The 2.6.23-0.14.rc0.git1.fc8smp is actually 2.6.22-0.14.rc0.git1.fc8smp -Joseph
The same on x86_64 - 2.6.23-0.14.rc0.git1.fc8 is 2.6.22-0.14.rc0.git1.fc8 and its module signatures say so. Thus on a startup it attempts to find modules.dep and various modules in /lib/modules/2.6.22-0.14.rc0.git1.fc8/ but such directory does not exist. As a result quite a bit of a user space does not work at all. Something funny happened in build process?
Should be fixed in 2.6.23-0.20.rc0.git2.fc8.
> Should be fixed in 2.6.23-0.20.rc0.git2.fc8 Hopefuly you are right. Rawhide from today has 2.6.23-0.15.rc0.git2.fc8 and it suffers from the same affliction.
Its definitely fixed in rawhide (was a minor oversight of one more spec change needed for the new versioning scheme), but there are some compile issues keeping the latest rawhide kernel bits from actually building. This particular problem is definitely fixed in cvs though. *** This bug has been marked as a duplicate of 248189 ***