platform/upstream/gcc.git
2 years agolibstdc++: ppc: conditionalize vsx-only simd intrinsics
Alexandre Oliva [Fri, 6 May 2022 18:43:42 +0000 (15:43 -0300)]
libstdc++: ppc: conditionalize vsx-only simd intrinsics

libstdc++'s bits/simd.h section for PowerPC, guarded by __ALTIVEC__,
defines various intrinsic vector types that are only available with
__VSX__: 64-bit long double, double, (un)signed long long, and 64-bit
(un)signed long.

experimental/simd/standard_abi_usable{,_2}.cc tests error out
reporting the unmet requirements when the target cpu doesn't enable
VSX.  Make the reported instrinsic types conditional on __VSX__ so
that <experimental/simd> can be used on PowerPC variants that do not
support VSX.

for  libstdc++-v3/ChangeLog

* include/experimental/bits/simd.h [__ALTIVEC__]: Require VSX
for double, long long, and 64-bit long intrinsic types.
[__ALTIVEC__] (__intrinsic_type): Mention 128-bit in
preexisting long double diagnostic, adjust no-VSX double
diagnostic to cover 64-bit long double as well.

2 years agocompiler: error for duplicate bool map keys
Ian Lance Taylor [Tue, 3 May 2022 22:48:23 +0000 (15:48 -0700)]
compiler: error for duplicate bool map keys

For golang/go#35945
Fixes golang/go#28104

Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/403954

2 years agors6000: Ignore fusion option flags for inlining test [PR102059]
Michael Meissner [Fri, 6 May 2022 16:39:13 +0000 (11:39 -0500)]
rs6000: Ignore fusion option flags for inlining test [PR102059]

The -mpower8-fusion and -mpower10-fusion options do not modify which
instructions we can generate, so ignore them when deciding whether we
can inline callee into caller.

2022-05-06   Michael Meissner  <meissner@linux.ibm.com>

gcc/
PR target/102059
* config/rs6000/rs6000.cc (rs6000_can_inline_p): Ignore -mpower8-fusion
and -mpower10-fusion options for inlining purposes.

gcc/testsuite/
PR target/102059
* gcc.target/powerpc/pr102059-4.c: New test.

2 years agoaarch64: remove useless GPF_TF_F16 iterator
Christophe Lyon [Tue, 5 Apr 2022 09:32:17 +0000 (10:32 +0100)]
aarch64: remove useless GPF_TF_F16 iterator

This iterator is not used anymore and is a leftover from previous
cleanup (r10-5890-gabbe1ed2735517).

2022-04-28  Christophe Lyon  <christophe.lyon@arm.com>

gcc/
* config/aarch64/iterators.md (GPF_TF_F16): Delete.

2 years agoOpenMP, libgomp: Add new runtime routine omp_target_is_accessible.
Marcel Vollweiler [Fri, 6 May 2022 14:28:26 +0000 (07:28 -0700)]
OpenMP, libgomp: Add new runtime routine omp_target_is_accessible.

gcc/ChangeLog:

* omp-low.cc (omp_runtime_api_call): Added target_is_accessible to
omp_runtime_apis array.

libgomp/ChangeLog:

* libgomp.map: Added omp_target_is_accessible.
* libgomp.texi: Tagged omp_target_is_accessible as supported.
* omp.h.in: Added omp_target_is_accessible.
* omp_lib.f90.in: Added interface for omp_target_is_accessible.
* omp_lib.h.in: Likewise.
* target.c (omp_target_is_accessible): Added implementation of
omp_target_is_accessible.
* testsuite/libgomp.c-c++-common/target-is-accessible-1.c: New test.
* testsuite/libgomp.fortran/target-is-accessible-1.f90: New test.

2 years agolibstdc++: Fix test that fails on Solaris [PR104731]
Jonathan Wakely [Fri, 6 May 2022 13:31:06 +0000 (14:31 +0100)]
libstdc++: Fix test that fails on Solaris [PR104731]

On Solaris the dirent::d_name member is a single char, causing this test
to fail with warnings about buffer overflow. Change the test to use a
union with additional space for writing a string to the d_name member.

libstdc++-v3/ChangeLog:

PR libstdc++/104731
* testsuite/27_io/filesystem/iterators/error_reporting.cc:
Use a trailing char array as storage for dirent::d_name.

2 years agolibstdc++: Do not include <cxxabi.h> in <stacktrace>
Jonathan Wakely [Fri, 6 May 2022 12:23:45 +0000 (13:23 +0100)]
libstdc++: Do not include <cxxabi.h> in <stacktrace>

This avoids polluting the global namespace with the "abi" namespace
alias.

libstdc++-v3/ChangeLog:

* include/std/stacktrace: Do not include <cxxabi.h>.
(__cxa_demangle): Declare.

2 years agolibstdc++: Do not use #include inside push visibility scope [PR99871]
Jonathan Wakely [Tue, 6 Apr 2021 14:11:15 +0000 (15:11 +0100)]
libstdc++: Do not use #include inside push visibility scope [PR99871]

libstdc++-v3/ChangeLog:

PR libstdc++/99871
* include/bits/specfun.h: Use visibility attribute on namespace,
instead of pragma push/pop.
* libsupc++/compare: Likewise.
* libsupc++/exception: Likewise.
* libsupc++/exception.h: Likewise.
* libsupc++/exception_ptr.h: Likewise.
* libsupc++/initializer_list: Likewise.
* libsupc++/nested_exception.h: Likewise.

2 years agolibstdc++: Update documentation about copyright and GPL notices in tests
Jonathan Wakely [Thu, 28 Apr 2022 12:28:06 +0000 (13:28 +0100)]
libstdc++: Update documentation about copyright and GPL notices in tests

There is no need to require FSF copyright for tests that are just
"self-evident" ways to check the API and behaviour of the library.
This is consistent with tests for the compiler, which do not have
copyright and licence notices either.

libstdc++-v3/ChangeLog:

* doc/xml/manual/test.xml: Remove requirement for copyright and
GPL notice in tests.
* doc/html/manual/test.html: Regenerate.

2 years agoAdd a restriction on allocate clause (OpenMP 5.0)
Hafiz Abid Qadeer [Fri, 18 Feb 2022 21:28:08 +0000 (21:28 +0000)]
Add a restriction on allocate clause (OpenMP 5.0)

An allocate clause in target region must specify an allocator
unless the compilation unit has requires construct with
dynamic_allocators clause.  Current implementation of the allocate
clause did not check for this restriction. This patch fills that
gap.

gcc/ChangeLog:

* omp-low.cc (omp_maybe_offloaded_ctx): New prototype.
(scan_sharing_clauses):  Check a restriction on allocate clause.

gcc/testsuite/ChangeLog:

* c-c++-common/gomp/allocate-2.c: Add tests.
* c-c++-common/gomp/allocate-8.c: New test.
* gfortran.dg/gomp/allocate-3.f90: Add tests.
* gcc.dg/gomp/pr104517.c: Update.

2 years agoUpdate gennews for GCC 12.
Jakub Jelinek [Fri, 6 May 2022 06:44:43 +0000 (08:44 +0200)]
Update gennews for GCC 12.

2022-05-06  Jakub Jelinek  <jakub@redhat.com>

* gennews (files): Add files for GCC 12.

2 years agoDaily bump.
GCC Administrator [Fri, 6 May 2022 00:16:26 +0000 (00:16 +0000)]
Daily bump.

2 years agolibgomp: Update docs to reflect Fortran support for non-rectangular loops
Sandra Loosemore [Thu, 5 May 2022 21:45:29 +0000 (14:45 -0700)]
libgomp: Update docs to reflect Fortran support for non-rectangular loops

libgomp/
* libgomp.texi (OpenMP 5.0): Feature is now fully supported.

2 years agoc++: wrong error with MVP and pushdecl [PR64679]
Marek Polacek [Tue, 3 May 2022 23:01:19 +0000 (19:01 -0400)]
c++: wrong error with MVP and pushdecl [PR64679]

This patch fixes the second half of 64679.  Here we issue a wrong
"redefinition of 'int x'" for the following:

  struct Bar {
    Bar(int, int, int);
  };

  int x = 1;
  Bar bar(int(x), int(x), int{x}); // #1

cp_parser_parameter_declaration_list does pushdecl every time it sees
a named parameter, so the second "int(x)" causes the error.  That's
premature, since this turns out to be a constructor call after the
third argument!

If the first parameter is parenthesized, we can't push until we've
established we're looking at a function declaration.  Therefore this
could be fixed by some kind of lookahead.  I thought about introducing a
lightweight variant of cp_parser_parameter_declaration_list that would
not have any side effects and would return as soon as it figures out
whether it's looking at a declaration or expression.  Since that would
require fairly nontrivial changes, I wanted something simpler.

Something like delaying the pushdecl until we've reached the ')'
following the parameter-declaration-clause.  But we must push the
parameters before processing a default argument, as in:

  Bar bar(int(a), int(b), int c = sizeof(a));  // valid

Moreover, this code should still be accepted

  Bar f(int(i), decltype(i) j = 42);

so this patch stashes parameters into a vector when parsing tentatively
only when pushdecl-ing a parameter would result in a clash and an error
about redefinition/redeclaration.  The stashed parameters are pushed at
the end of a parameter-declaration-clause if it's followed by a ')', so
that we still diagnose redefining a parameter.

PR c++/64679

gcc/cp/ChangeLog:

* parser.cc (cp_parser_parameter_declaration_clause): Maintain
a vector of parameters that haven't been pushed yet.  Push them at the
end of a valid parameter-declaration-clause.
(cp_parser_parameter_declaration_list): Take a new auto_vec parameter.
Do not pushdecl while parsing tentatively when pushdecl-ing a parameter
would result in a hard error.
(cp_parser_cache_defarg): Adjust the call to
cp_parser_parameter_declaration_list.

gcc/testsuite/ChangeLog:

* g++.dg/parse/ambig11.C: New test.
* g++.dg/parse/ambig12.C: New test.
* g++.dg/parse/ambig13.C: New test.
* g++.dg/parse/ambig14.C: New test.

