it's getting easier and easier to resync the new config tool to the ongoing changes. Took a few minutes earlier today and I was back on the air.
I have noticed that the testing of the first 16M of memory in auto.c is very slow, so I've turned it off on my version. My guess is for you guys it runs very fast. There must still be a difference somewhere. Can someone confirm that the K8 memory test in auto.c is fast?
A note on the SST 49LF00 flash parts, since we've all had trouble with them. Here's what I have found. The first time I flash there is always trouble. However, if I 'eraseall /dev/mtd0' a few times, then do the 'dd trick':
dd if=image of=/dev/mtd0 bs=65536 seek=$i skip=$i
i.e. program individual blocks, it almost always works.
From that point on, programming is a matter of
cat image > /dev/mtd0
So it seems there is something 'sticky' in these parts when new, cured by programming individual blocks. Flash is very odd ...