re PR libgomp/64635 (darwin produces libgomp-plugin-host_nonshm.1.dylib but tries to load libgomp-plugin-host_nonshm.so.1)

2015-01-28  Jack Howarth  <howarth.at.gcc@gmail.com>

PR libgomp/64635
* configure.tgt (*-*-aix*): Use config_path "aix posix".
(*-*-darwin*): Use config_path "bsd darwin posix".
(*-*-hpux*): Use config_path "hpux posix".
* target.c: Add include of plugin-suffix.h and use
SONAME_SUFFIX macro.
* config/aix/plugin-suffix.h: New file.
* config/darwin/plugin-suffix.h: New file.
* config/hpux/plugin-suffix.h: New file.
* config/posix/plugin-suffix.h: New file.

From-SVN: r220218

re PR target/64659 ([SH] Immedate values not used for atomic ops)

gcc/
PR target/64659
* config/sh/predicates.md (atomic_arith_operand,
atomic_logical_operand): Remove.
* config/sh/sync.md (fetchop_predicate, fetchop_constraint): Remove.
(atomic_arith_operand_0): New predicate.
(atomic_compare_and_swap<mode>): Use arith_reg_dest for output values.
Use atomic_arith_operand_0 for input values.
(atomic_compare_and_swapsi_hard, atomic_compare_and_swap<mode>_hard,
atomic_compare_and_swap<mode>_soft_gusa,
atomic_compare_and_swap<mode>_soft_tcb,
atomic_compare_and_swap<mode>_soft_imask): Use arith_reg_dest and
arith_reg_operand instead of register_operand.
(atomic_exchange<mode>): Use arith_reg_dest for output value.  Use
atomic_arith_operand_0 for newval input.
(atomic_exchangesi_hard, atomic_exchange<mode>_hard,
atomic_exchange<mode>_soft_gusa, atomic_exchange<mode>_soft_tcb,
atomic_exchange<mode>_soft_imask): Use arith_reg_dest and
arith_reg_operand instead of register_operand.
(atomic_arith_operand_1, atomic_logical_operand_1): New predicates.
fetchop_predicate_1, fetchop_constraint_1_llcs,
fetchop_constraint_1_gusa, fetchop_constraint_1_tcb,
fetchop_constraint_1_imask): New code iterator attributes.
(atomic_fetch_<fetchop_name><mode>): Use arith_reg_dest instead of
register_operand.  Use fetchop_predicate_1.
(atomic_fetch_<fetchop_name>si_hard,
atomic_fetch_<fetchop_name><mode>_hard): Use arith_reg_dest instead of
register_operand.  Use fetchop_predicate_1, fetchop_constraint_1_llcs.
(atomic_fetch_<fetchop_name><mode>_soft_gusa): Use arith_reg_dest
and arith_reg_operand instead of register_operand.  Use
fetchop_predicate_1, fetchop_constraint_1_gusa.
(atomic_fetch_<fetchop_name><mode>_soft_tcb): Use arith_reg_dest
and arith_reg_operand instead of register_operand.  Use
fetchop_predicate_1, fetchop_constraint_1_tcb.  Adjust asm sequence
to allow R0 usage.
(atomic_fetch_<fetchop_name><mode>_soft_imask): Use arith_reg_dest
and arith_reg_operand instead of register_operand.  Use
fetchop_predicate_1, fetchop_constraint_1_imask.  Adjust asm sequence
to allow R0 usage.
(atomic_fetch_nand<mode>): Use arith_reg_dest instead of
register_operand.  Use atomic_logical_operand_1.
(atomic_fetch_nandsi_hard, atomic_fetch_nand<mode>_hard,
atomic_fetch_nand<mode>_soft_gusa): Use arith_reg_dest and
arith_reg_operand instead of register_operand.
(atomic_fetch_nand<mode>_soft_tcb, atomic_fetch_nand<mode>_soft_imask):
Use arith_reg_dest and arith_reg_operand instead of register_operand.
Use logical_operand and rK08.  Adjust asm sequence to allow R0 usage.
(atomic_<fetchop_name>_fetch<mode>): Use arith_reg_dest instead of
register_operand.  Use fetchop_predicate_1.
(atomic_<fetchop_name>_fetchsi_hard,
atomic_<fetchop_name>_fetch<mode>_hard): Use arith_reg_dest and
arith_reg_operand instead of register_operand.  Use fetchop_predicate_1,
fetchop_constraint_1_llcs.
(atomic_<fetchop_name>_fetch<mode>_soft_gusa): Use arith_reg_dest and
arith_reg_operand instead of register_operand.  Use fetchop_predicate_1,
fetchop_constraint_1_gusa.
(atomic_<fetchop_name>_fetch<mode>_soft_tcb): Use arith_reg_dest and
arith_reg_operand instead of register_operand.  Use fetchop_predicate_1,
fetchop_constraint_1_tcb.  Adjust asm sequence to allow R0 usage.
(atomic_<fetchop_name>_fetch<mode>_soft_imask): Use arith_reg_dest and
arith_reg_operand instead of register_operand.  Use fetchop_predicate_1,
fetchop_constraint_1_imask.  Adjust asm sequence to allow R0 usage.
(atomic_nand_fetch<mode>): Use arith_reg_dest instead of
register_operand.  Use atomic_logical_operand_1.
(atomic_nand_fetchsi_hard, atomic_nand_fetch<mode>_hard,
atomic_nand_fetch<mode>_soft_gusa): Use arith_reg_dest and
arith_reg_operand instead of register_operand.
(atomic_nand_fetch<mode>_soft_tcb): Use arith_reg_dest and
arith_reg_operand instead of register_operand.  Use logical_operand
and K08.  Adjust asm sequence to allow R0 usage.
(atomic_nand_fetch<mode>_soft_imask): Use arith_reg_dest and
arith_reg_operand instead of register_operand.  Use logical_operand
and K08.

