-
Notifications
You must be signed in to change notification settings - Fork 5
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Use meson built in support for dependency to threads lib #10
base: tulio/dcsctp
Are you sure you want to change the base?
Commits on Oct 23, 2022
-
rtpfunnel: revert back to pexip version [pexhack]
Revisit when we do all the final changes around rtpsession and writing of TWCC seqnums.
Configuration menu - View commit details
-
Copy full SHA for 9b07c30 - Browse repository at this point
Copy the full SHA 9b07c30View commit details -
rtpmanager/rtpsource: minor warning log fix
Be a bit more consistent with how we log ssrcs
Configuration menu - View commit details
-
Copy full SHA for 91514cc - Browse repository at this point
Copy the full SHA 91514ccView commit details -
rtpptdemux: send lost-event to all pads
ptdemux can't know which of the payloadtypes the lost-event is actually meant for, and taking a guess of it being the "previous" one is going to be wrong in many cases.
Configuration menu - View commit details
-
Copy full SHA for bfd0be0 - Browse repository at this point
Copy the full SHA bfd0be0View commit details -
tests/rtpbin: add test for rtp and rtcp arriving simultaneously
And some cleanup/warning fixes.
Configuration menu - View commit details
-
Copy full SHA for f22c4a6 - Browse repository at this point
Copy the full SHA f22c4a6View commit details -
Configuration menu - View commit details
-
Copy full SHA for 7a6dc66 - Browse repository at this point
Copy the full SHA 7a6dc66View commit details -
Configuration menu - View commit details
-
Copy full SHA for 4e29484 - Browse repository at this point
Copy the full SHA 4e29484View commit details -
Configuration menu - View commit details
-
Copy full SHA for 5fe85b0 - Browse repository at this point
Copy the full SHA 5fe85b0View commit details -
vp9enc: disable VP9E_SET_COLOR_SPACE for now
...it crashes our libvpx...
Configuration menu - View commit details
-
Copy full SHA for ad9980a - Browse repository at this point
Copy the full SHA ad9980aView commit details -
rtpvp8depay: upgrade to use custom meta (FIXUP)
Need to sort out this mess later on!
Configuration menu - View commit details
-
Copy full SHA for eeca121 - Browse repository at this point
Copy the full SHA eeca121View commit details -
rtpmanager: add header-extension for Region of Interest.
It will take the GstVideoRegionOfInterestMeta, and turn one of its specified types into a RTP header-extension. Co-authored-by: Camilo Celis Guzman <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 850e6cf - Browse repository at this point
Copy the full SHA 850e6cfView commit details -
Configuration menu - View commit details
-
Copy full SHA for 382fedc - Browse repository at this point
Copy the full SHA 382fedcView commit details -
Configuration menu - View commit details
-
Copy full SHA for aeec498 - Browse repository at this point
Copy the full SHA aeec498View commit details -
rtph261depay: don't set the DELTA_UNIT flag for H.261
Basically, all frames are "i-frames" here.
Configuration menu - View commit details
-
Copy full SHA for 26037c3 - Browse repository at this point
Copy the full SHA 26037c3View commit details -
vpxenc: move temporal-scalability-layer to GValueArray
Can't have a mix of GstValueArray and GValueArray here.
Configuration menu - View commit details
-
Copy full SHA for 221af44 - Browse repository at this point
Copy the full SHA 221af44View commit details -
Configuration menu - View commit details
-
Copy full SHA for 90cb0dc - Browse repository at this point
Copy the full SHA 90cb0dcView commit details -
Configuration menu - View commit details
-
Copy full SHA for 3713ab5 - Browse repository at this point
Copy the full SHA 3713ab5View commit details -
Configuration menu - View commit details
-
Copy full SHA for bdd30fc - Browse repository at this point
Copy the full SHA bdd30fcView commit details -
Configuration menu - View commit details
-
Copy full SHA for 3dd78af - Browse repository at this point
Copy the full SHA 3dd78afView commit details -
Configuration menu - View commit details
-
Copy full SHA for 91e8ca5 - Browse repository at this point
Copy the full SHA 91e8ca5View commit details -
Configuration menu - View commit details
-
Copy full SHA for d188d25 - Browse repository at this point
Copy the full SHA d188d25View commit details -
Configuration menu - View commit details
-
Copy full SHA for 3120837 - Browse repository at this point
Copy the full SHA 3120837View commit details -
Configuration menu - View commit details
-
Copy full SHA for 0ae9161 - Browse repository at this point
Copy the full SHA 0ae9161View commit details -
Configuration menu - View commit details
-
Copy full SHA for 8ecc85d - Browse repository at this point
Copy the full SHA 8ecc85dView commit details -
Configuration menu - View commit details
-
Copy full SHA for e04fa91 - Browse repository at this point
Copy the full SHA e04fa91View commit details -
Configuration menu - View commit details
-
Copy full SHA for 0b2ed54 - Browse repository at this point
Copy the full SHA 0b2ed54View commit details -
SCTP: require lock to be held before changing state
The state change function will release the lock before emitting the property change event and relock afterwards. This simplifies the flow of calling code dramatically. Co-authored-by: John-Mark Bell <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 7c9a53c - Browse repository at this point
Copy the full SHA 7c9a53cView commit details -
Configuration menu - View commit details
-
Copy full SHA for 4d4a7a6 - Browse repository at this point
Copy the full SHA 4d4a7a6View commit details -
Configuration menu - View commit details
-
Copy full SHA for 8c29204 - Browse repository at this point
Copy the full SHA 8c29204View commit details -
SCTP: reset association when disconnected
Co-authored-by: John-Mark Bell <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 935b3c3 - Browse repository at this point
Copy the full SHA 935b3c3View commit details -
Configuration menu - View commit details
-
Copy full SHA for bbed90a - Browse repository at this point
Copy the full SHA bbed90aView commit details -
Configuration menu - View commit details
-
Copy full SHA for e08d7a7 - Browse repository at this point
Copy the full SHA e08d7a7View commit details -
Configuration menu - View commit details
-
Copy full SHA for 1cc9afb - Browse repository at this point
Copy the full SHA 1cc9afbView commit details -
SCTP: apply aggressive heartbeat using setsockopt
Additionally, reduce the numbers.
Configuration menu - View commit details
-
Copy full SHA for 5449d2a - Browse repository at this point
Copy the full SHA 5449d2aView commit details -
Configuration menu - View commit details
-
Copy full SHA for 1acf79d - Browse repository at this point
Copy the full SHA 1acf79dView commit details -
Configuration menu - View commit details
-
Copy full SHA for 65b695f - Browse repository at this point
Copy the full SHA 65b695fView commit details -
Configuration menu - View commit details
-
Copy full SHA for 0c7071d - Browse repository at this point
Copy the full SHA 0c7071dView commit details -
Configuration menu - View commit details
-
Copy full SHA for 99db846 - Browse repository at this point
Copy the full SHA 99db846View commit details -
Configuration menu - View commit details
-
Copy full SHA for 829f50b - Browse repository at this point
Copy the full SHA 829f50bView commit details -
Configuration menu - View commit details
-
Copy full SHA for eab4545 - Browse repository at this point
Copy the full SHA eab4545View commit details -
Configuration menu - View commit details
-
Copy full SHA for bc1419a - Browse repository at this point
Copy the full SHA bc1419aView commit details -
Configuration menu - View commit details
-
Copy full SHA for ef7cdd6 - Browse repository at this point
Copy the full SHA ef7cdd6View commit details -
Configuration menu - View commit details
-
Copy full SHA for 82232b3 - Browse repository at this point
Copy the full SHA 82232b3View commit details -
Co-authored-by: John-Mark Bell <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for b27bf31 - Browse repository at this point
Copy the full SHA b27bf31View commit details -
Configuration menu - View commit details
-
Copy full SHA for c200d69 - Browse repository at this point
Copy the full SHA c200d69View commit details -
usrsctp: port pexip/usrsctp 'Fix deadlock during shutdown'
commit a8b63ae927b6dddeeeae5d83803a0e8e1c7b2a3f Author: John Bassett <[email protected]> Date: Tue May 21 18:23:05 2019 +0100 Fix deadlock during shutdown Deadlock was caused by one thread calling close on a socket whilst another thread was blocked on a call to recvmsg. In this scenario the recv thread would wake up after the recvmsg had timed out but then attempt another read from the now closed socket. Normally this would result in an error, but if the file descriptor had been reused in the meantime the recvmsg call could end up blocking forever trying to read from the wrong fd. - Make sure the socket fds used are reset to -1 before closing to prevent other threads attempting to read from a socket that has already been closed. - Call shutdown on sockets before closing to guarantee other threads reading from the socket are unblocked immediately. - Check for return of 0 when reading from a socket to handle sockets that have been cleanly closed from another thread.
Configuration menu - View commit details
-
Copy full SHA for b06de7a - Browse repository at this point
Copy the full SHA b06de7aView commit details -
usrsctp: port pexip/usrsctp 'Forcibly cancel iterator thread'
commit 3a3cb88c388919b9348b828becb8576a580eccc4 Author: John-Mark Bell <[email protected]> Date: Thu Aug 15 22:22:59 2019 +0100 Forcibly cancel iterator thread
Configuration menu - View commit details
-
Copy full SHA for ccf8cc9 - Browse repository at this point
Copy the full SHA ccf8cc9View commit details -
usrsctp: port pexip/usrsctp 'Remove iterator thread'
commit 7d050e2140efd73b408eecac5f4cede116b7f1f1 Author: John-Mark Bell <[email protected]> Date: Sat Aug 17 19:15:30 2019 +0100 Remove iterator thread This is not necessary a world using AF_CONN only. Fix the one place where an iterator would be used to clean up directly in this case.
Configuration menu - View commit details
-
Copy full SHA for 65b2d89 - Browse repository at this point
Copy the full SHA 65b2d89View commit details -
usrsctp: port pexip/usrsctp 'init_random rename to init_usrsctp_rando…
…m as conflicts with cPython bindings' commit ecf2de543c3157e507e9fc2e7c8319bc443ede39 Author: Thomas Williams <[email protected]> Date: Thu Jun 17 14:36:26 2021 +0100 init_random rename to init_usrsctp_random as conflicts with cPython bindings
Configuration menu - View commit details
-
Copy full SHA for 3151696 - Browse repository at this point
Copy the full SHA 3151696View commit details -
Configuration menu - View commit details
-
Copy full SHA for 0917f10 - Browse repository at this point
Copy the full SHA 0917f10View commit details -
SCTP: retry after failure to bind with EADDRINUSE
Co-authored-by: John-Mark Bell <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 68a77ab - Browse repository at this point
Copy the full SHA 68a77abView commit details -
Configuration menu - View commit details
-
Copy full SHA for e727bea - Browse repository at this point
Copy the full SHA e727beaView commit details -
Configuration menu - View commit details
-
Copy full SHA for 6607f26 - Browse repository at this point
Copy the full SHA 6607f26View commit details -
Configuration menu - View commit details
-
Copy full SHA for 285b53a - Browse repository at this point
Copy the full SHA 285b53aView commit details -
Configuration menu - View commit details
-
Copy full SHA for 4319b91 - Browse repository at this point
Copy the full SHA 4319b91View commit details -
Configuration menu - View commit details
-
Copy full SHA for 111de24 - Browse repository at this point
Copy the full SHA 111de24View commit details -
[pexhack] re-add gst_debug_syslog option
Changes lost in monorepo merge; originally from 7bbcd29
Configuration menu - View commit details
-
Copy full SHA for 72649e8 - Browse repository at this point
Copy the full SHA 72649e8View commit details -
Configuration menu - View commit details
-
Copy full SHA for fa474f6 - Browse repository at this point
Copy the full SHA fa474f6View commit details -
Configuration menu - View commit details
-
Copy full SHA for 2cb91c8 - Browse repository at this point
Copy the full SHA 2cb91c8View commit details -
gstreamer/harness: make pull timesout configurable
Although 60 seconds is often enough, we have seen cases where the harnessed pipeline could taken longer than that to process a buffer. This is especially true when we run under additional tools like Valgrind or Debug builds.
Configuration menu - View commit details
-
Copy full SHA for 7e6b9f7 - Browse repository at this point
Copy the full SHA 7e6b9f7View commit details -
rtprtxsend: add "limiting" and "stuffing"
For "limiting" this means using the token bucket algorithm, allowing a sender to control the bitrate being consumed by rtx-packets, in case of an "out-of-control" receiver asking for way too much packets. 'max-kbps' sets the maximum kilobit per second that the RTX element will give out based on requests, limiting the ability for a receiver to congest the available network bandwidth. 'max-bucket-size' is related to the token bucket algorithm, and will allow control of the "burstiness" of the RTX requests, preventing a huge spike in the network. For "stuffing" this means adding RTX packets as stuffing to meet the total target bitrate 'stuffing-kbps'. There are cases when output rate sent on the network needs to be near constant, but the media isn't (Encoders producing variable bitrates) Also, for probing the network for available bitrate, the stuffing can be used to create redundant information at a certain bitrate. Co-authored-by: Tulio Beloqui <[email protected]> Co-authored-by: Stian Selnes <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 56b7ddf - Browse repository at this point
Copy the full SHA 56b7ddfView commit details -
Configuration menu - View commit details
-
Copy full SHA for 9f38255 - Browse repository at this point
Copy the full SHA 9f38255View commit details -
[pexhack] rtphdrextroi: tmp support of multiple roi-types
Squashed commit of the following: commit c2e3ed1 Author: Camilo Celis Guzman <[email protected]> Date: Tue Nov 2 22:30:12 2021 +0900 rtphdrextroi: correct max. number of roi-types commit ac3c6c4 Author: Camilo Celis Guzman <[email protected]> Date: Fri Oct 29 23:00:14 2021 +0900 rtphdrextroi: add test for multiple roi-types commit 1fd0bc8 Author: Camilo Celis Guzman <[email protected]> Date: Thu Oct 28 19:42:04 2021 +0900 rtphdrextroi: WIP: multiple roi-types support commit 4b9f2d8 Author: Camilo Celis Guzman <[email protected]> Date: Thu Oct 28 17:19:57 2021 +0900 rtphdrextroi: add test for case of multiple RoI metas with the same roi-type The expected behaviour is that only one of them is payloaded commit b17f460 Author: Camilo Celis Guzman <[email protected]> Date: Thu Oct 28 17:03:55 2021 +0900 rtphdrextroi: make roi-type construct only
Configuration menu - View commit details
-
Copy full SHA for 6d29786 - Browse repository at this point
Copy the full SHA 6d29786View commit details -
Configuration menu - View commit details
-
Copy full SHA for f131923 - Browse repository at this point
Copy the full SHA f131923View commit details -
check: Fix valgrid suppresion for debug function list
Fix suppresion to support release and debug builds. Here is debug build call stack: ``` ==10707== by 0x48B5520: g_malloc (gmem.c:106) ==10707== by 0x48D19DC: g_slice_alloc (gslice.c:1069) ==10707== by 0x48D3947: g_slist_copy_deep (gslist.c:619) ==10707== by 0x48D38B8: g_slist_copy (gslist.c:567) ==10707== by 0x4ADC90B: gst_debug_remove_with_compare_func (gstinfo.c:1504) ``` In release build `g_slist_copy (gslist.c:567)` got inlined: ``` ==15419== by 0x48963E0: g_malloc (gmem.c:106) ==15419== by 0x48AA382: g_slice_alloc (gslice.c:1069) ==15419== by 0x48AB732: g_slist_copy_deep (gslist.c:619) ==15419== by 0x4A39B8F: gst_debug_remove_with_compare_func (gstinfo.c:1504) ```
Configuration menu - View commit details
-
Copy full SHA for d10dfdd - Browse repository at this point
Copy the full SHA d10dfddView commit details -
Configuration menu - View commit details
-
Copy full SHA for bdde997 - Browse repository at this point
Copy the full SHA bdde997View commit details -
gstvideodecoder: set all-headers to TRUE in the GstForceKeyUnit when
doing a sync point request all_headers set to FALSE stops the FIR request in gstrtpsession.
