RFR (S) 8241668: Shenandoah: make ShenandoahHeapRegion not derive from ContiguousSpace

Roman Kennke rkennke at redhat.com
Thu Mar 26 17:04:59 UTC 2020


Hi Aleksey,

Nice cleanup.

I don't think we need block_start() block_size() and block_is_obj()
(ShHeap and ShHeapRegion) for anything. Correct me if I'm wrong.

Other than that, it looks good.

Thanks,
Roman

> RFE:
>   https://bugs.openjdk.java.net/browse/JDK-8241668
> 
> This is similar to what G1 did in the relevant cleanup (JDK-8189737). There is no reason to carry
> the cruft from ContiguousSpace superclass, when we just need to pull down a few fields and auxiliary
> methods from there.
> 
> sizeof(ShenandoahHeapRegion) went down from 328 to 264 bytes.
> 
> Webrev:
>   https://cr.openjdk.java.net/~shade/8241668/webrev.01/
> 
> Testing: hotspot_gc_shenandoah, serviceability/sa with Shenandoah
> 




More information about the hotspot-gc-dev mailing list