2 years agolibsanitizer: cherry-pick commit b226894d475b from upstream
H.J. Lu [Thu, 5 May 2022 20:59:16 +0000 (13:59 -0700)]
libsanitizer: cherry-pick commit b226894d475b from upstream

cherry-pick:

b226894d475b [sanitizer] [sanitizer] Correct GetTls for x32

2 years agolibstdc++: Fixes for tests that fail with -fno-rtti
Jonathan Wakely [Thu, 5 May 2022 08:50:53 +0000 (09:50 +0100)]
libstdc++: Fixes for tests that fail with -fno-rtti

This disables a use of dynamic_cast that is not valid for -fno-rtti and
adjusts some tests so they don't FAIL with -fno-rtti. Some tests are
skipped completely, and others just make use of typeid conditional on
the __cpp_rtti macro. A couple of tests were using typeid to verify
typedefs denote the right type, which can be done at compile-time using
templates instead.

libstdc++-v3/ChangeLog:

* include/experimental/memory_resource [!__cpp_rtti]
(__resource_adaptor_imp::do_is_equal): Do not use dynamic_cast
when RTTI is disabled.
* testsuite/17_intro/freestanding.cc: Require RTTI.
* testsuite/18_support/exception/38732.cc: Likewise.
* testsuite/18_support/exception_ptr/rethrow_exception.cc:
Likewise.
* testsuite/18_support/nested_exception/68139.cc: Likewise.
* testsuite/18_support/nested_exception/rethrow_if_nested.cc:
Likewise.
* testsuite/18_support/type_info/103240.cc: Likewise.
* testsuite/18_support/type_info/fundamental.cc: Likewise.
* testsuite/18_support/type_info/hash_code.cc: Likewise.
* testsuite/20_util/any/assign/emplace.cc: Likewise.
* testsuite/20_util/any/cons/in_place.cc: Likewise.
* testsuite/20_util/any/misc/any_cast.cc: Likewise.
* testsuite/20_util/any/observers/type.cc: Likewise.
* testsuite/20_util/function/1.cc: Likewise.
* testsuite/20_util/function/2.cc: Likewise.
* testsuite/20_util/function/3.cc: Likewise.
* testsuite/20_util/function/4.cc: Likewise.
* testsuite/20_util/function/5.cc: Likewise.
* testsuite/20_util/function/6.cc: Likewise.
* testsuite/20_util/function/7.cc: Likewise.
* testsuite/20_util/function/8.cc: Likewise.
* testsuite/20_util/polymorphic_allocator/resource.cc: Likewise.
* testsuite/20_util/shared_ptr/casts/1.cc: Likewise.
* testsuite/20_util/shared_ptr/casts/rval.cc: Likewise.
* testsuite/20_util/shared_ptr/cons/unique_ptr_deleter_ref_2.cc:
Likewise.
* testsuite/20_util/shared_ptr/misc/get_deleter.cc: Likewise.
* testsuite/20_util/typeindex/comparison_operators.cc: Likewise.
* testsuite/20_util/typeindex/comparison_operators_c++20.cc:
Likewise.
* testsuite/20_util/typeindex/hash.cc: Likewise.
* testsuite/20_util/typeindex/hash_code.cc: Likewise.
* testsuite/20_util/typeindex/name.cc: Likewise.
* testsuite/22_locale/ctype/is/string/89728_neg.cc: Likewise.
* testsuite/22_locale/global_templates/standard_facet_hierarchies.cc:
Likewise.
* testsuite/22_locale/global_templates/user_facet_hierarchies.cc:
Likewise.
* testsuite/22_locale/locale/13630.cc: Check type without using
RTTI.
* testsuite/23_containers/array/requirements/non_default_constructible.cc:
Require RTTI.
* testsuite/27_io/basic_ostream/emit/1.cc: Likewise.
* testsuite/27_io/fpos/14320-1.cc: Check type without using RTTI.
* testsuite/27_io/fpos/mbstate_t/12065.cc: Require RTTI.
* testsuite/27_io/ios_base/failure/dual_abi.cc: Likewise.
* testsuite/experimental/any/misc/any_cast.cc: Likewise.
* testsuite/experimental/any/observers/type.cc: Likewise.
* testsuite/experimental/memory_resource/resource_adaptor.cc:
Likewise.
* testsuite/lib/libstdc++.exp (check_effective_target_rtti):
Define new proc.
* testsuite/tr1/3_function_objects/function/1.cc: Likewise.
* testsuite/tr1/3_function_objects/function/2.cc: Likewise.
* testsuite/tr1/3_function_objects/function/3.cc: Likewise.
* testsuite/tr1/3_function_objects/function/4.cc: Likewise.
* testsuite/tr1/3_function_objects/function/5.cc: Likewise.
* testsuite/tr1/3_function_objects/function/6.cc: Likewise.
* testsuite/tr1/3_function_objects/function/7.cc: Likewise.
* testsuite/tr1/3_function_objects/function/8.cc: Likewise.
* testsuite/tr2/bases/value.cc: Likewise.
* testsuite/tr2/direct_bases/value.cc: Likewise.
* testsuite/util/exception/safety.h [!__cpp_rtti]: Don't print
types without RTTI.

2 years ago[PATCH] i386: Cleanup -m32 usage in the testuite.
Uros Bizjak [Thu, 5 May 2022 19:42:12 +0000 (21:42 +0200)]
[PATCH] i386: Cleanup -m32 usage in the testuite.

Use conditional compilation for ia32 target istead.

2022-05-05  Uroš Bizjak  <ubizjak@gmail.com>

gcc/testsuite/ChangeLog:

* gcc.target/i386/pr103611-2.c (dg-do): Compile for target ia32.
(dg-options): Remove -m32.
* gcc.target/i386/pr105032.c (dg-do): Compile for taget ia32.
(dg-additional-options): Remove.
* gcc.target/i386/pr104732.c (dg-options): Remove -m32.
* gcc.target/i386/pr99753.c (dg-options): Ditto.

2 years agoFortran: Add support for OMP non-rectangular loops.
Sandra Loosemore [Thu, 5 May 2022 18:37:16 +0000 (11:37 -0700)]
Fortran: Add support for OMP non-rectangular loops.

This patch adds support for OMP 5.1 "canonical loop nest form" to the
Fortran front end, marks non-rectangular loops for processing
by the middle end, and implements missing checks in the gimplifier
for additional prohibitions on non-rectangular loops.

Note that the OMP spec also prohibits non-rectangular loops with the TILE
construct; that construct hasn't been implemented yet, so that error will
need to be filled in later.

gcc/fortran/
* gfortran.h (struct gfc_omp_clauses): Add non_rectangular bit.
* openmp.cc (is_outer_iteration_variable): New function.
(expr_is_invariant): New function.
(bound_expr_is_canonical): New function.
(resolve_omp_do): Replace existing non-rectangularity error with
check for canonical form and setting non_rectangular bit.
* trans-openmp.cc (gfc_trans_omp_do): Transfer non_rectangular
flag to generated tree structure.

gcc/
* gimplify.cc (gimplify_omp_for): Update messages for SCHEDULED
and ORDERED clause conflict errors.  Add check for GRAINSIZE and
NUM_TASKS on TASKLOOP.

gcc/testsuite/
* c-c++-common/gomp/loop-6.c (f3): New function to test TASKLOOP
diagnostics.
* gfortran.dg/gomp/collapse1.f90: Update expected messages.
* gfortran.dg/gomp/pr85313.f90: Remove dg-error on non-rectangular
loops that are now accepted.
* gfortran.dg/gomp/non-rectangular-loop.f90: New file.
* gfortran.dg/gomp/canonical-loop-1.f90: New file.
* gfortran.dg/gomp/canonical-loop-2.f90: New file.

2 years agoRegenerate gcc.pot
Joseph Myers [Thu, 5 May 2022 17:02:40 +0000 (17:02 +0000)]
Regenerate gcc.pot

* gcc.pot: Regenerate.

2 years agoRemove loop-incremented dead code.
Martin Liska [Tue, 12 Apr 2022 13:06:40 +0000 (15:06 +0200)]
Remove loop-incremented dead code.

gcc/ChangeLog:

* genautomata.cc (create_composed_state): Remove dead code.
* graphite-poly.cc (print_pdrs): Likewise.
* lto-wrapper.cc (run_gcc): Likewise.
* tree-switch-conversion.cc (switch_decision_tree::balance_case_nodes):
Likewise.

2 years agoprofile: Unify identifier names for profiling
Martin Liska [Tue, 29 Mar 2022 08:04:53 +0000 (10:04 +0200)]
profile: Unify identifier names for profiling

gcc/ChangeLog:

* tree-profile.cc (gimple_gen_ic_profiler): Prefix names with
PROF_*.
(gimple_gen_time_profiler): Likewise.

2 years agoRemove sanity checking in stream_out_histogram_value.
Martin Liska [Thu, 5 May 2022 12:44:16 +0000 (14:44 +0200)]
Remove sanity checking in stream_out_histogram_value.

gcc/ChangeLog:

* value-prof.cc (stream_out_histogram_value): Remove sanity
checking.

2 years agotree-optimization/104162 - CSE of &MEM[ptr].a[i] and ptr + CST
Richard Biener [Wed, 26 Jan 2022 14:34:54 +0000 (15:34 +0100)]
tree-optimization/104162 - CSE of &MEM[ptr].a[i] and ptr + CST

This adds the capability to value-numbering of treating complex
address expressions where the offset becomes invariant as equal
to a POINTER_PLUS_EXPR.  This restores CSE that is now prevented
by early lowering of &MEM[ptr + CST] to a POINTER_PLUS_EXPR.

Unfortunately this regresses gcc.dg/asan/pr99673.c again, so
the testcase is adjusted accordingly.

2022-01-26  Richard Biener  <rguenther@suse.de>

PR tree-optimization/104162
* tree-ssa-sccvn.cc (vn_reference_lookup): Handle
&MEM[_1 + 5].a[i] like a POINTER_PLUS_EXPR if the offset
becomes invariant.
(vn_reference_insert): Likewise.

* gcc.dg/tree-ssa/ssa-fre-99.c: New testcase.
* gcc.dg/asan/pr99673.c: Adjust.

2 years ago[Committed] PR testsuite/105486: Use "signed char" in gcc.dg/pr102950.c
Roger Sayle [Thu, 5 May 2022 12:30:27 +0000 (08:30 -0400)]
[Committed] PR testsuite/105486: Use "signed char" in gcc.dg/pr102950.c

Although the automated regression testing scripts for powerpc64 appear
to be somewhat garbled at the moment, they've correctly identified that
my new test case for pr102950.c is failing on powerpc64, as char by
default is unsigned on this target.  This patch tweaks the new testcase
by explicitly using "signed char" so that it's testing the intended EVRP
behaviour portably.  Committed as obvious.

2022-05-05  Roger Sayle  <roger@nextmovesoftware.com>

gcc/testsuite/ChangeLog
PR testsuite/105486
* gcc.dg/pr102950.c: Use explicit "signed char" in test case.

2 years agolibsanitizer: update LOCAL_PATCHES.
Martin Liska [Thu, 5 May 2022 11:25:16 +0000 (13:25 +0200)]
libsanitizer: update LOCAL_PATCHES.

libsanitizer/ChangeLog:

* LOCAL_PATCHES: Update.

2 years agolibsanitizer: Apply local patches
Martin Liska [Thu, 5 May 2022 10:52:44 +0000 (12:52 +0200)]
libsanitizer: Apply local patches

2 years agolibsanitizer: merge from master (75f9e83ace52773af65dcebca543005ec8a2705d).
Martin Liska [Thu, 5 May 2022 10:51:36 +0000 (12:51 +0200)]
libsanitizer: merge from master (75f9e83ace52773af65dcebca543005ec8a2705d).

2 years agoEmbed real_value into REAL_CST
Richard Biener [Wed, 27 Apr 2022 11:48:49 +0000 (13:48 +0200)]
Embed real_value into REAL_CST

The following removes the indirection to real_value from REAL_CST
which doesn't seem to serve any useful purpose.  Any sharing can
be achieved by sharing the actual REAL_CST (which is what usually
happens when copying trees) and sharing of real_value amongst
different REAL_CST doesn't happen as far as I can see and would
only lead to further issues like mismatching type and real_value.

2022-04-27  Richard Biener  <rguenther@suse.de>

* tree-core.h (tree_real_cst::real_cst_ptr): Remove pointer
to real_value field.
(tree_real_cst::value): Add real_value field.
* tree.h (TREE_REAL_CST_PTR): Adjust.
* tree.cc (build_real): Remove separate allocation.
* tree-streamer-in.cc (unpack_ts_real_cst_value_fields):
Likewise.

gcc/cp/
* module.cc (trees_in::core_vals): Remove separate allocation
for REAL_CST.

2 years agorewrite undefined overflow to defined in ifcombine
Richard Biener [Mon, 4 Apr 2022 09:29:14 +0000 (11:29 +0200)]
rewrite undefined overflow to defined in ifcombine

When we make stmts to execute unconditionally in ifcombine we have
to make sure to rewrite stmts that can invoke undefined behavior
on overflow into a form with defined overflow.  That's possible
for all but signed division for which we have to avoid the transform.

2022-04-04  Richard Biener  <rguenther@suse.de>

* tree-ssa-ifcombine.cc (bb_no_side_effects_p): Avoid executing
divisions with undefined overflow unconditionally.
(pass_tree_ifcombine::execute): Rewrite stmts with undefined
overflow to defined.

2 years agotestsuite/105486 - adjust testcase to avoid misaligned accesses
Richard Biener [Thu, 5 May 2022 08:41:08 +0000 (10:41 +0200)]
testsuite/105486 - adjust testcase to avoid misaligned accesses

This properly aligns data, increasing test coverage.

2022-05-05  Richard Biener  <rguenther@suse.de>

PR testsuite/105486
* gcc.dg/vect/bb-slp-pr104240.c: Align all data.

2 years agotree-optimization/105484 - VEC_SET and EH
Richard Biener [Thu, 5 May 2022 07:34:59 +0000 (09:34 +0200)]
tree-optimization/105484 - VEC_SET and EH

When the IL representation of VEC_SET is marked as throwing
(unnecessarily), we need to clean that when replacing it with
the .VEC_SET internal function call which cannot throw.

2022-05-05  Richard Biener  <rguenther@suse.de>

PR tree-optimization/105484
* gimple-isel.cc (gimple_expand_vec_set_expr): Clean EH, return
whether the CFG changed.
(gimple_expand_vec_exprs): When the CFG changed, clean it up.

* gcc.dg/torture/pr105484.c: New testcase.

2 years agotree-optimization/104595 - vectorization of COND_EXPR with bool load
Richard Biener [Mon, 21 Feb 2022 10:05:58 +0000 (11:05 +0100)]
tree-optimization/104595 - vectorization of COND_EXPR with bool load

The following fixes an omission in bool pattern detection that
makes it fail when check_bool_pattern fails for COND_EXPR.  That's
not what it should do, instead it should still pattern recog
to var != 0 even if no further adjustments to the def chain are
necessary when var is not a mask already.

2022-02-21  Richard Biener  <rguenther@suse.de>

PR tree-optimization/104595
* tree-vect-patterns.cc (vect_recog_bool_pattern): For
COND_EXPR do not fail if check_bool_pattern returns false.

* gcc.dg/vect/pr104595.c: New testcase.

2 years agoMAINTAINERS: Add myself as PowerPC port co-maintainer
Kewen Lin [Thu, 5 May 2022 05:14:31 +0000 (00:14 -0500)]
MAINTAINERS: Add myself as PowerPC port co-maintainer

Add myself as PowerPC port co-maintainer and to DCO section.

ChangeLog:

* MAINTAINERS: Add myself as PowerPC port co-maintainer and to DCO
section.

2 years agoDaily bump.
GCC Administrator [Thu, 5 May 2022 00:16:29 +0000 (00:16 +0000)]
Daily bump.

2 years agolibsanitizer: cherry-pick commit f52e365092aa from upstream
H.J. Lu [Wed, 4 May 2022 22:59:49 +0000 (15:59 -0700)]
libsanitizer: cherry-pick commit f52e365092aa from upstream

cherry-pick:

f52e365092aa [sanitizer] Use newfstatat for x32

2 years agoc++: alias CTAD refactoring [PR104470]
Jason Merrill [Tue, 26 Apr 2022 22:32:51 +0000 (18:32 -0400)]
c++: alias CTAD refactoring [PR104470]

In my previous PR104470 patch I added yet another place that needs to handle
dependent member rewriting for deduction guides; this patches centralizes
rewriting into maybe_dependent_member_ref.  tsubst_baselink still has its
own handling because that's simpler than teaching maybe_dependent_member_ref
about BASELINKs.

PR c++/104470

gcc/cp/ChangeLog:

* pt.cc (maybe_dependent_member_ref): Handle types.
(tsubst, tsubst_copy): Use it.
(tsubst_aggr_type, instantiate_alias_template): Don't handle
tf_dguide here.

2 years agoc++: ICE during aggr CTAD for member tmpl [PR105476]
Patrick Palka [Wed, 4 May 2022 21:08:08 +0000 (17:08 -0400)]
c++: ICE during aggr CTAD for member tmpl [PR105476]

Here we're crashing from maybe_aggr_guide ultimately because
processing_template_decl isn't set when partially instantiating the
guide's parameter list; this causes us to force completion of the
dependent type Visitor_functior<Fn>, which of course fails and results
in an unexpected error_mark_node (the instantation should always succeed).

PR c++/105476

gcc/cp/ChangeLog:

* pt.cc (maybe_aggr_guide): Set processing_template_decl when
partially instantiating the guide's parameter list.

gcc/testsuite/ChangeLog:

* g++.dg/cpp2a/class-deduction-aggr13.C: New test.
* g++.dg/cpp2a/class-deduction-aggr13a.C: New test.

2 years agoc-family: Tweak -Woverflow diagnostic
Marek Polacek [Wed, 30 Mar 2022 20:59:33 +0000 (16:59 -0400)]
c-family: Tweak -Woverflow diagnostic

When g++ emits

warning: overflow in conversion from 'int' to 'char' changes value from '300' to '',''

for code like "char c = 300;" it might raise a few eyebrows.  With this
warning we're not interested in the ASCII representation of the char, only
the numerical value, so convert constants of type char to int.  It looks
like this conversion only needs to be done for char_type_node.

gcc/c-family/ChangeLog:

* c-warn.cc (warnings_for_convert_and_check): Convert constants of type
char to int.

gcc/testsuite/ChangeLog:

* c-c++-common/Wconversion-1.c: New test.

2 years agoc++: wrong parse with functors [PR64679]
Marek Polacek [Fri, 29 Apr 2022 19:01:12 +0000 (15:01 -0400)]
c++: wrong parse with functors [PR64679]

Consider

  struct F {
    F(int) {}
    F operator()(int) const { return *this; }
  };

and

  F(i)(0)(0);

where we're supposed to first call the constructor and then invoke
the operator() twice.  However, we parse this as an init-declarator:
"(i)" looks like a perfectly valid declarator, then we see an '(' and
think it must be an initializer, so we commit and we're toast.  My
fix is to look a little bit farther before deciding we've seen an
initializer.

