re PR rtl-optimization/80474 (ipa-cp wrongly adding LO(symbol) twice)

PR rtl-optimization/80474
* reorg.c (update_block): Do not ignore instructions in a delay slot.

From-SVN: r249219

Use pthread effective-target in testsuite

* testsuite: Add dg-require-effective-target pthread to -pthread
tests.
Remove explicit target lists from dg-do and dg-options.

* testsuite/30_threads/async/forced_unwind.cc: Remove explit
target list from dg-options.
* testsuite/30_threads/packaged_task/forced_unwind.cc: Likewise.

* 30_threads/shared_mutex/cons/1.cc: Likewise.
Pass -pthread for all targets.
* 30_threads/shared_mutex/try_lock/1.cc: Likewise.
* 30_threads/shared_mutex/try_lock/2.cc: Likewise.
* 30_threads/shared_mutex/unlock/1.cc: Likewise.

* testsuite/30_threads/this_thread/57060.cc: Require c++11 via
dg-require-effective-target.

From-SVN: r249217

[ARM] Make gcc.target/arm/its.c more robust

2017-06-15  Thomas Preud'homme  <thomas.preudhomme@arm.com>

    gcc/testsuite/
    * gcc.target/arm/its.c: Check that no IT blocks has more than 2
    instructions in it rather than the number of IT blocks being 2.
    Transfer scan directive arm_thumb2 restriction to the whole
    testcase and restrict further to Thumb-only targets.

From-SVN: r249215

sdiv_costs_1.c: Use dg-add-options.

2017-06-15  Tamar Christina  <tamar.christina@arm.com>

* gcc.target/arm/sdiv_costs_1.c:
Use dg-add-options.

From-SVN: r249214

rs6000: Don't use REGNO in the add<mode>3 expander

The argument could be a subreg of reg instead, which means we have to
use reg_or_subregno instead of REGNO.

* config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
of REGNO.

From-SVN: r249213

Daily bump.

From-SVN: r249212

compiler: avoid crash on erroneous type

    If there is an error constructing the backend type, the GCC backend
    will report that the size is 1.  That will then cause construction of
    the ptrmask to crash.  Avoid that case by just generating an empty
    ptrmask.

    Noticed while compiling a broken package.  The policy I've been
    following is to not commit a test case for a compiler crash on invalid
    code, so no test case.

    Reviewed-on: https://go-review.googlesource.com/45775

From-SVN: r249208

MIPS16/GCC: Emit bounds checking as RTL in `casesi'

gcc/
* config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
(casesi): Emit bounds checking as RTL.
(casesi_internal_mips16_<mode>): Remove bounds checking.

gcc/testsuite/
* gcc.target/mips/data-sym-jump.c: Adjust for whitespace changes.
* gcc.target/mips/pr51513-1.c: New test.
* gcc.target/mips/pr51513-2.c: New test.

From-SVN: r249207

Fix libstdc++ testsuite failures with -D_GLIBCXX_USE_CXX11_ABI=0

* doc/xml/manual/test.xml: Correct instructions on running tests.
* testsuite/21_strings/basic_string/cons/char/deduction.cc: Adjust to
pass when -D_GLIBCXX_USE_CXX11_ABI=0 added to RUNTESTFLAGS.
* testsuite/21_strings/basic_string/cons/wchar_t/deduction.cc:
Likewise.
* testsuite/27_io/basic_ios/copyfmt/char/1.cc: Likewise.
* testsuite/27_io/basic_ios/exceptions/char/1.cc: Likewise.
* testsuite/27_io/basic_istream/extractors_arithmetic/char/
exceptions_failbit.cc: Likewise.
* testsuite/27_io/basic_istream/extractors_arithmetic/wchar_t/
exceptions_failbit.cc: Likewise.
* testsuite/27_io/basic_istream/extractors_other/char/
exceptions_null.cc: Likewise.
* testsuite/27_io/basic_istream/extractors_other/wchar_t/
exceptions_null.cc: Likewise.
* testsuite/27_io/basic_istream/sentry/char/12297.cc: Likewise.
* testsuite/27_io/basic_istream/sentry/wchar_t/12297.cc: Likewise.
* testsuite/27_io/basic_ostream/inserters_other/char/
exceptions_null.cc: Likewise.
* testsuite/27_io/basic_ostream/inserters_other/wchar_t/
exceptions_null.cc: Likewise.
* testsuite/27_io/ios_base/storage/2.cc: Likewise.

From-SVN: r249206

Makefile.def: Add check-gotools to go check targets.

* Makefile.def: Add check-gotools to go check targets.
* Makefile.in: Rebuild.

From-SVN: r249204

Makefile.am (libgosrcdir): Define.

* Makefile.am (libgosrcdir): Define.
(check-head, check-gccgo, check-go-tool): New targets.
(CHECK_ENV): Define.
(check): New target.
(mostlyclean-local): New target.
* Makefile.in: Rebuild.

From-SVN: r249203

gcc: xtensa: allow XCHAL_* macros to be non-constant

XCHAL_* macros from the xtensa-config.h are used in a number of places
that require them to be preprocessor constants. Rewrite these places so
that non-constant XCHAL_* definitions could be used there.

