On Thu, Nov 10, 2022 at 01:40:32PM -0700, Simon Glass wrote:
Hi Patrick,
On Thu, 10 Nov 2022 at 03:49, Patrick Delaunay patrick.delaunay@foss.st.com wrote:
Update in U-Boot env mmc backend with several cosmetic changes or corrections and 2 new features:
1/ CONFIG_ENV_MMC_USE_DT = no more use CONFIG_ENV_OFFSET in the mmc ENV backend when this config is activated.
Requested by the STM32MP STMicroelectronics boards which activate several ENV_IS_IN_XXX; the value of CONFIG_ENV_OFFSET is invalid for SD-Card / eMMC boot; this offset should only used in SPIFlash backend (sf.c) for SPI-NOR boot.
If this offset is used on mmc backend, when partition name in GPT is not aligned with U-Boot DT: "u-boot,mmc-env-partition", the behavior is difficult to debug: a partition is corrupted on 'env save' command.
2/ selects the GPT env partition by the "u-boot-env" type GUID introduced by the commit c0364ce1c695 ("doc/README.gpt: define partition type GUID for U-Boot environment")
This feature can also avoid issue when 'u-boot-env' partition name change in GPT partitioning but not in the U-Boot DT with "u-boot,mmc-env-partition"
Few check patch warnings remained in the series, but after check I can't remove them :
IS_ENABLED(ENV_MMC_HWPART_REDUND) is normally used as IS_ENABLED(CONFIG_ENV_MMC_HWPART_REDUND) => ENV_MMC_HWPART_REDUND is locally defined in this file it is not a real CONFIG but I can use the IS_ENABLED() macro as it is defined to 1
Use 'if (IS_ENABLED(CONFIG...))' instead of '#if or #ifdef' where possible
- CONFIG_PARTITION_TYPE_GUID => info.type_guid existence
- CONFIG_ENV_OFFSET_REDUND and CONFIG_ENV_MMC_USE_DT => only for define
As I miss the merge window, not targeted for v2023.01 but for next v2023.04.
Shouldn't this all move to device tree? Using CONFIG options is such a mess. We have the devices in DT so can indicate which ones have an environment and what the parameters are for each.
And there's already the Documentation/devicetree/bindings/nvmem/u-boot,env.yaml to build upon.