On Mon, 2 Dec 2024 18:53:12 GMT, Henry Jen <henry...@openjdk.org> wrote:
>> src/hotspot/share/runtime/threads.cpp line 654: >> >>> 652: initialize_java_lang_classes(main_thread, CHECK_JNI_ERR); >>> 653: // Update the java.vm.vendor property with >>> java.lang.VersionProps.VM_VENDOR >>> 654: Arguments::update_vm_vendor(VM_Version::vm_vendor()); >> >> It seems very ad hoc to update this after initPhase1 has run, need to think >> through if there is a better way to do this. > > I agree. This is to minimize change on behavior. > > Not sure what kind of requirement is there for java.vm.vendor, would it be OK > for VM not the set this value and wait until loading of VersionProps class? > > The update is needed for the assertion in statSampler, I don't see other > reference to the property in the repo. So the question is when we need the > property to be available and exposure scope. I'm a little lost in the sequence of events here as well. Why do we call `override_vm_vendor` in `initialize_java_lang_classes` and then `update_vm_vendor` here? ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/21964#discussion_r1866534288
