[PATCH 3/7] OpenMP 4.0 offloading infrastructure: Offload tables.

gcc/
* Makefile.in (GTFILES): Add omp-low.h to list of GC files.
* cgraphunit.c: Include omp-low.h.
* doc/tm.texi: Regenerate.
* doc/tm.texi.in (TARGET_RECORD_OFFLOAD_SYMBOL): Document.
* gengtype.c (open_base_files): Add omp-low.h to ifiles.
* lto-cgraph.c (output_offload_tables): New function.
(input_offload_tables): Likewise.
* lto-section-in.c (lto_section_name): Add "offload_table".
* lto-section-names.h (OFFLOAD_VAR_TABLE_SECTION_NAME): Define.
(OFFLOAD_FUNC_TABLE_SECTION_NAME): Likewise.
* lto-streamer-out.c (lto_output): Call output_offload_tables.
* lto-streamer.h (lto_section_type): Add LTO_section_offload_table.
(output_offload_tables, input_offload_tables): Declare.
* omp-low.c: Include common/common-target.h and lto-section-names.h.
(offload_funcs, offload_vars): New global <tree, va_gc> vectors.
(expand_omp_target): Add child_fn into offload_funcs vector.
(add_decls_addresses_to_decl_constructor): New function.
(omp_finish_file): Likewise.
* omp-low.h (omp_finish_file, offload_funcs, offload_vars): Declare.
* target.def (record_offload_symbol): New DEFHOOK.
* toplev.c: Include omp-low.h.
(compile_file): Call omp_finish_file.
* varpool.c: Include omp-low.h.
(varpool_node::get_create): Add decl into offload_vars vector.

gcc/lto/
* lto/lto.c (read_cgraph_and_symbols): Call input_offload_tables.

Co-Authored-By: Andrey Turetskiy <andrey.turetskiy@intel.com>
Co-Authored-By: Bernd Schmidt <bernds@codesourcery.com>
Co-Authored-By: Ilya Tocar <ilya.tocar@intel.com>
Co-Authored-By: Michael Zolotukhin <michael.v.zolotukhin@intel.com>
From-SVN: r217489

genmatch.c (dt_node::gen_kids): Fix placement of break statement.

2014-11-13  Richard Biener  <rguenther@suse.de>

* genmatch.c (dt_node::gen_kids): Fix placement of break statement.

From-SVN: r217487

[PATCH 2/7] OpenMP 4.0 offloading infrastructure: LTO streaming.

gcc/
* cgraph.c: Include context.h.
(cgraph_node::create): Set node->offloadable and g->have_offload if
decl have "omp declare target" attribute.
* cgraph.h (symtab_node): Add need_lto_streaming and offloadable flags.
* cgraphunit.c: Include lto-section-names.h.
(ipa_passes): Call ipa_write_summaries if there is something to write to
OFFLOAD_SECTION_NAME_PREFIX sections.
(symbol_table::compile): Set flag_generate_lto if there is something to
offload.
Replace flag_lto with flag_generate_lto before lto_streamer_hooks_init.
* context.c (gcc::context::context): Initialize have_offload with false.
* context.h (class context): Add have_offload flag.
* ipa-inline-analysis.c (inline_generate_summary): Do not exit under
flag_generate_lto.
(inline_free_summary): Always remove hooks.
* lto-cgraph.c (referenced_from_other_partition_p): Ignore references
from non-offloadable nodes while streaming a node into offload section.
(reachable_from_other_partition_p): Likewise.
(select_what_to_stream): New function.
(compute_ltrans_boundary): Do not call
lto_set_symtab_encoder_in_partition if the node should not be streamed.
* lto-section-names.h (OFFLOAD_SECTION_NAME_PREFIX): Define.
(section_name_prefix): Declare.
* lto-streamer.c (section_name_prefix): New variable.
(lto_get_section_name): Use section_name_prefix instead of
LTO_SECTION_NAME_PREFIX.
* lto-streamer.h (select_what_to_stream): Declare.
* omp-low.c: Include context.h.
(is_targetreg_ctx): New function.
(scan_sharing_clauses): Use offloadable flag, instead of an attribute.
(create_omp_child_function, check_omp_nesting_restrictions): Use new
is_targetreg_ctx function.  Replace usage of "omp declare target"
attribute with a cgraph_node flag offloadable.
(expand_omp_target): Set mark_force_output for offloadable functions.
(lower_omp_critical): Set offloadable flag for omp critical symbol.
* passes.c (ipa_write_summaries): New argument offload_lto_mode.  Call
select_what_to_stream.  Do not call lto_set_symtab_encoder_in_partition
if the node should not be streamed out.
* tree-pass.h (ipa_write_summaries): New bool argument.
* varpool.c: Include context.h.
(varpool_node::get_create): Set node->offloadable and g->have_offload if
decl have "omp declare target" attribute.

gcc/lto/
* lto-object.c (lto_obj_add_section): Use section_name_prefix instead of
LTO_SECTION_NAME_PREFIX.
* lto-partition.c (lto_promote_cross_file_statics): Call
select_what_to_stream.
* lto.c (lto_section_with_id): Use section_name_prefix instead of
LTO_SECTION_NAME_PREFIX.
(read_cgraph_and_symbols): Read OFFLOAD_SECTION_NAME_PREFIX sections, if
being built as an offload compiler.

Co-Authored-By: Andrey Turetskiy <andrey.turetskiy@intel.com>
Co-Authored-By: Bernd Schmidt <bernds@codesourcery.com>
Co-Authored-By: Michael Zolotukhin <michael.v.zolotukhin@intel.com>
From-SVN: r217486

[PATCH 1/7] OpenMP 4.0 offloading infrastructure: configure and make

* configure: Regenerate.
* configure.ac (--enable-as-accelerator-for)
(--enable-offload-targets): New configure options.
gcc/
* Makefile.in (real_target_noncanonical, accel_dir_suffix)
(enable_as_accelerator): New variables substituted by configure.
(libsubdir, libexecsubdir, unlibsubdir): Tweak for the possibility of
being configured as an offload compiler.
(DRIVER_DEFINES): Pass new defines DEFAULT_REAL_TARGET_MACHINE and
ACCEL_DIR_SUFFIX.
(install-cpp, install-common, install_driver, install-gcc-ar): Do not
install for the offload compiler.
* config.in: Regenerate.
* configure: Regenerate.
* configure.ac (real_target_noncanonical, accel_dir_suffix)
(enable_as_accelerator): Compute new variables.
(ACCEL_COMPILER): Define if the compiler is built as the accel compiler.
(OFFLOAD_TARGETS): List of target names suitable for offloading.
(ENABLE_OFFLOADING): Define if list of offload targets is not empty.
gcc/cp/
* Make-lang.in (c++.install-common): Do not install for the offload
compiler.
gcc/doc/
* install.texi (Options specification): Document
--enable-as-accelerator-for and --enable-offload-targets.
gcc/fortran/
* Make-lang.in (fortran.install-common): Do not install for the offload
compiler.
libgcc/
* Makefile.in (crtoffloadbegin$(objext)): New rule.
(crtoffloadend$(objext)): Likewise.
* configure: Regenerate.
* configure.ac (accel_dir_suffix): Compute new variable.
(extra_parts): Add crtoffloadbegin.o and crtoffloadend.o
if enable_offload_targets is not empty.
* offloadstuff.c: New file.
libgomp/
* config.h.in: Regenerate.
* configure: Regenerate.
* configure.ac: Check for libdl, required for plugin support.
(PLUGIN_SUPPORT): Define if plugins are supported.
(enable_offload_targets): Support Intel MIC targets.
(OFFLOAD_TARGETS): List of target names suitable for offloading.
lto-plugin/
* Makefile.am (libexecsubdir): Tweak for the possibility of being
configured for offload compiler.
(accel_dir_suffix, real_target_noncanonical): New variables substituted
by configure.
* Makefile.in: Regenerate.
* configure: Regenerate.
* configure.ac (accel_dir_suffix, real_target_noncanonical): Compute new
variables.

Co-Authored-By: Andrey Turetskiy <andrey.turetskiy@intel.com>
Co-Authored-By: Ilya Verbin <ilya.verbin@intel.com>
Co-Authored-By: Thomas Schwinge <thomas@codesourcery.com>
From-SVN: r217485

Use POINTER_SIZE to check for pointer size

PR tree-optimization/63828
* ipa-polymorphic-call.c (possible_placement_new): Check
POINTER_SIZE, instead of BITS_PER_WORD, for pointer size.

From-SVN: r217483

PR 60324 Unbounded stack allocations in libgfortran.

2014-11-13  Janne Blomqvist  <jb@gcc.gnu.org>

PR libfortran/60324
* configure: Regenerated.
* configure.ac (AM_CFLAGS): Add Werror=vla.
* libgfortran.h (gfc_alloca): Remove macro.
(fc_strdup_notrim): New prototype.
* intrinsics/access.c (access_func): Use fc_strdup rather than
stack allocation.
* intrinsics/chdir.c (chdir_i4_sub): Likewise.
(chdir_i8_sub): Likewise.
* intrinsics/chmod.c (chmod_internal): New function, move logic
here.
(chmod_func): Call chmod_internal.
* intrinsics/env.c (getenv): Use fc_strdup rather than stack
allocation.
(get_environment_variable_i4): Likewise.
* intrinsics/execute_command_line.c (execute_command_line):
Likewise.
* intrinsics/hostnm.c (hostnm_0): New function, use static buffer
rather than VLA.
(hostnm_i4_sub): Call hostnm_0.
(hostnm_i8_sub): Likewise.
(hostnm): Likewise.
* intrinsics/link.c (link_internal): New function, use fc_strdup
rather than stack allocation.
(link_i4_sub): Call link_internal.
(link_i8_sub): Likewise.
(link_i4): Likewise.
(link_i8): Likewise.
* intrinsics/perror.c (perror_sub): Use fc_strdup rather than
stack allocation.
* intrinsics/random.c (random_seed_i4): Use static buffer rather
than VLA, use _Static_assert to make sure it's big enough.
* intrinsics/rename.c (rename_internal): New function, use
fc_strdup rather than stack allocation.
(rename_i4_sub): Call rename_internal.
(rename_i8_sub): Likewise.
(rename_i4): Likewise.
(rename_i8): Likewise.
* intrinsics/stat.c (stat_i4_sub_0): Use fc_strdup rather than
stack allocation.
(stat_i8_sub_0): Likewise.
* intrinsics/symlink.c (symlnk_internal): New function, use
fc_strdup rather than stack allocation.
(symlnk_i4_sub): Call symlnk_internal.
(symlnk_i8_sub): Likewise.
(symlnk_i4): Likewise.
(symlnk_i8): Likewise.
* intrinsics/system.c (system_sub): Use fc_strdup rather than
stack allocation.
* intrinsics/unlink.c (unlink_i4_sub): Likewise.
* io/file_pos.c (READ_CHUNK): Make it a macro rather than variable.
* io/list_read.c (nml_get_obj_data): Use fixed stack buffer, fall
back to xmalloc/free for large sizes.
* io/read.c (read_f): Likewise.
* io/transfer.c (MAX_READ): Make it a macro rather than variable.
(WRITE_CHUNK): Likewise.
* io/write_float.def (write_float): Use fixed stack buffer, fall
back to xmalloc/free for large sizes.
* runtime/string.c (fc_strdup_notrim): New function.

