Hi Tze Yee,

On 28/4/2026 10:53 am, [email protected] wrote:
From: tzeyeeng <[email protected]>

The eMMC device tree inherits the configuration from
socfpga_agilex5_socdk.dts.

Changes:
- Set SDHCI Capabilities bit18 to enable 8-bit embedded device support.

Signed-off-by: tzeyeeng <[email protected]>

---
Changes in v2:
- Remove /delete-property/ sd-uhs-sdr50 and /delete-property/
   sd-uhs-sdr104 from mmc node
- Fix wording in commit message
- Update commit title
---
  arch/arm/dts/socfpga_agilex5_socdk_emmc.dts | 5 ++---
  1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/arch/arm/dts/socfpga_agilex5_socdk_emmc.dts 
b/arch/arm/dts/socfpga_agilex5_socdk_emmc.dts
index 92ff8b4acf2..52aa92e4292 100644
--- a/arch/arm/dts/socfpga_agilex5_socdk_emmc.dts
+++ b/arch/arm/dts/socfpga_agilex5_socdk_emmc.dts
@@ -18,7 +18,6 @@
  };
&mmc {
-       /delete-property/ sd-uhs-sdr104;


I have a concern regarding the v2 patch series that I would like you to verify.

In v2, patch 1/2 moves sd-uhs-sdr50 and sd-uhs-sdr104 from socfpga_agilex5_socdk-u-boot.dtsi into socfpga_agilex5_socdk.dts as a new &mmc node.

Since socfpga_agilex5_socdk_emmc.dts uses #include "socfpga_agilex5_socdk.dts" as a C preprocessor paste,

these SD UHS properties would be compiled into the eMMC DT, which is incorrect.

In v1, those properties lived only in the dtsi, which is not applied to the eMMC build since the eMMC build uses its own separate socfpga_agilex5_socdk_emmc-u-boot.dtsi.

This means the eMMC compiled DT was clean in v1.

To verify this, please perform the following test:

1. Revert socfpga_agilex5_socdk_emmc.dts back to its v1 state.

2. Remove the /delete-property/ lines for both sd-uhs-sdr50 and sd-uhs-sdr104 from socfpga_agilex5_socdk_emmc.dts.

3. Build the eMMC DT and inspect the compiled .dtb (e.g. via dtc -I dtb -O dts).

4. Confirm whether sd-uhs-sdr50 and sd-uhs-sdr104 are absent from the eMMC compiled DT.

If they are absent, it confirms that the /delete-property/ lines in the eMMC DTS were always no-ops.

In that case, the correct fix is simply to remove those no-op /delete-property/ lines that were introduced in v1 patch 2/2,

without moving any properties out of the dtsi into the base DTS.


Best regards,

Tien Fong

Reply via email to