Hi all, When talking with Serguei about JDK-8201655 <https://bugs.openjdk.java.net/browse/JDK-8201655>, we talked about why ThreadHeapSampler has an enabled/disabled when we could have just used the should_post_sampled_object_alloc to begin with.
Could I get a review for this: Webrev: http://cr.openjdk.java.net/~jcbeyler/8211980/webrev.00/ Bug: https://bugs.openjdk.java.net/browse/JDK-8211980 This passed my testing on my dev machine in release and fastdebug. The question I would like to raise here at the same time (in order to reduce email spam and because it should be included in the review I believe) is: - When I did the enable/disable, I used OrderAccess to do so after a reviewer asked for it - From what I can tell, JVMTI_SUPPORT_FLAG does not use it and does instead: #define JVMTI_SUPPORT_FLAG(key) \ private: \ static bool _##key; \ public: \ inline static void set_##key(bool on) { \ JVMTI_ONLY(_##key = (on != 0)); \ NOT_JVMTI(report_unsupported(on)); \ } \ inline static bool key() { \ JVMTI_ONLY(return _##key); \ NOT_JVMTI(return false); \ } Should it (ie in a future bug/webrev)? Thanks, Jc