Configuration menu - View commit details
-
Copy full SHA for 36dd564 - Browse repository at this point
Copy the full SHA 36dd564View commit details -
Configuration menu - View commit details
-
Copy full SHA for 9df5b9f - Browse repository at this point
Copy the full SHA 9df5b9fView commit details -
gstglcolorconver: only set target target texture if any
Prevents logical error as if defined as a stack variable it could contain garbage data if uninitialized.
Configuration menu - View commit details
-
Copy full SHA for 0e00c92 - Browse repository at this point
Copy the full SHA 0e00c92View commit details -
Configuration menu - View commit details
-
Copy full SHA for 9baf1cf - Browse repository at this point
Copy the full SHA 9baf1cfView commit details -
meson: [msvc] only add warnings 4013, 4101 and 4189 if glib_checks is
disabled If the checks are not built, it will leave behind some "unused" variables, so this warnings will always trigger.
Configuration menu - View commit details
-
Copy full SHA for 52268d8 - Browse repository at this point
Copy the full SHA 52268d8View commit details -
Configuration menu - View commit details
-
Copy full SHA for 593595c - Browse repository at this point
Copy the full SHA 593595cView commit details -
Configuration menu - View commit details
-
Copy full SHA for 0ab7d19 - Browse repository at this point
Copy the full SHA 0ab7d19View commit details -
Configuration menu - View commit details
-
Copy full SHA for 6066aa7 - Browse repository at this point
Copy the full SHA 6066aa7View commit details -
libcheck: use SIGABRT instead of SIGKILL on timeout
This allows user-level signal handler to catch this Co-authored-by: Frederik Vestre <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for ae95a29 - Browse repository at this point
Copy the full SHA ae95a29View commit details -
Configuration menu - View commit details
-
Copy full SHA for f8f9a09 - Browse repository at this point
Copy the full SHA f8f9a09View commit details -
Configuration menu - View commit details
-
Copy full SHA for 05f84e5 - Browse repository at this point
Copy the full SHA 05f84e5View commit details -
androidmedia: added gstacamdeviceprovider
Enables device enumeration and monitor using the ACameraManager API
Configuration menu - View commit details
-
Copy full SHA for 8c787bb - Browse repository at this point
Copy the full SHA 8c787bbView commit details -
Configuration menu - View commit details
-
Copy full SHA for 5e84e1a - Browse repository at this point
Copy the full SHA 5e84e1aView commit details -
SCTPDec: don't unref NULL association
The association can be NULL if the element was never started, or if the association ID was already in use. Be symmetrical with SCTPEnc and check this.
Configuration menu - View commit details
-
Copy full SHA for 3d53270 - Browse repository at this point
Copy the full SHA 3d53270View commit details -
gstamdeviceprovider: fixed a race condition in which the callback would
execute regardless of the provider being stopped.
Configuration menu - View commit details
-
Copy full SHA for 2a21d75 - Browse repository at this point
Copy the full SHA 2a21d75View commit details -
applemedia/vtenc: fix deadlock
Calling destroy_session() without first calling finish() deadlocks almost every time, due to the explanation given in gst_vtenc_finish_encoding()
Configuration menu - View commit details
-
Copy full SHA for 63cadcf - Browse repository at this point
Copy the full SHA 63cadcfView commit details -
SCTP: moving the removal of the associations from dispose() to finali…
…ze() - Decrementing number_of_associations in dispose() can be troublesome, as as per documentation, the dispose function can be called many times. - Also removed extra check if dispose is present, as parent class (GObject) ensures it is available.
Configuration menu - View commit details
-
Copy full SHA for 846c9b2 - Browse repository at this point
Copy the full SHA 846c9b2View commit details -
Configuration menu - View commit details
-
Copy full SHA for a48937f - Browse repository at this point
Copy the full SHA a48937fView commit details -
Configuration menu - View commit details
-
Copy full SHA for ab38595 - Browse repository at this point
Copy the full SHA ab38595View commit details -
Configuration menu - View commit details
-
Copy full SHA for cdb83e1 - Browse repository at this point
Copy the full SHA cdb83e1View commit details -
gstpad: PEXHACK: log chain(list)funcs that take longer than 100ms
Remove this before releasing please
Configuration menu - View commit details
-
Copy full SHA for 9434083 - Browse repository at this point
Copy the full SHA 9434083View commit details -
Configuration menu - View commit details
-
Copy full SHA for ebe3f59 - Browse repository at this point
Copy the full SHA ebe3f59View commit details -
amcvideoenc: fix mess around level and profile for H.264
Also specify the alignment as part of caps always.
Configuration menu - View commit details
-
Copy full SHA for 7ad6b42 - Browse repository at this point
Copy the full SHA 7ad6b42View commit details -
Configuration menu - View commit details
-
Copy full SHA for 786c51f - Browse repository at this point
Copy the full SHA 786c51fView commit details -
vpxenc: FIXME: revert upstream changes
These made one of our vpx test fail. GST_DEBUG=*vp*:LOG GST_CHECKS=vpx_enc_rate_limit_key_frames python3 /pexip/external/meson/meson.py test vpx -vv Look into why and how more closely!
Configuration menu - View commit details
-
Copy full SHA for e9058c6 - Browse repository at this point
Copy the full SHA e9058c6View commit details -
Configuration menu - View commit details
-
Copy full SHA for 2d42e81 - Browse repository at this point
Copy the full SHA 2d42e81View commit details -
Configuration menu - View commit details
-
Copy full SHA for 1ae1695 - Browse repository at this point
Copy the full SHA 1ae1695View commit details -
rtpsession: don't report same stats twice
The notify is already on a timer, and notifying again outside this breaks the contract of not receiving more stats.
Configuration menu - View commit details
-
Copy full SHA for 9eab3d2 - Browse repository at this point
Copy the full SHA 9eab3d2View commit details -
SCTP: fixing pthread_mutex_destroy call on a destroyed mutex
The initialization of the iterators locks depended on SCTP_ITERATOR definition, but sctp_pcb_finish() calls destroy on them without that check. Note that also these locks are accessed without any check on SCTP_ITERATOR so doing the initialization here fixes that problem too.
Configuration menu - View commit details
-
Copy full SHA for 08cb9f0 - Browse repository at this point
Copy the full SHA 08cb9f0View commit details -
Configuration menu - View commit details
-
Copy full SHA for e9eeba7 - Browse repository at this point
Copy the full SHA e9eeba7View commit details -
Configuration menu - View commit details
-
Copy full SHA for 46e1288 - Browse repository at this point
Copy the full SHA 46e1288View commit details -
SCTP: refactoring encoder and decoder
- Move the cast macro to do a basic cast instead of check gtype instance as that one access a global lock. - Configure and cleanup the sctp association only on state change from NULL->READY and READY->NULL. - Moved the start/stop of the srcpad task in the activate_mode function, rather than the state change. This fixes the task thread accessing invalid data. Co-Authored-By: Håvard Graff <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 83e8e04 - Browse repository at this point
Copy the full SHA 83e8e04View commit details -
SCTP: fixed race condition and memory leak
gst_sctp_association_get can call g_object_ref on a destroyed sctp association because the unref calls do not hold the associations_lock. To solve this we add an API to unref the association while holding the lock. Moved the calls to usrsctp_init/finish to get/unref() respectively to not depend on GObjects behavior, keep the counting of the associations in the init/finalize. Also added g_hash_table_destroy() calls to the association hash table after the last association is destroyed.
