VirtualBox

Opened 2 years ago

Closed 2 years ago

#21108 closed defect (fixed)

VM crashes when xHCI controller is reset while USB3 disk is attached

Reported by: mgrzegor Owned by:
Component: USB Version: VirtualBox 6.1.38
Keywords: USB, xHCI, crash Cc:
Guest type: Linux Host type: Windows

Description

Since VirtualBox version 6.1.36 my Ubuntu VM crashes if I reset the xHCI controller (which I occasionally have to do due to another issue) while one or more of my SuperSpeed (i.e. USB 3.0) external disks are attached to the VM.

The crash also occurs on system boot if one of those disks are attached (which I usually do not do).

Reproduction steps:

  1. Start the VM, wait for the GUI to come up.
  2. Start gnome-terminal.
  3. In the terminal window, type:
    sudo -i
    (to obtain root prompt).
  4. Via the VM menu -> Devices -> USB, attach a USB 3.0 disk to the VM.
  5. In the terminal window, type:
    echo -n "0000:00:0c.0" > /sys/bus/pci/drivers/xhci_hcd/unbind

Result: VM crashes with an "unhandled exception" message (log attached).

The crash does not occur if a high-speed (USB 2.0) disk is attached, or if no USB devices are attached.

In all previous versions of VirtualBox I have used (back to at least 5.0.40) I observed no such crashes.

A workaround is to:

  • not attach USB 3..0 devices while booting the guest
  • detach all USB 3.0 devices before resetting the xHCI controller
  • pray that no xHCI controller reset occurs while I am using any of my USB 3.0 disks (such resets have been observed to occur sporadically in all previous versions of VirtualBox, and have been harmless before.)

Attachments (2)

VBox.log.20220917-1.zip (50.9 KB ) - added by mgrzegor 2 years ago.
Windows10-2022-10-02-23-34-43.log.zip (23.5 KB ) - added by Sean Wang 2 years ago.
Reproduced in windows 10 guest

Download all attachments as: .zip

Change History (4)

by mgrzegor, 2 years ago

Attachment: VBox.log.20220917-1.zip added

by Sean Wang, 2 years ago

Reproduced in windows 10 guest

comment:1 by mgrzegor, 2 years ago

Appears to be fixed in 6.1.40 (doing the reproduction steps no longer results in a crash).

comment:2 by aeichner, 2 years ago

Resolution: fixed
Status: newclosed

Thanks for the feedback, there was indeed a fix related to this in 6.1.40.

Note: See TracTickets for help on using tickets.

© 2024 Oracle Support Privacy / Do Not Sell My Info Terms of Use Trademark Policy Automated Access Etiquette