[Prism54-users] Re: [Prism54-devel] Prism54 development update

[Sebastien]

> Jean-Baptiste Note <jean-baptiste.note at m4x.org> wrote:
> > We could even do more (and i'd like this to show in the code -- stay
> > tuned !). From what i can see, there's no reason at all to even run code
> > on the ARM. ei, it is certainly possible to control the whole radio
> > thing entirely from the host !
>
> That sounds very interesting, could you please share some details, even
> incomplete? :)

At least the ISL3880 and ISL3886 contain a mechanism called "memory window", 
which allows mapping any portion of the ARM memory (except for TCMs) to the 
PCI I/O space.
I used it to find out the addresses of the two general purpose communication 
registers (by writing 0xdeadbeef into them from PCI, and then scanning all 
the ARM memory for 0xdeadbeef using the memory window) to enable a basic 
communication with a small firmware (the first ISLMON) from which I did the 
rest of the reverse engineering.

Ant that this memory window could be used to directly access the radio's 
memory mapped registers from the host.