Comment 40 for bug 430694

As others have commented earlier, this bug appears to be somewhat random (a race condition?) and so a few successful boots don't mean the bug is fixed. If #32 doesn't do essentially more than #2, it fails every now and then at least here and as such I'm not surprised of comments #12, #31, #37.

I'm with #29 in this and would like to know if it's possible that the module load order doesn't matter as much as whether intel_agp has settled before i915 needs it. If so, could a work-around be udevadm settle or sleep 1 between the module loads? (Specifically, I'm now testing 'install i915 /sbin/modprobe -qb intel-agp ; sleep 1 ; /sbin/modprobe --ignore-install i915 $CMDLINE_OPTS' in /etc/modprobe.d/i915.conf, and at least it does no harm.)