2017-06-14  Max Filippov  <jcmvbkbc@gmail.com>
gcc/
* config/xtensa/xtensa.c (xtensa_option_override): Append
MASK_CONST16 to target_flags in the absence of TARGET_L32R.
(hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
 xtensa_doloop_hooks): Define unconditionally.
(xtensa_reorg_loops): Only call reorg_loops in the presence of
TARGET_LOOPS.
* config/xtensa/xtensa.h (TARGET_L32R): New definition.
(TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
for it in xtensa_option_override.
(HARD_FRAME_POINTER_IS_FRAME_POINTER,
 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.

From-SVN: r249202

Recognize '-' as special -MF argument (write to stdout)

Sometimes it is useful to generate pre-processed output to a file and
the dependency information to stdout for further analysis/processing.
For example:

g++ -E -MD -fdirectives-only -o test.ii test.cxx

This will generate the dependency information to test.d (as per the
documentation). While changing this behavior is probably unwise, one
traditional (e.g., supported by -o) way to handle this is to recognize
the special '-' file name as an instruction to write to stdout:

g++ -E -MD -fdirectives-only -o test.ii -MF - test.cxx

Currently this will create a file named '-'. The included patch
changes this behavior to write to stdout.

Note also that Clang has supported this from at least version 3.5.

gcc:
2017-06-14  Boris Kolpackov  <boris@codesynthesis.com>

* doc/cppopts.texi: Document '-' special value to -MF.

gcc/c-family:
2017-06-14  Boris Kolpackov  <boris@codesynthesis.com>

* c-opts.c (c_common_finish): Handle '-' special value to -MF.

From-SVN: r249201

Improve Cortex-A53 FP scheduler

The Cortex-A53 scheduler model of FMAC bypass is not quite right
for FMAC to FMAC forwarding.  Experiments also show the latencies of
FP operations are too high as well.  Rather than adding more bypasses,
adjust the latencies of FP instructions to get a better schedule on
average.  As a result SPECFP2006 is 1.1% faster.

    gcc/
* config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
(cortex_a53_fconst): Likewise.
(cortex_a53_fpmul): Likewise.
(cortex_a53_f_load_64): Likewise.
(cortex_a53_f_load_many): Likewise.
(cortex_a53_advsimd_alu): Likewise.
(cortex_a53_advsimd_alu_q): Likewise.
(cortex_a53_advsimd_mul): Likewise.
(cortex_a53_advsimd_mul_q): Likewise.
(fpmac bypass): Add new bypass for fpmac-fpmac case.
Add missing fmul, r2f_cvt and fconst cases.

From-SVN: r249200

cmd/cgo: make _cgo_flags consistent across runs

    The go tool will pass -I objdir as one of the flags, where objdir is
    the temporary build directory. Remove that from _cgo_flags: we don't
    need it, and it will be different each time.

    Sort the flags to avoid the unpredictable map iteration order.

    This matters for gccgo because for a package that uses cgo, the go
    tool when building for gccgo will store the _cgo_flags file in the
    archive. That means that we want to generate identical _cgo_flags for
    every run.

    The test for this is the cmd/go testsuite, to follow in a future CL.

    Reviewed-on: https://go-review.googlesource.com/45692

From-SVN: r249199

cmd/go: gccgo: consistent results, fix vendoring

    Pass the -fdebug-prefix-map and -gno-record-gcc-switches compiler
    options to gccgo to generate consistent results.

    Fix the vendoring code to look for /vendor/, not just /vendor, to
    avoid being confused by something like vendor/vendor.org.

    Tested by the cmd/go tests in a followup CL.

    Reviewed-on: https://go-review.googlesource.com/45695

From-SVN: r249198

cmd/go: skip tests that do not work with gccgo

    These tests fail for various reasons, most commonly because gccgo
    doesn't really have GOROOT, so things like `go build errors` fail.

    Reviewed-on: https://go-review.googlesource.com/45696

From-SVN: r249197

go/build: use GCCGOTOOLDIR to find gccgo tools

    Add the environment variable GCCGOTOOLDIR to permit overriding the default
    directory where tools like cgo are found when building with gccgo.

    This will be used by the cmd/go tests in a future CL.

    Reviewed-on: https://go-review.googlesource.com/45694

From-SVN: r249196

internal/testenv: support testing gccgo

    If GO_TESTING_GOTOOLS is set in the environment, permit tests using
    gccgo to run the go tool. Like GO_BUILDER_NAME, this should not be set
    normally. But it is needed when testing the go tool itself, and will
    be set by the gotools Makefile in a future CL.

    Reviewed-on: https://go-review.googlesource.com/45693

From-SVN: r249195

compiler: remove "DIR/../" when joining relative import path

    Otherwise if DIR does not exist, the path does not work. This matches
    what the gc cmd/compile tool does, because it calls path.Join.

    The test for this is the cmd/go tests, to be added in a follow-up CL.

    Reviewed-on: https://go-review.googlesource.com/45691

From-SVN: r249194

re PR tree-optimization/81083 (ICE: Unable to coalesce ssa_names 4 and 13 which are marked as MUST COALESCE)

2017-06-14  Richard Biener  <rguenther@suse.de>

PR tree-optimization/81083
* gcc.dg/torture/pr81083.c: Add prototypes.

From-SVN: r249193

re PR middle-end/81088 (UBSAN: false positive as a result of reassosiation)

2017-06-14  Richard Biener  <rguenther@suse.de>

PR middle-end/81088
* fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
literal constants.
(fold_binary_loc): When associating do not treat pre-existing
TREE_OVERFLOW on literal constants as a reason to allow
TREE_OVERFLOW on associated literal constants.

* c-c++-common/ubsan/pr81088.c: New testcase.

