make noop_move_p take a rtx_insn *

gcc/ChangeLog:

2015-05-08  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* rtlanal.c (noop_move_p): Change argument type to rtx_insn *.
* rtl.h: Adjust.

From-SVN: r222943

make add_shallow_copy_of_reg_note take an rtx_insn *

gcc/ChangeLog:

2015-05-08  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* rtlanal.c (add_shallow_copy_of_reg_note): Change argument type
to rtx_insn *.
* rtl.h: Adjust.

From-SVN: r222942

make remove_reg_equal_equiv_notes take an rtx_insn *

gcc/ChangeLog:

2015-05-08  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* rtlanal.c (remove_reg_equal_equiv_notes): Change argument type
to rtx_insn *.
* rtl.h: Likewise.

From-SVN: r222941

can_nonlocal_goto can take a rtx_insn *

gcc/ChangeLog:

2015-05-08  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* except.c (can_nonlocal_goto): Change type of argument to
rtx_insn *.
* rtl.h: Adjust.

From-SVN: r222940

computed_jump_p takes a rtx_insn *

gcc/ChangeLog:

2015-05-08  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* rtlanal.c (computed_jump_p): Cange argument type to rtx_insn *.
* rtl.h: Adjust.

From-SVN: r222939

change in_expr_list_p to in_insn_list_p

gcc/ChangeLog:

2015-05-08  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* rtlanal.c (in_insn_list_p): Renamed from in_expr_list_p.
* cfgrtl.c (can_delete_label_p): Adjust.
* rtl.h: likewise.

From-SVN: r222938

make stop_search_p take a rtx_insn *

gcc/ChangeLog:

2015-05-08  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* reorg.c (stop_search_p): Change argument to rtx_insn *.

From-SVN: r222937

make make_reg_eh_region_note and friend take rtx_insn *

gcc/ChangeLog:

2015-05-08  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* except.c (make_reg_eh_region_note): Change argument to
rtx_insn *.
(make_reg_eh_region_note_nothrow_nononlocal): Likewise.
* except.h: Adjust.

From-SVN: r222936

use rtx_insn * as the type of a local variable in mode-switching.c

gcc/ChangeLog:

2015-05-08  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* mode-switching.c (commit_mode_sets): Change type of local
variable from rtx to rtx_insn *.

From-SVN: r222935

decl2.c (mangling_aliases): New variable.

* decl2.c (mangling_aliases): New variable.
(note_mangling_alias, generate_mangling_aliases): New.
(cp_write_global_declarations): Call generate_mangling_aliases.
(generate_mangling_alias): Split out from...
* mangle.c (mangle_decl): ...here.
* cp-tree.h: Declare note_mangling_alias.

From-SVN: r222934

decl2.c (collect_candidates_for_java_method_aliases): Remove.

* decl2.c (collect_candidates_for_java_method_aliases): Remove.
(build_java_method_aliases): Adapt to use create_same_body_alias
instead of assemble_alias.  Move variable declarations to
definition and tidy up.
(cp_write_global_declarations): Call build_java_method_aliases
instead of collecting candidates first.

From-SVN: r222933

Daily bump.

From-SVN: r222931

install.texi (--enable-languages): Add missing jit and lto info.

* doc/install.texi (--enable-languages): Add missing jit and lto info.
Add ^ to grep command.
* doc/match-and-simplify.texi (GIMPLE API): Add missing fourth tree
arg to last gimple_simplify declaration.  Add missing gimple_build
declaration for built-in function case with four tree args.

From-SVN: r222923

libiberty/mkstemps.c: Include <time.h> if <sys/time.h> not available.

libiberty/ChangeLog:

        * mkstemps.c: #include <time.h> if HAVE_TIME_H is defined
        but not HAVE_SYS_TIME_H.

(fixes a build failure on LynxOS-178)

From-SVN: r222918

[PATCH 6/13] mips musl support

2015-05-08  Gregor Richards  <gregor.richards@uwaterloo.ca>
    Szabolcs Nagy  <szabolcs.nagy@arm.com>

* config/mips/linux.h (MUSL_DYNAMIC_LINKER32): Define.
(MUSL_DYNAMIC_LINKER64, MUSL_DYNAMIC_LINKERN32): Define.
(GNU_USER_DYNAMIC_LINKERN32): Update.

Co-Authored-By: Szabolcs Nagy <szabolcs.nagy@arm.com>
From-SVN: r222915

re PR tree-optimization/66036 (strided group loads are not vectorized)

2015-05-08  Richard Biener  <rguenther@suse.de>

        PR tree-optimization/66036
* tree-vect-data-refs.c (vect_compute_data_ref_alignment):
Handle strided group loads.
(vect_verify_datarefs_alignment): Likewise.
(vect_enhance_data_refs_alignment): Likewise.
(vect_analyze_group_access): Likewise.
(vect_analyze_data_ref_access): Likewise.
(vect_analyze_data_ref_accesses): Likewise.
* tree-vect-stmts.c (vect_model_load_cost): Likewise.
(vectorizable_load): Likewise.

* gcc.dg/vect/slp-41.c: New testcase.

From-SVN: r222914

trans-array.c (gfc_walk_elemental_function_args): Don't skip the advance to the next dummy argument when...

gcc/fortran/
* trans-array.c (gfc_walk_elemental_function_args):
Don't skip the advance to the next dummy argument when skipping
absent optional args.
gcc/testsuite/
* gfortran.dg/elemental_optional_args_7.f90: New.

From-SVN: r222913

rs6000.md: Require operand inequality in one of the peepholes.

* config/rs6000/rs6000.md: Require operand inequality in one
of the peepholes.

From-SVN: r222912

Franz Sirl <Franz.Sirl-kernel@lauterbach.com>

gcc/
2015-05-08  Richard Sandiford  <richard.sandiford@arm.com>
    Franz Sirl  <Franz.Sirl-kernel@lauterbach.com>

* config/i386/i386.md (<mode>_ldx, *<mode>_ldx): Remove mode
from (set ...).
* config/rx/rx.md (movdi, movdf): Likewise.
Likewise for define_peephole2s.

Co-Authored-By: Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
From-SVN: r222911

[AArch64] Idiomatic 64x1 comparisons in arm_neon.h

gcc/:

* config/aarch64/arm_neon.h (vceq_s64, vceq_u64, vceqz_s64, vceqz_u64,
vcge_s64, vcge_u64, vcgez_s64, vcgt_s64, vcgt_u64, vcgtz_s64, vcle_s64,
vcle_u64, vclez_s64, vclt_s64, vclt_u64, vcltz_s64, vtst_s64,
vtst_u64): Rewrite using gcc vector extensions.

gcc/testsuite/:

* gcc.target/aarch64/singleton_intrinsics_1.c: Generalize regex to
allow cmlt or sshr.

From-SVN: r222909

[AArch64] Add vcond(u?)didi pattern

* config/aarch64/aarch64-simd.md (aarch64_vcond_internal<mode><mode>,
vcond<mode><mode>, vcondu<mode><mode>): Add DImode variant.

