The race occurs if the transport is being destroyed at the same time as
a callback occurs on the usrsctp timer thread (for example, for a
retransmission). Fixed by slightly extending the scope of mutex
acquisition to include posting a task to the network thread, where it's
safe to do further work.
Bug: chromium:1162424
Change-Id: Ia25c96fa51cd4ba2d8690ba03de8af9e9f1605ea
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/202560
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Taylor <deadbeef@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#33048}
When deinitializing usrsctp, we attempt to call usrsctp_finish in a loop
for three seconds (it may fail because another sctp thread is holding a
reference to something). If the three seconds run out, usrsctp is left
in a still initialized state, and bad things happen down the road if
usrsctp_init is called in the state.
Bug: chromium:1138878
Change-Id: I9c24d51d5a274b06bdf4183261694fc2989136c5
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/189940
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Taylor <deadbeef@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#32467}
When processing an INIT for existing SCTP association, which may happen
when transferring to a new remote endpoint, if there are chunks still on
the output queue, an SCTP_SEND_FAILED_EVENT notification will be
delivered while holding the INP lock. When processing this notification,
WebRTC was calling usrsctp_getladdrs in order to look up the
SctpTransport* associated with the socket, but this also tries to
acquire the INP lock, resulting in a deadlock.
The temporary fix is to simply not subscribe to the
SCTP_SEND_FAILED_EVENT notification; it was only used for logging
anyway.
Bug: chromium:1137936
Change-Id: I077efb4c769d6f7855d2fee2d266f19396b16f9d
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/188502
Commit-Queue: Taylor <deadbeef@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#32427}
OnSctpInboundPacket is called not only for incoming packets, but for
notifications, which can be delivered on the usrsctp timer thread.
I suspect that these notifications can be delivered after we attempt to
close the socket, because if we attempt to close it while the timer
thread holds a reference, it isn't actually destroyed until the timer
thread finishes its operation.
Bug: chromium:1127774
Change-Id: Id6a883b14796e8f5bf1c2990f3d9d389d72c8a46
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/184240
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Taylor <deadbeef@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#32204}
The signature of send_cb was changed, adding ulp_info. This change makes
it easier to retrieve the SctpTransport pointer from the callback.
Bug: webrtc:11899
Change-Id: I12a4ccd2d0deb329f6be17a4c7208449833dc188
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/182984
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Taylor <deadbeef@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#32067}
This is to allow testing without using the singleton sctp library.
cricket::SctpTransportInternalFactory is renamed to webrtc::SctpTransportFactoryInterface and moved to the API folder to follow the API structure.
Tests can use test/pc/sctp/fake_sctp_transport.h to inject a faked data channel implementation.
patch 1 contain the original cl.
patch 2 modifications
Bug: none
Change-Id: Ic088da3eb7d9aada79e6d601dbf2d1aa2be777f6
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/182840
Reviewed-by: Taylor <deadbeef@webrtc.org>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Erik Språng <sprang@webrtc.org>
Commit-Queue: Per Kjellander <perkj@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#32024}
This reverts commit 4c0a381137c04fd80830af8a041e25e3428dd33f.
Reason for revert: Breaks downstream test
Original change's description:
> Make cricket::SctpTransportInternalFactory injectable through PeerConnectionFactory Deps
>
> This is to allow testing without using the singleton sctp library.
> cricket::SctpTransportInternalFactory is renamed to webrtc::SctpTransportFactoryInterface and moved to the API folder to follow the API structure.
> Tests can use test/pc/sctp/fake_sctp_transport.h to inject a faked data channel implementation.
>
> Bug: none
> Change-Id: I482241269463595062548870750d33f31238c6b1
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/182082
> Commit-Queue: Per Kjellander <perkj@webrtc.org>
> Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
> Reviewed-by: Taylor <deadbeef@webrtc.org>
> Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#32007}
TBR=deadbeef@webrtc.org,mbonadei@webrtc.org,kwiberg@webrtc.org,perkj@webrtc.org
Change-Id: I46d5ba89fe723caccd065b0ac41d77ed45373838
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: none
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/182802
Reviewed-by: Björn Terelius <terelius@webrtc.org>
Commit-Queue: Björn Terelius <terelius@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#32008}
This is to allow testing without using the singleton sctp library.
cricket::SctpTransportInternalFactory is renamed to webrtc::SctpTransportFactoryInterface and moved to the API folder to follow the API structure.
Tests can use test/pc/sctp/fake_sctp_transport.h to inject a faked data channel implementation.
Bug: none
Change-Id: I482241269463595062548870750d33f31238c6b1
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/182082
Commit-Queue: Per Kjellander <perkj@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Taylor <deadbeef@webrtc.org>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#32007}
A usrsctp regression is causing this callback to not be invoked, but
reducing the threshold (from 128KB to 64KB) seems to mitigate the issue.
Can set it back once the root cause is fixed, though this isn't
expected to have any performance implications.
Bug: webrtc:11824
Change-Id: I2f6a3183d298abf4d1ad3bbd3697b1879eb4d696
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/180841
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Taylor <deadbeef@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#31851}
Patch originally submitted by Lennart Grahl:
https://webrtc-review.googlesource.com/c/src/+/177527
SCTP message chunks and notifications are being delivered interleaved.
However, the way the code was structured previously, a notification
would interrupt reassembly of a message chunk and hand out the partial
message, thereby violating message integrity. This patch separates the
handling of notifications and reassembly of messages.
Additional changes:
- Remove illegal cast from non-validated u32 to enum (PPID)
- Drop partial messages if the SID has been changed but EOR not yet
received instead of delivering them. (This should never happen
anyway.)
- Don't treat TSN as timestamp (wat)
- Replace "usrsctplib/usrsctp.h" with <usrsctp.h>, allowing a hack
to be removed from media/BUILD.gn
Bug: webrtc:11708
Change-Id: I29733b03f67a3d840104b8608a7f0083466c2d0f
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/178469
Commit-Queue: Taylor <deadbeef@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#31641}
This reverts commit 6cbff752f52bf3f70168d551c33ce719bd8e0663.
Reason for revert: breaking downstream projects, Win MSVC x86 dbg and Win x86 Clang rel
Original change's description:
> Fix data channel message integrity violation
>
> SCTP message chunks and notifications are being delivered interleaved.
> However, the way the code was structured previously, a notification
> would interrupt reassembly of a message chunk and hand out the partial
> message, thereby violating message integrity. This patch separates the
> handling of notifications and reassembly of messages.
>
> Additional changes:
>
> - Remove illegal cast from non-validated u32 to enum (PPID)
> - Drop partial messages if the SID has been changed but EOR not yet
> received instead of delivering them. (This should never happen
> anyway.)
> - Don't treat TSN as timestamp (wat)
>
> Bug: webrtc:11708
> Change-Id: I4e2fe2262feda2a96d2ae3f6ce9b06370d9878ae
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/177527
> Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
> Reviewed-by: Taylor <deadbeef@webrtc.org>
> Reviewed-by: Harald Alvestrand <hta@webrtc.org>
> Commit-Queue: Tommi <tommi@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#31605}
TBR=deadbeef@webrtc.org,kwiberg@webrtc.org,tommi@webrtc.org,hta@webrtc.org,lennart.grahl@gmail.com
Change-Id: I6d6c5a11835f155f8c449b996d034f43b8db452c
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: webrtc:11708
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/178488
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#31606}
SCTP message chunks and notifications are being delivered interleaved.
However, the way the code was structured previously, a notification
would interrupt reassembly of a message chunk and hand out the partial
message, thereby violating message integrity. This patch separates the
handling of notifications and reassembly of messages.
Additional changes:
- Remove illegal cast from non-validated u32 to enum (PPID)
- Drop partial messages if the SID has been changed but EOR not yet
received instead of delivering them. (This should never happen
anyway.)
- Don't treat TSN as timestamp (wat)
Bug: webrtc:11708
Change-Id: I4e2fe2262feda2a96d2ae3f6ce9b06370d9878ae
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/177527
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Taylor <deadbeef@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Tommi <tommi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#31605}
WebRTC doesn't use these features, so disable them to reduce the
potential attack surface.
Bug: webrtc:11694
Change-Id: I093aa824c6da592852270534ae7415ceb19fca47
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/177360
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Commit-Queue: Taylor <deadbeef@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#31540}
We were using the address of the SctpTransport object as
the sconn_addr field in usrsctp, which is used to get access to
the SctpTransport object in various callbacks.
However, this address is sent in the clear in the SCTP cookie,
which is undesirable.
This change uses a monotonically increasing id instead, which
is mapped back to a SctpTransport using a SctpTransportMap helper
class.
Bug: chromium:1076703
Change-Id: Iffb23fdbfa13625e921a9fd5500fe772b4d4015f
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/176422
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Reviewed-by: Tommi <tommi@webrtc.org>
Commit-Queue: Taylor <deadbeef@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#31449}
Tests the behavior of the usrsctp library buffering a large message in
unordered mode. The expected behavior is that this message will be sent
when the socket becomes unblocked, but instead an SCTP_SEND_FAILED_EVENT
is fired by usrsctp library and the message is never sent. This test
will pass with a newer version of usrsctp lib, or if the send is in
ordered mode.
Bug: webrtc:10939
Change-Id: I3b4b05e7dcc7574bf3397991848a9ad7122adc0b
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/172480
Commit-Queue: Taylor <deadbeef@webrtc.org>
Reviewed-by: Seth Hampson <shampson@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30950}
Note that this wasn't actually making a difference since both do the
same thing effectively.
Bug: webrtc:11386
Change-Id: I49d84d363dce12eabeb3770b40abdfdb674a05ac
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/169433
Commit-Queue: Taylor <deadbeef@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30665}
This CL was generated by running:
git ls-files | grep ".cc" | xargs perl -i -ne 'BEGIN {undef $/}; s/("[\s\n]*<<[\s\n]*")/" "/g; print;'; git cl format
After that I manually edited modules/audio_processing/gain_controller2.cc to preserve its original
formatting.
This primary benefit of this change is a small reduction in binary size.
