Re: 10/amd64 lockup with lots of rpcbinds in fstchg

[Brian Buhrow <buhrow%nfbcal.org@localhost>]

	hello Edgar.  I think this is the same bug I'm running into with 10.99.12/amd64.  I
haven't figured out how to reproduce it at will as yet, but I think it is a race condition
related to the revoke(2) system call.  I'm still trying to figure out the exact set of
conditions which produce the issue, but they seem to have  something to do with the sshd
session driving a user pty terminating while the user processes on that pty are still running.
Jason thorpe suggested this might be a race condition in the pty driver and a work around might
be to use the ptyfs filesystem for pseudo termnals rather than the legacy /dev/pty devices, not
so much to solve the problem, but to keep the machine from locking up when the condition is
triggered.  That might be a way to keep things running when the bug fires and also give you
way to troubleshoot what's going on while some of the processes are in fstchg state.

I posted about this to the tech-kern mailing list a few weeks ago if you want to read about it
in more detail.

-thanks
-Brian