On Thu, Apr 10, 2003 at 09:42:42AM -0700, Brian Buhrow wrote:
> 	Hello folks.  Following up on this thread, I believe my understanding
> of things is correct, and that I'm actually seeing another manifestation of
> kern/20390.

Not in current (well not the same bug) the callout code is a rather
differnt version now!

Of course, if TCP (or anyone else for that matter) is forgetting
to stop a callout before freeing the data area all bets are off...

Since all the callout table lists are (IIRC) double linked through
a null item, it ought to be possible to verify the back pointers
whenever anything is removed from a list.

	David

-- 
David Laight: david@l8s.co.uk