This is only a half of c++/64679, the other part of the PR is unrelated:
there the problem is that we are calling pushdecl while parsing
tentatively (in cp_parser_parameter_declaration_list), which is bad.

PR c++/64679

gcc/cp/ChangeLog:

* parser.cc (cp_parser_init_declarator): Properly handle a series of
operator() calls, they are not part of an init-declarator.

gcc/testsuite/ChangeLog:

* g++.dg/parse/functor1.C: New test.

2 years agoc++: optimize reshape_init
Jason Merrill [Mon, 21 Mar 2022 13:58:28 +0000 (09:58 -0400)]
c++: optimize reshape_init

If the index of a constructor_elt is a FIELD_DECL, the CONSTRUCTOR is
already reshaped, so we can save time and memory by returning immediately.

gcc/cp/ChangeLog:

* decl.cc (reshape_init): Shortcut already-reshaped init.
 (reshape_init_class): Assert not getting one here.

2 years agoc++: use %D more
Jason Merrill [Fri, 18 Jun 2021 09:45:02 +0000 (05:45 -0400)]
c++: use %D more

There's no reason to call cxx_printable_name_translate here instead of using
%D in the format string.

gcc/cp/ChangeLog:

* error.cc (cp_print_error_function): Use %qD.
(function_category): Use %qD.

2 years agolibgomp/plugin/plugin-gcn.c: Use -foffload-options= in err msg
Tobias Burnus [Wed, 4 May 2022 16:39:28 +0000 (18:39 +0200)]
libgomp/plugin/plugin-gcn.c: Use -foffload-options= in err msg

While -foffload=-<flag> works (never documented legacy feature),
the documented way is to use -foffload-options=.

libgomp/ChangeLog:

* plugin/plugin-gcn.c (isa_matches_agent): Suggest -foffload-options.

2 years agoUpdate cpplib es.po
Joseph Myers [Wed, 4 May 2022 16:33:35 +0000 (16:33 +0000)]
Update cpplib es.po

* es.po: Update.

2 years agoOpenMP: Fix use_device_{addr,ptr} with in-data-sharing arg
Tobias Burnus [Wed, 4 May 2022 16:18:44 +0000 (18:18 +0200)]
OpenMP: Fix use_device_{addr,ptr} with in-data-sharing arg

For array-descriptor vars, the descriptor is assigned to a temporary. However,
this failed when the clause's argument was in turn in a data-sharing clause
as the outer context's VALUE_EXPR wasn't used.

gcc/ChangeLog:

* omp-low.cc (lower_omp_target): Fix use_device_{addr,ptr} with list
item that is in an outer data-sharing clause.

libgomp/ChangeLog:

* testsuite/libgomp.fortran/use_device_addr-5.f90: New test.

2 years agoc++: parse error with >= in template argument list [PR105436]
Marek Polacek [Fri, 29 Apr 2022 21:03:41 +0000 (17:03 -0400)]
c++: parse error with >= in template argument list [PR105436]

This patch fixes an oversight whereby we treated >= as the end of
a template argument.  This causes problems in C++14, because in
cp_parser_template_argument we go different ways for C++14 and C++17:

  /* It must be a non-type argument.  In C++17 any constant-expression is
     allowed.  */
  if (cxx_dialect > cxx14)
    goto general_expr;

so in this testcase in C++14 we get "N" as the template argument but in
C++17 it is the whole "N >= 5" expression.  So in C++14 the remaining
">= 5" triggered the newly-added diagnostic.

PR c++/105436

gcc/cp/ChangeLog:

* parser.cc (cp_parser_next_token_ends_template_argument_p): Don't
return true for CPP_GREATER_EQ.

gcc/testsuite/ChangeLog:

* g++.dg/parse/template31.C: New test.

2 years agolibstdc++: Add always_inline to the simplest std::array accessors [PR104719]
Jonathan Wakely [Wed, 27 Apr 2022 15:09:06 +0000 (16:09 +0100)]
libstdc++: Add always_inline to the simplest std::array accessors [PR104719]

libstdc++-v3/ChangeLog:

PR libstdc++/104719
* include/std/array (array::size(), array::max_size())
(array::empty(), array::data()): Add  always_inline attribute.

2 years agolibstdc++: Simplify std::array accessors [PR104719]
Jonathan Wakely [Fri, 25 Mar 2022 10:28:28 +0000 (10:28 +0000)]
libstdc++: Simplify std::array accessors [PR104719]

This removes the __array_traits::_S_ref and __array_traits::_S_ptr
accessors, which only exist to make the special case of std::array<T, 0>
syntactically well-formed.

By changing the empty type used as the std::array<T, 0>::_M_elems data
member to support operator[] and conversion to a pointer, we can write
code using the natural syntax. The indirection through _S_ref and
_S_ptr is removed for the common case, and a function call is only used
for the special case of zero-size arrays.

The invalid member access for zero-sized arrays is changed to use
__builtin_trap() instead of a null dereference. This guarantees a
runtime error if it ever gets called, instead of undefined behaviour
that is likely to get optimized out as unreachable.

libstdc++-v3/ChangeLog:

PR libstdc++/104719
* include/std/array (__array_traits::_S_ref): Remove.
(__array_traits::_S_ptr): Remove.
(__array_traits<T, 0>::_Type): Define operator[] and operator T*
to provide an array-like API.
(array::_AT_Type): Remove public typeef.
(array::operator[], array::at, array::front, array::back): Use
index operator to access _M_elems instead of _S_ref.
(array::data): Use implicit conversion from _M_elems to pointer.
(swap(array&, array&)): Use __enable_if_t helper.
(get<I>): Use index operator to access _M_elems.
* testsuite/23_containers/array/tuple_interface/get_neg.cc:
Adjust dg-error line numbers.

2 years agoc++: Remove cdtor_label
Jason Merrill [Fri, 25 Feb 2022 14:58:47 +0000 (09:58 -0500)]
c++: Remove cdtor_label

Jakub pointed out that cdtor_label is unnecessary, we should get all the
desired semantics with a normal return.

gcc/cp/ChangeLog:

* cp-tree.h (struct language_function): Remove x_cdtor_label.
(cdtor_label, LABEL_DECL_CDTOR): Remove.
* constexpr.cc (returns): Don't check LABEL_DECL_CDTOR.
(cxx_eval_constant_expression): Don't call returns.
* decl.cc (check_goto): Don't check cdtor_label.
(start_preparsed_function): And don't set it.
(finish_constructor_body, finish_destructor_body): Remove.
(finish_function_body): Don't call them.
* typeck.cc (check_return_expr): Handle cdtor_returns_this here.
* semantics.cc (finish_return_stmt): Not here.

2 years agotree-optimization/104658 - avoid mixing mask & non-mask vector defs
Richard Biener [Tue, 22 Feb 2022 15:02:27 +0000 (16:02 +0100)]
tree-optimization/104658 - avoid mixing mask & non-mask vector defs

When pattern recognition fails to sanitize all defs of a mask
producing operation and the respective def is external or constant
we end up trying to produce a VECTOR_BOOLEAN_TYPE_P constructor
which in turn ends up exposing stmts like

  <signed-boolean:1> _135 = _49 ? -1 : 0;

which isn't handled well in followup SLP and generates awful code.

We do rely heavily on pattern recognition to sanitize mask vs.
data uses of bools but that fails here which means we also should
fail vectorization.  That avoids ICEing because of such stmts
and it also avoids generating weird code which makes the
vectorization not profitable.

The following patch simply disallows external VECTOR_BOOLEAN_TYPE_P
defs and arranges the promote to external code to instead promote
mask uses to extern (that's just a short-cut here).

I've also looked at aarch64 and with SVE and a fixed vector length
for the gcc.target/i386/pr101636.c testcase.  I see similar vectorization
(using <signed-boolean:4>) there but it's hard to decide whether the
old, the new or no vectorization is better for this.  The code
generated with traditional integer masks isn't as awkward but we
still get the != 0 promotion done for each scalar element which
doesn't look like intended - this operation should be visible upfront.

That also means some cases will now become a missed optimization
that needs to be fixed by bool pattern recognition.  But that can
possibly be delayed to GCC 13.

2022-02-22  Richard Biener  <rguenther@suse.de>

PR tree-optimization/104658
* tree-vect-slp.cc (vect_slp_convert_to_external): Do not
create VECTOR_BOOLEAN_TYPE_P extern defs.  Reset the vector
type on nodes we promote.
(vectorizable_bb_reduc_epilogue): Deal with externalized
root.
* tree-vect-stmts.cc (vect_maybe_update_slp_op_vectype): Do
not allow VECTOR_BOOLEAN_TYPE_P extern defs.

* gcc.target/i386/pr104658.c: New testcase.

2 years agotree-optimization/103116 - SLP permutes and peeling for gaps
Richard Biener [Wed, 4 May 2022 08:43:07 +0000 (10:43 +0200)]
tree-optimization/103116 - SLP permutes and peeling for gaps

The testcase shows that we can end up with a contiguous access across
loop iterations but by means of permutations the elements accessed
might only cover parts of a vector.  In this case we end up with
GROUP_GAP == 0 but still need to avoid accessing excess elements
in the last loop iterations.  Peeling for gaps is designed to cover
this but a single scalar iteration might not cover all of the excess
elements.  The following ensures peeling for gaps is done in this
situation and when that isn't sufficient because we need to peel
more than one iteration (gcc.dg/vect/pr103116-2.c), fail the SLP
vectorization.

