gcc.git
11 years agotransfer.c (read_block_direct): Correct condition.
Tobias Burnus [Mon, 11 Mar 2013 18:55:18 +0000 (19:55 +0100)]
transfer.c (read_block_direct): Correct condition.

2013-03-11  Tobias Burnus  <burnus@net-b.de>

        * io/transfer.c (read_block_direct): Correct condition.
        * intrinsics/execute_command_line.c (execute_command_line):
        Remove dead code for the HAVE_FORK case.

From-SVN: r196603

11 years agogfortran.texi (STRUCTURE and RECORD): State more clearly how to convert them into...
Tobias Burnus [Mon, 11 Mar 2013 18:37:30 +0000 (19:37 +0100)]
gfortran.texi (STRUCTURE and RECORD): State more clearly how to convert them into derived types.

2013-03-11  Tobias Burnus  <burnus@net-b.de>

        * gfortran.texi (STRUCTURE and RECORD): State more clearly how
        to convert them into derived types.

From-SVN: r196602

11 years agore PR c++/56567 (ICE with lambda return type deduction and braced-init-list)
Jason Merrill [Mon, 11 Mar 2013 16:22:16 +0000 (12:22 -0400)]
re PR c++/56567 (ICE with lambda return type deduction and braced-init-list)

PR c++/56567
* typeck.c (check_return_expr): Disallow returning init list here.
* semantics.c (apply_deduced_return_type): Not here.

From-SVN: r196600

11 years ago* de.po, sv.po: Update.
Joseph Myers [Mon, 11 Mar 2013 15:24:43 +0000 (15:24 +0000)]
* de.po, sv.po: Update.

From-SVN: r196598

11 years agoRemove dg-excess-errors in gcc.dg/inline_[34].c and unroll_[234].c
Dominique d'Humieres [Mon, 11 Mar 2013 15:08:13 +0000 (16:08 +0100)]
Remove dg-excess-errors in gcc.dg/inline_[34].c and unroll_[234].c

2013-02-26  Dominique d'Humieres  <dominiq@lps.ens.fr>

* gcc.dg/inline_3.c: Remove target and dg-excess-errors.
* gcc.dg/inline_4.c: Likewise.
* gcc.dg/unroll_2.c: Likewise.
* gcc.dg/unroll_3.c: Likewise.
* gcc.dg/unroll_4.c: Likewise.

From-SVN: r196597

11 years agore PR target/56470 (ICE output_operand: invalid shift operand)
Richard Earnshaw [Mon, 11 Mar 2013 11:48:34 +0000 (11:48 +0000)]
re PR target/56470 (ICE output_operand: invalid shift operand)

PR target/56470
* arm.c (shift_op): Validate RTL pattern on the fly.
(arm_print_operand, case 'S'): Don't use shift_operator to validate
the RTL.

From-SVN: r196595

11 years agore PR testsuite/54119 (FAIL: gcc.dg/tree-ssa/vector-4.c scan-tree-dump-times gimple...
John David Anglin [Mon, 11 Mar 2013 01:18:18 +0000 (01:18 +0000)]
re PR testsuite/54119 (FAIL: gcc.dg/tree-ssa/vector-4.c scan-tree-dump-times gimple "VEC_PERM_EXPR <a, b, { 0, 4, 1, 5 }>;" 1)

PR testsuite/54119
* gcc.dg/tree-ssa/vector-4.c: xfail on 32-bit hppa*-*-*.

From-SVN: r196592

11 years agore PR debug/56307 (FAIL: gcc.dg/tree-ssa/pr55579.c scan-tree-dump esra "Created a...
John David Anglin [Mon, 11 Mar 2013 01:10:38 +0000 (01:10 +0000)]
re PR debug/56307 (FAIL: gcc.dg/tree-ssa/pr55579.c scan-tree-dump esra "Created a debug-only replacement for s")

PR debug/56307
* gcc.dg/tree-ssa/pr55579.c: xfail 32-bit hppa*-*-hpux*.

From-SVN: r196591

11 years agore PR target/40797 (ICE in df_refs_verify, at df-scan.c:4361)
Oleg Endo [Mon, 11 Mar 2013 01:04:13 +0000 (01:04 +0000)]
re PR target/40797 (ICE in df_refs_verify, at df-scan.c:4361)

PR target/40797
* gcc.c-torture/compile/pr40797.c: New.

From-SVN: r196590

11 years agopr44194-1.c: Skip compilation on hppa*64*-*-*.
John David Anglin [Mon, 11 Mar 2013 00:56:45 +0000 (00:56 +0000)]
pr44194-1.c: Skip compilation on hppa*64*-*-*.

* gcc.dg/pr44194-1.c: Skip compilation on hppa*64*-*-*.

From-SVN: r196589

11 years agore PR target/56347 (FAIL: gfortran.dg/integer_exponentiation_2.f90 -O2 execution...
John David Anglin [Mon, 11 Mar 2013 00:44:28 +0000 (00:44 +0000)]
re PR target/56347 (FAIL: gfortran.dg/integer_exponentiation_2.f90  -O2  execution test)

PR target/56347
* config/pa/pa.md (call_value): Check for calls to powf and direct to
new call patterns that clobber %fr12.
(call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
split and postreload patterns.
* config/pa/pa.c (pa_conditional_register_usage): Revert marking
registers %fr12 and %fr12R as call used.

From-SVN: r196588

11 years agoDaily bump.
GCC Administrator [Mon, 11 Mar 2013 00:18:29 +0000 (00:18 +0000)]
Daily bump.

From-SVN: r196587

11 years ago* de.po: Update.
Joseph Myers [Sun, 10 Mar 2013 15:21:47 +0000 (15:21 +0000)]
* de.po: Update.

From-SVN: r196581

11 years agore PR fortran/56575 (An invalid OO code causes ICE)
Paul Thomas [Sun, 10 Mar 2013 13:23:58 +0000 (13:23 +0000)]
re PR fortran/56575 (An invalid OO code causes ICE)

2013-03-10  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/56575
* expr.c (gfc_default_initializer): Check that a class declared
type has any components.
* resolve.c (resolve_fl_derived0): On failing the test for C437
set the type to BT_UNKNOWN to prevent repeat error messages.
2013-03-10  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/56575
* gfortran.dg/class_56.f90: New test.

