Commit graph

17724 commits

Author SHA1 Message Date
Nicolas Dufresne
0649b0ab0f v4l2allocator: Fix userptr importation
The length passed to the driver was always 0 instead of the size of
the memory. This would fail validation in videobuf2.
2018-07-13 22:58:35 -04:00
Nicolas Dufresne
38b68cbbd7 v4l2bufferpool: Remove duplicate check
We were calling gst_v4l2_is_buffer_valid() before and inside
gst_v4l2_buffer_pool_qbuf() as we needed to access the group. The second
check failed since the writability of the buffer get inherited from the
GstMemory, which lead to pipeline failure. As we cannot avoid the extra
ref, it would be racy otherwise, just pass the group to _dbuf() so it
does not have to call gst_v4l2_is_buffer_valid() again.

https://bugzilla.gnome.org/show_bug.cgi?id=796692
2018-07-12 15:17:44 -04:00
Havard Graff
ac6e77acad rtpsession: Don't start the RTCP thread until it's needed
Always wait with starting the RTCP thread until either a RTP or RTCP
packet is sent or received. Special handling is needed to make sure the
RTCP thread is started when requesting an early RTCP packet.

We want to wait with starting the RTCP thread until it's needed in order
to not send RTCP packets for an inactive source.

https://bugzilla.gnome.org/show_bug.cgi?id=795139
2018-07-12 18:37:33 +02:00
Nicolas Dufresne
b12b8b0b22 v4l2src: Try to avoid TRY_FMT when camera is streaming
Some camera firmware crash is TRY_FMT is called during streaming. As a
side effect. This try and detect that the same format as currently
running is about to be tried, and skip renegotiation.

https://bugzilla.gnome.org/show_bug.cgi?id=796789
2018-07-11 19:05:00 -04:00
Nicolas Dufresne
c87f9eb37b v4l2videodec: Protect double calls to set_format()
In some cases, set_format() may get called twice before the output
format is set. Running an allocation query in this case is both not
needed and will cause assertion due tot he NULL caps.
2018-07-09 14:01:00 -04:00
Thibault Saunier
57be4d2da3 scaletempo: Mark as Audio in classification 2018-07-08 20:10:12 -04:00
Edward Hervey
9be72b585d qtdemux: Store and propagate SEGMENT sequence numbers
* When receiving a segment in TIME, use that seqnum
* Only reset the stored sequence number when doing HARD reset
  (and not when we get a FLUSH event from upstream)
2018-07-06 15:21:33 +02:00
Michael Tretter
442c377b87 v4l2transform: Implement stable element name
The first converter to be found will now gain the name v4l2convert.
Other converters will be named after the m2m dev node end point they are
attached to.

https://bugzilla.gnome.org/show_bug.cgi?id=784958
2018-07-01 15:45:53 -04:00
Philippe Normand
4c10a1ba27 matroskademux: Set subtitle tag title from TrackName field
GUI applications can then use the title tag to set menu items or labels
representing the track.

https://bugzilla.gnome.org/show_bug.cgi?id=796567
2018-06-29 17:11:28 +01:00
Nicolas Dufresne
6704fbb044 v4l2videoenc: Only renegotiate with upstream
When the decoder get linked further, it will receive a renegotiation
event from downstream. This case is not supported and should be ignored.

This fixes issues when this encoder is used inside an GstRtspServer
pipeline.

https://bugzilla.gnome.org/show_bug.cgi?id=796525
2018-06-29 10:39:04 -04:00
Alicia Boya García
025a430d08 qtdemux: rework segment event pushing, again
This patch aims at fixing the recent regressions in the adaptive test
suite.

All segment pushing in push mode is now done with
gst_qtdemux_check_send_pending_segment(), which is idempotent and
handles both edit lists cases and cases where the upstream TIME segments
have to be sent directly.

