My patch for 64372 removed a bogus lvalue-rvalue conversion for one arm of a
?: expression where the other arm is a throw. But we still need to require
any overload to be resolved, even though we aren't getting that from
decay_conversion anymore.
* pt.c (resolve_nondeduced_context_or_error): Split out from...
* typeck.c (decay_conversion): ...here.
* call.c (build_conditional_expr_1): Use it.
From-SVN: r268058
The StackGrowsDown version being turned off for hppa targets.
After other fixes in the compiler, this allows core.thread unittests to
all pass, as well as the garbage collector to work correctly.
Backported from upstream druntime 2.084.
Reviewed-on: https://github.com/dlang/druntime/pull/2410
From-SVN: r268056
PR target/88734
* config/aarch64/arm_neon.h: Fix #pragma GCC target syntax - replace
(("..."))) with ("..."). Use arch=armv8.2-a+sha3 instead of
arch=armv8.2-a+crypto for vsha512hq_u64 etc. intrinsics.
From-SVN: r268049
PR middle-end/88273 - [8/9 Regression] warning: 'memcpy' offset [-527, -529]
is out of the bounds [0, 16]
gcc/ChangeLog:
PR middle-end/88273
* gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range):
Handle anti-ranges the same as no range at all.
gcc/testsuite/ChangeLog:
PR middle-end/88273
* gcc.dg/Warray-bounds-38.c: New test.
From-SVN: r268048
* c-typeck.c (digest_init): Revised the error message produced for
ill-formed cases of array initialization with a string literal.
(error_init): Make variadic.
Co-Authored-By: Jason Merrill <jason@redhat.com>
From-SVN: r268047
When we partially instantiate the constexpr if, we walk through its body to
see what it uses from the enclosing local_specializations. That walk was
overlooking the use of 'count' in the captures of the innermost lambda,
because we weren't walking into the capture list.
* tree.c (cp_walk_subtrees): Handle LAMBDA_EXPR.
From-SVN: r268046
PR c++/88699 reports an ICE within this assertion in add_method:
gcc_assert (!current_fns || !DECL_DESTRUCTOR_P (method));
when adding an overloaded operator to a class via a using_decl, due to
DECL_DESTRUCTOR_P requiring a FUNCTION_DECL, but "method" being a
USING_DECL.
This patch weakens the assertion to avoid testing DECL_DESTRUCTOR_P
for the case where "via_using" is true, fixing the ICE.
gcc/cp/ChangeLog:
PR c++/88699
* class.c (add_method): Don't use DECL_DESTRUCTOR_P on
USING_DECLs.
gcc/testsuite/ChangeLog:
PR c++/88699
* g++.dg/template/pr88699.C: New test.
From-SVN: r268041
* src/c++17/fs_ops.cc
(equivalent(const path&, const path&, error_code&))
[_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Use GetFileInformationByHandle to
compare files instead of relying on incomplete info returned by stat.
From-SVN: r268036
PR libstdc++/88884
* src/c++17/fs_ops.cc (absolute(const path&, error_code&)): Do nothing
if the path is already absolute.
[_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Normalize root-directory.
[!_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Pass error_code to current_path.
From-SVN: r268035
On Windows stat("foo/bar/../.") will resolve to "foo" even if that is a
non-directory and "foo/bar" does not exist. This is the expected
behaviour and consistent with boost::filesystem, so don't try to correct
it. The only unwanted behaviour is that stat("baz/") fails due to a
mingw bug (fixed in mingw-w64 v6.0.0) so add a workaround.
PR libstdc++/88881
* src/c++17/fs_ops.cc (canonical(const path&, error_code&))
[_GLIBCXX_FILESYSTEM_IS_WINDOWS]: Normalize path, to match behaviour
of filesystem::exists.
(create_directories(const path&, error_code&)): Add assertions.
(status(const path&, error_code&)) [_GLIBCXX_FILESYSTEM_IS_WINDOWS]:
Add workaround for bug in _wstat for paths with trailing slash.
* testsuite/27_io/filesystem/operations/create_directories.cc: Adjust
for expected behaviour on mingw.
