Re: [PATCH 0/3] [RESEND v4] VME Framework Fixes

["Emilio G. Cota" <cota@braap.org>]

On Wed, Aug 31, 2011 at 12:05:43 +0200, Manohar Vanga wrote:
> Hi Greg, Martyn,
> 
> I have reordered the patches to separate the bridge refcounting patch
> for now so that we can make some progress on getting the other patches
> into the tree. I will rework the refcounting patch and send it later.

(snip to 3rd patch comment)

> * staging: vme: make match() driver specific to improve non-VME64x
>     support
> 	Changes based on the moving of the bridge refcounting patch as well
> 	as changes based on comments from Emilio and Greg in the last resend.

I'm confused. AFAICT you should fix things here; as per Greg's post:

On Mon, Aug 29, 2011 at 11:54:03 -0700, Greg KH wrote:
> On Mon, Aug 29, 2011 at 02:40:12PM -0400, Emilio G. Cota wrote:
> > AFAIK currently a vme bridge driver can be freely removed with rmmod
> > even if it has devices under it. The patch avoids this by increasing
> > the refcount of the bridge module every time a device is registered
> > under it.
> >
> > What's the appropriate thing to do when rmmod tries to remove a bridge
> > module?
>
> Let it happen and remove all of the devices under it.
>
> That's the way all other bus drivers in the kernel work, why change this
> type of behavior?

When a bridge is removed, just remove the devices under it. Refcounts
in this case are not necessary; in fact refcounts where used to avoid
a bridge being removed while it had devices under it, which it turns
out was a bad idea.

		Emilio