Nvidia 6200 and ATI RS480/482 chipset incompatibilities

[netllama]

biosirq [IA-32] Use PCI BIOS calls to get the interrupt
routing table. These calls are known to be buggy
on several machines and they hang the machine
when used, but on other computers it's the only
way to get the interrupt routing table. Try
this option if the kernel is unable to allocate
IRQs or discover secondary PCI buses on your
motherboard.

usepirqmask [IA-32] Honor the possible IRQ mask stored
in the BIOS $PIR table. This is needed on
some systems with broken BIOSes, notably
some HP Pavilion N5400 and Omnibook XE3
notebooks. This will have no effect if ACPI
IRQ routing is enabled.

If using either of the two options helps, then you're hitting one or more BIOS bugs (without them).

The Composite extension was never enabled by default until very recent distributions (FC6, SLED10, etc). If disabling Composite now seems to help, that seems to be unrelated to whatever original problems people were reporting in the past.

[MMRRMM@terra.es]

Originally Posted by elinran

Trying pci=biosirq now. fps in glxgears normal speed and closed w/o freezing.
My freezing is so erratic, may be a while before I know anything. I have gone days without a freeze then 3 or 4 in a single day. Mine can freeze while I am typing or not even doing anything with no screensaver even enabled.

Update:
My granddaughter has been playing Tux-Racer for the last hour. Maybe there is hope for this fix??

Any news?

[elinran]

Not really. Tux-racer seems to run fine for as long as you want. I can run multiple glxgears. I tried enabling some GL screensavers they ran for a few hours then it froze with one of them on the screen. Then I tried only enabling non-GL screensavers and all was good all day but found the system frozen the next morning. Have screensavers disabled and not had any freezes for over 24 hours. I have not had any when typing or clicking links like I was having.
My computer, an HP a1310n, with integrated ATI Raedon Express 200 (with official ati driver) only froze when you shutdown or rebooted -- black screen if you tried ctrl/alt backspace also. I originally installed the nvidia card so that I could shutdown without having to use the power off on the tower. With the eGeForce7300GS it shuts down and reboots fine. The freezes seem to be under control by not enabling screensavers. Screensavers are not that big of an issue, but I have an AMD64 3700 and it should be able to do whatever I want to do.

[lithiumfx]

I'm using an Nvidia 6200SE, with TurboCache, on Linux 2.6.19 with X11R7 (7.1.1). I'm currently using the 9742 BETA release of the driver (this doesn't fix the issue, by the way), but have been using the driver since the 8xxx series. I experience the same lockups, freezes and frustration, and it's clear from my own findings and this topic that the issue is a software one, not hardware. I dual boot with Windows XP for gaming and specific software, and experience no problems with my hardware configuration or Nvidia drivers.
Although the problems we are experiencing are only occurring in an environment with the ATI chipset and an Nvidia card, it doesn't mean that this is the cause. It's simply a bug that is induced by environments with the ATI chipset. Nvidia will probably not fix this bug for a while, as they seem to have no idea as to how to fix it. I appreciate that Linux users with an ATI chipset are an extremely low percentage of the Nvidia user base, but a bug that causes system crashes should be resolved as soon as possible - imagine if we were using mission critical systems. Software bugs are obviously being fixed (slowly), as each new driver release improves stability. For this reason, if no other, this is obviously a software bug. If it was the hardware causing the problem, stability would likely be far worse, and would affect any OS.

Perhaps a petition is in order, as Nvidia obviously don't want to fix this problem - even after topics such as this one. It's a software issue, can be replicated easily, affects nearly all users with this hardware configuration, and, is a bug that, if on a Windows platform, would be fixed within hours.

[sniart]

I don't know if a petition will count for something, but let's do it.
lithiumfx, go forward and tells us where we sign.

[alaric]

Hi,

I want to join the ranks here. I have just read the whole thread and my problems are exactly the same (with a few differences outlined below).
Most notably:

I am not using 3D apps - but once I ran glxgears full screen and it locked up the machine in about 10 minutes.

The machine seems stable - but always locks up when I go to bed. I have had no lockups while working with the machine except while running that glxgears application, and once while watching a video - something froze, but I was able to kill mplayer and restart the video (fast forward and continue watching).

This machine is rock solid stable with a FX5200. The instability started immediately after I put in a brand new FX6200 (AGP8x card). I have a motherboard with AGP4x by the way.

A very important difference that might interest a few people here is that I am NOT using the ATI RS480/482 chipset. I am using the intel 845* (Brookdale) chip set. Perhaps Lonni can test with that too, please?

