Bug 2016

Summary:	nouveau driver panics the kernel (also, fbcon trouble)
Product:	Mageia	Reporter:	Richard Neill <mageia>
Component:	RPM Packages	Assignee:	Anssi Hannula <anssi.hannula>
Status:	RESOLVED OLD	QA Contact:
Severity:	critical
Priority:	Normal	CC:	marja11, stormi-mageia, tmb
Version:	1	Keywords:	NEEDINFO
Target Milestone:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Source RPM:	x11-driver-video-nouveau	CVE:
Status comment:
Attachments:	dmesg.old

Description Richard Neill 2011-07-03 21:29:27 CEST

I just updated my system (64-bit, 8 GB RAM, Nvidia 7600 GT) from Mandriva to Mageia with urpmi (sccessfully), and then rebooted into the new kernel
 (2.6.38.7-desktop-1.mga)

On rebooting, the machine got slightly past grub, began to print kernel info, a few seconds later it turned the text illegible(*), but I could see the green colour from the "OK" messages, and after about 20 seconds, it hung. Then the machine rebooted again [because I have "panic=60" on the command line].

Using the older Mandriva kernel (2.6.33), I could reboot the machine, and investigate....

There are THREE problems here, which combined to make this most unfortunate to debug.

Problem #1. The fbcon driver means that all that wonderfully helpful boot text becomes mangled (it looks like the LCD has lost horizontal sync), and it's impossible to see what's happening. A partial fix is to add "fbcon=map:2" to the kernel cmdline.

Problem #2. At the point where X begins to try to start [by guesswork - it's about 20 seconds into the boot, though the text is illegible], the machine will kernel panic. The solution is to blacklist the nouveau module. 

Problem #3. The nvidia drivers, as packaged by Mageia don't work. The solution (see other bug report) is to use the binary package provided by nvidia.com

Comment 1 Samuel Verschelde 2011-10-01 11:19:03 CEST

Updating the topic to make it only focused on the nouveau driver issue.

Please open another bug report for the fbcon issue.

Assigning to Anssi, maintainer of the nouveau driver. tmb in CC in case it's a kernel bug.

CC: (none) => stormi, tmb
Assignee: bugsquad => anssi.hannula
Source RPM: kernel-desktop-2.6.38.7-1.mga => x11-driver-video-nouveau

Comment 2 Anssi Hannula 2011-12-16 05:37:34 CET

Please provide the kernel log output,
a) by running e.g. "dmesg > some_file.txt", or
b) by booting to the broken kernel, then to a working kernel, and then attaching /var/log/dmesg.old

Comment 3 Richard Neill 2011-12-28 03:10:43 CET

Created attachment 1295 [details]
dmesg.old

Here is dmesg.old, from the broken kernel, which is fortunately still present. That's all I have.

Sorry I can't be much more helpful, but as a result of a disk problem, my desktop machine is now doing double-duty as a temporary webserver(!), and I can't take it down for lots of experiments.

Comment 4 Marja Van Waes 2012-04-22 11:38:46 CEST

3 monthly ping


(In reply to comment #3)
> Created attachment 1295 [details]
> dmesg.old
> 
> Here is dmesg.old, from the broken kernel, which is fortunately still present.
> That's all I have.
> 
> Sorry I can't be much more helpful, but as a result of a disk problem, my
> desktop machine is now doing double-duty as a temporary webserver(!), and I
> can't take it down for lots of experiments.

Maybe you have a seperate webserver now?

CC: (none) => marja11

Marja Van Waes 2012-04-22 11:39:12 CEST

Keywords: (none) => NEEDINFO

Comment 5 Richard Neill 2012-04-22 16:06:05 CEST

I think we should now abandon this bug...it's now sufficiently old that the kernel and nouveau drivers are both substantially advanced from where they were. 

(also, the machine concerned suffered a motherboard failure, and has been replaced by one with onboard intel graphics - so I will not be able to replicate any more)

Comment 6 Marja Van Waes 2012-04-22 16:43:48 CEST

(In reply to comment #5)
> I think we should now abandon this bug...it's now sufficiently old that the
> kernel and nouveau drivers are both substantially advanced from where they
> were. 
> 
> (also, the machine concerned suffered a motherboard failure, and has been
> replaced by one with onboard intel graphics - so I will not be able to
> replicate any more)

Thanks for the feedback, Richard.

Closing as OLD

Status: NEW => RESOLVED
Resolution: (none) => OLD