Commit graph

2236 commits

Author SHA1 Message Date
Guillaume Desmottes
7bf0c5b7be concat: fix qos event handling
We were shadowing the sinkpad variable resulting in:
- the QoS event to be send to all sink pads instead of the active one
- the pad to be leaked

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1543>
2022-01-20 11:37:50 +00:00
Jan Alexander Steffens (heftig)
516f9592e7 multiqueue: Fix query unref race on flush
If the query has already been destroyed at this point, GST_IS_QUERY will
read garbage, can return false and we will try to unref it again.
Instead, make note of whether the item is a query when we dequeue it.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1284>
2021-10-31 12:28:35 +00:00
Mathieu Duponchelle
7cec3538f5 multiqueue: fix obsolete comment re initial flow status
The initial single queue srcresult is OK, it hasn't been
NOT_LINKED since 2007.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1271>
2021-10-29 11:14:28 +00:00
Mathieu Duponchelle
33ea698ef3 multiqueue: never consider a queue that is not waiting
.. when computing the high id.

After a flush for instance, sq->srcresult is reset to OK,
yet it doesn't make sense to pick a non-existing position
id as the high id when a queue doesn't contain any items
in that situation either.

It is in any case completely OK to let the not-linked stream
get consumed without throttling at this stage, as any
first packet arriving on other single queues will get assigned
a higher position id.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1271>
2021-10-29 11:14:28 +00:00
Jan Alexander Steffens (heftig)
8eebe19cae input-selector: Use proper segments when cleaning cached buffers
We need to use the segment associated with the cached buffer, not the
current segment of the pad, otherwise we miscalculate the running time
of cached buffers from before a segment change.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1270>
2021-10-29 11:30:34 +01:00
Jan Schmidt
ac70fb78c4 multiqueue: Use running time of gap events for wakeups.
Use gap events to update the next_time of a queue the same
as buffers or segment events. Fixes problems where a group
consisting only of sparse streams primarily driven by
gap events would stall with a full multiqueue because
unlinked streams in the group were not being woken to
push data.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/881>
2021-09-06 09:57:59 +01:00
Edward Hervey
23171cb6af concat: Properly propagate seqnum of segment events
Was broken by https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/819

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/872>
2021-08-20 15:56:56 +01:00
Mathieu Duponchelle
53c68d92d8 concat: adjust running time offsets on events
When concat adjusts the base of the segments it forwards
downstream, it needs to also adjust the running time offsets,
as GstPad does when an offset is set by the application on a pad.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/823>
2021-05-14 19:27:02 +01:00
Mathieu Duponchelle
96800c0b4a concat: fix locking in SEGMENT event handler
concat->current_start_offset needs the lock taken for safe access,
as it can be accessed from outside of the streaming thread, eg
in release_pad.

An early break is also added for an error case.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/822>
2021-05-14 17:12:57 +01:00
Doug Nazar
ab4ec53d56 sparsefile: Fix sparsefile on Win32
When switching between read/write a fseek() or fflush() is required.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/821>
2021-05-14 14:36:57 +01:00
Doug Nazar
20d4184a69 downloadbuffer: close file before trying to remove
On Windows, the file handles must be closed before you can delete a file.
Also, it would cause an error if you try to close an already closed handle.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/821>
2021-05-14 14:36:57 +01:00
Doug Nazar
a16492a61f downloadbuffer: return flow error on read error
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/821>
2021-05-14 14:36:57 +01:00
Jan Alexander Steffens (heftig)
1dcdde9fdf multiqueue: Ensure peer pad exists when iterating internal links
The pads can be NULL when we're racing with pad removal, e.g. when the
pads get removed between `gst_pad_iterate_internal_links` acquiring the
parent element and `gst_multi_queue_iterate_internal_links` locking the
multiqueue.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/811>
2021-05-06 12:59:58 +01:00
Edward Hervey
b566cbe022 queue2: Refuse all serialized queries when posting buffering messages
When posting buffering messages there are no safe places or timing to avoid
deadlocks.

