RE: slurpd core dump on exit (ITS#2363)

[bh7506@sbc.com]

Howard Chu,

	My initial tests demonstrate that the core dump no longer occurs on
shutdown.
	Thanks.
	When will these changes make their way into an "official" release?

	Also, during testing of the HEAD I found a problem in the configure
script relating to Berkley DB version tests.
	On line 19210 a comment reads "Requires 4.0 or later" but the test
that follows has a preprocessor condition of:
	if (DB_VERSION_MAJOR >= 4 && DB_VERSION_MINOR >= 1)

	The DB_VERSION_MINOR test should actually be ">=0".
	
	Thanks again.

Benjamin Harristhal
 

-----Original Message-----
From: Howard Chu [mailto:hyc@highlandsun.com] 
Sent: Monday, March 24, 2003 7:58 AM
To: HARRISTHAL, BENJAMIN (SBCSI); openldap-its@OpenLDAP.org
Subject: RE: slurpd core dump on exit (ITS#2363)


This is now fixed in CVS HEAD, please test.

  -- Howard Chu
  Chief Architect, Symas Corp.       Director, Highland Sun
  http://www.symas.com               http://highlandsun.com/hyc
  Symas: Premier OpenSource Development and Support

> -----Original Message-----
> From: owner-openldap-bugs@OpenLDAP.org
> [mailto:owner-openldap-bugs@OpenLDAP.org]On Behalf Of
benjamin.harristhal@sbc.com

>
> Full_Name: Benjamin Harrsithal
> Version: 2.1.12
> OS: AIX 4.3
> URL: ftp://ftp.openldap.org/incoming/
> Submission from: (NULL) (144.160.98.28)
>
>
> I experience a core dump when my slurpd is shutting down on
> an AIX platform.
> I run my LDAP environment across several OS's and several
> platforms.  AIX is the
> only one where this problem occurs.  The core dump occurs
> just about every time
> I shut the slurpd down and it always has the same call stack.
>
> I have done several things in trying to diagnose this problem:
> 1. tested pthread_kill.
> 2. looked for memory leaks using purify on the slurpd process.
> 3. undefined HAVE_PTHREAD_KILL
> and other little tricks.
>
> My results:
> 1. pthread_kill works great in all my tests.
> 2. clean.
> 3. The process cored when it got to pthread_cond_broadcast()
>
> Here is my call stack at the time of the core dump:
> reading symbolic information ...
> [using memory image in core]
>
> Segmentation fault in spin_lock_global_ppc_mp at 0xd0019428
> 0xd0019428 (spin_lock_global_ppc_mp+0x8) 90c40154        stw
>  r6,0x154(r4)
> (dbx) where
> spin_lock_global_ppc_mp() at 0xd0019428
> pthread_kill(??, ??) at 0xd0014b48
> ldap_pvt_thread_kill(0x102, 0x1e), line 167 in "thr_posix.c"
> set_shutdown(0x2), line 164 in "fm.c"
> _vp_sleep(??, ??) at 0xd0012908
> _usched_dispatch(??, ??, ??) at 0xd0010fe8
>
>
> I have found this same behavior in OpenLdap versions
> 2.0.23,2.0.27, and 2.1.12
>