2022-05-04  Richard Biener  <rguenther@suse.de>

PR tree-optimization/103116
* tree-vect-stmts.cc (get_group_load_store_type): Handle the
case we need peeling for gaps even though GROUP_GAP is zero.

* gcc.dg/vect/pr103116-1.c: New testcase.
* gcc.dg/vect/pr103116-2.c: Likewise.

2 years agoRemove dead code.
Martin Liska [Wed, 4 May 2022 12:48:53 +0000 (14:48 +0200)]
Remove dead code.

gcc/ChangeLog:

* gengtype-state.cc (read_a_state_token): Remove dead code.
* ipa-profile.cc (ipa_profile_read_summary_section): Likewise.

2 years agoFold more vector constants early
Richard Biener [Fri, 25 Mar 2022 07:43:45 +0000 (08:43 +0100)]
Fold more vector constants early

In PR105049 we had

  return VIEW_CONVERT_EXPR<U>( VEC_PERM_EXPR < {<<< Unknown tree: compound_literal_expr
        V D.1984 = { 0 }; >>>, { 0 }} , {<<< Unknown tree: compound_literal_expr
        V D.1985 = { 0 }; >>>, { 0 }} , { 0, 0 } >  & {(short int) SAVE_EXPR <c>, (short int) SAVE_EXPR <c>});

where we gimplify the init CTORs to

  _1 = {{ 0 }, { 0 }};
  _2 = {{ 0 }, { 0 }};

instead of to vector constants.  The following makes sure to simplify the
CTORs to VECTOR_CSTs during gimplification by re-ordering the simplification
to after CTOR flag recomputation and gimplification of the elements.

2022-03-25  Richard Biener  <rguenther@suse.de>

* gimplify.cc (gimplify_init_constructor): First gimplify,
then simplify the result to a VECTOR_CST.

2 years agogenconditions: Add support for targets without non-trivial insn conditions
Jakub Jelinek [Wed, 4 May 2022 10:01:56 +0000 (12:01 +0200)]
genconditions: Add support for targets without non-trivial insn conditions

Somebody complained on IRC that when writing a new backend one can get
an error while compiling build/gencondmd.cc.
The problem is that when host compiler is g++ 3 or later (or when
bootstrapping), we compile it with g++ -std=c++11 -pedantic and
the generated insn_conditions array contains pairs
{ "cond", __builtin_constant_p (cond) ? (int) (cond) : -1 },
where cond is some non-trivial instruction condition.  Now if a target
uses "" for all the conditions (admittedly unlikely for non-trivial
target), the initializer for insn_conditions[] is {} and that is
pedantically rejected because C++ doesn't support zero-sized arrays.

The following patch fixes that by adding an artificial termination
element and skips that during the walk.

2022-05-04  Jakub Jelinek  <jakub@redhat.com>

* genconditions.cc (write_conditions): Append a { nullptr, -1 }
element at the end of insn_conditions.
(write_writer): Use ARRAY_SIZE (insn_conditions) - 1 instead of
ARRAY_SIZE (insn_conditions).

2 years agolibsanitizer: update test that mixes fake and real stack
Martin Liska [Mon, 2 May 2022 11:58:49 +0000 (13:58 +0200)]
libsanitizer: update test that mixes fake and real stack

gcc/testsuite/ChangeLog:

* c-c++-common/asan/alloca_loop_unpoisoning.c: Do not combine
fake and real stack.

2 years agolibsanitizer: Update Makefile.am files.
Martin Liska [Mon, 2 May 2022 11:20:06 +0000 (13:20 +0200)]
libsanitizer: Update Makefile.am files.

libsanitizer/ChangeLog:

* tsan/Makefile.am: Update Makefile.am files.
* hwasan/Makefile.am: Likewise.
* sanitizer_common/Makefile.am: Likewise.
* Makefile.in: Re-generate.
* asan/Makefile.in: Likewise.
* hwasan/Makefile.in: Likewise.
* interception/Makefile.in: Likewise.
* libbacktrace/Makefile.in: Likewise.
* lsan/Makefile.in: Likewise.
* sanitizer_common/Makefile.in: Likewise.
* tsan/Makefile.in: Likewise.
* ubsan/Makefile.in: Likewise.

2 years agolibsanitizer: Apply local patches
H.J. Lu [Tue, 20 Jul 2021 17:46:51 +0000 (10:46 -0700)]
libsanitizer: Apply local patches

2 years agolibsanitizer: merge from upstream (0a1bcab9f3bf75c4c5d3e53bafb3eeb80320af46).
Martin Liska [Tue, 3 May 2022 10:56:26 +0000 (12:56 +0200)]
libsanitizer: merge from upstream (0a1bcab9f3bf75c4c5d3e53bafb3eeb80320af46).

2 years agoarm: Restrict support of vectors of boolean immediates (PR target/104662)
Christophe Lyon [Tue, 19 Apr 2022 15:34:04 +0000 (17:34 +0200)]
arm: Restrict support of vectors of boolean immediates (PR target/104662)

This simple patch avoids the ICE described in the PR:
internal compiler error: in simd_valid_immediate, at config/arm/arm.cc:12866

with an early exit from simd_valid_immediate if we are trying to
handle a vector of booleans and MVE is not enabled.

We still get an ICE when compiling the existing
gcc.dg/rtl/arm/mve-vxbi.c without -march=armv8.1-m.main+mve:

error: unrecognizable insn:
(insn 7 5 8 2 (set (reg:V4BI 114)
        (const_vector:V4BI [
                (const_int 1 [0x1])
                (const_int 0 [0]) repeated x2
                (const_int 1 [0x1])
            ])) -1
     (nil))
during RTL pass: ira

but there's little we can do since the testcase explicitly creates
vectors of booleans which do need MVE.

That is the reason why I do not add a testcase.

2022-04-19  Christophe Lyon  <christophe.lyon@arm.com>

PR target/104662
* config/arm/arm.cc (simd_valid_immediate): Exit when input is a
vector of booleans and MVE is not enabled.

2 years agoc++: Don't emit deprecated warnings or unavailable errors on lambda declarations
Jakub Jelinek [Wed, 4 May 2022 08:07:36 +0000 (10:07 +0200)]
c++: Don't emit deprecated warnings or unavailable errors on lambda declarations

On the following testcase, we emit deprecated warnings or unavailable errors
even on merge declarations of those lambdas (the dg-bogus directives), while
IMHO we should emit them only when something actually calls those lambdas.

The following patch temporarily disables that diagnostics during
maybe_add_lambda_conv_op.

PR2173R1 also says that ambiguity between attribute-specifier-seq at the
end of requires-clause and attribute-specifier-seq from lambda-expression
should be resolved to attribute-specifier-seq for the latter.  Do we need
to do anything about that?  I mean, can a valid requires-clause end with
an attribute-specifier-seq?  Say operator int [[]] is valid primary
expression, but requires operator int [[]] isn't valid, nor is
requires operator int, no?

2022-05-04  Jakub Jelinek  <jakub@redhat.com>

* lambda.cc: Include decl.h.
(maybe_add_lambda_conv_op): Temporarily override deprecated_state to
UNAVAILABLE_DEPRECATED_SUPPRESS.

* g++.dg/cpp23/lambda-attr1.C: New test.
* g++.dg/cpp23/lambda-attr2.C: New test.

2 years agohwasan: support new dg-output format.
Martin Liska [Wed, 4 May 2022 07:57:17 +0000 (09:57 +0200)]
hwasan: support new dg-output format.

Supports change in libsanitizer where it newly reports:
READ of size 4 at 0xffffffffc3d4 tags: 02/01(00) (ptr/mem) in thread T0

So the 'tags' contains now 3 entries compared to 2 entries.

gcc/testsuite/ChangeLog:

* c-c++-common/hwasan/alloca-outside-caught.c: Update dg-output.
* c-c++-common/hwasan/heap-overflow.c: Likewise.
* c-c++-common/hwasan/hwasan-thread-access-parent.c: Likewise.
* c-c++-common/hwasan/large-aligned-1.c: Likewise.
* c-c++-common/hwasan/stack-tagging-basic-1.c: Likewise.

2 years agodebug/105158 - improve debug stmt retaining for forwarder removal
Richard Biener [Tue, 5 Apr 2022 13:23:54 +0000 (15:23 +0200)]
debug/105158 - improve debug stmt retaining for forwarder removal

Currently when we cannot move debug stmt from a forwarder to the
destination block we drop/reset them.  But in some cases as for
the testcase we can move them to the predecessor when that has
a single successor and we can insert after the last stmt of the
block.  That allows us to preserve debug info here.

2022-04-05  Richard Biener  <rguenther@suse.de>

PR debug/105158
* tree-cfgcleanup.cc (move_debug_stmts_from_forwarder):
Move debug stmts to the predecessor if moving to the
destination is not possible.
(remove_forwarder_block): Adjust.
(remove_forwarder_block_with_phi): Likewise.

2 years agoLTO plugin: modernize a bit.
Martin Liska [Wed, 5 Jan 2022 09:10:19 +0000 (10:10 +0100)]
LTO plugin: modernize a bit.

include/ChangeLog:

* plugin-api.h (enum ld_plugin_tag): Do not set implicit enum
values.

lto-plugin/ChangeLog:

* lto-plugin.c (struct plugin_objfile): Use bool for offset
field.
(exec_lto_wrapper): Assign true/false to bool variables.
(process_offload_section): Likewise.
(claim_file_handler): Likewise.
(onload): Likewise.

2 years agoDaily bump.
GCC Administrator [Wed, 4 May 2022 00:16:24 +0000 (00:16 +0000)]
Daily bump.

2 years agoUpdate gcc sv.po
Joseph Myers [Tue, 3 May 2022 22:16:26 +0000 (22:16 +0000)]
Update gcc sv.po

* sv.po: Update.

