1 2005-01-15 David Edelsohn <edelsohn@gnu.org>
3 * config/rs6000/aix52.h (CPLUSPLUS_CPP_SPEC): Revert previous change.
5 2005-01-15 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
8 * pa.c (pa_asm_output_aligned_common): Warn if specified alignment
9 exceeds maximum alignment for global common data.
10 * pa64-hpux.h (MAX_OFILE_ALIGNMENT): Define to 32768.
11 * som.h (MAX_OFILE_ALIGNMENT): Likewise.
13 2004-01-15 Roger Sayle <roger@eyesopen.com>
15 * tree-ssa-dom.c (extract_range_from_cond): Correct condition.
17 2004-01-15 Roger Sayle <roger@eyesopen.com>
19 * harg-reg-set.h (reg_class_names): Prototype global array.
20 * regclass.c (reg_class_names): Declare here and initialize to
22 (dump_regclass): Remove local declaration of reg_class_names.
24 * cfg.c (dump_flow_info): Likewise.
25 * ra-debug.c (reg_class_names): Likewise.
26 * regrename.c (reg_class_names): Likewise.
27 * reload.c (reg_class_names): Likewise.
28 * reload1.c (spill_failure): Likewise.
29 * config/m68hc11/m68hc11.c (reg_class_names): Likewise.
31 2005-01-15 Ulrich Weigand <uweigand@de.ibm.com>
33 * config/s390/s390.md ("reload_outti"): Remove predicate for
34 output operand. Abort if operand is not a MEM.
35 ("reload_outdi", "reload_outdf"): Likewise.
37 2005-01-15 Marc Espie <espie@openbsd.org>
39 * config.gcc (*-*-openbsd*): Set HAS_LIBC_R for OpenBSD <= 3.2.
40 (i[34567]86-*-openbsd*): Switch to ELF for OpenBSD >= 3.4.
41 * config/openbsd.h: Fix C++ includes for native configurations.
42 Add proper OS_CPP_BUILTINS. Fix libspec for recent OpenBSD.
43 Add trampoline support.
44 * config/i386/openbsdelf.h: New.
46 2005-01-15 Marc Espie <espie@openbsd.org>
48 * collect2.c (main): Explicitly parse -dynamic-linker option.
50 2005-01-15 Jakub Jelinek <jakub@redhat.com>
52 PR tree-optimization/19060
53 * tree-ssa-dom.c (extract_range_from_cond) <case LT_EXPR, GT_EXPR>:
54 Return 0 if op1 <= TYPE_MIN_VALUE () resp. op1 >= TYPE_MAX_VALUE ().
55 (simplify_cond_and_lookup_avail_expr): Add assert for dummy == 0
56 and handle extract_range_from_cond returning false.
57 * fold-const.c (fold): Optimize comparisons with min/max even for
58 width > HOST_BITS_PER_WIDE_INT.
60 2005-01-15 Ralf Corsepius <ralf.corsepius@rtems.org>
62 * config/mips/rtems.h (MIPS_DEFAULT_GVALUE): Set to 0.
63 * config/mips/t-rtems (MULTILIBS_DIRNAMES,MULTILIB_OPTIONS):
64 Remove little endian multilib variants.
65 Add mips32 multilib variant.
67 2005-01-15 Kazu Hirata <kazu@cs.umass.edu>
69 * explow.c (copy_all_reg, stabilize): Remove.
70 * expr.h: Remove the corresponding prototypes.
72 * rtlanal.c (regs_set_between_p): Remove.
73 * rtl.h: Remove the corresponding prototype.
75 2005-01-14 David Edelsohn <edelsohn@gnu.org>
77 * config/rs6000/rs6000.c (rs6k_nonimmediate_operand): Rename
78 to rs6000_nonimmediate_operand.
79 * config/rs6000/rs6000.h (PREDICATE_CODES): Rename
80 rs6k_nonimmediate_operand to rs6000_nonimmediate_operand.
81 * config/rs6000/rs6000.md (movsi_internal1): Rename predicate to
82 rs6000_nonimmediate_operand.
83 * config/rs6000/spe.md (movdf_e500_double): Rename predicate to
84 rs6000_nonimmediate_operand.
86 2005-01-14 Andrew Pinski <pinskia@physics.uc.edu>
88 * tree-ssa-alias.c (pass_may_alias): Add TODO_verify_stmts.
90 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
92 * config/rs6000/rs6000.h (CLASS_MAX_NREGS): DF goes in 1 register
94 (CANNOT_CHANGE_MODE_CLASS): Restrict DI mode changes on e500v2.
95 (PREDICATE_CODES): Add rs6k_nonimmediate_operand.
97 * config/rs6000/rs6000.c (invalid_e500_subreg): New.
98 (rs6k_nonimmediate_operand): New.
99 (rs6000_legitimate_offset_address_p): Handle DI modes on e500v2
101 (legitimate_lo_sum_address_p): Same.
102 (rs6000_legitimize_address): Same.
103 (rs6000_legitimize_reload_address): Same.
104 (rs6000_legitimate_address): Same.
105 (spe_build_register_parallel): Pass DF and DC modes in a DI
108 * config/rs6000/rs6000.md ("*movsi_internal1"): Change predicate
109 to rs6k_nonimmediate_operand.
111 * config/rs6000/spe.md ("*frob_df_di"): New.
112 ("*frob_di_df"): New.
113 ("*frob_di_df_2"): New.
114 ("*mov_sidf_e500_subreg0"): New.
115 ("*mov_sidf_e500_subreg4"): New.
116 ("*movdf_e500_double"): Change predicate to
117 rs6k_nonimmediate_operand.
119 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
121 * postreload.c (move2add_note_store): Only call
122 trunc_int_for_mode on scalar integers.
124 2005-01-14 Steven Bosscher <stevenb@suse.de>
126 * tree-ssa-dce.c (visited_control_parents): New sbitmap to
127 replace BB_VISITED uses.
128 (find_obviously_necessary_stmts): Don't clear BB_VISITED.
129 (propagate_necessity): Check the bitmap instead of BB_VISITED.
130 (tree_dce_done): Free visited_control_parents.
131 (perform_tree_ssa_dce): Allocate and clear it.
132 * tree-ssa-pre.c (compute_antic_aux): Make non-recursive.
133 (compute_antic): Iterate from here using a DFS. Use an sbitmap
134 instead of BB_VISITED.
136 2005-01-14 Kazu Hirata <kazu@cs.umass.edu>
138 * c-tree.h, coverage.h, langhooks-def.h, optabs.h, output.h,
139 rtl.h, targhooks.h, tree-flow.h, tree-mudflap.h, tree.h:
140 Remove unused prototypes.
142 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
144 * config/rs6000/rs6000.c (rs6000_stack_info): Fix caching of SPE
145 64-bit register usage.
146 (spe_func_has_64bit_regs_p): Add FIXME note.
148 2005-01-14 Aldy Hernandez <aldyh@redhat.com>
150 * config/rs6000/rs6000.c (rs6000_generate_compare): Replace
151 flag_finite_math_only with flag_unsafe_math_optimizations.
153 2005-01-14 J. D. Johnston <jjohnst@us.ibm.com>
155 * config/s390/tpf-unwind.h (s390_fallback_frame_state): Correct
157 (__tpf_eh_return): Copy TPF private stack area from the module
158 boundary stack frame. Add check for when module addresses are
159 equal, but no stub address is found.
161 2005-01-14 Richard Earnshaw <rearnsha@arm.com>
164 * arm.h (struct machine_function): Add call_via field.
165 (thumb_call_via_label): Declare.
166 * arm.c (thumb_call_via_label): New variable.
167 (thumb_call_reg_needed): New variable.
168 (arm_output_function_epilogue): For Thumb code, output any per-function
169 call-indirect trampolines.
170 (thumb_call_via_reg): New function.
171 (arm_file_end): New function.
172 (TARGET_ASM_FILE_END): Call arm_file_end.
173 (aof_file_end): Likewise.
174 * arm-protos.h (thumb_call_via_reg): Declare.
175 * arm.md (call_reg_thumb, call_value_reg_thumb): Call
176 thumb_call_via_reg in normal case.
178 2005-01-14 Jakub Jelinek <jakub@redhat.com>
181 PR rtl-optimization/19348
182 * recog.c (peephole2_optimize): Do global life update if some peephole
183 decides it doesn't need at least one of its inputs and that change
184 influences liveness at the start of the basic block.
186 * basic-block.h (EXECUTE_IF_AND_COMPL_IN_REG_SET): Needs 2 REGSET
187 arguments instead of 1.
189 2005-01-14 Eric Botcazou <ebotcazou@libertysurf.fr>
192 * varasm.c (initializer_constant_valid_p) <ADDR_EXPR>: Return
193 zero for nested functions needing a static chain or functions
194 with a non-constant address.
196 2005-01-13 Roger Sayle <roger@eyesopen.com>
198 * simplify-rtx.c (simplify_binary_operation) <AND>: Optimize
199 (and (sign_extend X) C) into (zero_extend (and X C)).
201 2005-01-13 David O'Brien <obrien@FreeBSD.org>
203 * config/freebsd-spec.h: Make KSE pthread lib logic the default.
205 2005-01-13 Richard Henderson <rth@redhat.com>
210 * config/i386/i386.md (cmpdf, cmpsf, bunordered, bordered, buneq,
211 bunge, bungt, bunle, bunlt, bltgt): Enable for TARGET_SSE_MATH,
213 (cmpfp_i_387): Rename from cmpfp_i. Move after sse patterns.
