Uploaded image for project: 'JDK'
  1. JDK
  2. JDK-6809798

SafePointScalarObject node placed into incorrect block during GCM

    Details

    • Type: Bug
    • Status: Resolved
    • Priority: P4
    • Resolution: Fixed
    • Affects Version/s: hs14
    • Fix Version/s: hs15
    • Component/s: hotspot
    • Labels:
      None
    • Subcomponent:
    • Resolved In Build:
      b03
    • CPU:
      generic
    • OS:
      generic

      Backports

        Description

        Running server VM fastdebug build with -XX:+DoEscapeAnalysis could lead to:

        # To suppress the following error report, specify this argument
        # after -XX: or in .hotspotrc: SuppressErrorAt=\sched.cpp:2293
        #
        # A fatal error has been detected by the Java Runtime Environment:
        #
        # Internal Error (c:\...\sources\hotspot\src\share\vm\opto\sched.cpp:2293), pid=4076, tid=2156
        # Error: assert(prev_selected_node != chosen_node,"Trace_Sched::select_node is in an infinite loop")
        #

        The code inspection has shown that the bug is in common code.
        This assert happened in the new port sources.

        Also it is not a big problem - the only condition to generate correct code
        is to have SafePointScalarObject node referencing only one Safepoint node
        for which it was created. And this is why there was the fix to make it pinned.

          Attachments

            Issue Links

              Activity

                People

                • Assignee:
                  kvn Vladimir Kozlov
                  Reporter:
                  epavlova Ekaterina Pavlova
                • Votes:
                  0 Vote for this issue
                  Watchers:
                  0 Start watching this issue

                  Dates

                  • Created:
                    Updated:
                    Resolved:
                    Imported:
                    Indexed: