The v4l2_sink implementation directly read the internal video_buffer
field "pending_frame_consumed", which is protected by the internal
video_buffer mutex. But this mutex was not locked, so reads were racy.
Lock using the v4l2_sink mutex in addition, and use a separate field to
avoid depending on the video_buffer internal data.
Commit 21d206f360 added mutex assertions.
However, the "locker" variable to trace the locker thread id was read
and written by several threads without any protection, so it was racy.
Reported by TSAN.
The options --no-display and --no-control are independent.
The controller was not initialized when no display was requested,
because it was assumed that no control could occur without display. But
that's not true (anymore): for example, it is possible to pass
--turn-screen-off.
Fixes#2426 <https://github.com/Genymobile/scrcpy/issues/2426>
Mouse motion events were forwarded as soon as any mouse button was
pressed.
Instead, only consider left-click (and also middle-click and right-click
if --forward-all-clicks is enabled).
Change the default push target from /sdcard/ to /sdcard/Download/.
Pushing to the root of /sdcard/ is not very convenient, many apps do not
expose its content directly.
It can still be changed by --push-target.
PR #2384 <https://github.com/Genymobile/scrcpy/pull/2384>
When removing the black borders (by double-clicking on them, or by
pressing MOD+w), the window is resized to fit the device screen, but its
top-left position was left unchanged.
Instead, move the window so that the new window area is at the center of
the old window area.
Refs #2387 <https://github.com/Genymobile/scrcpy/issues/2387>
It should not be necessary, since screen_render() is called just after
on SDL_WINDOWEVENT_EXPOSED, but in practice the window content might not
be correctly displayed on restored if a rotation occurred while
minimized.
Note that calling screen_render() twice in a row on
SDL_WINDOWEVENT_EXPOSED also "fixes" the issue.
From FFmpeg/doc/APIchanges:
2021-03-17 - f7db77bd87 - lavc 58.133.100 - codec.h
Deprecated av_init_packet(). Once removed, sizeof(AVPacket) will
no longer be a part of the public ABI.
Refs #2302 <https://github.com/Genymobile/scrcpy/issues/2302>
From FFmpeg/doc/APIchanges:
2021-03-17 - f7db77bd87 - lavc 58.133.100 - codec.h
Deprecated av_init_packet(). Once removed, sizeof(AVPacket) will
no longer be a part of the public ABI.
Refs #2302 <https://github.com/Genymobile/scrcpy/issues/2302>
From FFmpeg/doc/APIchanges:
2021-03-17 - f7db77bd87 - lavc 58.133.100 - codec.h
Deprecated av_init_packet(). Once removed, sizeof(AVPacket) will
no longer be a part of the public ABI.
Refs #2302 <https://github.com/Genymobile/scrcpy/issues/2302>
From FFmpeg/doc/APIchanges:
2021-03-17 - f7db77bd87 - lavc 58.133.100 - codec.h
Deprecated av_init_packet(). Once removed, sizeof(AVPacket) will
no longer be a part of the public ABI.
Remove the has_pending boolean, which can be replaced by:
stream->pending != NULL
Refs #2302 <https://github.com/Genymobile/scrcpy/issues/2302>
The argument for option --lock-video-orientation has been made optional
by 5af9d0ee0f.
With getopt_long(), contrary to mandatory arguments, optional arguments
must be given with a '=':
--lock-video-orientation 2 # wrong, parse error
--lock-video-orientation=2 # correct
The input manager was partially initialized statically, but a call to
input_manager_init() was needed anyway, so initialize all the fields
from the "constructor".
This is consistent with the initialization of the other structs.
The frame can be unref immediately after it is pushed to the frame
sinks.
It was not really a memory leak because the frame was unref every time
by avcodec_receive_frame() (and freed on close), but a reference was
unnecessarily kept for too long.
Add a new mode to the --lock-video-orientation option, to lock the
initial orientation of the device.
This avoids to pass an explicit value (0, 1, 2 or 3) and think about
which is the right one.
The screen may not be destroyed immediately on close to avoid undefined
behavior, because it may still receive events from the decoder.
But the visual window must still be closed immediately.
The destruction order is important, but tricky, because the screen is
open/close by the decoder, but destroyed by scrcpy.c on the main thread.
Add assertions to guarantee that the screen is not destroyed before
being closed.
The video buffer is now an internal detail of the screen component.
Since the screen is plugged to the decoder via the frame sink trait, the
decoder does not access to the video buffer anymore.
The fact that the recorder uses a separate thread is an internal detail,
so the functions _start(), _stop() and _join() should not be exposed.
Instead, start the thread on _open() and _stop()+_join() on close().
This paves the way to expose the recorder as a packet sink trait.