214 (cmpfp_i_mixed): Rename from cmpfp_i_sse; use for TARGET_MIX_SSE_I387.
215 (cmpfp_i_sse): Rename from cmpfp_i_sse_only; use for TARGET_SSE_MATH.
216 (cmpfp_iu_mixed, cmpfp_iu_sse, cmpfp_iu_387): Similarly.
217 (fp_jcc_1_mixed, fp_jcc_1_sse, fp_jcc_1_387): Similarly.
218 (fp_jcc_2_mixed, fp_jcc_2_sse, fp_jcc_2_387): Similarly.
219 (fp_jcc_3_387, fp_jcc_4_387, fp_jcc_5_387, fp_jcc_6_387,
220 fp_jcc_7_387, fp_jcc_8_387): Rename from fp_jcc_N.
221 (movdicc_c_rex64): Rename with '*'.
222 (movsfcc, movdfcc): Add checks for 387 and sse math to condition.
223 (movsfcc_1_sse_min, movsfcc_1_sse_max, movsfcc_1_sse): New.
224 (movsfcc_1_387): Rename from movsfcc_1.
225 (movdfcc_1_sse_min, movdfcc_1_sse_max, movdfcc_1_sse): New.
226 (movdfcc_1, movdfcc_1_rex64): Add check for 387.
227 (sminsf3, smaxsf3, smindf3, smaxdf3): New.
228 (minsf3, minsf, minsf_nonieee, minsf_sse, mindf3, mindf,
229 mindf_nonieee, mindf_sse, maxsf3, maxsf, maxsf_nonieee, maxsf_sse,
230 maxdf3, maxdf, maxdf_nonieee, maxdf_sse, sse_movsfcc, sse_movsfcc_eq,
231 sse_movdfcc, sse_movdfcc_eq, sse_movsfcc_const0_1,
232 sse_movsfcc_const0_2, sse_movsfcc_const0_3, sse_movsfcc_const0_4,
233 sse_movdfcc_const0_1, sse_movdfcc_const0_2, sse_movdfcc_const0_3,
234 sse_movdfcc_const0_4): Remove.
235 * config/i386/i386.c (ix86_expand_fp_movcc): For TARGET_SSE_MATH,
236 recognize min/max early. Update for changed sse cmove patterns.
237 (ix86_split_sse_movcc): New.
238 * config/i386/i386-protos.h: Update.
240 2005-01-13 Steven Bosscher <stevenb@suse.de>
242 * tree-ssa-dse.c (fix_phi_uses): Use SSA operand iterators.
243 (fix_stmt_v_may_defs): Likewise.
245 2005-01-13 Richard Henderson <rth@redhat.com>
247 * config/i386/i386.c (ix86_expand_fp_absneg_operator): Use elt_mode
248 for converting the mask.
250 2005-01-13 David Edelsohn <edelsohn@gnu.org>
252 * config/rs6000/aix52.h (CPLUSPLUS_CPP_SPEC): Change _XOPEN_SOURCE
255 2005-01-13 Richard Henderson <rth@redhat.com>
257 * config/i386/i386.c (IX86_BUILTIN_MOVQ, IX86_BUILTIN_LOADD,
258 IX86_BUILTIN_STORED, IX86_BUILTIN_MOVQ2DQ,
259 IX86_BUILTIN_MOVDQ2Q): Remove.
260 (IX86_BUILTIN_VEC_EXT_V4SI): New.
261 (ix86_init_mmx_sse_builtins, ix86_expand_builtin): Update to match.
262 (ix86_expand_vector_extract): For V4S[FI], extract element 0 after
264 * config/i386/sse.md (sse_concatv2sf): Accept zero operand 2.
265 (sse2_pextrw): Fix immediate constraint.
266 (sse2_loadq, sse2_loadq_rex64): Remove.
267 * config/i386/emmintrin.h (_mm_cvtsi128_si32, _mm_cvtsi128_si64x):
268 Use __builtin_ia32_vec_ext_<size>.
269 (_mm_cvtsi32_si128, _mm_cvtsi64x_si128): Use _mm_set_epi<size>.
271 2005-01-13 Aldy Hernandez <aldyh@redhat.com>
273 * function.c (assign_parm_setup_block): Look inside original
274 entry_parm when inspecting PARALLEL.
276 2005-01-13 Ralf Corsepius <ralf.corsepius@rtems.org>
277 Joel Sherrill <joel@oarcorp.com>
280 * gthr-rtems.h (__gthread_recursive_mutex_t): New type.
281 (__GTHREAD_RECURSIVE_MUTEX_INIT_FUNCTION): Define to
282 rtems_gxx_recursive_mutex_init.
283 (__gthread_recursive_mutex_lock): New function.
284 (__gthread_recursive_mutex_trylock): Likewise.
285 (__gthread_recursive_mutex_unlock): Likewise.
287 2005-01-13 Ralf Corsepius <ralf.corsepius@rtems.org>
289 * config/i386/t-rtems-i386: Multilib on -mtune instead of -mcpu.
291 2005-01-13 Jan Beulich <jbeulich@novell.com>
293 * config/ia64/ia64.md (ashlti3, ashlti3_internal): New.
294 (ashrti3_internal): Indicate output is early clobber. Generate result
295 into output rather than first input. Use move for low word of output
296 if shift count is exactly 64.
297 (lshrti3_internal): Likewise.
299 2005-01-13 Hans-Peter Nilsson <hp@bitrange.com>
303 * reload1.c (delete_output_reload): Don't delete an output reload
304 if the pseudo lives longer than a single basic block. Adjust and
305 improve wording of comment.
307 2005-01-12 Andrew Pinski <pinskia@physics.uc.edu>
310 * config/darwin.c (machopic_select_section): Use TYPE_SIZE_UNIT instead
311 of TYPE_SIZE where we mean the number of bytes.
313 2005-01-12 Aldy Hernandez <aldyh@redhat.com>
315 * function.c (assign_parm_setup_block): Relax condition on
316 multi-register optimization.
318 2005-01-12 Nick Clifton <nickc@redhat.com>
320 * config/sh/sh.md (udivsi3_sh2a, divsi3_sh2a): Give these patterns
321 an "in_delay_slot" attribute of "no" to prevent them being used in
322 delay slots. This is forbidden because they might generate
325 2005-01-12 Alan Modra <amodra@bigpond.net.au>
328 * config/rs6000/rs6000.md (movtf_internal): Replace r->o and m->r
331 2005-01-12 Nick Clifton <nickc@redhat.com>
333 * config/iq2000/iq2000.h (ASM_SPEC): Undefine (to stop -Qy being
334 passed on to GAS) but do not define, as GAS no longer supports or
335 needs the -m2000 option.
337 2005-01-11 Roger Sayle <roger@eyesopen.com>
339 * builtins.c (fold_builtin_fabs): Convert argument to the appropriate
340 type as args of unprototyped builtins aren't automatically promoted.
341 (fold_builtin_abs): Likewise.
343 2005-01-11 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
345 * Makefile.in: Set a `build-warn' variable.
347 2005-01-11 Zdenek Dvorak <dvorakz@suse.cz>
349 PR tree-optimization/17949
350 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): New function.
351 (find_interesting_uses_address): Use it.
353 2005-01-11 Aldy Hernandez <aldyh@redhat.com>
355 * regrename.c (kill_value): Handle subreg's that won't simplify.
357 2005-01-11 Richard Henderson <rth@redhat.com>
360 * config/i386/i386.h (enum ix86_builtins): Move ...
361 * config/i386/i386.c: ... here.
362 (IX86_BUILTIN_MOVDDUP, IX86_BUILTIN_MMX_ZERO, IX86_BUILTIN_PEXTRW,
363 IX86_BUILTIN_PINSRW, IX86_BUILTIN_LOADAPS, IX86_BUILTIN_LOADSS,
364 IX86_BUILTIN_STORESS, IX86_BUILTIN_SSE_ZERO, IX86_BUILTIN_PEXTRW128,
365 IX86_BUILTIN_PINSRW128, IX86_BUILTIN_LOADAPD, IX86_BUILTIN_LOADSD,
366 IX86_BUILTIN_STOREAPD, IX86_BUILTIN_STORESD, IX86_BUILTIN_STOREHPD,
367 IX86_BUILTIN_STORELPD, IX86_BUILTIN_SETPD1, IX86_BUILTIN_SETPD,
368 IX86_BUILTIN_CLRPD, IX86_BUILTIN_LOADPD1, IX86_BUILTIN_LOADRPD,
369 IX86_BUILTIN_STOREPD1, IX86_BUILTIN_STORERPD, IX86_BUILTIN_LOADDQA,
370 IX86_BUILTIN_STOREDQA, IX86_BUILTIN_CLRTI,
371 IX86_BUILTIN_LOADDDUP): Remove.
372 (IX86_BUILTIN_VEC_INIT_V2SI, IX86_BUILTIN_VEC_INIT_V4HI,
373 IX86_BUILTIN_VEC_INIT_V8QI, IX86_BUILTIN_VEC_EXT_V2DF,
374 IX86_BUILTIN_VEC_EXT_V2DI, IX86_BUILTIN_VEC_EXT_V4SF,
375 IX86_BUILTIN_VEC_EXT_V8HI, IX86_BUILTIN_VEC_EXT_V4HI,
376 IX86_BUILTIN_VEC_SET_V8HI, IX86_BUILTIN_VEC_SET_V4HI): New.
377 (ix86_init_builtins): Make static.
378 (ix86_init_mmx_sse_builtins): Update for changed builtins.
379 (ix86_expand_binop_builtin): Only use ix86_fixup_binary_operands
380 if all the modes match. Otherwise, fake it.