From-SVN: r222908

optabs.c: Make vector_compare_rtx cope with VOID mode constants (e.g. const0_rtx)

* optabs.c (vector_compare_rtx): Handle RTL operands having VOIDmode.

From-SVN: r222907

[PATCH 0/13] Add musl support to GCC

2015-05-08  Szabolcs Nagy  <szabolcs.nagy@arm.com>

* config/glibc-stdint.h (OPTION_MUSL): Define.
(INT_FAST16_TYPE, INT_FAST32_TYPE, UINT_FAST16_TYPE, UINT_FAST32_TYPE):
Change the definition based on OPTION_MUSL for 64 bit targets.
* config/linux.h (OPTION_MUSL): Redefine.
* config/alpha/linux.h (OPTION_MUSL): Redefine.
* config/rs6000/linux.h (OPTION_MUSL): Redefine.
* config/rs6000/linux64.h (OPTION_MUSL): Redefine.

From-SVN: r222905

[PATCH 2/13] musl libc config

2015-05-08  Gregor Richards  <gregor.richards@uwaterloo.ca>
    Szabolcs Nagy  <szabolcs.nagy@arm.com>

* config.gcc (LIBC_MUSL): New tm_defines macro.
* config/linux.h (OPTION_MUSL): Define.
(MUSL_DYNAMIC_LINKER, MUSL_DYNAMIC_LINKER32,)
(MUSL_DYNAMIC_LINKER64, MUSL_DYNAMIC_LINKERX32,)
(INCLUDE_DEFAULTS_MUSL_GPP, INCLUDE_DEFAULTS_MUSL_LOCAL,)
(INCLUDE_DEFAULTS_MUSL_PREFIX, INCLUDE_DEFAULTS_MUSL_CROSS,)
(INCLUDE_DEFAULTS_MUSL_TOOL, INCLUDE_DEFAULTS_MUSL_NATIVE): Define.
* config/linux.opt (mmusl): New option.
* doc/invoke.texi (GNU/Linux Options): Document -mmusl.
* configure.ac (gcc_cv_libc_provides_ssp): Add *-*-musl*.
(gcc_cv_target_dl_iterate_phdr): Add *-linux-musl*.
* configure: Regenerate.

Co-Authored-By: Szabolcs Nagy <szabolcs.nagy@arm.com>
From-SVN: r222904

PR target/48904 x86_64-knetbsd-gnu missing defs

2015-05-08  H.J. Lu  <hongjiu.lu@intel.com>
            Bernhard Reutner-Fischer  <aldot@gcc.gnu.org>

        PR target/48904
        * config.gcc (x86_64-*-knetbsd*-gnu): Add i386/knetbsd-gnu64.h.
        * config/i386/knetbsd-gnu64.h: New file.

Co-Authored-By: Bernhard Reutner-Fischer <aldot@gcc.gnu.org>
From-SVN: r222903

re PR c++/59012 (alignas does not support parameter pack expansions)

PR c++/59012
* parser.c (cp_parser_std_attribute_list): Handle attribute expansion.
(cp_parser_std_attribute_spec): Handle alignas pack expansion.
* decl2.c (is_late_template_attribute): An attribute exp is dependent.
* pt.c (make_pack_expansion): Allow TREE_LIST for attribute expansion.
(apply_late_template_attributes): Handle attribute pack expansion.

From-SVN: r222902

Daily bump.

From-SVN: r222901

re PR c/64918 (invalid (?) warning when initializing structure)

PR c/64918
* c.opt (Woverride-init-side-effects): New option.

* c-typeck.c (add_pending_init): Use OPT_Woverride_init_side_effects.
(output_init_element): Likewise.

* doc/invoke.texi: Document -Woverride-init-side-effects.

* gcc.dg/Woverride-init-side-effects-1.c: New test.
* gcc.dg/Woverride-init-side-effects-2.c: New test.

From-SVN: r222894

re PR c/65179 (Introduce new C warning: -Wshift-negative-value)

PR c/65179
* c-common.c (c_fully_fold_internal): Warn when left shifting a
negative value.
* c.opt (Wshift-negative-value): New option.
* c-opts.c (c_common_post_options): Set warn_shift_negative_value
when -Wextra and C99/C++11 mode.

* c-typeck.c (build_binary_op): Warn when left shifting a negative
value.

* typeck.c (cp_build_binary_op): Warn when left shifting a negative
value.

* doc/invoke.texi: Document -Wshift-negative-value.

* c-c++-common/Wshift-negative-value-1.c: New test.
* testsuite/c-c++-common/Wshift-negative-value-2.c: New test.
* testsuite/c-c++-common/Wshift-negative-value-3.c: New test.
* testsuite/c-c++-common/Wshift-negative-value-4.c: New test.
* testsuite/c-c++-common/Wshift-negative-value-5.c: New test.
* testsuite/c-c++-common/Wshift-negative-value-6.c: New test.
* testsuite/gcc.dg/c90-left-shift-1.c: New test.
* testsuite/gcc.dg/c99-const-expr-7.c: Add dg-error.
* testsuite/gcc.dg/c99-left-shift-1.c: New test.

From-SVN: r222889

gcov-tool.c (do_merge): Refactore to remove int ret.

* gcov-tool.c (do_merge): Refactore to remove int ret.
* ipa-icf.c (sem_item::hash_referenced_symbol_properties): Change
!type == FUNC to type != FUNC.
* reload.h (struct target_reload): Changee to type of
x_spill_indirect_levels from bool to unsigned char.

From-SVN: r222884

rtl.h (always_void_p): New function.

