Request for reviews (M): 6930772: JSR 292 needs to support SPARC C1
john.r.rose at oracle.com
Fri Apr 16 14:48:56 PDT 2010
I like it too; reviewed.
The use of L0 is tricky. You call it L7 in one comment. Be sure that all of these uses of L0 are linked together by comments.
The change touches one of my pet peeves: The SP adjustment tracking in the frame logic has always seemed awkward to me (not from your change). I don't see a clearer reformulation yet. (Chuck Rasbold made it better with the unextended_sp stuff.)
One partial cleanup would lead to some simplifications. It would change the convention by which the interpreter pops its outgoing arguments. Currently it pretends to keep the arguments on stack throughout the call; it should release them to the callee before completing the control transfer to the callee. Currently, as a result, a return to the interpreter resets to a saved SP that points to the argument list base (lowest address, not highest), the address of an argument list that (at that point) is long gone. The interpreter must then perform extra calculations to re-adjust the SP by the size of the absent argument list, calculations which add no value. And when stack walking, we have to continually remember that the unextended SP might overlap with stack storage owned by the callee, which is counterintuitive and therefore bug-prone.
On Apr 16, 2010, at 5:41 AM, Christian Thalinger wrote:
> On Thu, 2010-04-15 at 10:31 -0700, Tom Rodriguez wrote:
>> Could the is_method_handle_invoke become part of CodeEmitInfo instead?
>> That would make more sense to me.
> Right. This looks much better:
> -- Christian
More information about the hotspot-compiler-dev