Bug 19934 - kernel bug in nouveau driver ("BUG: unable to handle kernel paging request")
Summary: kernel bug in nouveau driver ("BUG: unable to handle kernel paging request")
Status: RESOLVED FIXED
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: Cauldron
Hardware: All Linux
Priority: Normal critical
Target Milestone: ---
Assignee: Kernel and Drivers maintainers
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2016-12-12 21:28 CET by PC LX
Modified: 2017-01-14 14:53 CET (History)
2 users (show)

See Also:
Source RPM:
CVE:
Status comment:


Attachments
output of "journalctl -b" (441.06 KB, text/x-log)
2016-12-12 21:28 CET, PC LX
Details

Description PC LX 2016-12-12 21:28:28 CET
Created attachment 8762 [details]
output of "journalctl -b"

Description of problem:

I'm running a up-to-date cauldron with Plasma using the nouveau driver. I'm experiencing desktop freezes and when looking at the logs to see if there was any clue as to what the problem was, I saw the following kernel messages several (7) times (full log in attachment):

kernel: resource sanity check: requesting [mem 0xcff9c000-0xd009bfff], which spans more than 0000:01:00.0 [mem 0xce000000-0xcfffffff 64bit pref]
kernel: caller nv50_instobj_boot+0xac/0x120 [nouveau] mapping multiple BARs
kernel: BUG: unable to handle kernel paging request at ffffc2ba00810000


I don't know if these messages are related to the freezes or not but they are an indication of a problem in the nouveau kernel driver.

The desktop can be unfrozen by switching to another virtual console and back, except when I start a second X session, then the freeze is permanent and I can't even switch to another virtual console. The sysrq keys still worked.




Version-Release number of selected component (if applicable):

# rpm -qa | grep -E 'kernel|nouveau' | sort
kernel-desktop-4.8.14-1.mga6-1-1.mga6
kernel-desktop-devel-4.8.14-1.mga6-1-1.mga6
kernel-desktop-devel-latest-4.8.14-1.mga6
kernel-desktop-latest-4.8.14-1.mga6
kernel-firmware-20160914-1.mga6
kernel-firmware-nonfree-20160914-2.mga6.nonfree
kernel-userspace-headers-4.8.14-1.mga6
lib64drm_nouveau2-2.4.74-1.mga6
lib64vdpau-driver-nouveau-13.0.2-2.mga6.tainted
libdrm_nouveau2-2.4.74-1.mga6
libvdpau-driver-nouveau-13.0.2-2.mga6.tainted
task-plasma5-minimal-5.8.4-1.mga6
x11-driver-video-nouveau-1.0.13-2.mga6



How reproducible:

Don't know how to trigger the freezes that can be unfrozen by a virtual console switch. These have happened several times so far.
The two times I tried starting a second user session, the desktop freezes and only a sysrq or hard reset can unfreeze it.



Steps to Reproduce:
1. Start a desktop Plasma session, in Xorg using the nouveau driver;
2. Start a second user session;
3. The desktop should freeze several seconds in to the session.
Marja Van Waes 2016-12-13 00:48:31 CET

CC: (none) => marja11
Assignee: bugsquad => kernel

Comment 1 Thomas Backlund 2017-01-12 13:25:04 CET
Is this still an issue with 4.9.2 ?

CC: (none) => tmb

Comment 2 PC LX 2017-01-14 14:53:40 CET
I'm no longer seeing the kernel messages mentioned in the issue description. I'm also no longer seeing desktop freezes, except when starting a second X session. 

The freezes when starting a second session are likely independent on this issue so I may open another issue for it.

For now, since no one else complained about this issue, I'm closing this issue as resolved+fixed.

# uname -a
Linux marte 4.9.3-desktop-1.mga6 #1 SMP Thu Jan 12 23:02:31 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
# rpm -qa | grep -E 'kernel|nouveau' | sort
kernel-desktop-4.9.3-1.mga6-1-1.mga6
kernel-desktop-devel-4.9.3-1.mga6-1-1.mga6
kernel-desktop-devel-latest-4.9.3-1.mga6
kernel-desktop-latest-4.9.3-1.mga6
kernel-firmware-20170101-1.mga6
kernel-firmware-nonfree-20170101-1.mga6.nonfree
kernel-userspace-headers-4.9.3-1.mga6
lib64drm_nouveau2-2.4.74-1.mga6
libdrm_nouveau2-2.4.74-1.mga6
x11-driver-video-nouveau-1.0.13-2.mga6

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


Note You need to log in before you can comment on or make changes to this bug.