[plug] _hard_ crashes with X

[Craig Ringer]

Hi all
I'm working on a remote-boot setup for work (its actually really easy to 
turn Debian into a great diskless system with X etc) and am running into 
a weird problem on one machine.

I've got this setup working on other machines so it is not broken for 
all, but on a spare Athlon 500 w AGP TNT2 X just crashes the machine 
hard. Same if I swap the video card for an S3 Savage4. I've set up a 
serial console and can confirm that the kernel isn't panicing unless 
I've done something wrong in the console setup. X -probeonly is alright, 
but when it tries to actually fire up the machine goes down (stops 
responding on the serial console, no response to pings, etc). Its 
driving me nuts. Same problem with the SVGA XF86v3 server.

This machine should have no problems with the setup, since I've had a 
486DX2/66 running well enough to happily display KDE + OpenOffice + 
Mozilla (*woohoo*) admittedly over an XDMCP-controlled remote X session.

Anybody seen anything like this before?

On a related note, the only problem I'm having with this system on _all_ 
machines is swap. Most of the machines I plan to use have 16mb of RAM 
and swapping to disk is not an option, nor is a RAM upgrade. Has anybody 
here worked with network swap? I gather that nbd can't do it, and there 
are known problems with swap over nfs - so is there any decent way to do 
it with linux at the moment?

Thanks all.

-- 
Craig Ringer
GPG Key Fingerprint: AF1C ABFE 7E64 E9C8 FC27  C16E D3CE CDC0 0E93 380D
	-- if it ain't broke, add features 'till it is. (or:)
	while (! broken) { features ++ ; broken = isBroken(features) }