gcc/rust/ChangeLog:
* ast/rust-ast-visitor.cc
(DefaultASTVisitor::visit): Only visit the path of an instance
of Visibility if the instance has a path.
* ast/rust-ast.h
(SimplePath::SimplePath): Make sure constructors are explicit.
* resolve/rust-early-name-resolver-2.0.cc
(Early::visit_attributes): Pass entire paths to
NameResolutionContext::resolve_path.
(Early::visit): Likewise and avoid copying a path.
* resolve/rust-forever-stack.hxx
(ForeverStack::resolve_path): Assert that at least one path
segment has been passed in.
Signed-off-by: Owen Avery <powerboat9.gamer@gmail.com>
Config has been updated upstream to correctly indent declaration.
gcc/rust/ChangeLog:
* rust-attribs.cc (handle_hot_attribute): Remove clang-format comment.
Signed-off-by: Marc Poulhiès <dkm@kataplop.net>
gcc/rust/ChangeLog:
* checks/errors/borrowck/rust-bir-builder-internal.h: Include
"rust-immutable-name-resolution-context.h" and "options.h".
(AbstractBuilder::resolve_label): Use the 2.0 name resolver when
it's enabled.
(AbstractBuilder::resolve_variable): Likewise.
(AbstractBuilder::resolve_variable_or_fn): Likewise.
Signed-off-by: Owen Avery <powerboat9.gamer@gmail.com>
This patch ensuers we reuse the Builder for new type to
ensure we create a new type from scratch ensuring consistent
new node-ids.
gcc/rust/ChangeLog:
* expand/rust-derive-default.cc (DeriveDefault::visit_struct): use builder
(DeriveDefault::visit_tuple): likewise
* expand/rust-derive-eq.cc (DeriveEq::visit_tuple): likewise
(DeriveEq::visit_struct): likewise
(DeriveEq::visit_enum): likewise
(DeriveEq::visit_union): likewise
gcc/testsuite/ChangeLog:
* rust/compile/nr2/exclude: these are fixed now
Signed-off-by: Philip Herron <herron.philip@googlemail.com>
Since r14-6076, the LANG_HOOKS_COMMON_ATTRIBUTE_TABLE should not be used and
LANG_HOOKS_ATTRIBUTE_TABLE replaces it.
Add the "cold" attribute to fix PR120018 (and the matching "hot" attribute).
Replace some gcc_assert() by explicit warnings (same as D frontend).
Add some clang-format off/on comment around code that's not correctly
formatted.
gcc/rust/ChangeLog:
PR rust/120018
* rust-attribs.cc (handle_noreturn_attribute): Reindent declaration.
(handle_leaf_attribute): Likewise.
(handle_const_attribute): Likewise.
(handle_malloc_attribute): Likewise.
(handle_pure_attribute): Likewise.
(handle_novops_attribute): Likewise.
(handle_nonnull_attribute): Likewise.
(handle_nothrow_attribute): Likewise.
(handle_type_generic_attribute): Likewise.
(handle_transaction_pure_attribute): Likewise.
(handle_returns_twice_attribute): Likewise.
(handle_fnspec_attribute): Likewise.
(handle_omp_declare_simd_attribute): Likewise.
(handle_cold_attribute): New.
(handle_hot_attribute): New.
(attribute_spec::exclusions attr_cold_hot_exclusions): New.
(grs_langhook_common_attributes): Make it static.
(grs_langhook_common_attribute_table): New.
(grs_langhook_gnu_attributes): New.
(grs_langhook_gnu_attribute_table): New.
(handle_malloc_attribute): Make it static.
(handle_fnspec_attribute): Likewise.
(handle_pure_attribute): Replace gcc_assert by explicit warning.
(handle_novops_attribute): Likewise.
(handle_nothrow_attribute): Likewise.
(handle_returns_twice_attribute): Likewise.
(handle_omp_declare_simd_attribute): Likewise and make it static.
* rust-lang.cc (grs_langhook_gnu_attribute_table): New.
(grs_langhook_common_attribute_table): Adjust type to new hook.
(LANG_HOOKS_COMMON_ATTRIBUTE_TABLE): Remove.
(LANG_HOOKS_ATTRIBUTE_TABLE): New.
