On Thu, 8 May 2025 11:22:30 GMT, Alan Bateman <[email protected]> wrote:
> Implementation changes for [JEP 500: Prepare to Make Final Mean > Final](https://openjdk.org/jeps/500). > > Field.set (and Lookup.unreflectSetter) are changed to allow/warn/debug/deny > when mutating a final instance field. JFR event recorded if final field > mutated. Spec updates to Field.set, Field.setAccessible and Module.addOpens > to align with the proposal in the JEP. > > HotSpot is updated to add support for the new command line options. To aid > diagnosability, -Xcheck:jni reports a fatal error when a mutating a final > field with JNI, and -Xlog:jni=debug can help identity when JNI code mutates > finals. For now, JNI code is allowed to set the "write-protected" fields > System.in/out/err, we can re-visit once we change the > System.setIn/setOut/setErr methods to not use JNI (I prefer to keep this > separate to this PR because there is a small startup regression to address > when changing System.setXXX). > > There are many new tests. A small number of existing tests are changed to run > /othervm as reflectively opening a package isn't sufficient. Changing the > tests to /othervm means that jtreg will launch the agent with the command > line options to open the package. > > Testing: tier1-6 Hotspot changes are okay in principle but the JNI checks could be more streamlined to avoid duplication. There is also a more general cleanup of `Klass*` versus `InstanceKlass*` that simplifies things somewhat, though much of that would be external to this PR. src/hotspot/share/prims/jni.cpp line 1870: > 1868: } > 1869: > 1870: static void log_debug_if_final_static_field(JavaThread* thread, const > char* func_name, Klass* klass, int offset) { If `thread` is required to be the current thread, please call it `current`. src/hotspot/share/prims/jni.cpp line 1873: > 1871: if (log_is_enabled(Debug, jni)) { > 1872: fieldDescriptor fd; > 1873: bool found = > InstanceKlass::cast(klass)->find_field_from_offset(offset, true, &fd); If you are assuming/expecting an `InstanceKlass` then that should be the type of the parameter. (The existing code needs a cleanup in this area too but that is outside the scope of this PR.) src/hotspot/share/prims/jniCheck.cpp line 1224: > 1222: WRAPPER_GetField(jdouble, Double, T_DOUBLE) > 1223: > 1224: static void checkCanSetInstanceField(JavaThread* thr, jfieldID fid, > jobject obj) { There is a lot of duplication here with logic already performed in `checkInstanceFieldID` - can you not just pass an extra argument `is_set` to that and include the new checks there? And similarly for the static case. src/hotspot/share/runtime/fieldDescriptor.cpp line 53: > 51: // write protected fields (JLS 17.5.4) > 52: if (is_final() && is_static() && ik == vmClasses::System_klass() && > 53: (offset() == java_lang_System::in_offset() || offset() == > java_lang_System::out_offset() || offset() == > java_lang_System::err_offset())) { I thought the ability to mutate these fields was restricted to the native implementations of `setIn` et al. and was not allowed via regular reflection? Otherwise how can these fields be considered as "trusted finals"?? test/jdk/java/lang/reflect/Field/mutateFinals/jni/JNIAttachMutatorTest.java line 29: > 27: * @summary Test native thread attaching to the VM with JNI > AttachCurrentThread and directly > 28: * invoking Field.set to set a final field > 29: * @requires (os.family == "linux" | os.family == "mac") This test should work for any non-Windows platform using pthreads. ------------- Changes requested by dholmes (Reviewer). PR Review: https://git.openjdk.org/jdk/pull/25115#pullrequestreview-3256235026 PR Review Comment: https://git.openjdk.org/jdk/pull/25115#discussion_r2371150541 PR Review Comment: https://git.openjdk.org/jdk/pull/25115#discussion_r2371151309 PR Review Comment: https://git.openjdk.org/jdk/pull/25115#discussion_r2371182004 PR Review Comment: https://git.openjdk.org/jdk/pull/25115#discussion_r2371207094 PR Review Comment: https://git.openjdk.org/jdk/pull/25115#discussion_r2371211878