381 (get_element_number, ix86_expand_vec_init_builtin,
382 ix86_expand_vec_ext_builtin, ix86_expand_vec_set_builtin): New.
383 (ix86_expand_builtin): Make static. Update for changed builtins.
384 (ix86_expand_vector_move_misalign): Use sse2_loadlpd with zero
385 operand instead of sse2_loadsd. Cast sse1 fallback to V4SFmode.
386 (ix86_expand_vector_init_duplicate): New.
387 (ix86_expand_vector_init_low_nonzero): New.
388 (ix86_expand_vector_init_one_var, ix86_expand_vector_init_general):
389 Split out from ix86_expand_vector_init; handle integer modes.
390 (ix86_expand_vector_init): Use them.
391 (ix86_expand_vector_set, ix86_expand_vector_extract): New.
392 * config/i386/i386-protos.h: Update.
393 * config/i386/predicates.md (reg_or_0_operand): New.
394 * config/i386/mmx.md (mov<MMXMODEI>_internal): Add 'r' variants.
395 (movv2sf_internal): Likewise. And a splitter to match them all.
396 (vec_dupv2sf, mmx_concatv2sf, vec_setv2sf, vec_extractv2sf,
397 vec_initv2sf, vec_dupv4hi, vec_dupv2si, mmx_concatv2si, vec_setv2si,
398 vec_extractv2si, vec_initv2si, vec_setv4hi, vec_extractv4hi,
399 vec_initv4hi, vec_setv8qi, vec_extractv8qi, vec_initv8qi): New.
400 (mmx_pinsrw): Fix operand ordering.
401 * config/i386/sse.md (movv4sf splitter): Use direct pattern,
402 rather than sse_loadss expander.
403 (movv2df splitter): Similarly.
404 (sse_loadss, sse_loadlss): Remove.
405 (vec_dupv4sf, sse_concatv2sf, sse_concatv4sf, vec_extractv4sf_0): New.
406 (vec_setv4sf, vec_setv2df): Use ix86_expand_vector_set.
407 (vec_extractv4sf, vec_extractv2df): Use ix86_expand_vector_extract.
408 (sse3_movddup): Rename with '*'.
409 (sse3_movddup splitter): Use gen_rtx_REG instead of gen_lowpart.
410 (sse2_loadsd): Remove.
411 (vec_dupv2df_sse3): Rename from sse3_loadddup.
412 (vec_dupv2df, vec_concatv2df_sse3, vec_concatv2df): New.
413 (sse2_pinsrw): Fix argument ordering.
414 (sse2_loadld, sse2_loadq): Add sse1 alternatives.
415 (sse2_stored): Remove 'r' destination.
416 (vec_dupv4si, vec_dupv2di, sse2_concatv2si, sse1_concatv2si,
417 vec_concatv4si_1, vec_concatv2di, vec_setv2di, vec_extractv2di,
418 vec_initv2di, vec_setv4si, vec_extractv4si, vec_initv4si,
419 vec_setv8hi, vec_extractv8hi, vec_initv8hi, vec_setv16qi,
420 vec_extractv16qi, vec_initv16qi): New.
422 * config/i386/emmintrin.h (__m128i, __m128d): Use typedef, not define.
423 (_mm_set_sd, _mm_set1_pd, _mm_setzero_pd, _mm_set_epi64x,
424 _mm_set_epi32, _mm_set_epi16, _mm_set_epi8, _mm_setzero_si128): Use
426 (_mm_load_pd, _mm_store_pd): Use plain dereference.
427 (_mm_load_si128, _mm_store_si128): Likewise.
428 (_mm_load1_pd): Use _mm_set1_pd.
429 (_mm_load_sd): Use _mm_set_sd.
430 (_mm_store_sd, _mm_storeh_pd): Use __builtin_ia32_vec_ext_v2df.
431 (_mm_store1_pd, _mm_storer_pd): Use _mm_store_pd.
432 (_mm_set_epi64): Use _mm_set_epi64x.
433 (_mm_set1_epi64x, _mm_set1_epi64, _mm_set1_epi32, _mm_set_epi16,
434 _mm_set1_epi8, _mm_setr_epi64, _mm_setr_epi32, _mm_setr_epi16,
435 _mm_setr_epi8): Use _mm_set_foo form.
436 (_mm_loadl_epi64, _mm_movpi64_epi64, _mm_move_epi64): Use _mm_set_epi64.
437 (_mm_storel_epi64, _mm_movepi64_pi64): Use __builtin_ia32_vec_ext_v2di.
438 (_mm_extract_epi16): Use __builtin_ia32_vec_ext_v8hi.
439 (_mm_insert_epi16): Use __builtin_ia32_vec_set_v8hi.
440 * config/i386/mmintrin.h (_mm_setzero_si64): Use plain cast.
441 (_mm_set_pi32): Use __builtin_ia32_vec_init_v2si.
442 (_mm_set_pi16): Use __builtin_ia32_vec_init_v4hi.
443 (_mm_set_pi8): Use __builtin_ia32_vec_init_v8qi.
444 (_mm_set1_pi16, _mm_set1_pi8): Use _mm_set_piN variant.
445 * config/i386/pmmintrin.h (_mm_loaddup_pd): Use _mm_load1_pd.
446 (_mm_movedup_pd): Use _mm_shuffle_pd.
447 * config/i386/xmmintrin.h (_mm_setzero_ps, _mm_set_ss,
448 _mm_set1_ps, _mm_set_ps, _mm_setr_ps): Use constructor form.
449 (_mm_cvtpi16_ps, _mm_cvtpu16_ps, _mm_cvtpi8_ps, _mm_cvtpu8_ps,
450 _mm_cvtps_pi8, _mm_cvtpi32x2_ps): Avoid __builtin_ia32_mmx_zero;
452 (_mm_load_ss, _mm_load1_ps): Use _mm_set* form.
453 (_mm_load_ps, _mm_loadr_ps): Use raw dereference.
454 (_mm_store_ss): Use __builtin_ia32_vec_ext_v4sf.
455 (_mm_store_ps): Use raw dereference.
456 (_mm_store1_ps): Use _mm_storeu_ps.
457 (_mm_storer_ps): Use _mm_store_ps.
458 (_mm_extract_pi16): Use __builtin_ia32_vec_ext_v4hi.
459 (_mm_insert_pi16): Use __builtin_ia32_vec_set_v4hi.
461 2005-01-11 Stan Shebs <shebs@apple.com>
463 * config/rs6000/rs6000.c (machopic_output_stub): Issue
464 ldu instead of lwzu if 64-bit -mdynamic-no-pic.
466 2005-01-11 Andrew Pinski <pinskia@physics.uc.edu>
469 * config/rs6000/rs6000.c (rs6000_special_round_type_align):
470 Skip all DECLs except for FIELD_DECLs.
472 2005-01-11 Andreas Krebbel <krebbel1@de.ibm.com>
474 * config/s390/s390.c (override_options): Return error if
475 -mbackchain, -mpacked-stack and -mhard-float are used together.
476 (s390_va_start): Remove the backchain && packed-stack special case.
477 (s390_gimplify_va_arg): Likewise.
478 * doc/invoke.texi: Remove the ABI incompatibility note.
480 2005-01-11 Andreas Krebbel <krebbel1@de.ibm.com>
482 * config/s390/s390.c (struct s390_frame_layout): Remove
484 (s390_frame_info, s390_emit_prologue): Replace occurrences of
485 save_backchain_p with TARGET_BACKCHAIN.
487 2005-01-11 Alan Modra <amodra@bigpond.net.au>
490 * builtins.c (std_gimplify_va_arg_expr): Adjust alignment of *ap.
491 * expr.h (struct locate_and_pad_arg_data): Add "boundary".
492 * function.c (locate_and_pad_parm): Set new field.
493 (assign_parm_find_stack_rtl): Use it instead of FUNCTION_ARG_BOUNDARY.
494 Tweak where_pad test to include "none". Always set mem align for
496 (assign_parm_adjust_stack_rtl): Discard stack_parm if alignment
497 not sufficient for type.
498 (assign_parm_setup_block): If stack_parm is zero on entry, always
499 make a new stack local. Block move old stack parm if necessary
500 to new aligned stack local.
501 (assign_parm_setup_stack): Use a block move to handle
502 potentially misaligned entry_parm.
503 (assign_parms_unsplit_complex): Specify required alignment when
504 creating stack local.
505 * calls.c (compute_argument_addresses): Override alignment of stack
506 arg calculated from its type with the alignment given by
507 FUNCTION_ARG_BOUNDARY.
508 (store_one_arg): Likewise.
510 2005-01-11 Jan Beulich <jbeulich@novell.com>
512 * config/ia64/ia64.md (zero_extendsidi2): Replace zxt4 by addp4.
513 Change respective itanium_class attribute to ialu.
514 (shladdp4_internal): New.
515 * config/ia64/predicates.md (shladd_log2_operand): New.
517 2005-01-11 Richard Henderson <rth@redhat.com>
519 * expr.c (store_constructor): Use rtvec_alloc instead of
520 alloca+gen_rtvec_v, and an incorrect number passed to alloca.
522 2005-01-11 Kazu Hirata <kazu@cs.umass.edu>
524 * config/alpha/alpha.c, config/i386/mmx.md: Fix comment typos.
526 2005-01-11 Alan Modra <amodra@bigpond.net.au>
528 * varasm.c (default_section_type_flags_1): Don't set SECTION_SMALL.
529 * config/ia64/ia64.c (TARGET_SECTION_TYPE_FLAGS): Define.
