Uploaded image for project: 'Core ReactOS'
  1. Core ReactOS
  2. CORE-14553

[NTOSKRNL] suspected bug: 0.4.9-dev-709-g953dc72 aka "Dropping the VACB lock" increased thread-count unexpectedly

    XMLWordPrintable

    Details

    • Type: Bug
    • Status: Untriaged
    • Priority: Trivial
    • Resolution: Unresolved
    • Fix Version/s: None
    • Component/s: None
    • Labels:
      None

      Description

      reactOS gcc dbg build, VBox 4.3.40, without VBox additions.
      AC97 was installed and working but is not required to see this.

      With 0.4.9-dev-709-g953dc72 aka "Dropping the VACB lock" the running-thread-count increased unexpectedly.
      Looking at those code-changes does not imply that this is a sane outcome.
      Eventually the "new threads" were existing also before and would have properly terminated then. And now hang with 0.4.9-dev-709 and newer builds.

      The snapshots were done after 4th stage bootup.

      In the past 3years the reported thread-count in taskmgr worked rather reliable for thousands of tries and never gave a delta > 1 thread. The change is significant.
      The reported thread-counts of our taskmgr and the 2k3taskmgr are identical.

      ros-build thread# Handle# process# commit comment
      0.4.9-dev-710-gfd3a6c1 115 1112 17 [NTOSKRNL] Properly reset VACB on free
      0.4.9-dev-709-g953dc72 115 1112 17 [NTOSKRNL] Drop the VACB lock
      0.4.9-dev-708-g40017a5 109 1069 17 [NTOSKRNL] Use interlocked operations...
      0.4.9-dev-705-gb54e5c6 108 1063 17 [NTOS:MM] Do not map two pages... ThFabba

      Taskmgr shows that the unexpected threads do run within the processes eventlog.exe (+5) and services.exe (+2).
      Compare 708a.png to 709a.png
      and 708.png to 709.png

      I doubt those will help, but here are also logs of bootup in 708.log and 709.log

      Ged noted that within an MSVC+WinDbg build the command "!stacks" might help to dump the running threads.

      -------
      Side-note: Handle-count also increased significantly by the same commit. Significantly means here that it exceeds the usual random fluctuation of reported handle-count +-10 that is known for ros since years. I suspect significant part to vanish as well when we fix the thread-count.
      -------

        Attachments

        1. 708.log
          117 kB
        2. 708.png
          708.png
          33 kB
        3. 708a.png
          708a.png
          36 kB
        4. 708crashEventlog_11Threads_40Handles.log
          12 kB
        5. 709.log
          141 kB
        6. 709.png
          709.png
          33 kB
        7. 709a.png
          709a.png
          36 kB
        8. 709crashEventLog_15Threads_52Handles.log
          16 kB
        9. dbghelp_Eventlog.diff
          2 kB

          Issue Links

            Activity

              People

              • Assignee:
                bug zilla Bug Zilla
                Reporter:
                reactosfanboy reactosfanboy
              • Votes:
                0 Vote for this issue
                Watchers:
                1 Start watching this issue

                Dates

                • Created:
                  Updated: