On Tue, 4 Apr 2023 18:42:57 GMT, Chris Plummer <cjplum...@openjdk.org> wrote:
>> test/jdk/com/sun/jdi/ThreadMemoryLeakTest.java line 74: >> >>> 72: // that get queued up, so we need to slow >>> it down a bit more >>> 73: // than we do for platform threads to avoid >>> getting OOME. >>> 74: Thread.sleep(100); >> >> I wonder if this time for sleep can still be not enough. > > I did a lot of testing on all platforms, including with product builds, but > yes, it is possible that on some some platforms with some flags it might not > be enough. I guess more testing will tell. Adjustments might be necessary. It > is important not too slow things down too much, or it's possible that if > there is a memory leak, the test won't catch it because the leak is not fast > enough. With the current sleep values, throughput for virtual threads is > still about 2x what it is for platform threads, so right now I'm not worried > about it having been slowed down too much. small nit. shorter to use: long timeToSleep = "Virtual".equals(mainWrapper) ? 100 : 50; ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/13246#discussion_r1157751992
