http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47402
--- Comment #4 from Alexandre Oliva <aoliva at gcc dot gnu.org> 2011-01-21 22:35:47 UTC --- Ok, regression now fixed by reversal of the bit that exposed the failure, but the fact that it regressed implies there's something iffy with the patch for bug 47106, so I'll keep this open while I try to figure that out. I'll unfortunately be away tomorrow (back on Sunday), so if the patch for 47106 causes further breakage, please revert the rest of it. Sorry, and thanks in advance,
