FYI: cross posted requests for review
john.r.rose at oracle.com
Tue Mar 30 15:00:17 PDT 2010
For JSR 292, I will be integrating more changes through the hotspot-comp subtree.
As before, I have posted the requests for review to hotspot-compiler-dev. Here are the messages.
To scope the discussion to one venue, please give review comments on the hotspot-compiler-dev list, if possible, or send them directly to me.
Request for reviews (XL): 6939134: JSR 292 adjustments to method handle invocation
Request for reviews (M): 6939196: method handle signatures off the boot class path get linkage errors
Request for reviews (L): 6939207: refactor constant pool index processing
Request for reviews (L): 6939203: JSR 292 needs method handle constants
The change for 6939207 may be incomplete. I'd like advice on whether to push the change farther. I think we should perform byte swapping at the point where data is loaded (in the [Raw]BytecodeStream), because the byte order depends on the type of stream and is therefore known most accurately in the stream performing the fetching. But currently, we swap byte order in the CP accessors, which is many module boundaries away from the stream. See the temporary function 'get_index_native_swap' and the FIXME comments nearby; I'd like to get rid of get_index_native_swap. Is there a current reason why we can't tie byte swapping decisions to the type of bytecode stream?
More information about the hotspot-dev