Configuration menu - View commit details
-
Copy full SHA for 419fd4e - Browse repository at this point
Copy the full SHA 419fd4eView commit details -
gstelement: do not hold the element lock when activating the pad in
gst_element_add_pad() - Fixes a deadlock in which we will hold the lock inside gst_element_add_pad() and then later in gst_element_post_message_default(). This is triggered by calling gst_pad_start_task() from the pads activate_mode() function. (cherry picked from commit 22cc976)
Configuration menu - View commit details
-
Copy full SHA for e876bbe - Browse repository at this point
Copy the full SHA e876bbeView commit details -
SCTP: add the srcpad to the sctpdec element before activing it
This fixes a potential NULL access on the pads parent inside gst_sctp_data_srcpad_loop (cherry picked from commit fc409b9)
Configuration menu - View commit details
-
Copy full SHA for 063c345 - Browse repository at this point
Copy the full SHA 063c345View commit details -
Configuration menu - View commit details
-
Copy full SHA for 5e0fed2 - Browse repository at this point
Copy the full SHA 5e0fed2View commit details -
Configuration menu - View commit details
-
Copy full SHA for 4b9f5dd - Browse repository at this point
Copy the full SHA 4b9f5ddView commit details -
Configuration menu - View commit details
-
Copy full SHA for bdcbfd2 - Browse repository at this point
Copy the full SHA bdcbfd2View commit details -
Configuration menu - View commit details
-
Copy full SHA for aa1f907 - Browse repository at this point
Copy the full SHA aa1f907View commit details -
Configuration menu - View commit details
-
Copy full SHA for c737130 - Browse repository at this point
Copy the full SHA c737130View commit details -
rtpjitterbuffer: Default do-lost to TRUE for rtpjitterbuffer [pexhack]
Just for convenience
Configuration menu - View commit details
-
Copy full SHA for a86619c - Browse repository at this point
Copy the full SHA a86619cView commit details -
Revert "rtpjitterbuffer: Ensure to not take caps with the wrong pt fo…
…r getting the clock-rate" This reverts commit 608b4ee.
Configuration menu - View commit details
-
Copy full SHA for bf9827b - Browse repository at this point
Copy the full SHA bf9827bView commit details -
Configuration menu - View commit details
-
Copy full SHA for ca55ad0 - Browse repository at this point
Copy the full SHA ca55ad0View commit details -
rtpjitterbuffer: rtspsrc: Rename seqnum-base caps field to seqnum-offset
RTP caps from payloaders use seqnum-offset, hence the jitterbuffer should use the same. Same for rtspsrc.
Configuration menu - View commit details
-
Copy full SHA for 408e689 - Browse repository at this point
Copy the full SHA 408e689View commit details -
rtpjitterbuffer: improve calculation of jitter and add more stats
Given a constant source of jitter, the total-avg-jitter will be the "correct" amount of jitter experienced. Comparing this number to the running average for many long-running tests, having a slightly larger window seems to approximate the jitter much better. In addition we avoid weighting 0 too much when the initial calculations are done by ramping up the window and not counting the very first measurement.
Configuration menu - View commit details
-
Copy full SHA for 00328bd - Browse repository at this point
Copy the full SHA 00328bdView commit details -
rtpjitterbuffer: Optimize stats by caching quarks
To avoid the GQuark global lock per entry.
Configuration menu - View commit details
-
Copy full SHA for 70d541e - Browse repository at this point
Copy the full SHA 70d541eView commit details -
Configuration menu - View commit details
-
Copy full SHA for 2d740fc - Browse repository at this point
Copy the full SHA 2d740fcView commit details -
rtpjitterbuffer: don't calculate packet-rate for jumping seqnums
As demonstrated by the test, you can end up in situations where your packet-rate gets insanly high, and you produce thousands of RTX-timers /lost events, when the big-gap logic is there to prevent exactly that.
Configuration menu - View commit details
-
Copy full SHA for 289ac2b - Browse repository at this point
Copy the full SHA 289ac2bView commit details -
rtpjitterbuffer: add failing test that reproduces a stall
So what happens here is that at some point we receive packet #10508, which triggers the big gap logic, placing that buffer inside the gap_packets GList: gst_rtp_jitter_buffer_chain: expected #10529, got #10508, gap of -21 handle_big_gap_buffer: buffer too old -21 < -20, first one - waiting Then packet #10550 arrives: gst_rtp_jitter_buffer_chain: expected #10530, got #10550, gap of 20 This causes too many timers to be pending, causing a reset: gst_rtp_jitter_buffer_chain: 182 pending timers > 180 - resetting gst_rtp_jitter_buffer_reset:< flush and reset jitterbuffer Inside the reset() code, the next_seqnum is set to the first buffer in gap_packets, #10508: gst_rtp_jitter_buffer_reset: setting next_seqnum to #10508 And reset() calls chain() with all its gap_packets, pushing it out: gst_rtp_jitter_buffer_chain: Received packet #10508 at time 123:45:05.364650711, discont 0, rtx 1 However, #10508 is actually an RTX packet, so it gets dropped: gst_rtp_jitter_buffer_chain:<rtpjitterbuffer0> Unsolicited RTX packet #10508 detected, dropping Next #10550 is pushed from reset(): gst_rtp_jitter_buffer_chain:<rtpjitterbuffer0> Received packet #10550 at time 123:45:05.516136868, discont 0, rtx 0 But because next_seqnum is set to #10508, we get this: handle_next_buffer: Sequence number GAP detected: expected 10508 instead of 10550 (42 missing) So now the jitterbuffer is completely stalled, all buffer arriving now will just queue up until it sees the promised #10508, which will be ~65000 packets from now, and we will trigger the "full" logic before then, causing a deadlock.
Configuration menu - View commit details
-
Copy full SHA for 863e22e - Browse repository at this point
Copy the full SHA 863e22eView commit details -
Configuration menu - View commit details
-
Copy full SHA for 190f6e6 - Browse repository at this point
Copy the full SHA 190f6e6View commit details -
Configuration menu - View commit details
-
Copy full SHA for 5b4754d - Browse repository at this point
Copy the full SHA 5b4754dView commit details -
Configuration menu - View commit details
-
Copy full SHA for 7284f57 - Browse repository at this point
Copy the full SHA 7284f57View commit details -
rtpjitterbuffer: fixing ulpfec packets used in pts and skew calculati…
…ons. Co-authored-by: Håvard Graff <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 60b199b - Browse repository at this point
Copy the full SHA 60b199bView commit details -
Configuration menu - View commit details
-
Copy full SHA for b39a2b6 - Browse repository at this point
Copy the full SHA b39a2b6View commit details -
gstrtpjitterbuffer: Improve avg_jitter calculation.
Ignore sequential packets with the same rtp time stamp during jitter estimation.
Configuration menu - View commit details
-
Copy full SHA for eb68f14 - Browse repository at this point
Copy the full SHA eb68f14View commit details -
Configuration menu - View commit details
-
Copy full SHA for 9c8ef95 - Browse repository at this point
Copy the full SHA 9c8ef95View commit details -
PEXHACK: revert gstvideorate to old version
This to unbreak simulcast tests
Configuration menu - View commit details
-
Copy full SHA for a6ee00d - Browse repository at this point
Copy the full SHA a6ee00dView commit details
Commits on Oct 27, 2022
-
Configuration menu - View commit details
-
Copy full SHA for 2e01bfd - Browse repository at this point
Copy the full SHA 2e01bfdView commit details
Commits on Oct 28, 2022
-
Configuration menu - View commit details
-
Copy full SHA for d56b7aa - Browse repository at this point
Copy the full SHA d56b7aaView commit details
Commits on Nov 1, 2022
-
Configuration menu - View commit details
-
Copy full SHA for 7472535 - Browse repository at this point
Copy the full SHA 7472535View commit details -
gstamcvideoenc: avoid flushing the encoder twice if we have already f…
…lushed - Fixes a java exception of flushing the encoder in an ilegal state
Configuration menu - View commit details
-
Copy full SHA for 180da6f - Browse repository at this point
Copy the full SHA 180da6fView commit details
Commits on Nov 3, 2022
-
Configuration menu - View commit details
-
Copy full SHA for 0943cbc - Browse repository at this point
Copy the full SHA 0943cbcView commit details -
ahc2src: add PROP_EXPOSURE_MODE stub to avoid error of not implementing
GstPhotography interface
Configuration menu - View commit details
-
Copy full SHA for a9fd896 - Browse repository at this point
Copy the full SHA a9fd896View commit details
Commits on Nov 17, 2022
-
good/jitterbuffer: initialize SSRC on RTCP chain method
It is possible to log out an uninitialized SSRC. Reported as by clang's scan build.
Configuration menu - View commit details
-
Copy full SHA for 0e5b1fb - Browse repository at this point
Copy the full SHA 0e5b1fbView commit details -
gstvalue/test: initialize unset string
It is possible to do a comparison when the ASSERT checks are disabled
Configuration menu - View commit details
-
Copy full SHA for 9995bb6 - Browse repository at this point
Copy the full SHA 9995bb6View commit details
Commits on Nov 21, 2022
-
Configuration menu - View commit details
-
Copy full SHA for 3c7dcac - Browse repository at this point
Copy the full SHA 3c7dcacView commit details
Commits on Nov 22, 2022
-
Configuration menu - View commit details
-
Copy full SHA for 8a36826 - Browse repository at this point
Copy the full SHA 8a36826View commit details
Commits on Nov 25, 2022
-
Configuration menu - View commit details
-
Copy full SHA for 09b8049 - Browse repository at this point
Copy the full SHA 09b8049View commit details
Commits on Dec 7, 2022
-
Configuration menu - View commit details
-
Copy full SHA for cb30324 - Browse repository at this point
Copy the full SHA cb30324View commit details
Commits on Dec 19, 2022
-
Configuration menu - View commit details
-
Copy full SHA for 2de52d4 - Browse repository at this point
Copy the full SHA 2de52d4View commit details -
Configuration menu - View commit details
-
Copy full SHA for 6f36269 - Browse repository at this point
Copy the full SHA 6f36269View commit details -
SCTP: cleaned up code in gst_sctp_association_change_state
Changed code to have gst_sctp_association_change_state() and unlocked() version.
Configuration menu - View commit details
-
Copy full SHA for cbff58d - Browse repository at this point
Copy the full SHA cbff58dView commit details -
Configuration menu - View commit details
-
Copy full SHA for c173e47 - Browse repository at this point
Copy the full SHA c173e47View commit details
Commits on Dec 22, 2022
-
SCTP: ensure thread-safe access in gst_sctp_association_reset_stream
We need to ensure that members like state, sctp_assoc_id and sctp_ass_sock are accessed with a lock.
Configuration menu - View commit details
-
Copy full SHA for 483e1f2 - Browse repository at this point
Copy the full SHA 483e1f2View commit details
Commits on Jan 2, 2023
-
SCTP: access sctp_ass_sock with a lock in gst_sctp_association_send_data
To be consistent with the rest of the code.
Configuration menu - View commit details
-
Copy full SHA for 258c2a0 - Browse repository at this point
Copy the full SHA 258c2a0View commit details -
Configuration menu - View commit details
-
Copy full SHA for fddc320 - Browse repository at this point
Copy the full SHA fddc320View commit details -
SCTP: replace g_object_bind_property with g_object_set
This fixes a very nasty double-free on the SCTP association in which a SCTP association weak-reference will be dangling around and the elements unref() will land in an unlocked gst_sctp_association_finalize() call. Other changes: - Added locking around the SCTP association in both encoder and decoder. - Added ref() API to the SCTP association, to be consistent with unref(). - Calling SCTP associations API with a ref() of the current association to safely avoid finalizing a last reference. - Moved maybe_set_state_to_ready to an unlocked() version, since in the contexts it is called there is no need to release and acquire the lock.
Configuration menu - View commit details
-
Copy full SHA for fda49d5 - Browse repository at this point
Copy the full SHA fda49d5View commit details
Commits on Jan 3, 2023
-
Configuration menu - View commit details
-
Copy full SHA for 94d85da - Browse repository at this point
Copy the full SHA 94d85daView commit details -
Configuration menu - View commit details
-
Copy full SHA for 1adde14 - Browse repository at this point
Copy the full SHA 1adde14View commit details -
rtptwcc: handle reordered packets within the same report-interval
No need to treat them as lost when we can accurately report the reordering in the TWCC feedback message.
Configuration menu - View commit details
-
Copy full SHA for 0b1ba8d - Browse repository at this point
Copy the full SHA 0b1ba8dView commit details -
rtptwcc: mimic chrome behavior in reporting reordered packets
Basically being able to jump "back" and report older seqnums, as well as reporting the same packet multiple times.
Configuration menu - View commit details
-
Copy full SHA for d60a9d6 - Browse repository at this point
Copy the full SHA d60a9d6View commit details -
Configuration menu - View commit details
-
Copy full SHA for b9bf7fb - Browse repository at this point
Copy the full SHA b9bf7fbView commit details -
Configuration menu - View commit details
-
Copy full SHA for 779bc8f - Browse repository at this point
Copy the full SHA 779bc8fView commit details -
* introduce a configurable window in which to compute various stats * bugfixes in the twcc report generating code * optimizations around using less lists of packets * much improved handling of reordered packets
Configuration menu - View commit details
-
Copy full SHA for f1d9900 - Browse repository at this point
Copy the full SHA f1d9900View commit details
Commits on Jan 4, 2023
-
Configuration menu - View commit details
-
Copy full SHA for 60f798b - Browse repository at this point
Copy the full SHA 60f798bView commit details
Commits on Jan 23, 2023
-
Revert "gstpad: PEXHACK: log chain(list)funcs that take longer than 1…
…00ms" This reverts commit 9434083.
Configuration menu - View commit details
-
Copy full SHA for ceea6d7 - Browse repository at this point
Copy the full SHA ceea6d7View commit details
Commits on Jan 27, 2023
-
Configuration menu - View commit details
-
Copy full SHA for d2d4c9f - Browse repository at this point
Copy the full SHA d2d4c9fView commit details
Commits on Feb 6, 2023
-
Configuration menu - View commit details
-
Copy full SHA for 59fdbac - Browse repository at this point
Copy the full SHA 59fdbacView commit details -
SCTP: refactor to fix encoder use-after-free bug
Because of the nature of not having a clear ownership between the association and the elements (encoder or decoder) is not possible to rely on the associations signals to have a valid reference on the encoder. This also applies for the decoder. Since g_signal_connect_object is not thread-safe (more on this in the documentation) and we can't hold the association mutex while emitting a signal, it is completely possible -as our stress tests demonstrate- that we ran into data races. To fix this situation, I introduce proper callback structures per element context and inside the association we always keep a reference to the element while calling the related callback.
Configuration menu - View commit details
-
Copy full SHA for 0c04553 - Browse repository at this point
Copy the full SHA 0c04553View commit details -
rtptwcc: update with scan-build fixes
(cherry picked from commit 1550d22)
Configuration menu - View commit details
-
Copy full SHA for 9729feb - Browse repository at this point
Copy the full SHA 9729febView commit details
Commits on Feb 17, 2023
-
rtptwcc: use the same size-per-packet as on the wire
Comparing with wireshark, this gives the correct size per packet on the wire vs. what is used internally. Also add the original RTP sequence-number to the logging for easy correlation.
Configuration menu - View commit details
-
Copy full SHA for 1b0d273 - Browse repository at this point
Copy the full SHA 1b0d273View commit details -
rtprtxsend: introduce stuffing-max-burst-packets and use padding
In order to limit the amount of packets that is used for stuffing, we both introduce a hard limit of how many packets a "burst" can consist of, but also use padding to make packets slightly bigger, and hence reduce the need to send quite so many.
Configuration menu - View commit details
-
Copy full SHA for 6a9c397 - Browse repository at this point
Copy the full SHA 6a9c397View commit details
Commits on Feb 28, 2023
-
funnel: only forward events on active pads
IF we are going to forward events on non-active pads, we need to send our sticky events prior to this event to make it valid. However, this will lead to the situation of getting a caps change downstream for any event being received on either pad in the funnel. In the case of a decoder sitting downsstream, this would mean potentially replacing that decoder on a caps-change, meaning an "irrelevant" event received on a non-active pad would cause a decoder to lose all state, potentially needing an I-frame or similar.
Configuration menu - View commit details
-
Copy full SHA for 729695c - Browse repository at this point
Copy the full SHA 729695cView commit details
Commits on Mar 1, 2023
-
Revert "rtpjitterbuffer: DTX implementation"
This reverts commit 7284f57.
Configuration menu - View commit details
-
Copy full SHA for c944289 - Browse repository at this point
Copy the full SHA c944289View commit details -
gstaudiodecoder: implemented DTX API
Subclasses can now start DTX by calling gst_audio_decoder_start_dtx(), this will spawn a task on the decoders srcpad and generate silent buffers until a new audio packet or a gap event arrives. An API for checking if the decoder is running DTX has been also added in case subclasses need context.
