atg failed on solaris x86, the stack trace looks different.
The machine is jtg-lx50-2, logs are under /bt

  ---- called from signal handler with signal 11 (SIGSEGV) ------
  [11] MarkFromRootsClosure::scanOopsInOop(0xdcaefd78, 0xd09513a0), at 0xdd68a39e
  [12] MarkFromRootsClosure::do_bit(0xdcaefd78, 0x3544e8), at 0xdd68a24b
  [13] BitMap::iterate(0x81780a8, 0xdcaefd78, 0x0, 0x1f00000), at 0xdd6524d8
  [14] CMSCollector::markFromRootsWork(0x8178040, 0x1), at 0xdd685ad3
  [15] CMSCollector::markFromRoots(0x8178040, 0x1), at 0xdd685917
  [16] CMSCollector::collect_in_background(0x8178040, 0x0), at 0xdd683930
  [17] ConcurrentMarkSweepThread::run(0x81b83e8), at 0xdd68c81e
  [18] _start(0x81b83e8), at 0xdd58f6a2
  [19] _thr_setup(0xdda60600), at 0xddb64a59
  [20] _lwp_start(), at 0xddb64ce0

This **MAY** be a regression between b21 and b22:
atg failed on solaris x86. The machine is jtg-lx50-2.