1 2005-01-26 Bernd Schmidt <bernd.schmidt@analog.com>
3 * function.c (expand_function_end): If current_function_calls_alloca,
4 emit stack restore in a place that is reached when the function falls
7 2005-01-26 Alexandre Oliva <aoliva@redhat.com>
9 * Makefile.in (../$(build_subdir)/fixincludes/fixincl): Add dummy
10 rule to avoid matching inappropriate suffix rule.
12 2005-01-26 Jan Hubicka <jh@suse.cz>
14 PR tree-optimization/19241
15 * tree-inline.c (copy_body_r): Do not walk subtrees after substituting.
17 2005-01-26 Hans-Peter Nilsson <hp@axis.com>
19 * config/cris/cris.h (LIB_SPEC) <sim*>: Use -lsyslinux, not
20 -lsyssim. Wrap -lc -lsyssim in --start-group --end-group instead
22 (STARTFILE_SPEC): Add crti.o, before crtbegin.o.
23 <sim*>: Always use crt1.o, regardless of N in simN.
24 <!sim>: Drop support for -pg and -p variants; always use crt0.o.
25 (ENDFILE_SPEC): Add crtn.o, after crtend.o.
26 (CRT_CALL_STATIC_FUNCTION): Remove.
27 * config/cris/aout.h (STARTFILE_SPEC): <!melinux && sim*>: Always
28 use crt1.o, regardless of N in simN.
29 <!sim>: Drop support for -pg and -p variants; always use crt0.o.
30 (LIB_SPEC): Do not define; default to the one in config/cris/cris.h.
31 (CRIS_CPP_SUBTARGET_SPEC) <melinux && !nostdinc>: Remove special
33 (MAX_OFILE_ALIGNMENT): Correct to 32, not 16.
35 2005-01-25 Ian Lance Taylor <ian@airs.com>
37 * gimple-low.c (block_may_fallthru): Correct handling of
38 SWITCH_EXPR--may fall through if SWITCH_LABELS is NULL.
40 2005-01-26 Hans-Peter Nilsson <hp@axis.com>
42 * config/cris/cris.h (CPP_SPEC): Adjust the various mtune=
43 handlers to match indentation; nested under the mtune=* handler.
44 Similar for mcpu= and march=.
46 2005-01-25 Daniel Jacobowitz <dan@codesourcery.com>
48 * config/mips/linux.h (LIB_SPEC): Remove -rpath-link.
49 * config/mips/linux64.h (LIB_SPEC): Likewise.
50 * config/mn10300/linux.h (LIB_SPEC): Likewise.
51 * config/sh/linux.h (LIB_SPEC): Likewise.
53 2005-01-25 Aldy Hernandez <aldyh@redhat.com>
55 * config/frv/frv.c (frv_legitimize_tls_address): New.
56 (TARGET_HAVE_TLS): Define.
57 (FRV_SYMBOL_REF_TLS_P): Define.
58 (frv_override_options): Handle new register classes.
59 (frv_legitimate_address_p): Reject tls addresses.
60 (frv_legitimize_address): Handle TLS addresses.
61 (gen_inlined_tls_plt): New.
63 (frv_legitimize_tls_address): New.
64 (unspec_got_name): Add TLS entries.
65 (got12_operand): Add R_FRV_TLSMOFF12 case.
66 (frv_emit_move): Fixup TLS addresses.
67 (frv_emit_movsi): Legitimize TLS addresses.
69 * config/frv/frv.h (MASK_BIG_TLS): New.
70 (TARGET_BIG_TLS): New.
71 (HAVE_AS_TLS): Define.
72 (TARGET_SWITCHES): Add -mTLS and -mtls options.
73 (enum reg_class): Add GR8_REGS, GR9_REGS, GR89_REGS.
74 (REG_CLASS_NAMES): Same.
75 (REG_CLASS_CONTENTS): Same.
76 (CONSTRAINT_LEN): New.
77 (REG_CLASS_FROM_CONSTRAINT): New.
78 (PREDICATE_CODES): Add symbolic_operand.
80 * config/frv/frv.md (define_constants): Add UNSPEC_GETTLSOFF,
81 UNSPEC_TLS_LOAD_GOTTLSOFF12, UNSPEC_TLS_INDIRECT_CALL,
82 UNSPEC_TLS_TLSDESC_LDD, UNSPEC_TLS_TLSDESC_LDD_AUX,
83 UNSPEC_TLS_TLSOFF_LD, UNSPEC_TLS_LDDI, UNSPEC_TLSOFF_HILO,
84 R_FRV_GOTTLSOFF_HI, R_FRV_GOTTLSOFF_LO, R_FRV_TLSMOFFHI,
85 R_FRV_TLSMOFFLO, R_FRV_TLSMOFF12, R_FRV_TLSDESCHI,
86 R_FRV_TLSDESCLO, R_FRV_GOTTLSDESCHI, R_FRV_GOTTLSDESCLO, GR8_REG,
87 GR9_REG, GR14_REG, LRREG.
88 (type): Add load_or_call attribute.
89 ("load_or_call"): New reservation.
90 ("call_gettlsoff"): New.
91 ("tls_indirect_call"): New.
92 ("tls_load_gottlsoff12"): New.
94 ("tls_tlsdesc_ldd"): New.
95 ("tls_tlsoff_ld"): New.
98 * config/frv/frv-protos.h (symbolic_operand): Protoize.
100 2005-01-25 Alexandre Oliva <aoliva@redhat.com>
102 * config/frv/frv.h (ASM_OUTPUT_DWARF_DTPREL): Define.
103 * config/frv/frv-protos.h (frv_output_dwarf_dtprel): Declare.
104 * config/frv/frv.c (TLS_BIAS): Define.
105 (frv_output_dwarf_dtprel): New.
107 2005-01-26 Jakub Jelinek <jakub@redhat.com>
109 PR rtl-optimization/19579
110 * ifcvt.c (noce_try_cmove_arith): If emitting instructions to set up
111 both A and B, see if they don't clobber registers the other expr uses.
113 2005-01-25 J"orn Rennecke <joern.rennecke@st.com>
115 * real.c (do_add): Initialize signalling and canonical members.
117 * real.c (real_from_integer): Zero out destination.
119 2005-01-18 Jan Hubicka <jh@suse.cz>
121 PR tree-optimize/19337
122 * tree-inline.c (initialize_inlined_parameters): Copy
123 saved_static_chain_decl only.
125 2005-01-25 Richard Henderson <rth@redhat.com>
129 * config/i386/i386.c (x86_inter_unit_moves): Zero.
130 (ix86_preferred_reload_class): Rewrite fp-constant section, with
131 80387 enabled, to return a proper subclass. Return the subset that
132 overlaps with GENERAL_REGS for PLUS.
133 * config/i386/i386.md (movsi_1, movdi_2): Set type to "mmx" for pxor.
135 2005-01-25 Richard Henderson <rth@redhat.com>
138 * tree-complex.c (expand_complex_div_wide): Use the correct formulae.
140 2005-01-25 Roger Sayle <roger@eyesopen.com>
142 * expmed.c (choose_multiplier): Change interface to return the
143 multiplier lowpart as an rtx instead of an unsigned HOST_WIDE_INT.
144 (expand_divmod): Update calls to choose_multiplier.
146 2005-01-24 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
149 * expmed.c (extract_bit_field): Use adjust_address instead of
150 gen_lowpart when op0 is a MEM.
152 2005-01-24 Steven Bosscher <stevenb@suse.de>
154 * cfgexpand.c (expand_gimple_tailcall): Fix typo.
156 2005-01-24 Richard Sandiford <rsandifo@redhat.com>
158 * config/mips/irix-csr.c: New file.
159 * config/mips/t-iris6 (irix-csr.o): New rule to build it.
160 (EXTRA_MULTILIB_PARTS): Add irix-csr.o.
161 * config/mips/iris6.h (ENDFILE_SPEC): Include it in n32 and n64
164 2005-01-24 Eric Botcazou <ebotcazou@libertysurf.fr>
167 * config.gcc (sparc-*-elf*): Remove sol2.h, sparc/sol2.h and
168 sparc/elf.h, add sparc/sp-elf.h.
169 (sparc-*-rtems*): Likewise.
170 (sparclite-*-elf*): Remove sol2.h, sparc/sol2.h, sparc/elf.h and
171 tm-dwarf2.h, add sparc/sp-elf.h.
172 (sparc86x-*-elf): Likewise.
173 (sparc64-*-elf*): Remove sol2.h, sparc/sol2.h and tm-dwarf2.h.
174 * config/sparc/liteelf.h (TARGET_SUB_OS_CPP_BUILTINS): Rename into
175 TARGET_OS_CPP_BUILTINS.
176 * config/sparc/sp86x-elf (TARGET_SUB_OS_CPP_BUILTINS): Likewise.
177 * config/sparc/rtemself.h (TARGET_SUB_OS_CPP_BUILTINS): Likewise.
179 * config/sparc/openbsd64.h (NO_IMPLICIT_EXTERN_C): Undefine.
180 * config/sparc/sp64-elf.h (NO_IMPLICIT_EXTERN_C): New macro.
181 (SWITCH_TAKES_ARG): Likewise.
182 (LOCAL_LABEL_PREFIX): Likewise.
183 (ASM_GENERATE_INTERNAL_LABEL): Likewise.
184 (TARGET_N_FORMAT_TYPES): Delete.
185 (TARGET_FORMAT_TYPES): Likewise.
186 (ASM_DECLARE_FUNCTION_SIZE): Likewise.
187 * config/sparc/elf.h: Delete.
188 * config/sparc/sp-elf.h: New file.
190 2005-01-24 Kazu Hirata <kazu@cs.umass.edu>
192 * tree-flow-inline.h (phi_arg_from_edge): Remove.
193 * tree-flow.h: Remove the corresponding prototype.
194 * tree-ssa-operands.h, tree-ssa-threadupdate.c, tree-ssa.c,
195 tree-vectorizer.c: Use dest_idx instead of phi_arg_from_edge.
197 2005-01-24 Jakub Jelinek <jakub@redhat.com>
200 * config/rs6000/rs6000.c (rs6000_file_start): Create toc section
201 for AIX ABI or ELF -fPIC.
202 (rs6000_emit_load_toc_table): Don't create toc_section here.
203 (rs6000_xcoff_file_start): Nor here.
205 2005-01-24 J"orn Rennecke <joern.rennecke@st.com>
207 * sh.c (ra.h): Don't #include.
208 (hard_regs_intersect_p): New function, resurrected from ra.c.
210 * sh.c: Fix 1996 Copyright.
212 2005-01-24 Richard Henderson <rth@redhat.com>
213 Aldy Hernandez <aldyh@redhat.com>
215 * regrename.c (note_sets): Handle subregs.
217 2005-01-24 Joseph S. Myers <joseph@codesourcery.com>
220 * genconditions.c (write_header, write_conditions): Elide file if
223 2005-01-24 Kazu Hirata <kazu@cs.umass.edu>
225 * c-opts.c, expmed.c, gengtype.c, tree-complex.c,
226 config/arm/iwmmxt.md, config/avr/avr.c, config/s390/s390.h:
229 2005-01-24 Jakub Jelinek <jakub@redhat.com>
231 * builtins.c (get_pointer_alignment, gimplify_va_arg_expr,
232 expand_builtin_printf, expand_builtin_fprintf,
233 expand_builtin_sprintf): Use POINTER_TYPE_P instead TREE_CODE
234 checking against POINTER_TYPE.
235 (validate_arglist): Handle POINTER_TYPE code by checking
238 2005-01-24 Paolo Bonzini <bonzini@gnu.org>
240 * rtlanal.c (non_rtx_starting_operands, for_each_rtx_1,
242 (for_each_rtx): Call for_each_rtx_1.
243 * rtl.h (init_rtlanal): Declare.
244 * toplev.c (backend_init): Call init_rtlanal.
246 2005-01-24 Jakub Jelinek <jakub@redhat.com>
248 * flow.c (propagate_one_insn): Formatting.
251 * flow.c (libcall_dead_p): Be more conservative if unsure.
252 If there are any instructions between insn and call, see if they are
253 all dead before saying the libcall is dead.
255 2005-01-24 Uros Bizjak <uros@kss-loka.si>
257 * config/i386/i386.md (*extendsfdf2_sse): Use "nonimmediate_operand"
258 operand constraint for operand 0.
260 2005-01-23 Richard Henderson <rth@redhat.com>
263 * tree-complex.c (expand_complex_div_wide): Set EDGE_FALLTHRU on
264 edges leading to join block.
266 2005-01-23 Roger Sayle <roger@eyesopen.com>
268 * expmed.c (expand_mult_highpart): Make static. Change type of
269 constant multiplier argument from unsigned HOST_WIDE_INT to rtx.
270 (expand_divmod): Updates calls to expand_mult_highpart by using
271 gen_int_mode to pass a CONST_INT rtx.
272 * rtl.h (expand_mult_highpart): Remove prototype.
274 2005-01-23 Kazu Hirata <kazu@cs.umass.edu>
276 * flow.c (regset_bytes, regset_size): Remove.
278 * cfganal.c: Fix a reference to Harvey's paper.
280 2005-01-23 Steven Bosscher <stevenb@suse.de>
282 PR rtl-optimization/19464
283 * tree-optimize.c (init_tree_optimization_passes): Add one more
284 copyrename pass just before out-of-ssa.
286 2005-01-23 Joseph S. Myers <joseph@codesourcery.com>
289 * recog.c (recog_memoized): Don't define if GENERATOR_FILE.
290 * ggc-none.c (ggc_free): Define.
292 2005-01-23 Kazu Hirata <kazu@cs.umass.edu>
294 * cse.c (max_reg, max_insn_uid): Remove.
295 (cse_main): Don't access max_reg or max_insn_uid.
297 * c-common.c (explicit_flag_signed_bitfields,
298 lang_statement_code_p, lang_expand_function_end): Remove.
299 * c-common.h: Remove the corresponding declarations.
300 * c-opts.c (c_common_handle_option): Remove all write access
301 to explicit_flag_signed_bitfields.
303 * cgraph.c (cgraph_varpool_n_nodes): Remove.
304 (cgraph_varpool_node): Don't access cgraph_varpool_n_nodes.
305 * cgraph.h: Remove the corresponding declaration.
307 * gcse.c (null_pointer_info): Remove.
309 2005-01-23 Roger Sayle <roger@eyesopen.com>
310 Eric Botcazou <ebotcazou@libertysurf.fr>
312 * combine.c (simplify_logical): Only simplify logical expressions
313 of the form ior(and(x,y),z) by the inverse distributive law if the
314 result is cheaper than the original.
316 2005-01-23 Kazu Hirata <kazu@cs.umass.edu>
318 * alias.c, c-common.h, c-incpath.c, c-incpath.h, expr.c,
319 fold-const.c, gimplify.c, params.h, tree-data-ref.c,
320 tree-if-conv.c, tree-nested.c, tree-outof-ssa.c,
321 tree-ssa-dom.c, tree-vectorizer.c, tree.def, config/darwin.c,
322 config/freebsd-spec.h, config/arm/arm.h,
323 config/h8300/h8300.md, config/i386/i386.md,
324 config/i386/predicates.md, config/i386/sse.md,
325 config/ia64/ia64.c, config/ip2k/ip2k.c, config/s390/s390.c,
326 config/vax/vax.md: Fix comment typos. Follow spelling
329 2005-01-23 Steven Bosscher <stevenb@suse.de>
331 * gengtype.c (walk_type): Produce `not equal to' compares for loop
332 tests, instead of `less than'.
334 2005-01-23 Richard Sandiford <rsandifo@redhat.com>
336 * function.c (assign_parm_setup_block): When creating a new stack slot
337 for a parameter, get its alignment from the parameter's DECL_ALIGN
338 rather than the type's TYPE_ALIGN. Make sure that the parameter
339 is at least word aligned.
341 2005-01-22 Roger Sayle <roger@eyesopen.com>
344 * config/avr/avr.c (avr_hard_regno_mode_ok): Rewrite.