gcc/
* rtl.h (always_void_p): New function.
* gengenrtl.c (always_void_p): Likewise.
(genmacro): Don't add a mode parameter to gen_rtx_foo if rtxes
with code foo are always VOIDmode.
* genemit.c (gen_exp): Update gen_rtx_foo calls accordingly.
* builtins.c, caller-save.c, calls.c, cfgexpand.c, combine.c,
compare-elim.c, config/aarch64/aarch64.c,
config/aarch64/aarch64.md, config/alpha/alpha.c,
config/alpha/alpha.md, config/arc/arc.c, config/arc/arc.md,
config/arm/arm-fixed.md, config/arm/arm.c, config/arm/arm.md,
config/arm/ldrdstrd.md, config/arm/thumb2.md, config/arm/vfp.md,
config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
config/c6x/c6x.md, config/cr16/cr16.c, config/cris/cris.c,
config/cris/cris.md, config/darwin.c, config/epiphany/epiphany.c,
config/epiphany/epiphany.md, config/fr30/fr30.c, config/frv/frv.c,
config/frv/frv.md, config/h8300/h8300.c, config/i386/i386.c,
config/i386/i386.md, config/i386/sse.md, config/ia64/ia64.c,
config/ia64/vect.md, config/iq2000/iq2000.c,
config/iq2000/iq2000.md, config/lm32/lm32.c, config/lm32/lm32.md,
config/m32c/m32c.c, config/m32r/m32r.c, config/m68k/m68k.c,
config/m68k/m68k.md, config/mcore/mcore.c, config/mcore/mcore.md,
config/mep/mep.c, config/microblaze/microblaze.c,
config/mips/mips.c, config/mips/mips.md, config/mmix/mmix.c,
config/mn10300/mn10300.c, config/msp430/msp430.c,
config/nds32/nds32-memory-manipulation.c, config/nds32/nds32.c,
config/nds32/nds32.md, config/nios2/nios2.c, config/nvptx/nvptx.c,
config/pa/pa.c, config/pa/pa.md, config/rl78/rl78.c,
config/rs6000/altivec.md, config/rs6000/rs6000.c,
config/rs6000/rs6000.md, config/rs6000/vector.md,
config/rs6000/vsx.md, config/rx/rx.c, config/rx/rx.md,
config/s390/s390.c, config/s390/s390.md, config/sh/sh.c,
config/sh/sh.md, config/sh/sh_treg_combine.cc,
config/sparc/sparc.c, config/sparc/sparc.md, config/spu/spu.c,
config/spu/spu.md, config/stormy16/stormy16.c,
config/tilegx/tilegx.c, config/tilegx/tilegx.md,
config/tilepro/tilepro.c, config/tilepro/tilepro.md,
config/v850/v850.c, config/v850/v850.md, config/vax/vax.c,
config/visium/visium.c, config/xtensa/xtensa.c, cprop.c, dse.c,
expr.c, gcse.c, ifcvt.c, ira.c, jump.c, lower-subreg.c,
lra-constraints.c, lra-eliminations.c, lra.c, postreload.c, ree.c,
reg-stack.c, reload.c, reload1.c, reorg.c, sel-sched.c,
var-tracking.c: Update calls accordingly.

From-SVN: r222883

DR 1467 PR c++/51747

DR 1467
PR c++/51747
* typeck2.c (digest_init_r): Fix single element list.

From-SVN: r222881

re PR middle-end/192 (String literals don't obey -fdata-sections)

PR middle-end/192
PR middle-end/54303
* varasm.c (function_mergeable_rodata_prefix): New function.
(mergeable_string_section): Use it.
(mergeable_constant_section): Use it.

gcc/testsuite/
* gcc.dg/fdata-sections-2.c: New file.

From-SVN: r222880

re PR testsuite/66046 (UBSan output pattern tests fail on target ARM board.)

PR testsuite/66046
* c-c++-common/ubsan/align-6.c: Don't match trailing newlines in the
last dg-output.
* c-c++-common/ubsan/align-7.c: Likewise.
* c-c++-common/ubsan/bounds-8.c: Likewise.
* c-c++-common/ubsan/float-cast-overflow-9.c: Likewise.
* c-c++-common/ubsan/load-bool-enum.c: Likewise.
* c-c++-common/ubsan/null-1.c: Likewise.
* c-c++-common/ubsan/null-10.c: Likewise.
* c-c++-common/ubsan/null-11.c: Likewise.
* c-c++-common/ubsan/null-2.c: Likewise.
* c-c++-common/ubsan/null-3.c: Likewise.
* c-c++-common/ubsan/null-4.c: Likewise.
* c-c++-common/ubsan/null-5.c: Likewise.
* c-c++-common/ubsan/null-6.c: Likewise.
* c-c++-common/ubsan/null-7.c: Likewise.
* c-c++-common/ubsan/null-8.c: Likewise.
* c-c++-common/ubsan/null-9.c: Likewise.
* c-c++-common/ubsan/object-size-1.c: Likewise.
* c-c++-common/ubsan/object-size-10.c: Likewise.
* c-c++-common/ubsan/object-size-4.c: Likewise.
* c-c++-common/ubsan/object-size-5.c: Likewise.
* c-c++-common/ubsan/object-size-7.c: Likewise.
* c-c++-common/ubsan/object-size-8.c: Likewise.
* c-c++-common/ubsan/object-size-9.c: Likewise.
* c-c++-common/ubsan/overflow-add-2.c: Likewise.
* c-c++-common/ubsan/overflow-int128.c: Likewise.
* c-c++-common/ubsan/overflow-mul-2.c: Likewise.
* c-c++-common/ubsan/overflow-mul-4.c: Likewise.
* c-c++-common/ubsan/overflow-negate-1.c: Likewise.
* c-c++-common/ubsan/overflow-negate-3.c: Likewise.
* c-c++-common/ubsan/overflow-sub-2.c: Likewise.
* c-c++-common/ubsan/overflow-sub-4.c: Likewise.
* c-c++-common/ubsan/pr59333.c: Likewise.
* c-c++-common/ubsan/pr59667.c: Likewise.
* c-c++-common/ubsan/pr60613-2.c: Likewise.
* c-c++-common/ubsan/pr60636.c: Likewise.
* c-c++-common/ubsan/pr63802.c: Likewise.
* c-c++-common/ubsan/recovery-1.c: Likewise.
* c-c++-common/ubsan/recovery-3.c: Likewise.
* c-c++-common/ubsan/shift-1.c: Likewise.
* c-c++-common/ubsan/shift-2.c: Likewise.
* c-c++-common/ubsan/shift-4.c: Likewise.
* c-c++-common/ubsan/shift-7.c: Likewise.
* c-c++-common/ubsan/undefined-2.c: Likewise.
* c-c++-common/ubsan/vla-1.c: Likewise.
* g++.dg/ubsan/null-1.C: Likewise.
* g++.dg/ubsan/null-3.C: Likewise.
* g++.dg/ubsan/null-4.C: Likewise.
* g++.dg/ubsan/vptr-8.C: Likewise.
* g++.dg/ubsan/vptr-9.C: Likewise.
* gcc.dg/ubsan/bounds-2.c: Likewise.
* gcc.dg/ubsan/object-size-9.c: Likewise.

From-SVN: r222878

match.pd (bit_and (plus/minus (convert @0) (convert @1) mask): New simplifier to narrow arithmetic.

* match.pd (bit_and (plus/minus (convert @0) (convert @1) mask): New
simplifier to narrow arithmetic.
* generic-match-head.c: (types_match, single_use): New functions.
* gimple-match-head.c: (types_match, single_use): New functions.

* gcc.dg/tree-ssa/shorten-1.c: New test.

From-SVN: r222877

combine.c (make_compound_operation): Remove checks for PLUS/MINUS rtx type.

2015-05-07  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>

        * combine.c (make_compound_operation): Remove checks for PLUS/MINUS
        rtx type.

From-SVN: r222874

re PR tree-optimization/66002 (paq8p benchmark 50% slower than clang on sandybridge)

2015-05-07  Richard Biener  <rguenther@suse.de>

PR tree-optimization/66002
* passes.def: Schedule another pass_merge_phi after ifcombine, right
before phiopt.