Bug: None
Change-Id: I689fa7ba9c717c314bb167e5d592c3c4e0871e29
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/165961
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Jonas Olsson <jonasolsson@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30251}
This is part of a CL series merging rtc::MessageQueue into rtc::Thread.
Bug: webrtc:9883
Change-Id: I3cb857cc707d5e897759366d1478cc1ec19bce9a
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/165344
Reviewed-by: Stefan Holmer <stefan@webrtc.org>
Commit-Queue: Sebastian Jansson <srte@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30180}
This CL propagates a "closed" signal from DTLS up to the
SCTP section of the data channel controller, where it causes
closing of all open datachannels.
Bug: chromium:1030631, webrtc:10360
Change-Id: I88bb9e1aff5c25f330edfd092ef609d4fcc3a9f8
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/162206
Reviewed-by: Steve Anton <steveanton@webrtc.org>
Commit-Queue: Harald Alvestrand <hta@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#30099}
There is still an TSAN issues, but they are inside
usrsctp library, that's why tests are still disabled
by default.
Bug: None
Change-Id: I55f7c66b4d9a5feccd2121e2dd3b131cf1564804
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/158522
Commit-Queue: Yura Yaroshevich <yura.yaroshevich@gmail.com>
Reviewed-by: Patrik Höglund <phoglund@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29640}
Merge GlobalLock and GlobalLockPod, make member private.
annotate creation of all GlobalLocks with ABSL_CONST_INIT
Bug: None
Change-Id: I29abcc86796ec0e45b15df7d26392309d1bf7324
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/156303
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29447}
Currently we set the EOR bit when sending a message through the sctp
library. This makes the send non atomic, meaning that message can be
partially accepted by the sctp socket. Currently we ignore the sent
amount result, but this change now checks that result and buffers the
remaining message to be sent later in the case that it was only
partially accepted by usrsctp.
Bug: webrtc:10922
Change-Id: I9ff563c40e2b7dbdeb19b40d07c43a15ff7c9b49
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150562
Commit-Queue: Seth Hampson <shampson@webrtc.org>
Reviewed-by: Steve Anton <steveanton@webrtc.org>
Reviewed-by: Amit Hilbuch <amithi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29051}
This involves catching another callback from usrsctp.
It also moves the definition of "connected" a little later
in the sequence: From "ready to send data" to the reception
of the SCTP_COMM_UP event.
Bug: chromium:943976
Change-Id: Ib9e1b17d0cc356f19cdfa675159b29bf1efdcb55
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/137435
Commit-Queue: Harald Alvestrand <hta@webrtc.org>
Reviewed-by: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Steve Anton <steveanton@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28004}
This is a reland of e779847fb6499ac2dc4757de8c625ac377e9d0d4
Original change's description:
> Change SimpleStringBuilder::Append to not use strcpyn and SIZE_UNKNOWN
>
> Also add explicit includes of rtc_base/string_utils.h in files depending on it.
>
> Bug: webrtc:6424
> Change-Id: Id6b53937ab2d185d092a5d8863018fd5f1a88e27
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/135744
> Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
> Commit-Queue: Niels Moller <nisse@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#27903}
Tbr: kwiberg@webrtc.org
Bug: webrtc:6424
Change-Id: Ic08d5d7fbc25ff89e4182d7c9cb3b0e8e356339a
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/135946
Reviewed-by: Niels Moller <nisse@webrtc.org>
Commit-Queue: Niels Moller <nisse@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27957}
This also changes the default when no max-message-size is set
to the protocol defined value of 64K, and prevents messages
from being sent when they are too large to send.
Bug: webrtc:10358
Change-Id: Iacc1dd774d1554d9f27315378fbea6351300b5cc
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/135948
Commit-Queue: Harald Alvestrand <hta@webrtc.org>
Reviewed-by: Steve Anton <steveanton@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27945}
Also add explicit includes of rtc_base/string_utils.h in files depending on it.
Bug: webrtc:6424
Change-Id: Id6b53937ab2d185d092a5d8863018fd5f1a88e27
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/135744
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Niels Moller <nisse@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27903}
Semi-automatically created with:
git grep -l " testing::" | xargs sed -i "s/ testing::/ ::testing::/g"
git grep -l "(testing::" | xargs sed -i "s/(testing::/(::testing::/g"
git cl format
After this, two .cc files failed to compile and I have fixed them
manually.
Bug: webrtc:10523
Change-Id: I4741d3bcedc831b6c5fdc04485678617eb4ce031
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/132018
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Mirko Bonadei <mbonadei@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27526}
This involves inserting an extra layer between jsep_transport_controller
and the cricket::SctpTransportInternal layer. The objects at this layer
are reference counted.
Bug: chromium:818643
Change-Id: Ibed57c4a538de981cee63e0f7f1f319f029cab39
Reviewed-on: https://webrtc-review.googlesource.com/c/123884
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Henrik Boström <hbos@webrtc.org>
Reviewed-by: Steve Anton <steveanton@webrtc.org>
Commit-Queue: Harald Alvestrand <hta@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#26889}