* testsuite/experimental/filesystem/operations/create_directories.cc:
Likewise.
* testsuite/27_io/filesystem/operations/temp_directory_path.cc: Use
"TMP" instead of "TMPDIR" and clean environment before each test. Do
not test permissions on mingw targets.
From-SVN: r268034
Add test-case to verify that libbacktrace can read debug info that was
compressed with dwz.
2019-01-17 Tom de Vries <tdevries@suse.de>
PR libbacktrace/82857
* configure.ac (DWZ): Set with AC_CHECK_PROG.
(HAVE_DWZ): Set with AM_CONDITIONAL.
* configure: Regenerate.
* Makefile.am (TESTS): Add btest_dwz.
* Makefile.in: Regenerate.
From-SVN: r268032
Handle DW_FORM_GNU_ref_alt which references the .debug_info section in the
.gnu_debugaltlink file.
2019-01-17 Tom de Vries <tdevries@suse.de>
PR libbacktrace/82857
* dwarf.c (enum attr_val_encoding): Add ATTR_VAL_REF_ALT_INFO.
(read_attribute): Handle DW_FORM_GNU_ref_alt using
ATTR_VAL_REF_ALT_INFO.
(read_referenced_name_from_attr): Handle DW_FORM_GNU_ref_alt.
From-SVN: r268031
Add a function that finds the unit given an offset into .debug_info.
2019-01-17 Tom de Vries <tdevries@suse.de>
* dwarf.c (struct unit): Add low_offset and high_offset fields.
(struct unit_vector): New type.
(struct dwarf_data): Add units and units_counts fields.
(find_unit): New function.
(find_address_ranges): Add and handle unit_tag parameter.
(build_address_map): Add and handle units_vec parameter.
(build_dwarf_data): Pass units_vec to build_address_map. Store resulting
units vector.
From-SVN: r268030
The GCN toolchain must use the LLVM assembler and linker because there's no
binutils port. The LLVM tools do not have the same diagnostic style as
binutils, so the "blank line(s) in output" tests are inappropriate (and very
noisy).
The LLVM tools also have different command line options, so it's not possible
to autodetect object formats in the same way.
This patch addresses both issues.
2019-01-17 Andrew Stubbs <ams@codesourcery.com>
gcc/
* doc/sourcebuild.texi: Document dg-require-effective-target
llvm_binutils and offload_gcn.
gcc/testsuite/
* lib/file-format.exp (gcc_target_object_format): Handle AMD GCN.
* lib/gcc-dg.exp (gcc-dg-prune): Ignore blank lines from the LLVM
linker.
* lib/target-supports.exp (check_effective_target_llvm_binutils): New.
From-SVN: r268027
The GCN/HSA loader ignores the load address and uses a random location, so we
build all GCN binaries as PIE, by default.
This patch makes the necessary testsuite adjustments to make this work
correctly.
2019-01-17 Andrew Stubbs <ams@codesourcery.com>
gcc/testsuite/
* gcc.dg/graphite/scop-19.c: Check pie_enabled.
* gcc.dg/pic-1.c: Disable on amdgcn.
* gcc.dg/pic-2.c: Disable on amdgcn.
* gcc.dg/pic-3.c: Disable on amdgcn.
* gcc.dg/pic-4.c: Disable on amdgcn.
* gcc.dg/pie-3.c: Disable on amdgcn.
* gcc.dg/pie-4.c: Disable on amdgcn.
* gcc.dg/uninit-19.c: Check pie_enabled.
* lib/target-supports.exp (check_effective_target_pie): Add amdgcn.
From-SVN: r268026
This patch contains the configuration adjustments needed to enable the GCN
back-end.
The new configure check for dlopen is required to allow building the new
gcn-run tool. This tool uses libdl to load the HSA runtime libraries, which
are required to run programs on the GPU. The tool is disabled if libdl is not
available.
2019-01-17 Andrew Stubbs <ams@codesourcery.com>
Kwok Cheung Yeung <kcy@codesourcery.com>
Julian Brown <julian@codesourcery.com>
Tom de Vries <tom@codesourcery.com>
Jan Hubicka <hubicka@ucw.cz>
Martin Jambor <mjambor@suse.cz>
* configure.ac: Likewise.