From-SVN: r196580

11 years agoDaily bump.
GCC Administrator [Sun, 10 Mar 2013 00:19:02 +0000 (00:19 +0000)]
Daily bump.

From-SVN: r196579

11 years agodse.c (delete_dead_store_insn): Respect TDF_DETAILS.
Steven Bosscher [Sat, 9 Mar 2013 22:04:15 +0000 (22:04 +0000)]
dse.c (delete_dead_store_insn): Respect TDF_DETAILS.

* dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
(canon_address, record_store, replace_read, check_mem_read_rtx,
scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
rest_of_handle_dse): Likewise.

From-SVN: r196576

11 years ago* de.po: Update.
Joseph Myers [Sat, 9 Mar 2013 21:46:26 +0000 (21:46 +0000)]
* de.po: Update.

From-SVN: r196575

11 years ago55463.cc: Avoid -Wunused warnings.
Paolo Carlini [Sat, 9 Mar 2013 11:45:48 +0000 (11:45 +0000)]
55463.cc: Avoid -Wunused warnings.

2013-03-09  Paolo Carlini  <paolo.carlini@oracle.com>

* testsuite/20_util/function_objects/mem_fn/55463.cc: Avoid
-Wunused warnings.

From-SVN: r196572

11 years agore PR middle-end/56524 (Compiler ICE when compiling with -mips16)
Richard Sandiford [Sat, 9 Mar 2013 07:54:02 +0000 (07:54 +0000)]
re PR middle-end/56524 (Compiler ICE when compiling with -mips16)

gcc/
PR middle-end/56524
* tree.h (tree_optimization_option): Rename target_optabs to optabs.
Add base_optabs.
(TREE_OPTIMIZATION_OPTABS): Update after previous field change.
(TREE_OPTIMIZATION_BASE_OPTABS): New macro.
(save_optabs_if_changed): Replace with...
(init_tree_optimization_optabs): ...this.
* optabs.c (save_optabs_if_changed): Rename to...
(init_tree_optimization_optabs): ...this.  Take the optimization node
as argument.  Do nothing if the base optabs are already correct.
Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
to recompute optabs.
* function.h (function): Remove optabs field.
* function.c (invoke_set_current_function_hook): Call
init_tree_optimization_optabs.  Use the result to initialize
this_fn_optabs.

gcc/c-family/
PR middle-end/56524
* c-common.c (handle_optimize_attribute): Don't call
save_optabs_if_changed.

gcc/testsuite/
PR middle-end/56524
* gcc.target/mips/pr56524.c: New test.

From-SVN: r196570

11 years agoDaily bump.
GCC Administrator [Sat, 9 Mar 2013 00:19:09 +0000 (00:19 +0000)]
Daily bump.

From-SVN: r196567

11 years agore PR c++/56565 (lambda function in NSMI fails to capture non-static data member.)
Paolo Carlini [Fri, 8 Mar 2013 22:52:36 +0000 (22:52 +0000)]
re PR c++/56565 (lambda function in NSMI fails to capture non-static data member.)

2013-03-08  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/56565
* g++.dg/cpp0x/lambda/lambda-nsdmi2.C: New.

From-SVN: r196564

11 years agore PR c++/51412 ([c++0x] Broken diagnostic with invalid lambda expressions)
Paolo Carlini [Fri, 8 Mar 2013 22:37:04 +0000 (22:37 +0000)]
re PR c++/51412 ([c++0x] Broken diagnostic with invalid lambda expressions)

/cp
2013-03-08  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/51412
* cxx-pretty-print.c (pp_cxx_expression): Handle LAMBDA_EXPR.
* error.c (dump_expr): Likewise.

/testsuite
2013-03-08  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/51412
* g++.dg/cpp0x/lambda/lambda-err3.C: New.

From-SVN: r196563

11 years agotrans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE if GTMA_HAS_NO_INSTR...
Aldy Hernandez [Fri, 8 Mar 2013 22:26:37 +0000 (22:26 +0000)]
trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE if GTMA_HAS_NO_INSTRUMENTATION.

* trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
if GTMA_HAS_NO_INSTRUMENTATION.
(generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
(ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
* gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
* gimple-pretty-print.c (dump_gimple_transaction): Handle
GTMA_HAS_NO_INSTRUMENTATION.

From-SVN: r196562

11 years agomemoryfwd.h: New.
Benjamin Kosnik [Fri, 8 Mar 2013 21:39:24 +0000 (21:39 +0000)]
memoryfwd.h: New.

* include/bits/memoryfwd.h: New. Centralize forward declarations.
* include/bits/algorithmfwd.h: Consistent comments.
* include/bits/localefwd.h: Same.
* include/ext/vstring_fwd.h: Same.
* include/parallel/algorithmfwd.h: Same.
* include/parallel/numericfwd.h: Same.
* include/std/iosfwd: Same.

* include/bits/alloc_traits.h: Include memoryfwd.h, remove allocator
forward decl.
* include/ext/alloc_traits.h: Remove allocator forward decl.
* include/bits/stl_construct.h: Same.
* include/bits/stringfwd.h: Include memoryfwd.h, remove allocator
forward decl. Consistent comments.

* doc/doxygen/user.cfg.in (TEMPLATE_RELATIONS): To NO.

* include/Makefile.am (bits_headers): Add memoryfwd.h.
* include/Makefile.in: Regenerate.
* configure: Same.

e enter the commit message for your changes. Lines starting

From-SVN: r196559

11 years agovector.tcc (vector<>operator=(const vector<>&): Reset pointers after deallocation...
François Dumont [Fri, 8 Mar 2013 20:08:20 +0000 (20:08 +0000)]
vector.tcc (vector<>operator=(const vector<>&): Reset pointers after deallocation when memory can be reused.