2 years agoc++: make finish_non_static_data_member SFINAE enabled [PR105351]
Patrick Palka [Tue, 3 May 2022 19:21:26 +0000 (15:21 -0400)]
c++: make finish_non_static_data_member SFINAE enabled [PR105351]

Here since finish_non_static_data_member isn't SFINAE enabled, we
incorrectly emit an error when considering the first overload rather
than silently discarding it:

sfinae33.C: In substitution of ‘template<class T> A<T::value> f() [with T = B]’:
sfinae33.C:11:7:   required from here
sfinae33.C:5:31: error: invalid use of non-static data member ‘B::value’
    5 | template<class T> A<T::value> f();
      |                               ^

This patch makes the function SFINAE enabled in the usual way: give it a
complain parameter, check it before emitting an error, and pass it through
appropriately.

PR c++/105351

gcc/cp/ChangeLog:

* cp-tree.h (finish_non_static_data_member): Add defaulted
complain parameter.
* pt.cc (tsubst_copy_and_build): Pass complain to
finish_non_static_data_member.
* semantics.cc (finish_non_static_data_member): Respect complain
parameter.
(finish_qualified_id_expr): Pass complain to
finish_non_static_data_member.

gcc/testsuite/ChangeLog:

* g++.dg/template/sfinae33.C: New test.

2 years agotestsuite: libgcc function name for PRU
Dimitar Dimitrov [Sat, 2 Apr 2022 09:13:16 +0000 (12:13 +0300)]
testsuite: libgcc function name for PRU

Update the match rules to accommodate the non-standard libgcc function
names for PRU backend.

gcc/testsuite/ChangeLog:

* gcc.c-torture/compile/attr-complex-method-2.c: Accept both __divdc3
and __gnu_divdc3 as valid libgcc function names.
* gcc.dg/complex-6.c: Ditto for __mulsc3.
* gcc.dg/complex-7.c: Ditto for __muldc3.

Signed-off-by: Dimitar Dimitrov <dimitar@dinux.eu>
2 years agotestsuite: Skip cases for default_packed targets
Dimitar Dimitrov [Fri, 15 Apr 2022 19:58:50 +0000 (22:58 +0300)]
testsuite: Skip cases for default_packed targets

The memchr test cases expect padding to be present in structures.  But
this is not true for targets which pack by default.  Skip these test
cases in order to avoid static assert errors when checking field offsets.

gcc/testsuite/ChangeLog:

* gcc.dg/memchr.c: Skip for default_packed targets.
* gcc.dg/memcmp-3.c: Ditto.

Signed-off-by: Dimitar Dimitrov <dimitar@dinux.eu>
2 years agotestsuite: Annotate Wattributes-8.c for default_packed
Dimitar Dimitrov [Sun, 3 Apr 2022 09:27:16 +0000 (12:27 +0300)]
testsuite: Annotate Wattributes-8.c for default_packed

Place markers in test case to handle targets which pack structures by
default. Validated on pru-none-elf.

gcc/testsuite/ChangeLog:

* gcc.dg/Wattributes-8.c: Add annotations for default_packed
targets.

Signed-off-by: Dimitar Dimitrov <dimitar@dinux.eu>
2 years agotestsuite: Mark that PRU lowers DI alu ops by itself
Dimitar Dimitrov [Sun, 3 Apr 2022 10:24:52 +0000 (13:24 +0300)]
testsuite: Mark that PRU lowers DI alu ops by itself

PRU target defines DI patterns for logical ALU operations.

gcc/testsuite/ChangeLog:

* gcc.dg/lower-subreg-1.c: Skip for PRU.

Signed-off-by: Dimitar Dimitrov <dimitar@dinux.eu>
2 years agotestsuite: Skip gcc.dg/Wno-frame-address.c for PRU
Dimitar Dimitrov [Sun, 3 Apr 2022 13:11:27 +0000 (16:11 +0300)]
testsuite: Skip gcc.dg/Wno-frame-address.c for PRU

Access to arbitrary stack frames is not supported on PRU.

gcc/testsuite/ChangeLog:

* gcc.dg/Wno-frame-address.c: Skip for PRU target.

Signed-off-by: Dimitar Dimitrov <dimitar@dinux.eu>
2 years agoPR tree-optimization/102950: Improved EVRP for signed BIT_XOR_EXPR.
Roger Sayle [Tue, 3 May 2022 18:38:50 +0000 (14:38 -0400)]
PR tree-optimization/102950: Improved EVRP for signed BIT_XOR_EXPR.

This patch fixes PR tree-optimization/102950, which is a P2 regression,
by providing better range bounds for BIT_XOR_EXPR, BIT_AND_EXPR and
BIT_IOR_EXPR on signed integer types.  In general terms, any binary
bitwise operation on sign-extended or zero-extended integer types will
produce results that are themselves sign-extended or zero-extended.
More precisely, we can derive signed bounds from the number of leading
redundant sign bit copies, from the equation:
clrsb(X op Y) >= min (clrsb (X), clrsb(Y))
and from the property that for any (signed or unsigned) range [lb, ub]
that clrsb([lb, ub]) >= min (clrsb(lb), clrsb(ub)).

These can be used to show that [-1, 0] op [-1, 0] is [-1, 0] or that
[-128, 127] op [-128, 127] is [-128, 127], even when tracking nonzero
bits would result in VARYING (as every bit can be 0 or 1).  This is
equivalent to determining the minimum type precision in which the
operation can be performed then sign extending the result.

One additional refinement is to observe that X ^ Y can never be
zero if the ranges of X and Y don't overlap, i.e. X can't be equal
to Y.

Previously, the expression "(int)(char)a ^ 233" in the PR was considered
VARYING, but with the above changes now has the range [-256, -1][1, 255],
which is sufficient to optimize away the call to foo.

2022-05-03  Roger Sayle  <roger@nextmovesoftware.com>

gcc/ChangeLog
PR tree-optimization/102950
* range-op.cc (wi_optimize_signed_bitwise_op): New function to
determine bounds of bitwise operations on signed types.
(operator_bitwise_and::wi_fold): Call the above function.
(operator_bitwise_or::wi_fold): Likewise.
(operator_bitwise_xor::wi_fold): Likewise.  Additionally, the
result can't be zero if the operands can't be equal.

gcc/testsuite/ChangeLog
PR tree-optimization/102950
* gcc.dg/pr102950.c: New test case.
* gcc.dg/tree-ssa/evrp10.c: New test case.

2 years agoObjective-C, NeXT: Adjust symbol marking to match host tools.
Iain Sandoe [Mon, 2 May 2022 18:42:49 +0000 (19:42 +0100)]
Objective-C, NeXT: Adjust symbol marking to match host tools.

Current host tools mark some additional symbols as 'no dead strip' and also
expose one additional group to the linker.  This does not affect older Darwin
versions or x86_64, but omitting these changes results in link errors for
aarch64.

Signed-off-by: Iain Sandoe <iain@sandoe.co.uk>
gcc/ChangeLog:

* config/darwin.cc (darwin_label_is_anonymous_local_objc_name): Make
protocol class methods linker-visible.

gcc/objc/ChangeLog:

* objc-next-runtime-abi-02.cc (next_runtime_abi_02_protocol_decl): Do
not dead-strip the runtime meta-data symbols.
(build_v2_classrefs_table): Likewise.
(build_v2_protocol_list_address_table): Likewise.

2 years ago[PR105324] libstdc++: testsuite: pr105324 requires FP from_char
Alexandre Oliva [Tue, 3 May 2022 17:57:15 +0000 (14:57 -0300)]
[PR105324] libstdc++: testsuite: pr105324 requires FP from_char

The floating-point overloads of from_char are only declared if
_GLIBCXX_HAVE_USELOCALE is #defined as nonzero.  That's exposed from
charconv as __cpp_lib_to_chars >= 201611L, so guard the test body with
that.

for  libstdc++-v3/ChangeLog

PR c++/105324
* testsuite/20_util/from_chars/pr105324.cc: Guard test body
with conditional for floating-point overloads of from_char.

2 years agoi386: Optimize _mm_storeu_si16 w/o SSE4 [PR105079]
Uros Bizjak [Tue, 3 May 2022 15:59:40 +0000 (17:59 +0200)]
i386: Optimize _mm_storeu_si16 w/o SSE4 [PR105079]

Optimize _mm_storeu_si16 to use MOVD from a SSE to an integer register
instead of PEXTRW from a low word of the SSE register to an integer reg.

Avoid the transformation when optimizing for size for targets without
TARGET_INTER_UNIT_MOVES_FROM_VEC capability, where the transformation
results in two moves via a memory location.

2022-05-03  Uroš Bizjak  <ubizjak@gmail.com>

gcc/ChangeLog:

PR target/105079
* config/i386/sse.md (*vec_extract<mode>_0_mem): New pre-reload
define_insn_and_split pattern.

gcc/testsuite/ChangeLog:

PR target/105079
* gcc.target/i386/pr105079.c: New test.
* gcc.target/i386/pr95483-1.c (dg-options): Use -msse4.1.

2 years agodfp.m4: Add missing license header
Christophe Lyon [Fri, 29 Apr 2022 13:11:11 +0000 (14:11 +0100)]
dfp.m4: Add missing license header

It seems the license header was omitted when dfp.m4 was originally
contributed in 2010 (commit 3c39bca6bbb5, r0-102573 or svn r163815.

This copies the license from libdecnumber/configure.ac since dfp.m4
was originally extracted from that file.

2022-04-29  Christophe Lyon  <christophe.lyon@arm.com>

config/
* dfp.m4: Add license header.

