Misha/handle wrapping base time #36
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
...it crashes our libvpx...
Basically, all frames are "i-frames" here.
Can't have a mix of GstValueArray and GValueArray here.
Changes lost in monorepo merge; originally from 7bbcd29
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.
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 <tulio@pexip.com> Co-authored-by: Stian Selnes <stian@pexip.com>
Squashed commit of the following: commit c2e3ed1 Author: Camilo Celis Guzman <camilo@pexip.com> Date: Tue Nov 2 22:30:12 2021 +0900 rtphdrextroi: correct max. number of roi-types commit ac3c6c4 Author: Camilo Celis Guzman <camilo@pexip.com> Date: Fri Oct 29 23:00:14 2021 +0900 rtphdrextroi: add test for multiple roi-types commit 1fd0bc8 Author: Camilo Celis Guzman <camilo@pexip.com> Date: Thu Oct 28 19:42:04 2021 +0900 rtphdrextroi: WIP: multiple roi-types support commit 4b9f2d8 Author: Camilo Celis Guzman <camilo@pexip.com> 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 <camilo@pexip.com> Date: Thu Oct 28 17:03:55 2021 +0900 rtphdrextroi: make roi-type construct only
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) ```
doing a sync point request all_headers set to FALSE stops the FIR request in gstrtpsession.
disabled If the checks are not built, it will leave behind some "unused" variables, so this warnings will always trigger.
This allows user-level signal handler to catch this Co-authored-by: Frederik Vestre <frederik.vestre@pexip.com>
Enables device enumeration and monitor using the ACameraManager API
execute regardless of the provider being stopped.
Calling destroy_session() without first calling finish() deadlocks almost every time, due to the explanation given in gst_vtenc_finish_encoding()
MishaBaranov
force-pushed
the
misha/handle_wrapping_base_time
branch
from
2fef5c2 to
145f6a7
Compare
In the test, adding and removing 50.000 pads, on my machine it goes from about 14 seconds to about 0,05 seconds.
…lement_dispose` `gst_element_dispose` could be called more then one time. So we have to handle it accordingly. Fix pexip/media#1062
Avoid leaking pad_name in error paths
MishaBaranov
force-pushed
the
misha/handle_wrapping_base_time
branch
from
133d320 to
71bbe6c
Compare
Temporal fix for the flickering on resizing
…3de656d16fd79b791414813ef9c8e Squashed commit of the following: commit f01baab Author: Will Miller <will.miller@pexip.com> Date: Mon Jun 30 21:19:32 2025 +0100 absl: remove safe int parse functions These choke on Windows, and we don't need them for dcsctp commit f74466f Author: Will Miller <will.miller@pexip.com> Date: Mon Jun 30 17:59:12 2025 +0100 dcsctp: fixes for darwin build commit 7bb17a8 Author: Will Miller <will.miller@pexip.com> Date: Mon Jun 30 14:53:42 2025 +0100 dcsctp: reapply non-void control fixups commit 1fb040f Author: Will Miller <will.miller@pexip.com> Date: Mon Jun 30 13:10:40 2025 +0100 dcsctp: updating to https://webrtc.googlesource.com/src/+/4f422cb71cf3de656d16fd79b791414813ef9c8e
There was a segfault when processing fec block with missing info for a redundnant packet
Allow directly sending a User-Initiated abort with a given string from the application level. This is useful to simulate scenarios of receiving such a chunk off the wire in a test, even when an implementation such as dcsctp might not natively ever provoke such a scenario.
We can't assume the association's socket will always exist, as it gets torn down when we experience a force-close (due to a timeout, for example), or a client-received abort chunk. Remove assertions as to the state of the socket where we can't guarantee it, and replace them with conditionals. Additionally, ensure we only ever access the socket member from async handlers to give thread safety.
These checks are no longer required because the underlying dcsctp library performs send queueing and doesn't need to be protected from calling the methods early.
This reverts commit a430663.
This reverts commit 34f32c7.
…ool" This reverts commit ca7b7c7.