530 (TARGET_RWRELOC): Define.
531 (ia64_rwreloc_section_type_flags): Delete.
532 (ia64_section_type_flags): New function.
533 * config/ia64/hpux.h (TARGET_SECTION_TYPE_FLAGS): Don't define.
534 (TARGET_RWRELOC): Define.
536 2005-01-10 David Mosberger <davidm@hpl.hp.com>
539 * config/ia64/ia64.c (process_set): For alloc insn, only call
540 process_epilogue is !frame_pointer_needed.
542 2005-01-10 Roger Sayle <roger@eyesopen.com>
545 * c-common.c (c_common_truthvalue_conversion): TRUTH_NOT_EXPR is a
546 unary operator and can't be treated as a binary/comparison operator.
548 2005-01-10 Richard Henderson <rth@redhat.com>
550 * config/i386/i386.c (ix86_function_value): Use type_natural_mode.
551 (ix86_return_in_memory): Likewise.
552 (function_arg_advance): Likewise. Mirror structure in function_arg
553 for choosing register to advance.
555 2005-01-10 Kazu Hirata <kazu@cs.umass.edu>
557 * tree-vectorizer.c, tree.def: Fix comment typos.
559 2005-01-10 Mark Dettinger <dettinge@de.ibm.com>
561 * config/s390/s390.c (struct processor_costs): 4 new fields:
562 ddbr, ddr, debr, der.
563 (s390_rtx_costs): More precise handling of divide instructions.
565 2005-01-10 David Edelsohn <edelsohn@gnu.org>
567 * config/rs6000/t-aix43 (BOOT_LDFLAGS): Define.
569 2005-01-10 Jan Beulich <jbeulich@novell.com>
571 * config/ia64/ia64.c (ia64_in_small_data_p): Also handle the section
572 names resulting from -ffunction-sections/-fdata-sections and linkonce
574 * varasm.c (default_section_type_flags_1): Also set SECTION_SMALL
575 based on the section name. Rearrange the section name comparison logic
576 slightly so that each section name is compared against at most once.
578 2005-01-10 Ben Elliston <bje@au.ibm.com>
580 * doc/invoke.texi (Code Gen Options): Add PowerPC to the list of
581 targets for which -fPIC reduces limitations on the GOT size.
583 2005-01-09 Falk Hueffner <falk@debian.org>
585 * fold-const.c (fold): Also handle EXACT_DIV_EXPR when folding
588 2005-01-09 David Edelsohn <edelsohn@gnu.org>
591 * collect2.c (main): Set aixrtl_flag for -brtl option.
592 (resolve_lib_name): Search for .so file extension before .a
595 2005-01-09 Dorit Naishlos <dorit@il.ibm.com>
597 * tree-vectorizer.c (vect_enhance_data_refs_alignment): Add dump prints.
598 (vect_analyze_data_refs_alignment): Add dump prints.
600 2005-01-09 Ira Rosen <irar@il.ibm.com>
602 * tree-vectorizer.c (vect_analyze_offset_expr): Use
603 expr_invariant_in_loop_p.
604 Initialize outputs first thing in the function.
605 (vect_update_ivs_after_vectorizer): Call initial_condition_in_loop_num.
606 (vect_is_simple_iv_evolution): Call initial_condition_in_loop_num.
607 (vect_analyze_pointer_ref_access): Check that the initial condition of
608 the access function is loop invariant.
610 2005-01-09 Richard Henderson <rth@redhat.com>
612 * config/i386/i386.c (bdesc_2arg): Update names for mmx_ prefixes.
613 (ix86_expand_builtin): Likewise. Frob MASKMOVQ wrt the input mem
614 just like MASKMOVDQU. Return plain zero for MMX_ZERO.
615 * config/i386/i386.md (MMXMODEI, mov<MMXMODEI>,
616 mov<MMXMODEI>_internal_rex64, mov<MMXMODEI>_internal, movv2sf,
617 movv2sf_internal_rex64, movv2sf_internal, MMXMODE,
618 movmisalign<MMXMODE>, mmx_pmovmskb, mmx_maskmovq, mmx_maskmovq_rex,
619 sse_movntdi, addv8qi3, addv4hi3, addv2si3, mmx_adddi3, ssaddv8qi3,
620 ssaddv4hi3, usaddv8qi3, usaddv4hi3, subv8qi3, subv4hi3, subv2si3,
621 mmx_subdi3, sssubv8qi3, sssubv4hi3, ussubv8qi3, ussubv4hi3,
622 mulv4hi3, smulv4hi3_highpart, umulv4hi3_highpart, mmx_pmaddwd,
623 sse2_umulsidi3, mmx_iordi3, mmx_xordi3, mmx_anddi3, mmx_nanddi3,
624 mmx_uavgv8qi3, mmx_uavgv4hi3, mmx_psadbw, mmx_pinsrw, mmx_pinsrw,
625 mmx_pextrw, mmx_pshufw, eqv8qi3, eqv4hi3, eqv2si3, gtv8qi3, gtv4hi3,
626 gtv2si3, umaxv8qi3, smaxv4hi3, uminv8qi3, sminv4hi3, ashrv4hi3,
627 ashrv2si3, lshrv4hi3, lshrv2si3, mmx_lshrdi3, ashlv4hi3, ashlv2si3,
628 mmx_ashldi3, mmx_packsswb, mmx_packssdw, mmx_packuswb, mmx_punpckhbw,
629 mmx_punpckhwd, mmx_punpckhdq, mmx_punpcklbw, mmx_punpcklwd,
630 mmx_punpckldq, emms, addv2sf3, subv2sf3, subrv2sf3, gtv2sf3, gev2sf3,
631 eqv2sf3, pfmaxv2sf3, pfminv2sf3, mulv2sf3, femms, pf2id, pf2iw,
632 pfacc, pfnacc, pfpnacc, pi2fw, floatv2si2, pfrcpv2sf2, pfrcpit1v2sf3,
633 pfrcpit2v2sf3, pfrsqrtv2sf2, pfrsqit1v2sf3, pmulhrwv4hi3, pswapdv2si2,
634 pswapdv2sf2): Move to mmx.md; rename as necessary with leading
636 (mmx_clrdi, pavgusb): Remove.
637 (ldmxcsr, stmxcsr, sfence, sfence_insn): Move to sse.md; rename
638 with leading sse_ prefix.
639 * config/i386/sse.md: Receive them.
640 * config/i386/mmx.md: New file.
641 (MMXMODE12, MMXMODE24, mmxvecsize): New.
642 (subrv2sf3): Turn into expander for normal subtraction.
643 (mmx_addv2sf3, mmx_mulv2sf3, mmx_smaxv2sf3, mmx_sminv2sf3,
644 mmx_eqv2sf3, mmx_mulv4hi3, mmx_smulv4hi3_highpart,
645 mmx_umulv4hi3_highpart, mmx_pmaddwd, mmx_pmulhrwv4hi3, sse2_umulsidi3,
646 mmx_umaxv8qi3, mmx_smaxv4hi3, mmx_uminv8qi3, mmx_sminv4hi3): Mark
647 commutative; use ix86_binary_operator_ok.
648 (mmx_add<MMXMODEI>3, mmx_ssadd<MMXMODE12>3, mmx_usadd<MMXMODE12>3,
649 mmx_sub<MMXMODEI>3, mmx_sssub<MMXMODE12>3, mmx_ussub<MMXMODE12>3
650 mmx_ashr<MMXMODE24>3, mmx_lshr<MMXMODE23>3, mmx_ashl<MMXMODE24>3
651 mmx_eq<MMXMODEI>3, mmx_gt<MMXMODEI>3, mmx_and<MMXMODEI>3,
652 mmx_nand<MMXMODEI>3, mmx_ior<MMXMODEI>3, mmx_xor<MMXMODEI>3):
653 Macroize from existing patterns; use ix86_binary_operator_ok.
654 (mmx_packsswb, mmx_packssdw, mmx_packuswb): Add memory alternative.
655 (mmx_punpckhbw, mmx_punpcklbw, mmx_punpckhwd, mmx_punpcklwd,
656 mmx_punpckhdq, mmx_punpckhdq, mmx_punpckldq): Likewise. Model
657 with vec_select+vec_concat.
658 (mmx_pshufw, mmx_pshufw_1): Likewise.
659 (mmx_uavgv8qi3): Merge pavgusb. Model correcty.
660 (mmx_uavgv4hi3): Model correctly.
661 * config/i386/mmintrin.h (_mm_and_si64, _mm_andnot_si64, _mm_or_si64,
662 _mm_xor_si64): Remove casts.
664 2005-01-09 Zdenek Dvorak <dvorakz@suse.cz>
666 PR tree-optimization/19224
667 * tree-scalar-evolution.c (get_instantiated_value,
668 set_instantiated_value): New functions.
669 (instantiate_parameters_1): Cache the results.
670 (instantiate_parameters, resolve_mixers): Initialize and free
673 2005-01-08 David Edelsohn <edelsohn@gnu.org>
675 * config/i386/i386.md (addhi_4): Correct reference in comment.
678 2005-01-08 Richard Henderson <rth@redhat.com>
680 * config/i386/emmintrin.h (_mm_cvtsi128_si32): Move earlier.
681 (_mm_cvtsi128_si64x): Likewise.
682 (_mm_srl_epi64, _mm_srl_epi32, _mm_srl_epi16, _mm_sra_epi32,
683 _mm_sra_epi16, _mm_sll_epi64, _mm_sll_epi32, _mm_sll_epi16): Use
684 the _mm_{srl,sll}i_foo counterpart, and _mm_cvtsi128_si32.
