Customer is observing a JVM crash on jdk 5.0u8 linux x86 when exercising the JVMTI RedefineClasses functionality. The test case fails consistently on linux-x86.

Even though the test fails only with RedefinClasses enabled, the stack trace suggests a GC issue (similar to one of the stack traces reported in bug 6337544). 

This problem does not occur on jdk 6.0

Failure stack trace: 

Stack: [0x6dd19000,0x6dd9a000),  sp=0x6dd992d8,  free space=512k
Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
V  [libjvm_g.so+0x4b33d5];;  oopDesc::follow_contents()+0x59
V  [libjvm_g.so+0x4b2060];;  MarkSweep::follow_stack()+0x80
V  [libjvm_g.so+0x551a16];;  PSMarkSweep::mark_sweep_phase1(int&, int)+0x1a2
V  [libjvm_g.so+0x5509f0];;  PSMarkSweep::invoke_no_policy(int*, int)+0x39c
V  [libjvm_g.so+0x564a22];;  PSScavenge::invoke(int*)+0x1d6
V  [libjvm_g.so+0x51ca54];;  ParallelScavengeHeap::failed_mem_allocate(int*, unsigned, int, int)+0x15a
V  [libjvm_g.so+0x640595];;  VM_ParallelGCFailedAllocation::doit()+0xa7
V  [libjvm_g.so+0x63ff7f];;  VM_Operation::evaluate()+0x67
V  [libjvm_g.so+0x63ee8e];;  VMThread::evaluate_operation(VM_Operation*)+0x48
V  [libjvm_g.so+0x63f296];;  VMThread::loop()+0x370
V  [libjvm_g.so+0x63ec53];;  VMThread::run()+0xf7
V  [libjvm_g.so+0x503cf6];;  _start(Thread*)+0x16c
C  [libpthread.so.0+0x4dac]