The video buffer took ownership of the producer frame (so that it could
swap frames quickly).
In order to support multiple sinks plugged to the decoder, the decoded
frame must not be consumed by the display video buffer.
Therefore, move the producer and consumer frames out of the video
buffer, and use FFmpeg AVFrame refcounting to share ownership while
avoiding copies.
This flag forced the decoder to wait for the previous frame to be
consumed by the display.
It was initially implemented as a compilation flag for testing, not
intended to be exposed at runtime. But to remove ifdefs and to allow
users to test this flag easily, it had finally been exposed by commit
ebccb9f6cc.
In practice, it turned out to be useless: it had no practical impact,
and it did not solve or mitigate any performance issues causing frame
skipping.
But that added some complexity to the codebase: it required an
additional condition variable, and made video buffer calls possibly
blocking, which in turn required code to interrupt it on exit.
To prepare support for multiple sinks plugged to the decoder (display
and v4l2 for example), the blocking call used for pacing the decoder
output becomes unacceptable, so just remove this useless "feature".
Double-click on extra mouse button to open the settings panel (a
single-click opens the notification panel).
This is consistent with the keyboard shortcut MOD+n+n.
PR #2264 <https://github.com/Genymobile/scrcpy/pull/2264>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The collapsing action collapses any panels.
By the way, the Android method is named collapsePanels().
PR #2260 <https://github.com/Genymobile/scrcpy/pull/2260>
Signed-off-by: Romain Vimont <rom@rom1v.com>
The shortcut "back on screen on" is a bit special: the control is
requested by the client, but the actual event injection (POWER or BACK)
is determined on the device.
To properly inject DOWN and UP events for BACK, transmit the action as
a control parameter.
If the screen is off:
- on DOWN, inject POWER (DOWN and UP) (wake up the device immediately)
- on UP, do nothing
If the screen is on:
- on DOWN, inject BACK DOWN
- on UP, inject BACK UP
A corner case is when the screen turns off between the DOWN and UP
event. In that case, a BACK UP event will be injected, so it's harmless.
As a consequence of this change, the BACK button is now handled by
Android on mouse released. This is consistent with the keyboard shortcut
(Mod+b) behavior.
PR #2259 <https://github.com/Genymobile/scrcpy/pull/2259>
Refs #2258 <https://github.com/Genymobile/scrcpy/pull/2258>
The screen receives callbacks from the decoder, fed by the stream.
The decoder is run from the stream thread, so waiting for the end of
stream is sufficient to avoid possible use-after-destroy.
When --no-display was passed, screen_destroy() was called while
screen_init() was never called.
In practice, it did not crash because it just freed NULL pointers, but
it was still incorrect.
A skipped frame is detected when the producer offers a frame while the
current pending frame has not been consumed.
However, the producer (in practice the decoder) is not interested in the
fact that a frame has been skipped, only the consumer (the renderer) is.
Therefore, notify frame skip via a consumer callback. This allows to
manage the skipped and rendered frames count at the same place, and
remove fps_counter from decoder.
As soon as the stream is started, the video buffer could notify a new
frame available.
In order to pass this event to the screen without race condition, the
screen must be initialized before the screen is started.
Video buffer is a tool between a frame producer and a frame consumer.
For now, it is used between a decoder and a renderer, but in the future
another instance might be used to swscale decoded frames.
It makes sense to extract default values for bitrate and port range
(which are arbitrary and might be changed in the future).
However, the default values for "max size" and "lock video orientation"
are naturally unlimited/unlocked, and will never be changed. Extracting
these options just added complexity for no benefit, so hardcode them.
After the struct screen is initialized, the window, the renderer and the
texture are necessarily valid, so there is no need to check in
screen_destroy().
There were only two frames simultaneously:
- one used by the decoder;
- one used by the renderer.
When the decoder finished decoding a frame, it swapped it with the
rendering frame.
Adding a third frame provides several benefits:
- the decoder do not have to wait for the renderer to release the
mutex;
- it simplifies the video_buffer API;
- it makes the rendering frame valid until the next call to
video_buffer_take_rendering_frame(), which will be useful for
swscaling on window resize.
The functions SDL_malloc(), SDL_free() and SDL_strdup() were used only
because strdup() was not available everywhere.
Now that it is available, use the native version of these functions.
Small unsigned integers promote to signed int. As a consequence, if v is
a uint8_t, then (v << 24) yields an int, so the left shift is undefined
if the MSB is 1.
Cast to uint32_t to yield an unsigned value.
Reported by USAN (meson x -Db_sanitize=undefined):
runtime error: left shift of 255 by 24 places cannot be represented
in type 'int'
The current process could be waited both by run_file_handler() and
file_handler_stop().