* configure: Regenerate.
* contrib/config-list.mk: Add amdgcn-amdhsa.
gcc/
* config.gcc: Add amdgcn*-*-amdhsa configuration.
* configure.ac: Check for dlopen.
* configure: Regenerate.
Co-Authored-By: Jan Hubicka <hubicka@ucw.cz>
Co-Authored-By: Julian Brown <julian@codesourcery.com>
Co-Authored-By: Kwok Cheung Yeung <kcy@codesourcery.com>
Co-Authored-By: Martin Jambor <mjambor@suse.cz>
Co-Authored-By: Tom de Vries <tom@codesourcery.com>
From-SVN: r268024
This patch contains the major part of the GCN back-end. The machine
description has been broken out to avoid the mailing list size limit.
The back-end contains various bits that support OpenACC and OpenMP, but the
middle-end and libgomp patches are missing, as is mkoffload. I include them
here because they're harmless and carving up the files seems like unnecessary
effort. The remaining offload support will be posted at a later date.
The gcn-run.c is a separate tool that can run a GCN program on a GPU using
the ROCm drivers and HSA runtime libraries.
2019-01-17 Andrew Stubbs <ams@codesourcery.com>
Kwok Cheung Yeung <kcy@codesourcery.com>
Julian Brown <julian@codesourcery.com>
Tom de Vries <tom@codesourcery.com>
Jan Hubicka <hubicka@ucw.cz>
Martin Jambor <mjambor@suse.cz>
gcc/
* common/config/gcn/gcn-common.c: New file.
* config/gcn/driver-gcn.c: New file.
* config/gcn/gcn-builtins.def: New file.
* config/gcn/gcn-hsa.h: New file.
* config/gcn/gcn-modes.def: New file.
* config/gcn/gcn-opts.h: New file.
* config/gcn/gcn-passes.def: New file.
* config/gcn/gcn-protos.h: New file.
* config/gcn/gcn-run.c: New file.
* config/gcn/gcn-tree.c: New file.
* config/gcn/gcn.c: New file.
* config/gcn/gcn.h: New file.
* config/gcn/gcn.opt: New file.
* config/gcn/t-gcn-hsa: New file.
Co-Authored-By: Jan Hubicka <hubicka@ucw.cz>
Co-Authored-By: Julian Brown <julian@codesourcery.com>
Co-Authored-By: Kwok Cheung Yeung <kcy@codesourcery.com>
Co-Authored-By: Martin Jambor <mjambor@suse.cz>
Co-Authored-By: Tom de Vries <tom@codesourcery.com>
From-SVN: r268023
This patch contains the machine description portion of the GCN back-end. I've
broken it out mainly to avoid the mailing list size limit.
2019-01-17 Andrew Stubbs <ams@codesourcery.com>
Kwok Cheung Yeung <kcy@codesourcery.com>
Julian Brown <julian@codesourcery.com>
Tom de Vries <tom@codesourcery.com>
Jan Hubicka <hubicka@ucw.cz>
Martin Jambor <mjambor@suse.cz>
gcc/
* config/gcn/constraints.md: New file.
* config/gcn/gcn-valu.md: New file.
* config/gcn/gcn.md: New file.
* config/gcn/predicates.md: New file.
Co-Authored-By: Jan Hubicka <hubicka@ucw.cz>
Co-Authored-By: Julian Brown <julian@codesourcery.com>
Co-Authored-By: Kwok Cheung Yeung <kcy@codesourcery.com>
Co-Authored-By: Martin Jambor <mjambor@suse.cz>
Co-Authored-By: Tom de Vries <tom@codesourcery.com>
From-SVN: r268022
This patch contains the GCN port of libgcc.
2019-01-17 Andrew Stubbs <ams@codesourcery.com>
Kwok Cheung Yeung <kcy@codesourcery.com>
Julian Brown <julian@codesourcery.com>
Tom de Vries <tom@codesourcery.com>
libgcc/
* config.host: Recognize amdgcn*-*-amdhsa.