From-SVN: r217480

tm.texi.in (SELECT_CC_MODE): Update example.

* doc/tm.texi.in (SELECT_CC_MODE): Update example.
(REVERSIBLE_CC_MODE): Fix example.
(REVERSE_CONDITION): Fix typo.
* doc/tm.texi: Regenerate.

From-SVN: r217476

Run pass_expand_omp_ssa after pass_paralellize_loops

2014-11-13  Tom de Vries  <tom@codesourcery.com>

* omp-low.c (pass_data_expand_omp): Set properties_provided to
PROP_gimple_eomp.
(pass_expand_omp::gate): Remove function.  Move gate expression to ...
(pass_expand_omp::execute): ... here, as new variable gate.  Add early
exit if gate is false.
(pass_data pass_data_expand_omp_ssa): New pass_data.
(class pass_expand_omp_ssa): New pass.
(make_pass_expand_omp_ssa): New function.
* passes.def (pass_parallelize_loops): Use PUSH_INSERT_PASSES_WITHIN
instead of NEXT_PASS.
(pass_expand_omp_ssa): Add after pass_parallelize_loops.
* tree-parloops.c (gen_parallel_loop): Remove call to omp_expand_local.
(pass_parallelize_loops::execute): Don't do cleanups TODO_cleanup_cfg
and TODO_rebuild_alias yet.  Add TODO_update_ssa.  Set
cfun->omp_expand_needed.
* tree-pass.h: Add define PROP_gimple_eomp.
(make_pass_expand_omp_ssa): Declare.

From-SVN: r217474

line-map.h: Include EXPR, so that unused variable warnings do not occur.

* include/line-map.h: Include EXPR, so that unused variable warnings
do not occur.

From-SVN: r217473

Enable libcc1 on Solaris

* configure.ac (libcc1_cv_lib_sockets): Check for -lsocket -lnsl.
* configure: Regenerate.
* connection.cc: Include <string.h>.
        * libcc1.cc (libcc1_compile): Use AF_UNIX instead of AF_LOCAL.

From-SVN: r217471

tree.h (TYPE_OVERFLOW_SANITIZED): Define.

* tree.h (TYPE_OVERFLOW_SANITIZED): Define.
* fold-const.c (fold_binary_loc): Use it.
* match.pd: Likewise.

From-SVN: r217470

cp-tree.h (cp_build_function_call): Remove prototype.

2014-11-13  Kai Tietz  <ktietz@redhat.com>

* cp-tree.h (cp_build_function_call): Remove prototype.
(cp_build_addr_expr_strict): Likewise.
(build_typed_address): Likewise.
* typeck.c (build_typed_address): Removed.
(cp_build_addr_expr_strict): Make static.
(cp_build_function_call): Likewise.

From-SVN: r217469

Rename struct in gcc/lra-lives.c

gcc/
* gcc/lra-lives.c (struct bb_data): Rename to ...
(struct bb_data_pseudos): ... this.
(initiate_live_solver): Update struct name.

From-SVN: r217468

target-supports.exp (error_h): New check.

gcc/testsuite/
* lib/target-supports.exp (error_h): New check.
(libc_has_complex_functions): Ditto.
(tgmath_h): Ditto.
* gcc.dg/builtins-59.c: Add libc_has_complex_functions check.
* gcc.dg/builtins-61.c: Likewise.
* gcc.dg/builtins-67.c: Disable test for Bionic.
* gcc.dg/strlenopt-14g.c: Likewise.
* gcc.dg/strlenopt-14gf.c: Likewise.
* gcc.dg/c99-tgmath-1.c: Add tgmath_h check.
* gcc.dg/c99-tgmath-2.c: Likewise.
* gcc.dg/c99-tgmath-3.c: Likewise.
* gcc.dg/c99-tgmath-4.c: Likewise.
* gcc.dg/dfp/convert-dfp-round-thread.c: Add error_h check.

From-SVN: r217467

match.pd: Implement conditional expression patterns.

2014-11-13  Richard Biener  <rguenther@suse.de>

* match.pd: Implement conditional expression patterns.
* tree-ssa-forwprop.c (forward_propagate_into_cond): Remove
them here.
(combine_cond_exprs): Remove.
(pass_forwprop::execute): Do not call combine_cond_exprs.
* fold-const.c (fold_ternary_loc): Remove patterns here.
(pedantic_omit_one_operand_loc): Remove.

From-SVN: r217465

re PR tree-optimization/61559 (FAIL: gcc.dg/builtin-bswap-8.c on i686 with -mmovbe)

2014-12-13  Richard Biener  <rguenther@suse.de>

PR middle-end/61559
* match.pd: Implement bswap patterns for transforms checked by
gcc.dg/builtin-bswap-8.c.

From-SVN: r217464

divmodhi.S: Add support for the G10 architecture.

* config/rl78/divmodhi.S: Add support for the G10 architecture.
Use START_FUNC and END_FUNC macros to enable linker garbage
collection.
* config/rl78/divmodqi.S: Likewise.
* config/rl78/divmodsi.S: Likewise.
* config/rl78/mulsi3.S: Likewise.
* config/rl78/lib2div.c: Remove G10 functions.
* config/rl78/lib2muls.c: Likewise.
* config/rl78/t-rl8 (HOST_LIBGCC2_CFLAGS): Define.
* config/rl78/vregs.h (START_FUNC): New macro.
(END_FUNC): New macro.

From-SVN: r217463

re PR libstdc++/63775 ([C++11] Regex range with leading dash (-) not working)

PR libstdc++/63775
* include/bits/regex_compiler.h (_Compiler<>::_M_expression_term,
_BracketMatcher<>::_M_make_range): Throw regex_erorr on invalid range
like [z-a]. Change _M_expression_term interface.
* include/bits/regex_compiler.tcc (
_Compiler<>::_M_insert_bracket_matcher,
_Compiler<>::_M_expression_term): Rewrite bracket expression parsing.
* testsuite/28_regex/algorithms/regex_match/cstring_bracket_01.cc:
Add testcases and move file out of extended.

From-SVN: r217461

lra.c (lra): Switch off rematerialization pass.

2014-11-13  Vladimir Makarov  <vmakarov@redhat.com>

* lra.c (lra): Switch off rematerialization pass.

From-SVN: r217459

common.opt (flra-remat): New.

2014-11-12  Vladimir Makarov  <vmakarov@redhat.com>

* common.opt (flra-remat): New.
* opts.c (default_options_table): Add entry for flra_remat.
* timevar_def (TV_LRA_REMAT): New.
* doc/invoke.texi (-flra-remat): Add description of the new
option.
* doc/passes.texi (-flra-remat): Remove lra-equivs.c and
lra-saves.c.  Add lra-remat.c.
* Makefile.in (OBJS): Add lra-remat.o.
* lra-remat.c: New file.
* lra.c: Add info about the rematerialization pass in the top
comment.
(collect_non_operand_hard_regs, add_regs_to_insn_regno_info):
Process unallocatable regs too.
(lra_constraint_new_insn_uid_start): Remove.
(lra): Add code for calling rematerialization sub-pass.
* lra-int.h (lra_constraint_new_insn_uid_start): Remove.
(lra_constrain_insn, lra_remat): New prototypes.
(lra_eliminate_regs_1): Add parameter.
* lra-lives.c (make_hard_regno_born, make_hard_regno_dead):
Process unallocatable hard regs too.
(process_bb_lives): Ditto.
* lra-spills.c (remove_pseudos): Add argument to
lra_eliminate_regs_1 call.
* lra-eliminations.c (lra_eliminate_regs_1): Add parameter.  Use it
for sp offset calculation.
(lra_eliminate_regs): Add argument for lra_eliminate_regs_1 call.
(eliminate_regs_in_insn): Add parameter.  Use it for sp offset
calculation.
(process_insn_for_elimination): Add argument for
eliminate_regs_in_insn call.
* lra-constraints.c (get_equiv_with_elimination):  Add argument
for lra_eliminate_regs_1 call.
(process_addr_reg): Add parameter.  Use it.
(process_address_1): Ditto.  Add argument for process_addr_reg
call.
(process_address): Ditto.
(curr_insn_transform): Add parameter.  Use it.  Add argument for
process_address calls.
(lra_constrain_insn): New function.
(lra_constraints): Add argument for curr_insn_transform call.

From-SVN: r217458

opts-global.c (postpone_unknown_option_warning): Fix spelling.

2014-11-13  Manuel López-Ibáñez  <manu@gcc.gnu.org>

* opts-global.c (postpone_unknown_option_warning): Fix spelling.
(print_ignored_options): Fix quoting.
* opts.c (common_handle_option): Likewise.
(set_debug_level): Likewise.
* toplev.c (process_options): Likewise.

From-SVN: r217457

Daily bump.

From-SVN: r217456

compiler: Do not count package uses from ambiguous lookups.

When using dot imports, it is possible to
have an imported symbol name that matches
the name of a struct field in a composite
literal.  Do not consider the imported
package to be used in this situation.
Fixes issue 6427.

From-SVN: r217453

std::shared_ptr atomic operations

PR libstdc++/57250
* config/abi/pre/gnu.ver: Export new symbols.
* include/Makefile.am: Add new header.
* include/Makefile.in: Regenerate.
* include/bits/shared_ptr_atomic.h: Define atomic access functions.
* include/std/memory: Include new header.
* src/c++11/shared_ptr.cc (_Sp_locker): Define and use mutex pool.
* testsuite/20_util/shared_ptr/atomic/1.cc: New.
* testsuite/20_util/shared_ptr/atomic/2.cc: New.
* testsuite/20_util/shared_ptr/atomic/3.cc: New.

From-SVN: r217452

re PR ipa/63838 (ipa-pure-const miscomputes can_throw)

PR ipa/63838
* ipa-pure-const.c (propagate_nothrow): Walk w->indirect_calls
chain instead of node->indirect_calls.  Put !can_throw into
conditions of all the loops.