346 2005-01-22 Richard Henderson <rth@redhat.com>
349 * config/i386/i386.md (movsfcc_1_sse_max): Use nonimmediate_operand
350 in both compare operands.
351 (movdfcc_1_sse_max): Likewise.
352 (movsfcc_1_sse): Likewise. Add earlyclobber for scratch.
353 (movdfcc_1_sse): Likewise.
354 * config/i386/i386.c (ix86_split_sse_movcc): Emit copies into the
355 scratch register as needed.
357 2005-01-22 Richard Henderson <rth@redhat.com>
359 * config/i386/i386.md (smaxsf3): Fix mnemonic typo.
361 * config/i386/i386.c (ix86_prepare_fp_compare_args): Fix is_sse test.
363 2005-01-22 Richard Henderson <rth@redhat.com>
365 * genopinit.c (optabs): Use smin/smax for floating point too.
366 * doc/md.texi: Update to match. Clarify that floating point
367 results are undefined for +0/-0 and NaN.
368 * doc/rtl.texi: Likewise.
369 * rtl.def (SMIN, SMAX): Likewise
370 * tree.def (MIN_EXPR, MAX_EXPR): Likewise.
372 * config/alpha/alpha.md (smaxdf3, smindf3, smaxsf3, sminsf3): Add
373 leading 's' to the name.
374 * config/ia64/ia64.md (smaxsf3, sminsf3, smaxdf3, smindf3,
375 smaxxf3, sminxf3): Likewise.
376 * config/rs6000/rs6000.md (smaxdf3, smindf3, smaxsf3, sminsf3):
379 2005-01-22 Paul Brook <paul@codesourcery.com>
381 * config/arm/arm.md: Use "Uy" constraint for wcgr load/stores.
383 2005-01-21 Mark Dettinger <dettinge@de.ibm.com>
385 * config/s390/s390.c (struct processor_costs): New fields
386 dlgr, dlr, dr, dsgfr, dsgr.
387 (z900_cost, z990_cost): Values for new fields.
388 (s390_rtx_costs): New cases MEM und COMPARE in switch
389 statement. Modified handling of SIGN_EXTEND, ZERO_EXTEND,
390 DIV, MOD, UDIV, UMOD.
392 2005-01-21 Ulrich Weigand <uweigand@de.ibm.com>
394 * config/s390/s390.md ("doloop_si64"): Reload input value directly
395 into the register being decremented.
396 ("doloop_si31", "doloop_di"): Likewise.
397 ("*doloop_si_long"): Adapt pattern.
399 2005-01-21 Ulrich Weigand <uweigand@de.ibm.com>
401 * config/s390/s390.h (HARD_REGNO_NREGS): Fix computation for
403 (CLASS_MAX_NREGS): Likewise.
405 2005-01-21 Daniel Berlin <dberlin@dberlin.org>
407 * doc/tree-ssa.texi (Statement Operands): Add example for new
408 must-def macro. Note deprecation of old operands interface.
410 2005-01-22 Richard Sandiford <rsandifo@redhat.com>
412 PR tree-optimization/19484
413 * tree-cfg.c (remove_fallthru_edge): New function.
414 (cleanup_control_flow): Remove fallthru edges from calls that are
415 now known not to return.
417 2005-01-20 Daniel Berlin <dberlin@dberlin.org>
419 Fix PR tree-optimization/19038
420 * tree-ssa-dom.c (cprop_operand): Don't replace loop invaeriant
421 copies with loop variant ones.
423 2005-01-22 Kazu Hirata <kazu@cs.umass.edu>
425 * cfganal.c, real.h, reorg.c, timevar.def, tree-ssa-ccp.c,
426 config/alpha/alpha-protos.h, config/alpha/alpha.h,
427 config/alpha/alpha.md, config/alpha/predicates.md,
428 config/sparc/freebsd.h, config/sparc/netbsd-elf.h,
429 config/sparc/sol2.h: Update copyright.
431 * tree-cfg.c (remove_forwarder_block_with_phi): Look at the
432 first label to see if it is a nonlocal label.
434 2005-01-22 David Edelsohn <edelsohn@gnu.org>
435 Andrew Pinski <pinskia@physics.uc.edu>
438 * config/rs6000/rs6000.c (rs6000_va_start): Saturate n_gpr at
439 maximum number of GPRs. Saturate n_fpr at maximum number of FPRs.
441 2005-01-22 Ralf Corsepius <ralf.corsepius@rtems.org>
443 * config/rs6000/rtems.h: Update copyright.
445 2005-01-22 Ralf Corsepius <ralf.corsepius@rtems.org>
448 * config/rs6000/rtems.h: Resurrect cpp_os_rtems_spec from gcc < 3.4.
449 (CPP_OS_RTEMS_SPEC): New (From gcc-3.3's config/rs6000/sys4.h).
450 (SUBSUBTARGET_EXTRA_SPECS): Use CPP_OS_RTEMS_SPEC.
452 2005-01-22 Volker Reichelt <reichelt@igpm.rwth-aachen.de>
455 * c-typeck.c (convert_arguments): Check for error_mark_node.
457 2005-01-21 Ian Lance Taylor <ian@airs.com>
459 PR tree-optimization/13000
460 * tree-inline.c: Include "tree-flow.h".
461 (expand_call_inline): If warn_return_type, warn if non-void inline
462 function falls through.
463 * tree-cfg.c (execute_warn_function_return): Don't warn about
464 control reaching end if TREE_NO_WARNING is set. Set
466 * gimple-low.c (block_may_fallthru): Don't assume that SWITCH_EXPR
468 * gimplify.c (shortcut_cond_expr): Don't emit a jump over the else
469 branch if we don't need one.
470 * c-typeck.c: Include "tree-flow.h"
471 (c_finish_bc_stmt): Don't add a goto if the current statement
472 list doesn't fall through to the current point.
474 2005-01-21 Roger Sayle <roger@eyesopen.com>
476 PR rtl-optimization/576
477 * real.c (real_arithmetic): Change return type from void to bool
478 to return an indication that the result may be inexact.
479 * real.h (real_arithmeric): Update prototype.
480 * fold-const.c (const_binop): Don't constant fold floating
481 point expressions when the user specifies -frounding-math and
482 the result may depend upon the run-time rounding mode.
483 (fold_convert_const_real_from_real): Clean-up.
484 (fold_initializer): Ignore flag_rounding_math for initializers.
485 * simplify-rtx.c (simplify_binary_operation): Likewise, don't
486 constant fold FP operations with flag_rounding_math if the
487 result may depend upon the run-time rounding mode.
489 2005-01-21 Tom Tromey <tromey@redhat.com>
491 * c-cppbuiltin.c (define__GNUC__): Correct assertion.
493 2005-01-21 Kazu Hirata <kazu@cs.umass.edu>
495 * tree-cfg.c: Fix comment typos.
497 2005-01-21 J"orn Rennecke <joern.rennecke@st.com>
499 * passes.c (rest_of_handle_flow2): Close / open the flow2 dump file
500 around the call to rest_of_handle_branch_target_load_optimize.
502 2005-01-21 Eric Botcazou <ebotcazou@libertysurf.fr>
504 * config.gcc (sparc-*-netbsdelf*): Include dbxelf.h
505 (sparc64-*-openbsd*): Likewise.
506 (sparclite-*-elf*): Include tm-dwarf2.h.
507 (sparc86x-*-elf*): Likewise.
508 (sparc64-*-elf*): Likewise.
509 * config/sparc/linux64.h (DWARF2_DEBUGGING_INFO): Delete.
510 (DBX_DEBUGGING_INFO): Likewise.
511 * config/sparc/liteelf.h (DWARF2_DEBUGGING_INFO): Likewise.
512 (PREFERRED_DEBUGGING_TYPE): Likewise.
513 * config/sparc/netbsd-elf.h (PREFERRED_DEBUGGING_TYPE): Likewise.
514 * config/sparc/sp64-elf.h (DBX_DEBUGGING_INFO): Likewise.
515 (PREFERRED_DEBUGGING_TYPE): Likewise.
516 * config/sparc/sp86x-elf.h (DWARF2_DEBUGGING_INFO): Likewise.
517 (PREFERRED_DEBUGGING_TYPE): Likewise.
518 * config/sparc/sparc.h (DBX_DEBUGGING_INFO): Likewise.
520 2005-01-21 Hans-Peter Nilsson <hp@bitrange.com>
523 * combine.c (combine_simplify_rtx): Revert change of 2004-12-31.
525 2005-01-20 Paul Brook <paul@codesourcery.com>
527 * doc/rtl.texi: Document value extension requirements for CONST_INT.
529 2005-01-20 Kazu Hirata <kazu@cs.umass.edu>
531 * tree-cfg.c (tree_verify_flow_info): Check that a nonlocal
532 label is first in a sequence of labels.
534 * tree-cfg.c (tree_verify_flow_info): Fix a typo.
536 2005-01-20 Janis Johnson <janis187@us.ibm.com>
537 Giovanni Bajo <giovannibajo@gcc.gnu.org>
539 * doc/sourcebuild.texi (Test Directives): New.
541 2005-01-20 Roger Sayle <roger@eyesopen.com>
543 * real.c (real_floor): Don't leave the result uninitialized when
544 mode is VOIDMode, but return the unrounded intermediate value.
545 (real_ceil): Likewise.
547 2005-01-20 Paul Brook <paul@codesourcery.com>
549 * config/arm/arm.md (insv): Use gen_int_mode.
551 2005-01-20 Eric Botcazou <ebotcazou@libertysurf.fr>
553 * config/sparc/freebsd.h (ASM_OUTPUT_INTERNAL_LABELREF): Delete.
554 * config/sparc/linux.h (ASM_OUTPUT_INTERNAL_LABELREF): Likewise.
555 * config/sparc/linux64.h (ASM_OUTPUT_INTERNAL_LABELREF): Likewise.
556 * config/sparc/netbsd-elf.h (ASM_OUTPUT_INTERNAL_LABELREF): Likewise.
557 * config/sparc/sol2.h (ASM_OUTPUT_INTERNAL_LABELREF): Likewise.
559 2005-01-20 Daniel Berlin <dberlin@dberlin.org>
562 * dwarf2out.c (concat_loc_descriptor): We don't know we can use
563 fbreg, so we have to assume we can't.
564 (loc_descriptor_from_tree_1): Ditto.
565 (containing_function_has_frame_base): New function.
566 (add_location_or_const_value_attribute): Use it.
567 Also try to generate a frame_base from a single element location
570 2005-01-20 Kazu Hirata <kazu@cs.umass.edu>
572 PR tree-optimization/15349
573 * timevar.def (TV_TREE_MERGE_PHI): New.
574 * tree-cfg.c (tree_forwarder_block_p): Add a new argument
576 (remove_forwarder_block, cleanup_forwarder_blocks): Adjust the
577 calls to tree_forwarder_block_p.
578 (remove_forwarder_block_with_phi, merge_phi_nodes,
579 gate_merge_phi, pass_merge_phi): New.
580 * tree-optimize.c (init_tree_optimization_passes): Add
582 * tree-pass.h: Add an extern for pass_merge_phi;
584 2005-01-20 Richard Henderson <rth@redhat.com>
587 * config/i386/emmintrin.h (_mm_castpd_ps, _mm_castpd_si128): New.
588 (_mm_castps_pd, _mm_castps_si128): New.
589 (_mm_castsi128_ps, _mm_castsi128_pd): New.
591 2005-01-20 Richard Henderson <rth@redhat.com>
594 * config/i386/mmintrin.h (_mm_cvtsi32_si64): Use
595 __builtin_ia32_vec_init_v2si.
596 (_mm_cvtsi64_si32): Use __builtin_ia32_vec_ext_v2si.
597 * config/i386/i386.c (IX86_BUILTIN_VEC_EXT_V2SI): New.
598 (ix86_init_mmx_sse_builtins): Create it.
599 (ix86_expand_builtin): Expand it.
600 (ix86_expand_vector_set): Handle V2SFmode and V2SImode.
601 * config/i386/mmx.md (vec_extractv2sf_0, vec_extractv2sf_1): New.
602 (vec_extractv2si_0, vec_extractv2si_1): New.
604 2005-01-20 Richard Henderson <rth@redhat.com>
607 * config/i386/i386.c (ix86_expand_push): New.
608 * config/i386/mmx.md (push<MMXMODE>1): New.
609 * config/i386/sse.md (push<SSEMODE>1): New.
610 * config/i386/i386-protos.h: Update.
612 2005-01-20 Kazu Hirata <kazu@cs.umass.edu>
614 * params.def, params.h, predict.c, tree-ssa-loop-im.c: Update
617 2005-01-20 Richard Henderson <rth@redhat.com>
619 * rtl.def (CONST_VECTOR): Use RTX_CONST_OBJ.
620 * rtl.h (CONSTANT_P): Don't special case CONST_VECTOR.
622 2005-01-19 Richard Henderson <rth@redhat.com>
625 * config/i386/i386.c (ix86_expand_vector_move_misalign): Convert
626 to V4SFmode in SSE1 fallback load path.
628 2005-01-19 Richard Henderson <rth@redhat.com>
630 * config/i386/i386.c (ix86_expand_vector_init_one_var): Fix typo
631 in QImode expansion to ix86_expand_vector_set.
632 (ix86_expand_vector_init_general): Fix typo in V8HImode recursive call.
634 2005-01-19 Richard Henderson <rth@redhat.com>
637 * config/i386/i386.c (ix86_preferred_reload_class): Return a proper
638 subclass of the input class.
639 (ix86_secondary_memory_needed): Always true for cross-MMX classes.
640 Always true for cross-SSE1 classes. Rationalize conditionals.
641 * config/i386/i386.h (SSE_CLASS_P, MMX_CLASS_P): Use straight equality.
642 * config/i386/i386.md (movsi_1): Add MMX/SSE zeros. Fix alternatives
643 for SSE1. Don't check TARGET_INTER_UNIT_MOVES.
644 (movdi_2): Add MMX/SSE zeros.
645 (movdi_1_rex64): Likewise. Don't check TARGET_INTER_UNIT_MOVES.
646 (movsf_1): Don't check TARGET_INTER_UNIT_MOVES.
647 (zero_extendsidi2_32, zero_extendsidi2_rex64): Likewise.
648 (movsi_1_nointernunit, movdi_1_rex64_nointerunit): Remove.
649 (movsf_1_nointerunit, zero_extendsidi2_32_1): Remove.
650 (zero_extendsidi2_rex64_1): Remove.
651 (MOV0 peephole): Check GENERAL_REG_P.
653 2005-01-19 Richard Henderson <rth@redhat.com>
656 * config/i386/i386.c (ix86_expand_vector_set): Fix third and fourth
658 (ix86_expand_vector_extract): Likewise.
660 2005-01-19 Richard Henderson <rth@redhat.com>
663 * expr.c (emit_move_change_mode): New argument force; use
664 simplify_gen_subreg if true.
665 (emit_move_via_alt_mode): Merge into ...
666 (emit_move_via_integer): ... here.
667 (emit_move_ccmode): Use emit_move_change_mode directly.
669 2005-01-19 Richard Henderson <rth@redhat.com>
672 * config/alpha/alpha.c (alpha_rtx_costs): Handle HIGH.
673 (alpha_preferred_reload_class): Handle CONST_VECTOR.
674 (alpha_emit_set_const_1): Add no_output parameter; don't emit
676 (alpha_emit_set_const): Likewise. Make static.
677 (alpha_emit_set_long_const): Make static.
678 (alpha_extract_integer): Split out from alpha_expand_mov.
679 (alpha_split_const_mov): Likewise.
680 (alpha_expand_mov): Use them. Handle CONST_VECTOR.
681 (alpha_legitimate_constant_p): New.
682 * config/alpha/alpha-protos.h: Update.
683 * config/alpha/alpha.h (REGISTER_MOVE_COST): Correct fp<->gp cost.
684 (LEGITIMATE_CONSTANT_P): Re-implement with a function.
685 * config/alpha/alpha.md (movsi): Add n alternative.
686 (movsi_nt_vms, movdi_er_nofix, movdi_er_fix, movdi_fix): Likewise.