* gcc.dg/vect/vect-125.c: New testcase.

From-SVN: r222873

c-ubsan.c (ubsan_instrument_bounds): Don't skip instrumenting flexible member array-like members if...

* c-ubsan.c (ubsan_instrument_bounds): Don't skip instrumenting
flexible member array-like members if SANITIZE_BOUNDS_STRICT.

* doc/invoke.texi: Document -fsanitize=bounds-strict.
* flag-types.h (enum sanitize_code): Add SANITIZE_BOUNDS_STRICT, or it
into SANITIZE_NONDEFAULT.
* opts.c (common_handle_option): Handle -fsanitize=bounds-strict.

* c-c++-common/ubsan/bounds-10.c: New test.

From-SVN: r222871

re PR target/66015 (align directives not propagated after __attribute__ ((__optimize__ ("O2"))))

PR target/66015
* config/alpha/alpha.c (alpha_override_options_after_change): New.
(TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): New.
(alpha_override_options): Move align_loops, align_jumps and
align_functions handling into alpha_override_options_after_change.

From-SVN: r222870

Daily bump.

From-SVN: r222869

compiler: Analyze escape information of interface parameters.

When analyzing function calls with interface parameters, all arguments
related to an interface parameter were assumed to escape via the
function arguments.  Now, if an argument is related to a non-escaping
parameter with an interface type, the argument may be stack-allocated.

From-SVN: r222865

re PR fortran/37131 (inline matmul for small matrix sizes)

2015-05-06  Thomas Koenig  <tkoenig@gcc.gnu.org>

PR fortran/37131
* gfortran.h (gfc_isym_id):  Add GFC_ISYM_FE_RUNTIME_ERROR.
(gfc_intrinsic_sym):  Add vararg.
* intrinsic.h (gfc_check_fe_runtime_error):  Add prototype.
(gfc_resolve_re_runtime_error):  Likewise.
Add prototype for gfc_is_reallocatable_lhs.
* trans-array.h (gfc_is_reallocatable_lhs):  Remove prototype.
* check.c (gfc_check_fe_runtime_error):  New function.
* intrinsic.c (add_sym_1p):  New function.
(make_vararg):  New function.
(add_subroutines):  Add fe_runtime_error.
(gfc_intrinsic_sub_interface): Skip sorting for variable number
of arguments.
* iresolve.c (gfc_resolve_fe_runtime_error):  New function.
* lang.opt (inline-matmul-limit):  New option.
(gfc_post_options): If no inline matmul limit has been set and
BLAS is called externally, use the BLAS limit.
* frontend-passes.c:  Include intrinsic.h.
(var_num):  New global counter for naming temporary variablbles.
(matrix_case):  Enum for differentiating the different matmul
cases.
(realloc_string_callback):  Add "trim" to the variable name.
(create_var): Add optional argument vname as part of the name.
Use var_num. Set dimension of result correctly. Split off block
creation into
(insert_block): New function.
(cfe_expr_0): Use "fcn" as part of temporary variable name.
(optimize_namesapce): Also set gfc_current_ns. Call
inline_matmul_assign.
(combine_array_constructor):  Use "constr" as part of
temporary name.
(get_array_inq_function):  New function.
(build_logical_expr):  New function.
(get_operand):  new function.
(inline_limit_check):  New function.
(runtime_error_ne):  New function.
(matmul_lhs_realloc):  New function.
(is_functino_or_op):  New function.
(has_function_or_op):  New function.
(freeze_expr):  New function.
(freeze_references):  New function.
(convert_to_index_kind):  New function.
(create_do_loop):  New function.
(get_size_m1):  New function.
(scalarized_expr):  New function.
(inline_matmul_assign):  New function.
* simplify.c (simplify_bound):  Simplify the case of the
lower bound of an assumed-shape argument.

2015-05-06  Thomas Koenig  <tkoenig@gcc.gnu.org>

PR fortran/37131
* gfortran.dg/dependency_26.f90: Add option to suppress inlining
matmul.
* gfortran.dg/function_optimize_1.f90:  Likewise.
* gfortran.dg/function_optimize_2.f90:  Likewise.
* gfortran.dg/function_optimize_5.f90:  Likewise.
* gfortran.dg/function_optimize_7.f90:  Likewise.
* gfortran.dg/inline_matmul_1.f90:  New test.
* gfortran.dg/inline_matmul_2.f90:  New test.
* gfortran.dg/inline_matmul_3.f90:  New test.
* gfortran.dg/inline_matmul_4.f90:  New test.
* gfortran.dg/inline_matmul_5.f90:  New test.
* gfortran.dg/inline_matmul_6.f90:  New test.

From-SVN: r222864

jit: add test-benchmark.c

gcc/testsuite/ChangeLog:
* jit.dg/harness.h (set_options): Wrap with
#ifndef TEST_ESCHEWS_SET_OPTIONS.
* jit.dg/jit.exp (is_testcase_meant_to_generate_a_reproducer):
Special-case test-benchmark.c as a negative.
* jit.dg/test-benchmark.c: New file.

From-SVN: r222863

oacc-init.c (acc_device_lock): Add explanatory comment.

* oacc-init.c (acc_device_lock): Add explanatory comment.
(resolve_device): Add comment about locking requirement.
(acc_init_1, acc_shutdown_1): Likewise. Add locking around
gomp_init_device and gomp_fini_device calls.
(acc_get_num_devices, acc_set_device_type, acc_get_device_type)
(acc_get_device_num, acc_set_device_num): Add locking around
resolve_device and gomp_init_device calls.

From-SVN: r222862

oacc-init.c (acc_shutdown_1): Call gomp_mutex_unlock for goacc_thread_lock on error paths.

* oacc-init.c (acc_shutdown_1): Call gomp_mutex_unlock for
goacc_thread_lock on error paths.
* oacc-mem.c (lookup_host): Remove locking from function. Note
locking requirement for caller in function comment.
(lookup_dev): Likewise.
(acc_free, acc_deviceptr, acc_hostptr, acc_is_present)
(acc_map_data, acc_unmap_data, present_create_copy, delete_copyout)
(update_dev_host, gomp_acc_insert_pointer, gomp_acc_remove_pointer):
Add locking.

From-SVN: r222860

unknown-elf.h (STARTFILE_SPEC): Add conditional linking of crtfastmath.o.

2015-05-06  Sandra Loosemore  <sandra@codesourcery.com>
    Chris Jones  <chrisj@nvidia.com>
    Joshua Conner  <jconner@nvidia.com>

gcc/
* config/arm/unknown-elf.h (STARTFILE_SPEC): Add conditional
linking of crtfastmath.o.
* config/arm/linux-eabi.h (STARTFILE_SPEC): Likewise.

libgcc/
* config.host (arm*-*-linux*): Add support for crtfastmath.o.
(arm*-*-uclinux*): Likewise.
(arm*-*-eabi* | arm*-*-rtems*): Likewise.
* config/arm/crtfastmath.c: New file.

