Skip to contentSkip to conversation

AM65x MMC power-off delay fix

PATCH Applied Judith Mendez 3 participants 4 messages Sep 9 – Sep 22

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v2 cover AM65x MMC power-off delay fix

This series adds power-off-delay-us in AM65x EVM DT to ensure proper MMC card power cycling during boot.

AM65x EVM VDD rail discharges slowly (~632.97ms) through RC circuit. Without sufficient power-off delay, SD card boot fails in bootloader stage.

AM65x SD boot failure was initially exposed by U-Boot MMC driver patch [0], which was correct but revealed EVM hardware limitation. There is no SD card boot failure in Linux yet, but proper power cycling is required regardless. Additionally, equivalent Linux kernel discussions for similar fix [0] are ongoing. Linux MMC driver now supports delay after MMC power-off [1] and this patch adds power-off-delay-us=650ms to AM65x EVM DT.

Links:
[0] U-Boot patch & debug discussion:
https://lore.kernel.org/u-boot/[email protected]/

[1] Merged power_off_delay_us support in Linux MMC:
https://lore.kernel.org/linux-mmc/[email protected]/

V1: https://lore.kernel.org/linux-devicetree/[email protected]/

Changes since v1:
- Fix commit subject and description
- Add cover letter to provide full context of implementation & history

Judith Mendez (1):
arm64: dts: ti: k3-am654-base-board: Add power-off delay for proper
MMC power cycle

 arch/arm64/boot/dts/ti/k3-am654-base-board.dts | 1 +
 1 file changed, 1 insertion(+)
Signature · 1 line
2.34.1
#1 lore raw

PATCH v2 1/1 arm64: dts: ti: k3-am654-base-board: Add power-off delay for proper MMC power cycle

The AM65x EVM MMC1 VDD rail discharges slowly through RC circuit. SD spec (section 6.4.1.3) requires VDD < 0.5V for minimum 1ms during power cycle. Measured discharge time on this board is ~632.97ms to reach < 0.5V threshold.

During boot, MMC power cycles between boot stages. Without power off delay, VDD doesn't fully discharge. Card retains 1.8V signaling state from previous boot stages. When MMC driver later initializes, card no longer asserts S18A in ACMD41 response (already switched to 1.8V). Host misinterprets card state and forces 1.8V anyway, falling into a bad state and causes MMC init failure with "unable to select a mode" signature.

Add power-off-delay-us = 650000 (650ms) to ensure proper VDD discharge during power cycles, allowing card to fully reset state between boot stages.

---
Changes since v1:
- Fixup commit header and description

---
 arch/arm64/boot/dts/ti/k3-am654-base-board.dts | 1 +
 1 file changed, 1 insertion(+)
arch/arm64/boot/dts/ti/k3-am654-base-board.dts +1 -0
index ca1300e130936..edcf039ccfbd0 100644
@@ -460,6 +460,7 @@ &sdhci0 {
 &sdhci1 {
 	status = "okay";
 	vmmc-supply = <&vdd_mmc1_sd>;
+	power-off-delay-us = <650000>;
 	pinctrl-names = "default";
 	pinctrl-0 = <&main_mmc1_pins_default>;
 	ti,driver-strength-ohm = <50>;
Signature · 1 line
2.34.1
#2 lore raw

Re: [PATCH v2 1/1] arm64: dts: ti: k3-am654-base-board: Add power-off delay for proper MMC power cycle

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] The `power-off-delay-us` property is placed directly in the SDHCI host node where it is silently…

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] The power-off-delay-us property is placed directly in the SDHCI host node where it is silently ignored by the kernel, rendering the intended fix ineffective. --

commit ad27b5f7c84e8187c60f3d6ec61cfd8ac2996675
Author: Judith Mendez <[email protected]>

arm64: dts: ti: k3-am654-base-board: Add power-off delay for proper MMC power cycle

This commit adds a power-off-delay-us property to the SDHCI1 node in the AM654 base board Devicetree. It intends to ensure the MMC1 VDD rail fully discharges during power cycles between boot stages.

arch/arm64/boot/dts/ti/k3-am654-base-board.dts &sdhci0 { · line 463

Quoted 9 lines
arch/arm64/boot/dts/ti/k3-am654-base-board.dts +1 -0
index ca1300e130936..edcf039ccfbd0 100644
@@ -460,6 +460,7 @@ &sdhci0 {
 &sdhci1 {
 	status = "okay";
 	vmmc-supply = <&vdd_mmc1_sd>;
+	power-off-delay-us = <650000>;

[Severity: High]
Will this property be parsed by the kernel when placed directly in the host node?

The power-off-delay-us property is specific to the mmc-pwrseq-simple binding and is parsed exclusively by drivers/mmc/core/pwrseq_simple.c. Neither the core MMC code (drivers/mmc/core/host.c) nor the SDHCI host drivers parse this property directly from the host controller node.

Because it is placed directly under &sdhci1 rather than inside an mmc-pwrseq-simple node, the kernel will silently ignore it. Does this leave the MMC initialization failure unresolved?

Quoted 3 lines
arch/arm64/boot/dts/ti/k3-am654-base-board.dts +0 -0
 	pinctrl-names = "default";
 	pinctrl-0 = <&main_mmc1_pins_default>;
 	ti,driver-strength-ohm = <50>;
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
maintainer Applied
#3 lore raw
In reply to Judith Mendez

Hi Judith Mendez,

Quoted 6 lines · Judith Mendez

This series adds power-off-delay-us in AM65x EVM DT to ensure proper MMC card power cycling during boot.

AM65x EVM VDD rail discharges slowly (~632.97ms) through RC circuit. Without sufficient power-off delay, SD card boot fails in bootloader stage.

Quoted 8 lines

AM65x SD boot failure was initially exposed by U-Boot MMC driver patch [0], which was correct but revealed EVM hardware limitation. There is no SD card boot failure in Linux yet, but proper power cycling is required regardless. Additionally, equivalent Linux kernel discussions for similar fix [0] are ongoing. Linux MMC driver now supports delay after MMC power-off [1] and this patch adds power-off-delay-us=650ms to AM65x EVM DT.

[...]

I have applied the following to branch ti-k3-dts-next on [1].

For future reference, we dont need a coverletter for a single patch.. the information can be plugged in the diffstat section.

Additionally, I see that power-off-delay-us is in next - if the patch with commit 555bb7b7c1497accc0894423220f772f006a9584 is dropped, please let me know so that this patch can be dropped from my tree since that is required for dtbs_check to be clean. Picking this patch as a valid required fix and helps users.

Thank you!

[1/1] arm64: dts: ti: k3-am654-base-board: Add power-off delay for proper MMC power cycle

      commit: 7dd4fb52749d9726302463dfa2ccf71db6bbed4c

All being well this means that it will be integrated into the linux-next tree (usually sometime in the next 24 hours) and sent up the chain during the next merge window (or sooner if it is a relevant bug fix), however if problems are discovered then the patch may be dropped or reverted.

You may get further e-mails resulting from automated or manual testing and review of the tree, please engage with people reporting problems and send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they should be sent as incremental updates against current git, existing patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying to this mail.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/ti/linux.git

Signature · 4 lines
Regards,
Nishanth Menon
Key (0xDDB5849D1736249D) / Fingerprint: F8A2 8693 54EB 8232 17A3  1A34 DDB5 849D 1736 249D
https://ti.com/opensource