Signed-off-by: Marc Poulhiès <dkm@kataplop.net>
Empty loops have no body which means this is a NULL_TREE during const
evaluation which needs a check. Fixes Rust-GCC #3618.
gcc/rust/ChangeLog:
* backend/rust-constexpr.cc (eval_constant_expression): Check if t is a NULL_TREE
gcc/testsuite/ChangeLog:
* rust/compile/issue-3618.rs: Test empty loops error properly.
Signed-off-by: Tom Schollenberger <tss2344@g.rit.edu>
When using impl traits in argument position (APIT), they are desugared into generics,
and supplying explicit generic arguments is not allowed. This commit adds the error
diagnostic E0632 for attempting to pass generic arguments to impl traits, completing
the implementation of the APIT feature.
gcc/rust/ChangeLog:
* ast/rust-desugar-apit.cc: track if this is a impl-trait generic
* ast/rust-item.h (class TypeParam): add field to track if from impl trait
* hir/rust-ast-lower-type.cc (ASTLowerGenericParam::visit): likewise
* hir/tree/rust-hir-item.cc (TypeParam::TypeParam): upate hir as well
(TypeParam::operator=): likewise
* hir/tree/rust-hir-item.h (class TypeParam): likewise
* typecheck/rust-tyty-subst.cc (SubstitutionParamMapping::get_generic_param): add error
* typecheck/rust-tyty-subst.h: add const getter for the associated TypeParm
gcc/testsuite/ChangeLog:
* rust/compile/nr2/exclude: nr2 cant handle this
* rust/compile/impl_trait_generic_arg.rs: New test.
Signed-off-by: Philip Herron <herron.philip@googlemail.com>
gcc/rust/ChangeLog:
* backend/rust-compile-pattern.cc
(CompilePatternCheckExpr::visit): Fix GENERIC generation in
light of enum layout changes since this code was written.
(CompilePatternBindings::handle_struct_pattern_ident_pat):
Delegate handling of child patterns to another
CompilePatternBindings::Compile call.
(CompilePatternBindings::make_struct_access): Make field name
parameter const qualified.
* backend/rust-compile-pattern.h
(CompilePatternBindings::make_struct_access): Likewise.
gcc/testsuite/ChangeLog:
* rust/execute/torture/struct-pattern-match.rs: New test.
Signed-off-by: Owen Avery <powerboat9.gamer@gmail.com>
gcc/rust/ChangeLog:
* resolve/rust-ast-resolve-item.cc
(ResolveItem::visit): Use the return values of
CanonicalPath::inherent_impl_seg and
CanonicalPath::trait_impl_projection_seg more directly.
* util/rust-canonical-path.h
(CanonicalPath::trait_impl_projection_seg): Append "<impl "
instead of "<" to the beginning of the returned path segment.
(CanonicalPath::inherent_impl_seg): Likewise.
Signed-off-by: Owen Avery <powerboat9.gamer@gmail.com>
Argument position impl traits are simply syntatic sugar for generics. This
adds a new desugar pass to do this. So for example:
fn foo(a: impl Value, b: impl Value) -> i32
Is desugared into:
fn foo<T: Value, U: Value> (a: T, b: U) -> i32
So it just works like any normal generic function. There are more complex cases such as:
fn foo(_value: impl Bar<Baz = impl Foo>) -> i32
Which has a generic argument binding which needs to be turned into a where constraint:
fn foo<T, U>(_value: T) -> i32
where
T: Bar<Baz = U>,
U: Foo,
Fixes Rust-GCC#2015
Fixes Rust-GCC#1487
Fixes Rust-GCC#3454
Fixes Rust-GCC#1482
gcc/rust/ChangeLog:
* Make-lang.in: new desugar file
* ast/rust-ast.cc (ImplTraitTypeOneBound::as_string): its a unique_ptr now
(FormatArgs::set_outer_attrs): reformat
* ast/rust-path.h: remove has_generic_args assertion (can be empty because of desugar)
* ast/rust-type.h (class ImplTraitTypeOneBound): add copy ctor and use unique_ptr
* hir/rust-ast-lower-type.cc (ASTLoweringType::visit): update to use unique_ptr
* parse/rust-parse-impl.h (Parser::parse_type): reuse the existing unique_ptr instead
(Parser::parse_type_no_bounds): likewise
(Parser::parse_pattern): likewise
* resolve/rust-ast-resolve-type.cc (ResolveType::visit): its a unique_ptr now
* rust-session-manager.cc (Session::compile_crate): call desugar
* ast/rust-desugar-apit.cc: New file.