2013-03-08  François Dumont  <fdumont@gcc.gnu.org>

* include/bits/vector.tcc (vector<>operator=(const vector<>&):
Reset pointers after deallocation when memory can be reused.
* testsuite/23_containers/vector/allocator/minimal.cc: Insert
elements to really challenge C++11 allocator integration.
* testsuite/23_containers/vector/allocator/copy.cc: Likewise.
* testsuite/23_containers/vector/allocator/copy_assign.cc:
Likewise.
* testsuite/23_containers/vector/allocator/move_assign.cc:
Likewise.
* testsuite/23_containers/vector/allocator/swap.cc: Likewise and
swap vector back before checks on memory/personality mapping are
performed.
* testsuite/23_containers/forward_list/allocator/minimal.cc:
Insert element to really challenge C++11 allocator integration.
* testsuite/23_containers/forward_list/allocator/copy.cc:
Likewise.
* testsuite/23_containers/forward_list/allocator/copy_assign.cc:
Likewise.
* testsuite/23_containers/forward_list/allocator/move_assign.cc:
Likewise.
* testsuite/23_containers/forward_list/allocator/swap.cc: Likewise
and swap forward_list back before checks on memory/personality
mapping are performed.

From-SVN: r196557

11 years agognu-user.h (LIBTSAN_EARLY_SPEC): Don't link against libasan_preinit.o.
Jakub Jelinek [Fri, 8 Mar 2013 17:16:45 +0000 (18:16 +0100)]
gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against libasan_preinit.o.

* config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
libasan_preinit.o.

From-SVN: r196554

11 years agore PR c++/52183 ([C++0x] Redeclaration of __this in lambda inside template member...
Jason Merrill [Fri, 8 Mar 2013 16:10:19 +0000 (11:10 -0500)]
re PR c++/52183 ([C++0x] Redeclaration of __this in lambda inside template member function when calling static class function)

PR c++/52183
Correct PR tag.

From-SVN: r196552

11 years agore PR c++/51884 ([C++11] ICE with local class/lambda template argument)
Jason Merrill [Fri, 8 Mar 2013 16:04:15 +0000 (11:04 -0500)]
re PR c++/51884 ([C++11] ICE with local class/lambda template argument)

PR c++/51884
* class.c (modify_all_vtables): Mangle the vtable name before
entering dfs_walk.

From-SVN: r196551

11 years agosemantics.c (lambda_expr_this_capture): In unevaluated context, just return the neare...
Jason Merrill [Fri, 8 Mar 2013 16:04:02 +0000 (11:04 -0500)]
semantics.c (lambda_expr_this_capture): In unevaluated context, just return the nearest 'this'.

* semantics.c (lambda_expr_this_capture): In unevaluated context,
just return the nearest 'this'.

From-SVN: r196550

11 years agore PR c++/51494 (Legal program rejection - capturing "this" when using static method...
Jason Merrill [Fri, 8 Mar 2013 16:03:48 +0000 (11:03 -0500)]
re PR c++/51494 (Legal program rejection - capturing "this" when using static method inside lambda)

PR c++/51494
PR c++/51884
PR c++/56222
* tree.c (maybe_dummy_object): Don't capture 'this'.
* semantics.c (maybe_resolve_dummy): New.
(finish_non_static_data_member): Use it.
(finish_qualified_id_expr): Don't test is_dummy_object.
* cp-tree.h: Declare maybe_resolve_dummy.
* call.c (build_new_method_call_1): Use it.

From-SVN: r196549

11 years agore PR c++/56567 (ICE with lambda return type deduction and braced-init-list)
Jason Merrill [Fri, 8 Mar 2013 15:55:18 +0000 (10:55 -0500)]
re PR c++/56567 (ICE with lambda return type deduction and braced-init-list)

PR c++/56567
* semantics.c (apply_deduced_return_type): Don't allow returning
std::initializer_list.

From-SVN: r196548

11 years agore PR tree-optimization/56478 (ICE: Floating point exception in tree_estimate_probabi...
Marek Polacek [Fri, 8 Mar 2013 14:41:14 +0000 (14:41 +0000)]
re PR tree-optimization/56478 (ICE: Floating point exception in tree_estimate_probability)

PR tree-optimization/56478
* predict.c (is_comparison_with_loop_invariant_p): Change the
type of loop_step to tree.
(predict_loops): Adjust.
(predict_iv_comparison): Perform the computations on double_ints.

Co-Authored-By: Jakub Jelinek <jakub@redhat.com>
From-SVN: r196547

11 years agore PR lto/56570 (ICE when streaming a TREE_BLOCK)
Richard Biener [Fri, 8 Mar 2013 13:46:18 +0000 (13:46 +0000)]
re PR lto/56570 (ICE when streaming a TREE_BLOCK)

2013-03-08  Richard Biener  <rguenther@suse.de>

PR tree-optimization/56570
* tree-cfg.c (verify_expr_location_1): Verify locations for
DECL_DEBUG_EXPR.
* tree-sra.c (create_access_replacement): Strip locations
from DECL_DEBUG_EXPRs.

From-SVN: r196546

11 years agobuiltins.exp: Add for mingw targets linker option --allow-multiple-definition.
Kai Tietz [Fri, 8 Mar 2013 13:10:20 +0000 (14:10 +0100)]
builtins.exp: Add for mingw targets linker option --allow-multiple-definition.

* gcc.c-torture/execute/builtins/builtins.exp: Add for mingw
targets linker option --allow-multiple-definition.

From-SVN: r196545

11 years agoFix typo
Kai Tietz [Fri, 8 Mar 2013 13:07:03 +0000 (14:07 +0100)]
Fix typo

From-SVN: r196544

11 years agotree-inline.c (expand_call_inline): Do not associate a BLOCK with the location in...
Richard Biener [Fri, 8 Mar 2013 11:29:06 +0000 (11:29 +0000)]
tree-inline.c (expand_call_inline): Do not associate a BLOCK with the location in BLOCK_SOURCE_LOCATION.

2013-03-08  Richard Biener  <rguenther@suse.de>

* tree-inline.c (expand_call_inline): Do not associate
a BLOCK with the location in BLOCK_SOURCE_LOCATION.
* tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.

From-SVN: r196542

11 years agopr14092-1.c: Mark intptr_t typedef to use extension.
Kai Tietz [Fri, 8 Mar 2013 08:46:49 +0000 (09:46 +0100)]
pr14092-1.c: Mark intptr_t typedef to use extension.

