Since the code measuring the RTT has been converted to using TimeDelta
which internally stores the duration in microseconds, from DurationMs
which uses milliseconds, the RTO calculation can use the higher
precision to calculate lower non-zero durations on really fast networks
such within a data center.
Before this CL, which is from the initial drop of dcSCTP, the RTO
calculation was done using the algorithm from the paper "V. Jacobson:
Congestion avoidance and control", but now we're using the original
algorith from https://tools.ietf.org/html/rfc4960#section-6.3.1, which
comes from https://datatracker.ietf.org/doc/html/rfc6298#section-2.
Two issues were found and corrected:
1. The min RTT variance that is specified in the config file was
previously incorrectly divided by 8. That was not its intention,
but we're keeping that behaviour as other clients have actually
measured a good value to put there. This represents "G" in
the "basic algorithm" above, and since that is multiplied with K,
which is four, the default value of 220 wouldn't make sense if it
wasn't scaled down, as that would make the RTO easily saturate to
the RTO_max (800ms).
2. The previous algorithm had large round-off errors (probably because
the code used milliseconds), which makes fairly big changes to the
calculated RTO in some situations.
Bug: webrtc:15593
Change-Id: I95a3e137c2bbbe7bf8b99c016381e9e63fd01d87
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/349000
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#42170}
Keeping the old setting for the total queue size
limit, which avoids breaking a downstream.
This reverts commit 47ce449afa
and relands commit 4c990e2e56.
Bug: chromium:40072842
Change-Id: I1e7d14b5d0026232d1fc9277172b6947b8be3490
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/343120
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#41907}
This reverts commit 4c990e2e56.
Reason for revert: Breaks downstream build.
Original change's description:
> dcsctp: Add per-stream-limit, refactor limits.
>
> The limits have been moved out from the Send Queue as they were enforced
> outside the queue anyway (in the socket). That was a preparation for
> adding even more limits; There is now also a per-stream limit, allowing
> individual streams to have one (global) limit, and the entire socket to
> have another limit.
>
> These limits are very small in the default options. In Chrome, the limit
> is 16MB per stream, so expect the defaults to be updated when the
> additional buffering outside dcSCTP is removed.
>
> Bug: chromium:41221056
> Change-Id: I9f835be05d349cbfce3e9235d34b5ea0e2fe87d1
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/342481
> Reviewed-by: Florent Castelli <orphis@webrtc.org>
> Commit-Queue: Victor Boivie <boivie@webrtc.org>
> Cr-Commit-Position: refs/heads/main@{#41895}
Bug: chromium:41221056
Change-Id: Icd57fbfca87d6b512cfc7f7682ae709000c2bcad
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/343080
Commit-Queue: Björn Terelius <terelius@webrtc.org>
Reviewed-by: Victor Boivie <boivie@webrtc.org>
Bot-Commit: rubber-stamper@appspot.gserviceaccount.com <rubber-stamper@appspot.gserviceaccount.com>
Cr-Commit-Position: refs/heads/main@{#41901}
The limits have been moved out from the Send Queue as they were enforced
outside the queue anyway (in the socket). That was a preparation for
adding even more limits; There is now also a per-stream limit, allowing
individual streams to have one (global) limit, and the entire socket to
have another limit.
These limits are very small in the default options. In Chrome, the limit
is 16MB per stream, so expect the defaults to be updated when the
additional buffering outside dcSCTP is removed.
Bug: chromium:41221056
Change-Id: I9f835be05d349cbfce3e9235d34b5ea0e2fe87d1
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/342481
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#41895}
https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-zero-checksum/06/
The previous implementation was for version 00, and since then changes
have been made. The chunk that is used to negotiate this capability has
now grown to include an additional property - the sender's alternate
error detection method.
Bug: webrtc:14997
Change-Id: I78043d187b79f40bbadbcba02eae6eedf54f30f9
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/336380
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#41759}
This reverts commit 45eae34693.
It was found not to be the root cause of the performance
regression, so it's safe to reland.