gcc/testsuite/
PR target/64659
* gcc.target/sh/sh.exp
(check_effective_target_atomic_model_soft_gusa_available,
check_effective_target_atomic_model_soft_tcb_available,
check_effective_target_atomic_model_soft_imask_available,
check_effective_target_atomic_model_hard_llcs_available): New.
* gcc.target/sh/pr64659-0.h: New.
* gcc.target/sh/pr64659-1.c: New.
* gcc.target/sh/pr64659-2.c: New.
* gcc.target/sh/pr64659-3.c: New.
* gcc.target/sh/pr64659-4.c: New.

From-SVN: r220217

re PR other/63504 (Issues found by --enable-checking=valgrind)

PR other/63504
* dwarf2out.c (add_AT_wide, mem_loc_descriptor, loc_descriptor):
Use ggc_alloc<wide_int> instead of ggc_cleared_alloc<wide_int>.
(attr_checksum, attr_checksum_ordered, hash_loc_operands): Checksum
only get_full_len HOST_WIDE_INTs from get_val () array rather than
all bits in *val_wide.

From-SVN: r220216

checksums and test-text

* fixincl.tpl: add a check sum acceptance test
* fixlib.h: enumerate it
* fixincl.c: handle it
* README: document it and document the handling of test_text

From-SVN: r220215

* lto-symtab.c (lto_varpool_replace_node): Merge TLS models.

From-SVN: r220214

varpool.c (tls_model_names): Fix names.

* varpool.c (tls_model_names): Fix names.
(varpool_node::dump): Dump tls- prefix for tls models.

From-SVN: r220213

Call release_input_file in claim_file_handler

PR lto/64837
* lto-plugin.c (release_input_file): New.
(claim_file_handler): Call release_input_file.
(onload): Set release_input_file.

From-SVN: r220212

nvptx mkoffload.

gcc/
* config/nvptx/mkoffload.c: New file.
* config/nvptx/t-nvptx: Add build rules for it.
* config.gcc <nvptx-*> [$enable_as_accelerator = yes]
(extra_programs): Add mkoffload.
* config/nvptx/nvptx.c (nvptx_record_offload_symbol): New
function.
(TARGET_RECORD_OFFLOAD_SYMBOL): Define macro to use it.

Co-Authored-By: Bernd Schmidt <bernds@codesourcery.com>
Co-Authored-By: Nathan Sidwell <nathan@codesourcery.com>
From-SVN: r220209

atomic-op-consume.c (scan-assember-times): Adjust scan-assembler-times pattern.

2015-01-28  Alex Velenko  <Alex.Velenko@arm.com>

gcc/testsuite/

* gcc.target/arm/atomic-op-consume.c (scan-assember-times): Adjust
scan-assembler-times pattern.

From-SVN: r220208

re PR libstdc++/64828 (libstdc++-v3/libsupc++/del_opvs.cc:30:1: warning: ‘void operator delete [](void*, std::size_t)’ is a usual (non-placement) deallocation function in C++14 (or with -fsized-deallocation) [-Wc++14-compat])

PR libstdc++/64828
* libsupc++/Makefile.am: Compile del_opvs.cc as C++14.
* libsupc++/Makefile.in: Regenerate.
* src/c++11/Makefile.in: Regenerate.

From-SVN: r220207

[libstdc++][testsuite][reverted] Remove check for truncation overflow

    * testsuite/lib/libstdc++.exp (v3_target_compile): Remove
    check for unsupported.
    (v3_target_compile_as_c): Likewise.

From-SVN: r220206

re PR middle-end/64809 (ICE at -O3 with -g enabled on x86_64-linux-gnu (in 32-bit mode))

gcc/

PR middle-end/64809
* cfgexpand.c (reorder_operands): Skip debug gimples.

gcc/testsuite/

PR middle-end/64809
* gcc.dg/pr64809.c: New test.

From-SVN: r220205

re PR tree-optimization/64277 (Incorrect warning "array subscript is above array bounds")

gcc/

PR tree-optimization/64277
* tree-ssa-loop-niter.c (record_nonwrapping_iv): Use base
range info when possible to refine estimation.

gcc/testsuite/

PR tree-optimization/64277
* gcc.dg/pr64277.c: New.

From-SVN: r220204

re PR tree-optimization/64718 (Bad 16-bit bswap replacement)

2015-01-28  Thomas Preud'homme  <thomas.preudhomme@arm.com>

    gcc/
    PR tree-optimization/64718
    * tree-ssa-math-opts.c (pass_optimize_bswap::execute): Make bswap_type
    be a 16bit unsigned integer when n->range is 16.
    (bswap_replace): Convert src to that type if necessary for all bswap
    sizes.  Fix rotation right notation in nearby comment.  Use bswap_type
    set in pass_optimize_bswap::execute ().

    gcc/testsuite/
    PR tree-optimization/64718
    * gcc.c-torture/execute/pr64718.c: New test.