* ast/rust-desugar-apit.h: New file.
gcc/testsuite/ChangeLog:
* rust/compile/issue-2015.rs: fully supported now
* rust/compile/nr2/exclude: nr2 cant handle some of these
* rust/compile/issue-1487.rs: New test.
* rust/compile/issue-3454.rs: New test.
* rust/execute/torture/impl_desugar-2.rs: New test.
* rust/execute/torture/impl_desugar.rs: New test.
* rust/execute/torture/impl_trait1.rs: New test.
* rust/execute/torture/impl_trait2.rs: New test.
* rust/execute/torture/impl_trait3.rs: New test.
* rust/execute/torture/impl_trait4.rs: New test.
* rust/execute/torture/issue-1482.rs: New test.
Signed-off-by: Philip Herron <herron.philip@googlemail.com>
Rust only allows impl traits to be used in the return position of
functions.
Fixes Rust-GCC#1485
gcc/rust/ChangeLog:
* hir/rust-ast-lower-implitem.cc (ASTLowerImplItem::visit): allow impl type
* hir/rust-ast-lower-item.cc (ASTLoweringItem::visit): likewise
* hir/rust-ast-lower-type.cc (ASTLoweringType::ASTLoweringType): new flag for impl trait
(ASTLoweringType::translate): pass flag
(ASTLoweringType::visit): track impl trait tag
(ASTLoweringType::emit_impl_trait_error): new diagnostic
* hir/rust-ast-lower-type.h: add new field
gcc/testsuite/ChangeLog:
* rust/compile/impl_trait_diag.rs: New test.
* rust/compile/issue-1485.rs: New test.
Signed-off-by: Philip Herron <herron.philip@googlemail.com>
Allow matching on a struct instance and rebinding its fields to new names:
match X {
Foo {
field0: new_name0,
field1: new_name1,
} => {
do_something(new_name0, new_name1);
},
}
This will enable us to finish derives for PartialEq and PartialOrd but
isn't a complete implementation of these patterns yet.
gcc/rust/ChangeLog:
* backend/rust-compile-pattern.cc (CompilePatternBindings::make_struct_access):
New function.
(CompilePatternBindings::visit): Properly implement patterns mentioned above
and call make_struct_accesss.
* backend/rust-compile-pattern.h: New declaration.
gcc/testsuite/ChangeLog:
* rust/execute/torture/struct_pattern1.rs: New test.
Regression-tested on x86_64-pc-linux-gnu.
OK for master?
-- >8 --
In some places, a write to an array descriptor offset field was
generated simply by adding a modification of a reference to it that
was already available. This change uses the existing setter
function instead in those places, to generate the same code. It
makes it more explicit that in those areas a write to the field is
generated.
gcc/fortran/ChangeLog:
* trans-array.cc (gfc_alloc_allocatable_for_assignment): Use the
offset setter instead of generating a write to the offset.
(gfc_conv_array_parameter): Use the offset setter instead of
generating a write to the value returned by the offset getter.
* trans-expr.cc (gfc_trans_alloc_subarray_assign): Likewise.
Regression-tested on x86_64-pc-linux-gnu.
OK for master?
-- >8 --
The function gfc_conv_descriptor_data_addr generates an address to the
data field of an array descriptor. It is only used once, and in the
single place where it is used, the address is immediately dereferenced.
This change replaces the single usage with a plain access to the data
field, and removes the function. As the previous patch removed the
usage of the data field to write to it, the data getter can be used.
gcc/fortran/ChangeLog:
* trans-array.cc (gfc_conv_descriptor_data_addr): Remove.
* trans-array.h (gfc_conv_descriptor_data_addr): Remove.
* trans-decl.cc (gfc_trans_deferred_vars): Use
gfc_conv_descriptor_data_get.
Regression-tested on x86_64-pc-linux-gnu.
OK for master?
-- >8 --
In some places, a write to an array descriptor data field was generated
simply by adding a modification of a reference to it that was already
available. This change uses the existing setter function instead in
those places, to generate the same code. It makes it more explicit that
in those areas a write to the field is generated.
