[Prism54-users] Re: [Prism54-devel] Prism54 development update
Sebastien
sebastien.bourdeauducq at gmail.com
Fri Sep 9 14:19:55 UTC 2005
> 1) make a arm .S file which, when compiled, gives you _exactly_ same
> image as reverse-engineered one.
>
> 2) heavily comment it (can be done by many people in parallel).
> Give symbolic names to labels, variables etc...
> (evolution: b 1234 -> b lbl_1234 -> b disable_tx)
>
> 3) when you feel it's more or less understood, try some
> more invasive changes (which change length of fw image and/or
> offsets of code/data).
>
> 4) if (3) runs without crashes, well, it's mostly done.
This looks good, but one problem remains : the uploaded firmware is a
self-extracing compressed binary. We'll *also* need to repack the firmware
and run it exactly the way the official unpacker does. And this unpacker
still accesses a few unknown hardware locations...
The USB endpoints and PCI transfers are nearly OK by now, only one problem
remains with DMA TX : only the three least significant bits are sent. But
probably I have a missing or wrong value in some registers.
With that and the radio (0xc00006xx registers - won't be easy...), a minimal
firmware should be usable, although problably highly suboptimal.
> I'm at "need to try that armulator thing" stage now.
Do you want my beginning of modified version ?
More information about the Prism54-devel
mailing list