NV86 (Quadro NVS 140M) errors in dmesg

NV86 (Quadro NVS 140M) errors in dmesg

[Jasper Bryant-Greene]

In case it's any use for debugging, I get these error messages in dmesg
on my NV86 (Quadro NVS 140M) card:

[   21.729435] [drm] Initialized drm 1.1.0 20060810
[   21.749414] nouveau 0000:01:00.0: power state changed by ACPI to D0
[   21.749425] nouveau 0000:01:00.0: PCI INT A -> GSI 16 (level, low) ->
IRQ 16
[   21.749432] nouveau 0000:01:00.0: setting latency timer to 64
[   21.749528] [drm] Detected an NV50 generation card (0x086900a2)
[   21.749530] [drm] Initialized nouveau 0.0.11
drm-2.3.0-1351-g4585787bd1a1d782b on minor 0
[   22.586502] [drm:nouveau_graph_trapped_channel] *ERROR* AIII,
invalid/inactive channel id 128
[   22.586509] [drm] PGRAPH_ERROR - nSource: PROTECTION_ERROR, nStatus:
[   22.586513] [drm] PGRAPH_ERROR - Ch -1/0 Class 0x0000 Mthd 0x0000
Data 0x00000000:0x00000000
[   22.586521] [drm] Allocating FIFO number 1
[   22.593281] [drm] nouveau_fifo_alloc: initialised FIFO 1
[   22.593918] [drm] Allocating FIFO number 2
[   22.600736] [drm] nouveau_fifo_alloc: initialised FIFO 2

I'm happy to do any further debugging if the above are real errors.

Also, Xvideo does not work properly, with mplayer printing a large
number of these:

X11 error: BadMatch (invalid parameter attributes)

... and the screen remaining at a corrupted view of the desktop while
audio plays.

Otherwise, 2D performance is great. Haven't attempted any 3D as I don't
really have any use for 3D rendering.

-jasper

Re: NV86 (Quadro NVS 140M) errors in dmesg

[Ben Skeggs]

Am Samstag, den 02.08.2008, 10:20 +1200 schrieb Jasper Bryant-Greene:
> In case it's any use for debugging, I get these error messages in dmesg
> on my NV86 (Quadro NVS 140M) card:
> 
> [   21.729435] [drm] Initialized drm 1.1.0 20060810
> [   21.749414] nouveau 0000:01:00.0: power state changed by ACPI to D0
> [   21.749425] nouveau 0000:01:00.0: PCI INT A -> GSI 16 (level, low) ->
> IRQ 16
> [   21.749432] nouveau 0000:01:00.0: setting latency timer to 64
> [   21.749528] [drm] Detected an NV50 generation card (0x086900a2)
> [   21.749530] [drm] Initialized nouveau 0.0.11
> drm-2.3.0-1351-g4585787bd1a1d782b on minor 0
All normal

> [   22.586502] [drm:nouveau_graph_trapped_channel] *ERROR* AIII,
> invalid/inactive channel id 128
> [   22.586509] [drm] PGRAPH_ERROR - nSource: PROTECTION_ERROR, nStatus:
> [   22.586513] [drm] PGRAPH_ERROR - Ch -1/0 Class 0x0000 Mthd 0x0000
> Data 0x00000000:0x00000000
That's a bug - I see it on my FX570M, it'll only happen on first X
start.  Will look into it when it annoys me enough.

> [   22.586521] [drm] Allocating FIFO number 1
> [   22.593281] [drm] nouveau_fifo_alloc: initialised FIFO 1
> [   22.593918] [drm] Allocating FIFO number 2
> [   22.600736] [drm] nouveau_fifo_alloc: initialised FIFO 2
Normal.

> 
> I'm happy to do any further debugging if the above are real errors.
> 
> Also, Xvideo does not work properly, with mplayer printing a large
> number of these:
> 
> X11 error: BadMatch (invalid parameter attributes)
Use a composite manager (not compiz, xcompmgr or something else that
uses XRENDER for compositing) if you want to use Xv on G80 for the
moment.

> 
> ... and the screen remaining at a corrupted view of the desktop while
> audio plays.
> 
> Otherwise, 2D performance is great. Haven't attempted any 3D as I don't
> really have any use for 3D rendering.
3D wont work anyway :)

Cheers,
Ben.
> 
> -jasper
> 
> _______________________________________________
> Nouveau mailing list
> Nouveau-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
> http://lists.freedesktop.org/mailman/listinfo/nouveau

Re: NV86 (Quadro NVS 140M) errors in dmesg

[Jasper Bryant-Greene]

On Sat, 2008-08-02 at 10:02 +1000, Ben Skeggs wrote:
> Am Samstag, den 02.08.2008, 10:20 +1200 schrieb Jasper Bryant-Greene:
> > [   22.586502] [drm:nouveau_graph_trapped_channel] *ERROR* AIII,
> > invalid/inactive channel id 128
> > [   22.586509] [drm] PGRAPH_ERROR - nSource: PROTECTION_ERROR, nStatus:
> > [   22.586513] [drm] PGRAPH_ERROR - Ch -1/0 Class 0x0000 Mthd 0x0000
> > Data 0x00000000:0x00000000
> That's a bug - I see it on my FX570M, it'll only happen on first X
> start.  Will look into it when it annoys me enough.

OK, let me know if you need any details from me when you get around to
looking at this. It doesn't seem to actually cause any problems.

Also, is there any way to force the DPI with nouveau? I'm specifying

 -dpi 96

... on the X command line, but I still see

 (**) NOUVEAU(0): DPI set to (129, 203)

... in the Xorg.0.log, and the font sizes are all weird in X. With the
nvidia binary driver I used to have to use

 Option DPI "96 x 96"

... in xorg.conf - is there something similar with nouveau?

I'm guessing its my laptop panel reporting incorrect values for
something, but I still need to work around it :)

Cheers,
Jasper

Re: NV86 (Quadro NVS 140M) errors in dmesg

[Stephane Marchesin]

On 8/2/08, Jasper Bryant-Greene <jasper-XrxwZhJB1KNIJsTUW+i+0g@public.gmane.org> wrote:
> On Sat, 2008-08-02 at 10:02 +1000, Ben Skeggs wrote:
>  > Am Samstag, den 02.08.2008, 10:20 +1200 schrieb Jasper Bryant-Greene:
>
> > > [   22.586502] [drm:nouveau_graph_trapped_channel] *ERROR* AIII,
>  > > invalid/inactive channel id 128
>  > > [   22.586509] [drm] PGRAPH_ERROR - nSource: PROTECTION_ERROR, nStatus:
>  > > [   22.586513] [drm] PGRAPH_ERROR - Ch -1/0 Class 0x0000 Mthd 0x0000
>  > > Data 0x00000000:0x00000000
>  > That's a bug - I see it on my FX570M, it'll only happen on first X
>  > start.  Will look into it when it annoys me enough.
>
>
> OK, let me know if you need any details from me when you get around to
>  looking at this. It doesn't seem to actually cause any problems.
>
>  Also, is there any way to force the DPI with nouveau? I'm specifying
>
>   -dpi 96
>
>  ... on the X command line, but I still see
>
>   (**) NOUVEAU(0): DPI set to (129, 203)
>

Your panel size should be reported in the X.Org log, check that it is correct.

>  ... in the Xorg.0.log, and the font sizes are all weird in X. With the
>  nvidia binary driver I used to have to use
>
>   Option DPI "96 x 96"
>
>  ... in xorg.conf - is there something similar with nouveau?
>
>  I'm guessing its my laptop panel reporting incorrect values for
>  something, but I still need to work around it :)
>

You can use the xrandr utility to force the panel size/dpi value.

Stephane

Re: NV86 (Quadro NVS 140M) errors in dmesg

[Jasper Bryant-Greene]

On Sat, 2008-08-02 at 04:10 +0200, Stephane Marchesin wrote:
> On 8/2/08, Jasper Bryant-Greene <jasper-XrxwZhJB1KNIJsTUW+i+0g@public.gmane.org> wrote:
> >  Also, is there any way to force the DPI with nouveau? [snip]

> Your panel size should be reported in the X.Org log, check that it is correct.

Yes, it is reported correctly in the Xorg log. The nvidia binary driver
also reports it correctly, and has the same bug with DPI unless I
specify 'Option DPI "96 x 96"' under Device in xorg.conf.

(**) NOUVEAU(0): Display dimensions: (330, 210) mm

> >  I'm guessing its my laptop panel reporting incorrect values for
> >  something, but I still need to work around it :)

> You can use the xrandr utility to force the panel size/dpi value.

I will try this, though I would prefer to be able to set it in the Xorg
configuration so that X doesn't start with the wrong dpi and then switch
to the correct one...

-jasper

Re: NV86 (Quadro NVS 140M) errors in dmesg

[Stephane Marchesin]

On 8/2/08, Jasper Bryant-Greene <jasper-XrxwZhJB1KNIJsTUW+i+0g@public.gmane.org> wrote:
> On Sat, 2008-08-02 at 04:10 +0200, Stephane Marchesin wrote:
>  > On 8/2/08, Jasper Bryant-Greene <jasper-XrxwZhJB1KNIJsTUW+i+0g@public.gmane.org> wrote:
>
> > >  Also, is there any way to force the DPI with nouveau? [snip]
>
>
>  > Your panel size should be reported in the X.Org log, check that it is correct.
>
>
> Yes, it is reported correctly in the Xorg log. The nvidia binary driver
>  also reports it correctly, and has the same bug with DPI unless I
>  specify 'Option DPI "96 x 96"' under Device in xorg.conf.
>
>  (**) NOUVEAU(0): Display dimensions: (330, 210) mm
>

Interesting, it seems the DPI is computed for a 1680x1680 video mode.
The size is the one used in case you use desktop rotation, but it
shouldn't be used for DPI computation.

Stephane

Re: NV86 (Quadro NVS 140M) errors in dmesg

[Jasper Bryant-Greene]

On Sat, 2008-08-02 at 10:02 +1000, Ben Skeggs wrote:
> Am Samstag, den 02.08.2008, 10:20 +1200 schrieb Jasper Bryant-Greene:
> > Also, Xvideo does not work properly, with mplayer printing a large
> > number of these:
> > 
> > X11 error: BadMatch (invalid parameter attributes)
> Use a composite manager (not compiz, xcompmgr or something else that
> uses XRENDER for compositing) if you want to use Xv on G80 for the
> moment.

Xvideo plays in slow motion (very entertaining!)

mplayer -vo x11 plays fine, so it is definitely something to do with Xv.

I'm using xfce4's compositing manager.

What can I do to debug this?

Cheers,
-jasper

Re: NV86 (Quadro NVS 140M) errors in dmesg

[Ben Skeggs]

Am Samstag, den 02.08.2008, 21:54 +1200 schrieb Jasper Bryant-Greene:
> On Sat, 2008-08-02 at 10:02 +1000, Ben Skeggs wrote:
> > Am Samstag, den 02.08.2008, 10:20 +1200 schrieb Jasper Bryant-Greene:
> > > Also, Xvideo does not work properly, with mplayer printing a large
> > > number of these:
> > > 
> > > X11 error: BadMatch (invalid parameter attributes)
> > Use a composite manager (not compiz, xcompmgr or something else that
> > uses XRENDER for compositing) if you want to use Xv on G80 for the
> > moment.
> 
> Xvideo plays in slow motion (very entertaining!)
Really?! That's rather odd, haven't seen this reported previously :)  Is
your kernel log filled with many many many many PROTECTION_ERROR
messages by any chance?  If so, restart X and try again.  I never
actually tried Xv while that issue was present on my laptop.

> 
> mplayer -vo x11 plays fine, so it is definitely something to do with Xv.
> 
> I'm using xfce4's compositing manager.
> 
> What can I do to debug this?
Check what I mentioned above, if that's the case then it'd probably be
because the card's gone insane.  Otherwise, give xcompmgr a try to see
if it's XFCE4's fault.

Ben.
> 
> Cheers,
> -jasper
>