i386/sse.md: New file.
1788 (SSEMODE12, SSEMODE24, SSEMODE124, SSEMODE248, ssevecsize): New.
1789 (sse_movups): Rename from sse_movups_1.
1790 (sse_loadlss): Rename from sse_loadss_1.
1791 (andv4sf3, iorv4sf3, xorv4sf3, andv2df3): Remove the sse prefix
1793 (negv4sf2): Use ix86_expand_fp_absneg_operator.
1794 (absv4sf2, negv2df, absv2df): New.
1795 (addv4sf3): Add expander to call ix86_fixup_binary_operands_no_copy.
1796 (subv4sf3, mulv4sf3, divv4sf3, smaxv4sf3, sminv4sf3, andv4sf3,
1797 iorv4sf3, xorv4sf3, addv2df3, subv2df3, mulv2df3, divv2df3,
1798 smaxv2df3, sminv2df3, andv2df3, iorv2df3, xorv2df3, mulv8hi3,
1799 umaxv16qi3, smaxv8hi3, uminv16qi3, sminv8hi3): Likewise.
1800 (sse3_addsubv4sf3): Model correctly.
1801 sse3_haddv4sf3, sse3_hsubv4sf3, sse3_addsubv2df3, sse3_haddv2df3,
1802 sse3_hsubv2df3, sse2_ashlti3, sse2_lshrti3): Likewise.
1803 (sse_movhlps): Model with vec_select+vec_concat.
1804 (sse_movlhps, sse_unpckhps, sse_unpcklps, sse3_movshdup,
1805 sse3_movsldup, sse_shufps, sse_shufps_1, sse2_unpckhpd, sse3_movddup,
1806 sse2_unpcklpd, sse2_shufpd, sse2_shufpd_1, sse2_punpckhbw,
1807 sse2_punpcklbw, sse2_punpckhwd, sse2_punpcklwd, sse2_punpckhdq,
1808 sse2_punpckldq, sse2_punpckhqdq, sse2_punpcklqdq, sse2_pshufd,
1809 sse2_pshufd_1, sse2_pshuflw, sse2_pshuflw_1, sse2_pshufhw,
1810 sse2_pshufhw_1): Likewise.
1811 (neg<SSEMODEI>2, one_cmpl<SSEMODEI>2): New.
1812 (add<SSEMODEI>3, sse2_ssadd<SSEMODE12>3, sse2_usadd<SSEMODE12>3,
1813 sub<SSEMODEI>3, sse2_sssub<SSEMODE12>3, sse2_ussub<SSEMODE12>3,
1814 ashr<SSEMODE24>3, lshr<SSEMODE248>3, sse2_eq<SSEMODE124>3,
1815 sse2_gt<SSEMODDE124>3, and<SSEMODEI>3, sse_nand<SSEMODEI>3,
1816 ior<SSEMODEI>3, xor<SSEMODEI>3): Macroize from existing patterns.
1817 (addv4sf3, sse_vmaddv4sf3, mulv4sf3, sse_vmmulv4sf3, smaxv4sf3,
1818 sse_vmsmaxv4sf3, sminv4sf3, sse_vmsminv4sf3, addv2df3, sse2_vmaddv2df3,
1819 mulv2df3, sse2_vmmulv2df3, smaxv2df3, sse2_vmsmaxv2df3, sminv2df3,
1820 sse2_vmsminv2df3, umaxv16qi3, smaxv8hi3, uminv16qi3
1821 sminv8hi3): Mark commutative
1822 operands. Use ix86_binary_operator_ok.
1823 (sse_unpckhps, sse_unpcklps, sse2_packsswb, sse2_packssdw,
1824 sse2_packuswb, sse2_punpckhbw, sse2_punpcklbw, sse2_punpckhwd,
1825 sse2_punpcklwd, sse2_punpckhdq, sse2_punpckldq, sse2_punpckhqdq,
1826 sse2_punpcklqdq): Allow operand2 in memory.
1827 (sse_movhlps, sse_movlhps, sse2_unpckhpd, sse2_unpcklpd
1828 sse2_movsd): Add memory alternatives.
1829 (sse_storelps): Turn expander into an insn; split after reload.
1830 (sse_storess, sse2_loadhpd, sse2_loadlpd): Add non-xmm inputs.
1831 (sse2_storehpd, sse2_storelpd): Add non-xmm outputs.
