The idea is to have all frames not in use be stored in free_frames_, and whenever a packet from a new frame arrives we can just pop a frame from free_frames_. When a frame is grabbed for decoding it will be removed from all lists, and will be added to free_frames_ when it's returned to the jitter buffer.
We should be able to remove the state enum completely later, as their state is defined by the list they are in. But I'll keep it around for now to simplify the cl.
TEST=try bots and vie_auto_test --automated
R=mikhal@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/1721004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4273 4adac7df-926f-26a2-2b94-8c16560cd09d
A new test target named 'modules_integrationtests' is created
and the following test targets were merged into it:
* audio_coding_module_test
* test_fec
* video_coding_integrationtests
* vp8_integrationtests
A couple of other targets were merged into modules_unittests:
* audio_coding_unittests
* audioproc_unittest
* common_unittests
* video_coding_unittests
* video_processing_unittests
* vp8_unittests
I wasn't able to merge audio_decoder_unittests and neteq_unittests due to
conflicts with different defines in these tests.
Some tests that have special requirements aren't merged into
modules_integrationtests yet. I took the opportunity to rename them
since the bot configs will need to be update anyway:
* audio_device_test_api -> audio_device_integrationtests
* video_capture_module_test -> video_capture_integrationtests
* video_render_module_test -> video_render_integrationtests
Exclude files were added for modules_integrationtests to make sure
the memcheck and tsan bots doesn't tests that are too slow
(audio_coding_module_test and vp8_integrationtests were previously
disabled on those bots).
Suppressions for AudioCodingModuleTest needed to be added to get
modules_integrationtests to pass memcheck (even if the test is
excluded from execution).
BUG=1843
TEST=local execution on Linux and trybots (passing except the merged tests of course)
R=andrew@webrtc.org, tina.legrand@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/1656004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4228 4adac7df-926f-26a2-2b94-8c16560cd09d
DecodedImageCallback is allowed to be called on a thread different from decoding thread. To avoid the deadlock in VCMDecodedFrameCallback::Decoded, VCMDecodedFrameCallback::_critSect should not be held while calling VCMReceiveCallback::FrameToRender.
BUG=1832
R=stefan@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/1570004
Patch from Wu-Cheng Li <wuchengli@chromium.org>.
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4162 4adac7df-926f-26a2-2b94-8c16560cd09d
Reason - leading suspect of video frame corruption tracked in http://b/9216252
Note that if this turns out to not be the cause, be sure to re-revert both this change and r4145.
> Refactor jitter buffer to use separate lists for decodable and incomplete frames.
>
> This changes the design of the jitter buffer to keeping track of decodable frames from the point when packets are inserted in the buffer, instead of searching for decodable frames when they are needed.
>
> To accomplish this the frame_list_, which previously contained all frames (incomplete or complete, continuous or not), is split into a list of decodable_frames_ (complete, continuous) and a list of incomplete_frames_ (either incomplete or non-continuous). These frame lists are updated every time a packet is inserted.
>
> This is another step in the direction of doing most of the work in the jitter buffer only once, when packets are inserted, instead of doing it every time we look for a frame or try to get a nack list.
>
> BUG=1798
> TEST=vie_auto_test, trybots
> R=mikhal@webrtc.org
>
> Review URL: https://webrtc-codereview.appspot.com/1522005TBR=stefan@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/1586007
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4146 4adac7df-926f-26a2-2b94-8c16560cd09d
This changes the design of the jitter buffer to keeping track of decodable frames from the point when packets are inserted in the buffer, instead of searching for decodable frames when they are needed.
To accomplish this the frame_list_, which previously contained all frames (incomplete or complete, continuous or not), is split into a list of decodable_frames_ (complete, continuous) and a list of incomplete_frames_ (either incomplete or non-continuous). These frame lists are updated every time a packet is inserted.
This is another step in the direction of doing most of the work in the jitter buffer only once, when packets are inserted, instead of doing it every time we look for a frame or try to get a nack list.
BUG=1798
TEST=vie_auto_test, trybots
R=mikhal@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/1522005
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4104 4adac7df-926f-26a2-2b94-8c16560cd09d
This was actually working as intended, but as r3970 changed when render timestamps were extrapolated to when a frame was taken out for decoding, the wraparound could have happened in the Update() step before it had happened in the ExtrapolateLocalTime() step. This causes render timestamps to be generated 13 hours into the future.
TEST=trybots
BUG=1787
R=mflodman@webrtc.org
Review URL: https://webrtc-codereview.appspot.com/1497004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@4055 4adac7df-926f-26a2-2b94-8c16560cd09d
