Re: [PATCH v2] tee: shm: fix slab page refcounting

[Sumit Garg]

On Fri, Feb 13, 2026 at 11:04:48PM +0100, Marco Felsch wrote:
> Hi Sumit,
> 
> On 26-02-13, Sumit Garg wrote:
> > Hi Marco,
> > 
> > On Thu, Feb 12, 2026 at 01:58:30PM +0100, Marco Felsch wrote:
> > > Hi Sumit,
> > > 
> > > TBH: I was hoping that you will take care of this since you're marked as
> > > maintainer for the tee-trusted-key and we noticed the warning with 6.14
> > > and still no fix available :/
> > 
> > Mathew did suggested a fix long back on which everybody agreed but
> 
> You agreed. I said that the current TEE API also allows non-slabed based
> backed memory and therefore I don't wanted to send this patch approach
> and instead asked you to do so since you're the maintainer and fine with
> the change.
> 
> > didn't got enough attention from you to test and report if that fixed
> 
> Why should it get attention from us? Maybe we do have different views of
> being a maintainer.

It's really the basic expectation I have put here which every reporter
of a bug needs to say if a suggested fix works for them or not.

> 
> > your issue. Since you insisted further, I have created a formal fix
> 
> Why is it our issue? It's everyones issue which uses the tee trusted-key
> driver.
> 
> > patch based on that here [1]. Care to test that?
> 
> A colleague of mine is going to test it and will reply on the patch.
> 
> > [1] https://lore.kernel.org/all/20260213113317.1728769-1-sumit.garg@xxxxxxxxxx/
> 
> ...
> 
> > > I checked the code once again and figured that we could drop/replace
> > > tee_shm_register_kernel_buf() with tee_shm_alloc_kernel_buf(). I don't
> > > see why a kernel driver needs to tee_shm_register_kernel_buf() in the
> > > first place, maybe this is legacy. The only users of
> > > tee_shm_register_kernel_buf() are trusted_tee.c and tee_stmm_efi.c.
> > 
> > No it's not legacy but allows for efficient memory reuse within the
> > kernel as to not create bounce buffers to share data with TEE.
> 
> To be hones, there are only two driver using the API. The tee_stmm_efi
> driver can do the alloc during the probe(). The trusted_tee has to use a
> bounce buffer, yes but how often do you assume that (un)sealing and rng
> ops have to be done during runtime? This shouldn't be a overhead at all.
> 
> Therefore my suggestion would be still to drop the internal kernel API
> and only use it for userspace pages, where it could really matter.

I don't disagree with what you are saying here but we really need to
promote efficient memory reuse for TEE clients. There will surely be
more use-cases coming in future which can benefit from the flexibility
to register buffer. One another kernel client being remoteproc subsystem
which is already under review for this API.

-Sumit

> 
> Regards,
>   Marco
> -- 
> #gernperDu 
> #CallMeByMyFirstName
> 
> Pengutronix e.K.                           |                             |
> Steuerwalder Str. 21                       | https://www.pengutronix.de/ |
> 31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
> Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-9    |