Configuration menu - View commit details
-
Copy full SHA for 62afe0a - Browse repository at this point
Copy the full SHA 62afe0aView commit details
Commits on Mar 6, 2023
-
rtptwcc: fixed wrap-around bug in the tx feedback code
This bug stops us from updating the "send" timestamp for sent packets and cause the log to be spammed with warnings for each packet in cases where the receiver stops sending us the twcc feedback report. (cherry picked from commit 2749324)
Configuration menu - View commit details
-
Copy full SHA for 69e35cc - Browse repository at this point
Copy the full SHA 69e35ccView commit details
Commits on Mar 17, 2023
-
rtpjitterbuffer: new packet spacing heuristic
- use RTP time instead of pts, the former is based on the sender clock and so provides a more precise metric. - instead of averaging spacings, update the packet spacing when we receive two consecutive identical spacings, discarding decreasing RTP times. This new heuristic should provide a more precise and constant packet spacing on audio streams. It will allow us to support discontinuous transmission (DTX) as the large spacings introduced by DTX will no longer increase the actual packet spacing. Co-Authored-By: Tulio Beloqui <[email protected]> (cherry picked from commit f68e9b9)
Configuration menu - View commit details
-
Copy full SHA for fd45849 - Browse repository at this point
Copy the full SHA fd45849View commit details
Commits on Mar 24, 2023
-
SCTP: do not assert on the decoders pad, as the association might be
disconnecting when on_receive() is called from usrsctplib (cherry picked from commit b50426e)
Configuration menu - View commit details
-
Copy full SHA for ac8fe87 - Browse repository at this point
Copy the full SHA ac8fe87View commit details -
SCTP: changes to handle usrsctp bugs
Workaround for sctplab/usrsctp#405: - Since the sctp socket can outlive the sctp assoacition, we need to only register/deregister the association when creating/closing the socket. This prevents invalid calls to sctp_packet_out() and receive_cb() in potential invalid states. Workaround for sctplab/usrsctp#383: - Retry usrsctp_finish() for 5seconds. This fixes a race condition between usrsctp_close() and usrsctp_finish() in which a dead socket will be accessed by the SCTP thread. (cherry picked from commit 9cf8ac2)
Configuration menu - View commit details
-
Copy full SHA for e655834 - Browse repository at this point
Copy the full SHA e655834View commit details -
Configuration menu - View commit details
-
Copy full SHA for 74f381c - Browse repository at this point
Copy the full SHA 74f381cView commit details
Commits on Mar 29, 2023
-
SCTP: do not hold the association mutex when calling GstSctpAssociati…
…onPacketReceivedCb Follow a consistent pattern when executing callbacks in handle_message().
Configuration menu - View commit details
-
Copy full SHA for a120437 - Browse repository at this point
Copy the full SHA a120437View commit details
Commits on Apr 3, 2023
-
Revert "SCTP: do not hold the association mutex when calling GstSctpA…
…ssociationPacketReceivedCb" This reverts commit a120437. We might not need this change after all.
Configuration menu - View commit details
-
Copy full SHA for ab2b82b - Browse repository at this point
Copy the full SHA ab2b82bView commit details
Commits on Apr 27, 2023
-
rtpjitterbuffer: test for non-utf8 cname
Co-authored-by: John Bassett <[email protected]> (cherry picked from commit 03e16b7)
Configuration menu - View commit details
-
Copy full SHA for 3f0b4fd - Browse repository at this point
Copy the full SHA 3f0b4fdView commit details
Commits on May 5, 2023
-
rtpvp8pay, rtpvp9pay: small refactor around PictureID
A set of unrelated but minor changes to VP8 and VP9 payloaders: VP8: - avoid resetting PictureID if NO_PICTURE_ID mode is set. VP9: - minor refactor of PictureID logic to bring the logic in line with the vp8pay. - add picture-id-offset property. - remove unused struct and variable on a test. VPX: - rename PictureID enum to be more inline with each other. - add picture-id property. - increment rather than resetting the PictureID on a FLUSH_START to maximize interop. Squashed commit of the following (upstream): commit 0cee3cd Author: Camilo Celis Guzman <[email protected]> Date: Thu May 4 02:41:09 2023 +0900 rtpvp8pay: rtpvp9pay: access picture_id property atomically Atomically set and get the picture_id. This changeset only atomically gets the picture-id when such property is queried on the element, on every other place where it is accessed internally it is accessed directly. This is because there is no MT scenario where we would be modifying this value and reading it internally in parallel. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4530> commit e4d8cda Author: Camilo Celis Guzman <[email protected]> Date: Wed May 3 20:53:41 2023 +0900 rtpvp8pay, rtpvp9pay: increment PictureID on FLUSH_START In recent versions of Chrome (M106) a change on their jitter buffer means that they are very susceptible to PictureID discontinuities. Then avoid at all cost resetting the PictureID. Moreover, according to the RFCs for VP8 and VP9 payloads; the PictureID can start off at any random value. So there is no logical problem of incrementing it here rather than resetting it, as long as it is a different PictureID. WebRTC's recent corruption issue: https://bugs.chromium.org/p/webrtc/issues/detail?id=15101 Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4530> commit f159fd8 Author: Camilo Celis Guzman <[email protected]> Date: Wed May 3 00:34:55 2023 +0900 rtpvp8pay, rtpvp9pay: expose picture-id as a property Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4530> commit 38d5899 Author: Camilo Celis Guzman <[email protected]> Date: Tue May 2 22:21:34 2023 +0900 rtpvp9pay: tests: remove unused struct and argument on test Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4530> commit 11187a8 Author: Camilo Celis Guzman <[email protected]> Date: Tue May 2 21:45:48 2023 +0900 rtpvp9pay: add picture-id-offset property Bring the VP9 payloader in sync in this regard to the VP8 payloader Allowing setting the picture id to a known value is useful when testing. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4530> commit 7cffb40 Author: Camilo Celis Guzman <[email protected]> Date: Tue Apr 25 15:36:45 2023 +0900 rtpvp9pay: minor refactor of PictureID logic This brings the logic inline with the vp8pay Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4530> commit a79616e Author: Camilo Celis Guzman <[email protected]> Date: Tue Apr 25 15:25:57 2023 +0900 rtpvp8pay: avoid reseting PictureID if NO_PICTURE_ID mode is set There is no logical change here, as `& (1 << nbits) - 1` would produce also 0 when NO_PICTURE_ID mode is choosen. However, this avoid computing a random integer that is actually unused. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4530> commit 7dd6375 Author: Camilo Celis Guzman <[email protected]> Date: Tue Apr 25 15:11:10 2023 +0900 rtpvp8pay, rtpvp9pay: use GType like name for PictureIDMode Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4530> (cherry picked from commit 9814555)
Configuration menu - View commit details
-
Copy full SHA for 8a46c90 - Browse repository at this point
Copy the full SHA 8a46c90View commit details
Commits on Jun 8, 2023
-
Configuration menu - View commit details
-
Copy full SHA for 1a32a60 - Browse repository at this point
Copy the full SHA 1a32a60View commit details -
rtpbasedepayload: set default buffer offset based on RTP timestamp
Set a default offset on buffers: the delta between this buffer's RTP timestamp and the first RTP timestamp. This can be used when forwarding (depayloading/repayloading) media for codecs without specific offset logic in their payloader subclasses (e.g. Opus). The default offset allows the symmetric payloader to reconstruct the original RTP timestamp deltas for its output RTP timestamps, rather than having to rely on the (potentially altered) buffer PTS. (cherry picked from commit a97f44f) (cherry picked from commit 52edb23)
Configuration menu - View commit details
-
Copy full SHA for 8f0a4ad - Browse repository at this point
Copy the full SHA 8f0a4adView commit details -
rtpbasedepayload: test: correct order of fields
The gist of the issue and fix are somewhat obscure and relate with how a buffer is mapped into before on different platforms. The test uses a va_args method to create, set and push an RTP buffer to the depayloader. However, due to the nature of the fields that can be set, and the use of the va_args. The ordering of this fields matter. This is because some of this fields are only set of a GstRTPBuffer and other are public GstBuffer properties. The problem lies when we set a GstRTPBuffer field, which requires a map of the buffer and **then** try to set a public property on the raw GstBuffer that has already been mapped! Fix this by: 1). correct the order of fields set on the buffer. 2). assert hard when trying to set a public property on the raw GstBuffer after it has been mapped. (cherry picked from commit 38571db)
Configuration menu - View commit details
-
Copy full SHA for d9885dd - Browse repository at this point
Copy the full SHA d9885ddView commit details
Commits on Jun 23, 2023
-
Configuration menu - View commit details
-
Copy full SHA for 63005d8 - Browse repository at this point
Copy the full SHA 63005d8View commit details -
vp8pay: introduce new property parse-frame
Controls weather or not we would want to parse the bitstream in order to extract partition information, which is currently only used if no custom meta is present.
