[Vm-dev] bad code in patch to bug 6555

[Rob Withers]

I am attempting to generate a VM with the 
InterpreterSimulator-fixes-mtf.1.cs patch loaded from 
http://bugs.squeak.org/view.php?id=6555.  The problem is with the following 
methods:

InterpreterPlugin>>#long32At: byteAddress
    "Simulation support.  Answer the 32-bit word at byteAddress which must 
be 0 mod 4."

    ^self getInterpreter long32At: byteAddress

and

InterpreterPlugin>>#long32At: byteAddress put: a32BitValue
    "Simulation support.  Store the 32-bit value at byteAddress which must 
be 0 mod 4."

    ^self getInterpreter long32At: byteAddress put: a32BitValue

When the plugin is being inlined, the recursive call to #long32At: and 
#long32At:put: is causing the inlining to enter an infinite loop.

With these methods removed, everything builds fine.

Rob