1833 2005-01-08 Eric Botcazou <ebotcazou@libertysurf.fr>
1835 * configure.ac (DWARF-2 debug_line): Use objdump.
1836 * configure: Regenerate.
1838 2005-01-08 Jeff Law <law@redhat.com>
1839 Diego Novillo <dnovillo@redhat.com>
1841 PR tree-optimization/18241
1842 * tree-nrv.c (tree_nrv): Ignore volatile return values.
1843 * tree-ssa-dse.c (dse_optimize_stmt): Do not optimize
1844 statements with volatile operands.
1845 * tree-ssa-operands.c (add_stmt_operand): Do add volatile
1846 operands after marking a statement with has_volatile_ops.
1848 2005-01-08 Roger Sayle <roger@eyesopen.com>
1850 * tree.c (int_fits_type_p): Always honor integer constant
1851 TYPE_MIN_VALUE and TYPE_MAX_VALUE if they exist.
1853 2005-01-08 Roger Sayle <roger@eyesopen.com>
1855 * ifcvt.c (find_if_case_1): Reinstate 2005-01-04 change, now that
1856 the latent bug in rtl_delete_block has been resolved.
1858 2005-01-08 Richard Sandiford <rsandifo@redhat.com>
1860 * config/mips/t-iris6 (MULTILIB_DIRNAMES): Use -mabi argument values.
1861 (MULTILIB_OSDIRNAMES): Use the standard lib, lib32 and lib64.
1863 2005-01-08 Richard Sandiford <rsandifo@redhat.com>
1865 * config/mips/t-slibgcc-irix (SHLIB_LINK): Install a copy of the
1866 library as @multilib_dir@/$(SHLIB_SONAME).
1868 2005-01-07 Eric Botcazou <ebotcazou@libertysurf.fr>
1870 * configure.ac (HAVE_AS_OFFSETABLE_LO10): Fix typo.
1871 * configure: Regenerate.
1873 2005-01-07 Jakub Jelinek <jakub@redhat.com>
1875 * c-common.c (handle_mode_attribute): For ENUMERAL_TYPE, also copy
1878 2005-01-07 David Edelsohn <edelsohn@gnu.org>
1881 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address):
1882 Convert non-word aligned offset address using ld/std into
1885 2005-01-07 Richard Henderson <rth@redhat.com>
1887 * config/i386/i386.md (sse_loadhps splitter): Fix operand number typo.
1889 2005-01-07 Richard SAndiford <rsandifo@redhat.com>
1891 * tree.h (IS_EXPR_CODE_CLASS): Use a straight-forward range check.
1893 2005-01-07 Jakub Jelinek <jakub@redhat.com>
1895 PR tree-optimization/19283
1896 * fold-const.c (fold_widened_comparison): Return NULL if shorter_type
1897 is not shorter than the original type.
1899 PR rtl-optimization/19012
1900 * config/i386/i386.md (addqi_1_slp): Set memory attribute.
1902 PR rtl-optimization/18861
1903 * cfgbuild.c (BLOCK_USED_BY_TABLEJUMP): Define.
1904 (FULL_STATE): Define.
1905 (mark_tablejump_edge): New function.
1906 (purge_dead_tablejump_edges): New function.
1907 (find_bb_boundaries): Use it.
1909 PR tree-optimization/18828
1910 * builtins.c (expand_builtin_next_arg): Remove argument and all
1911 the argument checking.
1912 (expand_builtin): Adjust caller.
1913 (expand_builtin_va_start): Likewise. Remove error for too many
1915 (fold_builtin_next_arg): Issue error for too many arguments.
1916 After checking arguments, replace them with magic arguments that
1917 prevent further checking of the args.
1919 2005-01-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1921 * pa64-hpux.h (STARTFILE_SPEC): Fix typo in spec.
1923 2005-01-06 Roger Sayle <roger@eyesopen.com>
1926 * gcc.c (input_suffix_matches): Tweak the semantics of %{.s:...}
1927 and %{.S:...} (and their negative variants) to test whether the
1928 input file is assembler or pre-processed-assembler independent of
1929 the actual filename extension.
1931 2005-01-06 Roger Sayle <roger@eyesopen.com>
1933 * simplify-rtx.c (simplify_subreg): Simplify truncations of shifts
1934 of sign or zero extended values.
1936 2005-01-06 Geoffrey Keating <geoffk@apple.com>
1938 * c-cppbuiltin.c (builtin_define_float_constants): Set __*_EPSILON__
1939 for IBM long double format correctly.
