Re: Re: Re: Valgrind says gtkmm is leaking

[american communist party gmail com]

On Jun 6, 2010 6:44pm, american communist party gmail com wrote:
> On Jun 6, 2010 4:14pm, Chris Vine chris cvine freeserve co uk> wrote:
> > I haven't looked at all your examples but I suspect your code is OK and
> > there aren't any leaks.  What you are most likely seeing is the effect
> > of single instance library allocations (which are usually not
> > deallocated because there is no reason to do so) and of glib memory
> > slices.
> > This could usefully form the start of a suppressions file.  See the
> > valgrind documentation for further information.  However to get
> > meaningful results you need to run your test program with the
> > environmental variable G_SLICE set to always-malloc.
> > Chris
> 
> With a debug (-ggdb) version of the same sample helloworld app,
> no suppression file;
> 
> $ G_SLICE=always-malloc;export G_SLICE
> $ valgrind --tool=memcheck --leak-check=yes -v ./a.out 2>&1 | tee valgrind.log
> ~~~
> ==6137== LEAK SUMMARY:
> ==6137==    definitely lost: 2,484 bytes in 6 blocks
> ==6137==    indirectly lost: 6,704 bytes in 211 blocks
> ==6137==      possibly lost: 89,034 bytes in 899 blocks
> ==6137==    still reachable: 978,790 bytes in 9,025 blocks
> ==6137==         suppressed: 0 bytes in 0 blocks
> ==6137== Reachable blocks (those to which a pointer was found) are not shown.
> ==6137== To see them, rerun with: --leak-check=full --show-reachable=yes
> ==6137== 
> ==6137== ERROR SUMMARY: 546 errors from 546 contexts (suppressed: 15 from 8)