* gcc.dg/pr14092-1.c: Mark intptr_t typedef to use extension.
* gcc.dg/pr24683.c: Avoid warning about casting constant string.
* gcc.dg/pr52549.c: Add LLP64 case.
* gcc.dg/pr53701.c: Use for uintptr_t typedef __UINTPTR_TYPE__.
* gcc.dg/pr56510.c: Adjust for LLP64 targets.
* gcc.dg/torture/pr51071-2.c: Likewise.
* gcc.dg/tree-ssa/vrp72.c: Likewise.
* gcc.dg/tree-ssa/vrp73.c: Likewise.
* gcc.dg/tree-ssa/vrp75.c: Likewise.
* gcc.dg/torture/pr53922.c: Skike for mingw-targets.
* gcc.dg/weak/weak-1.c: Likewise.
* gcc.dg/weak/weak-2.c: Likewise.
* gcc.dg/weak/weak-3.c: Likewise.
* gcc.dg/weak/weak-4.c: Likewise.
* gcc.dg/weak/weak-5.c: Likewise.
* gcc.dg/weak/weak-15.c: Likewise.
* gcc.dg/weak/weak-16.c: Likewise.

From-SVN: r196540

11 years agotree-ssa-ter.c (is_replaceable_p): Do not TER across location or block changes with...
Richard Biener [Fri, 8 Mar 2013 08:45:29 +0000 (08:45 +0000)]
tree-ssa-ter.c (is_replaceable_p): Do not TER across location or block changes with -Og.

2013-03-08  Richard Biener  <rguenther@suse.de>

* tree-ssa-ter.c (is_replaceable_p): Do not TER across location
or block changes with -Og.  Fix for location / block encoding
changes and PHI arguments with locations.

From-SVN: r196539

11 years agopr54486.c: Skip test for mingw-targets.
Kai Tietz [Fri, 8 Mar 2013 08:40:32 +0000 (09:40 +0100)]
pr54486.c: Skip test for mingw-targets.

* c-c++-common/pr54486.c: Skip test for mingw-targets.

From-SVN: r196538

11 years agore PR ada/52123 (gcc bootstrap with ada fails on mingw target)
Cesar Strauss [Fri, 8 Mar 2013 08:04:00 +0000 (08:04 +0000)]
re PR ada/52123 (gcc bootstrap with ada fails on mingw target)

PR ada/52123
* seh_init.c (Raise_From_Signal_Handler): Declare as no-return.
(__gnat_SEH_error_handler): Likewise.  Remove final return.

From-SVN: r196536

11 years agocpuinfo.c (get_intel_cpu): Fix cpuid codes for sandybridge processors.
Sriraman Tallam [Fri, 8 Mar 2013 01:02:29 +0000 (01:02 +0000)]
cpuinfo.c (get_intel_cpu): Fix cpuid codes for sandybridge processors.

2013-03-07  Sriraman Tallam  <tmsriram@google.com>

* config/i386/cpuinfo.c (get_intel_cpu): Fix cpuid codes for
sandybridge processors.

From-SVN: r196532

11 years agoDaily bump.
GCC Administrator [Fri, 8 Mar 2013 00:19:24 +0000 (00:19 +0000)]
Daily bump.

From-SVN: r196531

11 years agoclosures.c (is_emutramp_enabled [!FFI_MMAP_EXEC_EMUTRAMP_PAX]): Move default definiti...
Dave Korn [Thu, 7 Mar 2013 22:25:20 +0000 (22:25 +0000)]
closures.c (is_emutramp_enabled [!FFI_MMAP_EXEC_EMUTRAMP_PAX]): Move default definition outside enclosing #if scope.

* src/closures.c (is_emutramp_enabled [!FFI_MMAP_EXEC_EMUTRAMP_PAX]):
Move default definition outside enclosing #if scope.

From-SVN: r196527

11 years agobitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT for all counters.
Steven Bosscher [Thu, 7 Mar 2013 17:27:52 +0000 (17:27 +0000)]
bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT for all counters.

* bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
for all counters.
(struct output_info): Likewise.
(register_overhead): Remove bad gcc_assert.
(bitmap_find_bit): If there is only a single bitmap element, do not
count a miss as a search.
(print_statistics): Update for counter type changes.
(dump_bitmap_statistics): Likewise.  Print headers such that they
are properly lined up with the printed counters.

From-SVN: r196525

11 years agobaseline_symbols.txt: Update.
Jakub Jelinek [Thu, 7 Mar 2013 15:47:47 +0000 (16:47 +0100)]
baseline_symbols.txt: Update.

* config/abi/post/i386-linux-gnu/baseline_symbols.txt: Update.
* config/abi/post/i486-linux-gnu/baseline_symbols.txt: Likewise.
* config/abi/post/powerpc64-linux-gnu/32/baseline_symbols.txt:
Likewise.
* config/abi/post/powerpc64-linux-gnu/baseline_symbols.txt: Likewise.
* config/abi/post/powerpc-linux-gnu/baseline_symbols.txt: Likewise.
* config/abi/post/s390-linux-gnu/baseline_symbols.txt: Likewise.
* config/abi/post/s390x-linux-gnu/baseline_symbols.txt: Likewise.
* config/abi/post/x86_64-linux-gnu/32/baseline_symbols.txt: Likewise.
* config/abi/post/x86_64-linux-gnu/baseline_symbols.txt: Likewise.

From-SVN: r196523

11 years agore PR middle-end/56559 (ICE in propagate_op_to_single_use, at tree-ssa-reassoc.c...
Jakub Jelinek [Thu, 7 Mar 2013 14:34:15 +0000 (15:34 +0100)]
re PR middle-end/56559 (ICE in propagate_op_to_single_use, at tree-ssa-reassoc.c:1057)

PR tree-optimization/56559
* tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
check that it has only a single use.

* gcc.dg/tree-ssa/reassoc-26.c: New test.

From-SVN: r196520

11 years ago* gcc.dg/pr31490.c: Fix last change.
Andreas Schwab [Thu, 7 Mar 2013 10:29:45 +0000 (10:29 +0000)]
* gcc.dg/pr31490.c: Fix last change.

From-SVN: r196519

11 years agoinvoke.texi (fwhole-program): Discourage use in combination with -flto.
Richard Biener [Thu, 7 Mar 2013 08:54:49 +0000 (08:54 +0000)]
invoke.texi (fwhole-program): Discourage use in combination with -flto.

2013-03-07  Richard Biener  <rguenther@suse.de>

* doc/invoke.texi (fwhole-program): Discourage use in combination
with -flto.

