ARM: Fix crash in assembler interpreter

[Omair Majid]

On 01/31/2012 07:28 AM, Andrew Haley wrote:
> Xerxes Rånby reported that he'd found a very rare crash in the GC when
> running the ARM assembler interpreter in an interpreter-only
> environment.  The problem was tracked down to an incorrect saved
> bytecode pointer in the interpreter state.  The fix is very simple:
> write the pointer back to the interpreter state when returning from
> a method.
>
> We've tested this pretty extensively, so I'm committing this to trunk,
> but I think it should go in the 1.11 branch too.  OK?
>

Okay for 1.11.

Cheers,
Omair