Configuration menu - View commit details
-
Copy full SHA for 78a6af4 - Browse repository at this point
Copy the full SHA 78a6af4View commit details -
Configuration menu - View commit details
-
Copy full SHA for 93c956e - Browse repository at this point
Copy the full SHA 93c956eView commit details -
Configuration menu - View commit details
-
Copy full SHA for 17ab430 - Browse repository at this point
Copy the full SHA 17ab430View commit details
Commits on Jun 26, 2023
-
Configuration menu - View commit details
-
Copy full SHA for 92e2997 - Browse repository at this point
Copy the full SHA 92e2997View commit details -
check: compile libcompat when building statically
...or else we have to carry rt_lib as a dependency
Configuration menu - View commit details
-
Copy full SHA for 19cfd7c - Browse repository at this point
Copy the full SHA 19cfd7cView commit details -
Configuration menu - View commit details
-
Copy full SHA for cb66fcc - Browse repository at this point
Copy the full SHA cb66fccView commit details -
rtpsession: make an action-signal for requesting TWCC stats
This should be squashed with the previous window-work for twcc, to avoid those size/delay properties ever being introduced.
Configuration menu - View commit details
-
Copy full SHA for a1f4cda - Browse repository at this point
Copy the full SHA a1f4cdaView commit details -
rtptwcc: FIXUP for twcc-reorder patch
With the reordering-patch, we started keeping the last packet of the previous report around IF there was any packetloss reported in the previous report, and then start the new report with this packet. We found that this is exactly what WebRTC are doing, and we decided to mimic their behavior here. However, ever since the big "reordering-patch" went in, we have seen some strange discrepancies around what the netsim bitrate have been set to drop, and what has *actually* been reported as dropped. After a long investigation, we found that the "missing_run" variable in the RecvPacket struct could have contained a number from the previous run, but no this number was no longer valid for the next report, BUT WAS STILL BEING USED! This basically means that we were encoding a wildly different set of packets than we actually intended (and actually received), and hence the discrepancies in bitrates...
Configuration menu - View commit details
-
Copy full SHA for fcbdd40 - Browse repository at this point
Copy the full SHA fcbdd40View commit details -
pexhack: do not use macos-bison-binary subproject
A good enough version is available from macport. The version supplied by gstreamer is only for x86_64
Configuration menu - View commit details
-
Copy full SHA for 36c5a06 - Browse repository at this point
Copy the full SHA 36c5a06View commit details -
Configuration menu - View commit details
-
Copy full SHA for 7415949 - Browse repository at this point
Copy the full SHA 7415949View commit details -
glshader: unlock before notify, and use dispose instead of finalize
To avoid deadlock when shutting down.
Configuration menu - View commit details
-
Copy full SHA for e9ea58d - Browse repository at this point
Copy the full SHA e9ea58dView commit details -
Configuration menu - View commit details
-
Copy full SHA for a3b419c - Browse repository at this point
Copy the full SHA a3b419cView commit details -
amcvideoenc: default to -1 as I_FRAME_INTERVAL_DEFAULT
This will produce a single I-frame at the start and then after that only on request. This makes much more sense than "0", which means producing an I-frame for every single frame!
Configuration menu - View commit details
-
Copy full SHA for b50e077 - Browse repository at this point
Copy the full SHA b50e077View commit details -
Configuration menu - View commit details
-
Copy full SHA for 3092c8c - Browse repository at this point
Copy the full SHA 3092c8cView commit details -
Configuration menu - View commit details
-
Copy full SHA for dd72476 - Browse repository at this point
Copy the full SHA dd72476View commit details -
Configuration menu - View commit details
-
Copy full SHA for b9bfbc5 - Browse repository at this point
Copy the full SHA b9bfbc5View commit details -
The idea is basically that every n second, we delay each packet within a X ms window, so that they all burst out at the same time at the end of that window, recreating a throttling effect for a network, but without actually losing any packets.
Configuration menu - View commit details
-
Copy full SHA for a2caee0 - Browse repository at this point
Copy the full SHA a2caee0View commit details -
Configuration menu - View commit details
-
Copy full SHA for 877a349 - Browse repository at this point
Copy the full SHA 877a349View commit details -
Configuration menu - View commit details
-
Copy full SHA for a9271e8 - Browse repository at this point
Copy the full SHA a9271e8View commit details
Commits on Jul 4, 2023
-
rtpvp8pay: don't payload partitions when forced not to parse the frame
regression from 63005d (cherry picked from commit eb702a5)
Configuration menu - View commit details
-
Copy full SHA for 2aa47be - Browse repository at this point
Copy the full SHA 2aa47beView commit details
Commits on Jul 12, 2023
-
audiodecoder: do not account for upstream latency when producing DTX …
…packets - Removed the upstream latency as well, not needed anymore.
Configuration menu - View commit details
-
Copy full SHA for 86d1d53 - Browse repository at this point
Copy the full SHA 86d1d53View commit details -
Configuration menu - View commit details
-
Copy full SHA for 0c93862 - Browse repository at this point
Copy the full SHA 0c93862View commit details -
gstamcvideoenc: fixed minimum value for PROP_I_FRAME_INTERVAL_FLOAT
Fixes g_param_spec_float: assertion 'default_value >= minimum && default_value <= maximum' failed
Configuration menu - View commit details
-
Copy full SHA for 0f38a87 - Browse repository at this point
Copy the full SHA 0f38a87View commit details
Commits on Aug 29, 2023
-
applemedia/vtenc: reset the encoder session if we fail to encode a frame
- kVTInvalidSessionErr: common if the application goes between fg and bg mode. - kVTVideoEncoderMalfunctionErr: if the encoder has an internal error.
Configuration menu - View commit details
-
Copy full SHA for aa35370 - Browse repository at this point
Copy the full SHA aa35370View commit details -
Configuration menu - View commit details
-
Copy full SHA for 493bd28 - Browse repository at this point
Copy the full SHA 493bd28View commit details
Commits on Aug 30, 2023
-
Configuration menu - View commit details
-
Copy full SHA for a4b2b14 - Browse repository at this point
Copy the full SHA a4b2b14View commit details -
gl+vulkan/cocoa: rename _invoke_on_main to allow gstreamer-full to build
Having _invoke_on_main defined in separate plugins doesn't work if you want to built the statically.
Configuration menu - View commit details
-
Copy full SHA for 5505cfe - Browse repository at this point
Copy the full SHA 5505cfeView commit details
Commits on Aug 31, 2023
-
Configuration menu - View commit details
-
Copy full SHA for f14ef2f - Browse repository at this point
Copy the full SHA f14ef2fView commit details -
Configuration menu - View commit details
-
Copy full SHA for ef66754 - Browse repository at this point
Copy the full SHA ef66754View commit details
Commits on Sep 5, 2023
-
Configuration menu - View commit details
-
Copy full SHA for d9b6e73 - Browse repository at this point
Copy the full SHA d9b6e73View commit details
Commits on Sep 6, 2023
-
Configuration menu - View commit details
-
Copy full SHA for d6fcf4b - Browse repository at this point
Copy the full SHA d6fcf4bView commit details -
vulkan: gstvkwindow_cocoa use _gst_vulkan_cocoa_invoke_on_main rather
than dispatch_async on set_window_handle
Configuration menu - View commit details
-
Copy full SHA for 7728fae - Browse repository at this point
Copy the full SHA 7728faeView commit details
Commits on Sep 7, 2023
-
vulkan: fixed main thread checker warning
CreateMacOSSurface should only be called from the main thread.
