The allocator, hash function and equality function should all be
value-initialized by the default constructor of an unordered container.
Do it in the EBO helper, so we don't have to get it right in multiple
places.
Signed-off-by: Jonathan Wakely <jwakely@redhat.com>
libstdc++-v3/ChangeLog:
PR libstdc++/100863
PR libstdc++/65816
* include/bits/hashtable_policy.h (_Hashtable_ebo_helper):
Value-initialize subobject.
* testsuite/23_containers/unordered_map/allocator/default_init.cc:
Remove XFAIL.
* testsuite/23_containers/unordered_set/allocator/default_init.cc:
Remove XFAIL.
The expected number of allocations is different when copying COW
strings.
Signed-off-by: Jonathan Wakely <jwakely@redhat.com>
libstdc++-v3/ChangeLog:
PR libstdc++/96088
* testsuite/23_containers/unordered_map/96088.cc: Adjust
expected number of allocations.
* testsuite/23_containers/unordered_set/96088.cc: Likewise.
arc700 is legacy and there's no active development for it, so switch to
latest hs38_linux as default
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Claudiu Zissulescu <claziss@synopsys.com>
gcc/
2021-06-02 Vineet Gupta <vgupta@synopsys.com>
* config/arc/arc.h (TARGET_CPU_DEFAULT): Change to hs38_linux.
Since commit dd1ef00c45 ("Fix bug in the define_subst handling that
made match_scratch unusable for multi-alternative patterns.") the
workaround for that bug in *ashrdi3_31<setcc><cconly> is not only no
longer necessary, but actually breaks the build.
Get rid of it by using only one alternative in (match_scratch). It
will be replicated as many times as needed in order to match the
pattern with which (define_subst) is used.
gcc/ChangeLog:
* config/s390/s390.md(*ashrdi3_31<setcc><cconly>): Use a single
constraint.
* config/s390/subst.md(cconly_subst): Use a single constraint
in (match_scratch).
gcc/testsuite/ChangeLog:
* gcc.target/s390/ashr.c: New test.
The testcase without this FAILs on i686-linux, where we warn about returning
of a generic vector changing the ABI.
2021-06-02 Jakub Jelinek <jakub@redhat.com>
* g++.dg/ext/builtin-shufflevector-1.C: Add -Wno-psabi -w to
dg-options.
gcc/
* config/h8300/h8300-protos.h (compute_a_shift_length): Drop unused
argument from prototype.
(output_logical_op): Add rtx_code argument.
(compute_logical_op_length): Likewise.
* config/h8300/h8300.c (h8300_and_costs): Pass additional argument
to compute_a_shift_length.
(output_logical_op); New argument with the rtx code rather than
extracting it from an operand. Handle QImode too.
(compute_logical_op_length): Similary.
(compute_a_shift_length): Drop unused argument.
* config/h8300/h8300.md (logicals): New code iterator.
* config/h8300/logical.md (<code><mode>3 expander): Combine
the "and" expander with the "ior"/"xor" expander.
(bclr<mode>msx): Combine the QI/HI mode patterns.
(<logical><mode>3 insns): Use code iterator rather than match_operator.
Handle QImode as well. Update call to output_logical_op and
compute_logical_op_length to pass in rtx_code
Fix split condition on all define_insn_and_split patterns.
(one_cmpl<mode>2<cczn>): Use <cczn> to support both clobbering
the flags and setting ZN via existing define_subst.
* config/h8300/shiftrotate.md: Drop unused argument from
calls to compute_a_shift_length.
Signed-off-by: Jeff Law <jeffreyalaw@gmail.com>
In the case of value-initializing an object of class type T,
[dcl.init.general]/8 says:
- if T has either no default constructor ([class.default.ctor]) or
a default constructor that is user-provided or deleted, then the
object is default-initialized;
- otherwise, the object is zero-initialized and ... if T has a
non-trivial default constructor, the object is default-initialized;
But when determining whether to first zero-initialize the object,
expand_aggr_init_1 incorrectly considers the user-providedness of _all_
constructors rather than only that of the _default_ constructors. This
causes us to skip the zero-initialization step when the class type has a
defaulted default constructor alongside a user-defined constructor.