Fragmented files that start with a non-zero tfdt are also taken into
account, but their handling has been vastly simplified: now they are
handled as implicit default seeks so there is no need to extend the
GstSegment formulas as was being done before.

qtdemux->segment.duration is no longer modified when
upstream_format_is_time, respecting in this way the durations provided
by dashdemux and fixing bugs in reverse playback tests where mangled
durations appeared in the emitted segments.

https://bugzilla.gnome.org/show_bug.cgi?id=752603
2018-06-29 09:19:31 -04:00
Alicia Boya García
2c394304b7 qtdemux: Don't send EOS during upstream reverse playback
Upstream driving elements such as dashdemux often do reverse playback by
feeding qtdemux with the fragments containing the requested playback
range in reverse order.

But the requested playback range stop may be somewhere in the
middle of a fragment. In that case, a naive pts >= segment.stop
condition may declare end of segment prematurely when demuxing this
first fragment.

This used not to happen because there were places in moov parsing where
segment.stop was overwritten to GST_CLOCK_TIME_NONE even if
upstream_format_is_time -- resulting in this case in a segment with rate
< 0 and stop == -1 and hence not triggering the EOS check, but that was
likely an accident.

This patch modifies the EOS check to take this case into account, not
sending EOS when upstream_format_is_time if rate < 0.

This fixes adaptive.dash.playback.seek_end_live.DASHIF_livestream_testpic_2s

https://bugzilla.gnome.org/show_bug.cgi?id=752603
2018-06-29 09:19:31 -04:00
Peter Seiderer
8ab3b91a8b v4l2transform: fold property set/get PROP_OUTPUT_IO_MODE case into default
https://bugzilla.gnome.org/show_bug.cgi?id=796714
2018-06-28 16:50:28 -04:00
Nicolas Dufresne
5fc67b248b v4l2videoenc: Don't set colorimetry on capture
The colorimetry will be set along with the raw format and those fields
will then be copied from sink to src caps by the gst encoder.

https://bugzilla.gnome.org/show_bug.cgi?id=791471
2018-06-27 17:06:30 -04:00
Nicolas Dufresne
b677ae3cb6 v4l2object: Really always set colorimetry
This fixes patch dd1c5aed65 which
pretended to always set colorimetry but the patch was incomplete.
This is again best effort considering the spec says that for CAPTURE
you may only read this value.
2018-06-27 16:57:29 -04:00
Michael Tretter
35ab185d75 v4l2videodec: do not call streamon while pool is flushing
gst_v4l2_buffer_pool_flush() executes streamoff for the output, but
streamoff->streamon for the capture of the decoder.
gst_v4l2_buffer_pool_streamon() on capture assumes that is able to
resurrect the buffers from the pool, but acquiring buffers fails if the
buffer pool is still flushing.

The decoder needs to stop flushing the pools before calling
gst_v4l2_buffer_pool_flush() to restart the v4l2 device. Otherwise
starting the decoding thread might fail, because there are no buffers in
the capture pool.

This fixes a regression that was introduced in 97985a335c
("v4l2videodec: Add dynamic resolution change support").

https://bugzilla.gnome.org/show_bug.cgi?id=796681
2018-06-27 16:32:23 -04:00
Philipp Zabel
32fa1e7a65 v4l2object: use S_SELECTION instead of S_CROP in gst_v4l2_object_set_crop
The S_CROP call doesn't work on mem2mem output queues. Use the
S_SELECTION call to set the crop rectangle and only fall back to
S_CROP for ancient kernels.
This will allow v4l2videoenc to set the coded size on the output
queue via S_FMT and then set the visible size via the crop rectangle,
as required by the V4L2 codec API.

