[71] in testers
Re: RT 6.3 AFS
daemon@ATHENA.MIT.EDU (daemon@ATHENA.MIT.EDU)
Sun Jul 9 13:07:15 1989
From: <probe@ATHENA.MIT.EDU>
Date: Sun, 9 Jul 89 13:07:01 -0400
To: jfc@ATHENA.MIT.EDU
Cc: testers@ATHENA.MIT.EDU
In-Reply-To: John F Carr's message of Sat, 01 Jul 89 11:54:20 EDT,
Reply-To: Richard Basch <probe@ATHENA.MIT.EDU>
> From: <jfc@ATHENA.MIT.EDU>
> Date: Sat, 01 Jul 89 11:54:20 EDT
> Twice recently my kernel has gone into an infinite loop on
> "AFS tokens for 2931 have expired". This may or may not mean something,
> since I built my own kernel, using hc2 for all parts except AFS. I was
> able to stop it the first time by running aklog; the second time I had
> to reboot.
Well, it just happened to me too on Odie... I, too, have a custom
kernel, but I certainly did not mix compilers... it looks like a rare
case bug to me since it is the first time this happened.
Talking about bugs... I just got my first "pagein" bug on Odie too...
now if it can be reproduced...
Now that we are on memory issues; perhaps, we should try an experiment:
the only thing that has changed since 6.0C (6.0R?) about memory
management that may resemble the fragmentation, is Don's fix to the
"panic: bad rmfree" bug... sys/vm_{drum,text}.c. I think this may want
to be investigated; in fact, Don was pondering about his fix immediately
when I talked to him about the memory fragmentation. What happens to
the memory fragmentation when this code is backed out? (I know that
this will be an inadequate solution since the "bad rmfree" panic is
clearly unacceptable).
-Richard
