Nouveau - May 2015 - Handling GeForce GTX 850M GPU on Arch Linux

Hello all,

I've juste installed a new computer and I feel like I'm having trouble
with my NVIDIA GPU but it could be something else so I've enclosed
detailed output. As you can see in [lspci] output, my GPU is `NVIDIA
Corporation GM107M [GeForce GTX 850M] (rev a2)` whose specification can be
found in [specs].

The trouble I'd like to solve has the following symptoms:
 --> Sometimes during booting process, it issues a lot of lines like
`nouveau E[   PFIFO][0000:01:00.0] SCHED_ERROR [ UNK06 ]` (see [dmesg]
output for full details). This seems to appear randomly: sometimes the
computer has to reboot automatically two or three times before being able
to boot correctly, sometimes it just needs once and doesn't have nay
troubles.
 --> Some other times, windows appear to stop responding with no
reasons. I have to move the pointer off so the window display will
unfreeze and then I can put the pointer over again. I've found it seems to
be related to lines like this one `nouveau E[   PIBUS][0000:01:00.0] HUB0:
0x000001 0x00000002 (0xbad00100)` in [dmesg]. It starts at timestamp
28.115820.
 --> Sometimes, even my window manager (Gnome 3) freezes and I have to
go to a tty to have a chance to unfreeze it. It doesn't always work
then sometimes I have to reboot my computer because it has become
absolutely frozen (no mouse, no keyboard, even the keyboard CAPS light
responds no more).

I'm quite puzzled by this and I don't know whether the last symptom is a
bug or a feature (maybe multiple user simultaneous outputs). I've also
enclosed for your information my [Xorg.0.log] file and [lsmod] output. One
suggested me to share that with nouveau mailing list: maybe you don't have
any GeForce GTX 850M to test. I would agree to take time and perform test
for you if it can help to improve the way driver nouveau handles recent
GPU :)

Just feel free to ask any further data you'd need about my configuration.

Thank you in advance for you reply and have a good week!

Caocoa

--- references
[lspci] https://pastee.org/mtwkr
[specs] http://www.geforce.com/hardware/notebook-gpus/geforce-gtx-850m
[dmesg] https://pastee.org/zd4tt
[lsmod] https://pastee.org/t5z4e
[Xorg.0.log] https://pastee.org/r5tak

Your errors are most likely due to:

[    2.421428] nouveau  [     PFB][0000:01:00.0] RAM size: 1975398418 MiB

I'm guessing you don't *actually* have 1.9PB of VRAM. At least one
other person with a GM108 was seeing a similar issue. You're getting a
bunch of other failed reads, looks like we're somehow not bringing the
card up properly. The most immediate "fix" is to just disable nouveau
entirely -- boot with nouveau.modeset=0.

Cheers,

  -ilia


On Sun, May 17, 2015 at 2:31 PM,  <caocoa at mail2tor.com>
wrote:
> Hello all,
>
> I've juste installed a new computer and I feel like I'm having
trouble
> with my NVIDIA GPU but it could be something else so I've enclosed
> detailed output. As you can see in [lspci] output, my GPU is `NVIDIA
> Corporation GM107M [GeForce GTX 850M] (rev a2)` whose specification can be
> found in [specs].
>
> The trouble I'd like to solve has the following symptoms:
>  --> Sometimes during booting process, it issues a lot of lines like
> `nouveau E[   PFIFO][0000:01:00.0] SCHED_ERROR [ UNK06 ]` (see [dmesg]
> output for full details). This seems to appear randomly: sometimes the
> computer has to reboot automatically two or three times before being able
> to boot correctly, sometimes it just needs once and doesn't have nay
> troubles.
>  --> Some other times, windows appear to stop responding with no
> reasons. I have to move the pointer off so the window display will
> unfreeze and then I can put the pointer over again. I've found it seems
to
> be related to lines like this one `nouveau E[   PIBUS][0000:01:00.0] HUB0:
> 0x000001 0x00000002 (0xbad00100)` in [dmesg]. It starts at timestamp
> 28.115820.
>  --> Sometimes, even my window manager (Gnome 3) freezes and I have to
> go to a tty to have a chance to unfreeze it. It doesn't always work
> then sometimes I have to reboot my computer because it has become
> absolutely frozen (no mouse, no keyboard, even the keyboard CAPS light
> responds no more).
>
> I'm quite puzzled by this and I don't know whether the last symptom
is a
> bug or a feature (maybe multiple user simultaneous outputs). I've also
> enclosed for your information my [Xorg.0.log] file and [lsmod] output. One
> suggested me to share that with nouveau mailing list: maybe you don't
have
> any GeForce GTX 850M to test. I would agree to take time and perform test
> for you if it can help to improve the way driver nouveau handles recent
> GPU :)
>
> Just feel free to ask any further data you'd need about my
configuration.
>
> Thank you in advance for you reply and have a good week!
>
> Caocoa
>
> --- references
> [lspci] https://pastee.org/mtwkr
> [specs] http://www.geforce.com/hardware/notebook-gpus/geforce-gtx-850m
> [dmesg] https://pastee.org/zd4tt
> [lsmod] https://pastee.org/t5z4e
> [Xorg.0.log] https://pastee.org/r5tak
>
>
>
> _______________________________________________
> Nouveau mailing list
> Nouveau at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/nouveau

On 18.05.2015 01:46, Ilia Mirkin wrote:
> Your errors are most likely due to:
> 
> [    2.421428] nouveau  [     PFB][0000:01:00.0] RAM size: 1975398418 MiB
> 
> I'm guessing you don't *actually* have 1.9PB of VRAM. At least one
> other person with a GM108 was seeing a similar issue. You're getting a
> bunch of other failed reads, looks like we're somehow not bringing the
> card up properly. The most immediate "fix" is to just disable
nouveau
> entirely -- boot with nouveau.modeset=0.
> 
> Cheers,
> 
>   -ilia
> 

As the IGPs "tend" to "steal" system memory:

[   21.733864] nouveau  [     PFB][0000:01:00.0] RAM type: stolen system memory
[   21.733883] nouveau  [     PFB][0000:01:00.0] RAM size: 256 MiB
[   21.787222] nouveau  [     DRM] VRAM: 256 MiB

can we reverse the process in this case, and "steal" a few GB of VRAM
for a system memory?
I mean there are more than enough.