Re: Kernel Oops while disconnecting USB peripheral (always)

[BjÃrn Mork]

Daniel Mack <zonque@xxxxxxxxx> writes:
> On 23.07.2012 16:47, Alan Stern wrote:
>> On Mon, 23 Jul 2012, Sarbojit Ganguly wrote:
>>> That is why I provided two stacks,
>>>
>>> 1st one is when I tried to remove the USB hub (which connects a webcam
>>> + microphone)
>>> 2nd one is when I tried to remove an USB powered external HDD.
>>>
>>> Just to make sure whether the problem is with USB sound or the USB subsystem.
>> 
>> Do you stop all the programs that are using the USB devices before 
>> unplugging the hub?  Do you unmount the USB HDD first?
>> 
>> The first crash shows a problem in the snd-usb-audio driver.
>> 
>> The second crash shows a problem in the VFS layer or in ext3, not in 
>> the USB stack.
>
> I dare to doubt there are two severe bugs of that kind that are 100%
> reproducible. I haven't had a  hotplug crash in any of the two drivers
> for a long time, and I use both of them extensively.

Actually, based on the recent usb_wwan experience, I'd say that two such
bugs isn't as unlikely as it may seem at first.  Even three if we add
the now fixed usb_wwan (or six, if we count the three drivers affected
by the usb_wwan bug).  There are probably even more.

The reason is this change:

 0998d0631 device-core: Ensure drvdata = NULL when no driver is bound


It will make bugs like this suddenly 100% reproducible.  But the bugs
*are* in the drivers, and may have been there for a long time.  The
drivers have been accessing drvdata after unbinding.  They just didn't
crash prior to that commit.

But the commit is correct, and a very much needed improvement if my
assumptions are correct.  The drivers need fixing and this just makes it
evident.

> I rather assume there's something else failing, probably some host
> controller issue that corrupts mempory? Can anyone else reproduce this
> maybe?

You could of course do like me and bisect, but if you end up with commit
0998d0631 then you haven't really learned more than confirming my guess.


BjÃrn
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/