2 years agomiddle-end/105083 - check prerequesites in scev_initialize
Richard Biener [Tue, 3 May 2022 10:48:43 +0000 (12:48 +0200)]
middle-end/105083 - check prerequesites in scev_initialize

2022-05-03  Richard Biener  <rguenther@suse.de>

PR middle-end/105083
* tree-scalar-evolution.cc (scev_initialize): Verify we
have appropriate loop state.
* tree-ssa-dce.cc (perform_tree_ssa_dce): Re-order SCEV and
loop init and finalization.

2 years agomiddle-end/105461 - opts processing of -fvar-tracking
Richard Biener [Tue, 3 May 2022 08:36:30 +0000 (10:36 +0200)]
middle-end/105461 - opts processing of -fvar-tracking

The flag_var_tracking reset in finish_options doesn't match the
condition in process_options, in particular we fail to reset it
when the option was specified on the command line.  The following
fixes this and also alters the debug info level guard to match
the one in process_options.

2022-05-03  Richard Biener  <rguenther@suse.de>

PR middle-end/105461
* opts.cc (finish_options): Match the condition to
disable flag_var_tracking to that of process_options.

* gcc.dg/pr105461.c: New testcase.

2 years agoFixup OPTION_SET_P usage in finish_options
Richard Biener [Tue, 3 May 2022 08:23:51 +0000 (10:23 +0200)]
Fixup OPTION_SET_P usage in finish_options

When some code was moved from process_options to finish_options,
uses of OPTION_SET_P were not replaced with references to the
opts_set option set.  The following fixes this.

2022-05-03  Richard Biener  <rguenther@suse.de>

* opts.cc: #undef OPTIONS_SET_P.
(finish_options): Use opts_set instead of OPTIONS_SET_P.

2 years agotree-optimization/105394 - vector lowering of compares
Richard Biener [Fri, 29 Apr 2022 09:54:45 +0000 (11:54 +0200)]
tree-optimization/105394 - vector lowering of compares

The following fixes missing handling of non-integer mode but
masked (SVE or MVE) compares in vector lowering by using the
appropriate mask element width to extract the components and
adjust the index.

2022-04-29  Richard Biener  <rguenther@suse.de>

PR tree-optimization/105394
* tree-vect-generic.cc (expand_vector_condition): Adjust
comp_width for non-integer mode masks as well.

2 years agoOpenMP, libgomp: Add new runtime routine omp_get_mapped_ptr.
Marcel Vollweiler [Tue, 3 May 2022 06:56:44 +0000 (23:56 -0700)]
OpenMP, libgomp: Add new runtime routine omp_get_mapped_ptr.

This patch adds the OpenMP runtime routine "omp_get_mapped_ptr" which was
introduced in OpenMP 5.1.

gcc/ChangeLog:

* omp-low.cc (omp_runtime_api_call): Added get_mapped_ptr to
omp_runtime_apis array.

libgomp/ChangeLog:

* libgomp.map: Added omp_get_mapped_ptr.
* libgomp.texi: Tagged omp_get_mapped_ptr as supported.
* omp.h.in: Added omp_get_mapped_ptr.
* omp_lib.f90.in: Added interface for omp_get_mapped_ptr.
* omp_lib.h.in: Likewise.
* target.c (omp_get_mapped_ptr): Added implementation of
omp_get_mapped_ptr.
* testsuite/libgomp.c-c++-common/get-mapped-ptr-1.c: New test.
* testsuite/libgomp.c-c++-common/get-mapped-ptr-2.c: New test.
* testsuite/libgomp.c-c++-common/get-mapped-ptr-3.c: New test.
* testsuite/libgomp.c-c++-common/get-mapped-ptr-4.c: New test.
* testsuite/libgomp.fortran/get-mapped-ptr-1.f90: New test.
* testsuite/libgomp.fortran/get-mapped-ptr-2.f90: New test.
* testsuite/libgomp.fortran/get-mapped-ptr-3.f90: New test.
* testsuite/libgomp.fortran/get-mapped-ptr-4.f90: New test.

2 years agotestsuite: vect: update unaligned message
Alexandre Oliva [Wed, 27 Apr 2022 21:23:01 +0000 (18:23 -0300)]
testsuite: vect: update unaligned message

gcc.dg/vect/costmodel/ppc/costmodel-vect-31a.c covers ppc variants
that accept and reject misaligned accesses.  The message that it
expects for rejection was removed in the gcc-11 development cycle by
commit r11-1969.  The patch adjusted multiple tests to use the message
introduced in r11-1945, but missed this one.

for  gcc/testsuite/ChangeLog

* gcc.dg/vect/costmodel/ppc/costmodel-vect-31a.c: Update
the expected message for the case in which unaligned accesses
are not allowed.

2 years agoDaily bump.
GCC Administrator [Tue, 3 May 2022 00:16:25 +0000 (00:16 +0000)]
Daily bump.

2 years agoc++: improve template-id location
Jason Merrill [Sat, 30 Apr 2022 09:45:02 +0000 (05:45 -0400)]
c++: improve template-id location

On PR102629 I noticed that we were giving the entire lambda as the location
for this template-id.

gcc/cp/ChangeLog:

* pt.cc (tsubst_copy_and_build) [TEMPLATE_ID_EXPR]: Copy location.
(do_auto_deduction): Use expr location.

gcc/testsuite/ChangeLog:

* g++.dg/cpp2a/lambda-pack-init7.C: Check column number.

2 years agoc++: also note name used in enclosing class
Jason Merrill [Thu, 31 Mar 2022 20:18:58 +0000 (16:18 -0400)]
c++: also note name used in enclosing class

While looking at PR96645 I noticed that while we were diagnosing names
changing meaning in the full class context, we weren't doing this for
lookups in nested class bodies.

Note that this breaks current range-v3; I've submitted a pull request to fix
its violation of the rule.

gcc/cp/ChangeLog:

* class.cc (maybe_note_name_used_in_class): Note in all enclosing
classes.  Remember location of use.
(note_name_declared_in_class): Adjust.

gcc/testsuite/ChangeLog:

* g++.dg/lookup/name-clash13.C: New test.
* g++.dg/lookup/name-clash14.C: New test.
* g++.dg/lookup/name-clash15.C: New test.
* g++.dg/lookup/name-clash16.C: New test.

2 years agox86: Add missing .note.GNU-stack to assembly source
H.J. Lu [Fri, 29 Apr 2022 17:26:24 +0000 (10:26 -0700)]
x86: Add missing .note.GNU-stack to assembly source

Add .note.GNU-stack assembly source to avoid linker warning:

ld: warning: /tmp/ccPZSZ7Z.o: missing .note.GNU-stack section implies executable stack
ld: NOTE: This behaviour is deprecated and will be removed in a future version of the linker
FAIL: gcc.target/i386/iamcu/test_3_element_struct_and_unions.c compilation,  -O0

PR testsuite/105433
* gcc.target/i386/iamcu/asm-support.S: Add .note.GNU-stack.
* gcc.target/x86_64/abi/asm-support.S: Likewise.
* gcc.target/x86_64/abi/avx/asm-support.S: Likewise.
* gcc.target/x86_64/abi/avx512f/asm-support.S: Likewise.
* gcc.target/x86_64/abi/avx512fp16/asm-support.S: Likewise.
* gcc.target/x86_64/abi/avx512fp16/m256h/asm-support.S: Likewise.
* gcc.target/x86_64/abi/avx512fp16/m512h/asm-support.S: Likewise.
* gcc.target/x86_64/abi/ms-sysv/do-test.S: Likewise.

2 years agoFortran: Add location info to OpenMP tree nodes
Sandra Loosemore [Fri, 25 Mar 2022 04:02:34 +0000 (21:02 -0700)]
Fortran: Add location info to OpenMP tree nodes

gcc/fortran/
* trans-openmp.cc (gfc_trans_omp_critical): Set location on OMP
tree node.
(gfc_trans_omp_do): Likewise.
(gfc_trans_omp_masked): Likewise.
(gfc_trans_omp_do_simd): Likewise.
(gfc_trans_omp_scope): Likewise.
(gfc_trans_omp_taskgroup): Likewise.
(gfc_trans_omp_taskwait): Likewise.
(gfc_trans_omp_distribute): Likewise.
(gfc_trans_omp_taskloop): Likewise.
(gfc_trans_omp_master_masked_taskloop): Likewise.

2 years agoc++: uses_template_parms cleanups
Marek Polacek [Wed, 27 Apr 2022 22:36:29 +0000 (18:36 -0400)]
c++: uses_template_parms cleanups

gcc/cp/ChangeLog:

* cp-tree.h (uses_template_parms): Adjust declaration.
* pt.cc (uses_template_parms): Return bool.  Use a RAII sentinel.

2 years agoAssume a call is expensive when it mismatches
Richard Biener [Wed, 13 Apr 2022 13:43:01 +0000 (15:43 +0200)]
Assume a call is expensive when it mismatches

This makes sure to not consider calls to builtin decls with
mismatching arguments as inexpensive.

2022-04-13  Richard Biener  <rguenther@suse.de>

* tree-scalar-evolution.cc (expression_expensive_p):
Never consider mismatched calls as cheap.

2 years agotree-optimization/104240 - SLP discovery with swapped comparisons
Richard Biener [Mon, 2 May 2022 10:01:51 +0000 (12:01 +0200)]
tree-optimization/104240 - SLP discovery with swapped comparisons

The following extends SLP discovery to handle swapped operands
in comparisons.

2022-05-02  Richard Biener  <rguenther@suse.de>

PR tree-optimization/104240
* tree-vect-slp.cc (op1_op0_map): New.
(vect_get_operand_map): Handle compares.
(vect_build_slp_tree_1): Support swapped operands for
tcc_comparison.

* gcc.dg/vect/bb-slp-pr104240.c: New testcase.

