On 2007-11-16 11:18 +0000, Colin Tuckley wrote:
> For some time now we have had problems with JTAGing loon-3's using bflash.
> About 50% of the time it fails with a seg fault.
> 
> I've spent some time looking at it this morning and found that the seg fault
> is caused by it trying to continue with the programming after it fails to
> read the CFI data. I've now fixed the error handling.
> 
> This leaves us with the real problem.
> 
> FATAL: Failed to read CFI Data. Can't find NOR.
> 
> Looking at the query_cfi routine it expects the first three bytes to be "QRY".
> 
> Anyone got any ideas about what might be causing this?

This is a hardware problem. QRY is the string the CFI spec says should
be sent. The problem is that it fails to arrive, which is because
'something' is preventing the JTAG process from getting the bus and
reading data. The problem is that this could be almost anything that
is on the bus failing to get into the right state. Fixing it requires
a long session with scopes/bus analysers, so it's a lot easier to run
bflash again until it works for the time being.

[The above understanding extracted from David Bisset]

So yes there is something wrong and it is tiresome, but it's not
serious, hard to fix and easy to work round so it's quite a long way
down the list of things to do.

Wookey
-- 
Principal hats:  Balloonz - Toby Churchill - Aleph One - Debian
http://wookware.org/