2012-06-27 Richard Guenther <rguenther@suse.de>
PR tree-optimization/53774
* tree-ssa-reassoc.c (get_rank): All default defs have
precomputed rank.
(init_reassoc): Precompute rank for all SSA default defs.
From-SVN: r189012
We were always falling through to the memory default.
Also use standard_sse_constant_p on CONST_VECTOR.
* config/i386/i386.c (ix86_rtx_costs): Use standard_sse_constant_p
and don't fall thru from standard_80387_constant_p to the memory
fallback,
From-SVN: r189007
If we don't implement this pattern, the vectorizer is happy to
unpack the v4si and use the full mulv2di3. This results in
more element shuffling than is required.
* config/i386/i386.c (bdesc_args): Update. Change
IX86_BUILTIN_VEC_WIDEN_SMUL_ODD_V4SI to OPTION_MASK_ISA_SSE2.
(IX86_BUILTIN_VEC_WIDEN_SMUL_EVEN_V4SI): New.
(ix86_builtin_mul_widen_even): Use it.
(ix86_builtin_mul_widen_odd): Relax SMUL_ODD from sse4 to sse2.
(ix86_expand_mul_widen_evenodd): Handle signed for sse2.
* config/i386/sse.md (vec_widen_<s>mult_hi_<V124_AVX2>): Allow
for all SSE2.
(vec_widen_<s>mult_lo_<V124_AVX2>): Likewise.
(vec_widen_<s>mult_odd_<VI4_AVX2>): Likewise. Relax from V124_AVX2.
(vec_widen_smult_even_v4si): New.
From-SVN: r189006
Move the expansion code to i386.c next to mulv4si3. Eliminate
one shift by adding the highparts before shifting. Correct costs.
* config/i386/sse.md (mul<VI8_AVX2>3): Change from insn_and_split
to expander; move guts to ...
* config/i386/i386.c (ix86_expand_sse2_mulvxdi3): ... here. Add
highparts before shifting up.
* config/i386/i386-protos.h: Update.
From-SVN: r189005
* system.h (USE_COMMON_FOR_ONE_ONLY): Poison.
* defaults.h (USE_COMMON_FOR_ONE_ONLY): Do not provide default.
* config/darwin.h (USE_COMMON_FOR_ONE_ONLY): Do not define.
From-SVN: r188999
2012-06-26 Vincent Pucci <pucci@adacore.com>
* exp_ch3.adb (Build_Init_Statements): Don't check the parents
in the Rep Item Chain of the task for aspects Interrupt_Priority,
Priority, CPU and Dispatching_Domain.
* exp_ch9.adb (Expand_N_Task_Type_Declaration): fields _Priority,
_CPU and _Domain are present in the corresponding record type
only if the task entity has a pragma, attribute definition
clause or aspect specification.
(Make_Initialize_Protection): Don't check the parents in the Rep Item
Chain of the task for aspects Interrupt_Priority, Priority, CPU and
Dispatching_Domain.
* freeze.adb (Freeze_Entity): Use of Evaluate_Aspects_At_Freeze_Point
call replaced by Analyze_Aspects_At_Freeze_Point.
* sem_ch13.adb, sem_ch13.ads (Analyze_Aspects_At_Freeze_Point):
Renaming of Evaluate_Aspects_At_Freeze_Point.
2012-06-26 Yannick Moy <moy@adacore.com>
* sem_attr.adb (Analyze_Attribute): Detect if 'Old is used outside a
postcondition, and issue an error in such a case.
2012-06-26 Yannick Moy <moy@adacore.com>
* gnat_rm.texi: Minor editing.
2012-06-26 Tristan Gingold <gingold@adacore.com>
* raise-gcc.c: Minor cleanup: remove unused prototype.
* seh_init.c: Do not create an image wide unwind info to catch
SEH when SEH unwind info are emitted by the compiler.
From-SVN: r188995
* system.h (UNALIGNED_SHORT_ASM_OP, UNALIGNED_INT_ASM_OP,
UNALIGNED_LONG_ASM_OP, UNALIGNED_DOUBLE_INT_ASM_OP): Poison.
