FreeBSD problem

[Casm]

Guys,

I have the following hard/soft

Athlon XP 2000+
nForce chipset
FX5900XT

FreeBSD 5.2.1
XFree 4.3.0

When I installed nvidia-driver, X worked. But.

Mouse cursor became jerking. top displayed 10-20% CPU was utilized by interrupt. When I run X without nvidia driver interrupt used about 0%. When I switched to text tty while X was running interrupt used 0%.

I tried different forms of AGP, disabled MTTR and ACPI - nothing has help.

What can be the problem?

Thanks in advance,

Casm

[hppnq]

Your problem probably has to do something with interrupt latency. It seems your kernel has had some major interrupt latency surgery, and that might just have bitten your Nvidia driver. Hopefully someone else comes up with a better idea of what's wrong, maybe googling along the lines of "interrupt latency freebsd nvidia" will point you to a solution.

(If you're really stuck, post the relevant parts of the kernel log, X log, X configuration and the output of lspci -vv.)

[Casm]

kernel configuration, dmesg, xf86config, xf86log and scanpci -vv are attached.
Also I have the following in the LOADER.CONF

nvidia_load="YES"
machdep.disable_mtrrs="1"

And the following in the DEVICE.HINTS

...
hint.acpi.0.disabled="1"
hint.apm.0.disabled="1"
...

Any sugesstions?

[hppnq]

You seem to have two devices using irq 11 (sound card + unknown card). That could be a problem.

Can you post the Nvidia sysctl information? (See the README that came with the driver for more on that.) Also, I think in FreeBSD irq handlers show up in ps -- can you see which ones are active when you're having latency problems? And, finally, could you also post a dmesg with another graphics driver loaded (not nvidia.ko)?

[Casm]

I have removed pcm device from kernel but situation has not changed. Unknown card is a video card. It uses irq alone now as ps shows.

Also, there is no other graphics driver loaded - I use standard nv from XFree.

sysctls are:

hw.nvidia.agp.card.rates: 4x 2x 1x
hw.nvidia.agp.card.fw: supported
hw.nvidia.agp.card.sba: supported
hw.nvidia.agp.card.registers: 0x1f000217:0x00000000
hw.nvidia.version: NVIDIA FreeBSD x86 nvidia.ko Kernel Module 1.0-4365 Wed May 28 09:20:25 PDT 2003
hw.nvidia.registry.EnableVia4x: 0
hw.nvidia.registry.EnableALiAGP: 0
hw.nvidia.registry.EnableAGPSBA: 1
hw.nvidia.registry.EnableAGPFW: 1
hw.nvidia.registry.SoftEDIDs: 1
hw.nvidia.registry.Mobile: 4294967295
hw.nvidia.registry.ResmanDebugLevel: 4294967295
hw.nvidia.registry.FlatPanelMode: 0
hw.nvidia.registry.UpdateKernelAGP: 1
hw.nvidia.cards.0.model: Unknown
hw.nvidia.cards.0.irq: 11
hw.nvidia.cards.0.vbios: ??.??.??.??.??
hw.nvidia.cards.0.type: AGP

Configs and logs with nv driver are attached.

Where to find the problem instead of nvidia-driver???

[Casm]

By the way, hppnq, are you an official nvidia representaive or a free helper?

[hppnq]

You are experiencing problems now then too, using the nv driver?! Have you upgraded your BIOS to the latest level? Can you still post the ps output of the interrupt handlers when your system is going berserk?

And no, I am no official representative of Nvidia. I hope that doesn't disappoint you. ;-)

[Casm]

I haven't got problems with nv anytime. The only problem with nv is that it does not support 3d acceleration My BIOS is the latest. WinXP works rather good if that OS may work good

I asked about you because I want to know if it has meaning to send my problem to the official nvidia e-mail.

I'll post ps result later but I see nothing special there. As I think the problem is in a driver itself. When I used FreeBSD 5.0, GF3 and the first nvidia driver everything was okay.

[hppnq]

Okay, I thought you meant your nv driver was turning on you too. ;-)

Configure X to use the nvidia driver of course, and post the interrupts ps list when you're system is unstable again. It wouldn't hurt to also post the dmesg and scanpci output, X log and X config again, along with the Nvidia installer log, if you have that. Yeah, I'm asking a lot. ;-)

By all means contact Nvidia about your problem. You've got nothing to lose! ;-)

[Casm]

I've attached ps -axww log.

I'm not sure why firstly attached logs (in "kernel configuration, dmesg, xf86config, xf86log and scanpci -vv are attached." message) are not enough. And also there is no problem with stability. Simply when I use nvidia-driver AND run X with it's selected AND I'm in X's terminal I have this interrupt cpu usage. When anything in the specified condition is changed (for example, when I run X with nvidia driver I switch to usual text tty via Ctrl-Alt-F2 and run top there) I see interrupts take 0 per cent of CPU.

I've written a letter to nvidia some hours ago and wait now for the answer. But I have something to lose: my hope that this damn computer would work

At least I thank you for your attention to my problem.

[Casm]

Hah! It's interesting:

This is an automatically generated Delivery Status Notification.
Delivery to the following recipients failed.
PF-FreeBSD-GFX-Bugs@nvidia.com


Message/delivery-status

Reporting-MTA: dns;hqempubf01.nvidia.com
Received-From-MTA: dns;hqemfe01.nvidia.com
Arrival-Date: Fri, 14 May 2004 10:49:40 -0700
Final-Recipient: rfc822;PF-FreeBSD-GFX-Bugs@nvidia.com
Action: failed
Status: 5.2.1
X-Display-Name: FreeBSD-GFX-Bugs

They don't receive bugs anymore?

[hppnq]

Sorry mate, I don't think I can help you any more.;-( Try Nvidia once more and search the BSD forums...