LinuxMusicians

This was explained here:
viewtopic.php?f=27&t=21821

Drumfix wrote: ↑

Wed Aug 19, 2020 9:42 pm

Easy. All devices use the first allocated irq of an xhci host controller. Simply because the use of xhci secondary event rings and allocation of devices/enpoints to them has never been implemented.

Thanks! That explains a lot.

turbidh20 wrote: ↑

Fri Jul 31, 2020 10:39 am

As you can see, IRQ 47 is the busiest, even though I can find nothing immediately attached to it.

I bet that it's a sound card. But I don't know how to find mapping between devices connected to xhci controller and their IRQs.

The patch is upstream and will be included in linux 5.9:
https://git.kernel.org/pub/scm/linux/ke ... f21048a2af

Also note that udev-rtirq (https://github.com/jhernberg/udev-rtirq) doesn't work correctly for xhci.

turbidh20 your method of finding IRQ doesn't work for xHCI controllers. These controllers use MSI/MSI-X interrupts and allocate dedicated interrupt for each connected device (at least my old uPD720202 do). 02:00.0 USB controller: Renesas Technology Corp. uPD720202 USB 3.0 Host Controller (rev 02) (...

summers wrote: ↑

Sun Aug 02, 2020 11:19 am

So the synch type seems different from the other Gen 3 devices - seems a bit odd.

This is in line with Gen 2 devices: only interfaces with digital I/O work in asynchronous mode.

One notice that if I use PulseAudio and select for output to M2 and play Youtube in browser (chrome), after I switch to input to Motu youtube starts to hang It shows loading circle but it can't load video, the same happen if I open another tab. Fixed only when I switch input to another device and r...

lilith wrote: ↑

Sat May 16, 2020 5:03 pm

Is it normal that I cannot do any selection for the Clock Source setting? Does one need this?
My interface is the Roland Rubix 24.

Roland Rubix doesn't have a digital input or word clock input, so it can only use internal clock.

I guess it requires implicit feedback quirk, just like SSL2+
https://git.kernel.org/pub/scm/linux/ke ... 308f7ffcbf

Also I doubt that 18i8 will work correctly with my patch. I should probably add additional filtering of sample rates on the output interface for this device.

Updated patch. Added Scarlett 18i8 3rd Gen. Also slightly changed the way to get altsetting.

Mr (or Mrs) Puleglot. Thank you so much for your help with this. If there's any other testing I can help with please ask. Oh, just remembeerd - I get a lot of "unable to change format on ep #82" messages when the usb-audio module is loaded, but it doesn't seem to affect anything. Thanks for testing...

These errors: "Unable to change format on ep ..." should be fixed by the following patch:
https://git.kernel.org/pub/scm/linux/ke ... 80c63bf77f
But it is not fixing all the problems with implicit feedback AFAIS.

Patch for pcm.c with added check for subs->interface >= 0. @bluebell could you test it?