"size of symbol changed" warnings -- what do these mean?

"size of symbol changed" warnings -- what do these mean?

[Kenneth Streeter]

We are getting some unusual warning messages from the WRS
Tornado II version of the g++ compiler, and are hoping that
somebody here could explain in layman's terms what specifically
these messages mean.  We are getting the errors in compiling
the ACE 3rd party library, which contains tens of thousands of
lines of code and we are having difficulty even knowing where
to start.  Knowing exactly what the linker warning we are getting
means would be a big help.
  
Here is the linker output:

  Warning: size of symbol
  `dequeue__t17ACE_Message_Queue1Z14ACE_NULL_SYNCHRP17ACE_Message_BlockP14ACE_Time_Value'
  changed from 56 to 52 in Message_Queue.o

  Warning: size of symbol `activate__t17ACE_Message_Queue1Z14ACE_NULL_SYNCH' changed from
  56 to 52 in Message_Queue.o 

The symbols that the linker is actually complaining about are
function names.  The demangled function names are as follows:

  ACE_Message_Queue<ACE_NULL_SYNCH>::dequeue (ACE_Message_Block *&,
                                              ACE_Time_Value *)

  ACE_Message_Queue<ACE_NULL_SYNCH>::activate (void)

What do the "changed from 56 to 52" statements mean?  If we knew what
was of size 56 and changed to size 52, we might have a possibility of
tracking this error down and fixing it.  (We are concerned about fixing
this, as we seem to be suffering from memory corruption when using
Message_Queues in the resulting executables.)

We are using the Tornado II version of the linker on Solaris
for PPC targets.  (g++ppc cygnus-2.7.2-960126)

Any help or suggestions you could provide would be greatly appreciated.

--ken

-- 
Kenneth B. Streeter     | EMAIL: kenneth.b.streeter@lmco.com
BAE SYSTEMS, MER15-2415 |
PO Box 868              | Voice: (603) 885-1500
Nashua, NH 03061        | Fax:   (603) 885-2356

Re: "size of symbol changed" warnings -- what do these mean?

[andy]

I saw this when I was using  -g  and  -frepo options ( on a Netwinder Linux box )
The error went away when I changed the order of the options in my makefile
Andy

Kenneth Streeter wrote:

> We are getting some unusual warning messages from the WRS
> Tornado II version of the g++ compiler, and are hoping that
> somebody here could explain in layman's terms what specifically
> these messages mean.  We are getting the errors in compiling
> the ACE 3rd party library, which contains tens of thousands of
> lines of code and we are having difficulty even knowing where
> to start.  Knowing exactly what the linker warning we are getting
> means would be a big help.
>
> Here is the linker output:
>
>   Warning: size of symbol
>   `dequeue__t17ACE_Message_Queue1Z14ACE_NULL_SYNCHRP17ACE_Message_BlockP14ACE_Time_Value'
>   changed from 56 to 52 in Message_Queue.o
>
>   Warning: size of symbol `activate__t17ACE_Message_Queue1Z14ACE_NULL_SYNCH' changed from
>   56 to 52 in Message_Queue.o
>
> The symbols that the linker is actually complaining about are
> function names.  The demangled function names are as follows:
>
>   ACE_Message_Queue<ACE_NULL_SYNCH>::dequeue (ACE_Message_Block *&,
>                                               ACE_Time_Value *)
>
>   ACE_Message_Queue<ACE_NULL_SYNCH>::activate (void)
>
> What do the "changed from 56 to 52" statements mean?  If we knew what
> was of size 56 and changed to size 52, we might have a possibility of
> tracking this error down and fixing it.  (We are concerned about fixing
> this, as we seem to be suffering from memory corruption when using
> Message_Queues in the resulting executables.)
>
> We are using the Tornado II version of the linker on Solaris
> for PPC targets.  (g++ppc cygnus-2.7.2-960126)
>
> Any help or suggestions you could provide would be greatly appreciated.
>
> --ken
>
> --
> Kenneth B. Streeter     | EMAIL: kenneth.b.streeter@lmco.com
> BAE SYSTEMS, MER15-2415 |
> PO Box 868              | Voice: (603) 885-1500
> Nashua, NH 03061        | Fax:   (603) 885-2356

Re: "size of symbol changed" warnings -- what do these mean?

[Alexandre Oliva]

On Mar 20, 2001, Kenneth Streeter <kenneth.b.streeter@lmco.com> wrote:

> What do the "changed from 56 to 52" statements mean?

It means there are multiple (possibly weak) definitions of that
symbol, and they have different sizes, which is probably wrong.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me