Fixing missing override on EVM DeletegateCall instruction? I think it's missing...

[msooseth]

Description

I think the DelegateCall was missing the override check? @elopez what do you think?

Notice that the override will be reset by the delegateCall function:

        \xGas -> do
          resetCaller <- use $ #state % #resetCaller
          when resetCaller $ assign (#state % #overrideCaller) Nothing

But it does not assign the from caller to the value, instead, we do that in e.g.:

        OpStaticcall ->
          case stk of
            xGas:xTo:xInOffset:xInSize:xOutOffset:xOutSize:xs ->
[...]
                Just xTo' ->
                  case gasTryFrom xGas of
                    Left _ -> vmError IllegalOverflow
                    Right gas -> do
                      overrideC <- use $ #state % #overrideCaller
                      delegateCall this gas xTo' xTo' (Lit 0) xInOffset xInSize xOutOffset xOutSize xs $
                        \callee -> do
                          zoom #state $ do
                            assign #callvalue (Lit 0)
                            assign #caller $ fromMaybe self overrideC

[...]

And similarly in other places. But somehow OpDelegateCall is missing this.

Checklist

• tested locally
• added automated tests
• updated the docs
• updated the changelog

[elopez]

@msooseth DELEGATECALL is special as it normally preserves the context (ie the caller remains the same as it was before the call), I'm not sure if pranking should be touching the caller here. Have you looked at Foundry's behavior to see what it does in this particular case?

[msooseth]

Ah, interesting. So they have a version of prank with a 2nd argument that, if set to true, will also affect delegateCall. But it does not affect delegateCall otherwise.

Discussion: foundry-rs/foundry#8863
The PR that got merged: https://github.com/foundry-rs/foundry/pull/8863/files

So it's good as-is. I am changing this PR to add a note instead so it's documented what's going on.