It seems the predicate type_has_non_user_provided_default_constructor
accurately captures the above rule for when to first perform a
zero-initialization during value-initialization, so this patch adjusts
expand_aggr_init_1 to use this predicate instead.
PR c++/65816
gcc/cp/ChangeLog:
* init.c (expand_aggr_init_1): Check
type_has_non_user_provided_default_constructor instead of
type_has_user_provided_constructor.
gcc/testsuite/ChangeLog:
* g++.dg/cpp0x/constexpr-delegating3.C: New test.
* g++.dg/cpp0x/dc10.C: New test.
* g++.dg/cpp0x/initlist-base4.C: New test.
* g++.dg/cpp2a/constexpr-init22.C: New test.
libstdc++-v3/ChangeLog:
* testsuite/23_containers/deque/allocator/default_init.cc,
testsuite/23_containers/forward_list/allocator/default_init.cc,
testsuite/23_containers/list/allocator/default_init.cc,
testsuite/23_containers/map/allocator/default_init.cc,
testsuite/23_containers/set/allocator/default_init.cc,
testsuite/23_containers/vector/allocator/default_init.cc,
testsuite/23_containers/vector/bool/allocator/default_init.cc:
Remove xfail.
This is the first of series of patches to simplify phi-opt
to use match and simplify in many cases. This simplification
will more things to optimize.
This is what Richard requested in
https://gcc.gnu.org/pipermail/gcc-patches/2021-May/571197.html
and I think it is the right thing to do too.
OK? Bootstrapped and tested on x86_64-linux-gnu with no regressions.
gcc/ChangeLog:
PR tree-optimization/25290
* tree-ssa-phiopt.c (match_simplify_replacement):
New function.
(tree_ssa_phiopt_worker): Use match_simplify_replacement.
(two_value_replacement): Change the comment about
conditional_replacement.
(conditional_replacement): Delete.
The problem here is we don't have an assignment type any more
for empty structs as they were removed during gimplifcation.
This adds a special case where the assignment var does not exist
and the return decl is empty typed.
OK? Tested on aarch64-linux-gnu with no regressions.
Thanks,
Andrew Pinski
changes since v1:
v2: Use is_empty_type instead of zero-sized type.
gcc/ChangeLog:
PR tree-optimization/95481
* tree-tailcall.c (find_tail_calls): Handle empty typed
return decls.
gcc/testsuite/ChangeLog:
PR tree-optimization/95481
* gcc.dg/tree-ssa/tailcall-10.c: New test.
* gcc.dg/tree-ssa/tailcall-11.c: New test.
* gcc.dg/tree-ssa/tailcall-12.c: New test.
* gcc.dg/tree-ssa/tailcall-13.c: New test.
* gcc.dg/tree-ssa/tailrecursion-8.c: New test.
Instead of only removing assignments of zero-sized types,
assignments of all empty types should be removed during gimplification.
This moves to use is_empty_type which will be used in other places too.
OK? Bootstrapped and tested on aarch64-linux-gnu with no regressions.
Thanks,
Andrew Pinski
gcc/ChangeLog:
* gimplify.c (zero_sized_field_decl): Delete
(zero_sized_type): Delete
(gimplify_init_ctor_eval): Use is_empty_type instead
of zero_sized_field_decl.
(gimplify_modify_expr): Use is_empty_type instead of
zero_sized_type.
These tests use <string_view> so must not run for anything older than
C++17.
Signed-off-by: Jonathan Wakely <jwakely@redhat.com>
libstdc++-v3/ChangeLog:
* testsuite/23_containers/unordered_map/96088.cc: Change
effective target to c++17.
* testsuite/23_containers/unordered_set/96088.cc: Likewise.
The isblank class is not supported until C++11.
Signed-off-by: Jonathan Wakely <jwakely@redhat.com>
libstdc++-v3/ChangeLog:
* testsuite/22_locale/ctype/is/string/89728_neg.cc: Only test
isblank for C++11 and later.
The three-argument form of ranges::advance is supposed to return the
difference between the second argument and the distance the iterator was
advanced. When a non-random-access iterator is not advanced (because it
already equals the sentinel) we were returning 0 rather than n - 0.
libstdc++-v3/ChangeLog:
PR libstdc++/100833
* include/bits/ranges_base.h (ranges::advance(iter, n, sentinel)):
Fix return value for no-op case.
