prlw1@newn.cam.ac.uk (Patrick Welche) writes:

>It seems the alogorithm is

>- issue reset
>- write gencomm = magic_no
>- wait until gencomm = ~magic_no

>What I am seeing is

>- issue reset
>- write gencomm = magic_no
>- wait forever as gencomm == magic_no

>If I remove the write, I do read ~magic_no from gencomm => the firmware
>has finished initializing.


Sounds like the firmware has finished initializing before the write occurs,
which is reasonable, since we DELAY(1000) after issuing the reset.

The FreeBSD driver writes the magic value first, then issues the global
reset.

-- 
-- 
                                Michael van Elst
Internet: mlelstv@serpens.de
                                "A potential Snark may lurk in every tree."