Bug#574433: linux-image-2.6.32-3-amd64: unable to handle kernel NULL pointer dereference at (null)

[Jonathan Nieder <jrnieder@gmail.com>]

Hi,

Erik de Castro Lopo wrote:

> Shortly after coming out of hibernate I plugged in a USB mouse. When the mouse
> didn't work I found the following in dmesg.
[...]
> [23415.258904] usb 6-1: New USB device found, idVendor=062a, idProduct=0000
> [23415.258913] usb 6-1: New USB device strings: Mfr=0, Product=0, SerialNumber=0

Hmm.  This is your mouse.

> [23415.259062] BUG: unable to handle kernel NULL pointer dereference at (null)
> [23415.259072] IP: [<ffffffff8118b599>] strcmp+0x5/0x1a

Calling strcmp with second argument NULL, which means

[...]
> [23415.259466] Call Trace:
> [23415.259479]  [<ffffffff8113debc>] ? sysfs_find_dirent+0x1b/0x2f

the "name" parameter to sysfs_find_dirent is NULL, which usually would
mean

> [23415.259487]  [<ffffffff8113dfca>] ? __sysfs_add_one+0x18/0x84
> [23415.259504]  [<ffffffff8113e08c>] ? sysfs_add_one+0x19/0xe4
> [23415.259520]  [<ffffffff8113db24>] ? sysfs_add_file_mode+0x4e/0x7f
> [23415.259534]  [<ffffffff8113faf3>] ? internal_create_group+0xdc/0x15a

the name field of one of the attributes passed to sysfs_create_group
was NULL, so

> [23415.259551]  [<ffffffff8121296c>] ? device_add_groups+0x22/0x5d
> [23415.259560]  [<ffffffff81213060>] ? device_add+0x2d2/0x537

an attribute in dev->groups or dev->type->groups had a NULL name, but

> [23415.259584]  [<ffffffffa00811ec>] ? usb_new_device+0x125/0x186 [usbcore]
> [23415.259609]  [<ffffffffa00828f6>] ? hub_thread+0xc23/0x117d [usbcore]
> [23415.259621]  [<ffffffff81064a56>] ? autoremove_wake_function+0x0/0x2e
> [23415.259644]  [<ffffffffa0081cd3>] ? hub_thread+0x0/0x117d [usbcore]
> [23415.259653]  [<ffffffff81064789>] ? kthread+0x79/0x81

I don't see why that would happen.  I guess it would be nice to know
which attribute, and to know whether we can trust anything at all or if
there is some rampaging driver clearing memory.

Was this reproducible?  If so, is it still?  Colin, could you send
kernel messages from the similar event you experienced (as a followup
to this report if it involved plugging in a mouse after hibernate, or
as a new report otherwise)?

Thanks,
Jonathan