Received-SPF: pass (google.com: domain of linux-kernel+bounces-84868-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240223095133.109046-1-balint.dobszay@arm.com>
 <20240223095133.109046-2-balint.dobszay@arm.com> <CAFA6WYNW9-7gCZQSEaV=Gcr+GLdu25rQ8MpTg9yNpX7OwyZ0Tg@mail.gmail.com>
 <CAHUa44G70L9CnQD6V3ivWf1hYfK0iWFt-cfg98-JGB-s8TVd-A@mail.gmail.com>
 <CAFA6WYOq3tHq5n7=XYpsSgBghCF0RtAxs-sLxvLYMWwnQuFSeQ@mail.gmail.com> <CAHUa44F=0juXyyhA7Bu0SzCK+1igRYEWCn1XRL1sDG1ZWNhr3g@mail.gmail.com>
In-Reply-To: <CAHUa44F=0juXyyhA7Bu0SzCK+1igRYEWCn1XRL1sDG1ZWNhr3g@mail.gmail.com>
From: Sumit Garg <sumit.garg@linaro.org>
Date: Wed, 28 Feb 2024 16:50:02 +0530
Message-ID: <CAFA6WYOWQZqtAcdCKcBa7qnCY1ueBzU1_1HSbfHPaLeyPdtxbw@mail.gmail.com>
Subject: Re: [PATCH v2 1/3] tee: optee: Move pool_op helper functions
To: Jens Wiklander <jens.wiklander@linaro.org>
Cc: Balint Dobszay <balint.dobszay@arm.com>, op-tee@lists.trustedfirmware.org, 
	linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org, 
	linux-arm-kernel@lists.infradead.org, corbet@lwn.net, sudeep.holla@arm.com, 
	rdunlap@infradead.org, krzk@kernel.org, gyorgy.szing@arm.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Wed, 28 Feb 2024 at 14:11, Jens Wiklander <jens.wiklander@linaro.org> wr=
ote:
>
> On Wed, Feb 28, 2024 at 6:58=E2=80=AFAM Sumit Garg <sumit.garg@linaro.org=
> wrote:
> >
> > On Tue, 27 Feb 2024 at 21:20, Jens Wiklander <jens.wiklander@linaro.org=
> wrote:
> > >
> > > On Tue, Feb 27, 2024 at 7:06=E2=80=AFAM Sumit Garg <sumit.garg@linaro=
org> wrote:
> [snip]
> > > > > --- a/include/linux/tee_drv.h
> > > > > +++ b/include/linux/tee_drv.h
> > > > > @@ -275,6 +275,17 @@ void *tee_get_drvdata(struct tee_device *tee=
dev);
> > > > >  struct tee_shm *tee_shm_alloc_priv_buf(struct tee_context *ctx, =
size_t size);
> > > > >  struct tee_shm *tee_shm_alloc_kernel_buf(struct tee_context *ctx=
, size_t size);
> > > > >
> > > > > +int tee_shm_pool_op_alloc_helper(struct tee_shm_pool *pool, stru=
ct tee_shm *shm,
> > > > > +                                size_t size, size_t align,
> > > > > +                                int (*shm_register)(struct tee_c=
ontext *ctx,
> > > > > +                                                    struct tee_s=
hm *shm,
> > > > > +                                                    struct page =
**pages,
> > > > > +                                                    size_t num_p=
ages,
> > > > > +                                                    unsigned lon=
g start));
> > > > > +void tee_shm_pool_op_free_helper(struct tee_shm_pool *pool, stru=
ct tee_shm *shm,
> > > > > +                                int (*shm_unregister)(struct tee=
_context *ctx,
> > > > > +                                                      struct tee=
_shm *shm));
> > > > > +
> > > >
> > > > These rather belong to drivers/tee/tee_private.h as we shouldn't
> > > > expose them to other kernel client drivers.
> > >
> > > This is the right place, this .h file is for TEE drivers too.
> > >
> >
> > But this is shared with other kernel TEE client drivers and we
> > shouldn't expose internal APIs which aren't meant for them with a side
> > effect of API abuse too. Any particular reason to not use
> > drivers/tee/tee_private.h?
>
> drivers/tee/tee_private.h is supposed to be used internally by only
> the files in drivers/tee.

How about "struct tee_device" being in drivers/tee/tee_private.h?

> If you look in include/linux/tee_drv.h
> you'll find a few functions and other definitions that a kernel TEE
> client driver should not use, for instance, tee_device_register() and
> tee_device_unregister(). This kernel TEE client interface was
> introduced with commit 25559c22cef8 ("tee: add kernel internal client
> interface"). include/linux/tee_drv.h existed before we even had any
> kernel TEE client interface.

Anyhow, it looks like there is a chance for refactoring here. How
about splitting this header further in something like
include/linux/tee_core.h which will contain all the pieces relevant to
TEE drivers?

BTW, this patch series can keep using include/linux/tee_drv.h for the
time being.

-Sumit

>
> Cheers,
> Jens