https://bugzilla.gnome.org/show_bug.cgi?id=796672
2018-06-27 16:31:39 -04:00
Marian Mihailescu
58e9341384 v4l2videoenc: activate capture pool after output pool
Some drivers need output buffers set before capture buffers.
CODA cannot set output format if capture is streaming.
Exynos MFC fails on output STREAMON if capture is already streaming.
This patch delays capture activation until output is configured and
streaming

https://bugzilla.gnome.org/show_bug.cgi?id=796693
2018-06-27 16:28:10 -04:00
Tim-Philipp Müller
238a37295c Update for g_type_class_add_private() deprecation in recent GLib
https://gitlab.gnome.org/GNOME/glib/merge_requests/7
2018-06-23 23:44:19 +02:00
Edward Hervey
8842d51ff8 souphttpsrc: Protect input stream with lock
This was the last remaining place where modifying/unreffing the
input stream was not protected by the lock

https://bugzilla.gnome.org/show_bug.cgi?id=796639
2018-06-20 10:05:35 +02:00
Sebastian Dröge
f2f15a119f splitmuxsrc: Make sure events are writable before setting their seqnum 2018-06-18 12:13:48 +03:00
Nicolas Dufresne
90eade28f2 v4l2bufferpool: Drop truncated frames
Drop truncated frames regardless if they have the ERROR flag or not.
Truncated frame causes video frame map failure in many elements
including cluttersink, glupload etc.
2018-06-07 15:26:32 -04:00
Nicolas Dufresne
e943a11bab v4l2bufferpool: Try return input buffer soon
In this patch we use a non-blocking poll in order to return all input
buffers (buffers from v4l2-output queue). This prevent holding too long
on upstreaming buffer in importing.

https://bugzilla.gnome.org/show_bug.cgi?id=794904
2018-06-07 15:26:32 -04:00
Matthew Waters
ee046864d8 qt: also check for un-suffixed moc
e.g. Qt windows installer doesn't have suffixes
2018-06-07 14:28:37 +10:00
Thibault Saunier
1840159d0b qtdemux: Do not set INVALID seqnum on events 2018-06-06 11:44:33 -04:00
Seungha Yang
aecc17251d tests: qtdemux: Add checking exposed segment event
https://bugzilla.gnome.org/show_bug.cgi?id=796480
2018-06-06 11:19:25 -04:00
Seungha Yang
6b29a4507e qtdemux: Forward upstream time-format segment without mapping
Sample table based segment event (genereted by qtdemux) could break
presentation timeline. For example, qtdemux should not modify upstream
time format segment (e.g., adaptivedemux use case)

https://bugzilla.gnome.org/show_bug.cgi?id=796480
2018-06-06 10:54:55 -04:00
Edward Hervey
418a6270e5 rtspsrc: Seek handling is always done with a valid event
Remove the checks
2018-06-06 07:48:32 +02:00
Edward Hervey
14a9cc603e wavparse: Don't set invalid seqnum on events
Some codepath will call gst_wavparse_perform_seek without an event
and therefore without a valid seqnum
2018-06-06 07:48:32 +02:00
Thibault Saunier
0fbde2a079 qtdemux: Clarify field name about stream-encryption-system
This field is actually only informatory and the user can potentially
choose something else. EME tests in WebKit testsuite actually doesn't
take it into and force another encryption system to be used, and expects
to be given the occasion to do so.

This basically also reverts 3e063703b3.
2018-05-28 16:52:10 -04:00
Thiago Santos
ae7b531902 qtdemux: mark segment as sent after pushing when moov is received
Otherwise we would try to send it a second time if the same moov is
received or in any other situation that might trigger segment sending.

https://bugzilla.gnome.org/show_bug.cgi?id=752603
2018-05-28 11:26:36 -07:00
Thiago Santos
0de143fa3e tests: qtdemux: Avoid using data beyond array and improve error msg
Makes it easier to debug the failures as well as prevents problems
reading out of bounds data.
2018-05-28 11:25:13 -07:00
Nicolas Dufresne
f72c7131ae v4l2object: Don't open the device in get property
This is both racy and inefficient. This function is still missing some
locking which will be address in later patch.

https://bugzilla.gnome.org/show_bug.cgi?id=796185
2018-05-28 12:58:45 -04:00
Tim-Philipp Müller
48dd93662d tests: rtpstorage: fix potential crashes / test failures on 32-bit
Pass 64 bits to g_object_set() for 64-bit integer properties like
rtpstorage's "size-time" property.

