[OLPC-devel] Re: LinuxBIOS and integrated system debug status (Friday night).

[Richard Smith]

> Secondly, there's the fact that USB on this board is flaky as hell.
> We've seen all kinds of problems with it, and it's why I had to switch
> frequently with LinuxBIOS. That one concerns me a _lot_ more.

I've got a Beagle USB sniffer.  And since I've had to write my own
custom USB stack to talk to mass storage devices I've spent a good bit
of time looking at the traces.  I'll start grabbing trace logs when
mount the USB key and see if I can spot anything.

The beagle is not a compliance tester so there is a lot of things it
won't find but it has bit timing info and I can at least see if there
are errors on the bus or if the packets are failing CRC or are munged
up in some way.

You mentioned that you thought my mount errors might be this flakyness
and I get them at least once or twice every session so it shouldn't be
hard to catch.

Depending on what I find it may be a good idea to send the board to a
testing house and have the usb compliance tested.  Its not too
expensive.  That will tell us if the core stuff like  line impedance,
rise and fall times are within limits.

-- 
Richard A. Smith