new kernel ... / meaning of stable-tracking

[Werner Almesberger]

andrzej zaborowski wrote:
> kernels through u-boot seems quicker and for that I have to disconnect
> the debug board everytime.

Because you have only one USB port ? Or because of some other problem ?

By the way, I now usually flash via Linux, except when the kernel
doesn't come up. The added convenience is in not having to interrupt
u-boot at the right moment. Here's the script I'm using for GTA02:

#!/bin/sh -ex
DEV=/dev/mtd3
[ -r "$2" ] || exit
ifconfig usb0 192.168.0.200 up
scp $2 $1:
ssh $1 "flash_eraseall $DEV && nandwrite -p $DEV `basename $2` && sync && sleep 1 && /sbin/reboot -f"

> Yes, sounds good, and also I'm not suggesting that it be forced
> upstream.

Heh, those who think they can _force_ something upstream are usually
in for a surprise :-) But I think it's not outlandish enough that we
couldn't submit it - after all, it extends a concept that already
exists in the PCF50633 hardware, and the code will be separated from
the rest clearly enough. If this still isn't good enough, then we'll
have to move that function to platform code.

- Werner