Commit graph

503 commits

Author SHA1 Message Date
Sebastian Dröge
e0dee7f4c1 oggdemux: Answer POSITION query 2018-10-17 19:44:22 +03:00
Sebastian Dröge
3dd95b1311 oggdemux: Make sure that events are writable before changing their seqnum 2018-06-18 13:28:15 +03:00
Edward Hervey
5c118e5924 oggdemux: Properly relay seqnum of segments
Not all cases were handled regarding properly propagating the
seqnum of SEGMENT events on all downstream segment-related events
2018-06-05 17:24:05 +02:00
Edward Hervey
0d14819ef2 oggdemux: Handle invalid-sized packets
On invalid packets there is the possibility we might end up wanting
to trim/offset more than what is available.

oss-fuzz issue #5866
2018-02-01 10:51:21 +01:00
Edward Hervey
25fa4802fe oggdemux: Move mutex/cond initialization/release
We only need to initialize the mutex/cond once when creating the
element and then release them when we are done with the element.

Avoids weird "mutex_clear called when still locked" issues
2017-11-15 10:55:55 +01:00
Edward Hervey
895d884701 oggdemux: Solidify gst_ogg_demux_loop_push() some more
There were still some races going on where seeking events wouldn't
be properly intercepted/executed by this thread.

* Instead of always waiting for the GCond to be emitted, first just
  check if there is an event available
* Take ownership of the event *while* the lock is taken and not
  after releasing/reacquiring it
* Finally acquire lock at the very top and release it at the end
  to make it a bit more streamlined

This removes the remaining issues with seeks not being executed
2017-11-08 17:51:52 +01:00
Edward Hervey
9f678bb27f oggdemux: Don't double-unlock
The previous branch will release the lock in the call to
gst_ogg_demux_seek_back_after_push_duration_check_unlock()

Only unlock it if we didn't call that function
2017-11-08 17:51:51 +01:00
Edward Hervey
c86df789ed oggdemux: Drop data before new segment
When calculating duration in push-mode we seek to a certain position
and discard any data until we get data from that requested position.

The problem is that basing ourselves solely on offset to determine
whether we reached the target offset is wrong since the source might
be fast enough  to send us that target position *before* it processed
the requested seek.

This would end up in a situation where:
* We think we're done with duration estimate
* We fire a seek back to "0" in the loop thread
* We resume normal processing
* ... except that we're still getting data from too far ahead which
  we decide to process.
* And we start doing totally wrong granule/time/duration calculation
  and pushing wrong data.

Instead of this confusion, wait until we receive data from the requested
seek. We do that by using the fact that the seqnum in
seek_event_drop_til will be non-zero until the SEGMENT corresponding
to the requested SEEK has been received.

Bonus: makes startup slightly faster
2017-11-07 15:16:52 +01:00
Edward Hervey
0a49b93ade oggdemux: Wait for push loop to be started
Code using the push_loop_thread (using for sending seeks) assumes
that the thread was properly started, except that this isn't always
true and the thread might not have completely started.

Instead wait for the thread to properly start before doing anything
else.
2017-11-07 15:16:52 +01:00
Edward Hervey
1172e5efc7 oggdemux: Protect against invalid granule positions
Only valid values are -1, 0 or positive values. Anything else is
most likely corrupted data streams
2017-11-04 11:50:10 +01:00
Edward Hervey
33dfed5af8 oggdemux: Only track time for initialized streams
in push-mode we only can track time (or most operations on streams
for that matter) if the underlying GstOggMap was properly initialized.
2017-11-01 18:39:26 +01:00
Edward Hervey
eb8bf0f162 oggdemux: Fix chain leak in push mode
In some corner cases we end up with the building chain not being
properly tracked (and therefore not properly freed).