From-SVN: r196518

11 years agore PR libquadmath/55473 (quadmath.h should have extern "C" for C++ users)
Shakthi Kannan [Thu, 7 Mar 2013 07:53:10 +0000 (07:53 +0000)]
re PR libquadmath/55473 (quadmath.h should have extern "C" for C++ users)

PR libquadmath/55473
* quadmath.h: Add 'extern "C"' block for C++ use.

From-SVN: r196517

11 years agoDaily bump.
GCC Administrator [Thu, 7 Mar 2013 00:19:06 +0000 (00:19 +0000)]
Daily bump.

From-SVN: r196516

11 years agore PR c++/56534 (ICE Segfault on invalid code in check_elaborated_type_specifier)
Paolo Carlini [Wed, 6 Mar 2013 23:47:20 +0000 (23:47 +0000)]
re PR c++/56534 (ICE Segfault on invalid code in check_elaborated_type_specifier)

/cp
2013-03-06  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/56534
* parser.c (cp_parser_elaborated_type_specifier): Don't call
check_elaborated_type_specifier when TREE_CODE (decl) != TYPE_DECL.
* decl.c (check_elaborated_type_specifier): Tidy.

/testsuite
2013-03-06  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/56534
* g++.dg/template/crash115.C: New.

From-SVN: r196513

11 years ago* config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
Jakub Jelinek [Wed, 6 Mar 2013 23:29:08 +0000 (00:29 +0100)]
* config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.

From-SVN: r196512

11 years agore PR tree-optimization/56539 (ICE: verify_ssa failed: caused by -foptimize-sibling...
Jakub Jelinek [Wed, 6 Mar 2013 23:26:42 +0000 (00:26 +0100)]
re PR tree-optimization/56539 (ICE: verify_ssa failed: caused by -foptimize-sibling-calls)

PR tree-optimization/56539
* tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
instead of GSI_CONTINUE_LINKING as last argument to
force_gimple_operand_gsi.

* gcc.c-torture/compile/pr56539.c: New test.

From-SVN: r196511

11 years ago20090914-2_0.c: Skip for mingw and cygwin targets.
Kai Tietz [Wed, 6 Mar 2013 18:29:46 +0000 (19:29 +0100)]
20090914-2_0.c: Skip for mingw and cygwin targets.

* gcc.dg/lto/20090914-2_0.c: Skip for mingw and cygwin
targets.
* gcc.dg/lto/20091013-1_1.c: Set x64-mingw as xfail.
* gcc.dg/lto/20091013-1_2.c: Likewise.
* gcc.dg/pr31490.c: Adjust for LLP64 targets.

From-SVN: r196508

11 years agotrans.c (Attribute_to_gnu): Abort instead of erroring out for an unimplemented attribute.
Eric Botcazou [Wed, 6 Mar 2013 18:07:14 +0000 (18:07 +0000)]
trans.c (Attribute_to_gnu): Abort instead of erroring out for an unimplemented attribute.

* gcc-interface/trans.c (Attribute_to_gnu): Abort instead of erroring
out for an unimplemented attribute.

From-SVN: r196507

11 years agodecl.c (gnat_to_gnu_field): Remove the wrapper around a misaligned integral type...
Eric Botcazou [Wed, 6 Mar 2013 18:00:50 +0000 (18:00 +0000)]
decl.c (gnat_to_gnu_field): Remove the wrapper around a misaligned integral type if...

* gcc-interface/decl.c (gnat_to_gnu_field): Remove the wrapper around
a misaligned integral type if a size is specified for the field.

From-SVN: r196506

11 years agotrans.c (Raise_Error_to_gnu): Record the unpadded type of the index type on the RCI...
Eric Botcazou [Wed, 6 Mar 2013 17:49:33 +0000 (17:49 +0000)]
trans.c (Raise_Error_to_gnu): Record the unpadded type of the index type on the RCI stack.

* gcc-interface/trans.c (Raise_Error_to_gnu) <CE_Index_Check_Failed>:
Record the unpadded type of the index type on the RCI stack.

From-SVN: r196504

11 years agotrans.c (emit_range_check): Assert that the range type is a numerical type and remove...
Eric Botcazou [Wed, 6 Mar 2013 17:34:02 +0000 (17:34 +0000)]
trans.c (emit_range_check): Assert that the range type is a numerical type and remove useless local variables.

* gcc-interface/trans.c (emit_range_check): Assert that the range type
is a numerical type and remove useless local variables.

From-SVN: r196500

11 years agot-aarch64 (TM_H, [...]): Add aarch64-cores.def.
Jakub Jelinek [Wed, 6 Mar 2013 17:21:26 +0000 (18:21 +0100)]
t-aarch64 (TM_H, [...]): Add aarch64-cores.def.

* config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
aarch64-cores.def.

From-SVN: r196499

11 years agore PR middle-end/56548 (ICE in emit_move_insn, at expr.c:3486 with -march=pentium...
Jakub Jelinek [Wed, 6 Mar 2013 17:18:46 +0000 (18:18 +0100)]
re PR middle-end/56548 (ICE in emit_move_insn, at expr.c:3486 with -march=pentium{pro,2,3} -O3)

PR middle-end/56548
* expr.c (expand_cond_expr_using_cmove): When expanding cmove in
promoted mode, convert the result back to the original mode.

* gcc.dg/pr56548.c: New test.

From-SVN: r196498

11 years agore PR middle-end/56461 (GCC is leaking lots of memory)
Jakub Jelinek [Wed, 6 Mar 2013 16:18:40 +0000 (17:18 +0100)]
re PR middle-end/56461 (GCC is leaking lots of memory)

PR middle-end/56461
* internal.h (struct cpp_buffer): Add to_free field.
(_cpp_pop_file_buffer): Add third argument.
* files.c (_cpp_stack_file): Set buffer->to_free.
(_cpp_pop_file_buffer): Add to_free argument.  Free to_free
if non-NULL, and if equal to file->buffer_start, also clear
file->buffer{,_start,_valid}.
* directives.c (_cpp_pop_buffer): Pass buffer->to_free
to _cpp_pop_file_buffer.