From-SVN: r249192

cp-gimplify.c (cp_genericize_r): Turn most of the function into a switch (TREE_CODE (stmt)) statement from long...

* cp-gimplify.c (cp_genericize_r): Turn most of the function
into a switch (TREE_CODE (stmt)) statement from long else if
sequence.

From-SVN: r249191

sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.

* config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
(MASK_FEATURES): New macro.
* config/sparc/sparc.c (sparc_option_override): Remove the special
handling of -mfpu and generalize it to all MASK_FEATURES switches.

From-SVN: r249189

simplify-rtx.c (simplify_binary_operation_1): Do not simplify a division of 0 if non-call exceptions are enabled.

* simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
a division of 0 if non-call exceptions are enabled.

From-SVN: r249188

re PR target/71663 (aarch64 Vector initialization can be improved slightly)

PR target/71663
gcc
* config/aarch64/aarch64.c (aarch64_expand_vector_init):
Improve vector initialization code gen for only variable case.

gcc/testsuite
* gcc.target/aarch64/vect-init-1.c: Newtestcase.
* gcc.target/aarch64/vect-init-2.c: Likewise.
* gcc.target/aarch64/vect-init-3.c: Likewise.
* gcc.target/aarch64/vect-init-4.c: Likewise.
* gcc.target/aarch64/vect-init-5.c: Likewise.

Co-Authored-By: Naveen H.S <Naveen.Hurugalawadi@cavium.com>
From-SVN: r249187

re PR c++/58541 ([c++11] Bogus "error: redeclaration ... differs in ‘constexpr’")

2017-06-14  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/58541
* g++.dg/cpp0x/constexpr-58541.C: New.

From-SVN: r249186

* config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.

From-SVN: r249183

re PR tree-optimization/81083 (ICE: Unable to coalesce ssa_names 4 and 13 which are marked as MUST COALESCE)

2017-06-14  Richard Biener  <rguenther@suse.de>

PR tree-optimization/81083
* tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
as values.

* gcc.dg/torture/pr81083.c: New testcase.

From-SVN: r249182

vrp101.c: Fix scan-tree-dump regex.

* gcc.dg/tree-ssa/vrp101.c: Fix scan-tree-dump regex.
* gcc.dg/tree-ssa/builtin-sprintf-2.c: Likewise.  Adjust comment.

From-SVN: r249181

vrp101.c: Fix scan-tree-dump regex.

* gcc.dg/tree-ssa/vrp101.c: Fix scan-tree-dump regex.
* gcc.dg/tree-ssa/builtin-sprintf-2.c: Likewise.  Adjust comment.

From-SVN: r249180

Daily bump.

From-SVN: r249179

re PR sanitizer/80973 (ICE with lambda and -fsanitize=undefined)

PR c++/80973
* cp-gimplify.c (cp_genericize_r): Don't instrument MEM_REF second
argument even if it has REFERENCE_TYPE.

* g++.dg/ubsan/pr80973.C: New test.

From-SVN: r249174

re PR c++/80984 (ICE with label/variable ambiguity)

PR c++/80984
* cp-gimplify.c (cp_genericize): Only look for VAR_DECLs in
BLOCK_VARS (outer) chain.
(cxx_omp_const_qual_no_mutable): Likewise.

* g++.dg/opt/nrv18.C: New test.

From-SVN: r249173

re PR go/80964 (cross-gotools are not executable on host system)

PR go/80964
* configure.ac: Set NATIVE if host_alias = target_alias.
* configure: Rebuild.

From-SVN: r249172

re PR objc/80949 (ICE in do_warn_duplicated_branches_r)

PR objc/80949
* c-warn.c (do_warn_duplicated_branches): Return if any of the
branches is null.

From-SVN: r249171

rs6000: Comment fixes + some leftovers

* config/rs6000/rs6000.c: Update all comments that mentioned SPE.
(rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
* config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
* config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete.  Adjust former use.
* config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
* config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.

From-SVN: r249170

rs6000: Remove VECTOR_SPE

* config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
* config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.

From-SVN: r249169

rs6000: Remove FIXED_SCRATCH

* config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.

From-SVN: r249168

rs6000: Updates to t-rtems

* config/rs6000/t-rtems: Don't handle SPE.

From-SVN: r249167

rs6000: Updates to t-linux

* config/rs6000/t-linux: Don't handle SPE.

From-SVN: r249166

rs6000: Remove eabispe.h

* config/rs6000/eabispe.h: Delete file.

From-SVN: r249165

rs6000: Remove t-spe

* config/rs6000/t-spe: Delete file.

From-SVN: r249164

rs6000: Remove SPE_CONST_OFFSET_OK

* config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
(rs6000_legitimate_offset_address_p): Return false for anything in
V2SImode or V2SFmode.

From-SVN: r249163

rs6000: Sanitize vector modes

This removes the vector modes that were only used by SPE.  It also
rearranges things so it is easier to see what is there, and for what.

* config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
except V2SF and V2SI.  Rearrange the vector modes, and add comments.
* config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
and V4HImode.
(reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
(rs6000_legitimate_offset_address_p): Ditto.
(rs6000_emit_move): Ditto.
(rs6000_init_builtins): Remove V4HI_type_node.

From-SVN: r249162

compiler: containing small bfunction mixup in Gogo::write_globals

    Fix buglet in Gogo::write_globals-- in a couple of places the
    wrong Bfunction was being used for the containing (not target)
    function when creating calls for init functions.