To avoid the race condition, wait the process without closing, then
close with mutex locked.
There were two versions: process_wait() and process_wait_noclose().
Expose a single version with a flag (it was already implemented that way
internally).
The function process_wait() returned a bool (true if the process
terminated successfully) and provided the exit code via an output
parameter exit_code.
But the returned value was always equivalent to exit_code == 0, so just
return the exit code instead.
The size, point and position structs were defined in common.h. Move them
to coords.h so that common.h could be used for generic code to be
included in all source files.
The header libavformat/version.h was included, but not
libavcodec/version.h.
As a consequence, the LIBAVCODEC_VERSION_INT definition depended on the
caller includes.
On Linux, waitpid() both waits for the process to terminate and reaps it
(closes its handle). On Windows, these actions are separated into
WaitForSingleObject() and CloseHandle().
Expose these actions separately, so that it is possible to send a signal
to a process while waiting for its termination without race condition.
This allows to wait for server termination normally, but kill the
process without race condition if it is not terminated after some delay.
Let the server terminate properly once all the sockets are closed.
If it does not terminate (this can happen if the device is asleep), then
kill it.
Note: since the server process termination is detected by a flag set
after waitpid() returns, there is a small chance that the process
terminates (and the PID assigned to a new process) before the flag is
set but before the kill() call. This race condition already existed
before this commit.
Fixes#1992 <https://github.com/Genymobile/scrcpy/issues/1992>
TerminateProcess() is "equivalent" to kill(), while
WaitForSingleObject() is "equivalent" to waitpid(), so the handle must
be closed after WaitForSingleObject().
On Windows, scrcpy paused on error before exiting to give the user a
chance to see the user message.
This was a hack and causes issues when using scrcpy from batch scripts.
Disable this pause from the scrcpy binary, and provide a batch wrapper
(scrcpy-console.bat) to pause on error.
Fixes#1875 <https://github.com/Genymobile/scrcpy/issues/1875>
Use "%Iu" on Windows. This fixes the following warning:
../app/src/sys/win/command.c:17:14: warning: unknown conversion type character ‘l’ in format [-Wformat=]
17 | LOGE("Command too long (%" PRIsizet " chars)", len - 1);
A new "repeat" field has been added by
3c1ed5d86c, but it was not initialized in
every code path.
As a consequence, keycodes generated by shortcuts were sent with an
undetermined value, breaking some shortcuts (especially HOME) randomly.
Fixes#1643 <https://github.com/Genymobile/scrcpy/issues/1643>
This avoids to pass specific options values individually. Since these
function are static (internal to the file), this is not a problem to
make them depend on scrcpy_options.
Refs #1623 <https://github.com/Genymobile/scrcpy/pull/1623>
Signed-off-by: Romain Vimont <rom@rom1v.com>
Send COPY and CUT on MOD+c and MOD+x (only supported for Android >= 7).
The shortcuts Ctrl+c and Ctrl+x should generally also work (even before
Android 7), but the active Android app may use them for other actions
instead.
Do not explicitly set the clipboard text if it already contains the
expected content.
Even if copy-paste loops are avoided by the previous commit, this avoids
to trigger a clipboard change on the computer-side.
Refs #1580 <https://github.com/Genymobile/scrcpy/issues/1580>
Pressing Ctrl+v on the device will typically paste the clipboard
content.
Before sending the key event, synchronize the computer clipboard to the
device clipboard to allow seamless copy-paste.
Now that the scrcpy shortcut modifier is Alt by default (and can be
configured), forward Ctrl to the device.
This allows to trigger Android shortcuts.
Fixes#555 <https://github.com/Genymobile/scrcpy/issues/555>
Pressing Alt+c generates a text event containing "c", so "c" was sent to
the device when --prefer-text was enabled.
Ignore text events when the mod state matches a shortcut modifier.
Remove the Cmd modifier on macOS, which was possible only for some
shortcuts but not all.
This paves the way to make the shortcut modifier customizable.
Touch events were HiDPI-scaled twice:
- once because the position (provided as floats between 0 and 1) were
converted in pixels using the drawable size (not the window size)
- once due to screen_convert_to_frame_coords()
One possible fix could be to compute the position in pixels from the
window size instead, but this would unnecessarily round the event
position to the nearest window coordinates (instead of drawable
coordinates).
Instead, expose two separate functions to convert to frame coordinates
from either window or drawable coordinates.
Fixes#1536 <https://github.com/Genymobile/scrcpy/issues/1536>
Refs #15 <https://github.com/Genymobile/scrcpy/issues/15>
Refs e40532a376
The header scrcpy.h is intended to be the "public" API. It should not
depend on other internal headers.
Therefore, declare all required structs in this header and adapt
internal code.
