Uros Bizjak [Mon, 14 Nov 2011 20:25:14 +0000 (21:25 +0100)]
avx-recip-vec.c: New test.
* gcc.target/i386/avx-recip-vec.c: New test.
* gcc.target/i386/avx-lrintf-vec.c: Ditto.
* gcc.target/i386/avx-lrint-vec.c: Ditto.
* gcc.target/i386/avx-ceilf-vec.c: Include sse4_1-ceilf-vec.c.
* gcc.target/i386/avx-ceil-vec.c: Include sse4_1-ceil-vec.c.
* gcc.target/i386/avx-floorf-vec.c: Include sse4_1-floorf-vec.c.
* gcc.target/i386/avx-floor-vec.c: Include sse4_1-floor-vec.c.
* gcc.target/i386/avx-rintf-vec.c: Include sse4_1-rintf-vec.c.
* gcc.target/i386/avx-rint-vec.c: Include sse4_1-rint-vec.c.
* gcc.target/i386/avx-roundf-vec.c: Include sse4_1-roundf-vec.c.
* gcc.target/i386/avx-round-vec.c: Include sse4_1-round-vec.c.
* gcc.target/i386/avx-truncf-vec.c: Include sse4_1-truncf-vec.c.
* gcc.target/i386/avx-trunc-vec.c: Include sse4_1-trunc-vec.c.
* gcc.target/i386/avx-copysignf-vec.c: Include sse-copysignf-vec.c.
* gcc.target/i386/avx-copysign-vec.c: Include sse2-copysign-vec.c.
Ed Smith-Rowland [Mon, 14 Nov 2011 20:05:12 +0000 (20:05 +0000)]
re PR c++/51107 ([C++11] Accepts invalid literal operator with void argument list.)
PR c++/51107
* typeck.c (check_literal_operator_args): Add processing_specialization
to check for void template fn. Test for exact arity for non-template fn.
Fabien ChĂȘne [Mon, 14 Nov 2011 17:55:57 +0000 (18:55 +0100)]
re PR c++/6936 (member "using" binds wrong)
PR c++/6936
PR c++/25994
PR c++/26256
PR c++/30195
* search.c (lookup_field_1): Look through USING_DECL.
(lookup_field_r): Call lookup_fnfields_slot instead of
lookup_fnfields_1.
* semantics.c (finish_member_declaration): Remove the check that
prevents USING_DECLs from being verified by
pushdecl_class_level. Call add_method for using declarations that
designates functions if the using declaration is in a template
class. Set DECL_IGNORED_P on class-scope using declarations.
* typeck.c (build_class_member_access_expr): Handle USING_DECLs.
* class.c (check_field_decls): Keep using declarations.
(add_method): Remove two diagnostics about conflicting using
declarations.
* parser.c (cp_parser_nonclass_name): Handle USING_DECLs.
* decl.c (start_enum): Call xref_tag whenever possible.
* cp-tree.h (strip_using_decl): Declare, and reident the previous
function.
* name-lookup.c (strip_using_decl): New function.
(supplement_binding_1): Call strip_using_decl on decl and
bval. Perform most of the checks with USING_DECLs stripped. Also
check that the target decl and the target bval does not refer to
the same declaration. Allow pushing an enum multiple times in a
template class. Adjustment to diagnose using redeclarations. Call
diagnose_name_conflict.
(push_class_level_binding): Call strip_using_decl on decl and
bval. Perform most of the checks with USING_DECLs stripped. Return
true if both decl and bval refer to USING_DECLs and are dependent.
(diagnose_name_conflict): New function.
* config/i386/i386.h (processor_costs): Add second dimension to
stringop_algs array.
* config/i386/i386.c (cost models): Initialize second dimension of
stringop_algs arrays.
(core_cost): New costs based on generic64 costs with updated stringop
values.
(promote_duplicated_reg): Add support for vector modes, add
declaration.
(promote_duplicated_reg_to_size): Likewise.
(processor_target): Set core costs for core variants.
(expand_set_or_movmem_via_loop_with_iter): New function.
(expand_set_or_movmem_via_loop): Enable reuse of the same iters in
different loops, produced by this function.
(emit_strset): New function.
(expand_movmem_epilogue): Add epilogue generation for bigger sizes,
use SSE-moves where possible.
(expand_setmem_epilogue): Likewise.
(expand_movmem_prologue): Likewise for prologue.
(expand_setmem_prologue): Likewise.