We have to be careful because in some of the places, the pointer that is
to be modified is an array descriptor field only under some conditions.
Add a conditional in those places to separate the cases, and keep
generating the reference modification as before if the conditions are
not met.
gcc/fortran/ChangeLog:
* trans.cc (gfc_finalize_tree_expr): Use the data setter instead
of writing to the value returned by the data getter.
* trans-decl.cc (gfc_trans_deferred_vars): Likewise.
* trans-stmt.cc (trans_associate_var): Use the data setter
instead of writing to the value dereferenced from the data
address.
The i386 high-register patterns used things like:
(match_operator:SWI248 2 "extract_operator"
[(match_operand 0 "int248_register_operand" "Q")
(const_int 8)
(const_int 8)])
to match an extraction of a high register such as AH from AX/EAX/RAX.
This construct is used in contexts where only the low 8 bits of the
value matter. This is either done explicitly using a (subreg:QI ... 0)
or implicitly by assigning to an 8-bit zero_extract destination.
extract_operator therefore matches both sign_extract and zero_extract,
since the signedness of the extension beyond 8 bits is irrelevant.
But the fact that only the low 8 bits of the value are significant
means that a shift right by 8 is as good as an extraction. Shifts
right would already be used for things like:
struct s {
long a:8;
long b:8;
long c:48;
};
struct s f(struct s x, long y, long z) {
x.b = (y & z) >> 8;
return x;
}
but are used more after g:965564eafb721f8000013a3112f1bba8d8fae32b.
This patch therefore replaces extract_operator with a new predicate
called extract_high_operator that matches both extractions and shifts.
The predicate checks the extraction field and shift amount itself,
so that patterns only need to match the first operand.
Splitters used match_op_dup to preserve the choice of extraction.
But the fact that the extractions (and now shifts) are equivalent
means that we can just as easily canonicalise on one of them.
(In theory, canonicalisation would also promote CSE, although
that's unlikely in practice.) The patch goes for zero_extract,
for consistency with destinations.
gcc/
PR target/121306
* config/i386/predicates.md (extract_operator): Replace with...
(extract_high_operator): ...this new predicate.
* config/i386/i386.md (*cmpqi_ext<mode>_1, *cmpqi_ext<mode>_2)
(*cmpqi_ext<mode>_3, *cmpqi_ext<mode>_4, *movstrictqi_ext<mode>_1)
(*extzv<mode>, *insvqi_2, *extendqi<SWI24:mode>_ext_1)
(*addqi_ext<mode>_1_slp, *addqi_ext<mode>_1_slp, *addqi_ext<mode>_0)
(*addqi_ext2<mode>_0, *addqi_ext<mode>_1, *<insn>qi_ext<mode>_2)
(*subqi_ext<mode>_1_slp, *subqi_ext<mode>_2_slp, *subqi_ext<mode>_0)
(*subqi_ext2<mode>_0, *subqi_ext<mode>_1, *testqi_ext<mode>_1)
(*testqi_ext<mode>_2, *<code>qi_ext<mode>_1_slp)
(*<code>qi_ext<mode>_2_slp. *<code>qi_ext<mode>_0)
(*<code>qi_ext2<mode>_0, *<code>qi_ext<mode>_1)
(*<code>qi_ext<mode>_1_cc, *<code>qi_ext<mode>_1_cc)
(*<code>qi_ext<mode>_2, *<code>qi_ext<mode>_3, *negqi_ext<mode>_1)
(*one_cmplqi_ext<mode>_1, *ashlqi_ext<mode>_1, *<insn>qi_ext<mode>_1)
(define_peephole2): Replace uses of extract_operator with
extract_high_operator, matching only the first operand.
Use zero_extract rather than match_op_dup when splitting.
The following removes hybrid SLP detection - it existed as sanity
check that all stmts are covered by SLP, but it proved itself
incomplete at that. Its job is taken by early terminating SLP
build when SLP discovery fails for one root and the hope that
we now do catch all of them.
* tree-vectorizer.h (vect_relevant::hybrid): Remove.
* tree-vect-loop.cc (vect_analyze_loop_2): Do not call
vect_detect_hybrid_slp.
* tree-vect-slp.cc (maybe_push_to_hybrid_worklist): Remove.