685 * config/i386/i386-modes.def: Add V16HI, V32QI, V4DF, V8SF.
686 * config/i386/i386-protos.h: Update.
687 * config/i386/i386.c (print_operand): Add 'H'.
688 (ix86_fixup_binary_operands): Split out from ...
689 (ix86_expand_binary_operator): ... here.
690 (ix86_fixup_binary_operands_no_copy): New.
691 (ix86_expand_fp_absneg_operator): Handle vector mode results.
692 (bdesc_2arg): Update names for sse{,2,3}_ prefixes.
693 (ix86_init_mmx_sse_builtins): Remove *maskncmp* special cases.
694 (safe_vector_operand): Use CONST0_RTX.
695 (ix86_expand_binop_builtin): Use ix86_fixup_binary_operands.
696 (ix86_expand_builtin): Merge CODE_FOR_sse2_maskmovdqu_rex64 and
697 CODE_FOR_sse2_maskmovdqu. Special case SSE version of MASKMOVDQU
698 expansion. Update names for sse{,2,3}_ prefixes. Remove *maskncmp*
700 * config/i386/i386.h (IX86_BUILTIN_CMPNGTSS): New.
701 (IX86_BUILTIN_CMPNGESS): New.
702 * config/i386/i386.md (UNSPEC_FIX_NOTRUNC): New.
703 (attr type): Add sselog1.
704 (attr unit, attr memory): Handle it.
705 (movti, movti_internal, movti_rex64): Move near other integer moves.
706 (movtf, movtf_internal): Move near other fp moves.
707 (SSEMODE, SSEMODEI, vec_setv2df, vec_extractv2df, vec_initv2df,
708 vec_setv4sf, vec_extractv4sf, vec_initv4sf, movv4sf, movv4sf_internal,
709 movv2df, movv2df_internal, mov<SSEMODEI>, mov<SSEMODEI>_internal,
710 movmisalign<SSEMODE>, sse_movups_1, sse_movmskps, sse_movntv4sf,
711 sse_movhlps, sse_movlhps, sse_storehps, sse_loadhps, sse_storelps,
712 sse_loadlps, sse_loadss, sse_loadss_1, sse_movss, sse_storess,
713 sse_shufps, addv4sf3, vmaddv4sf3, subv4sf3, vmsubv4sf3, negv4sf2,
714 mulv4sf3, vmmulv4sf3, divv4sf3, vmdivv4sf3, rcpv4sf2, vmrcpv4sf2,
715 rsqrtv4sf2, vmrsqrtv4sf2, sqrtv4sf2, vmsqrtv4sf2, sse_andv4sf3,
716 sse_nandv4sf3, sse_iorv4sf3, sse_xorv4sf3, sse2_andv2df3,
717 sse2_nandv2df3, sse2_iorv2df3, sse2_xorv2df3, sse2_andv2di3,
718 sse2_nandv2di3, sse2_iorv2di3, sse2_xorv2di3, maskcmpv4sf3,
719 vmmaskcmpv4sf3, sse_comi, sse_ucomi, sse_unpckhps, sse_unpcklps,
720 smaxv4sf3, vmsmaxv4sf3, sminv4sf3, vmsminv4sf3, cvtpi2ps, cvtps2pi,
721 cvttps2pi, cvtsi2ss, cvtsi2ssq, cvtss2si, cvtss2siq, cvttss2si,
722 cvttss2siq, addv2df3, vmaddv2df3, subv2df3, vmsubv2df3, mulv2df3,
723 vmmulv2df3, divv2df3, vmdivv2df3, smaxv2df3, vmsmaxv2df3, sminv2df3,
724 vmsminv2df3, sqrtv2df2, vmsqrtv2df2, maskcmpv2df3, vmmaskcmpv2df3,
725 sse2_comi, sse2_ucomi, sse2_movmskpd, sse2_pmovmskb, sse2_maskmovdqu,
726 sse2_maskmovdqu_rex64, sse2_movntv2df, sse2_movntv2di, sse2_movntsi,
727 cvtdq2ps, cvtps2dq, cvttps2dq, cvtdq2pd, cvtpd2dq, cvttpd2dq,
728 cvtpd2pi, cvttpd2pi, cvtpi2pd, cvtsd2si, cvtsd2siq, cvttsd2si,
729 cvttsd2siq, cvtsi2sd, cvtsi2sdq, cvtsd2ss, cvtss2sd, cvtpd2ps,
730 cvtps2pd, addv16qi3, addv8hi3, addv4si3, addv2di3, ssaddv16qi3,
731 ssaddv8hi3, usaddv16qi3, usaddv8hi3, subv16qi3, subv8hi3, subv4si3,
732 subv2di3, sssubv16qi3, sssubv8hi3, ussubv16qi3, ussubv8hi3, mulv8hi3,
733 smulv8hi3_highpart, umulv8hi3_highpart, sse2_umulsidi3,
734 sse2_umulv2siv2di3, sse2_pmaddwd, sse2_uavgv16qi3, sse2_uavgv8hi3,
735 sse2_psadbw, sse2_pinsrw, sse2_pextrw, sse2_pshufd, sse2_pshuflw,
736 sse2_pshufhw, eqv16qi3, eqv8hi3, eqv4si3, gtv16qi3, gtv8hi3,
737 gtv4si3, umaxv16qi3, smaxv8hi3, uminv16qi3, sminv8hi3, ashrv8hi3,
738 ashrv4si3, lshrv8hi3, lshrv4si3, lshrv2di3, ashlv8hi3, ashlv4si3,
739 ashlv2di3, sse2_ashlti3, sse2_lshrti3, sse2_unpckhpd, sse2_unpcklpd,
740 sse2_packsswb, sse2_packssdw, sse2_packuswb, sse2_punpckhbw,
741 sse2_punpckhwd, sse2_punpckhdq, sse2_punpcklbw, sse2_punpcklwd,
742 sse2_punpckldq, sse2_punpcklqdq, sse2_punpckhqdq, sse2_movupd,
743 sse2_movdqu, sse2_movdq2q, sse2_movdq2q_rex64, sse2_movq2dq,
744 sse2_movq2dq_rex64, sse2_loadd, sse2_stored, sse2_storehpd,
745 sse2_loadhpd, sse2_storelpd, sse2_loadlpd, sse2_movsd, sse2_loadsd,
746 sse2_loadsd_1, sse2_storesd, sse2_shufpd, sse2_clflush, sse2_mfence,
747 mfence_insn, sse2_lfence, lfence_insn, mwait, monitor, addsubv4sf3,
748 addsubv2df3, haddv4sf3, haddv2df3, hsubv4sf3, hsubv2df3, movshdup,
749 movsldup, lddqu, loadddup, movddup): Move to sse.md. Any with
750 non-optabs meanings renamed with an "sse{,2,3}_" prefix at the
752 (SSEPUSH, push<SSEPUSH>): Remove.
753 (MMXPUSH, push<MMXPUSH>): Remove.
754 (sse_movaps, sse_movaps_1, sse_movups): Remove.
755 (sse2_movapd, sse2_movdqa, sse2_movq): Remove.
756 (sse2_andti3, sse2_nandti3, sse2_iorti3, sse2_xorti3): Remove.
757 (sse_clrv4sf, sse_clrv2df, sse2_clrti): Remove.
758 (maskncmpv4sf3, vmmaskncmpv4sf3): Remove.
759 (maskncmpv2df3, vmmaskncmpv2df3): Remove.
760 (ashrv8hi3_ti, ashrv4si3_ti, lshrv8hi3_ti, lshrv4si3_ti): Remove.
761 (lshrv2di3_ti, ashlv8hi3_ti, ashlv4si3_ti, ashlv2di3_ti): Remove.
762 * config/i386/athlon.md (athlon_sselog_load): Handle sselog1.
763 (athlon_sselog_load_k8, athlon_sselog, athlon_sselog_k8): Likewise.
764 * config/i386/ppro.md (ppro_sse_div_V4SF_load): Fix memory attr.
765 (ppro_sse_log_V4SF_load): Similarly. Handle sselog1.
766 (ppro_sse_log_V4SF): Handle sselog1.
767 * config/i386/predicates.md (const_0_to_1_operand): New.
768 (const_0_to_255_mul_8_operand): New.
769 (const_1_to_31_operand): Rename from const_int_1_31_operand.
770 (const_2_to_3_operand, const_4_to_7_operand): New.
771 * config/i386/sse.md: New file.
772 (SSEMODE12, SSEMODE24, SSEMODE124, SSEMODE248, ssevecsize): New.
773 (sse_movups): Rename from sse_movups_1.
774 (sse_loadlss): Rename from sse_loadss_1.
775 (andv4sf3, iorv4sf3, xorv4sf3, andv2df3): Remove the sse prefix
777 (negv4sf2): Use ix86_expand_fp_absneg_operator.
778 (absv4sf2, negv2df, absv2df): New.
779 (addv4sf3): Add expander to call ix86_fixup_binary_operands_no_copy.
780 (subv4sf3, mulv4sf3, divv4sf3, smaxv4sf3, sminv4sf3, andv4sf3,
781 iorv4sf3, xorv4sf3, addv2df3, subv2df3, mulv2df3, divv2df3,
782 smaxv2df3, sminv2df3, andv2df3, iorv2df3, xorv2df3, mulv8hi3,
783 umaxv16qi3, smaxv8hi3, uminv16qi3, sminv8hi3): Likewise.
784 (sse3_addsubv4sf3): Model correctly.
