Message-ID: <db00f6a9-263d-9c47-486e-7080ffc5b3c9@linaro.org>
Date:   Fri, 2 Sep 2022 12:56:58 +0530
From:   Sumit Garg <sumit.garg@...aro.org>
To:     Maximilian Luz <luzmaximilian@...il.com>
Cc:     Konrad Dybcio <konrad.dybcio@...ainline.org>,
        Rob Herring <robh+dt@...nel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Steev Klimaszewski <steev@...i.org>,
        Shawn Guo <shawn.guo@...aro.org>,
        Sudeep Holla <sudeep.holla@....com>,
        Cristian Marussi <cristian.marussi@....com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        linux-arm-msm@...r.kernel.org, linux-efi@...r.kernel.org,
        devicetree@...r.kernel.org, linux-kernel@...r.kernel.org,
        Srinivas Kandagatla <srinivas.kandagatla@...aro.org>,
        Andy Gross <agross@...nel.org>,
        Bjorn Andersson <bjorn.andersson@...aro.org>,
        Ard Biesheuvel <ardb@...nel.org>
Subject: Re: [PATCH 0/4] firmware: Add support for Qualcomm UEFI Secure
 Application

Hi Maximilian,

On 02/08/22 18:52, Maximilian Luz wrote:
>
>
> On 8/2/22 13:51, Srinivas Kandagatla wrote:
>> Hi Maximilian,
>>
>> On 23/07/2022 23:49, Maximilian Luz wrote:
>>> On modern Qualcomm platforms, access to EFI variables is restricted to
>>> the secure world / TrustZone, i.e. the Trusted Execution Environment
>>> (TrEE or TEE) as Qualcomm seems to call it. To access EFI variables, we
>>> therefore need to talk to the UEFI Secure Application (uefisecapp),
>>> residing in the TrEE.
>>>
>>> This series adds support for accessing EFI variables on those 
>>> platforms.
>>>
>>> To do this, we first need to add some SCM call functions used to manage
>>> and talk to Secure Applications. A very small subset of this interface
>>> is added in the second patch (whereas the first one exports the 
>>> required
>>> functions for that). Interface specifications are extracted from [1].
>>> While this does not (yet) support re-entrant SCM calls (including
>>> callbacks and listeners), this is enough to talk to the aforementioned
>>> uefisecapp on a couple of platforms (I've tested this on a Surface 
>>> Pro X
>>> and heard reports from Lenovo Flex 5G, Lenovo Thinkpad x13s, and Lenovo
>>> Yoga C630 devices).
>>>
>>> The third patch adds a client driver for uefisecapp, installing the
>>> respective efivar operations. The application interface has been 
>>> reverse
>>> engineered from the Windows QcTrEE8180.sys driver.
>>>
>>> Apart from uefisecapp, there are more Secure Applications running that
>>> we might want to support in the future. For example, on the Surface Pro
>>> X (sc8180x-based), the TPM is also managed via one.
>>>
>>> I'm not sure whether this should go to drivers/firmware or to
>>> drivers/soc/qcom. I've put this into firmware as all of this is
>>> essentially an interface to the secure firmware running in the 
>>> TrustZone
>>> (and SCM stuff is handled here already), but please let me know if I
>>> should move this.
>>
>>  From what I see so far is that this is adapted from downstream 
>> qseecom driver, this approach could work for a limited usecases but 
>> not scalable, as we cannot add drivers for each Qualcomm specific TA 
>> in kernel.
>> This has to be handled in much generic way using Linux TEE framework, 
>> and let the userspace side deal with TA specific bits.
>
> I generally agree with the sentiment, however UEFI variables should 
> IMHO be
> handled by the kernel. Moving handling of those to userspace breaks 
> things like
> EFI-based pstore and efivarfs. The latter will in turn break some 
> user-space
> tools (most notably efibootmgr used by e.g. GRUB and I think fwupdmgr 
> which
> needs to set some capsule variables). Ideally, we would find a way to 
> not break
> these, i.e. have them work out-of-the-box.
>
> A similar argumentation might apply to the TPM app.

See below, there is already an existing TPM app driver [2] in kernel 
although the app is based on OP-TEE.

>
>> AFAIU, Qualcomm is moving away from qseecom interface to new 
>> smc-invoke interface, most of Qualcomm SoCs starting from SDM660 
>> already have support to this.
>>
>> This interface provides a better abstracted IPC mechanism to talk to 
>> TA. Most of these TA specific interfaces are packed in closed 
>> userspace source.
>> Having said that QTEE smcinvoke driver can be modeled as a proper TEE 
>> driver with Userspace driving the TA specific bits using existing tee 
>> uapis.
>> This also brings in other features like loading, Listeners aka 
>> callbacks, secure memory allocations..etc.
>>
>> In the past, I have tried to do a prototype of this smcinvoke driver 
>> as a proper tee driver, incase you are interested patches are at 
>> https://git.linaro.org/landing-teams/working/qualcomm/kernel.git/log/?h=tracking-qcomlt-qcomtee
>> Plan is to discuss with Qualcomm and send it for upstream review.
>
> Thanks for this information! So as far as I understand it, this is 
> currently an
> interface to user-space only, i.e. does not allow in-kernel drivers 
> for apps?

The Linux TEE framework already provides an in-kernel interface to TEE 
as well via TEE bus [1]. There are already multiple kernel drivers [2] 
[3] [4] [5] [6] [7] using it. So an EFI driver can be an addition to that.

Now coming on to TEE implementations, the drivers I mentioned are based 
on OP-TEE where devices are queried/enumerated during OP-TEE probe here 
[8]. So in similar manner QTEE smcinvoke driver should be able to 
register devices on the TEE bus.

[1] 
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/staging/tee.rst#n56

[2] drivers/char/tpm/tpm_ftpm_tee.c

[3] drivers/char/hw_random/optee-rng.c

[4] drivers/firmware/arm_scmi/optee.c

[5] security/keys/trusted-keys/trusted_tee.c

[6] drivers/firmware/broadcom/tee_bnxt_fw.c

[7] drivers/rtc/rtc-optee.c

[8] drivers/tee/optee/device.c

-Sumit

PS. TBH, I haven't looked into detail workings for the QTEE smcinvoke 
driver.

> It would be great if this could then be extended to handle (the bare 
> minimum
> of) in-kernel drivers (i.e. only things that the kernel itself needs, 
> like EFI
> variables). Alternatively, I'm happy to hear suggestions on how we not 
> break
> the aforementioned things while moving handling off to userspace.
>
>> I think its worth exploring if uefisecapp can talk smcinvoke.
>> I can ping Qualcomm engineers to see if that is doable.
>
> I think that would be great! Thanks!
>
> Regards,
> Max
>

Powered by blists - more mailing lists