Re: [PATCH v3] arm64: defconfig: Enable QCOMTEE on Qualcomm SM8650+

[Sumit Garg]

On Wed, Dec 10, 2025 at 02:35:19PM +0530, Harshal Dev wrote:
> Hi Bjorn,
> 
> On 12/10/2025 9:21 AM, Bjorn Andersson wrote:
> > On Mon, Dec 8, 2025 at 10:14 PM Dmitry Baryshkov
> > <dmitry.baryshkov@xxxxxxxxxxxxxxxx> wrote:
> >> On Tue, 9 Dec 2025 at 07:58, Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote:
> >>>
> >>> On 08/12/2025 13:18, Harshal Dev wrote:
> >>>> Enable QCOMTEE driver on Qualcomm SM8650+ SoCs to facilitate communication
> >>>> with the Qualcomm Trusted Execution Environment (QTEE).
> >>>> (No enablement required in DTS files since QCOMTEE device is dynamically
> >>>> registered by the QCOM_SCM firmware driver)
> >>>>
> >>>> Signed-off-by: Harshal Dev <harshal.dev@xxxxxxxxxxxxxxxx>
> >>>> ---
> >>>> Changes in v3:
> >>>> - Updated the commit message to reflect the supported Qualcomm platforms.
> >>>> - Link to v2: https://lore.kernel.org/r/20251205-qcom_qcomtee_defconfig-v2-1-c92560b0346e@xxxxxxxxxxxxxxxx
> >>>>
> >>>
> >>> I gave you the exact example to follow. Maybe it is not that important
> >>> for others, so I will not object, but OTOH it is important for me, thus
> >>> I will not give reviewed by. I damn asked VERY CLEARLY:
> >>>
> >>> "Just mention which UPSTREAM boards (which you called Qualcomm
> >>> platforms) use this driver."
> >>
> >> +1 Here. Defconfig changes mention devices, not SoC families.
> >>
> > 
> > I don't agree that you have to mention a specific board, if the
> > feature is used by all boards. But I think the commit message should
> > make _that_ clear.
> > 
> 
> Thanks for this input Bjorn. I gather we are now aligned that the board
> information is not required.
> 
> Then the other part to this is how to provide information on the particular
> SoCs using this.
> 
> > On the contrary, the commit message says that we're enabling
> > CONFIG_QCOMTEE because it's used on "SM8560+". What does the plus
> > mean?
> 
> I took reference from similar commits merged earlier where the plus seemed
> to indicate that all Qualcomm SoCs from 'SM8650' on-wards, that is, SM8750,
> SM8850 and so on. It felt that the plus sign is self-explanatory since it
> has been used already. But sure, maybe we can be explicit from now on and say
> from 'Qualcomm SM8650 onwards'.
> 
> commit c5d02bbaa217b2454ba1ce7528113aa2ecf14f3c
> 
> > Also, the driver isn't enabled "on Qualcomm SM8650+", it's enabled in
> > the Am64 defconfig, i.e. it's enabled on all Arm64 boards - the
> > question that should be answered by the commit message is "why?".
> > 
> 
> Even though we are enabling this via the arm64 defconfig, it is not true that
> the driver is applicable for all arm64 boards. The simple reason being that
> the QTEE firmware OS that the driver communicates with runs only on Qualcomm
> SoCs using arm64 CPUs with ARM TrustZone technology.
> 
> This is why I would try to avoid a commit message which claims the the driver
> is applicable to all arm64 boards.
> 
> Based on all this, I am thinking perhaps it would be better to say that the
> patch enables QCOMTEE driver for Qualcomm SoCs with arm64 CPUs? We could drop
> mentions of specific SM8x50 models with HDK/MTP boards since the feature is
> agnostic to those?

AFAIK, the QCOMTEE driver works on the Qcom SoCs based on arm64 which supports
the SMCInvoke protocol. So we should be explicit about it. Regarding
mention of reference publicly available boards, I can see how it can be useful
for the community to test QTEE based apps. If you can mention say
example RB3Gen2 supports QTEE driver at least then it will be helpful.

-Sumit

[...]