Re: [PATCH v1 00/17] tee: Use bus callbacks instead of driver callbacks

[Sumit Garg <sumit.garg@kernel.org>]

On Tue, Dec 16, 2025 at 12:08:38PM +0100, Uwe Kleine-König wrote:
> Hello,
> 
> On Tue, Dec 16, 2025 at 01:08:38PM +0530, Sumit Garg wrote:
> > On Mon, Dec 15, 2025 at 3:02 PM Uwe Kleine-König
> > <u.kleine-koenig@baylibre.com> wrote:
> > > On Mon, Dec 15, 2025 at 04:54:11PM +0900, Sumit Garg wrote:
> > > > Feel free to make the tee_bus_type private as the last patch in the series
> > > > such that any followup driver follows this clean approach.
> > >
> > > There is a bit more to do for that than I'm willing to invest. With my
> > > patch series applied `tee_bus_type` is still used in
> > > drivers/tee/optee/device.c and drivers/tee/tee_core.c.
> > 
> > Oh I see, I guess we need to come with some helpers around device
> > register/unregister from TEE subsystem as well. Let's plan that for a
> > followup patch-set, I don't want this patch-set to be bloated more.
> 
> Don't consider me in for that. But it sounds like a nice addition.
>

No worries, the current cleanup is fine for now.

> > > Maybe it's
> > > sensible to merge these two files into a single one.
> > 
> > It's not possible as the design for TEE bus is to have TEE
> > implementation drivers like OP-TEE, AMD-TEE, TS-TEE, QTEE and so on to
> > register devices on the bus.
> 
> So only OP-TEE uses the bus for devices and the other *-TEE don't. Also
> sounds like something worth to be fixed.

The TEE bus is common for all the TEE implementation drivers which need
to support kernel TEE client drivers. I am aware there will be QTEE and
TS-TEE from past discussion that they will be exposing devices on the
TEE bus.

> 
> > > The things I wonder about additionally are:
> > >
> > >  - if CONFIG_OPTEE=n and CONFIG_TEE=y|m the tee bus is only used for
> > >    drivers but not devices.
> > 
> > Yeah since the devices are rather added by the TEE implementation driver.
> > 
> > >
> > >  - optee_register_device() calls device_create_file() on
> > >    &optee_device->dev after device_register(&optee_device->dev).
> > >    (Attention half-knowledge!) I think device_create_file() should not
> > >    be called on an already registered device (or you have to send a
> > >    uevent afterwards). This should probably use type attribute groups.
> > >    (Or the need_supplicant attribute should be dropped as it isn't very
> > >    useful. This would maybe be considered an ABI change however.)
> > 
> > The reasoning for this attribute should be explained by commit:
> > 7269cba53d90 ("tee: optee: Fix supplicant based device enumeration").
> > In summary it's due to a weird dependency for devices we have with the
> > user-space daemon: tee-supplicant.
> 
> From reading that once I don't understand it. (But no need to explain
> :-)
> 
> Still the file should better be added before device_add() is called.

Noted, let me see if I can get to fix this until someone jumps in before
me.

> 
> > >  - Why does optee_probe() in drivers/tee/optee/smc_abi.c unregister all
> > >    optee devices in its error path (optee_unregister_devices())?
> > 
> > This is mostly to take care of if any device got registered before the
> > failure occured. Let me know if you have a better way to address that.
> 
> Without understanding the tee stuff, I'd say: Don't bother and only undo
> the things that probe did before the failure.
> 

True, but this is special case where if there is any leftover device
registered from the TEE implementation then it is likely going to cause
the corresponding kernel client driver crash.

-Sumit