Hi Mandy, > On 15 Feb 2018, at 20:10, mandy chung <mandy.ch...@oracle.com> wrote: > > Hi Robin, > > Do you want a shutdown event for every call to Runtime::exit regardless where > it is in the shutdown sequence? or do you expect to get an event of the > first thread calling JVM_Halt? or the first thread starting the shutdown > sequence but it may not be the thread halting?
I was aiming for an event from the first thread starting the shutdown sequence, as I think that’s probably the one you are looking for if you are diagnosing an unexpected shutdown. Any event generated after that point may not make it into the recording as the tracing framework shuts down from a shutdown hook. If finalizers on exit go away, is it still possible that the thread starting the shutdown sequence isn’t the one eventually calling JVM_Halt (in the absence of calls to Runtime.halt)? Best regards, Robin > > Mandy > > On 2/15/18 5:35 AM, Robin Westberg wrote: >> Hi again, >> >> Here’s the (hopefully) final version: >> >> Full: http://cr.openjdk.java.net/~rwestberg/8041626/webrev.02/ >> <http://cr.openjdk.java.net/%7Erwestberg/8041626/webrev.02/> >> Incremental: http://cr.openjdk.java.net/~rwestberg/8041626/webrev.01-02/ >> <http://cr.openjdk.java.net/%7Erwestberg/8041626/webrev.01-02/> >> >> Best regards, >> Robin >