687 (mov<VEC>_fix, mov<VEC>_nofix): Add i alternative.
688 (splitters for all of the above): Use alpha_split_const_mov.
689 * config/alpha/predicates.md (non_add_const_operand): New.
690 (non_zero_const_operand): New.
691 (input_operand): Use alpha_legitimate_constant_p after reload.
693 2005-01-19 Zdenek Dvorak <dvorakz@suse.cz>
695 PR tree-optimization/19038
696 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): New function.
697 (add_candidate): Add ivs with increment in latch only if
698 allow_ip_end_pos_p is true.
699 (determine_iv_cost): Use empty_block_p.
701 2005-01-19 Daniel Berlin <dberlin@dberlin.org>
703 * cfganal.c (compute_dominance_frontiers_1): Replace with new algorithm
704 (compute_dominance_frontiers): Ditto.
706 2005-01-19 Ralf Corsepius <ralf.corsepius@rtems.org>
709 * config/sh/t-rtems: New.
710 * config.gcc (sh-*-rtems*): Reflect having added config/sh/t-rtems.
712 2005-01-19 Zdenek Dvorak <dvorakz@suse.cz>
714 * tree-ssa-ccp.c (maybe_fold_offset_to_component_ref): Always subtract
715 the offset of the selected field.
717 2005-01-19 Kazu Hirata <kazu@cs.umass.edu>
719 * tree-cfg.c (remove_forwarder_block): Fix the check to
720 prevent a nonlocal label from appearing in the middle of a
723 2005-01-19 Hans-Peter Nilsson <hp@axis.com>
725 PR rtl-optimization/19462
726 * reorg.c (find_end_label): Create return insn only if
727 current_function_epilogue_delay_list is empty.
729 2005-01-19 Paolo Bonzini <bonzini@gnu.org>
731 * doc/passes.texi: Remove paragraph mentioning new-ra.
733 2005-01-19 Jakub Jelinek <jakub@redhat.com>
735 PR rtl-optimization/15139
736 * combine.c: Include params.h.
737 (count_rtxs): New function.
738 (record_value_for_reg): If replace_rtx would replace at least
739 2 occurrences of REG in VALUE and TEM is really large, replace REG with
740 (clobber (const_int 0)) instead of TEM.
741 * params.def (PARAM_MAX_LAST_VALUE_RTL): New.
742 * params.h (MAX_LAST_VALUE_RTL): New.
743 * Makefile.in (combine.o): Depend on $(PARAMS_H).
744 * doc/invoke.texi (--param max-last-value-rtl=N): Document.
747 * c-typeck.c (digest_init): Only call build_vector if all constructor
748 elements are *_CST nodes.
749 * gimplify.c (gimplify_init_constructor): Likewise.
752 * c-typeck.c (digest_init): Only call build_vector if inside_init
755 2005-01-18 Tobias Schl"uter <tobias.schlueter@physik.uni-muenchen.de>
757 * toplev.c (init_asm_output): Remove dead #ifdef.
759 2005-01-18 Kaz Kojima <kkojima@gcc.gnu.org>
761 * config/sh/linux.h (TARGET_C99_FUNCTIONS): Define.
763 2005-01-18 Eric Botcazou <ebotcazou@libertysurf.fr>
765 * calls.c (expand_call): Check DECL_BUILT_IN_CLASS before
766 accessing DECL_FUNCTION_CODE.
767 * dojump.c (do_jump): Likewise.
768 * gimplify.c (gimplify_call_expr): Likewise.
769 * predict.c (expr_expected_value): Likewise.
770 (strip_builtin_expect): Likewise.
771 * tree-inline.c (estimate_num_insns_1): Likewise.
772 * tree-ssa-loop-im.c (stmt_cost): Likewise
773 * fold-const.c (fold): Test for BUILT_IN_NORMAL.
774 (tree_expr_nonnegative_p): Likewise.
776 2005-01-18 Eric Botcazou <ebotcazou@libertysurf.fr>
778 * config/sparc/sparc.c (load_pic_register): Emit the appropriate
779 variant of the load_pcrel_sym pattern.
780 * config/sparc/sparc.md (P macro): Move to the top.
781 (load_pcrel_sym): Macroize using P.
782 (save_register_window): Likewise.
784 2005-01-18 Aldy Hernandez <aldyh@redhat.com>
786 * config/rs6000/rs6000.md ("sunordered"): Disable for e500.
789 2005-01-18 Andrew Pinski <pinskia@physics.uc.edu>
792 * c-typeck.c (build_asm_expr): Strip nops off
793 input memory operands.
795 2005-01-18 David Edelsohn <edelsohn@gnu.org>
797 * config/rs6000/aix43.h (CPLUSPLUS_CPP_SPEC): Delete
798 _XOPEN_SOURCE, _XOPEN_SOURCE_EXTENDED, _LARGE_FILE_API.
799 * config/rs6000/aix51.h (CPLUSPLUS_CPP_SPEC): Same.
800 * config/rs6000/aix52.h (CPLUSPLUS_CPP_SPEC): Same.
802 2005-01-18 Joel Sherrill <joel.sherrill@oarcorp.com>
805 * config/i386/i386.c (override_options): If the 80387 is disabled,
806 then do not return FP values using FP registers.
808 2005-01-18 Kazu Hirata <kazu@cs.umass.edu>
810 * basic-block.h, c-common.c, c-cppbuiltin.c, c-lang.c,
811 c-tree.h, cfgbuild.c, cgraph.c, cgraph.h, collect2.c,
812 combine.c, config.gcc, coverage.h, cse.c, cselib.c,
813 defaults.h, df.c, dwarf2asm.c, dwarf2out.c, explow.c, expr.c,
814 flow.c, fold-const.c, gcse.c, ggc-page.c, gimple-low.c,
815 gimplify.c, ifcvt.c, langhooks-def.h, lcm.c, optabs.h,
816 output.h, postreload-gcse.c, postreload.c, recog.c,
817 resource.c, rtl.def, rtlanal.c, sched-deps.c, sched-rgn.c,
818 targhooks.h, toplev.c, tree-data-ref.c, tree-eh.c,
819 tree-flow.h, tree-loop-linear.c, tree-mudflap.h, tree-nrv.c,
820 tree-optimize.c, tree-outof-ssa.c, tree-pass.h,
821 tree-scalar-evolution.c, tree-ssa-copy.c, tree-ssa-dce.c,
822 tree-ssa-dse.c, tree-ssa-loop-ivopts.c, tree-ssa-operands.c,
823 tree-ssa-pre.c, tree-ssa.c, tree-vectorizer.c, tree.def,
824 unwind-dw2-fde-darwin.c, var-tracking.c: Update copyright.
826 * config/darwin.c, config/darwin.h, config/freebsd-spec.h,
827 config/alpha/alpha.c, config/arm/vfp.md,
828 config/i386/emmintrin.h, config/i386/i386-modes.def,
829 config/i386/i386-protos.h, config/i386/i386.h,
830 config/i386/pmmintrin.h, config/i386/ppro.md,
831 config/i386/predicates.md, config/i386/xmmintrin.h,
832 config/i860/i860.c, config/ia64/ia64-modes.def,
833 config/ia64/ia64-protos.h, config/ia64/ia64.h,
834 config/ia64/ia64.md, config/ia64/itanium1.md,
835 config/ia64/itanium2.md, config/ia64/predicates.md,
836 config/iq2000/iq2000.h, config/mips/linux64.h,
837 config/rs6000/aix41.h, config/rs6000/aix43.h,
838 config/rs6000/aix52.h, config/rs6000/darwin-fallback.c,
839 config/rs6000/rs6000.c, config/rs6000/rs6000.h,
840 config/rs6000/spe.md, config/sh/sh.md, config/sparc/linux.h,
841 config/sparc/linux64.h, config/sparc/litecoff.h,
842 config/sparc/sp64-elf.h, config/sparc/sparc.c,
843 config/sparc/sparc.h, config/sparc/sparc.md,
844 config/sparc/sysv4-only.h: Update copyright.
846 2005-01-18 Richard Henderson <rth@redhat.com>
849 * config/i386/i386.c (ix86_expand_fp_movcc): Fail for LTGT and UNEQ.
851 2005-01-18 Marc Espie <espie@openbsd.org>
852 * config/i386/openbsdelf.h: Typo.
854 2005-01-18 Richard Henderson <rth@redhat.com>
856 * config/i386/i386.c (ix86_expand_fp_movcc): Remove TARGET_IEEE_FP
857 special case for sse.
858 * config/i386/predicates.md (sse_comparison_operator): Likewise.
860 2005-01-18 Dorit Naishlos <dorit@il.ibm.com>
862 * tree-inline.c (estimate_num_insns_1): Added cases for
863 ALIGN_INDIRECT_REF, MISALIGNED_INDIRECT_REF, and REALIGN_LOAD_EXPR.
865 2005-01-18 Eric Botcazou <ebotcazou@libertysurf.fr>
867 PR rtl-optimization/19296
868 * combine.c (simplify_comparison): Rewrite the condition under
869 which a non-paradoxical SUBREG of a PLUS can be lifted when
870 compared against a constant.
872 2005-01-18 Andi Kleen <ak@muc.de>
874 * c-typeck.c: (convert_for_assignment): Check warn_pointer_sign.
875 * c.opt (-Wpointer-sign): Add.
876 * doc/invoke.texi: (-Wpointer-sign): Add.
878 2005-01-18 Uros Bizjak <uros@kss-loka.si>
880 * config/i386/i386.c (override_options): Revert 2004-11-24 change.
881 * config/i386/i386.md (fmodsf3, dremsf3, *sinsf2, *cossf2, sincossf3,
882 *tansf3_1, tansf2, atan2sf3_1, atan2sf3, atansf2, asinsf2, acossf2,
883 logsf2, log10sf2, log2sf2, log1psf2, logbsf2, ilogbsf2, expsf2,
884 exp10sf2, exp2sf2, expm1sf2, rintsf2, floorsf2, ceilsf2, btruncsf2,
885 nearbyintsf2): Disable for TARGET_SSE_MATH. Leave patterns enabled
886 for TARGET_MIX_SSE_I387.
887 (fmoddf3, dremdf3, *sindf2, *sinextendsfdf2, *cosdf2, *cosextendsfdf2,
888 sincosdf3, *sincosextendsfdf3, *tandf3_1, tandf2, atan2df3_1,
889 atan2df3, atandf2, asindf2, acosdf2, logdf2, log10df2, log2df2,
890 log1pdf2, logbdf2, expdf2, exp10df2, exp2df2, expm1df2, rintdf2,
891 floordf2, ceildf2, btruncdf2, nearbyintdf2): Disable for
892 (TARGET_SSE2 && TARGET_SSE_MATH). Leave patterns enabled for
894 (atan2sf3, atan2df3, atan2xf3): Remove register constraints
897 2005-01-18 Uros Bizjak <uros@kss-loka.si>
900 * config/i386/mmx.md (*movv2sf_internal_rex64, *movv2sf_internal):
901 Add movaps alternative for xmm reg->reg move.
903 2005-01-17 Jeff Law <law@redhat.com>
905 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Clear SSA_NAME_VALUE
906 at the end of the main DOM loop rather than just before DOM exits.
908 2005-01-17 Diego Novillo <dnovillo@redhat.com>
910 PR tree-optimization/19121
911 * tree-ssa-alias.c (compute_flow_sensitive_aliasing): When
912 adding aliases to a name tag, also add them to the pointer's
914 * tree-ssa-copy.c (merge_alias_info): Do not merge flow
915 sensitive alias info at all. Only check that the two pointers
916 have compatible pointed-to sets.
917 * tree-ssa.c (verify_name_tags): Verify that the alias set of
918 a pointer's type tag is a superset of the alias set of the
921 2005-01-17 James E Wilson <wilson@specifixinc.com>
924 * config/ia64/ia64.md (movxf): Handle general register source. Adjust
925 comment to document why.
927 2005-01-17 Richard Henderson <rth@redhat.com>
929 * config/i386/sse.md (smaxv4sf3_finite, sse_vmsmaxv4sf3_finite,
930 sminv4sf3_finite, sse_vmsminv4sf3_finite, smaxv2df3_finite,
931 sse2_vmsmaxv2df3_finite, sminv2df3_finite,
932 sse2_vmsminv2df3_finite): New.
933 (smaxv4sf3, sse_vmsmaxv4sf3, sminv4sf3, sse_vmsminv4sf3, smaxv2df3,
934 sse2_vmsmaxv2df3, sminv2df3, sse2_vmsminv2df3): Remove commutative.
935 Force op1 into register.
937 2005-01-17 Kelley Cook <kcook@gcc.gnu.org>
939 * Makefile.in (STAGEMOVESTUFF): Stage all the stamp files to prevent
940 bubblestrap from always rebuilding build directory.
942 2005-01-17 Ulrich Weigand <uweigand@de.ibm.com>
944 * config/s390/s390.md ("*clc"): Remove incorrect '='.
945 ("*cmpmem_short"): Likewise.
947 2005-01-17 Steven Bosscher <stevenb@suse.de>
949 * basic-block.h: Document BB_* flags.
950 * regrename.c (copyprop_hardreg_forward): Don't use BB_VISITED,
951 use an sbitmap instead.
952 * sched-rgn.c (compute_trg_info): Likewise.
954 2005-01-17 Richard Sandiford <rsandifo@redhat.com>
956 * config.gcc (mips64*-*-linux*): Set the default abi to n32. Remove
957 redundant target_cpu_default and extra_parts lines. Use the default
959 * config/mips/linux64.h (DRIVER_DEFAULT_ABI_SELF_SPEC): Delete.
960 (SUBTARGET_EXTRA_SPECS): Delete.
961 (DRIVER_SELF_SPECS): Remove architecture lines. Set the ABI directly.
962 (SUBTARGET_ASM_SPEC): Remove -mabi=n32 and -mabi=64 mappings.
963 * config/mips/t-linux64 (MULTILIB_DIRNAMES): Use "32" for -mabi=32
964 and "n32" for -mabi=n32.
966 2005-01-17 Ian Lance Taylor <ian@airs.com>
969 * c-typeck.c (build_c_cast): Revert patch of 2003-02-16: don't
970 fold constant variables into initial values.
972 2005-01-17 Kazu Hirata <kazu@cs.umass.edu>
974 * tree-cfg.c (tree_can_merge_blocks_p): Reorder two checks.
976 * tree-cfg.c (tree_forwarder_block_p): Speed up by walking
977 through the statements backward.
979 2005-01-17 Ian Lance Taylor <ian@airs.com>
982 * tree-inline.c (expand_call_inline): Set TREE_NO_WARNING on
983 a variable set to the return value of the inlined function.
985 2005-01-17 Mark Dettinger <dettinge@de.ibm.com>
987 * rtlanal.c (rtx_cost): Assign cost of 0 to a SUBREG
988 when modes are tieable.
990 2005-01-17 Ranjit Mathew <rmathew@hotmail.com>
992 * gthr-posix.h (__gthread_active_p): Use pthread_cancel instead
993 of pthread_create to find out if threads are enabled.
994 * gthr-posix95.h (__gthread_active_p): Likewise.
996 2005-01-17 Paolo Bonzini <bonzini@gnu.org>
998 * common.opt (-fnew-ra): Remove.
1000 * toplev.h (flag_new_regalloc): Remove.
1001 * Makefile.in (ra*.*): Don't mention.
1002 * passes.c (rest_of_handle_new_regalloc): Remove.
1003 (rest_of_handle_combine, rest_of_compilation): Always consider
1004 flag_new_regalloc as false.
1005 * doc/invoke.texi: Don't document -fnew-ra.
1007 2005-01-17 Paolo Bonzini <bonzini@gnu.org>
1009 * bb-reorder.c (fix_edges_for_rarely_executed_code): Remove
1010 last parameter to reg_scan.
1011 * loop.c (loop_optimize): Likewise.
1012 * passes.c (rest_of_handle_tracer, rest_of_handle_if_conversion,
1013 rest_of_handle_web, rest_of_handle_cfg, rest_of_handle_jump_bypass,
1014 rest_of_handle_life, rest_of_handle_cse, rest_of_handle_cse2,
1015 rest_of_handle_gcse, rest_of_handle_loop_optimize,
1016 rest_of_handle_loop2, rest_of_handle_jump2): Likewise.
1017 * regclass.c (reg_scan): Likewise, for the declaration.
1018 * rtl.h (reg_scan): Likewise, for the prototype.
1020 2005-01-17 Kazu Hirata <kazu@cs.umass.edu>
1022 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Don't call
1023 mark_dfs_back_edges.
1025 2005-01-16 Daniel Berlin <dberlin@dberlin.org>
1027 * tree-ssa-pre.c (add_to_sets): s1 may be NULL.
1028 (compute_avail): Uses don't go in tmp_gen.
1030 2005-01-16 Steven Bosscher <stevenb@suse.de>
1032 * ggc-page.c (ggc_alloc_stat): Use __builtin_ctzl instead of a
1033 loop to look for a free slot in a page entry.
1035 2005-01-16 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1038 * defaults.h (TARGET_DEFERRED_OUTPUT_DEFS): Provide default.
1039 * toplev.c (compile_file): Call process_pending_assemble_output_defs
1040 just before targetm.asm_out.file_end.
1041 * tree.h (process_pending_assemble_output_defs): Declare.
1042 * varasm.c (assemble_output_def, process_pending_assemble_output_defs):
1044 (assemble_alias): Defer generation of assembly code for defines when
1045 TARGET_DEFERRED_OUTPUT_DEFS is true.
1046 * config/rs6000/aix41.h (TARGET_DEFERRED_OUTPUT_DEFS): Define.
1047 * config/rs6000/aix43.h (TARGET_DEFERRED_OUTPUT_DEFS): Define.
1048 * doc/tm.texi (TARGET_DEFERRED_OUTPUT_DEFS): document.
1050 2005-01-15 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1053 * pa.c (pa_scalar_mode_supported_p): New function.
1054 (TARGET_SCALAR_MODE_SUPPORTED_P): Define.
1055 * pa.h (MIN_UNITS_PER_WORD): Add comment.
1057 * pa-protos.h (prefetch_operand): Delete.
1058 (prefetch_cc_operand, prefetch_nocc_operand): New declations.
1059 * pa.c (prefetch_operand): Delete.
1060 (prefetch_cc_operand, prefetch_nocc_operand): New functions.
1061 * pa.h (EXTRA_CONSTRAINT): Add `W' constraint.
1062 (PREDICATE_CODES): Delete prefetch_operand. Add prefetch_cc_operand
1063 and prefetch_nocc_operand.
1064 * pa.md (prefetch): Rework to avoid reload problems handling short
1065 displacements when a cache control completer needs to be provided.
1066 (prefetch_32, prefetch_64): Delete.
1067 (prefetch_cc, prefetch_nocc): New patterns.
1069 2005-01-15 David Edelsohn <edelsohn@gnu.org>
1071 * config/rs6000/aix52.h (CPLUSPLUS_CPP_SPEC): Revert previous change.
1073 2005-01-15 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1076 * pa.c (pa_asm_output_aligned_common): Warn if specified alignment
1077 exceeds maximum alignment for global common data.
1078 * pa64-hpux.h (MAX_OFILE_ALIGNMENT): Define to 32768.
1079 * som.h (MAX_OFILE_ALIGNMENT): Likewise.
1081 2004-01-15 Roger Sayle <roger@eyesopen.com>
1083 * tree-ssa-dom.c (extract_range_from_cond): Correct condition.
1085 2004-01-15 Roger Sayle <roger@eyesopen.com>
1087 * harg-reg-set.h (reg_class_names): Prototype global array.
1088 * regclass.c (reg_class_names): Declare here and initialize to
1090 (dump_regclass): Remove local declaration of reg_class_names.
1091 (regclass): Likewise.
1092 * cfg.c (dump_flow_info): Likewise.
1093 * ra-debug.c (reg_class_names): Likewise.
1094 * regrename.c (reg_class_names): Likewise.
1095 * reload.c (reg_class_names): Likewise.
1096 * reload1.c (spill_failure): Likewise.
1097 * config/m68hc11/m68hc11.c (reg_class_names): Likewise.
1099 2005-01-15 Ulrich Weigand <uweigand@de.ibm.com>
1101 * config/s390/s390.md ("reload_outti"): Remove predicate for
1102 output operand. Abort if operand is not a MEM.
1103 ("reload_outdi", "reload_outdf"): Likewise.
1105 2005-01-15 Marc Espie <espie@openbsd.org>
1107 * config.gcc (*-*-openbsd*): Set HAS_LIBC_R for OpenBSD <= 3.2.
1108 (i[34567]86-*-openbsd*): Switch to ELF for OpenBSD >= 3.4.
1109 * config/openbsd.h: Fix C++ includes for native configurations.
1110 Add proper OS_CPP_BUILTINS. Fix libspec for recent OpenBSD.
1111 Add trampoline support.
1112 * config/i386/openbsdelf.h: New.
1114 2005-01-15 Marc Espie <espie@openbsd.org>
1116 * collect2.c (main): Explicitly parse -dynamic-linker option.
1118 2005-01-15 Jakub Jelinek <jakub@redhat.com>
1120 PR tree-optimization/19060
1121 * tree-ssa-dom.c (extract_range_from_cond) <case LT_EXPR, GT_EXPR>:
1122 Return 0 if op1 <= TYPE_MIN_VALUE () resp. op1 >= TYPE_MAX_VALUE ().
1123 (simplify_cond_and_lookup_avail_expr): Add assert for dummy == 0
1124 and handle extract_range_from_cond returning false.
1125 * fold-const.c (fold): Optimize comparisons with min/max even for
1126 width > HOST_BITS_PER_WIDE_INT.
1128 2005-01-15 Ralf Corsepius <ralf.corsepius@rtems.org>
1130 * config/mips/rtems.h (MIPS_DEFAULT_GVALUE): Set to 0.
1131 * config/mips/t-rtems (MULTILIBS_DIRNAMES,MULTILIB_OPTIONS):
1132 Remove little endian multilib variants.
1133 Add mips32 multilib variant.
1135 2005-01-15 Kazu Hirata <kazu@cs.umass.edu>
1137 * explow.c (copy_all_reg, stabilize): Remove.
1138 * expr.h: Remove the corresponding prototypes.
1140 * rtlanal.c (regs_set_between_p): Remove.
1141 * rtl.h: Remove the corresponding prototype.
1143 2005-01-14 David Edelsohn <edelsohn@gnu.org>
1145 * config/rs6000/rs6000.c (rs6k_nonimmediate_operand): Rename
1146 to rs6000_nonimmediate_operand.
1147 * config/rs6000/rs6000.h (PREDICATE_CODES): Rename
1148 rs6k_nonimmediate_operand to rs6000_nonimmediate_operand.
1149 * config/rs6000/rs6000.md (movsi_internal1): Rename predicate to
1150 rs6000_nonimmediate_operand.
1151 * config/rs6000/spe.md (movdf_e500_double): Rename predicate to
1152 rs6000_nonimmediate_operand.
1154 2005-01-14 Andrew Pinski <pinskia@physics.uc.edu>
1156 * tree-ssa-alias.c (pass_may_alias): Add TODO_verify_stmts.
1158 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
1160 * config/rs6000/rs6000.h (CLASS_MAX_NREGS): DF goes in 1 register
1162 (CANNOT_CHANGE_MODE_CLASS): Restrict DI mode changes on e500v2.
1163 (PREDICATE_CODES): Add rs6k_nonimmediate_operand.
1165 * config/rs6000/rs6000.c (invalid_e500_subreg): New.
1166 (rs6k_nonimmediate_operand): New.
1167 (rs6000_legitimate_offset_address_p): Handle DI modes on e500v2
1169 (legitimate_lo_sum_address_p): Same.
1170 (rs6000_legitimize_address): Same.
1171 (rs6000_legitimize_reload_address): Same.
1172 (rs6000_legitimate_address): Same.
1173 (spe_build_register_parallel): Pass DF and DC modes in a DI
1176 * config/rs6000/rs6000.md ("*movsi_internal1"): Change predicate
1177 to rs6k_nonimmediate_operand.
1179 * config/rs6000/spe.md ("*frob_df_di"): New.
1180 ("*frob_di_df"): New.
1181 ("*frob_di_df_2"): New.
1182 ("*mov_sidf_e500_subreg0"): New.
1183 ("*mov_sidf_e500_subreg4"): New.
1184 ("*movdf_e500_double"): Change predicate to
1185 rs6k_nonimmediate_operand.
1187 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
1189 * postreload.c (move2add_note_store): Only call
1190 trunc_int_for_mode on scalar integers.
1192 2005-01-14 Steven Bosscher <stevenb@suse.de>
1194 * tree-ssa-dce.c (visited_control_parents): New sbitmap to
1195 replace BB_VISITED uses.
1196 (find_obviously_necessary_stmts): Don't clear BB_VISITED.
1197 (propagate_necessity): Check the bitmap instead of BB_VISITED.
1198 (tree_dce_done): Free visited_control_parents.
1199 (perform_tree_ssa_dce): Allocate and clear it.
1200 * tree-ssa-pre.c (compute_antic_aux): Make non-recursive.
1201 (compute_antic): Iterate from here using a DFS. Use an sbitmap
1202 instead of BB_VISITED.
1204 2005-01-14 Kazu Hirata <kazu@cs.umass.edu>
1206 * c-tree.h, coverage.h, langhooks-def.h, optabs.h, output.h,
1207 rtl.h, targhooks.h, tree-flow.h, tree-mudflap.h, tree.h:
1208 Remove unused prototypes.
1210 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
1212 * config/rs6000/rs6000.c (rs6000_stack_info): Fix caching of SPE
1213 64-bit register usage.
1214 (spe_func_has_64bit_regs_p): Add FIXME note.
1216 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
1218 * config/rs6000/rs6000.c (rs6000_generate_compare): Replace
1219 flag_finite_math_only with flag_unsafe_math_optimizations.
1221 2005-01-14 J. D. Johnston <jjohnst@us.ibm.com>
1223 * config/s390/tpf-unwind.h (s390_fallback_frame_state): Correct
1225 (__tpf_eh_return): Copy TPF private stack area from the module
1226 boundary stack frame. Add check for when module addresses are
1227 equal, but no stub address is found.
1229 2005-01-14 Richard Earnshaw <rearnsha@arm.com>
1232 * arm.h (struct machine_function): Add call_via field.
1233 (thumb_call_via_label): Declare.
1234 * arm.c (thumb_call_via_label): New variable.
1235 (thumb_call_reg_needed): New variable.
1236 (arm_output_function_epilogue): For Thumb code, output any per-function
1237 call-indirect trampolines.
1238 (thumb_call_via_reg): New function.
1239 (arm_file_end): New function.
1240 (TARGET_ASM_FILE_END): Call arm_file_end.
1241 (aof_file_end): Likewise.
1242 * arm-protos.h (thumb_call_via_reg): Declare.
1243 * arm.md (call_reg_thumb, call_value_reg_thumb): Call
1244 thumb_call_via_reg in normal case.
1246 2005-01-14 Jakub Jelinek <jakub@redhat.com>
1249 PR rtl-optimization/19348
1250 * recog.c (peephole2_optimize): Do global life update if some peephole
1251 decides it doesn't need at least one of its inputs and that change
1252 influences liveness at the start of the basic block.
1254 * basic-block.h (EXECUTE_IF_AND_COMPL_IN_REG_SET): Needs 2 REGSET
1255 arguments instead of 1.
1257 2005-01-14 Eric Botcazou <ebotcazou@libertysurf.fr>
1260 * varasm.c (initializer_constant_valid_p) <ADDR_EXPR>: Return
1261 zero for nested functions needing a static chain or functions
1262 with a non-constant address.
1264 2005-01-13 Roger Sayle <roger@eyesopen.com>
1266 * simplify-rtx.c (simplify_binary_operation) <AND>: Optimize
1267 (and (sign_extend X) C) into (zero_extend (and X C)).
1269 2005-01-13 David O'Brien <obrien@FreeBSD.org>
1271 * config/freebsd-spec.h: Make KSE pthread lib logic the default.
1273 2005-01-13 Richard Henderson <rth@redhat.com>
1278 * config/i386/i386.md (cmpdf, cmpsf, bunordered, bordered, buneq,
1279 bunge, bungt, bunle, bunlt, bltgt): Enable for TARGET_SSE_MATH,
1280 not just TARGET_SSE.
1281 (cmpfp_i_387): Rename from cmpfp_i. Move after sse patterns.
1282 (cmpfp_i_mixed): Rename from cmpfp_i_sse; use for TARGET_MIX_SSE_I387.
1283 (cmpfp_i_sse): Rename from cmpfp_i_sse_only; use for TARGET_SSE_MATH.
1284 (cmpfp_iu_mixed, cmpfp_iu_sse, cmpfp_iu_387): Similarly.
1285 (fp_jcc_1_mixed, fp_jcc_1_sse, fp_jcc_1_387): Similarly.
1286 (fp_jcc_2_mixed, fp_jcc_2_sse, fp_jcc_2_387): Similarly.
1287 (fp_jcc_3_387, fp_jcc_4_387, fp_jcc_5_387, fp_jcc_6_387,
1288 fp_jcc_7_387, fp_jcc_8_387): Rename from fp_jcc_N.
1289 (movdicc_c_rex64): Rename with '*'.
1290 (movsfcc, movdfcc): Add checks for 387 and sse math to condition.
1291 (movsfcc_1_sse_min, movsfcc_1_sse_max, movsfcc_1_sse): New.
1292 (movsfcc_1_387): Rename from movsfcc_1.
1293 (movdfcc_1_sse_min, movdfcc_1_sse_max, movdfcc_1_sse): New.
1294 (movdfcc_1, movdfcc_1_rex64): Add check for 387.
1295 (sminsf3, smaxsf3, smindf3, smaxdf3): New.
1296 (minsf3, minsf, minsf_nonieee, minsf_sse, mindf3, mindf,
1297 mindf_nonieee, mindf_sse, maxsf3, maxsf, maxsf_nonieee, maxsf_sse,
1298 maxdf3, maxdf, maxdf_nonieee, maxdf_sse, sse_movsfcc, sse_movsfcc_eq,
1299 sse_movdfcc, sse_movdfcc_eq, sse_movsfcc_const0_1,
1300 sse_movsfcc_const0_2, sse_movsfcc_const0_3, sse_movsfcc_const0_4,
1301 sse_movdfcc_const0_1, sse_movdfcc_const0_2, sse_movdfcc_const0_3,
1302 sse_movdfcc_const0_4): Remove.
1303 * config/i386/i386.c (ix86_expand_fp_movcc): For TARGET_SSE_MATH,
1304 recognize min/max early. Update for changed sse cmove patterns.
1305 (ix86_split_sse_movcc): New.
1306 * config/i386/i386-protos.h: Update.
1308 2005-01-13 Steven Bosscher <stevenb@suse.de>
1310 * tree-ssa-dse.c (fix_phi_uses): Use SSA operand iterators.
1311 (fix_stmt_v_may_defs): Likewise.
1313 2005-01-13 Richard Henderson <rth@redhat.com>
1315 * config/i386/i386.c (ix86_expand_fp_absneg_operator): Use elt_mode
1316 for converting the mask.
1318 2005-01-13 David Edelsohn <edelsohn@gnu.org>
1320 * config/rs6000/aix52.h (CPLUSPLUS_CPP_SPEC): Change _XOPEN_SOURCE
1323 2005-01-13 Richard Henderson <rth@redhat.com>
1325 * config/i386/i386.c (IX86_BUILTIN_MOVQ, IX86_BUILTIN_LOADD,
1326 IX86_BUILTIN_STORED, IX86_BUILTIN_MOVQ2DQ,
1327 IX86_BUILTIN_MOVDQ2Q): Remove.
1328 (IX86_BUILTIN_VEC_EXT_V4SI): New.
1329 (ix86_init_mmx_sse_builtins, ix86_expand_builtin): Update to match.