* testsuite/24_iterators/range_operations/advance.cc: Test
return values of three-argument overload.
Before C++20 added destroying operator delete, by the time we called
operator delete for a pointer, the object would already be gone. But that
isn't true for destroying delete. Since the optimizers' assumptions about
operator delete are based on either DECL_IS_REPLACEABLE_OPERATOR (which
already is not set) or CALL_FROM_NEW_OR_DELETE_P, let's avoid setting the
latter flag in this case.
PR c++/91859
gcc/ChangeLog:
* tree.h (CALL_FROM_NEW_OR_DELETE_P): Adjust comment.
gcc/cp/ChangeLog:
* call.c (build_op_delete_call): Don't set CALL_FROM_NEW_OR_DELETE_P
for destroying delete.
* init.c (build_delete): Don't clobber before destroying delete.
gcc/testsuite/ChangeLog:
* g++.dg/cpp2a/destroying-delete5.C: New test.
-Wdeprecated-copy was depending only on the state of the warning at the
point where we call the function, making it hard to use #pragma diagnostic
to suppress the warning for a particular implicitly declared function.
But checking whether the warning is enabled at the location of the implicit
declaration turned out to be a bit complicated; option_enabled only tests
whether it was enabled at the start of compilation, the actual test only
existed in the middle of diagnostic_report_diagnostic. So this patch
factors it out and adds a new warning_enabled function to diagnostic.h.
gcc/ChangeLog:
PR c++/94492
* diagnostic.h (warning_enabled_at): Declare.
* diagnostic.c (diagnostic_enabled): Factor out from...
(diagnostic_report_diagnostic): ...here.
(warning_enabled_at): New.
gcc/cp/ChangeLog:
PR c++/94492
* decl2.c (cp_warn_deprecated_use): Check warning_enabled_at.
gcc/testsuite/ChangeLog:
PR c++/94492
* g++.dg/cpp0x/depr-copy4.C: New test.
Andrew's last set of changes fixes the bootstrap problem on i686 when
global ranges are exported from evrp. The specific patch that fixes the
problem is 715914d3:
Author: Andrew MacLeod <amacleod@redhat.com>
Date: Mon May 31 12:13:50 2021 -0400
Do not calculate new values when evaluating a debug statement.
Add a flag to enable/disable immediately improving poor values found during
cache propagation. Then disable it when processing debug statements.
This patch reverts commit 2364b58 now that exporting of global ranges works.
Tested on x86-64 Linux with default flags, and on i686 with the flags in
the PR: --enable-clocale=gnu --with-system-zlib --with-demangler-in-ld --with-fpmath=sse --enable-languages=c,c++ --enable-cet i686-linux --enable-bootstrap --with-fpmath=sse --disable-libcc1 --disable-libcilkrts --disable-libsanitizer
gcc/ChangeLog:
* gimple-ssa-evrp.c: Enable exporting of global ranges.
gcc/testsuite/ChangeLog:
* gcc.dg/Wstringop-overflow-55.c: Adjust for global ranges changes.
* gcc.dg/pr80776-1.c: Same.
gcc/c-family/ChangeLog:
PR other/100759
* c-attribs.c (handle_optimize_attribute): Limit sanity check
to a situation where we are not in processing of an optimize
pragma.
* c-pragma.c (handle_pragma_pop_options): Restore target
options.
When no shared library is installed, the new code to determine the name
of the -gdb.py file yields an empty string. Use the name of the static
library in that case.
libstdc++-v3/ChangeLog:
PR libstdc++/99453
* python/Makefile.am: Use archive name for printer hook if no
dynamic library name is available.
* python/Makefile.in: Regenerate.
Add a flag to enable/disable immediately improving poor values found during
cache propagation. Then disable it when processing debug statements.
gcc/
PR tree-optimization/100781
* gimple-range-cache.cc (ranger_cache::ranger_cache): Enable new
value calculation by default.
(ranger_cache::enable_new_values): New.
(ranger_cache::disable_new_values): New.
(ranger_cache::push_poor_value): Check if new values are allowed.
* gimple-range-cache.h (class ranger_cache): New member/methods.
* gimple-range.cc (gimple_ranger::range_of_expr): Check for debug
statement, and disable/renable new value calculation.
gcc/testsuite/
PR tree-optimization/100781
* gcc.dg/pr100781.c: New.