From-SVN: r220203

[Patch AArch64] Make integer vabs intrinsics UNSPECs

gcc/

* config/aarch64/aarch64-simd.md (aarch64_abs<mode>): New.
* config/aarch64/aarch64-simd-builtins.def (abs): Split by
integer and floating point variants.
* config/aarch64/iterators.md (unspec): Add UNSPEC_ABS.

gcc/testsuite/

* gcc.target/aarch64/abs_2.c: New.

From-SVN: r220202

re PR libstdc++/64798 (g++.old-deja/g++.eh/badalloc1.C FAILs)

2015-01-28  Richard Biener  <rguenther@suse.de>

PR libstdc++/64798
* libsupc++/eh_alloc.cc (struct allocated_entry): Align
data member.
(pool::allocate): Adjust allocation size and alignment to
that change.
(pool::free): Adjust pointer offsetting.

From-SVN: r220201

Prohibit vector modes in accumulators.

gcc/
* config/mips/mips.c (mips_hard_regno_mode_ok_p): Prohibit accumulators
for all vector modes.

From-SVN: r220200

Regression cleanup for MIPS nan2008 toolchain.

gcc/testsuite/
* lib/target-supports.exp (check_effective_target_mips_nanlegacy): New.
* gcc.target/mips/loongson-simd.c: Require legacy NaN support.
* gcc.target/mips/mips.exp (mips-dg-options): Imply -mnan=legacy for
ISA rev < 2.

From-SVN: r220199

re PR bootstrap/64612 (profiledbootstrap failures)

PR bootstrap/64612
* lib/target-supports.exp (check_effective_target_comdat_group): New.
* g++.dg/ipa/pr64612.C: Guard scan-assembler test with
{ target comdat_group }.

* doc/sourcebuild.texi (comdat_group): Document.

From-SVN: r220198

* config/arm/thumb1.md (*thumb1_movpc_insn): New insn pattern.

From-SVN: r220197

Daily bump.

From-SVN: r220196

re PR c++/58597 (ICE with lambda in default argument of template function)

PR c++/58597
* lambda.c (maybe_add_lambda_conv_op): Check cfun rather than
current_function_decl.

From-SVN: r220192

toplev: gcc version information for jit

gcc/ChangeLog:
* toplev.c (print_version): Add param "show_global_state", and
only print GGC and plugin information if it is true.
(init_asm_output): Pass in "true" for the new param when calling
print_version.
(process_options): Likewise.
(toplev::main): Likewise.
* toplev.h (print_version): Add new param to decl.

gcc/jit/ChangeLog:
* docs/internals/test-hello-world.exe.log.txt: Add example version
lines.
* jit-common.h (gcc::jit::dump::get_file): New accessor.
* jit-logging.c: Include toplev.h.
(gcc::jit::logger::logger): Log the GCC version.
* jit-recording.c: Include toplev.h.
(gcc:jit::recording::context::dump_reproducer_to_file): Log the
GCC version.

From-SVN: r220190

re PR fortran/63861 (OpenACC coarray ICE  (also with OpenMP?))

2015-01-27  Tobias Burnus  <burnus@net-b.de>

        PR fortran/63861
gcc/fortran/
        * trans-openmp.c (gfc_has_alloc_comps, gfc_trans_omp_clauses):
        Fix handling for scalar coarrays.
        * trans-types.c (gfc_get_element_type): Add comment.
gcc/testsuite/
        * gfortran.dg/goacc/coarray_2.f90: New.

From-SVN: r220189

re PR ipa/60871 (internal compiler error: in possible_polymorphic_call_targets, at ipa-devirt.c:1510)

PR ipa/60871
PR ipa/64139
* tree.c (lookup_binfo_at_offset): New function.
(get_binfo_at_offset): Use it.

* g++.dg/torture/pr64139.C: New testcase.
* g++.dg/torture/pr60871.C: Likewise.

From-SVN: r220185

re PR c++/63889 (Ice with redundant static in class scope constexpr variable template.)

PR c++/63889
* pt.c (finish_template_variable): Move from semantics.c.
Handle multiple template arg levels.  Handle coercion here.
(lookup_template_variable): Not here.

From-SVN: r220183

re PR fortran/64771 (ICE(segfault) when passing coarrays around; ICE in gfc_zero_size_array in arith.c:1637)

2015-01-27  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>

        PR fortran/64771
        * interface.c: Remove <algorithm>.
        (check_dummy_characteristics): Use MAX instead of std::max.

From-SVN: r220182

re PR fortran/64230 (Invalid memory reference in a compiler-generated finalizer for allocatable component)

2015-01-27  Janus Weil  <janus@gcc.gnu.org>

PR fortran/64230
* gfortran.dg/class_allocate_18.f90: Remove -fsanitize option to
prevent linking errors.

From-SVN: r220181

Fix build under cygwin/64.

2015-01-27  Bernd Edlinger  <bernd.edlinger@hotmail.de>

        Fix build under cygwin/64.
        * adaint.h: Add check for __CYGWIN__.
        * mingw32.h: Prevent windows.h from including x86intrin.h in GCC.

