Created attachment 345101 [details] [review]
validate: pad-monitor: fix caps leaks

These caps were not released when disposing the pad monitor.

Created attachment 345102 [details] [review]
validate: fix leaks in pad monitor test

- monitors were never released
- reports were leaked
- GstValidateMediaDescriptor was leaked
- caps were leaked: gst_check_setup_events_with_stream_id() and
  gst_event_new_caps() don't consume the caps
- srcpad were never released

Created attachment 345103 [details] [review]
validate: fix leak in overrides test

The runner was never released.

Created attachment 345104 [details] [review]
valdate: fix element leak in overide-registry

The elements were never released once created.

Review of attachment 345101 [details] [review]:

OK

Attachment 345101 [details] pushed as 4c5ba39 - validate: pad-monitor: fix caps leaks
Attachment 345102 [details] pushed as 2b33489 - validate: fix leaks in pad monitor test
Attachment 345103 [details] pushed as ca7f091 - validate: fix leak in overrides test
Attachment 345104 [details] pushed as 1cebf3a - valdate: fix element leak in overide-registry

I am still getting:

  1 =================                                                                                                                                                                                              
Test name: check.gst-devtools.validate.padmonitor.eos_without_segment
Command: '/home/thiblahute/devel/gstreamer/gst-build/build/subprojects/gst-devtools/validate/tests/check/validate/padmonitor'
=================

Running suite(s): padmonitor
 0%: Checks: 1, Failures: 1, Errors: 0 
 ../subprojects/gst-devtools/validate/tests/check/validate/padmonitor.c:49:F:padmonitor:eos_without_segment:0: bin (0x5654990980b0) refcount is 2 instead of 1


Unexpected critical/warning: Leaks detected
 ../subprojects/gstreamer/libs/gst/check/libcheck/check_msg.c:79: No messaging setup
~

Humm weird, it's all fine here:

$ GST_DEBUG="GST_TRACER:7" GST_TRACERS="leaks" /home/cassidy/dev/meson/mesontest.py -C build --suite  gst-devtools 
ninja: Entering directory `/home/cassidy/dev/gst/master/gst-build/build'
ninja: no work to do.
1/5 gst-devtools / validate/padmonitor      OK       7.81 s
2/5 gst-devtools / validate/monitoring      OK       2.03 s
3/5 gst-devtools / validate/reporting       OK       4.96 s
4/5 gst-devtools / validate/overrides       OK       2.17 s
5/5 gst-devtools / validate/launcher_tests  OK       1.12 s

OK:         5
FAIL:       0
SKIP:       0
TIMEOUT:    0

(In reply to Guillaume Desmottes from comment #8)
> Humm weird, it's all fine here:
> 
> $ GST_DEBUG="GST_TRACER:7" GST_TRACERS="leaks"
> /home/cassidy/dev/meson/mesontest.py -C build --suite  gst-devtools 
> ninja: Entering directory `/home/cassidy/dev/gst/master/gst-build/build'
> ninja: no work to do.
> 1/5 gst-devtools / validate/padmonitor      OK       7.81 s
> 2/5 gst-devtools / validate/monitoring      OK       2.03 s
> 3/5 gst-devtools / validate/reporting       OK       4.96 s
> 4/5 gst-devtools / validate/overrides       OK       2.17 s
> 5/5 gst-devtools / validate/launcher_tests  OK       1.12 s
> 
> OK:         5
> FAIL:       0
> SKIP:       0
> TIMEOUT:    0

It is racy it seems.

Just ran checks forever in a loop for 8 minutes with leaks tracer, without problems. Double-checked tracer was active.

In any case, patches have been merged, so let's close this.

If you can still reproduce this, let's make a new bug for that leak, with details.