Aldy Hernandez [Mon, 3 Dec 2012 16:11:21 +0000 (16:11 +0000)]
re PR middle-end/55401 (uninstrumented path in TM clones are still instrumented)
PR middle-end/55401
* trans-mem.c (get_tm_region_blocks): Exclude uninstrumented
blocks from vector if requested.
(collect_bb2reg): Pass new argument to
get_tm_region_blocks.
(get_bb_regions_instrumented): Add INCLUDE_UNINSTRUMENTED_P
argument, and pass it to expand_regions.
(execute_tm_mark): Pass new argument to
get_bb_regions_instrumented.
(execute_tm_edges): Same.
Jakub Jelinek [Mon, 3 Dec 2012 15:04:57 +0000 (16:04 +0100)]
asan.c (instrument_mem_region_access): Don't instrument if base doesn't have pointer type or len integral type.
* asan.c (instrument_mem_region_access): Don't instrument
if base doesn't have pointer type or len integral type.
Add cast if len doesn't have size_t compatible type.
(instrument_builtin_call): Don't instrument BUILT_IN_ATOMIC_LOAD,
BUILT_IN_ATOMIC_TEST_AND_SET, BUILT_IN_ATOMIC_CLEAR,
BUILT_IN_ATOMIC_EXCHANGE, BUILT_IN_ATOMIC_COMPARE_EXCHANGE
and BUILT_IN_ATOMIC_STORE.
Steven Bosscher [Mon, 3 Dec 2012 13:16:55 +0000 (13:16 +0000)]
rtl.h (print_insn_with_notes): Prototype.
* rtl.h (print_insn_with_notes): Prototype.
* sched-vis.c (print_insn_with_notes): Export it.
* gimple-pretty-print.h (gimple_dump_bb_for_graph): Prototype.
* gimple-pretty-print.c (print_gimple_expr): Flush the buffer.
(pp_gimple_stmt_1): Don't do it here.
(gimple_dump_bb_for_graph): New function.
* tree-pretty-print.c (print_generic_expr): Flush the buffer here.
(dump_generic_node): Don't flush the buffer here.
* graph.h (print_rtl_graph_with_bb): Rename to print_graph_cfg.
* graph.c: Include gimple.h, dumpfile.h, and gimple-pretty-print.h.
(draw_cfg_node): Handle GIMPLE basic blocks also.
(print_rtl_graph_with_bb): Rename to print_graph_cfg.
* passes.c (finish_optimization_passes): Don't finish graph dumps here.
(execute_function_dump): Use print_graph_cfg. Enable dumping the CFG
for GIMPLE also.
(pass_init_dump_file): Wrap in TV_DUMP. Set up CFG graph dumps.
(pass_fini_dump_file): Wrap in TV_DUMP. Finish graph dumps.
(execute_one_pass): Don't set up graph dumps here.
Jakub Jelinek [Mon, 3 Dec 2012 11:49:41 +0000 (12:49 +0100)]
asan-dg.exp (asan_get_gtest_test_list, [...]): New procedures.
* lib/asan-dg.exp (asan_get_gtest_test_list,
asan_get_gtest_expect_death_list, asan-gtest): New procedures.
(proc ${tool}_load): Remember [asan_get_gtest_test_list "$output"]
and [asan_get_gtest_expect_death_list "$output"] in global vars.
(asan_symbolize): Sanitize [] characters from key.
* g++.dg/asan/asan_test_config.h: New file.
* g++.dg/asan/asan_globals_test.cc: New file.
* g++.dg/asan/asan_test_utils.h: New file.
* g++.dg/asan/dejagnu-gtest.h: New file.
* g++.dg/asan/asan_test.cc: New file.
* g++.dg/asan/asan_test.C: New test.
Paul Thomas [Sun, 2 Dec 2012 15:23:30 +0000 (15:23 +0000)]
resolve.c (resolve_allocate_deallocate, [...]): Recover revisions 193568 and 193778...
2012-12-02 Paul Thomas <pault@gcc.gnu.org>
* resolve.c (resolve_allocate_deallocate,
resolve_typebound_intrinsic_op): Recover revisions 193568 and
193778, which were accidentally reverted by the previous patch.
gcc/
* doc/extend.texi: Various corrections to punctuation and grammar
throughout the file. Use consistent terminology and proper names.
Correct some minor markup issues.
re PR fortran/46897 ([OOP] type-bound defined ASSIGNMENT(=) not used for derived type component in intrinsic assign)
2012-12-01 Alessandro Fanfarillo <alessandro.fanfarillo@gmail.com>
Paul Thomas <pault@gcc.gnu.org>
PR fortran/46897
* gfortran.h : Add bit field 'defined_assign_comp' to
symbol_attribute structure.
Add primitive for gfc_add_full_array_ref.
