<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
<body bgcolor="#ffffff" text="#000000">
<font face="monospace">I am very sorry about my English. i should work
hard on English.<span class="moz-smiley-s6"><span> :-[ </span></span><br>
thanks for the suggestion.<br>
"write a minimalistic NAND loader for GTA02 (using the relevant code<br>
from u-boot as an inspiration)"<br>
<div id="result_box" dir="ltr"><font face="monospace">It is my
understanding that we just build a first-stage bootloader for <br>
kboot, use the mini loader to boot KBOOT or REAL SYSTEM. <br>
so this loader don't need command like in uboot. just to boot the KBOOT.<br>
<span class="moz-smiley-s1"><span> :-) </span></span><br>
<font face="monospace">Werner Almesberger wrote:</font>
<blockquote cite="mid:20080602210203.GF6685@almesberger.net" type="cite">
<pre wrap="">Wolfgang Spraul wrote:
<pre wrap="">Let's try this: Please settle on a specific u-boot/Kboot/Kexec task for
Xiangfu. He has a gta01 and gta02.
Andy and I discussed this. What we suggest is that Xiangfu continues
the work towards kboot, namely:
- write a minimalistic NAND loader for GTA02 (using the relevant code
from u-boot as an inspiration)
- move initialization that is currently in u-boot but that is not part
of the minimalistic NAND loader into the kernel
- as a proof of concept, boot a kernel image with a "hello world"
I've described this with more context in my mail with the general work
breakdown for making kboot happen, so I'd suggest to look there for
details. If anything needs clarification or if Xiangfu hits some nasty
problems, we won't bite his head off for asking ;-)
Andy has pointed out that Xiangfu may have some difficulties getting
things to build and run in the first place. For this, I'd suggest to
replace u-boot with some simple program, e.g., one that just blinks a
LED, and gradually build things up from there.
As development platform, I suggest to use GTA02, since it has a u-boot
in NOR, so it's easy to replace the NAND content during development.
Regarding the dynamic partitions: Andy and I feel that it's better if
we tackle these between ourselves.