From-SVN: r220180

re PR ipa/64282 (ICE in gimple_get_virt_method_for_vtable, at gimple-fold.c:5635)

PR ipa/64282
* gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert
on vtable being vtable.
* g++.dg/torture/pr64282.C: New testcase.

From-SVN: r220177

[AArch64][test] Fix FAIL: gcc.target/aarch64/store-pair-1.c scan-assembler stp\tw[0-9]+, w[0-9]+

* gcc.target/aarch64/store-pair-1.c: Update scan-assembler to check
for wzr reg.

From-SVN: r220176

sendmsg.c: Add prototypes for __objc_get_forward_imp and get_imp.

2015-01-27  Matthias Klose  <doko@ubuntu.com>

        * sendmsg.c: Add prototypes for __objc_get_forward_imp and get_imp.

From-SVN: r220175

Add myself to maintaners.

/
* MAINTAINERS (Write After Approval): Add myself.

From-SVN: r220172

extend.texi: s/390: Update documentation of hotpatch attribute.

2015-01-27  Dominik Vogt  <vogt@linux.vnet.ibm.com>

        * doc/extend.texi: s/390: Update documentation of hotpatch attribute.
        * doc/invoke.texi (-mhotpatch): s/390: Update documentation of
        -mhotpatch= option.
        * config/s390/s390.opt (mhotpatch): s/390: Remove -mhotpatch and
        -mno-hotpatch options.  Change syntax of -mhotpatch= option.
        * config/s390/s390.c (s390_hotpatch_trampoline_halfwords_default):
        Renamed.
        (s390_hotpatch_trampoline_halfwords_max): Renamed.
        (s390_hotpatch_hw_max): New name.
        (s390_hotpatch_trampoline_halfwords): Renamed.
        (s390_hotpatch_hw_before_label): New name.
        (get_hotpatch_attribute): Removed.
        (s390_hotpatch_hw_after_label): New name.
        (s390_handle_hotpatch_attribute): Add second parameter to hotpatch
        attribute.
        (s390_attribute_table): Ditto.
        (s390_function_num_hotpatch_trampoline_halfwords): Renamed.
        (s390_function_num_hotpatch_hw): New name.
        Remove special handling of inline functions and hotpatching.
        Return number of nops before and after the function label.
        (s390_can_inline_p): Removed.
        (s390_asm_output_function_label): Emit a configurable number of nops
        after the function label.
        (s390_option_override): Update -mhotpatch= syntax and remove -mhotpatch.
        (TARGET_CAN_INLINE_P) Removed.
        (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): New.

2015-01-27  Dominik Vogt  <vogt@linux.vnet.ibm.com>

* gcc/testsuite/gcc.target/s390/hotpatch-13.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-14.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-15.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-16.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-17.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-18.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-19.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-20.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-10.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-11.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-12.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-13.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-14.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-15.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-16.c: New testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-9.c: New
testcase.
* gcc/testsuite/gcc.target/s390/hotpatch-1.c: Testcase adjusted to
new -mhotpatch.
* gcc/testsuite/gcc.target/s390/hotpatch-10.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-11.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-12.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-2.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-3.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-4.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-5.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-6.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-7.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-8.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-9.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-1.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-2.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-3.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-4.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-5.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-6.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-7.c: Likewise.
* gcc/testsuite/gcc.target/s390/hotpatch-compile-8.c: Likewise.

From-SVN: r220171

[AArch64] Improve bit-test-branch pattern to avoid unnecessary register clobber

2015-01-27  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
            Jiong Wang  <jiong.wang@arm.com>

  gcc/
    * config/aarch64/aarch64.md (tb<optab><mode>1): Clobber CC reg instead of
    scratch reg.
    (cb<optab><mode>1): Likewise.
    * config/aarch64/iterators.md (bcond): New define_code_attr.

  gcc/testsuite/
    * gcc.dg/long_branch.c: New testcase.

From-SVN: r220170

s390.c (s390_memory_move_cost): Increase costs for memory accesses.

2015-01-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>

* config/s390/s390.c (s390_memory_move_cost): Increase costs for
memory accesses.

From-SVN: r220169

s390.c (s390_register_move_cost): Increase costs for FPR->GPR moves.

2015-01-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>

* config/s390/s390.c (s390_register_move_cost): Increase costs for
FPR->GPR moves.

From-SVN: r220168

configure.ac: Move AM_ENABLE_MULTILIB before AC_PROG_CC.

2015-01-26  Matthias Klose  <doko@ubuntu.com>

        * configure.ac: Move AM_ENABLE_MULTILIB before AC_PROG_CC.
        * configure: Regenerate.

From-SVN: r220165

tree-vrp.c (update_value_range): Intersect the range with old recorded SSA name range information.

2015-01-27  Richard Biener  <rguenther@suse.de>

* tree-vrp.c (update_value_range): Intersect the range with
old recorded SSA name range information.

From-SVN: r220164

rl78.c (rl78_expand_prologue): In G10 mode push the BC, DE and HL registers directly, not via AX.

* config/rl78/rl78.c (rl78_expand_prologue): In G10 mode push the
BC, DE and HL registers directly, not via AX.
When decrementing the stack pointer by a large amount, transfer SP
into AX and perform the subtraction there.
(rl78_expand_epilogue): Perform the inverse of the above
enhancements.

