Re: [Simulavr-devel] Settings for stackpointer

[Onno Kortmann]

Hey,

Am 21.02.2016 um 15:25 schrieb Albrecht Frenzel:
> How many users of simulavr recently use  it (or plan) to simulate
> tiny11/12/15 or at90s1200?
>
> Does anybody know, how many people really use simulavr?
>

I am using simulavr from time to time (maybe every couple months) and I believe 
(am not 100% sure) I
was also the guy who wrote the current ThreeLevelStack implementation.

I used simulavr to simulate an ATTiny15 (+ some external HW modelled using the 
verilog cosimulation
part).

Is the code in the way of some refactoring or similar?

IMO, we should try to keep the parts as loosely coupled as possible from each 
other - I also don't
think that, if decoupled from the rest, a tiny15 implementation would harm - or 
would it?

I am not sure whether the tiny15 is still in the current code base - is it?

I also think there is some value in supporting obsolete HW. I personally think 
it is a good
compromise to replace, if possible, simulation of obsolete HW with an 
implementation that is a
strict *superset* of the old behavior (e.g. a fully compatible timer 
implementation). That keeps
old-HW code working, except maybe not catching illegal register accesses and 
the like.

The peculiarities of the HW stack make it quite difficult to replace it with a 
pointer-based stack.
Isn't the implementation just a couple lines anyways?

Cheers,

Onno