From-SVN: r196497

11 years agothunk1.C bikeshedding
Jason Merrill [Wed, 6 Mar 2013 16:14:53 +0000 (11:14 -0500)]
thunk1.C bikeshedding

From-SVN: r196496

11 years agofix dg-skip-if in thunk1.C
Jason Merrill [Wed, 6 Mar 2013 16:00:10 +0000 (11:00 -0500)]
fix dg-skip-if in thunk1.C

From-SVN: r196495

11 years agore PR debug/53363 (g++.dg/debug/dwarf2/thunk1.C FAILs)
Rainer Orth [Wed, 6 Mar 2013 15:34:11 +0000 (15:34 +0000)]
re PR debug/53363 (g++.dg/debug/dwarf2/thunk1.C FAILs)

PR debug/53363
* g++.dg/debug/dwarf2/thunk1.C: Skip on darwin.

From-SVN: r196493

11 years agore PR c++/56543 (ICE on valid code in copy_node_stat)
Jakub Jelinek [Wed, 6 Mar 2013 15:19:11 +0000 (16:19 +0100)]
re PR c++/56543 (ICE on valid code in copy_node_stat)

PR c++/56543
* tree.c (strip_typedefs): Don't copy args if they are NULL.

* g++.dg/template/typename20.C: New test.

From-SVN: r196491

11 years agoUse thr_stksegment to determine Solaris stack base
Rainer Orth [Wed, 6 Mar 2013 15:08:58 +0000 (15:08 +0000)]
Use thr_stksegment to determine Solaris stack base

* os_dep.c [SOLARIS_STACKBOTTOM] (GC_solaris_stack_base): New
function.
[!BEOS && !AMIGA && !MSWIN32 && !MSWINCE && !OS2 && !NOSYS &&
!ECOS] (GC_get_stack_base): Use it.
* include/private/gcconfig.h [SPARC && SUNOS5]
(SOLARIS_STACKBOTTOM): Define.
(STACKBOTTOM, HEURISTIC2): Remove.
[I386 && SUNOS5]
(SOLARIS_STACKBOTTOM): Define.
(STACKBOTTOM): Remove.

From-SVN: r196490

11 years agoFilter out basever symbols on Solaris
Rainer Orth [Wed, 6 Mar 2013 13:34:35 +0000 (13:34 +0000)]
Filter out basever symbols on Solaris

* scripts/extract_symvers.pl: Omit symbols bound to base
versions.
* scripts/extract_symvers.in: Likewise.
* config/abi/post/solaris2.9/baseline_symbols.txt: Regenerate.
* config/abi/post/solaris2.9/sparcv9/baseline_symbols.txt: Likewise.
* config/abi/post/solaris2.10/baseline_symbols.txt: Likewise.
* config/abi/post/solaris2.10/amd64/baseline_symbols.txt: Likewise.
* config/abi/post/solaris2.10/sparcv9/baseline_symbols.txt: Likewise.

From-SVN: r196489

11 years agore PR tree-optimization/56294 (BOOT_CFLAGS='-O2 -g -fno-ipa-sra' leads to bootstrap...
Richard Biener [Wed, 6 Mar 2013 11:24:07 +0000 (11:24 +0000)]
re PR tree-optimization/56294 (BOOT_CFLAGS='-O2 -g -fno-ipa-sra' leads to bootstrap comparison failure)

2013-03-06  Richard Biener  <rguenther@suse.de>

PR middle-end/56294
* tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
(insert_updated_phi_nodes_compare_uids): New function.
(update_ssa): Sort symbols_to_rename after UID before
traversing it to insert PHI nodes.

From-SVN: r196488

11 years agore PR lto/50494 (gcc.dg/vect/vect-reduc-2char.c fails spuriously on ppc with -flto)
Richard Biener [Wed, 6 Mar 2013 08:38:46 +0000 (08:38 +0000)]
re PR lto/50494 (gcc.dg/vect/vect-reduc-2char.c fails spuriously on ppc with -flto)

2013-03-06  Richard Biener  <rguenther@suse.de>

PR middle-end/50494
* tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.

Revert
2013-02-13  Richard Biener  <rguenther@suse.de>

PR lto/50494
* varasm.c (output_constant_def_1): Get the decl representing
the constant as argument.
(output_constant_def): Wrap output_constant_def_1.
(make_decl_rtl): Use output_constant_def_1 with the decl
representing the constant.
(build_constant_desc): Optionally re-use a decl already
representing the constant.
(tree_output_constant_def): Adjust.

From-SVN: r196487

11 years agomake-obstacks-texi.pl: New.
Tobias Burnus [Wed, 6 Mar 2013 08:38:22 +0000 (09:38 +0100)]
make-obstacks-texi.pl: New.

contrib/
2013-03-06  Tobias Burnus  <burnus@net-b.de>

        * make-obstacks-texi.pl: New.

libiberty/
2013-03-06  Tobias Burnus  <burnus@net-b.de>

        * libiberty.texi: Update comment, remove lowersections.
        * obstacks.texi: Regenerate.

From-SVN: r196486

11 years agore PR lto/50293 (-flto fails if GCC is installed in directory with space in path...
Joey Ye [Wed, 6 Mar 2013 06:32:03 +0000 (06:32 +0000)]
re PR lto/50293 (-flto fails if GCC is installed in directory with space in path name)

2013-03-06  Joey Ye  <joey.ye@arm.com>

    PR lto/50293
    * gcc.c (convert_white_space): New function.
    (main): Handles white space in function name.

From-SVN: r196485

11 years agore PR target/56529 ([SH] Calls to __sdivsi3_i4i and __udivsi3_i4i are generated on...
Oleg Endo [Wed, 6 Mar 2013 00:41:25 +0000 (00:41 +0000)]
re PR target/56529 ([SH] Calls to __sdivsi3_i4i and __udivsi3_i4i are generated on SH2)

PR target/56529
* config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
to SH_DIV_CALL_TABLE for TARGET_SH2.
* config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
list.
* doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
call-table options.

PR target/56529
* config/sh/lib1funcs.S (udivsi3_i4i, sdivsi3_i4i): Add __SH2A__ to
inclusion list.

From-SVN: r196484

11 years agoDaily bump.
GCC Administrator [Wed, 6 Mar 2013 00:19:47 +0000 (00:19 +0000)]
Daily bump.