From-SVN: r220163

cmpsi2.S: Use function start and end macros.

* config/rl78/cmpsi2.S: Use function start and end macros.
(__gcc_bcmp): New function.
* config/rl78/lshrsi3.S: Use function start and end macros.
* config/rl78/mulsi3.S: Add support for G10.
(__mulqi3): New function for G10.
* config/rl78/signbit.S: Use function start and end macros.
* config/rl78/t-rl78 (LIB2ADD): Add bit-count.S, fpbit-sf.S and
fpmath-sf.S.
(LIB2FUNCS_EXCLUDE): Define.
(LIB2FUNCS_ST): Define.
* config/rl78/trampoline.S: Use function start and end macros.
* config/rl78/vregs.h (START_FUNC): New macro.
(START_ANOTHER_FUNC): New macro.
(END_FUNC): New macro.
(END_ANOTHER_FUNC): New macro.
* config/rl78/bit-count.S: New file.  Contains assembler
implementations of the bit counting functions: ___clzhi2,
__clzsi2, ctzhi2, ctzsi2, ffshi2, ffssi2, __partityhi2,
__paritysi2, __popcounthi2 and __popcountsi2.
* config/rl78/fpbit-sf.S: New file.  Contains assembler
implementationas of the math functions: __negsf2, __cmpsf2,
__eqsf2, __nesf2, __lesf2, __ltsf2, __gesf2, gtsf2, __unordsf2,
__fixsfsi,  __fixunssfsi, __floatsisf and __floatunssisf.
* config/rl78/fpmath-sf.S: New file.  Contains assembler
implementations of the math functions: __subsf3, __addsf3,
__mulsf3 and __divsf3

From-SVN: r220162

re PR target/64368 (Several libstdc++ test failures on non-linux platforms after r218964.)

PR libstdc++/64368
* include/std/shared_mutex (shared_timed_mutex::try_lock_for,
shared_timed_mutex::try_lock_until): Only define when POSIX thread
timeouts option is supported.
(shared_timed_mutex::try_shared_lock_for,
shared_timed_mutex::try_shared_lock_until): Likewise.

From-SVN: r220161

Move Solaris/x86 CRT_GET_RFIB_DATA definition to libgcc

gcc:
* config/i386/sysv4.h (CRT_GET_RFIB_DATA): Remove.

libgcc:
* config.host (i[34567]86-*-solaris2*, x86_64-*-solaris2.1[0-9]*):
Add i386/elf-lib.h to tm_file.
* config/i386/elf-lib.h: Fix comment.
* unwind-dw2-fde-dip.c (_Unwind_IteratePhdrCallback) [__x86_64__
&& __sun__ && __svr4__]: Remove workaround.

From-SVN: r220160

re PR sanitizer/64741 (Incorrect size of UBSan type descriptors)

2015-01-27  Jakub Jelinek  <jakub@redhat.com>
    Yury Gribov  <y.gribov@samsung.com>

PR ubsan/64741
* ubsan.c (ubsan_source_location): Refactor code.
(ubsan_type_descriptor): Update type size. Refactor code.

Co-Authored-By: Yury Gribov <y.gribov@samsung.com>
From-SVN: r220159

Correctly check for 32-bit x86

* testsuite/lib/libffi.exp: Load target-supports.exp.
(run-many-tests): Only set targetabis for ia32.

From-SVN: r220158

re PR tree-optimization/56273 (Bogus -Warray-bounds warning)

2015-01-27  Richard Biener  <rguenther@suse.de>

PR tree-optimization/56273
PR tree-optimization/59124
PR tree-optimization/64277
* tree-vrp.c (vrp_finalize): Emit array-bound warnings only
from the first VRP pass.

* g++.dg/warn/Warray-bounds-6.C: New testcase.
* gcc.dg/Warray-bounds-12.c: Likewise.
* gcc.dg/Warray-bounds-13.c: Likewise.

From-SVN: r220157

re PR ipa/64776 (FAIL: gcc.dg/ipa/pr64307.c (internal compiler error) on x86_64-apple-darwin14)

PR ipa/64776
* cgraphunit.c (cgraph_node::expand_thunk): If not this_adjusting,
handle the first argument in the same loop as all the other arguments.

From-SVN: r220156

re PR rtl-optimization/61058 (ICE: RTL check: expected elt 3 type 'B', have '0' (rtx barrier) in distance_agu_use_in_bb, at config/i386/i386.c:16740 with __builtin_unreachable())

PR rtl-optimization/61058
* jump.c (cleanup_barriers): Update basic block boundaries
if BLOCK_FOR_INSN is non-NULL on PREV.

* gcc.dg/pr61058.c: New test.

From-SVN: r220155

tree-chkp.c (chkp_call_returns_bounds_p): Fix handling of bounds narrowing...

gcc/

* tree-chkp.c (chkp_call_returns_bounds_p): Fix handling of
bounds narrowing, already instrumented calls and calls to
not instrumentable functions.

gcc/testsuite/

* gcc.target/i386/chkp-narrow-bounds.c: New.

From-SVN: r220154

re PR tree-optimization/64807 (Wrong-code because of wide-int division)

PR tree-optimization/64807
* wide-int.cc (wi::divmod_internal): Clear
b_dividend[dividend_blocks_needed].

