On Fri, Mar 30, 2018 at 3:55 AM, Alexandre Oliva <aol...@redhat.com> wrote: > Types defined within a __builtin_offsetof argument don't always get > properly recorded as members of their context types, so if they're > anonymous, we may fail to assign them an anon type index for mangling > and ICE. > > We shouldn't allow types to be introduced in __builtin_offsetof, I > think, so I've arranged for us to reject them. Even then, we still > parse the definitions and attempt to assign mangled names to its > member functions, so the ICE remains. Since we've already reported an > error, we might as well complete the name assignment with an arbitrary > index, thus avoiding the ICE. > > > Regstrapped on i686- and x86_64-linux-gnu, regressing > g++.dg/parse/semicolon3.C, which defines a (named) struct in > builtin_offsetof. I suppose this means I should look for another > solution that doesn't involve rejecting these definitions, eh?
Hmm, I'm afraid so. The C standard defines offsetof in terms of a notional declaration static /type/ t; and a type definition is allowed in such a declaration. Incidentally, it would be nice to replace all the type_definition_forbidden stuff with defining-type-specifier as per DR 2141... (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50169) Jason
