WLAN: known issues and how to help - uptime test - more results

Werner Almesberger werner at openmoko.org
Mon Feb 9 02:47:25 CET 2009


Philip Rhoades wrote:
> NAND read: device 0 offset 0x80000, size 0x200000

Funny. It didn't take the increase. But after the reboot it did, so
we have a happy ending ;-)

> However if I use reboot, the FR starts up and I can connect to  
> 192.168.0.202 via USB again.

Perfect !

> I have then rebooted but usb0 remains at 192.168.0.202

Unless you changed the init script, a reboot will return usb0 to its
default settings. If you want a different address, you can edit
/sbin/init. It's a shell script.

> and the USB link does not remain up long enough for me to  
> look at things (maybe because of conflict with my 192.169.0.* LAN?) -  

Could it be that USB locked up for some reason ? I'm seeing this
from time to time. Just unplugging and reconnecting USB and then
ssh'ing in again usually solves this. (The usb0 IP address is
preserved.)

> can you do a rootfs that uses DHCP to set up eth0 so I don't need to  
> mess around with usb0?

Hmm, I don't understand ... how would using DHCP for eth0 help with
usb0 ? In fact, part of the purpose of this rootfs is to keep things
as manual as possible, so that funny interactions of system services
can be debugged one step at a time.

Anyway, I think you could just try to use your normal rootfs. If
the problem is the assertion failure, then this will be the easiest
way to reproduce it and we won't need any of the debugging
capabilities of the wlan-trial rootfs.

> I presume it also uses memory that would normally be available for  
> something else?

Nope, it just reads 3 MB instead of 2 MB from the 8 MB kernel
partition. As far as RAM size is concerned, a "small" kernel will
only use what it needs and happily ignore all the rest that u-boot
has loaded from NAND.

> OK but would it be a good idea for normal use?  Faster?

Faster, leaner, meaner, cleaner ;-) However, it can also be a bit
of a learning experience. So perhaps it's better to change one
thing at a time to avoid confusion.

- Werner



More information about the devel mailing list