* expr.c (gfc_add_full_array_ref): New function.
(gfc_lval_expr_from_sym): Call new function.
* resolve.c (add_comp_ref): New function.
(build_assignment): New function.
(get_temp_from_expr): New function
(add_code_to_chain): New function
(generate_component_assignments): New function that calls all
the above new functions.
(resolve_code): Call generate_component_assignments.
(check_defined_assignments): New function.
(resolve_fl_derived0): Call check_defined_assignments.
(gfc_resolve): Reset component_assignment_level in case it is
left in a bad state by errors.
2012-12-01 Alessandro Fanfarillo <alessandro.fanfarillo@gmail.com>
Paul Thomas <pault@gcc.gnu.org>
PR fortran/46897
* gfortran.dg/defined_assignment_1.f90: New test.
* gfortran.dg/defined_assignment_2.f90: New test.
* gfortran.dg/defined_assignment_3.f90: New test.
* gfortran.dg/defined_assignment_4.f90: New test.
* gfortran.dg/defined_assignment_5.f90: New test.
Co-Authored-By: Paul Thomas <pault@gcc.gnu.org>
From-SVN: r194016
Jakub Jelinek [Fri, 30 Nov 2012 23:51:26 +0000 (00:51 +0100)]
re PR c++/55542 (g++ segmentation fault)
PR c++/55542
* pt.c (make_ith_pack_parameter_name): Return NULL if
name is NULL.
(tsubst_decl): Call make_ith_pack_parameter_name even if
DECL_NAME is NULL.
David Edelsohn [Fri, 30 Nov 2012 17:54:07 +0000 (17:54 +0000)]
xcoffout.c (xcoff_tls_data_section_name): Define.
* xcoffout.c (xcoff_tls_data_section_name): Define.
* xcoffout.h (xcoff_tls_data_section_name): Declare.
* config/rs6000/rs6000.c (tls_data_section): Define.
(TARGET_USE_BLOCKS_FOR_DECL_P): Define.
(rs6000_legitimize_tls_address_aix): New function.
(rs6000_legitimize_tls_address): Use new function for AIX.
(rs6000_cannot_force_const_mem): No sum in TLS TOC symbols.
Allow TLS symbol in constant pool other than ELF.
(rs6000_legitimate_address_p): Allow TLS symbol other than ELF.
(rs6000_assemble_visibility): Do not emit anything on AIX.
(output_toc): Handle alias of TLS general-dynamic symbols.
Emit TLS decorations on symbols.
(rs6000_use_blocks_for_decl_p): New function.
(rs6000_xcoff_output_tls_section_asm_op): New function.
(rs6000_xcoff_asm_init_sections): Initialize tls_data_section.
(rs6000_xcoff_select_section): Choose tls_data_section for
thread-local storage.
(rs6000_xcoff_file_start): Generate xcoff_tls_data_section_name.
(rs6000_legitimate_constant_p): Allow TLS symbol other than ELF.
* config/rs6000/rs6000.md (tls_tls_): Restrict to ELF.
(tls_get_tpointer): New.
(tle_get_tpointer_internal): New.
(tls_get_addr<mode>): New.
(tls_get_addr_internal<mode>): New.
Teresa Johnson [Fri, 30 Nov 2012 16:47:04 +0000 (16:47 +0000)]
Revised patch to ensure that histograms from the profile summary are streamed...
Revised patch to ensure that histograms from the profile summary are streamed
through the LTO files so that the working set can be computed for use in
downstream optimizations.
2012-11-30 Teresa Johnson <tejohnson@google.com>
* lto-cgraph.c (output_profile_summary): Stream out sum_all
and histogram.
(input_profile_summary): Stream in sum_all and histogram.
(merge_profile_summaries): Merge sum_all and histogram, and
change to use RDIV.
(input_symtab): Call compute_working_sets after merging
summaries.
* gcov-io.c (gcov_histo_index): Make extern for compiler.
* gcov-io.h (gcov_histo_index): Ditto.
* profile.c (compute_working_sets): Remove static keyword.
* profile.h (compute_working_sets): Ditto.
* Makefile.in (lto-cgraph.o): Depend on profile.h.
Martin Jambor [Fri, 30 Nov 2012 16:11:33 +0000 (17:11 +0100)]
re PR middle-end/52890 (Revision 185336 causes 10% degradation on cpu2000 benchmark 252.eon)
2012-11-30 Martin Jambor <mjambor@suse.cz>
PR middle-end/52890
PR tree-optimization/55415
PR tree-optimization/54386
PR target/55448
* ipa-prop.c (ipa_modify_call_arguments): Be optimistic when
get_pointer_alignment_1 returns false and the base was not a
dereference.
* tree-sra.c (access_precludes_ipa_sra_p): New parameter req_align,
added check for required alignment. Update the user.
