http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52760
--- Comment #5 from rguenther at suse dot de <rguenther at suse dot de> 2012-03-29 10:50:49 UTC --- On Thu, 29 Mar 2012, jakub at gcc dot gnu.org wrote: > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52760 > > --- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-03-29 > 10:39:24 UTC --- > But where else? The op is copied from the stmts (where it is certainly ok as > is) just a few lines above this. > Note, using VCE if the types have different sizes is invalid anyway. Yes, what I'm saying is that we are not expecting different sizes here. Thus, similar to handling COND_EXPR specially we should handle shifts specially and promote their shift amount. Richard.