https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92264
--- Comment #3 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> --- Similar to what Richard says, this sounds like a latent bug. One of the effects of that rev was to prevent unnecessary invalidation of equivalences based on the stack pointer and frame pointer (which stay the same after a call). It sounds like this has triggered something that could have happened anyway if there weren't any calls in the way. If you have handy access to the reproducer, is it -g that makes the difference? var-tracking can generate very large cselib tables and I'm wondering if it comes from there.
