Request for review (L): 6917766: JSR 292 needs its own deopt handler

[John Rose]

Reviewed.  But, can you factor the new assert code for frames some more?  I like what it does, but it is hard to read, because it uses a scratch frame to carry an argument, and it is repetitive.  It would be great if the 2-3 checks repeated could be merged into a single debug-only sanity check routine.

Thanks for chasing down this bug.

-- John

On Jan 21, 2010, at 11:02 AM, Christian Thalinger wrote:

> http://cr.openjdk.java.net/~twisti/6917766/webrev.01/
> 
> One thing that may pop into your eye is that I removed the guarantee in
> Deoptimization::fetch_unroll_info_helper:
> 
> http://cr.openjdk.java.net/~twisti/6917766/webrev.01/src/share/vm/runtime/deoptimization.cpp.udiff.html
> 
> I discussed this with Tom and it seems this check is just a duplication
> of the logic in frame.  I also removed the Event::log.  
> 
> Another way would be to adjust unpack_sp:
> 
> http://cr.openjdk.java.net/~twisti/6917766/webrev.00/src/share/vm/runtime/deoptimization.cpp.udiff.html
> 
> -- Christian
>