harmony-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ilya Berezhniuk (JIRA)" <j...@apache.org>
Subject [jira] Created: (HARMONY-5434) [drlvm][jit] Crash in reliability.api.kernel.thread.ThreadSuspendResume.ThreadSuspendResume
Date Mon, 28 Jan 2008 02:49:35 GMT
[drlvm][jit] Crash in reliability.api.kernel.thread.ThreadSuspendResume.ThreadSuspendResume
-------------------------------------------------------------------------------------------

                 Key: HARMONY-5434
                 URL: https://issues.apache.org/jira/browse/HARMONY-5434
             Project: Harmony
          Issue Type: Bug
          Components: DRLVM
         Environment: Linux/x86_64
            Reporter: Ilya Berezhniuk


I've got an intermittent  crash in ThreadSuspendResume test on 16-core Linux/x86_64 machine.
It occurs every ~5-10 run for me.
I've checked on 4-core machine - the same crash occured on 35th run.

Crash occurs for me in address 0x00002aaabd473da0; it belongs to memory area "2aaaae568000-2aaabd56b000
---p", i.e. without read/write/execution rights.
Investigation shows that this address is called from the code:
----------------
0x2aaabe473cb5: test   %al,(%rax)
0x2aaabe473cb7: add    %al,(%rax)
0x2aaabe473cb9: add    %al,(%rax)
0x2aaabe473cbb: nop    
0x2aaabe473cbc: callq  0x2aaabe473da0   <=================
0x2aaabe473cc1: add    $0x8,%rsp
0x2aaabe473cc8: mov    %eax,%r12d
----------------

This code belongs to java.util.Random.nextDouble method.
So the crash is probably because of race condition in compilation - or probably in something
else...

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message