* gcc.dg/pr64807.c: New test.

From-SVN: r220153

re PR middle-end/64766 (internal compiler error: tree check: expected block, have error_mark in lower_function_body, at gimple-low.c:122)

PR c/64766
* c-typeck.c (store_init_value): Don't overwrite DECL_INITIAL
of FUNCTION_DECLs with error_mark_node.

* gcc.dg/pr64766.c: New test.

From-SVN: r220152

rl78.c (move_elim_pass): Don't optimize away volatile memory references.

* config/rl78/rl78.c (move_elim_pass): Don't optimize away
volatile memory references.

From-SVN: r220151

Add PR jit/64257 to previous ChangeLog entry

From-SVN: r220150

update_web_docs_svn: support the JIT documentation

maintainer-scripts/ChangeLog:
* update_web_docs_svn: Don't delete gcc/jit/docs or
gcc/jit/jit-common.h, gcc/jit/notes.txt. Special case the
building of the JIT docs (using sphinx-build).  Special case
copying them up.

From-SVN: r220149

Daily bump.

From-SVN: r220148

Fix gcc_jit_context_get_last_error docs

gcc/jit/ChangeLog:
* docs/topics/compilation.rst (gcc_jit_result_get_code): Fix typo.
* docs/topics/contexts.rst (gcc_jit_context_get_last_error): The
error buffer is only valid until the next call to the context.
* libgccjit.h (gcc_jit_context_get_first_error): Reword the
comment to omit mention of compiling.
(gcc_jit_context_get_last_error): The error buffer is only valid
until the next call to the context.

From-SVN: r220145

re PR target/49263 (SH Target: underutilized "TST #imm, R0" instruction)

gcc/
PR target/49263
* config/sh/sh.c (sh_split_treg_set_expr): Invoke emit_insn before
remove_insn.
* config/sh/sh.md (tstsi_t): Don't try to optimize constant with right
shifts if it already fits into K08.

gcc/testsuite/
PR target/49263
* gcc.target/sh/pr49263-4.c: New.

From-SVN: r220144

re PR target/64368 (Several libstdc++ test failures on non-linux platforms after r218964.)

PR libstdc++/64368
* config/locale/gnu/numeric_members.cc (numpunct<char>::~numpunct(),
numpunct<wchar_t>::~numpunct()): Do not set _M_data->_M_grouping.
* src/c++11/cxx11-shim_facets.cc (numpunct_shim): Remove _M_grouping
and use cache's _M_grouping field.
(__numpunct_fill_cache): Likewise.
(__moneypunct_fill_cache): Improve comments.

From-SVN: r220143

configure.ac: Move AM_ENABLE_MULTILIB before AC_PROG_CC.

2015-01-26  Matthias Klose  <doko@ubuntu.com>

        * configure.ac: Move AM_ENABLE_MULTILIB before AC_PROG_CC.
        * configure: Regenerate.

From-SVN: r220141

re PR fortran/62044 (ICE in USE statement with RENAME for extended derived type)

2015-01-26  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/62044
* resolve.c (resolve_allocate_expr): If the default initializer
is NULL, keep the original MOLD expression so that the correct
typespec is available.

2015-01-26  Paul Thomas  <pault@gcc.gnu.org>

PR fortran/62044
* gfortran.dg/allocate_with_mold_1.f90: New test

From-SVN: r220140

re PR ipa/64730 (g++.dg/ipa/pr64049-1.C ICE: SEGV when printing NULL)

PR ipa/64730
* ipa-inline.c (inline_small_functions): Print "unknown" even
if edge->call_stmt is non-NULL, but has builtins or unknown
location.

From-SVN: r220139

re PR c/64778 (ICE on invalid code on x86_64-linux-gnu: tree check: expected class ‘type’, have ‘exceptional’ (error_mark) in create_tmp_from_val, at gimplify.c:491)

PR c/64778
* c-typeck.c (convert_arguments): Return -1 if there are
error_args, even if we've diagnosed too many arguments.

* gcc.dg/pr64778.c: New test.

From-SVN: r220138

re PR middle-end/64421 (Incorrect vector function name generated for log)

PR middle-end/64421
* omp-low.c (simd_clone_mangle): If DECL_ASSEMBLER_NAME starts
with asterisk, skip the first character.

* gcc.dg/vect/pr64421.c: New test.

From-SVN: r220137

re PR fortran/64771 (ICE(segfault) when passing coarrays around; ICE in gfc_zero_size_array in arith.c:1637)

2015-01-26  Tobias Burnus  <burnus@net-b.de>

        PR fortran/64771
gcc/fortran/
        * interface.c (check_dummy_characteristics): Fix coarray
        * handling.

testsuite/
        * gfortran.dg/coarray_36.f: New.
        * gfortran.dg/coarray_37.f90: New.

From-SVN: r220136

io.c (gfc_match_inquire): Replace "-1" by a defined constant.

2015-01-26  Tobias Burnus  <burnus@net-b.de>

        * io.c (gfc_match_inquire): Replace "-1" by a defined constant.

From-SVN: r220135

PR jit/64708: remove libgccjit.so from COMPILERS

gcc/jit/ChangeLog:
PR jit/64708
* config-lang.in (compilers): Drop "libgccjit.so".

