i.MX ProcessorsのトピックRe: dd'ing u-boot doesn't wipe u-boot variables?

dd'ing u-boot doesn't wipe u-boot variables?

entropyro — Wed, 17 Apr 2024 22:50:00 GMT

Working with an iMX8M Plus Quad.

Looking at this manual on ~page 11: i.MX_Linux_Users_Guide_Rev_LF6.6.3_1.0.0_2024-03-29.pdf (https://www.nxp.com/docs/en/user-guide/IMX_LINUX_USERS_GUIDE.pdf)

Issuing: dd if=imx-boot of=/dev/mmcblk0 bs=1k seek=32 conv=fsync

Plenty of normal-looking output.

But when I reboot and interrupt to get back into u-boot, my test env variable is still present! Shouldn't it be blown away? What am I doing wrong?

I've also tried with /dev/mmcblk0boot0 and zero'ing it out before the dd and same thing! The only thing that's managed to blow away my test variable is using uuu.exe to blast on u-boot and the .wic image.

I'm sure I'm missing something fundamental here...any thoughts?

Thank you for your time!

Re: dd'ing u-boot doesn't wipe u-boot variables?

Zhiming_Liu — Thu, 18 Apr 2024 06:42:01 GMT

Hi @entropyro

I tested same behavior on i.MX8MM, below steps can work, Uboot 2022 can be replaced by Uboot2023. You can refer it.

echo 0 > /sys/block/mmcblk2boot0/force_ro dd if=imx-boot-imx8mm-lpddr4-evk-sd.bin-flash_evk of=/dev/mmcblk2boot0 bs=1k seek=33 conv=fsync echo 1 > /sys/block/mmcblk2boot0/force_ro

Re: dd'ing u-boot doesn't wipe u-boot variables?

entropyro — Thu, 18 Apr 2024 15:40:54 GMT

@Zhiming_Liu, OK this is strange, that's basically exactly what we are doing, but our test u-boot variable persists after flashing it and power-cycling the system! Any idea what I'm missing?

What seek offset should I use for an i.MX8 Plus Quad? I've tried both with no change, but it'd be good to know.

Re: dd'ing u-boot doesn't wipe u-boot variables?

entropyro — Thu, 18 Apr 2024 23:32:48 GMT

@Zhiming_Liu, thank you for your help. I figured out what was going on - after a saveenv my u-boot environment variables were being written to an area of the disk not touched by re-blasting the u-boot image. I am actually compiling my own u-boot environment variables into the u-boot image, so I didn't want this feature. I am working around it by blasting 0s into the section of memory where the u-boot environment variables are stored, then loading the image so that the "default variables" are always from the binary.

Thank you!

Re: dd'ing u-boot doesn't wipe u-boot variables?

roke — Fri, 19 Apr 2024 13:16:13 GMT

dd if=/dev/zero of=/dev/mmcblk0 bs=16k seek=256 count=1

Re: dd'ing u-boot doesn't wipe u-boot variables?

entropyro — Fri, 19 Apr 2024 14:35:17 GMT

Where are you pulling the 4096000 (16k*256) number?

Re: dd'ing u-boot doesn't wipe u-boot variables?

roke — Fri, 19 Apr 2024 15:47:37 GMT

https://github.com/nxp-imx/uboot-imx/blob/lf-5.15.71-2.2.2/configs/imx8mp_evk_defconfig

CONFIG_ENV_SIZE=0x4000
CONFIG_ENV_OFFSET=0x400000
CONFIG_ENV_SECT_SIZE=0x10000

Re: dd'ing u-boot doesn't wipe u-boot variables?

roke — Fri, 19 Apr 2024 15:53:03 GMT

16k*256 =16*1024*256 = 4194304

https://man7.org/linux/man-pages/man1/dd.1.html

 N and BYTES may be followed by the following multiplicative
       suffixes: c=1, w=2, b=512, kB=1000, K=1024, MB=1000*1000,
       M=1024*1024, xM=M, GB=1000*1000*1000, G=1024*1024*1024, and so on
       for T, P, E, Z, Y, R, Q.  Binary prefixes can be used, too:
       KiB=K, MiB=M, and so on.  If N ends in 'B', it counts bytes not
       blocks.