Re: input0: bad kobj_uevent_env content in show_uevent()

Jean Delvare [off-list ref] writes:

On Sun, 2 Sep 2012 16:34:02 +0800, Fengguang Wu wrote:

quoted

It's more than the input device that can trigger the panic.
Here is another trace which panics on accessing dmi/id/uevent:

[   32.380795] sysfs_read_file: /sys/devices/virtual/dmi/id/uevent
[   32.399379] dmi id: uevent 73 env[1677721600]: id/.../id
[   32.400780] uevent 0 env[0]: (null)
[   32.401695] uevent 7 env[1]: (null)
...
[   32.478415] general protection fault: 0000 [#1] PREEMPT 
[   32.479953] CPU 0 
[   32.480433] Pid: 114, comm: trinity-child0 Not tainted 3.6.0-rc4-bisect2-00001-gb6d86d3-dirty #26 Bochs Bochs
[   32.482995] RIP: 0010:[<ffffffff81266c33>]  [<ffffffff81266c33>] strnlen+0x23/0x70
...
[   32.486234] Call Trace:
[   32.486234]  [<ffffffff81268d5c>] string.isra.4+0x4c/0x120
[   32.486234]  [<ffffffff81269f19>] vsnprintf+0x2c9/0x900
[   32.486234]  [<ffffffff8126a9c9>] vscnprintf+0x19/0x50
[   32.486234]  [<ffffffff8105f8c7>] vprintk_emit+0xe7/0x750
[   32.486234]  [<ffffffff81a49ba9>] printk+0x4f/0x58
[   32.486234]  [<ffffffff8150522b>] show_uevent+0x1cb/0x220
[   32.486234]  [<ffffffff81501dcb>] dev_attr_show+0x2b/0x90
[   32.486234]  [<ffffffff81174bea>] ? sysfs_read_file+0x18a/0x320
[   32.486234]  [<ffffffff810f2974>] ? __get_free_pages+0x24/0xc0
[   32.486234]  [<ffffffff81174c2d>] sysfs_read_file+0x1cd/0x320

I can't reproduce this, sorry, so I can't look into it. If it takes
special steps to reproduce, please tell me, as I wasn't part of the
original discussion.

The "only" step required is disabling CONFIG_HOTPLUG.

But that can be challenging.  I believe Fengguang Wu already posted some
configs.  I am attaching the config I used as well, booting this under
qemu-kvm using virtio-net-pci and ide devices.  Although my config isn't
optimized for fast builds or anything and probably enables many
unnecessary drivers.


Bjørn