(vect_detect_hybrid_slp): Likewise.
The following records the alternate SLP instance entries coming from
stmts with stores that have no SSA def, like OMP SIMD calls without LHS.
There's a bit of fallout with having a SLP tree with a NULL vectype,
but nothing too gross.
PR tree-optimization/121395
* tree-vectorizer.h (_loop_vec_info::alternate_defs): New member.
(LOOP_VINFO_ALTERNATE_DEFS): New.
* tree-vect-stmts.cc (vect_stmt_relevant_p): Populate it.
(vectorizable_simd_clone_call): Do not register a SLP def
when there is none.
* tree-vect-slp.cc (vect_build_slp_tree_1): Allow a NULL
vectype when there's no LHS. Allow all calls w/o LHS.
(vect_analyze_slp): Process LOOP_VINFO_ALTERNATE_DEFS as
SLP graph entries.
(vect_make_slp_decision): Handle a NULL SLP_TREE_VECTYPE.
(vect_slp_analyze_node_operations_1): Likewise.
(vect_schedule_slp_node): Likewise.
* gcc.dg/vect/pr59984.c: Adjust.
The following renames loop_vect to not_vect, removes the unused
HYBRID_SLP_STMT macro and rewords the slp_vect_type docs to clarify
STMT_SLP_TYPE is mainly used for BB vectorization, tracking what is
vectorized and what not.
* tree-vectorizer.h (enum slp_vect_type): Rename loop_vect
to not_vect, clarify docs.
(HYBRID_SLP_STMT): Remove.
* tree-vectorizer.cc (vec_info::new_stmt_vec_info): Adjust.
* tree-vect-loop.cc (vect_analyze_loop_2): Likewise.
We now never have hybrid SLP (if we have, we fail immediately).
* tree-vect-data-refs.cc (vect_get_data_access_cost): Use
ncopies == 1.
* tree-vect-slp.cc (vect_remove_slp_scalar_calls): Remove
hybrid/loop SLP skip.
* tree-vect-stmts.cc (vectorizable_store): Remove pure SLP assert.
IVOPTs, when replacing an IV, inserts the computation of the new IVs
step in the loop preheader without considering the case of the loop
not iterating. This means we have to ensure the step computation
does not invoke UB. There is also SCEV which does not care about
signed arithmetic UB when re-associating expressions to form CHRECs,
so even when we know the loop iterates this is required.
PR tree-optimization/121382
* tree-ssa-loop-ivopts.cc (create_new_iv): Rewrite the IV
step to defined form.
* gcc.dg/torture/pr121382.c: New testcase.
When there is obvious UB involved in the process of re-associating
a series of IV increments to build up a CHREC, fail. This catches
a few degenerate cases where SCEV introduces UB with its inherent
re-associating of IV increments.
PR tree-optimization/121370
* tree-scalar-evolution.cc (scev_dfs::add_to_evolution_1):
Avoid UB integer overflow in accumulating CHREC_RIGHT.
* gcc.dg/torture/pr121370.c: New testcase.
In LoongArch psABI, large _BitInt(N) (N > 64) objects are only
extended to fill the highest 8-byte chunk that contains any used bit,
but the size of such a large _BitInt type is a multiple of their
16-byte alignment. So there may be an entire unused 8-byte
chunk that is not filled by extension, and this chunk shouldn't be
checked when testing if the object is properly extended.
The original bitintext.h assumed that all bits within
sizeof(_BitInt(N)) beyond used bits are filled by extension.
This patch changes that for LoongArch and possibly
any future ports with a similar behavior.
P.S. For encoding this test as well as type-generic programming,
it would be nice to have a builtin function to obtain "N" at
compile time from _BitInt(N)-typed expressions. But here
we stick to existing ones (__builtin_clrsbg / __builtin_clzg).
* gcc.dg/bitintext.h (S, CEIL, PROMOTED_SIZE): Define.
(BEXTC): Generalize to only check extension within PROMOTED_SIZE bits.
For targets that set the "extended" flag in TARGET_C_BITINT_TYPE_INFO,
we assume small _BitInts to be internally extended after arithmetic
operations. In this case, an extra extension during RTL expansion
can be avoided.
