https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113596
--- Comment #11 from Jakub Jelinek <jakub at gcc dot gnu.org> --- In reply to Richard Biener from comment #10) > We can make ->calls_alloca more precise though of course > we usually also do not want to inline functions with VLAs. Yes. Or remember while inlining whether we've seen a non-VLA alloca call in the id structure and then just emit both calls afterwards to the corresponding blocks. > IIRC a VLA forces a frame pointer for the caller then. The patch doesn't change whether to inline such functions or not, that would be for incremental discussions. Supposedly sometimes it could be worth it even when needing a frame pointer because of that, but it should be taken into account in the cost heuristics or something.