785 sse3_haddv4sf3, sse3_hsubv4sf3, sse3_addsubv2df3, sse3_haddv2df3,
786 sse3_hsubv2df3, sse2_ashlti3, sse2_lshrti3): Likewise.
787 (sse_movhlps): Model with vec_select+vec_concat.
788 (sse_movlhps, sse_unpckhps, sse_unpcklps, sse3_movshdup,
789 sse3_movsldup, sse_shufps, sse_shufps_1, sse2_unpckhpd, sse3_movddup,
790 sse2_unpcklpd, sse2_shufpd, sse2_shufpd_1, sse2_punpckhbw,
791 sse2_punpcklbw, sse2_punpckhwd, sse2_punpcklwd, sse2_punpckhdq,
792 sse2_punpckldq, sse2_punpckhqdq, sse2_punpcklqdq, sse2_pshufd,
793 sse2_pshufd_1, sse2_pshuflw, sse2_pshuflw_1, sse2_pshufhw,
794 sse2_pshufhw_1): Likewise.
795 (neg<SSEMODEI>2, one_cmpl<SSEMODEI>2): New.
796 (add<SSEMODEI>3, sse2_ssadd<SSEMODE12>3, sse2_usadd<SSEMODE12>3,
797 sub<SSEMODEI>3, sse2_sssub<SSEMODE12>3, sse2_ussub<SSEMODE12>3,
798 ashr<SSEMODE24>3, lshr<SSEMODE248>3, sse2_eq<SSEMODE124>3,
799 sse2_gt<SSEMODDE124>3, and<SSEMODEI>3, sse_nand<SSEMODEI>3,
800 ior<SSEMODEI>3, xor<SSEMODEI>3): Macroize from existing patterns.
801 (addv4sf3, sse_vmaddv4sf3, mulv4sf3, sse_vmmulv4sf3, smaxv4sf3,
802 sse_vmsmaxv4sf3, sminv4sf3, sse_vmsminv4sf3, addv2df3, sse2_vmaddv2df3,
803 mulv2df3, sse2_vmmulv2df3, smaxv2df3, sse2_vmsmaxv2df3, sminv2df3,
804 sse2_vmsminv2df3, umaxv16qi3, smaxv8hi3, uminv16qi3
805 sminv8hi3): Mark commutative
806 operands. Use ix86_binary_operator_ok.
807 (sse_unpckhps, sse_unpcklps, sse2_packsswb, sse2_packssdw,
808 sse2_packuswb, sse2_punpckhbw, sse2_punpcklbw, sse2_punpckhwd,
809 sse2_punpcklwd, sse2_punpckhdq, sse2_punpckldq, sse2_punpckhqdq,
810 sse2_punpcklqdq): Allow operand2 in memory.
811 (sse_movhlps, sse_movlhps, sse2_unpckhpd, sse2_unpcklpd
812 sse2_movsd): Add memory alternatives.
813 (sse_storelps): Turn expander into an insn; split after reload.
814 (sse_storess, sse2_loadhpd, sse2_loadlpd): Add non-xmm inputs.
815 (sse2_storehpd, sse2_storelpd): Add non-xmm outputs.
817 2005-01-08 Eric Botcazou <ebotcazou@libertysurf.fr>
819 * configure.ac (DWARF-2 debug_line): Use objdump.
820 * configure: Regenerate.
822 2005-01-08 Jeff Law <law@redhat.com>
823 Diego Novillo <dnovillo@redhat.com>
825 PR tree-optimization/18241
826 * tree-nrv.c (tree_nrv): Ignore volatile return values.
827 * tree-ssa-dse.c (dse_optimize_stmt): Do not optimize
828 statements with volatile operands.
829 * tree-ssa-operands.c (add_stmt_operand): Do add volatile
830 operands after marking a statement with has_volatile_ops.
832 2005-01-08 Roger Sayle <roger@eyesopen.com>
834 * tree.c (int_fits_type_p): Always honor integer constant
835 TYPE_MIN_VALUE and TYPE_MAX_VALUE if they exist.
837 2005-01-08 Roger Sayle <roger@eyesopen.com>
839 * ifcvt.c (find_if_case_1): Reinstate 2005-01-04 change, now that
840 the latent bug in rtl_delete_block has been resolved.
842 2005-01-08 Richard Sandiford <rsandifo@redhat.com>
844 * config/mips/t-iris6 (MULTILIB_DIRNAMES): Use -mabi argument values.
845 (MULTILIB_OSDIRNAMES): Use the standard lib, lib32 and lib64.
847 2005-01-08 Richard Sandiford <rsandifo@redhat.com>
849 * config/mips/t-slibgcc-irix (SHLIB_LINK): Install a copy of the
850 library as @multilib_dir@/$(SHLIB_SONAME).
852 2005-01-07 Eric Botcazou <ebotcazou@libertysurf.fr>
854 * configure.ac (HAVE_AS_OFFSETABLE_LO10): Fix typo.
855 * configure: Regenerate.
857 2005-01-07 Jakub Jelinek <jakub@redhat.com>
859 * c-common.c (handle_mode_attribute): For ENUMERAL_TYPE, also copy
862 2005-01-07 David Edelsohn <edelsohn@gnu.org>
865 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address):
866 Convert non-word aligned offset address using ld/std into
869 2005-01-07 Richard Henderson <rth@redhat.com>
871 * config/i386/i386.md (sse_loadhps splitter): Fix operand number typo.
873 2005-01-07 Richard SAndiford <rsandifo@redhat.com>
875 * tree.h (IS_EXPR_CODE_CLASS): Use a straight-forward range check.
877 2005-01-07 Jakub Jelinek <jakub@redhat.com>
879 PR tree-optimization/19283
880 * fold-const.c (fold_widened_comparison): Return NULL if shorter_type
881 is not shorter than the original type.
883 PR rtl-optimization/19012
884 * config/i386/i386.md (addqi_1_slp): Set memory attribute.
886 PR rtl-optimization/18861
887 * cfgbuild.c (BLOCK_USED_BY_TABLEJUMP): Define.
888 (FULL_STATE): Define.
889 (mark_tablejump_edge): New function.
890 (purge_dead_tablejump_edges): New function.
891 (find_bb_boundaries): Use it.
893 PR tree-optimization/18828
894 * builtins.c (expand_builtin_next_arg): Remove argument and all
895 the argument checking.
896 (expand_builtin): Adjust caller.
897 (expand_builtin_va_start): Likewise. Remove error for too many
899 (fold_builtin_next_arg): Issue error for too many arguments.
900 After checking arguments, replace them with magic arguments that
901 prevent further checking of the args.
903 2005-01-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
905 * pa64-hpux.h (STARTFILE_SPEC): Fix typo in spec.
907 2005-01-06 Roger Sayle <roger@eyesopen.com>
910 * gcc.c (input_suffix_matches): Tweak the semantics of %{.s:...}
911 and %{.S:...} (and their negative variants) to test whether the
912 input file is assembler or pre-processed-assembler independent of
913 the actual filename extension.
915 2005-01-06 Roger Sayle <roger@eyesopen.com>
917 * simplify-rtx.c (simplify_subreg): Simplify truncations of shifts
918 of sign or zero extended values.
920 2005-01-06 Geoffrey Keating <geoffk@apple.com>
922 * c-cppbuiltin.c (builtin_define_float_constants): Set __*_EPSILON__
923 for IBM long double format correctly.
925 2005-01-06 Daniel Berlin <dberlin@dberlin.org>
927 Fix PR tree-optimization/18792
929 * tree-data-ref.c (build_classic_dist_vector): Change first_loop
930 to first_loop_depth, and use loop depth instead of loop number.
931 (build_classic_dir_vector): Ditto.
932 (compute_data_dependences_for_loop): Use depth, not loop number.
933 * tree-loop-linear.c (try_interchange_loops): Use loop depth, not loop
934 number. Pass in loops, instead of loop numbers.
935 (gather_interchange_stats): Ditto.
936 (linear_transform_loops): Ditto.
938 2005-01-06 Richard Sandiford <rsandifo@redhat.com>
941 * loop.c (get_monotonic_increment, biased_biv_fits_mode_p,
942 biv_fits_mode_p, extension_within_bounds_p): New functions.
943 (check_ext_dependent_givs): Use them.
945 2005-01-06 Roger Sayle <roger@eyesopen.com>
947 * cfgrtl.c (rtl_delete_block): A basic block may be followed by
948 more than one barrier, in which case we should delete them all.
950 2005-01-06 Tobias Schlueter <tobias.schlueter@physik.uni-muenchen.de>
952 * gcc.c (process_command): Change year in 'gcc --version' to 2005.
954 2005-01-05 Daniel Berlin <dberlin@dberlin.org>
956 Fix PR middle-end/19286
958 * dwarf2out.c (gen_subprogram_die): If we've already tried to
959 output this subprogram, simply ignore this attempt to do it again.
960 (add_abstract_origin_attribute): Don't abort trying to add the abstract
961 origin attribute if it's not possible.
962 (gen_block_die): Don't ignore subblocks of "unused" blocks.
963 (decls_for_scope): Ditto.
964 * gimple-low.c (mark_blocks_with_used_subblocks): Remove.
965 (mark_used_blocks): Don't call mark_blocks_with_used_subblocks.
967 2005-01-05 Richard Henderson <rth@redhat.com>
970 * config/i386/i386.c (BUILTIN_DESC_SWAP_OPERANDS): New.
971 (bdesc_2arg): Use it.
972 (ix86_expand_binop_builtin): Force operands into registers
974 (ix86_expand_unop_builtin, ix86_expand_unop1_builtin,
975 ix86_expand_sse_compare, ix86_expand_sse_comi,
976 ix86_expand_builtin): Likewise.
