On Wed, 2003-06-04 at 20:01, Andrew MacLeod wrote: > Thats interesting... they didn't fail on my run. must be related to or > triggered by what you did.. seems odd tho. > *shrug* Could be. On tls/opt-6.c we're ICEing due to an unrecognized insn. On gcc.dg/20020531-1.c we are ICEing in output_constant_pool. Easy enough to fix. Diego.