When starting Weblogic 6.1 beta, JVM core dumps during Weblogic's initialization. As Weblogic attempts to make a JDBC connection, the VM
dies and leaves this stack trace:

=>[1] _lwp_kill(0x0, 0x9, 0x0, 0xff33800c, 0x1975c, 0xff2caf90), at 0xff31b0ec
  [2] raise(0x6, 0x0, 0x0, 0xffffffff, 0xff33c404, 0xc), at 0xff2caf98
  [3] abort(0xff33800c, 0xf21fdbd8, 0x0, 0x4, 0x0, 0xf21fdbf9), at 0xff2b5774
  [4] os::abort(0x1, 0xff0b034e, 0xf21fdc78, 0x0, 0xff133b24, 0xfefb819c), at 0xfefb96b8
  [5] os::handle_unexpected_exception(0xd9c70, 0xb, 0xfeccf284, 0xf21fe9b8, 0xfee1bbe8, 0x0), at 0xfefb820c
  [6] JVM_handle_solaris_signal(0xfeccf284, 0x1, 0xf21fe700, 0x1, 0xfeccf288, 0x0), at 0xfee1c45c
  [7] __sighndlr(0xb, 0xf21fe9b8, 0xf21fe700, 0xfee1baec, 0x0, 0x0), at 0xff3808a0
  [8] call_user_handler(0xc, 0xfee1baec, 0xf21fe700, 0xf21fe9b8, 0xb, 0x0), at 0xff37dc4c
  [9] sigacthandler(0xb, 0xf21fe9b8, 0xf21fe700, 0xcb4f4, 0x800, 0x46), at 0xff37de04
  ---- called from signal handler with signal 11 (SIGSEGV) ------
  [10] PhaseIdealLoop::build_loop_late(0xf21ff018, 0xf21fed48, 0xf21fed48, 0xf21fed38, 0x0, 0x0), at 0xfeccf284
  [11] PhaseIdealLoop::build_loop_late(0xf21ff018, 0xf21fed48, 0xf21fed48, 0xf21fed38, 0x0, 0x715c), at 0xfeccee50
  [12] PhaseIdealLoop::build_loop_late(0xf21ff018, 0xf21fed48, 0xf21fed48, 0xf21fed38, 0x0, 0x280), at 0xfeccee50
  [13] PhaseIdealLoop::build_loop_late(0xf21ff018, 0xf21fed48, 0xf21fed48, 0xf21fed38, 0x0, 0x7e8c3c), at 0xfeccee50
  [14] PhaseIdealLoop::build_loop_late(0xf21ff018, 0xf21fed48, 0xf21fed48, 0xf21fed38, 0x0, 0x7e8c3d), at 0xfeccee50
  [15] PhaseIdealLoop::build_loop_late(0xf21ff018, 0xf21fed48, 0xf21fed48, 0xf21fed38, 0x0, 0x7e8e3d), at 0xfeccee50
  [16] PhaseIdealLoop::build_loop_late(0xf21ff018, 0xf21fed48, 0xf21fed48, 0xf21fed38, 0x0, 0x7e8c3d), at 0xfeccee50
  [17] PhaseIdealLoop::PhaseIdealLoop(0x0, 0x0, 0x1, 0x0, 0x2, 0x0), at 0xff06b198
  [18] Compile::Optimize(0xf21ff628, 0xf21ff404, 0xf21ff544, 0x0, 0x0, 0x0), at 0xfedebdc4
  [19] Compile::Compile(0x358b3c, 0x9577f8, 0x0, 0x894a8, 0xba, 0x1), at 0xff03d83c
  [20] C2Compiler::compile_method(0x296b8, 0xf21ffd88, 0x0, 0x894a8, 0xba, 0x0), at 0xff03428c
  [21] CompileBroker::invoke_compiler_on_method(0x0, 0xba, 0xff12a23c, 0x894a8, 0xff1365e8, 0xd9c70), at 0xfede85e8
  [22] CompileBroker::compiler_thread_loop(0xd9c70, 0xd9c70, 0x75688, 0xda1b0, 0x2f30ec, 0xfee292d4), at 0xfee77f9c
  [23] JavaThread::run(0xd9c70, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfee292fc
  [24] _start(0xd9c70, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfee26314

This happens only in build 69; previous builds did not exhibit this problem.
The VM arguments are -server -Xmx512m; this bug doesn't happen if -client is
used.

Reproducing this requires an installation of Weblogic 6.1 that is configured to create a JDBC connection pool. I've set up such an installation in
/home/sdo/bea/wlserver6.1_beta/config/mydomain that can be used to track this
down (make any changes in /home/sdo/bea/- that are necessary). To run Weblogic, cd to that directory and execute the startWebLogic.sh script. After a few seconds, the VM will core dump; if it were to succeed in its initialization, it would print out the line

<Jun 27, 2001 4:09:56 PM PDT> <Notice> <WebLogicServer> <ListenThread listening on port 7001>

Core dumps are left in /home/sdo/bea/wlserver6.1_beta.