Re: Race condition between driver_probe_device and device_shutdown

[Alan Stern]

On Fri, 7 Dec 2012, Ming Lei wrote:

> > I guess the question is whether the callee is allowed to call
> > device_move(), if not, we're good.
> 
> Not only the callee, and other contexts can change device->parent
> too. Looks rfcomm_tty_open() which calls device_move() is called
> in open() context, so the parent might be changed before unlock(dev->parent)
> in __driver_attach().
> 
> >
> >> Your concern on device_remove() might be correct. Also, I am wondering
> >> if we can walk the 'dpm_list' backwards for device shutdown, which should
> >> be simpler and more reasonable.
> >
> > How would that help?
> 
> device_pm_lock() can prevent device_move() from being running.

That wouldn't prevent problems during unbinding.  Wedson is right; the 
places that lock dev->parent must save a local copy of dev->parent.

Alan Stern

--
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/