1941 2005-01-06 Daniel Berlin <dberlin@dberlin.org>
1943 Fix PR tree-optimization/18792
1945 * tree-data-ref.c (build_classic_dist_vector): Change first_loop
1946 to first_loop_depth, and use loop depth instead of loop number.
1947 (build_classic_dir_vector): Ditto.
1948 (compute_data_dependences_for_loop): Use depth, not loop number.
1949 * tree-loop-linear.c (try_interchange_loops): Use loop depth, not loop
1950 number. Pass in loops, instead of loop numbers.
1951 (gather_interchange_stats): Ditto.
1952 (linear_transform_loops): Ditto.
1954 2005-01-06 Richard Sandiford <rsandifo@redhat.com>
1957 * loop.c (get_monotonic_increment, biased_biv_fits_mode_p,
1958 biv_fits_mode_p, extension_within_bounds_p): New functions.
1959 (check_ext_dependent_givs): Use them.
1961 2005-01-06 Roger Sayle <roger@eyesopen.com>
1963 * cfgrtl.c (rtl_delete_block): A basic block may be followed by
1964 more than one barrier, in which case we should delete them all.
1966 2005-01-06 Tobias Schlueter <tobias.schlueter@physik.uni-muenchen.de>
1968 * gcc.c (process_command): Change year in 'gcc --version' to 2005.
1970 2005-01-05 Daniel Berlin <dberlin@dberlin.org>
1972 Fix PR middle-end/19286
1974 * dwarf2out.c (gen_subprogram_die): If we've already tried to
1975 output this subprogram, simply ignore this attempt to do it again.
1976 (add_abstract_origin_attribute): Don't abort trying to add the abstract
1977 origin attribute if it's not possible.
1978 (gen_block_die): Don't ignore subblocks of "unused" blocks.
1979 (decls_for_scope): Ditto.
1980 * gimple-low.c (mark_blocks_with_used_subblocks): Remove.
1981 (mark_used_blocks): Don't call mark_blocks_with_used_subblocks.
1983 2005-01-05 Richard Henderson <rth@redhat.com>
1986 * config/i386/i386.c (BUILTIN_DESC_SWAP_OPERANDS): New.
1987 (bdesc_2arg): Use it.
1988 (ix86_expand_binop_builtin): Force operands into registers
1990 (ix86_expand_unop_builtin, ix86_expand_unop1_builtin,
1991 ix86_expand_sse_compare, ix86_expand_sse_comi,
1992 ix86_expand_builtin): Likewise.
1994 2005-01-05 Richard Henderson <rth@redhat.com>
1996 * config/ia64/ia64.c (rtx_needs_barrier): Handle CONST_VECTOR
1998 * config/ia64/vect.md (mulv8qi3): Re-implement with mix_[rl].
1999 (mulv4hi3): Set itanium_class mmmul.
2000 (fpack_sfxf, fpack_xfsf, fpack_xfxf): Remove.
2001 (fpack): Rename from fpack_sfsf.
2003 2005-01-05 Richard Henderson <rth@redhat.com>
2006 * reload1.c (do_input_reload): Restrict the optimization deleteing
2007 a previous output reload to RELOAD_FOR_INPUT.
2009 2005-01-05 Steven Bosscher <stevenb@suse.de>
2011 * combine.c (expand_compound_operation) <ZERO_EXTRACT>: Add
2012 comment that we fall through after case.
2013 (mark_used_regs_combine): Don't expect a SIGN_EXTRACT in a SET_DEST.
2014 (distribute_links): Likewise.
2015 * cse.c (cse_insn): Likewise.
2016 * cselib.c (cselib_invalidate_mem): Likewise.
2017 * df.c: Update comments at the top of the file.
2018 (read_modify_subreg_p): Update comments here too.
2019 (df_def_record_1): Don't expect a SIGN_EXTRACT in a SET_DEST.
2020 * flow.c (mark_set_1): Likewise.
2021 (mark_used_regs): Likewise.
2022 * gcse.c (mems_conflict_for_gcse_p): Likewise.
2023 (canon_list_insert): Likewise.
2024 (mark_set): Likewise.
2025 (try_replace_reg): Likewise.
2026 (store_killed_in_insn): Likewise.
2027 * loop.c (count_one_set): Likewise.
2028 (basic_induction_var): Likewise.
