UDP case should not be changed.
Active TCPConnection will initiate Reconnect after OnClose and when Send or Ping fails.
Passive TCPConnection will prune itself as usual as the active side will create a new connection.
The Reconnect could make P2PCT choose a different best_connection in the case where connectivities exist b/w more than 1 Network.
Also, to avoid upper layer triggers ice restart, the WRITE_TIMEOUT caused by the socket disconnection is delayed to give the reconnect mechanism chance to kick in. The timeout event is only fired if the reconnect can't work in 5 sec. If the reconnect, there should be no ICE disconnected state trigger either in active or passive side.
BUG=1926
R=pthatcher@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/31359004
Cr-Commit-Position: refs/heads/master@{#8929}
This cl uses the YV12 buffers from Java without a copy if no rotation is needed. Buffers are returned to the camera when the encoder and renderers no longer needs them.
This add a new frame type WrappedI420Buffer based in that allows for wrapping existing memory buffers and getting a notification when it is no longer used.
AndroidVideoCapturer::FrameFactory::CreateAliasedFrame wraps frame received from Java. For each wrapped frame a new reference to AndroidVideoCapturerDelegate is held to ensure that the delegate can not be destroyed until all frames have been returned.
Some overlap exist in webrtcvideoframe.cc and webrtcvideengine.cc with https://webrtc-codereview.appspot.com/47399004/ that is expected to be landed before this cl.
BUG=1128
R=glaznev@webrtc.org, magjed@webrtc.orgTBR=mflodman@webrtc.org // For changes in webrtc/common_video/video_frame_buffer
Review URL: https://webrtc-codereview.appspot.com/49459004
Cr-Commit-Position: refs/heads/master@{#8923}
Instead just use MD5Context type directly. In C++ it is unnecessary to
alias the types using typedef, unline C (where if you don't you have to
spell out struct or enum infront of the user-type everytime you want to make a
variable).
So since WebRTC's base API is C++, it seems unnecessay to keep this
typedef around.
BUG=None
TEST=rtc_unittests --gtest_filter=Md5*
R=tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/46799004
Patch from Thiago Farina <tfarina@chromium.org>.
Cr-Commit-Position: refs/heads/master@{#8916}
And add a constructor for creating an uninitialized Buffer of a
specified size.
(I intend to follow up with more Buffer changes, but since it's rather
widely used, the rename is quite noisy and works better as a separate
CL.)
R=tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/48579004
Cr-Commit-Position: refs/heads/master@{#8841}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8841 4adac7df-926f-26a2-2b94-8c16560cd09d
Mostly this consists of marking functions with override when
applicable, and moving function bodies from .h to .cc files.
Not inlining virtual functions with simple bodies such as
{ return false; }
strikes me as probably losing more in readability than we gain in
binary size and compilation time, but I guess it's just like any other
case where enabling a generally good warning forces us to write
slightly worse code in a couple of places.
BUG=163
R=kjellander@webrtc.org, tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/47429004
Cr-Commit-Position: refs/heads/master@{#8656}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8656 4adac7df-926f-26a2-2b94-8c16560cd09d
The problem we were running into on the Mac 10.9 debug bot in Chrome turned out to be good ol'fashion memory corruption. Part of webrtc was being compiled with _DEBUG, another half without it. This caused the definition of some symbols to be out of sync (notably pthread_mutex_t) and would cause code built from common.gypi, to overwrite memory allocated via common types from base/base.gypi derived code. Fun stuff to track down. This was a problem in particular with base/criticalsection.h since it's inlined into multiple object files but will have different definitions of what a mutex is.
TBR=pbos,kjellander
BUG=
Review URL: https://webrtc-codereview.appspot.com/43659004
Cr-Commit-Position: refs/heads/master@{#8646}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8646 4adac7df-926f-26a2-2b94-8c16560cd09d
Local testing indicates that the pthread_t member variable might be causing alignment problems on the Chromium bots. After landing this (and once the Chromium tree is open again), I'll try a roll again to see if this has an effect.