* config/gcn/crt0.c: New file.
* config/gcn/lib2-divmod-hi.c: New file.
* config/gcn/lib2-divmod.c: New file.
* config/gcn/lib2-gcn.h: New file.
* config/gcn/sfp-machine.h: New file.
* config/gcn/t-amdgcn: New file.
Co-Authored-By: Julian Brown <julian@codesourcery.com>
Co-Authored-By: Kwok Cheung Yeung <kcy@codesourcery.com>
Co-Authored-By: Tom de Vries <tom@codesourcery.com>
From-SVN: r268021
This patch contains the GCN port of libgfortran. We use the minimal
configuration created for NVPTX. That's all that's required, besides the
target-independent bug fixes posted already.
2019-01-17 Andrew Stubbs <ams@codesourcery.com>
Kwok Cheung Yeung <kcy@codesourcery.com>
Julian Brown <julian@codesourcery.com>
Tom de Vries <tom@codesourcery.com>
libgfortran/
* configure.ac: Use minimal mode for amdgcn.
* configure: Regenerate.
Co-Authored-By: Julian Brown <julian@codesourcery.com>
Co-Authored-By: Kwok Cheung Yeung <kcy@codesourcery.com>
Co-Authored-By: Tom de Vries <tom@codesourcery.com>
From-SVN: r268020
This patch adds a tuning struct for the Arm Ares CPU and uses it for -m{cpu,tune}=ares.
The tunings are an initial attempt and may be improved upon in the future, but they serve
as a decent starting point for GCC 9.
With this I see a 1.3% improvement on SPEC2006 int and 0.3% on SPEC2006 fp with -mcpu=ares.
On SPEC2017 I see a 0.6% improvement in intrate and changes in the noise for fprate.
* config/aarch64/aarch64.c (ares_tunings): Define.
* config/aarch64/aarch64-cores.def (ares): Use the above.
From-SVN: r268015
2019-01-17 Richard Biener <rguenther@suse.de>
PR lto/86736
* dwarf2out.c (want_pubnames): Never generate pubnames sections
and friends for the LTO part of debug info.
From-SVN: r268010
PR rtl-optimization/88870
* dce.c (deletable_insn_p): Never delete const/pure calls that can
throw if we can't alter the cfg or delete dead exceptions.
(mark_insn): Don't call find_call_stack_args for such calls.
* gcc.dg/pr88870.c: New test.
From-SVN: r268008
2019-01-17 Kewen Lin <linkw@gcc.gnu.org>
gcc/ChangeLog
* doc/extend.texi: Add four new prototypes for vec_ld and seven new
prototypes for vec_st.
* config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add entries
for scalar address type variants of altivec_vec_ld/altivec_vec_st,
mainly on signed/unsigned long long and double.
gcc/testsuite/ChangeLog
* gcc.target/powerpc/altivec_vld_vst_addr.c: New test.
From-SVN: r268007
Concepts-checking and other kinds of early tsubsting may often take
place while location wrappers are suppressed, e.g. because we've
triggered template instantiation within template parameter lists.
With that, exprs that are usually wrapped by VIEW_CONVERT_EXPRs
location wrappers may end up wrapped by NON_LVALUE_EXPRs that are not
marked as location wrappers. If such NON_LVALUE_EXPRs tsubsted exprs
undergo another round of tsubsting, say for constraint checking, or
even for another round of specialization, they will be rejected by
tsubst_copy_and_build.
This patch arranges for suppress_location_wrappers to be saved and
reset when pushing to the top level, and restored when popping from
it.
for gcc/cp/ChangeLog
PR c++/87768
* cp-tree.h (saved_scope): Add suppress_location_wrappers.
* name-lookup.c (do_push_to_top_level): Save and reset it.
(do_pop_from_top_level): Restore it.
for gcc/testsuite/ChangeLog
PR c++/87768
* g++.dg/concepts/pr87768.C: New.
