Re: [Linphone-developers] Memory leak on outgoing call

[Simon Morlat]

Hi,

Your proposal is risky: what if the allocated PayloadType is still referenced 
by some other portion of code somewhere else ?
Taking the decision of freeing the PayloadType is only relevant to the 
application.
I prefer the application to check that the payload type slot is empty using 
rtp_profile_get_payload() before calling rtp_profile_set_payload().

Simon

Le Friday 18 April 2008 18:37:31 damico, vous avez écrit :
> Hi all,
>
> I found a little memory leak on outgoing mail: when in
> linphone_payload_is_supported the payload is cloned and then add it to
> profile. So, the function that add the payload to profile in oRTP is
>
> void rtp_profile_set_payload(RtpProfile *prof, int idx, PayloadType *pt){
>     if (idx<0 || idx>=RTP_PROFILE_MAX_PAYLOADS) {
>         ortp_error("Bad index %i",idx);
>         return;
>     }
>     prof->payload[idx]=pt;
> }
>
> That function doesn't care if prof->payload[idx] is yet fill and the
> flags is PAYLOAD_TYPE_ALLOCATED. When we consider the incoming call the
> remote profile is empty and the calls to rtp_profile_set_payload() don't
> generate any memory leak.
>
> I'm proposing to fix it in oRTP by
> void rtp_profile_set_payload(RtpProfile *prof, int idx, PayloadType *pt){
>     PayloadType *payload;
>     if (idx<0 || idx>=RTP_PROFILE_MAX_PAYLOADS) {
>         ortp_error("Bad index %i",idx);
>         return;
>     }
>     payload=rtp_profile_get_payload(prof,idx);
>     if (payload!=NULL && (payload->flags & PAYLOAD_TYPE_ALLOCATED))
>         payload_type_destroy(payload);
>     prof->payload[idx]=pt;
> }
>
>
> Any comments?
>
> Regards
>
> --Michele
>
>
> _______________________________________________
> Linphone-developers mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/linphone-developers