https://bugzilla.gnome.org/show_bug.cgi?id=796429
2018-05-27 20:30:46 +01:00
Thiago Santos
364a921d3f qtdemux: do not update segment.stop is it is not a valid time
Otherwise it overflows and starts having a meaningful and wrong value.

https://bugzilla.gnome.org/show_bug.cgi?id=752603
2018-05-25 08:34:29 -07:00
Thiago Santos
9c514f9c7c qtdemux: offset edts segments by the min timestamp of the stream
Otherwise if the stream is starting at timestamp=X it would wait
'X' to start playing.

https://bugzilla.gnome.org/show_bug.cgi?id=752603
2018-05-25 08:34:29 -07:00
Thiago Santos
bfd0e0225e qtdemux: rework segment event pushing
Instead of always keeping a safe segment (start=0) event from the beginning,
delay the creation of this event to when we really know the timestamp of the
first sample. This is important to properly start fragmented streams that
we might join in the middle or to play isolated fragment files that might
have an advanced tfdt.

https://bugzilla.gnome.org/show_bug.cgi?id=752603
2018-05-25 08:34:29 -07:00
Thibault Saunier
43a540b1cd qtdemux: Do not unref a NULL stream_tags
stream->stream_tags is reset to NULL once we expose the stream and
these have been consumed, we need to check that when cleaning up
the stream.
2018-05-25 10:54:54 +02:00
Thibault Saunier
3e063703b3 qtdemux: Do not run the preferred decryptor context query if no decryptor avalaible
Ultimately this avoids a segfault as the code expect a non NULL array
here.
2018-05-25 10:53:16 +02:00
Alicia Boya García
c2a0da8096 qtdemux: Allow edit lists on fragmented files on push mode
Fragmented files often use elst.duration=0 which before
ee78825eae was wrongly interpreted as
having no frames.

Since that issue has now been fixed, there is no reason to disable edit
lists in fragmented files. This commit enables them, therefore producing
correct stream time for files containing edit lists.

https://bugzilla.gnome.org/show_bug.cgi?id=793058
2018-05-24 15:53:48 +02:00
Alicia Boya García
5388d6e6a2 qtdemux: fix computation of first_duration for fragmented files in push mode
Since ca068865c3 the duration of the first
frame is not used for estimating the frame rate.

For this purpose, stream->first_duration was initialized with the
duration of the first frame. In fragmented files, this was previously
done by peeking the first moof, but that can only be done in pull mode.

Fortunately, we don't really need to do that, at least with the current
design: When we are estimating the frame rate we already have the
sample table, regardless of the scheduling mode and whether the file is
fragmented or not, so we can obtain first_duration there much more
reliably.

This fixes frame rate estimation for fragmented files in push mode.

https://bugzilla.gnome.org/show_bug.cgi?id=796384
2018-05-24 15:53:48 +02:00
Vivia Nikolaidou
d11339d616 splitmuxsink: Added new async-finalize mode
This mode is useful for muxers that can take a long time to finalize a
file. Instead of blocking the whole upstream pipeline while the muxer is
doing its stuff, we can unlink it and spawn a new muxer+sink combination
to continue running normally.

This requires us to receive the muxer and sink (if needed) as factories,
optionally accompanied by their respective properties structures. Also
added the muxer-added and sink-added signals, in case custom code has to
be called for them.

https://bugzilla.gnome.org/show_bug.cgi?id=783754
2018-05-24 12:47:24 +03:00
Alicia Boya García
d35f893715 qtdemux: Don't send gaps bigger than 1 second (now in push mode too)
This applies the same workaround to gaps that is being used in pull
mode.

https://bugzilla.gnome.org/show_bug.cgi?id=778426
2018-05-24 11:09:53 +02:00
Seungha Yang
f61c2bc301 qtdemux: Properly handle edit list in push mode
If there are empty segments in edit list, demux should
adjust "accumulated_base" to apply it into running time.