1330 (ix86_expand_vector_extract): For V4S[FI], extract element 0 after
1332 * config/i386/sse.md (sse_concatv2sf): Accept zero operand 2.
1333 (sse2_pextrw): Fix immediate constraint.
1334 (sse2_loadq, sse2_loadq_rex64): Remove.
1335 * config/i386/emmintrin.h (_mm_cvtsi128_si32, _mm_cvtsi128_si64x):
1336 Use __builtin_ia32_vec_ext_<size>.
1337 (_mm_cvtsi32_si128, _mm_cvtsi64x_si128): Use _mm_set_epi<size>.
1339 2005-01-13 Aldy Hernandez <aldyh@redhat.com>
1341 * function.c (assign_parm_setup_block): Look inside original
1342 entry_parm when inspecting PARALLEL.
1344 2005-01-13 Ralf Corsepius <ralf.corsepius@rtems.org>
1345 Joel Sherrill <joel@oarcorp.com>
1348 * gthr-rtems.h (__gthread_recursive_mutex_t): New type.
1349 (__GTHREAD_RECURSIVE_MUTEX_INIT_FUNCTION): Define to
1350 rtems_gxx_recursive_mutex_init.
1351 (__gthread_recursive_mutex_lock): New function.
1352 (__gthread_recursive_mutex_trylock): Likewise.
1353 (__gthread_recursive_mutex_unlock): Likewise.
1355 2005-01-13 Ralf Corsepius <ralf.corsepius@rtems.org>
1357 * config/i386/t-rtems-i386: Multilib on -mtune instead of -mcpu.
1359 2005-01-13 Jan Beulich <jbeulich@novell.com>
1361 * config/ia64/ia64.md (ashlti3, ashlti3_internal): New.
1362 (ashrti3_internal): Indicate output is early clobber. Generate result
1363 into output rather than first input. Use move for low word of output
1364 if shift count is exactly 64.
1365 (lshrti3_internal): Likewise.
1367 2005-01-13 Hans-Peter Nilsson <hp@bitrange.com>
1371 * reload1.c (delete_output_reload): Don't delete an output reload
1372 if the pseudo lives longer than a single basic block. Adjust and
1373 improve wording of comment.
1375 2005-01-12 Andrew Pinski <pinskia@physics.uc.edu>
1378 * config/darwin.c (machopic_select_section): Use TYPE_SIZE_UNIT instead
1379 of TYPE_SIZE where we mean the number of bytes.
1381 2005-01-12 Aldy Hernandez <aldyh@redhat.com>
1383 * function.c (assign_parm_setup_block): Relax condition on
1384 multi-register optimization.
1386 2005-01-12 Nick Clifton <nickc@redhat.com>
1388 * config/sh/sh.md (udivsi3_sh2a, divsi3_sh2a): Give these patterns
1389 an "in_delay_slot" attribute of "no" to prevent them being used in
1390 delay slots. This is forbidden because they might generate
1393 2005-01-12 Alan Modra <amodra@bigpond.net.au>
1396 * config/rs6000/rs6000.md (movtf_internal): Replace r->o and m->r
1399 2005-01-12 Nick Clifton <nickc@redhat.com>
1401 * config/iq2000/iq2000.h (ASM_SPEC): Undefine (to stop -Qy being
1402 passed on to GAS) but do not define, as GAS no longer supports or
1403 needs the -m2000 option.
1405 2005-01-11 Roger Sayle <roger@eyesopen.com>
1407 * builtins.c (fold_builtin_fabs): Convert argument to the appropriate
1408 type as args of unprototyped builtins aren't automatically promoted.
1409 (fold_builtin_abs): Likewise.
1411 2005-01-11 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
1413 * Makefile.in: Set a `build-warn' variable.
1415 2005-01-11 Zdenek Dvorak <dvorakz@suse.cz>
1417 PR tree-optimization/17949
1418 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): New function.
1419 (find_interesting_uses_address): Use it.
1421 2005-01-11 Aldy Hernandez <aldyh@redhat.com>
1423 * regrename.c (kill_value): Handle subreg's that won't simplify.
1425 2005-01-11 Richard Henderson <rth@redhat.com>
1428 * config/i386/i386.h (enum ix86_builtins): Move ...
1429 * config/i386/i386.c: ... here.
1430 (IX86_BUILTIN_MOVDDUP, IX86_BUILTIN_MMX_ZERO, IX86_BUILTIN_PEXTRW,
1431 IX86_BUILTIN_PINSRW, IX86_BUILTIN_LOADAPS, IX86_BUILTIN_LOADSS,
1432 IX86_BUILTIN_STORESS, IX86_BUILTIN_SSE_ZERO, IX86_BUILTIN_PEXTRW128,
1433 IX86_BUILTIN_PINSRW128, IX86_BUILTIN_LOADAPD, IX86_BUILTIN_LOADSD,
1434 IX86_BUILTIN_STOREAPD, IX86_BUILTIN_STORESD, IX86_BUILTIN_STOREHPD,
1435 IX86_BUILTIN_STORELPD, IX86_BUILTIN_SETPD1, IX86_BUILTIN_SETPD,
1436 IX86_BUILTIN_CLRPD, IX86_BUILTIN_LOADPD1, IX86_BUILTIN_LOADRPD,
1437 IX86_BUILTIN_STOREPD1, IX86_BUILTIN_STORERPD, IX86_BUILTIN_LOADDQA,
1438 IX86_BUILTIN_STOREDQA, IX86_BUILTIN_CLRTI,
1439 IX86_BUILTIN_LOADDDUP): Remove.
1440 (IX86_BUILTIN_VEC_INIT_V2SI, IX86_BUILTIN_VEC_INIT_V4HI,
1441 IX86_BUILTIN_VEC_INIT_V8QI, IX86_BUILTIN_VEC_EXT_V2DF,
1442 IX86_BUILTIN_VEC_EXT_V2DI, IX86_BUILTIN_VEC_EXT_V4SF,
1443 IX86_BUILTIN_VEC_EXT_V8HI, IX86_BUILTIN_VEC_EXT_V4HI,
1444 IX86_BUILTIN_VEC_SET_V8HI, IX86_BUILTIN_VEC_SET_V4HI): New.
1445 (ix86_init_builtins): Make static.
1446 (ix86_init_mmx_sse_builtins): Update for changed builtins.
1447 (ix86_expand_binop_builtin): Only use ix86_fixup_binary_operands
1448 if all the modes match. Otherwise, fake it.
1449 (get_element_number, ix86_expand_vec_init_builtin,
1450 ix86_expand_vec_ext_builtin, ix86_expand_vec_set_builtin): New.
1451 (ix86_expand_builtin): Make static. Update for changed builtins.
1452 (ix86_expand_vector_move_misalign): Use sse2_loadlpd with zero
1453 operand instead of sse2_loadsd. Cast sse1 fallback to V4SFmode.
1454 (ix86_expand_vector_init_duplicate): New.
1455 (ix86_expand_vector_init_low_nonzero): New.
1456 (ix86_expand_vector_init_one_var, ix86_expand_vector_init_general):
1457 Split out from ix86_expand_vector_init; handle integer modes.
1458 (ix86_expand_vector_init): Use them.
1459 (ix86_expand_vector_set, ix86_expand_vector_extract): New.
1460 * config/i386/i386-protos.h: Update.
1461 * config/i386/predicates.md (reg_or_0_operand): New.
1462 * config/i386/mmx.md (mov<MMXMODEI>_internal): Add 'r' variants.
1463 (movv2sf_internal): Likewise. And a splitter to match them all.
1464 (vec_dupv2sf, mmx_concatv2sf, vec_setv2sf, vec_extractv2sf,
1465 vec_initv2sf, vec_dupv4hi, vec_dupv2si, mmx_concatv2si, vec_setv2si,
1466 vec_extractv2si, vec_initv2si, vec_setv4hi, vec_extractv4hi,
1467 vec_initv4hi, vec_setv8qi, vec_extractv8qi, vec_initv8qi): New.
1468 (mmx_pinsrw): Fix operand ordering.
1469 * config/i386/sse.md (movv4sf splitter): Use direct pattern,
1470 rather than sse_loadss expander.
1471 (movv2df splitter): Similarly.
1472 (sse_loadss, sse_loadlss): Remove.
1473 (vec_dupv4sf, sse_concatv2sf, sse_concatv4sf, vec_extractv4sf_0): New.
1474 (vec_setv4sf, vec_setv2df): Use ix86_expand_vector_set.
1475 (vec_extractv4sf, vec_extractv2df): Use ix86_expand_vector_extract.
1476 (sse3_movddup): Rename with '*'.
1477 (sse3_movddup splitter): Use gen_rtx_REG instead of gen_lowpart.
1478 (sse2_loadsd): Remove.
1479 (vec_dupv2df_sse3): Rename from sse3_loadddup.
1480 (vec_dupv2df, vec_concatv2df_sse3, vec_concatv2df): New.
1481 (sse2_pinsrw): Fix argument ordering.
1482 (sse2_loadld, sse2_loadq): Add sse1 alternatives.
1483 (sse2_stored): Remove 'r' destination.
1484 (vec_dupv4si, vec_dupv2di, sse2_concatv2si, sse1_concatv2si,
1485 vec_concatv4si_1, vec_concatv2di, vec_setv2di, vec_extractv2di,
1486 vec_initv2di, vec_setv4si, vec_extractv4si, vec_initv4si,
1487 vec_setv8hi, vec_extractv8hi, vec_initv8hi, vec_setv16qi,
1488 vec_extractv16qi, vec_initv16qi): New.
1490 * config/i386/emmintrin.h (__m128i, __m128d): Use typedef, not define.
1491 (_mm_set_sd, _mm_set1_pd, _mm_setzero_pd, _mm_set_epi64x,
1492 _mm_set_epi32, _mm_set_epi16, _mm_set_epi8, _mm_setzero_si128): Use
1494 (_mm_load_pd, _mm_store_pd): Use plain dereference.
1495 (_mm_load_si128, _mm_store_si128): Likewise.
1496 (_mm_load1_pd): Use _mm_set1_pd.
1497 (_mm_load_sd): Use _mm_set_sd.
1498 (_mm_store_sd, _mm_storeh_pd): Use __builtin_ia32_vec_ext_v2df.
1499 (_mm_store1_pd, _mm_storer_pd): Use _mm_store_pd.
1500 (_mm_set_epi64): Use _mm_set_epi64x.
1501 (_mm_set1_epi64x, _mm_set1_epi64, _mm_set1_epi32, _mm_set_epi16,
1502 _mm_set1_epi8, _mm_setr_epi64, _mm_setr_epi32, _mm_setr_epi16,
1503 _mm_setr_epi8): Use _mm_set_foo form.
1504 (_mm_loadl_epi64, _mm_movpi64_epi64, _mm_move_epi64): Use _mm_set_epi64.
1505 (_mm_storel_epi64, _mm_movepi64_pi64): Use __builtin_ia32_vec_ext_v2di.
1506 (_mm_extract_epi16): Use __builtin_ia32_vec_ext_v8hi.
1507 (_mm_insert_epi16): Use __builtin_ia32_vec_set_v8hi.
1508 * config/i386/mmintrin.h (_mm_setzero_si64): Use plain cast.
1509 (_mm_set_pi32): Use __builtin_ia32_vec_init_v2si.
1510 (_mm_set_pi16): Use __builtin_ia32_vec_init_v4hi.
1511 (_mm_set_pi8): Use __builtin_ia32_vec_init_v8qi.
1512 (_mm_set1_pi16, _mm_set1_pi8): Use _mm_set_piN variant.
1513 * config/i386/pmmintrin.h (_mm_loaddup_pd): Use _mm_load1_pd.
1514 (_mm_movedup_pd): Use _mm_shuffle_pd.
1515 * config/i386/xmmintrin.h (_mm_setzero_ps, _mm_set_ss,
1516 _mm_set1_ps, _mm_set_ps, _mm_setr_ps): Use constructor form.
1517 (_mm_cvtpi16_ps, _mm_cvtpu16_ps, _mm_cvtpi8_ps, _mm_cvtpu8_ps,
1518 _mm_cvtps_pi8, _mm_cvtpi32x2_ps): Avoid __builtin_ia32_mmx_zero;
1520 (_mm_load_ss, _mm_load1_ps): Use _mm_set* form.
1521 (_mm_load_ps, _mm_loadr_ps): Use raw dereference.
1522 (_mm_store_ss): Use __builtin_ia32_vec_ext_v4sf.
1523 (_mm_store_ps): Use raw dereference.
1524 (_mm_store1_ps): Use _mm_storeu_ps.
1525 (_mm_storer_ps): Use _mm_store_ps.
1526 (_mm_extract_pi16): Use __builtin_ia32_vec_ext_v4hi.
1527 (_mm_insert_pi16): Use __builtin_ia32_vec_set_v4hi.
1529 2005-01-11 Stan Shebs <shebs@apple.com>
1531 * config/rs6000/rs6000.c (machopic_output_stub): Issue
1532 ldu instead of lwzu if 64-bit -mdynamic-no-pic.
1534 2005-01-11 Andrew Pinski <pinskia@physics.uc.edu>
1537 * config/rs6000/rs6000.c (rs6000_special_round_type_align):
1538 Skip all DECLs except for FIELD_DECLs.
1540 2005-01-11 Andreas Krebbel <krebbel1@de.ibm.com>
1542 * config/s390/s390.c (override_options): Return error if
1543 -mbackchain, -mpacked-stack and -mhard-float are used together.
1544 (s390_va_start): Remove the backchain && packed-stack special case.
1545 (s390_gimplify_va_arg): Likewise.
1546 * doc/invoke.texi: Remove the ABI incompatibility note.
1548 2005-01-11 Andreas Krebbel <krebbel1@de.ibm.com>
1550 * config/s390/s390.c (struct s390_frame_layout): Remove
1552 (s390_frame_info, s390_emit_prologue): Replace occurrences of
1553 save_backchain_p with TARGET_BACKCHAIN.
1555 2005-01-11 Alan Modra <amodra@bigpond.net.au>
1558 * builtins.c (std_gimplify_va_arg_expr): Adjust alignment of *ap.
1559 * expr.h (struct locate_and_pad_arg_data): Add "boundary".
1560 * function.c (locate_and_pad_parm): Set new field.
1561 (assign_parm_find_stack_rtl): Use it instead of FUNCTION_ARG_BOUNDARY.
1562 Tweak where_pad test to include "none". Always set mem align for
1564 (assign_parm_adjust_stack_rtl): Discard stack_parm if alignment
1565 not sufficient for type.
1566 (assign_parm_setup_block): If stack_parm is zero on entry, always
1567 make a new stack local. Block move old stack parm if necessary
1568 to new aligned stack local.
1569 (assign_parm_setup_stack): Use a block move to handle
1570 potentially misaligned entry_parm.
1571 (assign_parms_unsplit_complex): Specify required alignment when
1572 creating stack local.
1573 * calls.c (compute_argument_addresses): Override alignment of stack
1574 arg calculated from its type with the alignment given by
1575 FUNCTION_ARG_BOUNDARY.
1576 (store_one_arg): Likewise.
1578 2005-01-11 Jan Beulich <jbeulich@novell.com>
1580 * config/ia64/ia64.md (zero_extendsidi2): Replace zxt4 by addp4.
1581 Change respective itanium_class attribute to ialu.
1582 (shladdp4_internal): New.
1583 * config/ia64/predicates.md (shladd_log2_operand): New.
1585 2005-01-11 Richard Henderson <rth@redhat.com>
1587 * expr.c (store_constructor): Use rtvec_alloc instead of
1588 alloca+gen_rtvec_v, and an incorrect number passed to alloca.
1590 2005-01-11 Kazu Hirata <kazu@cs.umass.edu>
1592 * config/alpha/alpha.c, config/i386/mmx.md: Fix comment typos.
1594 2005-01-11 Alan Modra <amodra@bigpond.net.au>
1596 * varasm.c (default_section_type_flags_1): Don't set SECTION_SMALL.
1597 * config/ia64/ia64.c (TARGET_SECTION_TYPE_FLAGS): Define.
