Here's a long shot.
What if something is trying to write to the display when the display has
been blanked (ie in powersave mode)?
Theoretically it should power back up again. Maybe it doesn't.
I had a problem on another X system (hummingbird on NT) where the X server
on the client would crash after a while if the client's window was
minimised.
It turned out to be a blinking cursor where the remote server would send
blinking X signals to the client machine, whose X server could not put the
states out because of minimisation.
I disabled the blinking cursor and fixed that problem.
Similarly, if the NT box was switched off without logging out, the X process
(on a solaris box this time) would hog one entire CPU until I killed the
client's X process. At least on a dual CPU machine you can see what's going
on. On a single CPU machine you hae to re-boot....
Am I right in guessing that this is how X protocol is defined? ie if X state
changes are unable to get through due to dead client, disconnected machine
etc then the states are backed up in a stack somewhere in the hope that the
remote end might come back, rather like mail on internet.
If that's the case, then perhaps we need to re-visit the X implementation
and create a separate queue (ie separate from /tmp or swap) with a fixed
size so that the os does not suffer from fill-up. In the case of the
Hummingbird software, I believe that's exactly what they do as they know
they are running on an untrustworthy os!
- Jill.
___________________________________________
Jill Rowling
Snr Design Engineer & Unix System Administrator
Electronic Engineering Department, Aristocrat Technologies Australia
3rd Floor, 77 Dunning Ave Rosebery NSW 2018
Phone: (02) 9697-4484 Fax: (02) 9663-1412
Email: [EMAIL PROTECTED]
-----Original Message-----
From: Jeff Waugh [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, 5 September 2000 1:53
To: [EMAIL PROTECTED]
Subject: Re: [SLUG] Sawmill keeps logging out after periods of
inactivity
> James Wilkinson wrote:
> It only seems to happen when the kernel has blanked the screen, never
> while it's still running. Perhaps the interaction of the kernel and my
> video card and X are causing X to die?
--
SLUG - Sydney Linux User Group Mailing List - http://slug.org.au/
More Info: http://slug.org.au/lists/listinfo/slug
