On Mon, 15 Mar 2004, Garance A Drosihn wrote: > At 12:24 AM -0500 3/10/04, Garrett Wollman wrote: > >Once you get enough activity to start to recycle vnodes (and AFS > >vcache entries) it will probably fall over pretty fast. Adding > >WITNESS and DEBUG_VFS_LOCKS may make the bugs more obvious. > > I'm building a second machine for stress-testing this (my main > maachine has too much important data on it).
Well, now with 5.2.1 and a WITNESS, debug enabled kernel, and a new debugging build of afs, I can still reproduce the mkdir panic I was having earlier: 1) reboot, log in as root 2) kldload /usr/local/lib/openafs.ko 3) afsd 4) cd /afs/@cell/some_dir 5) mkdir foo (fails, read-only filesystem) 6) cd panic: vrele: negative ref cnt panic vrele kern_chdir chdir syscall Xint0x80_syscall all the numbers in this panic trace left out, since I don't feel like typing them over. Note that trying to create a _file_ on a RO FS doesn't trigger any problem, just directories. _______________________________________________ OpenAFS-devel mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-devel
