On 09/01/2026 09:38, Peter Wang (王信友) wrote:

On Fri, 2026-01-09 at 08:24 +0100, Krzysztof Kozlowski wrote:

quoted

On 09/01/2026 07:22, Peter Wang (王信友) wrote:

quoted

Is it sufficient for us to supplement the ABI document?
This ABI might affect the ability to reset and recover after 
an UFS error in upstream world.

In normal case yes, but I cannot imagine arguments justifying your
usage
of TI properties. Basically it would not pass review.

Best regards,
Krzysztof

Yes, this part is indeed because MediaTek’s reset hardware 
implementation is the same as TI’s. That’s why we used “compatible” 
instead of actually implementing MediaTek’s own reset controller.

So that's another purely downstream code. Additionally very poor quality
downstream code.

So, are you suggesting that we upstream a MediaTek reset controller,
even though the code is almost identical to TI’s?

If you ask about DT, this is already answered in writing bindings
document. You cannot use someone else's compatible. Was also re-iterated
on mailing list bazillions of times.

Best regards,
Krzysztof