Hi folks.

I'm very interested in java code optimization and threading-issue.

I think that following code optimization can occur in current implementation
of JVMs and/or future implementation of JVMs.

I mean java-compiler/hotspot optimizing here.

			

			1) reordering
			

			

			synchronized ( o1 ) {
			

			data.y = 1;
			

			data.x = 1;
			

			}
			

			

			

			synchronized ( o2 ) {
			

			data.i = x;
			

			data.j = y;
			

			}
			

			

			

			2) lock-merging
			

			

			

			synchronized ( obj ) {
			

			data.y = 1;
			

			synchronized ( obj2 ) {
			

			data.x = 1;
			

			}
			

			}
			

			

			

			synchronized ( obj ) {
			

			synchronized ( obj2 ) {
			

			data.x = 1;
			

			data.y = 1;
			

			}
			

			}
			

			

			3) lock-coarsing
			

			

			synchronized ( obj ) {
			

			data.y = 1;
			

			}
			

			

			synchronized ( obj ) {
			

			data.x = 1;
			

			}
			

			

			

			synchronized ( obj ) {
			

			data.y = 1;
			

			data.x = 1;
			

			}
			

			

			

			4) lock-coarsing #2
			

			

			synchronized ( obj ) {
			

			data.y = 1;
			

			}
			

			

			synchronized ( obj2 ) {
			

			data.x = 1;
			

			}
			

			

			

			synchronized ( obj ) {
			

			synchronized ( obj2 ) {
			

			data.y = 1;
			

			data.x = 1;
			

			}
			

		

Am I right?
Or is there anything wrong?

TIA.

Sincerely,
Min-Koo Seo.