Re: [PATCH] Fix ide unregister vs. driver model

[Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>]

On Monday 25 of August 2003 08:34, Benjamin Herrenschmidt wrote:
> On Mon, 2003-08-25 at 00:23, Bartlomiej Zolnierkiewicz wrote:
> > On Sunday 24 of August 2003 15:05, Benjamin Herrenschmidt wrote:
> > > Hi Bart !
> >
> > Hi,
> >
> > > This patch seem to have been lost, so here it is again. It fixes
> > > an Ooops on unregistering hwifs due to the device model now having
> > > mandatory release() functions. It also close the possible race we
> > > had on release if the entry was in use (by or /sys typically) by
> > > using a semaphore waiting for the release() to be called after
> > > doing an unregister.
> >
> > I can't see the race - all references to struct device should be dropped
> > by driver model before finally calling ->release() function...
>
> We have no race with the patch, that is we have no race when we wait
> for the semaphore after calling unregister(). We have a race if we
> don't as unregister() will drop a reference, but we may have pending
> ones from sysfs still... so if we don't wait for release() to be
> called, we may overwrite a struct device currently beeing used by
> sysfs.

Nope, I don't think struct device can be used by sysfs after execution
of device_unregister() (I've checked driver model and sysfs code).

--bartlomiej