    Reviewed-on: https://go-review.googlesource.com/45510

From-SVN: r249159

Implement no_sanitize function attribute

2017-06-13  Martin Liska  <mliska@suse.cz>

PR sanitize/78204
* c-c++-common/ubsan/attrib-2.c (float_cast2): Enhance the
test by adding no_sanitize attribute.
* gcc.dg/asan/use-after-scope-4.c: Likewise.
2017-06-13  Martin Liska  <mliska@suse.cz>

PR sanitize/78204
* c-attribs.c (add_no_sanitize_value): New function.
(handle_no_sanitize_attribute): Likewise.
(handle_no_sanitize_address_attribute): Use the function.
(handle_no_sanitize_thread_attribute): New function.
(handle_no_address_safety_analysis_attribute): Use
add_no_sanitize_value.
(handle_no_sanitize_undefined_attribute): Likewise.
* c-common.h: Declare new functions.
* c-ubsan.c (ubsan_instrument_division): Use sanitize_flags_p.
(ubsan_instrument_shift): Likewise.
(ubsan_instrument_bounds): Likewise.
(ubsan_maybe_instrument_array_ref): Likewise.
(ubsan_maybe_instrument_reference_or_call): Likewise.
2017-06-13  Martin Liska  <mliska@suse.cz>

PR sanitize/78204
* asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
(gate_asan): Likewise.
* asan.h (asan_no_sanitize_address_p): Remove the function.
(sanitize_flags_p): New function.
* builtins.def: Fix coding style.
* common.opt: Use renamed enum value.
* convert.c (convert_to_integer_1): Use sanitize_flags_p.
* doc/extend.texi: Document no_sanitize attribute.
* flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
to SANITIZE_UNDEFINED_NONDEFAULT.
* gcc.c (sanitize_spec_function): Use the renamed enum value.
* gimple-fold.c (optimize_atomic_compare_exchange_p):
Use sanitize_flags_p.
* gimplify.c (gimplify_function_tree): Likewise.
* ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
* opts.c (parse_no_sanitize_attribute): New function.
(common_handle_option): Use renamed enum value.
* opts.h (parse_no_sanitize_attribute): Declare.
* tree.c (sanitize_flags_p): New function.
* tree.h: Declared here.
* tsan.c: Use sanitize_flags_p.
* ubsan.c (ubsan_expand_null_ifn): Likewise.
(instrument_mem_ref): Likewise.
(instrument_bool_enum_load): Likewise.
(do_ubsan_in_current_function): Remove the function.
(pass_ubsan::execute): Use sanitize_flags_p.
* ubsan.h: Remove do_ubsan_in_current_function
* tree-cfg.c (print_no_sanitize_attr_value): New function.
(dump_function_to_file): Use it here.
2017-06-13  Martin Liska  <mliska@suse.cz>

PR sanitize/78204
* class.c (build_base_path): Use sanitize_flags_p.
* cp-gimplify.c (cp_genericize_r): Likewise.
(cp_genericize_tree): Likewise.
(cp_genericize): Likewise.
* cp-ubsan.c (cp_ubsan_instrument_vptr_p): Likewise.
* decl.c (compute_array_index_type): Likewise.
(start_preparsed_function): Likewise.
* decl2.c (one_static_initialization_or_destruction): Likewise.
* init.c (finish_length_check): Likewise.
* lambda.c (maybe_add_lambda_conv_op): Likewise.
* typeck.c (cp_build_binary_op): Likewise.
(build_static_cast_1): Likewise.
2017-06-13  Martin Liska  <mliska@suse.cz>

PR sanitize/78204
* c-convert.c (convert): Use sanitize_flags_p.
* c-decl.c (grokdeclarator): Likewise.
* c-typeck.c (convert_for_assignment): Likewise.
(c_finish_return): Likewise.
(build_binary_op): Likewise.

From-SVN: r249158

[ARM][Testsuite]make dump_stack function call conditional in cold-lc.c

After r249013, die () and dump_stack () are both in cold section. This makes
the compiler generate bl instruction for the function call, instead of
honoring the -mlong-calls option.

This patch changes the dump_stack function call conditional, which fixes the
regression.

gcc/testsuite/

* gcc.target/arm/cold-lc.c: Update coding style, call dump_stack
conditionally.

From-SVN: r249157

runtime: don't always show frames with no function in traceback

    If there is no function name, the traceback is generally
    uninformative.  In earlier versions we did not show such frames.
    Restore that behavior.  These frames can be seen with GOTRACEBACK=system.

    Reviewed-on: https://go-review.googlesource.com/45431

From-SVN: r249156

Fixed spelling mistake in gcc/ChangeLog

From-SVN: r249155

[PR80803 2/2] Diligent queuing in SRA grp_write prop

2017-06-13  Martin Jambor  <mjambor@suse.cz>

PR tree-optimization/80803
PR tree-optimization/81063
* tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
(propagate_subaccesses_across_link): Enqueue subtree whneve necessary
instead of relying on the caller.

testsuite/
gcc.dg/tree-ssa/pr80803.c: New test.
gcc.dg/tree-ssa/pr81063.c: Likewise.

From-SVN: r249154

[PR80803 1/2] Streamline SRA access enqueuing

2017-06-13  Martin Jambor  <mjambor@suse.cz>

* tree-sra.c (add_access_to_work_queue): Only enqueue accesses
that have a first_link.
(sort_and_splice_var_accesses): Do not check first_link before
enquing.
(subtree_mark_written_and_enqueue): Likewise.
(propagate_all_subaccesses): Likewise and do not stop at first
parent with a first_link.

