JDK-4778176 : Internal error handling illegal instruction exception
  • Type: Bug
  • Component: hotspot
  • Sub-Component: compiler
  • Affected Version: 1.4.1
  • Priority: P2
  • Status: Closed
  • Resolution: Duplicate
  • OS: solaris_8
  • CPU: sparc
  • Submitted: 2002-11-13
  • Updated: 2003-01-17
  • Resolved: 2003-01-17
Related Reports
Duplicate :  
Description
Running BEA WebLogic with a version of the Java PetStore, we see repeated crashes with this stack trace:

=>[1] _lwp_kill(0x0, 0x19, 0x0, 0xff33c000, 0xff386000, 0xff340430), at 0xff31ee1c
  [2] raise(0x6, 0x0, 0x0, 0xffffffff, 0xff3403bc, 0x0), at 0xff2cb900
  [3] abort(0xff33c000, 0x90c7dc70, 0x0, 0x4, 0x0, 0x90c7dc91), at 0xff2b58ec
  [4] os::abort(0x1, 0xff1c15a8, 0x90c7e4f8, 0xff16816d, 0xff16810e, 0xff00), at 0xff0802b8
  [5] report_error(0x90c7e51a, 0xea4, 0xff1d9788, 0xde, 0xff10a2ac, 0xff1562c6), at 0xfef8c26c
  [6] report_fatal(0xff156283, 0x14f, 0xff1562c7, 0x0, 0x0, 0x0), at 0xfef8b924
  [7] SharedRuntime::compute_exception_return_address(0x1, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfeeb4f04
  [8] 0xfac36d18(0xfac8da88, 0x90c7f538, 0x90c7f2d0, 0x0, 0x20, 0x0), at 0xfac36d17
  [9] CompiledCodeSafepointHandler::handle_illegal_instruction_exception(0xd29550, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfee21430
  [10] 0xfac32888(0x0, 0x0, 0x0, 0x14, 0x35, 0xa2e9d5fe), at 0xfac32887
  [11] 0xfaf2b4ec(0x9fa06750, 0xf78cb7a8, 0x0, 0x4, 0xffffffcb, 0xa2e9d63c), at 0xfaf2b4eb
  [12] 0xfaf2d44c(0x9fa06598, 0x9fa06598, 0x9fa06750, 0x0, 0x0, 0x1), at 0xfaf2d44b
  [13] 0xfb08a0e0(0xa2e9d550, 0xc98de338, 0xf69398d8, 0x0, 0xd, 0xfffffffe), at 0xfb08a0df
  [14] 0xfb1c8b38(0xf68865a0, 0x9fa06598, 0xa2e9d028, 0x9fa06750, 0xc98c3878, 0xffffffff), at 0xfb1c8b37
  [15] 0xfb01d06c(0x92cf32d8, 0x10, 0x17, 0x0, 0x2e, 0x9fa06916), at 0xfb01d06b
  [16] 0xfb132228(0xc98c3be0, 0x9fa06598, 0x9fa06750, 0x0, 0x35, 0xa2e9ce9e), at 0xfb132227
  [17] 0xfb1124c4(0xc98c3be0, 0x9fa06598, 0x9fa06750, 0x4, 0x48, 0xa2e9cedc), at 0xfb1124c3
  [18] 0xfaf2d644(0x9fa06598, 0x9fa06598, 0xf6817b78, 0xa2e9cf78, 0x0, 0x1), at 0xfaf2d643
  [19] 0xfb077c38(0xa2e9cdf0, 0xc98de338, 0xf69398d8, 0x0, 0xf6f47940, 0x0), at 0xfb077c37
  [20] 0xfb04cc10(0xc98de338, 0xf69a3250, 0x0, 0x52, 0x11, 0xca0a92c6), at 0xfb04cc0f
  [21] 0xfb1212f4(0xf684e250, 0xf69a5e50, 0xfa3ba000, 0xf684db58, 0x5ffe, 0x90c7fa28), at 0xfb1212f3
  [22] 0xfac46c78(0x90c7fb94, 0x0, 0xf841a908, 0xfac15e30, 0x8, 0x90c7faa8), at 0xfac46c77
  [23] 0xfac0010c(0x90c7fc24, 0x90c7fe98, 0xa, 0xf714e3e0, 0x4, 0x90c7fb38), at 0xfac0010b
  [24] JavaCalls::call_helper(0x90c7fe90, 0x90c7fcf0, 0x90c7fdb0, 0xe8e258, 0xe8e258, 0x90c7fd04), at 0xfed67fd8
  [25] JavaCalls::call_virtual(0xff184000, 0xd058d8, 0x90c7fda4, 0x90c7fda0, 0x90c7fdb0, 0xe8e258), at 0xfee65cb8
  [26] JavaCalls::call_virtual(0x90c7fe90, 0x90c7fe8c, 0x90c7fe80, 0x90c7fe78, 0x90c7fe70, 0xe8e258), at 0xfee78ccc
  [27] thread_entry(0xe8e258, 0xe8e258, 0xb8a7a0, 0xd058d8, 0x2fff88, 0xfee83f60), at 0xfee8afc0
  [28] JavaThread::run(0xe8e258, 0x0, 0x0, 0x0, 0x0, 0x0), at 0xfee83f88
  [29] _start(0xe8e258, 0xfe8d2e00, 0x0, 0x0, 0x0, 0x0), at 0xfee803d4

This occurs on 1.4.1 FCS and 1.4.1_01 (build 01). So far, we have not observed it in mantis build 3. A fix is needed for 1.4.1_02.

This also looks exactly like bug 467904, which was closed as not reproducible. On our lab setup, this is regularly reproducible; contact me for scheduling and details.

Comments
EVALUATION ###@###.### 2002-11-13 Similar failure mode to 4777734 and 4770828 ----- It appears that when attempt to come to a safepoint, the function handle_illegal_instruction_exception is seeing a SEGV of its own because the ThreadCodeBuffer corresponding to the ThreadSafepointState is NULL. Has the ThreadCodeBuffer been released, or has it not been allocated yet? Bug has same failure mode as 4720694. Am not yet closing as a duplicate, but will work on the other as it is a more tractable program. ###@###.### 2003-01-14
14-01-2003