On Mon, Mar 18, 2002 at 02:36:31PM -0800, Alfred Perlstein wrote: > * Dag-Erling Smorgrav <[EMAIL PROTECTED]> [020318 08:23] wrote: > > Alfred Perlstein <[EMAIL PROTECTED]> writes: > > > I think you're right, I'm pretty sure the fix is basically moving > > > the p->p_fd = NULL to after the closef will fix things [...] > > > > There will still be a race... > > Are you sure? :) > > Btw, is there a way to easily reproduce this bug?
The panic in tail was triggered by using -f (i.e. kqueue), but it's only happened once on the cluster..err..twice now (just happened again). Without your previous patch several cluster machines were failing several times per hour, in umount. You could probably trigger it by stressing these two code paths. I'll test your latest patch a bit later on. Kris
msg36321/pgp00000.pgp
Description: PGP signature