Uploaded image for project: 'JDK'
  1. JDK
  2. JDK-6814467

G1: small fixes related to concurrent marking verboseness

    Details

    • Type: Bug
    • Status: Resolved
    • Priority: P4
    • Resolution: Fixed
    • Affects Version/s: hs15
    • Fix Version/s: hs15
    • Component/s: hotspot
    • Labels:
    • Subcomponent:
      gc
    • Resolved In Build:
      b03
    • CPU:
      generic
    • OS:
      generic

      Backports

        Description

        This CR covers three very small fixes all of which cause some strangeness to the verbose GC output:

        a) there's one occurence in concurrentMarkThread.cpp where tty is used, instead of gclog_or_tty, when printing verbose GC output

        b) doConcurrentMark() was, for some reason, incorrectly called from checkConcurrentMark (this was causing the marking phase to be incorrectly woken up after a Full GC; no marking actually happened, but marking related verbose GC information was printed which was confusing)

        c) some fields of ConcurrentMark (_has_aborted, _restart_for_overflow, _concurrent_marking_in_progress, and _should_gray_objects) were not initialised. In some cases, and in conjuction with b), this was causing the marking thread to output "[GC ocnurrent-mark-restart-for-overflow]" several times a second and was bloating the GC log (which would rapidly reach the 2GB limit).

          Attachments

            Issue Links

              Activity

                People

                • Assignee:
                  tonyp Tony Printezis
                  Reporter:
                  tonyp Tony Printezis
                • Votes:
                  0 Vote for this issue
                  Watchers:
                  1 Start watching this issue

                  Dates

                  • Created:
                    Updated:
                    Resolved:
                    Imported:
                    Indexed: