On Thu, 17 Oct 2024 06:29:05 GMT, Jan Lahoda <jlah...@openjdk.org> wrote:
>> src/java.base/share/native/libjli/java.c line 1935: >> >>> 1933: NULL_CHECK(printConciseUsageMessage = >>> (*env)->GetStaticMethodID(env, cls, >>> 1934: "printConciseUsageMessage", >>> "(Z)V")); >>> 1935: (*env)->CallStaticVoidMethod(env, cls, printConciseUsageMessage, >>> printTo); >> >> Should we have a exception check here after the `CallStaticVoidMethod` >> returns? I realize that several places in the current launcher code, in this >> file, don't have those checks, but the JNI specific of >> `CallStaticVoidMethod` does note that this function could raise any >> exception thrown by the underlying Java method? > > I see `CHECK_EXCEPTION_LEAVE(1);` in the caller of this method. Given we > could only do `CHECK_EXCEPTION_RETURN();`, which returns if an exception is > thrown (so that the caller can handle the exception as they see fit), it > seems unnecessary for last calls. As for last calls, it is "return if an > exception is thrown, and return if it is not". Note it is not only this > function, `PrintJavaVersion`, `ShowSettings`, `ShowResolvedModules`, > `ListModules` and maybe other use the same pattern, where the caller handles > the exceptions, and the functions don't check the exceptions for the last > call. That's a good point about this being a last call before the control returns out of this function. I had missed that part. What you note about not needing an exception check here, makes sense. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/21411#discussion_r1804192084