From-SVN: r196483

11 years agore PR debug/55364 (ICE: in remove_addr_table_entry, at dwarf2out.c:4201 with -O ...
Sterling Augustine [Tue, 5 Mar 2013 23:08:45 +0000 (23:08 +0000)]
re PR debug/55364 (ICE: in remove_addr_table_entry, at dwarf2out.c:4201 with -O -gsplit-dwarf)

gcc/
PR debug/55364
* dwarf2out.c (resolve_addr): Don't call
remove_loc_list_addr_table_entries a second time for the same
expression.

Co-Authored-By: Cary Coutant <ccoutant@google.com>
From-SVN: r196480

11 years agore PR debug/56510 (More var-tracking scalability problems)
Jakub Jelinek [Tue, 5 Mar 2013 22:31:50 +0000 (23:31 +0100)]
re PR debug/56510 (More var-tracking scalability problems)

PR debug/56510
* cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
(avoid_complex_debug_insns): New function.
(expand_debug_locations): Call it.

* gcc.dg/pr56510.c: New test.

From-SVN: r196479

11 years agore PR rtl-optimization/56484 (ICE in assign_by_spills, at lra-assigns.c:1268)
Jakub Jelinek [Tue, 5 Mar 2013 22:25:43 +0000 (23:25 +0100)]
re PR rtl-optimization/56484 (ICE in assign_by_spills, at lra-assigns.c:1268)

PR rtl-optimization/56484
* ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
lifetimes of hard registers on small register class machines.

From-SVN: r196478

11 years agoAdd fast interrupt.
David Holsgrove [Tue, 5 Mar 2013 19:34:25 +0000 (19:34 +0000)]
Add fast interrupt.

*  config/microblaze/microblaze-protos.h: Rename
microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
*  config/microblaze/microblaze.c (microblaze_attribute_table): Add
fast_interrupt.
(microblaze_fast_interrupt_function_p): New function.
(microblaze_is_interrupt_handler): Rename to
microblaze_is_interrupt_variant and add fast_interrupt check.
(microblaze_must_save_register): Use microblaze_is_interrupt_variant.
(save_restore_insns): Likewise.
(compute_frame_size): Likewise.
(microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
(microblaze_globalize_label): Likewise.
*  config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
*  config/microblaze/microblaze.md: Use wrapper
microblaze_is_interrupt_variant.

From-SVN: r196474

11 years agore PR c++/50276 ([C++0x] Wrong "used uninitialized in this function" warning)
Jason Merrill [Tue, 5 Mar 2013 18:58:36 +0000 (13:58 -0500)]
re PR c++/50276 ([C++0x] Wrong "used uninitialized in this function" warning)

PR c++/50276
* g++.dg/cpp0x/lambda/lambda-template10.C: New.

From-SVN: r196472

11 years agosdbout.c (sdbout_one_type): Switch to current function's section supporting cold...
Kai Tietz [Tue, 5 Mar 2013 17:59:38 +0000 (18:59 +0100)]
sdbout.c (sdbout_one_type): Switch to current function's section supporting cold/hot.

* sdbout.c (sdbout_one_type): Switch to current function's section
supporting cold/hot.

From-SVN: r196471

11 years agoinvoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian, -mxl-reorder.
David Holsgrove [Tue, 5 Mar 2013 16:00:26 +0000 (16:00 +0000)]
invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian, -mxl-reorder.

* doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
-mxl-reorder.

From-SVN: r196470

11 years agore PR middle-end/56461 (GCC is leaking lots of memory)
Jakub Jelinek [Tue, 5 Mar 2013 15:51:48 +0000 (16:51 +0100)]
re PR middle-end/56461 (GCC is leaking lots of memory)

PR middle-end/56461
* ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
if VALGRIND_GET_VBITS is defined, temporarily make object
memory all defined, and restore previous valgrind addressability
and definability afterwards.  Free this_object at the end.

* c-pch.c (pch_init): Free target_validity at the end.

From-SVN: r196469

11 years agore PR middle-end/56461 (GCC is leaking lots of memory)
Jakub Jelinek [Tue, 5 Mar 2013 15:50:38 +0000 (16:50 +0100)]
re PR middle-end/56461 (GCC is leaking lots of memory)

PR middle-end/56461
* lra.c (lra): Call lra_clear_live_ranges if live_p,
right before calling lra_create_live_ranges, also call it
when clearing live_p.  Only call lra_clear_live_ranges
at the end if live_p.

From-SVN: r196468

11 years agore PR middle-end/56461 (GCC is leaking lots of memory)
Jakub Jelinek [Tue, 5 Mar 2013 15:49:43 +0000 (16:49 +0100)]
re PR middle-end/56461 (GCC is leaking lots of memory)

PR middle-end/56461
* sched-deps.c (delete_dep_node): Free DEP_REPLACE.

From-SVN: r196467

11 years agore PR middle-end/56526 (false positive for maybe-uninitialized)
Jakub Jelinek [Tue, 5 Mar 2013 15:16:49 +0000 (16:16 +0100)]
re PR middle-end/56526 (false positive for maybe-uninitialized)

PR middle-end/56526
* simple-object-mach-o.c (simple_object_mach_o_segment): Initialize
wrapper_sect_offset to avoid a warning.

From-SVN: r196466

11 years agore PR tree-optimization/56521 (Uninitialized value_id)
Richard Biener [Tue, 5 Mar 2013 15:02:58 +0000 (15:02 +0000)]
re PR tree-optimization/56521 (Uninitialized value_id)

2013-03-05  Richard Biener  <rguenther@suse.de>

PR tree-optimization/56521
* tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
value-id.

From-SVN: r196465

11 years agore PR c++/55135 (Segfault of gcc on a big file)
Steven Bosscher [Tue, 5 Mar 2013 14:45:23 +0000 (14:45 +0000)]
re PR c++/55135 (Segfault of gcc on a big file)

gcc/
PR c++/55135
* except.h (remove_unreachable_eh_regions): New prototype.
* except.c (remove_eh_handler_splicer): New function, split out
of remove_eh_handler.
(remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
warning about running it on many EH regions one at a time.
(remove_unreachable_eh_regions_worker): New function, walk the
EH tree in depth-first order and remove non-marked regions.
(remove_unreachable_eh_regions): New function.
* tree-eh.c (mark_reachable_handlers): New function, split out
from remove_unreachable_handlers.
(remove_unreachable_handlers): Use mark_reachable_handlers and
remove_unreachable_eh_regions.
(remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
and remove_unreachable_eh_regions.

