`StackOutputs.stack_top()` fails when `stack < STACK_TOP_SIZE` about miden-vm

[bobbinth]

This is a bit strange because stack output should always have at least 16 items (i.e., STACK_TOP_SIZE) in it. This is because we build stack outputs from the stack here and the stack should always have at least 16 items.

So, the bug is either in the processor and for some reason stack depth drops to below 16 at some point, or we are initializing StackOutputs somewhere else - and that's causing the issue.

If it is the latter, we should do the following: in the constructors of StackOutput pad the input items with ZERO to ensure that StackOutputs always has at least 16 elements. One thing to pay attention to is the order of elements (I don't remember whether we'd need to pad from the front or from the back).

[frisitano]

So, the bug is either in the processor and for some reason stack depth drops to below 16 at some point, or we are initializing StackOutputs somewhere else - and that's causing the issue.

If it is the latter, we should do the following: in the constructors of StackOutput pad the input items with ZERO to ensure that StackOutputs always has at least 16 elements. One thing to pay attention to is the order of elements (I don't remember whether we'd need to pad from the front or from the back).

Turns out the issue was related to construction of the StackOutputs struct. Padding appropriately and ordering fixed the issue.

[bobbinth]

Should we still update the StackOutputs constructor?

[frisitano]

Should we still update the StackOutputs constructor?

Yes I think that makes sense. Removes a potential footgun.

[bobbinth]

Closed by #1010.