prospero 48 =>java_g -server -Xrunhprof:cpu=samples -version
#
# HotSpot Virtual Machine Error, assertion failure
# Please report this error at
# http://java.sun.com/cgi-bin/bugreport.cgi
#
# assert(eden()->is_empty() && from()->is_empty() && to()->is_empty(), "DefNewGeneration should be empty")
#
# Error happened during: generation collection for allocation
#
# Error ID: /usr/re/hotspot/hotspot1.4/ws/solsparc/hotspot1.4beta/src/share/vm/memory/generation.cpp, 803 [ Patched ]
#
# Problematic Thread: prio=5 tid=0x105030 nid=0x4 runnable
#


I have seen this with both C2 and C1, although it easily
reproduces with C2 and I have only seen it once with C1.

Stack looks like:

  [1] report_assertion_failure(code_str = 0xfecaec80 "eden()->is_empty() && from()->is_empty() && to()->is_empty()", file_name = 0xfecaecc0 "/net/karenspc/files/merlinfixes/src/share/vm/memory/generation.cpp", line_no = 803, message = 0xfecaed04 "DefNewGeneration should be empty"), line 148 in "debug.cpp"
=>[2] DefNewGeneration::fill_newgen(this = 0xa8278), line 803 in "generation.cpp"
  [3] DefNewGeneration::gc_epilogue(this = 0xa8278, full = 0), line 797 in "generation.cpp"
  [4] GenGCEpilogueClosure::do_generation(this = 0xfbe81864, gen = 0xa8278), line 911 in "genCollectedHeap.cpp"
  [5] GenCollectedHeap::generation_iterate(this = 0xa81c0, cl = 0xfbe81864, old_to_young = 0), line 796 in "genCollectedHeap.cpp"
  [6] GenCollectedHeap::gc_epilogue(this = 0xa81c0, full = 0), line 926 in "genCollectedHeap.cpp"
  [7] GenCollectedHeap::do_collection(this = 0xa81c0, full = 0, clear_all_soft_refs = 0, size = 4096U, large_noref = 0, tlab = 1, max_level = 1, notify_ref_lock = 0), line 349 in "genCollectedHeap.cpp"
  [8] GenCollectedHeap::satisfy_failed_allocation(this = 0xa81c0, size = 4096U, large_noref = 0, tlab = 1, notify_ref_lock = 0), line 369 in "genCollectedHeap.cpp"
  [9] VM_GenCollectForAllocation::doit(this = 0xffbee184), line 111 in "vm_operations.cpp"
  [10] VM_Operation::evaluate(this = 0xffbee184), line 34 in "vm_operations.cpp"
  [11] VMThread::evaluate_operation(this = 0x1041c8, op = 0xffbee184), line 230 in "vmThread.cpp"
  [12] VMThread::loop(this = 0x1041c8), line 306 in "vmThread.cpp"
  [13] VMThread::run(this = 0x1041c8), line 165 in "vmThread.cpp"
  [14] _start(data = 0x1041c8), line 494 in "os_solaris.cpp"

In the situation I looked at, both eden_space and from_space
had _top == _end and _to_space had _bottom == _end.