Bug: webrtc:14997
Change-Id: I67c90752875bf4071cbdd5adfa462a37f4d4ceab
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/302162
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Bot-Commit: rubber-stamper@appspot.gserviceaccount.com <rubber-stamper@appspot.gserviceaccount.com>
Cr-Commit-Position: refs/heads/main@{#39910}
This reverts commit bd46bb7660.
Reason for revert: There is a slight performance degradation
pointing to this CL, so revert this to be able to confirm if
it is the culprit.
Original change's description:
> dcsctp: Support zero checksum packets
>
> If configured, the packet parser will allow packets with
> a set checksum of zero. In that case, the correct checksum
> will not even be calculated, avoiding a CPU intensive
> calculation.
>
> Also, if specified when building a packet, the checksum can
> be opted to be not calculated and written to the packet.
> This is to be used when draft-tuexen-tsvwg-sctp-zero-checksum
> has been negotiated, except for some packets during association
> establishment.
>
> This is mainly a preparation CL and follow-up CL will enable
> this feature.
>
> Low-Coverage-Reason: Affects debug logging code not run in tests
> Bug: webrtc:14997
> Change-Id: I3207ac3a626df039ee2990403c2edd6429f17617
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/298481
> Reviewed-by: Harald Alvestrand <hta@webrtc.org>
> Commit-Queue: Victor Boivie <boivie@webrtc.org>
> Cr-Commit-Position: refs/heads/main@{#39737}
Bug: webrtc:14997
Change-Id: Ie22267abb4bcd25d5af07875eb933c51ed5be853
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/301580
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#39878}
If configured, the packet parser will allow packets with
a set checksum of zero. In that case, the correct checksum
will not even be calculated, avoiding a CPU intensive
calculation.
Also, if specified when building a packet, the checksum can
be opted to be not calculated and written to the packet.
This is to be used when draft-tuexen-tsvwg-sctp-zero-checksum
has been negotiated, except for some packets during association
establishment.
This is mainly a preparation CL and follow-up CL will enable
this feature.
Low-Coverage-Reason: Affects debug logging code not run in tests
Bug: webrtc:14997
Change-Id: I3207ac3a626df039ee2990403c2edd6429f17617
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/298481
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#39737}
This is a reland of commit 17a02a31d7.
This is the first part of supporting stream priorities, and adds the API
and very basic support for setting and retrieving the stream priority.
This commit doesn't in any way change the actual packet sending - the
specified priority values are stored, but not acted on.
This is all that is client visible, so clients can start using the API
as written, and they would never notice that things are missing.
Bug: webrtc:5696
Change-Id: I04d64a63cbaec67568496ad99667e14eba85f2e0
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/264424
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#37081}
This reverts commit 17a02a31d7.
Reason for revert: Breaks downstream test
Original change's description:
> dcsctp: Add public API for setting priorities
>
> This is the first part of supporting stream priorities, and adds the API
> and very basic support for setting and retrieving the stream priority.
>
> This commit doesn't in any way change the actual packet sending - the
> specified priority values are stored, but not acted on.
>
> This is all that is client visible, so clients can start using the API
> as written, and they would never notice that things are missing.
>
> Bug: webrtc:5696
> Change-Id: I24fce8cbb6f3cba187df99d1d3f45e73621c93c6
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/261943
> Reviewed-by: Harald Alvestrand <hta@webrtc.org>
> Commit-Queue: Victor Boivie <boivie@webrtc.org>
> Cr-Commit-Position: refs/heads/main@{#37034}
Bug: webrtc:5696
Change-Id: If172d9c9dbce7aae72152abbbae1ccc77340bbc1
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/264444
Owners-Override: Björn Terelius <terelius@webrtc.org>
Bot-Commit: rubber-stamper@appspot.gserviceaccount.com <rubber-stamper@appspot.gserviceaccount.com>
Commit-Queue: Björn Terelius <terelius@webrtc.org>
Auto-Submit: Björn Terelius <terelius@webrtc.org>
Reviewed-by: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#37039}
This is the first part of supporting stream priorities, and adds the API
and very basic support for setting and retrieving the stream priority.
