RFR: 8356289: Shenandoah: Clean up SATB barrier runtime entry points
Y. Srinivas Ramakrishna
ysr at openjdk.org
Thu Aug 21 01:59:51 UTC 2025
On Tue, 19 Aug 2025 23:47:34 GMT, Cesar Soares Lucas <cslucas at openjdk.org> wrote:
> The runtime entry points for the SATB slow-paths currently take a JavaThread* argument. When @rkennke did the entry points for Graal ([JDK-8356075](https://bugs.openjdk.org/browse/JDK-8356075)), he figured that we do not really need them. The slow-path is only called rarely (whenever the local SATB buffer is full), and getting the current Thread* in the runtime would not be the expensive part.
>
> This PR is a patch to do that clean up. The changes were tested on Linux x64/aarch64 with JTREG tier1-3 using Shenandoah for all tests.
Marked as reviewed by ysr (Reviewer).
Same; performance impact would be good to know (i.e. that it's performance neutral or may be even a bit better).
-------------
PR Review: https://git.openjdk.org/jdk/pull/26850#pullrequestreview-3138778757
PR Comment: https://git.openjdk.org/jdk/pull/26850#issuecomment-3208682261
More information about the shenandoah-dev
mailing list