(expand_constant_movmem_prologue): Likewise.
(expand_constant_setmem_prologue): Likewise.
(decide_alg): Add new argument align_unknown. Fix algorithm of
strategy selection if TARGET_INLINE_ALL_STRINGOPS is set; Skip sse_loop
(decide_alignment): Update desired alignment according to chosen move
mode.
(ix86_expand_movmem): Change unrolled_loop strategy to use SSE-moves.
(ix86_expand_setmem): Likewise.
(ix86_slow_unaligned_access): Implementation of new hook
slow_unaligned_access.
* config/i386/i386.md (strset): Enable half-SSE moves.
* config/i386/sse.md (vec_dupv4si): Add expand for vec_dupv4si.
(vec_dupv2di): Add expand for vec_dupv2di.
Co-Authored-By: Jan Hubicka <jh@suse.cz>
From-SVN: r181357
Oleg Endo [Sun, 13 Nov 2011 23:00:10 +0000 (23:00 +0000)]
re PR target/50694 (SH Target: SH2A little endian does not actually work)
PR target/50694
* config/sh/sh.h (IS_LITTLE_ENDIAN_OPTION, UNSUPPORTED_SH2A):
New macros.
(DRIVER_SELF_SPECS): Use new macros to filter out
unsupported options taking the default configuration into
account.
* gcc.target/sh/pr21255-2-ml.c: Skip if -mb or -m5* is
specified. Remove redundant runtime checks.
* gcc.target/sh/20080410-1.c: Skip if -mb is specified.
Allow for other than -m4. Fix typos in comments.
Jason Merrill [Sun, 13 Nov 2011 05:09:36 +0000 (00:09 -0500)]
re PR c++/986 (g++ misses warning for reference on temporary that invokes undefined behaviour)
PR c++/986
* call.c (set_up_extended_ref_temp): Warn about references
bound to non-static reference members.
* init.c (perform_member_init): Pass in the member.
Jason Merrill [Sun, 13 Nov 2011 00:44:39 +0000 (19:44 -0500)]
re PR c++/51060 (Temporary object stack space is not re-used)
PR c++/51060
* gimplify.c (gimplify_target_expr): Add a clobber to the cleanup.
(gimplify_modify_expr): Don't try to simplify it.
* cp/cp-gimplify.c (cp_gimplify_expr): Leave clobbers alone.
builtins.c (expand_builtin_compare_and_swap): If target is const0, don't pass the target to expand_atomic_compare_and_swap.
* builtins.c (expand_builtin_compare_and_swap): If target is const0,
don't pass the target to expand_atomic_compare_and_swap.
(expand_builtin_atomic_compare_exchange): Likewise.
Jonathan Wakely [Sat, 12 Nov 2011 15:57:03 +0000 (15:57 +0000)]
re PR libstdc++/51083 (TR1 [tr.c99.cmath.over] and C++11 [cmplx.over] overloads not constrained)
PR libstdc++/51083
* include/ext/type_traits.h (__promote): Only define __type member
for integral and floating point types, to prevent math functions
participating in overload resolution for other types.
(__promote_2, __promote_3, __promote_4): Use __promote in default
template argument values, so deduction only succeeds for integral and
floating point types.
* testsuite/26_numerics/cmath/51083.cc: New.
* testsuite/26_numerics/complex/51083.cc: New.
* testsuite/tr1/8_c_compatibility/cmath/51083.cc: New.
* testsuite/tr1/8_c_compatibility/complex/51083.cc: New.
Iain Sandoe [Sat, 12 Nov 2011 15:06:18 +0000 (15:06 +0000)]
Makefile.in (stamp-gnatlib-$(RTSDIR)): Don't link s-oscons.ads.
ada:
* gcc-interface/Makefile.in (stamp-gnatlib-$(RTSDIR)): Don't link
s-oscons.ads.
(OSCONS_CPP, OSCONS_EXTRACT): New.
(./bldtools/oscons/xoscons): New Target.
($(RTSDIR)/s-oscons.ads): New Target.
(gnatlib): Depend on $(RTSDIR)/s-oscons.ads.
* gcc-interface/Make-lang.in (ada/s-oscons.ads) Remove as dependency.
* Make-generated.in: Remove machinery to generate xoscons and
ada/s-oscons.ads.
libada:
Makefile.in: Change dependency on oscons to depend on the generator
tool.
