Re: kernel oops

"Also sprach Peter T. Breuer:"
> "Also sprach [email protected]:"
> >  [] __invalidate_device+0x36/0x50
> >  [] enbd_soft_reset+0x95/0x100 [enbd]
> >  [] enbd_write_proc+0x218/0x240 [enbd]
> >  [] enbd_write_proc+0x0/0x240 [enbd]
> >  [] proc_file_write+0x29/0x40
> >  [] proc_file_write+0x0/0x40
> >  [] vfs_write+0x93/0x110
> >  [] sys_write+0x3c/0x70
> >  [] sysenter_past_esp+0x52/0x79
> 
> When did this happen? When removing the module in a 2.6 kernel? Don't do
> that then.

Actually, that trace looks more like "writing a 1 to /proc/nbdinfo" (I
haven't checked the code to see). I wonder when? On the way through the
module remove code the proc dir and the device are disabled at slightly
different moments, which might allow the proc code to access a device
that no longer existed for a brief millisecond, but that race is too
insignificant to worry about!

Anyway, tell me about it.

Peter