…ts fifo Fixes #1252
MishaBaranov
force-pushed
the
misha/handle_wrapping_base_time
branch
from
71bbe6c to
496f430
Compare
This is now a re-entrant mutex; so use to corresponding API This was introduced in https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/2740 to avoid deadlocks on repeated initializations. Fixup of d2742d3
…end by handling empty payloads explicitly
- Replace vector range constructor with explicit assign and empty check
- Avoid constructing std::vector from empty or invalid ranges
- Prevents compiler error and undefined behavior due to invalid pointer deletion
Reported by gcc as:
```
In member function ‘void std::__new_allocator<_Tp>::deallocate(_Tp*, size_type) [with _Tp = unsigned char]’,
inlined from ‘static void std::allocator_traits<std::allocator<_CharT> >::deallocate(allocator_type&, pointer, size_type) [with _Tp = unsigned char]’ at /opt/gcc-future/include/c++/14.3.0/bits/alloc_traits.h:550:23,
inlined from ‘std::vector<_Tp, _Alloc>::_M_realloc_append(_Args&& ...)::_Guard::~_Guard() [with _Args = {unsigned char}; _Tp = unsigned char; _Alloc = std::allocator<unsigned char>]’ at /opt/gcc-future/include/c++/14.3.0/bits/vector.tcc:616:18,
inlined from ‘void std::vector<_Tp, _Alloc>::_M_realloc_append(_Args&& ...) [with _Args = {unsigned char}; _Tp = unsigned char; _Alloc = std::allocator<unsigned char>]’ at /opt/gcc-future/include/c++/14.3.0/bits/vector.tcc:688:7,
inlined from ‘std::vector<_Tp, _Alloc>::reference std::vector<_Tp, _Alloc>::emplace_back(_Args&& ...) [with _Args = {unsigned char}; _Tp = unsigned char; _Alloc = std::allocator<unsigned char>]’ at /opt/gcc-future/include/c++/14.3.0/bits/vector.tcc:123:21,
inlined from ‘void std::vector<_Tp, _Alloc>::push_back(value_type&&) [with _Tp = unsigned char; _Alloc = std::allocator<unsigned char>]’ at /opt/gcc-future/include/c++/14.3.0/bits/stl_vector.h:1314:21,
inlined from ‘SctpSocket_SendStatus sctp_socket_send(SctpSocket*, const uint8_t*, size_t, uint16_t, uint32_t, bool, int32_t*, size_t*)’ at /opt/gstreamer/subprojects/gst-plugins-bad/ext/sctp/dcsctp/sctpsocket.cc:400:31:
/opt/gcc-future/include/c++/14.3.0/bits/new_allocator.h:172:33: error: ‘void operator delete(void*, std::size_t)’ called on pointer ‘<unknown>’ with nonzero offset [1, 9223372036854775807] [-Werror=free-nonheap-object]
172 | _GLIBCXX_OPERATOR_DELETE(_GLIBCXX_SIZED_DEALLOC(__p, __n));
| ^
In member function ‘_Tp* std::__new_allocator<_Tp>::allocate(size_type, const void*) [with _Tp = unsigned char]’,
inlined from ‘static _Tp* std::allocator_traits<std::allocator<_CharT> >::allocate(allocator_type&, size_type) [with _Tp = unsigned char]’ at /opt/gcc-future/include/c++/14.3.0/bits/alloc_traits.h:515:28,
inlined from ‘std::_Vector_base<_Tp, _Alloc>::pointer std::_Vector_base<_Tp, _Alloc>::_M_allocate(std::size_t) [with _Tp = unsigned char; _Alloc = std::allocator<unsigned char>]’ at /opt/gcc-future/include/c++/14.3.0/bits/stl_vector.h:383:33,
inlined from ‘void std::vector<_Tp, _Alloc>::_M_range_initialize_n(_Iterator, _Iterator, size_type) [with _Iterator = const unsigned char*; _Tp = unsigned char; _Alloc = std::allocator<unsigned char>]’ at /opt/gcc-future/include/c++/14.3.0/bits/stl_vector.h:1716:23,
inlined from ‘void std::vector<_Tp, _Alloc>::_M_range_initialize(_ForwardIterator, _ForwardIterator, std::forward_iterator_tag) [with _ForwardIterator = const unsigned char*; _Tp = unsigned char; _Alloc = std::allocator<unsigned char>]’ at /opt/gcc-future/include/c++/14.3.0/bits/stl_vector.h:1705:25,
inlined from ‘std::vector<_Tp, _Alloc>::vector(_InputIterator, _InputIterator, const allocator_type&) [with _InputIterator = const unsigned char*; <template-parameter-2-2> = void; _Tp = unsigned char; _Alloc = std::allocator<unsigned char>]’ at /opt/gcc-future/include/c++/14.3.0/bits/stl_vector.h:724:23,
inlined from ‘SctpSocket_SendStatus sctp_socket_send(SctpSocket*, const uint8_t*, size_t, uint16_t, uint32_t, bool, int32_t*, size_t*)’ at /opt/gstreamer/subprojects/gst-plugins-bad/ext/sctp/dcsctp/sctpsocket.cc:393:57:
/opt/gcc-future/include/c++/14.3.0/bits/new_allocator.h:151:55: note: returned from ‘void* operator new(std::size_t)’
151 | return static_cast<_Tp*>(_GLIBCXX_OPERATOR_NEW(__n * sizeof(_Tp)));
| ^
cc1plus: all warnings being treated as errors
```
Previously the wrapping of the 24-bit reference time was not handled correctly when transforming it into GstClockTime. Given the unit of 64ms the span that could be represented by 24 bits is 12 days and depending on the start value we could get a wrapping problem anytime within this time frame. This turned out to be particularly problematic for the GCC algorithm in gst-plugins-rs which tried to evict old packages based on the "oldest" timestamp, which due to wrapping problems could be in the future. Thus, the container managing the packets could grow without limits for a long time thereby creating both CPU and memory problems. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/7527>
With https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/7527 an attempt was made to fix the wrapping of the TWCC 24-bit reference time. Unfortunately this fix was not correct due to a late change and the lack of unit tests to catch it. This time unit tests are provided to make sure we have a correct fix for all the edge cases.
camilo-celis
force-pushed
the
misha/handle_wrapping_base_time
branch
from
496f430 to
5bfc76c
Compare
Contributor
|
(rebased to bring in gcc-future compilation fixes) |
havardgraff
force-pushed
the
main
branch
2 times, most recently
from
c56bbe7 to
f1a3cf8
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Cherry pick of:
2.. twcc: Fix reference timestamp wrapping (again)]
(c) Johan Sternerup