RFR: 8365885: Clean up constant pool reflection native code

David Holmes dholmes at openjdk.org
Tue Aug 26 07:04:38 UTC 2025


On Wed, 20 Aug 2025 21:29:29 GMT, Chen Liang <liach at openjdk.org> wrote:

> When I was trying to reuse this constant pool reflection for assembly phase indy argument validation, I noted the JNI code has a lot of confusing arguments. In particular, the JVM_ConstantPoolGetSize is wrong because of argument confusion. We should remove the unused arguments to reduce confusion.

Wow that was such a mess! It looks like the native methods were intended to be static - hence the initial unused argument. But how did this even work when the jvm.cpp implementations reversed the arguments ?? One was an instance of `ConstantPool` and the other the VM set `constantPoolOop`. !!

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

PR Comment: https://git.openjdk.org/jdk/pull/26870#issuecomment-3222886432


More information about the hotspot-dev mailing list