1598 (TARGET_RWRELOC): Define.
1599 (ia64_rwreloc_section_type_flags): Delete.
1600 (ia64_section_type_flags): New function.
1601 * config/ia64/hpux.h (TARGET_SECTION_TYPE_FLAGS): Don't define.
1602 (TARGET_RWRELOC): Define.
1604 2005-01-10 David Mosberger <davidm@hpl.hp.com>
1607 * config/ia64/ia64.c (process_set): For alloc insn, only call
1608 process_epilogue is !frame_pointer_needed.
1610 2005-01-10 Roger Sayle <roger@eyesopen.com>
1613 * c-common.c (c_common_truthvalue_conversion): TRUTH_NOT_EXPR is a
1614 unary operator and can't be treated as a binary/comparison operator.
1616 2005-01-10 Richard Henderson <rth@redhat.com>
1618 * config/i386/i386.c (ix86_function_value): Use type_natural_mode.
1619 (ix86_return_in_memory): Likewise.
1620 (function_arg_advance): Likewise. Mirror structure in function_arg
1621 for choosing register to advance.
1623 2005-01-10 Kazu Hirata <kazu@cs.umass.edu>
1625 * tree-vectorizer.c, tree.def: Fix comment typos.
1627 2005-01-10 Mark Dettinger <dettinge@de.ibm.com>
1629 * config/s390/s390.c (struct processor_costs): 4 new fields:
1630 ddbr, ddr, debr, der.
1631 (s390_rtx_costs): More precise handling of divide instructions.
1633 2005-01-10 David Edelsohn <edelsohn@gnu.org>
1635 * config/rs6000/t-aix43 (BOOT_LDFLAGS): Define.
1637 2005-01-10 Jan Beulich <jbeulich@novell.com>
1639 * config/ia64/ia64.c (ia64_in_small_data_p): Also handle the section
1640 names resulting from -ffunction-sections/-fdata-sections and linkonce
1642 * varasm.c (default_section_type_flags_1): Also set SECTION_SMALL
1643 based on the section name. Rearrange the section name comparison logic
1644 slightly so that each section name is compared against at most once.
1646 2005-01-10 Ben Elliston <bje@au.ibm.com>
1648 * doc/invoke.texi (Code Gen Options): Add PowerPC to the list of
1649 targets for which -fPIC reduces limitations on the GOT size.
1651 2005-01-09 Falk Hueffner <falk@debian.org>
1653 * fold-const.c (fold): Also handle EXACT_DIV_EXPR when folding
1656 2005-01-09 David Edelsohn <edelsohn@gnu.org>
1659 * collect2.c (main): Set aixrtl_flag for -brtl option.
1660 (resolve_lib_name): Search for .so file extension before .a
1663 2005-01-09 Dorit Naishlos <dorit@il.ibm.com>
1665 * tree-vectorizer.c (vect_enhance_data_refs_alignment): Add dump prints.
1666 (vect_analyze_data_refs_alignment): Add dump prints.
1668 2005-01-09 Ira Rosen <irar@il.ibm.com>
1670 * tree-vectorizer.c (vect_analyze_offset_expr): Use
1671 expr_invariant_in_loop_p.
1672 Initialize outputs first thing in the function.
1673 (vect_update_ivs_after_vectorizer): Call initial_condition_in_loop_num.
1674 (vect_is_simple_iv_evolution): Call initial_condition_in_loop_num.
1675 (vect_analyze_pointer_ref_access): Check that the initial condition of
1676 the access function is loop invariant.
1678 2005-01-09 Richard Henderson <rth@redhat.com>
1680 * config/i386/i386.c (bdesc_2arg): Update names for mmx_ prefixes.
1681 (ix86_expand_builtin): Likewise. Frob MASKMOVQ wrt the input mem
1682 just like MASKMOVDQU. Return plain zero for MMX_ZERO.
1683 * config/i386/i386.md (MMXMODEI, mov<MMXMODEI>,
1684 mov<MMXMODEI>_internal_rex64, mov<MMXMODEI>_internal, movv2sf,
1685 movv2sf_internal_rex64, movv2sf_internal, MMXMODE,
1686 movmisalign<MMXMODE>, mmx_pmovmskb, mmx_maskmovq, mmx_maskmovq_rex,
1687 sse_movntdi, addv8qi3, addv4hi3, addv2si3, mmx_adddi3, ssaddv8qi3,
1688 ssaddv4hi3, usaddv8qi3, usaddv4hi3, subv8qi3, subv4hi3, subv2si3,
1689 mmx_subdi3, sssubv8qi3, sssubv4hi3, ussubv8qi3, ussubv4hi3,
1690 mulv4hi3, smulv4hi3_highpart, umulv4hi3_highpart, mmx_pmaddwd,
1691 sse2_umulsidi3, mmx_iordi3, mmx_xordi3, mmx_anddi3, mmx_nanddi3,
1692 mmx_uavgv8qi3, mmx_uavgv4hi3, mmx_psadbw, mmx_pinsrw, mmx_pinsrw,
1693 mmx_pextrw, mmx_pshufw, eqv8qi3, eqv4hi3, eqv2si3, gtv8qi3, gtv4hi3,
1694 gtv2si3, umaxv8qi3, smaxv4hi3, uminv8qi3, sminv4hi3, ashrv4hi3,
1695 ashrv2si3, lshrv4hi3, lshrv2si3, mmx_lshrdi3, ashlv4hi3, ashlv2si3,
1696 mmx_ashldi3, mmx_packsswb, mmx_packssdw, mmx_packuswb, mmx_punpckhbw,
1697 mmx_punpckhwd, mmx_punpckhdq, mmx_punpcklbw, mmx_punpcklwd,
1698 mmx_punpckldq, emms, addv2sf3, subv2sf3, subrv2sf3, gtv2sf3, gev2sf3,
1699 eqv2sf3, pfmaxv2sf3, pfminv2sf3, mulv2sf3, femms, pf2id, pf2iw,
1700 pfacc, pfnacc, pfpnacc, pi2fw, floatv2si2, pfrcpv2sf2, pfrcpit1v2sf3,
1701 pfrcpit2v2sf3, pfrsqrtv2sf2, pfrsqit1v2sf3, pmulhrwv4hi3, pswapdv2si2,
1702 pswapdv2sf2): Move to mmx.md; rename as necessary with leading
1704 (mmx_clrdi, pavgusb): Remove.
1705 (ldmxcsr, stmxcsr, sfence, sfence_insn): Move to sse.md; rename
1706 with leading sse_ prefix.
1707 * config/i386/sse.md: Receive them.
1708 * config/i386/mmx.md: New file.
1709 (MMXMODE12, MMXMODE24, mmxvecsize): New.
1710 (subrv2sf3): Turn into expander for normal subtraction.
1711 (mmx_addv2sf3, mmx_mulv2sf3, mmx_smaxv2sf3, mmx_sminv2sf3,
1712 mmx_eqv2sf3, mmx_mulv4hi3, mmx_smulv4hi3_highpart,
1713 mmx_umulv4hi3_highpart, mmx_pmaddwd, mmx_pmulhrwv4hi3, sse2_umulsidi3,
1714 mmx_umaxv8qi3, mmx_smaxv4hi3, mmx_uminv8qi3, mmx_sminv4hi3): Mark
1715 commutative; use ix86_binary_operator_ok.
1716 (mmx_add<MMXMODEI>3, mmx_ssadd<MMXMODE12>3, mmx_usadd<MMXMODE12>3,
1717 mmx_sub<MMXMODEI>3, mmx_sssub<MMXMODE12>3, mmx_ussub<MMXMODE12>3
1718 mmx_ashr<MMXMODE24>3, mmx_lshr<MMXMODE23>3, mmx_ashl<MMXMODE24>3
1719 mmx_eq<MMXMODEI>3, mmx_gt<MMXMODEI>3, mmx_and<MMXMODEI>3,
1720 mmx_nand<MMXMODEI>3, mmx_ior<MMXMODEI>3, mmx_xor<MMXMODEI>3):
1721 Macroize from existing patterns; use ix86_binary_operator_ok.
1722 (mmx_packsswb, mmx_packssdw, mmx_packuswb): Add memory alternative.
1723 (mmx_punpckhbw, mmx_punpcklbw, mmx_punpckhwd, mmx_punpcklwd,
1724 mmx_punpckhdq, mmx_punpckhdq, mmx_punpckldq): Likewise. Model
1725 with vec_select+vec_concat.
1726 (mmx_pshufw, mmx_pshufw_1): Likewise.
1727 (mmx_uavgv8qi3): Merge pavgusb. Model correcty.
1728 (mmx_uavgv4hi3): Model correctly.
1729 * config/i386/mmintrin.h (_mm_and_si64, _mm_andnot_si64, _mm_or_si64,
1730 _mm_xor_si64): Remove casts.
1732 2005-01-09 Zdenek Dvorak <dvorakz@suse.cz>
1734 PR tree-optimization/19224
1735 * tree-scalar-evolution.c (get_instantiated_value,
1736 set_instantiated_value): New functions.
1737 (instantiate_parameters_1): Cache the results.
1738 (instantiate_parameters, resolve_mixers): Initialize and free
1741 2005-01-08 David Edelsohn <edelsohn@gnu.org>
1743 * config/i386/i386.md (addhi_4): Correct reference in comment.
1746 2005-01-08 Richard Henderson <rth@redhat.com>
1748 * config/i386/emmintrin.h (_mm_cvtsi128_si32): Move earlier.
1749 (_mm_cvtsi128_si64x): Likewise.
1750 (_mm_srl_epi64, _mm_srl_epi32, _mm_srl_epi16, _mm_sra_epi32,
1751 _mm_sra_epi16, _mm_sll_epi64, _mm_sll_epi32, _mm_sll_epi16): Use
1752 the _mm_{srl,sll}i_foo counterpart, and _mm_cvtsi128_si32.
1753 * config/i386/i386-modes.def: Add V16HI, V32QI, V4DF, V8SF.
1754 * config/i386/i386-protos.h: Update.
1755 * config/i386/i386.c (print_operand): Add 'H'.
1756 (ix86_fixup_binary_operands): Split out from ...
1757 (ix86_expand_binary_operator): ... here.
1758 (ix86_fixup_binary_operands_no_copy): New.
1759 (ix86_expand_fp_absneg_operator): Handle vector mode results.
1760 (bdesc_2arg): Update names for sse{,2,3}_ prefixes.
1761 (ix86_init_mmx_sse_builtins): Remove *maskncmp* special cases.
1762 (safe_vector_operand): Use CONST0_RTX.
1763 (ix86_expand_binop_builtin): Use ix86_fixup_binary_operands.
1764 (ix86_expand_builtin): Merge CODE_FOR_sse2_maskmovdqu_rex64 and
1765 CODE_FOR_sse2_maskmovdqu. Special case SSE version of MASKMOVDQU
1766 expansion. Update names for sse{,2,3}_ prefixes. Remove *maskncmp*
1768 * config/i386/i386.h (IX86_BUILTIN_CMPNGTSS): New.
1769 (IX86_BUILTIN_CMPNGESS): New.
1770 * config/i386/i386.md (UNSPEC_FIX_NOTRUNC): New.
1771 (attr type): Add sselog1.
1772 (attr unit, attr memory): Handle it.
1773 (movti, movti_internal, movti_rex64): Move near other integer moves.
1774 (movtf, movtf_internal): Move near other fp moves.
1775 (SSEMODE, SSEMODEI, vec_setv2df, vec_extractv2df, vec_initv2df,
1776 vec_setv4sf, vec_extractv4sf, vec_initv4sf, movv4sf, movv4sf_internal,
1777 movv2df, movv2df_internal, mov<SSEMODEI>, mov<SSEMODEI>_internal,
1778 movmisalign<SSEMODE>, sse_movups_1, sse_movmskps, sse_movntv4sf,
1779 sse_movhlps, sse_movlhps, sse_storehps, sse_loadhps, sse_storelps,
1780 sse_loadlps, sse_loadss, sse_loadss_1, sse_movss, sse_storess,
1781 sse_shufps, addv4sf3, vmaddv4sf3, subv4sf3, vmsubv4sf3, negv4sf2,
1782 mulv4sf3, vmmulv4sf3, divv4sf3, vmdivv4sf3, rcpv4sf2, vmrcpv4sf2,
1783 rsqrtv4sf2, vmrsqrtv4sf2, sqrtv4sf2, vmsqrtv4sf2, sse_andv4sf3,
1784 sse_nandv4sf3, sse_iorv4sf3, sse_xorv4sf3, sse2_andv2df3,
1785 sse2_nandv2df3, sse2_iorv2df3, sse2_xorv2df3, sse2_andv2di3,
1786 sse2_nandv2di3, sse2_iorv2di3, sse2_xorv2di3, maskcmpv4sf3,
1787 vmmaskcmpv4sf3, sse_comi, sse_ucomi, sse_unpckhps, sse_unpcklps,
1788 smaxv4sf3, vmsmaxv4sf3, sminv4sf3, vmsminv4sf3, cvtpi2ps, cvtps2pi,
1789 cvttps2pi, cvtsi2ss, cvtsi2ssq, cvtss2si, cvtss2siq, cvttss2si,
1790 cvttss2siq, addv2df3, vmaddv2df3, subv2df3, vmsubv2df3, mulv2df3,
1791 vmmulv2df3, divv2df3, vmdivv2df3, smaxv2df3, vmsmaxv2df3, sminv2df3,
1792 vmsminv2df3, sqrtv2df2, vmsqrtv2df2, maskcmpv2df3, vmmaskcmpv2df3,
1793 sse2_comi, sse2_ucomi, sse2_movmskpd, sse2_pmovmskb, sse2_maskmovdqu,
1794 sse2_maskmovdqu_rex64, sse2_movntv2df, sse2_movntv2di, sse2_movntsi,
1795 cvtdq2ps, cvtps2dq, cvttps2dq, cvtdq2pd, cvtpd2dq, cvttpd2dq,
1796 cvtpd2pi, cvttpd2pi, cvtpi2pd, cvtsd2si, cvtsd2siq, cvttsd2si,
1797 cvttsd2siq, cvtsi2sd, cvtsi2sdq, cvtsd2ss, cvtss2sd, cvtpd2ps,
1798 cvtps2pd, addv16qi3, addv8hi3, addv4si3, addv2di3, ssaddv16qi3,
1799 ssaddv8hi3, usaddv16qi3, usaddv8hi3, subv16qi3, subv8hi3, subv4si3,
1800 subv2di3, sssubv16qi3, sssubv8hi3, ussubv16qi3, ussubv8hi3, mulv8hi3,
1801 smulv8hi3_highpart, umulv8hi3_highpart, sse2_umulsidi3,
1802 sse2_umulv2siv2di3, sse2_pmaddwd, sse2_uavgv16qi3, sse2_uavgv8hi3,
1803 sse2_psadbw, sse2_pinsrw, sse2_pextrw, sse2_pshufd, sse2_pshuflw,
1804 sse2_pshufhw, eqv16qi3, eqv8hi3, eqv4si3, gtv16qi3, gtv8hi3,
1805 gtv4si3, umaxv16qi3, smaxv8hi3, uminv16qi3, sminv8hi3, ashrv8hi3,
1806 ashrv4si3, lshrv8hi3, lshrv4si3, lshrv2di3, ashlv8hi3, ashlv4si3,
1807 ashlv2di3, sse2_ashlti3, sse2_lshrti3, sse2_unpckhpd, sse2_unpcklpd,
1808 sse2_packsswb, sse2_packssdw, sse2_packuswb, sse2_punpckhbw,
1809 sse2_punpckhwd, sse2_punpckhdq, sse2_punpcklbw, sse2_punpcklwd,
1810 sse2_punpckldq, sse2_punpcklqdq, sse2_punpckhqdq, sse2_movupd,
1811 sse2_movdqu, sse2_movdq2q, sse2_movdq2q_rex64, sse2_movq2dq,
1812 sse2_movq2dq_rex64, sse2_loadd, sse2_stored, sse2_storehpd,
1813 sse2_loadhpd, sse2_storelpd, sse2_loadlpd, sse2_movsd, sse2_loadsd,
1814 sse2_loadsd_1, sse2_storesd, sse2_shufpd, sse2_clflush, sse2_mfence,
1815 mfence_insn, sse2_lfence, lfence_insn, mwait, monitor, addsubv4sf3,
1816 addsubv2df3, haddv4sf3, haddv2df3, hsubv4sf3, hsubv2df3, movshdup,
1817 movsldup, lddqu, loadddup, movddup): Move to sse.md. Any with
1818 non-optabs meanings renamed with an "sse{,2,3}_" prefix at the
1820 (SSEPUSH, push<SSEPUSH>): Remove.