Split the old functionality of ssa_name_in_bb into the components for
definition in a block, entry and exit range. Call these as appropriate.
* gimple-range-cache.cc (ranger_cache::ssa_range_in_bb): Delete.
(ranger_cache::range_of_def): New.
(ranger_cache::entry_range): New.
(ranger_cache::exit_range): New.
(ranger_cache::range_of_expr): Adjust.
(ranger_cache::range_on_edge): Adjust.
(ranger_cache::propagate_cache): Call exit_range directly.
* gimple-range-cache.h (class ranger_cache): Adjust.
Flatten and simplify gori-computes. Tweak debug output.
range-cache now provides range_of_expr and range_on_edge in the
standard formats, but in a "query what you have" mode rather than
"go figure out anything that is missing" mode.
* gimple-range-cache.cc (ranger_cache::ranger_cache): Adjust for
gori_compute being a member rather than base class.
dervied call to member call.
(ranger_cache::dump): No longer dump gori_map.
(ranger_cache::dump_bb): New.
(ranger_cache::get_non_stale_global_range): Adjust for gori_compute
being a member rather than base class.
(ranger_cache::set_global_range): Ditto.
(ranger_cache::ssa_range_in_bb): Ditto.
(ranger_cache::range_of_expr): New.
(ranger_cache::range_on_edge): New.
(ranger_cache::block_range): Adjust for gori_computes. Debug changes.
(ranger_cache::propagate_cache): Adjust debugging output.
(ranger_cache::fill_block_cache): Adjust for gori_computes. Debug
output changes.
* gimple-range-cache.h (class ranger_cache): Make gori_compute a
member, and inherit from range_query instead.
(ranger_cache::dump_bb): New. split from dump.
* gimple-range-gori.cc (gori_compute::ssa_range_in_bb): Delete.
(gori_compute::expr_range_at_stmt): Delete.
(gori_compute::compute_name_range_op): Delete.
(gori_compute::compute_operand_range_switch): Add fur_source.
(gori_compute::compute_operand_range): Add fur_source param, inline
old compute_name_range_op and optimize_logical_operands.
(struct tf_range): Delete.
(gori_compute::logical_combine): Adjust
(gori_compute::optimize_logical_operands): Delete.
(gori_compute::compute_logical_operands_in_chain): Delete.
(gori_compute::compute_logical_operands): Adjust.
(gori_compute::compute_operand1_range): Adjust to fur_source.
(gori_compute::compute_operand2_range): Ditto.
(gori_compute::compute_operand1_and_operand2_range): Ditto.
(gori_compute::outgoing_edge_range_p): Add range_query parameter,
and adjust to fur_source.
* gimple-range-gori.h (class gori_compute): Simplify and adjust to
range_query and fur_source.
* gimple-range.cc (gimple_ranger::range_on_edge): Query range_on_edge
from the ranger_cache..
(gimple_ranger::fold_range_internal): Adjust to base class change of
ranger_cache.
(gimple_ranger::dump_bb): Adjust dump.
* gimple-range.h (gimple_ranger):export gori computes object.
when a range evolves to the point where it becomes a constant, it is
marked as invariant. Rather than marking it as always_current in the
timestamp, give it the correct timestamp and just never flag it as stale.
This will allow other names which use this value to become stale and be
recomputed using the newly invariant value.
gcc/
PR tree-optimization/100774
* gimple-range-cache.cc (ranger_cache::get_non_stale_global_range):
Constant values are also not stale.
(ranger_cache::set_global_range): Range invariant values should also
have the correct timestamp.
gcc/testsuite
PR tree-optimization/100774
* g++.dg/pr100774.C: New.
Fix failing pch testcases. Use xstrdup to retain a reliable copy of the debug
format str containing the names (df_set_names is a static string var).
2021-05-31 Indu Bhagat <indu.bhagat@oracle.com>
gcc/c-family/
PR testsuite/100749
* c-pch.c (c_common_valid_pch): Use xstrdup for debug format set names.
'dg-do run' is pointless -1 due to dg-error. And it won't work except
by chance for gomp tests; as -2 only has depend(out:), a 'dg-do compile'
is sufficient.
gcc/testsuite/ChangeLog:
* gfortran.dg/gomp/depend-iterator-1.f90: Use dg-do compile.
