This error occurs in a stress test using the Oracle JDBC thin driver and other test code. It is not practical to provide a compact test case. Note this is pure Java code.


The failure did not reappear when memory limits were set higher.

@ #  SIGSEGV (0xb) at pc=0xb6eba2ba, pid=17006, tid=1833708464
@ #
@ # Java VM: Java HotSpot(TM) Server VM (1.5.0_02-b09 interpreted mode)
@ # Problematic frame:
@ # V  [libjvm.so+0x3a92ba]


Here is the result of a GDB where command.

#6  <signal handler called>
#7  0xb6eba2ba in MarkSweep::follow_stack ()
   from /ade_autofs/ade_linux/DBJAVA_MAIN_LINUX.rdd/050512/jdk15/jre/lib/i386/server/libjvm.so
#8  0xb6f27cde in PSMarkSweep::mark_sweep_phase1 ()
   from /ade_autofs/ade_linux/DBJAVA_MAIN_LINUX.rdd/050512/jdk15/jre/lib/i386/server/libjvm.so
#9  0xb6f26fee in PSMarkSweep::invoke_no_policy ()
   from /ade_autofs/ade_linux/DBJAVA_MAIN_LINUX.rdd/050512/jdk15/jre/lib/i386/server/libjvm.so
#10 0xb6f2b875 in PSScavenge::invoke ()
   from /ade_autofs/ade_linux/DBJAVA_MAIN_LINUX.rdd/050512/jdk15/jre/lib/i386/server/libjvm.so
#11 0xb6efdb1f in ParallelScavengeHeap::failed_mem_allocate ()

We can provide the hs_err_pid*.log file, the core file and the full text of the gdb where command
###@###.### 2005-06-09 20:57:19 GMT


###@###.### 2005-06-10 17:14:25 GMT