978 2005-01-05 Richard Henderson <rth@redhat.com>
980 * config/ia64/ia64.c (rtx_needs_barrier): Handle CONST_VECTOR
982 * config/ia64/vect.md (mulv8qi3): Re-implement with mix_[rl].
983 (mulv4hi3): Set itanium_class mmmul.
984 (fpack_sfxf, fpack_xfsf, fpack_xfxf): Remove.
985 (fpack): Rename from fpack_sfsf.
987 2005-01-05 Richard Henderson <rth@redhat.com>
990 * reload1.c (do_input_reload): Restrict the optimization deleteing
991 a previous output reload to RELOAD_FOR_INPUT.
993 2005-01-05 Steven Bosscher <stevenb@suse.de>
995 * combine.c (expand_compound_operation) <ZERO_EXTRACT>: Add
996 comment that we fall through after case.
997 (mark_used_regs_combine): Don't expect a SIGN_EXTRACT in a SET_DEST.
998 (distribute_links): Likewise.
999 * cse.c (cse_insn): Likewise.
1000 * cselib.c (cselib_invalidate_mem): Likewise.
1001 * df.c: Update comments at the top of the file.
1002 (read_modify_subreg_p): Update comments here too.
1003 (df_def_record_1): Don't expect a SIGN_EXTRACT in a SET_DEST.
1004 * flow.c (mark_set_1): Likewise.
1005 (mark_used_regs): Likewise.
1006 * gcse.c (mems_conflict_for_gcse_p): Likewise.
1007 (canon_list_insert): Likewise.
1008 (mark_set): Likewise.
1009 (try_replace_reg): Likewise.
1010 (store_killed_in_insn): Likewise.
1011 * loop.c (count_one_set): Likewise.
1012 (basic_induction_var): Likewise.
1013 * postreload-gcse.c (find_mem_conflicts): Likewise.
1014 * postreload.c (reload_combine_note_store): Likewise.
1015 (move2add_note_store): Likewise.
1016 * reload.c (find_equiv_reg): Likewise.
1017 (mark_referenced_resources): Likewise.
1018 * rtlanal.c (set_noop_p): Likewise.
1019 (note_stores): Likewise.
1020 (note_uses): Likewise.
1021 * sched-deps.c (sched_analyze_1): Likewise.
1022 * sched-rgn.c (check_live_1): Likewise.
1023 (update_live_1): Likewise.
1024 * config/i860/i860.c: Likewise.
1026 * rtl.dec (SIGN_EXTRACT): Document that this cannot appear as
1028 (ZERO_EXTRACT): Mention that this one can be an lvalue.
1030 * doc/rtl.texi: Update documentation for bit-fields and SET.
1032 2005-01-05 Roger Sayle <roger@eyesopen.com>
1034 * ifcvt.c (find_if_case_1): Revert 2005-01-04 change.
1036 2005-01-05 Jan Hubicka <jh@suse.cz>
1037 Richard Henderson <rth@redhat.com>
1040 * config/i386/i386.c (ix86_expand_move): Handle tls symbols
1043 2005-01-05 Richard Henderson <rth@redhat.com>
1046 * config/i386/i386.md (sse_movhps, sse_movlps): Remove.
1047 (sse_shufps): Change operand 3 to const_int_operand.
1048 (sse2_storelps): Fix typo in template.
1049 (sse_storehps, sse_loadhps, sse_storelps, sse_loadlps): New.
1050 * config/i386/i386.c (ix86_expand_vector_move_misalign): Use them.
1051 (ix86_expand_builtin): Likewise.
1053 2005-01-05 Julian Brown <julian@codesourcery.com>
1055 * config/arm/arm.c (arm_return_in_memory): Treat complex types
1056 as aggregates for AAPCS ABIs.
1058 2005-01-05 Stan Shebs <shebs@apple.com>
1060 * unwind-dw2-fde-darwin.c (examine_objects): Use 64-bit
1061 Mach-O getters if ppc64.
1062 * config/darwin.c (darwin_asm_output_dwarf_delta): Obey
1064 * config/darwin.h (ASM_PREFERRED_EH_DATA_FORMAT): Use
1065 signed four-byte field for global code case.
1066 (STARTFILE_SPEC): Avoid crt2.o for 64-bit compilation.
1068 2005-01-05 Roger Sayle <roger@eyesopen.com>
1071 * c-common.c: Include real.h.
1072 (c_common_truthvalue_conversion): Avoid destructively modifying expr.
1073 Correctly handle TREE_CONSTANT_OVERFLOW for INTEGER_CST.
1074 Correctly handle TREE_CONSTANT_OVERFLOW and NaNs for REAL_CST.
1075 * Makefile.in (c-common.o): Update dependencies.
1077 2005-01-05 Joseph S. Myers <joseph@codesourcery.com>
1079 * c-parse.in (asm_string): Add trailing semicolon.
1081 2005-01-05 Joseph S. Myers <joseph@codesourcery.com>
1083 * c-parse.in (asm_string): New. Don't allow wide strings in
1085 (simple_asm_expr, asm_argument, asm_operand, asm_clobbers): Use
1086 asm_string instead of STRING.
1088 2005-01-05 Joseph S. Myers <joseph@codesourcery.com>
1090 * c-typeck.c (constructor_no_implicit): Remove.
1091 (set_designator, process_init_element): Don't check
1092 constructor_no_implicit.
1094 2005-01-05 J"orn Rennecke <joern.rennecke@st.com>
1095 Kaz Kojima <kkojima@gcc.gnu.org>
1098 * lcm.c (create_pre_exit): New.
1099 (optimize_mode_switching): In MODE_ENTRY / MODE_EXIT case, set
1100 ENTRY_EXIT_EXTRA to 3. Use create_pre_exit.
1102 2004-01-05 Richard Earnshaw <rearnsha@arm.com>
1104 * arm.h (TARGET_OPTIONS): Correctly record -mhard-float and
1105 -msoft-float in target_float_switch.
1106 * arm.c (arm_override_options): Fix processing of target_float_switch.
1108 2004-01-05 Richard Earnshaw <rearnsha@arm.com>
1110 * arm/vfp.md (arm_movsi_vfp): Hide VFP register classes from register
1113 2004-01-05 Uros Bizjak <uros@kss-loka.si>
1115 * doc/invoke.texi (Intel 386 and AMD x86-64 Options):
1116 Replace i387 with 'i386 compiler' in -mfpmath=sse option.
1118 2005-01-04 Roger Sayle <roger@eyesopen.com>
1120 * ifcvt.c (find_if_case_1): Avoid creating an empty forwarder block,
1121 if deleting the then-block allows the test-block to fallthru to the
1124 2005-01-04 Andrew Pinski <pinskia@physics.uc.edu>
1127 * c-decl.c (diagnose_mismatched_decls): Accept "extern inline" declared
1128 after the full declaration if the are in two different TUs.
1130 2005-01-04 Richard Henderson <rth@redhat.com>
1133 * tree-sra.c (generate_one_element_init): Just
1134 call gimplify_and_add.
1135 (generate_element_init): Record the
1136 new referenced variables and mark them for renaming
1137 and split out to ...
1138 (generate_element_init_1): This.
1139 (scalarize_init): Don't call push_gimplify_context/
1140 pop_gimplify_context.
1142 2005-01-04 Geoffrey Keating <geoffk@apple.com>
1144 * toplev.c (get_src_pwd): Handle failure of getpwd().
1146 2005-01-04 Roger Sayle <roger@eyesopen.com>
1148 * fold-const.c (fold_single_bit_test): Delete unreachable handling
1150 (fold): Don't call fold_single_bit_test with a TRUTH_NOT_EXPR, as
1151 all the cases handled by it are inverted by invert_truthvalue.
1153 2005-01-04 Eric Botcazou <ebotcazou@libertysurf.fr>
1155 * config/sparc/sparc.md (save_register_windowdi): Add missing mode.
1156 (save_register_windowsi): Likewise.
1158 2005-01-04 Richard Henderson <rth@redhat.com>
1160 * tree-mudflap.c (mf_varname_tree): Fix thinko setting declname.
1162 2005-01-04 Uros Bizjak <uros@kss-loka.si>
1165 * cse.c (fold_rtx) [RTX_COMPARE, RTX_COMM_COMPARE]: Don't attempt
1166 any simplifications of vector mode comparison operators.
1167 * simplify-rtx.c (simplify_relational_operation): Fix variable name.
1169 2005-01-04 Paolo Bonzini <bonzini@gnu.org>
1170 Devang Patel <dpatel@apple.com>
1172 PR tree-optimization/18308
1173 * tree-if-conv.c (add_to_dst_predicate_list): Gimplify
1174 the operands before creating a new expression.
1175 * dojump.c (do_jump): Make drop_through_label available
1176 for all cases. Add expansion of COND_EXPR.
1178 2005-01-04 Ira Rosen <irar@il.ibm.com>
1180 * tree-vectorizer.c (vect_analyze_offset_expr): Test for
1181 INTEGER_CST instead of TREE_CONSTANT.
1182 (vect_gen_niters_for_prolog_loop): Test for INTEGER_CST
1183 instead of TREE_CONSTANT.
1184 (vect_analyze_pointer_ref_access): Test for INTEGER_CST
1185 instead of TREE_CONSTANT.
1187 2005-01-04 Eric Botcazou <ebotcazou@libertysurf.fr>
1189 * tree-eh.c (replace_goto_queue): Return early if the queue is empty.