* g++.dg/ipa/pr63838.C: New test.

From-SVN: r217449

Use gen_rtx_REG to set pic_offset_table_rtx

* config/i386/i386.c (x86_output_mi_thunk): Use gen_rtx_REG to
set pic_offset_table_rtx.

From-SVN: r217448

Implement MIPS o32 FPXX, FP64, FP64A ABI extensions.

2014-11-12  Matthew Fortune  <matthew.fortune@imgtec.com>

gcc/
* common/config/mips/mips-common.c (mips_handle_option): Ensure
that -mfp32, -mfp64 disable -mfpxx and -mfpxx disables -mfp64.
* config.gcc (--with-fp-32): New option.
(--with-odd-spreg-32): Likewise.
* config.in (HAVE_AS_DOT_MODULE): New config define.
* config/mips/mips-protos.h
(mips_secondary_memory_needed): New prototype.
(mips_hard_regno_caller_save_mode): Likewise.
* config/mips/mips.c (mips_get_reg_raw_mode): New static prototype.
(mips_get_arg_info): Assert that V2SFmode is only handled specially
with TARGET_PAIRED_SINGLE_FLOAT.
(mips_return_mode_in_fpr_p): Likewise.
(mips16_call_stub_mode_suffix): Likewise.
(mips_get_reg_raw_mode): New static function.
(mips_return_fpr_pair): O32 return values span two registers.
(mips16_build_call_stub): Likewise.
(mips_function_value_regno_p): Support both FP return registers.
(mips_output_64bit_xfer): Use mthc1 whenever TARGET_HAS_MXHC1.  Add
specific cases for TARGET_FPXX to move via memory.
(mips_dwarf_register_span): For TARGET_FPXX pretend that modes larger
than UNITS_PER_FPREG 'span' one register.
(mips_dwarf_frame_reg_mode): New static function.
(mips_file_start): Switch to using .module instead of .gnu_attribute.
No longer support FP ABI 4 (-mips32r2 -mfp64), replace with FP ABI 6.
Add FP ABI 5 (-mfpxx) and FP ABI 7 (-mfp64 -mno-odd-spreg).
(mips_save_reg, mips_restore_reg): Always represent DFmode frame
slots with two CFI directives even for O32 FP64.
(mips_for_each_saved_gpr_and_fpr): Account for fixed_regs when
saving/restoring callee-saved registers.
(mips_hard_regno_mode_ok_p): Implement O32 FP64A extension.
(mips_secondary_memory_needed): New function.
(mips_option_override): ABI check for TARGET_FLOATXX.  Disable
odd-numbered single-precision registers when using TARGET_FLOATXX.
Implement -modd-spreg and defaults.
(mips_conditional_register_usage): Redefine O32 FP64 to match O32 FP32
callee-saved behaviour.
(mips_hard_regno_caller_save_mode): Implement.
(TARGET_GET_RAW_RESULT_MODE): Define target hook.
(TARGET_GET_RAW_ARG_MODE): Define target hook.
(TARGET_DWARF_FRAME_REG_MODE): Define target hook.
* config/mips/mips.h (TARGET_FLOAT32): New macro.
(TARGET_O32_FP64A_ABI): Likewise.
(TARGET_CPU_CPP_BUILTINS): TARGET_FPXX is __mips_fpr==0. Add
_MIPS_SPFPSET builtin define.
(MIPS_FPXX_OPTION_SPEC): New macro.
(OPTION_DEFAULT_SPECS): Pass through --with-fp-32=* to -mfp and
--with-odd-spreg-32=* to -m[no-]odd-spreg.
(ISA_HAS_ODD_SPREG): New macro.
(ISA_HAS_MXHC1): True for anything other than -mfp32.
(ASM_SPEC): Pass through mfpxx, mfp64, -mno-odd-spreg and -modd-spreg.
(MIN_FPRS_PER_FMT): Redefine in terms of TARGET_ODD_SPREG.
(HARD_REGNO_CALLER_SAVE_MODE): Define.  Implement O32 FPXX extension
(HARD_REGNO_CALL_PART_CLOBBERED): Likewise.
(SECONDARY_MEMORY_NEEDED): Likewise.
(FUNCTION_ARG_REGNO_P): Update for O32 FPXX and FP64 extensions.
* config/mips/mips.md (define_attr enabled): Implement O32 FPXX and
FP64A ABI extensions.
(move_doubleword_fpr<mode>): Use ISA_HAS_MXHC1 instead of
TARGET_FLOAT64.
* config/mips/mips.opt (mfpxx): New target option.
(modd-spreg): Likewise.
* config/mips/mti-elf.h (DRIVER_SELF_SPECS): Infer FP ABI from arch.
* config/mips/mti-linux.h (DRIVER_SELF_SPECS): Likewise and remove
fp64 sysroot.
* config/mips/t-mti-elf: Remove fp64 multilib.
* config/mips/t-mti-linux: Likewise.
* configure.ac: Detect .module support.
* configure: Regenerate.
* doc/invoke.texi: Document -mfpxx, -modd-spreg, -mno-odd-spreg option.
* doc/install.texi (--with-fp-32, --with-odd-spreg-32): Document new
options.

gcc/testsuite/
* gcc.target/mips/args-1.c: Handle __mips_fpr == 0.
* gcc.target/mips/call-clobbered-1.c: New.
* gcc.target/mips/call-clobbered-2.c: New.
* gcc.target/mips/call-clobbered-3.c: New.
* gcc.target/mips/call-clobbered-4.c: New.
* gcc.target/mips/call-clobbered-5.c: New.
* gcc.target/mips/call-saved-4.c: New.
* gcc.target/mips/call-saved-5.c: New.
* gcc.target/mips/call-saved-6.c: New.
* gcc.target/mips/mips.exp: Support -mfpxx, -ffixed-f*,
and -m[no-]odd-spreg.  Use _MIPS_SPFPSET to determine default
odd-spreg option.  Account for -modd-spreg in minimum arch code.
* gcc.target/mips/movdf-1.c: New.
* gcc.target/mips/movdf-2.c: New.
* gcc.target/mips/movdf-3.c: New.
* gcc.target/mips/oddspreg-1.c: New.
* gcc.target/mips/oddspreg-2.c: New.
* gcc.target/mips/oddspreg-3.c: New.
* gcc.target/mips/oddspreg-4.c: New.
* gcc.target/mips/oddspreg-5.c: New.
* gcc.target/mips/oddspreg-6.c: New.

libgcc/
* config/mips/mips16.S: Set .module when supported.  Update O32
FP64 calling convention and use for FPXX when possible.  Add FPXX
calling convention fallback case.

From-SVN: r217446

Initialize PIC register for large PIC model C++ thunk

gcc/

PR target/63815
* config/i386/i386.c (ix86_init_large_pic_reg): New.  Extracted
from ...
(ix86_init_pic_reg): Here.  Use ix86_init_large_pic_reg.
(x86_output_mi_thunk): Set PIC register to %r11.  Call
ix86_init_large_pic_reg to initialize PIC register.

gcc/testsuite/

PR target/63815
* g++.dg/other/pr63815.C: New test.

From-SVN: r217445

DR 1510 PR c++/60420

/cp
2014-11-12  Paolo Carlini  <paolo.carlini@oracle.com>

DR 1510
PR c++/60420
* cp-tree.h (struct cp_decl_specifier_seq): Add decltype_p bool field.
* decl.c (grokdeclarator): Use it.
* parser.c (cp_parser_simple_type_specifier): Likewise.
* pt.c (tsubst, case DECLTYPE_TYPE): Use tf_ignore_bad_quals.

/testsuite
2014-11-12  Paolo Carlini  <paolo.carlini@oracle.com>

DR 1510
PR c++/60420
* g++.dg/cpp0x/decltype61.C: New.

From-SVN: r217444

sdbout.c (sdbout_symbol): Eliminate register only if decl isn't a global variable.

2014-11-12  Kai Tietz  <ktietz@redhat.com>

        * sdbout.c (sdbout_symbol): Eliminate register only
        if decl isn't a global variable.

From-SVN: r217443

Implement resolutions of LWG 2399, 2400 and 2401.

* include/bits/shared_ptr.h (shared_ptr, weak_ptr): Define
_Convertible alias template to simplify constraints.
(shared_ptr(unique_ptr<TD>&&)): Constrain (LWG 2399).
* include/bits/shared_ptr_base.h: Likewise.
(_Sp_counted_deleter::_M_get_deleter()): Use addressof (LWG 2400).
* include/std/functional (function::operator=(nullptr_t)): Add
noexcept (LWG 2401).
* testsuite/20_util/shared_ptr/cons/43820_neg.cc: Adjust dg-error.
* testsuite/20_util/shared_ptr/cons/void_neg.cc: Adjust dg-error.

From-SVN: r217442

Add a testcase for PR tree-optimization/63835

PR tree-optimization/63835
* gcc.dg/pr63835.c: New test.

From-SVN: r217441

Add bounds checking to vqdm*_lane intrinsics via a qualifier that also flips endianness

gcc/:

* config/aarch64/aarch64.c (aarch64_simd_lane_bounds): Display indices.

* config/aarch64/aarch64-builtins.c (enum aarch64_type_qualifiers): Add
qualifier_lane_index.
(aarch64_types_ternop_lane_qualifiers, TYPES_TERNOP_LANE): Rename to...
(aarch64_types_quadop_lane_qualifiers, TYPES_QUADOP_LANE): ...these.
(aarch64_types_ternop_lane_qualifiers, TYPES_TERNOP_LANE): New.

(aarch64_types_getlane_qualifiers): Rename to...
(aarch64_types_binop_imm_qualifiers): ...this.
(TYPES_SHIFTIMM): Follow renaming.
(TYPES_GETLANE): Rename to...
(TYPE_GETREG): ...this.

(aarch64_types_setlane_qualifiers): Rename to...
(aarch64_type_ternop_imm_qualifiers): ...this.
(TYPES_SHIFTINSERT, TYPES_SHIFTACC): Follow renaming.
(TYPES_SETLANE): Follow renaming above, and rename self to...
(TYPE_SETREG): ...this.

(enum builtin_simd_arg): Add SIMD_ARG_LANE_INDEX.
(aarch64_simd_expand_args): Add range check and endianness-flip.

(aarch64_simd_expand_builtin): Add mapping for qualifier_lane_index.

* config/aarch64/aarch64-simd.md
(aarch64_sq<r>dmulh_lane<mode>_internal *2): Rename to...
(aarch64_sq<r>dmulh_lane<mode>): ...this, and remove lane bounds check.
(aarch64_sqdmulh_lane<mode> *2, aarch64_sqrdmulh_lane<mode> *2): Delete.

