JDK-6754812 : FrameGC test sometimes fails with jvm crash
  • Type: Bug
  • Component: client-libs
  • Sub-Component: java.awt
  • Affected Version: 6u10
  • Priority: P2
  • Status: Closed
  • Resolution: Cannot Reproduce
  • OS: windows_vista
  • CPU: x86
  • Submitted: 2008-10-01
  • Updated: 2011-01-19
  • Resolved: 2008-10-30
Related Reports
Relates :  
Description
JDK              : JDK 6u10b31 b32
Platforms[s]     : x64-winvista-jvm64
testbase         : /net/vice.russia.sun.com/export/home0/dtf/dtf_ws/suites/6.0/awt

6u10b32: FrameGC test fails with crash on x64-winvista-jvm64
Results:
http://stt-13.russia/results/1.6.0_10/b32_j4b/awt/x64-winvista-jvm64/awt-awt_stt_suites_winvista-amd642008-09-16-22-36-26/Administrator.Windows_Vista.amd64/FramesGC/
crash log:
http://stt-13.russia/results/1.6.0_10/b32_j4b/awt/x64-winvista-jvm64/awt-awt_stt_suites_winvista-amd642008-09-16-22-36-26/hs_err_pid9716.log

6u10b32: FrameGC test fails without crash on x64-winvista-jvm64
Results:
http://stt-13.russia/results/1.6.0_10/b32_j4b/awt/x64-winvista-jvm64/rerun/x64-winvista-jvm64/awt-awt_stt_suites_winvista-amd642008-09-22-16-27-49/Administrator.Windows_Vista.amd64/FramesGC/

6u10b31: FrameGC test fails with crash  on x64-winvista-jvm64
Results:
http://stt-13.russia/results/1.6.0_10/b32_j4b/awt/x64-winvista-jvm64/rerun_on_1.6.0_10b31/x64-winvista-jvm64/awt-awt_stt_suites_winvista-amd642008-09-28-21-01-29/Administrator.Windows_Vista.amd64/FramesGC/
crash log:
http://stt-13.russia/results/1.6.0_10/b32_j4b/awt/x64-winvista-jvm64/rerun_on_1.6.0_10b31/x64-winvista-jvm64/awt-awt_stt_suites_winvista-amd642008-09-28-21-01-29/hs_err_pid3764.log

Comments
EVALUATION Closing as "not reproducible" in latest 6u10.
30-10-2008

EVALUATION It's possible that the crash is a side effect of the fix for 6722527 (Need manifest for every exe file in jre/bin directory) committed in 6u10 b31. The fix modifies the manifests of the jre binaries and puts AWT into the inconsistent state. The latest build 6u10 b33 includes fix for the problem - 6748350 (REGRESSION:Eclipse does not pick up the correct theme for Vista) and I couldn't reproduce the crash using both 6u10 b33 and latest 6u12 build. Can you please check if the test still fails with the crash using 6u10 b33 / latest 6u12 build?
20-10-2008

EVALUATION The stack trace of the crash is Java frames: (J=compiled Java code, j=interpreted, Vv=VM code) j java.awt.Cursor.finalizeImpl(J)V+0 j java.awt.Cursor.access$000(J)V+1 j java.awt.Cursor$CursorDisposer.dispose()V+13 j sun.java2d.Disposer.run()V+26 j java.lang.Thread.run()V+11 v ~StubRoutines::call_stub The issue is a duplicate of 6507549, the fix for 6507549 has been backport'ed to 6u4, need to find out why the bug still exists in 6u10.
02-10-2008