[BUG] ALSA FireWire: poll timeout causes JACK to stop (6.10 – 6.17-rc2)

[BUG] ALSA FireWire: poll timeout causes JACK to stop (6.10 – 6.17-rc2)

[Илья Кузнецов]

------------------------------

Hello ALSA developers,

I would like to report a reproducible issue with *snd-fireface* on Linux
kernels 6.10 through 6.17-rc2.
------------------------------
System Information

   -

   *Kernel versions tested:* 6.10 → 6.17-rc2
   -

   *Audio interface:* RME Fireface 800
   -

   *Driver in use:* snd-fireface
   -

   *Audio server:* JACK (tested with jackd and QJackCtl)
   -

   *Distribution:* Debian 12
   -

   *Kernel config:* ALSA FireWire drivers enabled, no FFADO in use
   - *Chip 1394: *Texas Instrumets

------------------------------
Steps to Reproduce

   1.

   Boot kernel 6.10 … 6.17-rc2 with snd-fireface loaded.
   2.

   Start JACK with ALSA backend, for example:

   jackd -d alsa -d hw:Fireface -p 64 -n 3

   3. Use the system normally — during playback, recording, or even when
   idle. The issue may occur at any time without apparent external cause.

   Actual Behavior

   After some idle time, ALSA stops responding and JACK prints:

   ERROR: ALSA: poll time out, polled for ... usecs
   ERROR: JackAudioDriver::ProcessAsync: read error, stopping...

   JACK immediately terminates. The Fireface 800 does not recover until
   JACK is restarted.
   ------------------------------
   Expected Behavior

   The ALSA FireWire driver should maintain the stream and allow JACK to
   continue running without manual restart, even when idle.
   ------------------------------
   Notes
   4.

   With *FFADO* backend (jackd -d firewire), the same hardware works
   reliably and does not exhibit this issue.
   5.

   This suggests that the problem lies in the ALSA FireWire streaming
   engine rather than the hardware itself.
   6.

   The problem has been reproducible across all tested kernels (vanila, rt,
   liquorix) from *6.10 through 6.17-rc2*.

------------------------------
Request

Could you please confirm if this issue is known?
Is there any ongoing work on the ALSA FireWire streaming engine (
snd-fireface) to address these poll timeout conditions?

I can provide additional logs (dmesg, journalctl) or test patches if needed.

Thank you,

Ilya Kuznetsov.

[BUG] ALSA FireWire: poll timeout causes JACK to stop (6.10 – 6.17-rc2)

[Илья Кузнецов]

________________________________

Hello ALSA developers,

I would like to report a reproducible issue with snd-fireface on Linux
kernels 6.10 through 6.17-rc2.

________________________________

System Information

Kernel versions tested: 6.10 → 6.17-rc2

Audio interface: RME Fireface 800

Driver in use: snd-fireface

Audio server: JACK (tested with jackd and QJackCtl)

Distribution: Debian 12

Kernel config: ALSA FireWire drivers enabled, no FFADO in use

Chip 1394: Texas Instrumets

________________________________

Steps to Reproduce

Boot kernel 6.10 … 6.17-rc2 with snd-fireface loaded.

Start JACK with ALSA backend, for example:

jackd -d alsa -d hw:Fireface -p 64 -n 3

Use the system normally — during playback, recording, or even when
idle. The issue may occur at any time without apparent external cause.

Actual Behavior

After some idle time, ALSA stops responding and JACK prints:

ERROR: ALSA: poll time out, polled for ... usecs
ERROR: JackAudioDriver::ProcessAsync: read error, stopping...

JACK immediately terminates. The Fireface 800 does not recover until
JACK is restarted.

________________________________

Expected Behavior

The ALSA FireWire driver should maintain the stream and allow JACK to
continue running without manual restart, even when idle.

________________________________

Notes

With FFADO backend (jackd -d firewire), the same hardware works
reliably and does not exhibit this issue.

This suggests that the problem lies in the ALSA FireWire streaming
engine rather than the hardware itself.

The problem has been reproducible across all tested kernels (vanila,
rt, liquorix) from 6.10 through 6.17-rc2.

________________________________

Request

Could you please confirm if this issue is known?
Is there any ongoing work on the ALSA FireWire streaming engine
(snd-fireface) to address these poll timeout conditions?

I can provide additional logs (dmesg, journalctl) or test patches if needed.

Thank you,

Ilya Kuznetsov.

Re: [BUG] ALSA FireWire: poll timeout causes JACK to stop (6.10 – 6.17-rc2)

[Takashi Sakamoto]

Hi,

I'm sorry for your inconvenience.

If you have ever used your device with Pipewire, I would like to see
your experiences how it works, since both servers uses the different
ways to handle ALSA PCM device in respect to no-period-wakeup mode..

