crash: rh8/driver 3123

[lnxxprt]

hi there

i have this problem a longer time now. my system
crashes with the nvidia drivers since the beta of
rh8. it is not easy to reproduce. it all begins with
a restart of the X server. after that you can bet
that the system will crash hard. no log entries,
no mouse movement, X frozen. only chance is to
reboot. i think it has something to do with switching
to fullscreen mode with an arbitrary app.

today i experienced another strange problem with
the xmms plugin iris which uses opengl. it crashes
the X server. sysrq+e helps to get rid of the dead
X but a reboot is necessary.

I am not very happy about the nvidia driver :(

system: 384mb ram, tnt2, pIII 667mhz, rh8.

can someone please fix this?

[junkieclown]

Knowing the name of the motherboard chipset is much more important than knowing your cpu rating.

Does the ouptut of cat /proc/interrupts show other devices on the same irq as the nvidia card? you will only see the nvidia card in /proc/interrupts after starting X.
nvidia cards do not like to share their irqs -and some other cards are very badly behaved in that regard too, so when they meet there is trouble.

What agp driver option are you using? Some motherboards can use the Nvidia supplied agp driver and some must use the linux kernel supplied agpgart driver. (this is why it's important to know your mobo chipset's name) rh8 comes with agpgart compiled as a module (agp.o) so you don't need to mess with recompiling the kernel source at least.
the AGP driver option is specified in /etc/X11/XF86Config as
Option "NvAgp=somenumber"
the default behavior is to try FIRST to use the linux kernel agpgart driver then fall back to nvidia's agp, if agpgart doesn't initialize.
This default is the same as "NvAGP=3"

the other options are :
"NvAgp =2" which means that only the linux kernel agpgart driver will be used instead of the nvidia module

"NvAGP= 1" use only the nvidia module and never try the linux agp.o driver

"NvAGP=0" use NO agp driver at all.

If you turn off agp altogether and still get X crashes as I did, _and_ your nvidia card is sharing an irq with another device(s) then you probably have a pci steering irq logjam.
Try moving your pci cards to different slots and enabling ESCD in your bios so they can be rerecognized in their new locations.
I can't get my nvidia card to an irq of its own, but getting the Promise ata adapter card off that same irq seems to help stability for me.


Make sure X is using the appropriate agp driver. a list of boards can be found
at the gentoo linux nvidia troubleshooting guide
------------------------------------
Intel

440BX NVAGP / AGPGART # doesn't work for some people
810 NVAGP / AGPGART
815(e) NVAGP / AGPGART # w/ 815 patch

VIA

Apollo AGPGART / NVAGP
MVP3 AGPGART
MVP4 AGPGART
KX133 NVAGP
KT133 NVAGP

Ali

... AGPGART # 2.4.0+, 2.2.18 w/ ALi patch

SiS

... AGPGART # ?

---------------------------

I retrieved that from google's cache, as the page seems to be down.

Also there should be no options enabled in your XF86Config regarding DRI. There shouldn't be a reference to GLcore in your config. On the other hand the option Load glx should be present . that much is clear from the readme that accompanies the nvidia glx package.

Hope that helps.

[lnxxprt]

hi junkieclown

thanks for your fast reply.

what makes me wonder is that i didnt change anything in my system and it was stable until i upgrade to the beta of rh8. configuration/irq/...
is all equal.

and the nvidia driver works flawless e.g. after
a reboot.
*only* in case i restart X or i use opengl then
the system will crash. and not immediatly. it
happens after some time. i dont know what triggers the crash.

so i am not sure that it has something to do
with my system settings.

lnxxprt

---------------------------
some more system infos:

output of /proc/interrupts:
10: 11004184 XT-PIC eth0, nvidia

i am using agpgart.