From-SVN: r249153

Trivial fix in dump_access_tree_1

2017-06-13  Martin Jambor  <mjambor@suse.cz>

* tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
instead of f.

From-SVN: r249152

match.pd: New pattern.

2017-06-13  Yury Gribov  <tetra2005@gmail.com>

gcc/
* match.pd: New pattern.

gcc/testsuite/
* c-c++-common/fold-masked-cmp-3.c: New test.

From-SVN: r249151

tree-vrp.c (is_masked_range_test): New function.

2017-06-13  Yury Gribov  <tetra2005@gmail.com>

gcc/
* tree-vrp.c (is_masked_range_test): New function.
(register_edge_assert_for): Determine ranges for
some bit tests.

From-SVN: r249150

re PR tree-optimization/67328 (range test rather than single bit test for code testing enum values)

2017-06-13  Yury Gribov  <tetra2005@gmail.com>

gcc/
PR tree-optimization/67328
* fold-const.c (maskable_range_p): New function.
(build_range_check): Generate bittests if possible.

gcc/testsuite/
PR tree-optimization/67328
* c-c++-common/fold-masked-cmp-1.c: New test.
* c-c++-common/fold-masked-cmp-2.c: Likewise.
* gcc.dg/pr46309.c: Fix pattern.
* gcc.dg/pr46309-2.c: Likewise.

From-SVN: r249149

sdiv_costs_1.c: Require arm_arch_v8a_ok and add march option.

2017-06-13  Tamar Christina  <tamar.christina@arm.com>

* gcc.target/arm/sdiv_costs_1.c:
Require arm_arch_v8a_ok and add march option.

From-SVN: r249148

Enhance dump_probability function.

2017-06-13  Martin Liska  <mliska@suse.cz>

* gimple-pretty-print.c (dump_probability): Add new argument.
(dump_edge_probability): Dump both probability and count.
(dump_gimple_label): Likewise.
(dump_gimple_bb_header): Likewise.
2017-06-13  Martin Liska  <mliska@suse.cz>

* gcc.dg/tree-ssa/builtin-sprintf-2.c: Adjust scanned pattern.
* gcc.dg/tree-ssa/dump-2.c: Likewise.
* gcc.dg/tree-ssa/vrp101.c: Likewise.

From-SVN: r249147

re PR target/81072 ([avr] Support some XMEGA devices with flash seen in RAM address space.)

PR target/81072
* config/avr/avr-devices.c: Fix indentation.
* config/avr/gen-avr-mmcu-specs.c: Dito.

From-SVN: r249146

tree-vect-loop.c (vect_model_reduction_cost): Do not fail, instead get vector type from stmt_info.

2017-06-13  Richard Biener  <rguenther@suse.de>

* tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
instead get vector type from stmt_info.
(vectorizable_reduction): Adjust.  Remove dead code.

From-SVN: r249145

re PR sanitizer/81065 (UBSAN: false positive as a result of distribution involving different types)

2017-06-13  Richard Biener  <rguenther@suse.de>

PR middle-end/81065
* fold-const.c (extract_muldiv_1): Remove bogus distribution
case of C * (x * C2 + C3).
(fold_addr_of_array_ref_difference): Properly fold index difference.

* c-c++-common/ubsan/pr81065.c: New testcase.

From-SVN: r249144

runtime: clear isSystemGoroutine in goexit

    Otherwise it may be set when the g struct is reused via gfput/gfget.

    Test is golang.org/x/net/http2 with GOMAXPROCS=12.

    Reviewed-on: https://go-review.googlesource.com/45430

From-SVN: r249143

Daily bump.

From-SVN: r249142

runtime: ignore _Gscan bit when checking status in CgocallDone

    Also always access the atomicstatus field atomically.

    The effect of not checking the _Gscan bit is that if the GC decides to
    scan the stack just as the goroutine is leaving the system call, the
    goroutine might fail to call exitsyscall.  Then then typically causes
    a runtime assertion failure later on.  If we do call exitsyscall as we
    should, it will stall (in casgstatus) until the _Gscan bit is cleared.

    No separate test.  I've observed causing sporadic failures running the
    misc/cgo tests, but we don't currently have a way to run those
    routinely for gccgo.  I should fix that.

    Reviewed-on: https://go-review.googlesource.com/45392

From-SVN: r249138

Finish implementing P0426R1 "Constexpr for std::char_traits" for C++17

As discussed in PR c++/80265 ("__builtin_{memcmp,memchr,strlen} are
not usable in constexpr functions"), use __builtin_constant_p to tell
whether we can defer to a constexpr algorithm.

I used __always_inline__ just to be thorough.  It isn't really really
necessary as far as I could determine.

Changes like these:

 if (__n == 0)
   return 0;
 - return wmemcmp(__s1, __s2, __n);
 + else
 +   return wmemcmp(__s1, __s2, __n);

are necessary otherwise G++ complains that we're calling a
non-constexpr function, which looks like a a manifestation of PR67026
to me.

libstdc++-v3:
2017-06-12  Pedro Alves  <palves@redhat.com>

* doc/xml/manual/status_cxx2017.xml: Update C++17 constexpr
char_traits status.
* doc/html/*: Regenerate.