R=magjed@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/48449004
Cr-Commit-Position: refs/heads/master@{#8644}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8644 4adac7df-926f-26a2-2b94-8c16560cd09d
Clang version changed 223108:230914
Details: e144d30..6fdb142/tools/clang/scripts/update.sh
Removes the OVERRIDE macro defined in:
* webrtc/base/common.h
* webrtc/typedefs.h
The majority of the source changes were done by running this in src/:
perl -0pi -e "s/virtual\s([^({;]*(\([^({;]*\)[^({;]*))(OVERRIDE|override)/\1override/sg" `find {talk,webrtc} -name "*.h" -o -name "*.cc*" -o -name "*.mm*"`
which converted all:
virtual Foo() OVERRIDE
functions to:
Foo() override
Then I manually edited:
* talk/media/webrtc/fakewebrtccommon.h
* webrtc/test/fake_common.h
Remaining uses of OVERRIDE was fixed by search+replace.
Manual edits were done to fix virtual destructors that were
overriding inherited ones.
Finally a build error related to the pure virtual definitions of
Read, Write and Rewind in common_types.h required a bit of
refactoring in:
* webrtc/common_types.cc
* webrtc/common_types.h
* webrtc/system_wrappers/interface/file_wrapper.h
* webrtc/system_wrappers/source/file_impl.cc
This roll should make it possible for us to finally re-enable deadlock
detection for TSan on the buildbots.
BUG=4106
R=pbos@webrtc.org, tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/41069004
Cr-Commit-Position: refs/heads/master@{#8596}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8596 4adac7df-926f-26a2-2b94-8c16560cd09d
Currently, I420VideoFrame uses three webrtc::Plane to store pixel data, and WebRtcVideoFrame uses WebRtcVideoFrame::FrameBuffer/webrtc::VideoFrame. The two subclasses WebRtcTextureVideoFrame and TextureVideoFrame use a NativeHandle to store pixel data, and there is also a class WebRtcVideoRenderFrame that wraps an I420VideoFrame.
This CL replaces these classes with a new interface VideoFrameBuffer that provides the common functionality. This makes it possible to remove deep frame copies between cricket::VideoFrame and I420VideoFrame.
Some additional minor changes are:
* Disallow creation of 0x0 texture frames.
* Remove the half-implemented ref count functions in I420VideoFrame.
* Remove the Alias functionality in WebRtcVideoFrame
The final goal is to eliminate all frame copies, but to limit the scope of this CL, some planned changes are postponed to follow-up CL:s (see planned changes in https://webrtc-codereview.appspot.com/38879004, or https://docs.google.com/document/d/1bxoJZNmlo-Z9GnQwIaWpEG6hDlL_W-bzka8Zb_K2NbA/preview). Specifically, this CL:
* Keeps empty subclasses WebRtcTextureVideoFrame and TextureVideoFrame, and just delegates the construction to the superclass.
* Keeps the deep copies from cricket::VideoFrame to I420VideoFrame.
BUG=1128
R=mflodman@webrtc.org, pbos@webrtc.org, perkj@webrtc.org, tommi@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/42469004
Cr-Commit-Position: refs/heads/master@{#8580}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8580 4adac7df-926f-26a2-2b94-8c16560cd09d
> Ensure only temporary IPv6 address is selected as the best IP.
>
> The current logic of IPv6 selection could still have a small chance for non-temporary address to be selected for candidate. The scenario is that when there is no non-deprecated temporary IP, the global ones could be selected.
>
> Global ones don't necessarily carry MAC. However, instead of comparing whether it has the MAC in it (sometimes 5 out of 6 elements from a MAC are the same, only one diffs), we should just err on the safe side.
