[mythtv-users] mythbackend still eats memory: the current status

[Udo van den Heuvel]

Jan Ceuleers wrote:
> Udo van den Heuvel wrote:
>> Valgrind ran bu the load of mythbackend *AND** valgrind was too much.
> 
> I just ran mythbackend under valgrind, twice, on my Via EN12000 (having 
> shut down mythfrontend and even X and during a time when nothing was 
> scheduled to record), and got the output shown below.

Interesting!
I run the backend mainly at runlevel 3, so the recording is the 
bottlenceck during my tries with valgrind then?

(............)
> ==7734== LEAK SUMMARY:
> ==7734==    definitely lost: 596 bytes in 5 blocks.
> ==7734==    indirectly lost: 64 bytes in 1 blocks.
> ==7734==      possibly lost: 1,440 bytes in 10 blocks.
> ==7734==    still reachable: 975,140 bytes in 23,367 blocks.
> ==7734==         suppressed: 0 bytes in 0 blocks.
> ==7734== Reachable blocks (those to which a pointer was found) are not 
> shown.
> ==7734== To see them, rerun with: --leak-check=full --show-reachable=yes

So indeed stuff is going on, without even going into my own setup.
Could we somehow make valgrind/Mythbackend work OK with a recording(s) 
going on?
Or is that really too much for the EN12000?

Thanks,
Udo