My BIOS code says 6A69VH29C. This means:
6: Medallion BIOS v6.00 (in fact, that is Award BIOS v6.00PG)
A: ISA/PCI
6: Pentium class
9: Intel
V: (together with the above): intel 845 845E 845G 845GE 845GL 845PE Brookdale chipset.
H2: manufactorer: Holco (Shuttle)

I am currently using BIOS AB40S01I

Another difference is that I need to use Option "NvAGP" "0"
I would very much like to know if this makes a difference for others here as well? Are your instabilities related to the use of AGP? What rate are you using (4x, 8x ?). If I use AGP (only could test agpgart) then the system is extremely unstable (Xid's).

Finally - because my machine was dead in the morning every time. Today I exitted X before I went to bed. When I woke up my machine was still running, but there had been harddisk errors?!? Partitions where mounted read-only. Rebooting the system result in fixing the partitions (ext3) except for one: my /var partition had severe damage. Fortunately - it seems to have been isolated to mail (/var/spool/mail).

I investigated my logs and found that harddisk errors had started IMMEDIATELY after inserting the FX6200 -- BEFORE starting X OR LOADING THE NVIDIA DRIVERS!

In other words, no way in hell that I can blame the nvidia drivers for this.

Here are the logs. The FX6200 was put in just before the reboot at Dec 9 05:15:49. As you can see there are hda dma_intr problems before (69 seconds later) the nvidia kernel module is loaded.

$ sudo egrep 'Initializing CPU|kernel: Linux version|loading NVIDIA|Putting AGP|hd[ab].*(error|dma_intr|fail)' /var/log/kern.log

PUT IN FX6200 HERE
Dec  9 05:15:49 taryn kernel: Linux version 2.6.17-2-686 (Debian 2.6.17-9) (waldi@debian.org) (gcc version 4.1.2 20060901 (prerelease) (Debian 4.1.1-13)) #1 SMP Wed Sep 13 16:34:10 UTC 2006
Dec  9 05:15:49 taryn kernel: Initializing CPU#0
Dec  9 05:15:49 taryn kernel: hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
Dec  9 05:15:49 taryn kernel: hda: dma_intr: error=0x10 { SectorIdNotFound }, LBAsect=15371698469964, high=916224, low=10517580, sector=10517580
Dec  9 05:15:49 taryn kernel: hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
Dec  9 05:15:49 taryn kernel: hda: dma_intr: error=0x10 { SectorIdNotFound }, LBAsect=175971583182620, high=10488723, low=11847452, sector=11847452
Dec  9 05:15:58 taryn kernel: NVRM: loading NVIDIA Linux x86 Kernel Module  1.0-9631  Thu Nov  9 17:38:10 PST 2006

For the full output of that grep, see the attached file.

Ok, I need a plan now... what to do next?

Alaric

[netllama]

alaric,
The problem that you're having is not the same as what is being reported in this thread. This thread is specific to PCI-Express GeForce 6200TC cards, not AGP cards (which are not TC). I'd suggest creating a new thread for your issue.

thanks,
Lonni

[lithiumfx]

Originally Posted by sniart

I don't know if a petition will count for something, but let's do it.
lithiumfx, go forward and tells us where we sign.

How about here? I have no idea how to start a petition, at least not against a corporation such as Nvidia.

Originally Posted by alaric

This machine is rock solid stable with a FX5200. The instability started immediately after I put in a brand new FX6200 (AGP8x card). I have a motherboard with AGP4x by the way.

I investigated my logs and found that harddisk errors had started IMMEDIATELY after inserting the FX6200 -- BEFORE starting X OR LOADING THE NVIDIA DRIVERS!

Alaric,

As Lonni says, this isn't the problem we're discussing in this thread. The hard disk errors might be a sign of a failing hard disk - my FreeBSD box displayed similar errors shortly before disk death.

Originally Posted by netllama

alaric,
The problem that you're having is not the same as what is being reported in this thread. This thread is specific to PCI-Express GeForce 6200TC cards, not AGP cards (which are not TC). I'd suggest creating a new thread for your issue.

thanks,
Lonni

How about replying to the post about this not being a hardware issue, Lonni. Ignoring the complaints of your customers won't make the problem go away.

[netllama]

lithiumfx,
I'm not sure what kind of reply you're seeking. I commented on the progress of this issue 2 days ago.

Thanks,
-Lonni

[lithiumfx]

Originally Posted by netllama

lithiumfx,
I'm not sure what kind of reply you're seeking. I commented on the progress of this issue 2 days ago.

Thanks,
-Lonni

I'm seeking a reply that tells us, with some sort of official statement, that you're not shelving the bug but are actively seeking to fix it and that it will be fixed in an upcoming release (i.e. the next release) of the driver.

[netllama]

This bug is currently under active investigation by development, however, it will not be resolved in the next driver release.

Thanks,
Lonni

[lithiumfx]

Originally Posted by netllama

This bug is currently under active investigation by development, however, it will not be resolved in the next driver release.

Thanks,
Lonni

When will it be resolved? Do you have a time frame set out for this issue? Can you assure me that this issue is being investigated seriously, and that you're not simply casting it aside?