On 1/21/26 5:24 PM, Jakub Jelinek wrote:
Hi!The following testcase ICEs when we evaluate a = {CLOBBER} stmt. The code assumes that if type is an aggregate type and *valp is non-NULL, then it must be a CONSTRUCTOR. That is usually the case, but there is one exception, *valp can be a PTRMEM_CST if TYPE_PTRMEMFUNC_P (type) and in that case CONSTRUCTOR_ELTS (*valp) obviously ICEs or misbehaves. Now, while I could do something like if (*valp && (!TYPE_PTRMEMFUNC_P (type) || TREE_CODE (*valp) != PTRMEM_CST)) just making sure TREE_CODE (*valp) == CONSTRUCTOR seems much easier and more readable. Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
OK
2026-01-21 Jakub Jelinek <[email protected]> PR c++/123677 * constexpr.cc (cxx_eval_store_expression): Only clear CONSTRUCTOR_ELTS (*valp) if *valp is CONSTRUCTOR. * g++.dg/cpp2a/pr123677.C: New test. --- gcc/cp/constexpr.cc.jj 2026-01-15 16:33:46.971098572 +0100 +++ gcc/cp/constexpr.cc 2026-01-20 17:33:25.928219375 +0100 @@ -8197,7 +8197,7 @@ cxx_eval_store_expression (const constex { if (AGGREGATE_TYPE_P (type)) { - if (*valp) + if (*valp && TREE_CODE (*valp) == CONSTRUCTOR) CONSTRUCTOR_ELTS (*valp) = nullptr; else *valp = build_constructor (type, nullptr); --- gcc/testsuite/g++.dg/cpp2a/pr123677.C.jj 2026-01-20 17:37:35.056959819 +0100 +++ gcc/testsuite/g++.dg/cpp2a/pr123677.C 2026-01-20 17:37:13.330331372 +0100 @@ -0,0 +1,7 @@ +// PR c++/123677 +// { dg-do compile { target c++20 } } + +struct B { void foo (); }; +typedef void (B::*A) (); +struct C { constexpr C (A d) { auto e = new A (d); e->~A (); } }; +C c { &B::foo }; Jakub