Configuration menu - View commit details
-
Copy full SHA for 8cdd248 - Browse repository at this point
Copy the full SHA 8cdd248View commit details
Commits on Sep 11, 2023
-
gstaes: fixed missing symbols and rework on dependencies
In our dynamic build winsock2 symbols are missing.
Configuration menu - View commit details
-
Copy full SHA for d1db003 - Browse repository at this point
Copy the full SHA d1db003View commit details -
gstvkwindow: do not depend on close signal handlers to close the window
gstvkwindow_cocoa: call close on the internal window Fixes window leak.
Configuration menu - View commit details
-
Copy full SHA for 3ce91a7 - Browse repository at this point
Copy the full SHA 3ce91a7View commit details
Commits on Sep 12, 2023
-
- Moved internal gst_ks_video_device_parse_win32_error into a common function ks_parse_win32_error to reuse and add more logging in other places. - Removed unused function gst_ks_video_src_get_device_name_values. - Removed priority list, it doesn't make much sense for cameras. By doing so we also fixed a strange issue with CreateFile failing on a second attempt. Co-Authored-By: Knut Saastad <[email protected]>
Configuration menu - View commit details
-
Copy full SHA for 3b42686 - Browse repository at this point
Copy the full SHA 3b42686View commit details
Commits on Sep 13, 2023
-
SCTP: acquire lock in gst_sctp_association_incoming_packet()
To fix calling usrsctp_conninput() on a dead socket. user_socketvar.h: added KASSERT to make sure we do not try to reference a dead socket
Configuration menu - View commit details
-
Copy full SHA for de8447b - Browse repository at this point
Copy the full SHA de8447bView commit details
Commits on Sep 18, 2023
-
Configuration menu - View commit details
-
Copy full SHA for cfcc705 - Browse repository at this point
Copy the full SHA cfcc705View commit details
Commits on Sep 24, 2023
-
Configuration menu - View commit details
-
Copy full SHA for da7df33 - Browse repository at this point
Copy the full SHA da7df33View commit details -
Configuration menu - View commit details
-
Copy full SHA for abc12ab - Browse repository at this point
Copy the full SHA abc12abView commit details -
Configuration menu - View commit details
-
Copy full SHA for b98e07d - Browse repository at this point
Copy the full SHA b98e07dView commit details -
Configuration menu - View commit details
-
Copy full SHA for 6b945e4 - Browse repository at this point
Copy the full SHA 6b945e4View commit details -
Configuration menu - View commit details
-
Copy full SHA for a951ab7 - Browse repository at this point
Copy the full SHA a951ab7View commit details -
Configuration menu - View commit details
-
Copy full SHA for f8e8529 - Browse repository at this point
Copy the full SHA f8e8529View commit details -
Configuration menu - View commit details
-
Copy full SHA for c4c8fd2 - Browse repository at this point
Copy the full SHA c4c8fd2View commit details -
Configuration menu - View commit details
-
Copy full SHA for 98d1876 - Browse repository at this point
Copy the full SHA 98d1876View commit details -
Configuration menu - View commit details
-
Copy full SHA for e05f45d - Browse repository at this point
Copy the full SHA e05f45dView commit details -
Configuration menu - View commit details
-
Copy full SHA for daea53f - Browse repository at this point
Copy the full SHA daea53fView commit details
Commits on Sep 25, 2023
-
Configuration menu - View commit details
-
Copy full SHA for adf6651 - Browse repository at this point
Copy the full SHA adf6651View commit details
Commits on Sep 27, 2023
-
Configuration menu - View commit details
-
Copy full SHA for d204c51 - Browse repository at this point
Copy the full SHA d204c51View commit details -
Configuration menu - View commit details
-
Copy full SHA for 206f648 - Browse repository at this point
Copy the full SHA 206f648View commit details -
Configuration menu - View commit details
-
Copy full SHA for 1bd8e7a - Browse repository at this point
Copy the full SHA 1bd8e7aView commit details -
Configuration menu - View commit details
-
Copy full SHA for a00e360 - Browse repository at this point
Copy the full SHA a00e360View commit details -
Configuration menu - View commit details
-
Copy full SHA for b264084 - Browse repository at this point
Copy the full SHA b264084View commit details -
Configuration menu - View commit details
-
Copy full SHA for 9c97874 - Browse repository at this point
Copy the full SHA 9c97874View commit details -
Configuration menu - View commit details
-
Copy full SHA for 7d7e832 - Browse repository at this point
Copy the full SHA 7d7e832View commit details -
Configuration menu - View commit details
-
Copy full SHA for 9ab1ccd - Browse repository at this point
Copy the full SHA 9ab1ccdView commit details -
Configuration menu - View commit details
-
Copy full SHA for 743d551 - Browse repository at this point
Copy the full SHA 743d551View commit details -
Configuration menu - View commit details
-
Copy full SHA for 71e1e55 - Browse repository at this point
Copy the full SHA 71e1e55View commit details -
Configuration menu - View commit details
-
Copy full SHA for 85cff5c - Browse repository at this point
Copy the full SHA 85cff5cView commit details -
moved sctpsocket layer into dcsctp lib itself
So we don't have to involve C++ in gst land! Also, link dcsctp statically
Configuration menu - View commit details
-
Copy full SHA for bad7b7f - Browse repository at this point
Copy the full SHA bad7b7fView commit details
Commits on Sep 28, 2023
-
Configuration menu - View commit details
-
Copy full SHA for 0e567d3 - Browse repository at this point
Copy the full SHA 0e567d3View commit details -
Configuration menu - View commit details
-
Copy full SHA for 8d885a8 - Browse repository at this point
Copy the full SHA 8d885a8View commit details -
make the decoder copy data for now
to avoid mixing C++ and glib allocators
Configuration menu - View commit details
-
Copy full SHA for 5f574b3 - Browse repository at this point
Copy the full SHA 5f574b3View commit details -
Configuration menu - View commit details
-
Copy full SHA for 5b2de73 - Browse repository at this point
Copy the full SHA 5b2de73View commit details
Commits on Sep 29, 2023
-
implemented timeouts single main loop for every sctp association associations lifetime is now only managed by the factory
Configuration menu - View commit details
-
Copy full SHA for 9da94b0 - Browse repository at this point
Copy the full SHA 9da94b0View commit details -
Configuration menu - View commit details
-
Copy full SHA for b620721 - Browse repository at this point
Copy the full SHA b620721View commit details
Commits on Oct 2, 2023
-
Configuration menu - View commit details
-
Copy full SHA for f2aaab7 - Browse repository at this point
Copy the full SHA f2aaab7View commit details -
Configuration menu - View commit details
-
Copy full SHA for 4eaac80 - Browse repository at this point
Copy the full SHA 4eaac80View commit details -
- renamed gst_sctp_association_start to connect(), to keep consistency with disconnect - handling and logging SctpSocket_Error, closing the socket on error - configure the socket for aggressive-heartbeat, not quite working.. yet!
Configuration menu - View commit details
-
Copy full SHA for 4b4889e - Browse repository at this point
Copy the full SHA 4b4889eView commit details
Commits on Oct 3, 2023
-
Configuration menu - View commit details
-
Copy full SHA for 253fcd1 - Browse repository at this point
Copy the full SHA 253fcd1View commit details -
Configuration menu - View commit details
-
Copy full SHA for d29c712 - Browse repository at this point
Copy the full SHA d29c712View commit details -
Configuration menu - View commit details
-
Copy full SHA for a6f19cd - Browse repository at this point
Copy the full SHA a6f19cdView commit details
Commits on Oct 6, 2023
-
Configuration menu - View commit details
-
Copy full SHA for 05903a9 - Browse repository at this point
Copy the full SHA 05903a9View commit details -
Configuration menu - View commit details
-
Copy full SHA for ab08edb - Browse repository at this point
Copy the full SHA ab08edbView commit details -
Configuration menu - View commit details
-
Copy full SHA for 2b2460a - Browse repository at this point
Copy the full SHA 2b2460aView commit details -
Use meson built in support for dependency to threads lib
-lpthread directly just didn't work on my setup. Using the meson built-in dependency seems to be the way other libs do it.
Configuration menu - View commit details
-
Copy full SHA for 9c05dd5 - Browse repository at this point
Copy the full SHA 9c05dd5View commit details