Co-Authored-By: Chris Jones <chrisj@nvidia.com>
Co-Authored-By: Joshua Conner <jconner@nvidia.com>
From-SVN: r222857

rs6000.md (cstore<mode>4_signed_imm): New expander.

* config/rs6000/rs6000.md (cstore<mode>4_signed_imm): New expander.
(cstore<mode>4_unsigned_imm): New expander.
(cstore<mode>4): Remove empty constraint strings.  Use the new
expanders.

From-SVN: r222855

re PR target/64208 ([iwmmxt] ICE: internal compiler error: Max. number of generated reload insns per insn is achieved (90))

gcc/
2015-05-06  Yvan Roux  <yvan.roux@linaro.org>

PR target/64208
* config/arm/iwmmxt.md ("*iwmmxt_arm_movdi"): Cleanup redundant
alternatives.

gcc/testsuite/
2015-05-06  Yvan Roux  <yvan.roux@linaro.org>

PR target/64208
* gcc.target/arm/pr64208.c: New test.

From-SVN: r222853

[AArch64] Fix geniterators.sh to use standard BRE syntax in sed

On behalf of Szabolcs Nagy.

2015-05-06  Szabolcs Nagy  <szabolcs.nagy@arm.com>

* config/aarch64/geniterators.sh: Use standard BRE in sed.

From-SVN: r222852

re PR target/66033 (rs6000 nops removed by rtl_dce)

PR target/66033
* config/rs6000/rs6000.md (nop): Use an unspec pattern.
(UNSPEC_NOP): Define.
(reload_vsx_from_gpr<mode>): Add missing DONE.
(reload_gpr_from_vsx<mode>): Likewise.
* config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
(vsx_div_v2di, vsx_udiv_v2di): Likewise.

From-SVN: r222851

re PR target/66020 (FAIL: gcc.target/powerpc/ppc64-abi-2.c execution test)

PR target/66020
* gcc.target/powerpc/ppc64-abi-2.c (my_mcount): Rewrite.
(gparms): Make volatile.

From-SVN: r222850

re PR fortran/62283 (basic-block vectorization fails)

2015-05-06  Richard Biener  <rguenther@suse.de>

PR tree-optimization/62283
* gcc.dg/vect/bb-slp-32.c: Remove XFAIL.

From-SVN: r222849

re PR target/66015 (align directives not propagated after __attribute__ ((__optimize__ ("O2"))))

2015-05-06  Christian Bruel  <christian.bruel@st.com>

PR target/66015
* config/aarch64/aarch64.c (aarch64_override_options): Move align_loops,
align_jumps, align_functions into aarch64_override_options_after_change.

From-SVN: r222847

tree-vect-slp.c (vect_supported_load_permutation_p): Use vect_transform_slp_perm_load to check if...

2015-05-06  Richard Biener  <rguenther@suse.de>

* tree-vect-slp.c (vect_supported_load_permutation_p): Use
vect_transform_slp_perm_load to check if we support a permutation
for basic-block vectorization.

* gcc.dg/vect/bb-slp-34.c: New testcase.

From-SVN: r222846

rl78.c (need_to_save): Save register 22 if it is used, even if it is not being used as a frame pointer.

* config/rl78/rl78.c (need_to_save): Save register 22 if it is
used, even if it is not being used as a frame pointer.

From-SVN: r222845

re PR fortran/62283 (basic-block vectorization fails)

2015-05-06  Richard Biener  <rguenther@suse.de>

PR tree-optimization/62283
* gcc.dg/vect/bb-slp-14.c: Adjust.

From-SVN: r222843

c.opt (Wterminate): New.

gcc/c-family/
* c.opt (Wterminate): New.
gcc/cp/
* cp-gimplify.c (cp_genericize_r): Track TRY_BLOCK and
MUST_NOT_THROW_EXPR, warn about a THROW_EXPR directly within a
MUST_NOT_THROW_EXPR.
(cp_genericize_data): Add try_block field.
(cp_genericize_tree): Initialize it.
* except.c (expand_end_catch_block): Set TREE_NO_WARNING on
implicit rethrow.

From-SVN: r222842

Fix obj-c++.dg/encode-10.mm with cxx_dialect == cxx11.

* constexpr.c (potential_constant_expression_1) [AT_ENCODE_EXPR]:
Return false.

From-SVN: r222841

Fix g++.dg/torture/Wsizeof-pointer-memaccess2.C with -std=c++11

Fix g++.dg/torture/Wsizeof-pointer-memaccess2.C with -std=c++11
* semantics.c (finish_call_expr): Check complain.

From-SVN: r222840

Fix gdb.cp/anon-struct.cc with -std=c++11.

gcc/
* dwarf2out.c (gen_member_die): Don't emit anything for an
anonymous class constructor.
gcc/cp/
* decl2.c (reset_type_linkage_2): Update the DECL_NAME of a
maybe-in-charge constructor.

From-SVN: r222839

decl.c (start_decl): Don't push the plain VAR_DECL for a variable template.

* decl.c (start_decl): Don't push the plain VAR_DECL for a
variable template.

From-SVN: r222837

DR 1518 DR 1630 PR c++/54835 PR c++/60417

DR 1518
DR 1630
PR c++/54835
PR c++/60417
* call.c (convert_like_real): Check value-initialization before
explicit.
* typeck2.c (process_init_constructor_record): Don't set
CONSTRUCTOR_IS_DIRECT_INIT.
(process_init_constructor_array): Likewise.
* init.c (build_vec_init): Likewise.

From-SVN: r222836

Daily bump.

From-SVN: r222834

* sv.po: Update.

From-SVN: r222829

compiler: Don't make temporaries for constant string concatentation.

Fixes golang/go#10642.

From-SVN: r222828

Fix indentation issues seen by -Wmisleading-indentation

gcc/ChangeLog:
* auto-profile.c (afdo_find_equiv_class): Fix indentation so
that it reflects the block structure.
(afdo_propagate_edge): Likewise.
(afdo_calculate_branch_prob): Likewise.
(afdo_annotate_cfg): Likewise.
* cfgcleanup.c (equal_different_set_p): Likewise.
(try_crossjump_to_edge): Likewise.
* cgraph.c (cgraph_node::verify_node): Likewise.
* cgraphunit.c (expand_all_functions): Likewise.
* config/i386/i386.c (ix86_expand_copysign): Likewise.
(exact_dependency_1): Likewise.
* dwarf2asm.c (dw2_output_indirect_constants): Likewise.
* dwarf2out.c (tree_add_const_value_attribute_for_decl): Likewise.
* gensupport.c (process_define_subst): Likewise.
* lto-wrapper.c (merge_and_complain): Likewise.
* tree-if-conv.c (if_convertible_bb_p): Likewise.
* tree-ssa-loop-prefetch.c (find_or_create_group): Likewise.
* tree-ssa-tail-merge.c (gsi_advance_fw_nondebug_nonlocal): Likewise.
* tree-vect-data-refs.c (vect_grouped_load_supported): Likewise.
* tree-vect-loop.c (vectorizable_reduction): Likewise.
* tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
* tree-vect-stmts.c (vectorizable_shift): Likewise.
* tree-vrp.c (vrp_finalize): Likewise.
* tree.c (variably_modified_type_p): Likewise.