* expr.cc (expand_expr_real_1): Do not call
reduce_to_bit_field_precision if the target assumes the _BitInt
results to be already extended.
(EXTEND_BITINT): Same.
* expr.h (bitint_extended): Declare the cache variable.
* function.cc (prepare_function_start): Initialize it.
For targets that treat small _BitInts like the fundamental
integral types, we should allow their machine modes to be promoted
in the same way.
* explow.cc (promote_function_mode): Add a case for
small/medium _BitInts.
(promote_mode): Same.
On Mon, Aug 04, 2025 at 11:33:17AM -0400, Patrick Palka wrote:
> > @@ -1693,6 +1697,8 @@ export namespace std
> > {
> > using std::ranges::advance;
> > using std::ranges::distance;
> > + using std::ranges::iter_move;
> > + using std::ranges::iter_swap;
>
> Actually a few lines above we already do:
>
> // _Cpo is an implementation detail we can't avoid exposing; if we do the
> // using in ranges directly, it conflicts with any friend functions of the
> // same name, which is why the customization points are in an inline
> // namespace in the first place.
> namespace ranges::inline _Cpo
> {
> using _Cpo::iter_move;
> using _Cpo::iter_swap;
> }
>
> So I think we don't want to export iter_move and iter_swap directly... Sorry
> for not catching this sooner :/
Here is a patch which does that.
2025-08-05 Jakub Jelinek <jakub@redhat.com>
PR libstdc++/121373
* src/c++23/std.cc.in (std::ranges::iter_move, std::ranges::iter_swap):
Remove exports.
Another easy part from the paper.
Part of the CWG2579 has been already done in an earlier paper (with
test commits by Marek) and the remaining part is implemented correctly,
we diagnose as error when token pasting doesn't form a valid token.
Except that message
pasting """" and """" does not give a valid preprocessing token
looked weird and so I've updated the message to use %< and %> instead
of \" quoting.
2025-08-05 Jakub Jelinek <jakub@redhat.com>
PR preprocessor/120778
* macro.cc (paste_tokens): Use %< and %> instead of \" in
diagnostics around %.*s.
* g++.dg/DRs/dr2579.C: New test.
* c-c++-common/cpp/va-opt-6.c: Expect ' rather than \" around
tokens in incorrect pasting diagnostics.
* gcc.dg/c23-attr-syntax-6.c: Likewise.
* gcc.dg/cpp/paste12.c: Likewise.
* gcc.dg/cpp/paste12-2.c: Likewise.
* gcc.dg/cpp/paste14.c: Likewise.
* gcc.dg/cpp/paste14-2.c: Likewise.
The previous code-gen of scalar unsigned SAT_MUL, aka usmul.
Leverage the mulhs by mistake, it should be mulhu for the
hight bit result of mul. Thus, this patch would like to make
it correct.
gcc/ChangeLog:
* config/riscv/riscv.cc (riscv_expand_xmode_usmul): Take
umulhu for high bits mul result.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/sat/sat_u_mul-1-u32-from-u64.c: Add mulhu
asm check.
* gcc.target/riscv/sat/sat_u_mul-1-u64-from-u128.c: Ditto.
Signed-off-by: Pan Li <pan2.li@intel.com>
Here's a previously accepted testcase that is now ambiguous after
r16-2771-gb9f1cc4e119da, since the uninstantiated constraints are
equivalent but the partially instantiated constraints aren't, so
the two member functions no longer correspond.
PR c++/121351
gcc/testsuite/ChangeLog:
* g++.dg/cpp2a/concepts-using6.C: New test.
The b.f(42) calls in the below testcases started to get rejected as
ambiguous after r15-3740 which corrected our inheritedness tiebreaker to
only apply to constructors (and not all member functions) as per CWG2273.
But arguably these calls should still be valid regardless of the
tiebreaker because B::f corresponds to and therefore hides A::f, so
there should only be a single candidate in the first place. This
doesn't happen because when determining correspondence we compare
the members' uninstantiated constraints instead of their partially
substituted constraints as in other declaration matching situations.
It doesn't really make sense to compare uninstantiated constraints
from two different template contexts.
This patch fixes this by substituting in outer template arguments before
comparing constraints of two potentially corresponding member functions.