* include/bits/char_traits.h (_GLIBCXX_ALWAYS_INLINE): Define if
not already defined.
(__cpp_lib_constexpr_char_traits): Uncomment.
(__constant_string_p, __constant_char_array_p): New.
(std::char_traits<char>, std::char_traits<wchar_t>): Add
_GLIBCXX17_CONSTEXPR on compare, length and find and use
__constant_string_p, __constant_char_array_p and
__builtin_constant_p to defer to __gnu_cxx::char_traits at compile
time.

* testsuite/21_strings/char_traits/requirements/
constexpr_functions_c++17.cc: Uncomment
__cpp_lib_constexpr_char_traits tests.  Uncomment
test_compare<char>, test_length<char>, test_find<char>,
test_compare<wchar_t>, test_length<wchar_t> and test_find<wchar_t>
static_assert tests.

From-SVN: r249137

stl_tree.h (_Rb_tree_impl()): Restore _Node_allocator default init.

2017-06-12  François Dumont  <fdumont@gcc.gnu.org>

* include/bits/stl_tree.h (_Rb_tree_impl()): Restore _Node_allocator
default init.
* testsuite/util/testsuite_allocator.h
(__gnu_test::default_init_allocator<>) New.
* testsuite/23_containers/set/allocator/default_init.cc: New.
* testsuite/23_containers/map/allocator/default_init.cc: New.

From-SVN: r249136

More refinements to fixing sparc's PR target/80968.

gcc/

PR target/80968
* config/sparc/sparc.md (return expander): Emit frame blockage if
function uses alloca.

From-SVN: r249134

compiler: avoid orphaning Bexpressions when processing conversions

    The method Type_conversion_expression::do_get_backend was (in some
    circumstances) creating a Bexpression for the source expression of the
    conversion and then throwing it away before using it. Fix up this
    method to insure that the call to get_backend() on the source
    expression is only made when the result will be used.

    Reviewed-on: https://go-review.googlesource.com/45350

From-SVN: r249131

PR libstdc++/55917 do not handle exceptions in std::thread

PR libstdc++/55917
* src/c++11/thread.cc (execute_native_thread_routine): Remove
try-block so that exceptions propagate out of the thread and terminate
is called by the exception-handling runtime.
(execute_native_thread_routine_compat): Likewise.
* testsuite/30_threads/thread/cons/terminate.cc: New.

From-SVN: r249130

split-1.c: Require split_stack, don't require freorder.

* gcc.dg/tree-prof/split-1.c: Require split_stack, don't require
freorder.  Update comment to explain test.

From-SVN: r249129

Fix pessimistic DImode handling in combine.c:make_field_assignment

The make_field_assignment code:

      src = force_to_mode (src, mode,
       GET_MODE_PRECISION (mode) >= HOST_BITS_PER_WIDE_INT
       ? HOST_WIDE_INT_M1U
       : (HOST_WIDE_INT_1U << len) - 1,
       0);

would ignore the field length len for DImode, even though DImode can be
handled using HWIs.  I think the code should be testing len instead.

2017-06-12  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
* combine.c (make_field_assignment): Check len rather than the mode
precision when calling force_to_mode.

From-SVN: r249128

inline-lrint_1.c: Broaden regexp.

2017-06-12  Tamar Christina  <tamar.christina@arm.com>

* gcc.target/aarch64/inline-lrint_1.c: Broaden regexp.
* gcc.target/aarch64/inline-lrint_2.c: Likewise.
* gcc.target/aarch64/no-inline-lrint_1.c: Likewise.
* gcc.target/aarch64/no-inline-lrint_2.c: Likewise.

From-SVN: r249127

sdiv_costs_1.c: Require arm_v8_vfp_ok.

2017-06-12  Tamar Christina  <tamar.christina@arm.com>

* gcc.target/arm/sdiv_costs_1.c: Require arm_v8_vfp_ok.

From-SVN: r249125

Support multilibs and devices that see flash in RAM address range.

gcc/
Support multilibs and devices that see flash in RAM address range.

PR target/81072
* config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
(avr_mcu_t) <flash_pm_offset>: New field.
(avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
* config/avr/avr.h (AVR_SHORT_CALLS): New define.
(AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
(AVR_TINY_PM_OFFSET): Remove macro.
* config/avr/avr.opt (-mshort-calls): New option.
* config/avr/gen-avr-mmcu-specs.c (print_mcu)
[*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
* config/avr/avr-c.c (avr_cpu_cpp_builtins)
<__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
<__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
instead of avr_arch->have_jmp_call.
<__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
[AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
avr_arch->flash_pm_offset to define.
* config/avr/avr-devices.c (avr_arch_types): Add initializers for
new field flash_pm_offset.  Add entry for avrxmega3.
(avr_texinfo): Add entry for avrxmega3.
* config/avr/avr-mcus.def: Add entries for: avrxmega3,
attiny212, attiny214,
attiny412, attiny414, attiny416, attiny417,
attiny814, attiny816, attiny817,
attiny1614, attiny1616, attiny1617,
attiny3214, attiny3216, attiny3217.
* config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
(avr_print_operand_address) [AVR_TINY]: Same.
(avr_asm_init_sections) <readonly_data_section>: Only patch
callback if avr_arch->flash_pm_offset = 0.
(avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
for rodata if avr_arch->flash_pm_offset != 0.
(avr_encode_section_info) [AVR_TINY]: Adjust comment.
* config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
(opts) [AVR_ISA_RCALL]: Append opt_rcall.
(m_options): Append opt_rcall.
(m_dirnames): Append dir_rcall.
* config/avr/t-multilib: Regenerate.
* configure.ac [target=avr]: Check whether avrxmega3 default
linker description file works as needed.
* configure: Regenerate.
* doc/avr-mmcu.texi: Regenerate.
* doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
<__AVR_ARCH__>: Document avrxmega3 and 103.
<__AVR_HAVE_JMP_CALL__>: Adjust documentation.
<__AVR_SHORT_CALLS__>: Document it.
<__AVR_PM_BASE_ADDRESS__>: Document it.
* doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
(AVR Variable Attributes) <progmem>: Document this is
not needed for avrxmega3.
(AVR Named Address Spaces) <__flash>: Dito.