Add a FIXME so it can later be fixed, but for now just fix the leak
2017-11-01 11:18:12 +01:00
Edward Hervey
90106b6bf2 oggdemux: Don't forget to reacquire lock when needed
Fixup to ef93130cf0

I overlooked the issue. There is a case when the lock is released
and we need to reacquire it
2017-10-26 10:07:15 +02:00
Edward Hervey
6fd8d78d8b oggdemux: Don't drop sticky events
Previous commit was wrong. We should still send all events to the
pad (so that sticky events get attached to it and sent when pad
gets added).
2017-10-24 11:05:20 +02:00
Edward Hervey
e8a60b3de9 oggdemux: Improve handling of EOS without source pads
We might have a chain to use, but it might not have any active pads

Properly detect that and send an error message on EOS
2017-10-24 10:56:00 +02:00
Edward Hervey
f902286a26 oggdemux: Error out on EOS if we have no chains to use
There are not active and pending chains, if we get EOS we need to
inform the user via an error message
2017-10-20 18:41:52 +02:00
Edward Hervey
ef93130cf0 oggdemux: Don't double lock
The lock was already taken just before this block and is released after
2017-10-20 18:41:46 +02:00
Vincent Penquerc'h
bb0abf8558 oggdemux: fix artifacts at chain boundaries
https://bugzilla.gnome.org/show_bug.cgi?id=782132
2017-05-29 16:22:04 +01:00
Vincent Penquerc'h
523e98396f oggdemux: fix clipping more samples than exist in the first packet
This can happen in Opus (and maybe other codecs ?), and would cause
failure to play.

https://bugzilla.gnome.org/show_bug.cgi?id=782157
2017-05-26 14:41:47 +01:00
Thibault Saunier
099ac9faf2 docs: Convert gtkdoc comments to markdown
Modernizing the documentation, making it simpler to read an
modify and allowing us to possibly switch to hotdoc in the
future.
2017-03-10 18:19:17 -03:00
Jan Schmidt
8596ec23cb oggdemux: Fix reverse playback
Fix various issues with reverse playback by clearing tracking
vars when working in reverse, and where possible using the
timestamp interpolation code to generate timestamps for
outgoing buffers. Make sure to mark things as discontinuous
only when looping backward to a new position and fix seeking
to the next page when starting.
2017-03-04 00:30:37 +11:00
Jan Schmidt
fe1f47aa17 oggdemux: Timestamp tracking fixes
In gst_ogg_demux_do_seek() when calculating the
keyframe time, account for a non-zero start-time

Handle a discontinuous first packet in
gst_ogg_demux_setup_first_granule() because that's pretty
normal after a seek. Also differentiate between a genuinely
truncated first packet and just bailing out early, by not using
granule = -1 as an error code.

Make the debug output logs clearer about which timestamps
are stream times (PTS) and which are ogg timestamps.
2017-03-04 00:30:37 +11:00
Jan Schmidt
342132a700 oggdemux: Don't arbitrarily guess a timestamp of 0
When we haven't managed to manufacture a timestamp for
a packet, don't just guess '0', leave it at none and
let downstream decide
2017-03-04 00:30:37 +11:00
Sebastian Dröge
732ecf0925 oggdemux: Don't end up ignoring caps just because there are no headers for this stream
https://bugzilla.gnome.org/show_bug.cgi?id=775459
2016-12-01 15:12:59 +02:00
Vincent Penquerc'h
31438ef49b oggdemux: safety for failing to determine time length in push mode
If we can't find a valid granule near the end of the file, we
disable seeking. This guards against the whole file being then
read and never going to PLAYING.

https://bugzilla.gnome.org/show_bug.cgi?id=770314
2016-09-05 11:42:44 +01:00
Vincent Penquerc'h
6f856cb54d oggdemux: increase EOS granpos detection chunk size
This can be too small on some files to find a valid granule.

