On Tue, 15 Dec 2020 10:44:45 +0100, Marco Giunta wrote:
Hi, recently I've bought a usb webcam with integrated mic:
Jieli Technology USB PHY 2.0 (1224:2a25)
The video part works well, but the mic sound is speedups, "like minions" (cit.). When I connect the camera, these are the dmesg messages:
kernel: usb 1-8: current rate 0 is different from the runtime rate 8000 kernel: usb 1-8: current rate 0 is different from the runtime rate 16000 kernel: usb 1-8: current rate 0 is different from the runtime rate 44100 kernel: usb 1-8: current rate 0 is different from the runtime rate 48000 kernel: usb 1-8: Warning! Unlikely big volume range (=4096), cval->res is probably wrong. kernel: usb 1-8: [3] FU [Mic Capture Volume] ch = 1, val = 0/4096/1 kernel: usbcore: registered new interface driver snd-usb-audio
and after a while, dmesg log is filled, every 5 seconds, with:
kernel: retire_capture_urb: 84 callbacks suppressed kernel: retire_capture_urb: 1714 callbacks suppressed
A guy reports on ArchLinux bug website the same problem (https://bugs.archlinux.org/task/68141?opened=12995&status%5B0%5D=) and provides a patch to fix the sound issue. I've applied the patch on kernel 5.9.13 (Fedora 33 x86_64) and now the mic works, no more minions voice effect. Now dmesg messages are only:
kernel: usb 1-8: Warning! Unlikely big volume range (=4096), cval->res is probably wrong. kernel: usb 1-8: [3] FU [Mic Capture Volume] ch = 1, val = 0/4096/1 kernel: usbcore: registered new interface driver snd-usb-audio
the retire_capture_urb messages are gone.
All credits for the patch go to him but I don't know how to contact that guy nor I don't know if he has already contacted you, so my question is if you could review his patch and finally apply upstream.
If you need other information or you need a tester, I'm here.
Thanks for the patch. The still remaining warnings are about the mixer, and your patch doesn't touch about it. You may apply the similar change in volume_control_quirks() like other webcams.
And now about the patch:
--- a/sound/usb/format.c 2020-10-01 18:36:35.000000000 +0300 +++ b/sound/usb/format.c 2020-10-04 02:10:21.678685952 +0300 @@ -217,6 +217,21 @@ (chip->usb_id == USB_ID(0x041e, 0x4064) || chip->usb_id == USB_ID(0x041e, 0x4068))) rate = 8000;
// hack for "Jieli Technology USB PHY 2.0" webcamif (chip->usb_id == USB_ID(0x1224, 0x2a25)) {switch (rate) {case 8000:fp->datainterval += 4;break;case 16000:fp->datainterval += 3;break;default:fp->datainterval += 1;break;}}
Modifying datainterval at this point doesn't look intuitive. What's the original datainterval value for those altsettings? The value is retrieved in snd_usb_parse_datainterval() in helper.c, and if any, the correction there would be more sensible.
fp->rate_table[fp->nr_rates] = rate; if (!fp->rate_min || rate < fp->rate_min)--- a/sound/usb/endpoint.c 2020-10-01 18:36:35.000000000 +0300 +++ b/sound/usb/endpoint.c 2020-10-04 02:09:09.471978982 +0300 @@ -882,6 +882,8 @@ if (snd_usb_get_speed(ep->chip->dev) != USB_SPEED_FULL) { packs_per_ms = 8 >> ep->datainterval; max_packs_per_urb = MAX_PACKS_HS;
if (!packs_per_ms)packs_per_ms = 1;
This rather indicates that the datainterval is somehow wrong.
} else { packs_per_ms = 1; max_packs_per_urb = MAX_PACKS;--- a/sound/usb/quirks.c 2020-10-01 18:36:35.000000000 +0300 +++ b/sound/usb/quirks.c 2020-10-04 02:14:04.532196519 +0300 @@ -1516,6 +1516,7 @@ case USB_ID(0x1901, 0x0191): /* GE B850V3 CP2114 audio interface */ case USB_ID(0x21b4, 0x0081): /* AudioQuest DragonFly */ case USB_ID(0x2912, 0x30c8): /* Audioengine D1 */
- case USB_ID(0x1224, 0x2a25): /* Jieli Technology USB PHY 2.0 */ return true; }
This looks fine.
thanks,
Takashi