http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47800
--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-02-18 15:11:16 UTC --- Created attachment 23394 --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23394 gcc46-pr47800-1.patch The problem is that for QImode and HImode the peephole2 creates the first of the two insns with mode mismatch. Either we can make the first insn zero extend from QI/HI to SImode, as in this patch...
