Skip to contentSkip to conversation

USB Disconnect BUG with RT Kernels (linux-2.6.23-rt1)

DISCUSSION Esben Stien 2 participants 2 messages Oct 22, 2007

lore
Share & tools
lore
#0 lore raw

I also saw this in 2.6.22-rt9. When I disconnect a USB device the whole USB system goes down and I'm not able to insert any other USB device until I reboot.

This happens every time. I'm on a P4.

usb 1-1.3: USB disconnect, address 11

BUG: unable to handle kernel paging request at virtual address 00100100
 printing eip:
c031c3d9
*pde = 00000000
stopped custom tracer.
Oops: 0000 [#1]
PREEMPT 
Modules linked in: snd_seq_midi snd_rtctimer rtc radeon snd_hdsp snd_rawmidi firmware_class snd_hwdep
CPU:    0
EIP:    0060:[<c031c3d9>]    Not tainted VLI
EFLAGS: 00010206   (2.6.23-rt1 #1)

EIP is at evdev_disconnect+0x5b/0x96
eax: 00000000 ebx: 000ffcf0 ecx: 00000073 edx: c1e03450
esi: f7540000 edi: f754006c ebp: f7bb9800 esp: c1e07e84
ds: 007b es: 007b fs: 0000 gs: 0000 ss: 0068 preempt:00000001
Process khubd (pid: 82, ti=c1e06000 task=c1e03450 task.ti=c1e06000)

Stack: f7f1d400 f7ea4000 f7ea47f4 f7ea7000 c0319a64 00000000 f7ea6000 f7ea7000 
       f7bb9800 f6ac3b00 f7ea7c58 c032a139 f7ea601c f7ea6000 f7ea7000 c032d697 
       f7bb9800 f7bb981c c04afd40 f66f3858 c02fd799 f7bb981c c04afd88 00000000 
Call Trace:
 [<c0319a64>] input_unregister_device+0x67/0xfc
 [<c032a139>] hidinput_disconnect+0x2e/0x47
 [<c032d697>] hid_disconnect+0x76/0xce
 [<c02fd799>] usb_unbind_interface+0x2d/0x6e
 [<c02b7ba4>] __device_release_driver+0x71/0x8e
 [<c02b7c65>] device_release_driver+0x18/0x21
 [<c02b76f3>] bus_remove_device+0x70/0x80
 [<c02b5ddf>] device_del+0x1bb/0x22f
 [<c02fb4f1>] usb_disable_device+0x5c/0xbb
 [<c02f807d>] usb_disconnect+0x7a/0xd3
 [<c02f85a9>] hub_thread+0x30f/0x9ea
 [<c010d79a>] update_stats_wait_end+0x9b/0xbe
 [<c03c4c32>] __schedule+0x27e/0x36d
 [<c01214e5>] autoremove_wake_function+0x0/0x35
 [<c02f829a>] hub_thread+0x0/0x9ea
 [<c012142e>] kthread+0x36/0x5c
 [<c01213f8>] kthread+0x0/0x5c
 [<c0102913>] kernel_thread_helper+0x7/0x10
 =======================
Code: e8 3c c5 ff ff 8b 5e 64 eb 1b 8d 83 08 04 00 00 b9 06 00 02 00 ba 1d 00 00 00 e8 d7 fc e3 ff 8b 9b 10 04 00 00 81 eb 10 04 00 00 <8b> 83 10 04 00 00 0f 18 00 90 8d 93 10 04 00 00 8d 46 64 39 c2 
EIP: [<c031c3d9>] evdev_disconnect+0x5b/0x96 SS:ESP 0068:c1e07e84
Signature · 5 lines
Esben Stien is b0ef@e     s      a             
         http://www. s     t    n m
          irc://irc.  b  -  i  .   e/%23contact
           sip:b0ef@   e     e 
           jid:b0ef@    n     n
#1 lore raw
Quoted 13 lines · Esben Stien
 [<c0319a64>] input_unregister_device+0x67/0xfc
 [<c032a139>] hidinput_disconnect+0x2e/0x47
 [<c032d697>] hid_disconnect+0x76/0xce
 [<c02fd799>] usb_unbind_interface+0x2d/0x6e
 [<c02b7ba4>] __device_release_driver+0x71/0x8e
 [<c02b7c65>] device_release_driver+0x18/0x21
 [<c02b76f3>] bus_remove_device+0x70/0x80
 [<c02b5ddf>] device_del+0x1bb/0x22f
 [<c02fb4f1>] usb_disable_device+0x5c/0xbb
 [<c02f807d>] usb_disconnect+0x7a/0xd3
 [<c02f85a9>] hub_thread+0x30f/0x9ea
 [<c010d79a>] update_stats_wait_end+0x9b/0xbe
 [<c03c4c32>] __schedule+0x27e/0x36d
Quoted 8 lines
 [<c01214e5>] autoremove_wake_function+0x0/0x35
 [<c02f829a>] hub_thread+0x0/0x9ea
 [<c012142e>] kthread+0x36/0x5c
 [<c01213f8>] kthread+0x0/0x5c
 [<c0102913>] kernel_thread_helper+0x7/0x10
 =======================
Code: e8 3c c5 ff ff 8b 5e 64 eb 1b 8d 83 08 04 00 00 b9 06 00 02 00 ba 1d 00 00 00 e8 d7 fc e3 ff 8b 9b 10 04 00 00 81 eb 10 04 00 00 <8b> 83 10 04 00 00 0f 18 00 90 8d 93 10 04 00 00 8d 46 64 39 c2 
EIP: [<c031c3d9>] evdev_disconnect+0x5b/0x96 SS:ESP 0068:c1e07e84

Please try the latest git. Dmitry has pushed patches that implement proper locking in input layer, which are very likely going to fix the bug you are seeing.

Namely commits 8006479, 6addb1d and 464b241, so please try to reproduce with any kernel that has them.

Thanks,

Signature · 1 line
Jiri Kosina