Re: 答复: [PATCH] xhci: sideband: check vdev liveness before removing endpoints on unregister

[Wesley Cheng]

On 9/15/2026 1:35 AM, 胡连勤 wrote:
> Hi Michal,
>
> > > Unfortunately, no. The reproduction rate is very low, and we have
> > > not yet found a stable way to reproduce it — it requires the
> > > device to be in a specific state where the port is disabled after
> > > autosuspend resume, which happens rarely in the field.
> >
> > Why are offloaded devices being suspended at all? Obviously, no
> > "sideband entity" can be using a suspended device, so it seems that:
> >
> > - if the device is unused and meant to be suspended, sideband can be
> >    removed before suspending
> > - if the device becomes suspended while in use, obviously it's a bug
>
> You raise a valid point. Looking at the code:
>
> The snd-usb-audio driver sets supports_autosuspend = 1
> (card.c:1347), so the device can autosuspend when no audio stream
> is active.
>
> During suspend, usb_audio_suspend() calls
> platform_ops->suspend_cb → qc_usb_audio_offload_suspend(),
> which sends a QMI disconnect indication to the ADSP and waits for
> dev->in_use to be cleared.
>
> However, qc_usb_audio_offload_suspend() does not call
> xhci_sideband_remove_endpoint() or xhci_sideband_unregister().
> The sideband registration stays active during suspend — the
> endpoint pointers to xHCI ring buffers remain in place.

Hi Lianqin,

Sorry for interjecting here, but just wanted to point out the detail that 
when we call qc_usb_audio_offload_suspend() --> 
uaudio_send_disconnect_ind() this will ensure that the audio stream is 
idle/not active.  For example, if there is currently audio playback 
happening, then this will issue a command to the aDSP to stop the audio 
playback and disable the offload stream by issuing 
handle_uaudio_stream_req(req_msg->enable = 0).  This will invoke the 
sideband remove endpoint path.

If the audio stream is already idle, then we shouldn't have an active 
endpoint registered to xhci sideband.


> This is the root design issue: the sideband client assumes the
> xHCI ring buffers remain valid across suspend/resume cycles. But
> on reset-resume, usb_reset_and_verify_device() → hub_port_init()
> → xhci_discover_or_reset_device() frees and reallocates those
> rings, invalidating the sideband's pointers.

Should be only mapped when audio offload is active.  Also, we explicitly 
take a runtime PM reference on enable_audio_stream() using 
snd_usb_autoresume() and remove this reference on disable_audio_stream(), 
so minus a PM suspend transition, we shouldn't see the 
qc_usb_audio_offload_suspend() callback executing when audio playback is 
happening.

Thanks
Wesley Cheng