http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45794
--- Comment #2 from janus at gcc dot gnu.org 2010-10-02 11:57:49 UTC --- I think this regression is due to r153793, which was Tobias' fix for PR41850. The reason for the ICE is that the formal argument "mask" of "_gfortran_mmaxloc0_4_r4" has "as = NULL" (while the actual argument has nonzero rank). I'm not sure if the missing array spec is special to MAXLOC, or if this is the case for all intrinsics.