This commit doesn't in any way change the actual packet sending - the
specified priority values are stored, but not acted on.
This is all that is client visible, so clients can start using the API
as written, and they would never notice that things are missing.
Bug: webrtc:5696
Change-Id: I24fce8cbb6f3cba187df99d1d3f45e73621c93c6
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/261943
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#37034}
It wasn't correct and not enabled by default, so just remove it.
Bug: webrtc:12943
Change-Id: Idd426abd0da4ae259e519dd01239b4303296756a
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/232609
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35075}
This is mentioned in
https://datatracker.ietf.org/doc/html/rfc4960#section-6.3.1 and further
described in https://datatracker.ietf.org/doc/html/rfc6298#section-4.
The TCP RFCs mentioned G as the clock granularity, but in SCTP it should
be set much higher, to account for the delayed ack timeout (ATO) of the
peer (as that can be seen as a very high clock granularity). That one is
set to 200ms by default in many clients, so a reasonable default limit
could be set to 220 ms.
If the measured variance is higher, it will be used instead.
Bug: webrtc:12943
Change-Id: Ifc217daa390850520da8b3beb0ef214181ff8c4e
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/232614
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#35068}
The previous limits were taken from Oracles SCTP stack[1] as they were
more up-to-date than the suggested ones in RFC4960. However, after
having evaluated them for a while, it's evident that they are a bit too
aggressive and likely have their origin from a wired LAN network.
Let's do a re-take. These values have been taken from Solaris TCP
stack[2]. They are even less aggressive than Linux defaults. This can be
iterated even more, and is always possible to override by the client.
It's generally the increase of rto_min that is helping here, as the
delayed SACK and RTT jitter require that the RTO.min is quite much
higher than the delayed SACK timeout of the peer (which isn't in control
by us, but one can assume it's 200ms or less). And with a too low
RTO.min, it's increased risk of getting spurious retransmissions and
decreasing the congestion window.
[1] https://docs.oracle.com/cd/E93309_01/docs.466/SIGTRAN/GUID-2136614F-4BED-407C-87B0-7EE10E0FF534.htm
[2] https://docs.oracle.com/cd/E19120-01/open.solaris/819-2724/chapter4-69/index.html
Bug: webrtc:12943
Change-Id: I9678ac4396286a55c251c5f57589379da70fd27d
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/231139
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#34927}
By allowing the max timer backoff duration to be limited, a socket can
fast recover in case of intermittent network issues. Before this CL, the
exponential backoff algorithm could result in very long retry durations
(in the order of minutes), when connection has been lost or been flaky
for a long while.
Note that limiting the maximum backoff duration might require
compensating the maximum retransmission limit to avoid closing the
socket prematurely due to reaching the maximum retransmission limit much
faster than previously.
Bug: webrtc:13129
Change-Id: Ib94030d666433e3fa1a2c8ef69750a1afab8ef94
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/230702
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#34913}
The restart limit for timers can already be limitless, but the
RetransmissionErrorCounter didn't support this. With this change, the
max_retransmissions and max_init_retransmits can be absl::nullopt to
indicate that there should be infinite retries.
Bug: webrtc:13129
Change-Id: Ia6e91cccbc2e1bb77b3fdd7f37436290adc2f483
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/230701
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#34882}
The congestion window is unlikely to be even divisible by the size
of a packet, so when the congestion window is almost full, there is
often just a few bytes remaining in it. Before this change, a small
packet was created to fill the remaining bytes in the congestion window,
to make it really full.
Small packets don't add much. The cost of sending a small packet is
often the same as sending a large one, and you usually get lower
throughput sending many small packets compared to few larger ones.'
This mode will only be enabled when the congestion window is large, so
if the congestion window is small - e.g. due to poor network conditions,
it will allow packets to become fragmented into small parts, in order to
fully utilize the congestion window.