2029 * postreload-gcse.c (find_mem_conflicts): Likewise.
2030 * postreload.c (reload_combine_note_store): Likewise.
2031 (move2add_note_store): Likewise.
2032 * reload.c (find_equiv_reg): Likewise.
2033 (mark_referenced_resources): Likewise.
2034 * rtlanal.c (set_noop_p): Likewise.
2035 (note_stores): Likewise.
2036 (note_uses): Likewise.
2037 * sched-deps.c (sched_analyze_1): Likewise.
2038 * sched-rgn.c (check_live_1): Likewise.
2039 (update_live_1): Likewise.
2040 * config/i860/i860.c: Likewise.
2042 * rtl.dec (SIGN_EXTRACT): Document that this cannot appear as
2044 (ZERO_EXTRACT): Mention that this one can be an lvalue.
2046 * doc/rtl.texi: Update documentation for bit-fields and SET.
2048 2005-01-05 Roger Sayle <roger@eyesopen.com>
2050 * ifcvt.c (find_if_case_1): Revert 2005-01-04 change.
2052 2005-01-05 Jan Hubicka <jh@suse.cz>
2053 Richard Henderson <rth@redhat.com>
2056 * config/i386/i386.c (ix86_expand_move): Handle tls symbols
2059 2005-01-05 Richard Henderson <rth@redhat.com>
2062 * config/i386/i386.md (sse_movhps, sse_movlps): Remove.
2063 (sse_shufps): Change operand 3 to const_int_operand.
2064 (sse2_storelps): Fix typo in template.
2065 (sse_storehps, sse_loadhps, sse_storelps, sse_loadlps): New.
2066 * config/i386/i386.c (ix86_expand_vector_move_misalign): Use them.
2067 (ix86_expand_builtin): Likewise.
2069 2005-01-05 Julian Brown <julian@codesourcery.com>
2071 * config/arm/arm.c (arm_return_in_memory): Treat complex types
2072 as aggregates for AAPCS ABIs.
2074 2005-01-05 Stan Shebs <shebs@apple.com>
2076 * unwind-dw2-fde-darwin.c (examine_objects): Use 64-bit
2077 Mach-O getters if ppc64.
2078 * config/darwin.c (darwin_asm_output_dwarf_delta): Obey
2080 * config/darwin.h (ASM_PREFERRED_EH_DATA_FORMAT): Use
2081 signed four-byte field for global code case.
2082 (STARTFILE_SPEC): Avoid crt2.o for 64-bit compilation.
2084 2005-01-05 Roger Sayle <roger@eyesopen.com>
2087 * c-common.c: Include real.h.
2088 (c_common_truthvalue_conversion): Avoid destructively modifying expr.
2089 Correctly handle TREE_CONSTANT_OVERFLOW for INTEGER_CST.
2090 Correctly handle TREE_CONSTANT_OVERFLOW and NaNs for REAL_CST.
2091 * Makefile.in (c-common.o): Update dependencies.
2093 2005-01-05 Joseph S. Myers <joseph@codesourcery.com>
2095 * c-parse.in (asm_string): Add trailing semicolon.
2097 2005-01-05 Joseph S. Myers <joseph@codesourcery.com>
2099 * c-parse.in (asm_string): New. Don't allow wide strings in
2101 (simple_asm_expr, asm_argument, asm_operand, asm_clobbers): Use
2102 asm_string instead of STRING.
2104 2005-01-05 Joseph S. Myers <joseph@codesourcery.com>
2106 * c-typeck.c (constructor_no_implicit): Remove.
2107 (set_designator, process_init_element): Don't check
2108 constructor_no_implicit.
2110 2005-01-05 J"orn Rennecke <joern.rennecke@st.com>
2111 Kaz Kojima <kkojima@gcc.gnu.org>
2114 * lcm.c (create_pre_exit): New.
2115 (optimize_mode_switching): In MODE_ENTRY / MODE_EXIT case, set
2116 ENTRY_EXIT_EXTRA to 3. Use create_pre_exit.
2118 2004-01-05 Richard Earnshaw <rearnsha@arm.com>
2120 * arm.h (TARGET_OPTIONS): Correctly record -mhard-float and
2121 -msoft-float in target_float_switch.
2122 * arm.c (arm_override_options): Fix processing of target_float_switch.
