When screen is zoomed in/out, OSX only updates the parts of Rects currently
displayed on screen, with relative location to current top-left on screen.
This will cause problems when we copy the dirty regions to the captured
frame. So we invalidate the whole screen to copy all the screen contents.
- With CGI method, the zooming will be ignored and the whole screen contents
will be captured as before.
- With IOSurface method, the zoomed screen contents will be captured.
Since we can't know the zooming level and focusing location, so we have
to copy the whole screen region for each frame during rooming. And this
will impact peformance a bit (with IOSurface capturer about 5-10 fps
down on MBP.)
Bug: chromium:911862
Change-Id: Icf123cde4d686ab7ce28fa731bc8dac6925492c8
Reviewed-on: https://webrtc-review.googlesource.com/c/113101
Reviewed-by: Jamie Walch <jamiewalch@chromium.org>
Commit-Queue: Brave Yao <braveyao@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25936}
This class exposes Wait()-Set() logic to synchronize events.
- There is a bug in checking EventWrapper::Wait() as it returns [1,2]. Negating
these values cause us to always pass timeout checks.
- There is a general problem in this class with waiter. There are 2 scenarios:
1) Lock()-Unlock()-DisplaysReconfigured()
In this scenario, Wait() in DisplaysReconfigured() immediately passes as event
is already signaled. Next Lock() call won't continue until Set() is called in
DisplaysReconfigured(). This blocks capture thread from accessing display until
reconfiguration completes.
2) Lock()-DisplaysReconfigured()-Unlock()
In this scenario, Wait() in DisplaysReconfigured() passes when Unlock() called.
Capture thread accesses display while reconfiguration happens. Note that we are
only delaying the OS delegate thread here. As an experiment, adding Sleep() in
DisplaysReconfigured() results in no change, because it looks like OS uses this
thread for only delegates but not for the actual display switch.
Overall, (1) doesnt seem necessary as (2) already accesses display while
reconfiguration happens. (2) doesn't seem necessary as blocking system delegate
thread doesn't help. Therefore, I changed the class to only protect from race
condition on |desktop_configuration_|.
Bug: chromium:796889
Change-Id: I37263305e5ac629e21ff9e977952cf4a21bae19f
Reviewed-on: https://webrtc-review.googlesource.com/c/108560
Commit-Queue: Emircan Uysaler <emircan@webrtc.org>
Reviewed-by: Tommi <tommi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#25437}
Thanks to the CL https://webrtc-review.googlesource.com/c/src/+/83822
there is now no need to destroy the stream asynchronously.
But the CL above introduced a leak, the streams were stopped but
not destroyed. This CL essentially fixes the leak, it is now safe
to destroy the stream right after being stopped as everything happen
in the same capture thread.
Bug: chromium:851883
Change-Id: I4bf7409246f3957d90040d0d8cf09e98f28d6559
Reviewed-on: https://webrtc-review.googlesource.com/96621
Reviewed-by: Brave Yao <braveyao@webrtc.org>
Reviewed-by: Zijie He <zijiehe@chromium.org>
Commit-Queue: Brave Yao <braveyao@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24519}
This CL separates the files under sdk/objc into logical directories, replacing
the previous file layout under Framework/.
A long term goal is to have some system set up to generate the files under
sdk/objc/api (the PeerConnection API wrappers) from the C++ code. In the shorter
term the goal is to abstract out shared concepts from these classes in order to
make them as uniform as possible.
The separation into base/, components/, and helpers/ are to differentiate between
the base layer's common protocols, various utilities and the actual platform
specific components.
The old directory layout that resembled a framework's internal layout is not
necessary, since it is generated by the framework target when building it.
Bug: webrtc:9627
Change-Id: Ib084fd83f050ae980649ca99e841f4fb0580bd8f
Reviewed-on: https://webrtc-review.googlesource.com/94142
Reviewed-by: Kári Helgason <kthelgason@webrtc.org>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Rasmus Brandt <brandtr@webrtc.org>
Reviewed-by: Henrik Andreassson <henrika@webrtc.org>
Commit-Queue: Anders Carlsson <andersc@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24493}
This cl is to move the RegisterRefreshAndMoveHandlers to be done on
capture thread, and reverse some execution order of releasing processing,
also remove a lock since the handler is on capturing thread too.
