One of our nightly tests failed on Solaris Nevada build 44 (amd64). It was running java
1.5.0_09 under collect. Here is a part of stack trace reported by dbx:

  ---- called from signal handler with signal 11 (SIGSEGV) ------
  [14] frame::entry_frame_is_first(0xfffffd7fffdfbe40, 0xfffffd7ff700032d, 0xfffffd7fffdfbdb8, 0x6, 0x0, 0xfffffd7ff700032d), at 0xfffffd7ffd7b480c
  [15] frame::is_first_frame(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7ffd76f042
  [16] vframeStreamCommon::fill_from_frame(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7ffd7b8675
  [17] vframeStreamForte::vframeStreamForte(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7ffda53ad7
  [18] forte_fill_call_trace_given_top(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7ffda54419
  [19] AsyncGetCallTrace(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7ffda5472a
  [20] __collector_ext_jstack_unwind(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7fff337b06
  [21] __collector_get_frame_info(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7fff337e92
  [22] __collector_ext_profile_handler(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7fff32e4b8
  [23] collector_sigprof_dispatcher(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7fff32b694
  [24] __sighndlr(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7fff29a426
  [25] call_user_handler(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7fff28e5c2
  [26] sigacthandler(0x0, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfffffd7fff28e78a
  ---- called from signal handler with signal 29 (SIGPROF) ------

I created a tar file, that includes "core" file, and other files that were in the
test directory:

ultima4% ls -l /net/lee2.sfbay/export/home/mpmt/TEST/*.gz
-r--r--r-- 1 nikm staff 5946680 Aug 1 15:13 /net/lee2.sfbay/export/home/mpmt/TEST/ultima4_i386_5.11-jvmlib.FZ_DBG,FZ,FZ_STABS,FZ_JS,J150,JVMPI,B64.tar.gz

Please, see "Comments" section for additional details. Let me know if any additional
information is needed.

I don't think it is a promptly reproducible problem, but we can try to create a
test that will run this command several times (like 100 times or more).