Bug 55948 - [NVD9 optimus] drm/nouveau/bios: attempt to fetch entire acpi rom image in one shot breaks module load
Summary: [NVD9 optimus] drm/nouveau/bios: attempt to fetch entire acpi rom image in on...
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/nouveau (show other bugs)
Version: git
Hardware: x86-64 (AMD64) Linux (All)
: medium major
Assignee: Ben Skeggs
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2012-10-13 10:50 UTC by Jan Vesely
Modified: 2013-08-25 07:24 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:

Attachments
kernel log of failed modprobe (6.05 KB, text/plain)
2012-10-13 10:50 UTC, Jan Vesely 		no flags Details
successful module load (23.57 KB, text/plain)
2012-10-13 10:52 UTC, Jan Vesely 		no flags Details
DSDT acpi table (33.99 KB, text/plain)
2012-10-16 07:20 UTC, Jan Vesely 		no flags Details
acpidump (213.39 KB, text/plain)
2012-10-16 10:23 UTC, Jan Vesely 		no flags Details
fallback to slow _ROM access if checksum fails (720 bytes, patch)
2012-10-17 01:39 UTC, Ben Skeggs 		no flags Details | Splinter Review
modprobe nouveau debug=debug (4.56 KB, text/plain)
2012-10-17 08:34 UTC, Jan Vesely 		no flags Details
View All

Description Jan Vesely 2012-10-13 10:50:50 UTC 
Created attachment 68517 [details]
kernel log of failed modprobe

modprobe nouveau hangs with repeated message:
nouveau W[   VBIOS][0000:01:00.0] DCB *ver 0x87 unknown (see attached log)

the card is: NVIDIA Corporation GF119 [Quadro NVS 4200M]
on dell Latitude E6420.
Comment 1 Jan Vesely 2012-10-13 10:52:04 UTC 
Created attachment 68518 [details]
successful module load
Comment 2 Ben Skeggs 2012-10-16 00:12:30 UTC 
I don't suppose you can copy /sys/firmware/acpi/tables/DSDT somewhere (copy might be necessary to avoid getting a 0-byte file) and attach it for me?
Comment 3 Jan Vesely 2012-10-16 07:20:04 UTC 
Created attachment 68601 [details]
DSDT acpi table
Comment 4 Peter Wu 2012-10-16 09:23:36 UTC 
The DSDT does not contain a _ROM method. It must be in one of the SSDT files, can you run:

    acpidump > acpidump.txt
and attach the acpidump.txt file? You need to install acpidump first.
Comment 5 Jan Vesely 2012-10-16 10:23:46 UTC 
Created attachment 68619 [details]
acpidump
Comment 6 Ben Skeggs 2012-10-17 01:39:43 UTC 
Created attachment 68662 [details] [review]
fallback to slow _ROM access if checksum fails

How do you go if you apply this?
Comment 7 Jan Vesely 2012-10-17 08:34:37 UTC 
Created attachment 68676 [details]
modprobe nouveau debug=debug

The patch works. modprobe is OK. Haven't tested any functionality as rc1 hangs my SNB graphics.

I noticed these erros while unloading:
Oct 17 10:08:13 [kernel] nouveau E[   PFIFO][0000:01:00.0] write fault at 0x0000000000 [PAGE_NOT_PRESENT] from PCOPY0/PCOPY0 on channel 0x001fe1b000
Oct 17 10:08:15 [kernel] nouveau E[   PFIFO][0000:01:00.0] channel 0 kick timeout

They don't appear when I use config=PGRAPH=1.
Comment 8 Ilia Mirkin 2013-08-21 00:07:59 UTC 
Can you retest with a more recent kernel? I believe NVD9 optimus works relatively well now.
Comment 9 Jan Vesely 2013-08-22 07:38:46 UTC 
(In reply to comment #8)
> Can you retest with a more recent kernel? I believe NVD9 optimus works
> relatively well now.

The patch fixed this issue for me.
Everything has been working OK after the unrelated issue regarding intel graphics (mentioned in comment 7) has been fixed.
Comment 10 Martin Peres 2013-08-25 07:24:12 UTC 
(In reply to comment #9)
> (In reply to comment #8)
> > Can you retest with a more recent kernel? I believe NVD9 optimus works
> > relatively well now.
> 
> The patch fixed this issue for me.
> Everything has been working OK after the unrelated issue regarding intel
> graphics (mentioned in comment 7) has been fixed.

Perfect, thanks. I'm closing the bug as the patch has been committed.

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.