From-SVN: r220133

Revert the last P_POPCNT order change

PR target/64806
* config/i386/i386 (feature_priority): Revert the last P_POPCNT
order change.

From-SVN: r220131

re PR target/64795 (too many memory references for `lea')

        PR target/64795
* config/i386/i386.md (*movdi_internal): Also check operand 0
to determine TYPE_LEA operand.
(*movsi_internal): Ditto.

From-SVN: r220128

Fix incorrect ChangeLog formatting.

From-SVN: r220126

re PR fortran/64230 (Invalid memory reference in a compiler-generated finalizer for allocatable component)

2015-01-26  Janus Weil  <janus@gcc.gnu.org>

PR fortran/64230
* class.c (finalize_component): New argument 'sub_ns'. Insert code to
check if 'expr' is associated.
(generate_finalization_wrapper): Rename 'ptr' symbols to 'ptr1' and
'ptr2'. Pass 'sub_ns' to finalize_component.

2015-01-26  Janus Weil  <janus@gcc.gnu.org>

PR fortran/64230
* gfortran.dg/class_allocate_18.f90: New.

From-SVN: r220125

[ARM/AArch64][testsuite] Add vpaddl tests.

2015-01-26  Christophe Lyon  <christophe.lyon@linaro.org>

* gcc.target/aarch64/advsimd-intrinsics/vpaddl.c: New file.

From-SVN: r220124

[ARM/AArch64][testsuite] Add vpadal tests.

2015-01-26  Christophe Lyon  <christophe.lyon@linaro.org>

* gcc.target/aarch64/advsimd-intrinsics/vpadal.c: New file.

From-SVN: r220123

[ARM/AArch64][testsuite] Add vmnv tests.

2015-01-26  Christophe Lyon  <christophe.lyon@linaro.org>

* gcc.target/aarch64/advsimd-intrinsics/vmvn.c: New file.

From-SVN: r220122

[ARM/AArch64][testsuite] Add vmovl tests.

2015-01-26  Christophe Lyon  <christophe.lyon@linaro.org>

    * gcc.target/aarch64/advsimd-intrinsics/vmovl.c: New file.

From-SVN: r220121

rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_QUAD_MEMORY_ATOMIC.

* config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add
OPTION_MASK_QUAD_MEMORY_ATOMIC.

From-SVN: r220120

[ARM/AArch64][testsuite] Add vpadd, vpmax and vpmin tests.

2015-01-26  Christophe Lyon  <christophe.lyon@linaro.org>

* gcc.target/aarch64/advsimd-intrinsics/vpXXX.inc: New file.
* gcc.target/aarch64/advsimd-intrinsics/vpadd.c: New file.
* gcc.target/aarch64/advsimd-intrinsics/vpmax.c: New file.
* gcc.target/aarch64/advsimd-intrinsics/vpmin.c: New file.

From-SVN: r220119

[ARM/AArch64][testsuite] Add vmla_n and vmls_n tests.

2015-01-26  Christophe Lyon  <christophe.lyon@linaro.org>

* gcc.target/aarch64/advsimd-intrinsics/vmlX_n.inc: New file.
* gcc.target/aarch64/advsimd-intrinsics/vmla_n.c: New file.
* gcc.target/aarch64/advsimd-intrinsics/vmls_n.c: New file.

From-SVN: r220118

[ARM/AArch64][testsuite] Add vsubhn, vraddhn and vrsubhn tests. Split vaddhn.c into vXXXhn.inc and vaddhn.c to share code with other new tests.

2015-01-26  Christophe Lyon  <christophe.lyon@linaro.org>

* gcc.target/aarch64/advsimd-intrinsics/vXXXhn.inc: New file.
* gcc.target/aarch64/advsimd-intrinsics/vraddhn.c: New file.
* gcc.target/aarch64/advsimd-intrinsics/vrsubhn.c: New file.
* gcc.target/aarch64/advsimd-intrinsics/vsubhn.c: New file.
* gcc.target/aarch64/advsimd-intrinsics/vaddhn.c: Use code from
vXXXhn.inc.

From-SVN: r220117

[AARCH64]Fix TLS local exec model addressing code generation inconsistency.

gcc/

* config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Correct
the comment.
* config/aarch64/aarch64.md * (tlsle_small_<mode>): Add left shift
12-bit for higher part.

From-SVN: r220116

configure.ac: Move AM_ENABLE_MULTILIB before AC_PROG_CC.

2015-01-24  Matthias Klose  <doko@ubuntu.com>

        * configure.ac: Move AM_ENABLE_MULTILIB before AC_PROG_CC.
        * configure: Regenerate.

From-SVN: r220113

Move wrong ChangeLog entry from the toplevel ChangeLog to
libbacktrace.

From-SVN: r220112

re PR middle-end/64764 (internal compiler error: in is_value_included_in, at tree-ssa-uninit.c:942)

2015-01-26  Richard Biener  <rguenther@suse.de>

PR middle-end/64764
* tree-ssa-uninit.c (is_pred_expr_subset_of): Handle
combining two BIT_AND_EXPR predicates.

* gcc.dg/uninit-19.c: New testcase.

From-SVN: r220111

Make SPARC a vect64 target

* lib/target-supports.exp (check_effective_target_vect64): Add
sparc*-*-*.

From-SVN: r220109

Initialize ruid in new_var_info

PR bootstrap/64754
* tree-ssa-structalias.c (new_var_info): Initialize ruid.