* vmsdbgout.c (UNALIGNED_SHORT_ASM_OP, UNALIGNED_INT_ASM_OP,
UNALIGNED_LONG_ASM_OP, UNALIGNED_DOUBLE_INT_ASM_OP): Rename to
VMS_UNALIGNED_SHORT_ASM_OP, VMS_UNALIGNED_INT_ASM_OP,
VMS_UNALIGNED_LONG_ASM_OP, and VMS_UNALIGNED_DOUBLE_INT_ASM_OP.
* config/microblaze/microblaze.h (UNALIGNED_SHORT_ASM_OP,
UNALIGNED_INT_ASM_OP, UNALIGNED_DOUBLE_INT_ASM_OP): Do not define.
* doc/tm.texi.in: Remove UNALIGNED_INT_ASM_OP reference from the
documentation.
* doc/tm.texi: Regenerate.
* doc/cpp.texi: Make example for #error generic.
* config/frv/frv.h: Fix example text to match tm.texi.
From-SVN: r188991
PR tree-optimization/53729
PR tree-optimization/53636
* tree-vect-slp.c (vect_slp_analyze_bb_1): Delay call to
vect_verify_datarefs_alignment until after statements have
been marked as relevant/irrelevant.
* tree-vect-data-refs.c (vect_verify_datarefs_alignment):
Skip irrelevant statements.
(vect_enhance_data_refs_alignment): Use STMT_VINFO_RELEVANT_P
instead of STMT_VINFO_RELEVANT.
(vect_get_data_access_cost): Do not check for supportable
alignment before calling vect_get_load_cost/vect_get_store_cost.
* tree-vect-stmts.c (vect_get_store_cost): Do not abort when
handling unsupported alignment.
(vect_get_load_cost): Likewise.
From-SVN: r188979
PR tree-optimization/53748
* tree-ssa-phiopt.c (conditional_replacement): Only optimize
if arg0/arg1 have integral or pointer types.
* gcc.c-torture/compile/pr53748.c: New test.
From-SVN: r188975
* lib/profopt.exp: Make prof_option_list local to profopt-execute.
* g++.dg/tree-prof/tree-prof.exp (PROFOPT_OPTIONS): Define after
including profopt.opt; save and restore existing value.
* g++.dg/bprob/bprob.exp: Likewise.
* gcc.dg/matrix/matrix.exp: Likewise.
* gcc.dg/tree-prof/tree-prof.exp: Likewise.
* gcc.misc-tests/bprob.exp: Likewise; also replace formerly-ignored
PROFOPT_OPTIONS.
From-SVN: r188964
Now that we support mult_even/odd hooks, the vectorizer can
generate the exact same code for plain sse dot_prod by itself,
as well as other reductions other than plus.
From-SVN: r188960
Prepares for exposing builtin_mul_widen_even/odd hooks
for more efficient reduction. Adds QImode multiplication.
Shares code between mulv4si3 and the widening multiplies.
From-SVN: r188957
PR target/53759
* config/i386/sse.md (sse_loadlps): Use x m x constraints instead
of x x x in the vmovlps load alternative.
* gcc.target/i386/pr53759.c: New test.
From-SVN: r188937
gcc/
PR debug/53740
* df.h (dead_debug_add): Remove third argument.
* df-problems.c (dead_debug_add): Likewise. Use the REGNO of the
REG that we want to replace instead.
(dead_debug_insert_temp): Use the REGNO of the reg that we want
to replace instead of DF_REF_REGNO. Require there to always be
at least one such use. Check for cases where the same location
has more than df_ref associated with it.
(df_note_bb_compute): Remove third dead_debug_add argument.
* dce.c (word_dce_process_block): Likewise.
From-SVN: r188934
Richard Guenther
Nick Clifton
Richard Henderson
GCC Administrator
Steven Bosscher
Alexandre Oliva
Janis Johnson
Arnaud Charlet
Sterling Augustine
Sterling Augustine
Bill Schmidt
Matt Turner
Jan Hubicka
Dehao Chen
Ulrich Weigand
Jakub Jelinek
Jason Merrill
Edward Smith-Rowland
Benjamin Kosnik
Christophe Lyon
Florian Weimer
Matthew Gretton-Dann
Ian Lance Taylor
Richard Sandiford