Previously the code was trying to be "smart" by only forwarding serialized
queries if the queue was empty ... but that could happen when queue2 hadn't yet
posted a 100% buffering message. Meaning the pipeline might be paused and
pushing a serialized query downstream might never complete.

Therefore let's completely disable forwarding of serialized queries when
`queue2` is used as a buffering element (meaning `ALLOCATION` and `DRAIN`
queries).

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/797>
2021-04-17 09:28:46 +01:00
Doug Nazar
daf4e5b410 clocksync: Fix providing system clock by default
clocksync defaults to sync=true so should advertise it by default.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/795>
2021-04-13 11:19:54 +01:00
Matthew Waters
fdf6a793dc gst: don't use volatile to mean atomic
volatile is not sufficient to provide atomic guarantees and real atomics
should be used instead.  GCC 11 has started warning about using volatile
with atomic operations.

https://gitlab.gnome.org/GNOME/glib/-/merge_requests/1719

Discovered in https://gitlab.freedesktop.org/gstreamer/gst-plugins-good/-/issues/868

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/793>
2021-04-13 01:58:54 +01:00
Bing Song
b39a06065a identity/clocksync: Also provide system clock if sync=false
identity should provide when sync=true. Don't provide when sync=false.

Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/630

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/697>
2020-11-20 14:00:16 +00:00
Jonathan Matthew
d1fb776782 typefind: copy seqnum to new segment event
Fixes: #635
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/701>
2020-11-16 10:47:28 +00:00
Jonathan Matthew
39df723ba5 queue2: Fix modes in scheduling query handling
Create a new query to send upstream and copy the flags across from it,
rather than reusing the same query, as this allows us to prevent use
of pull mode when we don't have a download file.

Fixes: #629
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/696>
2020-11-04 10:00:18 +00:00
Sebastian Dröge
bc81eaf103 input-selector: Wake up blocking pads when releasing them
Otherwise deactivating them will cause a deadlock as they're blocking
inside the streaming thread.

Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/601

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/652>
2020-09-30 15:54:39 +01:00
Thibault Saunier
9d195da36c inputselector: Push event on selected pad only when one is selected
Calling `gst_input_selector_get_active_sinkpad` triggers sinkpad
selection but won't notify about it, leading  to breaking code that
relying on it. This new code added as part of
63ccf45395 was thought to be triggered only
when a pad was already selected and not change the behavior otherwise
so this commit makes sure it is actually the case.

Fixes https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/issues/766

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/522>
2020-06-15 17:33:22 +00:00
Camilo Celis Guzman
3ac3a01298 queue: protect against lost wakeups for iterm_del condition
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/513>
2020-06-06 17:41:20 +00:00
Mathieu Duponchelle
0c1ce6e8c4 utils: expose GstPluginAPIFlags
These can be passed to gst_type_mark_as_plugin_api, to inform
plugin cache generation.

For now a single flag is specified, "IGNORE_ENUM_MEMBERS", it
can be used for dynamically generated enums to avoid documenting
environment-specific enumeration members. An example is
GstX265EncTune.
2020-06-06 00:38:29 +02:00
Mathieu Duponchelle
d46e902273 identity, clocksync: implement provide_clock
Since those are using the clock for sync, they need to also
provide a clock for good measure. The reason is that even if
downstream elements provide a clock, we don't want to have
that clock selected because it might not be running yet.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/509>
2020-06-05 11:28:12 +00:00
Jan Schmidt
8e670a23b0 queue2: Defer downstream bitrate query to the streaming thread.
When we want to perform a downstream bitrate query, just
set the reconfigure flag on the srcpad and get the streaming
thread to do it. That avoids emitting a downstream query
when receiving the upstream RECONFIGURE event - which can
lead to deadlocks if downstream is sending the event from
within a lock - e.g. input-selector.

If querying the downstream bitrate changes the cached
value, then make sure to update our buffering state
and potentially post a BUFFERING message to the application.

