Closed iam-TJ closed 3 years ago
I copied just the bootloader from the 20210108
image (the first 10240 sectors) to the micro-SD card with the 20210730
release and that boots, so the issue is in the bootloader.
One data-point - when it was failing the Armory was getting very hot very quickly, including the micro-SD card, which might suggest it was in a loop re-reading the card.
Recently U-Boot enforced the use of their "driver model" drivers. As a consequence the SD card configuration is taken from the devicetree file in arch/arm/dts/imx53-usbarmory.dts
.
Such file is missing the card detect (CD) pin configuration, and therefore the card is not detected by the bootloader.
The following patch will be submitted to U-Boot to fix this issue:
diff --git a/arch/arm/dts/imx53-usbarmory.dts b/arch/arm/dts/imx53-usbarmory.dts
index f34993a490..433b62e736 100644
--- a/arch/arm/dts/imx53-usbarmory.dts
+++ b/arch/arm/dts/imx53-usbarmory.dts
@@ -91,6 +91,7 @@
&esdhc1 {
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_esdhc1>;
+ broken-cd;
status = "okay";
};
Ran into the same issue today, and it took a while to debug because I thought I had a corrupt image somehow.
Any guidance on when the next pre-compiled release will be available? In the meantime, it seems it would be prudent to pull the last Mk 1 release since it is in effect defective.
Earlier today I installed the
usbarmory-mark-one-usd-debian_buster-base_image-20210730
image to micro-SD card and found the Armory Mark 1 failed to boot. No sign of life, host PC doesn't see the CDC device, no heartbeat LED. Thinking I'd done something wrong I hacked about with checking SHA sums, including comparing the .raw image hash with the bytes on the micro-SD - all matched. Eventually I thought to try an older image and found thatusbarmory-mark-one-usd-debian_buster-base_image-20210108
boots correctly. This seems to suggest the boot-loader on the 20210730 image is broken.