Created attachment 327146 [details] [review]
elementfactory: fix factory leak in test

Created attachment 327147 [details] [review]
fakesink: fix pipeline leak in test

Created attachment 327148 [details] [review]
filesrc: fix buffer leaks in tests

gst_check_setup_sink_pad() internally uses gst_check_chain_func() so we
should call gst_check_drop_buffers() when tearing down tests to free the
buffers which have been exchanged through the pipeline.

Created attachment 327149 [details] [review]
selector: fix pad leaks in tests

setup_input_pad() creates a new pad so we should unref it once we're
done.

Created attachment 327150 [details] [review]
streamiddemux: fix list and event leaks in test

Created attachment 327151 [details] [review]
pipeline: fix bus leak in seek test

gst_bus_add_signal_watch_full() keeps a ref on the bus which should
be released using gst_bus_remove_signal_watch().

(In reply to Guillaume Desmottes from comment #0)
> I kept feeding unit tests to my experimental leaks detection tool and fixed
> a bunch of leaks while doing so. :)

Where can your tool be found? :)

Attachment 327146 [details] pushed as 1e297d7 - elementfactory: fix factory leak in test
Attachment 327147 [details] pushed as 1064bdb - fakesink: fix pipeline leak in test
Attachment 327148 [details] pushed as 7c1a091 - filesrc: fix buffer leaks in tests
Attachment 327149 [details] pushed as d456f8e - selector: fix pad leaks in tests
Attachment 327150 [details] pushed as 0e42f92 - streamiddemux: fix list and event leaks in test
Attachment 327151 [details] pushed as 8af52df - pipeline: fix bus leak in seek test

(In reply to Sebastian Dröge (slomo) from comment #7)
> (In reply to Guillaume Desmottes from comment #0)
> > I kept feeding unit tests to my experimental leaks detection tool and fixed
> > a bunch of leaks while doing so. :)
> 
> Where can your tool be found? :)

"Coming soon" ;)

Any idea why valgrind couldn't catch those?

we make it ignore stuff that's still reachable (e.g. in global variables).