| Summary: | GM206: MMIO write of 800000ec FAULT at 10eb14 [ IBUS ] | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | xorg | Reporter: | Johnny B. Goode <adamos> | ||||||||
| Component: | Driver/nouveau | Assignee: | Nouveau Project <nouveau> | ||||||||
| Status: | RESOLVED MOVED | QA Contact: | Xorg Project Team <xorg-team> | ||||||||
| Severity: | normal | ||||||||||
| Priority: | medium | CC: | adamos | ||||||||
| Version: | unspecified | ||||||||||
| Hardware: | x86-64 (AMD64) | ||||||||||
| OS: | Linux (All) | ||||||||||
| Whiteboard: | |||||||||||
| i915 platform: | i915 features: | ||||||||||
| Attachments: |
|
||||||||||
|
Description
Johnny B. Goode
2018-12-08 16:07:26 UTC
Interesting. The only write I see to that register is in the GM107 fan control logic. This is not used for GM200+, as controlling the fan can't be done anymore, except via secure firmware means. So ... my *guess* is that this is a write failure that happens as a result of the firmware that gets loaded, which is not supplied by nouveau. (Always preferable to blame someone else...) Can you get a dmesg with nouveau.debug=trace which should place that error more exactly in the initialization flow. Created attachment 142762 [details]
GM206 nouveau.debug trace
journalctl -b -1 --no-hostname -o short-monotonic | grep nouveau
Created attachment 142763 [details]
GM206 nouveau.debug trace2
Once again, because the first one was without -b -1 parameter, so without closing nouveau.
journalctl -b -1 --no-hostname -o short-monotonic | grep nouveau
(In reply to Johnny B. Goode from comment #3) > Created attachment 142763 [details] > GM206 nouveau.debug trace2 > > Once again, because the first one was without -b -1 parameter, so without > closing nouveau. > journalctl -b -1 --no-hostname -o short-monotonic | grep nouveau I could be mistaken, but there's no MMIO errors in there... which means it's intermittent, perhaps based on boot state or something else. I do think it's the secure firmware doing this. Created attachment 142870 [details]
Rom
Rom bios because Asus made few cards GTX 950 with different bioses hxxps://www.techpowerup.com/vgabios/?architecture=NVIDIA&manufacturer=Asus&model=GTX+950&interface=&memType=&memSize=&since=
Mine is 84.06.3d.00.6f
(In reply to Johnny B. Goode from comment #5) > Created attachment 142870 [details] > Rom > > Rom bios because Asus made few cards GTX 950 with different bioses > hxxps://www.techpowerup.com/vgabios/ > ?architecture=NVIDIA&manufacturer=Asus&model=GTX+950&interface=&memType=&memS > ize=&since= > Mine is 84.06.3d.00.6f I was referring to the firmware that's supplied by NVIDIA in linux-firmware, which has to be loaded to operate any of the acceleration components of the board. -- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/xorg/driver/xf86-video-nouveau/issues/472. |
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.