From-SVN: r249124

* cgraph.c (cgraph_node::dump): Complain about profile insanities.

From-SVN: r249123

sdiv_costs_1.c: Disable on softfloat.

2017-06-12  Tamar Christina  <tamar.christina@arm.com>

* gcc.target/arm/sdiv_costs_1.c: Disable on softfloat.

From-SVN: r249122

config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".

2017-06-12  Doug Rupp  <rupp@adacore.com>

gcc/
* config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
Append vxworks-stdint.h to the tm_file list.
* config/vxworks-stdint.h: New file.

fixincludes/
* inclhack.def (AAB_vxworks_stdint): Remove hack.
* fixincl.x: Regenerate.

From-SVN: r249121

Create an extra BB in profile-generate (PR tree-optimization/81041).

2017-06-12  Martin Liska  <mliska@suse.cz>

PR tree-optimization/81041
* tree-profile.c (gimple_gen_ic_func_profiler):
Create an extra BB in profile-generate
(gimple_gen_time_profiler): Likewise.

From-SVN: r249120

re PR bootstrap/80897 (gnat bootstrap broken on SPARC64/Linux)

PR bootstrap/80897
* exp_ch3.adb (Make_Predefined_Primitive_Specs): Use Positive index.

From-SVN: r249119

rs6000: Remove e500.h from config.gcc some more

* config.gcc: Remove rs6000/e500.h from tm_file for all targets.

From-SVN: r249117

re PR ada/81070 (build failure for s-intrr.adb)

PR ada/81070
* s-interr-hwint.adb: Reinstate.
* gcc-interface/Makefile.in (RTEMS): Use it again.

From-SVN: r249115

re PR tree-optimization/81003 (ICE: tree check: expected ssa_name, have integer_cst in optimize_range_tests_to_bit_test, at tree-ssa-reassoc.c:2782)

PR tree-optimization/81003
* tree-ssa-reassoc.c (force_into_ssa_name): New function.
(update_range_test): Use it instead of force_gimple_operand_gsi.

* gcc.c-torture/compile/pr81003.c: New test.

From-SVN: r249114

re PR tree-optimization/81053 (ICE on valid code at -O3 on x86_64-linux-gnu: in as_a, at is-a.h:192)

2017-06-12  Richard Biener  <rguenther@suse.de>

PR tree-optimization/81053
* tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
with backedge value not defined in loop.  Simplify def stmt
compute.

* gcc.dg/torture/pr81053.c: New testcase.

From-SVN: r249113

elf.c (backtrace_initialize): Always set *fileline_fn.

* elf.c (backtrace_initialize): Always set *fileline_fn.
* ttest.c: New file.
* btest.c: Move support functions into testlib.c.  Change calls to
check to pass file name.
* testlib.c: New file, copied from (part of) btest.c.
* testlib.h: New file, declarations for testlib.c.
* edtest.c: Use testlib.h and testlib.c.
* configure.ac: Test for -pthread, set HAVE_PTHREAD conditional.
* Makefile.am (btest_SOURCES): Add testlib.c.
(edtest_SOURCES): Likewise.
(CHECK_PROGRAMS): Add ttest if HAVE_PTHREAD.
(ttest_SOURCES, ttest_CFLAGS, ttest_LDADD): Define.
* configure, Makefile.in: Rebuild.

From-SVN: r249111

Daily bump.

From-SVN: r249110

Disable constant pool for nvptx

2017-06-11  Tom de Vries  <tom@codesourcery.com>

PR target/79939
* config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
Return true.
(TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
nvptx_cannot_force_const_mem.

From-SVN: r249106

opts.c (finish_options): Move test for flag_split_stack after it has been initialized.

* opts.c (finish_options): Move test for flag_split_stack after
it has been initialized.

From-SVN: r249105

* error.c (dump_expr): Use is_this_parameter.

From-SVN: r249104

tree.h (id_equal): New.

* tree.h (id_equal): New.

* dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
instead of strcmp of IDENTIFIER_POINTER.
c-family/
* c-ada-spec.c, c-pragma.c: Use it.
cp/
* cp-tree.h, decl2.c, mangle.c, parser.c, pt.c, semantics.c: Use it.

From-SVN: r249103

Daily bump.

From-SVN: r249102

ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h

* ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
(mark_all_inlined_calls_cdtor): Fix formating.
(inline_transform): Rescale profile before inlining.

From-SVN: r249098

cgraph.h (cgraph_edge::clone): Update prototype.

* cgraph.h (cgraph_edge::clone): Update prototype.
* cgraphclones.c (cgraph_edge::clone): Update profile scaling.
(cgraph_node::create_clone): Update.
(cgraph_node::create_version_clone): Update.
* tree-inline.c (copy_bb): Update.
(expand_call_inline): Update.

