Re: [PATCH 7/8] rust: add revocable objects

[Danilo Krummrich <dakr@redhat.com>]

On Tue, Mar 26, 2024 at 07:16:51PM +0100, Greg KH wrote:
> On Tue, Mar 26, 2024 at 06:07:36PM +0100, Danilo Krummrich wrote:
> > On Mon, Mar 25, 2024 at 07:21:32PM +0100, Greg KH wrote:
> > > On Mon, Mar 25, 2024 at 06:49:11PM +0100, Danilo Krummrich wrote:
> > > > From: Wedson Almeida Filho <wedsonaf@gmail.com>
> > > > 
> > > > This implements the Revocable and AsyncRevocable types.
> > > > 
> > > > Revocable allows access to objects to be safely revoked at run time.
> > > > 
> > > > This is useful, for example, for resources allocated during device probe;
> > > > when the device is removed, the driver should stop accessing the device
> > > > resources even if other state is kept in memory due to existing
> > > > references (i.e., device context data is ref-counted and has a non-zero
> > > > refcount after removal of the device).
> > > 
> > > Again, device removal is different from unbinding a driver from a
> > > device.  Please fix this all up.
> > 
> > I think this isn't an "again", "[PATCH 8/8] rust: add device::Data" actually
> > phrased things differently (which indeed isn't great and needs to be fixed).
> > 
> > This comment indeed implies that device removal actually means unbinding a
> > driver from a device. I think "device remove" is meant here as the drivers
> > remove() callback is called.
> > 
> > Your comment seems to imply you have a different understanding of what "device
> > remove" means?
> 
> Yes, see my other comments.  A device has a lifespan that exceeds that
> of a driver, and "remove" does not mean it is gone from the system just
> because the remove() callback of a single driver is called.  It still
> can, and does, live outside of a driver quite well, and in fact, the
> driver is not in control of its lifespan at all!

Yes, that's clear. But what does "device remove" mean for you then?

I think this one here mostly seems to be about terminology and I want to make
sure I don't create any other misunderstandings e.g. by using "device remove"
for unbinding a driver from a device.

> 
> All a driver can know is that it has a valid reference to a device for
> the amount of time from probe() to remove().  Outside of that a driver
> can not touch it, BUT the device structure lives on and can do lots of
> other things that do not involve the driver at all.
> 
> thanks,
> 
> greg k-h
>