[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: make test fails on test02-glue (core dump) (ITS#2311)



I just wanted to let you know that I finally got this test to pass after
apply Sun patch #111308-03 and now it works with -lmtmalloc ...

Thanks for your help,

Joseph

PS -  This patch did not fix the ch_malloc problem I defined in ITS #2270,
though.

----- Original Message -----
From: <joseph.tingiris@cox.net>
To: <openldap-its@OpenLDAP.org>
Sent: Friday, February 14, 2003 6:14 PM
Subject: Re: make test fails on test02-glue (core dump) (ITS#2311)


> Sorry, thought I included the backtrace ... here it is:
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x414d4520 in ?? ()
> (gdb) bt
> #0  0x414d4520 in ?? ()
> #1  0x32250 in slap_startup (be=0x758c58) at init.c:234
> #2  0x37f60 in slap_tool_init (name=0xffffffff <Address 0xffffffff out of
> bounds>, tool=2, argc=7, argv=0xffbef7d4)
>     at slapcommon.c:281
> #3  0x11e34 in main (argc=7, argv=0xffbef7d4) at slapadd.c:38
> (gdb)
>
> ----- Original Message -----
> From: "Howard Chu" <hyc@highlandsun.com>
> To: "'Joseph Tingiris'" <joseph.tingiris@cox.net>;
> <openldap-its@OpenLDAP.org>
> Sent: Friday, February 14, 2003 5:36 PM
> Subject: RE: make test fails on test02-glue (core dump) (ITS#2311)
>
>
> >
> > > -----Original Message-----
> > > From: Joseph Tingiris [mailto:joseph.tingiris@cox.net]
> >
> > > Ok, I verified this is only happening on a Sun Fire 3800 when
> > > I try to link
> > > it with -lmtmalloc.  When I use the identical environment on
> > > a Sun Fire 280R
> > > and make test it does not core when linked with mtmalloc.
> > > I've double,
> > > triple, and even quadruple checked the environments.  They
> > > are identical
> > > with the exception of the hardware.  Same version of Solaris,
> > > same patches,
> > > same compiler packages and libraries.  Same soft everything.
> >
> > Given the hardware dependency, I suggest you file a bug report with Sun.
> >
> > > Here is the gdb output you requested:
> >
> > I asked for a stack backtrace. That is, when the program crashes in gdb,
> > issue the "bt" command and send us that output. However, I believe it
will
> be
> > a moot point; most likely the trace will show that the crash occurred in
> > libmtmalloc.
> >
> >   -- Howard Chu
> >   Chief Architect, Symas Corp.       Director, Highland Sun
> >   http://www.symas.com               http://highlandsun.com/hyc
> >   Symas: Premier OpenSource Development and Support
> >
>
>
>