using kboot for booting / bootloader minimization

Andy Green andy at
Thu Apr 24 11:50:57 CEST 2008

Hash: SHA1

Somebody in the thread at some point said:

| Could I say roughly:
| kboot = minimize uboot(firmware) with 2-stage booting?
| If using kboot instead of uboot, the image size and booting time would
| not be improved.

Your summary is correct AIUI: but the booting time would be improved.
In the normal case, you don't use kboot, you only use it to get into --->

| what else would get from kboot, a diagnostic environment with rich
utility? in the normal case, you just go booloader -> final kernel in one
step and do not use kboot.

Boot time improvements are threfore not coming from kboot, they are
coming from chopping down U-Boot, not initializing Glamo / LCM there, etc.

Still, current setup with U-Boot and NAND is ferociously complicated and
also in production test is complicated.  Wolfgang mentioned
simplification in this area.  That would be a good thing here and for
future products.

It currently looks like despite our best hopes to ditch NAND going on,
we won't be able to and we will end up with another GTA02-type situation
as far as NAND.  So an effort nuking bootloader / NAND back to a bare
minimum is worth having with or without kboot.

- -Andy
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora -


More information about the openmoko-kernel mailing list