[jdk21u-dev] RFR: 8343377: Performance regression in reflective invocation of native methods

[Ayman]

On Fri, 6 Dec 2024 17:01:13 GMT, Chen Liang <liach at openjdk.org> wrote:

>> The reimplementation of core reflection with Method Handles blanket-forced all native methods to use the native accessor due to bootstrapping issues. However, these are already covered by a `VM.isJavaLangInvokeInited()` check, and this blanket enforcement causes performance regression in reflective invocation of native methods like `Object::hashCode` or `Thread::currentThread`.
>> 
>> The patch is not clean due to the comment update to `useNativeAccessor` - it edits a previously added comment not in JDK 21. The code changes are all clean.
>
> Can anyone help run higher-tier tests besides those available from GHA? I will write an approval request after.

Hello @liach, Is there a reason why this PR got closed?

-------------

PR Comment: https://git.openjdk.org/jdk21u-dev/pull/1207#issuecomment-3049072502