2124 2004-01-05 Richard Earnshaw <rearnsha@arm.com>
2126 * arm/vfp.md (arm_movsi_vfp): Hide VFP register classes from register
2129 2004-01-05 Uros Bizjak <uros@kss-loka.si>
2131 * doc/invoke.texi (Intel 386 and AMD x86-64 Options):
2132 Replace i387 with 'i386 compiler' in -mfpmath=sse option.
2134 2005-01-04 Roger Sayle <roger@eyesopen.com>
2136 * ifcvt.c (find_if_case_1): Avoid creating an empty forwarder block,
2137 if deleting the then-block allows the test-block to fallthru to the
2140 2005-01-04 Andrew Pinski <pinskia@physics.uc.edu>
2143 * c-decl.c (diagnose_mismatched_decls): Accept "extern inline" declared
2144 after the full declaration if the are in two different TUs.
2146 2005-01-04 Richard Henderson <rth@redhat.com>
2149 * tree-sra.c (generate_one_element_init): Just
2150 call gimplify_and_add.
2151 (generate_element_init): Record the
2152 new referenced variables and mark them for renaming
2153 and split out to ...
2154 (generate_element_init_1): This.
2155 (scalarize_init): Don't call push_gimplify_context/
2156 pop_gimplify_context.
2158 2005-01-04 Geoffrey Keating <geoffk@apple.com>
2160 * toplev.c (get_src_pwd): Handle failure of getpwd().
2162 2005-01-04 Roger Sayle <roger@eyesopen.com>
2164 * fold-const.c (fold_single_bit_test): Delete unreachable handling
2166 (fold): Don't call fold_single_bit_test with a TRUTH_NOT_EXPR, as
2167 all the cases handled by it are inverted by invert_truthvalue.
2169 2005-01-04 Eric Botcazou <ebotcazou@libertysurf.fr>
2171 * config/sparc/sparc.md (save_register_windowdi): Add missing mode.
2172 (save_register_windowsi): Likewise.
2174 2005-01-04 Richard Henderson <rth@redhat.com>
2176 * tree-mudflap.c (mf_varname_tree): Fix thinko setting declname.
2178 2005-01-04 Uros Bizjak <uros@kss-loka.si>
2181 * cse.c (fold_rtx) [RTX_COMPARE, RTX_COMM_COMPARE]: Don't attempt
2182 any simplifications of vector mode comparison operators.
2183 * simplify-rtx.c (simplify_relational_operation): Fix variable name.
2185 2005-01-04 Paolo Bonzini <bonzini@gnu.org>
2186 Devang Patel <dpatel@apple.com>
2188 PR tree-optimization/18308
2189 * tree-if-conv.c (add_to_dst_predicate_list): Gimplify
2190 the operands before creating a new expression.
2191 * dojump.c (do_jump): Make drop_through_label available
2192 for all cases. Add expansion of COND_EXPR.
2194 2005-01-04 Ira Rosen <irar@il.ibm.com>
2196 * tree-vectorizer.c (vect_analyze_offset_expr): Test for
2197 INTEGER_CST instead of TREE_CONSTANT.
2198 (vect_gen_niters_for_prolog_loop): Test for INTEGER_CST
2199 instead of TREE_CONSTANT.
2200 (vect_analyze_pointer_ref_access): Test for INTEGER_CST
2201 instead of TREE_CONSTANT.
2203 2005-01-04 Eric Botcazou <ebotcazou@libertysurf.fr>
2205 * tree-eh.c (replace_goto_queue): Return early if the queue is empty.
2207 2005-01-04 Uros Bizjak <uros@kss-loka.si>
2210 * config/i386/i386.md (*fop_df_1_i387): Disable for TARGET_SSE_MATH.
2211 (*fop_df_1_i387): Disable for (TARGET_SSE2 && TARGET_SSE_MATH).
2213 2005-01-03 Richard Henderson <rth@redhat.com>
2215 * fold-const.c (force_fit_type): Cope with types larger than 2 HWI.
2216 (fold_convert_const_int_from_int, fold_convert_const_int_from_real,
2217 fold_convert_const_real_from_real): Split out from ...
2218 (fold_convert_const): ... here.
2220 2005-01-03 Richard Henderson <rth@redhat.com>
2223 * config/i386/i386.md (movdi_2): Separate SSE1 and SSE2 alternatives.
2224 (mov<MMXMODEI>_internal): Likewise.
2225 (movdf_nointeger): Prefer Y while not preferring, but allowing, x.