(aarch64_sq<r>dmulh_laneq<mode>_internal): Rename to...
(aarch64_sq<r>dmulh_lane<mode>): ...this.

(aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal *2): Rename to...
(aarch64_sqdml<SBINQOPS:as>l_lane<mode>): ...this.

(aarch64_sqdml<SBINQOPS:as>l_laneq<mode>_internal *2): Rename to...
(aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): ...this.

(aarch64_sqdmull_lane<mode>_internal *2): Rename to...
(aarch64_sqdmull_lane<mode>): ...this.

(aarch64_sqdmull_laneq<mode>_internal *2): Rename to...
(aarch64_sqdmull_laneq<mode>): ...this.

(aarch64_sqdmulh_laneq<mode>, aarch64_sqrdmulh_laneq<mode>,
(aarch64_sqdmlal_lane<mode>, aarch64_sqdmlal_laneq<mode>,
aarch64_sqdmlsl_lane<mode>, aarch64_sqdmlsl_laneq<mode>,
aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Delete.

(aarch64_sqdmlal2_lane<mode>, aarch64_sqdmlal2_laneq<mode>,
aarch64_sqdmlsl2_lane<mode>, aarch64_sqdmlsl2_laneq<mode>,
aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>): Remove
bounds check and lane flip.

* config/aarch64/aarch64-simd-builtins.def (be_checked_get_lane,
get_dregoi, get_dregci, getdregxi, get_qregoi,get_qregci, get_qregxi,
set_qregoi, set_qregci, set_qregxi): Change qualifiers to GETREG.

(sqdmlal_lane, sqdmlsl_lane, sqdmlal_laneq, sqdmlsl_laneq,
sqdmlal2_lane, sqdmlsl2_lane, sqdmlal2_laneq, sqdmlsl2_laneq): Follow
renaming of TERNOP_LANE to QUADOP_LANE.

(sqdmull_lane, sqdmull_laneq, sqdmull2_lane, sqdmull2_laneq,
sqdmulh_lane, sqdmulh_laneq, sqrdmulh_lane, sqrdmulh_laneq): Set
qualifiers to TERNOP_LANE.

gcc/testsuite/:

         * gcc.target/aarch64/simd/vqdmlal_high_lane_s16_indices_1.c: New test.
         * gcc.target/aarch64/simd/vqdmlal_high_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlal_high_laneq_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlal_high_laneq_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlal_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlal_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlal_laneq_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlal_laneq_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlalh_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlals_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlsl_high_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlsl_high_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlsl_high_laneq_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlsl_high_laneq_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlsl_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlsl_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlsl_laneq_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlsl_laneq_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlslh_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmlsls_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulh_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulh_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulh_laneq_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulh_laneq_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulhh_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulhq_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulhq_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulhq_laneq_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulhq_laneq_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulhs_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmull_high_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmull_high_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmull_high_laneq_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmull_high_laneq_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmull_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmull_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmull_laneq_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmull_laneq_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmullh_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqdmulls_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqrdmulh_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqrdmulh_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqrdmulh_laneq_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqrdmulh_laneq_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqrdmulhh_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqrdmulhq_lane_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqrdmulhq_lane_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqrdmulhq_laneq_s16_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqrdmulhq_laneq_s32_indices_1.c: Likewise.
         * gcc.target/aarch64/simd/vqrdmulhs_lane_s32_indices_1.c: Likewise.

From-SVN: r217440

Fix ChangeLog of previous commit, r217438:

        * Makefile.def: Make all-gcc depend on all-isl.

From-SVN: r217439

Makefile.def: Remove CLooG.

2014-11-11  Tobias Burnus  <burnus@net-b.de>

        * Makefile.def: Remove CLooG.
        * Makefile.in: Regenerate.

From-SVN: r217438

387-1.c (dg-skip-if): Use *-*-* target selector.

* gcc.target/i386/387-1.c (dg-skip-if): Use *-*-* target selector.
* gcc.target/i386/387-2.c (dg-skip-if): Ditto.
* gcc.target/i386/387-5.c (dg-skip-if): Ditto.
* gcc.target/i386/387-6.c (dg-skip-if): Ditto.
* gcc.target/i386/cmov7.c (dg-skip-if): Ditto.
* gcc.target/i386/funcspec-2.c (dg-skip-if): Ditto.
* gcc.target/i386/funcspec-3.c (dg-skip-if): Ditto.
* gcc.target/i386/funcspec-8.c (dg-skip-if): Ditto.
* gcc.target/i386/funcspec-9.c (dg-skip-if): Ditto.
* gcc.target/i386/gcc-have-sync-compare-and-swap-1.c (dg-skip-if):
Ditto.
* gcc.target/i386/gcc-have-sync-compare-and-swap-2.c (dg-skip-if):
Ditto.
* gcc.target/i386/isa-1.c (dg-skip-if): Ditto.
* gcc.target/i386/isa-6.c (dg-skip-if): Ditto.
* gcc.target/i386/isa-9.c (dg-skip-if): Ditto.
* gcc.target/i386/lea.c (dg-skip-if): Ditto.
* gcc.target/i386/memcpy-strategy-1.c (dg-skip-if): Ditto.
* gcc.target/i386/memcpy-strategy-2.c (dg-skip-if): Ditto.
* gcc.target/i386/memcpy-vector_loop-1.c (dg-skip-if): Ditto.
* gcc.target/i386/memcpy-vector_loop-2.c (dg-skip-if): Ditto.
* gcc.target/i386/memset-vector_loop-1.c (dg-skip-if): Ditto.
* gcc.target/i386/memset-vector_loop-2.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-10.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-2.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-3.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-4.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-5a.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-5b.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-6a.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-6b.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-7.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-8.c (dg-skip-if): Ditto.
* gcc.target/i386/pad-9.c (dg-skip-if): Ditto.
* gcc.target/i386/pentium4-not-mull.c (dg-skip-if): Ditto.
* gcc.target/i386/pr42589.c (dg-skip-if): Ditto.
* gcc.target/i386/sse-19.c (dg-skip-if): Ditto.
* gcc.target/i386/sse2-init-v2di-2.c (dg-skip-if): Ditto.
* gcc.target/i386/ssefn-1.c (dg-skip-if): Ditto.
* gcc.target/i386/ssetype-1.c (dg-skip-if): Ditto.
* gcc.target/i386/ssetype-2.c (dg-skip-if): Ditto.
* gcc.target/i386/ssetype-5.c (dg-skip-if): Ditto.
* gcc.target/i386/vect-double-1.c (dg-skip-if): Ditto.

From-SVN: r217437

Makefile.in (CLOOGLIBS, CLOOGINC): Remove.

2014-11-12  Tobias Burnus  <burnus@net-b.de>

        * Makefile.in (CLOOGLIBS, CLOOGINC): Remove.
        * configure.ac: Ditto.
        * graphite-interchange.c: Remove HAVE_CLOOG block.
        * config.in: Regenerate.
        * configure: Regenerate.

From-SVN: r217436

pr45852.c: Simplify target selector.

* gcc.target/i386/pr45852.c: Simplify target selector.
* gcc.target/i386/pr63495.c: Ditto.
* gcc.target/i386/pr39496.c: Ditto.
* gcc.target/i386/pr36613.c: Ditto.

From-SVN: r217432

[AArch64] Let LR register alloctable

  gcc/
    * config/aarch64/aarch64.h (CALL_USED_REGISTERS): Mark LR as caller-save.
    (EPILOGUE_USES): Guard the check by epilogue_completed.
    * config/aarch64/aarch64.c (aarch64_layout_frame): Explictly check for LR.
    (aarch64_can_eliminate): Check LR_REGNUM liveness.

  gcc/testsuite/
    * gcc.target/aarch64/lr_free_1.c: New testcase for -fomit-frame-pointer.
    * gcc.target/aarch64/lr_free_2.c: New testcase for leaf
    -fno-omit-frame-pointer.

From-SVN: r217431

Fix typo in *<arith_shift_insn>_shiftsi

From-SVN: r217430

Cast strtol return to unsigned long

* testsuite/test-strtol.c (run_tests): Cast strtol return to
unsigned long.

From-SVN: r217429

gcc-dg.exp (${tool}_load): Call prune_file_path instead of prune_gcc_output.

* lib/gcc-dg.exp (${tool}_load): Call prune_file_path instead
of prune_gcc_output.
* lib/prune.exp (prune_file_path): New procedure.

From-SVN: r217428

fold-const.c (fold_binary_loc): Don't fold if the result is undefined.

* fold-const.c (fold_binary_loc): Don't fold if the result
is undefined.
* match.pd (A + (-B) -> A - B, A - (-B) -> A + B,
-(-A) -> A): Likewise.

* c-c++-common/ubsan/overflow-sub-4.c: New test.
* c-c++-common/ubsan/overflow-sub-2.c: Adjust dg-output.
* c-c++-common/ubsan/overflow-int128.c: Likewise.

From-SVN: r217427

pr52252-atom-1.c: Extend assembler scan on AVX2 case.

gcc/testsuite

* gcc.target/i386/pr52252-atom-1.c: Extend assembler scan on AVX2 case.

From-SVN: r217423

[multiple changes]

2014-11-12  Richard Biener  <rguenther@suse.de>

Merge from match-and-simplify branch
2014-11-04  Prathamesh Kulkarni  <bilbotheelffriend@gmail.com>

* genmatch.c (user_id): Add new member is_oper_list.
(user_id::user_id): Add new default argument.
(parser::parse_operator_list): New function.
(parser::parse_for): Allow operator-list.
(parser::parse_pattern): Call parser::parse_operator_list.
(parser::parse_operation): Reject operator-list.
* match-builtin.pd: Define operator lists POWs, CBRTs and SQRTs.

2014-10-31  Prathamesh Kulkarni  <bilbotheelffriend@gmail.com>

* genmatch.c (parser::parse_c_expr): Mark user-defined ops as used.

2014-10-30  Prathamesh Kulkarni  <bilbotheelffriend@gmail.com>

* genmatch.c (parser::parse_op): Check if predicate is used in
result operand.

2014-10-29  Prathamesh Kulkarni  <bilbotheelffriend@gmail.com>

* genmatch.c (parser::parse_for): Make sure to have a valid
token to report errors at.

2014-10-28  Prathamesh Kulkarni  <bilbotheelffriend@gmail.com>

* genmatch.c (parser): Add new member parsing_match_operand.
(parser::parse_operation): Check for conditional convert in result
operand.
(parser::parse_expr): Check for commutative operator in result operand.
Check for :type in match operand.
(parser::parse_simplify): Set/unset parsing_match_operand.
(parser::parser): Initialize parsing_match_operand.