Bug: webrtc:12943
Change-Id: I8522459174bc72df569edd57f5cc4a494a4b93a8
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/228526
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34778}
Some deployments, e.g. Chromium, has a limited send buffer. It's
reasonable that it's quite small, as it avoids queuing too much, which
typically results in increased latency for real-time communication. To
avoid SCTP to fill up the entire buffer at once - especially when doing
fast retransmissions - limit the amount of packets that are sent in one
go.
In a typical scenario, SCTP will not send more than three packets for
each incoming packet, which is is the case when a SACK is received which
has acknowledged two large packets, and which also adds the MTU to the
congestion window (due to in slow-start mode), which then may result in
sending three packets. So setting this value to four makes any
retransmission not use that much more of the send buffer.
This is analogous to usrsctp_sysctl_set_sctp_fr_max_burst_default in
usrsctp, which also has the default value of four (4).
Bug: webrtc:12943
Change-Id: Iff76a1668beadc8776fab10312ef9ee26f24e442
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/228480
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34744}
This is useful in tests and in scenarios where the connection is
monitored externally and the heartbeat monitoring would be of no use.
Bug: webrtc:12614
Change-Id: Ida4f4e2e40fc4d2aa0c27ae9431f434da4cc8313
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/220766
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34164}
The minimum RTO time shouldn't be lower than the delayed ack timeout
of the peer to avoid sending retransmissions before the peer has
actually intended to reply.
In usrsctp, the default delayed ack timeout is 200ms and configurable
using the `sctp_delayed_sack_time_default` option. In dcsctp, it's
min(RTO/2, 200ms), to avoid this issue.
Bug: webrtc:12614
Change-Id: Ie84c331334af660d66b1a7d90d20f5cf7e2a5103
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/219100
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34026}
Due to a limit socket send buffer, it's quite easy to fill it up when
using exponential slow start, which results in dropping a lot of packets
and having to retransmit those.
Disabling this, to align it to how SCTP normally behaves, and then try
to stabilize it later. With SCTP slow start, it will increase with one
MTU for each RTT when there is no packet loss. Even this mode will
experience packet loss, but not as much will be lost, and it will
stabilize quicker.
Bug: webrtc:12614
Change-Id: Ibc484b19b7e708fe5bd837bbef178a2f69b7211f
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/218203
Reviewed-by: Florent Castelli <orphis@webrtc.org>
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#33969}
This completes the basic implementation of the dcSCTP library. There
are a few remaining commits to e.g. add compatibility tests and
benchmarks, as well as more support for e.g. RFC8260, but those are not
strictly vital for evaluation of the library.
The Socket contains the connection establishment and teardown sequences
as well as the general chunk dispatcher.
Bug: webrtc:12614
Change-Id: I313b6c8f4accc144e3bb88ddba22269ebb8eb3cd
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/214342
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Tommi <tommi@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#33890}
The socket can measure the round-trip-time (RTT) by two different
scenarios:
* When a sent data is ACKed
* When a HEARTBEAT has been sent, which as been ACKed.
The RTT will be used to calculate which timeout value that should be
used for e.g. the retransmission timer (T3-RTX). On connections with a
low RTT, the RTO value will be low, and on a connection with high RTT,
the RTO value will be high. And on a connection with a generally low
RTT value, but where it varies a lot, the RTO value will be calculated
to be fairly high, to not fire unnecessarily. So jitter is bad, and is
part of the calculation.
Bug: webrtc:12614
Change-Id: I64905ad566d5032d0428cd84143a9397355bbe9f
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/214045
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#33832}
Clients will use this API for all their interactions with this library.
It's made into an interface (of which there will only be a single
implementation) simply for documentation purposes. But that also allows
clients to mock the library if wanted or to have a thread-safe wrapper,
as the library itself is not thread-safe, while keeping the same
interface.
Bug: webrtc:12614
Change-Id: I346af9916e26487da040c01825c2547aa7a5d0b7
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/213348
Commit-Queue: Victor Boivie <boivie@webrtc.org>
Reviewed-by: Tommi <tommi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#33648}