1191 2005-01-04 Uros Bizjak <uros@kss-loka.si>
1194 * config/i386/i386.md (*fop_df_1_i387): Disable for TARGET_SSE_MATH.
1195 (*fop_df_1_i387): Disable for (TARGET_SSE2 && TARGET_SSE_MATH).
1197 2005-01-03 Richard Henderson <rth@redhat.com>
1199 * fold-const.c (force_fit_type): Cope with types larger than 2 HWI.
1200 (fold_convert_const_int_from_int, fold_convert_const_int_from_real,
1201 fold_convert_const_real_from_real): Split out from ...
1202 (fold_convert_const): ... here.
1204 2005-01-03 Richard Henderson <rth@redhat.com>
1207 * config/i386/i386.md (movdi_2): Separate SSE1 and SSE2 alternatives.
1208 (mov<MMXMODEI>_internal): Likewise.
1209 (movdf_nointeger): Prefer Y while not preferring, but allowing, x.
1210 Add V2SF case; use it for SSE1; don't use TI.
1211 (movdf_integer): Likewise.
1212 (mov<SSEMODEI>_internal, movti_internal): Force V4SF for SSE1.
1214 2005-01-03 Ira Rosen <irar@il.ibm.com>
1216 * tree-vectorizer.c (vect_strip_conversions): New function.
1217 (vect_analyze_offset_expr): Call vect_strip_conversions. Add
1218 check for binary class.
1220 2005-01-03 Daniel Berlin <dberlin@dberlin.org>
1224 * dwarf2out.c (block_ultimate_origin): Follow decl origin if origin
1226 * gimple-low.c (mark_blocks_with_used_vars): New function.
1227 (mark_blocks_with_used_subblocks): Ditto.
1228 (mark_used_blocks): Ditto.
1229 (pass_mark_used_blocks): New pass.
1230 * tree-inline.c: Include debug.h.
1231 (expand_call_inline): Call outlining_inline_function here.
1232 * tree-optimize.c (init_tree_optimization_passes): Add
1233 pass_mark_used_blocks.
1234 * tree-pass.h (pass_mark_used_blocks): New.
1235 * Makefile.in (tree-inline.o): Add debug.h dependency.
1237 2005-01-03 Geoffrey Keating <geoffk@apple.com>
1239 * config/darwin.c (darwin_handle_weak_import_attribute): Permit
1240 VAR_DECLs to have weak_import attribute.
1242 * config/rs6000/darwin-fallback.c: Use 'ucontext_t' rather than
1245 2004-01-03 Steven Bosscher <stevenb@suse.de>
1247 * passes.c (rest_of_compilation): Don't run regmove if only
1248 flag_expensive_optimizations, require flag_regmove instead.
1250 2005-01-03 Eric Botcazou <ebotcazou@libertysurf.fr>
1252 * config/sparc/litecoff.h (TARGET_OS_CPP_BUILTINS): Do not
1254 * config/sparc/sysv4-only.h (TARGET_OS_CPP_BUILTINS): Likewise.
1255 * config/sparc/rtemself.h (TARGET_SUB_OS_CPP_BUILTINS): Likewise.
1256 * config/sparc/sol2-64.h (TARGET_SUB_OS_CPP_BUILTINS): Delete.
1257 * config/sparc/sp64-elf.h (TARGET_SUB_OS_CPP_BUILTINS): Likewise.
1259 2005-01-03 Richard Henderson <rth@redhat.com>
1261 * config/ia64/ia64.c (TARGET_VECTOR_MODE_SUPPORTED_P): New.
1262 (ia64_const_ok_for_letter_p): New.
1263 (ia64_const_double_ok_for_letter_p): New.
1264 (ia64_extra_constraint): New.
1265 (ia64_expand_vecint_compare): New.
1266 (ia64_expand_vcondu_v2si): New.
1267 (ia64_expand_vecint_cmov): New.
1268 (ia64_expand_vecint_minmax): New.
1269 (ia64_print_operand): Add 'v'.
1270 (ia64_preferred_reload_class): New.
1271 (ia64_vector_mode_supported_p): New.
1272 * config/ia64/ia64.h (UNITS_PER_SIMD_WORD): New.
1273 (PREFERRED_RELOAD_CLASS): Move to function.
1274 (CONST_OK_FOR_LETTER_P): Move to function.
1275 (CONST_DOUBLE_OK_FOR_LETTER_P): Move to function.
1276 (CONSTRAINT_OK_FOR_Q, CONSTRAINT_OK_FOR_R): Remove.
1277 (CONSTRAINT_OK_FOR_S, CONSTRAINT_OK_FOR_T): Remove.
1278 (EXTRA_CONSTRAINT): Move to function.
1279 * config/ia64/ia64.md: Include vect.md.
1280 (itanium_class): Add mmalua.
1282 * config/ia64/itanium1.md (1_mmalua): New. Add it to bypasses.
1284 * config/ia64/itanium2.md (2_mmalua, 2b_mmalua): Similarly.
1285 * config/ia64/predicates.md (gr_reg_or_0_operand): Accept any
1287 (const_int_2bit_operand): New.
1288 (fr_reg_or_0_operand): New.
1289 * config/ia64/ia64-modes.def: Add vector modes.
1290 * config/ia64/ia64-protos.h: Update.
1291 * config/ia64/vect.md: New file.
1293 2005-01-03 Richard Henderson <rth@redhat.com>
1295 * simplify-rtx.c (simplify_binary_operation): Handle VEC_CONCAT.
1297 2005-01-03 Uros Bizjak <uros@kss-loka.si>
1300 * config/i386/i386.c (log1psf2): Change mode of operands[0,1]
1302 (log1pdf2): Change mode of operands[0,1] to DFmode.
1304 2005-01-03 Eric Botcazou <ebotcazou@libertysurf.fr>
1306 * config/sparc/sparc.h (SPARC_RELAXED_ORDERING): Define to false.
1307 * config/sparc/linux.h (SPARC_RELAXED_ORDERING): Define to true.
1308 * config/sparc/linux64.h (SPARC_RELAXED_ORDERING): Likewise.
1309 * config/sparc/sparc.c (TARGET_RELAXED_ORDERING): Define to
1310 SPARC_RELAXED_ORDERING.
1312 2005-01-03 Richard Henderson <rth@redhat.com>
1313 Uros Bizjak <uros@kss-loka.si>
1316 * config/i386/i386.c (ix86_expand_builtin): [IX86_BUILTIN_PINSRW,
1317 IX86_BUILTIN_PINSRW128]: Fix wrong selector range in error message.
1318 * config/i386/i386.md (mmx_pinsrw, sse2_pinsrw): Fix selector
1320 (*mmx_pinsrw, *sse2_pinsrw): New patterns.
1321 * config/i386/i386/predicates.md (const_pow2_1_to_8_operand,
1322 const_pow2_1_to_128_operand): New predicates.
1324 2005-01-02 Greg McGary <greg@mcgary.org>
1326 * tree-mudflap.c (mf_varname_tree): decl_printable_name handles
1328 (mf_xform_derefs_1) [RESULT_DECL, STRING_CST]: Handle as innermost
1329 object. [ptr_type]: Remove unused variable.
1330 Remember nearest addressable array-element or record-component
1331 when checking bitfield components. Tolerate empty BIND_EXPR.
1333 2005-01-01 Richard Henderson <rth@redhat.com>
1335 * tree-vectorizer.c (vect_analyze_offset_expr): Strip conversions
1336 that don't narrow the value. Fail for other conversions.
1338 2005-01-01 Richard Henderson <rth@redhat.com>
1341 * c-decl.c (pop_file_scope): Call maybe_apply_pending_pragma_weaks.
1342 * c-lang.c (finish_file): Don't do it here.
1343 * objc/objc-act.c (objc_finish_file): Likewise.
1345 * cgraph.c (decl_assembler_name_equal): New.
1346 (cgraph_node_for_asm, cgraph_varpool_node_for_asm): New.
1347 (cgraph_varpool_node): Actually link up cgraph_varpool_nodes.
1348 * cgraph.h (struct cgraph_varpool_node): Add next.
1349 (cgraph_node_for_asm, cgraph_varpool_node_for_asm): Declare.
1350 * varasm.c (assemble_alias): Mark the target as needed.
1352 2005-01-01 Andrew Pinski <pinskia@physics.uc.edu>
1355 * function.c (get_arg_pointer_save_area): Use entry_of_function
1356 instead of get_insns.
1358 2005-01-01 Roger Sayle <roger@eyesopen.com>
1359 Andrew Pinski <pinskia@physics.uc.edu>
1360 James E. Wilson <wilson@specifixinc.com>
1362 PR rtl-optimization/12092
1363 * loop.c (emit_prefetch_instructions): Do nothing if PREFETCH_BLOCK
1366 2005-01-01 Roger Sayle <roger@eyesopen.com>
1367 Olivier Hainque <hainque@act-europe.fr>
1369 * tree.c (int_fits_type_p): A narrower type always fits in a
1370 wider one, except for negative values into unsigned types.
1372 2005-01-01 Roger Sayle <roger@eyesopen.com>
1374 * tree.c (int_fits_type_p): Compare the result of force_fit_type
1375 with the original constant rather than require TREE_OVERFLOW.
1377 2005-01-01 Steven Bosscher <stevenb@suse.de>
1380 * c-decl.c (finish_function): If compiling C99, annotate the
1381 compiler generated return with the current file name and line 0.
1382 * tree-cfg.c (remove_useless_stmts_warn_notreached): Only warn if
1383 the source line is greater than 0.
1384 (remove_bb): Likewise.
1386 See ChangeLog.12 for earlier changes.