linux usb kernel crashes when vbox is resumed

[karlp]

If I start up a VM, and use USB, it all works just fine. if I close the vbox session, and choose "Save the machine state", when I resume it later (like the next day) the usb subsystem is broken. Any attempt to access a usb device (even lsusb) hangs completely.

/var/log/messages contains the following entries:

Oct 20 17:19:24 dragnet kernel: [114714.960939] e1000: eth1 NIC Link is Down
Oct 20 17:19:24 dragnet kernel: [114714.970867] e1000: eth0 NIC Link is Down
<<<<< this is where it resumes....>>>>>>>
Oct 21 09:35:54 dragnet kernel: [114719.244611] ohci_hcd 0000:00:06.0: ED unlink timeout
Oct 21 09:35:54 dragnet kernel: [114719.381110] usb 2-2: USB disconnect, address 3
Oct 21 09:35:55 dragnet kernel: [114719.716299] usb 2-2: new full speed USB device using ohci_hcd and address 5
Oct 21 09:35:55 dragnet kernel: [114719.840586] e1000: eth1 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
Oct 21 09:35:55 dragnet kernel: [114719.841699] e1000: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
Oct 21 09:39:56 dragnet kernel: [114961.077228] khubd         D 01b47ee3     0   155      2 0x00000000
Oct 21 09:39:56 dragnet kernel: [114961.077235]  f6d53b80 00000046 f65c7dec 01b47ee3 00000292 c141d100 c141d100 c14186ac
Oct 21 09:39:56 dragnet kernel: [114961.077242]  f6d53d3c c2508100 00000001 00000292 00000000 f7e519e0 f6fd2c00 df329380
Oct 21 09:39:56 dragnet kernel: [114961.077249]  c25036ac f6d53d3c 01b47ed0 00000000 df329380 00000000 00000000 00000000
Oct 21 09:39:56 dragnet kernel: [114961.077256] Call Trace:
Oct 21 09:39:56 dragnet kernel: [114961.077292]  [<f7d14efb>] ? usb_kill_urb+0x84/0xa0 [usbcore]
Oct 21 09:39:56 dragnet kernel: [114961.077310]  [<c104402e>] ? autoremove_wake_function+0x0/0x2d
Oct 21 09:39:56 dragnet kernel: [114961.077322]  [<f7d15f1c>] ? usb_start_wait_urb+0x62/0x92 [usbcore]
Oct 21 09:39:56 dragnet kernel: [114961.077332]  [<f7d1535e>] ? usb_init_urb+0x1a/0x26 [usbcore]
Oct 21 09:39:56 dragnet kernel: [114961.077343]  [<f7d16157>] ? usb_control_msg+0xf7/0x111 [usbcore]
Oct 21 09:39:56 dragnet kernel: [114961.077355]  [<c126cd97>] ? printk+0xe/0x17
Oct 21 09:39:56 dragnet kernel: [114961.077365]  [<f7d10c42>] ? hub_port_init+0x2cd/0x650 [usbcore]
Oct 21 09:39:56 dragnet kernel: [114961.077376]  [<f7d12b2a>] ? hub_thread+0x819/0xe53 [usbcore]
Oct 21 09:39:56 dragnet kernel: [114961.077382]  [<c104402e>] ? autoremove_wake_function+0x0/0x2d
Oct 21 09:39:56 dragnet kernel: [114961.077392]  [<f7d12311>] ? hub_thread+0x0/0xe53 [usbcore]
Oct 21 09:39:56 dragnet kernel: [114961.077397]  [<c1043dfc>] ? kthread+0x61/0x66
Oct 21 09:39:56 dragnet kernel: [114961.077402]  [<c1043d9b>] ? kthread+0x0/0x66
Oct 21 09:39:56 dragnet kernel: [114961.077409]  [<c1003d47>] ? kernel_thread_helper+0x7/0x10

The khubd backtrace repeates itself every few minutes for about 20 minutes, then ceases. My only workaround so far is to reboot the VM

Guest OS is debian 6.0.3, Linux dragnet 2.6.32-5-686 #1 SMP Mon Oct 3 04:15:24 UTC 2011 i686 GNU/Linux

I can't remember when I started seeing this, but it was definitely in 4.1.2, and is also in 4.1.4

[karlp]

Seems similar to ​http://thread.gmane.org/gmane.linux.kernel/999410