[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#758460: xserver-xorg-video-nouveau: Crash and freeze with GeForce 6150SE.

2015-05-05 19:38 GMT+03:00 Agustin Martin <agmartin@debian.org>:
> On Mon, May 04, 2015 at 12:49:52PM +0200, Agustin Martin wrote:
>> Hi,
>> Just to confirm a similar problem in a GeForce 6150SE graphics engine,
>> 00:0d.0 VGA compatible controller: NVIDIA Corporation C61 [GeForce 6150SE nForce 430] (rev a2)
>> Similar messages in syslog,
>> [  711.849697] nouveau E[    PBUS][0000:00:0d.0] MMIO write of 0x00870001 FAULT at 0x00b010
>> and random freeze. Last seen thing is some sort of zig-zag random pattern
>> apparently with colors used at that time. Nothing but hard reboot helps
>> once the system is frozen, not even Magic SysRq. No relevant info seems
>> to appear in syslog.
>> An important info is that this does not happen if I reboot with old
>> 3.2.04-amd64 (3.2.65-1+deb7u2) wheezy kernel. So this seems to be a problem
>> with the nouveau kernel module.
>> Noticed that something similar has been reported in
>> http://bugs.debian.org/758460 for a GeForce 6150LE. Not merging both bug
>> reports in case they are different issues for different chip variants, but
>> cc'ing. Following advice there I tried with kernel boot parameter
>> "nouveau.config=NvMSI=0" in the command line and with
>> options nouveau NvMSI=0
>> in a /etc/modprobe.d/local-video.conf file (also built a new initramfs in
>> case this is needed), problem is still here, no luck.
> Hi,
> I was thinking about building a kernel with patch mentioned in
> http://bugs.debian.org/758460
>  http://cgit.freedesktop.org/~darktama/nouveau/commit/?id=b7387454bda64a16350fabee4d81bec3b8de8fa2
> applied, but I then noticed that it is already in jessie's kernel.
> Searching kernel changelogs, I found this in
> https://www.kernel.org/pub/linux/kernel/v3.x/ChangeLog-3.18.4
> commit 715f7c8214c7822833af757109022006c3f01d6a
> Author: Ilia Mirkin <imirkin@alum.mit.edu>
> Date:   Tue Dec 16 13:55:38 2014 -0500
>     drm/nv4c/mc: disable msi
>     commit 4761703bd04bbdf56396d264903cc5a1fdcb3c01 upstream.
>     Several users have, over time, reported issues with MSI on these IGPs.
>     They're old, rarely available, and MSI doesn't provide such huge
>     advantages on them. Just disable.
>     Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=87361
>     Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=74492
>     Fixes: fa8c9ac72fe ("drm/nv4c/mc: nv4x igp's have a different msi rearm
>     register")
>     Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
>     Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
>     Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> also in https://www.kernel.org/pub/linux/kernel/v3.0/ChangeLog-3.14.30 (LTS?)
> Now it is clear why setting NvMSI=0 did not make any difference in the box
> with the GeForce 6150SE where I hit the bug (Did not try with noaccel).
> Martin-Éric, has jessie's kernel fixed the problem in your box?


There's two NVDIA boxes involved (IIRC, I filed separate bugs reports).

Neither box works with the Nouveau X.org driver in Jessie. In both
cases, I've had to remove xserver-xorg-video-nouveau and let Xserver
use the generic modeline driver.


Reply to: