According to Jan Harkes:
>
>Emacs is sending itself signals that tend to interrupt long running
>upcalls to userspace. As far as I know our kernel code knows about this
>and explicitly ignores the signals that emacs is triggering for a while.
>

Hmmm - I did try tracing the process and it looked like it was looping
on doing select/poll/gettime calls, it is hard for me to tell whether
this was part of the normal X event loop or if emacs is stuck
somewhere else.  Hmmm I just tried emacs on a plain ascii console and
that seems to behave itself.

>
>That is very unusual, I can't really explain this part. Maybe upcalls
>take too long because of the optimization that takes out duplicate
>stores from the CML.

Maybe - the laptop is not exactly the fastest machine about (it's a
P150 DEC HiNote Ultra II), I can try this on my server which has a bit
more grunt and see if I see the same problem.  At the moment I am
suspecting it is more a wierd emacs+X+coda interaction, perhaps there
is a race condition in there somewhere I am triggering.

> How large is your client cache?
>

20Mb max, I am using about 143Kbytes at the moment.

-- 
===============================================================================
Brett Lymn, Computer Systems Administrator, BAE SYSTEMS
===============================================================================