RFR 8245124: Shenandoah: optimize code root evacuation/update during concurrent class unloading
zgu at redhat.com
Mon May 18 18:03:45 UTC 2020
>> Updated: http://cr.openjdk.java.net/~zgu/JDK-8245124/webrev.01/index.html
> Good. Any reason why ShenandoahEvacOOMScope is not inside heal_nmethod_metadata()? Is there any
> problem to call _bs->disarm(nm) when holding that scope?
This is something I want to get rid of, especially on code root
unlinking path. Moving ShenandoahEvacOOMScope up is another noticeable
improvement, unfortunately, unload_nmethod_caches() triggers barrier.
More information about the hotspot-gc-dev