On Thu, Jul 06, 2006 at 10:25:40AM -0400, Joseph Jezak wrote:
> 
> As far as I can tell, the 4.x firmware isn't all that different,
> it's the initvals that are causing the trouble.  

Ok. Let's look at the first ten initvals. I don't care about 11+13 
initvals at the moment.

4.80.9.1                3.130..

d11b0g0initvals4        d11b0g0initvals2
d11b0g0bsinitvals4      d11b0g0bsinitvals2
d11a0g0initvals4        d11a0g0initvals2
d11a0g0bsinitvals4      d11a0g0bsinitvals2
d11b0g0initvals5        d11b0g0initvals5
d11b0g0bsinitvals5      d11b0g0bsinitvals5
d11a0g0initvals5        d11a0g0initvals5
d11a0g1initvals5        d11a0g1initvals5
d11a0g0bsinitvals5      d11a0g0bsinitvals5
d11a0g1bsinitvals5      d11a0g1bsinitvals5

This is absolutely ok. I don't see a differnt order. Keep also in mind 
that rev2 and rev4 initvals are identical. It's just another name for 
the same initvals block. Where do you see a different order?

> I guess what I'm trying to say is that instead of assuming that a
> certain type of initval follows another, perhaps it would be best to
>   keep the offset of each initval independent, like what is done for
> the uCode.  That and it would be nice to have the a/b/g/n/revision
> for the initial values instead of the arbitrary numbering scheme we
> have now.

We have to create new code, if we want to extract 11+13 initvals.
This is a good chance to introduce something like .iv_map for our 
struct. There we can define which numbering scheme is used. We can get 
rid of REVERSE_ORDER_INITVALS and MISSING_INITVAL_08 flags, too. But 
adding a bunch of offsets instead is painful. We should avoid that.

Basically, the initvals are always arranged in the same order. I do not 
think that the 4.x problems are caused by any wrong initvals. 


Martin
_______________________________________________
Bcm43xx-dev mailing list
[email protected]
http://lists.berlios.de/mailman/listinfo/bcm43xx-dev

Reply via email to