https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94779
--- Comment #11 from Jakub Jelinek <jakub at gcc dot gnu.org> --- Related PR is PR89059 though, while we can have a useful range info already in the early opts from evrp, in many cases we can get much better info after inlining. So, if we during the first switchconv pass could just perform analysis and stick on the gswitch info that e.g. the inlining cost computations could use and only perform lowering later, or if we could perform the conversion right away but in some way note for the second switchconv pass that this and this came from switch conversion and that if the range info is narrowed second switchconv pass should attempt to adjust it, it would be really appreciated.