Fixes: https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/566
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/501>
2020-06-05 10:29:19 +00:00
Edward Hervey
f2478618b5 inputselector: Avoid deadlock when requesting pads
The deadlock was the following:
* One thread requests a new pad, the internal lock is kept while adding the pad
* Another thread (or the same one) requests the internal links of a pad (could
be that pad)... which also requires that lock.

That internal lock is not required when adding the pad to the element (which is
the last action when requesting a new pad). The fact it will be actually used
will be *after* the request pad function is released.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/512>
2020-06-05 09:59:15 +00:00
Sebastian Dröge
5bdbde32da plugins: Use gst_type_mark_as_plugin_api() for all non-element plugin types 2020-06-04 15:49:24 -04:00
Edward Hervey
08d2d3cfd7 queue2: Avoid races when posting buffering messages
When posting a buffering message succesfully:
* Remember the *actual* percentage value that was posted
* Make sure we only reset the percent_changed variable if the value we just
  posted is indeed different from the current value

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/511>
2020-06-04 11:24:39 +02:00
Mathieu Duponchelle
07fcd4d1b5 queue2: don't post unnecessary buffering message, refine locking
This is a follow up to review comments in !297

+ The posting of the buffering message in READY_TO_PAUSED isn't
  needed, removing it made the test fail, but the correct fix
  was simply to link elements together

+ Move code to relock the queue and set last_posted_buffering_percent
  and percent_changed inside the buffering_post_lock in create_write().
  This makes locking consistent with post_buffering()

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/297>
2020-05-27 13:07:03 +00:00
Carlos Rafael Giani
6d4ad80b82 queue2: Fix missing/dropped buffering messages at startup
This fixes a bug that occurs when an attempt is made to post a buffering
message before the queue2 was assigned a bus. One common situation where
this happens is when the use-buffering property is set to TRUE before the
queue2 was added to a bin.

If the result of gst_element_post_message() is not checked, and the
aforementioned situation occurs, then last_posted_buffering_percent and
percent_changed will still be updated, as if posting the message succeeded.
Later attempts to post again will not do anything because the code then
assumes that a message with the same percentage was previously posted
successfully and posting again is redundant.

Updating these variables only if posting succeed and explicitely
posting a buffering message in the READY->PAUSED state change ensure that
a buffering message is posted as early as possible.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/297>
2020-05-27 13:07:03 +00:00
dhilshad
319f81bc60 sparsefile: fix possible crash when seeking
In gst_sparse_file_clear function we were closing a file and
reopening it using closed file descriptor.

Fix: Removed closing and reopening of file.

Fixes #512

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/372>
2020-05-25 08:35:32 +00:00
Mathieu Duponchelle
742291c799 fdsink: do not supress legitimate errors when unlocking
Instead, only wait_preroll when writev_* returns FLUSHING

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/476>
2020-05-20 14:10:14 +00:00
Mathieu Duponchelle
2812064734 fdsink: remove unused struct member bytes_written
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/476>
2020-05-20 14:10:14 +00:00
Mathieu Duponchelle
710ce84d0c filesink: port over unlock code from fdsink
See also: 5216322d39

The previous code was causing "random" flushing returns
in scenarios with intensive state changes such as within
a buffering pipeline.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/476>
2020-05-20 14:10:14 +00:00
Thibault Saunier
24d789a9b5 input-selector: Ensure events are forwarded only once per pad
The code was prepared to do it but was missing to fill the pushed_pads
list.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/485>
2020-05-15 13:51:43 +00:00
Thibault Saunier
63ccf45395 inputselector: Ensure that events are pushed first on active pad
Making it less random and fixing a race in a GES test where we have
as pipeline:

```
videotestsrc ! output-selector name=s ! input-selector name=i s. ! timecodestamper ! i.
```

which we seek, leading to the seek reaching the video testsrc
without going through the timecodestamper and generating a buffer
even before timecodestamper gets the seek which means that its internal
state is wrong compared to the datastream it gets and attaches wrong
timecode metas.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/485>
2020-05-15 13:51:43 +00:00
Thibault Saunier
9f2e9ccd7e inputselector: Never reset active pad set from the user
This was leading to interesting races in a GES test.
Related to: https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/issues/108

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/481>
2020-05-14 22:15:12 +00:00
Thibault Saunier
f41f84538a output-selector: Drop duplicated seek events
When we get a seek event on several source pads, we should drop
the duplicated ones as any element that has several srcpads (like
demuxers).