https://bugzilla.gnome.org/show_bug.cgi?id=778426
2018-05-24 11:09:53 +02:00
Mathieu Duponchelle
d9235404aa matroska-mux: write colorimetry
This is a straightforward translation of 5dd39d8, can be trivially
checked by running:

gst-launch-1.0 -v videotestsrc ! video/x-raw, colorimetry=2:4:7:1 ! \
matroskamux ! matroskademux ! fakesink

and verifying that the colorimetry is correctly preserved.

https://bugzilla.gnome.org/show_bug.cgi?id=796344
2018-05-23 15:14:23 +02:00
Alicia Boya García
ee78825eae qtdemux: fix buggy duration in edits with duration=0 in fragmented files without a mehd
https://bugzilla.gnome.org/show_bug.cgi?id=794858
2018-05-23 14:25:56 +02:00
Tim-Philipp Müller
db688c5504 docs: fix typos 2018-05-23 13:14:27 +01:00
Seungha Yang
c69ccf60d8 qtdemux: Clarify variable name
As defined by spec, use "empty edit". It's more straightforward.

https://bugzilla.gnome.org/show_bug.cgi?id=778426
2018-05-22 11:29:44 +02:00
Xabier Rodriguez Calvar
ee4b45da24 qtdemux: add context for a preferred protection
qtdemux selected the first system corresponding to a working GStreamer
decryptor. With this change, before selecting that decryptor, qtdemux
will check if it has context (a preferred decryptor id) and if not, it
will request it.

The request includes track-id, available key system ids for the
available decryptors and even the events so that the init data is
accessible.

[eocanha@igalia.com: select the preferred protection system even if not available]

Test "4. ClearKeyVideo" in YouTube leanback EME conformance tests 2016 for
H.264[1] uses a media file[2] with cenc encryption which embeds 'pssh' boxes
with the init data for the Playready and Widevine encryption systems, but not
for the ClearKey encryption system (as defined by the EMEv0.1b spec[3] and with
the encryption system id defined in [4]).

Instead, the ClearKey encryption system is manually selected by the web page
code (even if not originally detected by qtdemux) and the proper decryption key
is dispatched to the decryptor, which can then decrypt the video successfully.

[1] http://yt-dash-mse-test.commondatastorage.googleapis.com/unit-tests/2016.html?test_type=encryptedmedia-test&webm=false
[2] http://yt-dash-mse-test.commondatastorage.googleapis.com/unit-tests/media/car_cenc-20120827-86.mp4
[3] https://dvcs.w3.org/hg/html-media/raw-file/eme-v0.1b/encrypted-media/encrypted-media.html#simple-decryption-clear-key
[4] https://www.w3.org/Bugs/Public/show_bug.cgi?id=24027#c2

https://bugzilla.gnome.org/show_bug.cgi?id=770107
2018-05-21 17:36:17 +02:00
Enrique Ocaña González
844423ff99 qtdemux: also push buffers without encryption info instead of dropping them
Test "17. PlayReadyH264Video" in YouTube leanback EME conformance tests 2016
for H.264[1] uses a media file[2] with cenc encryption whose first two 'moof'
boxes have no encryption information (no 'saiz' and 'saio' boxes).

Those boxes are actually not encrypted and the current qtdemux implementation
was just dropping them, breaking the test use case.

This patch detects those kind of situations and just lets the unencrypted
buffers pass. Of course, this needs some collaboration by the decryptors,
which should also do the same and not to try to decrypt those clear buffers.

[1] http://yt-dash-mse-test.commondatastorage.googleapis.com/unit-tests/2016.html?test_type=encryptedmedia-test&webm=false
[2] http://yt-dash-mse-test.commondatastorage.googleapis.com/unit-tests/media/oops_cenc-20121114-142.mp4

https://bugzilla.gnome.org/show_bug.cgi?id=770107
2018-05-21 17:36:17 +02:00