Commit graph

1349 commits

Author SHA1 Message Date
Young Han Lee
75205828c7 validate:launcher: Install config.py for non-development mode
Running installed gst-validate-launcher aborted with the following error.

File "lib/gst-validate-launcher/python/launcher/baseclasses.py", line 28, in <module>
  import config
ImportError: No module named config

This is because config.py is added but not installed
in ba6d209b3f.

https://bugzilla.gnome.org/show_bug.cgi?id=747087
2015-03-31 09:29:16 +02:00
Young Han Lee
4725f4ec09 validate:launcher: Show timeout seconds for timeout result message
Current timeout message doesn't show how many seconds a test took and
it is timeouted by normal timeout or hard timeout.

This patch changes the message like following.

1. normal timeout
old : validate.http.playback.reverse_playback.raw_video_mov: Timeout (Application timed out)
new : validate.http.playback.reverse_playback.raw_video_mov: Timeout (Application timed out: 120 secs)

2. hard timeout
old : validate.http.playback.reverse_playback.raw_video_mov: Timeout (Application timed out)
new : validate.http.playback.reverse_playback.raw_video_mov: Timeout (Hard timeout reached: 600 secs)

https://bugzilla.gnome.org/show_bug.cgi?id=746957
2015-03-30 13:32:57 +02:00
Young Han Lee
4c7ec6b8b1 validate:launcher: Fix wrong test number with -j option
When '-j n' option is given, first n tests print test number 0.
This is caused by test_num part of 919db98605.

https://bugzilla.gnome.org/show_bug.cgi?id=747006
2015-03-30 13:02:39 +02:00
Thibault Saunier
ba6d209b3f validate:launcher: Avoid depending on PyGObject
Summary:
And rely on our knowledge of the configuration to figure out where the
suppression file has been installed

Reviewers: gdesmott

Differential Revision: http://phabricator.freedesktop.org/D61
2015-03-28 23:35:07 +01:00
Guillaume Desmottes
625fe7b5df validate: media-descriptor-writer: don't leak info and streaminfo 2015-03-26 15:42:11 +01:00
Guillaume Desmottes
779e44dd36 validate: media-check: don't leak output_file and expected_file 2015-03-26 15:40:50 +01:00
Guillaume Desmottes
4f805422a7 validate: check VALGRIND_ERROR_CODE in Test as well
We were doing it only in GstValidateTest which was overriding the default
implementation.
2015-03-26 13:59:30 +01:00
Guillaume Desmottes
9ad79ffcbc validate: don't increase hard_timeout is if it's None
Some tests, like the media check ones, have None as hard_timeout.
2015-03-26 13:57:34 +01:00
Guillaume Desmottes
daaa32676c validate: increase VALGRIND_TIMEOUT_FACTOR
5 wasn't enough for my poor laptop.
2015-03-26 11:29:26 +01:00
Guillaume Desmottes
66a675cde8 validate: increase the normal timeout as well when using valgrind 2015-03-26 11:29:06 +01:00
Guillaume Desmottes
32eac4db44 validate: media-descriptor: fix caps leak
gst_pad_get_current_caps() returns a reffed caps.
2015-03-26 10:32:09 +01:00
Guillaume Desmottes
093390981c also call gst_deinit() in media-check and transcoding
More valgrind friendly.
2015-03-26 09:18:44 +01:00
Guillaume Desmottes
f1985bc738 validate: raise an error if valgrind detected issues
Differential Revision: http://phabricator.freedesktop.org/D53
2015-03-26 09:18:27 +01:00
Guillaume Desmottes
5456ddbae0 call gst_deinit() when we are done
More valgrind friendly.
2015-03-23 13:41:03 +01:00
Guillaume Desmottes
a35614d108 validate: call gst_bus_remove_signal_watch()
We are supposed to call gst_bus_remove_signal_watch() for each gst_bus_add_signal_watch() call to prevent leaks.
2015-03-23 13:41:03 +01:00
Guillaume Desmottes
6e776b8f0f validate: media-descriptor-parser: fix string leak
_set_content() doesn't actually consume @content so the caller is responsible
freeing it.
2015-03-23 13:41:03 +01:00
Guillaume Desmottes
d9a13b5563 validate: pad-monitor: fix buffers list leak 2015-03-23 13:41:03 +01:00
Guillaume Desmottes
3f0cb5fedf validate: pad-monitor: fix caps leak 2015-03-23 13:41:03 +01:00
Guillaume Desmottes
81a33d5f6a validate: fix a bunch of GstBus leaks 2015-03-23 13:41:03 +01:00
Guillaume Desmottes
c23bac2912 validate: store valgrind logs to its own file 2015-03-23 13:41:03 +01:00
Guillaume Desmottes
2aeaa1b279 validate: report: fix invalid read when destroying Report
Summary:
@report was invalid when we were trying to clear the mutex.

validate: scenario: remove weak pointer when destroying action

Free an invalid read when the scenario is destroyed after the action.

