RFR: 8327799: JFR view: the "Park Until" field of jdk.ThreadPark is invalid if the parking method is not absolute

Long Yang lyang at openjdk.org
Thu Mar 14 08:30:39 UTC 2024


On Tue, 12 Mar 2024 09:32:28 GMT, Erik Gahlin <egahlin at openjdk.org> wrote:

>> Hi all
>> 
>> Could I have a review of this patch for [JDK-8327799](https://bugs.openjdk.org/browse/JDK-8327799)?
>> 
>> 
>> UNSAFE.park(true, System.currentTimeMillis() + 1000);
>> UNSAFE.park(false, 2000L * 1000 * 1000);
>> 
>> 
>> 
>> jfr view jdk.ThreadPark test.jfr
>> 
>>                                                     Java Thread Park
>> Start Time Duration Event Thread Stack Trace             Class Parked On Park Timeout Park Until Address of Object Pa...
>> ---------- -------- ------------ ----------------------- --------------- ------------ ---------- -----------------------
>> 20:13:21     1.00 s main         jdk.internal.misc.Un... N/A                      N/A 20:13:22   0x00000000
>> 20:13:22     2.00 s main         jdk.internal.misc.Un... N/A                   2.00 s 08:05:43   0x00000000
>> 
>> 
>> If the parking method is not absolute (the second event), the real value of "until" in JFR event is `Long.MIN_VALUE`, which will be convert back to `java.time.Instant.MIN`, but `JFR view` displays this value as '08:05:43' of my timezone. This is somewhat misleading, better to show as `N/A`, just like what `jfr print --events` does.
>> 
>> Testing:
>> test/jdk/jdk/jfr/tool/TestView.java
>> test/jdk/jdk/jfr/jcmd/TestJcmdView.java
>> 
>> All passed.
>> 
>> Thanks
>
> I can sponsor change.

@egahlin Please help sponsor when it's convenient for you. Thanks.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/18214#issuecomment-1996843283


More information about the hotspot-jfr-dev mailing list