2226 Add V2SF case; use it for SSE1; don't use TI.
2227 (movdf_integer): Likewise.
2228 (mov<SSEMODEI>_internal, movti_internal): Force V4SF for SSE1.
2230 2005-01-03 Ira Rosen <irar@il.ibm.com>
2232 * tree-vectorizer.c (vect_strip_conversions): New function.
2233 (vect_analyze_offset_expr): Call vect_strip_conversions. Add
2234 check for binary class.
2236 2005-01-03 Daniel Berlin <dberlin@dberlin.org>
2240 * dwarf2out.c (block_ultimate_origin): Follow decl origin if origin
2242 * gimple-low.c (mark_blocks_with_used_vars): New function.
2243 (mark_blocks_with_used_subblocks): Ditto.
2244 (mark_used_blocks): Ditto.
2245 (pass_mark_used_blocks): New pass.
2246 * tree-inline.c: Include debug.h.
2247 (expand_call_inline): Call outlining_inline_function here.
2248 * tree-optimize.c (init_tree_optimization_passes): Add
2249 pass_mark_used_blocks.
2250 * tree-pass.h (pass_mark_used_blocks): New.
2251 * Makefile.in (tree-inline.o): Add debug.h dependency.
2253 2005-01-03 Geoffrey Keating <geoffk@apple.com>
2255 * config/darwin.c (darwin_handle_weak_import_attribute): Permit
2256 VAR_DECLs to have weak_import attribute.
2258 * config/rs6000/darwin-fallback.c: Use 'ucontext_t' rather than
2261 2004-01-03 Steven Bosscher <stevenb@suse.de>
2263 * passes.c (rest_of_compilation): Don't run regmove if only
2264 flag_expensive_optimizations, require flag_regmove instead.
2266 2005-01-03 Eric Botcazou <ebotcazou@libertysurf.fr>
2268 * config/sparc/litecoff.h (TARGET_OS_CPP_BUILTINS): Do not
2270 * config/sparc/sysv4-only.h (TARGET_OS_CPP_BUILTINS): Likewise.
2271 * config/sparc/rtemself.h (TARGET_SUB_OS_CPP_BUILTINS): Likewise.
2272 * config/sparc/sol2-64.h (TARGET_SUB_OS_CPP_BUILTINS): Delete.
2273 * config/sparc/sp64-elf.h (TARGET_SUB_OS_CPP_BUILTINS): Likewise.
2275 2005-01-03 Richard Henderson <rth@redhat.com>
2277 * config/ia64/ia64.c (TARGET_VECTOR_MODE_SUPPORTED_P): New.
2278 (ia64_const_ok_for_letter_p): New.
2279 (ia64_const_double_ok_for_letter_p): New.
2280 (ia64_extra_constraint): New.
2281 (ia64_expand_vecint_compare): New.
2282 (ia64_expand_vcondu_v2si): New.
2283 (ia64_expand_vecint_cmov): New.
2284 (ia64_expand_vecint_minmax): New.
2285 (ia64_print_operand): Add 'v'.
2286 (ia64_preferred_reload_class): New.
2287 (ia64_vector_mode_supported_p): New.
2288 * config/ia64/ia64.h (UNITS_PER_SIMD_WORD): New.
2289 (PREFERRED_RELOAD_CLASS): Move to function.
2290 (CONST_OK_FOR_LETTER_P): Move to function.
2291 (CONST_DOUBLE_OK_FOR_LETTER_P): Move to function.
2292 (CONSTRAINT_OK_FOR_Q, CONSTRAINT_OK_FOR_R): Remove.
2293 (CONSTRAINT_OK_FOR_S, CONSTRAINT_OK_FOR_T): Remove.
2294 (EXTRA_CONSTRAINT): Move to function.
2295 * config/ia64/ia64.md: Include vect.md.
2296 (itanium_class): Add mmalua.
2298 * config/ia64/itanium1.md (1_mmalua): New. Add it to bypasses.
2300 * config/ia64/itanium2.md (2_mmalua, 2b_mmalua): Similarly.
2301 * config/ia64/predicates.md (gr_reg_or_0_operand): Accept any
2303 (const_int_2bit_operand): New.
2304 (fr_reg_or_0_operand): New.
2305 * config/ia64/ia64-modes.def: Add vector modes.
2306 * config/ia64/ia64-protos.h: Update.
2307 * config/ia64/vect.md: New file.
