https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109945
--- Comment #22 from Richard Biener <rguenth at gcc dot gnu.org> --- Ping. Jakub/Jason - we now have three open bugs because of this. The patch in comment#12 works but Jakub doesn't like it. Instead as far as I understand we need a flag to tell us whether there's guaranteed copy elision (in which case the programmer can expect using the return slot address will "work", maybe be even well-defined). As can be seen with the testcases TREE_ADDRESSABLE on the type does not cover all cases where that's appearantly so. Maybe those types simply need to have TREE_ADDRESSABLE set as otherwise RTL expansion or gimplification might introduce an additional copy, breaking copy elision?