gcc/cp/ChangeLog:
* parser.c (cp_parser_asm_definition): Only test for
        error_mark_node if "outputs" was just set.
(cp_parser_asm_definition): Likewise for "inputs".

gcc/fortran/ChangeLog:
* expr.c (check_inquiry): Fix indentation so that it reflects the
block structure.
* interface.c (compare_parameter): Likewise.
* parse.c (parse_oacc_structured_block): Likewise.
* target-memory.c (expr_to_char): Likewise.
* trans-types.c (gfc_init_kinds): Likewise.

libcpp/ChangeLog:
* pch.c (cpp_valid_state): Fix indentation so that it reflects the
block structure.

From-SVN: r222823

config.gcc: Use darwin9.h, darwin10.h and darwin12.h on darwin12 and later.

2015-05-05  Jack Howarth  <howarth.at.gcc@gmail.com>

* config.gcc: Use darwin9.h, darwin10.h and darwin12.h
on darwin12 and later.
* config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Add
file to pass -rdynamic on darwin12 and later.
* config/darwin.opt (rdynamic): Add.

From-SVN: r222822

extend.texi (C Extensions): Update menu for moved Variable Attributes and Type Attributes sections.

        * doc/extend.texi (C Extensions): Update menu for moved Variable
        Attributes and Type Attributes sections.

From-SVN: r222819

re PR target/65990 (ICE: in extract_insn, at recog.c:2341 (unrecognizable insn) with -mmemcpy-strategy=rep_8byte:-1:noalign -m32 -mtune=btver2)

PR target/65990
* config/i386/i386.c (ix86_parse_stringop_strategy_string): Error out
if rep_8byte stringop strategy was specified for 32-bit target.

testsuite/ChangeLog:

PR target/65990
* gcc.target/i386/pr65990.c: New test.

From-SVN: r222817

re PR go/66016 (Accessing nil Func's name results in crash)

PR go/66016
runtime: Don't crash in Func.Name if the Func is nil.

Related to Go issue 10696

From-SVN: r222816

re PR target/65915 (FAIL: gcc.target/i386/avx512f-vrndscalepd-2.c (internal compiler error))

PR target/65915
* config/i386/i386.md (vector convert to float spltiter): Check for
xmm16+, when splitting scalar float conversion.
* config/i386/sse.md (sse2_cvtsi2sd): Support EVEX version.

testsuite/ChangeLog:

PR target/65915
* gcc.target/i386/pr65915.c: New.

From-SVN: r222811

msp430-opts.h (enum msp430_regions): New.

* config/msp430/msp430-opts.h (enum msp430_regions): New.
* config/msp430/msp430.c (msp430_override_options): Complain if
-mcode-region or -mdata-region is used on a non MSP430X.
(msp430_section_attr): New function.  Checks lower, upper and
either attributes.
(msp430_attribute_table): Add lower, upper and either.
(gen_prefix): New function.  Generates a prefix for a section
name.
(msp430_select_section): New function - handles the choice of
section for an object.  Takes into account memory region
attributes and options.
(msp430_function_section): Use gen_prefix.
(TARGET_SECTION_TYPE_FLAGS): Define.
(msp430_section_type_flags): New function.
(TARGET_ASM_UNIQUE_SECTION): Define.
(msp430_unique_section): New function.
(msp430_output_aligned_decl_common): New function.
(msp430_do_not_relax_short_jumps): New function.
* config/msp430/msp430.h (USE_SELECT_SECTION_FOR_FUNCTIONS):
Define.
(ASM_OUTPUT_ALIGNED_DECL_COMMON): Define.
* config/msp430/msp430-protos.h
(msp430_do_not_relax_short_jumps): New prototype.
(msp430_output_aligned_decl_common): New prototype.
* config/msp430/msp430.md (length): New attribute.
(cbranchhi4_real): If msp430_do_not_relax_short_jumps is true
then use a long code sequence for short jumps.
* config/msp430/msp430.opt (mcode-region): New.
(mdata-region): New.
* doc/invoke.texi: Document new options.
* doc/extend.texi: Document new attributes.

From-SVN: r222810

libcpp: Improvements to comments in line-map.h/c

libcpp/ChangeLog:
* include/line-map.h: Fix comment at the top of the file.
(source_location): Rewrite and expand the comment for this
typedef, adding an ascii-art table to clarify how source_location
values are allocated.
* line-map.c: Fix comment at the top of the file.

From-SVN: r222806

[AArch64] Add branch-cost to cpu tuning information.

* gcc/config/aarch64-protos.h (struct cpu_branch_cost): New.
        (tune_params): Add field branch_costs.
(aarch64_branch_cost): Declare.
* gcc/config/aarch64.c (generic_branch_cost): New.
(generic_tunings): Set field cpu_branch_cost to generic_branch_cost.
(cortexa53_tunings): Likewise.
(cortexa57_tunings): Likewise.
(thunderx_tunings): Likewise.
(xgene1_tunings): Likewise.
(aarch64_branch_cost): Define.
* gcc/config/aarch64/aarch64.h (BRANCH_COST): Redefine.

From-SVN: r222805

i386.c: Use HOST_WIDE_INT_1 instead of (HOST_WIDE_INT) 1 and HOST_WIDE_INT_1U instead...

* config/i386/i386.c: Use HOST_WIDE_INT_1 instead of (HOST_WIDE_INT) 1
and HOST_WIDE_INT_1U instead of (unsigned HOST_WIDE_INT) 1.
* config/i386/i386.md: Ditto.
* config/i386/winnt.c: Ditto.

From-SVN: r222804

[docs] Update __atomic builtins documentation.

* doc/extend.texi (__atomic Builtins): Move implementation details
to the end of the description, rewrite opening paragraphs, state
difference with __sync builtins, state C11/C++11 assumptions,
weaken itemized descriptions, add explanation of memory model
behaviour, expand description of compare-exchange, simplify text.

From-SVN: r222803

Xfail gcc.dg/tree-ssa/stdarg-2.c f15 scans

2015-05-05  Tom de Vries  <tom@codesourcery.com>

* gcc.dg/tree-ssa/stdarg-2.c: Xfail f15 scans which test for presence of
'va_list escapes 0'.

From-SVN: r222802

Undo incorrect fix in tree-ssa/stdarg-2.c f15 scans

2015-05-05  Tom de Vries  <tom@codesourcery.com>

* gcc.dg/tree-ssa/stdarg-2.c: Undo incorrect fix in f15 scans committed
in r222173.

