JDK-4946823 : CMS: Tomcat and atg failed intermittently with tiger b25
  • Type: Bug
  • Component: hotspot
  • Sub-Component: gc
  • Affected Version: 5.0
  • Priority: P4
  • Status: Closed
  • Resolution: Duplicate
  • OS: generic
  • CPU: generic
  • Submitted: 2003-10-31
  • Updated: 2004-01-26
  • Resolved: 2004-01-07
Related Reports
Relates :  
JDK-4947814 - CMS: assert during mark-word restoration
Description
On j2se-app.west, tomcat and atg failed intermittenly with tiger b25.

atg was run with -d64 -Xcomp, the failure happened after 161 hours.
tomcat was run with -d64. the failure happened after 130 hours.

stack traces from both failures are similiar
  ---- called from signal handler with signal 11 (SIGSEGV) ------
  [10] ScanMarkedObjectsAgainCarefullyClosure::do_object_careful(0x8000000000, 0xffffffff312db738, 0xffffffff2c2ff8b0, 0xffffffff312db738, 0x27, 0x1), at 0xffffffff7e47cc7c
  [11] CompactibleFreeListSpace::object_iterate_mem_careful(0xffffffff312dba00, 0xffffffff2c2ff758, 0xffffffff2c2ff8b0, 0xffffffff312d9b08, 0xffffffff312db738, 0x1), at 0xffffffff7e45f710
  [12] CMSCollector::preclean_card_table(0xffffffff312d9a00, 0x1002e4fb0, 0x10029bd30, 0x10029baa0, 0xffffffff32000000, 0xffffffff312dba00), at 0xffffffff7e477ed4
  [13] CMSCollector::preclean_work(0x1002e4fb0, 0xf800, 0xffffffff7e93999c, 0x1, 0xfb38, 0xffffffff7e8ba000), at 0xffffffff7e477448
  [14] CMSCollector::preclean(0x1002e4fb0, 0x1f4, 0xffffffff7e975368, 0xffffffff7e75ad78, 0xcc00, 0xffffffff7e8ba000), at 0xffffffff7e476c2c
  [15] CMSCollector::collect_in_background(0x1002e4fb0, 0x10, 0x0, 0xffffffff7e4739a8, 0x4, 0x1002e53c8), at 0xffffffff7e473c18
  [16] ConcurrentMarkSweepThread::run(0x0, 0x0, 0xffffffff7e934bf0, 0x1001151c0, 0x7d0, 0xf538), at 0xffffffff7e4813a8
  [17] _start(0x1001151c0, 0x9220, 0x585cbc, 0x100388590, 0x9000, 0xffffffff7e8ba000), at 0xffffffff7e334480


###@###.### 2003-10-30

The failure did not happen with -client and -server.

###@###.### 2003-10-30

I am able to reproduce the failure with build b26 with Tomcat after 47 hours.

###@###.### 2003-11-04

With build 26, the failure happened with atg with -server flag after 4 days.

###@###.### 2003-11-06

With build beta2+4962516 the problem happened with atg with -server flag
after ~1 day. Will reopen this issue in the form of a new bug (#4975054)
and transfer this info into that bug.

#0  0xffffe002 in ?? ()
#1  0x40467b7f in os::message_box ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#2  0x40535766 in VMError::show_message_box ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#3  0x40534d42 in VMError::report_and_die ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#4  0x4046a168 in JVM_handle_linux_signal ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#5  0x404691e4 in signalHandler ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#6  <signal handler called>
#7  0x4020a52c in ScanMarkedObjectsAgainCarefullyClosure::do_object_careful ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#8  0x401ed85d in CompactibleFreeListSpace::object_iterate_mem_careful ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#9  0x402068fe in CMSCollector::preclean_card_table ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#10 0x40205a19 in CMSCollector::preclean_work ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#11 0x402052fa in CMSCollector::preclean ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#12 0x40203511 in CMSCollector::collect_in_background ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#13 0x4020f927 in ConcurrentMarkSweepThread::run ()
   from /usr/j2se/jre/lib/i386/server/libjvm.so
#14 0x4046507f in _start () from /usr/j2se/jre/lib/i386/server/libjvm.so


###@###.### 2004-01-07
Comments
EVALUATION Under investigation; cause of problem not yet known. ###@###.### 2004-01-06: We have not reproduced the problem since the fix for 4959508 went in. However, that does not mean that we know for sure that the bug is fixed (certainly similar crashes seen with GCBahser seem to have disappeared) since the only other time it was seen was after 9 days with tomcat/atg. As such i am marking this a duplicate of 4959508. If future long-lived testing reproduces the problem, please open a new bug with a back-reference to this bug or reopen this bug.
11-06-2004
WORK AROUND try -XX:-CMSPrecleaningEnabled (this seems related to 4959508, but we don't know for sure yet)
11-06-2004