On Tue, Sep 02, 2025 at 03:11:52PM +0500, Илья Кузнецов wrote:
> ________________________________
> 
> Hello ALSA developers,
> 
> I would like to report a reproducible issue with snd-fireface on Linux
> kernels 6.10 through 6.17-rc2.
> 
> ________________________________
> 
> System Information
> 
> Kernel versions tested: 6.10 → 6.17-rc2
> 
> Audio interface: RME Fireface 800
> 
> Driver in use: snd-fireface
> 
> Audio server: JACK (tested with jackd and QJackCtl)
> 
> Distribution: Debian 12
> 
> Kernel config: ALSA FireWire drivers enabled, no FFADO in use
> 
> Chip 1394: Texas Instrumets
> 
> ________________________________
> 
> Steps to Reproduce
> 
> Boot kernel 6.10 … 6.17-rc2 with snd-fireface loaded.
> 
> Start JACK with ALSA backend, for example:
> 
> jackd -d alsa -d hw:Fireface -p 64 -n 3
> 
> Use the system normally — during playback, recording, or even when
> idle. The issue may occur at any time without apparent external cause.
> 
> Actual Behavior
> 
> After some idle time, ALSA stops responding and JACK prints:
> 
> ERROR: ALSA: poll time out, polled for ... usecs
> ERROR: JackAudioDriver::ProcessAsync: read error, stopping...
> 
> JACK immediately terminates. The Fireface 800 does not recover until
> JACK is restarted.
> 
> ________________________________
> 
> Expected Behavior
> 
> The ALSA FireWire driver should maintain the stream and allow JACK to
> continue running without manual restart, even when idle.
> 
> ________________________________
> 
> Notes
> 
> With FFADO backend (jackd -d firewire), the same hardware works
> reliably and does not exhibit this issue.
> 
> This suggests that the problem lies in the ALSA FireWire streaming
> engine rather than the hardware itself.
> 
> The problem has been reproducible across all tested kernels (vanila,
> rt, liquorix) from 6.10 through 6.17-rc2.
> 
> ________________________________
> 
> Request
> 
> Could you please confirm if this issue is known?
> Is there any ongoing work on the ALSA FireWire streaming engine
> (snd-fireface) to address these poll timeout conditions?
> 
> I can provide additional logs (dmesg, journalctl) or test patches if needed.
> 
> Thank you,
> 
> Ilya Kuznetsov.


Regards

Takashi Sakamoto

Re: [BUG] ALSA FireWire: poll timeout causes JACK to stop (6.10 – 6.17-rc2)

[Илья Кузнецов]

Hi Takashi,

Thank you for your reply. I tried PipeWire in the past, but my overall
experience was rather negative. With its ALSA drivers I could not achieve
the same low latency as with FFADO. When using PipeWire together with
FFADO, the results were even worse, with many clearly audible xruns.

Because of this, I stopped using PipeWire for ProAudio purposes. As far as
I know, there is still an unresolved issue with about 90 ms of additional
latency, which makes it unsuitable for my workflow.

I am aware of the ongoing work on the Linux FireWire stack, but my
understanding was that with ALSA I should not be getting xruns on my
system. For reference: I am running a Ryzen 7 7700 CPU and tested with both
RT and Liquorix kernels.

Best regards,
Ilya Kuznetsov

ср, 3 сент. 2025 г., 04:45 Takashi Sakamoto <o-takashi@sakamocchi.jp>:

> Hi,
>
> I'm sorry for your inconvenience.
>
> If you have ever used your device with Pipewire, I would like to see
> your experiences how it works, since both servers uses the different
> ways to handle ALSA PCM device in respect to no-period-wakeup mode..
>
> On Tue, Sep 02, 2025 at 03:11:52PM +0500, Илья Кузнецов wrote:
> > ________________________________
> >
> > Hello ALSA developers,
> >
> > I would like to report a reproducible issue with snd-fireface on Linux
> > kernels 6.10 through 6.17-rc2.
> >
> > ________________________________
> >
> > System Information
> >
> > Kernel versions tested: 6.10 → 6.17-rc2
> >
> > Audio interface: RME Fireface 800
> >
> > Driver in use: snd-fireface
> >
> > Audio server: JACK (tested with jackd and QJackCtl)
> >
> > Distribution: Debian 12
> >
> > Kernel config: ALSA FireWire drivers enabled, no FFADO in use
> >
> > Chip 1394: Texas Instrumets
> >
> > ________________________________
> >
> > Steps to Reproduce
> >
> > Boot kernel 6.10 … 6.17-rc2 with snd-fireface loaded.
> >
> > Start JACK with ALSA backend, for example:
> >
> > jackd -d alsa -d hw:Fireface -p 64 -n 3
> >
> > Use the system normally — during playback, recording, or even when
> > idle. The issue may occur at any time without apparent external cause.
> >
> > Actual Behavior
> >
> > After some idle time, ALSA stops responding and JACK prints:
> >
> > ERROR: ALSA: poll time out, polled for ... usecs
> > ERROR: JackAudioDriver::ProcessAsync: read error, stopping...
> >
> > JACK immediately terminates. The Fireface 800 does not recover until
> > JACK is restarted.
> >
> > ________________________________
> >
> > Expected Behavior
> >
> > The ALSA FireWire driver should maintain the stream and allow JACK to
> > continue running without manual restart, even when idle.
> >
> > ________________________________
> >
> > Notes
> >
> > With FFADO backend (jackd -d firewire), the same hardware works
> > reliably and does not exhibit this issue.
> >
> > This suggests that the problem lies in the ALSA FireWire streaming
> > engine rather than the hardware itself.
> >
> > The problem has been reproducible across all tested kernels (vanila,
> > rt, liquorix) from 6.10 through 6.17-rc2.
> >
> > ________________________________
> >
> > Request
> >
> > Could you please confirm if this issue is known?
> > Is there any ongoing work on the ALSA FireWire streaming engine
> > (snd-fireface) to address these poll timeout conditions?
> >
> > I can provide additional logs (dmesg, journalctl) or test patches if
> needed.
> >
> > Thank you,
> >
> > Ilya Kuznetsov.
>
>
> Regards
>
> Takashi Sakamoto
>