** Description changed:
SRU Justification:
[ Impact ]
- xhci_endpoint_disable() clears host_ep->hcpriv = NULL, which breaks
- xhci_endpoint_reset(). When a USB driver (e.g. uvcvideo) calls
- usb_set_interface(), submits URBs that make host sequence state non-zero,
- then calls usb_clear_halt(), the device clears its sequence state but
- xhci_endpoint_reset() bails out because hcpriv is NULL. The next URB
+ USB camera is lost after suspend. xhci_endpoint_disable() clears
+ host_ep->hcpriv = NULL, which breaks xhci_endpoint_reset(). When uvcvideo
+ calls usb_set_interface(), submits URBs that make host sequence state
+ non-zero, then calls usb_clear_halt(), the device clears its sequence state
+ but xhci_endpoint_reset() bails out because hcpriv is NULL. The next URB
malfunctions: USB2 loses one packet, USB3 gets Transaction Error or may
not complete at all on some host controllers from ASMedia and AMD.
- This is triggered by uvcvideo on bulk video devices.
[ Fix ]
Cherry-pick upstream mainline commit:
- 25e531b422dc ("usb: xhci: Make usb_host_endpoint.hcpriv survive
endpoint_disable()")
Fixes: 18b74067ac78 ("xhci: Fix use-after-free regression in xhci clear hub
TT implementation")
Cc: [email protected]
[ Test Plan ]
- 1. System with USB3 bulk video device (e.g. USB webcam using uvcvideo)
- 2. Use the camera with an application (e.g. cheese, guvcview)
- 3. Trigger usb_set_interface + usb_clear_halt sequence
- 4. Verify no Transaction Errors or packet loss in dmesg
- 5. Verify endpoint_reset works correctly after endpoint_disable
+ 1. System with USB camera (uvcvideo bulk video device)
+ 2. Suspend and resume the system
+ 3. Verify camera is still available (/dev/video0 exists)
+ 4. Verify no Transaction Errors in dmesg
+ 5. Verify camera works in application (e.g. cheese, guvcview)
[ Where problems could occur ]
The fix removes one line (host_ep->hcpriv = NULL) from
xhci_endpoint_disable().
- Risk is very low — the commit message explains hcpriv should only be NULL on
+ Risk is very low - the commit message explains hcpriv should only be NULL on
emulated root hub endpoints, and core should not try to reset dropped
endpoints.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2153966
Title:
USB camera lost after suspend due to xhci endpoint_reset failure
To manage notifications about this bug go to:
https://bugs.launchpad.net/hwe-next/+bug/2153966/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
