https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122686
--- Comment #13 from Richard Biener <rguenth at gcc dot gnu.org> --- So what's probably most interesting is the GIMPLE stmt we are currently expanding, the topmost thing we are trying to compute the range on and the CFG context with the stmts possibly looked at by ranger. From the GIMPLE stmt and the initial part of the -fdump-rtl-expand-details dump the rest might be infered. So, can you attach the expand dump (it will be not "finished", but it should show relevant parts) and see to print the GIMPLE stmt that is expanded (expand_gimple_stmt)?
