Channel mapping randomly change on usb soundcard and sucpicious dmesg log

[sprudent]

Hello erveryone,

I am getting random channel swap when using speaker test to output sequencially a pink noise on every channel of the soundcard. I tested two distinct usb soundcard getting the same issue. I could identify a suspicious log showing at the same time the channels swapped. Tested and reproduced on a recent archlinux image (see system information).

[Aug16 18:26] xhci_hcd 0000:00:14.0: ERROR Transfer event TRB DMA ptr not part of current TD ep_index 11 comp_code 1
[  +0.000134] xhci_hcd 0000:00:14.0: Looking for event-dma 0000000014ed98e0 trb-start 0000000014ed98f0 trb-end 0000000014ed98f0 seg-start 0000000014ed9000 seg-end 0000000014ed9ff0

system information

uname -r

5.18.7-arch1-1

aplay -l

**** List of PLAYBACK Hardware Devices ****
card 0: Audio [Intel HDMI/DP LPE Audio], device 0: HdmiLpeAudio [Intel HDMI/DP LPE Audi]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 0: Audio [Intel HDMI/DP LPE Audio], device 1: HdmiLpeAudio [Intel HDMI/DP LPE Audi]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 0: Audio [Intel HDMI/DP LPE Audio], device 2: HdmiLpeAudio [Intel HDMI/DP LPE Audi]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 1: ICUSBAUDIO7D [ICUSBAUDIO7D], device 0: USB Audio [USB Audio]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 2: bytchtes8316 [bytcht-es8316], device 0: Audio (*) []
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 2: bytchtes8316 [bytcht-es8316], device 1: Deep-Buffer Audio (*) []
  Subdevices: 1/1
  Subdevice #0: subdevice #0

alsa-info.txt

Test with dmix

Description

First test was made with alsa dmix plugin. Multiple processes running in parrallel and outputing to dmix pcm device.

Alsa configuration

pcm.play_8
{
    @args [ CARD ]
    @args.CARD 
    {
        type string
    }

    type dmix

    ipc_key 5678

    slave
    {
        pcm $CARD
        channels 8
        period_time 0
        period_size 1024
        buffer_size 4096
        rate 44100
    }

    bindings {
        0 0
        1 1
        2 2
        3 3
        4 4
        5 5
        6 6
        7 7
    } 
}

Step to reproduce

Run this script to simulate multiple processes writing to dmix device at the same time

for i in $(seq 20)
do
        speaker-test -r44100 -D"play_8:\"hw:1\"" -c8 -t pink &
done

Then on another tty, start a speaker-test command outputing a sine wave in order to identify channel mapping by correlating speaker test output with audio output. speaker-test -r44100 -D"play_8:\"hw:1\"" -c8 -t sine

 0 - Front Left
 4 - Front Center
 1 - Front Right
 3 - Side Right
 7 - Side Right
 2 - Side Left
 6 - Side Left
 5 - LFE
Time per period = 21.998519
...

Expected result

Audio output to the same channels during all the test

Actual result

Random audio channel swap correlated with dmesg log mentioned above.

Second test with direct output

Alsa configuration

No specific configuration required

Step to reproduce

Run speaker-following test command and wait for channel swap speaker-test -r44100 -Dsurround71:ICUSBAUDIO7D -c8

Expected result

Audio output to the same channels during all the test

Actual result

Random audio channel swap correlated with dmesg log mentioned above. The chanel swap happen far less often than with dmix.

Interpretation

The DMA transfer ring buffer get messed up after a while (see dmesg log) and interleaved audio frames get shifted in soundcard buffer. If I unplug/replug usb soundcard and try again, channel mapping goes back to the correct order.

My alsa, kernel and usb driver knowledge is limited, so I come to you for a little help. Tell me if you need more information to reproduce.

Thank's, Sylvain

[perexg]

This does not look like an alsa-lib's issue. Please, create a driver bug at https://bugzilla.kernel.org (Audio group). Although this issue may be also USB core related.

[sahi1422]

@sprudent I am facing the same issue with kernel 5.10 Did you get the fix?

[sahi1422]

@sprudent , with kernel 4.19 it used to work fine.

[frno7]

Hello. I’ve got Linux 6.1.72 with this ICUSBAUDIO7D, messing up the channels randomly as described above. My trigger is the mpv media player, and the channels rearrange randomly when fast forwarding for example a video with 5.1 surround sound. As mentioned, the physical USB hardware has to be unplugged and replugged to correct the channels. Is there any known fix for this?

[frno7]

Turns out the problem with my ICUSBAUDIO7D was caused by a misbehaving USB hub, more specifically a Dell Computer dock. Removing this USB hub fixed it.