PR c++/121351
PR c++/119859
gcc/cp/ChangeLog:
* class.cc (add_method): Substitute outer template arguments
into constraints before comparing them if the declarations are
from different classes.
gcc/testsuite/ChangeLog:
* g++.dg/cpp2a/concepts-using5.C: New test.
* g++.dg/cpp2a/concepts-using5a.C: New test.
Reviewed-by: Jason Merrill <jason@redhat.com>
r13-3299 changed our internal declaration of __dynamic_cast to reside
inside the abi/__cxxabiv1:: namespace instead of the global namespace,
matching the real declaration. This inadvertently made us now attempt
constexpr evaluation of user-written calls to abi::__dynamic_cast since
cxx_dynamic_cast_fn_p now also returns true for them, but we're not
prepared to handle arbitrary calls to __dynamic_cast, and therefore ICE.
This patch restores cxx_dynamic_cast_fn_p to return true only for
synthesized calls to __dynamic_cast, which can be distinguished by
DECL_ARTIFICIAL, since apparently the synthesized declaration of
__dynamic_cast doesn't get merged with the actual declaration.
PR c++/120620
gcc/cp/ChangeLog:
* constexpr.cc (cxx_dynamic_cast_fn_p): Return true only
for synthesized __dynamic_cast.
gcc/testsuite/ChangeLog:
* g++.dg/cpp2a/constexpr-dynamic19.C: New test.
* g++.dg/cpp2a/constexpr-dynamic1a.C: New test.
Reviewed-by: Jason Merrill <jason@redhat.com>
The old GET_MODE_SIZE (DImode) (i.e. 64) made sense before
64-bitters. Now the default is just a trap: when using the
default 64, things like TImode (128 bits) still mostly works,
but general corner cases related to computing large-size objects
numbers, like (1 << 64)/8 break, as exposed by
gcc.dg/pr105094.c.
So, keep the floor at 64 for 32-bitters and smaller targets, but
for larger targets, make it 2 * BITS_PER_WORD. Also, express it
more directly with focus on BITS_PER_WORD, not the size of a
mode. Add "by GCC internally" in an attempt to tell that this
is when gcc cooks something up, not when plain input uses a type
with such a mode.
* defaults.h (MAX_FIXED_MODE_SIZE): Default to 2 * BITS_PER_WORD
for larger-than-32-bitters.
* doc/tm.texi.in (MAX_FIXED_MODE_SIZE): Adjust accordingly. Tweak
wording.
* doc/tm.texi: Regenerate.
On Sat, Aug 02, 2025 at 09:05:07PM +0200, Jakub Jelinek wrote:
> Wonder how to automatically discover other missing exports (like in PR121373
> std::byteswap), maybe one could dig that stuff somehow from the raw
> dump (look for identifiers in std namespace (and perhaps inlined namespaces
> thereof at least) which don't start with underscore.
To answer that question, I wrote a simple plugin which just dumps the names
(which do not start with underscore) in std namespace (and its inlined
namespaces) and for non-inline namespaces in there which do not start with
underscore also recurses on those namespaces.
Plugin source in
https://gcc.gnu.org/pipermail/libstdc++/2025-August/062859.html
I went through it all now, using cppreference as a quick check for stuff
removed in C++17/C++20 and for everything added verified it is in
corresponding eel.is/c++-draft/*.syn etc. and looked it up in the libstdc++
headers for guarding macros.
After all the additions I've compiled std.cc with -std=c++20, -std=c++23 and
-std=c++26, the first one revealed std::ranges::shift_{left,right} emitted an
error in that case, the patch fixes that too.
2025-08-04 Jakub Jelinek <jakub@redhat.com>
hexne <printfne@gmail.com>
PR libstdc++/121373
* src/c++23/std.cc.in (std::ranges::shift_left,
std::ranges::shift_right): Only export for C++23 and later.
(std::ranges::fold_left_first_with_iter_result,
std::ranges::fold_left_with_iter_result): Export.
(std::byteswap): Export for C++23 and later.
(std::ranges::iter_move, std::ranges::iter_swap): Export.
(std::projected_value_t): Export for C++26 and later.
(std::out_ptr_t, std::inout_ptr_t): Export.
(std::ranges::iota_result): Export.
(std::regex_constants): Export a lot of constants.
(std::is_scoped_enum, std::is_scoped_enum_v): Export.