* gfortran.dg/gomp/depend-iterator-2.f90: Use dg-do compile.
This adds support for the clang __builtin_shufflevector extension to
the C and C++ frontends. The builtin is lowered to VEC_PERM_EXPR.
Because VEC_PERM_EXPR does not support different sized vector inputs
or result or the special permute index of -1 (don't-care)
c_build_shufflevector applies lowering by widening inputs and output
to the widest vector, replacing -1 by a defined index and
subsetting the final vector if we produced a wider result than
desired.
Code generation thus can be sub-optimal, followup patches will
aim to fix that by recovering from part of the missing features
during RTL expansion and by relaxing the constraints of the GIMPLE
IL with regard to VEC_PERM_EXPR.
2021-05-21 Richard Biener <rguenther@suse.de>
PR c++/88601
gcc/c-family/
* c-common.c: Include tree-vector-builder.h and
vec-perm-indices.h.
(c_common_reswords): Add __builtin_shufflevector.
(c_build_shufflevector): New funtion.
* c-common.h (enum rid): Add RID_BUILTIN_SHUFFLEVECTOR.
(c_build_shufflevector): Declare.
gcc/c/
* c-decl.c (names_builtin_p): Handle RID_BUILTIN_SHUFFLEVECTOR.
* c-parser.c (c_parser_postfix_expression): Likewise.
gcc/cp/
* cp-objcp-common.c (names_builtin_p): Handle
RID_BUILTIN_SHUFFLEVECTOR.
* cp-tree.h (build_x_shufflevector): Declare.
* parser.c (cp_parser_postfix_expression): Handle
RID_BUILTIN_SHUFFLEVECTOR.
* pt.c (tsubst_copy_and_build): Handle IFN_SHUFFLEVECTOR.
* typeck.c (build_x_shufflevector): Build either a lowered
VEC_PERM_EXPR or an unlowered shufflevector via a temporary
internal function IFN_SHUFFLEVECTOR.
gcc/
* internal-fn.c (expand_SHUFFLEVECTOR): Define.
* internal-fn.def (SHUFFLEVECTOR): New.
* internal-fn.h (expand_SHUFFLEVECTOR): Declare.
* doc/extend.texi: Document __builtin_shufflevector.
gcc/testsuite/
* c-c++-common/builtin-shufflevector-2.c: New testcase.
* c-c++-common/torture/builtin-shufflevector-1.c: Likewise.
* g++.dg/ext/builtin-shufflevector-1.C: Likewise.
* g++.dg/ext/builtin-shufflevector-2.C: Likewise.
The mma_assemble_input_operand predicate does not accept reg+reg indexed
addresses which can lead to ICEs. The lxv and lxvp instructions have
indexed forms (lxvx and lxvpx), so the simple solution is to just allow
indexed addresses in the predicate.
2021-05-30 Peter Bergner <bergner@linux.ibm.com>
gcc/
PR target/99842
* config/rs6000/predicates.md(mma_assemble_input_operand): Allow
indexed form addresses.
gcc/testsuite/
PR target/99842
* g++.target/powerpc/pr99842.C: New.
The original link redirects to a very generic page that advertises
lots of tools and other things, just not this.
libstdc++-v3/ChangeLog:
2021-05-31 Gerald Pfeifer <gerald@pfeifer.com>
* doc/xml/manual/abi.xml: Remove dead reference to "Intel
Compilers for Linux: Compatibility with GNU Compilers" article.
* doc/html/manual/abi.html: Regenerate.
We use getline, so leverage gnulib to provide fallback implementation.
ChangeLog:
* configure.ac: Add gnulib to configdirs for sim.
* configure: Regenerate.
I forgot to add default(none) and defaultmap(none) wherever possible on the
testcases to make sure none of the required clauses are added implicitly (because
in that case it doesn't work with these none arguments of those default* clauses
or works differently with other default* settings.
And that revealed we didn't add shared on parallel for linear clause
on parallel master taskloop simd, so this patch fixes that too.
2021-05-29 Jakub Jelinek <jakub@redhat.com>
PR middle-end/99928
* gimplify.c (gimplify_scan_omp_clauses): For taskloop simd
combined with parallel, make sure to add shared clause to
parallel for explicit linear clause.