1821 (MMXPUSH, push<MMXPUSH>): Remove.
1822 (sse_movaps, sse_movaps_1, sse_movups): Remove.
1823 (sse2_movapd, sse2_movdqa, sse2_movq): Remove.
1824 (sse2_andti3, sse2_nandti3, sse2_iorti3, sse2_xorti3): Remove.
1825 (sse_clrv4sf, sse_clrv2df, sse2_clrti): Remove.
1826 (maskncmpv4sf3, vmmaskncmpv4sf3): Remove.
1827 (maskncmpv2df3, vmmaskncmpv2df3): Remove.
1828 (ashrv8hi3_ti, ashrv4si3_ti, lshrv8hi3_ti, lshrv4si3_ti): Remove.
1829 (lshrv2di3_ti, ashlv8hi3_ti, ashlv4si3_ti, ashlv2di3_ti): Remove.
1830 * config/i386/athlon.md (athlon_sselog_load): Handle sselog1.
1831 (athlon_sselog_load_k8, athlon_sselog, athlon_sselog_k8): Likewise.
1832 * config/i386/ppro.md (ppro_sse_div_V4SF_load): Fix memory attr.
1833 (ppro_sse_log_V4SF_load): Similarly. Handle sselog1.
1834 (ppro_sse_log_V4SF): Handle sselog1.
1835 * config/i386/predicates.md (const_0_to_1_operand): New.
1836 (const_0_to_255_mul_8_operand): New.
1837 (const_1_to_31_operand): Rename from const_int_1_31_operand.
1838 (const_2_to_3_operand, const_4_to_7_operand): New.
1839 * config/i386/sse.md: New file.
1840 (SSEMODE12, SSEMODE24, SSEMODE124, SSEMODE248, ssevecsize): New.
1841 (sse_movups): Rename from sse_movups_1.
1842 (sse_loadlss): Rename from sse_loadss_1.
1843 (andv4sf3, iorv4sf3, xorv4sf3, andv2df3): Remove the sse prefix
1845 (negv4sf2): Use ix86_expand_fp_absneg_operator.
1846 (absv4sf2, negv2df, absv2df): New.
1847 (addv4sf3): Add expander to call ix86_fixup_binary_operands_no_copy.
1848 (subv4sf3, mulv4sf3, divv4sf3, smaxv4sf3, sminv4sf3, andv4sf3,
1849 iorv4sf3, xorv4sf3, addv2df3, subv2df3, mulv2df3, divv2df3,
1850 smaxv2df3, sminv2df3, andv2df3, iorv2df3, xorv2df3, mulv8hi3,
1851 umaxv16qi3, smaxv8hi3, uminv16qi3, sminv8hi3): Likewise.
1852 (sse3_addsubv4sf3): Model correctly.
1853 sse3_haddv4sf3, sse3_hsubv4sf3, sse3_addsubv2df3, sse3_haddv2df3,
1854 sse3_hsubv2df3, sse2_ashlti3, sse2_lshrti3): Likewise.
1855 (sse_movhlps): Model with vec_select+vec_concat.
1856 (sse_movlhps, sse_unpckhps, sse_unpcklps, sse3_movshdup,
1857 sse3_movsldup, sse_shufps, sse_shufps_1, sse2_unpckhpd, sse3_movddup,
1858 sse2_unpcklpd, sse2_shufpd, sse2_shufpd_1, sse2_punpckhbw,
1859 sse2_punpcklbw, sse2_punpckhwd, sse2_punpcklwd, sse2_punpckhdq,
1860 sse2_punpckldq, sse2_punpckhqdq, sse2_punpcklqdq, sse2_pshufd,
1861 sse2_pshufd_1, sse2_pshuflw, sse2_pshuflw_1, sse2_pshufhw,
1862 sse2_pshufhw_1): Likewise.
1863 (neg<SSEMODEI>2, one_cmpl<SSEMODEI>2): New.
1864 (add<SSEMODEI>3, sse2_ssadd<SSEMODE12>3, sse2_usadd<SSEMODE12>3,
1865 sub<SSEMODEI>3, sse2_sssub<SSEMODE12>3, sse2_ussub<SSEMODE12>3,
1866 ashr<SSEMODE24>3, lshr<SSEMODE248>3, sse2_eq<SSEMODE124>3,
1867 sse2_gt<SSEMODDE124>3, and<SSEMODEI>3, sse_nand<SSEMODEI>3,
1868 ior<SSEMODEI>3, xor<SSEMODEI>3): Macroize from existing patterns.
1869 (addv4sf3, sse_vmaddv4sf3, mulv4sf3, sse_vmmulv4sf3, smaxv4sf3,
1870 sse_vmsmaxv4sf3, sminv4sf3, sse_vmsminv4sf3, addv2df3, sse2_vmaddv2df3,
1871 mulv2df3, sse2_vmmulv2df3, smaxv2df3, sse2_vmsmaxv2df3, sminv2df3,
1872 sse2_vmsminv2df3, umaxv16qi3, smaxv8hi3, uminv16qi3
1873 sminv8hi3): Mark commutative
1874 operands. Use ix86_binary_operator_ok.
1875 (sse_unpckhps, sse_unpcklps, sse2_packsswb, sse2_packssdw,
1876 sse2_packuswb, sse2_punpckhbw, sse2_punpcklbw, sse2_punpckhwd,
1877 sse2_punpcklwd, sse2_punpckhdq, sse2_punpckldq, sse2_punpckhqdq,
1878 sse2_punpcklqdq): Allow operand2 in memory.
1879 (sse_movhlps, sse_movlhps, sse2_unpckhpd, sse2_unpcklpd
1880 sse2_movsd): Add memory alternatives.
1881 (sse_storelps): Turn expander into an insn; split after reload.
1882 (sse_storess, sse2_loadhpd, sse2_loadlpd): Add non-xmm inputs.
1883 (sse2_storehpd, sse2_storelpd): Add non-xmm outputs.
1885 2005-01-08 Eric Botcazou <ebotcazou@libertysurf.fr>
1887 * configure.ac (DWARF-2 debug_line): Use objdump.
1888 * configure: Regenerate.
1890 2005-01-08 Jeff Law <law@redhat.com>
1891 Diego Novillo <dnovillo@redhat.com>
1893 PR tree-optimization/18241
1894 * tree-nrv.c (tree_nrv): Ignore volatile return values.
1895 * tree-ssa-dse.c (dse_optimize_stmt): Do not optimize
1896 statements with volatile operands.
1897 * tree-ssa-operands.c (add_stmt_operand): Do add volatile
1898 operands after marking a statement with has_volatile_ops.
1900 2005-01-08 Roger Sayle <roger@eyesopen.com>
1902 * tree.c (int_fits_type_p): Always honor integer constant
1903 TYPE_MIN_VALUE and TYPE_MAX_VALUE if they exist.
1905 2005-01-08 Roger Sayle <roger@eyesopen.com>
1907 * ifcvt.c (find_if_case_1): Reinstate 2005-01-04 change, now that
1908 the latent bug in rtl_delete_block has been resolved.
1910 2005-01-08 Richard Sandiford <rsandifo@redhat.com>
1912 * config/mips/t-iris6 (MULTILIB_DIRNAMES): Use -mabi argument values.
1913 (MULTILIB_OSDIRNAMES): Use the standard lib, lib32 and lib64.
1915 2005-01-08 Richard Sandiford <rsandifo@redhat.com>
1917 * config/mips/t-slibgcc-irix (SHLIB_LINK): Install a copy of the
1918 library as @multilib_dir@/$(SHLIB_SONAME).
1920 2005-01-07 Eric Botcazou <ebotcazou@libertysurf.fr>
1922 * configure.ac (HAVE_AS_OFFSETABLE_LO10): Fix typo.
1923 * configure: Regenerate.
1925 2005-01-07 Jakub Jelinek <jakub@redhat.com>
1927 * c-common.c (handle_mode_attribute): For ENUMERAL_TYPE, also copy
1930 2005-01-07 David Edelsohn <edelsohn@gnu.org>
1933 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address):
1934 Convert non-word aligned offset address using ld/std into
1937 2005-01-07 Richard Henderson <rth@redhat.com>
1939 * config/i386/i386.md (sse_loadhps splitter): Fix operand number typo.
1941 2005-01-07 Richard SAndiford <rsandifo@redhat.com>
1943 * tree.h (IS_EXPR_CODE_CLASS): Use a straight-forward range check.
1945 2005-01-07 Jakub Jelinek <jakub@redhat.com>
1947 PR tree-optimization/19283
1948 * fold-const.c (fold_widened_comparison): Return NULL if shorter_type
1949 is not shorter than the original type.
1951 PR rtl-optimization/19012
1952 * config/i386/i386.md (addqi_1_slp): Set memory attribute.
1954 PR rtl-optimization/18861
1955 * cfgbuild.c (BLOCK_USED_BY_TABLEJUMP): Define.
1956 (FULL_STATE): Define.
1957 (mark_tablejump_edge): New function.
1958 (purge_dead_tablejump_edges): New function.
1959 (find_bb_boundaries): Use it.
1961 PR tree-optimization/18828
1962 * builtins.c (expand_builtin_next_arg): Remove argument and all
1963 the argument checking.
1964 (expand_builtin): Adjust caller.
1965 (expand_builtin_va_start): Likewise. Remove error for too many
1967 (fold_builtin_next_arg): Issue error for too many arguments.
1968 After checking arguments, replace them with magic arguments that
1969 prevent further checking of the args.
1971 2005-01-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
1973 * pa64-hpux.h (STARTFILE_SPEC): Fix typo in spec.
1975 2005-01-06 Roger Sayle <roger@eyesopen.com>
1978 * gcc.c (input_suffix_matches): Tweak the semantics of %{.s:...}
1979 and %{.S:...} (and their negative variants) to test whether the
1980 input file is assembler or pre-processed-assembler independent of
1981 the actual filename extension.
1983 2005-01-06 Roger Sayle <roger@eyesopen.com>
1985 * simplify-rtx.c (simplify_subreg): Simplify truncations of shifts
1986 of sign or zero extended values.
1988 2005-01-06 Geoffrey Keating <geoffk@apple.com>
1990 * c-cppbuiltin.c (builtin_define_float_constants): Set __*_EPSILON__
1991 for IBM long double format correctly.
1993 2005-01-06 Daniel Berlin <dberlin@dberlin.org>
1995 Fix PR tree-optimization/18792
1997 * tree-data-ref.c (build_classic_dist_vector): Change first_loop
1998 to first_loop_depth, and use loop depth instead of loop number.
1999 (build_classic_dir_vector): Ditto.
2000 (compute_data_dependences_for_loop): Use depth, not loop number.
2001 * tree-loop-linear.c (try_interchange_loops): Use loop depth, not loop
2002 number. Pass in loops, instead of loop numbers.
2003 (gather_interchange_stats): Ditto.
2004 (linear_transform_loops): Ditto.
2006 2005-01-06 Richard Sandiford <rsandifo@redhat.com>
2009 * loop.c (get_monotonic_increment, biased_biv_fits_mode_p,
2010 biv_fits_mode_p, extension_within_bounds_p): New functions.
2011 (check_ext_dependent_givs): Use them.
2013 2005-01-06 Roger Sayle <roger@eyesopen.com>
2015 * cfgrtl.c (rtl_delete_block): A basic block may be followed by
2016 more than one barrier, in which case we should delete them all.
2018 2005-01-06 Tobias Schlueter <tobias.schlueter@physik.uni-muenchen.de>
2020 * gcc.c (process_command): Change year in 'gcc --version' to 2005.
2022 2005-01-05 Daniel Berlin <dberlin@dberlin.org>
2024 Fix PR middle-end/19286
2026 * dwarf2out.c (gen_subprogram_die): If we've already tried to
2027 output this subprogram, simply ignore this attempt to do it again.
2028 (add_abstract_origin_attribute): Don't abort trying to add the abstract
2029 origin attribute if it's not possible.
2030 (gen_block_die): Don't ignore subblocks of "unused" blocks.
2031 (decls_for_scope): Ditto.
2032 * gimple-low.c (mark_blocks_with_used_subblocks): Remove.
2033 (mark_used_blocks): Don't call mark_blocks_with_used_subblocks.
2035 2005-01-05 Richard Henderson <rth@redhat.com>
2038 * config/i386/i386.c (BUILTIN_DESC_SWAP_OPERANDS): New.
2039 (bdesc_2arg): Use it.
2040 (ix86_expand_binop_builtin): Force operands into registers
2042 (ix86_expand_unop_builtin, ix86_expand_unop1_builtin,
2043 ix86_expand_sse_compare, ix86_expand_sse_comi,
2044 ix86_expand_builtin): Likewise.
2046 2005-01-05 Richard Henderson <rth@redhat.com>
2048 * config/ia64/ia64.c (rtx_needs_barrier): Handle CONST_VECTOR
2050 * config/ia64/vect.md (mulv8qi3): Re-implement with mix_[rl].
2051 (mulv4hi3): Set itanium_class mmmul.
2052 (fpack_sfxf, fpack_xfsf, fpack_xfxf): Remove.
2053 (fpack): Rename from fpack_sfsf.
2055 2005-01-05 Richard Henderson <rth@redhat.com>
2058 * reload1.c (do_input_reload): Restrict the optimization deleteing
2059 a previous output reload to RELOAD_FOR_INPUT.
2061 2005-01-05 Steven Bosscher <stevenb@suse.de>
2063 * combine.c (expand_compound_operation) <ZERO_EXTRACT>: Add
2064 comment that we fall through after case.
2065 (mark_used_regs_combine): Don't expect a SIGN_EXTRACT in a SET_DEST.
2066 (distribute_links): Likewise.
2067 * cse.c (cse_insn): Likewise.
2068 * cselib.c (cselib_invalidate_mem): Likewise.
2069 * df.c: Update comments at the top of the file.
2070 (read_modify_subreg_p): Update comments here too.
2071 (df_def_record_1): Don't expect a SIGN_EXTRACT in a SET_DEST.
2072 * flow.c (mark_set_1): Likewise.
2073 (mark_used_regs): Likewise.
2074 * gcse.c (mems_conflict_for_gcse_p): Likewise.
2075 (canon_list_insert): Likewise.
2076 (mark_set): Likewise.
2077 (try_replace_reg): Likewise.
2078 (store_killed_in_insn): Likewise.
2079 * loop.c (count_one_set): Likewise.
2080 (basic_induction_var): Likewise.
2081 * postreload-gcse.c (find_mem_conflicts): Likewise.
2082 * postreload.c (reload_combine_note_store): Likewise.
2083 (move2add_note_store): Likewise.
2084 * reload.c (find_equiv_reg): Likewise.
2085 (mark_referenced_resources): Likewise.
2086 * rtlanal.c (set_noop_p): Likewise.
2087 (note_stores): Likewise.
2088 (note_uses): Likewise.
2089 * sched-deps.c (sched_analyze_1): Likewise.
2090 * sched-rgn.c (check_live_1): Likewise.
2091 (update_live_1): Likewise.
2092 * config/i860/i860.c: Likewise.
2094 * rtl.dec (SIGN_EXTRACT): Document that this cannot appear as
2096 (ZERO_EXTRACT): Mention that this one can be an lvalue.
2098 * doc/rtl.texi: Update documentation for bit-fields and SET.
2100 2005-01-05 Roger Sayle <roger@eyesopen.com>
2102 * ifcvt.c (find_if_case_1): Revert 2005-01-04 change.