From-SVN: r222801

[PATCH][AARCH64]Use mov for add with large immediate.

gcc/

2015-05-05  Renlin Li  <renlin.li@arm.com>

* config/aarch64/aarch64.md (add<mode>3): Use mov when allowed.

From-SVN: r222800

[PR testsuite/65205, libgomp/65993] Fix dg-shouldfail usage in OpenACC libgomp tests

In dg-output, don't expect "0x" prefix for "%p" format specifier, don't expect
"(nil)" for NULL pointer.

PR testsuite/65205
PR libgomp/65993
libgomp/
* testsuite/libgomp.oacc-c-c++-common/clauses-2.c: In dg-output,
don't expect "0x" prefix for "%p" format specifier, don't expect
"(nil)" for NULL pointer.
* testsuite/libgomp.oacc-c-c++-common/lib-16.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-17.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-18.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-20.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-21.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-22.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-23.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-25.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-26.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-27.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-28.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-29.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-30.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-34.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-35.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-36.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-39.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-40.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-42.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-43.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-44.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-47.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-48.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-52.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-53.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-54.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-57.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/lib-58.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/data-already-1.c: More
accurately specify what we're looking for.
* testsuite/libgomp.oacc-c-c++-common/data-already-2.c: Likewise.
* testsuite/libgomp.oacc-c-c++-common/data-already-8.c: Likewise.
* testsuite/libgomp.oacc-fortran/data-already-1.f: Likewise.
* testsuite/libgomp.oacc-fortran/data-already-2.f: Likewise.
* testsuite/libgomp.oacc-fortran/data-already-8.f: Likewise.

From-SVN: r222799

pr65067.c: Require Thumb2 effective target.

gcc/testsuite/
2015-05-05  Yvan Roux  <yvan.roux@linaro.org>

* gcc.target/arm/pr65067.c: Require Thumb2 effective target.
* gcc.target/arm/pr65924.c: Likewise.

From-SVN: r222798

aarch64-elf-raw.h (CA53_ERR_843419_SPEC): Define.

2015-05-05  Yvan Roux  <yvan.roux@linaro.org>

* config/aarch64/aarch64-elf-raw.h (CA53_ERR_843419_SPEC): Define.
(LINK_SPEC): Include CA53_ERR_843419_SPEC.
* config/aarch64/aarch64-linux.h (CA53_ERR_843419_SPEC): Define.
(LINK_SPEC): Include CA53_ERR_843419_SPEC.
* config/aarch64/aarch64.opt (mfix-cortex-a53-843419): New option.
* configure: Regenerate.
* configure.ac: Add --enable-fix-cortex-a53-843419 option.
* doc/install.texi (aarch64*-*-*): Document new
--enable-fix-cortex-a53-843419 option.
* doc/invoke.texi (AArch64 Options): Document -mfix-cortex-a53-843419
and -mno-fix-cortex-a53-843419 options.

From-SVN: r222797

re PR target/65871 (bzhi builtin/intrinsic wrongly assumes bzhi instruction doesn't set the ZF flag)

PR target/65871
* config/i386/i386.md (*bmi_andn_<mode>_ccno): New pattern.

testsuite/ChangeLog:

PR target/65871
* gcc.target/i386/pr65871-3.c: New test.

From-SVN: r222795

Improve the test in bitfields.m4

Using a named bitfield with a width more than 0 means we won't hit
weirdness caused by the bitfield not really needing to exist.  Changing
int to long long means we won't have trouble with some arch where size
of int is 1 or 2.

libobjc/ChangeLog:

2015-05-04  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* configure: Regenerate.

config/ChangeLog:

2015-05-04  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* bitfields.m4: Change int to long long, and use bitfields of
width 1 instead of 0.

From-SVN: r222794

re PR libstdc++/51617 ([C++0x] async(f) isn't.)

PR libstdc++/51617
* include/std/future (async): Change default policy to launch::async.

From-SVN: r222793

tree.c (verify_type): Check various uses of TYPE_MAXVAL...

* tree.c (verify_type): Check various uses of TYPE_MAXVAL;
fix overactive TYPE_MIN_VALUE check and add FIXME for type
compatibility problems.

From-SVN: r222792

The changes are made in the patch for optimized usage of pcmpne/pcmpeq instructions.

The changes are made in the patch for optimized usage of pcmpne/pcmpeq
instructions. The xor with register to register is replaced with pcmpeq
/pcmpne instructions and for immediate check still the xori will be used.
The purpose of the change is to acheive the aggressive usage of pcmpne
/pcmpeq instructions instead of xor being used for comparison.

ChangeLog:
2015-05-04  Ajit Agarwal  <ajitkum@xilinx.com>

* config/microblaze/microblaze.md (cbranchsi4): Added immediate
constraints.
(cbranchsi4_reg): New.
* config/microblaze/microblaze.c
(microblaze_expand_conditional_branch_reg): New.
* config/microblaze/microblaze-protos.h
(microblaze_expand_conditional_branch_reg): New prototype.

From-SVN: r222791

The changes are made in the patch for optimized usage of fint instruction.

The sequence of fint/cond_branch is replaced with fcmp/cond_branch. The
fint instruction takes 6/7 cycles as compared to fcmp instruction which
takes 1 cycles. The conversion from float to int with fint instruction
is not required and can directly compared with fcmp instruction which
takes 1 cycle as compared to 6/7 cycles with fint instruction.

ChangeLog:
2015-03-04  Ajit Agarwal  <ajitkum@xilinx.com>

* config/microblaze/microblaze.md (peephole2): New.

From-SVN: r222790

Daily bump.

From-SVN: r222788

Revert:

2015-05-04  Jeff Law  <law@redhat.com>

        * match.pd (bit_and (plus/minus (convert @0) (convert @1) mask): New
        simplifier to narrow arithmetic.
        * generic-match-head.c: (types_match, single_use): New functions.
        * gimple-match-head.c: (types_match, single_use): New functions.

Revert:
2015-05-04  Jeff Law  <law@redhat.com>
* gcc.dg/tree-ssa/shorten-1.c: New test.

From-SVN: r222784

re PR target/65987 ([SH] Wrong jump generated for gcc.dg/tree-prof/va-arg-pack-1.c with -fprofile-use)

PR target/65987
* config/sh/sh.c (output_far_jump, split_branches): Take into account
  sectiion crossing jumps.

From-SVN: r222783

re PR c++/66007 (Narrowing conversion inside { } results in all zero elements in C++11 mode with -Wno-error=narrowing)

/cp
2015-05-04  Paolo Carlini  <paolo.carlini@oracle.com>
    Jakub Jelinek  <jakub@redhat.com>

PR c++/66007
* typeck2.c (check_narrowing): Check by-hand that the pedwarn didn't
result in an actual error.

/testsuite
2015-05-04  Paolo Carlini  <paolo.carlini@oracle.com>
    Jakub Jelinek  <jakub@redhat.com>

PR c++/66007
* g++.dg/cpp0x/Wnarrowing4.C: New.

