On Thu, Jun 9, 2011 at 12:50 PM, Christian Thalinger <[email protected]> wrote: > Sorry I'm a bit behind but is this still a problem or have you fixed it > already? > > It looks like it works: > > @ 24 > java.lang.invoke.MethodHandle::invokeExact (32 bytes) inline (hot) > @ 28 constant::method__0$RUBY$foo (7 > bytes) inline (hot) > \-> discounting inline depth > @ 1 > java.lang.invoke.MethodHandle::invokeExact (9 bytes) inline (hot) > @ 2 > sun.invoke.util.ValueConversions::identity (2 bytes) inline (hot) > \-> discounting inline depth
I need to do an updated run. It's certainly possible it got fixed in the interim. I was not looking at the inlining graph (is this PrintInlining or LogCompilation, btw? I've never seen this output before with the nice "discounting inline depth"), I was looking at the resulting assembly. There, it had at least one callq operation right where I'd expect to see the inlined constant. Does the assembly output for this region look ok to you? - Charlie _______________________________________________ mlvm-dev mailing list [email protected] http://mail.openjdk.java.net/mailman/listinfo/mlvm-dev
