From owner-freebsd-questions  Wed May 16 14:31:22 2001
Delivered-To: freebsd-questions@freebsd.org
Received: from MPI-Softtech.Com (mpi.mpi-softtech.com [208.60.120.177])
	by hub.freebsd.org (Postfix) with ESMTP id CB53B37B423
	for <freebsd-questions@FreeBSD.ORG>; Wed, 16 May 2001 14:31:15 -0700 (PDT)
	(envelope-from dleimbac@MPI-Softtech.Com)
Received: from mpi.mpi-softtech.com (mpi.mpi-softtech.com [208.60.120.177]); 
          by MPI-Softtech.Com (8.9.3/8.9.3/MPI-Softtech/evision: 1.3 $) with SMTP; 
          id QAA04973; Wed, 16 May 2001 16:31:05 -0500 (CDT)
Message-Id: <200105162131.QAA04973@MPI-Softtech.Com>
Date: Wed, 16 May 2001 16:31:05 -0500 (CDT)
From: Dave Leimbach <dleimbac@MPI-Softtech.Com>
Reply-To: Dave Leimbach <dleimbac@MPI-Softtech.Com>
Subject: Re: mproff/c++
To: freebsd-questions@FreeBSD.ORG, kmarx@bigshed.com
Cc: kmarx@ankimo.bigshed.com
MIME-Version: 1.0
Content-Type: TEXT/plain; charset=us-ascii
Content-MD5: zGwqt4PMWXwzKjrlFAiXbg==
X-Mailer: dtmail 1.3.0 CDE Version 1.3 SunOS 5.7 sun4u sparc 
Sender: owner-freebsd-questions@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

I have a guess....

C++ uses name mangling to achieve the ability known as overloading. 

This name mangling is applied to the symbols that represent function names in 
your object file.  C doesn't have overloading capabilities and thusly doesn't 
mangle the names.

Its possible that mproff only works with non-mangled symbols for function names.

If this is the case you can declare all of your functions in C++ to be
extern "C".

That will effectively disable name mangling.

I have never used mproff so I can't give you a definitive answer.

Dave
>Delivered-To: freebsd-questions@freebsd.org
>Subject: mproff/c++
>To: freebsd-questions@FreeBSD.ORG (questions freeBSD)
>Date: Wed, 16 May 2001 14:27:58 -0700 (PDT)
>Cc: kmarx@ankimo.bigshed.com (Kenneth J Marx)
>From: Ken Marx <kmarx@bigshed.com>
>MIME-Version: 1.0
>Content-Transfer-Encoding: 7bit
>X-Loop: FreeBSD.ORG
>
>Hi,
>
>Is there some reason why mprof barfs on c++ binaries?
>
>See below for reproduction.
>
>(The man page does reference problems with using
>valloc and memalign, but I put my test binary
>in gdb with breakpoints for these and they
>never got hit. Both these symbols are defined
>in the binary even if I take out the c++ code.)
>
>thanks,
>k.
>
>-- 
>Ken Marx, kmarx@bigshed.com
>OK. This is a tough pill to swallow but we must recognize who our customers are 
>and motivate the troops for the solution space.
>		- http://cgi.bigshed.com/~kmarx/cgi-bin/speak.cgi
>
>------------------------------------------------------------------------------- 
>The source:
>
>[sploot] /tmp <356> cat jnk.c
>#include "stdlib.h"
>
>class dog
>{
>public:
>    dog();
>    ~dog();
>};
>
>dog::dog()
>{
>}
>
>dog::~dog()
>{
>}
>
>main()
>{
>    int i=3;
>    static char *s = "fyuc";
>    char *m = (char *)malloc( 10 );
>    dog *breath = new dog();
>}
>----------------------------------<snip>---------------------------------------
>
>the compile, run, mprof phases:
>
>[sploot] /tmp <357> g++ -g jnk.c -lc_mp
>[sploot] /tmp <358> a.out
>[sploot] /tmp <359> mprof -verbose a.out
>--c20+--v3.0+--m3+--+--+--+--+--+--+--+--+ MPROF 
+--+--+--+--+--+--+--s0+--f0+--d0+--l0+
>
>
>--------- Allocation Bins with possible Types ------------
>
>      size:    allocs     bytes (%)      frees      kept (%)     types   
>
>          1         1         1 (9)          0         1 (9)    dog          
>         10         1        10 (90)         0        10 (90)   
>     > 1024         0         0              0         0        
>
>    <TOTAL>         2        11              1        11        
>
>
>
>--------- Partial Dynamic Call Paths for Memory Leaks ------------
>
>Total bytes not freed: 11
>
>kept bytes (%)     allocs     bytes (%)     path    
>
>        10 (90)         1        10 (90)   || > _start > main 
>Segmentation fault (core dumped)
>
>------------------------------------------------------------------------------- 
>
>[sploot] /tmp <365> wc mprof.data
>    2083    2176    4507 mprof.data
>
>
>To Unsubscribe: send mail to majordomo@FreeBSD.org
>with "unsubscribe freebsd-questions" in the body of the message


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message