Hi Jens,

We reviewed your more recent patches. We will give it a try.

Thanks

-thomas

From: Jens Wiklander [mailto:jens.wiklander@linaro.org]
Sent: Tuesday, January 23, 2018 12:55 AM
To: Thomas M Zeng <tzeng@codeaurora.org>
Cc: tee-dev@lists.linaro.org; Zhen Kong <zkong@quicinc.com>
Subject: Re: (Gen Tee Driver API in the Linux Kernel) Reopen the Topic of TEE Kernel Client API

Hi Thomas,

On Tue, Jan 23, 2018 at 3:27 AM, Thomas M Zeng <tzeng@codeaurora.org> wrote:

Hello TEE POCs:

We came across Jens’ 2015 patch:

https://github.com/linaro-swg/linux/commit/b8d05a0934b2

What is the status of the kernel client API below? Is it slated to go upstream in the 4.16 or 4.17 kernel?

I posted that patch on the kernel mailing list some time ago (https://lkml.org/lkml/2017/7/17/520), but since I didn't get any feedback I didn't take it any further.

I ask because we have use cases that, if ported to the Generic TEE interface, may require tee_shm_alloc() and friends to be available for use cases such as the Linux soc-specific kernel entropy manager requesting TEE side to provide PRNG data. The above patch would be a good starting point.

For instance, we were looking at using the following two APIs from Jens’ patch to open and close context objects. They will enable the clients to use “tee_shm_alloc”.

I haven't tested or used this myself, so it's basically only a proposal from my side so far. Does the patch work well for you, or does it need some changes?

This patch can make it into the 4.17 kernel earliest. Before that can happen I will need at least some proof that it's used inside the kernel. A tested by tag and a sentence with some description of the test would probably be good enough.

Thanks,

Jens