2309 2005-01-03 Richard Henderson <rth@redhat.com>
2311 * simplify-rtx.c (simplify_binary_operation): Handle VEC_CONCAT.
2313 2005-01-03 Uros Bizjak <uros@kss-loka.si>
2316 * config/i386/i386.c (log1psf2): Change mode of operands[0,1]
2318 (log1pdf2): Change mode of operands[0,1] to DFmode.
2320 2005-01-03 Eric Botcazou <ebotcazou@libertysurf.fr>
2322 * config/sparc/sparc.h (SPARC_RELAXED_ORDERING): Define to false.
2323 * config/sparc/linux.h (SPARC_RELAXED_ORDERING): Define to true.
2324 * config/sparc/linux64.h (SPARC_RELAXED_ORDERING): Likewise.
2325 * config/sparc/sparc.c (TARGET_RELAXED_ORDERING): Define to
2326 SPARC_RELAXED_ORDERING.
2328 2005-01-03 Richard Henderson <rth@redhat.com>
2329 Uros Bizjak <uros@kss-loka.si>
2332 * config/i386/i386.c (ix86_expand_builtin): [IX86_BUILTIN_PINSRW,
2333 IX86_BUILTIN_PINSRW128]: Fix wrong selector range in error message.
2334 * config/i386/i386.md (mmx_pinsrw, sse2_pinsrw): Fix selector
2336 (*mmx_pinsrw, *sse2_pinsrw): New patterns.
2337 * config/i386/i386/predicates.md (const_pow2_1_to_8_operand,
2338 const_pow2_1_to_128_operand): New predicates.
2340 2005-01-02 Greg McGary <greg@mcgary.org>
2342 * tree-mudflap.c (mf_varname_tree): decl_printable_name handles
2344 (mf_xform_derefs_1) [RESULT_DECL, STRING_CST]: Handle as innermost
2345 object. [ptr_type]: Remove unused variable.
2346 Remember nearest addressable array-element or record-component
2347 when checking bitfield components. Tolerate empty BIND_EXPR.
2349 2005-01-01 Richard Henderson <rth@redhat.com>
2351 * tree-vectorizer.c (vect_analyze_offset_expr): Strip conversions
2352 that don't narrow the value. Fail for other conversions.
2354 2005-01-01 Richard Henderson <rth@redhat.com>
2357 * c-decl.c (pop_file_scope): Call maybe_apply_pending_pragma_weaks.
2358 * c-lang.c (finish_file): Don't do it here.
2359 * objc/objc-act.c (objc_finish_file): Likewise.
2361 * cgraph.c (decl_assembler_name_equal): New.
2362 (cgraph_node_for_asm, cgraph_varpool_node_for_asm): New.
2363 (cgraph_varpool_node): Actually link up cgraph_varpool_nodes.
2364 * cgraph.h (struct cgraph_varpool_node): Add next.
2365 (cgraph_node_for_asm, cgraph_varpool_node_for_asm): Declare.
2366 * varasm.c (assemble_alias): Mark the target as needed.
2368 2005-01-01 Andrew Pinski <pinskia@physics.uc.edu>
2371 * function.c (get_arg_pointer_save_area): Use entry_of_function
2372 instead of get_insns.
2374 2005-01-01 Roger Sayle <roger@eyesopen.com>
2375 Andrew Pinski <pinskia@physics.uc.edu>
2376 James E. Wilson <wilson@specifixinc.com>
2378 PR rtl-optimization/12092
2379 * loop.c (emit_prefetch_instructions): Do nothing if PREFETCH_BLOCK
2382 2005-01-01 Roger Sayle <roger@eyesopen.com>
2383 Olivier Hainque <hainque@act-europe.fr>
2385 * tree.c (int_fits_type_p): A narrower type always fits in a
2386 wider one, except for negative values into unsigned types.
2388 2005-01-01 Roger Sayle <roger@eyesopen.com>
2390 * tree.c (int_fits_type_p): Compare the result of force_fit_type
2391 with the original constant rather than require TREE_OVERFLOW.
2393 2005-01-01 Steven Bosscher <stevenb@suse.de>
2396 * c-decl.c (finish_function): If compiling C99, annotate the
2397 compiler generated return with the current file name and line 0.
2398 * tree-cfg.c (remove_useless_stmts_warn_notreached): Only warn if
2399 the source line is greater than 0.
2400 (remove_bb): Likewise.
2402 See ChangeLog.12 for earlier changes.