From-SVN: r249097

rs6000: Factor out emit_vrsave_prologue

This is a second patch to make the prologue code more manageable.

* config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
factored out from ...
(rs6000_emit_prologue): ... here.

From-SVN: r249096

rs6000: Factor out emit_split_stack_prologue

This is a first patch to make the prologue code more manageable.

* config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
factored out from ...
(rs6000_emit_prologue): ... here.

From-SVN: r249095

predict.c (drop_profile): Also drop individual bb/edge and cgraph edge counts.

* predict.c (drop_profile): Also drop individual bb/edge and cgraph
edge counts.
(handle_missing_profiles): Fix computation of tp_first_run.
(counts_to_freqs): Do not touch freqs when count is 0.

From-SVN: r249094

cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch profile.

* cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
profile.

From-SVN: r249093

re PR fortran/80988 (171_swim fails)

2017-06-10  Thomas Koenig  <tkoenig@gcc.gnu.org>

PR fortran/80988
* frontend-passes.c (traverse_io_block):  Also
check for variables occurring as indices multiple
time in a single implied DO loop.

2017-06-10  Thomas Koenig  <tkoenig@gcc.gnu.org>

PR fortran/80988
* gfortran.dg/implied_do_io_3.f90:  New test.

From-SVN: r249092

Add effective target signal

2017-06-10  Tom de Vries  <tom@codesourcery.com>

* lib/target-supports.exp (check_effective_target_signal): New proc.
* lib/gcc.exp (gcc_target_compile): Remove appending of
-DSIGNAL_SUPPRESS to additional_flags.
* gcc.c-torture/execute/ieee/ieee.exp: Add -DSIGNAL_SUPPRESS to
additional_flags if effective target signal is not supported by the
target.
* gcc.c-torture/execute/20101011-1.c: Same.
* gcc.dg/c99-stdint-1.c: Same.
* gcc.dg/c99-stdint-2.c: Same.
* gcc.dg/c99-stdint-5.c: Same.
* gcc.dg/c99-stdint-6.c: Same.
* gcc.dg/stdint-width-1.c: Same.

* doc/sourcebuild.texi (Effective-Target Keywords, Environment
attributes): Document signal effective target.

From-SVN: r249091

Add effective target stack_size

2017-06-10  Tom de Vries  <tom@codesourcery.com>

* lib/target-supports.exp (check_effective_target_stack_size)
(dg-effective-target-value): New proc.
* lib/gcc.exp (gcc_target_compile): Remove adding of DSTACK_SIZE to
additional_flags.
* lib/objc.exp (objc_target_compile): Same.
* lib/profopt.exp (profopt-get-options): Allow dg-additional-options.
* g++.old-deja/g++.eh/badalloc1.C: Add -DSTACK_SIZE=<value> using
dg-additional-options.
* gcc.c-torture/compile/20080806-1.c: Same.
* gcc.c-torture/compile/920723-1.c: Same.
* gcc.c-torture/compile/930621-1.c: Same.
* gcc.c-torture/compile/991214-2.c: Same.
* gcc.c-torture/compile/bcopy.c: Same.
* gcc.c-torture/compile/memtst.c: Same.
* gcc.c-torture/compile/msp.c: Same.
* gcc.c-torture/compile/stuct.c: Same.
* gcc.c-torture/execute/20011008-3.c: Same.
* gcc.c-torture/execute/20030209-1.c: Same.
* gcc.c-torture/execute/20031012-1.c: Same.
* gcc.c-torture/execute/20040805-1.c: Same.
* gcc.c-torture/execute/920410-1.c: Same.
* gcc.c-torture/execute/920501-7.c: Same.
* gcc.c-torture/execute/921113-1.c: Same.
* gcc.c-torture/execute/921202-1.c: Same.
* gcc.c-torture/execute/921208-2.c: Same.
* gcc.c-torture/execute/930106-1.c: Same.
* gcc.c-torture/execute/930406-1.c: Same.
* gcc.c-torture/execute/950221-1.c: Same.
* gcc.c-torture/execute/960521-1.c: Same.
* gcc.c-torture/execute/980605-1.c: Same.
* gcc.c-torture/execute/comp-goto-1.c: Same.
* gcc.c-torture/execute/comp-goto-2.c: Same.
* gcc.c-torture/execute/memcpy-1.c: Same.
* gcc.c-torture/execute/multi-ix.c: Same.
* gcc.c-torture/execute/nestfunc-4.c: Same.
* gcc.c-torture/execute/pr20621-1.c: Same.
* gcc.c-torture/execute/pr23135.c: Same.
* gcc.c-torture/execute/pr28982b.c: Same.
* gcc.dg/loop-3.c: Same.
* gcc.dg/struct-ret-3.c: Same.
* gcc.dg/torture/stackalign/comp-goto-1.c: Same.
* gcc.dg/torture/stackalign/non-local-goto-4.c: Same.
* gcc.dg/tree-prof/comp-goto-1.c: Same.
* gcc.dg/tree-prof/pr44777.c: Same.

* doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
Document effective target stack_size.

From-SVN: r249090

Missing bits from N4268, constant evaluation for all non-type args.

* call.c (build_converted_constant_expr): Rename from
build_integral_nontype_arg_conv, handle all types.
* pt.c (convert_nontype_argument): In C++17 call it for all types.
Move NOP stripping inside pointer case, don't strip ADDR_EXPR.
* cvt.c (strip_fnptr_conv): Also strip conversions to the same type.

From-SVN: r249089