On 2/26/26 13:38, Quentin Schulz wrote:
Hi Heinrich, Igor,
On 2/25/26 7:31 AM, Heinrich Schuchardt wrote:
Symbol CONFIG_PHY_MTK_UFS does not exist.
Don't select it.
Signed-off-by: Heinrich Schuchardt <[email protected]>
---
CONFIG_PHY_MTK_TPHY seems to be the only Mediatek PHY driver symbol
in U-Boot.
This binds against mediatek,mt6878-ufshci compatible property. We have no DT in
U-Boot with that compatible. I couldn't find any mention of this compatible or
this SoC's support in the kernel mailing list(s) either...
The PHY symbol was actually defined in patch 1 of the series, see
https://lore.kernel.org/u-boot/20251011-mtk-ufs-uboot-v1-0-a05f991ee...@mentallysanemainliners.org/
Considering the kernel binding for the Mediatek UFS controller requires PHYs,
I'm assuming the series shouldn't have been partially merged, especially the
controller part without the PHY. (Note that the kernel binding still doesn't
list MT6878 as supported).
@Igor, maybe you can work on a v2 for the first patch in your series so we
don't need to revert the UFS controller part since it's unusable today as is
(first because no DT with this compatible is found, and that even if it was,
there's no PHY driver while it is required for the controller to work). If the
PHY is not required by the code, then it doesn't follow the Mediatek UFS
upstream binding (providing mediatek,ufs.yaml is where MT6878 support would be)
and shouldn't have been accepted as is. We do not want to drift apart from the
kernel bindings any more than we already are.
I should've probably merged the PHY driver at the same time.
Neil
Cheers,
Quentin
