-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
krjw wrote:
| The issue I've seen is reproducable like so:
| 1) boot linux box
| 2) enable use of kernel AGPGART (not internal nvidia AGPGART code) in
| XF86Config (load agpgart kernel module if needed)
| 3) Start X
| 4) Leave X
| 5) enable use of internal nvidia AGPGART code in XF86Config
| 6) Start X
| 7) choke. have to log in from remote box to reboot.
|
| (2) and (5) imply changing the value of the option NvAgp in XF86Config.
Nope, I remember I have tried both the nvidia internal agp support and
agpgart, but I think I always rebooted between each change. The
"mem=nopentium" option, albeit not an optimal solution, cured the
lockups, still I had "spontaneous" session ending (especially when using
xvideo). Now, the problem I have with the latest nvidia driver is much
more serious, and it seems I'm not the only one experiencing this (Mike
Stilson in another reply is confirming the same problem, with a
different kernel, different compiler, different distribution and no agp).
| Why it chokes, I don't know. Maybe it has something to do with the AMD
| AGP issue you quoted and the "inadvertant trashing of AGP memory".
| Does anyone else know more about this[1]? Perhaps the NV guys on the
| list? How does this problem manifest itself? Segfaults? System
| lockups?
This particular problem ("solved" with the mem=nopentium option)
manifested itself with an X lockup, the machine was still running fine
and nothing was registered in the logs, only a reboot recovered the
graphic subsystem, though. I don't know if there were other problems, I
didn't notice.
The current problem that leaves the message "kernel BUG at page_alloc.c:
xxx" is more subtle: everything seems to be running fine but then
something strange happens, e.g: I cannot su or I cannot login or I
cannot list the content of a directory (the ls locks up and is
unkillable). When that happens, I go looking in the logs and there it
is, a "kernel BUG" some time (1 minute, 1 hour, 5 hours..) before the
strangeness began.
"shutdown -r" doesn't work (halfway through it locks up, on occasions it
never starts) and a reset is the only option.
When kernel memory is corrupted everything could happen.
Bye
- --
Luca Olivetti
Note.- This message reached you today, it may not tomorrow if you
are using MAPS services. They arbitrarily include in their lists
IP addresses not related in any way to spam, and in so doing are
disrupting Internet connectivity. Please stop supporting them.
See http://slashdot.org/article.pl?sid=01/05/21/1944247
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: http://www.gnupg.org
iD8DBQE9PHYaCQPXTRx9NmQRAtklAJ4lH+8jf9sPLMHqnqxDpVKW4MYyIQCaA5sI
UNWOeOFmpMLDSvZXd+Oqfh4=
=LrZV
-----END PGP SIGNATURE-----
_______________________________________________
Xpert mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/xpert
