> On July 10, 2014, 11 a.m., Corey Farrell wrote: > > Have you done leak testing? I suspect the change to __frame_free will > > result in leaks that will be reported by REF_DEBUG. > > opticron wrote: > I have been watching for leaks here. There is nothing major that I see, > but I have some further investigation to do. > > Matt Jordan wrote: > This will definitely cause leaks. Easiest example is in ex_adpcm.h - > frame is statically allocated and bumps the ref count to the format. > > We either need to require static frames to bump the reference count to > formats they use, or not. If not, then static frames have to have a *very* > short lifetime (not all do). So I'm not sure we can get away with this.
That is the exact situation I needed to investigate further. I'll pull this change out of the patch for now. - opticron ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviewboard.asterisk.org/r/3736/#review12540 ----------------------------------------------------------- On July 10, 2014, 11:24 a.m., opticron wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviewboard.asterisk.org/r/3736/ > ----------------------------------------------------------- > > (Updated July 10, 2014, 11:24 a.m.) > > > Review request for Asterisk Developers, Corey Farrell, Joshua Colp, and Matt > Jordan. > > > Bugs: ASTERISK-23960 > https://issues.asterisk.org/jira/browse/ASTERISK-23960 > > > Repository: Asterisk > > > Description > ------- > > This contains the fixes necessary to get a PJSIP call up and playing files > from the Playback() application with a translation path in the mix. > > * channel.c: The set format helper functions were not actually setting the > newly chosen formats resulting in no audio. > * frame.c: The format on the frame was being overzealously unreffed resulting > in a crash. > * sorcery.c: The codec retrieval code was using the wrong level of > indirection for ast_format_cap structures resuting in a crash for "pjsip show > endpoint x" > * translate.c: The chosen codecs were being set backward on set vs native for > what was actually desired causing incorrect codecs to be chosen. > * res_pjsip_sdp_rtp.c: An ast_rtp_codecs struct was not being initialized > properly causing a crash. > > > Diffs > ----- > > team/group/media_formats-reviewed-trunk/res/res_pjsip_sdp_rtp.c 418253 > team/group/media_formats-reviewed-trunk/main/translate.c 418253 > team/group/media_formats-reviewed-trunk/main/sorcery.c 418253 > team/group/media_formats-reviewed-trunk/main/frame.c 418253 > team/group/media_formats-reviewed-trunk/main/channel.c 418253 > > Diff: https://reviewboard.asterisk.org/r/3736/diff/ > > > Testing > ------- > > Call testing. > > > Thanks, > > opticron > >
-- _____________________________________________________________________ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- asterisk-dev mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-dev
