]> git.ipfire.org Git - thirdparty/gcc.git/commit
c++: constexpr init of union sub-aggr w/ base [PR105491]
authorPatrick Palka <ppalka@redhat.com>
Mon, 9 May 2022 13:53:27 +0000 (09:53 -0400)
committerPatrick Palka <ppalka@redhat.com>
Mon, 9 May 2022 13:53:27 +0000 (09:53 -0400)
commit0c7bce0ac184c057bacad9c8e615ce82923835fd
tree0d9ce377d64297ac2850a6089423a6476efd7187
parentfcda0efccad41eba9134c1bd9d024a93d93fb82f
c++: constexpr init of union sub-aggr w/ base [PR105491]

Here ever since r10-7313-gb599bf9d6d1e18, reduced_constant_expression_p
in C++11/14 is rejecting the marked sub-aggregate initializer (of type S)

  W w = {.D.2445={.s={.D.2387={.m=0}, .b=0}}};
                     ^
ultimately because said initializer has CONSTRUCTOR_NO_CLEARING set,
hence the function must verify that all fields of S are initialized.
And before C++17 it doesn't expect to see base class fields (since
next_initializable_field skips over them), so the presence thereof
causes r_c_e_p to return false.

The reason r10-7313-gb599bf9d6d1e18 causes this is because in that
commit we began using CONSTRUCTOR_NO_CLEARING to precisely track whether
we're in middle of activating a union member.  This ends up affecting
clear_no_implicit_zero, which recurses into sub-aggregate initializers
only if the outer initializer has CONSTRUCTOR_NO_CLEARING set.  After
that commit, the outer union initializer above no longer has the flag
set at this point and so clear_no_implicit_zero no longer recurses into
the marked inner initializer.

But arguably r_c_e_p should be able to accept the marked initializer
regardless of whether CONSTRUCTOR_NO_CLEARING is set.  The primary bug
therefore seems to be that r_c_e_p relies on next_initializable_field
which skips over base class fields in C++11/14.  To fix this, this patch
introduces a new helper function next_subobject_field which is like
next_initializable_field except that it never skips base class fields,
and makes r_c_e_p use it.  This patch then renames next_initializable_field
to next_aggregate_field (and makes it skip over vptr fields again).

PR c++/105491

gcc/cp/ChangeLog:

* call.cc (field_in_pset): Adjust after next_initializable_field
renaming.
(build_aggr_conv): Likewise.
(convert_like_internal): Likewise.
(type_has_extended_temps): Likewise.
* class.cc (default_init_uninitialized_part): Likewise.
(finish_struct): Likewise.
* constexpr.cc (cx_check_missing_mem_inits): Likewise.
(reduced_constant_expression_p): Use next_subobject_field
instead.
* cp-gimplify.cc (get_source_location_impl_type): Adjust after
next_initializable_field renaming.
(fold_builtin_source_location): Likewise.
* cp-tree.h (next_initializable_field): Rename to ...
(next_aggregate_field): ... this.
(next_subobject_field): Declare.
* decl.cc (next_aggregate_field): Renamed from ...
(next_initializable_field): ... this.  Skip over vptr fields
again.
(next_subobject_field): Define.
(reshape_init_class): Adjust after next_initializable_field
renaming.
* init.cc (build_value_init_noctor): Likewise.
(emit_mem_initializers): Likewise.
* lambda.cc (build_capture_proxy): Likewise.
* method.cc (build_comparison_op): Likewise.
* pt.cc (maybe_aggr_guide): Likewise.
* tree.cc (structural_type_p): Likewise.
* typeck2.cc (split_nonconstant_init_1): Likewise.
(digest_init_r): Likewise.

gcc/testsuite/ChangeLog:

* g++.dg/cpp0x/constexpr-union7.C: New test.
* g++.dg/cpp0x/constexpr-union7a.C: New test.
* g++.dg/cpp2a/constinit17.C: New test.
15 files changed:
gcc/cp/call.cc
gcc/cp/class.cc
gcc/cp/constexpr.cc
gcc/cp/cp-gimplify.cc
gcc/cp/cp-tree.h
gcc/cp/decl.cc
gcc/cp/init.cc
gcc/cp/lambda.cc
gcc/cp/method.cc
gcc/cp/pt.cc
gcc/cp/tree.cc
gcc/cp/typeck2.cc
gcc/testsuite/g++.dg/cpp0x/constexpr-union7.C [new file with mode: 0644]
gcc/testsuite/g++.dg/cpp0x/constexpr-union7a.C [new file with mode: 0644]
gcc/testsuite/g++.dg/cpp2a/constinit17.C [new file with mode: 0644]