lspci:
00:00.0 Host bridge: VIA Technologies, Inc. VT82C693A/694x [Apollo PRO133x] (rev c2)
00:01.0 PCI bridge: VIA Technologies, Inc. VT82C598/694x [Apollo MVP3/Pro133x AGP]
00:07.0 ISA bridge: VIA Technologies, Inc. VT82C686 [Apollo Super South] (rev 1b)
00:07.1 IDE interface: VIA Technologies, Inc. VT82C586B PIPC Bus Master IDE (rev 06)
00:07.3 USB Controller: VIA Technologies, Inc. USB (rev 0e)
00:07.4 Host bridge: VIA Technologies, Inc. VT82C686 [Apollo Super ACPI] (rev 20)
00:0c.0 Multimedia audio controller: Ensoniq ES1371 [AudioPCI-97] (rev 07)
00:0e.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8029(AS)
00:0f.0 SCSI storage controller: LSI Logic / Symbios Logic 53c875 (rev 26)
00:10.0 Multimedia video controller: Brooktree Corporation Bt848 Video Capture (rev 11)
01:00.0 VGA compatible controller: nVidia Corporation NV5 [Riva TnT2] (rev 15)

[junkieclown]

Originally posted by lnxxprt
hi junkieclown

thanks for your fast reply.

what makes me wonder is that i didnt change anything in my system and it was stable until i upgrade to the beta of rh8. configuration/irq/...
is all equal.

and the nvidia driver works flawless e.g. after
a reboot.
*only* in case i restart X or i use opengl then
the system will crash. and not immediatly. it
happens after some time. i dont know what triggers the crash.

so i am not sure that it has something to do
with my system settings.

lnxxprt

---------------------------
some more system infos:

output of /proc/interrupts:
10: 11004184 XT-PIC eth0, nvidia

i am using agpgart.

lspci:
00:00.0 Host bridge: VIA Technologies, Inc. VT82C693A/694x [Apollo PRO133x] (rev c2)
00:01.0 PCI bridge: VIA Technologies, Inc. VT82C598/694x [Apollo MVP3/Pro133x AGP]
00:07.0 ISA bridge: VIA Technologies, Inc. VT82C686 [Apollo Super South] (rev 1b)
00:07.1 IDE interface: VIA Technologies, Inc. VT82C586B PIPC Bus Master IDE (rev 06)
00:07.3 USB Controller: VIA Technologies, Inc. USB (rev 0e)
00:07.4 Host bridge: VIA Technologies, Inc. VT82C686 [Apollo Super ACPI] (rev 20)
00:0c.0 Multimedia audio controller: Ensoniq ES1371 [AudioPCI-97] (rev 07)
00:0e.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8029(AS)
00:0f.0 SCSI storage controller: LSI Logic / Symbios Logic 53c875 (rev 26)
00:10.0 Multimedia video controller: Brooktree Corporation Bt848 Video Capture (rev 11)
01:00.0 VGA compatible controller: nVidia Corporation NV5 [Riva TnT2] (rev 15)

Hmm. looks like we may have the same problem --a second device on the nvidia interrupt.

You didn't change something about the system, but did you upgrade the nvidia driver set at the same time? The default NvAGP option has changed:

"...The default is 3 (the default was 1 until after 1.0-1251"

So it used to default you to using the NvAGP (if you had no nvagp option selected in XF86Config) but now it tries
kernel agpgart first. Of course it could be that you have had an option for nvagp set in your configuration file all along so that would be totally irrelevant....but it's the only thing I can think of that could explain how without overtly changing anything you might have different agp used at different revisions of the nvidia drivers.

If you have an open pci slot, you might try rearranging your pci cards and enabling escd reset or 'reset configuration data' in your bios. IRQ sharing is a "known problem" as they say. Unfortunately it's never clear that there's a known solution. I don't think I'll ever manage to get the nvidia card to its own irq in my system because there are no open pci slots to move things to.

[lnxxprt]

hi junkieclown,

yes, i always upgraded the driver to the newest version.
today i edited the x config and changed to nvagp 1
option. same result.

the iris opengl plugin still crashes X.
and the x restart also causes a crash after some time.

i wish nvidia would make the driver open source so
that someone could fix this piece if they cannot do
it. i think it is a bug in the driver itself. the (rh-)kernel and the driver dont play well together.

i am really disappointed now.

lnxxprt