https://bugzilla.gnome.org/show_bug.cgi?id=770314
2016-09-05 11:41:43 +01:00
Thibault Saunier
bc6aae6ca7 Use the new API to post flow ERROR messages on the bus
https://bugzilla.gnome.org/show_bug.cgi?id=770158
2016-08-26 19:23:24 -03:00
Vincent Penquerc'h
273d35ce20 oggdemux: remove eos avoidance workaround
This workaround tried to avoid an EOS event when seeking to the
end of an Ogg stream in order to find its duration. At some point,
an EOS event there would cause any queue2 upstream to pause and
not restart on a seek back to the beginning. This now appears to
not be the case anymore, and so the workaround can be removed.

https://bugzilla.gnome.org/show_bug.cgi?id=767689
2016-08-05 07:50:35 +02:00
Vincent Penquerc'h
714e3b9741 oggdemux: fix unknown duration playing Ogg over HTTP
If the duration is not known from the chain, it might be known
by the startup seek.

This fixes failure to seek.

Merged with a patch from Tim-Philipp Müller <tim@centricular.com>

https://bugzilla.gnome.org/show_bug.cgi?id=768991
2016-07-21 10:40:07 +01:00
Vincent Penquerc'h
528f030eff oggdemux: demote an expected error to debug
Dropping a buffer because we have a seek pending is normal,
and will now happen when we trigger a seek while going through
the packets in a page. So this should not be an error.
2016-06-23 10:24:55 +01:00
Vincent Penquerc'h
2ac5bd293b oggdemux: fix audio glitches with low bitrate vorbis
A low bitrate stream which can pack more than 2 seconds of audio
in a page would cause the stream's position to be updated not
often enough, and would trigger a spurious "jump" via a GAP
event. Instead, we update the stream position after calculating
the new overall segment position.

https://bugzilla.gnome.org/show_bug.cgi?id=764966
2016-06-16 11:10:08 +01:00
Edward Hervey
98c9eb9858 oggdemux: Reset keyframe_granule when needed
This avoids ending up with bogus values when doing flushing seeks
in push-mode.

https://bugzilla.gnome.org/show_bug.cgi?id=766467
2016-05-15 14:39:25 +02:00
Vineeth TM
44b70ca3a1 base: use new gst_element_class_add_static_pad_template()
https://bugzilla.gnome.org/show_bug.cgi?id=763075
2016-03-24 14:25:41 +02:00
Vincent Penquerc'h
603e2fe24a oggdemux: fix chaining causing running time to restart from 0
This fixes:
gst-play-1.0 http://relay-nyc.gameowls.com:8000/chiptune.ogg

https://bugzilla.gnome.org/show_bug.cgi?id=758282
2016-03-04 12:18:53 +00:00
Sebastian Dröge
a135868262 oggdemux: Add GstAudioClippingMeta for Opus for accurate start/end clipping
https://bugzilla.gnome.org/show_bug.cgi?id=757153
2015-11-03 20:35:33 +02:00
Sebastian Dröge
8a3be7323a oggdemux: Allow start clipping for Opus
The granulepos does not have the pre-skip subtracted while timestamps do,
and the last granulepos will be shorter by the number of samples that should
be dropped because of padding in the end.

As such, extrapolating the granule of the beginning of the first frame will
lead to a negative value, which is not a problem but intentional.

https://bugzilla.gnome.org/show_bug.cgi?id=757153
2015-11-03 20:35:33 +02:00
Luis de Bethencourt
799020804f oggdemux: Use GstClockTimeDiff and print signed integer in debug logs
Use GstClockTimeDiff and Clock macros to print signed integer time
differences in the debug logs.

https://bugzilla.gnome.org/show_bug.cgi?id=757480
2015-11-02 16:10:07 +00:00
Olivier Crête
4665c0802a oggdemux: Set chain pointers to NULL
Otherwise, they will refer to freed memory

https://bugzilla.gnome.org/show_bug.cgi?id=753078
2015-08-06 17:40:40 +01:00
Olivier Crête
315857dce6 oggdemux: Return FLUSHING if pad if flushing
If the initial seek fails because the pad is
flushing, then return GST_FLOW_FLUSHING instead
of an error.
2015-07-30 18:43:19 -04:00
Guillaume Desmottes
a5dcce98aa oggdemux: set building_chain to NULL when deactivating chain
The chain is about to be invalidated so we shouldn't keep it around.
Prevent a double free crash when the demuxer is being finalized.

https://bugzilla.gnome.org/show_bug.cgi?id=751000
2015-06-22 14:09:51 +02:00
Guillaume Desmottes
1c10b58cce oggdemux: fix chain leak
Don't leak the building_chain when destroying.