As we doubt the existing sequence may be the cause of a crash due to
race conditions at end of capture.
Bug: chromium:851883
Change-Id: I2254a69815144415424a77b4c82f150cfc369585
Reviewed-on: https://webrtc-review.googlesource.com/83822
Commit-Queue: Brave Yao <braveyao@webrtc.org>
Reviewed-by: Zijie He <zijiehe@chromium.org>
Cr-Commit-Position: refs/heads/master@{#23647}
This is a reland of 89653d5db46419d2a80898635cb27fed64898db2
Original change's description:
> [desktopCapture] Unify the position info in DIP coordinates on Mac.
>
> On OSX, the logical(DIP) and physical coordinates are used mixingly.
> For example, the captured image has its size in physical pixels(2x) and
> location in logical(DIP) pixels. Same to the cursor position. This
> causes trouble when we check the relative position of image and cursor
> when there are multiple monitors with different DIP setting connected.
>
> This cl proposed a solution to use DIP pixel for any location info,
> i.e. top-left of a frame and cursor position. Also propose a method to
> get the current scale factor of a window across multiple monitors. And
> save the current scale factor in DPI of the capture frame.
> Then we can check relative position of cursor and frame correctly
> in DIP pixel and compose them in physical pixel.
>
> Bug: webrtc:9178
> Change-Id: I3c076aeac2d6f2c1f63d000d7fff03500aa375ac
> Reviewed-on: https://webrtc-review.googlesource.com/71621
> Reviewed-by: Jamie Walch <jamiewalch@chromium.org>
> Reviewed-by: Zijie He <zijiehe@chromium.org>
> Commit-Queue: Brave Yao <braveyao@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#23263}
Bug: webrtc:9178
Change-Id: I97d9150f7b9a4ed6671733b75613ea9c315d5c1d
Reviewed-on: https://webrtc-review.googlesource.com/77481
Reviewed-by: Zijie He <zijiehe@chromium.org>
Commit-Queue: Brave Yao <braveyao@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23289}
This reverts commit 89653d5db46419d2a80898635cb27fed64898db2.
Reason for revert:
Tentatively revert since I believe this break remoting unittests on Asan/Tsan
https://chromium-review.googlesource.com/c/chromium/src/+/1063330https://chromium-swarm.appspot.com/task?id=3d8692bedcc85c10&refresh=10&show_raw=1
Original change's description:
> [desktopCapture] Unify the position info in DIP coordinates on Mac.
>
> On OSX, the logical(DIP) and physical coordinates are used mixingly.
> For example, the captured image has its size in physical pixels(2x) and
> location in logical(DIP) pixels. Same to the cursor position. This
> causes trouble when we check the relative position of image and cursor
> when there are multiple monitors with different DIP setting connected.
>
> This cl proposed a solution to use DIP pixel for any location info,
> i.e. top-left of a frame and cursor position. Also propose a method to
> get the current scale factor of a window across multiple monitors. And
> save the current scale factor in DPI of the capture frame.
> Then we can check relative position of cursor and frame correctly
> in DIP pixel and compose them in physical pixel.
>
> Bug: webrtc:9178
> Change-Id: I3c076aeac2d6f2c1f63d000d7fff03500aa375ac
> Reviewed-on: https://webrtc-review.googlesource.com/71621
> Reviewed-by: Jamie Walch <jamiewalch@chromium.org>
> Reviewed-by: Zijie He <zijiehe@chromium.org>
> Commit-Queue: Brave Yao <braveyao@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#23263}
TBR=zijiehe@chromium.org,jamiewalch@chromium.org,perkj@webrtc.org,braveyao@webrtc.org
Change-Id: Ica02365925623e21b256d20a21b5625e7ed6f49b
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: webrtc:9178
Reviewed-on: https://webrtc-review.googlesource.com/77461
Reviewed-by: Per Kjellander <perkj@webrtc.org>
Commit-Queue: Per Kjellander <perkj@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23280}
On OSX, the logical(DIP) and physical coordinates are used mixingly.
For example, the captured image has its size in physical pixels(2x) and
location in logical(DIP) pixels. Same to the cursor position. This
causes trouble when we check the relative position of image and cursor
when there are multiple monitors with different DIP setting connected.
