Nouveau - Feb 2010 - nouveau interrupts firewire ISO streaming

Hi!

After switching from UMS (dated 20090701+c0bf670) to KMS (as of
2010-02-01), almost any X activity causes buffer underruns on my
firewire soundcard.

When the X pointer changes from arrow to the vertical text selection
bar, firewire streaming is interrupted. Same with scrolling in firefox
or hovering over a link, thus changing the pointer to "hand".

Looks like nouveau is disabling interrupts now and then, which leads to
stalled firewire ISO streaming. I don't know what nouveau's actually
doing, I'm just guessing. ;) 

Firewire audio streaming (www.ffado.org) requires precisely timed DMA
transfers. If the kernel/controller/scheduler fails to deliver those
packets in time, e.g.  due to a ressource with high interrupt load, the
audio stream is interrupted and you get audible glitches.

I see this behaviour on 

06:00.0 VGA compatible controller: nVidia Corporation G71 [GeForce 7300
GS] (rev a1)

with Dual-Head (VGA+DVI).


I've also uploaded kern.log and Xorg log, though I guess they're not
really helpful to trace this problem:

   http://adi.loris.tv/nouveau-xorg.log

   http://adi.loris.tv/nouveau-kern.log

Note that kern.log also includes an Oops at Feb  1 20:22:43, but I'm not
really reporting this issue here. ;) (I rmmod'ed nouveau remotely)



Cheerio



PS: I (sometimes) have two mouse cursors. My screen layout is 

     xrandr --output DVI-I-1 --right-of VGA-1

and when moving the mouse from the left screen to the right, a static
(not moving) cursor stays on the left screen. It changes its appearance
simultaneously with the real (moving) cursor on the right screen, that
is, if the right pointer is a hand, the left pointer also becomes a
hand, a.s.o.

Moving the mouse back to the left screen re-activates movement of the
left cursor and no additional cursor is left on the right screen.

But I'm not reporting this issue here, the firewire stuff is more
important.


PPS: If you prefer Bugzilla, I can re-report there. I just started with
the mailinglist in case the ISO streaming requires some extended
discussion.

-- 
mail: adi at thur.de  	http://adi.thur.de	PGP/GPG: key via keyserver

On Tue, Feb 2, 2010 at 2:15 PM, Adrian Knoth <adi at
drcomp.erfurt.thur.de> wrote:
> Hi!
>
> After switching from UMS (dated 20090701+c0bf670) to KMS (as of
> 2010-02-01), almost any X activity causes buffer underruns on my
> firewire soundcard.
>
> When the X pointer changes from arrow to the vertical text selection
> bar, firewire streaming is interrupted. Same with scrolling in firefox
> or hovering over a link, thus changing the pointer to "hand".
>
> Looks like nouveau is disabling interrupts now and then, which leads to
> stalled firewire ISO streaming. I don't know what nouveau's
actually
> doing, I'm just guessing. ;)
I don't think interrupts are disabled for uploading a cursor, more
likely is that your firewire driver depends on the big kernel lock. At
the moment we still hold that. That said, i don't think nouveau can
make any kind of latency garuantees. We do intend to go to a design
that doesn't rely on the BKL at some point, but i wouldn't expect that
to happen this month.
>
> Firewire audio streaming (www.ffado.org) requires precisely timed DMA
> transfers. If the kernel/controller/scheduler fails to deliver those
> packets in time, e.g. ?due to a ressource with high interrupt load, the
> audio stream is interrupted and you get audible glitches.
>
> I see this behaviour on
>
> 06:00.0 VGA compatible controller: nVidia Corporation G71 [GeForce 7300
> GS] (rev a1)
>
> with Dual-Head (VGA+DVI).
>
>
> I've also uploaded kern.log and Xorg log, though I guess they're
not
> really helpful to trace this problem:
>
> ? http://adi.loris.tv/nouveau-xorg.log
>
> ? http://adi.loris.tv/nouveau-kern.log
>
> Note that kern.log also includes an Oops at Feb ?1 20:22:43, but I'm
not
> really reporting this issue here. ;) (I rmmod'ed nouveau remotely)
>
>
>
> Cheerio
>
>
>
> PS: I (sometimes) have two mouse cursors. My screen layout is
>
> ? ? xrandr --output DVI-I-1 --right-of VGA-1
>
> and when moving the mouse from the left screen to the right, a static
> (not moving) cursor stays on the left screen. It changes its appearance
> simultaneously with the real (moving) cursor on the right screen, that
> is, if the right pointer is a hand, the left pointer also becomes a
> hand, a.s.o.
>
> Moving the mouse back to the left screen re-activates movement of the
> left cursor and no additional cursor is left on the right screen.
>
> But I'm not reporting this issue here, the firewire stuff is more
> important.
>
>
> PPS: If you prefer Bugzilla, I can re-report there. I just started with
> the mailinglist in case the ISO streaming requires some extended
> discussion.
>
> --
> mail: adi at thur.de ? ? ? http://adi.thur.de ? ? ?PGP/GPG: key via
keyserver
> _______________________________________________
> Nouveau mailing list
> Nouveau at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/nouveau
>