Iain Sandoe [Sat, 12 Nov 2011 14:12:26 +0000 (14:12 +0000)]
re PR target/45233 (FAIL: gcc.c-torture/compile/pr44707.c)
gcc:
PR target/45233
* config/rs6000/rs6000.c (rs6000_legitimize_reload_address):
Only expand a symbol ref. into an access when the entity is defined
in the TU.
Jakub Jelinek [Fri, 11 Nov 2011 19:58:10 +0000 (20:58 +0100)]
i386-protos.h (ix86_maybe_emit_epilogue_vzeroupper): New prototype.
* config/i386/i386-protos.h (ix86_maybe_emit_epilogue_vzeroupper):
New prototype.
* config/i386/i386.c (ix86_maybe_emit_epilogue_vzeroupper): New
function.
(ix86_expand_epilogue): Use it.
* config/i386/i386.md (return, simple_return): Call it in the
expanders.
Jakub Jelinek [Fri, 11 Nov 2011 19:56:13 +0000 (20:56 +0100)]
re PR tree-optimization/51058 (ICE: gimple check: expected gimple_assign(error_mark), have gimple_call() in gimple_assign_rhs_code, at gimple.h:1992)
PR tree-optimization/51058
* tree-vect-slp.c (vect_remove_slp_scalar_calls): New function.
(vect_schedule_slp): Call it.
* tree-vect-stmts.c (vectorizable_call): If slp_node != NULL,
don't replace scalar calls with setting of their lhs to zero here.
* gcc.dg/vect/fast-math-vect-call-1.c: Add f4 test.
* gfortran.dg/vect/pr51058-2.f90: New test.
Jakub Jelinek [Fri, 11 Nov 2011 08:10:07 +0000 (09:10 +0100)]
tree-vect-slp.c (vect_free_slp_tree): Also free SLP_TREE_CHILDREN vector.
* tree-vect-slp.c (vect_free_slp_tree): Also free SLP_TREE_CHILDREN
vector.
(vect_create_new_slp_node): Don't allocate node before checking stmt
type.
(vect_free_oprnd_info): Remove FREE_DEF_STMTS argument, always
free def_stmts vectors and additionally free oprnd_info.
(vect_build_slp_tree): Adjust callers. Call it even if
stop_recursion. If vect_create_new_slp_node or
vect_build_slp_tree fails, properly handle freeing memory.
If it succeeded, clear def_stmts in oprnd_info.
final.c, output.h (fprint_whex, [...]): New functions serving as fast replacements for fprintf() integer to string...
* final.c, output.h (fprint_whex, fprint_w, fprint_ul, sprint_ul):
New functions serving as fast replacements for fprintf() integer
to string conversions. They were used in the following changes.
* final.c (sprint_ul_rev): Internal helper for the above.
(output_addr_const): case CONST_INT: don't use fprintf().
* elfos.h (ASM_GENERATE_INTERNAL_LABEL): Don't use sprintf("%u"),
use sprint_ul() and stpcpy() which are much faster.
(TARGET_ASM_INTERNAL_LABEL): Define as default_elf_internal_label.
(ELF_ASCII_ESCAPES, ELF_STRING_LIMIT): Are the old ESCAPES and
STRING_LIMIT macros.
(ASM_OUTPUT_LIMITED_STRING, ASM_OUTPUT_ASCII): Macros now just
call respective functions that provide the same
functionality. Those are default_elf_asm_output_limited_string()
and default_elf_asm_output_ascii() in varasm.c.
* varasm.c: Fixed some whitespace inconsistencies.
(default_elf_asm_output_limited_string)
(default_elf_asm_output_ascii): The above macros from elfos.h are
implemented here as these functions, avoiding superfluous calls to
fprintf().
(default_elf_internal_label): Hook for
targetm.asm_out.internal_label and ASM_OUTPUT_DEBUG_LABEL.
* i386.c: Don't call fprintf("%u") but fprint_ul() instead.
* defaults.h (ASM_OUTPUT_LABEL, ASM_OUTPUT_INTERNAL_LABEL):
Expanded the macros on multiple lines for readability.
(ASM_OUTPUT_LABELREF): Have two calls to fputs() instead of one to
asm_fprintf().
* dwarf2asm.c (dw2_assemble_integer, dw2_asm_output_data)
(dw2_asm_data_uleb128, dw2_asm_delta_uleb128)
(dw2_asm_delta_sleb128): Convert fprintf() calls to the new
faster functions.
* dwarf2out.c (dwarf2out_source_line): Convert fprintf() calls to
the new faster functions.
PR c++/50973
* decl2.c (mark_used): Defer synthesis of virtual functions.
* method.c (use_thunk): Make sure the target function has
DECL_INTERFACE_KNOWN.
testsuite
* gcc.dg/atomic-invalid.c: Add test for invalid __atomic_clear models.
* gcc.dg/atomic-flag.c: New. Test __atomic_test_and_set and
__atomic_clear.
Eric Botcazou [Thu, 10 Nov 2011 19:45:17 +0000 (19:45 +0000)]
fe.h (Serious_Errors_Detected): New macro.
* fe.h (Serious_Errors_Detected): New macro.
* gcc-interface/gigi.h (build_atomic_load): Declare.
(build_atomic_store): Likewise.
* gcc-interface/trans.c (atomic_sync_required_p): New predicate.
(call_to_gnu): Add ATOMIC_SYNC parameter. Use local variable.
Build an atomic load for an In or In Out parameter if needed.
Build an atomic store for the assignment of an Out parameter if needed.
Build an atomic store to the target if ATOMIC_SYNC is true.
(present_in_lhs_or_actual_p): New predicate.
(gnat_to_gnu) <N_Identifier>: Build an atomic load if needed.
<N_Explicit_Dereference>: Likewise.
<N_Indexed_Component>: Likewise.
<N_Selected_Component>: Likewise.
<N_Assignment_Statement>: Adjust call to call_to_gnu.
Build an atomic store to the LHS if needed.
<N_Function_Call>: Adjust call to call_to_gnu.
* gcc-interface/utils2.c: Include toplev.h.
(resolve_atomic_size): New static function.
(build_atomic_load): New function.
(build_atomic_store): Likewise.
* gcc-interface/Make-lang.in (ada/utils2.o): Add toplev.h.
Nathan Sidwell [Thu, 10 Nov 2011 19:17:13 +0000 (19:17 +0000)]
gcov.c (struct function_info): Make src an index, not a pointer.
* gcov.c (struct function_info): Make src an index, not a pointer.
(struct source_info): Remove index and next source fields.
(fn_end): New static var.
(sources_index): Remove.
(sources): Now a pointer to an array, not a list.
(n_sources, a_sources): New.
(process_file): Adjust for changes to read_graph_file. Insert
functions into source lists and check line numbers here.
(generate_results): Only allocate lines for sources with
contents. Adjust for source array.
(release_structures): Likewise.
(find_source): Return source index, adjust for source array.
(read_graph_file): Return function list. Don't insert into source
lists here.
(read_count_file): Take list of functions.
(solve_flow_graph): Reverse the arc lists here.
(add_line_counts): Adjust for source array.
Andrew MacLeod [Thu, 10 Nov 2011 16:39:32 +0000 (16:39 +0000)]
re PR rtl-optimization/51040 (ICE: RTL check: access of elt 1 of 'not' with last elt 0 in gen_rtx_fmt_ee_stat, at ./genrtl.h:33 with __atomic_nand_fetch())
PR rtl-optimization/51040
* optabs.c (expand_atomic_fetch_op): Patchup code for NAND should be AND
followed by NOT.
* builtins.c (expand_builtin_atomic_fetch_op): Patchup code for NAND
should be AND followed by NOT.
* testsuite/gcc.dg/atomic-noinline[-aux].c: Test no-inline NAND and
patchup code.
Jakub Jelinek [Thu, 10 Nov 2011 15:36:48 +0000 (16:36 +0100)]
re PR rtl-optimization/51023 (likely integer wrong code bug)
PR rtl-optimization/51023
* combine.c (simplify_comparison) <case SIGN_EXTEND>: Don't use
val_signbit_known_clear_p for signed comparison narrowing
optimization. Don't check for non-VOIDmode, use
HWI_COMPUTABLE_MODE_P macro.
<case ZERO_EXTEND>: Don't check for non-VOIDmode.
Optimize even when const_op is equal to GET_MODE_MASK (mode),
don't optimize if const_op is negative.
Richard Guenther [Thu, 10 Nov 2011 13:46:27 +0000 (13:46 +0000)]
re PR tree-optimization/51070 (ICE verify_gimple failed)
2011-11-10 Richard Guenther <rguenther@suse.de>
PR tree-optimization/51070
* tree-loop-distribution.c (generate_builtin): Do not replace
the loop with a builtin if the partition contains statements which
results are used outside of the loop.