Differential Revision: http://phabricator.freedesktop.org/D44
2015-03-20 15:13:30 +01:00
Guillaume Desmottes
7aa5f85ccd validate: report: fix invalid read when destroying Report
@report was invalid when we were trying to clear the mutex.
2015-03-20 14:49:24 +01:00
Guillaume Desmottes
08a9034576 validate: fix string arguments leaks
We are responsible of freeing the string arguments parsed by GOptionContext.
2015-03-20 12:15:03 +01:00
Guillaume Desmottes
ba175368d1 validate: scenario: don't borrow @structure in _fill_action()
@structure was borrowed in some code path and wasn't in some other. Make it
clearer, and fix a leak, by always copying it.
2015-03-20 11:39:32 +01:00
Guillaume Desmottes
9691136887 validate: override-registry: fix structs list leak
The list returned by _lines_get_strutures() needs to be deeply freed.
2015-03-20 11:33:01 +01:00
Guillaume Desmottes
a99c13da4b validate: scenario: fix scenarios leak 2015-03-20 11:27:29 +01:00
Guillaume Desmottes
38342dfe2e validate: reporter: fix message leak 2015-03-20 11:25:39 +01:00
Guillaume Desmottes
e9aeca28a8 validate: report: don't shadow the GString variable
We were leaking the GString as it's freed outside of the block.
2015-03-20 11:24:27 +01:00
Guillaume Desmottes
3a275d2212 validate: report: fix GStrv leak 2015-03-20 11:24:04 +01:00
Guillaume Desmottes
7373286968 validate: report: fix GStrv leak
We borrow the content of the GStrv but were leaking the array itself.
2015-03-20 11:23:29 +01:00
Guillaume Desmottes
cb8348c7b1 launcher: try using gst.supp as valgrind suppressions file
https://bugzilla.gnome.org/show_bug.cgi?id=746465
2015-03-20 10:34:49 +01:00
Guillaume Desmottes
3024b3682f validate: install gst.supp
Will be used when running tests inside Valgrind.

https://bugzilla.gnome.org/show_bug.cgi?id=746465
2015-03-20 10:34:49 +01:00
Guillaume Desmottes
271f9f3c8e launcher: add valgrind support
Add a --valgrind option to gst-validate-launcher to run the tests inside
Valgrind and tune GLib's memory allocator accordingly.

Fix https://bugzilla.gnome.org/show_bug.cgi?id=746465
2015-03-20 10:34:49 +01:00
Guillaume Desmottes
2778e501c4 validate: move scenarios to validate/scenarios/
https://bugzilla.gnome.org/show_bug.cgi?id=746465
2015-03-20 10:34:48 +01:00
Guillaume Desmottes
e7762d13cc validate:launcher: Fix small typo 2015-03-19 12:22:39 +01:00
Thibault Saunier
ad5b5f5369 validate:launcher: Make sure to show apps specific options in the help 2015-03-18 17:06:14 +01:00
Thibault Saunier
e9da098e33 validate:launcher: Fix test number printing 2015-03-18 17:05:57 +01:00
Thibault Saunier
7079d1910d validate:launcher Rename _other_testsuite_for_tester
To _check_tester_has_other_testsuite
2015-03-17 12:41:26 +01:00
Thibault Saunier
919db98605 validate:launcher: Cache all the tests in the runner
This way we do not have to re ask all the test managers
what tests should be run.
2015-03-17 12:41:26 +01:00
Thibault Saunier
d1411cfc6a validate:launcher: Add a way to simply run validate default tests on uris
Summary:
This allows us to easily run all the scenarios on a particular file doing:

  $ gst-validate-launcher validate --validate-check-uri file:///some/media/file.webm

Reviewers: Mathieu_Du

Differential Revision: http://phabricator.freedesktop.org/D36
2015-03-13 18:04:42 +00:00
Thibault Saunier
b8c085b319 validate: keep executing actions even after linking up following execution
When linking actions execution without waiting on execution context, then
idle callback should keep being called so following action keep being
executed.
2015-03-13 18:04:42 +00:00
Thibault Saunier
e11646dcca validate: Add more files to .gitignore
Differential Revision: http://phabricator.freedesktop.org/D34
2015-03-10 11:32:15 +01:00
Thibault Saunier
49b2ed5d7a validate:scenario: Rename get_position to execute_next_action
That function was wrongly called and did not correspond to what it
actually does.
2015-03-10 11:32:15 +01:00
Vincent Penquerc'h
3bb6ecd6fa Revert "validate: add non flushing seek support"
This reverts commit 3ff55dcc31.

Regressions on the test server, apparently linked to this patchset.
2015-03-09 18:26:37 +00:00
Vincent Penquerc'h
09988a97bf Revert "validate: use segments to detect success of flushing seeks too"
This reverts commit c47cc7ba90.

Regressions on the test server, apparently linked to this patchset.
2015-03-09 18:26:33 +00:00
Vincent Penquerc'h
d78c20322f Revert "validate: expect a buffer with discontinuity after a seek"
This reverts commit 87064b6994.

Regressions on the test server, apparently linked to this patchset.
2015-03-09 18:26:06 +00:00
Thibault Saunier
08d8e01f21 validate:launcher: Make sure TIMEOUTs do not get converted to ERROR
This was a regression introduced in c0e3d2e4f1
2015-03-09 18:41:54 +01:00
Vincent Penquerc'h
87064b6994 validate: expect a buffer with discontinuity after a seek
https://bugzilla.gnome.org/show_bug.cgi?id=744783
2015-03-09 10:06:34 +00:00
Vincent Penquerc'h
c47cc7ba90 validate: use segments to detect success of flushing seeks too
https://bugzilla.gnome.org/show_bug.cgi?id=744783
2015-03-09 10:06:34 +00:00
Vincent Penquerc'h
3ff55dcc31 validate: add non flushing seek support
and a couple scenarios using them

https://bugzilla.gnome.org/show_bug.cgi?id=744783
2015-03-09 10:06:34 +00:00