Bug 23373

Summary: X goes to 100% cpu on start with GeForce 9400 (0x086a, C79)
Product: xorg Reporter: Anssi Hannula <anssi>
Component: Driver/nouveauAssignee: Nouveau Project <nouveau>
Status: RESOLVED FIXED QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium    
Version: git   
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
GF9400: dmesg
none
GF9400: X.org log none

Description Anssi Hannula 2009-08-17 11:43:27 UTC
Created attachment 28710 [details]
GF9400: dmesg

When starting X.org server on GeForce 9400 (0x086a, C79, x86_64, DVI monitor with 1360x768 resolution), the screen is completely white (except for mouse cursor) and X starts running on 100% cpu (this is SMP system, so it doesn't render the machine unusable). X seems to be locked up (e.g. numlock doesn't work), but mouse cursor continues moving.

This happens both with and without KMS. DRM from 20090816 and DDX from 20090815. KMS console works correctly (though I do see "GPU lockup - switching to software fbcon").

Attached dmesg and X.org log, with KMS enabled.
Comment 1 Anssi Hannula 2009-08-17 11:44:19 UTC
Created attachment 28711 [details]
GF9400: X.org log
Comment 2 Ben Skeggs 2009-08-17 15:20:05 UTC
Could I get you to grab a mmio-trace of the NVIDIA binary driver initialising your card please?  There's detailed instructions at http://nouveau.freedesktop.org/wiki/MmioTrace

Thanks!
Comment 3 Anssi Hannula 2009-08-20 14:22:16 UTC
It took a bit longer than I had hoped, but I finally got the mmiotrace (yes, I didn't forget how to make those in just one week ;).
I sent it to the mmio dumps address, and also put it here:
http://stuff.onse.fi/nouveau/10de-086a/

It seems there was a kernel WARNING "kmmio_page_table not empty at cleanup, any further tracing will leak memory", but as that happened after stopping trace, I guess it didn't affect the trace itself.
Comment 4 Pekka Paalanen 2009-08-20 22:22:34 UTC
(In reply to comment #3)
> It seems there was a kernel WARNING "kmmio_page_table not empty at cleanup, any
> further tracing will leak memory", but as that happened after stopping trace, I
> guess it didn't affect the trace itself.

Oh shoot. Maybe someone broke it. I haven't checked if it works for a long time.

Yeah, that one does not affect the trace, but I wonder if something else is broken, too. Gah. At least the size of your trace is decent, and people will probably tell me, if there's something really wrong in the tool.
Comment 5 Anssi Hannula 2009-12-23 13:30:49 UTC
Seems to be working with current git.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.