2104 2005-01-05 Jan Hubicka <jh@suse.cz>
2105 Richard Henderson <rth@redhat.com>
2108 * config/i386/i386.c (ix86_expand_move): Handle tls symbols
2111 2005-01-05 Richard Henderson <rth@redhat.com>
2114 * config/i386/i386.md (sse_movhps, sse_movlps): Remove.
2115 (sse_shufps): Change operand 3 to const_int_operand.
2116 (sse2_storelps): Fix typo in template.
2117 (sse_storehps, sse_loadhps, sse_storelps, sse_loadlps): New.
2118 * config/i386/i386.c (ix86_expand_vector_move_misalign): Use them.
2119 (ix86_expand_builtin): Likewise.
2121 2005-01-05 Julian Brown <julian@codesourcery.com>
2123 * config/arm/arm.c (arm_return_in_memory): Treat complex types
2124 as aggregates for AAPCS ABIs.
2126 2005-01-05 Stan Shebs <shebs@apple.com>
2128 * unwind-dw2-fde-darwin.c (examine_objects): Use 64-bit
2129 Mach-O getters if ppc64.
2130 * config/darwin.c (darwin_asm_output_dwarf_delta): Obey
2132 * config/darwin.h (ASM_PREFERRED_EH_DATA_FORMAT): Use
2133 signed four-byte field for global code case.
2134 (STARTFILE_SPEC): Avoid crt2.o for 64-bit compilation.
2136 2005-01-05 Roger Sayle <roger@eyesopen.com>
2139 * c-common.c: Include real.h.
2140 (c_common_truthvalue_conversion): Avoid destructively modifying expr.
2141 Correctly handle TREE_CONSTANT_OVERFLOW for INTEGER_CST.
2142 Correctly handle TREE_CONSTANT_OVERFLOW and NaNs for REAL_CST.
2143 * Makefile.in (c-common.o): Update dependencies.
2145 2005-01-05 Joseph S. Myers <joseph@codesourcery.com>
2147 * c-parse.in (asm_string): Add trailing semicolon.
2149 2005-01-05 Joseph S. Myers <joseph@codesourcery.com>
2151 * c-parse.in (asm_string): New. Don't allow wide strings in
2153 (simple_asm_expr, asm_argument, asm_operand, asm_clobbers): Use
2154 asm_string instead of STRING.
2156 2005-01-05 Joseph S. Myers <joseph@codesourcery.com>
2158 * c-typeck.c (constructor_no_implicit): Remove.
2159 (set_designator, process_init_element): Don't check
2160 constructor_no_implicit.
2162 2005-01-05 J"orn Rennecke <joern.rennecke@st.com>
2163 Kaz Kojima <kkojima@gcc.gnu.org>
2166 * lcm.c (create_pre_exit): New.
2167 (optimize_mode_switching): In MODE_ENTRY / MODE_EXIT case, set
2168 ENTRY_EXIT_EXTRA to 3. Use create_pre_exit.
2170 2004-01-05 Richard Earnshaw <rearnsha@arm.com>
2172 * arm.h (TARGET_OPTIONS): Correctly record -mhard-float and
2173 -msoft-float in target_float_switch.
2174 * arm.c (arm_override_options): Fix processing of target_float_switch.
2176 2004-01-05 Richard Earnshaw <rearnsha@arm.com>
2178 * arm/vfp.md (arm_movsi_vfp): Hide VFP register classes from register
2181 2004-01-05 Uros Bizjak <uros@kss-loka.si>
2183 * doc/invoke.texi (Intel 386 and AMD x86-64 Options):
2184 Replace i387 with 'i386 compiler' in -mfpmath=sse option.
2186 2005-01-04 Roger Sayle <roger@eyesopen.com>
2188 * ifcvt.c (find_if_case_1): Avoid creating an empty forwarder block,
2189 if deleting the then-block allows the test-block to fallthru to the
2192 2005-01-04 Andrew Pinski <pinskia@physics.uc.edu>
2195 * c-decl.c (diagnose_mismatched_decls): Accept "extern inline" declared
2196 after the full declaration if the are in two different TUs.
2198 2005-01-04 Richard Henderson <rth@redhat.com>
2201 * tree-sra.c (generate_one_element_init): Just
2202 call gimplify_and_add.
2203 (generate_element_init): Record the
2204 new referenced variables and mark them for renaming
2205 and split out to ...
2206 (generate_element_init_1): This.
2207 (scalarize_init): Don't call push_gimplify_context/
2208 pop_gimplify_context.
2210 2005-01-04 Geoffrey Keating <geoffk@apple.com>
2212 * toplev.c (get_src_pwd): Handle failure of getpwd().
2214 2005-01-04 Roger Sayle <roger@eyesopen.com>
2216 * fold-const.c (fold_single_bit_test): Delete unreachable handling
2218 (fold): Don't call fold_single_bit_test with a TRUTH_NOT_EXPR, as
2219 all the cases handled by it are inverted by invert_truthvalue.
2221 2005-01-04 Eric Botcazou <ebotcazou@libertysurf.fr>
2223 * config/sparc/sparc.md (save_register_windowdi): Add missing mode.
2224 (save_register_windowsi): Likewise.
2226 2005-01-04 Richard Henderson <rth@redhat.com>
2228 * tree-mudflap.c (mf_varname_tree): Fix thinko setting declname.
2230 2005-01-04 Uros Bizjak <uros@kss-loka.si>
2233 * cse.c (fold_rtx) [RTX_COMPARE, RTX_COMM_COMPARE]: Don't attempt
2234 any simplifications of vector mode comparison operators.
2235 * simplify-rtx.c (simplify_relational_operation): Fix variable name.
2237 2005-01-04 Paolo Bonzini <bonzini@gnu.org>
2238 Devang Patel <dpatel@apple.com>
2240 PR tree-optimization/18308
2241 * tree-if-conv.c (add_to_dst_predicate_list): Gimplify
2242 the operands before creating a new expression.
2243 * dojump.c (do_jump): Make drop_through_label available
2244 for all cases. Add expansion of COND_EXPR.
2246 2005-01-04 Ira Rosen <irar@il.ibm.com>
2248 * tree-vectorizer.c (vect_analyze_offset_expr): Test for
2249 INTEGER_CST instead of TREE_CONSTANT.
2250 (vect_gen_niters_for_prolog_loop): Test for INTEGER_CST
2251 instead of TREE_CONSTANT.
2252 (vect_analyze_pointer_ref_access): Test for INTEGER_CST
2253 instead of TREE_CONSTANT.
2255 2005-01-04 Eric Botcazou <ebotcazou@libertysurf.fr>
2257 * tree-eh.c (replace_goto_queue): Return early if the queue is empty.
2259 2005-01-04 Uros Bizjak <uros@kss-loka.si>
2262 * config/i386/i386.md (*fop_df_1_i387): Disable for TARGET_SSE_MATH.
2263 (*fop_df_1_i387): Disable for (TARGET_SSE2 && TARGET_SSE_MATH).
2265 2005-01-03 Richard Henderson <rth@redhat.com>
2267 * fold-const.c (force_fit_type): Cope with types larger than 2 HWI.
2268 (fold_convert_const_int_from_int, fold_convert_const_int_from_real,
2269 fold_convert_const_real_from_real): Split out from ...
2270 (fold_convert_const): ... here.
2272 2005-01-03 Richard Henderson <rth@redhat.com>
2275 * config/i386/i386.md (movdi_2): Separate SSE1 and SSE2 alternatives.
2276 (mov<MMXMODEI>_internal): Likewise.
2277 (movdf_nointeger): Prefer Y while not preferring, but allowing, x.
2278 Add V2SF case; use it for SSE1; don't use TI.
2279 (movdf_integer): Likewise.
2280 (mov<SSEMODEI>_internal, movti_internal): Force V4SF for SSE1.
2282 2005-01-03 Ira Rosen <irar@il.ibm.com>
2284 * tree-vectorizer.c (vect_strip_conversions): New function.
2285 (vect_analyze_offset_expr): Call vect_strip_conversions. Add
2286 check for binary class.
2288 2005-01-03 Daniel Berlin <dberlin@dberlin.org>
2292 * dwarf2out.c (block_ultimate_origin): Follow decl origin if origin
2294 * gimple-low.c (mark_blocks_with_used_vars): New function.
2295 (mark_blocks_with_used_subblocks): Ditto.
2296 (mark_used_blocks): Ditto.
2297 (pass_mark_used_blocks): New pass.
2298 * tree-inline.c: Include debug.h.
2299 (expand_call_inline): Call outlining_inline_function here.
2300 * tree-optimize.c (init_tree_optimization_passes): Add
2301 pass_mark_used_blocks.
2302 * tree-pass.h (pass_mark_used_blocks): New.
2303 * Makefile.in (tree-inline.o): Add debug.h dependency.
2305 2005-01-03 Geoffrey Keating <geoffk@apple.com>
2307 * config/darwin.c (darwin_handle_weak_import_attribute): Permit
2308 VAR_DECLs to have weak_import attribute.
2310 * config/rs6000/darwin-fallback.c: Use 'ucontext_t' rather than
2313 2004-01-03 Steven Bosscher <stevenb@suse.de>
2315 * passes.c (rest_of_compilation): Don't run regmove if only
2316 flag_expensive_optimizations, require flag_regmove instead.
2318 2005-01-03 Eric Botcazou <ebotcazou@libertysurf.fr>
2320 * config/sparc/litecoff.h (TARGET_OS_CPP_BUILTINS): Do not
2322 * config/sparc/sysv4-only.h (TARGET_OS_CPP_BUILTINS): Likewise.
2323 * config/sparc/rtemself.h (TARGET_SUB_OS_CPP_BUILTINS): Likewise.
2324 * config/sparc/sol2-64.h (TARGET_SUB_OS_CPP_BUILTINS): Delete.
2325 * config/sparc/sp64-elf.h (TARGET_SUB_OS_CPP_BUILTINS): Likewise.
2327 2005-01-03 Richard Henderson <rth@redhat.com>
2329 * config/ia64/ia64.c (TARGET_VECTOR_MODE_SUPPORTED_P): New.
2330 (ia64_const_ok_for_letter_p): New.
2331 (ia64_const_double_ok_for_letter_p): New.
2332 (ia64_extra_constraint): New.
2333 (ia64_expand_vecint_compare): New.
2334 (ia64_expand_vcondu_v2si): New.
2335 (ia64_expand_vecint_cmov): New.
2336 (ia64_expand_vecint_minmax): New.
2337 (ia64_print_operand): Add 'v'.
2338 (ia64_preferred_reload_class): New.
2339 (ia64_vector_mode_supported_p): New.
2340 * config/ia64/ia64.h (UNITS_PER_SIMD_WORD): New.
2341 (PREFERRED_RELOAD_CLASS): Move to function.
2342 (CONST_OK_FOR_LETTER_P): Move to function.
2343 (CONST_DOUBLE_OK_FOR_LETTER_P): Move to function.
2344 (CONSTRAINT_OK_FOR_Q, CONSTRAINT_OK_FOR_R): Remove.
2345 (CONSTRAINT_OK_FOR_S, CONSTRAINT_OK_FOR_T): Remove.
2346 (EXTRA_CONSTRAINT): Move to function.
2347 * config/ia64/ia64.md: Include vect.md.
2348 (itanium_class): Add mmalua.
2350 * config/ia64/itanium1.md (1_mmalua): New. Add it to bypasses.
2352 * config/ia64/itanium2.md (2_mmalua, 2b_mmalua): Similarly.
2353 * config/ia64/predicates.md (gr_reg_or_0_operand): Accept any
2355 (const_int_2bit_operand): New.
2356 (fr_reg_or_0_operand): New.
2357 * config/ia64/ia64-modes.def: Add vector modes.
2358 * config/ia64/ia64-protos.h: Update.
2359 * config/ia64/vect.md: New file.
2361 2005-01-03 Richard Henderson <rth@redhat.com>
2363 * simplify-rtx.c (simplify_binary_operation): Handle VEC_CONCAT.
2365 2005-01-03 Uros Bizjak <uros@kss-loka.si>
2368 * config/i386/i386.c (log1psf2): Change mode of operands[0,1]
2370 (log1pdf2): Change mode of operands[0,1] to DFmode.
2372 2005-01-03 Eric Botcazou <ebotcazou@libertysurf.fr>
2374 * config/sparc/sparc.h (SPARC_RELAXED_ORDERING): Define to false.
2375 * config/sparc/linux.h (SPARC_RELAXED_ORDERING): Define to true.
2376 * config/sparc/linux64.h (SPARC_RELAXED_ORDERING): Likewise.
2377 * config/sparc/sparc.c (TARGET_RELAXED_ORDERING): Define to
2378 SPARC_RELAXED_ORDERING.
2380 2005-01-03 Richard Henderson <rth@redhat.com>
2381 Uros Bizjak <uros@kss-loka.si>
2384 * config/i386/i386.c (ix86_expand_builtin): [IX86_BUILTIN_PINSRW,
2385 IX86_BUILTIN_PINSRW128]: Fix wrong selector range in error message.
2386 * config/i386/i386.md (mmx_pinsrw, sse2_pinsrw): Fix selector
2388 (*mmx_pinsrw, *sse2_pinsrw): New patterns.
2389 * config/i386/i386/predicates.md (const_pow2_1_to_8_operand,
2390 const_pow2_1_to_128_operand): New predicates.
2392 2005-01-02 Greg McGary <greg@mcgary.org>
2394 * tree-mudflap.c (mf_varname_tree): decl_printable_name handles
2396 (mf_xform_derefs_1) [RESULT_DECL, STRING_CST]: Handle as innermost
2397 object. [ptr_type]: Remove unused variable.
2398 Remember nearest addressable array-element or record-component
2399 when checking bitfield components. Tolerate empty BIND_EXPR.
2401 2005-01-01 Richard Henderson <rth@redhat.com>
2403 * tree-vectorizer.c (vect_analyze_offset_expr): Strip conversions
2404 that don't narrow the value. Fail for other conversions.
2406 2005-01-01 Richard Henderson <rth@redhat.com>
2409 * c-decl.c (pop_file_scope): Call maybe_apply_pending_pragma_weaks.
2410 * c-lang.c (finish_file): Don't do it here.
2411 * objc/objc-act.c (objc_finish_file): Likewise.
2413 * cgraph.c (decl_assembler_name_equal): New.
2414 (cgraph_node_for_asm, cgraph_varpool_node_for_asm): New.
2415 (cgraph_varpool_node): Actually link up cgraph_varpool_nodes.
2416 * cgraph.h (struct cgraph_varpool_node): Add next.
2417 (cgraph_node_for_asm, cgraph_varpool_node_for_asm): Declare.
2418 * varasm.c (assemble_alias): Mark the target as needed.
2420 2005-01-01 Andrew Pinski <pinskia@physics.uc.edu>
2423 * function.c (get_arg_pointer_save_area): Use entry_of_function
2424 instead of get_insns.
2426 2005-01-01 Roger Sayle <roger@eyesopen.com>
2427 Andrew Pinski <pinskia@physics.uc.edu>
2428 James E. Wilson <wilson@specifixinc.com>
2430 PR rtl-optimization/12092
2431 * loop.c (emit_prefetch_instructions): Do nothing if PREFETCH_BLOCK
2434 2005-01-01 Roger Sayle <roger@eyesopen.com>
2435 Olivier Hainque <hainque@act-europe.fr>
2437 * tree.c (int_fits_type_p): A narrower type always fits in a
2438 wider one, except for negative values into unsigned types.
2440 2005-01-01 Roger Sayle <roger@eyesopen.com>
2442 * tree.c (int_fits_type_p): Compare the result of force_fit_type
2443 with the original constant rather than require TREE_OVERFLOW.
2445 2005-01-01 Steven Bosscher <stevenb@suse.de>
2448 * c-decl.c (finish_function): If compiling C99, annotate the
2449 compiler generated return with the current file name and line 0.
2450 * tree-cfg.c (remove_useless_stmts_warn_notreached): Only warn if
2451 the source line is greater than 0.
2452 (remove_bb): Likewise.
2454 See ChangeLog.12 for earlier changes.