[Bug c++/114426] [14 regression] ICE when building log4cxx on arm (cxx_eval_call_expression, at cp/constexpr.cc:3242) since r14-6507

["jakub at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114426

--- Comment #11 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Actually I had another look.
Jason said in the c++: fix in-charge parm in constexpr mail back in December
(as well as in the r14-6507 commit message):
"Since a class with vbases can't have constexpr 'tors there isn't actually a
need for an in-charge parameter in a destructor"
but the ICE is because the destructor is marked implicitly constexpr.
https://eel.is/c++draft/dcl.constexpr#3.2 says that a destructor of a class
with virtual bases is not constexpr-suitable, but we were actually implementing
this just for constructors, so clearly my fault from the
https://wg21.link/P0784R7 implementation.  That paper clearly added that
sentence in there and removed similar sentence just from the constructor case.
--- gcc/cp/constexpr.cc.jj      2024-04-09 09:29:04.708521907 +0200
+++ gcc/cp/constexpr.cc 2024-04-12 11:45:08.845476718 +0200
@@ -262,18 +262,15 @@ is_valid_constexpr_fn (tree fun, bool co
        inform (DECL_SOURCE_LOCATION (fun),
                "lambdas are implicitly %<constexpr%> only in C++17 and
later");
     }
-  else if (DECL_DESTRUCTOR_P (fun))
+  else if (DECL_DESTRUCTOR_P (fun) && cxx_dialect < cxx20)
     {
-      if (cxx_dialect < cxx20)
-       {
-         ret = false;
-         if (complain)
-           error_at (DECL_SOURCE_LOCATION (fun),
-                     "%<constexpr%> destructors only available"
-                     " with %<-std=c++20%> or %<-std=gnu++20%>");
-       }
+      ret = false;
+      if (complain)
+       error_at (DECL_SOURCE_LOCATION (fun),
+                 "%<constexpr%> destructors only available with "
+                 "%<-std=c++20%> or %<-std=gnu++20%>");
     }
-  else if (!DECL_CONSTRUCTOR_P (fun))
+  else if (!DECL_CONSTRUCTOR_P (fun) && !DECL_DESTRUCTOR_P (fun))
     {
       tree rettype = TREE_TYPE (TREE_TYPE (fun));
       if (!literal_type_p (rettype))
patch fixes the ICE too, just will need to add testcase coverage and see what
regresses in the testsuite...