2014-10-28  Richard Biener  <rguenther@suse.de>

* genmatch.c (parser::parse_for): Properly check for already
defined operators.

2014-10-28  Prathamesh Kulkarni  <bilbotheelffriend@gmail.com>

* genmatch.c (error_cb): Adjust for printing warnings.
(warning_at): New function.
(user_id): Add new member used.
(get_operator): Mark user_id as used.
(parse_for): Warn for unused operators.

From-SVN: r217422

match.pd: Implement simple complex operations cancelling.

2014-11-12  Richard Biener  <rguenther@suse.de>

* match.pd: Implement simple complex operations cancelling.
* fold-const.c (fold_unary_loc): Remove them here.

* gcc.dg/tree-ssa/ssa-fre-32.c: Disable forwprop.

From-SVN: r217421

Predefine __NO_MATH_ERRNO__ for -fno-math-errno.

This patch adds a predefined macro __NO_MATH_ERRNO__ for when
-fno-math-errno is passed or implied.  This allows math.h to provide a
more accurate definition of the C99 math_errhandling macro that takes
this option into account, and allows for choice of libm functions to
be optimized at compile time based on this option.

(There may be a case for such interfaces for -fno-rounding-math
(default) and -fno-trapping-math as well, but as C99 standard pragmas
would allow those to vary on a per-block basis, predefined macros
would be problematic as the interface; you can't select a
-fno-trapping-math or -fno-rounding-math version of a function in a
standard header if a conforming program could then use "#pragma STDC
FENV_ACCESS ON" to require a -ftrapping-math -frounding-math version
in a particular block.  So built-in functions might be a better way of
providing access to information about those options.)

Bootstrapped with no regressions on x86_64-unknown-linux-gnu.  OK to
commit?

* cppbuiltin.c (define_builtin_macros_for_compilation_flags):
Define __NO_MATH_ERRNO__ if -fno-math-errno.
* doc/cpp.texi (__NO_MATH_ERRNO__): Document predefined macro.

c-family:
* c-cppbuiltin.c (c_cpp_builtins_optimize_pragma): Define and
undefine __NO_MATH_ERRNO__ based on changes to -fmath-errno state.

testsuite:
* gcc.dg/no-math-errno-1.c, gcc.dg/no-math-errno-2.c,
gcc.dg/no-math-errno-3.c, gcc.dg/no-math-errno-4.c: New tests.

From-SVN: r217420

Avoid tail call in c-c++-common/asan/strlen-overflow-1.c

PR testsuite/63830
* c-c++-common/asan/strlen-overflow-1.c (main): Avoid tail call.

From-SVN: r217417

genmatch.c (::gen_transform): Add capture_info and expand_compares arguments.

2014-11-12  Richard Biener  <rguenther@suse.de>

* genmatch.c (::gen_transform): Add capture_info and
expand_compares arguments.
(struct expr): Add is_generic flag.
(lower_cond): New functions lowering [VEC_]COND_EXPR
conditions to a GENERIC and a GIMPLE variant.
(lower): Call lower_cond.
(cmp_operand): Also compare the is_generic flag.
(capture_info::cinfo): Add cond_expr_cond_p flag.
(capture_info::capture_info): Pass down whether the
expression argument is a COND_EXPR condition.
(capture_info::walk_match): Likewise, mark captures
capturing COND_EXPR conditions with cond_expr_cond_p.
(expr::gen_transform): Pass down whether we need to
expand compares from COND_EXPR conditions.
(capture::gen_transform): Expand compares substituted
from COND_EXPR conditions into non-COND_EXPR conditions.
(dt_operand::gen_gimple_expr): Handle explicitely marked
GENERIC expressions as generic.
(dt_simplify::gen): Pass whether we need to expand
conditions to gen_transform.  Handle capture results
which are from COND_EXPR conditions.
(main): Pass gimple flag down to lower.

From-SVN: r217416

re PR c/59708 (clang-compatible checked arithmetic builtins)

PR c/59708
* builtin-attrs.def (ATTR_NOTHROW_TYPEGENERIC_LEAF): New attribute.
* builtins.c (fold_builtin_arith_overflow): New function.
(fold_builtin_3): Use it.
* builtins.def (BUILT_IN_ADD_OVERFLOW, BUILT_IN_SUB_OVERFLOW,
BUILT_IN_MUL_OVERFLOW, BUILT_IN_SADD_OVERFLOW, BUILT_IN_SADDL_OVERFLOW,
BUILT_IN_SADDLL_OVERFLOW, BUILT_IN_SSUB_OVERFLOW,
BUILT_IN_SSUBL_OVERFLOW, BUILT_IN_SSUBLL_OVERFLOW,
BUILT_IN_SMUL_OVERFLOW, BUILT_IN_SMULL_OVERFLOW,
BUILT_IN_SMULLL_OVERFLOW, BUILT_IN_UADDL_OVERFLOW,
BUILT_IN_UADDLL_OVERFLOW, BUILT_IN_USUB_OVERFLOW,
BUILT_IN_USUBL_OVERFLOW, BUILT_IN_USUBLL_OVERFLOW,
BUILT_IN_UMUL_OVERFLOW, BUILT_IN_UMULL_OVERFLOW,
BUILT_IN_UMULLL_OVERFLOW): New built-in functions.
* builtin-types.def (BT_PTR_UINT, BT_PTR_ULONG, BT_PTR_LONGLONG,
BT_FN_BOOL_INT_INT_INTPTR, BT_FN_BOOL_LONG_LONG_LONGPTR,
BT_FN_BOOL_LONGLONG_LONGLONG_LONGLONGPTR, BT_FN_BOOL_UINT_UINT_UINTPTR,
BT_FN_BOOL_ULONG_ULONG_ULONGPTR,
BT_FN_BOOL_ULONGLONG_ULONGLONG_ULONGLONGPTR, BT_FN_BOOL_VAR): New.
* expr.c (write_complex_part): Remove prototype, no longer static.
* expr.h (write_complex_part): New prototype.
* function.c (aggregate_value_p): For internal functions return 0.
* gimple-fold.c (arith_overflowed_p): New functions.
(gimple_fold_call): Fold {ADD,SUB,MUL}_OVERFLOW internal calls.
* gimple-fold.h (arith_overflowed_p): New prototype.
* tree-ssa-dce.c: Include tree-ssa-propagate.h and gimple-fold.h.
(find_non_realpart_uses, maybe_optimize_arith_overflow): New
functions.
(eliminate_unnecessary_stmts): Transform {ADD,SUB,MUL}_OVERFLOW
into COMPLEX_CST/COMPLEX_EXPR if IMAGPART_EXPR of the result is
never used.
* gimplify.c (gimplify_call_expr): Handle gimplification of
internal calls with lhs.
* internal-fn.c (get_range_pos_neg, get_min_precision,
expand_arith_overflow_result_store): New functions.
(ubsan_expand_si_overflow_addsub_check): Renamed to ...
(expand_addsub_overflow): ... this.  Add LOC, LHS, ARG0, ARG1,
UNSR_P, UNS0_P, UNS1_P, IS_UBSAN arguments, remove STMT argument.
Handle ADD_OVERFLOW and SUB_OVERFLOW expansion.
(ubsan_expand_si_overflow_neg_check): Renamed to ...
(expand_neg_overflow): ... this.  Add LOC, LHS, ARG1, IS_UBSAN
arguments, remove STMT argument.  Handle SUB_OVERFLOW with
0 as first argument expansion.
(ubsan_expand_si_overflow_mul_check): Renamed to ...
(expand_mul_overflow): ... this.  Add LOC, LHS, ARG0, ARG1,
UNSR_P, UNS0_P, UNS1_P, IS_UBSAN arguments, remove STMT argument.
Handle MUL_OVERFLOW expansion.
(expand_UBSAN_CHECK_ADD): Use expand_addsub_overflow, prepare
arguments for it.
(expand_UBSAN_CHECK_SUB): Use expand_addsub_overflow or
expand_neg_overflow, prepare arguments for it.
(expand_UBSAN_CHECK_MUL): Use expand_mul_overflow, prepare arguments
for it.
(expand_arith_overflow, expand_ADD_OVERFLOW, expand_SUB_OVERFLOW,
expand_MUL_OVERFLOW): New functions.
* internal-fn.def (ADD_OVERFLOW, SUB_OVERFLOW, MUL_OVERFLOW): New
internal functions.
* tree-vrp.c (check_for_binary_op_overflow): New function.
(extract_range_basic): Handle {REAL,IMAG}PART_EXPR if the operand
is SSA_NAME set by {ADD,SUB,MUL}_OVERFLOW internal functions.
(simplify_internal_call_using_ranges): Handle {ADD,SUB,MUL}_OVERFLOW
internal functions.
* optabs.def (umulv4_optab): New optab.
* config/i386/i386.md (umulv<mode>4, <u>mulvqi4): New define_expands.
(*umulv<mode>4, *<u>mulvqi4): New define_insns.
* doc/extend.texi (Integer Overflow Builtins): Document
__builtin_*_overflow.
c-family/
* c-common.c (check_builtin_function_arguments): Handle
BUILT_IN_{ADD,SUB,MUL}_OVERFLOW.
testsuite/
* c-c++-common/builtin-arith-overflow-1.c: New test.
* c-c++-common/torture/builtin-arith-overflow-10.c: New test.
* c-c++-common/torture/builtin-arith-overflow-11.c: New test.
* c-c++-common/torture/builtin-arith-overflow-12.c: New test.
* c-c++-common/torture/builtin-arith-overflow-12.h: New file.
* c-c++-common/torture/builtin-arith-overflow-13.c: New test.
* c-c++-common/torture/builtin-arith-overflow-14.c: New test.
* c-c++-common/torture/builtin-arith-overflow-15.c: New test.
* c-c++-common/torture/builtin-arith-overflow-16.c: New test.
* c-c++-common/torture/builtin-arith-overflow-17.c: New test.
* c-c++-common/torture/builtin-arith-overflow-18.c: New test.
* c-c++-common/torture/builtin-arith-overflow-1.c: New test.
* c-c++-common/torture/builtin-arith-overflow-1.h: New file.
* c-c++-common/torture/builtin-arith-overflow-2.c: New test.
* c-c++-common/torture/builtin-arith-overflow-3.c: New test.
* c-c++-common/torture/builtin-arith-overflow-4.c: New test.
* c-c++-common/torture/builtin-arith-overflow-5.c: New test.
* c-c++-common/torture/builtin-arith-overflow-6.c: New test.
* c-c++-common/torture/builtin-arith-overflow-7.c: New test.
* c-c++-common/torture/builtin-arith-overflow-8.c: New test.
* c-c++-common/torture/builtin-arith-overflow-9.c: New test.
* c-c++-common/torture/builtin-arith-overflow.h: New file.
* gcc.dg/builtin-arith-overflow-1.c: New test.
* gcc.dg/builtin-arith-overflow-2.c: New test.

