On Thu, Mar 24, 2022 at 03:44:41AM +0100, Petr Janecek wrote:
> Hi Alan,
>
> On 3/22/22 19:03, Alan Stern wrote:
> > On Sun, Mar 20, 2022 at 02:15:40AM +0100, Petr Janecek wrote:
> > > Hi,
> > > I'm trying to use Apogee ONEv2. From usb dumps under
> > > macos or windows it seems like a usb Class Audio 2.0
> > > device, but in linux, it keeps resetting every two seconds
> > > or so. It keeps resetting even when the snd-usb-audio
> > > driver is disabled, so the problem is probably at a lower
> > > level.
> >
> > Can you post the log output on a system where snd-usb-audio is disabled?
>
> [ 3412.279063] usb 3-8: new high-speed USB device number 6 using xhci_hcd
> [ 3412.470003] usb 3-8: New USB device found, idVendor=0c60, idProduct=0017,
> bcdDevice= 1.05
> [ 3412.470017] usb 3-8: New USB device strings: Mfr=1, Product=2,
> SerialNumber=3
> [ 3412.470023] usb 3-8: Product: ONEv2
> [ 3412.470028] usb 3-8: Manufacturer: Apogee
> [ 3412.470033] usb 3-8: SerialNumber: 0C12FF2020204652334D513A7A2A9B
> [ 3413.217053] usb 3-8: USB disconnect, device number 6
> [ 3413.629046] usb 3-8: new high-speed USB device number 7 using xhci_hcd
> [ 3413.829760] usb 3-8: New USB device found, idVendor=0c60, idProduct=0017,
> bcdDevice= 1.05
> [ 3413.829766] usb 3-8: New USB device strings: Mfr=1, Product=2,
> SerialNumber=3
> [ 3413.829768] usb 3-8: Product: ONEv2
> [ 3413.829770] usb 3-8: Manufacturer: Apogee
> [ 3413.829772] usb 3-8: SerialNumber: 0C12FF2020204652334D513A7A2A9B
> [ 3414.421964] usb 3-8: USB disconnect, device number 7
> [ 3414.838981] usb 3-8: new high-speed USB device number 8 using xhci_hcd
> [ 3415.029947] usb 3-8: New USB device found, idVendor=0c60, idProduct=0017,
> bcdDevice= 1.05
> [ 3415.029961] usb 3-8: New USB device strings: Mfr=1, Product=2,
> SerialNumber=3
> [ 3415.029968] usb 3-8: Product: ONEv2
> [ 3415.029973] usb 3-8: Manufacturer: Apogee
> [ 3415.029978] usb 3-8: SerialNumber: 0C12FF2020204652334D513A7A2A9B
> [ 3415.627052] usb 3-8: USB disconnect, device number 8
Nothing particularly suspicious there.
>
> > > The messages below are from v5.16.16, but it behaves
> > > the same no matter what I plug it into.
> > >
> > > [ 253.708616] usb 3-8: new high-speed USB device number 6 using xhci_hcd
> > > [ 253.899363] usb 3-8: New USB device found, idVendor=0c60, idProduct=0017,
> > > bcdDevice= 1.05
> > > [ 253.899370] usb 3-8: New USB device strings: Mfr=1, Product=2,
> > > SerialNumber=3
> > > [ 253.899373] usb 3-8: Product: ONEv2
> > > [ 253.899375] usb 3-8: Manufacturer: Apogee
> > > [ 253.899377] usb 3-8: SerialNumber: 0C12FF2020204652334D513A7A2A9B
> > > [ 253.960901] mc: Linux media interface: v0.10
> >
> > Any idea where that line came from?
>
> That's from CONFIG_MEDIA_CONTROLLER, which is selected for usb
> webcam. It also selects CONFIG_SND_USB_AUDIO_USE_MEDIA_CONTROLLER.
> Compiled it out, rebooted. Now the line is missing, but no other
> change.
>
> > You should try capturing a usbmon trace showing what happens when the
> > device is plugged in and then resets. Preferably on a system where
> > snd-usb-audio is disabled.
>
> Trace from wireshark is attached.
The trace doesn't help much, unfortunately. It shows normal device
initialization, plus a couple of extra string descriptor reads. About
1/2 second afterward, the device disconnects itself electronically from
the USB bus and reconnects 20 ms later.
I get the feeling that the device expects to receive some specific
messages from the host within that one-half second, and disconnects if
it doesn't see them. And presumably it doesn't get what it's looking
for even when the USB sound drivers are enabled.
Although I doubt it will show anything helpful, you might try collecting
a usbmon trace with the sound drivers enabled. Perhaps it will suggest
something to the USB audio developers.
> > One other thing you might try: Disable runtime PM for USB ("echo -1
> > > /sys/module/usbcore/parameters/autosuspend" before plugging in the
> > device).
>
> Makes no difference -- almost: after few resets it once took much
> longer (>10x) to reconnect.
Oh well, it was worth a try. Some devices don't like runtime suspend.
But obviously that isn't the problem here.
Alan Stern