Co-Authored-By: Jakub Jelinek <jakub@redhat.com>
From-SVN: r222778

common.opt (fdelete-null-pointer-checks): Init to -1.

2015-05-04  Sandra Loosemore  <sandra@codesourcery.com>

gcc/
* common.opt (fdelete-null-pointer-checks): Init to -1.
* config/nios2/elf.h (SUBTARGET_OVERRIDE_OPTIONS): Define to
override flag_delete_null_pointer_checks default.
* doc/invoke.texi (-fdelete-null-pointer-checks): Clarify
behavior re address zero.  Better document target-specific behavior.
(-fisolate-errneous-paths-dereference): Mention relationship to
-fdelete-null-pointer-checks.

gcc/testsuite/
* lib/target-supports.exp
(check_effective_target_keeps_null_pointer_checks): Clarify that
this is for targets that disable -fdelete-null-pointer-checks,
not default it to off.
* gcc.dg/ipa/ipa-pta-14.c: Make dependence on
-fdelete-null-pointer-checks explicit.
* gcc.dg/tree-ssa/20030730-1.c: Likewise.
* gcc.dg/tree-ssa/20030730-2.c: Likewise.
* gcc.dg/tree-ssa/isolate-1.c: Likewise.
* gcc.dg/tree-ssa/isolate-2.c: Likewise.
* gcc.dg/tree-ssa/isolate-3.c: Likewise.
* gcc.dg/tree-ssa/isolate-4.c: Likewise.
* gcc.dg/tree-ssa/isolate-5.c: Likewise.
* gcc.dg/tree-ssa/nonzero-1.c: Likewise.
* gcc.dg/tree-ssa/pr20318.c: Likewise.
* gcc.dg/tree-ssa/pr20701.c: Likewise.
* gcc.dg/tree-ssa/pr20702.c: Likewise.
* gcc.dg/tree-ssa/pr21086.c: Likewise.
* gcc.dg/tree-ssa/pr21090.c: Likewise.
* gcc.dg/tree-ssa/pr58480.c: Likewise.
* gcc.dg/tree-ssa/pta-escape-1.c: Likewise.
* gcc.dg/tree-ssa/pta-escape-2.c: Likewise.
* gcc.dg/tree-ssa/pta-escape-3.c: Likewise.
* gcc.dg/tree-ssa/ssa-vrp-thread-1.c: Likewise.
* gcc.dg/tree-ssa/unreachable.c: Likewise.
* gcc.dg/tree-ssa/vrp02.c: Likewise.
* gcc.dg/tree-ssa/vrp07.c: Likewise.
* gcc.dg/tree-ssa/vrp08.c: Likewise.
* gcc.dg/tree-ssa/vrp55.c: Likewise.
* g++.dg/cpp0x/static_assert9.C: Likewise.
* g++.dg/tree-ssa/nonzero-1.C: Likewise.
* g++.dg/tree-ssa/pr19476-1.C: Likewise.
* g++.dg/tree-ssa/pr19476-2.C: Likewise.
* g++.dg/tree-ssa/pr19476-5.C: Likewise.
* g++.dg/tree-ssa/pr26406.C: Likewise.

From-SVN: r222777

re PR middle-end/65984 (ICE: definition in block 4 does not dominate use in block 2 with -fnon-call-exceptions -fsanitize=enum)

PR tree-optimization/65984
* ubsan.c: Include tree-cfg.h.
(instrument_bool_enum_load): Use stmt_ends_bb_p instead of
stmt_could_throw_p test, rename can_throw variable to ends_bb.

* c-c++-common/ubsan/pr65984.c: New test.

From-SVN: r222775

i386.c: Change GET_CODE (...) == CONST_DOUBLE check to CONST_DOUBLE_P predicate.

* config/i386/i386.c: Change GET_CODE (...) == CONST_DOUBLE check
to CONST_DOUBLE_P predicate.
(standard_sse_constant_p): Return 0 for !TARGET_SSE.
(ix86_legitimate_constant_p) <case CONST_WIDE_INT>: For 32bit targets,
allow only operands that satisfy standard_sse_constant_p predicate.
* config/i386/i386.md: Change GET_CODE (...) == CONST_DOUBLE check
to CONST_DOUBLE_P predicate.

From-SVN: r222774

match.pd (bit_and (plus/minus (convert @0) (convert @1) mask): New simplifier to narrow arithmetic.

* match.pd (bit_and (plus/minus (convert @0) (convert @1) mask): New
simplifier to narrow arithmetic.
* generic-match-head.c: (types_match, single_use): New functions.
* gimple-match-head.c: (types_match, single_use): New functions.

* gcc.dg/tree-ssa/shorten-1.c: New test.

From-SVN: r222771

arm.c: Restore bootstrap.

2015-05-04  Andreas Tobler  <andreast@gcc.gnu.org>

    * config/arm/arm.c: Restore bootstrap.

From-SVN: r222768

i386.h (TARGET_SUPPORTS_WIDE_INT): New define.

* config/i386/i386.h (TARGET_SUPPORTS_WIDE_INT): New define.
* config/i386/i386.c (ix86_legitimate_constant_p): Handle TImode
as CONST_WIDE_INT, not CONST_DOUBLE.
(ix86_cannot_force_const_mem): Handle CONST_WIDE_INT.
(output_pic_addr_const): Do not handle VOIDmode CONST_DOUBLEs.
(ix86_find_base_term): Do not check for CONST_DOUBLE.
(ix86_print_operand): Do not handle non-FPmode CONST_DOUBLEs.
(ix86_build_signbit_mask): Rewrite using wide ints.
(ix86_split_to_parts) [HOST_BITS_PER_WIDE_INT < 64]: Remove.
(ix86_rtx_costs): Handle CONST_WIDE_INT.
(find_constant): Ditto.
* config/i386/i386.md (bts, btr, btc peepholes): Rewrite
using gen_int_mode.
* config/i386/predicates.md (x86_64_immediate_operand)
<case CONST_INT>: Remove HOST_BITS_PER_WIDE_INT == 32 code.
(x86_64_zext_immediate_operand): Remove CONST_DOUBLE handling.
<case CONST_INT>: Remove HOST_BITS_PER_WIDE_INT == 32 code.
(const0_operand): Also match const_wide_int.
(constm1_operand): Ditto.
(const1_operand): Ditto.

From-SVN: r222767

re PR fortran/44735 (ICE on FORALL with character array pointer)

PR fortran/44735
* gfortran.dg/pr44735.f90: New test.

From-SVN: r222766

re PR middle-end/65965 (Straight-line memcpy/memset not vectorized when equivalent loop is)

2015-05-04  Richard Biener  <rguenther@suse.de>

PR tree-optimization/65965
* tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Split
store groups at gaps.

* gcc.dg/vect/bb-slp-33.c: New testcase.

From-SVN: r222765