* Handle cases of long mute/rollover of time stamp.
There are cases where the track is muted for long enough for timestamp
roll over to happen. There are no packets in that window (typically
there should be black frames (for video) or silence (for audio)). But,
maybe the pause based implementation of mute is causing this.
Anyhow, use time since last packet to gauge how much roll over should
have happened and use that to update time stamp. There will be really
edge cases where this could also fail (for e. g. packet time is affected
by propagation delay, so it could theoretically happen that mute/unmute
+ packet reception could happen exactly around that rollover point and
miscalculate, but should be rare).
As this happen per packet on receive side, changing time to `UnixNano()`
to make it more efficient to check this.
* spelling
* tests
* test util
* tests
* Add support for "abs-capture-time" extension.
Currently, it is just passed through from publisher -> subscriber side.
TODO: Need to store in sequencer and restore for retransmission.
* abs-capture-time in retransmissions
* clean up
* fix test
* more test fixes
* more test fixes
* more test fixes
* log only when size is non-zero
* log on both sides for debugging
* add marshal/unmarshal
* normalize abs capture time to SFU clock
* comment out adding abs-capture-time from registered extensions
* Reduce heap for dependency descriptor in forwarding path.
Marshaled dependency descriptor is held in sequencer adding heap objcts.
Store DD bytes in sequencer to avoid heap usage.
Also, accomodating over sized objects via storing in slice and using it
in case the bytes do not fit in the internal array.
NOTE: Marshal DD still does a make([]byte...), but I think it should be
on the stack given the short use of it. Have to verify.
* fix test and also add cases for oversized codec/dd bytes
* Split RTPStats into receiver and sender.
For receiver, short types are input and need to calculate extended type.
For sender (subscriber), it can operate only in extended type.
This makes the subscriber side a little simpler and should make it more
efficient as it can do simple comparisons in extended type space.
There was also an issue with subscriber using shorter type and
calculating extended type. When subscriber starts after the publisher
has already rolled over in sequence number OR timestamp, when
subsequent publisher side sender reports are used to adjust subscriber
time stamps, they were out of whack. Using extended type on subscriber
does not face that.
* fix test
* extended types from sequencer
* log
* Sequencer small optimisations
1. Use range map to exclude padding only packets. Should take lesser
space as we are not using slice to hold pointer to actual data.
2. Avoid `time.Now()` when adding each packet. Just use the arrival time
as it should be close enough. `time.Now()` was showing up in
profile.
* remove debug
* correct comment
With SVC codecs, input marker and fowarded marker could be different.
So, cache it in sequence and use it on retransmit.
@cndderrauber - this could have affected SVC under packet loss.
* Notify max layer taking into account overshoot.
An attempt to handle case of FF stopping layer 0, but not layer 1.
When max subscribed layer is layer 0, server allows overshoot to layer 1
to ensure continued streaming when the channel is not congested.
But, dynacast could have reported maximum subscribed layer as layer 0.
This is a very simple attempt to address that by taking overshoot
into account. Needs testing if this works well or not.
NOTE: When subsriber/down track is unmuted, it will report
the max subscribed layer as the max required layer. In those cases,
if the client does not start layer 0, there will still be an issue.
IOW, server is not keeping track of client behaviour that the client
has stopped layer 0 and publishing only layer 1. Server is just
accommodating overshoot with this change.
* Fix a couple of tests and change reflect.DeepEqual -> require.Equal as
much as possible
- ticker.Stop always
- clean up timer func (if they are added) on participant close
- sequencer test enhancement to add a real packet after a pdding packet
Padding packets do not need the full structure. They just
need a placeholder in the sequencer array. So, use pointers
(with padding slots filled by nil) to save some memory.
Also, don't need padding for audio (yet). As padding packets
are used only for probing and we do not probe using audio tracks (yet).
* Separate from ion-sfu
changes:
1. extract pkg/buffer, twcc, sfu, relay, stats, logger
2. to solve cycle import, move ion-sfu/pkg/logger to pkg/sfu/logger
3. replace pion/ion-sfu => ./
reason: will change import pion/ion-sfu/pkg/* to livekit-server/pkg/*
after this pr merged. Just not change any code in this pr, because it
will confused with the separate code from ion-sfu in review.
* Move code from ion-sfu to pkg/sfu
* fix build error for resovle conflict
Co-authored-by: cnderrauber <zengjie9004@gmail.com>
Raja Subramanian