Steven Bosscher [Fri, 30 Nov 2012 09:46:02 +0000 (09:46 +0000)]
coretypes.h (struct pretty_print_info): Make a coretype.
* coretypes.h (struct pretty_print_info): Make a coretype.
(typedef pretty_printer): Likewise.
* pretty-print.h (typedef pretty_printer): Not needed here anymore.
* pretty-print.c (pp_write_text_as_dot_label_to_stream): New function.
* rtl.h (str_pattern_slim): New prototype.
(print_value, print_pattern, print_insn): Adjust prototypes to take
a pretty-printer rather than a char buffer.
* sched-vis.c (safe_concat): Remove.
(print_exp): Print into a pretty-printer.
(print_value): Likewise.
(print_pattern): Likewise.
(print_insn): Likewise.
(print_insn_with_notes): New static function.
(init_rtl_slim_pretty_print): New function.
(dump_value_slim): Simplify. Print into and flush a pretty-printer.
(dump_insn_slim): Likewise.
(dump_rtl_slim): Likewise.
(str_pattern_slim): New function.
* haifa-sched.c (model_recompute): Use str_pattern_slim instead of
static buffers.
(model_record_pressures): Likewise.
(schedule_insn): Likewise.
* sel-sched-dump.c (dump_insn_rtx_1): Likewise.
(sel_prepare_string_for_dot_label): Refer to graph.c CFG dumper code.
* graph.c: Include pretty-print.h.
(init_graph_slim_pretty_print): New function.
(print_escaped_line): Removed here, and reincarnated as
pp_write_text_as_dot_label_to_stream.
(draw_cfg_node): Print into a pretty printer.
(draw_cfg_node_succ_edges): Likewise.
(print_rtl_graph_with_bb): Likewise.
* Makefile.in (graph.o): Fix dependencies.
(tree-optimize.o, toplev.o, sched-vis.o): Likewise.
Jakub Jelinek [Fri, 30 Nov 2012 08:48:02 +0000 (09:48 +0100)]
tsan.c (is_load_of_const_p): Removed.
* tsan.c (is_load_of_const_p): Removed.
(instrument_expr): Use result of get_inner_reference
instead of get_base_address, avoid some unnecessary tests,
use !pt_solution_includes and !may_be_aliased tests to
check whether base might escape current function.
Jason Merrill [Thu, 29 Nov 2012 20:12:58 +0000 (15:12 -0500)]
re PR c++/53137 (g++ segfault)
PR c++/53137
* pt.c (tsubst_expr) [DECL_EXPR]: Set LAMBDA_EXPR_THIS_CAPTURE here.
(tsubst_copy_and_build) [LAMBDA_EXPR]: And clear it here.
(instantiate_class_template_1): Not here.
Vladimir Makarov [Thu, 29 Nov 2012 19:00:38 +0000 (19:00 +0000)]
re PR middle-end/55456 (454.calculix in SPEC CPU 2006 is miscompiled)
2012-11-29 Vladimir Makarov <vmakarov@redhat.com>
PR middle-end/55456
* lra-int.h (lra_new_regno_start): New external.
* lra.c (lra_new_regno_start): New global.
(lra): Set up lra_new_regno_start.
* lra-constraints.c (match_reload): Sync values only for original
pseudos.
Richard Earnshaw [Thu, 29 Nov 2012 17:51:40 +0000 (17:51 +0000)]
re PR target/55073 (Wrong Neon code generation at -O2 caused by -fschedule-insns)
PR target/55073
* arm/neon.md (neon_vtrn<mode>_internal): Split into expand
and insn patterns. Re-order insn arguments to tie inputs to
outputs.
(neon_vzip<mode>_internal): Likewise.
(neon_vuzp<mode>_internal): Likewise.
Richard Biener [Thu, 29 Nov 2012 13:53:42 +0000 (13:53 +0000)]
tree-ssa-pre.c (get_expr_value_id): Do not add expr to the set of value expressions here.
2012-11-29 Richard Biener <rguenther@suse.de>
* tree-ssa-pre.c (get_expr_value_id): Do not add expr
to the set of value expressions here.
(add_to_exp_gen, make_values_for_phi): Fold into ...
(compute_avail): ... here, and avoid useless work. Dump
avail sets in processing order.
(do_pre): Do not dump avail sets here.
Oleg Raikhman [Thu, 29 Nov 2012 03:05:23 +0000 (03:05 +0000)]
epiphany.h (FIXED_REGISTERS, [...]): Make r40..r43 call-used.
* config/epiphany/epiphany.h (FIXED_REGISTERS, CALL_USED_REGISTERS):
Make r40..r43 call-used.
(REG_ALLOC_ORDER): Sort r40..r43 into the call-used registers.