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/478>
2020-05-13 17:59:06 -04:00
Thibault Saunier
cb212c9fd7 identity: Handle seeking with single_segment=True
Identity was ignoring seek and flush events even when using
a single segment. In the end it means that we couldn't compute
buffers running-time and stream time after seeks.

This commits adds support for flushing seeks only as I have no idea
what to do for non flushing ones.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/450>
2020-05-06 14:24:36 +00:00
Thibault Saunier
b517b59171 identity,clocksync: Fix timestamping inside single segment in reverse playback
In reverse playback, buffers are played back from buffer.stop
(buffer.pts + buffer.duration) to buffer.pts running times which
mean that we need to use the buffer end running time as a buffer
timestsamp, not the buffer pts when using a single segment in reverse
playback.

This is now being tested in
`validate.test.identity.reverse_single_segment`

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/450>
2020-05-06 14:24:36 +00:00
Edward Hervey
d854fb12e9 queue2: Use explicit limit checking
When we know we'll only be checking the real limits, use a clearer/simpler
macro.

CID #1037148

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/464>
2020-05-06 08:04:28 +02:00
Sebastian Dröge
687c65994e clocksync: Remove handoff signals
They're not really useful on this element and were just a leftover from
identity.

Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/-/issues/540

Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/443>
2020-05-04 09:21:02 +03:00
Mathieu Duponchelle
d2299180c9 multiqueue: fix link-like syntax in doc 2020-04-28 00:33:22 +02:00
dhilshad
c8bcb9db18 downloadbuffer:fix pushing buffer before stream start event
downloadbuffer source pad pushes the first buffer before pushing
Stream Start and Segment event, when working in Push mode.

Fix:Pushing Stream Start and Segment after coming out of
wait for data, and before pushing the buffer to next element.

Fixes #534
2020-04-17 15:16:18 +00:00
Vivia Nikolaidou
4679c2ed6f multiqueue: Add current-level-{buffers, bytes, time} pad properties
To get the current buffers/bytes/time levels of the corresponding
internal queue
2020-04-09 18:03:35 +03:00
Vivia Nikolaidou
74938f07c2 multiqueue: Add stats property
The returned "stats" structure contains, for now, one array called
"queues" with one GstStructure per internal queue, containing said
queue's current level of bytes, buffers, and time.
2020-04-09 17:57:43 +03:00
Vivia Nikolaidou
cbdc259865 downloadbuffer, multiqueue, queue2: Fix watermark docs
It is not explicitly specified anywhere in the docs that 0% buffering is
at low-watermark and 100% buffering is at high-watermark. It was
specified only in the sources.
2020-04-06 18:19:06 +03:00
Seungha Yang
13bee5ae0c filesink: Fix for updating the index of memory to write in the next iteration
current_buf_mem_idx stands for the index of memory of the corresponding
buffer which is scheduled to be written in the next iteration.
If all memory objects were scheduled to be written in the current
iteration, reset the index to zero so that starting from the first
memory object of the next buffer.
2020-03-31 19:05:30 +09:00
Seungha Yang
5c74c8c4f8 filesink: Fix crash caused by zero-size memory allocation
If size of vector is greater than one, we are allocating zero-size
memory and trying invalid memcpy operation
2020-03-28 16:20:51 +09:00
Sebastian Dröge
dcb2be3282 filesink: Add a new full buffer mode to filesink
Previously the default and full modes were the same. Now the default
mode is like before: it accumulates all buffers in a buffer list until
the threshold is reached and then writes them all out, potentially in
multiple writes.

The new full mode works by always copying memory to a single memory area
and writing everything out with a single write once the threshold is
reached.
2020-03-26 11:31:03 +00:00