Instruction for computing absolute address of a local variable about miden-vm HOT 3 CLOSED

Just to reiterate - procedures will always be trusted, right?

Yes, this is correct.

How much do you expect this to come up, though? I wonder if it's worth doing it now vs. waiting to see to what extent we really want this & in what contexts, since different contexts could drive different preferred solutions.

Good point! We can wait a little on this. I think u256 operations, EC signatures, and hash functions might need something like this - but good to wait until we actually start working on them.

from miden-vm.

The only downside of this I can think of is that it would enable potential abuse of fmp where a procedure could return absolute address of its local (which would be a logical error).

Just to reiterate - procedures will always be trusted, right? (With functions used for "untrusted" procedures, where parameters passed via the stack, as discussed in #73 )

We could add assembly instructions to access fmp and then the caller would be able to compute the absolute address, but i wonder whether having a more specialized instruction is more convenient.

Yes, I prefer passing via env as proposed rather than requiring the caller to figure it out themselves. The complexity is minimized this way & kept in one place (where it will in fact be straightforward to handle).

In some situations it may be useful to get the absolute memory location of a local variable.

How much do you expect this to come up, though? I wonder if it's worth doing it now vs. waiting to see to what extent we really want this & in what contexts, since different contexts could drive different preferred solutions

from miden-vm.

Closed by #124

from miden-vm.