* c-c++-common/gomp/pr99928-1.c: Add default(none) to constructs
combined with parallel, teams or taskloop and defaultmap(none)
to constructs combined with target.
* c-c++-common/gomp/pr99928-2.c: Likewise.
* c-c++-common/gomp/pr99928-3.c: Likewise.
* c-c++-common/gomp/pr99928-4.c: Likewise.
* c-c++-common/gomp/pr99928-5.c: Likewise.
* c-c++-common/gomp/pr99928-6.c: Likewise.
* c-c++-common/gomp/pr99928-7.c: Likewise.
* c-c++-common/gomp/pr99928-8.c: Likewise.
* c-c++-common/gomp/pr99928-9.c: Likewise.
* c-c++-common/gomp/pr99928-10.c: Likewise.
* c-c++-common/gomp/pr99928-13.c: Likewise.
* c-c++-common/gomp/pr99928-14.c: Likewise.
The patch converting evrp to the get_range_query(fun) API broke i686
bootstrap (commit 57bf37515). The problem seems to be in a subsequent
pass that has more up-to-date global ranges. I won't be able to look at
this until next week, so I am reverting the problematic bit of the
patch-- the exporting of global ranges once evrp finishes. The use of
the new API remains.
Reverting the behavior shouldn't be a problem as we never used to export
global ranges from ranger. This was new behavior in the patchset.
Tested on x86-64 Linux with a bootstrap and regtest, and on x86-32 with
only a bootstrap and the configure flags from the PR:
--enable-clocale=gnu --with-system-zlib --with-demangler-in-ld --with-fpmath=sse --enable-languages=c,c++ --enable-cet i686-linux --enable-bootstrap --with-fpmath=sse --disable-libcc1 --disable-libcilkrts --disable-libsanitizer
gcc/ChangeLog:
PR tree-optimization/100787
* gimple-ssa-evrp.c: Disable exporting of global ranges.
gcc/testsuite/ChangeLog:
* gcc.dg/Wstringop-overflow-55.c:
* gcc.dg/pr80776-1.c:
Like my recent patch to add ovl_range and lkp_range in the C++ front end,
this patch adds the tsi_range adaptor for using C++11 range-based 'for' with
a STATEMENT_LIST, e.g.
for (tree stmt : tsi_range (stmt_list)) { ... }
This also involves adding some operators to tree_stmt_iterator that are
needed for range-for iterators, and should also be useful in code that uses
the iterators directly.
The patch updates the suitable loops in the C++ front end, but does not
touch any loops elsewhere in the compiler.
gcc/ChangeLog:
* tree-iterator.h (struct tree_stmt_iterator): Add operator++,
operator--, operator*, operator==, and operator!=.
(class tsi_range): New.
gcc/cp/ChangeLog:
* constexpr.c (build_data_member_initialization): Use tsi_range.
(build_constexpr_constructor_member_initializers): Likewise.
(constexpr_fn_retval, cxx_eval_statement_list): Likewise.
(potential_constant_expression_1): Likewise.
* coroutines.cc (await_statement_expander): Likewise.
(await_statement_walker): Likewise.
* module.cc (trees_out::core_vals): Likewise.
* pt.c (tsubst_expr): Likewise.
* semantics.c (set_cleanup_locs): Likewise.
This avoids vectorizing a possibly trapping operation when lanes
are handled in different BBs. I spotted this when working on the
originally reported issue in PR100778.
2021-05-28 Richard Biener <rguenther@suse.de>
PR tree-optimization/100778
* tree-vect-slp.c (vect_build_slp_tree_1): Prevent possibly
trapping ops in different BBs.
* gcc.dg/vect/bb-slp-pr100778-1.c: New testcase.
Jason Merrill
Jonathan Wakely
Claudiu Zissulescu
Ilya Leoshkevich
Tobias Burnus
Jakub Jelinek
Martin Liska
Jeff Law
GCC Administrator
Patrick Palka
Andrew Pinski
Bill Schmidt
Aldy Hernandez
liuhongt
Andrew MacLeod
Indu Bhagat
Richard Biener
Peter Bergner
Gerald Pfeifer
Bernd Edlinger
Jeff Law
Mike Frysinger