From-SVN: r196464

11 years agoparser.c (cp_parser_braced_list): For {} initialize *non_constant_p to false.
Jakub Jelinek [Tue, 5 Mar 2013 14:14:16 +0000 (15:14 +0100)]
parser.c (cp_parser_braced_list): For {} initialize *non_constant_p to false.

* parser.c (cp_parser_braced_list): For {} initialize
*non_constant_p to false.

From-SVN: r196463

11 years agore PR middle-end/56525 (Access to ggc_freed memory in flow_loops_find)
Richard Biener [Tue, 5 Mar 2013 12:51:28 +0000 (12:51 +0000)]
re PR middle-end/56525 (Access to ggc_freed memory in flow_loops_find)

2013-03-05  Richard Biener  <rguenther@suse.de>

PR middle-end/56525
* loop-init.c (fix_loop_structure): Remove loops in two stages,
not freeing them until the end.

From-SVN: r196462

11 years agos390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
Andreas Krebbel [Tue, 5 Mar 2013 12:02:06 +0000 (12:02 +0000)]
s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.

2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>

* config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.

From-SVN: r196461

11 years agore PR c++/56530 (Bogus warning with -Wsign-conversion (again))
Paolo Carlini [Tue, 5 Mar 2013 11:15:17 +0000 (11:15 +0000)]
re PR c++/56530 (Bogus warning with -Wsign-conversion (again))

2013-03-05  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/56530
* g++.dg/warn/Wsign-conversion-2.C: New.

From-SVN: r196460

11 years agore PR c++/56530 (Bogus warning with -Wsign-conversion (again))
Paolo Carlini [Tue, 5 Mar 2013 11:14:28 +0000 (11:14 +0000)]
re PR c++/56530 (Bogus warning with -Wsign-conversion (again))

2013-03-05  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/56530
* g++.dg/warn/Wsign-conversion-2.C: New.

From-SVN: r196459

11 years agore PR tree-optimization/56270 (loop over array of struct float causes compiler error...
Richard Biener [Tue, 5 Mar 2013 09:54:29 +0000 (09:54 +0000)]
re PR tree-optimization/56270 (loop over array of struct float causes compiler error: segmentation fault)

2013-03-05  Richard Biener  <rguenther@suse.de>

PR tree-optimization/56270
* tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
of loads after scheduling an SLP instance.

* gcc.dg/vect/slp-38.c: New testcase.

From-SVN: r196458

11 years agoMakefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and tic6x.exp.
Jakub Jelinek [Tue, 5 Mar 2013 09:46:34 +0000 (10:46 +0100)]
Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and tic6x.exp.

* Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
tic6x.exp.
(check_gcc_parallelize): Run guality.exp as a separate job from
vect.exp with unsorted.exp and $(dg_target_exps) separately from
struct-layout-1.exp with stackalign.exp.

From-SVN: r196457

11 years ago* alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
Jakub Jelinek [Tue, 5 Mar 2013 09:40:38 +0000 (10:40 +0100)]
* alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.

From-SVN: r196456

11 years agore PR middle-end/56461 (GCC is leaking lots of memory)
Jakub Jelinek [Tue, 5 Mar 2013 09:39:43 +0000 (10:39 +0100)]
re PR middle-end/56461 (GCC is leaking lots of memory)

PR middle-end/56461
* tree-vect-slp.c (vect_supported_load_permutation_p): Free
load_index sbitmap even if some bit in it isn't set.

From-SVN: r196455

11 years agore PR middle-end/56461 (GCC is leaking lots of memory)
Jakub Jelinek [Tue, 5 Mar 2013 09:38:48 +0000 (10:38 +0100)]
re PR middle-end/56461 (GCC is leaking lots of memory)

PR middle-end/56461
* tree-ssa-loop-niter.c (bb_queue): Remove typedef.
(discover_iteration_bound_by_body_walk): Change queues to
vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
spelling in comment.  Call safe_push on queues[bound_index] directly.
Release queues[queue_index] in every iteration unconditionally.
Release bounds vector.

From-SVN: r196454

11 years agore PR middle-end/56461 (GCC is leaking lots of memory)
Jakub Jelinek [Tue, 5 Mar 2013 09:37:56 +0000 (10:37 +0100)]
re PR middle-end/56461 (GCC is leaking lots of memory)

PR middle-end/56461
* tree-vect-stmts.c (free_stmt_vec_info_vec): Call
free_stmt_vec_info on any left-over stmt_vec_info in the vector.
* tree-vect-loop.c (vect_create_epilog_for_reduction): Release
inner_phis vector.

From-SVN: r196453

11 years agore PR lto/56515 (location references block not in block tree, verify_gimple failed...
Richard Biener [Tue, 5 Mar 2013 09:20:30 +0000 (09:20 +0000)]
re PR lto/56515 (location references block not in block tree, verify_gimple failed (LTO + profile))

2013-03-05  Richard Biener  <rguenther@suse.de>

PR lto/56515
* tree-inline.c (remap_blocks_to_null): New function.
(expand_call_inline): When expanding a call stmt without
an associated block inline remap all callee blocks to NULL.

From-SVN: r196452

11 years agore PR rtl-optimization/56494 (ICE in simplify_truncation, at simplify-rtx.c:619)
Jakub Jelinek [Tue, 5 Mar 2013 06:04:14 +0000 (07:04 +0100)]
re PR rtl-optimization/56494 (ICE in simplify_truncation, at simplify-rtx.c:619)

PR rtl-optimization/56494
* simplify-rtx.c (simplify_truncation): If C is narrower than A,
optimize (truncate:A (subreg:B (truncate:C X) 0)) into
(subreg:A (truncate:C X) 0) instead of (truncate:A X).

* gcc.dg/pr56494.c: New test.

From-SVN: r196451

11 years agore PR middle-end/56461 (GCC is leaking lots of memory)
Jakub Jelinek [Tue, 5 Mar 2013 06:02:36 +0000 (07:02 +0100)]
re PR middle-end/56461 (GCC is leaking lots of memory)

PR middle-end/56461
* sel-sched-ir.c (free_sched_pools): Release
succs_info_pool.stack[succs_info_pool.max_top] vectors too
if succs_info_pool.max_top isn't -1.

From-SVN: r196450