RFR (S) 8247296: Optimize JVM_GetDeclaringClass
david.holmes at oracle.com
Fri Jul 24 08:23:07 UTC 2020
Thanks for taking a look.
On 24/07/2020 5:22 pm, Aleksey Shipilev wrote:
> On 7/24/20 5:49 AM, David Holmes wrote:
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8247296
>> webrev: http://cr.openjdk.java.net/~dholmes/8247296/webrev/
> This looks fine to me.
> So we need to be careful that naked oop is still valid at each use, which at least implies there are
> no allocations (and associated safepoints) happen anywhere in between? The patch looks safe in that
Right - no safepoints (or handshakes?) allowed. But these cases are all
okay. Places that can cross safepoints tend to wrap the oop straight
into a Handle to maintain safe access.
More information about the hotspot-runtime-dev