On Wed, 27 Jan 2021 14:17:22 GMT, Daniel D. Daugherty <dcu...@openjdk.org> wrote:
>>> I thought the comments regarding the test were to use an if statement to >>> check iter.hasNext() and return after seeing one Monitor, rather than >>> keeping the while-loop and reporting every single Monitor found. The latter >>> seems a waste of timer as we don't know how many Monitors to expect, so it >>> adds no value to report them all. >> >> I thought we should test whether we can get ObjectMonitor via SA at first - >> in other words, the test would pass when we get 1 ObjectMonitor at least. >> However all of information of ObjectMonitors might be useful for diagnostic >> purposes as @dcubed-ojdk said. In this test, we use LingeredAppWithLock for >> debugee, so it will not waste time to iterate. (we might estimate num of >> locks we will observe: we can see 3 ObjectMonitors now) > > The number should be small. I typically see 2 or 3 depending on the java > options. If I happen to see more, then that's a clue (to me anyway) that > something has changed and might be amiss. I ran this patch thru Mach5 Tier[1-3]: - Tier1 - TestObjectMonitorIterate.java executed and passed 4 times - Tier2 - TestObjectMonitorIterate.java does not execute in Tier2 - Tier3 - TestObjectMonitorIterate.java executed and passed 13 times ------------- PR: https://git.openjdk.java.net/jdk/pull/1910