From-SVN: r217415

genmatch.c (capture_info::capture_info): Add missing COND_EXPR handling.

2014-11-12  Richard Biener  <rguenther@suse.de>

* genmatch.c (capture_info::capture_info): Add missing
COND_EXPR handling.
(capture_info::walk_match): Fix COND_EXPR handling.
(capture_info::walk_result): Likewise.

From-SVN: r217414

LWG DR 2315. weak_ptr should be movable

* include/bits/shared_ptr.h (weak_ptr): Add move constructor and
assignment.
* include/bits/shared_ptr_base.h (__weak_count, __weak_ptr): Likewise.
Use nullptr and injected class name.
* testsuite/20_util/shared_ptr/cons/43820_neg.cc: Adjust dg-error.
* testsuite/20_util/shared_ptr/cons/void_neg.cc: Adjust dg-error.

From-SVN: r217413

Add -Wno-deprecated to dg-options.

PR c++/33911
* testsuite/20_util/shared_ptr/assign/auto_ptr.cc: Use -Wno-deprecated.
* testsuite/20_util/shared_ptr/assign/auto_ptr_neg.cc: Likewise.
* testsuite/20_util/shared_ptr/cons/43820_neg.cc: Likewise.
* testsuite/20_util/shared_ptr/cons/auto_ptr.cc: Likewise.
* testsuite/20_util/shared_ptr/cons/auto_ptr_neg.cc: Likewise.
* testsuite/20_util/unique_ptr/cons/auto_ptr.cc: Likewise.
* testsuite/20_util/unique_ptr/cons/auto_ptr_neg.cc: Likewise.
* testsuite/ext/array_allocator/variadic_construct.cc: Likewise.

From-SVN: r217412

re PR bootstrap/63821 (ICE in verify_gimple during libgcc build starting with r217349)

2014-11-12  Richard Biener  <rguenther@suse.de>

PR middle-end/63821
* match.pd: Add missing conversion to the -(T)-X pattern.

* gfortran.dg/pr63821.f90: New testcase.

From-SVN: r217411

re PR middle-end/63819 (Cannot build compiler with --enable-gather-detailed-mem-stats (error: default argument given for parameter 5 of ‘void* ggc_internal_cleared_alloc))

2014-11-12  Richard Biener  <rguenther@suse.de>

PR bootstrap/63819
* hash-table.h: Include ggc.h also for generator programs.
* genmatch.c (ggc_internal_cleared_alloc): Properly define
using MEM_STAT_DECL instead of CXX_MEM_STAT_INFO.

From-SVN: r217410

[multiple changes]

2014-11-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>

    gcc/
    PR tree-optimization/63761
    * tree-ssa-math-opts.c (bswap_replace): Construct gsi from cur_stmt
    rather than taking it as a parameter. Add some comments to explain the
    gsi_move_before in case of load and why canonicalization of bswap into
    a rotation is only done for 16bit values.
    (pass_optimize_bswap::execute): Adapt for loop via gsi to make gsi
    refer to the statement just before cur_stmt. Ignore 16bit bswap that
    are already in canonical form. Adapt bswap_replace to removal of its
    gsi parameter.

    2014-11-12  Thomas Preud'homme  <thomas.preudhomme@arm.com>

    gcc/testsuite/
    PR tree-optimization/63761
    * gcc.c-torture/compile/pr63761.c: New test.

From-SVN: r217409

[PATCH][dejagnu] truncate absolute file path into relative for dg-output

  gcc/testsuite/
    * lib/gcc-dg.exp (${tool}_load): Truncate gcc output.
    * lib/prune.exp (prune_gcc_output): New absolute path to relative path
    truncation pattern.

From-SVN: r217408

rtl.h (rtx_function, [...]): Delete.

gcc/
* rtl.h (rtx_function, for_each_rtx, for_each_rtx_in_insn): Delete.
* rtlanal.c (non_rtx_starting_operands, for_each_rtx_1, for_each_rtx):
(for_each_rtx_in_insn): Delete.
(init_rtlanal): Remove initialization of non_rtx_starting_operands.
* df-core.c: Remove reference to for_each_rtx in comment.

From-SVN: r217407

[AArch64] Remove unnecessary files.

2014-11-12  Tejas Belagod  <tejas.belagod@arm.com>

        * Makefile.in (TEXI_GCC_FILES): Remove arm-acle-intrinsics.texi,
        arm-neon-intrinsics.texi, aarch64-acle-intrinsics.texi.
        * doc/aarch64-acle-intrinsics.texi: Remove.
        * doc/arm-acle-intrinsics.texi: Remove.
        * doc/arm-neon-intrinsics.texi: Remove.
        * doc/extend.texi: Consolidate sections AArch64 intrinsics,
        ARM NEON Intrinsics, ARM ACLE Intrinsics into one ARM C Language
        Extension section. Add references to public ACLE specification.

From-SVN: r217406

[AArch64] Add reference to ACLE and consolidate documentation.

2014-11-12  Tejas Belagod  <tejas.belagod@arm.com>

        * Makefile.in (TEXI_GCC_FILES): Remove arm-acle-intrinsics.texi,
        arm-neon-intrinsics.texi, aarch64-acle-intrinsics.texi.
        * doc/aarch64-acle-intrinsics.texi: Remove.
        * doc/arm-acle-intrinsics.texi: Remove.
        * doc/arm-neon-intrinsics.texi: Remove.
        * doc/extend.texi: Consolidate sections AArch64 intrinsics,
        ARM NEON Intrinsics, ARM ACLE Intrinsics into one ARM C Language
        Extension section. Add references to public ACLE specification.

From-SVN: r217405

libgo: Update libtool support to files in current GCC trunk.

From-SVN: r217404

eabi1.c: Replace arm*-*-*eabi* and arm*-*-symbianelf* with arm_eabi.

2014-11-12  Andreas Tobler  <andreast@gcc.gnu.org>

        * gcc.target/arm/eabi1.c: Replace arm*-*-*eabi* and arm*-*-symbianelf*
with arm_eabi.
* g++.dg/abi/arm_rtti1.C: Likewise.
* g++.dg/abi/key1.C: Likewise.
* g++.dg/ext/visibility/arm1.C: Likewise.
* g++.dg/ext/visibility/arm2.C: Likewise.
* g++.dg/ext/visibility/arm3.C: Likewise.

From-SVN: r217403

Fix dump scanning in g++.dg/ipa/devirt-42.C

* g++.dg/ipa/devirt-42.C: Fix dump scanning for "return 2".

From-SVN: r217402

1.cc: Remove unused header.

* testsuite/experimental/optional/constexpr/observers/1.cc: Remove
unused header.
* testsuite/experimental/optional/constexpr/observers/2.cc: Likewise.
* testsuite/experimental/optional/constexpr/observers/3.cc: Likewise.
* testsuite/experimental/optional/constexpr/observers/4.cc: Likewise.
* testsuite/experimental/optional/constexpr/observers/5.cc: Likewise.

From-SVN: r217401

VRP: Simplify logic for checking if any asserts need to be inserted

2014-11-11  Patrick Palka  <ppalka@gcc.gnu.org>

* tree-vrp.c (register_edge_assert_for_2): Change return type to
void and adjust accordingly.
(register_edge_assert_for_1): Likewise.
(register_edge_assert_for): Likewise.
(find_conditional_asserts): Likewise.
(find_switch_asserts): Likewise.
(find_assert_locations_1): Likewise.
(find_assert_locations): Likewise.
(insert_range_insertions): Inspect the need_assert_for bitmap.

From-SVN: r217400

Daily bump.

From-SVN: r217399

optional (_Has_addressof): Check for non-member operator&.

* include/experimental/optional (_Has_addressof): Check for non-member
operator&.
* testsuite/experimental/optional/observers/2.cc: Add operator&.
* testsuite/experimental/optional/constexpr/observers/2.cc: Likewise.

From-SVN: r217397

Define __void_t and SFINAE-friendly iterator_traits.

* include/std/type_traits (__void_t): Define new alias template.
(_GLIBCXX_HAS_NESTED_TYPE): Redefine using __void_t.
* include/std/functional (_Maybe_get_result_type): Likewise.
* include/bits/stl_iterator_base_types.h (__iterator_traits): Likewise.
* include/bits/uses_allocator.h (__uses_allocator_helper): Likewise.
* testsuite/20_util/bind/ref_neg.cc: Adjust dg-error.
* testsuite/20_util/reference_wrapper/typedefs-3.cc: Adjust to changes
in _GLIBCXX_HAS_NESTED_TYPE.

From-SVN: r217395

Bug target/61997

2014-11-11  Andrew Pinski  <apinski@cavium.com>

        Bug target/61997
        * config.gcc (aarch64*-*-*): Set target_gtfiles to include
        aarch64-builtins.c.
        * config/aarch64/aarch64-builtins.c: Include gt-aarch64-builtins.h
        at the end of the file.

From-SVN: r217394

Make std::mem_fn work with varargs functions.

* include/std/functional (_Mem_fn_traits): Add partial specializations
for varargs functions.
(_Mem_fn_base): Do not check arguments are convertible for varargs.
(_Bind_check_arity): Add partial specializations for varargs functions.
* include/std/type_traits (__bool_constant): Add alias template.
* testsuite/20_util/bind/ref_neg.cc: Adjust dg-error.
* testsuite/20_util/bind/refqual.cc: New, test ref-qualifiers.
* testsuite/20_util/declval/requirements/1_neg.cc: Adjust dg-error.
* testsuite/20_util/function_objects/mem_fn/refqual.cc: Test varargs.
* testsuite/20_util/make_signed/requirements/typedefs_neg.cc: Adjust
dg-error.
* testsuite/20_util/make_unsigned/requirements/typedefs_neg.cc: Adjust
dg-error.

