Hummingboard/Custom Carrier Board IoT Boot Issues

[admin11]

This thread continues an earlier email conversation between a number of parties. It relates to IoT Core boot freezes where a Quad Core iMX6 Hummingboard has been placed in a third-party designed carrier board, this one known as Scout.

[henrybeberman‌, jonnettleton‌, jnettlet‌, (@robert_solomon), Ragan_Dunham‌, chirag.shah@harman.com‌, ChiragShah‌, (@marek_vinkler)]

[henrybeberman]

Hi Andrew,

(original log attached for context)

Looking at the log I believe your device is hanging as it probes for the presence of an eMMC. This change was upstreamed into U-Boot by SolidRun in the middle of last year. During the December public preview we rebased our changes on the latest U-Boot which pulled in this commit:

https://github.com/ms-iot/u-boot/commit/19ed6063a567c6924dbfc358bf4ce9a60a31c567#diff-376009f6384c7a...

Issues with the eMMC may also explain the TA panic error you encountered once we turned on the secure UEFI config by default.

I’ve uploaded a zip with two firmwares: https://github.com/hbeberman/imx-iotcore/blob/debug_print_init/temp/mmc_troubleshoot_7-26.zip

Experiment 1 is a firmware_fit.merged that is functionally the same as the previous test but with more debug prints so we can pinpoint where it’s getting hung up.

Experiment 2 is both a firmware_fit.merged and a uefi.fit. The U-Boot bypasses the eMMC check and uefi.fit has the TAs disabled so they won’t try to touch eMMC.

My recommendation is to work with SolidRun to validate how your eMMC is connected to the SoM and how it’s initialized in U-Boot.

Thanks,

Henry

[admin11]

henrybeberman‌, Sorry for the really late acknowledgement here. Your comment helped our hardware engineer narrow in on the issue, and we have now solved it. Turns out the our carrier board wasn't explicitly powering the eMMC because our carrier board doesn't provide this. However, the iMX6 has its own eMMC, and this is powered using the same interface. So the SOMs eMMC wasn't operating, which led to the boot failure.

From our engineer:

Issue was power to the SD3 interface. In the original Scout hardware design I had left the SD3 interface power disconnected, as this line powers the hummingboard eMMC chip, which isn't placed on Scout. But SD3 also powers the SOM onboard eMMC which caused the boot issue.

Thanks again for your support. Much appreciated.

[admin11]

moderator‌: Can I please request fast approval of this post. A number of NXP and Microsoft staff are participating in this new discussion. Thanks.