On Mon, Dec 20, 2021 at 12:54 PM Krzysztof Kozlowski
<krzysztof.kozlowski@xxxxxxxxxxxxx> wrote:
> ----------------------------------------------------------------
> Samsung SoC drivers changes for v5.17
>
> 1. Exynos ChipID: add Exynos7885 support.
> 2. Exynos PMU: add Exynos850 support.
> 3. Minor bindings cleanup.
> 4. Add Exynos USIv2 (Universal Serial Interface) driver. The USI block is
> a shared IP block between I2C, UART/serial and SPI. Basically one has
> to choose which feature the USI block will support and later the
> regular I2C/serial/SPI driver will bind and work.
> This merges also one commit with dt-binding headers from my dts64
> pull request.
>
> Together with a future serial driver change, this will break the ABI.
>
> Affected: Serial on ExynosAutov9 SADK and out-of-tree ExynosAutov9 boards
>
> Why: To properly and efficiently support the USI with new hierarchy
> of USI-{serial,SPI,I2C} devicetree nodes.
>
> Rationale:
> Recently added serial and USI support was short-sighted and did not
> allow to smooth support of other features (SPI and I2C). Adding
> support for USI-SPI and USI-I2C would effect in code duplication.
> Adding support for different USI versions (currently supported is
> USIv2 but support for v1 is planned) would cause even more code
> duplication and create a solution difficult to maintain.
> Since USI-serial and ExynosAutov9 have been added recently, are
> considered fresh development features and there are no supported
> products using them, the code/solution is being refactored in
> non-backwards compatible way. The compatibility is not broken yet.
> It will be when serial driver changes are accepted.
> The ABI break was discussed with only known users of ExynosAutov9 and
> received their permission.
Thanks a lot for the detailed description, very helpful!
I've applied pull requests 1 through 4, though it seems that once more
the automated emails did not go out.
I can't find the two defconfig patches you mentioned in the introductory
mail, neither in patchwork nor in my inbox, I assume these were
numbered 5/6 and 6/6?
Arnd