This cl proposed a solution to use DIP pixel for any location info,
i.e. top-left of a frame and cursor position. Also propose a method to
get the current scale factor of a window across multiple monitors. And
save the current scale factor in DPI of the capture frame.
Then we can check relative position of cursor and frame correctly
in DIP pixel and compose them in physical pixel.
Bug: webrtc:9178
Change-Id: I3c076aeac2d6f2c1f63d000d7fff03500aa375ac
Reviewed-on: https://webrtc-review.googlesource.com/71621
Reviewed-by: Jamie Walch <jamiewalch@chromium.org>
Reviewed-by: Zijie He <zijiehe@chromium.org>
Commit-Queue: Brave Yao <braveyao@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23263}
Always having the latest iosurface before invalidating a region.
Otherwise if CaptureFrame() happens in between, the capture result
may not be fully refreshed. Also we can't add lock since it will
impact performance.
Bug: webrtc:8652
Change-Id: Ib23105b16065018c691685083b76a771ce8771d3
Reviewed-on: https://webrtc-review.googlesource.com/74643
Reviewed-by: Zijie He <zijiehe@chromium.org>
Commit-Queue: Brave Yao <braveyao@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#23154}
While investigating some screen-capture-track-end-in-meeting issues, the
relevant rtc error logs are not uploaded to server as other webrtc
modules do, which cause great hardness to identify the reason.
This cl is to use existing trace event methods to store error logs of
desktop capturers.
Bug: chromium:831756
Change-Id: Id0c1b439f9b63916fb9417cf4e6f2b8f3c556fcd
Reviewed-on: https://webrtc-review.googlesource.com/69783
Reviewed-by: Jamie Walch <jamiewalch@chromium.org>
Commit-Queue: Brave Yao <braveyao@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#22866}
The given IOSurfaceRef was ignored until now. Wrap it into the new
DesktopFrameIOSurface. The new DesktopFrameProvider object is there
to manage them as it has to be done per display id.
From initial measurement this speed-up the frame capture by 2.
Disabled by default for now but it can be enabled by calling
options.set_use_iosurface. This CL will allow to do some advanced
tests.
Bug: webrtc:8652
Change-Id: Ia9ac0b69b30098774941cb378804b45cb1710119
Reviewed-on: https://webrtc-review.googlesource.com/33014
Commit-Queue: Zijie He <zijiehe@chromium.org>
Reviewed-by: Zijie He <zijiehe@chromium.org>
Cr-Commit-Position: refs/heads/master@{#22801}
No functional change. This makes the code more generic
and this reduces the size of screen_capturer_mac.mm
Bug: webrtc:8652
Change-Id: I37743ba385fea5e1b40df3b094bfc321b8d796ae
Reviewed-on: https://webrtc-review.googlesource.com/65082
Commit-Queue: Julien Isorce <julien.isorce@chromium.org>
Commit-Queue: Zijie He <zijiehe@chromium.org>
Reviewed-by: Zijie He <zijiehe@chromium.org>
Cr-Commit-Position: refs/heads/master@{#22684}
- Remove unsued ScopedPixelBufferObject that was used for the
capture using OpenGL.
- Also remove InvertedDesktopFrame for the same reason.
- Replace several occurrences of assert by RTC_DCHECK
Bug: webrtc:8652
Change-Id: I262db0a445f2211cde7476a6cadfb1c19a3e814f
Reviewed-on: https://webrtc-review.googlesource.com/64883
Commit-Queue: Zijie He <zijiehe@chromium.org>
Reviewed-by: Zijie He <zijiehe@chromium.org>
Cr-Commit-Position: refs/heads/master@{#22632}
1- git mv screen_capturer_mac.mm mac/screen_capturer_mac.mm
2- extract class ScreenCapturerMac declaritions to its own header
3- extract static CreateRawScreenCapturer to screen_capturer_darwin.mm
(Using 'darwin' instead of 'mac' allows to make happy the command
git log --follow mac/screen_capturer_mac.mm)
4- git cl format
Bug: webrtc:8652
Change-Id: Ibb13bd5dec61aa9b92c9f5f30fedd0508a727dd9
Reviewed-on: https://webrtc-review.googlesource.com/64680
Commit-Queue: Julien Isorce <julien.isorce@chromium.org>
Reviewed-by: Zijie He <zijiehe@chromium.org>
Cr-Commit-Position: refs/heads/master@{#22620}