From-SVN: r220108

2.cc: Add test for non-transparent comparison function.

* testsuite/23_containers/set/operations/2.cc: Add test for
non-transparent comparison function.

From-SVN: r220107

arm.c (arm_file_start): Update the assignment of Tag_ABI_HardFP_use.

2015-01-26  Terry Guo  <terry.guo@arm.com>

* config/arm/arm.c (arm_file_start): Update the assignment of
Tag_ABI_HardFP_use.

From-SVN: r220106

re PR testsuite/64712 (FAIL: gnat.dg/unchecked_convert1.adb execution test (x86_64/-m32))

PR testsuite/64712
* gnat.dg/unchecked_convert1.adb (Unchecked_Convert1): Initialize A.

From-SVN: r220104

[ARM] Wire up the new scheduler description for the ARM Cortex-A57  processor

* config/arm/arm-cores.def (cortex-a57): Use the new Cortex-A57
pipeline model.
config/arm/arm.md: Include the new Cortex-A57 model.
(generic_sched): Don't use generic_sched when tuning for
Cortex-A57.

From-SVN: r220103

scd42-1.c: Improve the check for UAL and non-UAL cases.

2015-01-26  Tony Liu  <tony.liu@arm.com>

        * gcc.target/arm/scd42-1.c: Improve the check for UAL and
        non-UAL cases.

From-SVN: r220102

Daily bump.

From-SVN: r220101

PR 64770 Make testcase work properly under DejaGNU.

2015-01-26  Janne Blomqvist  <jb@gcc.gnu.org>

PR libfortran/64770
* gfortran.dg/open_new_segv.f90: Add dg- stuff to make test work
correctly, clean up afterwards.

From-SVN: r220098

re PR fortran/62044 (ICE in USE statement with RENAME for extended derived type)

fortran/
PR fortran/62044
* decl.c (gfc_match_derived_decl): Don't insert a new symtree element.
* module.c (MOD_VERSION): Bump.
(write_module): Don't write list of extensions.
(read_module): Don't jump over list of extensions;
don't load list of extensions.
(load_derived_extensions, write_dt_extensions,
 write_derived_extensions): Remove.

testsuite/
PR fortran/62044
* gfortran.dg/use_rename_7.f90: New.

From-SVN: r220097

* config/i386/i386.c (get_builtin_code_for_version): Revert unwanted change.

From-SVN: r220096

i386.c (get_builtin_code_for_version): Add support for BMI and BMI2 multiversion functions.

gcc/ChangeLog:

* config/i386/i386.c (get_builtin_code_for_version): Add
support for BMI and BMI2 multiversion functions.
(fold_builtin_cpu): Add F_BMI and F_BMI2.

libgcc/ChangeLog:

* config/i386/cpuinfo.c (enum processor_features): Add FEATURE_BMI and
FEATURE_BMI2.
(get_available_features): Detect FEATURE_BMI and FEATURE_BMI2.

testsuite/ChangeLog:

* gcc.target/i386/funcspec-5.c: Test new multiversion targets.
* g++.dg/ext/mv17.C: Test BMI/BMI2 multiversion dispatcher.

Co-Authored-By: Uros Bizjak <ubizjak@gmail.com>
From-SVN: r220095

re PR libstdc++/29366 (atomics config for sh is weird)

libstdc++-v3/
PR target/29366
* config/cpu/sh/atomicity.h (__exchange_and_add, __atomic_add):
Remove SH4A inline asm and lock based implementations and use the
defaults from ext/atomicity.h.

From-SVN: r220094

re PR target/54236 ([SH] Improve addc and subc insn utilization)

gcc/testsuite/
PR target/54236
* gcc.target/sh/pr54236-4.c: New.

From-SVN: r220093

emit-rtl.h (store_bit_field): Move prototype to expmed.h.

2015-01-25  Prathamesh Kulkarni  <prathamesh.kulkarni@linaro.org>

        * emit-rtl.h (store_bit_field): Move prototype to expmed.h.
        (extract_bit_field): Likewise.
        (extract_low_bits): Likewise.
        (expand_mult): Likewise.
        (expand_mult_highpart_adjust): Likewise.

From-SVN: r220092

Make whitespace consistent in MAINTAINERS

2015-01-25  Tom de Vries  <tom@codesourcery.com>

* MAINTAINERS: Make whitespace consistent.

From-SVN: r220091

Support new Intel processor model numbers

gcc/

* config/i386/driver-i386.c (host_detect_local_cpu): Check new
Silvermont, Haswell, Broadwell and Knights Landing model numbers.
* config/i386/i386.c (processor_model): Add
M_INTEL_COREI7_BROADWELL.
(arch_names_table): Add "broadwell".

gcc/testsuite/

* gcc.target/i386/builtin_target.c (check_intel_cpu_model): Add
Silvermont, Ivy Bridge, Haswell and Broadwell tests.  Update Sandy
Bridge test.

2015-01-24  H.J. Lu  <hongjiu.lu@intel.com>

* config/i386/cpuinfo.c (processor_subtypes): Add
INTEL_COREI7_BROADWELL.
(get_intel_cpu): Support new Silvermont, Haswell and Broadwell
model numbers.

From-SVN: r220090

Daily bump.

From-SVN: r220089