In message 1115544065.6600.7.camel@localhost you wrote:
- erase 0x100000 0x3fffff (0x100000--0x3fffff is flash room for kernel)
- loadb 0x100000
You cannot load directly to flash. Youmust always write to RAM first, and then use the "cp" command (or cp.b etc.) to write to flash.
- using kermit to transfer uImage from ttyS0
- using "iminfo 0x100000", and i found "Bad Magic Number"
This means that your image was not written to flash (which is not a surprise, as "loadb" is NOT listed as a flash programming command in the manual -- ummm.... You did read the manual, didn't you? Before posting???
- setenv bootcmd bootm 0x100000
- boot, but i found "## Booting image at 00100000 ... Bad Magic Number"
This is to be expeted, too. If there was no valid image after download, it would take a certain amount of magic to be there now.
i used this uImage file to boot kernel from memory, so kernel image is right! and i tried to use "tftp 0x100000" to download image to flash,and the result is identical.
Of course. Because "tftp" is NOT listed as a flash programming command in the manual .. etc.
I REALLY recommend you to READ the manual.
and i did more testing:
- tftp 0x10000000 (0x10000000 is ram address)
- erase 0x100000 0x3fffff
- cp.b 0x10000000 0x100000 $(filesize)
- bootm 0x100000
the kernel boot successed! ...
Of course this works. That's the documented way to do things.
... but if i rebooted, and "bootm 0x100000", the kernel stoped when mounted nfsroot fs. the boot information was: .............................. <6>Freeing initrd memory: 4096K Freeing initrd memory: 4096K <5>Looking up port of RPC 100003/2 on 172.16.6.79 Looking up port of RPC 100003/2 on 172.16.6.79
(172.16.6.79 is nfsroot server)
Probably a configuration problem with your NFS server?
Best regards,
Wolfgang Denk