(pass_loop_distribution): Verify and collect.
Richard Guenther [Thu, 10 Nov 2011 12:24:52 +0000 (12:24 +0000)]
re PR tree-optimization/51030 (PHI opt does not handle value-replacement with a transfer function)
2011-11-10 Richard Guenther <rguenther@suse.de>
PR tree-optimization/51030
* tree-ssa-phiopt.c (jump_function_from_stmt): New function.
(value_replacement): Use it to handle trivial non-empty
intermediate blocks.
Jakub Jelinek [Thu, 10 Nov 2011 08:29:52 +0000 (09:29 +0100)]
re PR tree-optimization/51000 (ICE: in vect_get_store_cost, at tree-vect-stmts.c:923 on powerpc-apple-darwin9)
PR tree-optimization/51000
* tree-vect-patterns.c (vect_recog_bool_pattern): If adding
a pattern stmt for a bool store, adjust DR_STMT too.
Don't handle bool conversions to single bit precision lhs.
* tree-vect-stmts.c (vect_remove_stores): If next is a pattern
stmt, remove its related stmt and free its stmt_vinfo.
(free_stmt_vec_info): Free also pattern stmt's vinfo and
pattern def stmt's vinfo.
* tree-vect-loop.c (destroy_loop_vec_info): Don't try to
free pattern stmt's vinfo here.
(vect_transform_loop): When calling vect_remove_stores,
do gsi_next first and don't call gsi_remove. If not strided
store, free stmt vinfo for gsi_stmt (si) rather than stmt.
Jakub Jelinek [Wed, 9 Nov 2011 21:37:27 +0000 (22:37 +0100)]
re PR target/50911 (assertion failure in expand_vec_perm_interleave2 with -msse)
PR target/50911
* config/i386/i386.c (expand_vec_perm_interleave2): If d->vmode is
V4SImode, !TARGET_SSE2 and punpck[lh]* is needed, change dremap.vmode
to V4SFmode.
* gcc.dg/torture/vshuf-16.inc: Add interleave low and high
permutations.
* gcc.dg/torture/vshuf-32.inc: Likewise.
* gcc.dg/torture/vshuf-4.inc: Likewise.
* gcc.dg/torture/vshuf-8.inc: Likewise.
Jakub Jelinek [Wed, 9 Nov 2011 21:28:57 +0000 (22:28 +0100)]
function.h (requires_stack_frame_p): New prototype.
* function.h (requires_stack_frame_p): New prototype.
* function.c (requires_stack_frame_p): No longer static.
* config/i386/i386.c (ix86_finalize_stack_realign_flags): If
stack_realign_fp was just a conservative guess for a function
which doesn't use sp/fp/argp at all, clear frame_pointer_needed
and stack realignment.
Jonathan Wakely [Wed, 9 Nov 2011 21:14:03 +0000 (21:14 +0000)]
allocator.h (__shrink_to_fit_aux::_S_do_it): Create the new object with the same allocator.
* include/bits/allocator.h (__shrink_to_fit_aux::_S_do_it): Create
the new object with the same allocator.
* testsuite/23_containers/vector/capacity/shrink_to_fit2.cc: New.
libgcc/:
* generic-morestack.c: Include <string.h>.
(uintptr_type): Define.
(struct initial_sp): Add dont_block_signals field. Reduce size of
extra array by 1.
(allocate_segment): Set prev field to NULL. Don't set
__morestack_current_segment or __morestack_segments.
(__generic_morestack): Update current->prev and *pp after calling
allocate_segment.
(__morestack_block_signals): Don't do anything if
dont_block_signals is set.
(__morestack_unblock_signals): Likewise.
(__generic_findstack): Check for initial_sp == NULL. Add casts to
uintptr_type.
(__splitstack_block_signals): New function.
(enum __splitstack_content_offsets): Define.
(__splitstack_getcontext, __splitstack_setcontext): New functions.
(__splitstack_makecontext): New function.
(__splitstack_block_signals_context): New function.
(__splitstack_find_context): New function.
* config/i386/morestack.S (__morestack_get_guard): New function.
(__morestack_set_guard, __morestack_make_guard): New functions.
* libgcc-std.ver.in: Add new functions to GCC_4.7.0.
gcc/testsuite/:
* lib/target-supports.exp (check_effective_target_ucontext_h): New
procedure.
* gcc.dg/split-5.c: New test.