>
> BUG=4348
> R=juberti@webrtc.org, pthatcher@webrtc.org
>
> Review URL: https://webrtc-codereview.appspot.com/38289004TBR=guoweis@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/38319004
Cr-Commit-Position: refs/heads/master@{#8534}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8534 4adac7df-926f-26a2-2b94-8c16560cd09d
The current logic of IPv6 selection could still have a small chance for non-temporary address to be selected for candidate. The scenario is that when there is no non-deprecated temporary IP, the global ones could be selected.
Global ones don't necessarily carry MAC. However, instead of comparing whether it has the MAC in it (sometimes 5 out of 6 elements from a MAC are the same, only one diffs), we should just err on the safe side.
BUG=4348
R=juberti@webrtc.org, pthatcher@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/38289004
Cr-Commit-Position: refs/heads/master@{#8532}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8532 4adac7df-926f-26a2-2b94-8c16560cd09d
Mostly, it's about moving constructors and descructors to the .cc
files, so that they won't be inlined everywhere.
The reason this CL is so big is that a lot of code was using
common_types.h without declaring a dependency on webrtc_common, which
broke the build once common_types.h started to depend on
common_types.cc.
BUG=163
R=kjellander@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/26089004
Cr-Commit-Position: refs/heads/master@{#8516}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8516 4adac7df-926f-26a2-2b94-8c16560cd09d
WebRTC binds to individual NICs and listens for incoming Stun packets. Sending stun through this specific NIC binding could make OS route the packet differently hence exposing non-VPN public IP.
The fix here is
1. to bind to any address (0:0:0:0) instead. This way, the routing will be the same as how chrome/http is.
2. also, remove the any all 0s addresses which happens when we bind to all 0s.
BUG=4276
R=juberti@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/39129004
Cr-Commit-Position: refs/heads/master@{#8418}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8418 4adac7df-926f-26a2-2b94-8c16560cd09d
VirtualSocketServer, when binding to any address (all 0s), will assign a unique IP address by incrementing the IP address, resulted in 0.0.0.1. However, this breaks the testing of 4276 where we bind to all 0s and expect the local address should remain all 0s.
BUG=4276
R=pthatcher@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/35189004
Cr-Commit-Position: refs/heads/master@{#8370}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8370 4adac7df-926f-26a2-2b94-8c16560cd09d
PeerConnectionTest.java currently works, but only on a device with
network interfaces up. This is not a problem for desktop, but it is a
problem when running on Android devices since the devices in the lab
generally don't have network (due to the chaotic radio environment in
the device labs, devices are simply kept in flight mode).
The test does work if one modifies this line in the file
webrtc/base/network.cc:
bool ignored = ((cursor->ifa_flags & IFF_LOOPBACK) ||
IsIgnoredNetwork(*network));
If we remove the IFF_LOOPBACK clause, the test starts working on
an Android device in flight mode. This is nice - we're running the
call and packets interact with the OS network stack, which is good
for this end-to-end test. We can't just remove the clause though since
having loopback is undesirable for everyone except the test (right)?
so we need to make this behavior configurable.
This CL takes a stab at a complete solution where we pass a boolean
all the way through the Java PeerConnectionFactory down to the
BasicNetworkManager. This comes as a heavy price in interface
changes though. It's pretty out of proportion, but fundamentally we
need some way of telling the network manager that it is on Android
and in test mode. Passing the boolean all the way through is one way.
Another way might be to put the loopback filter behind an ifdef and
link a custom libjingle_peerconnection.so with the test. That is hacky
but doesn't pollute the interfaces. Not sure how to solve that in GYP
but it could mean some duplication between the production and
test .so files.
It would have been perfect to use flags here, but then we need to
hook up gflags parsing to some main() somewhere to make sure the
flag gets parsed, and make sure to pass that flag in our tests.
I'm not sure how that can be done.
Making the loopback filtering conditional is exactly how we solved the
equivalent problem in content_browsertests in Chrome, and it worked
great.
That's all I could think of.
BUG=4181
R=perkj@webrtc.org, pthatcher@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/36769004
Cr-Commit-Position: refs/heads/master@{#8344}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8344 4adac7df-926f-26a2-2b94-8c16560cd09d