2 years agolibstdc++: Don't use std::tolower in <charconv> [PR103911]
Patrick Palka [Mon, 2 May 2022 11:01:33 +0000 (07:01 -0400)]
libstdc++: Don't use std::tolower in <charconv> [PR103911]

As with std::isdigit in r12-6281-gc83ecfbe74a5cf, we shouldn't be using
std::tolower in <charconv> either.

PR libstdc++/103911

libstdc++-v3/ChangeLog:

* src/c++17/floating_from_chars.cc (find_end_of_float): Accept
two delimeters for the exponent part in the form of a possibly
NULL string of length two.  Don't use std::tolower.
(pattern): Adjust calls to find_end_of_float accordingly.

2 years agolibstdc++: case-sensitivity in hexfloat std::from_chars [PR105441]
Patrick Palka [Mon, 2 May 2022 11:00:48 +0000 (07:00 -0400)]
libstdc++: case-sensitivity in hexfloat std::from_chars [PR105441]

The hexfloat parser for binary32/64 added in r12-6645-gcc3bf3404e4b1c
overlooked that the exponent part can also begin with an uppercase 'P'.

PR libstdc++/105441

libstdc++-v3/ChangeLog:

* src/c++17/floating_from_chars.cc (__floating_from_chars_hex):
Also accept 'P' as the start of the exponent.
* testsuite/20_util/from_chars/7.cc: Add corresponding testcase.

2 years agoexpand: Throw away non-external decls without varpool nodes [PR105415]
Jakub Jelinek [Mon, 2 May 2022 09:30:58 +0000 (11:30 +0200)]
expand: Throw away non-external decls without varpool nodes [PR105415]

The following testcase fails -fcompare-debug on aarch64-linux.  The problem
is that for the n variable we create a varpool node, then remove it again
because the var isn't really used, but it keeps being referenced in debug
stmts/insns with -g.  Later during sched1 pass we ask whether the n var
can be modified through some store to an anchored variable and with -g
we create a new varpool node for it again just so that we can find its
ultimate alias target.  Even later on we create some cgraph node for the
loop parallelization, but as there has been an extra varpool node creation
in between, we get higher node->order with -g than without.

The patch fixes that by throwing variables without varpool nodes away
during expansion time, they are very unlikely to actually end up with
useful debug info anyway.

I've bootstrapped/regtested the following on x86_64-linux and i686-linux,
then bootstrapped with the patch reverted, reapplied the patch and did make
cc1plus in stage3.  The debug section sizes are identical, .debug_info and
.debug_loc is identical too, so I think we don't lose any debug info through
it.
So at least on cc1plus it makes no difference.

2022-05-02  Jakub Jelinek  <jakub@redhat.com>

PR debug/105415
* cfgexpand.cc (expand_debug_expr): Don't make_decl_rtl_for_debug
if there is no symtab node for the VAR_DECL.

* gcc.dg/pr105415.c: New test.

2 years agogcov: Fix first time gcov info dump
Sebastian Huber [Mon, 2 May 2022 06:44:16 +0000 (08:44 +0200)]
gcov: Fix first time gcov info dump

This patch fixes an issue introduced by commit
ef9a53feae5701953da9161afef2aea0329ec8b2:

gcc --coverage main.c && ./a.out
libgcov profiling error:a-main.gcda:Error writing

gcc/ChangeLog:

* gcov-io.cc (gcov_rewrite):  Clear the file error status.

2 years agotree-optimization/105437 - BB vect with extern defs of throwing stmts
Richard Biener [Mon, 2 May 2022 07:30:27 +0000 (09:30 +0200)]
tree-optimization/105437 - BB vect with extern defs of throwing stmts

We have to watch out for vectorized stmt insert locations if the
def from the last stmt alters control flow.  We constrain region
building so we know the def is outside of the current region
and thus we can insert at the region start point.

2022-05-02  Richard Biener  <rguenther@suse.de>

PR tree-optimization/105437
* tree-vect-slp.cc (vect_schedule_slp_node): Handle the
case where last_stmt alters control flow.

* g++.dg/vect/pr105437.cc: New testcase.

2 years agoUse CASE_CONVERT in a few more cases
Richard Biener [Wed, 12 Jan 2022 09:42:32 +0000 (10:42 +0100)]
Use CASE_CONVERT in a few more cases

This uses CASE_CONVERT more which eases eventual removal of NOP_EXPR.

2022-04-29  Richard Biener  <rguenther@suse.de>

gcc/cp/
* constexpr.cc (fold_simple_1): Use CASE_CONVERT.
* cp-gimplify.cc (cp_fold): Likewise.
* pt.cc (tsubst_copy): Likewise.

gcc/
* dojump.cc (do_jump): Use CASE_CONVERT.
* tree-ssa-dom.cc (edge_info::derive_equivalences): Likewise.

2 years agosystem.h: Include <initializer_list> in system.h unconditionally
Jakub Jelinek [Mon, 2 May 2022 05:10:14 +0000 (07:10 +0200)]
system.h: Include <initializer_list> in system.h unconditionally

On Sun, May 01, 2022 at 07:06:53PM +0100, Jonathan Wakely wrote:
> > >> the reason is that "gcc/analyzer/region-model.cc” uses initializer_lists, and it seems that <initializer_list>
> > >> is not transitively included by any used headers for _LIBCPP_VERSION < 4000.  I fixed that locally by
> > >> adding initializer_list into system.h (and adding INCLUDE_INITIALIZER_LIST to the top of gcc/analyzer/region-model.cc)
> > >> - with that change those versions do bootstrap and test OK***
> > >
> > > From what I can see, with libstdc++ it works because <utility> which is
> > > included by system.h includes <initializer_list>.
> > > If I rename initializer_list in analyzer/region-model.ii to initializer_listx, I
> > > also get:
> > > ../../gcc/analyzer/region-model.cc: In function ‘void ana::selftest::test_binop_svalue_folding()’:
> > > ../../gcc/analyzer/region-model.cc:4966:48: error: deducing from brace-enclosed initializer list requires ‘#include <initializer_list>’
> > > 4508 |
> > >  +++ |+#include <initializer_list>
> > > 4509 | static void
> > > ......
> > > 4966 |     for (auto op : {BIT_IOR_EXPR, TRUTH_OR_EXPR})
> > >      |                                                ^
> > > ../../gcc/analyzer/region-model.cc:4978:49: error: deducing from brace-enclosed initializer list requires ‘#include <initializer_list>’
> > > 4978 |     for (auto op : {BIT_AND_EXPR, TRUTH_AND_EXPR})
> > >      |                                                 ^
> > >
> > > I think we have 2 options, one is do what you wrote above,
> > > INCLUDE_INITIALIZER_LIST defined before system.h to get #include <initializer_list>.
> > > The other option is just to include that unconditionally, it is a very small
> > > header.  For libstdc++ it will make no difference as it is included anyway
> > > and the header is really small there, libc++ includes <cstddef> which isn't
> > > normally included and system.h includes <stddef.h> instead.
> >
> > I’d say unconditionally would be OK. I suppose the chance that any host
> > C++ is good enough to build GCC as-is but fails to provide
> > <initializer_list> is zero?
> >
>
> Yes, definitely.
>
> > I’d be OK to do this change without a new RC even.

2022-05-02  Jakub Jelinek  <jakub@redhat.com>

* system.h: Include initializer_list.

2 years agolibgcov: add ATTRIBUTE_UNUSED for dump_string
Martin Liska [Mon, 2 May 2022 04:46:05 +0000 (06:46 +0200)]
libgcov: add ATTRIBUTE_UNUSED for dump_string

Mitigates the following clang warning:
libgcc/libgcov-driver.c:416:1: warning: unused function 'dump_string' [-Wunused-function]

libgcc/ChangeLog:

* libgcov-driver.c: Add ATTRIBUTE_UNUSED.

2 years agoDaily bump.
GCC Administrator [Mon, 2 May 2022 00:16:24 +0000 (00:16 +0000)]
Daily bump.

2 years agors6000: "Y" is DS-form, not DQ-form
Segher Boessenkool [Tue, 26 Apr 2022 21:30:44 +0000 (21:30 +0000)]
rs6000: "Y" is DS-form, not DQ-form

2022-05-01  Segher Boessenkool  <segher@kernel.crashing.org>

* config/rs6000/constraints.md (Y constraint): Fix comment.

2 years agoDenormalize VR_VARYING to VR_RANGE before passing it to set_range_info_raw.
Aldy Hernandez [Fri, 29 Apr 2022 20:46:25 +0000 (22:46 +0200)]
Denormalize VR_VARYING to VR_RANGE before passing it to set_range_info_raw.

We are ICEing in set_range_info_raw because value_range_kind cannot be
VR_VARYING, since SSA_NAME_RANGE_TYPE can only hold VR_RANGE /
VR_ANTI_RANGE.  Most of the time setting a VR_VARYING as a global
range makes no sense.  However, we can have a range spanning the
entire domain (VR_RANGE of [MIN,MAX] which is essentially a
VR_VARYING), if the nonzero bits are set.

This was working before because set_range_info_raw allows setting
VR_RANGE of [MIN, MAX].  However, when going through an irange, we
normalize this to a VR_VARYING, thus causing the ICE.  It's
interesting that other calls to set_range_info with an irange haven't
triggered this.

One solution would be to just ignore VR_VARYING and bail, since
set_range_info* is really an update of the current range semantic
wise.  After all, we keep the nonzero bits which provide additional
info.  But this would be a change in behavior, so not suitable until
after GCC 12 is released.  So in order to keep with current behavior
we can just denormalize the varying to VR_RANGE.

Tested on x86-64 Linux.

    PR tree-optimization/105432

gcc/ChangeLog:

* tree-ssanames.cc (set_range_info): Denormalize VR_VARYING to
VR_RANGE before passing a piecewise range to set_range_info_raw.