[Question] How is image-b used on CRB
Closed this issue · 1 comments
mine260309 commented
On CRB with 64MiB flash chip for BMC, the build uses below partitions:
Partition | Size |
---|---|
u-boot | 512KiB |
image-a | 27.5MiB |
sofs | 2MiB |
rwfs | 6MiB |
image-b | 27.5MiB |
From the build scripts in obmc-phosphor-full-fitimage.bbclass and image_types_phosphor_auto.bbclass, the image-a/b
are the identical fitImages containing the full rofs.
During boot, u-boot loads the fitImage
addressing from image-a
, so image-a
is always used.
I was expecting that the image-b
will be used when image-a
fails to boot, but I could not find related code.
So my questions are:
- Does the build support fail-over on BMC boot?
- If yes, when
image-a
fails to boot, how does it detect and switch to theimage-b
partition?
Thanks!
leiyu-bytedance commented
It's figured out that the update script will flash the image to the other image-a/b and update the uboot env to boot from the newly flashed fitImage.