From-SVN: r268006
dwarf2out recognizes unspecified auto types by the identifier. C++
template class placeholders are unspecified auto types that take the
identifier of the class rather than those used by preexisting auto
types, so dwarf2out ICEs when it finds one of those. Alas, they may
be visible to dwarf2out, since the types of e.g. static data members
of templates are only deduced at member instantiation, i.e., if the
data member is actually referenced, but the data member is added as a
field, still with unspecified auto placeholder type, when the
enclosing class is instantiated.
I've changed placeholder creator to use an auto identifier instead,
which allowed dropping the placeholder test in C++'s is_auto (alas, it
can't be used in dwarf2out, think LTO). To avoid losing information
in error messages and dumps and whatnot, I've added code to recognize
placeholders for template classes say A and print them out as
A<...auto...>.
for gcc/cp/ChangeLog
PR c++/86648
* pt.c (make_template_placeholder): Use auto_identifier.
(is_auto): Drop CLASS_PLACEHOLDER_TEMPLATE test.
* error.c (dump_type): Handle template placeholders.
* cxx-pretty-print.c (pp_cx_unqualified_id): Likewise.
for gcc/testsuite/ChangeLog
PR c++/86648
* gcc.dg/cpp1z/pr86648.C: New.
From-SVN: r268005
Diagnostics for testsuite/g++.dg/cpp0x/inh-ctor32.C varied across
platforms. Specifically, on ARM, the diagnostics within the subtest
derived_ctor::inherited_derived_ctor::constexpr_noninherited_ctor did
not match those displayed on other platforms, and the test failed.
The difference seemed to have to do with locations assigned to ctors,
but it was more subtle: on ARM, the instantiation of bor's template
ctor was nested within the instantiation of bar's template ctor
inherited from bor. The reason turned out to be related with the
internal return type of ctors: arm_cxx_cdtor_returns_this is enabled
for because of AAPCS, while cxx.cdtor_returns_this is disabled on most
other platforms. While convert_to_void returns early with a VOID
expr, the non-VOID return type of the base ctor CALL_EXPR causes
convert_to_void to inspect the called decl for nodiscard attributes:
maybe_warn_nodiscard -> cp_get_fndecl_from_callee ->
maybe_constant_init -> cxx_eval_outermost_constant_expr ->
instantiate_constexpr_fns -> nested instantiation.
The internal return type assigned to a cdtor should not affect
instantiation (constexpr or template) decisions, IMHO. We know it
affects diagnostics, but I have a hunch this might bring deeper issues
with it, so I've arranged for the CALL_EXPR handler in convert_to_void
to disregard cdtors, regardless of the ABI.
for gcc/cp/ChangeLog
PR c++/88146
* cvt.c (convert_to_void): Handle all cdtor calls as if
returning void.
From-SVN: r268004
PR target/87306
* gcc.dg/vect/bb-slp-pow-1.c: Modify to reflect that
the loop is not vectorized on POWER unless hardware
misaligned loads are available.
From-SVN: r268003
Currently, we dropg (which clears gp.m) after we CAS the g status
to _Grunnable or _Gwaiting. Immediately after CASing the g status,
another thread may CAS it to _Gscan status and scan its stack.
With precise stack scan, it accesses gp.m in order to switch to g
and back (in doscanstackswitch). This races with dropg. If
doscanstackswitch reads gp.m, then dropg runs, when we restore
the m at the end of the scan it will set to a stale value. Worse,
if dropg runs after doscanstackswitch sets the new m, gp will be
running with a nil m.
To fix this, we do dropg before CAS g status to _Grunnable or
_Gwaiting. We can do this safely if we are CASing from _Grunning,
as we own the g when it is in _Grunning. There is one case where
we CAS from _Gsyscall to _Grunnable. It is not safe to dropg when
it is in _Gsyscall, as precise stack scan needs to read gp.m in
order to signal the m. So we need to introduce a transient state,
_Gexitingsyscall, between _Gsyscall and _Grunnable, where the GC
should not scan its stack.
In is a little unfortunate that we have to add another g status.
We could reuse an existing one (e.g. _Gcopystack), but it is
clearer and safer to just use a new one, as Austin suggested.
Reviewed-on: https://go-review.googlesource.com/c/158157
From-SVN: r268001
