Define SDP memory area in embedded DTB

[omasse-linaro]

As an improvement, could it be possible to define the Secure Data Path memory region in an embedded dtb instead of the scattered array ?
Probably something like that:

	sdp_mem {
		compatible = "optee_sdp";
		region = <0x8C000000 0x02000000>;
	}

This memory region would not rely on a core_mmu_phys_mem structure definition, but a dtb one.

[jenswi-linaro]

Adding stuff to DT takes a bit of consideration. We don't want to define our own standard in secure world.
Is there anything already defined that can be used for this purpose?

[omasse-linaro]

I do not thing that something has already been defined for SDP in DT. But what about reserved memory which could be used to define the SDP memory ?

reserved-memory {
	#address-cells = <1>;
	#size-cells = <1>;

	sdp_mem {
		compatible = "optee-sdp";
		reg = <0xCC000000 0x02000000>;
	};
};

[jforissier]

@omasse-linaro yes it is done in linaro-swg/linux branch optee, see for instance commit linaro-swg/linux@2a766ee (plus the no-map property is added in a subsequent commit).

Edit: sorry what I mentioned is the reserved shared memory not SDP :-/ but yeah could be similar.

Edit 2
Given that the ION/dmabuf code on which SDP was based has been modified/removed upstream (see discussion here), I'm not sure discussing DT for SDP even makes sense until a proper implementation supported upstream is proposed.

[omasse-linaro]

Edit 2 Given that the ION/dmabuf code on which SDP was based has been modified/removed upstream (see discussion here), I'm not sure discussing DT for SDP even makes sense until a proper implementation supported upstream is proposed.

@jforissier you are right, optee-test should now rely on DMA buff heaps to test SDP.
@jbech-linaro confirm us that @ruchi393 has this on the roadmap.

in the meantime, optee-os could use DT to define SDP and will be tested with next version of optee-test.
(cause SDP is still a used feature in optee)

[jbech-linaro]

@jbech-linaro confirm us that @ruchi393 has this on the roadmap.

Yes, fixing SDP is on @ruchi393 roadmap, but it has lower priority for the moment IIRC.

[omasse-linaro]

Please find a first draft of the modifications here #5149

Tested on imx board. Unfortunately I do not have an hickey board.

[jenswi-linaro]

So far we have managed to avoid defining our own bindings in OP-TEE, instead we've been able to reuse already established bindings. With this you're proposing something new. I'm not sure of the best way of doing such a thing. Are we sure there is nothing to reuse?
If not: How should it be reviewed? Who should review it?

[github-actions]

This issue has been marked as a stale issue because it has been open (more than) 30 days with no activity. Remove the stale label or add a comment, otherwise this issue will automatically be closed in 5 days. Note, that you can always re-open a closed issue at any time.

[omasse-linaro]

#5177 tested on hikey.

[github-actions]

This issue has been marked as a stale issue because it has been open (more than) 30 days with no activity. Remove the stale label or add a comment, otherwise this issue will automatically be closed in 5 days. Note, that you can always re-open a closed issue at any time.