Quote:
Originally Posted by natemcnutty
ACL, I've been trying to find a way to get recovery loaded from our NBH. There are only two things I can think of.
First, cat an image of recovery out to the sdcard and then cat it onto the end of the payload. I'd have to figure out padding to get it to write to the correct offset, and I'm not sure if it would mess anything else up since we'd be writing blank or padded values to that protected area of NAND.
The other alternative is that we include /recovery in initrd and have init check /dev/block/mtdblock1 to see if recovery is there yet. If not, deploy it from the initrd and launch into it. This way we can still update recovery, but it gets deployed if it doesn't exist.
What do you think?
|
I dont know about this padding stuff. From what wozzer told me it looks like this thing can end up in different locations so its impossible to predict where. So right now only testing will answer our questions.
I tried in the past to embed the recovery in the initrd, but its simply too big. This is a potential solution if we use LK, but we are too far from that right now.
Another temporary option is to build a small recovery kernel( make a new defconfig) and just disable all the unneeded options. If we can bring the kernel down to 1 meg, then recovery can fit fine in intrd. We just need a kernel to help us hop into the real kernel.. im open to ideas, this is new grown for all of us.