From-SVN: r217393

re PR driver/36312 (should refuse to overwrite input file with output file)

gcc/testsuite/ChangeLog:

2014-11-11  Anthony Brandon  <anthony.brandon@gmail.com>
    Manuel López-Ibáñez  <manu@gcc.gnu.org>

PR driver/36312
* gcc.misc-tests/output.exp: New test case for identical input and
output files.

include/ChangeLog:

2014-11-11  Anthony Brandon  <anthony.brandon@gmail.com>
    Manuel López-Ibáñez  <manu@gcc.gnu.org>

PR driver/36312
* filenames.h: Add prototype for canonical_filename_eq.

gcc/ChangeLog:

2014-11-11  Anthony Brandon  <anthony.brandon@gmail.com>
    Manuel López-Ibáñez  <manu@gcc.gnu.org>

PR driver/36312
* diagnostic-core.h: Add prototype for fatal_error.
* diagnostic.c (fatal_error): New function fatal_error.
* gcc.c (store_arg): Remove have_o_argbuf_index.
(process_command): Check if input and output files are the same.
* toplev.c (init_asm_output): Check if input and output files are
the same.

libiberty/ChangeLog:

2014-11-11  Anthony Brandon  <anthony.brandon@gmail.com>
    Manuel López-Ibáñez  <manu@gcc.gnu.org>

PR driver/36312
* filename_cmp.c (canonical_filename_eq): New function to check if
file names are the same.
* functions.texi: Updated with documentation for new function.

Co-Authored-By: Manuel López-Ibáñez <manu@gcc.gnu.org>
From-SVN: r217391

re PR fortran/44054 (Handle -Werror, -Werror=, -fdiagnostics-show-option, !GCC$ diagnostic (pragmas) and color)

libcpp/ChangeLog:

2014-11-11  Manuel López-Ibáñez  <manu@gcc.gnu.org>

PR fortran/44054
* include/line-map.h (linemap_position_for_loc_and_offset):
Declare.
* line-map.c (linemap_position_for_loc_and_offset): New.

gcc/fortran/ChangeLog:

2014-11-11  Manuel López-Ibáñez  <manu@gcc.gnu.org>

PR fortran/44054
* gfortran.h (warn_use_without_only): Remove.
(gfc_diagnostics_finish): Declare.
* error.c: Include tree-diagnostics.h
(gfc_format_decoder): New.
(gfc_diagnostics_init): Use gfc_format_decoder. Set default caret
char.
(gfc_diagnostics_finish): Restore tree diagnostics defaults, but
keep gfc_diagnostics_starter and finalizer. Restore default caret.
* options.c: Remove all uses of warn_use_without_only.
* lang.opt (Wuse-without-only): Add Var.
* f95-lang.c (gfc_be_parse_file): Call gfc_diagnostics_finish.
* module.c (gfc_use_module): Use gfc_warning_now_2.
* parse.c (decode_statement): Likewise.
(decode_gcc_attribute): Likewise.
(next_free): Likewise.
(next_fixed): Likewise.

gcc/testsuite/ChangeLog:

2014-11-11  Manuel López-Ibáñez  <manu@gcc.gnu.org>

PR fortran/44054
* lib/gfortran-dg.exp: Update regexp to match locus and message
without caret.
* gfortran.dg/use_without_only_1.f90: Add column numbers.
        * gfortran.dg/warnings_are_errors_1.f: Update.

From-SVN: r217383

* reorg.c (fill_slots_from_thread): Do not copy frame-related insns.

From-SVN: r217381

associative-containers.cc: Add new file missed from earlier commit.

2014-11-11  Siva Chandra Reddy  <sivachandra@google.com>

* testsuite/libstdc++-xmethods/associative-containers.cc: Add new
file missed from earlier commit.

From-SVN: r217378

re PR target/61535 (SIGBUS in gen_group_rtx compiling 64-bit gcc.dg/vect/vect-singleton_1.c)

PR target/61535
* config/sparc/sparc.c (function_arg_vector_value): Deal with vectors
smaller than 8 bytes.
(sparc_function_arg_1): Tweak.
(sparc_function_value_1): Tweak.

From-SVN: r217377

Merger of dmalcolm/jit branch from git

ChangeLog:
* ChangeLog.jit: New.
* MAINTAINERS (Various Maintainers): Add myself as jit maintainer.

contrib/ChangeLog:
* ChangeLog.jit: New.
* jit-coverage-report.py: New file: a script to print crude
code-coverage information for the libgccjit API.

gcc/ChangeLog:
* ChangeLog.jit: New.
* Makefile.in (doc_build_sys): New variable, set to "sphinx" if
sphinx is installed, falling back to "texinfo" otherwise.
(FULL_DRIVER_NAME): New variable, adapted from the
install-driver target.  New target, a symlink within the builddir,
linked to "xgcc", for use when running the JIT library from the
builddir.
(MOSTLYCLEANFILES): Add FULL_DRIVER_NAME.
(install-driver): Use $(FULL_DRIVER_NAME) rather than spelling it
out.
* configure.ac (doc_build_sys): New variable, set to "sphinx" if
sphinx is installed, falling back to "texinfo" otherwise.
(GCC_DRIVER_NAME): Generate a gcc-driver-name.h file containing
GCC_DRIVER_NAME for the benefit of jit/internal-api.c.
* configure: Regenerate.
* doc/install.texi (--enable-host-shared): Specify that this is
required when building libgccjit.
(Tools/packages necessary for modifying GCC): Add Sphinx.
* timevar.def (TV_JIT_REPLAY): New.
(TV_ASSEMBLE): New.
(TV_LINK): New.
(TV_LOAD): New.

gcc/java/ChangeLog:
* gcc/ChangeLog.jit: New.

gcc/jit/ChangeLog:
* ChangeLog.jit: New.
* ChangeLog: New.
* Make-lang.in: New.
* TODO.rst: New.
* config-lang.in: New.
* docs/Makefile: New.
* docs/_build/texinfo/Makefile: New.
* docs/_build/texinfo/factorial.png: New.
* docs/_build/texinfo/libgccjit.texi: New.
* docs/_build/texinfo/sum-of-squares.png: New.
* docs/conf.py: New.
* docs/examples/tut01-hello-world.c: New.
* docs/examples/tut02-square.c: New.
* docs/examples/tut03-sum-of-squares.c: New.
* docs/examples/tut04-toyvm/Makefile: New.
* docs/examples/tut04-toyvm/factorial.toy: New.
* docs/examples/tut04-toyvm/fibonacci.toy: New.
* docs/examples/tut04-toyvm/toyvm.c: New.
* docs/index.rst: New.
* docs/internals/index.rst: New.
* docs/intro/factorial.png: New.
* docs/intro/index.rst: New.
* docs/intro/sum-of-squares.png: New.
* docs/intro/tutorial01.rst: New.
* docs/intro/tutorial02.rst: New.
* docs/intro/tutorial03.rst: New.
* docs/intro/tutorial04.rst: New.
* docs/topics/contexts.rst: New.
* docs/topics/expressions.rst: New.
* docs/topics/functions.rst: New.
* docs/topics/index.rst: New.
* docs/topics/locations.rst: New.
* docs/topics/objects.rst: New.
* docs/topics/results.rst: New.
* docs/topics/types.rst: New.
* dummy-frontend.c: New.
* jit-builtins.c: New.
* jit-builtins.h: New.
* jit-common.h: New.
* jit-playback.c: New.
* jit-playback.h: New.
* jit-recording.c: New.
* jit-recording.h: New.
* libgccjit++.h: New.
* libgccjit.c: New.
* libgccjit.h: New.
* libgccjit.map: New.
* notes.txt: New.

gcc/testsuite/ChangeLog:
* ChangeLog.jit: New.
* jit.dg/all-non-failing-tests.h: New.
* jit.dg/harness.h: New.
* jit.dg/jit.exp: New.
* jit.dg/test-accessing-struct.c: New.
* jit.dg/test-accessing-union.c: New.
* jit.dg/test-array-as-pointer.c: New.
* jit.dg/test-arrays.c: New.
* jit.dg/test-calling-external-function.c: New.
* jit.dg/test-calling-function-ptr.c: New.
* jit.dg/test-combination.c: New.
* jit.dg/test-dot-product.c: New.
* jit.dg/test-empty.c: New.
* jit.dg/test-error-accessing-field-in-other-struct.c: New.
* jit.dg/test-error-adding-to-terminated-block.c: New.
* jit.dg/test-error-array-as-pointer.c: New.
* jit.dg/test-error-bad-cast.c: New.
* jit.dg/test-error-block-in-wrong-function.c: New.
* jit.dg/test-error-call-through-ptr-with-mismatching-args.c: New.
* jit.dg/test-error-call-through-ptr-with-non-function.c: New.
* jit.dg/test-error-call-through-ptr-with-non-pointer.c: New.
* jit.dg/test-error-call-through-ptr-with-not-enough-args.c: New.
* jit.dg/test-error-call-through-ptr-with-too-many-args.c: New.
* jit.dg/test-error-call-with-mismatching-args.c: New.
* jit.dg/test-error-call-with-not-enough-args.c: New.
* jit.dg/test-error-call-with-too-many-args.c: New.
* jit.dg/test-error-dereference-field-of-non-pointer.c: New.
* jit.dg/test-error-dereference-read-of-non-pointer.c: New.
* jit.dg/test-error-get-type-bad-enum.c: New.
* jit.dg/test-error-index-not-a-numeric-type.c: New.
* jit.dg/test-error-mismatching-types-in-assignment.c: New.
* jit.dg/test-error-mismatching-types-in-call.c: New.
* jit.dg/test-error-missing-return.c: New.
* jit.dg/test-error-new-binary-op-bad-op.c: New.
* jit.dg/test-error-new-function-bad-kind.c: New.
* jit.dg/test-error-new-unary-op-bad-op.c: New.
* jit.dg/test-error-null-passed-to-api.c: New.
* jit.dg/test-error-return-within-void-function.c: New.
* jit.dg/test-error-unreachable-block.c: New.
* jit.dg/test-error-unterminated-block.c: New.
* jit.dg/test-error-value-not-a-numeric-type.c: New.
* jit.dg/test-expressions.c: New.
* jit.dg/test-factorial.c: New.
* jit.dg/test-fibonacci.c: New.
* jit.dg/test-functions.c: New.
* jit.dg/test-fuzzer.c: New.
* jit.dg/test-hello-world.c: New.
* jit.dg/test-linked-list.c: New.
* jit.dg/test-long-names.c: New.
* jit.dg/test-nested-contexts.c: New.
* jit.dg/test-nested-loops.c: New.
* jit.dg/test-operator-overloading.cc: New.
* jit.dg/test-quadratic.c: New.
* jit.dg/test-quadratic.cc: New.
* jit.dg/test-reading-struct.c: New.
* jit.dg/test-string-literal.c: New.
* jit.dg/test-sum-of-squares.c: New.
* jit.dg/test-threads.c: New.
* jit.dg/test-types.c: New.
* jit.dg/test-using-global.c: New.
* jit.dg/test-volatile.c: New.

include/ChangeLog:
* ChangeLog.jit: New.

libbacktrace/ChangeLog:
* ChangeLog.jit: New.

libcpp/ChangeLog:
* ChangeLog.jit: New.

libdecnumber/ChangeLog:
* ChangeLog.jit: New.

libiberty/ChangeLog:
* ChangeLog.jit: New.

zlib/ChangeLog:
* ChangeLog.jit: New.