Fix leaks with the validate.http.playback.reverse_playback.vorbis_theora_1_ogg
scenario.

https://bugzilla.gnome.org/show_bug.cgi?id=748964
2015-05-26 08:28:23 +01:00
Tim-Philipp Müller
0a3b584aa0 Revert "oggdemux: Prevent seeks when _SCHEDULING_FLAG_SEQUENTIAL is set"
This reverts commit 76647f2710.

Avoiding pull mode activation is a feature regression, and
demuxers should always use pull mode where that is possible,
e.g. if there's an upstream queue2 with a ring buffer or
a download buffer.

This patch made reverse playback no longer possible over http.

If the goal is to minimise seeks, then that can still be done
by making the demuxer behave differently in pull mode if
the SEQUENTIAL flag is set. If there are bugs, like the demuxer
needlessly scanning the entire file on start-up in pull mode,
then those should be fixed instead.

https://bugzilla.gnome.org/show_bug.cgi?id=746010
2015-05-20 10:28:30 +01:00
Tim-Philipp Müller
ec5c93f169 docs: update element example pipelines
- gst-launch -> gst-launch-1.0
- use autoaudiosink and audiovideosink more often
- review pipeline examples and descriptions
2015-05-10 11:38:19 +01:00
Young Han Lee
bd40d2e33f oggdemux: remove unused code
These lines have done nothing for about 10 years.

https://bugzilla.gnome.org/show_bug.cgi?id=748820
2015-05-03 13:33:01 +01:00
Guillaume Desmottes
d31472fde7 oggdemux: fix event leaks
gst_event_replace() takes its own reference on the event so we should drop
ours after creating and storing an event using it.

This fix leaks which can be reproduced using the
validate.http.media_check.vorbis_theora_1_ogg scenario.

https://bugzilla.gnome.org/show_bug.cgi?id=748247
2015-04-23 11:02:45 +01:00
Vincent Penquerc'h
a44258dbe8 oggdemux: fix wrong duration on partial streams with a skeleton index
When a stream has a skeleton index, the stream time is taken from that
index. However, when part of the stream is captured, the index is
invalid as its offsets are now wrong. To avoid this, we ignore the index
when the last offset points beyond the end of the stream (when its
byte length is known).

https://bugzilla.gnome.org/show_bug.cgi?id=744070
2015-04-03 16:40:04 +01:00
Guillaume Desmottes
452d7b04e3 oggdemux: don't use GST_ERROR() for debug messages
Fix https://bugzilla.gnome.org/show_bug.cgi?id=746457
2015-04-01 16:06:42 +01:00
Mark Nauwelaerts
d1f91723be oggdemux: resurrect some flow return handling
https://bugzilla.gnome.org/show_bug.cgi?id=744572
2015-03-29 13:51:05 +02:00
Wonchul Lee
b45941219e oggdemux: Fix compiler warning
gstoggdemux.c:1233:11: error: format specifies type 'long' but the argument has type 'ogg_int64_t' (aka 'long long') [-Werror,-Wformat]
          granule);
          ^~~~~~~

https://bugzilla.gnome.org/show_bug.cgi?id=746512
2015-03-23 09:15:02 +01:00
Vincent Penquerc'h
1ee2fccfdf oggdemux: fix playback regression on streams with clipped data at start
The code that was calculating the start granule from packet durations
was interpreting a negative value as an error, but this is actually a
valid case, to indicate clipping of data at start.

https://bugzilla.gnome.org/show_bug.cgi?id=743900
2015-03-16 11:59:19 +00:00