Uploaded image for project: 'JDK'
  1. JDK
  2. JDK-8282314

nsk/jvmti/SuspendThread/suspendthrd003 may leak memory

XMLWordPrintable

    • b14

      In my recent stress testing of jdk-19 bits, I've run into
      StressWrapper_suspendthrd003 failing in slowdebug configs
      due to "OutOfMemoryError: Java heap space". The first
      occurrence was in my jdk-19+8 stress testing and the failures
      reproduced in jdk-19+9 and jdk-19+10 stress testing.

      I added an RSS monitor for the StressWrappers to my stress
      kit setup and captured RSS data in my jdk-19+10 stress testing.
      I wrote a script to extract the RSS data and another script to do
      some initial analysis on that data.

      In do_all_check_stress_results.jdk-19+10_1.log:

      StressWrapper_StopAtExit release:
      n_values=101
      max_value=13102752
      min_value=80516
      max_increase=542264
      min_increase=780
      Value increased 100 times
      WARNING: there might be a leak.

      StressWrapper_suspendthrd003 release:
      n_values=102
      max_value=260024
      min_value=43676
      max_increase=17628
      min_increase=8436
      Value increased 101 times
      WARNING: there might be a leak.

      StressWrapper_suspendthrd003 fastdebug:
      n_values=101
      max_value=431132
      min_value=193884
      max_increase=18416
      min_increase=6908
      Value increased 100 times
      WARNING: there might be a leak.

      StressWrapper_suspendthrd003 slowdebug:
      n_values=74
      max_value=1019140
      min_value=203728
      max_increase=29192
      min_increase=13156
      Value increased 73 times
      WARNING: there might be a leak.


      In do_all_check_stress_results.jdk-19+10_2.log:

      StressWrapper_suspendthrd003 release:
      n_values=102
      max_value=326184
      min_value=45896
      max_increase=18260
      min_increase=11096
      Value increased 101 times
      WARNING: there might be a leak.

      StressWrapper_suspendthrd003 fastdebug:
      n_values=102
      max_value=478212
      min_value=193176
      max_increase=18756
      min_increase=5872
      Value increased 101 times
      WARNING: there might be a leak.

      StressWrapper_suspendthrd003 slowdebug:
      n_values=74
      max_value=1029096
      min_value=198076
      max_increase=27360
      min_increase=13056
      Value increased 73 times
      WARNING: there might be a leak.


      do_all_check_stress_results.jdk-19+10_3.log:

      StressWrapper_StopAtExit release:
      n_values=101
      max_value=13230152
      min_value=78128
      max_increase=532040
      min_increase=1764
      Value increased 100 times
      WARNING: there might be a leak.


      StressWrapper_suspendthrd003 release:
      n_values=102
      max_value=173640
      min_value=42908
      max_increase=10428
      min_increase=10428
      Value increased 101 times
      WARNING: there might be a leak.

      StressWrapper_suspendthrd003 fastdebug:
      n_values=101
      max_value=479636
      min_value=193480
      max_increase=18764
      min_increase=5864
      Value increased 100 times
      WARNING: there might be a leak.

      StressWrapper_suspendthrd003 slowdebug:
      n_values=75
      max_value=1032804
      min_value=204916
      max_increase=29068
      min_increase=12572
      Value increased 74 times
      WARNING: there might be a leak.

      Since the possible leak reproduced in all three configs
      in all three runs of StressWrapper_suspendthrd003, I'm
      going to focus on that wrapper first.

      StressWrapper_StopAtExit did report a possible leak in
      the 'release' bits config in 2 of the 3 runs so I'm not as
      sure that is a real sighting.

            dcubed Daniel Daugherty
            dcubed Daniel Daugherty
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

              Created:
              Updated:
              Resolved: