More testing reveals (Yitzhak Bar Geva). Some good news

[Yitzhak B. G. <yit...@ya...>]

Nice work on the stress testing, James.
We must get to the bottom of this one of these days.
Nvidia has not been responsive so far on our repeated
reports of X crashes with multiple PCI boards. They
don't even have a debugging procedure for multiple
local X servers despite my and Mike Pardee's repeated
requests.
We have applied Aivils' patch to the Xorg X server and
the results are very good. Specifically, Aivils didn't
solve anything, rather he provided a workaround
whereby logout (sending SIGHUP to the X server) will
no longer force reset of the hardware video device. I
doubt you will see a difference when performing
<crtl><alt><bksp>, since that does force a hardware
reset, but repeated crashes resulting from logins and
logouts will either not occur at all or will become
less frequent.
Anyone wanting help in implementing aivils' patch is
welcome to approach me for assistance.
Yitzhak