Re: [PATCH] fix warning treated as error

[John Levon <levon@mo...>]

On Wed, Apr 04, 2007 at 06:24:00PM -0500, Maynard Johnson wrote:

> >>-	chown(destination.c_str(), buf.st_uid, buf.st_gid);
> >>+	retval = chown(destination.c_str(), buf.st_uid, buf.st_gid);
> >
> >(void) chown(...
> >
> >is the usual trick for this sort of thing. Does gcc work out what that
> >means correctly?
> Nope.  I get the same error as with original source.

Grumble, does tradition mean nothing round here!!

Patch is fine.

thanks
john

Re: [PATCH] fix warning treated as error

[Maynard Johnson <maynardj@us...>]

John Levon wrote:
> On Wed, Apr 04, 2007 at 02:09:56PM -0500, Maynard Johnson wrote:
> 
> 
>>-	chown(destination.c_str(), buf.st_uid, buf.st_gid);
>>+	retval = chown(destination.c_str(), buf.st_uid, buf.st_gid);
> 
> 
> (void) chown(...
> 
> is the usual trick for this sort of thing. Does gcc work out what that
> means correctly?
Nope.  I get the same error as with original source.

-Maynard
> 
> regards
> john

Re: [PATCH] fix warning treated as error

[John Levon <levon@mo...>]

On Wed, Apr 04, 2007 at 02:09:56PM -0500, Maynard Johnson wrote:

> -	chown(destination.c_str(), buf.st_uid, buf.st_gid);
> +	retval = chown(destination.c_str(), buf.st_uid, buf.st_gid);

(void) chown(...

is the usual trick for this sort of thing. Does gcc work out what that
means correctly?

regards
john

[PATCH] fix warning treated as error

[Maynard Johnson <maynardj@us...>]

Problem:  OProfile CVS fails to build when packaged as a src rpm and 
built using rpmbuild.  The compilation error is:
	cc1plus: warnings being treated as errors
	file_manip.cpp: In function 'bool copy_file(
		const std::string&, const std::string&)':
	file_manip.cpp:52: warning: ignoring return value of
		'int chown(const char*, __uid_t, __gid_t)',
		declared with attribute warn_unused_result

This is happening on at least two recent distros I tried (SLES 10 and 
RHEL5) and on multiple architectures (Intel P4, IBM POWER5).

Cause: The call to chown() in libutil++/file_manip.cpp does not catch 
the return value.  A manual build of OProfile doesn't complain about 
this, but rpmbuild does complain because of how the optflags macro is 
set in the rpm build environment.  Running 'rpm --eval "%{optflags}"' on 
the distros mentioned above shows _FORTIFY_SOURCE=2 is defined.  And 
since chown() is declared with __attribute_warn_unused_result__ in 
unistd.h, the warning surfaces when FORTIFY_LEVEL > 0.

Fix:  The attached patch simply stores and ignores the result returned 
from chown().

Regards,
-Maynard

Re: [PATCH] Also load symbols from dynamic symbol table

[Maurice van Swaaij <maurice@bl...>]

I followed the discusion about loading symbols from a dynamic shared 
library.
We are trying to profile an animation package (maya) which is built 
almost entirely from dso's, hence my interest.

Any opinion on if it is worth building my own oprofile with Josef 
Weidendorfer's patch?

Thanks.
-- 

Maurice van Swaaij - Manager Technical Software - Blue Sky Studios NY

    [ http://www.blueskystudios.com | maurice@... ]

callgraph support in oprofile.

[Bikash K Agarwalla <bikash@cc...>]

I installed oprofile, but when I pass callgraph option to opcontrol,
it complains that machine/architecture is not supported.

uname gives "2.6.9-42.0.2.ELsmp #1 SMP Wed Aug 23 00:17:26 CDT 2006 i686
i686 i386 GNU/Linux" output for the machine.

Can someone please confirm if callgraph profiling will work for this
configuration or not.

Thanks,
Bikash Agarwalla