From-SVN: r217374

re PR libstdc++/61107 (stl_algo.h: std::__inplace_stable_partition() doesn't process the whole data range)

2014-11-11  François Dumont  <fdumont@gcc.gnu.org>

PR libstdc++/61107
* include/bits/stl_algo.h (__inplace_stable_partition): Delete.
(__stable_partition_adaptive): Return __first if range length is 1.
(__stable_partition): Adapt.
* testsuite/util/testsuite_new_operators.h: New.
* testsuite/25_algorithms/stable_sort/1.cc: Test algo in simulated
constraint memory context.
* testsuite/25_algorithms/inplace_merge/1.cc: Likewise.
* testsuite/25_algorithms/stable_partition/1.cc: Likewise.

From-SVN: r217370

re PR target/63610 (Fixing the libtool bug for Yosemite (darwin14))

toplevel:
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* libtool.m4: Fix globbing of darwin versions.

boehm-gc/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

gcc/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libatomic/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libbacktrace/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libcc1/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libcilkrts/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libffi/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libgfortran/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libgomp/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libitm/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libobjc/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libquadmath/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libsanitizer/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libssp/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libstdc++-v3/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libvtv/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

lto-plugin/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libjava/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

libjava/classpath/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

zlib/
2014-11-11  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>

PR target/63610
* configure: Regenerate.

From-SVN: r217366

cloog.m4: Remove.

2014-11-11  Tobias Burnus  <burnus@net-b.de>

        * config/cloog.m4: Remove.
        * Makefile.def: Remove CLooG.
        * Makefile.tpl: Ditto.
        * configure.ac: Ditto.
        * configure: Regenerate.
        * Makefile.in: Ditto.

From-SVN: r217364

[Patch AArch64] Fix up BSL expander for floating point types

gcc/

* config/aarch64/aarch64-simd.md
(aarch64_simd_bsl<mode>_internal): Remove float cases, canonicalize.
(aarch64_simd_bsl<mode>): Add gen_lowpart expressions where we
are punning between float vectors and integer vectors.

gcc/testsuite/

* gcc.target/aarch64/vbslq_f64_1.c: New.
* gcc.target/aarch64/vbslq_f64_2.c: Likewise.
* gcc.target/aarch64/vbslq_u64_1.c: Likewise.
* gcc.target/aarch64/vbslq_u64_2.c: Likewise.

From-SVN: r217362

re PR c++/63265 (Constexpr variables can trigger spurious compiler warnings)

/cp
2014-11-11  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/63265
* pt.c (tsubst_copy_and_build, case COND_EXPR): Maybe fold to
constant the condition.

/testsuite
2014-11-11  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/63265
* g++.dg/cpp0x/constexpr-63265.C: New.

From-SVN: r217361

alpha.c (alpha_emit_conditional_branch): Replace open-coded swap with std::swap to swap values.

* config/alpha/alpha.c (alpha_emit_conditional_branch): Replace
open-coded swap with std::swap to swap values.
(alpha_emit_setcc): Ditto.
(alpha_emit_conditional_move): Ditto.
(alpha_split_tmode_pair): Ditto.

From-SVN: r217360

tree-vect-data-refs.c (vect_shift_permute_load_chain): Extend shift permutations on power of 2 cases.

gcc/

        * tree-vect-data-refs.c (vect_shift_permute_load_chain): Extend shift
permutations on power of 2 cases.

gcc/testsuites/

* gcc.target/i386/pr52252-atom-1.c: New.

From-SVN: r217359

[AArch64] Properly guard CUMULATIVE_ARGS definition and remove 'enum' from machine_mode in aarch64.h

* config/aarch64/aarch64.h (MACHMODE): Remove 'enum' keyword.
(CUMULATIVE_ARGS): Guard on !defined(USED_FOR_TARGET).

From-SVN: r217358

mklog: Symbol '}' stops search for changes.

contrib/

2014-11-11  Marat Zakirov  <m.zakirov@samsung.com>

* mklog: Symbol '}' stops search for changes.

From-SVN: r217357

tree-core.h (pedantic_lvalues): Remove.

2014-11-11  Richard Biener  <rguenther@suse.de>

* tree-core.h (pedantic_lvalues): Remove.
* fold-const.c (pedantic_lvalues): Likewise.
(pedantic_non_lvalue_loc): Remove conditional non_lvalue_loc call.

c/
* c-decl.c (c_init_decl_processing): Do not set pedantic_lvalues
to true.

From-SVN: r217356

re PR bootstrap/63622 (Bootstrap fails on x86_64-apple-darwin1[34] after revision r216305)

PR ipa/63622

        PR ipa/63795
* g++.dg/ipa/ipa-icf-4.C: Add more precise dump scan.
* g++.dg/ipa/ipa-icf-5.C: Add condition for target
with symbol alias support.
* ipa-icf.c (sem_function::merge): Add new target symbol alias
support guard.
(sem_variable::merge): Likewise.
* ipa-icf.h (target_supports_symbol_aliases_p): New function.

From-SVN: r217355

match.pd: Implement patterns from associate_plusminus and factor in differences from the...

2014-11-11  Richard Biener  <rguenther@suse.de>

* match.pd: Implement patterns from associate_plusminus
and factor in differences from the fold-const.c implementation.
* fold-const.c (fold_binary_loc): Remove patterns here.
* tree-ssa-forwprop.c (associate_plusminus): Remove.
(pass_forwprop::execute): Don't call it.
* tree.c (tree_nop_conversion_p): New function, factored
from tree_nop_conversion.
* tree.h (tree_nop_conversion_p): Declare.

From-SVN: r217349

system.h: Include algorithm and utility.

* system.h: Include algorithm and utility.
* rtl.h: Do not include utility here.
* wide-int.h: Ditto.
* tree-vect-data-refs.c (swap): Remove template.
(vect_prune_runtime_alias_test_list): Use std::swap instead of swap.

From-SVN: r217348

re PR ada/42978 (gnatmake doesn't report the options it actually provides to ranlib)

PR ada/42978
* mlib-utl.adb (ar): Output the options passed to ranlib.

From-SVN: r217346

[Fortran] PR 63701 Make sure variable is always used initialised

PR fortran/63701
* trans-expr.c (gfc_get_tree_for_caf_expr): Initialise found to false.

From-SVN: r217345

xmethods.py: Add xmethods for associative containers.

2014-11-11  Siva Chandra Reddy  <sivachandra@google.com>

* python/libstdcxx/v6/xmethods.py: Add xmethods for associative
containers.
* testsuite/libstdc++-xmethods/associative-containers.cc: New file.

From-SVN: r217344

re PR libstdc++/63811 (r217322 breaks bootstrap)

PR libstdc++/63811
* include/bits/stl_deque.h (_Deque_base::_M_move_impl()): Avoid using
badname.

From-SVN: r217343

re PR bootstrap/63699 (Bootstrap fails in libcc1 on darwin14)

PR bootstrap/63699
PR bootstrap/63750

* system.h: Include <string> before "safe-ctype.h"
* wide-int.h (wi::smin, wi::smax, wi::umin, wi::umax): Prefix
calls to min/max with wi namespace.
* ipa-chkp.c: Don't include <string>.

* plugin.cc: Don't include <string>.

From-SVN: r217342

invoke.texi (-masm-syntax-unified): Reword and fix typo.

2014-11-11  Terry Guo  <terry.guo@arm.com>

* doc/invoke.texi (-masm-syntax-unified): Reword and fix typo.
* config/arm/thumb1.md (*thumb_mulsi3): Use movs to move low
registers.
(*thumb1_movhf): Likewise.

From-SVN: r217341

sreal.c (sreal::to_int): Use INTTYPE_MAXIMUM (int64_t) instead of INT64_MAX.

* sreal.c (sreal::to_int): Use INTTYPE_MAXIMUM (int64_t)
instead of INT64_MAX.

From-SVN: r217340

download_prerequisites: Stop downloading CLooG.

2014-11-11  Tobias Burnus  <burnus@net-b.de>

contrib/
        * download_prerequisites: Stop downloading CLooG.
gcc/
        * doc/install.texi (Prerequisites): Remove CLooG.

From-SVN: r217339

Add illegal cilk checks to C++ front.

Add calls for several illegal Cilk cases to the C++ frontend.
C++ usually doesn't ICE unlike C on illegal cilk, but it's
better to match C in what is allowed and what is not.

if (_Cilk_spawn ...) is still not errored, but at least it doesn't ICE.

gcc/cp/:

2014-11-10  Andi Kleen  <ak@linux.intel.com>

* semantics.c (finish_goto_stmt): Call check_no_cilk.
(finish_while_stmt_cond): Dito.
(finish_do_stmt): Dito.
(finish_for_cond): Dito.
(finish_switch_cond): Dito.

From-SVN: r217337

Error out for Cilk_spawn or array expression in forbidden places

_Cilk_spawn or Cilk array expressions are only allowed on their own,
but not in for(), if(), switch, do, while, goto, etc.
The C parser didn't always check for that, which lead to ICEs earlier
for invalid code.

Add a generic helper that checks this and call it where needed
in the C frontend.

I chose to allow spawn/array for for init and increment expressions.
While the Cilk spec could be interpreted to forbid it there too
there didn't seem any reason to not allow it.

One dark corner is spawn, array in statement expressions not at
the end. Right now that's forbidden too.

gcc/c-family/:

2014-11-10  Andi Kleen  <ak@linux.intel.com>

PR c/60804
* c-common.h (check_no_cilk): Declare.
* cilk.c (get_error_location): New function.
(check_no_cilk): Dito.

gcc/c/:

2014-11-10  Andi Kleen  <ak@linux.intel.com>

PR c/60804
* c-parser.c (c_parser_statement_after_labels): Call
check_no_cilk.
(c_parser_if_statement): Dito.
(c_parser_switch_statement): Dito.
(c_parser_while_statement): Dito.
(c_parser_do_statement): Dito.
(c_parser_for_statement): Dito.
* c-typeck.c (c_finish_loop): Dito.

From-SVN: r217336

Fix some of the existing Cilk tests for the new errors.

gcc/testsuite/:

2014-11-10  Andi Kleen  <ak@linux.intel.com>

* c-c++-common/cilk-plus/AN/misc.c (main): Handle
new cilk errors.

From-SVN: r217335