Bugzilla – Bug 1212019
Thunderbolt USB audio disconnecting with Kernel 6.3.4
Last modified: 2023-11-08 14:22:03 UTC
I have a laptop with a TB2 dock connected via TB3 using the TB3->TB2 Apple adapter. I'm using USB audio on this dock and since kernel 6.3.4, I get a random disconnection that can happen 1h or a few hours after boot: [Jun 5 11:25] xhci_hcd 0000:09:00.0: WARN Event TRB for slot 1 ep 1 with no TDs queued? [ +0.033032] xhci_hcd 0000:09:00.0: WARN Event TRB for slot 1 ep 1 with no TDs queued? [ +0.906993] xhci_hcd 0000:09:00.0: xHCI host controller not responding, assume dead [ +0.000037] xhci_hcd 0000:09:00.0: HC died; cleaning up [ +0.000041] usb 3-4: USB disconnect, device number 2 When this happens and audio is playing, it stops playing via the dock and switches to my laptop speakers. Replugging the TB3 adapter usually make it work again (if audio is playing it switches automatically from speakers to TB USB audio), for a while. That setup was rock solid until updated to kernel 6.3.4.
So when you boot with 6.3.3 kernel on the same system now, the problem doesn't happen?
Also, I'm building a test kernel with the backport of the thunderbolt fix patch in the upstream subsystem tree. It's being built in OBS home:tiwai:bsc1212019 repo. Once after the build finishes (it takes an hour or so), could you check whether it works better?
Unfortunately, I moved place today for 2 weeks and do not have my TB dock for testing as I only use it at home. I will be able to test when I'm back.
I have access to my TB dock again and this issue just happened with kernel 6.3.7. Should I test older kernel at home:tiwai:bsc1212019 ?
Not sure whether 6.3.7 contains the backport. At best, check with the kernel in OBS Kernel:stable tree instead. That's the very latest one.
Still happening with Kernel 6.3.9: [ 2422.754635] xhci_hcd 0000:09:00.0: xHCI host controller not responding, assume dead [ 2422.754674] xhci_hcd 0000:09:00.0: HC died; cleaning up [ 2422.754716] usb 3-4: USB disconnect, device number 2
Could you check 6.4 kernel in OBS Kernel:stable? A known regression of TW seems to have been already addressed there.
I just installed 6.4 and will report.
Still happened after with 6.4 after about 2h of playing audio, although with different logging: [Jun28 12:42] xhci_hcd 0000:09:00.0: WARN Event TRB for slot 1 ep 1 with no TDs queued? [ +0.007571] xhci_hcd 0000:09:00.0: ERROR Transfer event TRB DMA ptr not part of current TD ep_index 1 comp_code 1 [ +0.000011] xhci_hcd 0000:09:00.0: Looking for event-dma 00000001dba553e0 trb-start 00000001dba553d0 trb-end 00000001dba553d0 seg-start 00000001dba55000 seg-end 00000001dba55ff0 [ +0.000007] xhci_hcd 0000:09:00.0: ERROR Transfer event TRB DMA ptr not part of current TD ep_index 1 comp_code 1 [ +0.000004] xhci_hcd 0000:09:00.0: Looking for event-dma 00000001dba553f0 trb-start 00000001dba553d0 trb-end 00000001dba553d0 seg-start 00000001dba55000 seg-end 00000001dba55ff0 [ +0.000005] xhci_hcd 0000:09:00.0: ERROR Transfer event TRB DMA ptr not part of current TD ep_index 1 comp_code 1 [ +0.000004] xhci_hcd 0000:09:00.0: Looking for event-dma 00000001dba55400 trb-start 00000001dba553d0 trb-end 00000001dba553d0 seg-start 00000001dba55000 seg-end 00000001dba55ff0 [ +0.000004] xhci_hcd 0000:09:00.0: ERROR Transfer event TRB DMA ptr not part of current TD ep_index 1 comp_code 1 [ +0.000004] xhci_hcd 0000:09:00.0: Looking for event-dma 00000001dba55410 trb-start 00000001dba553d0 trb-end 00000001dba553d0 seg-start 00000001dba55000 seg-end 00000001dba55ff0 [ +0.000005] xhci_hcd 0000:09:00.0: ERROR Transfer event TRB DMA ptr not part of current TD ep_index 1 comp_code 1 [ +0.000003] xhci_hcd 0000:09:00.0: Looking for event-dma 00000001dba55420 trb-start 00000001dba553d0 trb-end 00000001dba553d0 seg-start 00000001dba55000 seg-end 00000001dba55ff0 [ +0.036511] xhci_hcd 0000:09:00.0: WARN Event TRB for slot 1 ep 1 with no TDs queued? [ +1.746876] xhci_hcd 0000:09:00.0: xHCI host controller not responding, assume dead [ +0.000038] xhci_hcd 0000:09:00.0: HC died; cleaning up I'm wondering if my old TB2 dock is not just dying...
I'm testing the possibility that the 135mv undervolt that I apply to my laptop CPU is causing this issue and will report.
Confirming it is still happening with no CPU undervolt.
Still happening with Kernel 6.4.3 although it took a tad longer to happen this time. Only the USB hub part of that TB dock is disconnecting. I also have a DisplayPort monitor connected to that dock and it is not affected.
The report should be better moved to the upstream bug tracker, I suppose. e.g. bugzilla.kernel.org. Could you try it?
I will if it happens again once I have updated to 6.4.4. With 6.4.3 is only happened once in several days of usage. With 6.3.x it was much more frequent. Still wonder if my old TB2 dock is not the problem, overheating or something.
Still happening regularly with kernel 6.4.4. Issue submitted: https://bugzilla.kernel.org/show_bug.cgi?id=217715
I have not seen this issue happening since Kernel 6.5.6, so closing it.